Temperature is a major factor governing the distribution and seasonal behaviour of plants. Being sessile, plants are highly responsive to small differences in temperature and adjust their growth and development accordingly. The suite of morphological and architectural changes induced by high ambient temperatures, below the heat-stress range, is collectively called thermomorphogenesis. Understanding the molecular genetic circuitries underlying thermomorphogenesis is particularly relevant in the context of climate change, as this knowledge will be key to rational breeding for thermo-tolerant crop varieties. Until recently, the fundamental mechanisms of temperature perception and signalling remained unknown. Our understanding of temperature signalling is now progressing, mainly by exploiting the model plant Arabidopsis thaliana. The transcription factor PHYTOCHROME INTERACTING FACTOR 4 (PIF4) has emerged as a critical player in regulating phytohormone levels and their activity. To control thermomorphogenesis, multiple regulatory circuits are in place to modulate PIF4 levels, activity and downstream mechanisms. Thermomorphogenesis is integrally governed by various light signalling pathways, the circadian clock, epigenetic mechanisms and chromatin-level regulation. In this Review, we summarize recent progress in the field and discuss how the emerging knowledge in Arabidopsis may be transferred to relevant crop systems.
The year 2015 is on track to surpass 2014 as the warmest year ever recorded since systematic temperature measurements began more than a century ago1. In fact, the 10 warmest years on record all occurred after 1998. The fifth report of the Intergovernmental Panel on Climate Change2 projects an increase of 0.8–4.8 °C in global mean surface temperature within the twenty-first century. Such figures are alarming as it is expected that this will strongly affect plant distribution and survival, and therefore threaten biodiversity3,
Likewise, crop productivity will probably suffer greatly from global warming, while food production is required to increase significantly to sustain a growing and more demanding world population9,13,
Breeding for crop-level adaptations to cope with high temperatures could potentially reverse this negative trend9,13,
Although abundant literature is available on how plants tolerate extreme heat stress (reviewed in refs 9,21), we are only beginning to understand the molecular mechanisms underlying thermomorphogenesis in response to moderately increased temperatures. A key breakthrough was the identification of the bHLH (basic helix–loop–helix) transcription factor PHYTOCHROME INTERACTING FACTOR 4 (PIF4) as a central regulator of ambient temperature signalling in Arabidopsis22. Recent findings have implicated important roles for light signalling pathways, the circadian clock23,
Here we discuss and integrate recent findings on the molecular networks driving thermomorphogenic adaptations. We will highlight missing links and suggest how the knowledge on Arabidopsis could be transferred to crops. In addition to thermomorphogenesis, adaptation to high ambient temperature also involves physiological processes such as photosynthetic acclimation, respiration and changes in carbon balance. For discussions of these topics as well as on phenological changes including premature flowering, we refer the reader to reviews elsewhere20,37,
Growth and developmental effects of high temperature
To the best of our knowledge, the term ‘thermomorphogenesis’ was coined by Erwin and colleagues16, in analogy to photomorphogenesis (light-mediated growth), to describe the effects of temperature on plant morphology. Here, we will define it as the suite of morphological changes that together are likely to contribute to adaptive growth acclimation to otherwise detrimental high ambient temperature conditions.
Elongation of the hypocotyl is one of the earliest thermomorphogenic effects seen in seedlings across Arabidopsis accessions in response to high ambient temperature22,
Rosette leaves and cotyledons exhibit marked petiole elongation upon sensing of high ambient temperatures17,
PIF4 is a hub in ambient-temperature signalling
Changes in plant morphology initiated by high ambient temperature and by vegetation shade are very similar55, indicating the possibility of shared signalling elements. This idea led to the identification of the bHLH transcription factor PIF4 as a key regulator of thermomorphogenic phenotypes including hyponasty, hypocotyl and petiole elongation22,29,30,32,56,57. As discussed below, PIF4 (and to a lesser extent PIF5) performs its pivotal function in high-temperature signalling by orchestrating transcriptional changes that subsequently trigger primarily phytohormone-induced elongation responses.
Shortly after shifting plants to high ambient temperature, a notable increase in PIF4 transcript has been observed, triggering thermomorphogenesis22,30,32. But thermomorphogenesis needs to be precisely timed and restrained, in order, for example, to balance elongation growth against biomass production58. A complex circuitry of PIF4 regulation is therefore at play that includes gene expression, epigenetic regulation, protein stability, protein sequestration, promoter access and promoter competition (Fig. 2). This tight control of PIF4 activity and other coordinating factors is indispensable for the integration of various environmental signals into plant morphogenesis and growth control.
Transcriptional regulation of PIF4. Expression of PIF4 itself is rhythmic and tightly regulated by the circadian clock (Fig. 2a)59,
In the light, PIF4 restriction probably involves a similar repression mechanism facilitated at least partially by the bZIP transcription factor LONG HYPOCOTYL 5 (HY5)67,
Post-translational regulation of PIF4 protein levels. In addition to control at the transcriptional level, PIF4 is also subjected to post-translational control. PIF4 interacts with several proteins, which can affect its activity or stability. The name-giving interaction with phytochrome B (phyB) in the light, for example, results in phosphorylation and subsequent ubiquitination followed by proteasomal degradation of PIFs71 (Fig. 2b). The kinase BRASSINOSTEROID-INSENSITIVE 2 (BIN2) has also been shown to phosphorylate PIF4 preferentially in the light, restricting the daytime impact of PIF4 by depleting protein levels72. As high temperature triggers accumulation of phosphorylated PIF4 in red and blue light, however, light-mediated phosphorylation does not necessarily result in degradation of the protein58. Possibly, differential phosphorylation patterns by independent kinases occur in response to distinct stimuli, resulting in different fates of the protein.
Recently, interaction of PIFs with DET1, a repressor of photomorphogenesis, has been shown to stabilize PIFs and counteract their degradation73,74 (Fig. 2b). Whether this process directly contributes to the regulation of PIF activity in response to elevated temperatures remains to be elucidated. However, det1 mutants are impaired in temperature-induced hypocotyl elongation41, which could very well indicate a dual role of DET1 in temperature-dependent PIF regulation through direct interaction/stabilization, and also DET1-COP1-mediated HY5 degradation.
Interaction with other proteins can also sequester free PIF4 protein, preventing its DNA-binding and downstream transcriptional regulation48,58,75. Among these, LONG HYPOCOTYL IN FAR-RED 1 (HFR1), which accumulates in a CRYPTOCHROME 1 (CRY1)-dependent manner, acts as a negative regulator in temperature responses under monochromatic blue light58. This process may also contribute to PIF4 regulation in blue-light-rich white light conditions (Fig. 2b).
In addition, PIF4 access to target promoters seems to be under tight control as well. Here, competition for mutual regulatory DNA-binding sites can occur among PIF4 and HY5, which differentially affects the transcriptional activity of target genes69. As increasing temperatures result in decreased HY5 and increased PIF4 protein levels22,32,69, the alteration in protein ratios can quantitatively affect target gene expression levels.
PIF4-mediated regulation of phytohormones
Phytohormone biosynthesis and signalling genes represent prominent PIF4 targets32, thereby connecting PIF4 activity with the long-known essential role of phytohormones in thermomorphogenesis31 (Fig. 2c).
Auxin and auxin signalling are required and sufficient for PIF4-mediated hypocotyl elongation induced by high temperature, and for other thermomorphogenic responses29,
Increased intracellular auxin levels initiate gene expression changes by means of the TRANSPORT INHIBITOR 1/AUXIN SIGNALING F-BOX proteins (TIR1/AFBs) signalling pathway77. Auxin binding by a co-receptor complex formed by TIR1/AFBs and members of the AUXIN/INDOLE-3-ACETIC ACID (AUX/IAA) protein family results in the subsequent degradation of AUX/IAAs and the initiation of transcriptional auxin responses78. Accordingly, mutants defective in one or more of the partially redundant TIR1/AFBs show reduced temperature-induced hypocotyl elongation31,41.
Among the temperature-inducible auxin response genes are the SMALL AUXIN UP RNA 19–24 (SAUR19–24) and SAUR61–68 subfamilies29,32. Several members of this gene family have been shown to regulate elongation growth, probably by increasing H+-ATPase activity at the plasma membrane79,
In addition to auxin, brassinosteroids (BR) and gibberellins (GA) play crucial roles in high-temperature-induced hypocotyl elongation22,31,
Gibberellin (GA) presence leads to degradation of growth-repressive DELLA proteins that inhibit PIF action in light signalling75,85. Moreover, Stavang and colleagues32 demonstrated a rapid up-regulation of the major GA biosynthesis genes AtGA20ox1 and AtGA3ox1 in Arabidopsis seedlings subjected to elevated temperatures, whereas the prominent catabolism gene AtGA2ox1 was down-regulated. Consistent with these observations, detailed mutant analyses showed that both GA biosynthesis and signalling are required for the promotion of thermomorphogenesis32. This suggests that the GA pathway is more active at high ambient temperatures, putatively as a result of increased GA levels and release of DELLA-dependent PIF4 sequestering. In contrast to auxin, however, the GA pathway seems insufficient to induce thermomorphogenesis, because quintuple della mutant seedlings still show a partial hypocotyl elongation response22,32. Interestingly, GA-mediated cell elongation requires BRs, auxin, BZR1 and PIF4, and it was shown34,83 that DELLA growth repressors directly interact with BZR1 and ARF6. GA presence releases DELLA-mediated repression of BZR1 and ARF6 to allow BAP-module function and subsequent induction of hypocotyl elongation34,83 (Fig. 2c). Hence, GA seems permissive, rather than regulatory, by modulation of PIF4 activity.
Multiple signalling pathways converge at PIF4
Tight regulation of PIF4 and its downstream auxin biosynthesis and signalling targets is required to assure that cooling capacity is achieved, while physiological imbalance and exaggerated elongation growth is prevented. Therefore, several signal transduction pathways converge on PIF4 in addition to the (post-)transcriptional regulatory mechanisms discussed above.
One such pathway involves feedback regulation by AUX/IAA auxin signalling genes (Fig. 2c). Various AUX/IAAs (for example, IAA4 and IAA29) are induced under high ambient temperatures in a PIF4-dependent manner22,31. Auxin-mediated degradation of AUX/IAAs and subsequent release of ARF transcription factors is essential for thermomorphogenesis. Yet, the TIR1/AFB-independent direct and rapid induction of the genes encoding AUX/IAA transcriptional repressors by PIF4 also provides the possibility of a fast and timely attenuation of the auxin stimulus when auxin levels decrease. Consistent with this idea, gain-of-function mutations in several AUX/IAAs (for example, IAA3/SHY2 and IAA19/MSG2) can suppress PIF4-mediated hypocotyl elongation at high temperatures30,47.
A recent study described the involvement of epigenetic silencing of the auxin biosynthesis gene YUC8 to attenuate thermomorphogenesis36. Mutants in the RNA-binding protein FLOWERING TIME CONTROL PROTEIN A (FCA) exhibited increased PIF4 binding to the YUC8 promoter (Fig. 2c). Accordingly, fca mutants displayed increased auxin levels and exhibited enhanced hypocotyl and petiole elongation as well as hyponasty under both control and elevated temperatures36. Furthermore, enhanced levels of the activating epigenetic histone mark H3K4me2 on chromatin of the YUC8 promoter were observed at high temperatures, which was further stimulated in the fca mutant background36. Taken together, the results suggest that PIF4 binds to the YUC8 promoter and stimulates auxin biosynthesis driving thermomorphogenesis shortly after high temperature sensing, followed by PIF4-mediated recruitment of FCA. This leads to removal of activating H3K4me2 marks and subsequent dissociation of PIF4 from the YUC8 locus, resulting in attenuation of thermomorphogenesis36.
Additional regulation of PIF4 may be conferred through HLH factors (Fig. 2c). The non DNA-binding HLH factor PHYTOCHROME RAPIDLY REGULATED 1 (PAR1) attenuates high-temperature-mediated elongation responses through direct inactivation of PIF448, resulting in decreased high-temperature-induced hypocotyl elongation48. Furthermore, the BAP module stimulates the expression of another non-DNA-binding HLH factor PACLOBUTRAZOL RESISTANCE 1 (PRE1)34,83. PRE1 acts as a positive regulator of thermomorphogenesis as part of a module of three HLH/bHLH factors, together with ILI1 BINDING BHLH1 (IBH1) and HOMOLOG OF BEE2 INTERACTING WITH IBH1 (HBI1)34,44,83. Sequestration of IBH1 by PRE1 enables the binding of HBI1 to the promoters of EXPANSIN genes44, promoting cell-wall loosening and hypocotyl elongation (Fig. 2c). Consistent with this model, high-temperature-induced hypocotyl elongation is severely reduced in transgenic lines displaying reduced PRE1/HBI1 or enhanced IBH1 levels34,44,83.
Modelling-based integration of light, circadian and temperature signals in the control of thermomorphogenesis. The studies outlined above illustrate that PIF4 associates with a number of proteins that collectively integrate multiple environmental and endogenous stimuli to control thermomorphogenesis. Although we already have detailed knowledge of some molecular events, we are still some way from understanding how the network operates at a whole system level. When striving to do this, lab-to-lab variation in experimental regimes, and limited access to quantitative data, can provide additional obstacles. Thus, linking new and published data to gain a comprehensive understanding of thermoregulation is not a trivial process. Despite these constraints, mathematical modelling has emerged as a valuable approach to learning how complex biological systems work. Modelling provides a formal means to consolidate knowledge, challenge our current understanding and derive new and experimentally testable hypotheses. Recently, a combination of modelling and experimental approaches was successfully applied to address the complex regulatory circuitry underlying morphogenesis by connecting the circadian clock, light and temperature to identify new regulators and interconnections and to explain regulatory switches in response to multiple conflicting stimuli27,43,86.
Initial groundwork in this area was laid by Rausenberger and colleagues87, who constructed the first kinetic model for light signalling. This model captured key aspects of phyB photochemistry including photoreceptor protein dynamics to hypocotyl length87,88. The model also highlighted the combined network features that were required to deliver fluence-rate dependency of phyB. A more recent study extended the Rausenberger87 model to incorporate PIF control of hypocotyl elongation43. This revised model provided a framework to understand how changes in the light and temperature environment alter signalling through the phyB–PIF circuit. The study revealed that temperature has a strong impact on how light regulates hypocotyl elongation by showing that fluence-rate-dependent hypocotyl elongation is attenuated at 22 °C compared with 17 °C. Furthermore, at 27 °C increasing fluence rates do not inhibit but instead promote elongation above a low irradiance threshold. This implies that temperature can completely switch the mode of light action, possibly by increased photoconversion between active Pfr and inactive Pr forms at higher fluence rates, resulting in less efficient phyB signalling. This scenario predicts that phyB would be less effective in degrading PIF proteins at increased fluence rates at 27 °C. However, this is not the case, as a strong fluence-rate-dependent depletion of PIF4 (and PIF3) protein levels was observed at both 22 °C and 27 °C (ref. 43). Model analysis provided an alternative hypothesis: that fluence-rate-dependent factors are required to modulate PIF activity. At moderate temperatures these factors suppress PIF action, but at higher temperatures they activate PIFs. This hypothesis was partially validated, as HY5 was shown to be a strong PIF suppressor at cooler temperatures, particularly as fluence rates increase43,69. Nevertheless, the molecular or biochemical entity that mediates light activation of PIFs at higher temperatures has yet to be determined.
Although such steady-state hypocotyl models provide useful formats to conduct network structure–function analyses, rhythmicity of PIF-mediated hypocotyl elongation requires integration of the circadian clock and natural photoperiods59,
Based on the described examples, combining modelling and experimental approaches has proved to be important in deciphering biological complexity. The highlighted studies27,43,68 provide conceptual frameworks to understand how the mode of PIF4 control by light is switched by temperature, and to understand the temperature-dependent nocturnal rise in PIF4 transcription in a diurnal cycle. The study by Seaton and colleagues27 also provides a systems-level understanding of how temperature and photoperiodic signals integrate to control growth.
Chromatin-level regulation of thermomorphogenesis
Temperature influences virtually every biological process, and a key feature of investigating the impact of temperature on any given organism is that passive, thermodynamic effects of temperature on biomolecules need to be separated from active thermal perception and signalling89. Among the processes that are tightly controlled by temperature are gene transcription and mRNA degradation. Sidaway-Lee and colleagues90 noted that both transcription and mRNA decay rates passively increased in response to higher ambient temperatures in Arabidopsis. In an effort to dissect active and passive thermal regulation, they found that active temperature-directed changes in mRNA abundances could be assigned to temperature-mediated regulation of transcription, rather than to mRNA decay90. The authors next determined which epigenetic modifications were related to temperature-mediated transcriptional regulation and found that H3K27me3 was associated with genes exhibiting both high and low levels of temperature-dependent transcriptional regulation. This epigenetic mark was depleted from genes showing passive temperature-mediated regulation only90. Global changes in several other epigenetic marks, including H3K4me3, H3K9Ac and DNA methylation, were not implicated in active thermoregulation of gene expression90, but contribution of these marks to the expression of specific thermomorphogenesis-regulating genes cannot be excluded. The prominent role for epigenetic modifications in thermomorphogenesis control was recently supported by the example of FCA-mediated H3K4me2 removal from the YUC8 promoter, described above, which probably restricts PIF4 binding and thereby attenuates thermomorphogenesis36.
In addition to epigenetic modifications, chromatin remodelling has a prominent role in thermomorphogenesis. ACTIN RELATED PROTEIN 6 (ARP6) controls H2A.Z-nucleosome incorporation into chromatin91, and plants carrying mutations in ARP6 display several aspects consistent with a constitutive thermomorphogenic response, such as longer hypocotyls and petioles and a transcriptome profile typical for high ambient temperatures, even at lower growth temperatures35 (Fig. 2c). This implies a role for H2A.Z-containing nucleosomes in thermal regulation of transcription. H2A.Z-nucleosomes are highly enriched at the beginning of genes at the +1 position, adjacent to the transcription start site. For some genes, such as HEAT SHOCK PROTEIN 70 (HSP70), it has been shown that the occupancy of the +1 H2A.Z-nucleosome is rate-limiting for expression. Consequently, HSP70 was more highly expressed in the arp6 background compared with wild type at low ambient temperatures. Based on these observations, it was hypothesized that the observed high-temperature-induced H2A.Z eviction may provide thermal information to the cell by allowing better accessibility for transcriptional regulators that ultimately orchestrate thermomorphogenesis35. H2A.Z eviction therefore seems to enable temperature-dependent expression for at least some — and possibly many — genes. A key question is whether H2A.Z-nucleosome eviction is a direct response to temperature (suggesting that it is thermosensory) or whether it is mediated indirectly, for example by means of a temperature-responsive chromatin remodelling factor. Notably, however, arp6 mutants still show an increase in hypocotyl elongation at warmer temperatures, suggesting that H2A.Z-nucleosomes themselves do not transmit all temperature information.
Future challenges, knowledge transfer and conclusions
Numerous open questions about temperature signalling and response networks remain to be resolved before comprehensive understanding of thermomorphogenesis regulation is reached. It is likely that many relevant thermomorphogenesis regulators remain to be identified and their signalling hierarchies need to be investigated to understand how multiple conflicting signals are integrated in coordinated plant growth and development. Importantly, the thermomorphogenesis mechanisms described here are probably operating across a broad range of non-damaging temperatures, beyond the somewhat rigid temperature range of ∼20 to ∼29 °C normally used in thermomorphogenesis research in Arabidopsis (Table 1). To fully understand plant acclimation to warmer temperatures, a broader temperature range needs to be taken into account. Above all, however, the exact mechanisms by which small changes in ambient temperature are sensed remain enigmatic. H2A.Z eviction and subsequent changes in chromatin suggest a possible temperature-sensing mechanism, but this needs to be confirmed. The data are consistent with a model whereby H2A.Z-nucleosomes at the transcriptional start site35 and/or the gene body90 may be rate-limiting for the expression of other key genes in the thermomorphogenesis pathway, such as PIF4, or PIF4 targets. Alternatively, the enhanced elongation phenotype of arp6 may arise from a parallel pathway.
Our currently rather limited understanding of ambient temperature perception is in contrast to many other signal transduction pathways. This may be due in part to the involvement of numerous processes, prohibiting the elucidation of a ‘temperature receptor’. Among these, temperature effects on transcriptional rates, protein–protein interaction, protein turn-over, changes in subcellular localization and changes in rates of metabolism might intricately contribute to altered physiological read-outs of known and unknown signalling processes. The recent identification of natural CRYPTOCHROME 2 alleles and their role in thermomorphogenesis50 emphasizes that the identification of additional, as yet unknown rate-limiting and crucial signalling hubs within this network of sensors and response elements constitutes a considerable challenge, as does experimental design and interpretation. In this respect, the role of metabolism in thermomorphogenesis deserves more attention. Carbon starvation occurs in plants shifted to high ambient temperatures, and this correlates with thermomorphogenesis phenotypes54. Moreover, PIFs including PIF4 are required for sucrose-induced hypocotyl elongation, and PIF5 has been shown to be stabilized by sucrose92,93. Sugars induce auxin biosynthesis by stimulating auxin biosynthesis genes94, an effect that might potentially be counteracted or enhanced by PIFs depending on specific growth conditions. Such data underscore that temperature, light, sugars, PIFs and auxin are part of a complex and not yet well-understood circuitry integrating environmental and metabolic cues into a coordinated growth response. Genetic analysis can be used to provide insight into the complex molecular networks underlying thermomorphogenesis, but major advances will require the combination of wet-lab genetic, physiological and biochemical approaches together with in silico modelling of dynamic structural plant phenotypes and the underlying genetic circuitries.
One aspect that needs particular consideration is the interaction of thermomorphogenesis with other environmental stresses. The relationship with drought deserves more attention, as thermomorphogenesis assists cooling by enhancing transpiration, which is only favourable under well-irrigated conditions17. Water is already growth-limiting in many parts of the world95, and high temperatures and drought often occur simultaneously, suggesting that thermomorphogenic acclimation is not beneficial and can be even detrimental in these conditions. Accordingly, when combined, high temperatures and drought result in more severe inhibition of growth in plants than observed for only one individual stress53. Both stresses have impacts on growth through partly separate and partly parallel mechanisms that become additive when experienced together. Therefore, it is important to assess the contribution of thermomorphogenesis-regulatory networks on plant acclimation to other stresses and their combinations.
Climate change has already caused large-scale changes in distribution and behaviour of wild species, and unseasonably hot weather led to global disruptions in crop productivity, for example in 2003 and 2012. Further temperature increases during this century are forecast to exacerbate these problems3,
Potential targets for mining of favourable natural alleles could include the receptor-like kinase ERECTA, which was recently shown to play a critical role in high-temperature stress tolerance97. ERECTA probably acts by protecting against temperature-induced cellular damage, since overexpression of ERECTA conferred high temperature tolerance on Arabidopsis, tomato and rice in greenhouse and field conditions, without compromising growth and yield. Key thermomorphogenesis regulators such as PIF4 and elements of the EC are also good candidates. Allelic variation in ELF3, ELF4, LUX and other clock components, for example, has contributed to the domestication of several crop species in terms of flowering time adaptation98. Based on the experimental work in Arabidopsis, allelic variation of EC components can significantly affect thermomorphogenesis under controlled environmental conditions26,28. It remains to be investigated whether these alleles also cause differential temperature responses under natural environmental conditions, and whether similar differences can be observed in different crop species. On the bright side, the observation that H2A.Z-nucleosome-mediated temperature responses in the monocot model species Brachypodium distachyon99 are similar to those observed in the dicot Arabidopsis suggests that at least some of the main molecular circuitries underlying thermomorphogenesis are functionally conserved.
Meeting future challenges to plant productivity imposed by globally increasing temperatures will require basic research in model plant species as well as applied approaches in crops. Integration of these ends of the spectrum will require directed efforts from the academic plant research community and private companies. Further development of thermomorphogenesis as a research area could ultimately provide efficient and timely leads for the initiation of appropriate breeding efforts to generate much-needed thermo-tolerant crops.
We thank J. Bellstädt for graphical support and J. Trenner for the photographs in Fig. 3 (Leibniz Institute of Plant Biochemistry, Halle (Saale), Germany). This work was supported by a grant from the Deutsche Forschungsgemeinschaft to M.Q. (Qu 141/3-1) and NWO VENI grant 863.11.008 to M.v.Z.