l e t t e r s

Overexpression of receptor-like kinase ERECTA improves thermotolerance in rice and tomato

Hui Shen1,2,9, Xiangbin Zhong1,9, Fangfang Zhao1, Yanmei Wang1, Bingxiao Yan1, Qun Li1, Genyun Chen1, Bizeng Mao3, Jianjun Wang4, Yangsheng Li5, Guoying Xiao6, Yuke He1, Han Xiao1, Jianming Li7 & Zuhua He1,8

The detrimental effects of global warming on crop productivity Overexpression of HSP101 increased tolerance to heat shock or threaten to reduce the world’s food supply1–3. Although plant prolonged heat stress in transgenic Arabidopsis, tobacco and cot- responses to changes in temperature have been studied4, ton9,14, although transgenic cotton produced fewer bolls and seeds genetic modification of crops to improve thermotolerance than control cotton plants under normal growth conditions14. has had little success to date. Here we demonstrate that On the other hand, engineering to increase the tolerance of plants to overexpression of the Arabidopsis thaliana receptor-like high temperatures during multiple hot summer seasons at different kinase ERECTA (ER) in Arabidopsis, rice and tomato locations has not been reported. In particular, there are few confers thermotolerance independent of water loss and that reports of engineering or breeding of thermotolerant staple crop Arabidopsis er mutants are hypersensitive to heat. A loss- species, except for a recent study, which reported that natural alleles of of-function mutation of a rice ER homolog and reduced a gene encoding proteasome α2 subunit from African rice contribute expression of a tomato ER allele decreased thermotolerance of to thermotolerance15. Therefore, genes other than HSP101 are critical both species. Transgenic tomato and rice lines overexpressing to thermotolerance improvement in crops16. Arabidopsis ER showed improved heat tolerance in the The model plant Arabidopsis is sensitive to high temperatures, 17,18 greenhouse and in field tests at multiple locations in China probably owing to local adaptation , and cannot survive prolonged during several seasons. Moreover, ER-overexpressing transgenic heat stress19. Following our previous studies of Arabidopsis heat and Arabidopsis, tomato and rice plants had increased biomass. harmattan responses19–21, we report in this study that two ecotypes, Our findings could contribute to engineering or breeding Columbia-0 (Col-0) and Landsberg erecta (Ler), were substantially dif- thermotolerant crops with no growth penalty. ferent from each other in their tolerance to prolonged extreme heat in greenhouse conditions (40 °C; Fig. 1a). Ler plants began wilting 12 h Nature America, Inc. All rights reserved. America, Inc. © 201 5 Nature The temperature increases associated with global warming reduce after transfer from 22 °C to 40 °C and completely wilted or died 48 h plant growth and crop productivity3. In particular, a harmattan climate, after heat treatment. Col-0 plants, however, only began wilting 36 h which brings desert-like conditions, causes immediate death of plants after heat treatment, and a fraction (~50%) of their leaves retained npg or their tissues. It has been estimated that rice grain yield declines osmotic pressure 48 h after heat treatment, resulting in a significantly by 10% for each 1 °C increase in local night temperature in the dry higher survival rate (47.2%) compared with Ler (7.8%) 48 h after heat season1 and that global warming has a negative impact on the yield treatment (Supplementary Fig. 1a). This difference in heat tolerance of other major grain crops including wheat, maize and barley2. indicates that the Col-0 strain contains one or more gene(s) for heat The heat wave during the 2013 summer in eastern China had a dev- tolerance (Hat). To map Hat loci in Col-0, we first screened a set astating impact on crops. Plants respond to temperature changes by of recombinant inbred lines (RILs) derived from a cross of Col and reprogramming their growth and development4–8, and progress has Ler. These lines were classified into three groups based on survival been made towards genetic modification of plants to increase their rates: lines with high tolerance (similar to Col-0), lines with high tolerance to heat stress. However, these studies have mainly been sensitivity (similar to Ler) and lines with intermediate properties restricted to the model plant Arabidopsis9–14. Many studies have (Supplementary Fig. 1a). Based on the RIL phenotype, two major shown that expression of heat shock proteins (HSPs) can improve quantitative trait loci (QTL) were detected: qHat2-1, with a single peak the tolerance of transgenic plants to heat shock (short exposure to having an LOD score of 6.6 and 30% contribution on chromosome 2, high temperatures)15. In particular, HSP101 is required for thermo- and qHat2-2, with three small peaks with an LOD score of 5.0–6.9 tolerance in plants and has potential as a tool for crop improvements. and 23–28% contribution on chromosome 4 (Supplementary Fig. 1b

1National Key Laboratory of Plant Molecular Genetics and National Center of Plant Gene Research, Institute of Plant Physiology & Ecology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China. 2Shanghai Chenshan Plant Science Research Center, Chinese Academy of Sciences, Shanghai Chenshan Botanical Garden, Shanghai, China. 3College of Agriculture and Biotechnology, Zhejiang University, Hangzhou, China. 4Zhejiang Academy of Agricultural Sciences, Hangzhou, China. 5College of Life Science, Wuhan University, Hubei, China. 6Institute of Subtropical Agriculture, Chinese Academy of Sciences, Changsha, China. 7Shanghai Center for Plant Stress Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China. 8Collaborative Innovation Center of Genetics and Development, Institute of Plant Physiology & Ecology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China. 9These authors contributed equally to this work. Correspondence should be addressed to Z.H. ([email protected]).

Received 10 February; accepted 21 July; published online 17 August 2015; doi:10.1038/nbt.3321

996 VOLUME 33 NUMBER 9 SEPTEMBER 2015 nature biotechnology l e t t e r s

and Supplementary Table 1). We decided to fine-map the qHat2-1 whether ER is the qHat2-1 locus, we first measured the heat locus by constructing chromosome segment substitution lines sensitivity of er mutants in Col-0 and observed that a null er mutant, (CSSLs), with Ler as the recurrent parent, that cover chromosome 2 er-105 (ref. 22), was more sensitive to heat than the wild-type Col-0 with five CSSLs (Supplementary Fig. 1c). The line CSSL-2-3 was (Supplementary Fig. 2a,b). A similar difference in heat sensitivity associated with a considerably higher survival ratio compared was also observed between Ler and its wild-type Landsberg (Lan) with other CSSLs (Supplementary Fig. 1d), indicating that CSSL-2-3 (Supplementary Fig. 2c,d). We next transformed er-105 with the contains the qHat2-1 locus for heat tolerance. Col-0 wild-type ER gene driven by its native promoter (pER::ER) and The line CSSL-2-3 covers a 493-kb region of chromosome 2, and found that the pER::ER transgene completely restored heat tolerance the middle of this segment contains the ER gene (Supplementary of the null mutant (Supplementary Fig. 2a,b). Similar results were Fig. 1d), which encodes a receptor-like kinase that was mutated in obtained by complementation with the same pER::ER transgene in Ler and has been implicated in diverse growth and development Ler lines (Supplementary Fig. 2c,d). Thus, we conclude that ER is the processes22–28, including warmth-induced abnormal adaxial-abaxial functional gene for qHat2-1, and that er loss-of-function mutations polarity in leaf patterning29 and photosynthetic capacity30. To evaluate greatly reduce heat tolerance.

a Heat treatment (40 °C) d e f g 0 h 12 h 24 h 36 h 48 h Markerer-105Col-0LM13LM17LM10 25 22 °C 22 °C 22 °C 170 kDa 0.5 25 ** ER-MH(α-MYC) 20 ** ** ** 130 kDa 0.4 * 20 15 100 kDa ** 0.3 15 ** 10 Col-0 0.2 10 per plant

70 kDa plant (g) **

5 0.1 per plant (g) 5 55 kDa Seed weight per Actin 0 0 Shoot dry weight 0 Inflorescence number

Col-0 L7-1 Col-0 L2-3 L7-1 Col-0 L2-3 L7-1 er-105 er-105 er-105 L er Genotype Genotype Genotype 22 °C h i j 100 b 40 °C 40 °C Col-0 ** ** Col-0 er-105 L2-3 L7-1 Col-0 L7-1 80 * * 30 °C 60 ** 40 er-105 ** ** 20 Survival rate (%) 0 L2-3 40 °C 48 h 30 °C 30 d Col-0 L2-3L7-1 Col-0 L2-3L7-1 Col-0 Col-0L7-1 er-105 er-105 er-105LM13LM17LM10 Genotype L7-1 22 °C 22 °C 22 °C 35S::ER l ** k 5 m 3.5 ** 0.20 ** 4 3.0 Nature America, Inc. All rights reserved. America, Inc. © 201 5 Nature * 0.15 2.5

/s) 3 c 2 2.0 30 °C Col-0 er-105 L7-1 0.10 2 1.5 * (mol/m ( µ mol/mmol) WUE (mg/g) npg 1.0 * 0.05 1 Instantaneous WUE 0.5 Stomatal conductance 0 0 0

Col-0 L7-1 Col-0 L7-1 Col-0 L7-1 Col-0 L7-1 er-105 er-105 Genotype Genotype Genotype

Figure 1 ER plays an important role in heat tolerance in Arabidopsis. (a) Pictures of 2-week-old Arabidopsis plants grown in a 40 °C growth chamber for 0, 12, 24, 36 and 48 h. Scale bar, 1 cm. (b) Pictures of 4-week-old plants grown at 22 °C (left panels) of the wild-type Col-0, er-105 and two ER-OE lines (L2-3 and L7-1) and their corresponding individual leaves of different developmental stages (right panels). Scale bars, 1 cm. (c) Mature plants of Col-0, er-105 and L7-1 grown at 22 °C. Scale bar, 1 cm. (d) Immunoblot analysis of the ER-MH abundance in wild-type Col-0, er-105 and three selected p35S::ER-MH er1-05 transgenic lines, LM13 (wild-type-looking), and L10 and L17 (bigger than Col-0) (see Supplementary Fig. 4 for pictures of plants and their leaves). Total proteins extracted from 2-week-old plants grown at 22 °C were separated by SDS-PAGE and analyzed by immunoblotting using anti-myc and anti-actin (for loading control) antibodies. (e–g) Average numbers of inflorescent stems (e), average seed production (f) per plant after maturation and average shoot dry weight (g) of 4-week-old wild-type Col-0, er-105 and two ER-OE lines (L2-3 and L7-1) grown at 22 °C, as shown in box plots (n =30). *P < 0.05 or **P < 0.01, Student’s t-test and Bonferroni correction for multiple tests (two comparisons for e, three comparisons for f,g). (h) Pictures of 2-week-old plants of the wild-type Col-0, er-105 and the two ER-OE lines after 2 d treatment at 40 °C. Scale bar, 1 cm. (i) Pictures of mature plants of the wild-type Col-0 and L7-1 grown at 30 °C for 30 d. (j) Survival rates of the wild-type Col-0, er-105 and the ER-OE lines after exposure to a hot (40 °C, 2 d) (h) and a long warm temperature (30 °C, 30 d) (i), as shown with dots of three replicates (30 plants each). *P < 0.05 or **P < 0.01, by Student’s t-test Bonferroni correction for multiple tests in the independent temperature experiments, with three comparisons for Col-0, er-105, L2-3 and L2-7 at 40 °C, and four comparisons for Col-0, er-105, LM10, LM13 and LM7 at 40 °C. (k) Stomatal conductance of 4-week-old wild-type Col-0 and L7-1 grown at 22 °C, as shown with dot plots (n ≥ 8). **P < 0.01, Student’s t-test. (l) Quantitative analysis of instantaneous water use efficiency of 4-week-old plants of the wild-type Col-0 and L7-1 grown at 22 °C, as shown with dot plots (n ≥ 8). **P < 0.01, Student’s t-test. (m) Quantitative measurement of water use efficiency of 4-week-old plants of the wild-type Col-0, er-105 and L7-1 grown at 22 °C or 30 °C, shown as box plots (upper and lower quartiles and median; n = 25). *P < 0.05 or **P < 0.01, by Student’s t-test and Bonferroni correction for multiple (two comparisons) tests in the independent temperature experiments. (See source data.)

nature biotechnology VOLUME 33 NUMBER 9 SEPTEMBER 2015 997 l e t t e r s

We next investigated whether overexpressing the ER gene could eight of ten p35S::ER-MH er-105 transgenic lines were phenotypically improve heat tolerance of wild-type plants. We used the strong and rescued, two lines (LM10 and LM17) that accumulated higher levels constitutively active 35S promoter from cauliflower mosaic virus with of the transgene transcripts and ER-MH protein grew bigger and had two copies of an enhancer element to drive the expression of a promot- a higher thermotolerance than the wild-type control (Fig. 1d,j and erless ER transgene in wild-type Col-0 plants. Out of p35S::ER trans- Supplementary Fig. 4). This suggests that the observed growth and genic lines, we identified nine lines with increased biomass as measured physiological phenotypes of ER-OE plants were most likely caused by rosette leaf sizes, inflorescence numbers (Fig. 1e), shoot weight by increased abundance of the ER protein. Whereas er mutations and seed production (Fig. 1b,c,e–g and Supplementary Fig. 3c,d). decrease the size of leaf epidermal cells and increase the stomatal Quantitative real-time RT-PCR (qPCR) analysis showed that ER density without affecting the stomatal index (the ratio of number of transcript abundance was stably increased by ~20- to 90-fold in rosette stomata to total number of leaf epidermal cells)28,30, our ER-OE plants leaves of nine 2-week-old transgenic lines (referred to as ER-OE, had larger leaf epidermal cells and decreased stomatal density but hereafter) compared with wild-type Col-0 (Supplementary Fig. 3a). there was no change in their stomatal index (Supplementary Fig. 5), qPCR analysis using RNAs isolated from different tissues of the providing additional support for the hypothesis that ER abundance wild-type Col-0 and L7-1, one of the ER-OE lines, revealed that the was increased in the ER-OE lines and for a role for ER in regulating ER transcript accumulated to very high levels (Supplementary Fig. 3) cell growth and stomata development. in both rosette and cauline leaves (Supplementary Fig. 3b). Further, To directly test the effect of ER overexpression on heat tolerance, we elevated ER expression had no detectable effect on the transcript subjected 2-week-old Arabidopsis plants to a 2-d 40 °C treatment in a levels of two ER homologs, ERL1 and ERL2, at different developmen- growth chamber and observed (P < 0.05 or P < 0.01) that the ER-OE tal stages of 2-week-old and 6-week-old plants or in different tissues, plants grew much better (Fig. 1h), with significantly higher survival such as rosette and cauline leaves (Supplementary Fig. 3e). Despite rates (65–75%) compared with wild-type Col-0 plants (~48%)(Fig. 1h,j). numerous attempts, we failed to obtain an anti-ER antibody that We tested plant tolerance to prolonged exposure to a moderate would allow us to directly test if increased levels of the ER transcript increase in ambient temperature ( from 22 °C to 30 °C). After 30-d led to increased accumulation of the ER protein. Instead, we added growth at 30 °C the leaves of wild-type Col-0 became yellowed, and a myc-His (MH) tag to the p35S::ER transgene, transformed er-105 most plants (~85%) wilted and died. In contrast, most leaves on the with the resulting p35S::ER-MH transgene and found that although L7-1 line grown at 30 °C for 30 d remained green (Fig. 1i), and a

40 °C 40 °C Col-0 Col-0 a 30–40% RH 90–95% RH b c d e er-105 er-105 40 °C, 0 h 40 °C, 12 h 40 °C, 24 h 80 L2-3 * v v v * L7-1 100 100 pm pm pm 100 cw

60 ** * cw cw 80 80 Col-0 v v 90 v 40 * v 60 60 v v cw 80 pm pm 40 20 * 40 cw er-105

Survival rate (%) cw 20 v v disruption (%) 70 pm Plasma membrane lon leakage (%) 20 0 0 v v v pm 60 0 cw cw Relative water content (%) pm pm L7-1 L7-1 L7-1 v cw cw Col-0 Col-0 Col-0 0 12 24 0 12 24 36 48 L7-1 er-105 Nature America, Inc. All rights reserved. America, Inc. © 201 5 Nature er-105 er-105 v pm Genotype Time (h) Time (h) Genotype

f g 40 °C, 0 h 40 °C, 24 h 40 °C, 48 h h i 30 Col-0 er-105 L7-1 30 er-105 er-105

npg 25 25 Col-0 Col-0

Col-0 20 20 L7-1 L7-1 15 15 40 ° C, 0 h 10 10 levels of Bl1 er-105 5 5 levels of HSFA1a Relative expression Relative expression 0 0 40 ° C, 48 h L7-1 0 h 2 h 4 h 8 h 0 h 2 h 4 h 12 h 24 h 36 h 8 h 12 h 24 h 36 h 40 min 40 min

Figure 2 Analysis of leaf water content and heat-induced cell damages. (a) Quantification of survival rates of 2-week-old heat-treated (40 °C for 2 d) plants of the wild-type Col-0, er-105 and L7-1, grown under low (30–40%) and high (90–95%) RH conditions, shown in dots of three replicates (30 plants each). *P < 0.05, by Student’s t-test and Bonferroni correction for multiple tests with two comparisons in the independent RH experiments. (b) Relative water content of 2-week-old wild-type Col-0 and er-105 plants grown at 22 °C treated with a 40 °C growth condition for 0 h, 12 h and 24 h, shown as dots (n = 12). No difference in water content was detected between heat-treated er-105 and wild-type Col-0 but a statistically significant difference was detected between non-heat-treated plants of the wild-type Col-0 and er-105 by Student’s t-test (**P < 0.01). (c) Measurements of cellular ion leakage over a time course of 48 h under heat treatment (40 °C). Leaves of 2-week-old plants grown at 22 °C of the wild-type Col-0, er-105 and the ER-OE lines L2-3 and L7-1 treated with or without 48 h growth at 40 °C were measured. Values are means ± s.d. (n = 15). (d) TEM subcellular observation of leaf cell collapse during the 40 °C heat treatment for 0–24 h. The plasma membrane (pm) became blebbed and collapsed more quickly in the er-105 cells than in the Col-0 cells; blebbing and collapsing of the plasma membrane were observed less often in the ER overexpression L7-1 cells. Red arrowheads indicate pm stacking and twinning, and red arrows indicate PM blebbing in er-105. Scale bars, 500 nm. (e) Quantification of the occurrence of plasma membrane disruption (blebbing and collapsing) in leaf cells after 24 h growth at 40 °C, shown as dots of three replicates (≥50 cells observed each). *P < 0.05, by Student’s t-test and Bonferroni correction for two comparisons. (f) Visualization of cell death using trypan blue staining of rosette leaf cells of 2-week-old plants grown at 22 °C and treated with heat (40 °C) for 0 and 48 h. Scale bars, 2 mm. (g) H2O2 accumulation, detected by the DAB reaction, was more obvious in er-105 than in Col-0. Scale bars, 2 mm. (h,i) qPCR analysis of the heat shock factor gene HSFA1a (h) and the cell death regulator gene BI1 (i) with total RNAs isolated from leaves of 2-week-old plants grown at 22 °C over the course of 36 h heat (40 °C) treatment. ACTIN2 was used as a control to normalize expression levels. Values are means ± s.d. (n = 3). (a,b,e). cw, cell wall; pm, plasma membrane; v, vacuole. (See source data.)

998 VOLUME 33 NUMBER 9 SEPTEMBER 2015 nature biotechnology l e t t e r s

Figure 3 ER confers heat tolerance in tomato. a b e (a) qPCR analysis of the transgenic ER 10,000 35S::ER 2012 summer, Shangahi transcript in selected p35S::ER transgenic Vector L20 L22 L25 tomato lines (4-week-old, grown at 22 °C). 8,000 L20 L22 L25 6,000 Vector The tomato SleIF4a6 gene was used as a 4,000 control to normalize expression levels. Values 1 100 2 3 are means ± s.d. (n = 3). (b) Pictures of mature 0

branches from 6-week-old plants grown at Relative expression L20 L22 L25 22 °C of the vector control and p35S::ER Vector transgenic tomato. Scale bar, 2 cm. (c) Pictures c f of 6-week-old transgenic tomato plants grown Vector L20 L22 L25 at 22 °C (upper panels) and subsequently for 10 d under a 40 °C (38–42 °C recorded due to daily fluctuation)/28 °C (day/night) cycle (middle panels), followed by 2-d recovery

growth at 22 °C (lower panels). (d) Survival Pre-treatment rates of transgenic tomato lines after heat treatment in growth chamber (c), shown as g 2013 summer, dots of three replicates (~30 plants each). Shanghai 2014 summer, 2014 summer, **P < 0.01, by Student’s t-test and Bonferroni 100 ** Shanghai ** Wuhan correction for three comparisons. (e) Pictures of ** post-summer mature tomato plants (~10-week- 80 ** ** 40 ° C/28 C 10 days ** ** ** old after the summer) grown at the Shanghai 60 experimental station in 2012. (f) Field-grown ** transgenic tomato plants (6 weeks old) in the 40 Survival rate (%)

summers (July–August) of 2013 and 2014 at Recovery 20 the Shanghai station and the 2014 summer at L20 L22 L25 L20 L22 L25 L20 L22 L25 the Wuhan station. (g) Post-summer survival Vector Vector Vector rates of the field-grown transgenic tomato d 80 h 12 d i c c lines tested at Shanghai (2013 and 2014) ** 100 ** cd b b and Wuhan (2014), shown as dots of replicates 60 9 c 80 ** a (30–50 plants each). **P < 0.01, by Student’s 6 b 60 a a 40 40 t-test and Bonferroni correction for three 3 a a a 20 comparisons in the independent summer field 20 levels of S/ER Survival rate (%) Survival rate (%)

Relative expression 0 0 tests. (h) SlER transcript levels in selected 0 tomato accessions (4 weeks old, grown at M82 M82 L20 L22 L25 LA1589LA1781LA1580 LA1589LA1781LA1580 28 °C), as shown in dots of three replicates. Vector Heinz1706 Heinz1706 Borgo cellano Borgo cellano The tomato SleIF4a6 gene was used as Djena gold girl Djena gold girl control to normalize expression levels. (i) Seedling survival rates of the six chosen tomato accessions plus LA1589 after 3-d heat treatment (40 °C/28 °C (day/night)) and 2-d 22 °C recovery growth, shown as dots of three replicates (30 plants each). Different letters at the top of dots indicate a significant difference at P < 0.05, by Welch’s ANOVA and Bonferroni correction for six tests (h,i). Survival rates for normal growth

Nature America, Inc. All rights reserved. America, Inc. © 201 5 Nature temperature (22 °C) controls are 100% (d,i). (See source data.)

higher proportion of the L7-1 plants (~43%) survived after recovery by reduced transpiration. We undertook these experiments because npg under 22 °C compared with the wild-type control (~13%) (Fig. 1j). ER-OE lines had lower leaf conductance compared with the wild Similarly, the two ER-MH lines (LM10 and LM17) with higher type, which could counteract the effect of higher temperatures on ER-MH abundance were more thermotolerant than the wild-type control transpiration rates. We carried out 40 °C heat treatments for 48 h for (Fig. 1j). Taken together, these experiments show that ER overexpres- er-105, L7-1 and the wild-type Col-0 under low (30–40%) and high sion substantially increased heat tolerance in Arabidopsis. or near-saturated (90–95%) relative humidity (RH) because humidity The ER locus is known to be a determinant of transpiration is known to be inversely related to transpiration; we then measured efficiency (the rate of carbon fixation to water loss) or water use plant survival rates. Counterintuitively, the survival rate of all three efficiency30. Loss-of-function er mutations have decreased transpira- genotypes was higher (31–71%) under low RH than under high RH tion efficiency, likely owing to increased leaf porosity and decreased (16–63%; Fig. 2a), possibly due to a faster transpiration rate, result- photosynthetic capacity30. Consistent with this, ER overexpression ing in lower leaf surface temperature at low RH. No difference in in Arabidopsis lines L2-3 and L7-1 decreased stomatal density and water content between er-105 and wild-type Col-0 incubated at 40 °C enlarged epidermal cells (Supplementary Fig. 5), therefore decreas- for 24 h was detected (Fig. 2b). Notably, the difference in relative ing stomatal conductance overall (Fig. 1k). Importantly, we found survival rate of L7-1/the wild type/er-105 was more significant under that ER overexpression increased both instantaneous and integrated high RH (predicted to cause slower transpiration rate with higher leaf water-use efficiency relative to the wild-type Col-0 (Fig. 1l,m). surface temperature) (~4:2:1) than under low RH (~2:1.6:1) (Fig. 2a), In addition, ER-OE lines were more tolerant to drought than the indicating that ER had a stronger impact on thermotolerance under wild-type (Supplementary Fig. 6). The fewer stomata and lower high RH than under low RH. Taken together, our data indicate that transpiration rate might result in increased drought tolerance in the ER enhanced plant thermotolerance through an unknown cellular ER-OE plants. However, an alternate explanation is that the ER-mediated mechanism that is independent of its effect on leaf transpiration30. thermotolerance mechanism occurs to affect drought tolerance. Supporting an independent role for ER in thermotolerance, an earlier To better understand how ER improves heat tolerance, we first deter- study showed that ER is also required for thermoprotection in the leaf mined whether the ER-mediated thermotolerance was simply caused AS1/AS2 pathway for adaxial-abaxial polarity formation29.

nature biotechnology VOLUME 33 NUMBER 9 SEPTEMBER 2015 999 l e t t e r s

42 °C a b 35S::ER d e 100 28 °C Vector L40 L36 L38 ** Vector L40 L36 L38 80 ** ** 8,000 60

7,000 40 C ° 6,000 20 28 Seed setting (%) 5,000 0

4,000 L40 L36 L38 L40 L36 L38 Vector Vector 3,000 2014 summer, Shanghai

Relative expression 2,000 f 2013 2013 Vector 35S::ER L36 1,000 c 100 summer, 2013 summer, * ** Shanghai summer, Changsha C

0 ° 80 * * * Wuhan * 42 L40 L36 L38 60 * Vector 40 Figure 4 ER enhances 42 °C 20 heat tolerance in rice. Seed setting in eld (%) 0 (a) qPCR analysis of the transgenic ER L40 L36 L38 L38 L38 L40 L36 L38 Vector Vector Vector Vector expression in 4-week-old 2014 transgenic rice grown at 28 °C. g summer, h i The rice ACTIN1 gene was used Shanghai 100 Wild type Oser1 28 °C as control to normalize expression 25 ** 2013 **** 80 levels. Values are means ± s.d. 20 summer, Shanghai 60 (n = 3). (b) Pictures of 1-week-old 42 °C 15 2013 seedlings of transgenic rice plants ****** summer, 2013 40 grown at 28 °C. (c) Pictures of 10 Wuhan summer, in eld (g) Changsha Seed setting (%) 20 ** mature transgenic rice plants 5 *

Grain yield per plant 0 after 10-d growth with a 42 °C 0 (40–43 °C recorded due to daily L40L36L38 L38 L38 L40L36L38 42 °C Oser1 Oser1 fluctuation)/35°C (day/night) cycle Vector Vector Vector Vector Wild type Wild type and 1-week recovery growth at 28 °C. (d) Shown here are panicles and seeds collected from transgenic plants grown at constant 28 °C or subjected to the 10-d growth with a 42 °C/35 °C (day/night) cycle while flowering (c). Note that aberrant spikelets (white) were greatly decreased in the 35S::ER lines. (e) Quantification of seed setting after heat treatment (c) or grown under normal conditions (28 °C) in growth chambers as shown in box plots (n = 15). **P < 0.01, by Student’s t-test and Bonferroni correction for three comparisons’ test at 42 °C. (f) Quantification of seed-setting rates of transgenic rice plants in field tests at three different locations in the summer of 2013 (Shanghai, Wuhan and Changsha), and in the summer of 2014 at the Shanghai station, shown as box plots (n ≥ 30). (g) Analysis of grain yield of transgenic rice plants tested in the listed four field experiments, shown as box plots (n ≥ 30). Filled grains were weighed for each plant. **P < 0.01, by Student’s t-test and Bonferroni correction for three comparisons in the Shanghai (2013 and 2014) summer field test, or *P < 0.05, by Student’s t-test in the Wuhan and Changsha summer field tests (f,g). (h) Shown here are panicles of the flowering wild-type Nature America, Inc. All rights reserved. America, Inc. © 201 5 Nature Dongjin and OsER1 mutant plants that were grown for 10 d at 42 °C/35 °C (day/night) in the growth chamber. Note that spikelets of the OsER1 mutant plants were almost dead. (i) Quantification of seed setting of the OsER1 mutant and wild-type plants after heat treatment (h) or grown under normal conditions (28 °C) in growth chambers, shown as box plots (n = 15). **P < 0.01, by Student’s t-test (**P < 0.01). (See source data.) npg

To better understand how ER enhances plant thermotolerance er-105 was rapidly broken without typical plasmolysis (Fig. 2d and at the cellular level, we analyzed the cellular responses of er-105, Supplementary Fig. 8). Earlier and more severe membrane blebbing ER-OE and the wild-type Col-0 to 40 °C heat stress. We found that and collapsing were also detected in the chloroplasts and mitochon- the surfaces of wild-type Col-0 and er-105 leaves were similar at dria of er-105 cells subjected to heat stress (Supplementary Fig. 8). normal temperatures but differed when lines were shifted to 40 °C Consistent with the observed membrane damage, trypan blue stain- (Supplementary Fig. 7). 12 h after heat treatment, the entire leaf ing revealed increased cell death in er-105 (Fig. 2f); we also observed surface of er-105 was wrinkled and rough, whereas the leaf surface of increased accumulation of H2O2, which is associated with cell damage, Col-0 was only just starting to become wrinkled, which suggests that relative to the wild-type control (Fig. 2g). By contrast, cells in the leaf cells in er mutants are more easily damaged or killed under heat ER-OE plants were resistant to heat-induced cellular damage and stress. It is already known that the stability of the cell membrane is remained healthy 24 h after heat treatment, with less ion leakage and essential for plant thermotolerance and that high temperatures dis- cell damage compared with wild type (Fig. 2c–f and Supplementary rupt cell membranes. We found that er-105 leaf cells had more ion Fig. 8). These data suggested the hypothesis that ER overexpression leakage than those in the wild-type Col-0 (82% vs. 60%) (Fig. 2c). might protect membrane integrity during heat stress. Transcript lev- Transmission electron microscopy (TEM) revealed that the plasma els of HSFA1a31 and BI1 (ref. 32), both of which are involved in heat membrane seemed to be blebbing and broken in er-105 cells 12 h after response and cell death inhibition, increased in L7-1 and decreased in heat treatment, whereas wrinkled and broken plasma membrane was er-105 (Fig. 2h,i). Interestingly, heat stress suppressed transcription only detected in WT Col-0 cells 24 h after heat treatment (Fig. 2d and of ER (Supplementary Fig. 9). Taken together, these results might Supplementary Fig. 8), with a higher occurrence of plasma membrane indicate an important role for ER in protecting plant cells from disruption in er-105 than in Col-0 (71% vs. 46%) 24 h after heat heat-induced cellular damage and death. treatment) (Fig. 2e). Plasma membrane disruption was often associ- Next, we evaluated whether ER could be used to improve heat tol- ated with plasmolysis in Col-0; however, the plasma membrane in erance in crops. We transformed tomato (dicot) and rice (monocot)

1000 VOLUME 33 NUMBER 9 SEPTEMBER 2015 nature biotechnology l e t t e r s

with the p35S::ER genomic transgene and analyzed the growth (Fig. 3i). It therefore seems likely that ER overexpression could greatly phenotypes and heat tolerance of the resulting transgenic crop lines improve thermotolerance in commercially relevant tomato lines. in both laboratory and field conditions. We used the small red-fruit We also introduced the p35S:ER transgene into the model rice vari- tomato species, Solanum pimpinellifolium (accession LA1589, the ety Nipponbare (Oryza sativa L. japonica) and identified three lines closest wild relative of the cultivated tomato), which is known to that accumulated high levels of transgenic ER transcripts (Fig. 4a). be hypersensitive to chilling, high temperature and high humid- These ER-OE transgenic rice seedlings developed longer roots and ity, compared with modern tomato varieties. Similar to Arabidopsis, bigger leaves at the seedling stage compared to the control transgenic the ER-OE transgenic tomato lines were also bigger than the wild rice plants carrying the empty vector (Fig. 4b and Supplementary type, with larger leaves (nearly twice as large as wild type, likely Fig. 14). The ER-OE transgenic rice plants also had increased resulting from enlarged cells) (Fig. 3a,b and Supplementary Fig. 10), heat tolerance. After 10 d of growth at 42°C/35°C (day/night) followed decreased stomatal density, decreased stomatal conductance and by 1-week recovery growth at 28 °C, many leaves and tillers of the increased transpiration efficiency (Supplementary Figs. 10 and 11). ER-OE transgenic plants remained green and alive, whereas those This result is consistent with an earlier finding of reduced vegeta- of the vector control lines were dried and almost dead (Fig. 4c). tive growth by dominant-negative disruption of the ER-mediated Because pollination, and hence spikelet fertility and seed setting, is signaling pathway in transgenic tomatoes33. Importantly, the ER-OE the most heat-sensitive trait of rice, we measured seed-setting rates transgenic tomato plants had significantly (P < 0.01) increased heat after the prolonged heat treatment. ER-OE transgenic rice plants had tolerance as measured by survival rates (Fig. 3c,d) when grown significantly higher seed setting (55–70%) than the control transgenic under heat stress at 40 °C/28 °C (day/night) for 10 d in the growth line (~35%), after heat stress (Fig. 4d,e). chamber, and also decreased cell death under heat treatment for Field tests of transgenic rice plants in the Shanghai and Wuhan 48 h (Supplementary Fig. 11). stations during the summer heat wave of 2013 showed that ER-OE Field experiments in three summer seasons (2012, 2013 and 2014) transgenic rice had higher seed-setting rates and yield potential than and at three planting locations (Shanghai, Wuhan and Hainan) of the the control transgenic plants (Fig. 4f,g and Supplementary Fig. 15). transgenic lines L20, L22 and L23 confirmed that ER overexpression Similar results were also observed in the summer of 2014 at the improved heat tolerance in tomato. We found that >50% of transgenic Shanghai station (Fig. 4f,g). We also observed significantly higher tomato plants survived the summer (July and August) heat waves in (~49%; P < 0.05) seed setting of ER-OE transgenic rice than the control 2012 and 2013 at our experimental station in Shanghai (30–50 plants line (~45%) in a field test during the 2013 summer in Changsha (another each line, three replicates), where the daily high temperature of the ‘furnace’ city in China); however, the grain yield of the ER-OE line was summer often surpassed 36 °C (Supplementary Data Set 1), whereas not significantly higher than the vector control’s, probably because >70% of control transgenic lines carrying the empty vector died or more seeds developed on the ER-OE plants in this location and thus almost died (Fig. 3e–g and Supplementary Fig. 12a). Similar results the ER-OE line produced grains that were slightly smaller than those were observed in the 2014 summer at the Shanghai station (30–50 of the vector control (Fig. 4f,g and Supplementary Fig. 15). When plants each line, three replicates) (Fig. 3f,g), whereas no significant grown in the autumn of 2013 (flowered in early October) with normal difference in the survival rate between the two groups was observed in growth temperatures in the Shanghai station, the ER-OE transgenic the early summer of 2014 when the growth temperature was normal lines had the same seed-setting rates as those of the control line (early June to early July) (Supplementary Fig. 12d). The increased (Supplementary Fig. 15). Therefore, ER could increase rice toler- thermotolerance of transgenic tomato was also observed for two sum- ance to heat. Additional genetic support for a role of ER in heat toler- Nature America, Inc. All rights reserved. America, Inc. © 201 5 Nature mer field tests in 2013 and 2014 in a mid-eastern China city Wuhan ance came from our studies of two rice T-DNA insertional mutants, (one of the seven hottest cities, which includes Shanghai, widely one carrying a T-DNA null allele in Os06g0203800 (referred to as known as the ‘seven furnaces’ in China) (Fig. 3f,g and Supplementary OsER1) (Supplementary Fig. 16a) and the other with a T-DNA npg Fig. 12b), and an early-summer field test in 2013 on Hainan Island, insertion in Os06g0130100 (referred to as OsER2) (Supplementary the southernmost province in China (Supplementary Fig. 12c). The Fig. 16b). Consistent with our overexpression data, the panicles of tomato species used in this study produced only a few small fruits that the OsER1 plants died quickly when grown at high temperatures in often shriveled under our growth conditions. We could obtain only a growth chamber, leading to significantly (P < 0.01) decreased seed enough fruits and seeds for our experiments by hand pollination and setting (Fig. 4h,i). By contrast, no difference in heat tolerance was with extreme care. Thus, we were not able to accurately measure yields detected between OsER2 and its wild-type control (Supplementary of tomato fruit of the wild-type control and transgenic lines. Under Fig. 16c), suggesting that OsER2 is a different ER gene that does not normal growth conditions in our greenhouse (22 °C), we observed function in heat tolerance in rice. Taken together, our transgenic no difference in the flower number or weight of fruits (produced by and genetic results clearly demonstrate that ER is also an important hand pollination) between the ER-OE lines and the wild-type plants thermotolerance regulator in rice. (Supplementary Fig. 13). We have shown that ER is a major QTL for heat tolerance, thereby As the parental strain LA1589 is not a widely cultivated crop, we revealing a new function for this well-studied receptor-like kinase in searched a set of modern tomato germplasm and identified four the stress response25. It is intriguing that no difference was observed in tomato (Solanum lycopersicum) varieties, M82, Djena gold girl, water content between 40 °C-treated er-105 and its wild-type control, Borgo Cellano and Heinz 1706 (the genome-sequenced cultivated despite the fact that ER positively regulates transpiration efficiency, tomato), that carry high-expression alleles (more than fourfold indicating that ER-mediated thermotolerance is likely unrelated to higher compared to LA1589) of SlER, the tomato gene orthologous plant water status. Further support for a water loss–independent, ER- to ER, and two others (LA1580 and LA1781) with expression levels mediated thermotolerance mechanism came from our experiment of SlER similar to that of LA1589 (Fig. 3h). Heat treatment showed showing that the relative heat tolerance of the ER-OE plants in com- that the four varieties with higher SlER expression levels were more parison with the wild-type Col-0 plants was higher when grown under heat tolerant as measured by survival rates, whereas the two low-SlER- high relative humidity, which should reduce plant transpiration. expression accessions were more susceptible to high temperatures ER likely confers heat tolerance by protecting cells from heat-induced

nature biotechnology VOLUME 33 NUMBER 9 SEPTEMBER 2015 1001 l e t t e r s

cellular damage and cell death. This hypothesis is consistent with AUTHOR CONTRIBUTIONS the reported effects of ER on chloroplast electron transport30. H.S., X.Z., F.Z., Y.W., G.C., Y.H., H.X., J.L. and Z.H. conceived the research project, Further detailed research should shed light on a better mechanistic designed experiments and analyzed the data. H.S., X.Z., F.Z., Y.W., B.Y., Q.L., G.C., B.M., J.W., Y.L. and G.X. conducted the experiments. Z.H. and J.L. oversaw the understanding of the ER-mediated, water loss–independent, ther- entire study and wrote the manuscript. motolerance process. Given that crop production is often hard hit during hot summers1–3, COMPETING FINANCIAL INTERESTS engineering heat tolerance in plants could prove important. However, The authors declare no competing financial interests. until now engineering approaches have not been particularly successful, Reprints and permissions information is available online at http://www.nature.com/ mainly because either increased stress tolerance has been accompa- reprints/index.html. nied by a growth and/or yield penalty, or because thermotolerance 1. Peng, S. et al. Rice yields decline with higher night temperature from global of engineered crops has not translated from the greenhouse to the warming. Proc. Natl. Acad. Sci. USA 101, 9971–9975 (2004). field at multiple locations. 2. Lobell, D.B. & Field, C.B. Global scale climate-crop yield relationships and the impacts of recent warming. Environ. Res. Lett. 2, 014002 (2007). This study has shown, to our knowledge for the first time, that 3. Long, S.P. & Ort, D.R. More than taking the heat: crops and global change. overexpressing ER significantly increased the tolerance of plants to Curr. Opin. Plant Biol. 13, 241–248 (2010). 4. Penfield, S. Temperature perception and signal transduction in plants. New Phytol. heat and prolonged hot temperatures, not only in the model plant 179, 615–628 (2008). Arabidopsis but also in two important food crop species (or crop 5. Fitter, A.H. & Fitter, R.S. Rapid changes in flowering time in British plants.Science relative) with no growth and/or yield penalty. Consistent with our 296, 1689–1691 (2002). 6. Samach, A. & Wigge, P.A. Ambient temperature perception in plants. Curr. Opin. transgenic studies, low expression SlER alleles in tomato and a loss-of- Plant Biol. 8, 483–486 (2005). function rice ER gene were associated with reduced thermotoler- 7. Kumar, S.V. & Wigge, P.A. H2A.Z-containing nucleosomes mediate the thermosensory ance. Importantly, many of the ER-OE transgenic plants also had response in Arabidopsis. Cell 140, 136–147 (2010). 8. Posé, D.S.H. et al. Temperature-dependent regulation of flowering by antagonistic increased biomass and enhanced water use efficiency, which further FLM variants. Nature 503, 414–417 (2013). substantiates an earlier claim that ER is a major locus for transpiration 9. Queitsch, C., Hong, S.W., Vierling, E. & Lindquist, S. Heat shock protein 101 plays 30 a crucial role in thermotolerance in Arabidopsis. Plant Cell 12, 479–492 (2000). efficiency . This is especially important because global warming is 10. Finka, A., Cuendet, A.F.H., Maathuis, F.J.M. & Saidi, Y. Plasma membrane cyclic causing a worldwide water deficiency. nucleotide gated calcium channels control land plant thermal sensing and acquired Our findings indicate that overexpression of ER is likely superior thermotolerance. Plant Cell 24, 3333–3348 (2012). 9,14 11. Iba, K. Acclimative response to temperature stress in higher plants: approaches of to using HSPs to improve thermotolerance . Plants engineered to gene engineering for temperature tolerance. Annu. Rev. Plant Biol. 53, 225–245 produce HSP101 have not been tested in multiple locations or through (2002). several growing seasons and HSP overexpression in model plants has 12. Kim, M., Lee, U., Small, I., des Francs-Small, C.C. & Vierling, E. Mutations in an Arabidopsis mitochondrial transcription termination factor-related protein enhance 34 been associated with growth penalties under nonstress conditions . thermotolerance in the absence of the major molecular chaperone HSP101. Finally, HSP genes are often functionally redundant, and no single Plant Cell 24, 3349–3365 (2012). 13. Guan, Q., Yue, X., Zeng, H. & Zhu, J. The protein phosphatase RCF2 and its HSP locus has been reported as a major QTL in the gene-to-field test interacting partner NAC019 are critical for heat stress–responsive gene regulation 34 for crop thermotolerance improvement . and thermotolerance in Arabidopsis. Plant Cell 26, 438–453 (2014). Our study indicates that a single ER gene-mediated heat tolerance 14. Burke, J.J. & Chen, J. Enhancement of reproductive heat tolerance in plants. PLoS ONE 10, e0122933 (2015). pathway is likely to be conserved in higher plants, which is consistent 15. Li, X. et al. Natural alleles of a proteasome α2 subunit gene contribute to with a recent finding that a poplar ER homolog could increase biomass thermotolerance and adaptation of African rice. Nat. Genet. 47, 827–833 and water use efficiency in transgenic Arabidopsis35. ER-like genes are (2015). Nature America, Inc. All rights reserved. America, Inc. © 201 5 Nature 16. Grover, A., Mittal, D., Negi, M. & Lavania, D. Generating high temperature tolerant 36 widely distributed in plants , which is a good basis for improving transgenic plants: Achievements and challenges. Plant Sci. 205-206, 38–47 crop thermotolerance through identification of ER alleles with higher (2013). 17. Hancock, A.M. et al. Adaptation to climate across Arabidopsis thaliana genome. expression levels or higher activity from available crop germplasm Science 334, 83–86 (2011). npg collections. Future work will include introgression of identified elite 18. Fournier-Level, A. et al. A map of local adaptation in Arabidopsis thaliana. Science ER alleles into elite crop varieties through marker-assisted selection 334, 86–89 (2011). 19. Yan, C., Shen, H., Li, Q. & He, Z. A novel ABA-hypersensitive mutant in Arabidopsis to improve thermotolerance of major crops. defines a genetic locus that confers tolerance to xerothermic stress. Planta 224, 889–899 (2006). Methods 20. Lin, L., Zhong, S.H., Cui, X.F., Li, J. & He, Z.H. Characterization of temperature- sensitive mutants reveals a role for receptor-like kinase CRAMBLED/STRUBBELIG Methods and any associated references are available in the online in coordinating cell proliferation and differentiation during Arabidopsis leaf version of the paper. development. Plant J. 72, 707–720 (2012). 21. Zhong, S. et al. Warm temperatures induce transgenerational epigenetic release of Note: Any Supplementary Information and Source Data files are available in the RNA silencing by inhibiting siRNA biogenesis in Arabidopsis. Proc. Natl. Acad. Sci. 110, 9171–9176 (2013). online version of the paper. USA 22. Torii, K.U. et al. The Arabidopsis ERECTA gene encodes a putative receptorprotein kinase with extracellular leucine-rich repeats. Plant Cell 8, 735–746 (1996). Acknowledgments 23. Shpak, E.D., Berthiaume, C.T., Hill, E.J. & Torii, K.U. Synergistic interaction of We thank J.-S. Jeon and C.Y. Wu for the rice T-DNA mutants, X.L. Wang for the three ERECTA-family receptor-like kinases controls Arabidopsis organ growth and er-105 line, X.G. Zhu, Y.J. Zhang and M.Z. Lv for help in statistical analysis, flower development by promoting cell proliferation. Development 131, 1491–1501 D.Y. Sun and H.X. Lin for helpful discussions, and L. Lin for help in experiments. (2004). This work was supported by the National Key Basic Research and Development 24. Shpak, E.D., McAbee, J.M., Pillitteri, L.J. & Torii, K.U. Stomatal patterning and Program (2011CB100700 to H.Z.), the National Natural Science Foundation differentiation by synergistic interactions of receptor kinases. Science 309, of China (3130061 to H.Z.), the Ministry of Science and Technology of China 290–293 (2005). 25. van Zanten, M., Snoek, L.B., Proveniers, M.C. & Peeters, A.J. The many functions (2012AA10A302 to Z.H.), the National GMO project (2013ZX08009-003-001 of ERECTA. Trends Plant Sci. 14, 214–218 (2009). to Z.H.), the National High Technology Research and Development Program 26. Lee, J.S. et al. Direct interaction of ligand-receptor pairs specifying stomatal of China (2012AA100104-6 to H.X.), the Chinese Academy of Sciences (KSCX2- patterning. Genes Dev. 26, 126–136 (2012). EW-N-01 to H.Z. and 2009OHTP07 to H.X.), the Shanghai Committee of Science 27. Uchida, N. et al. Regulation of inflorescence architecture by intertissue layer ligand- and Technology (11PJ1410900 to H.X.), and the National Key Basic Research and receptor communication between endodermis and phloem. Proc. Natl. Acad. Sci. Development Program (2015CB150104 to G.C.). USA 109, 6337–6342 (2012).

1002 VOLUME 33 NUMBER 9 SEPTEMBER 2015 nature biotechnology l e t t e r s

28. Tisné, S. et al. Combined genetic and modeling approaches reveal that 32. Ishikawa, T., Uchimiya, H. & Kawai-Yamada, M. The role of plant Bax inhibitor-1 in epidermal cell area and number in leaves are controlled by leaf and plant suppressing H2O2-induced cell death. Methods Enzymol. 527, 239–256 (2013). developmental processes in Arabidopsis. Plant Physiol. 148, 1117–1127 33. Villagarcia, H., Morin, A.C., Shpaket, E.D. & Khodakovskaya, M.V. Modification of (2008). tomato growth by expression of truncated ERECTA protein from Arabidopsis thaliana. 29. Qi, Y., Sun, Y., Xu, L., Xu, Y. & Huang, H. ERECTA is required for protection against J. Exp. Bot. 63, 6493–6504 (2012). heat-stress in the AS1/AS2 pathway to regulate adaxial-abaxial leaf polarity in 34. Mickelbart, M.V., Hasegawa, P.M. & Bailey-Serres, J. Genetic mechanisms of abiotic Arabidopsis. Planta 219, 270–276 (2004). stress tolerance that translate to crop yield stability. Nat. Rev. Genet. 16, 237–251 30. Masle, J., Gilmore, S.R. & Farquha, G.D. The ERECTA gene regulates (2015). plant transpiration efficiency in Arabidopsis. Nature 436, 866–870 35. Xing, H.T., Guo, P., Xia, X.L. & Yin, W.L. PdERECTA, a leucine-rich repeat receptor- (2005). like kinase of poplar, confers enhanced water use efficiency in Arabidopsis. Planta 31. Li, S. et al. HEAT-INDUCED TAS1 TARGET1 mediates thermotolerance via HEAT 234, 229–241 (2011). STRESS TRANSCRIPTION FACTOR A1a–directed pathways in Arabidopsis. Plant 36. Shiu, S.H. et al. Comparative analysis of the receptor-like kinase family in Cell 26, 1764–1780 (2014). Arabidopsis and rice. Plant Cell 16, 1220–1234 (2004). Nature America, Inc. All rights reserved. America, Inc. © 201 5 Nature npg

nature biotechnology VOLUME 33 NUMBER 9 SEPTEMBER 2015 1003 ONLINE METHODS and early summer (late April–early June) on Hainan Island for field tests of heat Plant materials and treatment. Seeds of the Arabidopsis ecotypes Col-0, Ler tolerance. Tomato plants were also grown in the early summer of 2014 (early and Lan were stratified at 4 °C for 3 d and germinated at 22 °C for 1 week on June–early July) at the Shanghai station under normal temperature conditions. 1/2× MS medium. Seedlings were transplanted into soil and grown at 22 °C In 2013 and 2014 in the Shanghai station, 30–50 6-week-old plants of each line in a growth room with a 16 h/8 h (day/night) photoperiod and ~ 65% relative with three biological repeats (total 30–50 × 3 plants each line) were grown for humidity (RH). For the heat treatment experiment, 2-week-old plants grown at about 1 month, by which time the control plants were almost dead. Survival 22 °C were transferred into a growth chamber with 80% RH and temperatures rates (%) were measured (alive plants/total plants as for Arabidopsis) Because gradually increasing from 22–30 °C for 12 h and then 36 °C for 12 h, followed of the strict governmental administration on plant numbers of transgenic by 40 °C for 48 h, or simply grown at 30 °C as previously reported18,19, for plant test, the additional field tests conducted in 2012 (Shanghai) and 2013 survival rate analysis. Two-week-old plants were also grown at 40 °C for 48 h (Hainan, Wuhan) used ~50 plants/line for direct survival rate measurement. under low RH (30–40%) or high RH (90–95%, near saturated with water vapor) Because there were no replicates, we did not perform statistical significance to determine the effect of transpiration on heat tolerance. Plants were watered analysis in these additional field tests. every day (40 °C) or every 2 d (30 °C) to keep the soil wet during heat treat- ment. Each experiment was conducted with 30 plants each line, three replicates QTL mapping and plant transformation. A set of 30 recombinant inbred lines (30 × 3), total 90 plants each line. Plants grown at 22 °C were used as heat (RILs) derived from a cross of Col and Ler38 obtained from the Nottingham treatment controls. For drought treatment, 4-week-old plants were subjected Arabidopsis Stock Centre (http://Arabidopsis.info/) was screened for heat tol- to gradual drought stress until er-105 or Ler was dry dead, and the plants were erance. Thirty plants per line with three biological replicates (for each line, then rewatered with the same amount of water to permit recovery; 30 plants there were three sets of 30 plants) were tested, divided by phenotype into three each line, three replicates (30 × 3), total 90 plants each line were analyzed to groups, and genotyped with PCR-based indel makers. QTLs were detected determine the survival rate. using Windows QTL Cartographer version 2.5 and a LOD threshold of 2.5 For rice heat treatments in growth chambers, heading plants were grown in (α = 0.05). Based on RIL screening, CSSLs from the BC3F2 populations of the growth chamber with a 16 h/8 h photoperiod, a 42 °C (40–43 °C recorded Col-0 × Ler with Ler as the recurrent parent were developed and screened for due to daily fluctuation)/35 °C (day/night) temperature cycle and 80–90% RH qHat loci, and subsequent fine mapping of qHat2-1 was performed using the for 10 d, with the control experiments performed at 28 °C. The plants were then selected CSSL and its progeny. allowed to recover at 28 °C in the greenhouse for one week. Spikelet fertility, known to be sensitive to heat stress37, was statistically analyzed with 30 panicles Generation of transgene constructs and plant transformation. A 7,832-bp from 15 individual plants for each line with three replicates. For tomato heat genomic fragment including the entire wild-type ER coding region and its treatments in growth chamber, 6-week-old plants or cutting-propagated plants native promoter (1,802-bp) and 3′-noncoding region (501 bp) was isolated were grown in the growth chamber with a 16 h /8 h photoperiod, a 40 °C (38–42 °C from BAC T1D16 by digestion using EcoRI and SnaBI and inserted into the recorded because of daily temperature fluctuation)/28 °C (day/night) tem- EcoRI and SmaI sites of the cloning vector pBluescript II SK(−). The fragment perature cycle and 80–90% RH for 10 d followed by 2-d recovery growth was then released from SK using KpnI and BamHI digestion and inserted into at 22 °C, with the control experiments performed at 22 °C. For testing the the binary vector pCAMBIA1301 to generate the pER::ER genomic transgene. heat tolerance of tomato SlER alleles, 3-week-old seedlings were grown at This plasmid was used to transform Ler or er-105 to generate >10 independent 40 °C for 3 d followed by 2-d recovery growth at 22 °C to determine survival complementation lines using Agrobacterium tumefaciens-mediated transfor- rates. About 30 plants each line, three replicates (30 × 3), total 90 plants each mation with floral dipping39. To generate the p35S::ER construct, a 412-bp line were treated, and survival rate was measured for tomato heat tolerance. 5′ ER fragment with a 42-bp 5′ nontranslation region was PCR-amplified from Rice and tomato plants were watered daily to keep the soil wet during heat BAC T1D16 using the primers 35S-ER-5′-F and 35S-ER-5″-R (Supplementary treatment. Flower numbers per inflorescence were recorded from 10 plants Table 2), digested with ClaI/SphI to release a 140-bp fragment with the (3 inflorescences per plant), and fruit weights were measured for 20 fruits 42-bp 5′ nontranslation region, which was cloned into the ClaI/SphI-digested

Nature America, Inc. All rights reserved. America, Inc. © 201 5 Nature produced by hand pollination, with well maintenance in the plants grown in pER::ER plasmid. The resulting plasmid was sequenced to ensure there was the greenhouse (22 °C). no PCR-introduced error and digested with ClaI and BamHI to release the promoter-less ER genomic fragment, which was subsequently cloned into a Field experiments. Field tests in multiple growing seasons and at different overexpression vector pCAMBIA1301-35SN (35S- C1301) that contains the npg locations for rice and tomato were carried out during 2012–2014. Plants were 35S promoter of the cauliflower mosaic virus with two copies of an enhancer maintained under the same field conditions and were well irrigated and pro- element to drive the expression of the transgenes40, to generate the p35S::ER tected from pests to ensure meaningful field tests. For the rice field tests, rice overexpression transgene. The plasmid p35S::ER was introduced into Col-0, plants were grown in the summer to flower at mid-August to early September and the resulting transgenic T1 lines were screened by qPCR for ER- (2013) at three well-controlled biological stations located in Shanghai overexpressing lines. Homozygous ER-overexpressing T3-T4 lines were used (eastern China), Wuhan (mid-eastern China) and Changsha (southeastern for phenotypic analyses. All transgene constructs were fully sequenced to China), widely known as furnace cities in China because of their hot summer ensure no PCR/cloning-introduced error. weather, where the recorded daily temperatures during the hot summer often The p35S::ER construct was used to transform tomato germplasm (Solanum reach 36–40 °C. Rice plants were also tested in the summer of 2014 in the pimpinellifolium LA1589)41 that was previously shown to be sensitive to Shanghai station. Daily records of high, low and average temperatures and temperature stress42, to generate >15 independent transgenic lines (T0) by relative humidity (RH) during the experimental seasons of 2012–2014 were Agrobacterium tumefaciens-mediated cotyledon transformation. The resulting obtained from the local weather stations (Supplementary Data Set 1). Rice transgenic lines were screened by qPCR to identify ER-overexpressing lines was also grown to flower in the autumn (early October) of 2013 at our Shanghai whose subsequent homozygous progeny (T1–T5) were used for heat tolerance experimental station to test performance under normal growth temperatures. studies and microscopic analyses. The p35S::ER transgene was also introduced 50–60 plants each line, three replicates (50–60 × 3) were grown. qPCR was into the rice variety Nipponbare to generate >20 lines (T0) by Agrobacterium used to check ER expression in all transgenic plants grown. About 30–50 tumefaciens-mediated transformation, and the resulting transgenic lines were plants each line, three replicates (30–50 × 3) were analyzed for agronomic screened by qPCR to find ER-overexpression lines. The T1–T5 homozygous traits. Due to government regulations we could test line 38 and the control progeny plants of the selected ER-overexpressing lines were assayed for heat vector line only in net-protected fields in Wuhan and Changsha (2013). tolerance and agronomic traits. Seed-setting rate and other important agronomic traits including grain To generate an epitope-tagged ER transgene, the p35S::ER was fused in- (full-filled) yield per plant were quantified during these field trials (see source frame with a 7× myc–6× His (MH) fragment between the last codon and data associated with Fig. 4, and Supplementary Figs. 14 and 15). the stop codon to create the p35S::ER-MH plasmid by PCR fragment fusion Tomato plants were transplanted at our experimental stations during the using the primers (Supplementary Table 2), which was introduced into er-105 summers (July–August) of 2012–2014 in Shanghai and 2013, 2014 in Wuhan to generate >30 lines. The resulting transgenic lines were screened

nature biotechnology doi:10.1038/nbt.3321 morphologically to identify lines with increased biomass and qPCR for Physiological assay. Gas exchange parameters including the net photosyn- ER-MH overexpression lines, and by immunoblotting to identify lines with thetic rate, the transpiration rate and the stomatal conductance of Arabidopsis higher ER-MH levels. Thermotolerance of the homozygous progeny (T2 and T3) (4-week-old) and tomato (6-week-old) plants grown at 22 °C were measured of the chosen lines with increased biomass were compared with wild-type in situ using a LI-6400 gas exchange system (LI-COR Biosciences). Newly Col-0 as described above. fully expanded leaves at the same development stage of each genotype (eight to ten plants per genotype, three biological replicates) were measured at the Identification of rice T-DNA mutants and tomato SlER alleles. A homozygous conditions of 300 µmol photon/m/s and 400 µmol/mol ambient CO2 concen- T-DNA knockout mutant of the rice ER homolog gene (Os06g0203800, OsER1) tration. The temperature of the leaf chamber was maintained at 22 °C. The leaf in Dongjin (japonica) background was obtained from the Korean rice T-DNA to air vapor difference was kept at about 1.0 KPa during measurement. The mutant center (the National Crop Experiment Station, RDA, Suwon, Korea) instantaneous water-use efficiency was calculated as the ratio of net photosyn- whereas a homozygous T-DNA insertional mutant in the second rice ER thetic rate to transpiration rate. Relative water content (RWC) was measured as homolog (Os06g013010, OsER2) in Zhonghua 11 background (japonica) was previously described44. In brief, 2-week-old Arabidopsis plants grown at 22 °C generously provided by the Chinese rice T-DNA mutant center (the Huazhong were heat-treated (40 °C) in the growth chamber. Twelve mature rosette leaves rice mutant center, http://rmd.ncpgr.cn/). The molecular effect of the T-DNA from 12 individuals (1 leaf /plant) were sampled at 0, 12, 24, 36 and 48 h with insertions on the two ER homologs was analyzed by qPCR using the OsER1- three biological replicates. Fresh weight (FW) was measured immediately after RT8F/R and OsER2-RT3F/R primer sets (Supplementary Table 2). The sampling. Leaves were immersed in deionized water for 4 h, and then blotted confirmed homozygous OsER1 and OsER2 mutants and the corresponding dry and weighted (turgid weight, TW). Leaves were then oven-dried at 80 °C wild-type rice plants (~15) were heat-treated in the growth chamber, and seed- for 48 h and weighed (dry weight, DW). RWC was calculated with the follow- setting rates were measured and statistically analyzed with three biological ing formula: RWC = (FW – DW) /(TW – DW) × 100%. replicates, as described above. WUE was calculated by using a gravimetric method45. Four Arabidopsis A set of tomato germplasm including >100 modern varieties/inbreed seedlings of each line were grown in one container supplemented with 350 ± 5 g lines were screened for transcript abundance of the tomato ER orthologous soil (wet weight) with a total of 25 containers and three biological replicates. gene (SlER, accession no. LOC101266437) by qPCR with the SlER-RT3F/R After 1 week, two of the four plants were oven-dried at 65 °C for 1 week and primer set (Supplementary Table 2). The qPCR-based screen identified four weighed to obtain the initial dry weight (IDW). Then, the containers with the varieties (M82, Djena gold girl, Borgo Cellano and the sequenced reference other two plants were grown at 22 °C or 30 °C and covered with plastic wrap to variety Heinz 1706, ref. 41) with increased transcription levels of SlER in prevent evaporation from the soil surface. Each container was weighed before comparison with LA1589 used for generating transgenic tomato plants, and and after watering every 6 d to determine moisture loss, and the difference in two accessions (LA1580 and LA1781) accumulating similar levels of SlER as weight was corrected by adding water to the same weight. The total water loss LA1589. Seedlings of the six varieties together with LA1589 were heat-treated (TWL) was summed over a period of 3 weeks. In addition, the control water in the growth chamber, and survival rates (used for measurement of heat loss (CWL) of the same container without plants was measured to calculate tolerance) were calculated with 30 plants and three biological replicates, as the water loss by evaporation. At the end of the experiment, plants were described above. oven-dried and weighed to obtain the terminal dry weight (TDW). The con- tainers with dead plants during 30 °C growth were removed from calculation. Microscopy and histology. Scanning electron microscopy (SEM) and the The water use efficiency was calculated as increased dry weight divided preparation of resin-embedded thin sections were performed, as previously by cumulative water loss with the following formula: water use efficiency = described20. In brief, leaf tissues of 2-week-old Arabidopsis and 6-week-old (TDW – IDW)/(TWL – CWL). tomato plants with or without heat treatment were fixed in FAA (50% ethanol/ Ion leakage was measured as previously described46. In brief, 15 leaves with acetic acid/formaldehyde = 9:0.5:0.5). The samples were dehydrated in an three biological replicates of 2-week-old Arabidopsis plants grown at 22 °C ethanol series and then dried by supercritical fluid drying with CO2. The dried with heat treatment (40 °C ) for 0, 12, 24, 36 and 48 h from each line were

Nature America, Inc. All rights reserved. America, Inc. © 201 5 Nature samples were mounted on copper supports and sputter-coated with gold, and cut into two pieces from the midrib, and the leaf tissues were incubated in then observed under SEM (JEOL, JSM-6360LV). The stomatal density was ten tubes (three leaf pieces/tube) containing 3 ml deionized water overnight determined using SEM images of the abaxial surfaces of expended rosette with shaking at 28 °C. The initial ion leakage (ILi) was determined using a leaves, with at least five leaves per sample with three biological replicates, and conductivity meter (FE30, Mettler Toledo). The samples were boiled for 10 min npg the stomatal index was calculated from the total cells in the images of the leaf under high pressure (0.21 Mpa) and then cooled to room temperature to abaxial surfaces. Cell size (cell length and width) was statistically analyzed measure the total ion leakage (ILt). The relative ion leakage was expressed as by measuring 30 cells of the same-age leaves from each line with three a percentage of the total electrolytes. biological replicates. Arabidopsis plants were grown at 22 °C to mature, leaf and branch number, For observation by TEM, new mature leaves were cut into pieces, fixed in seed weight per plant were analyzed with 30 individual plants per line. Thirty 2.5% glutaraldehyde and then fixed in 1% osmic acid overnight at 4 °C. After plants of 4-week-old plants grown at 22 °C were measured for aboveground rinsing with phosphate buffer (pH 7.2) three times, the samples were dehy- biomass (shoot dry weight). Leaf sizes (in cm2) were determined for ten rosette drated in an ethanol series, treated with propylene oxide, and then infiltrated leaves of 4-week-old Arabidopsis plants grown at 22 °C or for three same-aged and embedded with Epon812 resin. The resin sections were stained with ura- full-expended leaves of 6-week-old tomato plants grown at 22 °C with ten nyl acetate and lead citrate and then observed under a transmission electron individual plants/line. microscope (CM120, Phillips). The disruption of cells including blebbing and broken plasma membrane, plasmolysis, and the collapse of chloroplasts and Analysis of gene expression and protein accumulation. Total RNAs were mitochondria were recorded and statistically analyzed by observing at least isolated from leaves or different tissues of Arabidopsis, rice and tomato plants 50 cells with three biological section replicates. using TRIzol reagent according to the manufacturer’s instructions (Invitrogen). Cell death was observed using trypan blue staining by boiling the leaves For ER, ERL1 and ERL2 developmental expression, rosette leaves of 2-week-old for 1 min in a lactophenol solution (lactic acid:phenol:glycerol = 1:1:1) plants, and rosette and cauline leaves and inflorescence organs of 6-week-old containing 100 µg/ml trypan blue, followed by tissue clearing in chlo- plants (flowering stage) grown at 22 °C were collected for RNA preparation. ral hydrate solution (2.5 g chloral hydrate dissolved in 1 ml H2O) over- For tomato and rice ER gene expression, RNA was prepared from leaves of night at room temperature and mounting in 70% glycerol for microscopic 4-week-old rice and tomato seedlings grown at 28 °C in greenhouses. Total 43 inspection . The accumulation of reactive oxygen species (H2O2) was visualized RNAs were also prepared from rosette levels of 2-week-old Col-0 plants by 3,3′-diaminobenzidine (DAB) staining as previously described38. treated by heat (40 °C) for a time course of 24 h for analysis of ER expression Leaves from plants grown at 22 °C immediately before moving to heat in response to heat stress. For qPCR analysis of gene expression, 2 µg of treatment were used as the heat treatment controls (0 h) in these total RNAs were converted into cDNA using the SuperScript III First-Strand microscopic experiments. cDNA Synthesis kit (Invitrogen), and 2 µl of the resulting cDNA was used as

doi:10.1038/nbt.3321 nature biotechnology a template for qPCR analysis using the primer sets listed in Supplementary 37. Jagadish, S.V. et al. Physiological and proteomic approaches to address heat tolerance Table 2. The qPCR reactions were performed in triplicate using a Rotor- during anthesis in rice (Oryza sativa L.). J. Exp. Bot. 61, 143–156 (2010). 38. Lister, C. & Dean, C. Recombinant inbred lines for mapping RFLP and phenotypic Gene 2000 fluorometric thermal cycler (Corbett Research) with SYBR markers in Arabidopsis thaliana. Plant J. 4, 745–750 (1993). Green real-time PCR master mix (Toyobo). For investigating the molecular 39. Clough, S.J. & Bent, A.F. Floral dip: a simplified method forAgrobacterium -mediated mechanism of the ERECTA-mediated thermotolerance, the marker genes, transformation of Arabidopsis thaliana. Plant J. 16, 735–743 (1998). HSFA1a and BI1, functioning in heat tolerance and cell death, respectively30,31, 40. Wang, E.T. et al. Control of rice grain-filling and yield by a gene with a potential signature of domestication. Nat. Genet. 40, 1370–1374 (2008). were analyzed for their expression in 2-week-old Arabidopsis plants during 41. The Tomato Genome Consortium. The tomato genome sequence provides insights heat treatment (40 °C). The relative expression levels were calculated into fleshy fruit evolution. Nature 485, 635–641 (2012). from three replicate qPCR experiments and normalized using the signal 42. Pennycooke, J.C. et al. The low temperature-responsive, Solanum CBF1 genes derived from ACTIN2 (At03g8780) in Arabidopsis, SleIF4a6 in tomato47 and maintain high identity in their upstream regions in a genomic environment undergoing gene duplications, deletions, and rearrangements. Plant Mol. Biol. 67, ACTIN1 in rice. 483–497 (2008). For immunoblotting detection of the ER-MH fusion protein, total proteins 43. Zhang, H., Zhang, X., Mao, B., Li, Q. & He, Z. Alpha-picolinic acid, a fungal toxin extracted from entire rosettes of 2-week-old homozygous T2/T3 plants were and mammal apoptosis-inducing agent, elicits hypersensitive-like response and enhances disease resistance in rice. Cell Res. 14, 27–33 (2004). separated by 7.5% SDS-PAGE (for ER-MH) or 12% SDS-PAGE (for Actin) and 44. Parida, A.K., Dasgaonkar, V.S., Phalak, M.S., Umalkar, G.V. & Aurangabadkar, L.P. analyzed by immunoblotting using anti-myc antibody (Millipore, #05-724) or Alterations in photosynthetic pigments, protein, and osmotic components in cotton anti-Actin antibody (ABclonal Biotechnology, #AC009). genotypes subjected to short-term drought stress followed by recovery. Plant Biotechnol. Rep. 1, 37–48 (2007). 45. Karaba, A. et al. Improvement of water use efficiency in rice by expression of Statistical analysis. Statistical analysis was performed by Student’s t-test fol- HARDY, an Arabidopsis drought and salt tolerance gene. Proc. Natl. Acad. Sci. USA lowed by the most conserved Bonferroni correction that adjusts alpha level 104, 15270–15275 (2007). based on comparisons for testing multiple hypotheses. We also used Welch’s 46. Charng, Y.-Y. et al. A heat-inducible transcription factor, HsfA2, is required for extension ANOVA followed by Bonferroni correction for the SlER gene expression of acquired thermotolerance in Arabidopsis. Plant Physiol. 143, 251–262 (2007). 47. Weng, L. et al. The zinc finger transcription factor SlZFP2negatively regulates and survival rate analysis of the seven tomato accessions (varieties) and the abscisic acid biosynthesis and fruit ripening in tomato. Plant Physiol. 167, 931–949 Arabidopsis CSSL lines. (2015). Nature America, Inc. All rights reserved. America, Inc. © 201 5 Nature npg

nature biotechnology doi:10.1038/nbt.3321