The CsHSP17.2 molecular chaperone is essential for thermotolerance in Camellia sinensis

Small heat shock proteins (sHSPs) play important roles in responses to heat stress. However, the functions of sHSPs in tea plants (Camellia sinensis) remain uncharacterized. A novel sHSP gene, designated CsHSP17.2, was isolated from tea plants. Subcellular localization analyses indicated that the CsHSP17.2 protein was present in the cytosol and the nucleus. CsHSP17.2 expression was significantly up-regulated by heat stress but was unaffected by low temperature. The CsHSP17.2 transcript levels increased following salt and polyethylene glycol 6000 treatments but decreased in the presence of abscisic acid. The molecular chaperone activity of CsHSP17.2 was demonstrated in vitro. Transgenic Escherichia coli and Pichia pastoris expressing CsHSP17.2 exhibited enhanced thermotolerance. The transgenic Arabidopsis thaliana exhibited higher maximum photochemical efficiencies, greater soluble protein proline contents, higher germination rates and higher hypocotyl elongation length than the wild-type controls. The expression levels of several HS-responsive genes increased in transgenic A. thaliana plants. Additionally, the CsHSP17.2 promoter is highly responsive to high-temperature stress in A. thaliana. Our results suggest that CsHSP17.2 may act as a molecular chaperone to mediate heat tolerance by maintaining maximum photochemical efficiency and protein synthesis, enhancing the scavenging of reactive oxygen species and inducing the expression of HS-responsive genes.

Heat shock (HS) is induced by a higher temperature, which is approximately 10-15 °C above the optimal growth temperature 1, 2 . After exposure to HS, prokaryotic and eukaryotic cells produce a group of heat shock proteins (HSPs) 2,3 . The HSPs can be divided into the following five categories based on their approximate molecular weights: HSP100, HSP90, HSP70, HSP60, and small HSPs (sHSPs). The sHSPs are the most prevalent in plants, and their importance is implied by their unusual abundance and diversity 4 . Six classes of sHSPs have been identified based on their amino acid sequence similarities, immunological cross-reactivities, and intracellular localizations 4,5 . Members of Classes I and II are localized in the cytosol and the nucleus. Classes III, IV, V, and VI consist of sHSPs present in the chloroplast, endoplasmic reticulum, mitochondrion, and membrane, respectively 6,7 .
Molecular chaperones are responsible for protein folding, assembly, translocation and degradation in many normal cellular processes. They also stabilize proteins and membranes, and assist in protein refolding under stress conditions 4 . Previous studies revealed that a wide range of sHSPs have molecular chaperone activities in vitro and in vivo 8,9 . Analyses of sHSP transcript levels indicated that some sHSPs were undetectable in the absence of stress, but rapidly accumulated following exposure to HS 10,11 . In addition to HS, the expression of sHSP genes can be induced by other abiotic stress, including salinity 12 , low temperature 13 and drought 14 . Furthermore, some plant sHSPs are only produced in specific tissues or during particular developmental stages 10,15 . Previous studies have demonstrated that heterologous expression of sHSP genes confers abiotic stress tolerance in transgenic A. thaliana 16 , rice 17 , tomato 18 , potato 19 , and tobacco 20 , suggesting that sHSPs are involved in stress tolerance. Ruibal et al. 21 found that a functional sHSP PpHsp16.4 was essential for recovery from heat, salt and osmotic stress in Physcomitrella patens. However, overexpression of AsHSP17, a new sHSP gene from creeping bentgrass (Agrostis stolonifera), negatively regulate plant responses to adverse environmental stresses in transgenic A. thaliana 22 .

Subcellular localization of CsHSP17.2 protein.
To clarify the CsHSP17.2 biological functions, subcellular localization investigations were completed using onion epidermal cells and tobacco leaves producing the CsHSP17.2:GFP fusion protein. The 35S:GFP signals were distributed throughout the cytosol and the nucleus in onion (Fig. 1A) and tobacco epidermal cells (Fig. 1B). As predicted by the protein subcellular localization prediction software WoLF PSORT 28 , diffuse 35S:CsHSP17.2:GFP signals were detected in the cytosol and nucleus of onion (Fig. 1A) and tobacco epidermal cells (Fig. 1B).

CsHSP17.2 transcription profiles under heat shock (HS) and recovery conditions.
To determine whether CsHSP17.2 expression under HS conditions is time-dependent, tea plants were incubated at high temperatures (40 °C) for different times ( Fig. 2A). The abundance of CsHSP17.2 transcripts in plants exposed to HS initially increased very quickly, with the highest levels (approximately 1,450-fold higher than pre-treatment levels) occurring within 1 h. Transcript levels significantly decreased after 6 h (P < 0.01). Following the HS treatment for 1 h, the tea plants were returned to normal conditions to recover from the heat stress. The CsHSP17.2 transcript levels decreased significantly within 1 h of recovery and were almost undetectable after 3 h (P < 0.01) (Fig. 2B).
Analysis of CsHSP17.2 expression following exposure to cold, salinity and exogenous abscisic acid (ABA). To investigate whether the CsHSP17.2 transcript levels were regulated by other abiotic stress, tea plants were exposed to various treatments. Under cold stress conditions, there were no significant changes in CsHSP17.2 expression levels within the first 24 h (Fig. 2C). The abundance of CsHSP17.2 transcripts in salt-stressed plants increased within 1 h of treatment, decreased significantly after 6 and 12 h (P < 0.05), and then peaked at 24 h (Fig. 2D). When exposed to drought stress, CsHSP17.2 transcript levels were high at 1 h and then decreased significantly after 6 h (P < 0.05) (Fig. 2E). Finally, exogenous ABA treatment significantly inhibited the CsHSP17.2 transcription levels in tea plants (P < 0.05) (Fig. 2F). These results indicated that CsHSP17.2 influenced drought and salinity stress responses and ABA signal transduction pathways.

CsHSP17.2 overexpression in E. coli cells.
We completed reverse transcription polymerase chain reaction (RT-PCR) experiments to ensure that appropriate transgenic E. coli cells had been generated. The resulting amplicons were 202 and 655 bp long, which corresponded to the control and CsHSP17.2-overexpressing lines, respectively (Fig. S2a). The bands corresponding to the expected 22-kDa thioredoxin/histidine tag were larger in the control E. coli lines (i.e., those harboring the empty vector) treated with 0.2 mM isopropyl-β-D-thiogalac-topyranoside (IPTG) than in the non-induced control E. coli lines. Additionally, an approximately 36-kDa band was detected for CsHSP17.2-overexpressing lines induced with IPTG, which matched the expected size of the CsHSP17.2 protein fused to the thioredoxin/histidine tag (Fig. 3A).
Under the premise that the pET E. coli overexpression system works normally, we performed cell viability assays to investigate the possible functions of CsHSP17.2. There was no obvious difference in the growth rates of the pET and pET-HSP17.2 strains, indicating that CsHSP17.2 overexpression did not affect E. coli growth under normal conditions (Fig. 3B). Under heat stresses, the E. coli survival decreased rapidly, but the CsHSP17.2-overexpressing cells were more viable than the control cells (P < 0.01). After a 90-min heat treatment, more than 95% of the pET cells died, whereas approximately 51% of pET-HSP17.2 cells survived (Fig. 3C). This observation suggested that CsHSP17.2 increased the thermotolerance of the transgenic E. coli cells.
Constitutive expression of CsHSP17.2 confers thermotolerance in transgenic P. pastoris. To confirm that CsHSP17.2 influenced thermotolerance, we examined the expression of CsHSP17.2 in a eukaryotic organism. First, RT-PCR was used to verify that the recombinant plasmids were present in the transformed P. pastoris cells. The amplified fragments were 221 and 548 bp long, which were the expected sizes for the control and CsHSP17.2-overexpressing lines, respectively (Fig. S2b). The positive strains were used for subsequent thermotolerance assays. Similar growth rates were observed between the control and the CsHSP17.2-overexpressing lines at 30 °C. However, the growth rates were higher for the CsHSP17.2-overexpressing strains than the control after HS treatment for 30 or 60 min (Fig. 4). These results indicated that the constitutive expression of CsHSP17.2 confers thermotolerance in P. pastoris.

CsHSP17.2 exhibits molecular chaperone activity in vitro.
After the thioredoxin/hi-stidine-tagged CsHSP17.2 protein was bound to a nickel-charged affinity resin column, recombinant CsHSP17.2 (approximately 17 kDa) was separated from the fusion tag using the tobacco etch virus protease and then eluted from the column and R3 indicate recovery from heat shock (HS) for 1, 2, and 3 h, respectively. Tea plants exposed to low temperature (4 °C) (C), 300 mM NaCl (D), 20% (w/v) PEG6000 (E) and 50 µM ABA (F). Different letters indicate significant differences (P < 0.05) between treatments and the control (0 h).
( Fig. 5A). To investigate whether CsHSP17.2 has molecular chaperone activity in vitro, citrate synthase (CS) were chemically denatured and renatured. When 300 nM CsHSP17.2 was added to the denatured CS under renaturing conditions, approximately 78% of the CS activity (threefold relative to the control) was recovered. In contrast, in the presence of H 2 O, only 26% of the CS activity was recovered after 75 min (Fig. 5B). These results suggested that CsHSP17.2 functioned as a molecular chaperone in vitro.
Variations in thermotolerance, maximum photochemical efficiency (F v /F m ), soluble protein and free proline content. An RT-PCR analysis indicated that CsHSP17.2 was substantially expressed in six independent transgenic A. thaliana lines (i.e., OE-8, OE-18, OE-21, OE-28, OE-29, and OE-30) but not in the wild-type (WT) plants (Fig. 6A). The three transgenic lines with the highest transgene expression levels (i.e., OE-8, OE-21, and OE-30) were used for subsequent experiments. The survival rates were higher in transgenic lines than in the WT plants after HS for 45 min, suggesting that CsHSP17.2 overexpression enhanced thermotolerance in A. thaliana plants (Fig. 6B). To estimate the level of HS-induced damage to A. thaliana plants, related physiological indices were recorded after a 4-h HS treatment. Exposure to HS resulted in a significant decrease in the F v /F m value in the CsHSP17.2-overexpressing and WT plants (P < 0.05) (Fig. 6C), but photosystem II was more active in the transgenic lines than in the WT plants, implying that CsHSP17.2 overexpression stabilized F v /F m under HS conditions. Similarly, the soluble protein content decreased more in WT plants than in the CsHSP17.2-overexpressing lines following HS treatment (Fig. 6D). This change indicated that CsHSP17.2 overexpression resulted in increased amounts of soluble protein in transgenic plants. In addition, after exposure to HS, more free proline accumulated in the transgenic lines than in WT plants ( Fig. 6E), indicating that CsHSP17.2 overexpression promoted proline synthesis.

CsHSP17.2 overexpression in A. thaliana promotes seed germination and hypocotyl development under HS conditions.
To determine whether CsHSP17.2 affected seed vigor, we examined the germination rates of WT and transgenic A. thaliana seeds under HS conditions. Without the HS treatment, nearly 100% of WT and transgenic seeds incubated at 22 °C germinated within 3 days (Figs 7A and S3a), indicating the high quality of the seeds used for the germination assays. However, the germination rates of transgenic and WT seeds gradually decreased with increasing duration of HS treatment. Additionally, the transgenic seeds exhibited enhanced vigor compared with the WT seeds after HS treatment (Figs 7B-D and S3b-d). Only 95.9% and 91.6% of WT seeds germinated after 1 and 2 h of heat treatment, respectively, whereas 98.7% of the transgenic seeds germinated even after 2 h. The germination rates of transgenic and WT seeds were 95.9-100.0% and 72.5% after a 3-h heat treatment, respectively. These results implied that CsHSP17.2 overexpression in plants enhanced the basal thermotolerance of seeds.
Hypocotyls were much longer in the CsHSP17.2-overexpressing lines than in the WT plants after exposure to high temperatures for 1 or 2 h (Fig. 7E), indicating that CsHSP17.2 plays an important role in hypocotyl development under HS conditions.
Cloning of the putative promoter of CsHSP17.2. The 863-bp upstream sequence of the CsHSP17.2 translation initiation site was obtained via thermal asymmetric interlaced PCR (TAIL-PCR) (Fig. S4), and the presence of core promoter elements was assessed using the PlantCARE database 29 . The presence of hormone-responsive motifs together with abiotic stress-related motifs and developmental stage specific elements was observed in the putative promoter sequence ( Fig. 8 and Table S1). The presence of these regulatory elements suggested that the promoter region of CsHSP17.2 might be responsive to a wide variety of abiotic stress and hormones and induced in some developmental stages.  Responsiveness of the CsHSP17.2 promoter to abiotic stress and hormones. The induction of the GUS under abiotic stress and hormones was monitored in 3-week-old transgenic A. thaliana plants via quantitative real-time PCR (qRT-PCR) and GUS histochemical staining. The GUS transcript gradually increased and peaked at approximately 12 h of heat stress treatment; thereafter, the transcription levels decreased significantly (Fig. 9A). The GUS histochemical assay revealed protein accumulation until 24 h, suggesting that although the transcript might not be required in high amounts, the corresponding protein accumulated (Fig. S5).
In addition to heat stress, the CsHSP17.2 promoter activity was also induced by simulated drought (10% PEG6000) and plant hormones (40 μM MeJA or 20 μM GA3) (Figs 9B and S5). These findings were consistent with the presence of multiple abiotic stress-and hormone-responsive elements in the CsHSP17.2 promoter sequence.

Discussion
Since the first sHSP was discovered in Drosophila melanogaster, numerous sHSPs have been identified in various plant species 5,30 . Plant sHSPs are classified into different subfamilies based on amino acid sequence similarities and localization to distinct subcellular compartments. In this study, the green fluorescence of the CsHSP17.2:GFP fusion protein was detected in the cytosol and nucleus, in agreement with our results from analyses of homology and phylogenetic relationships. These findings suggested that CsHSP17.2 is a Class I sHSP.
Numerous studies have previously demonstrated that sHSPs are regulated by various abiotic stress. A Primula sHSP gene, PfHSP17.1, is induced by heat, cold, salt, PEG-induced drought stress, and oxidative stress 14 . Similarly, a Tamarix hispida sHSP gene, ThHSP18.3, is induced by cold, heat, salt, and drought stress 12 . In addition, a David Lily sHSP gene, LimHSP16.45 is triggered by cold and heat stress 16 . Our observations revealed that the expression of CsHSP17.2 was differentially regulated by heat, cold, drought, salinity, and ABA treatments, suggesting that it has different roles in various abiotic stress responses. However, CsHSP17.2 expression was not obviously affected by cold stress, which suggested that CsHSP17.2 might not participate in the cold stress response. Interestingly, CsHSP17.2 transcription was down-regulated by ABA treatment, which was consistent with the results of Sun et al. 31 , suggesting that CsHSP17.2 may function in an ABA-dependent manner.
Several recent studies have used the expression of sHSP genes in E. coli cells to investigate their possible functions in vivo. The heterologous expression of a Medicago sativa mitochondrial HSP23 protected E. coli from salinity and arsenic stresses 32 , and the overexpression of a Rosa chinensis RcHSP17.8 enhanced E. coli viability during exposure to heat and cold stresses 33 . Additionally, as a major eukaryotic model organism, P. pastoris produces sHSPs and is often used to investigate sHSP functions 33,34 . Similarly, the E. coli and P. pastoris cells Scientific RepoRts | 7: 1237 | DOI:10.1038/s41598-017-01407-x constitutively expressing CsHSP17.2 were more viable when exposed to heat stress than the control cells, implying that CsHSP17.2 overexpression results in greater thermotolerance.
The F v /F m value is a good indicator of the photosynthetic functions of plants under adverse environmental conditions. A previous study concluded that overexpression of a tomato (Lycopersicon esculentum) HSP21 protected photosystem (PS) II from the effects of temperature-dependent oxidative stress 35 . In our study, the F v /F m values of CsHSP17.2-overexpressing plants were significantly higher than those of WT plants under HS conditions, suggesting that CsHSP17.2 has a role in protecting of PSII against oxidative stress created by high temperature. Previous studies have demonstrated that protein synthesis positively correlates with stress tolerance, and heat-tolerant plants maintain a higher protein synthesis rate and a lower protein degradation rate than heat-intolerant plants 36   resulted in increased scavenging of H 2 O 2 under HS conditions (Fig. S6), suggesting that CsHSP17.2 overexpression confers thermo-tolerance in A. thaliana. Ascorbate peroxidases (APXs) and peroxidases (PODs) are considered to have essential roles in scavenging reactive oxygen species (ROS) and protecting cells from oxidative damage in higher plants 40,41 . During exposure to HS, the transcription levels of AtAPX1 and AtPOD were higher in the CsHSP17.2-overexpressing lines than in the WT plants (P < 0.05) (Fig. S7). Consequently, we hypothesize that CsHSP17.2 increases the capacity for ROS-scavenging by elevating AtAPX1 and AtPOD transcript levels. Proline confers osmotic tolerance during exposure to stress conditions, and AtP5CS2 and AtProT1 are two key genes involved in proline biosynthesis and transport 42,43 . The transcription levels of these two genes were higher in the transgenic lines than in WT plants under normal and HS conditions (Fig. S7), which was consistent with the content of proline, implying that CsHSP17.2 overexpression may enhance proline biosynthesis and transport and ultimately increase the thermotolerance in A. thaliana.
Heat shock transcription factors (HSFs) are transcription factors that activate the expression of genes in response to stress, thereby playing a central role in cellular homeostatic control mechanisms 44 . An HSF binds to a DNA sequence motif, the heat shock element (HSE), which is characterized by an array of inverted repeats of the motif nGAAn. Besides an HSE motif, AGGGG motif (designated as STRE) and CCAAT box were also found in the promoter region of CsHSP17. 2, which have been demonstrated to act cooperatively with the HSEs to enhance HSP promoter activity under high temperature 45,46 . Therefore, these combined results strongly suggest that CsHSP17.2 is regulated by HSFs and induced by heat stress conditions. The transcript levels of AtHSF genes (especially AtHSFA4a and AtHSFC1) were higher in CsHSP17.2 transgenic lines than in WT plants. Thus, we speculated that CsHSP17.2 overexpression might enhance the thermotolerance of transgenic plants by co-regulating HSF gene expression.
Molecular chaperones, including sHSPs, are important for maintaining cellular homeostasis under optimal and adverse growth conditions. They can bind to partially denatured proteins to prevent further denaturation or aggregation and promote the correct refolding of proteins 4, 47 . Our results indicated that CsHSP17.2 functions as a molecular chaperone in vitro by inducing the renaturation and reactivation of chemically denatured CS, suggesting that CsHSP17.2 may help to maintain proteins in their functional conformations and to prevent the aggregation of non-native proteins in host cells. AtHSP17. 4 and AtHSP70, which are associated with molecular chaperone activities 11,48,49 , were more actively transcribed in transgenic A. thaliana plants under HS conditions (Fig. S7). These findings suggested that CsHSP17.2 overexpression induces HSP synthesis and confers thermotolerance in plants.
The sHSPs have protective roles in seed germination. Transgenic A. thaliana seeds expressing NnHSP17.5 displayed enhanced seed germination vigor to heat stress 50 . Similarly, ectopic expression of LimHSP16.45 enhanced the seed viability of A. thaliana exposed to high temperature, salinity, and oxidative stress 51 . In this study, seeds from CsHSP17.2-overexpressing lines exhibited enhanced germination rates under HS conditions, suggesting that CsHSP17.2 may contribute to basal thermotolerance during seed germination. Induced thermotolerance is defined as the capacity of an organism to survive a normally lethal temperature if it is first conditioned by pretreatment at a milder temperature and can be measured in hypocotyl elongation assays 52 . CsHSP17.2-overexpressing lines exhibited higher hypocotyl elongation lengths, suggesting that CsHSP17.2 overexpression conferred induced thermotolerance in transgenic plants. To the best of our knowledge, this report is the first demonstrating that a Class I sHSP is correlated with induced thermotolerance. Thus, we speculate that CsHSP17.2 is essential for basal and induced thermotolerance in transgenic A. thaliana.
Previous studies have demonstrated that overexpression of sHSPs have different roles in plant growth and development. Overexpression of TaHSP26 in A. thaliana produced higher biomass and seed yield than WT plants 3 . Besides, transgenic tomato (Lycopersicon esculentum) plants constitutively expressing LeHSP21 promoted fruit ripening under normal growth temperature 35 . However, overexpression of NnHSP17.5 had no positive or detrimental effects on plant development in A. thaliana. Similarly, overexpressing CsHSP17.2 did not cause any obvious phenotypic changes. The germination times and rates, growth rates, time to flowering, and seed yields of the transgenic plants were similar to those of WT plants (Fig. S8). For example, under normal conditions, there were no significant differences between WT and CsHSP17.2-overexpressing plants in terms of the fresh weights of rosette leaves from 4-or 5-week-old seedlings (Fig. S9). These results suggested that heterologous expression of CsHSP17.2 may be used for enhancing thermotolerance in economically important crops.
In summary, our results demonstrate that CsHSP17.2 is a heat-inducible gene, and CsHSP17.2 has molecular chaperone activities in vitro. CsHSP17.2 has the capacity to confer thermotolerance not only on E. coli and P. pastoris, but also on A. thaliana under heat stress. We propose that CsHSP17.2 increases plant thermotolerance through several pathways, including the maintenance of photosynthetic rates and protein synthesis, enhancement of ROS-scavenging, and expression of HS-responsive genes. These findings should help to clarify the complex mechanisms and roles of HSPs in regulating plant responses to environmental stresses.

Plant materials and stress treatments.
Two-year-old C. sinensis cv. 'Yingshuang' plants were grown in a light incubator with a 12-h light (200 μmol·m −2 ·s −1 ; 24 °C)/12-h dark (20 °C) photoperiod for 30 days before treatments. For HS treatments, plants were placed in a light incubator (40 °C) for 24 h. For recovery treatments, the plants (HS for 1 h) were allowed to recover at 24 °C for 3 h, and samples were collected every hour. For the low-temperature treatment, tea plants were transferred to another chamber maintained at 4 °C. To simulate high salinity and drought stresses, tea plants were treated with 300 mM NaCl and 20% (w/v) PEG6000, respectively. For the exogenous ABA treatment, tea leaves were sprayed with 50 µM ABA solution. All treatments were completed under normal conditions (as described above), unless otherwise indicated. Additionally, the third fully expanded leaves from the top buds were harvested at 0, 1, 6, 12, and 24 h after various treatments, immediately frozen with liquid nitrogen, and stored at −70 °C for subsequent RNA extraction.
Scientific RepoRts | 7: 1237 | DOI:10.1038/s41598-017-01407-x Molecular cloning of CsHSP17.2 gene and its promoter. Total RNA was isolated from tea leaves using a Quick RNA Isolation Kit (Huayueyang, Beijing, China), and 1 μg of total RNA was reverse transcribed to generate first-strand cDNA using Reverse Transcriptase M-MLV (RNase H−) (TaKaRa, Dalian, China) according to the manufacturer's instructions. To obtain the ORF of CsHSP17.2, a primer pair (ORF-F/-R, Table S2) was designed and employed for PCR amplification. The resulting amplicons were purified and cloned into the pEASY-T1 Simple Cloning Vector (Transgen, Beijing, China) for sequencing (Genscript, Nanjing, China).
Genomic DNA was extracted from tea buds using Plant Genomic DNA Kit (TIANGEN, Beijing, China) according to the manufacturer's instructions. The promoter of CsHSP17.2 gene was amplified using Genome Walking Kit (TaKaRa, Dalian, China) with three gene-specific primers (SP1, SP2 and SP3, Table S2). The target products were cloned into the pEASY-T1 simple vector and sequenced. A search for regulatory elements in the promoter was performed using the PlantCARE (Plant Cis-Acting Regulatory Element) database 29 . Analyses of sequences and phylogenetic relationships. The ClustalX program was used for multiple sequence alignments 53 , and a phylogenetic tree was constructed using MEGA5 according to the neighbor-joining method 54 .

Subcellular localization of the CsHSP17.2 protein in onion (Allium cepa) epidermal cells and tobacco (Nicotiana benthamiana) leaves. To construct a transient expression vector, the ORF of the
CsHSP17.2 gene (without the stop codon) was amplified using the sub-F/-R primer pair (Table S2). The amplified CsHSP17.2 coding region was inserted into the pCAMBIA2300-GFP vector at the Bam HI and Xba І sites to generate the CsHSP17.2::GFP construct. Onion epidermal cells were transformed with the recombinant plasmid or the empty vector using a PDS-1000/He particle delivery system (Bio-Rad, Hercules, CA, USA). The transformed onion epidermal cells were then incubated on Murashige and Skoog (MS) agar medium for 16 h at 25 °C in darkness. Finally, the GFP signal was detected using a Zeiss LSM700 confocal laser-scanning microscope (Carl Zeiss Inc., USA).
For transient expression in tobacco cells, Agrobacterium tumefaciens strain GV3101 cells were independently transformed with the recombinant plasmid or empty vector. The transformed A. tumefaciens cells were infiltrated into tobacco leaves 55 , and then plants were incubated for 3 days in darkness. The GFP signal was detected as described above.
Gene expression analysis by qRT-PCR. Total RNAs were extracted from tea leaves exposed to different treatments using the Quick RNA Isolation Kit (Huayueyang, Beijing, China). After assessing the quality and determining the concentration of total RNA using the ONE Drop OD-1000+ spectrophotometer (ONE Drop, USA), 1 μg of total RNA was reverse transcribed to single-stranded cDNA using the PrimeScript RT Reagent Kit with gDNA Eraser (TaKaRa, Dalian, China) following the manufacturer's instructions. The CsHSP17.2-qF/-qR primer pair (Table S2) was used to analyze CsHSP17.2 expression levels. The C. sinensis β-actin gene (GenBank accession number: HQ420251, Table S2) was used as the reference gene.
A. thaliana transgenic plants (3-week-old) were exposed to various abiotic stress and hormone treatments. For heat stress treatment, plants were cultured in a growth cabinet at 40 °C (dark) for different time periods. Drought stress was provided by treating with 10% PEG6000 for 4 h, and anaerobic treatment by immersing seedlings in distilled water for 3 h. For hormone treatments, leaves of plants were sprayed with 40 μM methyl jasmonate (MeJA) or 20 μM gibberellin A3 (GA3) for 24 h. After these treatments, the A. thaliana rosette leaves were sampled and flash-frozen in liquid nitrogen and stored at −80 °C for RNA extraction. All of these treatments were performed under a growth regime of 16/8 h light/darkness at 22 °C unless otherwise mentioned. qRT-PCR was performed with GUS-specific primers (GUS-qF/-qR, Table S2), and AtACTIN2 (AT3G18780) was selected as a housekeeping gene.
To analyze the expression profiles of HS-responsive genes, total RNA was extracted from the leaves of 4-week-old A. thaliana plants that had been incubated at 45 °C for 0, 0.5, 1, 2, and 4 h. The expression profiles of twelve HS-responsive genes were determined by qRT-PCR, with the A. thaliana ACTIN2 (AT3G18780) gene used as the housekeeping gene. All of the relevant primers are listed in Table S3.
The qRT-PCR was completed using SYBR Premix Ex Taq II (Tli RnaseH Plus) (TaKaRa, Dalian, China) and an iQ5 Multicolor Real-Time PCR Detection System (Bio-Rad). Each 20 μl qRT-PCR sample consisted of 10 μl of SYBR Premix Ex Taq II (2x), 0.2 μM of each primer, and 10 ng of cDNA template. The PCR program was as follows: 95 °C for 30 s, 40 cycles at 95 °C for 5 s, and 58 °C for 30 s. All experiments were repeated three times with independent RNA samples, and the relative expression levels were calculated using the 2 −ΔΔCT method 56 .

CsHSP17.2 overexpression in E. coli cells.
The full-length CsHSP17.2 coding region was amplified using the pro-F/-R primer pair (Table S2)  Thermotolerance of transgenic E. coli exposed to HS. Thermotolerance assays were conducted according to the method described by Soto et al. 58 with some modifications. Briefly, when the cultures (grown at 37 °C) of E. coli harboring the pET-CsHSP17.2 plasmid or the empty vector reached an optical density (at 600 nm, OD 600 ) of 1.0, they were diluted 100-fold with fresh Luria-Bertani liquid medium supplemented with ampicillin (100 μg·ml −1 final concentration). After the E. coli cells were treated with 1 mM IPTG for 4 h, 1-ml samples were transferred to a temperature-controlled water bath (50 °C), and 30-μl aliquots were added to Luria-Bertani agar plates after 0, 30, 60, and 90 min. Cell viability was estimated by counting the number of colony-forming units after an overnight incubation at 37 °C.
CsHSP17.2 purification and chemical denaturation and renaturation experiments. Recombinant proteins were purified using a nickel-charged affinity resin column (Qiagen GmbH, Germany) according to the method described by Liu et al. 8 . The fusion proteins were cleaved with tobacco etch virus protease for 4 h at 30 °C and then analyzed by SDS-PAGE. Gels were stained with Coomassie brilliant blue R-250 and photographed. The concentrations of the purified proteins were determined using the Bradford method 59 with bovine serum albumin (BSA) as the standard.
Refolding assays involving chemically denatured citrate synthase (CS) were conducted as described by Collada et al. 60 to determine whether CsHSP17.2 exhibited molecular chaperone activities. First, 15 μM CS (Sigma, St. Louis, MO, USA) was denatured with 6 M guanidine hydrochloride for 2 h, and then diluted 100-fold in refolding buffer (100 mM Tris-HCl, pH 8.0) supplemented with 300 nM CsHSP17.2 protein or distilled water. We then analyzed 20-μl aliquots for CS activity by monitoring the breakage of the thioester bond of acetyl CoA, which absorbs at 233 nm.
For thermotolerance assays, cultures of P. pastoris carrying pPIC3.5K-HSP17.2 (OD 600 = 1.5) were incubated at 50 °C for 0, 30, and 60 min, and 10-μl aliquots of 10-fold serial dilutions were spotted onto yeast extract/peptone/dextrose (YEPD) agar medium. After incubation at 30 °C for 3 days, samples were observed and photographed using a digital camera (Canon, Japan). The P. pastoris strains harboring the pPIC3.5 K were used as controls. For characterization, the putative CsHSP17.2 promoter (PHSP17.2) was cloned into the pEASY-T1 Simple Cloning Vector with the primer pair pHSP17.2-F/-R (Table S2). Then, the amplified product was digested with Hind III and Bam HI and inserted into the pBI121 vector (Invitrogen) to generate the PHSP17.2:GUS fusion vector (Fig. S10b). Agrobacterium strain EHA105 harboring the binary vector PHSP17.2:GUS was used for A. thaliana transformation. The independent transformants were selected on 1/2 MS agar medium containing 50 μg·ml −1 kanamycin and the T 3 homozygous plants were raised for subsequent histochemical staining and GUS transcription assays. The PCR primers used to confirm the transgenic A. thaliana were T-pHSP17.2-F and T-pHSP17.2-R (Table S2). Thermotolerance test. Plates containing 1-week-old A. thaliana plants were submerged in a water bath at 45 °C for 45 min according to the method described by Charng et al. 62 .

Plasmid construction and transformation of
Chlorophyll fluorescence, soluble protein and proline content measurements and H 2 O 2 detection. The chlorophyll fluorescence of rosette leaves was measured using an Imaging-PAM Chlorophyll Fluorometer (M-series; Heinz Walz GmbH, Germany). All A. thaliana plants were incubated in darkness for 10 min immediately before measurements, and the F v /F m was calculated automatically. The soluble protein contents were analyzed using Coomassie brilliant blue G250 according to a published procedure 37 . The free proline content was determined using the acid-ninhydrin method 63 . We detected H 2 O 2 in A. thaliana rosette leaves according to the method described by Orozco-Cardenas and Ryan 64 . For these experiments, 4-week-old seedlings were incubated at 45 °C for 4 h, and each experiment was performed three times.
Germination assays. Seeds of WT and CsHSP17.2-overexpressing plants (i.e., OE-8, OE-21, and OE-30) were plated on 1/2 MS medium. After a 3-day cold treatment (4 °C) in darkness, the plates were transferred to a water bath at 45 °C and incubated for 0, 1, 2, and 3 h. The plates were then placed in a light incubator under normal conditions for 7 days, and the germination rates were calculated every day. After the 7-day incubation, representative plates were photographed.
Hypocotyl elongation assays. Hypocotyl elongation assays were conducted according to a published method 52 with minor modifications. For all assays, seeds were plated on 1/2 MS medium, and the plates were covered with foil. After a 2.5-day cold treatment, the plates were incubated at 22 °C for an additional 2.5 days. The seeds were then incubated at 38 °C for 90 min, followed by a 2-h recovery period at 22 °C and then a 1-or 2-h treatment at 45 °C (i.e., HS treatment). After another 2.5 days, the hypocotyl lengths of all seedlings were measured. The relative hypocotyl length was calculated using the following formula: (hypocotyl lengths of 5-day-old seedlings under HS conditions-hypocotyl lengths of 2.5-day-old seedlings)/hypocotyl lengths of 2.5-day-old seedlings. Experiments included at least 10 seedlings from each line and were repeated at least three times. Phenotypic analysis. The phenotypes of each A. thaliana line were observed and photographed after 7 days (grown in 1/2 MS), 14 days (grown in 1/2 MS), 5 weeks (grown in soil), and 7 weeks (grown in soil), respectively. Additionally, the fresh weights of rosette leaves were measured after 4 and 5 weeks of cultivation under normal conditions. Histochemical GUS staining. The GUS histochemical staining of transgenic A. thaliana plants containing the PHSP17.2:GUS fusion construct followed the method described previously 65 . The explants were then observed with a bright field microscope and photographed (Leica Q500MC, Cambridge, England). Statistical analysis. All data were statistically analyzed with SPSS 17.0 software (SPSS Inc., Chicago, IL, USA) using Duncan's multiple range test at a 0.05 level of significance.