High temperatures induce the generation of reactive oxygen species (ROS) in plants. ROS function in signaling and, in excessive amounts, cause oxidative damage to proteins, lipids, and nucleic acids, which impairs cellular and metabolic functions and leads to cell death (Miller et al., 2010). This universal phenomenon is known as heat stress (HS)-induced oxidative stress (Vallelian-Bindschedler et al., 1998). Understanding how plants cope with HS-induced oxidative stress during growth is crucial for improving HS endurance in crops to increase agricultural productivity (Hall, 2001).

The HS response (HSR) is primarily regulated at the transcriptional level by heat shock factors (Hsfs), which are activated by HS. Activated Hsfs specifically bind to palindromic heat shock elements (HSEs: 5′-nGAAnnTTCn-3′), leading to the transcription of these genes. This process functions during both basal and acquired HSRs in plants (Xuan et al., 2010). In the Arabidopsis thaliana genome, a total of 21 Hsfs genes have been identified, encoding Hsfs assigned to three major classes (A, B, and C) based on the characteristics of their oligomerization domains (Nover et al., 2001). Arabidopsis class A Hsfs function as positive regulators of environmental stress-responsive genes, which results in the constitutive expression of HS proteins (HSPs) and other proteins in biochemical pathways that increase thermotolerance levels (Panikulangara et al., 2004). By contrast, class B Hsfs may act as coactivators and/or repressors, while no clear activation or repression functions have been identified for class C Hsfs (Ikeda et al., 2011).

Arabidopsis contains 15 class A Hsfs, including HsfA1a and HsfA1b, which play important roles in the early phase of the HSR (Nover et al., 2001), as well as HsfA2, which maintains HSP expression during extended acquired thermotolerance responses (Charng et al., 2007). HsfA3 transcript levels increase under HS and excess light, and the hsfA3 T-DNA insertion mutant shows substantially reduced thermotolerance, while overexpression of HsfA3 increases thermotolerance (Jung et al., 2013; Yoshida et al., 2008). In vivo and in vitro experiments have demonstrated that HsfA3 functions directly downstream of dehydration-responsive element-binding factor 2A (DREB2A) and/or DREB2C, important transcription factors involved in plant responses to heat and salt stress (Chen et al., 2010; Yoshida et al., 2008). In addition to regulating HSP genes, DREB2-induced HsfA3 directly regulates cytosolic ascorbate peroxidase 2 (APX2), encoding a key enzyme in the ROS scavenging system of Arabidopsis (Hwang et al., 2012; Jung et al., 2013; Shin et al., 2013). Additionally, HsfA3 regulates the expression of many heat-inducible genes in the transcriptional cascade downstream of the K homology (KH) domain-containing nucleus-localized putative RNA-binding protein (RCF3) stress-regulatory system, an important upstream regulator of the HSR in Arabidopsis (Guan et al., 2013). These findings suggest that HsfA3 plays important roles in the regulation of HS-induced oxidative stress signaling; however, such roles remain to be demonstrated.

HS-induced oxidative stress brings about the accumulation of a number of metabolic intermediates in plant cells that act as antioxidants and/or signals (Nishizawa et al., 2008). Arabidopsis HsfA1b and HsfA2 directly regulate galactinol synthase (GolS) 1 and GolS2, encoding a key enzyme in the biosynthesis of raffinose family oligosaccharides (RFOs), which function as osmoprotectants and/or antioxidants in the abiotic stress tolerance response (ElSayed et al., 2014; Panikulangara et al., 2004). Galactinol is formed from UDP-galactose and myo-inositol via the activity of GolS, a key regulator of this pathway (Keller and Pharr 1996). As GolS genes are Hsf-dependent, overexpressing these genes increases tolerance to drought, high salinity and osmotic stresses in Arabidopsis (Busch et al., 2005; Sun et al., 2013; Taji et al., 2002). There is substantial evidence that galactinol and RFOs function as signals that mediate stress responses (Kim et al., 2008; Valluru and den Ende, 2011). However, such roles for galactinol and RFOs remain to be demonstrated (ElSayed et al., 2014).

In the current study, we determined that oxidative damage triggers the expression of HsfA3, which is correlated with the upregulation of GolS genes. These findings suggest that HsfA3 activates the expression of GolS genes, which in turn regulate the biosynthesis of galactinol to increase oxidative stress tolerance in plants.

Plant materials and growth conditions

Arabidopsis thaliana L. Heynh. ecotype Columbia (Col-0) plants (WT) and HsfA3-overexpressing transgenic Arabidopsis lines were grown on phytohormone-free MS medium (MSO) containing 2% sucrose and 0.25% Phyta-gel (pH 5.8) at 22°C under a 16 h light/8 h dark cycle (light intensity 100 μE m^?2 s^?1). To induce synchronous germination, the seeds were primed at 4°C for 3 days in the dark, followed by transfer to a growth chamber (Hwang et al., 2012).

A T-DNA insertional mutant line containing a single T-DNA insertion in the HsfA3 gene was identified in the SALK T-DNA collection (SALK_011107). To identify mutants homozygous for the T-DNA insertion, genomic DNA was obtained from seedlings and subjected to PCR genotyping using the following HsfA3 primer sets: HsfA3 forward (P1: 5′-ATGAGCCCAAAAAAAGATGC-3′) and reverse primer (P2: 5′-CTAAGGATCATTCATTGG-3′); and T-DNA right (P3: 5′-TGGGAAAACGGGCGTTACCCAACTTAAT-3′) and left border primer (P4: 5′-GTGATGGTTCACGTAGTGGGCCATCG-3′)(Supplementary Figs. 1A and 1C).

RNA extraction and reverse transcription (RT)-PCR

Total RNA was isolated from 10-day-old WT and HsfA3 transgenic lines using TRIzol reagent (Invitrogen, USA). Complementary DNA synthesis and RT-PCR were performed as described by Hwang et al. (2012) using HsfA3- and GolS-specific primer pairs (Supplementary Table 1). PCR products were sequenced to confirm that the amplified sequences were identical to the predicted sequences of the respective mRNAs based on Arabidopsis genomic data. Relative levels of HsfA3 and GolS expression were determined using ImageJ software (http://rsb.info.nih.gov/ij).

Construction of transgenic plants

To amplify full-length HsfA3 cDNA, RT-PCR was performed as described above (in Plant materials and growth conditions) using P1 forward and P2 reverse primers, and the RT-PCR product was cloned into pGEM-T Easy (Promega, USA) using TA-overhangs. The integrity of the construct was verified by sequencing. Subsequently, HsfA3 cDNA was digested with XbaI and BamHI, and the resulting 1,239 bp fragment was inserted between the cauliflower mosaic virus (CaMV) 35S promoter and nopaline synthase terminator in pBI121 (Clontech, USA) (Supplementary Fig. 1B). The recombinant plasmid (35SΩ:HsfA3) was introduced into Agrobacterium tumefaciens LBA4404 and used to transform Arabidopsis plants via the floral dip method (Martinez-Trujillo et al., 2004). Homozygous T₃ lines containing a single T-DNA insertion were used for further analysis as described by Lim et al. (2007) (Supplementary Fig. 1D).

Stress treatments

To analyze the transcriptional expression of HsfA3 in response to oxidative stress, primed WT seeds were germinated on MSO medium without or with 20 mM H₂O₂ (Sigma-Aldrich, USA) or 25 μM 2,5-dibromo-3-methyl-6-isopropyl-p-benzoquinone (DBMIB; Santa Cruz Biotechnology, USA). The seeds of WT and transgenic Arabidopsis plants were harvested and stored as described by Je et al. (2014).

To analyze the oxidative stress tolerance of plants by phenotypic comparison, primed Col-0, transgenic 35SΩ:HsfA3 (L12 and L20), and HsfA3 knockout mutant (hsfa3) seeds were germinated on MSO plates containing H₂O₂ (4, 5, and 6 mM) or methyl viologen (MV; 0.3, 0.5, and 0.7 μM). The resulting plants were allowed to grow for 10 days before analysis.

Fresh weight, chlorophyll fluorescence, and ion leakage assay

Untransformed WT, 35SΩ:HsfA3, and hsfa3 seeds were germinated and grown on MSO medium under normal growth conditions and exposed to oxidative stress treatments. After 10 days, the plants were weighed and their chlorophyll fluorescence was measured according to Yu et al. (2002). The plants were dark-adapted for 15 min just before harvest, and chlorophyll fluorescence was measured at room temperature using a plant efficiency analyzer (Handy PEA, Hansatech Instruments, UK). Ion leakage was measured according to Scarpeci et al. (2008). Ten-day-old Col-0 and transgenic plants grown on solid MSO plates as above were uprooted, thoroughly rinsed with de-ionized water, and suspended in de-ionized water. The conductivity of the suspension solution was measured with a conductance meter (Orion 3-Star Plus, Thermo Scientific, USA) before and after autoclaving (121°C for 15 min) to release all electrolytes.

Transient transactivation assay

The effector construct contained HsfA3 cDNA fused in-frame at its N-terminus to a Flag-tag sequence (F), which was expressed constitutively from a chimeric promoter consisting of the CaMV35S promoter and the TMV translation enhancing omega element (Ω). This construct was designated 35SΩ:FHsfA3 (Fig. 4A). For reporter constructs, putative promoter sequences of the GolS genes were amplified by PCR from Col-0 genomic DNA using GolS gene-specific primer sets (Supplementary Table 2). An approximately 1.0 kb fragment of each of the GolS upstream regulatory regions was inserted upstream of the β-glucuronidase reporter gene (GUS) in a pUC19-derived plasmid vector (Kim et al., 2007) to yield reporter constructs Prom_GolS1:GUS, Prom_GolS2:GUS, Prom_GolS3:GUS, and Prom_GolS4:GUS (Fig. 4A).

Transient transactivation assays were performed as described by Song et al. (2014) using Arabidopsis leaf protoplasts (300 μl, 5 × 10⁶/ml), 15 μg of effector construct, 15 μg of GUS reporter construct, and 5 μg of Prom_35S:LUC construct. Quantitative GUS activity assays were performed on protein extracts using the fluorogenic substrate 4-methy-lumbellyferyl-β-D-glucuronide (4-MUG; Sigma-Aldrich, USA), as described previously (Jefferson et al., 1987). Luciferase assays were conducted using the Luciferase Assay System (Promega, USA) and measured on a 20/20ⁿ Luminometer (Turner Bio-Systems, USA) as described by Chen et al. (2010). Promoter activity was calculated as units of GUS activity per unit of luciferase activity.

Electrophoretic mobility shift assay (EMSA)

Electrophoretic mobility shift assays were performed as described by Chen et al. (2010). Synthetic double-stranded (ds) oligonucleotides spanning the HSE in each of the native GolS1, GolS2, and GolS4 promoters were designated GolS1-wHSE, GolS2-wHSE, and GolS4-wHSE, respectively (Fig. 4). One variant of each ds-oligonucleotide probe, containing mutations in the HSE, was also synthesized; these probes were designated GolS1-mHSE and GolS2-mHSE (Fig. 4A). ³²P-labeling of the probes and purification of recombinant glutathione-S-transferase (GST)-HsfA3 were carried out as described by Lim et al. (2007).

Measurement of galactinol levels

Galactinol was extracted and measured according to Nishizawa et al. (2008). The sixth and seventh leaves of 4-week-old plants were weighed to obtain fresh weight (FW), ground to a fine powder in liquid N₂, and homogenized in 10 ml of 80% ethanol at 80°C. The homogenate was boiled for 10 min at 90°C, centrifuged for 10 min at 7,000 × g, and extracted twice in 1 ml of 80% ethanol at 90°C. The extracts were dried and dissolved in water (0.5 mg/ml). Galactinol levels were analyzed by high performance anion exchange chromatography using a CarboPac PA100 column (3 × 250 mm) on a Dionex ICS-3000 gradient system coupled with pulsed amperometric detection system (Thermo Scientific). Fifteen microliter samples were injected onto the CarboPac PA100 column and separated in a NaOH gradient. The flow rate was 1 ml/min. Identification and quantification were performed against a galactinol chemical standard (Fluka, USA).

HsfA3 transcription is induced by oxidative stress treatment

To verify the effect of oxidative stress on HsfA3 expression, we exposed 10-day-old WT plants to exogenous H₂O₂ or to the endogenous H₂O₂ propagator DBMIB (?lesak et al., 2003) under low light conditions. As determined by RT-PCR, HsfA3 transcript levels reached a peak at 3 h of H₂O₂ treatment, followed by a rapid decline. The peak HsfA3 transcript level was approximately 7-fold higher than that under untreated conditions (Fig. 1). Although its expression levels were lower than under H₂O₂ treatment, HsfA3 expression rapidly increased within 1 h of DBMIB treatment, followed by a rapid return to baseline levels after 6 h of treatment. Other studies have also demonstrated that treatment with the superoxide anion propagator MV, or low light plus DBMIB, rapidly induces the expression of HsfA3 (Guan et al., 2013; Jung et al., 2013). These findings demonstrate that HsfA3 is an oxidative stress-responsive gene, and they suggest that the transcription factor HsfA3 regulates the HS-induced oxidative stress response (Chen et al., 2010).

Overexpressing HsfA3 increases oxidative stress tolerance in Arabidopsis

Since the expression of HsfA3 was triggered by oxidative stress, we investigated whether HsfA3 overexpression is associated with increased oxidative stress tolerance. We compared a number of morphological phenotypes of WT and HsfA3 transgenic Arabidopsis plants, including an HsfA3 overexpressor (35SΩ:HsfA3) and knockout mutant (hsfa3), which were grown in vitro in the absence or presence of various concentrations of H₂O₂ or MV (Fig. 2). Under normal growth conditions (no H₂O₂ or MV), the HsfA3 transgenic plants exhibited no differences in morphology or growth compared to WT (Figs. 2A and 2F). However, as the concentrations of H₂O₂ or MV increased, the growth inhibition and degree of leaf yellowing gradually became more severe in both the control and transgenic plants, but the growth inhibition was much more severe in WT and hsfa3 than in the HsfA3 overexpressors (Figs. 2B and 2G). For example, the FW of the overexpressors was reduced by approximately 45% in response to 4 mM H₂O₂ treatment, 75% in response to 5 mM H₂O₂ treatment, and 90% in response to 6 mM H₂O₂ treatment. In WT, the FW was reduced by 50%, 80%, and 98% in response to 4, 5, and 6 mM H₂O₂ treatment, respectively. The reduction in FW was much more severe in hsfa3 than in WT, with a reduction of 90% in response to 5 mM H₂O₂ treatment (Fig. 2C). The chlorophyll contents were altered in a similar manner (Fig. 2D). Ion leakage measurements revealed that although 5 and 6 mM H₂O₂ treatment caused considerable ion leakage in all plants examined, ion leakage rates were approximately 10% lower in H₂O₂-treated HsfA3 overexpressors compared to WT or hsfa3 plants (Fig. 2E). When WT and HsfA3 transgenic plants were grown on MSO agar medium supplemented with various concentrations of MV for 10 days, the FW of the HsfA3 overexpressors was reduced by approximately 70% on 0.3 μM MV, 80% in 0.5 μM MV, and 90% on 0.7 μM MV. By contrast, the FW of WT plants was reduced by 80%, 90%, and 95% on 0.3 μM, 0.5 μM, and 0.7 μM MV, respectively (Fig. 2H). The response of hsfa3 was more severe than that of WT. The chlorophyll contents in the HsfA3 overexpressors after 10 d of MV treatment were higher than those of WT and hsfa3 plants: the chlorophyll contents were almost twice as high as both WT and hsfa3 levels after 0.7 μM MV treatment (Fig. 2I), suggesting that HsfA3 prevents chlorophyll degradation to maintain plant photosynthesis under oxidative stress conditions. While in MV-treated WT and hsfa3 plants, ion leakage increased by 50?60% after 0.7 μM MV treatment, this value increased by only 30?40% in the overexpressors (Fig. 2J), indicating that there was a much lower degree of ROS-induced damage in the HsfA3 overexpressors than in WT and knockout plants. These results indicate that HsfA3 increases oxidative stress tolerance during seed germination and seedling growth.

Expression of GolS genes in HsfA3 overexpressors

Oxidative stress is neutralized by ROS detoxifying enzymes and by the biosynthesis of compatible solutes, such as amino acids, quaternary ammonium compounds, amines, and several sugars, such as RFOs (Jung et al., 2013). HsfA3 can directly bind to the HSEs of genes encoding ROS detoxifying enzymes, especially the ascorbate peroxidase 2 (APX2) gene (Hwang et al., 2012; Jung et al., 2013). By contrast, little is known about how HsfA3 regulates the biosynthetic genes of compatible solutes. Therefore, we performed quantitative RT-PCR to compare the expression levels of galactinol synthase (GolS; EC 2.4.1.123) genes in WT and transgenic plants, as GolS catalyzes the first committed step in the biosynthesis of RFOs (Lahuta et al., 2014). Arabidopsis contains seven GolS genes (GolS1 to GolS7) (Taji et al., 2002), including GolS1 to GolS4, which were shown to be induced by several abiotic stress treatments; experimental proof of the protective role of RFOs came from the analysis of transgenic plants (Nishizawa et al., 2008). Therefore, we examined the expression of GolS1 to GolS4 in WT and HsfA3 transgenic seedlings under normal growth conditions. As shown in Fig. 3A, GolS1, 2, and 4, but not GolS3, were induced in HsfA3 transgenic seedlings. The levels of GolS1, 2, and 4 transcripts were approximately 2.5?9.5-fold higher in the HsfA3 overexpressors than in WT and hsfa3 seedlings (Fig. 3B).

Since these GolS transcripts were upregulated in the HsfA3 overexpressors, we investigated whether the upregulated GolS transcripts were associated with galactinol biosynthesis. Specifically, we examined galactinol levels in the leaves of WT, overexpressors, and hsfa3 plants under normal growth conditions (Table 1). In the WT, the level of galactinol was 22.9 ± 2.8 nmol/g FW. Galactinol clearly accumulated in the HsfA3 overexpressors (36.6 ± 0.8 and 102.1 ± 11.3 nmol/g FW), whereas hsfa3 contained only 13.8 ± 1.8 nmol/g FW galactinol. These results suggest that HsfA3 activates the transcription of GolS genes and that increased GolS transcript levels lead to increased galactinol biosynthesis during seed germination and seedling growth.

HsfA3 trans-activates GolS gene expression

Based on the oxidative stress tolerance phenotype of the HsfA3 overexpressors (Fig. 2) and the observation that GolS expression was induced in these plants (Fig. 3), we reasoned that oxidative stress triggers the expression of HsfA3 and that HsfA3 in turn activates the expression of GolSs. To test this hypothesis, we carried out transient promoter activation assays. An approximately 1 kb region upstream of the putative transcriptional start site of GolS1 contains a tandem inverted repeats of the HSE (5′-TTCCAGAACTTTC-3′; core HSE is underlined), and both GolS2 (5′-GAAGCTTC-3′) and GolS4 (5′-TTCATGAA-3′) contain a symmetrical HSE, but the GolS3 promoter sequence does not (Fig. 4A). As shown in Fig. 4B, the GUS/LUS activity of the GolS1 construct (Prom_GolS1:GUS) increased 26-fold in transformation reactions including the 35SΩ:F-HsfA3 effector compared to reactions lacking the effector construct, while that of the GolS2 construct (Prom_GolS2:GUS) increased 23.7-fold and that of the GolS4 construct (Prom_GolS4:GUS) increased 8.5-fold. These results indicate that HsfA3 trans-activates gene expression from the GolS1, 2, and 4 promoters. The GolS3 construct (Prom_GolS3:GUS) did not exhibit GUS/LUS activity, as it did not contain a HSE in the promoter sequence.

To directly assess the ability of HsfA3 to physically bind to the promoter of GolS1, GolS2, or GolS4, we purified bacterially expressed GST-HsfA3 fusion protein and characterized the DNA-binding ability of the recombinant protein to the HSE motifs in GolS promoters using EMSAs. As shown in Fig. 4C, radiolabeled bands with retarded mobility were observed when ³²P-GolS1-wHSE and GolS2-wHSE were used as probes with GST-HsfA3 but not with GST, indicating that HsfA3 specifically binds to the probe. The signal intensity of the retarded bands was reduced in a concentration-dependent manner by the addition of unlabeled wild-type HSEs (wHSEs), but not unlabeled mutated HSEs (mHSEs), to the binding reaction, indicating that the mutated bases are required for interaction with HsfA3. No retarded band was observed in reactions using ³²P-GolS4-wHSE as a probe, indicating that other factors are involved in the regulation of GolS promoter activity in vivo.

Oxidative stress arising from an imbalance in the generation and removal of ROS is a challenge faced by all aerobic organisms (Scarpeci et al., 2008). Hsfs play a central role in ROS sensing in plants (Davletova et al., 2005). However, little is known about the target genes of Arabidopsis Hsfs and their contribution to plant oxidative stress responses.

In the present study, we functionally characterized an Arabidopsis class A Hsf gene, HsfA3, under oxidative stress conditions. HsfA3 expression was induced in response to exogenous H₂O₂ or DBMIB application in Arabidopsis seedlings (Fig. 1), and ectopic overexpression of HsfA3 improved oxidative stress tolerance in these plants, as revealed by analysis of morphological and biochemical traits (Fig. 2). These findings suggest that HsfA3 is involved in the plant response to oxidative stress and plays a substantial role in oxidative stress signaling. We previously reported that the antioxidant gene APX2 is upregulated by HsfA3 and that increased APX2 activity reduces the accumulation of H₂O₂, thereby increasing plant tolerance to oxidative damage (Hwang et al., 2012). Additionally, GolS and raffinose synthase 2 (RS2) genes, encoding committed enzymes in the RFO biosynthetic pathway (Peterbauer and Richter, 2001), are upregulated by HsfA1a, HsfA1b, and HsfA2 in Hsf class A-overexpressing transgenic Arabidopsis plants, thereby increasing plant tolerance to oxidative damage due to increased galactinol and raffinose levels (Busch et al., 2005; Nishizawa et al., 2006; 2008; Panikulangara et al., 2004). In this study, we also found that the expression of GolS1, GolS2, and GolS4 was markedly induced in HsfA3-overexpressing transgenic Arabidopsis plants (Fig. 3), and intracellular galactinol levels were higher (Table 1) in these plants than in WT plants under control growth conditions. Furthermore, GolS1 and GolS2 transcript levels increased in response to salinity and drought stress, and GolS2-overexpressing Arabidopsis plants exhibited increased tolerance to drought stress (Taji et al., 2002). These findings indicate that in HsfA1b-, HsfA2-, and HsfA3-overexpressing transgenic Arabidopsis plants, GolS genes are upregulated, resulting in increased levels of galactinol and RFOs. Although it is unclear which products are formed by the reaction between ROS and RFOs, Nishizawa et al. (2008) reported that the enhanced levels of galactinol and raffinose in plants under stressful conditions may be closely related to the maintenance of ascorbate and glutathione levels in these plants. Hence, all of these findings suggest that several HsfAs regulate the expression of oxidative stress-responsive genes and that the target genes of HsfAs may contribute to increased biosynthesis of galactinol and RFOs in plants under stressful conditions. It is possible that galactinol and RFOs act as osmoprotectants and antioxidants, and that they play a role in protecting plants from oxidative damage caused by several abiotic stresses, especially HS.

In the current study, promoter transactivation assays and competitive EMSAs demonstrated that recombinant HsfA3 activates GolS1, GolS2, and GolS4 transcription and directly binds to GolS1 and GolS2 via the HSE motifs in their promoters (Fig. 4). GolS3 was not induced by HsfA3 effector and failed to physically bind to HsfA3, as GolS3 lacks a HSE in its promoter sequence. Similarly, Nishizawa et al. (2006) reported that HsfA2 can only activate HSE-dependent transcription of GolS1 and GolS2. Thus, both HsfA2 and HsfA3 appear to activate GolS1 and GolS2 expression through interaction with the HSEs in their promoters. Interestingly, the GolS1 promoter contains a tandem inverted repeat of the short consensus sequence nGAAn (GolS1-wHSE; 5′-nTTCnnGAAnnTTCn-3′, referred to as the “perfect” HSE) (Guo et al., 2008), whereas the GolS2 promoter contains an adjacent motif (GolS2-wHSE; 5′-GAAnnTCC-3). The signal generated by binding of HsfA3 to GolS1-wHSE was stronger than that with GolS2-wHSE, providing additional evidence that HsfA3 binds more efficiently to tandem invert repeats of the HSE motif than to single adjacent HSE motifs. Interestingly, co-transformation with the 35SΩ:FHsfA3 construct induced the expression of Prom_GolS4:GUS in promoter transactivation assays, with an approximately 8.5-fold increase in GUS activity (Fig. 4B), but HsfA3 did not physically bind to the GolS4 promoter in EMSAs (Fig. 4C). This result indicates that GolS4-wHSE HSE sequences do not directly bind to HsfA3. This finding supports the notion that other transcription factors downstream of HsfA3 are involved in the regulation of GolS4 promoter activity.

In light of these results, we propose a model for the role of HsfA3 in the oxidative stress response in Arabidopsis. HsfA3 binds to the promoters of GolS genes and induces their expression. This, in turn, results in an increase in galactinol and RFO biosynthesis, which leads to a decrease in ROS levels. This regulatory cascade enhances the ability of the plant to adapt to oxidative stress.