Changes in light intensity reveal a major role for carbon balance in Arabidopsis responses to high temperature


D. Vile. Fax: +33467522116; e-mail:


High temperature (HT) is a major limiting factor for plant productivity. Because some responses to HT, notably hyponasty, resemble those encountered in low light (LL), we hypothesized that plant responses to HT are under the control of carbon balance. We analysed the interactive effects of HT and irradiance level on hyponasty and a set of traits related to plant growth in natural accessions of Arabidopsis thaliana and mutants affected in heat dissipation through transpiration (NCED6-OE, ost2) and starch metabolism (pgm). HT induced hyponasty, reduced plant growth and modified leaf structure. LL worsened the effects of HT, while increasing light restored trait values close to levels observed at control temperature. Leaf temperature per se did not play a major role in the observed responses. By contrast, a major role of carbon balance was supported by hyponastic growth of pgm, as well as morphological, physiological (photosynthesis, sugar and starch contents) and transcriptional data. Carbon balance could be a common sensor of HT and LL, leading to responses specific of the shade avoidance syndrome. Hyponasty and associated changes in plant traits could be key traits conditioning plant performance under competition for light, particularly in warm environments.


High temperature (HT) is among the most damaging factors for plant productivity (Jones 1992). For most plant species, even a moderate increase in temperature leads to significant changes in leaf structure and morphology (Poorter et al. 2009). HT affects central processes such as photosynthesis, leaf expansion, germination, buds and flower abortion or cell division (Berry & Björkman 1980; Penfield 2008). Indeed, the rates of numerous plant processes increase with temperature up to an optimum above which dramatic physiological and developmental changes occur, leading to a rapid decrease of these rates (Ong 1983; Gillooly et al. 2001; Granier et al. 2002).

HT could affect plant carbon balance because carbohydrate demand increases while its supply decreases: rates of physiological processes increase while photosynthetic yield decreases (Berry & Björkman 1980; Kobza & Edwards 1987). Accordingly, tolerance to warm temperatures is increased at high CO2 concentration in C3 plants (Huxman et al. 1998; Taub, Seemann & Coleman 2000). Furthermore, the allocation of carbohydrates into costly processes, such as the biosynthesis of protection proteins (notably heat shock proteins; Heckathorn et al. 1996), raises with increasing temperature. In line with this, Heckathorn et al. (1996) reported that plant tolerance to heat stress is decreased at low nitrogen supply because of a limited production of nitrogen-costly heat shock proteins. Because photosynthesis is a major driver of plant carbon balance, light availability should also be taken into account to investigate plant responses to HT.

Because leaf orientation directly determines light interception, it has been proposed that leaf phototropism could be part of plant responses to temperature. For instance, in several species, changes in leaf angle avoid blade overheating when light intensity is maximal (Fu & Ehleringer 1989; King 1997; Falster & Westoby 2003). In Arabidopsis thaliana, as well as in other species, hyponasty (i.e. upward leaf movements) (Kang 1979) is one of the first morphological responses to HT (Koini et al. 2009). Hyponastic response varies widely among natural accessions of Arabidopsis and is related to the daily temperature variation encountered in the collection sites, suggesting an adaptive role for this trait (Van Zanten et al. 2009). Recently, Franklin (2010) also reported that Arabidopsis rosettes displaying hyponastic growth have a higher transpiration rate. HT-induced hyponasty could therefore contribute to optimize leaf cooling by increasing transpiration (possibly through an increase of boundary layer conductance), reducing the stress caused by excess irradiance or by repositioning the photosynthesizing tissues away from the heated soil (Gray et al. 1998).

Hyponasty is also a typical response to low light (LL) intensity and to decreased red to far-red ratio (Hangarter 1997; Maliakal et al. 1999; Smith 2000), occurring under the control of the phytochrome and cryptochrome pathways (Smalle et al. 1997; Vandenbussche et al. 2003; Kozuka et al. 2005; Millenaar et al. 2009). Hyponasty has therefore been proposed to be a morphological response typical of the shade avoidance syndrome, allowing plants to reach more light and maximize carbon gain as in the case of competition for light under a canopy (Pierik et al. 2004; Mullen, Weinig & Hangarter 2006; Van Zanten et al. 2010a). Van Zanten et al. (2009) showed that HT-induced and LL-induced hyponasty display very similar responses in terms of kinetics and amplitude. Interestingly, LL-induced hyponasty was also observed in multiple loss-of-function photoreceptor mutants (Millenaar et al. 2009). Taken together, these results suggest a tight link between hyponasty and carbon balance.

Here, we aimed at deciphering the role of carbon balance into plant responses to HT. To this end, we investigated to what extent Arabidopsis responses to HT are driven by light intensity. Specifically, we tested whether hyponasty induced by HT prevents elevation of leaf temperature or is an anticipated response against carbon depletion. For this purpose, the responses of A. thaliana accessions and mutants affected in carbon balance and regulation of leaf temperature via transpiration were studied at three light intensities under prolonged elevated temperature. We chose the physiologically relevant HT of 30 °C, which is known to affect growth and hyponasty in Arabidopsis (Van Zanten et al. 2009). This temperature was unlikely to cause mortality because it has been described as the temperature of basal thermotolerance of the Col-0 reference accession (Ludwig-Muller, Krishna & Forreiter 2000). In parallel with leaf temperature, leaf carbon status was investigated through sugars and starch contents, CO2 exchanges, chlorophyll fluorescence and a transcriptional analysis of targeted genes. Finally, using a multilevel analysis of plant traits, we highlighted that changes in growth and development induced by HT are tightly related to changes in carbon status.


Plant material

Four A. thaliana (L.) Heynh accessions (NASC numbers in brackets) were chosen for their variability in the phenotypic responses to growth conditions (Millenaar et al. 2005; Tonsor et al. 2008): Col-0 (N1092), An-1 (N944), Cvi-0 (N902) and Ler (NW20). Ler carries a mutation at ERECTA that affects multiple plant traits also affected by HT and light (Masle, Gilmore & Farquhar 2005; Tisnéet al. 2010). This gene is also involved in the control of ethylene-induced hyponasty (Van Zanten et al. 2010b). Therefore, LER, a complemented accession homozygous for Col-0 allele at ERECTA was included in our analysis (Torii et al. 1996). A starch synthesis-deficient mutant pgm (Caspar, Huber & Somerville 1985), and two mutants affected in stomata opening: an ‘open stomata’, ost2 (Merlot et al. 2002) and a ‘closed stomata’, NCED6-OE (Lefebvre et al. 2006), all in Col-0 background, completed this selection.

Growth conditions and treatments

Five seeds of each genotype were sown in 225 mL pots filled with a damped mixture (1:1, v/v) of loamy soil and organic compost (Neuhaus N2), and placed at 4 °C in the dark for 48 h. After germination, plants were grown in a chamber at 20 °C air temperature and 12/12 h photoperiod under a photosynthetic active radiation (PAR) of 175 µmol m−2 s−1 supplied from a bank of HQi lamps until the emergence of the two first leaves (stage 1.02 in Boyes et al. 2001) (Supporting Information Table S1). The pots were moved daily to avoid boundary effects. From stage 1.02 onwards, air temperature was set to 20/17 °C day/night in a first growth chamber and to 30/25 °C in three others. PAR was maintained at 175 µmol m−2 s−1 until the sixth leaf emergence (stage 1.06; Supporting Information Table S3). Then, HT-treated plants were grown under LL (70 µmol m−2 s−1), moderate (ML, 175 µmol m−2 s−1) and high (HL) (330 µmol m−2 s−1) light. In each chamber, vapour pressure deficit (VPD) was maintained at 0.6/0.4 kPa during day/night. Each pot was weighed daily and watered with a one-tenth-strength Hoagland's solution (Hoagland & Arnon 1950) to maintain soil water content at a well-watered level of 0.35 g H2O g−1 dry soil equivalent to a pre-dawn water potential of −0.3 MPa (Granier et al. 2006; Hummel et al. 2010). Six to eight plants were harvested at first silique emergence (stage 6.02).

Measurement of plant traits

Whole plant and leaf traits

Total length, blade length and tip height of the youngest fully expanded leaf were determined three times per week in all genotypes during the 2 weeks following light treatments. Measurements were performed at different times of the day on randomly selected plants to avoid the effects of weak changes occurring along the day caused by the endogenous rhythms (Mullen et al. 2006), and during the vegetative stage to avoid the effects of drastic changes in carbon status caused by floral transition (Christophe et al. 2008). Blade ratio was calculated as blade length to total leaf length. Leaf insertion angle (degree) was calculated as θ = arcsine(leaf tip height/leaf length) × 180/π. Mean values of leaf angle and blade ratio were calculated for the 2 weeks period and used in further analyses.

Plants were harvested shortly after flowering when the first silique emerged (stage 6.01; from 35 to 100 d after sowing). Rosettes were cut and immediately weighed (FW, mg) after the removal of inflorescences. The rosettes were wrapped in moist paper and placed into Petri dishes at 4 °C in darkness overnight to achieve complete rehydration. Water-saturated fresh weight (SW) was then determined. Total leaf number (LN) was determined, and leaf blades were separated from their petiole and scanned for area measurements before being oven-dried at 65 °C for 48 h to determine their dry weight (DW). Rosette area (RA, cm2) was determined as the sum of individual leaf blade areas measured with an image analyser (BioScan-Optimas 4.10, Edmond, WA, USA). Relative water content [RWC = (FW − DW)/(SW − DW), %], leaf dry matter content (DW/FW, mg g−1) and specific leaf area (RA/DW, cm2 g−1) were calculated at the rosette level. Mean leaf thickness (µm) was estimated as SWblade/RA (Vile et al. 2005). Epidermal imprints of the sixth leaf were placed under a microscope (Leitz DM RB; Leica, Wetzlar, Germany) coupled to an image analyser. Mean cell density and stomatal density were determined in two 0.12 mm2 zones. Stomatal index was calculated as 100 × stomatal number/(stomatal number + stomatal number × 2 + cell number).

Transpiration and leaf temperature

Transpirational water loss was determined on five to eight plants of Col-0, Ler, NCED6-OE and ost2 at bolting by successive weighing of the pots over 3 d and nights. Soil evaporation was prevented by sealing soil surface with four layers of a plastic film. Whole-plant transpiration rate (mg H2O h−1) was estimated as the slope of the linear relationship between weight and time, and then expressed per dissected RA (mg H2O h−1 cm−2). Leaf temperature (°C) was determined at two points of four to six rosettes by infrared imaging (ThermaCAM B20HSV; FLIR Systems, Wilsonville, OR, USA).

Net photosynthesis, respiration and dark-adapted chlorophyll fluorescence

Rate of CO2 assimilation was measured on four Col-0 and Ler plants at bolting (between stages 3.90 and 5.01 of Boyes et al. 2001; c. 15 d after the beginning of light treatments) using a single leaf chamber designed for Arabidopsis connected to an infrared gas analyser system (CIRAS 2; PP Systems, Amesbury, MA, USA). Dark respiration and chlorophyll fluorescence were measured using a fluorescence module supplied by the manufacturer on plants dark adapted for at least 20 min and submitted to a saturating light flash to estimate photosystem II (PSII) yield capacity as Fv/Fm, where Fv is the difference between the maximum (Fm) and the minimum fluorescence signals (Maxwell & Johnson 2000). Carbon fluxes were determined at steady state (approximately 15 min after light was switched on or off) under 390 ppm CO2.

Sugars and starch contents

Four samples containing two to four rosettes of Col-0, Ler or pgm were harvested 3 d after the beginning of light treatments at the end of the day or night, and immediately frozen in liquid nitrogen. Starch and soluble sugars (as the sum of glucose, fructose and sucrose) contents were analysed by enzymatic assay as in Hummel et al. (2010).

Gene expression

Three hyponastically expanding leaves of Col-0 and Ler grown at HT were harvested on four plants (stage 1.06) and immediately frozen in liquid nitrogen. A first harvest was performed in the middle of the morning, before any light treatment (t0). Two subsequent harvests were performed 1 h (t1) and 24 h (t24) after light treatments. Another harvest was performed 3 d later at the end of the daytime or night-time. RNA was isolated using NucleoSpin RNA Plant (Macherey-Nagel, Düren, Germany). Reverse transcription and amplification of cDNA were performed as described in Supporting Information Table S4. Real-time quantification of target cDNA was performed in a LightCycler 480 (Roche, Penzberg, Germany) using specific primers (Supporting Information Table S4). Cycle threshold (Ct) values were determined by the fit point method. PCR efficiency (E) was deduced from a standard dilution series as E = −1/slope. Relative quantification was determined using the delta delta Ct method with E correction. Two reference genes (CIPK23, At1g30270; TUB4, At5g44340.1) were selected for normalization on the basis of their expression stability. Finally, all expression values at t1 and t24 were normalized by the gene expression at t0 (before any light treatment).

Statistical analyses

Genotype, temperature and light effects on traits were analysed in analyses of variance (anovas) and Kruskall–Wallis tests for multiple comparisons. Gene expression was analysed in a hierarchical clustering analysis using Euclidean distances after log-transformation and plotted as a heat map. A principal component analysis (PCA) was performed to study the relationships between traits, genotypes and environments. All statistical tests were performed using R 2.10 (R Development Core Team 2009).


HT-induced hyponasty is modulated by light intensity

A strong HT-induced hyponasty (i.e. an increase in leaf insertion angle) was observed in all accessions and the complemented line LER (Fig. 1). At the same light level (175 µmol m−2 s−1), leaf angle was more than doubled at 30 °C compared to 20 °C, and varied significantly between genotypes (P < 0.001; Table 1) from 2.9-fold in Col-0 to 3.5-fold in Cvi-0. Remarkably, the HT-induced hyponasty was significantly increased under LL (70 µmol m−2 s−1) in all genotypes, with Col-0 showing the highest response. Conversely, a significant decrease in HT-induced hyponasty was found under HL (330 µmol m−2 s−1). On the other hand, the blade ratio tended to decrease in response to HT, particularly at LL (Table 1; Supporting Information Fig. S1f). These results clearly show that HT-induced hyponastic growth and the proportion of leaf blade are modulated by light levels.

Figure 1.

Hyponastic growth response to high temperature (HT) and light intensity of four Arabidopsis accessions and the complemented line at ERECTA (LER). Leaf angle is the average of six values measured within 2 weeks after the beginning of light treatments on plants grown at 20 °C under moderate light (ML) intensity (175 µmol m−2 s−1; white bars), and at HT (30 °C) under low (LL) (70 µmol m−2 s−1; black bars), ML (175 µmol m−2 s−1; dark grey bars) and high light (HL) intensity (330 µmol m−2 s−1; light grey bars). Bars are means ± SE (n = 6–10). Letters indicate significant differences following Kruskal–Wallis non-parametric test (P < 0.05).

Table 1.  Results of univariate analyses of variance (anovas)
TraitModerate light (ML) intensity (temperature effect)High temperature (HT) (light effect)Light and temperature treatments merged
  • Numbers are mean squares (type III) from the partitioning of phenotypic variation among five genotypes (G) including four accessions (An-1, Col-0, Cvi-0, Ler) and the complemented LER line grown at 20 °C under ML intensity (175 µmol m−2 s−1), and at HT (30 °C) under low (LL) (70 µmol m−2 s−1), ML (175 µmol m−2 s−1) and high light (HL) intensity (330 µmol m−2 s−1). Hypothesis testing was based on F-ratios from type III mean squares for all anovas.

  • Mean squares in bold typeface followed by a symbol were significant at ¤, P < 0.10; *, P < 0.05; **, P < 0.01; ***, P < 0.001. Mean squares reported are those from the full model including all interactions. All non-significant terms (normal typeface) are reported, but were removed from the final model.

  • a

    ln-transformed variable.

Leaf insertion angle1287***13 517***172.0***0.95***1202***6217***159.8***0.79***1843***12 132***142.5***0.90***
Leaf blade ratio323.7***363.8***43.99***0.72***495.8***1469***62.96***0.85***705.2***1398***58.18***0.84***
Vegetative duration1134***82.3687.67¤0.59***2304***1448***196.3***0.71***2398***1247***166.2***0.73***
Leaf number (LN) at flowering674.0***211.2***68.74***0.67***475.7***255.3***37.69**0.63***785.1***322.5***44.31***0.69***
Total fresh massa3.532***1.759***0.148**0.78***3.884***2.648***0.196***0.78***4.839***3.039***0.159***0.81***
Specific leaf area3630***5665***736***0.70***6434***41 516***193.00.82***6698***33 360***462***0.84***
Leaf dry matter content2.751***4.061***1.023*0.35***3.177***26.45***0.2920.56***3.550***19.43***0.5360.56***
Leaf thickness25 176***43 302***28 190***0.57***23 064***42 647***3170**0.58***27.52***59 676***4265***0.64***
Cell densitya1.307***0.320**0.496***0.59***1.496***1.435***0.084¤0.65***1.528***1.008***0.225***0.66***
Stomatal densitya1.471***3.739***0.589***0.72***2.483***0.430***0.123*0.67***2.255***1.286***0.298***0.70***
Stomatal index4.220***131.1***9.476**0.42***17.44***38.68***4.1270.37***15.05***106.6***6.026**0.57***

Hyponasty does not coincide with leaf temperature

At HT, leaf temperature of Col-0 and Ler was higher under HL and lower under LL compared to ML (Fig. 2a). Accordingly, after 1 h exposure to HL or LL, a strong transcriptional induction or repression, respectively, of the heat stress marker gene heat shock protein 101 (HSP101) was found in both accessions (Fig. 2b). This response was maintained after 24 h exposure to the light treatments. Therefore, exposure to HL induced a higher leaf temperature that superimposed with that of elevated air temperature, but did not coincide with higher leaf angle.

Figure 2.

Leaf temperature sensing in response to temperature and light. (a) Surface temperature of hyponastic leaves measured by infrared imaging in Col-0 and Ler accessions, and two mutants impaired in stomata opening (NCED6-OE) and closing (ost2). Plants were grown at 20 °C under moderate light (ML ) intensity (175 µmol m−2 s−1; white bars), and at high temperature (HT) (30 °C) under low (LL) (70 µmol m−2 s−1; black bars), ML (175 µmol m−2 s−1; dark grey bars) and high light (HL) intensity (330 µmol m−2 s−1; light grey bars). Bars are means ± SE (n = 5–16). Letters indicate significant differences following Kruskal–Wallis non-parametric test (P < 0.05). (b) Expression of HSP101 at HT at 0, 1 and 24 h after light treatment. Plants were grown until the emergence of leaf 6 at HT (30 °C) under ML, and then transferred under LL, HL or left under ML. Hyponastic leaves were harvested 1 h (t1) and 24 h (t24) after light treatment. Each expression was normalized according to t0 (i.e. before transfer under LL or HL).

To further rule out the hypothesis that leaf temperature solely determines the hyponastic response, we analysed two mutants impaired in stomata opening and closing. As expected, the open (ost2) and closed (NCED6-OE) stomata mutants were, respectively, cooler and warmer compared to the wild-type Col-0 in the control irradiance level (Fig. 2a). Those differences were related to differences in transpiration rates significantly higher in ost2 and lower in NCED6-OE (Fig. 3). In all genotypes, HT and HL induced a higher transpiration compared to control temperature (Fig. 3). Interestingly, in Col-0 and Ler, this trend held true during day and night, despite the absence of heat gain from irradiance. However, this increase in latent heat dissipation through transpiration under HL was not yet sufficient to counterbalance the conjugated effects on leaf temperature of lower leaf angle and higher heat gain caused by irradiance. Finally, leaf temperature did not positively correlate with leaf angle, neither within a given environment nor within a given genotype. Indeed, we observed that the hyponastic response of NCED6-OE and ost2 did not differ significantly from Col-0 (P > 0.54, Fig. 4).

Figure 3.

Night and day transpiration rates of Col-0 and Ler accessions, and of two mutants impaired in stomata opening (NCED6-OE) and closing (ost2). Transpiration was determined gravimetrically in plants at bolting stage grown at 20 °C under moderate light (ML) intensity (175 µmol m−2 s−1; white bars), and at high temperature (HT) (30 °C) under low (LL) (70 µmol m−2 s−1; black bars), ML (175 µmol m−2 s−1; dark grey bars) and high light (HL) intensity (330 µmol m−2 s−1; light grey bars). Bars are means ± SE (n = 5–10). Letters indicate significant differences (P < 0.05) following Kruskal–Wallis non-parametric tests independently performed for night and day.

Figure 4.

Leaf insertion angle of the wild-type Col-0 and mutants impaired in stomata opening (NCED6-OE) and closing (ost2), and in starch synthesis (pgm). Leaf angle is the average of six values measured within 2 weeks after the beginning of light treatments on plants grown at 20 °C under moderate light (ML) intensity (175 µmol m−2 s−1; white bars), and at high temperature (HT) (30 °C) under low (LL) (70 µmol m−2 s−1; black bars), ML (175 µmol m−2 s−1; dark grey bars) and high light (HL) intensity (330 µmol m−2 s−1; light grey bars). Bars are means ± SE (n = 6–10). Letters indicate significant differences following Kruskal–Wallis non-parametric test (P < 0.05).

Hyponasty interplays with starch metabolism

Because leaf temperature was not the primary determinant of hyponasty and that LL worsened hyponasty, we investigated the involvement of carbon balance using a genetic manipulation. The pgm mutant, strongly impaired in starch synthesis (Caspar et al. 1985), showed a very different response compared to the wild-type Col-0. This mutant not only displayed a significantly steeper leaf angle than Col-0 under HT (Fig. 4; Table 1), but light level had also no significant effect on its HT-induced hyponastic response. Consistent with previous studies (Gibon et al. 2004a), the starchless (Fig. 5c,d) pgm mutant displayed a significantly higher sugar concentration at the end of the day compared to Col-0 whatever the growth condition (Fig. 5a). Sugar contents at the end of the night were similarly low between pgm and Col-0, although significantly different under ML (Fig. 5b). The tight link between carbon metabolism and HT-induced hyponasty evidenced by the pgm mutant prompted us to further investigate the involvement of carbon balance in plant responses to HT.

Figure 5.

Sugar and starch contents of the Col-0 and Ler accessions, and the starch-deficient mutant pgm in response to temperature and light. Plants were grown at 20 °C under moderate light (ML) intensity (175 µmol m−2 s−1; white bars), and at high temperature (HT) (30 °C) under low (LL) (70 µmol m−2 s−1; black bars), ML (175 µmol m−2 s−1; dark grey bars) and high light (HL) intensity (330 µmol m−2 s−1; light grey bars). Sugar concentration (a, b) and starch content (c, d) were determined 3 d after the beginning of light treatments, at the end of the day and at the end of the night, respectively. Sugar concentration is the sum of sucrose, glucose and fructose. Sugar and starch contents are expressed in C6 equivalents. Bars are means ± SE (n = 4). Letters indicate significant differences following Kruskal–Wallis non-parametric test (P < 0.05). No detectible levels of starch were found in pgm.

Light modulates the deleterious effects of HT on carbon status

Because increasing incident light intensity reverted HT-induced hyponasty and pgm mutant had altered responses, we hypothesized that leaf carbon status could be a good candidate to unify both LL- and HT-induced responses. Changes in carbon assimilation and status induced by variations in temperature and light were thus investigated. Overall, plants under HT accumulated less carbohydrates during the day and were more carbon depleted at the end of the night, while HL restored the contents encountered at the control temperature (Fig. 5). Although LL did not affect significantly sugar concentration under HT, starch content was significantly lower.

The net photosynthetic rate was significantly reduced by HT in Col-0, but not in Ler (Fig. 6a), while dark respiration was not significantly affected by HT in both accessions. Not surprisingly, net photosynthesis increased with light intensity in both accessions. PSII yield capacity, as evaluated by chlorophyll fluorescence (Fv/Fm), was reduced at HT in both accessions, whereas increasing light intensity led to a recovery of Fv/Fm levels close to those encountered in control conditions in Col-0 (Fig. 6b). Although increasing light caused a slight increase in leaf temperature, a shift from 70 to 330 µmol m−2 s−1 PAR was sufficient to balance and even abolish the negative effects of HT on net carbon assimilation and PSII quantum efficiency.

Figure 6.

Photosynthetic performance of Col-0 and Ler accessions acclimated to contrasted temperature and light conditions. (a) Net photosynthetic rate and dark respiration. (b) Photosystem II (PSII) yield capacity estimated by chlorophyll fluorescence. CO2 fluxes and chlorophyll fluorescence after dark adaptation were measured in ambient conditions at 20 °C under moderate light (ML) intensity (175 µmol m−2 s−1; white bars), and at high temperature (HT) (30 °C) under low (LL) (70 µmol m−2 s−1; black bars), ML (175 µmol m−2 s−1; dark grey bars) and high light (HL) intensity (330 µmol m−2 s−1; light grey bars). Bars are means ± SE (n = 7–11 for CO2 flues, n = 3–5 for fluorescence). Letters indicate significant differences following Kruskal–Wallis non-parametric test (P < 0.05).

The carbon status of the plants under HT was also investigated through the expression of specific marker genes (Blasing et al. 2005). DIN10 and DIN6 (Fujiki et al. 2001), and TPS8 (Hummel et al. 2010) were selected as markers of carbon limitation, while CPN60A and CPN60B (Hummel et al. 2010) were used as markers of high carbon supply. Dynamics of relative transcript abundance of each gene were compared to the levels encountered at the end of the day or night. These latter stages have been well described as bringing the rosette to a high and low sugar status, respectively (Gibon et al. 2004b).

As indicated by the proximity of Col-0 and Ler in the clustering, the genotypic effect on the expression of all genes was negligible under all environmental conditions (Fig. 7). As expected, DIN10, DIN6 and TPS8 were enhanced, and CPN60A and CPN60B were repressed at the end of the night when carbon is limiting, while the opposite trend was true at the end of the day (Fig. 7). Therefore, these genes can reasonably be used as indicators of leaf carbon status. Within 1 h following changes in light conditions, the expression of genes indicative of high carbon supply was clearly repressed under LL and enhanced under HL. On the other hand, the expression of genes indicative of carbon limitation was strongly enhanced 1 h after exposure to LL and after 24 h, yet to a lesser extent. The reverse was true under HL. Overall, transcript levels under LL mimicked those encountered at the end of the night, whereas under HL these levels resembled those encountered at the end of the day.

Figure 7.

Expression of marker genes of carbon starvation of Col-0 and Ler accessions grown at high temperature (HT) under contrasting light intensities. Overlay heat map showing expression levels of genes induced (DIN6, DIN10, TPS8) or repressed (CPN60A, CPN60B) by carbon starvation in plants grown until emergence of leaf 6 at HT (30 °C) under moderate light (ML) intensity (175 µmol m−2 s−1), and then transferred under low light (LL) intensity (70 µmol m−2 s−1; black boxes of the left side bar), high light (HL) intensity (330 µmol m−2 s−1; light grey boxes of the left side bar) or kept under ML (dark grey boxes of the left side bar). Hyponastic leaves were harvested 1 h (t1) and 24 h (t24) after light treatment. Two days after the beginning of light treatments, plants were harvested at the end of the day and night. Each box represents the effect of change in light intensity on relative expression level of each marker gene identified on the right side. Level values are log-transformed values. The dendrogram represents the proximity between each gene expression pattern, as calculated from a hierarchical clustering analysis using Euclidean distances after log-transformation.

In summary, results at the metabolic, photosynthetic and transcriptional levels converge to indicate that carbon status is significantly impaired under HT, but can be improved by increasing light intensity, while reducing light intensity leads to a worsened carbon balance.

Interaction between HT and light on growth: a multi-scale analysis of plant traits

Our data clearly indicated that HT and light interact in the regulation of leaf hyponasty and carbon status. We therefore extended our analysis to other growth-related traits. A PCA was performed on morphological and anatomical traits from the cellular to the leaf and whole-plant levels measured in the four accessions Col-0, Cvi-0, An-1 and Ler, and the complemented line LER (Table 1; Fig. 8; Supporting Information Fig. S1). The complemented line LER was included in the analysis because no detectible effect of ERECTA was found to modify the interpretation of the results. For instance, no significant difference in the hyponastic response to both temperature and light was found between Ler and LER (Fig. 1a; Supporting Information Table S2; Supporting Information Fig. S1f). However, Ler was characterized by high epidermal cell density, which was significantly decreased in LER (Supporting Information Fig. S1g), as expected (Masle et al. 2005; Tisnéet al. 2010). LER also exhibited a marginally significant weaker leaf angle than Ler at 20 °C, and a lower leaf blade ratio whatever the environmental condition (Supporting Information Table S2 and Supporting Information Fig. S1f).

Figure 8.

Principal component analysis (PCA) on multiple plant traits measured on Arabidopsis accessions in contrasted temperature and light treatments. The first two axes are shown which account for 70% of the total inertia. (a) Projection of the variables. (b) Projection of individual plants (grey symbols) and centres of gravity for each treatment and each accession (An-1, circles; Col-0, diamonds; Cvi-0, squares; Ler, triangles; LER, upside-down triangles). Plants were grown at 20 °C under moderate light (ML) intensity (175 µmol m−2 s−1; white symbols), and at high temperature (HT) (30 °C) under low (LL) (70 µmol m−2 s−1; black symbols), ML (175 µmol m−2 s−1; dark grey symbols) and high light (HL) intensity (330 µmol m−2 s−1; light grey symbols). LDMC, leaf dry matter content; SLA, specific leaf area.

The first and second principal components (PCs) explained 53 and 17% of the total variance, respectively. PC1 was positively correlated with leaf angle and specific leaf area, and negatively correlated with total fresh weight, LN, leaf thickness, leaf dry matter content, stomatal index and blade ratio (Fig. 8a; Supporting Information Table S2 for loadings). PC2 was mainly explained by vegetative stage duration. Epidermal cell density was poorly represented on the first two PCs, but explained the main proportion of PC3.

Projection of individuals revealed significant effects of temperature and light (P < 0.001, anova on PC coordinates; Fig. 8b). More interestingly, PC1 discriminated the individuals in a consistent way according to the environment, with a strong effect of HT under LL and a progressive recovery with increasing light intensity (Fig. 8b). Not only this gradient was represented by the hyponastic response as previously characterized, but it was also explained by an increase in specific leaf area and a decrease in plant fresh mass, leaf thickness, leaf dry matter content and stomatal index. HT caused the production of thinner leaves, but increasing light intensity allowed plants to re-allocate assimilates into thicker and denser leaves. By contrast, reducing light intensity amplified the effects of HT observed on leaf structure. HT significantly reduced plant size, but increasing light intensity resulted in larger plants (Supporting Information Fig. S1). The same trend held true for the other traits on PC1. Within the groups discriminated by the temperature and light treatments, individuals were mainly separated by vegetative stage duration on PC2 and a composite axis represented by rosette fresh weight and cell density, but to a lesser extent. This discrimination was driven by a significant genotype effect (Table 1). For instance, An-1 had significantly smaller rosette and shorter vegetative duration than Col-0 or Cvi-0. Despite some differences in plant size, very similar responses to the treatments were found in the ‘open’ (ost2) and the ‘closed stomata’ (NCED6-OE) mutants compared to the wild-type Col-0. In addition to its contrasted hyponastic response, pgm was significantly smaller than Col-0 and displayed a clear delay in flowering.

Overall, our results show that increasing light intensity under HT not only restores leaf angle close to levels encountered under control temperature, but also restores many other traits related to leaf structure, plant growth and development.


HT and LL-induced hyponasty: does the same consequence arise from the same cause?

An HT of 30 °C induced hyponastic growth in all Arabidopsis accessions we investigated here. This response was significantly increased under LL (70 µmol m−2 s−1), which is consistent with previous findings at 38 °C and light intensity <20 µmol m−2 s−1 (Millenaar et al. 2005; Van Zanten et al. 2009). Remarkably, we found that HL reversed the effects of HT on hyponasty, leading to leaf angle values similar to those encountered under control temperature (20 °C).

Different hypotheses may explain the interacting effects of HT and light on leaf angle. HT-induced hyponasty could be triggered by leaf temperature itself, contributing to leaf cooling by: (1) decreasing incoming radiant heat (Fu & Ehleringer 1991; Falster & Westoby 2003); (2) decreasing conductive and radiative heat transfer by moving the leaf away from the heated soil as suggested by Gray et al. (1998) for hypocotyl elongation; and (3) increasing transpiration through an increased boundary layer conductance. Here, the HT-induced hyponastic responses of two mutants impaired in stomata closure (ost2) and opening (NCED6-OE) were not different from that of the wild type, although these mutants had, respectively, cooler and warmer leaves because of differential transpiration (Merlot et al. 2002; Fig. 3; Lefebvre et al. 2006). Furthermore, under HL, leaves were warmed by 1.1 °C despite a higher transpiration rate and lower insertion angles than leaves under low and moderate irradiance. Leaf warming was confirmed by the induction of HSP101, which acts as a virtual thermometer (Young et al. 2001). If leaf temperature was the only trigger of HT-induced hyponasty, increasing light would have led to increased hyponasty. Our results clearly rule out this assumption pointing towards other possible roles of light in hyponasty.

Several studies reported a role for photoreceptors in hyponasty under LL (Somers et al. 1991; Robson, Whitelam & Smith 1993; Morelli & Ruberti 2002; Vandenbussche et al. 2005; Mullen et al. 2006; Millenaar et al. 2009) and HT (Koini et al. 2009; Van Zanten et al. 2009). However, while hyponasty is delayed in photoreceptor mutants during the first hours following HT or LL treatments, a response similar to the wild type was observed afterwards (Van Zanten et al. 2009). Millenaar et al. (2009) also found that a prolonged exposure to LL led to hyponastic growth induction even in multiple loss-of-function photoreceptor mutants. Here, leaf angles were measured during 2 weeks after the beginning of light treatments (i.e. after the recovery period of hyponasty in the photoreceptor mutants described in Van Zanten et al. 2009); therefore, excluding a major role for photoreceptors in the patterns observed.

Sugars act both as signal and carbon supply for several plant processes, including differential petiole-to-blade leaf growth (Kozuka et al. 2005). Previous studies have also shown that regulators of starch metabolism or derived signals are integrators of plant metabolism and growth (Sulpice et al. 2009). Here, we found that changes in leaf inclination following changes in light level fitted in a consistent way with leaf carbon status, as measured by sugar and starch contents. Specifically, starch content and leaf angle were negatively correlated along the environmental conditions (Supporting Information Fig. S3). Moreover, constitutive HT-induced hyponasty was found in pgm, a starch-deficient mutant whose diurnal physiological state resembles that of a wild-type plant exposed to an extended night (Gibon et al. 2004a). Provided that HT and LL induce carbon starvation, this could explain why a prolonged exposure of the wild type to these environments led to a similar response than that of pgm. Millenaar et al. (2009) also reported that a pharmacological inhibition of the photosynthetic electron transport chain induced hyponasty under non-inducing light intensity and control temperature. Expression patterns of genes related to carbon status were in agreement with a possible role of carbon status on hyponasty. They also revealed that plants sense a carbon-limiting environment largely before they really experience carbon depletion. In Arabidopsis, diurnal changes in leaf angle are negligible regarding to the changes in leaf angle (Mullen et al. 2006) caused by the environment, but they follow the diurnal pattern of carbohydrate availability and starch content that is tightly linked to the circadian clock (Blasing et al. 2005; Graf et al. 2010). Indeed, leaf inclination is enhanced at night, when carbon supply relies on starch, and reduced at dawn when photosynthesis resumes. Overall, changes in carbon metabolism could be a common signal of both LL- and HT-induced hyponasty, with sugars or starch degradation products acting either as a primary signal or in a parallel pathway following exposure to unfavourable growth conditions.

The possible role of carbon status in hyponasty does not preclude a molecular crosstalk with hormonal and photocontrol regulation. Here, no significant changes in transcript levels of genes related to ethylene biosynthesis or signalling were found (data not shown), in agreement with Millenaar et al. (2009) under LL and Van Zanten et al. (2009) at HT. In these alternative pathways, ETHYLENE-INSENSITIVE 3 (EIN3) could be the receptor of ethylene signalling linking ethylene-induced hyponasty and sugars, given that EIN3 is degraded in the presence of glucose in interaction with light (Yanagisawa, Yoo & Sheen 2003; Lee, Deng & Kim 2006) and cooperates with the PHYTOCHROME-INTERACTING FACTOR 1 (PIF1) to prevent photo-oxidation and promote greening (Zhong et al. 2009). PIF4, the function of which is important in both LL- (Cole, Kay & Chory 2011) and HT-induced hyponasty (Koini et al. 2009), appears also as a candidate in the crosstalk between carbon status and phytochrome pathways.

Our results support the view that the primary cause of leaf hyponasty under moderately HT is related to the shade avoidance syndrome, and suggest that leaf temperature and transpiration per se have a minor role in this response. HT-induced hyponasty is therefore likely part of plant response selected to counteract carbon starvation rather than leaf warming itself.

Plant responses to HT mimic a carbon starvation

In addition to the effects of HT on hyponasty, our results show that its deleterious effects on plant growth are partially abolished with increasing light intensity. As shown for hyponasty, we hypothesized that HT-induced responses reflect an altered plant carbon status that may be counteracted by light intensity, at least for a moderately increased temperature.

There are several reasons why plants under elevated temperature would be carbon limited. For instance, Morison & Lawlor (1999) showed that assimilate demand could be increased, while photosynthetic capacity becomes limited in warm conditions. The kinetics of numerous plant processes are known to increase with temperature until an optimum above which rates strongly decrease before lethality (Jacobs & Pearson 1999; Gillooly et al. 2001; Parent et al. 2010). Accordingly, this study provides evidences that leaf heating is associated to an unbalanced carbon supply/demand. As indicated by sugar and starch contents, plant carbon status under HT was significantly impaired. Furthermore, the induction of heat response genes such as HSP101 and associated downstream metabolic pathways suggests that higher carbon allocation to maintenance was required under HT. This net carbon loss translated into reduced structural growth as indicated by lower leaf dry matter content and thickness, and higher specific leaf area at HT (Chabot & Chabot 1977; Atkin et al. 2006; Fig. 8). Hence, it is not surprising that plant tolerance to HT was increased at higher CO2 concentration (Huxman et al. 1998; Albert et al. 2011).

Here, we demonstrated that plant carbon status under HT was also significantly improved with increasing irradiance. This was indicated by an increase in sugar and starch contents, and the expression of specific marker genes. HL counterbalanced the deleterious effects of HT on net photosynthesis and PSII quantum efficiency, although higher light intensity induced higher leaf temperature, higher HSP101 expression and a slight increase in respiration rate. A global recovery of HT damages was found with increasing light intensity, whereas they were worsened under LL. As a result, plants grown at 30 °C under HL were bigger and morphologically more similar to plants grown at 20 °C, but under ML intensity.

Furthermore, changes in many plant traits observed in response to HT were similar to changes associated with the shade avoidance syndrome. For instance, hyponastic leaf growth and blade ratio decreases are typical responses to HT and LL (Gray et al. 1998; Tsukaya, Kozuka & Kim 2002; Franklin & Whitelam 2005; Koini et al. 2009; Van Zanten et al. 2009, 2010b; Heydarian et al. 2010). Several other changes in leaf and whole-plant traits are related to shading. Leaf structure is strongly altered by light intensity, and leaves developed in LL are thinner and tender, which can result in a better light harvesting (Chabot & Chabot 1977; Yano & Terashima 2001; Kim et al. 2005). These changes are well represented by the variations in specific leaf area (Witkowski & Lamont 1991; Poorter et al. 2009). which further increased under LL. Interestingly, increasing light at HT restored specific leaf area values close to the control values (Supporting Information Fig. S1). By increasing the density of photosynthetic tissues (Hassiotou et al. 2010) together with lower leaf angle (i.e. higher light interception), the side effects of HL could act synergistically to enhance net carbon gain under HT. Consistently, Foreman et al. (2011) have shown that light receptor action is critical for maintaining plant biomass at warm temperatures.

Flowering was delayed at HT and LL, in association with a decrease in leaf production rate (see also Mendez-Vigo et al. 2010). This contrasts with studies reporting that shade avoidance and HT accelerate flowering in Arabidopsis (Devlin et al. 1999; Sparks, Jeffree & Jeffree 2000; Botto & Smith 2002), but no data are available on their interactive effect. Our results are, however, in accordance with the negative effects of LL on Arabidopsis developmental rate under control temperature (Chenu et al. 2005), and could be interpreted as a symptom of decreased carbon availability under LL, because flowering is a major carbon sink (Christophe et al. 2008). In agreement, flowering time was clearly delayed in the starch-deficient mutant pgm (Supporting Information Fig. S2), irrespectively of the light conditions, in line with a disturbed carbon balance (Corbesier, Lejeune & Bernier 1998).

Natural variability and ecological consequences of temperature and light interactions

Plants have to manage a trade-off between improvement of photosynthesis with a higher light interception and limitation of radiant heat gain. This trade-off is a typical issue that a plant could encounter under a shaded, warm canopy – an environment highly competitive for carbon fixation. Indeed, phenotypic plasticity to light and temperature is an important trait for plants to achieve carbon assimilation and growth (Kim et al. 2005; Atkin et al. 2006), while plasticity in response to light can be considered as an adaptive response determining competitive ability in a plant canopy (Schmitt 1997; Dorn, Pyle & Schmitt 2000). In Arabidopsis, Van Zanten et al. (2009) suggested that HT-induced hyponasty is an adaptive response because it is negatively correlated to diurnal temperature range at the accession collection site. Here, variation between accessions was large enough to highlight the possibility for natural selection to act on the syndrome of traits described in this study.

Incident light levels used in this study were relatively low compared to those encountered in natural conditions, or to those that induce profound changes in the photosynthetic machinery (Bailey et al. 2001). Here, increasing PAR from 175 to 330 µmol m−2 s−1 appeared sufficient to counterbalance the negative effects of a 10 °C elevation of air temperature. The results found here should stand up to light levels that induce photosystem breakdown or until damages caused by heat gain from radiation become predominant over the improvement of photosynthesis. Whether HT-induced hyponasty would be observed in such conditions is still an open question. Further, as also suggested by Morison & Lawlor (1999), the results presented here warn us that the low levels of light used in many laboratory experiments testing the effects of HT may have altered the genuine response induced by HT. Nonetheless, hyponasty and subsequent changes in plant growth and development could be key traits conditioning plant performance under competition for light, particularly in a warming world.


The deleterious effects of HT on plants have been extensively studied, but few reports have taken into account the interacting effect of light intensity to interpret the observed responses. Here, we demonstrated that light strongly interacts with plant responses to HT by modulating its carbon balance. Temperature elevation induces a decrease in carbon assimilation and an increase in assimilate demand because of the over-activation of certain molecular and physiological processes. These energetically costly pathways would modify the carbon balance, which is, respectively, worsened under LL and restored with increasing light intensity. Because the dose–response to combined light and temperature varies between genotypes and between species, it is likely to play a key role in plant strategies and community dynamics.


We thank Christine Granier, Bertrand Muller and Matthew Hannah for comments on the paper; Myriam Dauzat, Alexis Bediee, Crispulo Balsera, Phillippe Clair and Gaëlle Rolland for help during the experiments. F.V. was funded by a CIFRE grant (ANRT, French Ministry of Research) supported by BAYER Crop Science (contract 0398/2009 – 09 42 008). F.P. was funded by French Ministry of Research.