Gene expression in response to endoplasmic reticulum stress in Arabidopsis thaliana

Authors


  • Database
    The nucleotide sequence data for soybean SEL-1L are available in the DDBJ/EMBL/GenBank databases under accession number AB197676.

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

Abstract

Eukaryotic cells respond to the accumulation of unfolded proteins in the endoplasmic reticulum (ER). In this case, so-called unfolded protein response (UPR) genes are induced. We determined the transcriptional expression of Arabidopsis thaliana UPR genes by fluid microarray analysis of tunicamycin-treated plantlets. Two hundred and fifteen up-regulated genes and 17 down-regulated ones were identified. These genes were reanalyzed with functional DNA microarrays, using DNA fragments cloned through fluid microarray analysis. Finally, 36 up-regulated and two down-regulated genes were recognized as UPR genes. Among them, the up-regulation of genes related to protein degradation (HRD1, SEL-1L/HRD3 and DER1), regulation of translation (P58IPK), and apoptosis (BAX inhibitor-1) was reconfirmed by real-time reverse transcriptase-PCR. The induction of SEL-1L protein in an Arabidopsis membrane fraction on tunicamycin-treatment was demonstrated. Phosphorylation of initiation factor-2α, which was inhibited by P58IPK, was decreased in tunicamycin-treated plantlets. However, regulatory changes in translation caused by ER stress were not detected in Arabidopsis. Plant cells appeared to have a strategy for overcoming ER stress through enhancement of protein folding activity, degradation of unfolded proteins, and regulation of apoptosis, but not regulation of translation.

Abbreviations
AARE

amino acid response element

ATF6

activating transcription factor 6

AZC

l-azetidine-2-carboxylic acid

BI-1

Bax inhibitor-1

eIF2α

initiation factor-2α

Endo H

endoglycosidase H

ER

endoplasmic reticulum

ERAD

ER-associated protein degradation

ERSE

ER stress response element

MS

Murashige and Skoog medium

PDI

protein disulfide isomerase

PKR

double stranded RNA-activated protein kinase

P-UPRE

plant-specific UPR element

RAMP4

ribosomal-associated membrane protein 4

TM

tunicamycin

UPR

unfolded protein response

UPRE

UPR cis-acting regulatory element

XBP-1

X-box binding factor

A nascent polypeptide synthesized on the rough endoplasmic reticulum (ER) is translocated and folded with the assistance of molecular chaperones and other folding factors such as glycosylation/modification enzymes and disulfide oxidoreductases within the ER. However, the folding of nascent polypeptides occasionally does not occur, resulting in the accumulation of unfolded or misfolded proteins in the ER (ER stress). To solve this problem, eukaryotic cells sense ER stress and induce a set of genes called unfolded protein response (UPR) genes. In the budding yeast Saccharomyces cerevisiae, ER transmembrane protein kinase/riboendonuclease Ire 1p is activated by ER stress [1,2], and nonconventionally splices mRNA of basic leucine zipper transcription factor Hac 1p [3–5]. Hac 1p is translated from the spliced mRNA and induces the UPR genes, having a UPR cis-acting regulatory element [6–8]. On DNA microarray analysis, 381 genes have been identified as UPR ones induced by both tunicamycin (TM) and dithiothreitol [9]. These comprise ≈ 6% of the total yeast genes encoding 173 unknown proteins and 208 proteins related to folding, glycosylation/modification, translocation, protein degradation, vesicle trafficking/transport, vacuolar protein sorting, cell wall biogenesis, and lipid/inositol metabolism.

In comparison with those of yeast, the UPR genes of mammalian cells are induced through a much more complicated mechanism, which has been shown to be triggered by at least three transcription factors, X-box binding factor (XBP-1), activating transcription factor 6 (ATF6), and ATF4 [10]. The mammalian paralog of yeast Ire 1p is activated by ER stress and splices the invalid mRNA into mature mRNA encoding 371-amino acid XBP-1 [11,12]. XBP-1 translated from the spliced mRNA is translocated to the nucleus [13], where it binds to its target sequence in the regulatory regions of the P58IPK, ERdj4, HEDJ, EDEM, protein disulfide isomerase (PDI)-P5, ribosomal-associated membrane protein 4 (RAMP4), DnaJ/HSP40-like genes, etc. [14]. ATF6 is an ER transmembrane protein that senses ER stress through its luminal domain, and then moves to Golgi bodies to be cleaved by site-1 and site-2 proteases [15–17]. The cleaved ATF6 cytoplasmic domain is released from Golgi membranes into the nucleus, where it induces, in the presence of nuclear factor Y, ER chaperone genes including BIP, GRP94, Calreticulin and ORP150, which have an ER stress response element (ERSE) in their regulatory regions [18,19]. PERK is an interferone-induced double stranded RNA-activated protein kinase (PKR)-related protein that senses ER stress through its luminal domain and then phosphorylates initiation factor-2α (eIF2α), resulting in inhibition of bulk protein translation [20,21] and stimulation of translation of ATF4 [22]. ATF4 is a basic leucine zipper transcription factor that induces the transcription of many amino acid synthetic enzymes and amino acid transporters by binding to the amino acid response element (AARE) in the regulatory regions of these genes [23]. ATF4 has also been shown to stimulate the transcription of CHOP, which is important for apoptotic cell death [24].

In contrast to the UPR mechanism(s) in yeast and animal cells, that of plant cells is not well understood. Putative plant paralogs of yeast Ire1p have been found in Arabidopsis thaliana and Oryza sativa[25,26]. Their N-terminal luminal domains have each been shown to function as a sensor for ER stress in yeast. However, neither target mRNAs of transcription factors for plant Ire1p nor target genes induced by this system have been identified. On the other hand, the mRNAs of BiP, calreticulin, calnexin and PDI have been shown to be induced on treatment with TM and dithiothreitol in Arabidopsis, Zea mays, Phaseolus vulgalis, Glycine max and Nicotiana tabacum on northern analysis [27–31]. The 21 UPR genes up-regulated by the stress induced by both TM and dithiothreitol have been identified among 8297 genes of the ≈ 27 000 protein-coding genes of Arabidopsis with an Affimetrix GeneChips [32].

In this paper, we present a list of the UPR genes of Arabidopsis identified among all the protein-coding genes. In order to increase the accuracy of the list, the genes selected on fluid microarray analysis were reanalyzed by functional DNA microarray analysis. In addition to the genes related to protein folding and degradation, genes related to protein translation and apoptosis are also included in the list.

Results

Fluid microarray analysis of gene expression on TM-treatment

To identify UPR genes among all the genes expressed in Arabidopsis, we adopted the fluid microarray method, by which target genes can be cloned from selected fluid microarray beads. The fluid microarray beads and probes for array analysis were prepared using the mRNA from plantlets treated with or without TM for 6 h. BiP mRNA, a representative UPR gene, in TM-treated plantlets, was shown to increase 5.7 times compared to the level in untreated plantlets on real-time RT-PCR analysis. For the control experiment, competitive hybridization and sorting of the beads with a cell sorter were performed on 4 × 104 beads with a 1 : 1 mixture of the probes, which had been prepared from noninduced plantlets, and differentially labeled with Cy5 and fluorescein. In the control experiment, almost all of the beads after the control hybridization were sorted in the diagonal line region, the fluorescence intensities for fluorescein and Cy5 being the same (Fig. 1A). Based on the distribution of beads in this experiment, we set three gates to collect beads, i.e., for ones more heavily labeled with Cy5 (U1 and U2) and fluorescein (D). For differential gene expression analysis, probes from TM-treated plantlets were labeled with Cy5. Probes from nontreated plantlets were labeled with fluorescein. Then 4 × 105 beads were hybridized with a 1 : 1 mixture of the two types of probes. For analysis of differential gene expression, 1473 and 1703 beads were collected in fractions U1 and U2 of the up-regulated genes, and 3550 beads in fraction D of the down-regulated genes (Figs 1B and 2A). The DNA fragments on beads in these fractions were amplified by PCR and then sequenced. In the up-regulated fractions, 215 genes (Table S1) were found as clusters of clones, which were identified on more than two beads, and 412 as singlet clones, which were identified on single beads (Table S2). For the down-regulated fraction, 10% of the total beads were analyzed to reveal 17 genes as clusters of clones (Table 1) and 34 as singlet clones (Table S2).

Figure 1.

Competitive hybridization on fluid microarrays. (A) Control hybridization: 4 × 104 beads were hybridized with a 1 : 1 mixture of differentially labeled probes from noninduced plantlets. (B) Competitive hybridization: 4 × 105 beads were hybridized with a 1 : 1 mixture of cDNA probes prepared from induced (Cy5) and noninduced plantlets (fluorescein) as described under Experimental procedures. After hybridization, beads that went to gates U1, U2 and D were collected and subjected to gene analysis as described under Experimental procedures.

Figure 2.

Overview of the fluid microarray and functional microarray analyses. (A) Gene selection by fluid microarray analysis. Gates, U1, U2 and D were set as shown in Fig. 1B. Singlet, a gene identified on a single bead. Cluster, a gene identified on more than two beads. (B) Analysis with functional DNA microarrays. The genes selected in (A) were analyzed. Two hundred and thirty-two genes (215 up-regulated cluster and 17 down-regulated cluster genes) were spotted on functional DNA microarrays. The functional DNA microarray analysis was carried out with target DNA fragments prepared from the mRNA of control plantlets or plantlets treated with tunicamycin (TM), dithiothreitol (DTT) or l-azetidine-2-carboxylic acid (AZC) as described under Experimental procedures. The numbers are the numbers of genes that showed an expression difference between control plantlets and plantlets treated with TM, DTT or AZC. The numbers in the ‘Overlap’ row are the numbers of overlapping up-regulated genes or down-regulated genes upon treatments with the three reagents. (C) Venn diagram of the numbers of overlapping and nonoverlapping putative UPR cis-acting regulatory elements of the 36 up-regulated genes selected in (B). The numbers in parentheses are the numbers of genes that have a cis-acting regulatory element. Bold letters are the numbers of overlapping genes. ERSE, CCAAT-N9-(A/C)CACG; XbpI, TGACGTG(G/T); P-UPRE, ATTGG(T/G)CCACGTCAT; AARE, TT(G/T)CATCA.

Table 1.  Genes recovered at gate D and functional DNA microarray analysis of them. Tunicamycin (TM), dithiothreitol (DTT) and l-azetidine-2-carboxylic acid (AZC) values are means for six experiments. Control ratio obtained on competitive hybridization with Cy5- and Cy3-labeled control mRNA; values are means for six experiments. SD, standard deviation; n.d., not determined.
AGI geneDescriptionFluid microbead array (number of beads)Functional DNA microarray (fold variation)
U1U2DTMDTTAZCControl (SD)
At5g24770Vegetative storage protein Vsp2001820.200.780.380.99 (0.04)
At5g24780Vegetative storage protein Vsp100120.190.790.120.98 (0.05)
At2g39330Putative mylosinase-binding protein00150.361.10n.d.1.10 (0.05)
At5g50960Nucleotide-binding protein1040.971.205.640.93 (0.08)
At3g04120Glyceraldehyde-3-phosphate dehydrogenase C subunit0020.680.990.981.01 (0.03)
At5g64120Peroxidase002n.d.n.d.0.051.01 (0.03)
At4g34490Adenylyl cyclase-associated protein CAP20020.940.940.021.09 (0.14)
At4g37410Cytochrome P4500020.281.030.840.94 (0.04)
At5g04140Ferredoxin-dependent glutamate synthase0050.830.97n.d.3.89 (6.12)
At4g22470Extensin-like protein0050.270.910.021.03 (0.03)
At5g47930Ribosomal protein S270031.150.950.691.09 (0.06)
At1g01060Similar to DNA binding protein CCA10020.240.820.010.89 (0.08)
At2g07671Unknown00310.210.936.561.30 (0.15)
At4g32610Unknown0020.920.960.741.10 (0.09)
At2g07707Unknown00120.280.902.161.04 (0.07)
At3g02200Unknown0040.991.000.320.93 (0.06)
At5g51190Unknown0081.120.992.9518.08 (0)

Analysis with functional DNA microarrays

In order to increase the accuracy of the list of UPR genes, we reanalyzed the genes selected on fluid microarray analysis with functional DNA microarrays. The functional DNA microarrays were prepared by spotting PCR fragments from the 215 up-regulated cluster genes (Table S1) and the 17 down-regulated cluster genes (Table 1) cloned on fluid microarray analysis. Singlet genes were omitted from the functional DNA microarray analysis, because the list of singlet genes was predicted to contain missorted non-UPR genes at a high frequency. Functional DNA microarray analyses were performed with mRNA preparations from plantlets treated with or without TM, dithiothreitol or l-azetidine-2-carboxylic acid (AZC). AZC is a proline analog that is incorporated in nascent polypeptides instead of proline and prevents the folding of the polypeptides [33]. Induction of BiP mRNA by dithiothreitol- or AZC-treatment (3 h or 17 h, respectively) was confirmed to be 3.4 or 22-times higher than that in untreated plantlets on real-time RT-PCR analysis. To identify the up-regulated UPR genes, it was required that candidate UPR genes show a mean fold variation of greater than 1.2-fold with all the treatments with TM, dithiothreitol and AZC. In addition, from the list, we eliminated the genes in which the degree of variation was lower than the sum of the background variation and twice the standard deviation. The degree of background variation was obtained by means of a self/self hybridization experiment with Cy5 or Cy3-labeled target DNA fragments prepared from nontreated plantlet mRNA. Thus, the expression difference between selected genes was regarded as being significant below a probability of error of 5%. Thirty-six genes were confirmed to be induced under the three different inductive conditions, because these genes satisfied this criterion (Fig. 2B and Table 2). These genes comprised 30 for which some functional information was available and six for which no information was available. Among them, 27 genes were putative paralogs that have been reported to be UPR genes in yeast and/or mammalian cells. The functional categories comprise protein folding (13 genes), translocation (six genes), ER-associated protein degradation (ERAD) (three genes; HRD1-like, SEL-1L/HRD3-like, and DER1-like), protein glycosylation and modification (two genes), regulation of translation (P58IPK) [34], and vesicle trafficking (two genes). The induction of HRD1-like, SEL-1L/HRD3-like, DER1-like, and P58IPK mRNA was confirmed by real-time RT-PCR analysis (Fig. 3). In addition, we found that an antiapoptosis protein, Bax inhibitor-1 (BI-1) [35,36], was also included in the list of up-regulated UPR genes. Induction of this paralog by ER stress in organisms other than plants has not been reported. The induction of BI-1 mRNA by ER stress in Arabidopsis was confirmed by real-time RT-PCR analysis (Fig. 3). Furthermore, the induction (1.5-fold variation) of Homo sapiens BI-1 by TM-treatment for 24 h was confirmed in Hep G2 cells, a cell line derived from a human hepatoma, by real-time RT-PCR (data not shown).

Table 2.  Genes up-regulated by ER stress. Tunicamycin (TM), dithiothreitol (DTT) and l-azetidine-2-carboxylic acid (AZC) values are means for six experiments. Control ratio obtained on competitive hybridization with Cy5- and Cy3-labeled control mRNA; values are means for six experiments. SD, standard deviation.
AGI geneDescriptionFluid microbead array (number of beads)Functional DNA microarray (fold variation)
U1U2TMDTTAZCControl (SD)
  1. a Gene identified by Martìnez and Chrispeels [32]. b Genes identified by Noh et al. [54].

PROTEIN FOLDING
At1g09080BiP-likea,b0238.2438.35508.733.13 (0.33)
At5g28540BiPa,b302913.814.0636.221.14 (0.14)
At5g42420BiP140264.124.0953.671.21 (0.10)
At5g61790Calnexin 1a,b82123.422.6325.260.27 (0)
At5g07340Calnexin 2a,b072.372.3410.480.97 (0.09)
At1g56340Calreticulin 10222.162.071.941.02 (0.15)
At1g09210Calreticulin 2a,b16442.432.021.510.95 (0.05)
At4g24190AtHsp90-7a,b3433.862.656.881.05 (0.12)
At2g47470Similar to PDIa5542.112.283.531.05 (0.19)
At1g77510Similar to PDI3023.822.6811.980.91 (0.07)
At2g32920Similar to PDIa202.462.5010.301.01 (0.08)
At1g04980Similar to PDI113.223.0010.460.95 (0.13)
At5g58710AtCYP20-1 (cyclophilin ROC7)031.431.342.131.01 (0.08)
GLYCOSYLATION/MODIFICATION
At2g02810UDP-glucose/UDP-galactose transportera,b233.532.0821.950.94 (0.04)
At2g41490UDP-GlcNac:dolichol phosphate N-acetyl-glucosamine-1-phosphate transferasea021.551.536.671.01 (0.09)
TRANSLOCATION
At5g50460SEC61 gamma subunit2221.941.665.541.07 (0.13)
At1g29310Similar to SEC61 alpha subunit041.51.6111.890.97 (0.10)
At2g34250Similar to SEC61 alpha subunit021.271.382.110.96 (0.07)
At1g27330Similar to SERP1/RAMP42031202.421.8213.480.98 (0.04)
At1g27350Similar to SERP1/RAMP4a,b0132.051.7210.611.08 (0.19)
At3g51980Similar to ER chaperone SIL 12342.393.0052.880.98 (0.12)
PROTEIN DEGRADATION
At1g65040Similar to HRD17113.362.336.480.99 (0.07)
At4g21810Similar to DER1a,b071.671.594.231.09 (0.07)
At1g18260Similar to SEL-1L/HRD3031.541.549.160.96 (0.11)
TRANSLATION
At5g03160P58IPK2142.061.7610.760.94 (0.11)
VESICLE TRAFFICKING
At3g07680Similar to Emp24p041.471.292.730.98 (0.05)
At4g21730Similar to NEM-sensitive fusion protein037.529.68688.381.05 (0.19)
ANTI-APOPTOSIS
At5g47120BI-1022.301.7386.700.99 (0.07)
UNCLASSIFIED
At2g25110Similar to stroma cell-derived factora032.142.089.170.93 (0.03)
At5g09410Similar to anther ethylene-up-regulated calmodulin-binding protein ER1111.201.203.590.89 (0.05)
UNKNOWN
At5g18090 231.201.2514.310.90 (0.06)
At1g56580 5701.951.989.661.04 (0.02)
At5g64510 8312.745.87181.561.29 (0.32)
At5g14890 173.605.6046.171.05 (0.05)
At3g22235 041.512.491.281.11 (0.09)
At1g29060 031.791.7833.691.02 (0.07)
Figure 3.

Confirmation of transcriptional induction of six genes by real-time RT-PCR analysis. The amounts of actin, BiP, HRD1, SEL-1L, DER1, p58IPK and BI-1 mRNAs in total RNA from Arabidopsis plantlets treated with TM (black bars), DTT (hatched bars), or AZC (white bars) for 2 h were determined by real-time RT-PCR as described under Experimental procedures. The value for each mRNA was standardized to the value for actin mRNA in the corresponding total RNA preparation. Fold expression change was calculated as the ratio of mRNA in the plantlets treated and untreated with a stress reagent. Each value represents the mean for two experiments.

To identify the down-regulated UPR genes, we required that candidate UPR genes show a mean fold variation of lower than 0.8-fold with all the treatments with TM, dithiothreitol and AZC. Two genes encoding vegetative storage proteins, Vsp1 and Vsp2 [37,38], satisfied this criterion. Vsp2-beads comprised 58% of the beads collected and were analyzed at gate D.

Putative cis-acting regulatory element of UPR genes

In yeast, ER stress activates Ire1p, which triggers the nonconventional splicing of HAC1 mRNA [3–5]. Hac1p produced from the spliced mRNA induces the transcription of UPR genes by binding to their UPR cis-acting regulatory element (UPRE), CAGCGTG [6–8]. In mammals, four kinds of cis-acting regulatory elements, which respond to ER stress, are known. Mammalian UPRE (TGACGTG-T/G) has been shown to be the specific cis-acting regulatory element for XBP1 and is referred to as the XBP1 binding site [39,40]. ERSE (CCAAT-N9-CCACG) has been found to be recognized by both ATF6 and XBP1 in vitro[41]. ERSEII (ATTGG-N-CCACG) has also been demonstrated to be a target of ATF6 [42]. Binding of ATF6 to these cis-acting regulatory elements occurs in collaboration with general transcription factor nuclear factor-Y [43,44]. AARE (C/EBT-ATF) (TT-G/T-CATCA), which was discovered in the CHOP promoter, is recognized by ATF4, translation of which is accelerated by ER stress [24]. In plants, a plant-specific UPR element (P-UPRE) (ATTGGTCCACGTCATC), which contains two mammalian UPR cis-acting regulatory elements such as ERSEII and XBP1 binding sequences, was found in the 5′ upstream regions of the BiP and calnexin genes [45]. Furthermore, complementary sequences to the mammalian ERSE and XBP1 binding sequences have been found in the 5′ upstream regions of several genes that are induced by TM- or dithiothreitol-treatment [32,45]. Therefore, we searched for P-UPRE, the XBP1 binding sequence, ERSE, AARE, or complementary sequences in the 5′ upstream regions (up to 1000 nucleotides) of the UPR genes. Single or plural putative cis-acting regulatory elements were found in the 5′ upstream regions of 28 of the 36 up-regulated genes (Fig. 2C and Table 3). No cis-acting regulatory element sequence was found in the 5′ upstream regions of the two down-regulated genes.

Table 3.  Putative cis-acting regulatory elements of genes up-regulated by ER stress. Position designated from the 5′ terminus of the ATG initiation codon. Lowercase letters in sequences correspond to N9 in ERSE-like sequence CCAAT-N9-(A/C) CACG.
AGI geneDescriptioncis-Acting regulatory element
Motif-likePosition
At1g09080BiP-likeERSE-likeCGTGTcaagaagtgATTGG(142–124)
ERSE-likeCGTGTctgcttgtgATTGG(220–202)
At5g28540BiPP-UPREATTGGTCCACGTCAT(168–154)
At5g42420BiPXbp1 binding-likeCCACGTCA(187–180)
P-UPREATTGGACCACGTCAT(193–179)
At5g61790Calnexin 1ERSE-likeCGTGGcctgttatgATTGG(237–219)
Xbp1 binding-likeTGACGTGG(240–233)
At5g07340Calnexin 2P-UPREATTGGGCCCAGGTCA(290–274)
At1g56340Calreticulin 1ERSE-likeCGTGTatttaactaATTGG(147–129)
At1g09210Calreticulin 2ERSE-likeCGTGTcggttacctACCGG(178–160)
At4g24190AtHsp90-7ERSE-likeCCAATacaaaactaCCACG(229–211)
Xbp1 binding-likeCCACGTCA(253–246)
At2g47470Similar to PDIXbp1 binding-likeCCACGTCA(139–132)
At1g77510Similar to PDIERSE-likeCCAATgaaaactctCCACG(158–140)
At2g32920Similar to PDI 
At1g04980Similar to PDIERSE-likeCGTGTgacaatatcATTGG(128–110)
Xbp1 binding-likeTGACGTGT(131–124)
At5g58710AtCYP20-1 (cyclophilin ROC7)Xbp1 binding-likeTGACGTGG(83–76)
ERSE-likeCCAATtacaattgtACACG(134–116)
At2g02810Similar to UDP-glucose/UDP-galactose transporter 
At2g41490UDP-GlcNac:dolichol phosphate N-acetylglucosamine-1-phosphate transferaseERSE-likeCGTGGcaaatccttATTGG(128–110)
 
At5g50460SEC61 gamma subunitXbp1 binding-likeTGACGTGT(171–164)
Xbp1 binding-likeTGACGTGT(322–315)
At1g29310Similar to SEC61 alpha subunitERSE-likeCGTGTatccgtattATTGG(439–420)
At2g34250Similar to SEC61 alpha subunit 
At1g27330Similar to SERP1/RAMP4ERSE-likeCCAATcactgaccgCCACG(223–205)
At1g27350Similar to SERP1/RAMP4ERSE-likeCCAATtatagacggCCACG(269–251)
At3g51980Similar to ER chaperone SIL 1Xbp1 binding-likeTGACGTGT(149–142)
ERSE-likeCGTGTaataatataATTGG(146–128)
At1g65040Similar to HRD1ERSE-likeCGTGTcgttatatcATTGG(338–320)
At4g21810Similar to DER1 
At1g18260Similar to SEL-1L/HRD3ERSE-likeCGTGGccggttactATTGG(176–158)
At5g03160P58IPKERSE-likeCGTGGgtcataacgATTGG(244–226)
ERSE-likeCGTGTttaattatcATTGG(304–286)
At3g07680Similar to Emp24pERSE-likeCCAATgatataacgCCACG(437–419)
Xbp1 binding-likeTGACGTGG(477–470)
Xbp1 binding-likeACACGTCA(609–602)
At4g21730Similar to NEM-sensitive fusion protein 
At5g47120BI-1ERSE-likeCGTGGatgattcttATTGG(298–280)
At2g25110Similar to stroma cell-derived factor 
At5g09410Similar to anther ethylene-up-regulated calmodulin-binding protein ER1ERSE-likeCGTGTcggaggtttATTGG(271–253)
Xbp1 binding-likeTGACGTGG(396–389)
At5g18090UnknownAARE-likeTTTCATCA(154–161)
At1g56580UnknownAARE-likeTTTCATCA(271–278)
At5g64510Unknown 
At5g14890Unknown 
At3g22235Unknown 
At1g29060UnknownERSE-likeCCAATattaaaacgCCACG(233–215)

Increase in putative SEL-1L due to ER stress in Arabidopsis

In yeast and mammalian cells, the HRD1/HRD3 (SEL-1L) ubiquitination system coupled to protein degradation by 26S proteasomes is known to be induced to remove unfolded proteins under ER stress [9,46]. Plant paralogs of these genes have not been identified yet. In this study, the transcriptional induction of genes homologous to mammalian HRD1 and SEL-1L[47–49] was observed (Fig. 3). Then, HRD1- and SEL-1L-like cDNAs were cloned with mRNA of Arabidopsis plantlets by RT-PCR. Their nucleotide sequences coincided with those presented in the database of ‘The Arabidopsis Information Resource’ (http://www.arabidopsis.org/). The putative amino acid sequence of an HRD1-like protein contained an N-terminal signal sequence and five membrane-spanning regions (data not shown). The recombinant luminal domain of the HRD1-like protein was expressed in Escherichia coli and purified. Unfortunately, autoubiquitination activity was not detected for the recombinant HRD1-like protein. On the other hand, the putative amino acid sequence of Arabidopsis SEL-1L (At SEL-1L) contained an N-terminal signal sequence (Met1–Glu20), two N-glycosylation consensus sequences, and a membrane-spanning region (Phe623–Arg643) near the C-terminus (data not shown). The amino acid sequence of a soybean paralog of SEL-1L, which was deduced from the nucleotide sequence of cDNA cloned from young leaves by RT-PCR, was closely similar to Arabidopsis ones (data not shown). Anti-(At SEL-1L) serum was prepared with the recombinant luminal domain (Phe21–Val622) of At SEL-1L, which was expressed in E. coli and isolated. The antiserum only immunoreacted with a 74 kDa protein of control plantlets on western blotting analysis (Fig. 4A). With TM-treatment, the 74 kDa protein gradually decreased and a 70 kDa band began to appear at 4 h after the treatment. During the next 24 h, the 70 kDa band significantly increased. The size of the 74 kDa band decreased to 70 kDa on endoglycosidase H (Endo H) digestion. On the other hand, the 70 kDa band was insensitive to Endo H (Fig. 4B). From these results, the 70 kDa protein was thought to be a nonglycosylated form of At SEL-1L. On cell fractionation, At SEL-1L was assumed to be a membrane protein, as judging from the existence of a putative membrane spanning region (Fig. 4C). The 70 kDa band of plantlets treated with TM for 24 or 48 h was denser than the 74 kDa band of the control plantlets (Fig. 4B). Thus, it was suggested that At SEL-1L polypeptides were synthesized from the At SEL-1L mRNA induced by ER stress, but that N-glycosylation of newly synthesized At SEL-1L molecules was inhibited by TM.

Figure 4.

Increase in At SEL-1L in the membranes of Arabidopsis plantlets on TM-treatment. (A) Plantlets were incubated in the presence (lanes 7–12) or absence of TM (lanes 1–6) for the indicated times. Proteins were extracted and then subjected to SDS/PAGE. At SEL-1L was stained by western blotting with antiserum as described under Experimental procedures. (B) Plantlets were incubated in the presence (lanes 3 and 4) or absence of TM (lanes 1 and 2) for 48 h. Proteins were extracted, digested with (lanes 2 and 4) or without (lanes 1 and 3) Endo H, and then subjected to SDS/PAGE. At SEL-1L was stained by western blotting with antiserum as described under Experimental procedures. (C) The total (lane 1), supernatant (lane 2), and membrane (lane 3) fractions obtained from the plantlets treated with TM for 48 h on centrifugation at 100 000 g were subjected to SDS/PAGE, and At SEL-1L was stained by western blotting with antiserum as described under Experimental procedures.

ER stress and phosphorylation of eIF2α

In this study, we found that the mRNA of P58IPK was induced by ER stress (Table 2 and Fig. 3). P58IPK was first identified as an inhibitor of interferon-induced PKR in mammalian cells [50]. The PKR family responds to different stress signals and attenuates translation by phosphorylating the specific serine residue of eIF2α[51] to protect cells from the stress. P58IPK inhibits PKR-mediated translational arrest by inactivating the kinase by binding to the domain of PKR family members. In mammals, ER stress also causes translational arrest through phosphorylation of eIF2α by PKR-like ER kinase, PERK [20,52]. Mammalian P58IPK has been shown to be induced at a later phase of ER stress [53]. Deletion of P58IPK has been reported to result in an increase in phosphorylated eIF2α. Hence P58IPK is thought to function as a feedback regulator for translational regulation in the later phase of ER stress. The phosphorylated Ser51 of eIF2α in plantlets was examined during ER stress by western blot analysis (Fig. 5A). The level of phosphorylated eIF2α (Ser51) in the plantlets treated with TM was lower than that in untreated plantlets. The phosphorylated eIF2α increased again on removal of TM from the medium after 6 h of treatment. However, the protein synthesis in plantlets, which was assayed as the incorporation of [35S]-labeled Met and Cys into nascent proteins, was not affected by TM-treatment (Fig. 5B).

Figure 5.

Effect of TM-treatment on phosphorylation of eIF2α. Plantlets were incubated in the medium with TM for 6 h (lane 2), 7 h (lane 3), or 9 h (lane 4), or without TM for 9 h (lane 1) as described under Experimental procedures. In other experiments, plantlets were incubated in the medium with TM for 6 h and then incubated in the medium without TM for an additional 1 h (lane 5) or 2 h (lane 6). (A) After the incubation, the proteins were extracted from the plantlets and subjected to SDS/PAGE. Phosphorylated Ser51 of eIF2α was determined by western blot analysis as described under Experimental procedures. (B) After the incubation, proteins of the plantlets were metabolically labeled with [35S]Met and [35S]Cys for 20 min at 25 °C. Then, the proteins were extracted and subjected to SDS/PAGE. Labeled proteins were determined by fluorography as described under Experimental procedures.

Discussion

In this study, we tried to make a list of the UPR genes in Arabidopsis. In total, 215 up-regulated and 17 down-regulated cluster genes were cloned from mRNA of Arabidopsis plantlets treated with TM on fluid microarray analysis. A functional DNA array was prepared by using the cloned gene fragments, and then used for analysis. Among the 215 up-regulated cluster genes, only 63 showed statistically positive signals on functional DNA array analysis, showing differences in the expression of target mRNA of the plantlets treated with or without TM. Because the fluid microarray beads included a large number with highly expressed housekeeping genes, some of them might be missorted at the gates, which would expand the list of genes. Of the beads collected at gates U1 and D on fluid microarray analysis, 89 and 87% were regarded as up-regulated and down-regulated genes on functional DNA microarray analysis, respectively. On the other hand, 38% of the beads collected at gate U2 were regarded as up-regulated genes on functional DNA microarray analysis. This suggests that the discrepancy between the values obtained in the two analyses is mainly due to the beads missorted at gate U2. However, of the rest, the 50 genes that showed no up-regulated signal for the plantlets treated with TM showed an up-regulated signal in the plantlets treated with dithiothreitol and/or AZC. In addition, 23 of the genes that showed no difference in expression on DNA microarray analysis between plantlets treated and untreated with TM had putative UPR cis-acting regulatory elements in their upstream regions. Furthermore, 27 of the 63 genes were eliminated on functional DNA microarray analysis from the list by setting some criteria. Therefore, the remaining 36 genes, which satisfied these criteria, were considered to be reliable up-regulated UPR genes. Among these 36 genes, 12 coincided with up-regulated UPR genes previously identified on analysis with an Affimetrix GeneChips loaded with 8297 Arabidopsis probe sets [32,54]. Two down-regulated genes, Vsp1 and Vsp2, which satisfied all the criteria, are known to be for temporary nitrogen-storage proteins [38], and are subject to regulation by sugars, light, phosphates, nitrogen, wounding, auxins, jasmonates and oxidative-stress [55]. The down-regulation of Vsp1 and Vsp2 may result in an increase in the intracellular amino acid pool, which may play an important role in the recovery from ER stress. In mammalian cells, ER stress affects cellular amino acid metabolism via the PERK/ATF4-mediated signaling pathway, which induces some amino acid synthesis- and transport-related genes [23]. No putative UPR cis-acting regulatory element was found in the 5′ upstream regions of Arabidopsis Vsp1 and Vsp2. Therefore, it is not clear whether these genes are directly regulated by the UPR system or down-regulated by a metabolic disorder caused by ER stress.

Thirteen genes, which encode six protein families responsible for protein folding, are included in the UPR gene list. Among them, BiP (three genes), calnexin (two genes), calreticulin (two genes), and AtHSP 90-7 (one gene) have been shown to be induced by ER stress on northern blotting [25,54]. Four genes encoding PDI families are also included in the list. PDI and its family members are characterized by the presence of one or two thioredoxin homologous motifs per molecule. Yeast and mammalian PDIs are known as multifunctional folding catalysts and molecular chaperones, which catalyze the formation and rearrangement of disulfide bonds between correct pairs of cysteine residues in nascent polypeptide chains in the ER [56]. Mammalian PDI functions not only as a catalytic enzyme but also as a subunit of microsome triacylglycerol transfer protein [57] and prolylhydroxylase [58]. Mammalian PDI family ER-60/ERp57, which also exhibits protein oxidoreductase activity, interacts and cooperates with calnexin or calreticulin for oxidative folding of N-glycosylated proteins [59–61]. The genes of these PDI families are UPR genes [41]. In the Arabidopsis genome, 13 genes encoding putative PDI-related proteins, i.e. At1g04980 (NP 171990), At1g07960 (NP172274), At1g15020 (NP 172955), At1g35620 (NP 564462), At1g21750 (NP 173594), At1g52260 (NP 175636), At1g77510 (NP 177875), At2g01270 (NP 565258), At2g32920 (NP 180851), At2g47470 (NP182269), At3g54960 (NP 191056), At3g16110 (NP 188232), and At5g60640 (NP 568926), were found. Identification and characterization of these PDI family proteins were not carried out. However, they were supposed to play important roles in protein folding, as four PDI-related genes among the above 13 genes were confirmed to be induced by ER stress. A gene encoding cyclophilin family protein ATCYP20-1 was identified as a UPR gene. Twenty-nine genes encoding cyclophilin family members were found in the Arabidopsis genome [62]. Among them, five gene products are assumed to be targeted to the ER lumen with N-terminal signal peptides. Among them, ATCYP20-1 has the amino acid sequence RFWH, which is an essential sequence for peptidyl prolyl cis, trans isomerase activity. Hence, it is suggested that ATCYP20-1 may participate in the folding of proteins in the ER.

The genes of six translocation-related proteins were found to be induced. In mammalian cells and yeast, translocon subunit proteins are thought to be induced to enhance retrotranslocation of unfolded proteins from the ER to the cytosol [63]. The retrotranslocated proteins are degraded by 26S proteasomes. Recently, in tobacco, a GFP-fusion protein containing the P region of calreticulin, which is a model of a misfolded protein in the ER, was shown to be retrotranslocated to the cytosol, ubiquitinated, and then degraded [64]. The induction of translocon subunits by ER stress in Arabidopsis suggests that an ERAD system similar to those of yeast or mammalian cells may remove misfolded proteins produced in the ER of plant cells. This is supported by our finding that the genes encoding putative plant DER1, HRD1 and SEL-1L/HRD3 were also induced by ER stress. DER1 is a hydrophobic protein that is localized to the ER. In yeast, deletion of DER1 prevents degradation of unfolded proteins, suggesting that the function of DER1 may be specifically required for ERAD [65]. Yeast HRD1 is an ER-membrane-anchored ubiquitin ligase, which is required for the degradation and ubiquitination of several ERAD substrates, and is associated with relevant ubiquitin-conjugating enzymes [46]. At HRD1, which has the same nucleotide sequence as that registered in ‘The Arabidopsis Information Resource’, was cloned by RT-PCR with mRNA from Arabidopsis. Six transmembrane regions and a RING-H2 domain of Arabidopsis HRD1 (At HRD1) showed high sequence homology with those of yeast and human HRD1s. Unfortunately, it is unclear whether or not At HRD1 functions as an ubiquitin ligase, as the cytosolic domain of At HRD1, which was expressed in E. coli and isolated, showed no self-ubiquitination activity with an in vitro assay system involving commercial human E1 and yeast E2 (UbCH5c). Yeast HRD3 is an ER-resident glycoprotein with a single span near the C-terminus, which stabilizes HRD1 and regulates the cytosolic HRD1 RING-H2 domain through interaction with the HRD1 transmembrane domain [66]. We showed that At SEL-1L was a membrane-anchored glycoprotein and that it increased under ER stress. In order to clarify the details of the mechanism of plant ERAD, functional characterization of these proteins must be performed.

In mammalian cells, ER stress responses are composed of three steps, i.e., enhancement of the refolding and degradation of unfolded proteins, attenuation of translation [20,21], and apoptosis [24]. ER stress has not been found to cause attenuation of translation in plants. In this study, we found that the P58IPK gene was up-regulated by ER stress. Mammalian P58IPK is induced at a later phase of ER stress and inhibits PKR-mediated translational arrest by binding to the kinase domain of the PKR family [53]. However, bulk protein translation of Arabidopsis was not affected by ER stress, even though the phosphorylation of eIF2α (Ser51) was partially inhibited by ER stress. The phosphorylation of eIF2α (Ser51) increases the translational efficiency of yeast GCN4 mRNA and mammalian ATF4 mRNA, which have four and two upstream open reading frames in the 5′ noncoding portion, respectively [67,68]. Induction of Arabidopsis P58IPK followed by a decrease in the phosphorylation of eIF2α (Ser51) may increase the translational efficiency for unidentified gene(s).

It is unclear whether apoptosis may function as a UPR in plants, although inhibition of ER-type IIA Ca2+-pumps has been reported to induce ER stress and apoptosis in soybean cells [69]. In this study, we identified apoptosis-related gene BI-1 as a UPR gene. BI-1 is an evolutionarily conserved integral membrane protein localized in the ER [35,36]. In mammalian cells, BI-1 affords protection from apoptosis induced by ER stress by inhibiting BAX activation and translocation to mitochondria, by preserving the mitochondrial membrane potential, and by suppressing caspase activation [70]. BAX and Bcl2, and their relatives were not found in plants. However, in rice and barley, BI-1 has been shown to suppress fungal elicitor-induced apoptosis [71,72].

Experimental procedures

Plant materials and treatments

Sterile seeds of Arabidopsis thaliana (Columbia) were germinated in 0.5× Murashige and Skoog medium [73] containing 1% (w/v) sucrose (MS), and cultured for two weeks. To prepare a cDNA tagged library and probes for transcriptome analysis with fluid microarrays or functional DNA microarrays, whole plantlets were treated by immersing their roots in MS containing 5 µg·mL−1 TM, 1 mm dithiothreitol or 50 mm AZC for the indicated times. For the control experiment, plantlets were treated with MS without stress reagents. For relative quantification of mRNA by real-time RT-PCR, and pulse-labeling experiments with [35S]Met and [35S]Cys, the upper parts of plants were cut off from their roots and immersed in MS with a stress reagent.

Real-time RT-PCR analysis

Total RNA was isolated with an RNeasy Plant Mini kit (Qiagen, Valencia, CA) from plant tissues treated with or without TM for 6 h. Relative quantification of mRNA was carried out by the real-time RT-PCR method with an ABI PRISM 7000 Sequence Detection System (Applied Biosystems, Foster City, CA). Forward primers, 5′-AAGTCGTTGCACCTCCTGAGA-3′, 5′-TCAAGGACGCTGTTGTCACTGT-3′, 5′-ACACGGCAAATAACGTTCATCTCTA-3′, 5′-GGACTGCTTTCATCTGGCTTGT-3′, 5′-TCTCTGTTGGGTTTATCTCTTTGGTT-3′, 5′-TGATGGAAGAAGCAGTGGATGA-3′ and 5′-CGTAGAAGAGTGGTACAAGCAGATG-3′, were used for detection of the mRNAs of actin, BiP, P58IPK, BI-1, At HRD1, At SEL-1L or Arabidopsis DER1 (At DER1), respectively.

Reverse primers, 5′-ATCGACGGGCCTGACTCAT-3′, 5′-CAACATTGAGCCCAGCAATAAC-3′, 5′-CAGCTATTTAAGCCGTCTTTTCCA-3′, 5′-GATAGATGCAGAGCCACCAAAGA-3′, 5′-CGGACATGAGAGAGCAAAGTCA-3′, 5′-CAGCTGCAAATTATGGTGAAG-3′ and 5′-ACCCGACGGTGGTGACTACA-3′, were used for detection of the mRNAs of actin, BiP, P58IPK, BI-1, At HRD1, At SEL-1L and At DER1, respectively.

TaqMan probes (Applied Biosystems), 5′-VIC-CAGTACCTTCCAGCAGATGTGGATCGC-TAMRA-3′, 5′-FAM-CCAGCTTACTTACTTCAATGATGCTCAAAGGC-TAMRA-3′, 5′-FAM-CTATGCAAGGTCTCAGTCAGGCTCGGC-TAMRA-3′, 5′-FAM-ATGCTAATGTGGCTCCAGTTTGCCTCT-TAMRA-3′, 5′-FAM-TCCACTCTCTTTTGAGCCATCCAATGC-TAMRA-3′, 5′-FAM-AACGACTTGCTTTTGCTCTTCTCTCGC-TAMRA-3′ and 5′-FAM-ATTATAACCCGGTCGTATCTCACGGC-TAMRA-3′, were used for detection of the mRNAs of actin, BiP, P58IPK, BI-1, At HRD1, At SEL-1L and At DER1, respectively.

Preparation of fluid microarrays

A cDNA tagged library was constructed according to Brenner's method [74]. In brief, mRNA was extracted from plant tissues, except roots, treated with or without TM for 6 h. A total of 2.5 µg of mRNA from plants treated with or without TM was combined and converted to cDNA with a 5′-biotin-conjugated anchored (dT19) primer containing a BsmBI restriction sequence as a primer and a dNTP mixture containing 5-methyl dCTP as a substrate. The DNA fragments were digested with DpnII and BsmBI, and then ligated into a tag vector (tag library plasmid) (Takara Bio Co. Ltd, Kyoto, Japan).

DNA fragments for loading onto antitag microbeads were prepared by PCR using the tagged library as a template and a 6-carboxyl-fluorescein-labeled reverse primer (BD Biosciences Clontech, Palo Alto, CA). The DNA fragments were digested with PacI and then treated with T4 DNA polymerase in the presence of dGTP to expose the tags as single strands. The DNA fragments were loaded onto antitag microbeads. Microbeads combined with cDNA were selected with a cell sorter, MoFlo™ (DacoCytomation, Glostrup, Denmark), and then treated with T4 DNA polymerase and T4 DNA ligase to fill the gap between the cDNA and the tag. 6-Carboxyl-fluorescein was removed by DpnII digestion. Then the antisense strand of cDNA on the microbeads was labeled by ligation with an adaptor carrying 3′-6-carboxyl-fluorescein and removed by treatment with 150 mm NaOH. The microbeads that carried antisense strands were removed from the microbeads that carried sense strands using the cell sorter.

Fluid microarray analysis

For analysis of differentially expressed mRNA in Arabidopsis treated with or without TM, fluid microarray analysis was performed. Probes for competitive hybridization on the fluid microrrays were prepared from the same mRNA sources as those used for the preparation of the cDNA tagged library. In brief, mRNA was converted to cDNA using a flanking oligo dT primer carrying a T7 promoter sequence for first strand synthesis. The probes were synthesized from cDNA derived from control or TM-treated plantlets by T7 RNA polymerase reaction in the presence of fluorescein-UTP or Cy5-UTP. A mixture of probes was then hybridized with a mixture of 4 × 105 fluid microarray beads prepared from the control or TM-treated plantlets at 50 °C overnight. Labeled fluid microarray beads were washed in 1× NaCl/Cit (0.15 m NaCl, 0.015 m sodium citrate, pH 7)/0.1% (w/v) SDS and 0.1× NaCl/Cit/0.1% (w/v) SDS at 65 °C for 15 min [75]. The distribution of microbeads in Fig. 1A allowed us to set gates for collecting microbeads that were more heavily labeled with Cy5 or fluorscein (Fig. 1B). The polygons in Fig. 1B represent the gates at which microbeads carrying up-regulated or down-regulated clones (D) were collected. The up-regulated clone fraction was further separated at two gates (U1 and U2) to divide the beads fraction into two. DNA fragments on the sorted beads were amplified by PCR, subcloned into pT7Blue-2 (Novagen, Darmstadt, Germany), and then sequenced by the Dye Terminator method. Sequences of more than 300 nucleotides were adopted as useful data from the sequence data obtained. Filtering of sequence data and trimming of the vector sequence were carried out with the Paracel Filtering Package (Paracel, Inc., Pasadena, CA). The obtained sequence was searched for the sequence data in ‘The Arabidopsis Information Resource’ (http://www.arabidopsis.org/index.jsp). Then, clustering of the sequence was performed with the Paracel Clustering Package (Paracel, Inc.).

Functional DNA microarray analysis

Functional DNA microarrays were prepared by spotting the PCR fragments derived from the genes selected as up- or down-regulated genes on fluid microarray analysis. The PCR fragments were amplified using the cDNA fragments subcloned into pT7Blue-2 as a template. Each fragment was spotted at two sites on a slide glass. Target DNA fragments were synthesized by in vitro reverse transcription reaction using Cy3-dUTP or Cy5-dUTP from 1.5 µg of mRNA of control plantlets or plantlets treated with TM, dithiothreitol or AZC for 6 h, 3 h or 17 h. The labeled targets were hybridized to a functional DNA microarray in 6× NaCl/Cit/0.2% (w/v) SDS/5× Denhardt's solution/carrier DNA at 65 °C for 14 h [75], and then washed in 1.2× NaCl/Cit/0.2% (w/v) SDS, 2.2× NaCl/Cit/0.2% (w/v) SDS and then 3.2× NaCl/Cit/0.2% (w/v) SDS at 55 °C for 5 min. The functional DNA microarray was rinsed once with 0.05× NaCl/Cit. The fluorescence was scanned with a GeneChip® Scanner 428 (Affymetrix, Inc., Santa Clara, CA). The same experiments were carried out using three functional DNA microarrays. The data were analysed using BioDiscovery imagene Ver. 4.2 (BioDiscovery, El Segundo, CA). The mean Cy5 : Cy3 ratio values were calculated as the Cy5 value divided by both the correction value and the raw Cy3 value. Calculation of the correction value was carried out as described below:

(a) Spots were selected according to the following criteria: [signal mean] − [background mean] more than 60 000, and [signal mean] more than [background mean] + 2 × [background standard deviation (SD)]; (b) Log (Cy5 : Cy3) of the spots selected in (a) was calculated; (c) Mean value I SD of (b) was calculated; (d) Spots were selected according to the following criteria: Log (Cy5 : Cy3) ranged within the mean values I SD obtained in (c); (e) Mean Log (Cy5 : Cy3) of the spots selected in (d) was calculated; (f) Mean Log (Cy5 : Cy3) in (e) was converted to a natural value, which corresponds to the correction value.

Control experiments (self/self hybridization) to obtain a spot-specific background Cy5 : Cy3 ratio for judgment of significant differences in the Cy5 : Cy3 ratio were carried out. Target DNA fragments were synthesized using Cy3-dUTP or Cy5-dUTP from the mRNA of control plantlets, and then hybridized competitively to a functional DNA microarray under the same conditions as for the comparative experiments. Mean control Cy5 : Cy3 ratios and their SD were calculated from the six values obtained in triplicate control experiments on two spots on functional DNA microarray assays. Most of the background [Cy5 : Cy3 I SD] values were in the range of 1.2–0.8. Hence, it was judged as a significant difference when the Cy5 : Cy3 ratio was more than 1.2 and also more than [background Cy5 : Cy3 ratio + 2 × SD], or less than 0.8 and also less than [background mean] + 2 × [background standard deviation (SD)].

Construction of an expression vector for the putative luminal domain of At SEL-1L

The cDNA encoding At SEL-1L was cloned by RT-PCR with a forward primer, 5′-ACGTCGCTGCAGCGATCTGATCACTGAGAAAC-3′, and a reverse primer, 5′-AAAGCCGGTACCCTCTGCTATTACAATGACGAAAACGATTATC-3′, using mRNA from Arabidopsis plantlets treated with TM for 6 h. The obtained fragments were digested with PstI and KpnI, and then cloned into pBluescript (Stratagene, La Jolla, CA) digested with PstI and KpnI. The insert in the vector was sequenced by the fluorescence dideoxy chain termination method (Applied Biosystems). An expression vector for the putative luminal domain of At SEL-1L, which corresponds to residues 21–621, was constructed as described below. For cloning into an expression vector, two kinds of DNA fragments were amplified by PCR with two sets of primers. One set comprised a forward primer as the DNA sequence encoding the N-terminus of the luminal domain of At SEL-1L containing an NdeI restriction site, 5′-ACGTCTGACATATGTTTGGCGTTCACGCTCGTCCC-3′, and a reverse primer corresponding to the sequence containing a XhoI restriction site in At SEL-1L, 5′-AAATCTTCATCCTCCTCGCCTCGAG-3′. The other set comprised a forward primer corresponding to the sequence containing a XhoI restriction site in At SEL-1L, 5′-AAAGGTGCTCTAAGGAAATCTCGAG-3′, and a reversed primer as the DNA sequence encoding the C-terminus of the luminal domain of At SEL-1L containing a XhoI restriction site, GTGGTGCTCGAGCACCACATTCTCTATCCAAGTCTC-3′. The former or latter PCR fragments produced were digested with NdeI and XhoI, or XhoI, respectively, and then cloned into pET-30Xa/LIC digested with NdeI and XhoI. Expression vector pET-30/At SEL-1L allows the fusion of the histidine tag LEHHHHHH to the C-terminus of a recombinant protein.

Expression and purification of the recombinant luminal domain of At SEL-1L BL21(DE3) cells were transformed with pET-30/At SEL-1L

The expression of the putative luminal domain of At SEL-1L was induced by the addition of 0.4 mm isopropyl thio-β-d-galactoside for 4 h. The recombinant protein was produced as inclusion bodies in E. coli. The cells from 2 L culture broth were collected by centrifugation, disrupted by sonication in 40 mL of 20 mm Tris/HCl buffer, pH 7.9, containing 5 mm imidazole, 0.5 m NaCl and 1 mm CaCl2 (binding buffer), and then centrifuged at 10 000 g for 30 min at 4 °C. The pellet was suspended in the binding buffer containing 6 m urea and 5 mm 2-mercaptoethanol (urea-binding buffer) by sonication, and dissolved by adjusting the pH to 9 with 1 m NaOH and then readjusting it to 8 with 1 m HCl. A sample was applied to a His-Bind quick cartridge (Novagen) equilibrated with the urea-binding buffer. After washing the cartridge with the urea-binding buffer, the luminal domain of At SEL-1L was eluted with the urea-binding buffer containing 1 m imidazole, and then concentrated with a Centriprep-10 (Millipore, Billerica, MA). The purified luminal domain of At SEL-1L was used for the preparation of rabbit antiserum. The recombinant protein was confirmed to have an initial methionine residue by N-terminal sequencing.

Analysis of At SEL-1L in Arabidopsis

Plantlets treated with TM for the indicated times were frozen in liquid nitrogen and then ground into a fine powder with a micropestle SK-100 (Tokken, Inc., Chiba, Japan). Proteins were extracted from 100 mg of the tissue with 80 µL of Laemmli's SDS/PAGE buffer [76] containing a 1% (v/v) cocktail of protease inhibitors (Sigma, Inc., St. Louìs, MO) by boiling for 5 min. For Endo H treatment, 300 mg of plant tissue was ground, suspended in 800 µL of 100 mm tricine/KOH buffer, pH 7.5, containing 0.5 m sucrose, 1 mm EDTA and a 1% (v/v) cocktail of protease inhibitors, and then filtered through a cell strainer (BD Biosciences, Bedford, MA). The filtrate was centrifuged at 1000 g for 10 min at 4 °C to remove tissue debris. The supernatant obtained was centrifuged at 100 000 g for 1 h at 4 °C. The pellet was dissolved in 16 µL 0.1 m phosphate buffer, pH 5.5, containing 0.2% (w/v) SDS and 0.5% (v/v) 2-mercaptoethanol by boiling for 5 min. The resulting solution was diluted with four volumes of 0.1 m phosphate buffer and then digested with 15 mU Endo H (Sigma, Inc.) at 37 °C overnight. After digestion, proteins were treated with the SDS/PAGE buffer. For cell fraction analysis, the supernatant and pellet fraction obtained on centrifugation at 100 000 g were treated with the SDS/PAGE buffer. Twenty-five micrograms of protein was subjected to SDS/PAGE and then blotted onto a poly(vinylidene difluoride) membrane. The At SEL-1L protein was then immunostained with 1 : 1000-diluted anti-At SEL-1L serum and horseradish peroxidase-conjugated rabbit Ig antiserum (Promega, Madison, WI) as secondary antibodies, using Western Lightning Chemiluminescence Reagent (PerkinElmer Life Sciences, Boston, MA).

Pulse labeling of proteins

For pulse labeling of proteins, plant tissues cut from the roots were treated with or without TM for the indicated times, and then incubated in 1 mL of MS containing 50 µCi (1850 kBq) each of [35S]Met and [35S]Cys (NEN Life Science Products, Inc., Boston, MA) for 20 min at 25 °C. The labeled plant tissues were rinsed with MS, frozen with liquid nitrogen, and then ground with an electrical homogenizer, S-203, equipped with a spindle (Inouchi-Seieidou Ltd, Osaka, Japan). The disrupted sample was boiled for 2 min in SDS/PAGE buffer containing a 10% (v/v) cocktail of protease inhibitors. Then, 35 µg of protein was subjected to SDS/PAGE. [35S]Met and [35S]Cys in the gel were detected by fluorography with Enlightning (NEN Life Science Products, Inc.).

Detection of phosphorylated Ser51 of eIF2α

Plant tissues cut off from the roots were treated as described above with or without TM for the indicated times. Plant proteins (30 µg of proteins) were separated by SDS/PAGE and then blotted onto poly(vinylidene difluoride) membranes. Phosphorylated Ser51 of eIF2α was immunostained with rabbit eIF2α phospho-specific polyclonal antibodies (Biosource International, Camarillo, CA).

Protein measurement

The concentrations of proteins were measured using an RC DC protein assay kit (Bio-Rad Laboratories, Hercules, CA), with γ-immunoglobulin as an internal standard.

Acknowledgements

We greatly thank Dr Makoto Kito, Emeritus Professor of Kyoto University, for the critical reading of the manuscript, valuable advice and warm encouragement. This study was supported by a Grant for the Program for Promotion of Basic Research Activities for Innovative Biosciences.

Supplementary material

The following supplementary material for this article is available online:

Table S1. Up-regulated genes selected for functional DNA microarray analysis.

Table S2. List of singlet genes identified on fluid microarray analysis.

Ancillary