Identification and expression analysis of OsHsfs in rice

Abstract

Heat stress transcription factors (Hsfs) are the central regulators of defense response to heat stress. We identified a total of 25 rice Hsf genes by genome-wide analysis of rice (Oryza sativa L.) genome, including the subspecies of O. japonica and O. indica. Proteins encoded by OsHsfs were divided into three classes according to their structures. Digital Northern analysis showed that OsHsfs were expressed constitutively. The expressions of these OsHsfs in response to heat stress and oxidative stress differed among the members of the gene family. Promoter analysis identified a number of stress-related cis-elements in the promoter regions of these OsHsfs. No significant correlation, however, was found between the heat-shock responses of genes and their cis-elements. Overall, our results provide a foundation for future research of OsHsfs function.

INTRODUCTION

Heat stress transcription factors (Hsfs) are the central regulators of defense response. They control the expression of heat shock proteins (HSPs) in plants by specific binding to the highly conserved heat shock element (HSE) characterized by palindromic motifs of nGAAn (Miller and Mittler, 2006). The basic structure and promoter recognition of Hsfs are highly conserved throughout the eukaryotic kingdom (von Koskull-Döring et al., 2007). A typical Hsf protein contains a modular structure with an N-terminal DNA-binding domain (DBD), an adjacent bipartite oligomerization domain composed of heptads repeat of hydrophobic amino acid residues (HR-A/B), a nuclear localization signal (NLS) essential for nuclear uptake of the protein, a nuclear export signal (NES), and in many cases a less conserved C-terminal activation domain (CTAD) rich in aromatic, hydrophobic and acidic amino acids (AHA) that have been reported to be crucial for activation function (Nover et al., 2001). Based on the conservative DBD and the HR-A/B regions, 21 putative Hsfs from the Arabidopsis, 23 from rice, and 18 from tomato have been identified through the genome-wide analysis (Baniwal et al., 2004). Plant Hsf gene family is divided into three classes, HsfA, HsfB, and HsfC, according to their protein structures (Nover et al., 2001). HsfA and HsfC have insertions of 21 and 7 amino acids, between the hydrophobic regions HR-A and HR-B, respectively. HsfB and HsfC are also characterized by lack of AHA motifs in their C-terminal regions (CTRs).

Evidence from several important representatives in tomato and Arabidopsis displayed that plant Hsfs diversify in their biological functions (Kotak et al., 2007b; von Koskull-Döring et al., 2007). HsfA members are capable of transcriptional activation, while HsfB members act as repressors or as co-activators (e.g., HsfB1) of HsfA members (Bharti et al., 2004; Czarnecka-Verner et al., 2004). However, AtHsfA4 activity was reported to be repressed by AtHsf5, which belongs to class A Hsfs (Baniwal et al., 2007). Overexpression of Hsf genes in transgenic plants resulted in an up-regulation of heat stress-associated genes and an enhancement of thermotolerance, whereas the down-regulation of Hsf genes leads to a reduction in the thermotolerance (Charng et al., 2007; Mishra et al., 2002; Schramm et al., 2008). In addition to control of heat stress response, Hsfs have also been reported to be involved in the defense response to pathogen attack, oxidative stress, heavy metals, dehydration and salinity, and in certain processes of development and differentiation (Larkindale et al., 2005; von Koskull-Döring et al., 2007).

Rice is the most important cereal crop, which feeds more than a half of the world’s population (Jeon et al., 2008). Molecular dissection of rice Hsf gene family would help to unravel the stress response mechanism in rice. Compared with the extensive studies done in Arabidopsis Hsf genes, only a few researches have involved monocots, such as rice and maize (Fu et al., 2006; Yamanouchi et al., 2002; Yokotani et al., 2008). Although 23 Hsfs were identified in the Oryza japonica previously, the structure and expression profile of these OsHsfs have not been elucidated. In this study, we identified and classified 25 rice Hsf genes from both O. japonica and O. indica genomes. In addition, the expression of the individual genes was investigated through both digital expression analysis and semi-quantitative reverse-transcript polymerase chain reaction (RT-PCR). Our work will facilitate the function analysis of the OsHsfs genes.

MATERIALS AND METHODS

Search for Hsf genes in rice genome and gene annotation

Consensus amino acid sequences of Arabidopsis heat shock factors, including the DNA-binding domain and HR-A/B region, were used to search the GenBank (National Center for Biotechnology Information (NCBI), Bethesda, MD, USA; http://www.ncbi.nlm.nih.gov), the International Rice Genome Sequencing Project (IRGSP; http://rgp.dna.affrc.go.jp), and Beijing Genomics Institute (BGI; http://btn.genomics.org.cn/rice), using an E-value cutoff of 1.0. The full length of cDNA sequence, 1 kb DNA sequence upstream of the start-codon, and bacterial artificial chromosomes (BACs) or phage artificial chromosomes (PACs) containing OsHsf genes were obtained from the NCBI, IRGSP, or BGI. Expressed sequence tag (EST) sequences of all OsHsf genes collected from dbEST database (http://www.ncbi.nlm.nih.gov/dbEST/) were used for the identification of the tissue specific expression patterns of the OsHsfs (Audic and Claverie, 1997). Finally, we compared all the OsHsf genes to identify redundant sequences. Promoters were analyzed by using PlantCARE (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/).

Sequence alignment and phylogenic analysis

Amino acid sequences of DBD and HR-A/B regions were used for multiple alignments by using ClustalX version 1.83 (Hicks et al., 1997). To produce preferable alignments, the parameters were set as followings: for pairwise parameters, gap opening cost=30, gap extension cost=0.3; for multiple parameters, gap opening cost=20, gap extension cost=0.15; the Gonnet series were applied for protein weight matrices and defaulting parameters were used for other settings. Phylogenetic tree of OsHsf gene family was constructed using the N-J method. The ScHsf1 gene from Saccharomyces cerevisiae was selected as outgroup and bootstrap analysis was performed to measure the robustness of all nodes.

Digital expression analysis

Digital expression of the OsHsfs was performed using the rice dbEST database. All ESTs were sorted by the library source, and normalized libraries were delimited for expression analysis. Frequencies of the ESTs in the corresponding library were calculated to represent the gene expression level.

RT-PCR analysis

Sixty rice seedlings (Oryza sativa L. cv. Nipponbare) were grown hydroponically with a 12-h photoperiod (250 µmol photons/(m²·s)) and a day/night temperature of 28/22 °C. The 7-d-old seedlings were transferred to a growth chamber at 37 °C for heat shock treatment or immersed in 20 mmol/L H₂O₂ solution for H₂O₂ stress induction. The shoots of the seedlings were sampled at different time points (0, 1, 6, and 24 h after treatments) for isolating RNA. Total RNA was extracted with Trizol reagent (Gibco-BRL, Gaithersburg, MD, USA) according to the manufacturer’s instructions. cDNA was synthesized from 5 μg of the total RNA using moloney murine leukemia virus (M-MLV) reverse transcriptase (Promega, Madison, WI, USA). Rice ACTIN gene OsACT1 (accession No. X63830) was served as a housekeeping gene control. Gene specific primers for corresponding OsHsf genes were designed for semi-quantitative RT-PCR analysis (Table A1).

RESULTS

Identification of OsHsf genes in rice genome

In order to identify rice Hsf genes, we used the conserved DBD and the HR-A/B regions to conduct basic local alignment search tool (BLAST) searches against the rice genome in NCBI, IRGSP, and GPI, and identified a total of 26 homologs in the O. japonica genome, which were highly similar to HSF proteins from other plants, and 27 in the O. indica. Of these Hsf genes, two gene loci Loc_Os06g66210 (O. japonica) and OsIBCD008764 (O. indica) were manually discarded because of the truncation in DBD region and the absence of corresponding ESTs in the dbEST, respectively. Another two Hsf genes from gene loci OsIBCD028792 and OsIBCD044667, which were previously annotated as two distinct genes, were proved to be identical in their cDNA sequences (Table 1). The remaining 50 sequences were analyzed for redundancy by performing pairwise comparisons of the genomic sequences in the coding regions ranging from start to the stop codon in each cultivar. Genes that share more than 95% similarity are considered as the same gene. It was found that all Hsf genes in O. indica have corresponding orthologs in O. japonica.

Table 1.

Characteristic of the 25 OsHsf genes

Genes	Gene loci		BAC/PAC clone	Protein accession No.	Accession No. of cDNA	Total EST	ORF (aa)	Intron (bp)
	O. japonica	O. indica
OsHsfA1	Os03g63750	OsIBCD012688	AC120506	NP_001051938	AK100430	50	506	1386
			AP008209		AK106118
OsHsfA2a	Os03g53340	OsIBCD011954	AC092558	AAP13005	AK069579	15	376	81
			AP008209		AK109590
OsHsfA2b	Os07g08140	OsIBCD022526	AP003826	NP_001059028	AK101824	10	372	1323
OsHsfA2c	Os10g28340	OsIBCD030592	AC027658	NP_001064617	AK072391	10	358	1669
			AP008216
OsHsfA2d	Os03g06630	OsIBCD008955	AC105729	NP_001049047	AK066844	6	359	946
			AP008209
OsHsfA2e	Os03g58160	OsIBCD012247	AC092076	NP_001051552	AK068660	50	357	1199
			AP008209
OsHsfA3	Os02g32590	OsIBCD006671	AP004777	NP_001047003	AK101934	7	498	1298
OsHsfA4a	Os01g54550	OsIBCD003303	AP003076	NP_001044247	AK109856	21	440	582
OsHsfA4d	Os05g45410	OsIBCD018681	AC111015	NP_001056127	AK100412	35	459	237
			AP008211
OsHsfA5	Os02g29340	OsIBCD006509	AP004999	NP_001046889	AK072210	16	475	1475
			AP008208		AK065643
OsHsfA6	Os06g36930	OsIBCD021096	AP005456	NP_001057889	−	−	331	122
OsHsfA7	Os01g39020	OsIBCD002361	AP003308	NP_001043378	AK064271	7	402	1406
OsHsfA9	Os03g12370	OsIBCD009329	AC107226	NP_001049429	AK072571	24	406	1231
			AP008209
OsHsfB1	Os09g28354	OsIBCD028792a	AP006057	NP_001063364	AK101182	36	302	5880
		OsIBCD044667a	AP008215		AK061433
OsHsfB2a	Os04g48030	OsIBCD015287	AL663003	NP_001053591	AY344483	4	305	102
OsHsfB2b	Os08g43334	OsIBCD027305	AP004163	NP_001062423	AK101700	29	390	101
OsHsfB2c	Os09g35790	OsIBCD029172	AP005681	NP_001063726	AK106545	19	454	121
					AK106525
					AK105409
OsHsfB4a	Os08g36700	OsIBCD026821	AP004693	BAD09860	−	−	380	93
OsHSFB4b	Os07g44690	OsIBCD024341	AP005292	NP_001060424	AK063952	26	310	1665
			AP008213		AK099354
OsHsfB4c	Os09g28200	OsIBCD028788	AP005655	NP_001063356	AK241190	5	394	88
OsHsfB4d	Os03g25120	OsIBCD010276b	AC125784	ABF96133	−	1	305	1450
		OsIBCD010275b	AP008209
OsHsfC1a	Os01g43590	−	AP002744	AAQ23067	AK069479	1	348	109
OsHsfC1b	Os01g53220	OsIBCD003218	AP003309	NP_001044160	AK066316	9	250	79
OsHsfC2a	Os02g13800	OsIBCD005620	AP004070	NP_001046370	AK106488	9	298	86
OsHsfC2b	Os06g35960	OsIBCD021036	AP003682	NP_001057843	−	1	278	1128

BAC: bacterial artificial chromosome; PAC: phage artificial chromosome; EST: expressed sequence tag: ORF: open reading frame.

^aSame genes with two cDNAs separated by intron;

^bTwo OsHsfB4d gene loci annotated in O. indica genome

Protein structure and classification of OsHsfs

Protein sequence analysis detected a conserved DBD in the N-terminal region (Fig.1) and an adjacent HR-A/B region in each of the OsHsf proteins (Fig.2). The DBD domain contained three α-helix bundles and a small four-stranded antiparallel β-sheet as previously described in LpHsf24 (Fig.1) (Schultheiss et al., 1996). There was a conserved intron located near the 3′-end of the third helix (Fig.1) with variant length (Table 1) in all the genes.

Fig. 1.

Multiple sequence alignment of DNA binding domains

All OsHsfs have only one intron between the invariant tyrosine residue and glycine residue, which is indicated by arrow. The secondary structure elements are showed below the sequence alignment: α₁ to α₃, α-helix; β₁ to β₄, β-sheet

Fig. 2.

Multiple sequence alignment of HR-A/B domains

Based on the distance between HR-A and HR-B domains, OsHsfs are grouped into three classes: (a) OsHsfA, (b) OsHsfB, and (c) OsHsfC. The hydrophobic residues and unique amino acids (glycine and proline) are colored black and gray respectively. The positions a and d of heptapeptide repeats are marked below

Further analyses of the HR-A/B domain revealed that there were, in the majority of cases, three or more repeated heptads in the HR-A domain and two incompletely repeated heptads in the HR-B domain (Fig.2). HR-B in classes A and C had a very characteristic structure of overlapping sets of hydrophobic heptapeptide repeats (Fig.2). According to the linker region between the HR-A and HR-B domains, OsHsfs family (Nover et al., 2001) was divided into three classes, including 13 OsHsfAs, 8 OsHsfBs, and 4 OsHsfCs (Table 2).

Table 2.

Function motifs of rice OsHsfs

Hsfs	NLS	NES	AHA motifs
HsfA1	(231)RRR-5aa-KKRRLPK	(493)LTEQMGLL	AHA (449)DSFWEQFLVA
HsfA2a	(255)RNR-7aa-KKRRR	(366)LVQSIYHL	AHA (318)ESFWMQLLSL
HsfA2b	(239)RK-7aa-KKRRRR	(356)LSEKMGYL	AHA (328)DNFWEELLNE
HsfA2c	(227)KEKRK-7aa-KKRRR	(345)LAQQLGYL	AHA (315)DDFWAELLVE
HsfA2d	(223)KMK-7aa-KKRTR	(346)LAQKLGYL	AHA (315)DDFWEELLNE
HsfA2e	(224)RK-7aa-KKRRRR	(340)LLSQKMGYL	AHA (314)EDFWEDLLHE
HsfA3	(247)RQQK-7aa-KRKFLK	(472)LDDGDLQL	AHA (463)KFWELDFQAL
HsfA4a	(200)RKKRR	(429)EKLGLL	AHA1 (375)DGFWQQFLTE
			AHA2 (416)HLWWGKRNVE
HsfA4d	(210)KKRRVPK	(445)EQMGHL	AHA (397)DVFWERFLTE
HsfA5	(214)NKKRR	(470)VEQLKL	AHA (422)DKFWEQFLTE
HsfA6	(231)RKKRR	(319)LVQQIDCL	AHA1 (271)MIWYELLGEE
			AHA2 (306)EPWEEMGEEE
HsfA7	(282)KNGLRGAAKRQR	−	AHA1 (349)DDVWEELDAL
			AHA2 (384)CGWVDDCPYL
HsfA9	(246)QRR-9aa-KKRR	(371)QMGPPL	AHA (380)DYDFPQLEQD
HsfB1	(263)RKRAR	−	−
HsfB2a	(265)REGKVRR	(275)LSDLDVLAL	−
HsfB2b	(318)RKRMR	−	−
HsfB2c	(327)LKRTR	−	−
HsfB4a	(346)RKRS	−	−
HsfB4b	(291)RKKRAHR	−	−
HsfB4c	(269)RKKP	(363)LVLECDDLSL	−
HsfB4d	(286)KKRRVQL	−	−
HsfC1a	(229)RKRRR	−	−
HsfC1b	(209)KRRR	−	−
HsfC2a	(213)KRPRLLL	−	−
HsfC2b	(219)KRARLLL	−	−

Numbers in brackets indicate positions of the putative function motifs; NLS: nuclear localization signal; NES: nuclear export signal; AHA: aromatic, large hydrophobic and acidic amino acids. Aromatic and large hydrophobic residues of AHA are set in boldface type

Within the OsHsfs, the nuclear localization signals (NLS) composed of a cluster of arginine and lysine residues were detected adjacent to the HR-A/B domain (Table 2). There was a leucine-rich nuclear export signal (NES) at the C-terminus of all OsHsfAs except OsHsfA7, and of a small portion of OsHsfBs and OsHsfCs (Table 2). The C-terminal activation domain (CTAD) was only found in OsHsfA, suggesting that only class A Hsfs can activate autonomously.

Phylogenetic analysis of OsHsfs

To determine the phylogenetic relationship among the OsHsfs, neighbor-joining phylogenetic trees were constructed using the amino acid sequences of DBD, the HR-A/B region, and the linker between them (Fig.3). Arabidopsis Hsf genes were included in the phylogenetic tree as reference to classify rice Hsfs. As expected, the classes A, B and C Hsfs formed three individual clusters. Furthermore, the class A Hsfs were divided into two sub-clusters (Fig.3). In a previous study, the N-terminal part and C-terminal part of DBD and HR-A/B regions were used separately to draw phylogenetic trees (Nover et al., 2001). Although most proteins fixed their positions in the different phylogenetic trees, a few Hsfs changed their positions (Nover et al., 2001). Similar phenomenon was also observed on the OsHsfs (data not shown). A more convinced relationship of the Hsfs was revealed by combining the DBD, HR-A/B, and the flexible linker between DBD and HR-A/B (Fig.3).

Fig. 3.

Neighbor-joining phylogenetic trees of OsHsf and AtHsf genes constructed using ClustalX program

The tree was generated on the basis of the amino acid sequences of the N-terminal domains of Hsfs including the DBD region, the HR-A/B region and parts of the linker between both. ScHsf1 was set as the outgroup. Bar=substitutions/site

Cis-acting elements in OsHsf promoters

One kilobase upstream regions of the OsHsf genes were analyzed for cis-acting elements using PlantCARE software (Higo et al., 1999). A total of 9 different cis-acting elements commonly existed in the promoters of OsHsfs (Table 3). Seven of these cis-acting elements were related to stress responses. Two cis-acting elements with the highest frequency were G-box/Sp1 and abscisic acid (ABA) response element (ABRE). These two elements are involved in the light responsiveness and the ABA response, respectively (Table 3). CCGTCC/CAT-box, LTR, and ARE/GC-motif are cis-acting elements related to meristem expression, low temperature response, and anaerobic induction. Compared with classes B and C OsHsf genes, class A genes contained more these two cis-elements, suggesting that class A genes may play different roles from classes B and C genes (Table 3). In addition to these two elements, other cis-elements occurred at a similar frequency among the promoters of these three classes.

Table 3.

The high frequency cis-elements of OsHsf genes

Cis-elements	HsfA													T1	HsfB								HsfC				T2	Descriptions
	1a	2a	2b	2c	2d	2e	3	4a	4d	5	6	7	9		1	2a	2b	2c	4a	4	4c	4d	1a	1b	2a	2b
Skn-1 and GCN4-motif	2	2	0	1	0	0	1	1	1	1	1	1	4	15	1	1	3	0	0	3	0	2	1	4	2	0	17	Required for endosperm expression
CCGTCC and CAT-box	2	1	2	1	1	0	0	1	1	1	1	3	0	14	1	0	0	0	0	0	2	0	0	2	0	1	6	Related to meristem expression
G-box and Sp1	5	8	17	12	3	6	14	7	3	3	17	6	5	106	10	7	2	10	7	5	7	7	13	8	7	19	102	Involved in light response
HSE	0	3	1	2	2	2	0	1	0	0	2	1	0	14	3	1	2	1	0	1	1	0	2	0	1	1	13	Involved in heat stress response
LTR	0	2	1	1	1	0	1	0	1	0	0	0	0	7	0	0	0	0	1	1	0	0	0	0	0	0	2	Involved in low-temperature response
ARE and GC-motif	2	5	4	3	2	1	4	1	1	2	5	3	0	33	4	1	1	2	1	0	3	0	1	2	0	3	18	Involved in the anaerobic induction
MBS and CCAAT-box	1	1	0	1	2	1	0	2	1	1	3	2	1	16	1	1	1	2	3	2	1	3	1	1	0	1	17	MYB binding site, involved in drought-inducibility
ABRE	0	1	3	4	2	3	8	0	1	0	14	0	1	37	0	2	0	5	11	3	0	5	5	2	1	3	37	Involved in the abscisic acid response
CGTCA-motif	0	2	0	0	1	2	4	4	0	0	1	0	4	18	1	3	0	0	0	2	1	0	2	2	2	0	13	Involved in the MeJA-response

T ₁: total number of each cis-element in HsfA; T ₂: total number of each cis-element in HsfB and HsfC

Expression patterns of OsHsfs

The number of ESTs for a specific gene in a cDNA library is considered to be proportional to the transcript abundance of the mRNA. We used the EST sequences downloaded from the dbEST to analyze the expression pattern of OsHsfs in various plant tissues and different developmental stages. A total of 391 OsHsf ESTs corresponding to 23 of the 25 genes were identified except the OsHsfA6 and OsHsfB4a genes (Table 1). The majority of OsHsf genes had a low to moderate level of transcripts in most tissues, while some displayed a tissue-specific manner (Fig.4). OsHsfA4a and OsHsfA4d were dominantly expressed in panicle and adult leaves, respectively. Overall, OsHsfs were expressed at a higher level in panicle and flower than in other tissues.

Fig. 4.

Expression level of OsHsfs in different tissues

ESTs of each Hsf are collected from dbEST and the transcript rations of all Hsfs in each library are calculated. Transcript levels of all OsHsfs in per million transcripts are counted according to the ration. Normalized libraries and libraries with less than 5000 ESTs are excluded. Detail data are shown in Table A2

Under heat shock (HS) and H₂O₂ treatment, the expression of class A Hsfs was higher than those of class B and class C Hsfs (Fig.5). These results are consistent with those derived from the analyses of EST data. The lack of the detectable transcripts of several OsHsfs under our condition might have resulted from the tissue specific expression pattern (OsHsfA4d) or low expression levels of the genes (e.g., OsHsfB4a, OsHsfB4d, and so on) as suggested by the digital expression pattern. The transcription of OsHsfA1a, OsHsfA2b, OsHsfA3, OsHsfA7, OsHsfA9, OsHsfB2c, and OsHsfC1b was up-regulated by both HS and H₂O₂ treatments in a similar manner, while the up-regulation of OsHsfA2a, OsHsfA2c, OsHsfA2e, OsHsfA4a, OsHsfB2b, and OsSHFC1b was detected only under HS treatment, implying that there was a H₂O₂ independent HS responsive pathway in rice.

Fig. 5.

Expression levels of OsHsfs under heat shock (HS) and oxidative stress (H₂O₂) analyzed by RT-PCR

mRNA was extracted from shoots of 7-d-old O. japonica seedlings treated with heat (37 °C) and 20 mmol/L H₂O₂ for different time points. The transcript level of OsACTIN1 was used as the housekeep gene control. Class A Hsfs and ACTIN genes were amplified for 28 cycles, while classes B and C Hsfs were amplified for 30 cycles

DISCUSSION

Unlike yeast and Drosophila that only contain one Hsf gene, plant has a more complex family of Hsf genes (Baniwal et al., 2004; Miller and Mittler, 2006). In this study, we identified 25 OsHsfs based on the rice genome. Majority of the predicted OsHsfs were supported by the FL-cDNA and the EST sequences (Table 1). These 25 OsHsfs were divided into 3 classes (A, B and C) based on the protein structure and phylogenetic relationship (Figs.2~3). The sequence analysis identified the conserved DBD, the HR-A/B domain, and the NLS that were the common motifs of OsHSFs. The NES and the AHA motifs were also the common motifs of class A genes, but they were only detected in a small portion of classes B and C genes (Table 2).

Recently, great progress has been made in elucidating the functions of Hsfs (von Koskull-Döring et al., 2007). AtHsfA9 is involved in seed development and controlled by seed-specific transcription factor abscisic acid-insensitive 3 (ABI3) (Kotak et al., 2007a). The expression of AtHsfA3 is induced by drought and heat, and is dependent on the dehydration response element biding protein 2A (DREB2A) (Sakuma et al., 2006; Schramm et al., 2008). Promoter analysis showed that an array of OsHsf genes contains ABREs (Table 3). This indicates that an ABA pathway may be involved in Hsf induction and responsible for control of downstream processes, such as seed development or drought resistance (Kotak et al., 2007b). The transcripts of AtHsfA2 increased under several stress conditions, especially in response to high light plus heat stress (Nishizawa et al., 2006). During repeated cycles of heat stress and recovery, HsfA2 becomes a dominant Hsf in tomato and Arabidopsis (Mishra et al., 2002; Schramm et al., 2006). In contrast to tomato and Arabidopsis containing only one HsfA2, rice has five HsfA2 genes (von Koskull-Döring et al., 2007). Our results show that OsHsfA2a, OsHsfA2b, OsHsfA2c, and OsHsfA2e were induced by heat stress, while OsHsfA2b and OsHsfA2c were induced by H₂O₂ also (Fig.5). From the result of expression pattern, the OsHSfA2 subfamily was closely related with rice stress response. Overexpression of OsHsfA2e in Arabidopsis led to enhanced thermo and salt tolerance in transgenic plants (Yokotani et al., 2008). Further studies to elucidate the functions of the other members in the subfamily may also have potential for the development of transgenic plants with improved stress tolerance.

H₂O₂ is reported to be an essential component in the heat stress signaling pathway (Volkov et al., 2006), and the Hsf genes were considered as H₂O₂ sensors in plant (Miller and Mittler, 2006). Human and Drosophila Hsf1 was shown to sense hydrogen peroxide directly and assemble into a homotrimer in a reversible and redox-regulated manner (Storozhenko et al., 1998; Zhong et al., 1998). In Arabidopsis, the heat stress-induced H₂O₂ is required for effective expression of heat shock genes (Volkov et al., 2006), and the AtHsf4a may be an important sensors of H₂O₂ (Davletova et al., 2005). The intimate relationship between the HS and oxidative stress response suggests that some Hsfs might be responsive to both stresses. In this study, eight OsHsfs (OsHsfA1a, OsHsfA2b, OsHsfC1b, OsHsfA3, OsHsfA7, OsHsfA9, OsHsfB2c, and OsHsfC1b) were found to be induced by both heat shock and H₂O₂ (Fig.5). Heat stress led to the accumulation of H₂O₂ in tobacco and Arabidopsis culture cells (Volkov et al., 2006) and mustard seedlings (Dat et al., 1998). Hence, it is likely that Hsfs sense the H₂O₂ level in cells and transfer the stress signaling. Mutation of OsHsfA4d led to a lesion mimic phenotype in mature leaves (Yamanouchi et al., 2002). The phenotype may be caused by H₂O₂ accumulation that transferred a cell death signaling (Takahashi et al., 1999). It is likely that OsHsfA4d works in mature leaves to sense H₂O₂ levels as its homologue AtHsfA4a in Arabidopsis (Davletova et al., 2005). It is also worth noting that OsHsfA4a, the other member of OsHsfA4, has extremely high transcript abundance in panicle (Fig.4). Further research will help to elucidate the special functions of these genes.

APPENDIX: SUPPORTING ONLINE MATERIAL FOR IDENTIFICATION AND EXPRESSION ANALYSIS OF OsHsfsIN RICE

Table A1.

Gene specific primers used in this research

Hsfs	Forward (5′ to 3′)	Reverse (5′ to 3′)
OsHsfA1a	TATGCCAAATGGTCAAGGTC	CAGGCAAAGAGAATACACCC
OsHsfA2a	GGCTTCCTCCAGATGCTCGTC	CGTCGTCATCCTCCTCGTCGTT
OsHsfA2b	AGCAGAGGCAACAGCAGATG	CAGACCTCCAGCATCCATAG
OsHsfA2c	GCAGAAACAGGTCCAGATG	TCTACTTTACCCTTCCCCAG
OsHsfA2d	GAGGTTGGTCAGTTCGGATT	GGTTGAGAAATGGCACTATGT
OsHsfA2e	ATGGCATTCTTGTCACGAGT	GGTTCCTGGTATCCTCATCG
OsHsfA3	CAACACACTGAGAAGGGAGA	TGCTCTCTCCAGTGTGTTGT
OsHsfA4a	TGGCAGCAGTTTCTTACCGAG	AGGCACCAATGTCAGCGTTC
OsHsfA4d	CCCATCTCCATTTATCCACT	CATTTGCTCGGTGATCTGAT
OsHsfA5	GTAAGCCTATCCACAGCCAC	ATCTTGGTCTGTCGCTGCTC
OsHsfA7	TCCGAAAGGTCACTCCAGAT	TGGTCTGCTGTTGCCTTCTC
OsHsfA9	GGCACTACCAGCAAACATC	CCACTGGATTTACTTGACCT
OsHsfB1	GGACAACCAAACGCTGACGA	CGACGATGGAACGCTGACC
OsHsfB2a	GGCATACCGACGGCGATACC	CCTTCCTCCTCCTCCTCCTCCT
OsHsfB2b	GGCTGCTCTGCGAGATACACC	CTGCTGGGTGGAGGCGTACTTG
OsHsfB2c	GCAACAGAGCAACTTGTGA	CACCAACACACACACAAAGC
OsHsfB4b	GACGACGACGACGACGAT	TCATCACTCTTCACCAGCAT
OsHsfC1b	TACTTCAAGCACCGCAACT	CCGCTGCACCGCCTCGATG
OsHsfC2a	TGTTTGGACGAAGTGGCTG	ACAAGGCACACTCAACATTC
OsACTIN	TCAGCAACTGGGATGATATGGAG	GCCGTTGTGGTGAATGAGTAAC

Table A2.

Expression level of OsHsfs in different tissues

Gene	Transcripts per million
	Callus	Shoot	Adult leaf	Root	Seed	Flower	Panicle	Mix
OsHsfA1a	168	190	0	58	102	108	260	269
OsHsfA2a	10	10	0	0	0	43	80	136
OsHsfA2b	56	0	0	0	0	43	166	0
OsHsfA2c	37	0	0	0	0	43	0	30
OsHsfA2d	10	41	76	0	0	22	0	0
OsHsfA2e	19	82	0	0	102	65	332	0.018
OsHsfA3	37	61	0	0	0	0	0	0
OsHsfA4a	0	45	0	0	0	141	2268	0
OsHsfA4d	0	21	1892	0	0	43	215	30
OsHsfA5	0	61	0	58	0	0	432	170
OsHsfA7	143	0	0	0	0	0	166	142
OsHsfA9	37	46	0	58	0	43	216	280
OsHsfB1	37	0	0	523	953	0	166	85
OsHsfB2a	0	0	0	10	0	0	0	136
OsHsfB2b	37	108	0	0	0	43	0	136
OsHsfB2c	56	93	0	58	0	0	0	136
OsHsfB4b	75	139	0	58	183	117	216	30
OsHsfB4c	19	0	0	0	0	0	0	0
OsHsfB4d	67	0	0	0	0	0	0	0
OsHsfC1a	0	0	0	0	0	22	0	0
OsHsfC1b	19	41	0	116	0	141	338	42
OsHsfC2a	10	46	0	0	0	0	80	61
OsHsfC2b	0	0	0	0	0	264	0	0