SIZ1 small ubiquitin-like modifier E3 ligase facilitates basal thermotolerance in Arabidopsis independent of salicylic acid.

Small ubiquitin-like modifier (SUMO) conjugation/deconjugation to heat shock transcription factors regulates DNA binding of the peptides and activation of heat shock protein gene expression that modulates thermal adaptation in metazoans. SIZ1 is a SUMO E3 ligase that facilitates SUMO conjugation to substrate target proteins (sumoylation) in Arabidopsis (Arabidopsis thaliana). siz1 T-DNA insertional mutations (siz1-2 and siz1-3; Miura et al., 2005) cause basal, but not acquired, thermosensitivity that occurs in conjunction with hyperaccumulation of salicylic acid (SA). NahG encodes a salicylate hydroxylase, and expression in siz1-2 seedlings reduces endogenous SA accumulation to that of wild-type levels and further increases thermosensitivity. High temperature induces SUMO1/2 conjugation to peptides in wild type but to a substantially lesser degree in siz1 mutants. However, heat shock-induced expression of genes, including heat shock proteins, ascorbate peroxidase 1 and 2, is similar in siz1 and wild-type seedlings. Together, these results indicate that SIZ1 and, by inference, sumoylation facilitate basal thermotolerance through processes that are SA independent.

High temperature stress adversely affects organisms by causing membrane integrity loss, reactive oxygen species (ROS) production, protein inactivation and denaturation, and metabolic and cellular disequilibria, which ultimately lead to cell death (Berry and Björkman, 1980;Quinn, 1988;Lindquist, 1992;Dat et al., 1998b;Los and Murata, 2000;Iba, 2002). Plants have an innate capacity to survive high temperature stress (basal thermotolerance) and can sense and acclimate to high temperatures with metabolic and cellular adjustments that impart a capacity to tolerate heat extremes that were previously lethal (acquired thermotolerance; Vierling, 1991;Alfonso et al., 2001;Clarke et al., 2004;Larkindale et al., 2005). Acquired thermal tolerance responses are coordinated by signaling pathways that regulate heat tolerance determinants to limit stress damage and facilitate reestablishment of cellular homeostasis for survival and growth (Clarke et al., 2004;Larkindale et al., 2005). Thermal adaptation responses include membrane compositional changes necessary for maintenance of functional integrity, activation of oxidative defensive systems through ethylene and salicylic acid (SA), and production of heat shock proteins (HSPs) necessary for cellular protection (Quinn, 1988;Boston et al., 1996;Schöffl et al., 1998;Larkindale and Knight, 2002;Baniwal et al., 2004;Clarke et al., 2004;Larkindale et al., 2005).
HSPs are molecular chaperones that reduce protein denaturation, target denatured proteins for proteasome degradation, facilitate protein folding necessary for proper maturation or renaturation, and regulate the activity of HSFs to control HSP gene expression during thermotolerance acquisition (Johnson and Craig, 1997;Lee and Goldberg, 1998;Lee and Vierling, 2000;Frydman, 2001;Kim et al., 2002). Plant HSPs presumably mediate high temperature stress tolerance, but this is inferred largely because orthologs in other organisms have a thermal adaptive function (Vierling, 1991;Ellis, 2000;Hartl and Hayer-Hartl, 2002). Only the HSP100 family of plant HSPs, which are members of the ClpB chaperone family of ATPases that facilitate disaggregation of denatured proteins, are established functional determinants of acquired thermotolerance Queitsch et al., 2000;Lee et al., 2005). Hsa32 (heat shockassociated) protein is necessary for maintenance of acquired thermotolerance in Arabidopsis (Charng et al., 2006). HOT2, HOT3, and HOT4 are genetic loci that facilitate acquired thermotolerance in Arabidopsis but map to positions in the genome that do not encode HSPs .
Basal thermotolerance is comparatively less understood than acquired thermotolerance Larkindale et al., 2005). HSP101 is an essential determinant for basal thermotolerance of seed germination , and ethylene, SA, and ROS signaling functions in basal thermotolerance at different plant developmental stages (Clarke et al., 2004;Larkindale et al., 2005). The numerous cellular and metabolic processes involved in basal thermotolerance implicate that a signaling network composed of numerous regulators is necessary to exercise concerted control over effector determinants in a developmental context (Larkindale et al., 2005).
Arabidopsis SIZ1 is an ortholog of mammalian PIAS (protein inhibitor of activated signal transducer and activator of transcription) and yeast Siz family SUMO E3 ligases that facilitate sumoylation of transcription factors (Gill, 2005;Hay, 2005;Miura et al., 2005). Lossof-function analyses described herein establish that the independent dysfunctional T-DNA insertion alleles siz1-2 or siz1-3 (Miura et al., 2005) cause thermosensitivity. Experimental results indicate that SIZ1 is a positive regulator of processes that are necessary for basal thermotolerance through functions that are independent of SA. However, SIZ1, dependent or independent of sumoylation function, does not regulate acquired thermotolerance as it does in fungi and metazoans Hietakangas et al., 2003;Hilgarth et al., 2003Hilgarth et al., , 2004. Apparently, SUMO conjugation/deconjugation facilitates high temperature tolerance in plants through processes that have yet to be identified in other organisms.

RESULTS
The SUMO E3 Ligase SIZ1 Facilitates High Temperature Tolerance Ten-day-old siz1-2 and siz1-3 seedlings exhibited substantial thermosensitivity, as determined by using a heat shock survival assay (Fig. 1, A and B). siz1 seedlings rapidly developed severe heat shock symptoms in response to high temperature that were not evident in wild type. Substantial reduction in leaf surface area (shrinkage) was visible immediately after treatment, followed by severe chlorosis 24 h later. siz1 seedlings that developed these extreme symptoms did not survive (Fig. 1, A and B). The maximal heat shock temperature (4-h treatment) that wild-type and siz1 seedlings could survive was 43°C and 40°C, respectively (data not shown).
Thermosensitivity of siz1-2 and siz1-3 seedlings was evident also in a hypocotyl elongation assay (Fig. 1C). High soil temperatures early in seedling development can restrict hypocotyl elongation, which may prevent or delay shoot emergence from the soil (Lin et al., 1984;Hong and Vierling, 2000). Hypocotyl elongation of siz1 seedlings was sensitive to even a brief exposure to 39°C (Fig. 1C). siz1 seedlings also exhibited a reduction in hypocotyl elongation at 37°C and 38°C, but sensitivity relative to wild type was less than at 39°C. Similar results in heat shock-sensitive phenotypes caused by independent siz1 alleles ( Fig. 1) are indicative that SIZ1 functions in thermotolerance.

SIZ1 Mediates Basal But Not Acquired Thermotolerance
Plants, like most other organisms, exhibit both basal (innate) and acquired thermotolerance (Larkindale et al., 2005). The latter phenomenon is an acclimation that occurs in response to brief exposure to high temperature or longer-term exposure to increasing temperature and facilitates survival of thermal extremes that previously were lethal Hong et al., 2003). siz1 mutations cause heat sensitivity in assays that assess basal thermotolerance ( Fig. 3B; Col-0 and siz1-2, P , 0.01; Col-0 and siz1-3, P , 0.05); consequently, their effects on acquired thermotolerance were examined. siz1-2 and siz1-3 seedlings exhibited a similar capacity for acquired thermotolerance as wild type, as indicated from viability and hypocotyl elongation assays (Fig. 3, A and B). An exposure to 39°C for 90 min is a sublethal heat shock treatment for all the genotypes compared in this experiment. hot1-3 seedlings did not acclimate in response to the sublethal temperature pretreatment stress, because the mutation abrogates capacity for acquired thermotolerance (Hong and Vierling, 2001). By inference, SIZ1 function in high temperature adaptation seems restricted to basal thermotolerance.

SIZ1 Controls Heat Shock-Induced SUMO Conjugation
Heat shock induced an increase in conjugation of SUMO1/SUMO2 to substrate proteins at 34°C to 43°C (Figs. 5 and 6; Kurepa et al., 2003;Miura et al., 2005). A 39°C heat shock treatment induced sumoylation in wild type but to a lesser extent in siz1 seedlings (Figs. 5 and 6). Immunoblots were probed with anti-SUMO1 that detects both SUMO1 and SUMO2, and heat shock does not induce SUMO3 conjugation in Arabidopsis (Kurepa et al., 2003). Increased SUMO conjugation corresponded with a decrease in free SUMO1/2, indicating that accumulation is facilitated by increased sumoylation and not by reduced desumoylation. Together, these results indicate that SIZ1 mediates high temperature-induced accumulation of SUMO conjugation products. SUMO1/2 conjugation was substantially greater in NahG and hot1-3 seedlings compared to wild type after 30 min of heat shock exposure (Figs. 5 and 6). Interestingly, SUMO conjugation and deconjugation rates were different in NahG and hot1-3 seedlings (Fig. 7). NahG seedlings exhibited more rapid SUMO conjugation in response to heat shock and delayed SUMO deconjugation after return to ambient temperature (Fig. 7, center). SUMO deconjugation of Figure 2. siz1 increases thermosensitivity during seed germination. Stratified wild-type, siz1-2, siz1-3, NahG siz1-2, NahG, and hot1-3 seeds (3 d in the dark at 4°C) were immediately subjected to heat shock treatment of 45°C for 4 h or incubated at 24°C (normal), sown onto plates, and then maintained under a 16-h daily photoperiod at 24°C. Germination was assessed at the indicated intervals. A, Illustration of representative seeds/seedlings 6 d after heat treatment and B, seed germination data from three independent experiments, mean 6 SE, n 5 21.
hot1-3 seedlings was impaired during the recovery period. siz1-2 suppressed induction of sumoylation that is associated with NahG expression (Fig. 6). This is indicative that SIZ1-mediated SUMO conjugation may be a biochemical process by which the E3 ligase regulates thermotolerance responses independently of SA.

SIZ1-Mediated Thermotolerance Is Independent of HSF Regulon Expression
HSF complexity in plants is predicted to be substantially greater than in other organisms. Yeast, Drosophila, and C. elegans have one HSF and vertebrates have four (Morimoto, 1998;Nakai, 1999;Nover et al., 2001;Baniwal et al., 2004), while bioinformatic analyses predict 21, 18, and 23 HSFs in Arabidopsis, tomato, and rice (Oryza sativa), respectively, based on sequence similarity (Baniwal et al., 2004). Transcript abundance of HSF1, 3, 4, and 7 was similar in wild-type and siz1-2 seedlings, indicating that SIZ1-dependent sumoylation does not regulate expression of these genes (Fig. 8A). HSF1 and HSF3 are major regulators of high temperatureinduced HSP expression (Lohmann et al., 2004). Expression of heat shock-induced genes was also similar in wild-type and siz1-2 seedlings (Fig. 8B). Included in the survey were genes that encode ascorbate peroxidase (APX)1 and APX2, and HSPs, all of which are regulated by HSF1 and HSF3 and implicated to facilitate thermotolerance (Storozhenko et al., 1998;Hong and Vierling, 2000;Panchuk et al., 2002;Lohmann et al., 2004). APX1 and APX2 function as antioxidant enzymes that detoxify hydrogen peroxide, which is produced in response to heat stress and is presumed to be a major effector of cellular dysfunction (Panchuk et al., 2002). HSFs, APXs, and HSPs expression patterns were similar in siz1-2 and wild-type seedlings after heat shock treatment at 39°C (data not shown). Furthermore, heat shock-induced expression of these genes in NahG siz1-2 was similar to siz1-2 seedlings (data not shown), indicating that SIZ1 Figure 3. siz1-2 and siz1-3 seedlings exhibit reduced basal thermotolerance but not acquired thermotolerance. A, Ten-day-old wild-type, siz1 (siz1-2 and siz1-3), NahG siz1-2, NahG, and hot1-3 seedlings were not acclimated (normal, 22°C) or high-temperature acclimated by exposure to 39°C for 90 min. After a recovery period of 2 h at 22°C, seedlings were exposed to a heat shock of 45°C for 2 h under the conditions described in the Figure 1 legend. Illustrated are representative seedlings 4 d after heat shock treatment. B, Stratified seeds of genotypes described in A were incubated in the dark at 22°C/ 18°C (16 h/8 h). Three days thereafter, seedlings were high-temperature acclimated by exposure to 39°C for 90 min. After a recovery period of 2 h at 22°C, seedlings were exposed to a heat shock of 45°C for 2 h under conditions described in the Figure 1 legend and then grown for 2.5 d at 22°C/18°C (16 h/8 h). Illustrated are relative growth determinations (100% indicates the growth of genotypes at 22°C) from three independent experiments, mean 6 SE, n $ 18 seedlings/experiment. Figure 4. siz1 hyperaccumulates high levels of SA. Ten-day-old wildtype, siz1 (siz1-2 and siz1-3), NahG siz1-2, NahG, and snc1 seedlings grown at 24°C on medium were harvested. SA content was determined by HPLC analysis. Illustrated are data from three independent experiments, mean 6 SE (micrograms of SA per gram fresh weight). Experiments were repeated four times in the same condition. snc1 seedlings result hyperaccumulation of SA (Zhang et al., 2003). functions in thermal adaptation independent of the HSF-controlled regulon.

DISCUSSION
The results of this study implicate a function for SIZ1 SUMO E3 ligase in basal thermotolerance of Arabidopsis. siz1-2 and siz1-3 cause thermal hypersensitivity (Figs. 1-3) that is manifested during seed germination, hypocotyl elongation, and seedling survival, inferring that SIZ1-mediated sumoylation is necessary for heat shock tolerance at numerous developmental stages. SUMO conjugation/deconjugation, in other organisms, is necessary for both ambient and high temperature-dependent HSF interactions with heat shock elements that regulate HSP expression Hilgarth et al., 2003Hilgarth et al., , 2004Hietakangas et al., 2006). However, the biological role of sumoylation in heat stress responses is unresolved, because it is not evident if SUMO conjugation to HSFs positively or negatively regulates thermotolerance Hietakangas et al., 2006). SUMO1/2 conjugates accumulation is an early plant response (within minutes) to high temperature stress, implicating a function for sumoylation in thermotolerance (Kurepa et al., 2003;Miura et al., 2005). The evidence herein confirms that high temperature-induced SUMO1/2 conjugation is attributable to SIZ1 SUMO E3 ligase function (Fig. 5;Miura et al., 2005). However, these data do not resolve if overall increased SUMO1/2 conjugation is a phenomenon that is linked causally to thermotolerance. Together, these results indicate that SIZ1 controls basal, but not acquired, thermotolerance, and infer that sumoylation/desumoylation of specific substrates is necessary.
AtSIZ1 is, by domain composition and functional data, a member of the PIAS family of SUMO E3 ligases (Kurepa et al., 2003;Miura et al., 2005). AtSIZ1 regulates gene expression and root architecture responses that are caused by phosphate deprivation (Miura et al., 2005). The MYB transcription factor PHR1 that is a controller of phosphate starvation-induced gene expression is a sumoylation substrate of SIZ1 (Miura et al., 2005). SIZ1 is implicated also in both thermal adaptation (herein) and pathogen defense (Lee et al., 2006b). Together, these results indicate that sumoylation/desumoylation is a key control process in the regulation of signal networks in plants. It is still to be resolved in any organism how sumoylation/desumoylation regulates diverse biological processes. However, emerging evidence implicates a function in gene expression regulator that involves chromatin remodeling (Gill, 2005;Hay, 2005).
PIAS family members are implicated to function as transcriptional regulators independent of the SP-RING domain that facilitates E3 ligase activity (Lee et al., 2006a;Sharrocks, 2006). The SAP domain of PIAS proteins is associated with transcriptional regulation resulting from chromatin remodeling (Shuai and Liu, 2005;Sharrocks, 2006). Consequently, there is a foundation to support the notion that SIZ1 may regulate basal thermotolerance through a sumoylation-independent process. Alternatively, it is also possible that SIZ1 is a determinant of thermotolerance through sumoylationdependent and -independent processes. Figure 5. Heat shock-induced SUMO1/2 conjugation is suppressed by siz1-2 and siz1-3. Ten-day-old wild-type, siz1-2, siz1-3, and hot1-3 seedlings were exposed to a 30-min heat shock treatment (39°C, dark, 60% relative humidity). Total protein was extracted from untreated or heat shock-treated seedlings. Twenty micrograms of protein were loaded onto an SDS-PAGE, and the immunoblot was probed with anti-SUMO1, which detects both SUMO1 and SUMO2 (Kurepa et al., 2003). Figure 6. Heat shock-induced SUMO1/2 conjugation in NahG seedlings requires SIZ1. Ten-day-old wild-type, siz1-2, NahG siz1-2, and NahG seedlings were exposed to a 30-min heat shock at 39°C in the dark. Total protein was extracted from seedlings as described in Figure  5. Ten micrograms of protein were separated by SDS-PAGE and the immunoblot was probed with anti-SUMO1.

SIZ1 Controls Basal Thermotolerance through a SA-Independent Process(es)
A genetic analysis of basal and acquired thermotolerance in Arabidopsis implicated the involvement of numerous heat shock-response pathways that do not involve a HSF regulon (Larkindale et al., 2005). Evidence implicates SA, ethylene, and ROS as possible intermediary signal molecules (Dat et al., 1998a;Cronjé and Bornman, 1999;Larkindale and Knight, 2002;Clarke et al., 2004;Larkindale et al., 2005). Understanding of the biological role of SA in thermal adaptation is rudimentary, but SA regulates HSP17.6 expression in Arabidopsis and HSP70 in tomato (Cronjé and Bornman, 1999;Clarke et al., 2004). However, HSP expression (particularly that of HSP101 and Hsa32) is presumed to be associated with acquired thermotolerance, and SA is not considered to be a principal regulator of this process (Hong and Vierling,Figure 7. Heat stress-induced SUMO1/2 conjugation/deconjugation is impaired in NahG and hot1-3 seedlings. Ten-day-old wild-type, NahG, and hot1-3 seedlings were exposed to a 39°C heat stress for 15 or 30 min, returned to 24°C, and collected at indicated time points. Black arrow indicates the serial of heat shock. Total protein was extracted as described in Figure 5. Ten micrograms of protein were separated by SDS-PAGE, and the immunoblot was probed with anti-SUMO1. Figure 8. HSF and heat shock-induced gene expression is similar in wild-type and siz1 seedlings. Ten-day-old wild-type and siz1-2 seedlings were exposed to 37°C for the times indicated. Transcript abundance was determined by semiquantitative reverse transcription-PCR for four major HSFs (HSF1, 3, 4, and 7), and five different HSPs (HSP17.6A [class II sHSP], HSP18.2 [class I sHSP], HSP70, HSP83.1, HSP101) and APX1 and APX2. Black arrows indicate the period of heat shock. 2000, 2001Clarke et al., 2004;Larkindale et al., 2005;Charng et al., 2006). Our data further illustrate that NahG expression affects basal but not acquired thermotolerance and that SA does not regulate expression of numerous heat shock-regulated genes ( Fig. 3; data not shown). siz1 mutations cause SA accumulation in seedlings and plants, yet seed germination and seedling growth and survival are adversely affected by high temperature. NahG expression causes thermal sensitivity of Arabidopsis (Clarke et al., 2004;Larkindale et al., 2005;Figs. 2 and 3) and increases hypersensitivity resulting from siz1 mutations (i.e. NahG siz1-2), indicating that SA has some thermal protective function that is precluded by processes controlled by SIZ1. That is, SIZ1 negatively regulates SA biosynthesis and/or catabolism but positively regulates basal thermotolerance through a SA-independent process(es).
Heat shock-induced SUMO conjugation/deconjugation is impaired in NahG and hot1-3 seedlings. Thiobarbituric acid reactive substances are products of lipid oxidation and thiobarbituric acid reactive substance levels are induced by heat stress in NahG and hot1-3 seedlings (Larkindale et al., 2005). ROS are reported to increase SUMO conjugation (Saitoh and Hinchey, 2000;Kurepa et al., 2003), even though low concentrations of ROS cause desumoylation of most substrates (Bossis and Melchior, 2006). High concentrations of ROS cause inactivation of the SUMO isopeptidase SENP-1 that leads to accumulation of SUMO conjugation products (Bossis and Melchior, 2006). A paradigm is emerging that ROS species and/or levels constitute a regulatory system that controls sumoylation/desumoylation of protein substrates (Kurepa et al., 2003;Bossis and Melchior, 2006). These results support the notion that ROS may regulate differences in heat shock-induced sumoylation and desumoylation evident in NahG and hot1-3 seedlings that correlate with increased thermal sensitivity, although it is not possible to link SUMO conjugation/deconjugation of specific substrates based on the results presented herein.

SIZ1-Mediated Thermotolerance Is Not Due to HSF Regulon Expression
Although SIZ1 is apparently not essential (Miura et al., 2005), it is evident that the SUMO E3 ligase is necessary for sumoylation that is associated with plant stress responses, as are PIAS and Siz orthologs in yeast, Drosophila, C. elegans, and vertebrates (Schmidt and Mü ller, 2002;García-Estrada et al., 2003;Takahashi and Kikuchi, 2005). HSF-controlled gene expression is critical for high temperature tolerance in these organisms, and sumoylation of Xenopus and human HSFs regulates transactivation of HSP expression Hilgarth et al., 2004). Although sumoylation of hHSF1, hHSF2, and hHSF4b likely result in transcriptional repression, it is postulated that SUMO conjugation and deconjugation are dynamic regulatory processes that are necessary for fine tuning reg-ulation of basal and acquired thermotolerance Hietakangas et al., 2006).
At present, there is no evidence that Arabidopsis HSF family members are substrates for SUMO conjugation or that sumoylation/desumoylation of HSF is necessary for regulation of high temperature-induced HSP expression or thermotolerance in plants. It is important to note that redundant regulatory effect of AtHSF1 and AtHSF3 on HSP expression does not markedly influence thermotolerance (Lohmann et al., 2004) as does LeHSF1 in tomato (Mishra et al., 2002). Also, we detected no difference in high temperatureinduced mRNA expression patterns of HSPs in siz1 and wild-type seedlings (Fig. 8), indicating that SIZ1dependent sumoylation does not regulate acquired thermotolerance in plants (Fig. 3). Attempts at in vitro sumoylation of AtHSF1 or AtHSF3 were inconclusive, although the assay does mediate SUMO conjugation to the transcription factor PHR1 (Miura et al., 2005).
Together, our results establish that SIZ1 facilitates basal thermotolerance in Arabidopsis through a SA-independent process(es). SIZ1, independent or dependent of sumoylation function, does not regulate acquired thermal adaptation responses in plants, unlike in other organisms as diverse as human, yeast, and Xenopus. The protein targets of SIZ1-dependent sumoylation that mediate thermotolerance remain to be identified as do the processes that control basal thermotolerance. Determinants that control high temperature signaling that regulate sumoylation/desumoylation and control thermal adaptation await identification. However, the results herein indicate that SUMO conjugation is a necessary process for basal thermotolerance at different plant developmental stages.

Thermotolerance Assays
Seedlings were subjected to heat shock in the dark at 60% relative humidity in a plant growth chamber (E-30B, Percival Scientific) with the capacity to control temperature fluctuation control by 61°C. These conditions were used to minimize photooxidative and high humidity stresses (Larkindale and Knight, 2002;Zhou et al., 2004;Larkindale et al., 2005). Survival was monitored daily beginning 4 d after heat shock treatment. Stratified seeds were subjected to heat shock treatment in a temperature-controlled water bath (ISOTEMP 210, Fisher Scientific) and then sown onto medium. Seed germination was monitored every 24 h. The hypocotyl elongation assay was carried out as described by Hong and Vierling (2000). Briefly, 3-d-old etiolated seedlings were incubated at 22°C or heat shock-treated at 39°C or 45°C in dark for the indicated time. Plumule position of each seedling was recorded by marking the plate, and the plate was rewrapped and incubated at 22°C/18°C (16 h/8 h) in dark. Hypocotyl growth after heat shock treatment was measured after 2.5 d.

Total RNA Isolation and Semiquantitative Reverse Transcription-PCR Analysis
Total RNA from 10-d-old seedlings grown at 22°C or heat shock treated for the indicated time was isolated by using PureLink Micro-to-Midi Total RNA Purification system (no. 12183-018, Invitrogen). Two micrograms of total RNA were used as template for first-strand cDNA synthesis with ThermoScript reverse transcription-PCR system (no. 11146-016, Invitrogen) and an oligo(dT) 20 primer. Gene-specific primers were used to amplify PCR products of approximately 500 bp in length (Supplemental Table S1).

SA Quantification
Shoots of 10-d-old seedlings that were grown on medium under conditions described above were harvested and frozen in liquid nitrogen. Tissue (0.2 g fresh weight, without roots) was extracted in 4 mL of methanol for 24 h at 4°C and then in a solution of 2.4 mL of water plus 2 mL of chloroform with 40 mL of 5 mM 3,4,5-trimethoxy-trans-cinamic acid (internal standard) for 24 h at 4°C. Supernatants were dried by speed vacuum. The residue was resuspended in 0.4 mL of water:methanol (1:1, v/v), and SA was quantified by HPLC as described by Freeman et al. (2005).

Statistical Analysis
All data except germination rates were analyzed by Student's t test for pairwise comparison. Germination was compared at different time points using method of logistic regression with repeated measurements using SAS PHREG software and P values are indicated.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Table S1. Gene-specific primer sequences used to detect heat shock-related genes by RT-PCR.