Oryza sativa L. cv. Hinohikari rice grains were treated by 70% ethanol for 15 min, then soaked the grains in 0.2% sodium hypochlorite solution for 15 min, followed by washing five times in distilled water. The disinfected rice grains were then incubated in distilled water at 30˚C for 2 days, then moved to incubator at 25˚C for 12 days using a cycle with 16 h light (80 W∙m⁻²), 8 h dark, and 70% relative humidity.

As heat stress treatment, 14-day-old rice seedlings grown at 25˚C were subjected to 52˚C for 1, 3, 6, 9 h. For the treatment with GDA, rice seedlings were treated by 50 µM GDA for 3, 6, 8, 12, and 24 h. Each sample consisted of 0.5 g of plant tissue collected at each point in time, and was immediately placed in liquid N₂ and stored at −80˚C for use in subsequent analyses.

Total RNA was extracted from the plant tissue frozen in liquid N₂ using the SDS/phenol/LiCl method as previously described [12]. RT-PCR was performed on a PC-816 thermal cycler (ASTEC Co., Fukuoka, Japan) in a 20-µL reaction mixture containing 1 µL cDNA sample, 200 µM dNTPs, 400 nM F (Forward)- and R (Reverse)- primers (Table 1), and GoTaq Green Master Mix (Promega, MADISON WI, USA) in PCR reaction buffer under the following thermal cycle conditions: 94˚C for 2 min; 27 - 30 cycles of 94˚C for 10 sec, 58˚C for 10 sec, and 72˚C for 30 sec; a final extension at 72˚C for 1 min; and a final hold at 16˚C for 99 min. DNA was separated in 1.5% agarose gels and stained with ethidium bromide, then visualized using FluorChem (Alpha Innotech, San Leandro, CA, USA).

To prepare maltose binding protein (MBP)-fused OsHsfA1 and OsHsfA2, we constructed pMal-OsHsfA1 and -OsHsfA2. PCR fragments encoding full length OsHsfA1 and OsHsfA2 were amplified with KOD Plus DNA polymerase (TOYOBO, Tokyo, Japan), rice cDNA, and a gene-specific primer set (OsHsfA1-FL and OsHsfA2-FL) (Table 1), and then digested with BamHI and SalI. The resultant fragments were ligated into BamHI-SalI sites of pMal-c (New England Biolabs, Inc., Ipswich, MA) by DNA Ligation Kit v. 2 (Takara Bio Inc., Tokyo, Japan). The cloned OsHsfA1 and OsHsfA2 cDNAs were confirmed by DNA sequencing. The recombinant proteins, MBP, MBP-OsHsfA1, and MBPOsHsfA2, were expressed in Escherichia coli cells (BL21DE) containing pMal-c empty vector, pMalOsHsfA1 and pMal-OsHsfA2 grown over night at 37˚C in LB/Amp medium in the presence of 0.5 mM IPTG.

The recombinant proteins were purified from the E. coli cells with Amylose Resin column (NewEngland Lab Inc.), as described in the manufacturer’s instruction manual with minor modifications [13]. Protein concentrations were determined by measuring OD₅₉₅ with a Bio Rad protein assay kit (Bio Rad, Hercules, CA, USA), using 1 mg∙mL^–1 bovine serum albumin as a standard.

Table 1. Gene specific oligo DNA primers used.

RGI, rice gene indics were referred to The Rice Annotation Project (RAP); URL http://rapdb.dna.affrc.go.jp/

Rice genomic DNA was prepared from 14-day-old rice seedlings by CTAB method. Promoter regions (about 200 bp) of OsGolS1, OsHsfA2 (10 - 190), (200 - 400) and OsHsfA3 were amplified by PCR in a 50-µL 1 × GoTaq Green Master Mix reaction mixture containing 100 ng rice total genomic DNA, 400 nM Fand R-promoter specific primers sets (Table 1), under the following thermal cycle conditions: 94˚C for 2 min; 35 cycles of 94˚C for 10 sec, 58˚C for 10 sec, and 72˚C for 20 sec; a final extension at 72˚C for 5 min; and a final hold at 16˚C for 99 min. For a gel-shift assay, a 10 µL of reaction mixture containing 10 mM Tris-HCl pH7.5, 50 mM NaCl, 1 mM MgCl₂, 0.1% b-mercaptoethanol, 100 ng promoter DNA (OsGolS1 or OsHsfA3) and 1 to 3 µg recombinant protein (MBP, MBP-OsHsfA1 or MBP-OsHsfA2) was incubated at 25˚C for 30 min, then separated in 2% agarose gels [14]. PCR product coding a promoter region of OsHsfA3, which is reported to be induced under cold stress but not heat stress, was used as a negative control [13,15]. The DNA and DNA-protein complexes stained with ethidium bromide were visualized by FluorChem imager (AlphaInnotech, San Leandro, CA, USA).

The endogenous H₂O₂ content was measured according to the method of Veljovic-Jovanovic et al. [16]. Briefly, plant tissue (0.5 g) was ground in 0.5 mL of 0.2 M perchloric acid until completely homogenized, using a mortar and pestle. 2 mL of the homogenate was then centrifuged at 13,000 ×g for 15 min. The supernatant was transferred into 1.5 mL plastic tube with 0.5 mL of 4 M potassium hydroxide then centrifuged at 13,000 ×g for 3 min. The supernatant was tested using the peroxidase enzyme reaction. We created a compound solution by mixing 60 mL of 12 mM m-dimethylamino benzoic acid with 0.5 mL of 1.3 M 3-methyl-2-benzothiazole nonhydrazone hydrochloride salt-hydrate. We placed 1430 µL of the compound solution in a glass cuvette then added 50 µL of the supernatant and 20 µL of peroxidase (0.25 Unit). The solution was analyzed using a U-1800 spectrophotometer (Hitachi, Tokyo, Japan). The absorbance as a result of the reaction was measured after 5 min in the U-1800 spectrophotometer at 590 nm. We created a standard curve using 0.125, 0.25, 0.50, and 1.0 µM H₂O₂ solutions. Experiments were repeated for the three batches of 14-day-old rice seedlings which were grown at the same condition.

Plant tissue (0.5 g) was ground to become a fine powder in liquid N₂ and then homogenized with 5 mL of 80% ethanol (v/v) using a mortar and pestle. 300 µL of 15 mM lactose was added to the homogenate as a control. The homogenate was boiled for 30 min at 80˚C and then centrifuged for 15 min at 10,000 ×g. Extracts were dried and dissolved in 0.5 mL of distilled water. We then added 0.5 mL of Dowex (Muromachi Gagaku Kokyo, Japan) to adsorb the suspension at 4˚C for 60 min. Galactinol and other sugars were analyzed by high-performance ionexchange chromatography (LC-10A system, Shimazu, Tokyo, Japan) and refractive index detector (RID10A, Shimazu) and an NH₂P-50 4E column (Shodex, Tokyo, Japan), which is a polyamine-bonded silica-based amino column. The flow rate was set to 0.7 mL∙min⁻¹, with acetonitrile-H₂O (75:25 v/v) as the carrier, according to the method of Ishibashi et al. [17]. All measurements were repeated for each extracts from the three batches of plants.

A batch of 14-day-old rice seedlings grown at 25˚C was pre-treated by 5 µM geldanamycin for 12 h, and the other batch of rice seedlings was pre-treated by 50 mM raffinose for 12 h. After the pre-treatment, the rice seedlings were subjected to high temperature at 52˚C for 3 h, then moved to control condition. The appearance of the rice seedlings were observed at 7 days after heat stress.

To analyze the expression of heat-inducible genes, rice seedlings were subjected to high temperature as sublethal level at 52˚C for 3 h. The expression of OsGolS1 increased drastically in response to the severe heat stress compared to expression levels in the control (Figure 1) and remained at high level when the rice seedlings were continuously exposed to heat stress (data not shown). The expression of OsHsfA1 remained at a constant level whereas OsHsfA2 was highly up-regulated under heat stress (Figure 1). The mRNA levels of OsGolS2 increased slightly after heat stress. However, the mRNA levels of OsGolS1 were significantly higher than those of OsGolS2, indicating that OsGolS1 plays a major role in the biosynthetic pathway of galactinol and the heat-stress response.

Previous studies on Arabidopsis revealed that GDA, an Hsp90 inhibitor, disassembles Hsp90-HsfA1 complexes, leading to the activation of HsfA1; this activated form of HsfA1 is necessary for the subsequent up-regulation of heat-inducible genes [10,18]. These findings prompted us to analyze the effects of GDA on the expression of OsHsfA2 and OsGolS1. Interestingly, under GDA treatments, the expression of OsHsfA2 and OsGolS1 increased remarkably (Figure 2). The induction of OsHsfA2 was observed at 3 h after GDA treatment followed by up-regulation of OsGolS1 at 8 h, and then the

expression of both genes increased up to 12 h in the case of OsHsfA2 and 24 h in the case of OsGolS1. In contrast, we observed no significant change in OsHsfA1 mRNA levels (Figure 2). These results indicate that OsGolS1 induction was correlated with the expression profile of OsHsfA2 which was induced under the GDA treatment.

To determine whether OsHsfA2 is involved in regulation of OsGolS1 expression, nucleotide sequence motives related to heat shock elements in the promoter region of OsGolS1 and their interaction with OsHsfA2 were examined (Figure 3). The inspection of the putative promoter region of OsGolS1 revealed the TATA boxproximal 5’ flanking region of OsGolS1 gene indicates that several HSE-like consensus motives potentially associating with HSFs are identified at positions indicated by #1-10 (Figure 3(a)). Particularly, position #4 has a sequence closely similar to the efficient HSE consensus [nGAAnnTTCnnGAAn] and other #1-3 and #5-10 also contain the core consensus motives of [nGAAn] and [nTTCn] (Figure 3(a)) [19]. In contrast, an amplified PCR fragment of OsHsfA3 promoter region contains only 3 sites of the core consensus motives in “TTCGAACCCGAA” closed to TATA-box. A purified MBP-OsHsfA2 purified from E. coli was confirmed at a relative molecular mass of approximately 86 kDa by SDS-PAGE (Figure 3(b)). Interaction of the MBP-OsHsfA2 with the promoter regions of OsGolS1 and OsHsfA3 was determined by gel-shift assay. The two shifted bands indicated that MBP-OsHsfA2 formed a complex with the OsGolS1 promoter region, while there no shifted band was observed with regard to a mixture of MBP-OsHsfA2 and OsHsfA3 promoter region (Figure 3(c)). In addition, no mobility shift according to the DNA-protein complex was observed in the mixture of MBP with the OsGolS1 promoter region. An increase of the OsHsfA2 protein amounts in the assay mixtures resulted in the uppershifted bands of the OsHsfA2/DNA-complex, indicating that the OsGolS1 promoter contains multiple HSE consensus sites or different OsHsfA2-binding affinities of the HSE consensus-like motives dependent on OsHsfA2

protein levels. These data revealed the binding activity of the recombinant OsHsfA2 to the promoter region of OsGolS1 at specific HSEs, suggesting that OsHsfA2 is involved in regulation of OsGolS1 expression in response to heat stress.

To determine whether OsHsfA1 is involved in transcriptional regulation of OsHsfA2 and/or OsGolS1, interactions of the promoter regions of OsHsfA2 and OsGolS1 with OsHsfA1 were examined by gel shift assay (Figure 4). A recombinant MBP-OsHsfA1 purified from E. coli was confirmed at the relative molecular mass of approximately 100 kDa by SDS-PAGE (Figure 4(a)). Interaction of the MBP-OsHsfA1 with the promoter regions of OsHsfA2, OsGolS1 and OsHsfA3 was compared (Figures 4(b) and (c)). Two regions from −10 to −190, and from −200 to −400 at upstream of the TATA box of OsHsfA2 gene were amplified by PCR as probe. Single and significant band shifts of both Pro-OsHsfA2 (10 - 190) and (200 - 400) in the presence of MBP-OsHsfA1 indicated that MBP-OsHsfA1 formed a complex with the OsHsfA2 promoter regions, while there no shifted band was observed with regard to a mixture of MBP-OsHsfA1 and OsHsfA3 promoter region (Figure 4(b)). In addition, no mobility shift according to the DNA-protein complex was observed in the mixture of MBP with the OsHsfA2 promoter regions. An increase of the OsHsfA1 protein amounts in the assay mixtures resulted in upper-shift and increased signal of the OsHsfA1/DNA-complex, indicating that the OsHsfA2 promoter contains multiple HSE consensus sites. As same as the OsHsfA2 promoter fragments, significant signals according to the DNA-protein complex composed of MBP-OsHsfA1 and OsGolS1 promoter fragment were detected as mobility shift in the mixture of MBP-OsHsfA1 with the OsGolS1 promoter fragment, while the bands of the protein-unbound OsGolS1 promoter fragment disappeared in the presence of MBP-OsHsfA1 (Figure 4(c)). The data revealed that the OsHsfA1 protein binds to the promoter regions of both OsHsfA2 and OsGolS1, suggesting that OsHsfA1 is involved in transcriptional regulation at upstream of OsHsfA2 and OsGolS1 in response to heat stress.

It has been convincingly demonstrated that heat stress

increases the production of reactive oxygen species (ROS) such as superoxide anions (), hydrogen peroxide (H₂O₂), and hydroxyl radicals (OH⁻) in plant cells [20]. When rice seedlings were subjected to heat stress, the endogenous H₂O₂ level in the rice seedlings increased significantly (p < 0.01), from 0.24 µmol∙g⁻¹ at 0 h (control level) to 0.43 µmol∙g⁻¹ at 3 h, and then decreased slightly to remain stable at approximately 0.35 µmol∙g⁻¹ from 6 h, which was not significantly different from the control level (Figure 5).

To examine whether the OsGolS1 induction was correlated with galactinol accumulation, we treated 14-dayold rice seedlings with heat stress to measure the profiles of galactinol and other carbohydrates with the time course based on the induction of OsGolS1 which was described in the method. Under heat stress, the galactinol content increased significantly (p < 0.01), from 0.44 mg∙gFW⁻¹ at 0 h to 1.6 mg∙gFW⁻¹ after 9 h, which represents an increase to approximately four times the original value (Figure 6(a)). Accordingly, raffinose content also increased significantly (p < 0.01) from 0.33 mg∙gFW⁻¹ at 0h or control level to 1.74 mg∙gFW⁻¹ at 9 h which was slightly higher than the level of galactinol at 9 h (Figures 6(a), (b)). The sucrose content was also highly accumulated from 3.25 mg∙gFW⁻¹ at 0 h to 9.03 mg∙gFW⁻¹ (nearly three times the initial level) by 9 h (Figure 6(c)). Glucose and fructose levels did not change significantly during the heat stress treatments (Figures 6(d) and (e)).

To determine the effects of GDA and raffinose pre-

Figur 5. Changes in endogenous levels of H₂O₂ in rice seedlings under heat stress. For the heat stress treatment, 14-day-old rice seedlings grown at 25˚C were subjected to heat stress at 52˚C for 1, 3, 6 and 9 h; 0 h represents control. Bar is standard error (n = 3, (**): p < 0.01). FW, Fresh Weight.

treatment on thermotolerance of rice seedlings, we pretreated 14-day-old rice seedlings grown at 25˚C with 50 mM raffinose for 12 h and pre-treated another group of plants with 5 µM GDA for 12 h before subjecting the plants to heat stress at 52˚C for 3 h. After the heat stress treatment, the rice seedlings were grown under control conditions for 7 days to observe their growth and appearance. The rice seedlings pre-treated with raffinose or GDA remained green and did not show severe leaf injury, whereas the non-treated rice seedlings exhibited severe bleaching after the heat stress (Figures 7(a) and (b)). The difference is particularly clear in enlarged pictures (Figure 6(c)). In combination with the data from our present experiments (Figures 1-6), this suggests that the induction of OsGolS1 expression in response to the GDA treatments or heat stress resulted in the accumulation of galactinol which is converted to raffinose. The intracellular accumulation of these carbohydrates may have acted as osmoprotectants that helped the rice plants cope with severe heat stress.

The accumulated evidence concerning on heat-shock proteins and heat stress signaling in Arabidopsis and other organisms has revealed that; i) under normal conditions, Hsp90 maintains its association with HsfA1 in the cytoplasm, so that HsfA1 remains inactive; ii) Subsequently, under heat stress, the recruitment of Hsp90 by heat-denatured polypeptides interrupts the interaction between Hsp90 and HsfA1, leading to trimerization and subsequent activation of HsfA1; iii) A trimer complex of HsfA1 recognized HSE in promoter regions of several heat-inducible genes as observed under heat stress [10,18, 19,21]. Treatment of Arabidopsis plants with GDA or radicicol, which are specific inhibitors for Hsp90, enhanced the expression of heat-inducible genes in the absence of heat stress [18]. In silico expression analysis revealed that a set of Hsf genes class HsfA, HsfB, HsfC families identified in the rice genome are regulated in response to high temperature, drought, and oxidative stress in complex manners [22]. However, the expression-correlation network among those rice Hsfs and putative target genes remains to be clarified.

HsfA1 is considered to be a house keeping gene in plants and a master regulator functioning on heat-induced expression of multiple HSF and HSP genes [1,18]. In the present study, the treatment of the rice seedlings with GDA resulted in the induction of OsHsfA2 and OsGolS1 expression, but the mRNA levels of OsHsfA1 re-

mained relatively constant (Figure 2). Interestingly, the mRNA level of OsHsfA2 appeared to be induced earlier than that of OsGolS1 in response to the GDA treatment (Figure 2). A study of the stress-responsive cis-element in the putative promoter region (within 1000 bp upstream of the coding region) of AtHsfA2 indicated the presence of HSEs that are recognized by some class A of the AtHsf family [10]. Our study on rice seedlings revealed that OsHsfA2 was up-regulated under both an Hsp90 inhibitor treatment and heat stress conditions (Figures 1, 2). Based on the expression profiles of OsHsfA2 and OsGolS1 under heat stress, it is hypothesized that OsHsfA2 is involved in the up-regulation of OsGolS1 in response to heat stress (Figure 8).

Previous studies have revealed that the expression of AtGolS1 and AtGolS2 in Arabidopsis is regulated by AtHsfA1a, AtHsfA1b, and/or AtHsfA2, which is also consistent with the results in transgenic Arabidopsis plants over-expressing AtHsfA2 whose AtGolS1 up-regulation was correlated with AtHsfA2 expression [5,8,9,23]. The GDA treatment on rice seedlings remarkably induced the expression of OsHsfA2, which was followed by the induction of OsGolS1. The gel shift assay revealed that recombinant MBP-OsHsfA2 bound to the promoter region of OsGolS1 containing potent HSE motives, resembling to a strong consensus [nGAAnnTTCnnGAAn], and a number of core consensus motives [nGAAn], [nTTCn]) resulting in the significant mobility shifts (Figures 3(a) and (c)) [19]. In addition, there was no interaction between the recombinant OsHsfA2 and the OsHsfA3 promoter region. It could be expected based on the previous observations that OsHsfA3 is induced under cold stress but not under heat stress [13,15]. Furthermore, gel-shift assay revealed that OsHsfA1 binds the promoter regions of OsHsfA2 and OsGolS1 (Figures 4(b) and (c)). In Arabidopsis, HsfA1d and HsfA1e appeared to be involved in induction of HsfA2, which is a key regulator for aquired tolerance under various environmental stresses [24]. Therefore, it is reasonable that dissociation from the Hsp90-OsHsfA1 under heat stress or GDA treatment activates OsHsfA1 protein, leading to the transcriptional activation of OsHsfA2 and resultant induction of OsGolS1. It can be assumed that a sequential transcription cascade composing of OsHsfA1-OsHsfA2-OsGolS1 play a pivotal role in aquired thermotolerance of rice seedling via galactinol synthesis (Figure 8). Alternatively, it is conceivable that heat stress and ROS differentially induces OsHsfA1 and OsHsfA2, leading to transcriptional activation of OsGolS1 by OsHsfA1-mediated OsHsfA2 induction and or by direct interaction of OsHsfA1 to the OsGolS1 promoter.

Recent studies have demonstrated that the metabolic pathways for sucrosyl oligosaccharides, including RFOs and fructans, function in the tolerance of oxidative stress via ROS signaling pathways [25]. It has been reported that the induction of GolS expression in Arabidopsis precedes the accumulation of galactinol, which has com-

patible solute-like protective abilities, thus reduces the intracellular ROS and oxidant levels [5]. This is consistent with the observation that the H₂O₂ contents in rice leaf tissues increased transiently at 3 h during the heat stress (Figure 5). As the heat stress continued, the H₂O₂ contents in the rice leaf tissues decreased to the control level and then remained stable (Figure 4). The expression of OsGolS1 appeared to be highly induced during heat stress and correlated with the accumulation of galactinol (Figures 1 and 5(a)). Raffinose content also increased at approximately as high as the level of galactinol (Figure 6(b)). These results suggest that after the transient increase in intracellular H₂O₂, the H₂O₂ was consumed by up-regulated scavenging systems such as galactinol and raffinose, which accumulated under heat stress [5]. The reduction of ROS in plant cells enables the plants to cope with high temperature stress. In the present study, we also observed a drastic increase in the content of sucrose, but no significant change in the levels of glucose and fructose under heat stress (Figures 6(c)- (e)). Therefore, it can be supposed that together with galactinol and raffinose, sucrose accumulation also plays an important role as an osmoprotectant in the thermotolerance of rice seedlings, even though the increase in sucrose may be regulated in different manner from galactinol.

It was reported that high temperature treatment on Lycoris enhances oxidative stress leading to acceleration of senescence [26]. The observation that raffinose treatment suppressed early senescence of rice seedling after heat stress indicates that ROS-induced GolS1 plays important roles in aquired tolerance against oxidative damage induced by heat stress (Figures 7). Oxidative damage to the photosynthetic apparatus of Arabidopsis under heat and drought stress appeared to be prevented by over-expression of GolS or by GDA-induced heat-shock proteins [5,18,27]. Those previous studies are consistent with our observations that rice seedlings pre-treated with raffinose or GDA showed less damage under heat stress than the controls (Figures 7(b) and (c)). These results indicate that raffinose or GDA treatments can protect photosynthesis apparatus of rice under severe heat stress. It was reported that transgenic Arabidopsis with introduced OsHsfA2e expressed AtGolS1 at significant levels [28]. The transgenic Arabidpsis overexpressing OsHsfA2e showed enhanced the thermotolerance, similar to those in Arabidopsis with over-expressing AtHsfA2 [5]. Therefore, our results indicate that heat stress or the Hsp90 inhibitor possibly induce the expression of OsHsfA2 by binding of OsHsfA1 to the OsHsfA2 promoter regions, leading to galactinol accumulation by the resultant OsGolS1 induction, and that heat stressand/or ROS-mediated HsfA2 induction is a critical signal component that exerts the protection mechanisms of plant cells against sub-lethal oxidative stress caused by heat stress (Figure 8).

Increasing evidence suggests that the HSFA2 is a central signal component regulating a set of gene expression essential for the aquired thermotolerance of higher plants. Therefore, identifications and molecular application technique of the signaling molecules at upstream of the HSFA2 are important for breeding of stress-tolerant crops.