Hormonal regulation of temperature-induced growth in Arabidopsis


  • Jon A. Stavang,

    1. Department of Plant and Environmental Sciences, Norwegian University of Life Sciences, N1432 Ås, Norway,
    Search for more papers by this author
    • Present address: Bioforsk Vest Ullensvang, 5781 Lofthus, Norway.

    • These authors contributed equally to this work.

  • Javier Gallego-Bartolomé,

    1. Instituto de Biología Molecular y Celular de Plantas (CSIC-UPV), Edificio E8, Campus Universitat Politecnica de Valencia, Av de los Naranjos s/n, 46022-Valencia, Spain,
    Search for more papers by this author
    • These authors contributed equally to this work.

  • María D. Gómez,

    1. Instituto de Biología Molecular y Celular de Plantas (CSIC-UPV), Edificio E8, Campus Universitat Politecnica de Valencia, Av de los Naranjos s/n, 46022-Valencia, Spain,
    Search for more papers by this author
  • Shigeo Yoshida,

    1. RIKEN Plant Science Center, Yokohama City, Kanagawa 230-0045, Japan
    Search for more papers by this author
  • Tadao Asami,

    1. Graduate School of Agricultural and Life Sciences, University of Tokyo, Tokyo, Japan
    Search for more papers by this author
  • Jorunn E. Olsen,

    1. Department of Plant and Environmental Sciences, Norwegian University of Life Sciences, N1432 Ås, Norway,
    Search for more papers by this author
  • José L. García-Martínez,

    1. Instituto de Biología Molecular y Celular de Plantas (CSIC-UPV), Edificio E8, Campus Universitat Politecnica de Valencia, Av de los Naranjos s/n, 46022-Valencia, Spain,
    Search for more papers by this author
  • David Alabadí,

    Corresponding author
    1. Instituto de Biología Molecular y Celular de Plantas (CSIC-UPV), Edificio E8, Campus Universitat Politecnica de Valencia, Av de los Naranjos s/n, 46022-Valencia, Spain,
    Search for more papers by this author
  • Miguel A. Blázquez

    1. Instituto de Biología Molecular y Celular de Plantas (CSIC-UPV), Edificio E8, Campus Universitat Politecnica de Valencia, Av de los Naranjos s/n, 46022-Valencia, Spain,
    Search for more papers by this author

*For correspondence (fax +34 963 87 7859; e-mail dalabadi@ibmcp.upv.es).


Successful plant survival depends upon the proper integration of information from the environment with endogenous cues to regulate growth and development. We have investigated the interplay between ambient temperature and hormone action during the regulation of hypocotyl elongation, and we have found that gibberellins (GAs) and auxin are quickly and independently recruited by temperature to modulate growth rate, whereas activity of brassinosteroids (BRs) seems to be required later on. Impairment of GA biosynthesis blocked the increased elongation caused at higher temperatures, but hypocotyls of pentuple DELLA knockout mutants still reduced their response to higher temperatures when BR synthesis or auxin polar transport were blocked. The expression of several key genes involved in the biosynthesis of GAs and auxin was regulated by temperature, which indirectly resulted in coherent variations in the levels of accumulation of nuclear GFP–RGA (repressor of GA1) and in the activity of the DR5 reporter. DNA microarray and genetic analyses allowed the identification of the transcription factor PIF4 (phytochrome-interacting factor 4) as a major target in the promotion of growth at higher temperature. These results suggest that temperature regulates hypocotyl growth by individually impinging on several elements of a pre-existing network of signaling pathways involving auxin, BRs, GAs, and PIF4.


As sessile organisms, plants have to cope with ever-changing environmental conditions. In the wild, plants are exposed to a variety of temperatures ranging from values below 0°C to values above 40°C, and they have developed the ability to continuously sense this environmental variable to modulate their growth and development accordingly. Plants are able to tolerate extreme low and high temperatures by triggering cold acclimation and thermo-tolerance, respectively (Penfield, 2008). These processes involve massive changes in the physiology of the plant directed to avoid the potential damaging effects of extreme temperatures. Between these extremes there is a range of temperatures that are optimal for a plant’s life. For instance, exposure to low temperatures is critical for initiating several developmental programs, such as flowering or breaking bud dormancy (Dennis and Peacock, 2007; Horvath et al., 2003). In this way plants living in temperate regions guarantee that these important developmental transitions occur in spring rather than during the less favorable months of winter. Plants respond in different ways to changes within the optimal, physiological range of temperatures. On the one hand, they are able to buffer these changes in temperature to avoid any unwanted effects on their basic physiology, which is the case, for example, with the temperature compensation of the circadian clock (Hotta et al., 2007). On the other hand, they actively respond and take advantage of moderate changes in ambient temperature, for example to modulate flowering time (Blázquez et al., 2003), plant architecture (Halliday and Whitelam, 2003; Mazzella et al., 2000) or cell expansion (Gray et al., 1998). These responses illustrate well the high degree of plasticity in plant development, even under small changes in an environmental variable (Casal et al., 2004).

A good example of the plastic integration of multiple cues into a particular developmental decision is the interplay between endogenous and environmental signals in the control of hypocotyl elongation (Alabadí and Blázquez, 2009). In this process, as in other stages of development, hormones have been placed at the center of the regulatory network, often mediating the effect of different external cues. For instance, hormone action has been proposed as an output target of the circadian clock during the regulation of daily hypocotyl growth (Covington and Harmer, 2007; Michael et al., 2008). In a similar way, light targets gibberellin (GA) synthesis to repress hypocotyl elongation when photomorphogenesis is triggered (Achard et al., 2007; Alabadíet al., 2008). And the promotion of hypocotyl growth by increasing temperatures requires the concurrence of at least auxin and brassinosteroids (BRs) (Gray et al., 1998). There is also evidence for the relevance of GAs in integrating information about ambient temperature to regulate elongation in species other than Arabidopsis, such as pea (Stavang et al., 2007), apple (Steffens and Hedden, 1992), wheat (Tonkinson et al., 1997), and citrus (Vidal et al., 2003).

Therefore, study of the molecular interactions that occur during hypocotyl growth provides the possibility of elucidating the molecular basis for this plasticity and to find out whether it relies on the interplay between multiple signaling pathways. In particular, does temperature recruit the activity of different hormones independently or is the involvement of multiple hormones a consequence of mutual interactions between them? Does temperature activate any hormone-independent mechanism to modulate growth? In this work we address these issues with a combination of genetic and microarray analyses in hormone mutant backgrounds at different temperatures.


Activity of the gibberellin pathway is necessary for hypocotyl elongation in response to high temperature

Analysis of temperature-induced hypocotyl growth in the weak GA-deficient mutant ga4 led to the conclusion that the GA pathway was not involved in promoting growth at high temperatures in Arabidopsis (Gray et al., 1998). To unambiguously demonstrate if the GA pathway is necessary for hypocotyl elongation in this species in response to an increase in the ambient temperature, we studied how this response was affected by the severe inhibition of GA biosynthesis, both pharmacologically and genetically. As shown in Figure 1(a), increasing amounts of the GA biosynthesis inhibitor paclobutrazol (PAC) in the medium progressively reduced the elongation response to temperature, being negligible at 1 μm PAC. Consequently, the response was not affected when medium was simultaneously supplemented with PAC and GAs (Figure 1b). The blockage of the response at the highest PAC concentration was similar to the effect observed in the severe GA-deficient mutant ga1-3 (Figure 1c). These results indicate that GAs are necessary for hypocotyl elongation in response to high temperature. Genetic analyses with several GA receptor mutants (Griffiths et al., 2006) confirmed the involvement of this pathway and indicated that the response is almost fully dependent upon the activity of GID1a and GID1c (Figure 1d).

Figure 1.

 The activity of the gibberellin (GA) pathway is needed to elongate the hypocotyl in response to temperature.
(a), (b) Hypocotyl length of 7-day-old Col-0 wild-type seedlings grown in continuous light in the presence of increasing concentrations of paclobutrazol (a) (P; 0.01–1 μm), or in the presence of 1 μm paclobutrazol (P), 1 μm paclobutrazol plus 50 μm GA3 (P + GA), or mock solution (b) at the indicated temperatures. The meaning of the bar colors is indicated in the inset in (a) and applies for all graphs in the figure.
(c)–(f) Hypocotyl length of 7-day-old Ler (c, e, f) or Col-0 (d) wild-type and mutant seedlings grown in continuous light at the indicated temperatures. Experiments were repeated twice with similar results. Data from one representative experiment is shown. Error bars represent standard error of the mean (= 15–20).

Hypocotyl and stem growth responses to GA are largely regulated by the DELLA proteins GAI (gibberellic acid insensitive) RGA (repressor of GA1) (Dill and Sun, 2001; King et al., 2001). To investigate whether these two proteins are also key regulators in temperature-induced growth, we studied this response in plants carrying semi-dominant (gai-1 and rga-Δ17; referred to as gai-D and rga-D in Figure 1e) and null alleles (gai-t6 rga-24) of these genes (Dill et al., 2001; Dill and Sun, 2001; King et al., 2001; Peng et al., 1997). gai-1 and rga-Δ17 seedlings responded like the wild type (Figure 1e), suggesting that the block in GA signaling caused by single dominant DELLA alleles can be overcome by temperature. Consistent with this idea, null mutations of both genes in the triple mutant ga1-3 gai-t6 rga-24 were not enough to completely bypass the need for extra GAs to properly respond to the higher temperature, although they restored the wild-type hypocotyl length at 20°C (Figure 1c; King et al., 2001) indicating that other DELLAs have to be inactivated to complete extra growth under this condition. In fact, removal of GAI and RGA in the GA1 genetic background (gai-t6 rga-24) did not affect the response (Figure 1e), whereas seedlings deficient in four (rga-t2 gai-t6 rgl1-1 rgl2-1; Achard et al., 2006) or in all five DELLA genes (rga-t2 gai-t6 rgl1-1 rgl2-1 rgl3-1; Feng et al., 2008), in which the GA pathway is fully active, showed a partial response (Figures 1f and 2). All these results indicate that activity of the GA pathway is needed to promote hypocotyl growth when ambient temperature rises, albeit it is not sufficient since pentuple della mutant seedlings still show a partial response.

Figure 2.

 The auxin and brassinosteroid (BR) pathways control temperature-induced growth independently of the activity of the DELLA proteins.
(a), (b) Hypocotyl length of 7-day-old Ler (a) or Col-0 (b) wild-type and mutant seedlings grown in continuous light in mock media, or in media supplemented with 100 μm 1-naphthylphthalamic acid (NPA), 3 μm brassinazole (BRZ) or 1 μm paclobutrazol (PAC) at the indicated temperatures. Error bars represent the standard error of the mean (= 15–20). Experiments were repeated twice with similar results. Data from one representative experiment are shown.

The auxin and BR pathways control temperature-induced growth independently of the activity of the DELLA proteins

Seedlings need intact auxin, BR, or GA pathways to be able to increase hypocotyl length in response to temperature (Figure 1; Gray et al., 1998). In certain developmental processes cross-regulation exists between these hormone pathways, for example auxin and BR jointly control hypocotyl growth and the expression of many genes (Nemhauser et al., 2004; Vert et al., 2008), and the control that auxin exerts on root growth and/or apical hook maintenance requires proper GA signaling (Achard et al., 2003; Fu and Harberd, 2003). Therefore, we tried to establish if there is a hierarchy in the action of these three hormone pathways or if they act independently of each other to control this growth response. For example, we reasoned that if part of the auxin activity to promote hypocotyl growth is mediated by GAs, mutant seedlings with a fully active GA pathway should be resistant, or partially resistant, to chemical blockage of the auxin pathway. The same rationale was applied to investigate cross-regulation between BRs and GAs. Pentuple della mutant seedlings, which have a fully active GA pathway, responded in a similar way to the wild type to treatment with an inhibitor of polar auxin transport, 1-naphthylphthalamic acid (NPA), and to treatment with the BR biosynthesis inhibitor brassinazole (BRZ; Figure 2a). The growth-restraining effect of both inhibitors, therefore, is not mediated by the growth-repressing activity of the DELLA proteins. These results suggest that regulation of the hypocotyl growth in response to temperature by auxin and BRs is not mediated by GAs but through parallel pathways.

Triple gsk3 mutants, which lack three of the ten GSK3s, BIN2-like kinases that negatively regulate BR signaling and that therefore show a very active BR pathway (Vert and Chory, 2006), were more sensitive than the wild type to temperature, but they responded like the wild type to the NPA treatment (Figure 2b). This is consistent with the currently understood molecular mechanism for synergistic interaction of auxin and BRs: the highly active BR pathway sensitizes the seedling to the temperature-stimulated increase in the endogenous auxin levels, and this is abolished when auxin transport is inhibited (Nemhauser et al., 2004; Vert et al., 2008). Triple gsk3 mutants were not resistant to the inhibitory effect of PAC on temperature-induced growth (Figure 2b). In the absence of any molecular data that support a possible interaction, the most straightforward explanation is that GA activity is not mediated by activation of the BR pathway to properly promote growth in response to temperature.

Expression of hormone metabolism genes is modulated by ambient temperature

Hypocotyls of seedlings growing at 29°C have increased levels of both free and conjugated indole-3-acetic acid (IAA) when compared with hypocotyls of seedlings growing at 20°C, suggesting that temperature adjusts IAA levels to modulate growth (Gray et al., 1998). In order to gauge the possibility that temperature also targets GA and/or BR metabolism to promote hypocotyl growth, we analyzed the expression of genes encoding key enzymes in both pathways (Yamaguchi, 2008; Fujioka and Yokota, 2003), as well as those coding for enzymes involved in IAA biosynthesis (Benjamins and Scheres, 2008; Stepanova et al., 2008; Tao et al., 2008), in response to a temperature shift. For this purpose we made use of 4-day-old seedlings grown at 20°C and transferred to 29°C for different times; importantly, seedlings of this age were competent to respond to the temperature shift concerning promotion of hypocotyl growth (Figure S1 in Supporting Information). In addition, to get an idea of where in the seedling changes in gene expression take place we analyzed cotyledons and hypocotyls separately. The most remarkable changes in expression of GA metabolism genes were observed in the hypocotyl, where expression of two major biosynthetic genes in the pathway, AtGA20ox1 and AtGA3ox1, was rapidly upregulated after the transfer to the higher temperature (Figure 3). On the contrary, transcript levels of the gene encoding a major GA-inactivating enzyme, AtGA2ox1, decreased soon after the transfer, being 10-fold lower than in the controls 4 h later. In cotyledons the same trends were observed for the expression of both AtGA3ox1 and AtGA2ox1, although changes were smaller in the latter case. Conversely, transcript levels of AtGA20ox1 did not significantly change in this organ over the time course. The localization of the expression of these genes shown by the quantitative (q)RT-PCR analysis was further supported by the expression pattern of a transcriptional fusion of GUS under the control of the AtGA3ox1 promoter (Mitchum et al., 2006), which increased all over the cotyledons and in the expansion zone of the hypocotyl 2 h after the temperature shift (Figure 4a). The other members of the GA20ox family (AtGA20ox2–5) either did not change or showed little change during the time course (Figure S2a). Interestingly, transcript levels of other members of the GA2ox family tended to accumulate late in the time course in both organs, whereas that of AtGA3ox2 steadily decreased in hypocotyls after the temperature rise (Figure S2a), which could be a consequence of the feed-forward and feed-back regulatory mechanisms, respectively, that operate to control GA homeostasis (Hedden and Phillips, 2000). The increased expression of AtGA2ox4 may alternatively respond to a protective mechanism to avoid the flow of GAs to the shoot apical meristem (Jasinsky et al., 2005).

Figure 3.

 Temperature modulates the expression of genes encoding key metabolic enzymes of the auxin and gibberellin (GA) pathways. Four-day-old wild-type Col-0 seedlings were grown at 20°C under continuous light, as described in Experimental procedures, and transferred to 29°C for the indicated times in hours (h). Transcript levels in cotyledons and hypocotyls were determined by quantitative (q)RT-PCR. The expression level at time 0 for each gene was set to 1 and the level in the other samples was calculated relative to the corresponding time 0 value. Data represent the mean and standard deviation of three technical replicates. Experiments were repeated twice with similar results; results from one representative experiment are shown.

Figure 4.

 Spatial activation of the auxin and gibberellin (GA) pathways. Activity of the brassinosteroid (BR) signaling pathway.
(a) Seedlings were grown in continuous light for 4 days at 20°C and then kept at the same temperature or transferred to 29°C for 2 h. Seedlings were fixed and stained for β-glucuronidase activity as described in Experimental procedures. Representative seedlings are shown.
(b), (c) Five-day-old 35S:BES1-GFP and pBZR1:BZR1-CFP seedlings were grown in continuous light at 20°C and transferred to 29°C for several hours (4 and 8 h) or were grown continuously at either 20 or 29°C (5 days). The ratio of non-phosphorylated versus phosphorylated versions of BZR1 and BES1 (b), and the total amount of each protein (c) is shown. In (c), the amount of protein at time points 4 and 8 h in the time course was normalized to RPT5 and then to the amount at time 0; the amount of protein from seedlings continuously growing at 29°C was normalized to RPT5 and then to the amount in seedlings grown at 20°C. Total proteins were extracted from hypocotyls and the transgenic proteins detected by western blotting using an anti-GFP antibody. The RPT5 levels were used as a loading control. Quantification was carried out as described in Experimental procedures. Results are the mean of two independent biological repeats and error bars allude to standard deviation.

To assess if these changes in gene expression result in the activation of the GA signaling pathway, we analyzed the accumulation pattern of the DELLA protein RGA in the elongation zone of hypocotyls of 4-day-old seedlings transferred from 20 to 29°C. For that purpose, we made use of pRGA:GFP-RGA seedlings that express a GFP–RGA fusion that faithfully recapitulates the activity of the endogenous RGA protein (Silverstone et al., 2001). Confocal imaging showed that the nuclear fluorescence due to accumulation of GFP–RGA strongly decreased in the elongation zone of hypocotyls 4 h after the temperature rose, being barely detectable 8 h after the shift (Figure 5). The decrease in GFP–RGA levels was observed across several optical sections in the hypocotyls of seedlings transferred to 29°C (see average projections of image Z-stacks in Figure S2b, as well as the series of individual focal planes from the same images in Movies S1–S4). Importantly, accumulation of GFP–RGA was not modulated by temperature in hypocotyls of the GA-deficient mutant ga1-3, suggesting that changes observed in the wild-type GA1 genetic background are a consequence of alterations in GA levels triggered by the temperature shift (Figure 5). On the other hand, variations in temperature did not affect the endogenous or the transgenic RGA transcript level (data not shown). Taken together, these results suggest the GA pathway is more active at 29°C than at 20°C, most probably as a result of increased GA levels.

Figure 5.

 The GFP–RGA (repressor of GA1-3) disappears from nuclei of the hypocotyl elongation zone after increases in ambient temperature. Four-day-old pRGA:GFP-RGA and ga1-3 pRGA:GFP-RGA seedlings were grown at 20°C under continuous light, transferred to 29°C for the indicated times in hours (h), and fluorescence of GFP–RGA and chloroplasts in the upper, apical third of the hypocotyls detected by confocal imaging as described in Experimental procedures. The arrow indicates nuclear GFP–RGA fluorescence. Images are representative of two independent biological repeats each including 10–15 seedlings per time point and per genotype.

A less complex scenario seems to occur in the case of BR metabolism genes, since the expression of only two genes changed in response to the temperature shift: CPD and DWF4 were moderately upregulated in cotyledons and in hypocotyls, respectively (Figure S3a), suggesting a possible increase in the biosynthesis of BR. No significant changes were observed for other biosynthesis genes in the pathway, such as DET2, ROT3, and BR6ox1, whereas expression of BAS1, which encodes a BR-inactivating enzyme, slightly increased late in the time course, most probably as a consequence of the feed-forward mechanism that regulates BR homeostasis (Figure S3a). To test whether these changes in gene expression result in an activation of the BR pathway in hypocotyls, we investigated the phosphorylation status of the positive BR-signaling elements BZR1 and BES1 in response to an increase in temperature (Vert and Chory, 2006). Interestingly, the ratio of dephosphorylated (active) versus phosphorylated (inactive) versions of BZR1 and BES1 was higher in hypocotyls of 5-day-old seedlings continuously growing at 29°C than at 20°C (Figures 4b and S3b); however, BR signaling does not seem to be a primary target for temperature since we could not detect changes in this ratio during the first 8 h after a shift from 20 to 29°C (Figures 4b and S3b). Besides, part of the long-term effect of temperature upon BR signaling was exerted at the level of accumulation of BZR1 protein, because it was higher in seedlings growing at 29°C compared with those at 20°C (Figures 4c and S3b). However, an increase in ambient temperature did not affect the BES1 or BZR1 transcript levels (data not shown).

As mentioned above, a moderate temperature increase promotes the accumulation of IAA in hypocotyls (Gray et al., 1998). Genetic analyses indicated that this process involves at least two of the Trp-dependent IAA biosynthesis branches, the one defined by the genes CYP79B2 and CYP79B3 (Zhao et al., 2002) and the one defined by TAA1 (Stepanova et al., 2008; Tao et al., 2008). To estimate if the other Trp-dependent pathway, namely the one defined by the YUCCA (YUC) family (Zhao et al., 2001), could be involved in the increase in IAA biosynthesis at higher temperatures, we analyzed the expression of all members of the family after a temperature shift (Figure 3). The larger changes were observed in cotyledons. In this organ, YUC8 and YUC9 showed a strong increase in their expression levels, whereas they were only marginally higher in the hypocotyl of the shifted seedlings. To the best of our knowledge, there is no experimental evidence showing that these two genes participate in IAA biosynthesis. However, the strong sequence homology with YUC5 (Cheng et al., 2006), a bona fide IAA biosynthesis gene (Woodward et al., 2005), suggests they may also be actively involved in this pathway. The expression of all other members of the family, as well as that of TAA1 and one of its homologs, TAR2, either did not change or only slightly changed after the temperature shift (Figure S4); expression of TAR1, the other TAA1 homolog gene, was below the detection limit (data not shown). Upregulation of YUC8 and YUC9 genes in cotyledons in response to a moderate temperature increase is consistent with their participation in the promotion of IAA biosynthesis in this organ. In fact, the activity of the DR5:GUS reporter increased specifically in cotyledons shortly after the shift, which is probably the result of the local accumulation of newly synthesized IAA (Figure 4a).

In summary, these results suggest that the activity of auxin and GA pathways is quickly upregulated in response to an increase in the ambient temperature and that it may be the result of de novo accumulation of hormone pools, therefore promoting hypocotyl elongation.

Global changes in the transcriptome in response to differences in temperature

In an effort to identify the primary targets of temperature regulation and study their connection with the hormonal network described above, we analyzed the transcriptome of seedlings that have been subjected to a short-term increase in temperature, by using 70-mer oligonucleotide arrays that represent the majority of the Arabidopsis genes (http://www.ag.arizona.edu/microarray/). Three biological replicates were used for the analysis following the same experimental design described in the previous section, although samples were harvested from whole seedlings and only 2 h after the shift. A two-fold cutoff value allowed us to identify 113 genes that were differentially expressed in the shifted samples, 100 of which were upregulated, while 13 of them were downregulated (Table S1). Subsequently, we sought to identify any Gene Ontology (GO) term (Ashburner et al., 2000) over- or under-represented in the list of differentially expressed genes compared with the rest of the genes included in the array by using the FatiGO algorithm (Al-Shahrour et al., 2005). Several biological processes, one of the three GO categories, were over-represented among the genes upregulated in response to the temperature shift (Table 1). Four of the biological processes differentially represented, namely response to heat, response to light intensity, response to reactive oxygen species, and protein folding, mostly shared the same genes, encoding several heat shock proteins (Table S2). Even though 29°C may lie within the physiological range of growth temperatures, a step transfer from 20 to 29°C may also activate the heat stress response, which shares signaling components with other abiotic stress pathways such as light intensity or oxidative stresses (Kotak et al., 2007; Nishizawa et al., 2006; Rizhsky et al., 2004). More importantly, two biological processes related to growth were identified as over-represented (Tables 1 and S2): response to auxin stimulus and response to red or far-red light. The over-representation of auxin-regulated genes, though striking, may be a direct manifestation of auxin accumulation after the temperature shift. Seven out of eight auxin-regulated genes belong to the SAUR family (McClure and Guilfoyle, 1987). As far as we know, no direct role in growth promotion for any SAUR gene has been demonstrated. Nonetheless, expression of these genes is usually associated with elongating tissues, for instance expression of two SAUR genes is transiently induced by simulated shade, a condition that also promotes hypocotyl elongation (Roig-Villanova et al., 2007). On the other hand, over-representation of genes related to light signaling among those upregulated by the temperature shift pointed to an otherwise expected cross-regulation between the two pathways, which oppositely regulate hypocotyl elongation. Two genes encoding transcription factors whose activity promotes elongation, PIF4 and ATHB-2 (Huq and Quail, 2002; Steindler et al., 1999), were identified in this group.

Table 1.   Non-redundant Gene Ontology (GO) categories over-represented in the set of genes induced by temperature
GO termGO levelDif. expressed genes (%)Total genes (%)Corrected P value
Response to heat516.980.861.9 × 10−7
Response to auxin stimulus515.092.201.8 × 10−3
Response to light intensity620.000.459.9 × 10−9
Red or far-red light signaling pathway612.501.131.3 × 10−2
Protein folding615.002.272.9 × 10−2
Response to hydrogen peroxide722.580.331.1 × 10−8

PIF4 activity is critical for temperature-induced hypocotyl growth

Activity of PIF4 (phytochrome-interacting factor 4) has been demonstrated to be important in regulating elongating growth under several physiological contexts, usually acting in concert with other PIF proteins (Alabadíet al., 2008; de Lucas et al., 2008; Huq and Quail, 2002; Koini et al., 2009; Leivar et al., 2008a,b; Lorrain et al., 2008; Nozue et al., 2007). To investigate if the expression of other PIF genes involved in growth promotion, i.e. if PIF1 (Shen et al., 2005), PIF3 (Kim et al., 2003), and PIF5 (Nozue et al., 2007) were also targets for temperature regulation, we analyzed their expression by qRT-PCR in dissected seedlings subject to a temperature shift. As shown in Figure 6(a), the induction of PIF4 expression at high temperature was confirmed and detected in both cotyledons and hypocotyls, while expression of PIF5 was similarly increased only in hypocotyls. On the contrary, transcript levels of PIF1 and PIF3 were not altered in either organ. Interestingly, expression of PIF4 was maintained at higher levels in hypocotyls of seedlings continuously growing at 29°C compared with those growing at 20°C (Figure S5a). We assayed levels of PIF4-HA protein from hypocotyls of plants constitutively over-expressing the fusion protein (Nozue et al., 2007) grown at either 20 or 29°C in order to test whether there is any effect of temperature on the amount of PIF protein at the post-transcriptional level. The amount of PIF4-HA was similar at the two temperatures (Figure 6b), suggesting that post-translational regulation of PIF4 levels by temperature is negligible.

Figure 6.

 Temperature upregulates PIF4 expression.
(a) Four-day-old wild-type Col-0 seedlings were grown at 20°C under continuous light and transferred to 29°C for the indicated times, in hours (h). Transcript levels in cotyledons and hypocotyls were analyzed by quantitative (q)RT-PCR. The expression level at time 0 for each gene was set to 1 and level in the other samples was calculated relative to the corresponding time 0 value. Data represent the mean and standard deviation of three technical replicates. Experiments were repeated twice with similar results; results from one representative experiment are shown.
(b) Five-day-old 35S:PIF4-HA (Nozue et al., 2007) seedlings were grown at either 20 or 29°C in continuous light. Total proteins were extracted from hypocotyls and the transgenic proteins detected by western blotting using an anti-HA-peroxidase antibody. Total proteins from hypocotyls of 5-day-old wild-type Col-0 seedlings grown at 20°C were used as a non-transgenic control. Loading was checked by Ponceau staining.

To investigate the physiological relevance of changes in PIF4 and PIF5 expression induced by temperature, we measured hypocotyl lengths at 20 and 29°C in seedlings carrying null alleles in PIF genes. This analysis confirmed that the stimulation of hypocotyl growth by temperature is largely mediated by PIF4 activity, since pif4 null mutant seedlings hardly responded to temperature (Figure 7). The response of pif1, pif3, and pif5 null mutants was very similar to that of the wild type (Figure 7a,b), while the double mutant pif4 pif5 behaved like the pif4 single mutant (Figure 7a). Hence, genetic analysis clearly points to PIF4 as a critical element in the regulation of growth by temperature. The discrepancy between the lack of a defect in the response of pif5 mutants and the induction of PIF5 expression in hypocotyls may be due to redundancy with PIF activities other than PIF4, for instance PIF7, since seedlings over-expressing PIF5, like those that over-express PIF4, showed a reduced elongation response, suggesting that the growth pathway is partially saturated at the lower temperature (Figure 7c).

Figure 7.

 Activity of PIF4 is crucial for temperature-stimulated hypocotyl elongation.
(a)–(c) Hypocotyl length of 7-day-old Col-0 wild-type and mutant seedlings grown in continuous light at the indicated temperatures. Experiments were repeated twice with similar results. Error bars represent standard error of the mean (= 15–20).


The ability of plants to cope with different, ever-changing environmental conditions is key to their survival, and this is most significant at the seedling stage. Temperature is an environmental factor that plants can perceive on a daily basis, and one remarkable effect is that not only are the basic physiological processes like metabolism, carbon partitioning, or photosynthesis affected, but developmental decisions, such as growth, that ultimately govern the architecture of the plant, are also modulated according to small variations in ambient temperature (Penfield, 2008). Elucidating the molecular events that underlie this regulation is key to understanding the fundamentals of plasticity.

The results presented here show that temperature regulates growth by modulating the activity of a pre-existing network that involves auxin, BRs, GAs, and the transcription factor PIF4 – known to be part of this network through its regulatory interaction with GA-regulated DELLA proteins (de Lucas et al., 2008), and shown to act as a growth promoter in several developmental stages (Alabadíet al., 2008; Huq and Quail, 2002; Koini et al., 2009; Leivar et al., 2008a,b; Lorrain et al., 2008; Nozue et al., 2007).

Interestingly, the mechanism by which changes in ambient temperature enhance or downplay the network rely mostly on transcriptional regulation. Although transcriptional control is usually associated with long-term regulation during development and might seem difficult to reconcile with the plastic, reversible nature of the regulation of growth by temperature, two arguments support this transcriptional model. First, the changes in the expression level of GAs and auxin metabolism genes, as well as PIF4, occur very fast after the seedlings undergo a temperature shift (Figures 3 and 6). And second, it has been shown in pea plants that moderate drops in ambient temperature rapidly and reversibly stimulate the expression of the PsGA2ox2 gene that parallels the reversible decrease in the growth rate (Stavang et al., 2007). This growth-promoting mechanism seems to be widely conserved in plants, since the GA pathway is also targeted by temperature to promote growth in apple (Steffens and Hedden, 1992), wheat (Tonkinson et al., 1997), and citrus (Vidal et al., 2003). Besides, our results show that the expression of several heat shock protein (HSP) genes, such as the ones encoding the chaperones HSP70 and HSP101, is clearly affected by temperature changes within the range that modulates growth (i.e. between 20 and 29°C in this study). These genes are direct targets for the HSF family of transcription factors (Busch et al., 2005), and the consensus cis-elements recognized by these proteins are also present in the promoters of some of the GA, BR, and auxin metabolism genes identified here (data not shown), so it is likely that they are directly regulated as part of the general response to temperature.

All of the above indicates that articulating rapid changes in the transcriptional activity of key metabolic genes in certain hormone pathways may be a general strategy by which temperature modulates hypocotyl or stem growth rate. More importantly, the spatial localization of the transcriptional control of hormone genes occurs in the regions known to be natural sources for hormone synthesis (i.e. auxin in the cotyledons) or the hormone-responsive tissue where active expansion takes place, such as the top third of the hypocotyl (Figures 3–6).

The framework for the control of growth by temperature that emerges now presents two additional features characteristic of an adaptive response: plasticity and robustness. An important element that provides plasticity to the network is that each individual pathway is recruited by temperature and contributes to the precise response to environmental changes. And, moreover, a certain degree of cross-regulation is likely to contribute to this property of the network in wild-type plants, as proposed for auxin and BRs (Figure 2; Nemhauser et al., 2004; Vert et al., 2008) and for auxin and GAs (Frigerio et al., 2006). Furthermore, this regulatory mechanism includes a module that appears to be very robust: on the one hand, temperature promotes PIF4 activity at the transcriptional level as an early and direct response to this environmental cue, independently of hormonal homeostasis of the seedling (Figure S5b). On the other hand, temperature might act at the post-translational level through the transcriptional activation of the GA pathway, as a result of releasing the inhibitory effect of the DELLA proteins on PIF4’s DNA-binding activity (de Lucas et al., 2008). The similar phenotype of pentuple della mutants and PIF4 overexpressors highlights a prominent feature of the strategy to control growth by temperature, which is the independent effect on at least two pathways (namely GA and PIF4) that eventually merge and reinforce the response (Figure S6). According to this view, a simple explanation for the apparently paradoxical observation that pentuple della mutants still respond to an increase in temperature (Figures 1 and 2; Koini et al., 2009) is that the absolute level of PIF4 protein is higher at 29°C; while the observation that PIF4 overexpressors are taller at 29°C than at 20°C (Figure 7) would rely on the diminished negative interaction between PIF4 and DELLA proteins at the higher temperature. Moreover, other pathways are also triggered in these mutants, i.e. auxin and BR signaling, which would also participate in the concurrent regulation of growth by temperature. The same rationale might be applied to explain why yucca mutant seedlings still respond to temperature (Nemhauser et al., 2004). It is likely that the mechanisms through which GA and auxin/BR may control growth in response to temperature are different, based on the comparison between triple gsk3 and pentuple della mutants or the PIF4 over-expressing lines (Figures 1, 2, and 7). A model that accommodates these observation would be that GA regulate the downstream ‘growth genes’ mainly through the activity of PIF4 (de Lucas et al., 2008), whereas auxin and BR pathways regulate target ‘growth genes’ directly through the activity of ARFs and BES1/BZR1 transcription factors, respectively (Benjamins and Scheres, 2008; He et al., 2005; Li et al., 2009;Yin et al., 2005), and/or modulate each other’s activity to jointly undertake this task (Vert et al., 2008).

In natural environments ambient temperature usually fluctuates on a daily basis, with warm days and cool nights. The ability of seedlings to grow also oscillates diurnally, being highest at the end of the night due to the joint action of light signaling pathways and the circadian clock that phases the growth-promoting activities of PIF4 and PIF5 towards the end of the dark period (Nozue et al., 2007). Then, when is the growth-stimulating effect of temperature physiologically relevant? One can envision at least two possibilities: in certain natural niches this regulation might be relevant during the day, to take advantage of the warmer temperatures and thus to counteract the negative effect that light exerts on both PIF4 stability and on active GA levels (de Lucas et al., 2008; Nozue et al., 2007; Zhao et al., 2007); alternatively, in certain natural environments it could be relevant during the night, to enhance growth rate in the dark. Several lines of evidence support the first hypothesis: first, pea plants grow taller under a regime of warmer days than under warmer nights, due to a decreased rate of inactivation of GA during warm days (Grindal et al., 1998; Stavang et al., 2005); second, temperature drop treatments are more effective in reducing the growth rate in pea plants when applied in the middle of the light period rather than in the middle of the night, and this correlates with diminished levels of active GA (Stavang et al., 2007). Light and temperature signals regulate the same set of GA metabolism genes in opposite ways to control hypocotyl elongation in Arabidopsis (Figure 3; Alabadíet al., 2008), suggesting that the GA pathway may be indeed key to integrating both environmental cues during the day and to accordingly modulate growth rate (Franklin, 2008). Lastly, increased cell expansion at high temperatures may allow the seedling to bring cotyledons and the apical meristem away from the heated soil in certain landscapes (Gray et al., 1998).

In any case, it is worth mentioning that our microarray analysis provides additional support for the view that growth modulation is only a subset of the responses triggered by mild changes in ambient temperature, and the observation that genes like ASN1, encoding asparagine synthetase (Lam et al., 2003), or APRR7, encoding a pseudo-response regulator, are induced at higher temperatures may provide clues to understanding at the molecular level the regulation of carbon partitioning by temperature, and temperature compensation of the circadian clock (Salome and McClung, 2005), respectively.

In summary, our results indicate that ambient temperature influences hypocotyl elongation through the rapid modulation of auxin and GA hormone pathways, and of PIF4 activity, while the BR pathway seems to be required at later stages of growth. Remarkably, a seedling’s growth response to temperature partly relies on a dual feed-forward mechanism in which temperature robustly enhances elongation by simultaneously promoting activity of PIF4 at the transcriptional and post-translational levels, the latter mediated by the GA pathway.

Experimental procedures

Plant material, growth conditions and hypocotyl length measurements

Arabidopsis thaliana accessions Col-0 and Ler were used as the wild type. The pRGA:GFP-RGA ga1-3 line was obtained by genetic crosses and isolated from an F3 population. All seeds were surface sterilized for 5 min in 70% (v/v) ethanol and 0.01% (v/v) Triton X-100, followed by 5 min in 96% (v/v) ethanol. Seeds were sown on plates of 1/2 MS medium (Duchefa, http://www.duchefa.com/), 0.8% (w/v) agar, 1% (w/v) sucrose, and stratified at 4°C for 6 days in the dark. Stratification of ga1-3, ga1-3 rga-24 gai-t6 and pRGA:GFP-RGA ga1-3 seeds was done in water containing 10 μm GA3 (Duchefa) and then rinsed several times with water before sowing. Germination was induced by placing plates for 24 h under white fluorescent light (90–100 μmol m−2 sec−1) at 20°C in a Percival E-30B (http://www.percival-scientific.com/). In those experiments involving growth at two temperatures, one group of plates was kept at 20°C and the replica was placed at 29°C; in both cases the lighting regime was set to continuous white fluorescent light (20–30 μmol m−2 sec−1) and seedlings were grown for a total of 5 or 7 days. In experiments involving a temperature rise, germination was induced as above and plates were kept at 20°C and continuous white fluorescent light (20–30 μmol m−2 sec−1) for three additional days before transfer to 29°C for several hours; the lighting regime was kept constant.

In experiments involving pharmacological treatments, seeds were sown on sterile filter papers, placed on 1/2 MS plates and stratified as above. Filter papers harboring seeds were transferred to control or treatment plates by the end of the 24-h period of induction of germination. The PAC treatments were 0.01–1 μm (Duchefa) and the corresponding control plates contained 0.01% acetone (v/v, final concentration). The GA3 treatments were 50 μm and the control plates contained 0.014% ethanol (v/v, final concentration). The BRZ treatments were 3 μm and the control plates contained 0.01% DMSO (v/v, final concentration). The NPA treatments were 100 μm (Sigma, http://www.sigmaaldrich.com/) and the control plates contained 0.2% DMSO (v/v, final concentration).

In all cases plates were placed horizontally in the growing cabinets. To measure the hypocotyl length, seedlings were scanned and length was measured with ImageJ software (http://rsb.info.nih.gov/ij/).

Gene expression analysis by ‘real-time’ quantitative RT-PCR

Total RNA was isolated from either whole seedlings or separately from hypocotyls and cotyledons by using the E.Z.N.A.® Plant RNA Mini Kit (Omega Bio-tek, http://www.omegabiotek.com,). Complementary DNA synthesis and quantitative PCR as well as primer sequences for amplification of GA metabolism genes and PIF1, PIF3, PIF4, and EF1-α genes has been described (Alabadíet al., 2008; Frigerio et al., 2006). Sequences of other primers used in this study can be found in Table S3.

Gene expression analysis by long oligonucleotide microarrays

Wild-type Col-0 seeds were sterilized, sown, stratified, and germinated as described above. All plates were kept for three additional days at 20°C under constant, white fluorescent light (20–30 μmol m−2 sec−1). One set of plates was transferred to 29°C under continuous, white fluorescent light (20–30 μmol m−2 sec−1) for 2 h; control plates were kept at 20°C under the same lighting regime. Three independent biological replicates were used for the analysis. Total RNA from whole seedlings was extracted as above. RNA amplification, labeling, and hybridization of microarray slides were carried out as described in Bueso et al. (2007). Microarray slides were scanned with a GenePix 4000B scanner (Axon Molecular Devices, http://www.moleculardevices.com/). Spot intensities were quantified using Genepix Pro 6.0 software (Axon Molecular Devices) and those with a net intensity in both channels lower than the median signal background plus twice the standard deviations were removed as low signal spots. Data were normalized by median global intensity and with LOWESS correction (Yang et al., 2001) with Genepix Pro 6.0 and Acuity 4.0 software (Axon Molecular Devices), respectively. Microarray raw data have been deposited in the NCBI’s GEO database under accession GSE13822.

GUS staining

GUS staining was performed as described in Frigerio et al. (2006).

Protein extraction and western blots

Total proteins from hypocotyls of 35S:BES1-GFP (Yin et al., 2002), pBZR1:BZR1-CFP (Wang et al., 2002), and 35S:PIF4-HA (Nozue et al., 2007) transgenic lines were extracted by homogenizing ground frozen tissue in one volume of 62 mm 2-amino-2-(hydroxymethyl)-1,3-propanediol (TRIS)-HCl pH 6.8, 25% (v/v) glycerol, 2% (w/v) SDS, 0.02% (w/v) bromophenol blue, and DTT (20 mg ml−1), boiled for 5 min and centrifuged at maximum speed for 10 min at room temperature. Protein concentration in supernatants was quantified using the RC DC Protein Assay (Bio-Rad, http://www.bio-rad.com/). Twenty to forty micrograms of denatured proteins were separated in Precise® 8% TRIS-HEPES-SDS gel (Pierce, http://www.piercenet.com/) and transferred onto PVDF membranes (Bio-Rad). Signal from bound antibodies was detected using the ECL Advance Western Blotting Detection Kit (GE Healthcare, http://www1.gelifesciences.com/). The GFP and cyan fluorescent protein (CFP) fusion proteins were detected using an anti-GFP (clone JL-8, Clontech, http://www.clontech.com/), and PIF4-HA was detected using an anti-HA-peroxidase (clone 3F10, Roche, http://www.roche-applied-science.com/). Antibodies against RPT5 were used as a loading control (Yu et al., 2008). Quantification of the intensity of protein bands was carried out using a Luminescence Image Analyzer LAS-3000 (Fujifilm, http://www.fujifilmlifescienceusa.com/).

Confocal imaging

Fluorescence from the GFP–RGA fusion protein was detected using a Leica TCS SL confocal microscope (Leica Microsystems, http://www.leica-microsystems.com/). The excitation wavelength was 488 nm and GFP emission was detected between 505 and 530 nm. To discriminate fluorescence emitted by GFP from that emitted by chloroplasts, which is detected in the same window, chloroplasts were further visualized by detecting red fluorescence between 610 and 660 nm. Superimposition of both images shows orange chloroplasts due to autofluorescence and green nuclei due to GFP. The same settings were used to detect fluorescence from all samples. Images shown in Figure S2 (b) depict 30–60 μm Z-stacks and correspond to the sum of individual optical sections whose step size is 1 μm.


We thank G. Choi (KAIST, Daejeon, South Korea), C. Fankhauser (University of Lausanne, Lausanne, Switzerland), T. Guilfoyle (Department of Biochemistry, University of Missouri, MO, USA), N. P. Harberd (Department of Plant Sciences, University of Oxford, Oxford, UK), E. Huq (University of Texas, Austin, TX, USA), T-p Sun (Department of Biology, Duke University, Durham, USA), S. G. Thomas (Rothamsted Research, Hertfordshire, UK), G. Vert (Institut de Biologie Intégrative des Plantes, Montpellier, France), Z. Y. Wang (Department of Plant Biology, Carnegie Institution, Stanford, USA), Y. Yin (Plant Science Institute, Iowa State University, Ames, IA, USA), and the Arabidopsis Biological Resource Center for seeds; and X. W. Deng (Yale University, New Haven, CT, USA) for antibodies against RPT5. We also thank Dr Jorge Casal (Universidad de Buenos Aires, Buenos Aires, Argentina) for helpful suggestions on this work. Work in the authors’ laboratories is funded by grant BIO2007-60923 from the Spanish Ministry of Science and Innovation and by grant 167890/110 from the Norwegian Research Council. JG-B was supported by a JAE pre-doctoral fellowship from CSIC.