Introduction

Phytochromes are present in bacteria, cyanobacteria, fungi, algae, and land plants, and while in all cases they can perceive light, their photochemical properties vary largely among phyla1,2. In this review we will focus on land plant phytochromes, and in particular on Arabidopsis, for which we have a better understanding of the underlying molecular mechanisms. In land plants, phytochromes are red and far-red light receptors that exist in two forms. They are synthesized in the inactive Pr state, which upon light absorption converts to the active Pfr conformation. Pfr is inactivated upon far-red (FR) light absorption or through thermal relaxation, which depends on temperature, a process known as dark or thermal reversion. Phytochromes act as dimers, resulting in three possible phytochrome species: Pr–Pr, Pfr–Pr, and Pfr–Pfr3 (Fig. 1a). Pr and Pfr have different absorption maxima, but due to overlapping spectra both conformers are always present in the light while only prolonged darkness returns all phytochrome to Pr (Fig. 1b)3. Given that phytochrome responses depend on the proportion of Pfr conformers, signaling is influenced by a combination of light quantity, color, and temperature3,4,5.

Fig. 1
figure 1

Control of phytochrome activity. a Factors controlling phytochrome activity. Phytochromes exist in two conformations, Pr and Pfr, the latter being the active form. They exist as dimers so three species can be found. Each monomer can be activated by red light (R) and inactivated by far-red light (FR) or by thermal reversion, a process that depends on temperature (T). At least in the case of phyB, Pfr in heterodimers reverts much faster than that in homodimers, allowing phyB to perceive temperature both during the day and during the night. b Plant phytochrome absorption spectra of the Pr and Pfr conformations. In dark-adapted seedlings phytochromes are in the Pr form. Upon a saturating R pulse, due to overlapping absorption spectra of Pr and Pfr, only 87% of Pfr is achieved. c Action spectra for phyA and phyB in the control of hypocotyl elongation. Data from Klose et al.3. Fluence rate response curves are measured at different wavelengths and fluence rate that leads to 40% inhibition compared with dark control is determined. In order to specifically determine action spectra for phyA and phyB, for phyB the curve was performed with phyB-GFP/phyAphyB seedlings, and for phyA using phyB-5 seedlings. Values are relative to the response obtained at the most efficient wavelength in each case

Upon perception of inductive wavelengths, activated phytochromes together with blue (phototropins, cryptochromes, and Zeitlupes) and UV light receptors (UVR8) control plant physiology and development6,7. Once the seed is imbibed, active phytochromes promote germination6. When a seedling grows in the soil it adopts an etiolated morphology characterized by fast-growing hypocotyls and closed apical hook, which maximizes the chance to reach the surface. Once it reaches the light, activated phytochromes promote de-etiolation: hypocotyl growth slows down, the apical hook opens, cotyledons expand, and chloroplasts develop6. This initial response to light occurs even in poor light conditions as encountered under deep shade where blue and red light (R), which are essential for photosynthesis, are scarce6. Unfiltered sunlight has approximately equal amounts of R and FR resulting in a ratio of R to FR (R/FR) slightly above 1 and high phytochrome activity8. However, in environments with high plant density the R/FR drops since green tissues absorb mainly R and blue light and transmit or reflect FR. This results in reduced phytochrome activity triggering the shade-avoidance response in green seedlings. Stem and petiole elongation are promoted, leaves change their position and anatomy, root architecture is altered, while senescence is promoted8,9,10. In these conditions, plants allocate more resources to growing aerial parts11 and change their metabolism12. Increasing temperature promotes similar architectural changes as shade13 and a subset of these temperature responses depend on phytochromes4,5,14. Moreover, phytochromes are important to entrain the circadian clock and for the perception of the photoperiod6,15,16. These signals in addition to temperature and light quality control flowering time6,17,18,19.

In this review we summarize the current knowledge on early phytochrome signaling mechanisms in Arabidopsis with an emphasis on de-etiolation and shade avoidance, the photomorphogenic processes for which most information is available. Other phytochrome-controlled developmental processes are discussed in the context of long-distance signaling. Comparisons between Arabidopsis and other species for which molecular or genetic data are available is presented in a separate chapter. Due to space constraints we could not systematically refer to the older primary literature but we suggest more focused review articles where historical perspectives can be found.

The phytochrome family in Arabidopsis

Most land plants possess several phytochromes, and in most angiosperms three groups can be identified: phytochrome A (phyA), phyB, and phyC2,20. Arabidopsis has one member in the phyA and phyC groups while the phyB class is composed of phyB, phyD, and phyE6,14,21,22 (Table 1). Each phytochrome has different roles and their relative contributions vary depending on the environmental conditions and developmental stage of the plant (Table 1)14,21,22,23,24,25. Angiosperm phytochromes are further classified into two categories according to their capacity to trigger responses to specific light signals. Type I, represented by phyA in Arabidopsis, are light labile and allow germination and de-etiolation when light is scarce (Very Low Fluence Response or VLFR) or when the R/FR is very low (FR-High Irradiance Response, FR-HIR)6. Such conditions are encountered under a thin layer of soil and in deep shade. Type II phytochromes (phyB–phyE in Arabidopsis) are light stable but require a substantial fraction of Pfr to promote signaling. Therefore, they are active in more open environments over a wide range of irradiances where the R/FR is relatively high (low fluence response, LFR)6. Hence, despite very similar absorption spectra type I and type II phytochromes have different action spectra3 (Fig. 1b, c, see Box 1). The contribution of each phytochrome to light-regulated plant development can be explained by their different expression levels, protein structure, protein stability, and photochemical properties, as well as the mechanisms underlying their nuclear internalization3,6,14,24,26,27 (see Box 1 and Table 1 for details). In addition, each phytochrome has a different contribution to temperature responses with phyB having a major role. Active phyB is a Pfr–Pfr homodimer and the rate of thermal reversion of the Pfr–Pr heterodimer is much faster than that of the active Pfr–Pfr homodimer3 (Fig. 1a). The occurrence of this fast reversion allows phyB to perceive temperature not only during the night but also in the light4,5. According to modeling approaches temperature-enhanced phyB inactivation is expected to significantly antagonize light activation during cloudy days or in shaded environments28. In contrast, phyA is not expected to contribute to such temperature responses, as its thermal reversion rates are much slower than those of phyB at least in some accessions29. In addition to their individual roles, type II phytochromes can form heterodimers24,30. Importantly, in Arabidopsis phyC mostly works as a heterodimer with other phytochromes and the phyB–phyC heterodimer is particularly important to inhibit flowering in non-inductive photoperiods (Table 1)14,23. Despite their important roles throughout the plant life cycle and the strong phenotypes of higher order phytochrome mutants, Arabidopsis can complete its life cycle in the lab without phytochromes14,31,32.

Table 1 Functional diversification of the phytochrome family in Arabidopsis

Structure of phytochrome molecules and link to function

Plant phytochromes are dimeric, each monomer consisting of a ~1150 aminoacids covalently bound to its chromophore, a linear tetrapyrrole named phytochromobilin (PΦB) (Fig. 2) (for a comprehensive review on phytochrome structures we suggest Burgie et al.1). Crystal structures of the phyB N-terminal photosensory module (PSM), solid-state NMR analysis of Oat phyA PSM in the Pr and Pfr form, and structures of bacterial and cyanobacterial phytochromes are available and inform our mechanistic understanding of plant phytochrome structure and how it pertains to subsequent signaling steps33,34,35.

Fig. 2
figure 2

Structural domains of phytochromes and their role in perception of environmental signals and downstream signaling. NTE N-terminal extension, PSM photosensory module, CTM C-terminal module. Modified from Burgie et al.44

The apoprotein can be divided in the N-terminal PSM, which consists of the N-terminal extension (NTE), for which structural information remains scarce, and three structurally related domains Period/Arnt/SIM (PAS), cGMP phosphodiesterase/adenylyl cyclase/FhlA (GAF), and a phytochrome-specific domain (PHY) and a C-terminal module (CTM) comprising two PAS domains and a histidine kinase-related domain (HKRD)1 (Fig. 2). In the following paragraphs we will describe each part of the protein and its role in light-regulated protein interactions.

The chromophore is bound covalently to a conserved cysteine in the GAF domain, which has intrinsic chromophore lyase activity1,35. Light perception triggers a Z to E isomerization around the C15–C16 double bond of PΦB, which leads to a cascade of structural modifications in the protein1,34,35. PΦB interacts with the GAF domain covalently, but also has non-covalent interactions with residues from the NTE, PHY, and PAS domains. Hence all domains of the PSM are affected by the light-induced conformational change of the chromophore1,35. In particular, the hairpin (or tongue) structure in the PHY domain undergoes a beta-sheet to alpha-helix transition resulting in broader structural rearrangements1,35,36. This affects the figure-of-eight knot present between the PAS and GAF domains pulling them together, while separating the PHY domains from each monomer1,34,35. Initial conformational rearrangements in the PSM lead to structural changes also affecting the CTM. Ultimately, the light-induced Pfr conformer interacts with other proteins necessary for nuclear localization and signaling partners such as transcription factors of the PHYTOCHROME INTERACTING FACTOR (PIF) family and ubiquitin E3 ligase complexes37,38,39,40,41,42.

The NTE is longer in plant than bacterial phytochromes and variations in its sequence explains some differences between type I and type II phytochromes. In particular, type II phytochromes have a longer NTE, and in Arabidopsis replacing phyB NTE-PAS with that of phyA confers two features to this hybrid phytochrome that are unique to phyA; the capacity to accumulate in the nucleus in response to FR and rapid degradation in R43. This part of the protein is also involved in the control of thermal relaxation1,44 (Fig. 2). Both in phyA and phyB phosphorylation of the NTE modulates this reaction45,46,47,48,49. Several residues from the GAF and PHY domains control Pfr stability, thermal relaxation, and hence the ability of phyB to respond to light and temperature cues1,5,44. Finally, the PHY domain comprises key determinants distinguishing phyA from phyB, which may relate to different rates of thermal relaxation20,43. In summary, despite strong conservation several domains of the PSM contribute to subfunctionalisation of type I and type II phytochromes.

The presence of the HKRD suggested that phytochromes transduce their signal through the CTM. Although many bacterial and cyanobacterial phytochromes have a C-terminal histidine kinase domain and act as light-regulated histidine kinases, plant phytochromes are not histidine kinases and their role as Ser/Thr kinases remains contentious50,51. Actually, the PSM fused to a nuclear localization signal and a dimerization sequence is sufficient to restore most phyB functions, pointing to key signaling functions of the PSM52. Major roles of the plant phytochrome CTM are dimerization, nuclear import, and localization to sub-nuclear structures known as photobodies44,52,53. However, it was recently shown that the C-terminal part of phyB also engages in light-regulated interactions and regulation of PIF activity39,40. Moreover, the activity of the CTM is controlled by post-translational modification with SUMOylation limiting the ability of active phyB to interact with downstream signaling targets thereby limiting light responses54. In addition, the CTM modulates active (Pfr) phytochrome levels with the HKRD inhibiting the Pr–Pfr photoconversion while the PAS–PAS promotes thermal reversion44 (Fig. 2). Hence, while the division of plant phytochromes into PSM and a CTM helps describing the molecule, both parts of the photoreceptor contribute to the regulation of active Pfr levels and downstream signaling activities.

Light-controlled nuclear import of phytochromes

Upon light perception all five phytochromes in Arabidopsis are translocated from the cytosol to the nucleus. This is a key step that is required for almost all known phytochrome responses55,56,57,58, it is broadly conserved in all land plants studied and also in marine algae57,59,60. The molecular mechanisms as well as the dynamics of this transport are distinct for type I (phyA) and type II (mostly studied for phyB) phytochromes, which contributes to the unique signaling features of phyA27,57 (Box 1). While phyA nuclear import relies on interaction with NLS-containing proteins, how phyB enters the nucleus remains poorly understood57.

phyA enters the nucleus very quickly even in conditions leading to very low levels of phyA in the active conformation (PfrA) (VLFR and FR-HIR), hence the heterodimer PfrA–PrA is probably transported to the nucleus57. This translocation depends on the interaction between PfrA and the NLS-containing proteins FHY1 and FHL56. The NLS in FHY1 is recognized by importin α, allowing nuclear import of the complex61. According to modeling approaches, once in the nucleus phyA is released from FHY1 and has to undergo another cycle of photo-activation to initiate light responses27. FHY1 exits the nucleus through the exportin route, and its import or export depend on its phosphorylation state61. FHY1 is rapidly phosphorylated in R light thereby limiting phyA nuclear localization and providing an attenuation mechanism for phyA responses to R62. This import mechanism has deep evolutionary roots as an FHY1-like protein is present and required for phytochrome responses in the basal land plant Marchantia polymorpha63 and FHY1-mediated phytochrome nuclear import operates in the moss Physcomitrella patens60.

Currently there is no consensus about the molecular mechanisms for phyB–E nuclear localization57. phyB nuclear import is induced by the same signals capable to induce the LFR and both PfrB–PfrB and PfrB–PrB dimers can enter the nucleus3. It was proposed that phyB nuclear import is mediated by light-induced NLS-unmasking, but no canonical NLS sequence is recognizable in phyB53. phyB import is not controlled by FHY1 or FHL, but may occur following interaction with NLS-bearing PIF transcription factors64. Although less studied some information about phyC–E localization is available in Table 1. While the mechanisms of type II phytochromes nuclear import remain poorly understood, it is clear that the distinct modes of type I versus type II phytochrome nuclear import are an important feature underlying photoreceptor subfunctionalization.

Phytochrome signaling elicited through light-activated interactions

The light-induced Pfr conformer selectively interacts with several classes of transcription factors and with ubiquitin E3 ligases, which control the stability of transcriptional regulators. Collectively these light-regulated interactions result in rapid and global transcriptional reprograming41,58,65,66. We will start by describing the consequences of phytochrome–transcription factor interactions (Fig. 3) before a brief presentation of how phytochromes regulate ubiquitin E3 ligases (e.g. COP1/SPA, DDB1/DET1) and convergence points between both signaling branches (Figs. 3 and 4).

Fig. 3
figure 3

Simplified mechanism for phytochrome control of transcription factors in different light environments. a Below the soil surface during etiolated growth. For simplicity we consider that phytochromes remain inactive (Pr) below the soil surface, which results in accumulation of transcription factors PIFs, EIN3, and ARFs and subsequent induction of etiolation and auxin response genes. The COP1/SPA ubiquitin E3 ligase accumulates in dark and leads to proteasome-mediated degradation of HY5, a transcription factor that suppresses the expression of genes required for etiolation and induces expression of genes required for de-etiolation. b During de-etiolation upon light perception. Light perception activates phytochromes (Pfr) which promote de-etiolation by directly inhibiting PIFs and EIN3, and indirectly inhibiting ARFs by stabilizing Aux/IAA proteins. The Pfr form of either phyA or phyB interacts with SPA proteins, resulting in inhibition of COP1/SPA. This results in stabilization of HY5 leading to induction of de-etiolation related gene expression and repression of etiolation genes. c In a deetiolated plant in response to shade (reduced R/FR). Low R/FR in shade reduces the fraction of active phytochrome (Pfr/Ptot). PIFs accumulate and induce growth-promoting gene expression. In addition, PIFs induce a negative feedback loop exemplified by HFR1 expression. HFR1 (and other HLH proteins) binds to PIFs forming non-DNA-binding heterodimers. COP1/SPA is also involved in this loop by leading HFR1 to proteasome-mediated degradation. Arrows indicate positive regulation, blunt-ended arrows indicate negative regulation, and dotted-lined arrows indicate nucleo-cytoplasmatic relocalization

Fig. 4
figure 4

Mechanisms of phytochrome-mediated regulation of transcription factors. a From top to bottom, sequential steps by which Pfr inhibits PIFs. Top: PfrA interacts with PIF1 and PIF3 while PfrB interacts with PIF1–PIF8. Middle left: for PIF1, 3 and 4 phytochrome inhibits DNA binding. Middle right: Interaction with Pfr leads to phosphorylation of PIFs. Many kinases have been found to phosphorylate PIFs (see text) with PPKs phosphorylating PIFs in response to light. Bottom: after light-induced phosphorylation, PIF3 is degraded by LRBs and EBFs with phyB co-degradation occurring in the LRB-mediated process (left, center), phosphorylated PIF7 interacts with 14-3-3 proteins and remains in the cytoplasm (right). b Other mechanisms of transcriptional control by phytochromes. Left: PfrA and PfrB interact with SPA and inhibit the COP1/SPA complex. Center: PfrB interacts with EIN3 to promote ERF-mediated EIN3 degradation. Right: PfrA and PfrB interact with Aux/IAA to prevent their degradation by SCFTIR1/AFB. c Patterns of PIF abundance depending on the developmental state and growth conditions. In etiolated seedlings PIFs accumulate to high levels, promoting etiolated growth. Upon light exposure, PIFs are rapidly degraded in a phytochrome-dependent manner, with half-lives of ~5 min for PIF1 and PIF5, and ~10 min for PIF3 and PIF4 (left). In contrast, in light-grown seedlings PIFs are under strong transcriptional control, allowing accumulation of the protein even in conditions when phytochrome activity is predicted to be high (right), SD (short days), LD (long days)

PIFs are members the basic helix–loop–helix (bHLH) TF-superfamily that selectively interact with light-activated phytochromes41,65,67. Pfr interacts with PIFs through short domains located towards their amino-terminus. These short sequences are known as APB for Active PhyB binding and APA for Active PhyA binding domains. Arabidopsis has eight PIFs (PIF1, PIF2/PIL1, PIF3-PIF8) with an APB motif, whereas an APA motif is only present in PIF1 and PIF3 (refs. 41,65) (Fig. 4a, upper panel). The phy-PIF signaling module originated early in evolution since the liverwort Marchantia polymorpha has a single phytochrome and a single APA-containing PIF68. PIF–Pfr interaction results in PIF inactivation, which occurs through several mechanisms41,65,68,69 (Fig. 4a). First, phyB–PIF interaction results in blocking the DNA-binding capacity of several PIFs (shown for PIF1, PIF3, PIF4 while this was not reported for other PIFs)39,40,70. Second, interaction with the photoreceptor leads to rapid PIF phosphorylation which for most PIFs is followed by ubiquitination and proteasome-mediated degradation. PIF7 is an exception in this regard as phosphorylation facilitates its interaction with 14-3-3 proteins leading to cytoplasmic retention of the transcription factor41,65,71 (Fig. 4a). Moreover, PIFs are under strong transcriptional control72. This allows PIF accumulation during the day, even when phytochromes are active18,73 (Fig. 4c, right panel). In summary, light-activated phytochromes inhibit PIFs through multiple mechanisms, which presumably contribute to optimal light regulation of these transcription factors across a variety of light conditions40 (Fig. 4a, c).

The phytochrome-PIF regulon is an initial step underlying photoreceptor’s control of a wide array of physiological and developmental responses18,19,41,58,65 (Figs. 3 and 4). However, it was later discovered that PIFs regulate additional processes in response to other external and internal cues acting as signaling hubs41,65. This may explain why PIF phosphorylation and subsequent proteasome-mediated degradation is so complex41. As mentioned above, plant phytochromes contain a C-terminal HKRD but surprisingly a recent report concluded that the PSM of phytochromes (including phyB) directly phosphorylates PIFs50. However, phyB-mediated PIF3 phosphorylation was not observed in a subsequent study and based on the phyB PSM structure it is unclear how it may act as a protein kinase51. Moreover, while both the PSM and CTM of phyB bind and inhibit PIF3, it is binding to the CTM which results in PIF3 degradation, which typically follows PIF phosphorylation38,39,74. Members of Photoregulatory Protein Kinases (PPK1–4; formerly called MUT9-Like Kinases (MLKs)) are the best-characterized kinases phosphorylating PIF3 in a light-induced fashion41,51 (Fig. 4a). These protein kinases are recruited to the nuclear PfrB–PIF3 complex leading to PIF3 phosphorylation and subsequent degradation51. This study reports a large overlap between PIF3 phosphorylation sites determined in vivo and following in vitro phosphorylation by PPK1. Interestingly, phyB enhances the protein kinase activity of PPKs towards PIF3 thereby potentially acting as a pseudokinase rather than a genuine protein kinase51. While the role of PPKs in PIF3 phosphorylation and subsequent degradation is supported by mutant analysis, the growth phenotype of ppk mutants is somewhat counter-intuitive. This might be due to a broader role of these protein kinases including light-induced phosphorylation and subsequent degradation of phyB and cryptochrome 2 (refs. 51,75). However, precisely delimiting PPK function in phytochrome signaling is hampered by the essential nature of the PPK gene family51. Additional protein kinases have been implicated in PIF phosphorylation and subsequent proteasome-mediated degradation including BIN2, CK2, and MPK6 (refs. 73,76,77,78) (reviewed in ref. 41). BIN2 phosphorylates and regulates the stability of PIF4. However, this does not specifically pertain to phytochrome-mediated PIF regulation but rather to the PIFs-brassinosteroid signaling crosstalk, which is important for plant growth regulation65,73,79,80.

PIF ubiquitination and subsequent proteasome-mediated degradation is equally complex involving several CULLIN (CUL) RING UBIQUITIN LIGASEs (CRLs)74,81,82,83,84,85. Five different families of substrate recognition components have been implicated in the ubiquitination of different PIFs through three members of CRLs: CUL1, CUL3, and CUL4. PIF1 is ubiquitinated by CUL4 COP1/SPA and CUL1CTG10; PIF3 by CUL3LRB1/2/3 and CUL1EBF1/2; PIF4 by CUL3BOP1/2; and PIF5 by CUL4COP1/SPA74,81,82,83,84,85. LRB1-3 and EBF1/2 control light and phytochrome-dependent ubiquitination and subsequent degradation of PIF3 (Fig. 4a, lower panel). Interestingly, LRBs are preferentially involved in strong-light-induced PIF3 degradation in a process resulting in PIF3-phyB co-degradation, while EBFs are more important in response to low light and target PIF3 but not phyB74,81. Moreover, light-regulated PIF3 abundance is controlled transcriptionally by EIN3, which is itself destabilized by light through phyB-promoted interaction with CUL1EBF1/2 (ref. 86) (Fig. 4b, middle panel). The role of other E3 ligases in regulating PIF abundance is not as clearly light and phytochrome-regulated. COP1/SPA’s role is complex because the most striking feature of cop1 mutants is the reduction of PIF abundance in the dark, suggesting that this E3 ligase stabilizes PIFs in darkness41,84,87. COP1 controls PIF3 levels in etiolated seedlings indirectly through BIN2 kinase inhibition, thereby limiting BIN2-mediated PIF3 destabilization73,77. COP1 was also implicated in the light-induced degradation of the light-labile phyA. However, subsequent studies showed that when grown in more natural conditions (on soil) phyA degradation occurs normally in cop1 mutants demonstrating the difficulty of interpreting the rather pleiotropic cop1 phenotype88. BOPs control PIF4 ubiquitination and subsequent degradation but although the process is more efficient in the light it is not strictly light dependent and also contributes to the regulation of PIF4 abundance in darkness82. Finally, CTG10 was recently shown to control light-modulated PIF1 abundance in seeds and seedlings but the underlying mechanism remains to be established85. The involvement of multiple PIF degradation systems is intriguing and may suggest specific mechanisms dedicated to particular PIFs, cell types, organ, and/or stimuli regulating PIF abundance.

More recently, direct phytochrome control of transcription factors was extended to additional families. This includes phyB-EIN3 interaction, which facilitates EIN3-SCFEBF1/2 complex formation, EIN3 ubiquitination, and degradation86 (Figs. 3b and 4b, middle panel). There is some evidence for inhibition of the bHLH protein BES1, a central player in brassinosteroid signaling, following interaction with phyB89. Finally, several Aux/IAA regulators of auxin-controlled gene expression are modulated through interaction with phyA or phyB90,91 (Figs. 3b and 4b, right panel). In these last examples, Aux/IAAs were protected from SCFTIR/AFB auxin-regulated proteolytic degradation following interaction with the phytochromes90,91. Phytochrome-regulated PIF activity regulates plant hormone signaling by directly targeting the expression of key hormone biosynthetic and signaling genes17,66,92,93,94,95,96. This enables light-regulated hormonal responses which in some cases feedback on PIF activity. For example, phytochrome regulation of gibberellic acid (GA) levels controls the stability of the DELLAs family of growth-suppressing proteins97. Light-independent interaction between DELLAs and PIF1, PIF3, PIF4, and PIF5 leads to proteasome-mediated PIF degradation independently of phyB and LRBs98. Upon interaction with DELLAs PIF3 and PIF4 are also prevented from binding to DNA99,100. These mechanisms contribute to an important and complex crosstalk between light and hormonal-mediated growth and developmental responses17,41,60,65.

Phytochromes also exert a strong effect on gene expression through light-activated inhibition of ubiquitin E3 ligases in particular COP1/SPA and DDB1/DET1 (Figs. 3 and 4b, left panel)42,58. Inhibition of COP1/SPA is initiated by selective interaction of the Pfr form of either phyA or phyB with SPA proteins101,102. The mechanisms underlying photoreceptor-mediated inhibition of COP1/SPA and the main targets of this regulation are described in recent reviews42. COP1 and SPA proteins are present in green algae and all land plants but the central role of COP1/SPA in light signaling was proposed to originate in the ancestor of angiosperms103,104. We will briefly illustrate the role of COP1/SPA in phytochrome signaling and links between COP1/SPA and PIF-regulated transcription. The bZIP transcription factor ELONGATED HYPOCOTYL 5 (HY5) is a prominent target of this regulation. HY5 is stabilized following light stimulation and is an important regulator of light-controlled development (Fig. 3b). Interestingly, HY5 and PIFs share many target genes but control them in antagonistic fashion. This system allows for repression of photosynthetic genes in darkness followed by their light activation by a light-controlled exchange of major transcription factors105,106. Some members of B-box (BBX) protein zinc-finger transcription factors, including CONSTANS, a major determinant of photoperiod-induced flowering, are controlled by COP1 and modulate numerous aspects of photomorphogenesis42,107,108,109. Finally, several transcriptional regulators targeted by COP1 directly inhibit PIFs, for example, by forming non-DNA-binding heterodimers (e.g. HFR1). As PIFs directly control HFR1 expression this constitutes a negative feedback loop with important growth regulatory properties operating, for example, during the shade-avoidance response15,110,111 (Fig. 3c). This final example also illustrates a salient feature of phytochrome-regulated gene expression with the photoreceptor directly regulating multiple steps of a gene regulatory network41,42,58.

New modes of action of phytochromes

In addition to regulating a number of transcription factors, phytochromes regulate other aspects of transcription, post-transcriptional regulation, and translation. A genome-wide study indicates that phytochromes potentially regulate alternative promoter selection of more than 2000 genes112. This frequently results in modified protein amino-termini with predicted alteration of subcellular localization in several hundred cases112. Of particular interest is an essential photorespiration enzyme, glycerate kinase (GLYK) which changes its location from chloroplasts to the cytoplasm in shade-treated plants, thereby alleviating photoinhibition occurring in fluctuating light conditions112. ChIP data indicate binding of phyA and phyB to numerous promoters and phyB-binding sites are significantly enriched in genes showing phytochrome-mediated alternative promoter usage4,112,113. How phytochromes associate with chromatin, the functional consequence of this interaction and the mechanisms underlying phytochrome-mediated alternative promoter usage remain to be determined. Post-transcriptionally, phytochromes may control alternative splicing (AS) (extensively reviewed in Cheng et al.114). In Physcomitrella patens phytochromes appear to directly regulate AS115. In Arabidopsis, despite the identification of a number of splicing factors that can interact with phytochromes116,117, it remains debated whether AS is due to direct phytochrome regulation, is an indirect metabolic consequence of de-etiolation or a combination of both114,118. Light also regulates AS through chloroplast retrograde signaling, a process requiring normal chloroplasts119. Considering the recent discovery of a second-site mutation in the VENOSA4 gene controlling chloroplast size and photosynthetic traits in the frequently used phyB-9 allele (used in aforementioned studies)120, we conclude that the extent to which phytochromes rely on AS to control development requires further investigations.

De-etiolation is accompanied by extensive changes not only in transcript abundance and splicing but also translation. In etiolated seedlings hundreds of transcripts are present in cytosolic processing bodies (p-bodies) in a translationally repressed state121. Light cues, in a phytochrome-dependent manner, inhibit p-body formation/maintenance allowing translation of those mRNAs. Light-regulated suppression of translational inhibition is functionally important for seedling de-etiolation121. However, how phytochromes control this step remains poorly understood with the exception of light-induced inhibition of protochlorophyllide reductase A (PORA) translation that is mediated by a PENTA (PNT1)–PfrA complex122. Whether phytochromes are more broadly involved in translational control awaits further investigation. However, this may be a more downstream phytochrome signaling effect given that most tested physiological responses require nuclear rather than cytosolic phytochrome57.

The role of photobodies

Once in the nucleus all phytochromes locate in nuclear foci termed photobodies (PBs)57,58,60,123. During de-etiolation, a set of small PBs containing phyA and phyB appear after 2 min of irradiation, they disappear after 1 h and a new set mainly containing phyB is established after 2 h58,87. The abundance and size distribution of these late PBs can subsequently be modified by light conditions that change the fraction of active phyB. In de-etiolating seedlings, increasing R intensity or the R/FR ratio changes phyB localization from small PBs to increasingly large PBs58,123. Moreover, in adult plants phyB PBs are modulated by shade cues with low R/FR or low irradiance rapidly converting big PBs to many small PBs124. The same reduction in PB size is observed when light-grown seedlings are transferred into the dark125. Although light has a dominant effect, phyB PBs size distribution is also affected by ambient temperature5. Association with PBs largely depends on the C-terminus of phyB123. Moreover, phyB mutants with faster thermal reversion have smaller PBs126 while a constitutively active phyB forms PBs in the dark127. Finally, other proteins regulate the size and stability PBs. HEMERA (HMR), NUCLEAR CONTROL OF PEP ACTIVITY (NCP) and REGULATOR OF CHLOROPLAST BIOGENESIS (RCB), three dual targeted nuclear/chloroplast proteins are necessary for large PBs formation128,129,130. In addition, PHOTOPERIODIC CONTROL OF HYPOCOTYL 1 (PCH1) and its paralog PCH1-LIKE (PCHL) stabilize these sub-nuclear structures131,132,133. Recent reviews provide a more comprehensive list of PB functions and PB-associated proteins57,58.

Despite the tight correlation between phytochrome sub-nuclear localization and function, it is still not clear which molecular processes occur in each type of PBs. Early PBs have been proposed as the site of PIF degradation, since phyB PBs depend on PIF3 (ref. 87) and the abundance of those PBs correlates with PIF3 degradation58,128 (Fig. 4c, left panel). Late PBs are considered as sites maintaining active phyB3,131,132,133. PCH1 and PCHL are structural components of PBs which interact with phyB131,132,133. PCH1–phyB interaction inhibits Pfr thermal reversion in vitro133, while both PCH1 and PCHL reduce phyB thermal reversion in vivo, thereby restricting hypocotyl growth during the night period in short days and at elevated temperature131,132,133. Interestingly, PCH1 and PCHL expression is circadian regulated peaking at dusk correlating with large phyB PBs and possibly contributing to the diel regulation of PB formation131,134. While these results support the notion that PBs are a site for storage and maintenance of phyB Pfr, the presence of other proteins in these late large PBs points to additional functions. COP1 and SPA proteins co-localize with phytochromes in PBs. Originally this was interpreted as a possible role for PBs in phytochrome degradation. However recent work showed that the interaction of phytochromes with SPA proteins in PBs inactivates the COP1–SPA complex, allowing the accumulation of their targets, supporting the idea that PBs are site of phytochrome signal transduction101 (Figs. 3 and 4b). Additional proteins co-localize with PBs and support a role of PBs in phytochrome signaling. TZP, a chromatin-binding protein containing a zinc-finger domain and a PLUS3 domain, interacts with both phyA and phyB in a light-dependent manner and co-localizes to PBs consistent with a role of PBs in phytochrome-regulated transcriptional regulation135,136. SFPS interacts with PfrB and co-localizes with phyB and U2 small ribonucleoprotein (U2 snRNP)-associated proteins in nuclear bodies, suggesting that PBs could be a site of storage or modification of splicing factors and could have a role in the regulation of mRNA post-transcriptional modifications in response to light signals117. Finally, HMR, NCP, and RCB control PB size and are closely connected to PIF-mediated phytochrome signaling. All three proteins are required for light-induced degradation of PIF1 and PIF3 (refs.128,129,130). Moreover, HMR may also act as a transcriptional co-activator working with PIFs suggesting that at least in some cases PIF transactivation and stability are intimately linked58,137,138. Finally, HMR, NCP, and RCB control expression of chloroplastic gene expression in a process linking PIF degradation with activation of the plastidic PEP RNA polymerase129,130. These studies reveal how early phytochrome-PIF signaling steps lead to coordinated regulation of nuclear and chloroplastic gene expression which is particularly important for the transition towards photo-autotrophy.

Phytochrome signaling from the cellular to the whole-plant level

While we started by describing phytochrome signaling at the cellular level, not all phytochrome-mediated responses are local. Examples where the site of light perception is far from the physiological or developmental response are numerous and have been reviewed recently17,139,140. Many phytochrome-triggered long-distance signals rely on hormones. Phytochromes through PIF regulation control the expression of genes encoding rate-limiting catalytic enzymes of several phytohormones (e.g. auxin, gibberellins, abscisic acid), presumably at the site of light perception17. Regulated transport and response to these hormones underlies some of the distal responses elicited by phytochromes141. In other cases small proteins like HY5 or FLOWERING LOCUS T (FT) travel long distances to trigger a phytochrome-dependent response9,142. However, in many cases the long-distance signal remains to be established139,143. In the following paragraphs we will describe the most recent examples of phytochrome-mediated local and long-distance responses throughout development by starting with germination and finishing with the transition to reproduction.

In seeds, shortly after imbibition shade signals can inactivate phyB in the endosperm overriding phyA-dependent signaling in the embryo and preventing germination. This is achieved by controlling the gibberellin to abscisic acid balance and abscisic acid transport from the endosperm to the embryo17,144. Etiolated seedlings have a strong gravitropic response that depends on amyloplasts in the endodermis. It was recently shown that phyB-mediated light perception inhibits gravitropism by inhibiting PIFs, which allows the conversion of gravity-sensing amyloplast to chloroplast-like plastids in the endodermis. Unexpectedly this response requires phyB expression in the epidermis while PIF1 has to be present in the endodermis145. Moreover, PIF1 degradation in the endodermis occurs very efficiently even when phyB is selectively expressed in the epidermis145. This is surprising because all known mechanisms for phytochrome-mediated inhibition of PIFs depend on interaction of both proteins and it suggests that an unknown signal travels very fast from the epidermis to the endodermis.

After de-etiolation, shade signals promote hypocotyl growth while reducing cotyledon expansion8,95,146,147. This response largely depends on inactivation of phyB in the cotyledons which allows PIFs to promote the expression of YUCCA genes, coding for rate-limiting auxin synthesis enzymes. Auxin is then transported to the hypocotyl where it promotes elongation17,139,147. Shade signals also increase carbon allocation from cotyledons to hypocotyl by sucrose transport, which may contribute to the contrasting growth responses between cotyledons and hypocotyls11. In addition, local auxin production also contributes to hypocotyl elongation, but whether this relies on local shade perception or on another distal cotyledon-derived signal remains to be established148.

In adult plants shade signals can be more complex, for instance they might be caused by self-shading instead of shading by competitors. Consequently, shade responses are also more complex often involving interorgan signaling. Low R/FR perceived by phyB at the leaf tip promotes leaf hyponasty. Similar to young seedlings, the long-distance signaling relies on auxin synthesis promoted by PIF-mediated YUCCA expression in the blade and auxin transport towards the base of the petiole149,150. In contrast, a local drop in R/FR on the petiole does not promote hyponasty but instead promotes petiole elongation149,150. Gene expression analyses combined with physiological studies indicate that auxin biosynthesis and response are regulated differently in different organs and are both indispensable for robust shade-avoidance responses80,95,110,146,149,150,151. This includes a predominant function of PIF7 in the regulation of auxin biosynthesis genes while PIF4 and PIF5 also regulate auxin signaling genes80,93,94,96,110. Interestingly, in some cases PIF4 works as a member of tripartite TF-complex with AUXIN RESPONSE FACTOR 6 (ARF6) and BRASSINAZOLE RESISTANT 1 (BZR)179. Moreover, phyA and phyB potentially directly stabilize Aux/IAAs in the light90,91 (Figs. 3 and 4), suggesting another level of local regulation of auxin signaling by phytochromes. Under persistent shade new leaves develop less stomata in a phyB-dependent manner. This is independent of phyB expression in the stomata lineage, and phyB can regulate stomata development in young leaves depending on the light conditions perceived in older leaves152. In this case phyB activity controls the expression of genes related to stomata development as SPEECHLES and MUTE but the mobile signal has not yet been identified152.

In addition to a role in aerial organs shade signals also control root architecture. This phenomenon can be explained by two alternative hypotheses. Either phytochromes in the shoot perceive light and send a chemical signal to the roots, or light reaching the shoot is piped towards the root where it is perceived by root-localized phytochrome triggering a local response. So far there is evidence supporting both theories9,10. In both cases, the phytochrome downstream signaling molecule involved is HY5. This protein is known to travel long distances through the phloem153, it is stabilized in the shoot in response to light and is translocated to the root to control root architecture9. However, according to grafting experiments a full response requires a functional HY5 gene also in the root10. One possible explanation is that HY5 promotes its own expression154, hence shoot derived HY5 may promote HY5 expression in the root, but further studies are needed to address this model.

A final example of long distance signaling by a small protein is the case of flowering induction by FT. Light perceived by phytochrome in the leaves control FT abundance in response to photoperiod and shade signals. Upon the perception of the inductive signals FT accumulates in leaves and travels to the shoot apical meristem where it induces differentiation of the reproductive structures17,142. Thus, although phytochromes are expressed in all plant tissues and the early steps driving activation are conserved, phytochrome responses in the different organs presumably result from local, distal, or a combination of both types of signals.

Phytochrome signaling in other flowering plants

Many of the molecular mechanisms underlying phytochrome signaling have been discovered in Arabidopsis. However, phytochrome research has a long history and early studies have used a variety of plants leading to the discovery of these photosensory receptors and their functions155,156. Based on an ever-growing list of sequenced genomes we now know that many components of phytochrome signaling are deeply conserved in land plants103. Moreover, in a number of angiosperms including crops, genetic analyses identified light responses controlled by phytochrome signaling components20,157,158. In this section we will mention similarities and differences between Arabidopsis and other species in terms of phytochrome-mediated responses.

Most phytochrome responses described in Arabidopsis are shared by other plant species. This includes regulation of germination, de-etiolation, responses to shade and photoperiodic control of flowering time20,155,157,158. All angiosperms have phyA and phyB class phytochromes with variable numbers of family members in each class20. Functional divergence between both classes is conserved with, for example, phyA controlling the FR-HIR and phyB mediating reversible LFR responses20. Most angiosperms also have a phyC class, which was proposed to have a more prominent role in grasses and possibly have a different mode of action than Arabidopsis phyC159,160. However, both in grasses and Arabidopsis phyB plays a broader role than phyC while phyC plays a particularly important role for photoperiodic regulation of flowering presumably as a phyB–phyC heterodimer14,23,160,161. We therefore conclude that additional research is required to conclude whether the mode of action of phyC differs among species.

Numerous flowering plants are shade avoiders but some species are shade tolerant and therefore display different adaptive responses to shade cues155,162. Phytochromes control both types of responses to foliar shade, and comparative studies between related species with contrasting shade responses suggest mechanisms underlying different adaptations to the same light cue163,164. In Cardamine hirsuta, a species related to Arabidopsis, shade tolerance can in part be explained by higher PHYA transcript levels and phyA protein activity165. In other comparisons between related species or varieties of Geranium, rice, or conifers, the differential responses to shade have been related to hormonal signaling163,164,166. The ecological significance of the shade-avoidance response has been demonstrated but in crops it may also produce undesirable consequences because it diverts resources from storage organs, decreases the ability of plants to defend against pathogens and makes them more vulnerable to wind damage167. In some species domestication led to reduced shade-avoidance responses presumably due to the selection of yield at high planting density157. The role of several phyB class phytochromes for shade responses is established in a number of crops including tomato, maize, wheat, and rice161,168,169,170,171. While some shade responses in crops trigger yield penalties, responding to shade cues is also beneficial167. A good example is the control of leaf positioning in dense stands of maize or sunflowers as shade signals promotes canopy gap filling and hence better usage of the light172. This may explain why temperate maize inbred lines maintain robust shade responses168.

A faster transition to flowering by shade cues is common in many crops, but in Medicago sativa, a perennial species, flowering is delayed by additional FR light8,173. This suggests that in perennial species, the strategy to survive in suboptimal light environments is to save nutrients for the next season rather than to trigger flowering as in annual species. In crop species flowering time is determinant for yield and phytochromes affect photoperiodic control of flowering in rice, sorghum, barley, soybean, wheat, among others159,174,175,176,177. Another photoperiodic response tightly linked to yield is tuberization. In potato, (Solanum tuberosum) reallocating resources towards tubers is controlled by a phyB and a phyC class phytochrome178.

Studies in a wide variety of species identified role for phytochromes in processes which are absent in Arabidopsis. For example, in tomato phyA, phyB1, and phyB2 control fruit nutritional quality, carotenoid synthesis, and time to ripening179,180. Interestingly, during fruit ripening the light signal to induce carotenoid biosynthesis originates in the fruit itself, bringing a new layer of complexity to the phytochrome system which cannot be studied in other models179,180. Phytochromes also control biotic interactions, which is well studied in the context of plant pathogen interactions, exemplified by shade cues which through inhibition of phyB also inhibit plant defense167. Such interactions can be complex as in tritrophic interactions where the plant attracts predators of the herbivore. A beautiful example of such interactions controlled by phytochromes was described in Passiflora edulis where shade signals inhibit the production of extra floral nectar181. Another example of phytochrome-controlled interaction with beneficial organisms comes from Lotus japonicus where phyB promotes nodulation182. This study is also interesting in the context of phytochrome-controlled nutrient uptake which is well established for photosynthesis but also comprises light regulation of nutrients uptake by roots183.

In most flowering plants signaling events downstream of phytochrome activation remain poorly understood. However, rapid and widespread phytochrome control of gene expression is conserved as exemplified by studies in rice and Arabidopsis41,58,184. Rice PIF transcription factors control many phytochrome-mediated responses including de-etiolation, chlorophyll abundance, senescence, gravitropism, elongation growth, and interestingly grain size which is of agronomic interest41,185,186,187. However, the role of HY5 in rice differs at least partially from its function in Arabidopsis as a rice hy5 mutant is seedling lethal188. Characterization of PIF genes and mutants in maize shows that these bHLH transcription factors promote elongation of stem-like structures in etiolated seedlings and in response to shade as they do in Arabidopsis41,189. In tomato, PIFs control fruit ripening through a self-shading mechanism where light absorption by chlorophyll in the immature green fruit inhibits phytochrome activation179. At this stage low phytochrome activity allows PIF accumulation which inhibit carotenoid and tocopherol synthesis179,190. As fruits ripen, chlorophyll is degraded, which activates phytochromes triggering PIF degradation, thereby releasing PIFs inhibitory role on fruit maturation179,190. Regulation of tomato fruit quality involves a number of additional phytochrome signaling components including COP1, DET1, DDB1, and HY5 (ref. 158). Future studies will undoubtedly reveal how phytochrome signaling elements control light responses in a variety of species including agronomically important plants41.

Outlook

Phytochromes are among the most studied proteins in the history of plant biology but despite many years of research, important questions remain unsolved. In the following paragraphs, we outline some such questions ranging from molecular to physiological and ecological scales.

At the molecular level the complete structure of a plant phytochrome remains unresolved. Also, we lack knowledge of how light regulates interactions between plant phytochromes and immediate downstream partners. Addressing these questions will require the use of rigorous biochemistry. For example, the affinity constants for complexes between light-activated phytochromes and PIFs or any other interacting partners (e.g. SPA, FHY1) have not been systematically studied. A rather substantial list of proteins that can interact with phytochromes exists. However, many such complexes have not been biochemicaly characterized to, for example, determine which complexes are the most likely to form or to learn about competition between different proteins interacting with phytochromes. Ideally, structures of phytochrome-PIF (or other interacting factor) complexes showing how light changes interaction surfaces will provide insights into these early light-triggered events. Moreover, given that phytochromes can interact with PIF proteins both through their PSM and CTM, such studies may reveal how these interactions lead to different modes of PIF inhibition. Biochemistry is also required to provide more insight into the mode of action of other proteins important in phytochrome signaling such as PCH1 and HMR. It should be noted that the common practice of assembling plant phytochromes with the cyanobacterial instead of the plant chromophore renders Pfr more thermally stable, which has obvious consequences on, e.g., the study of light-regulated interactions44.

At the subcellular level, while most phytochrome-mediated responses require nuclear localization there are some light-triggered responses which are too fast to be explained through phytochrome-mediated transcriptomic changes (reviewed in ref. 191). Cytosolic functions for phytochromes appear to be more prevalent in cryptogams, for example, in Physcomitrella there is evidence that phytochrome in the cytoplasm interacts with phototropins (which are located at the plasma membrane) to modulate phototropism192. However, in Arabidopsis phytochromes primarily control phototropism by controlling gene expression140. Once in the nucleus, understanding the function(s) of PBs remains an important challenge. Substantial evidence points to their importance as sites of PIF protein degradation, transcriptional control, potentially splicing and to maintain a pool of light-activated phytochromes. We understand how light-activated phytochrome links with transcriptional changes in particular via PIFs. However, what distinguishes different mechanism of phytochrome-mediated PIF inhibition, whether they mostly occur in different cells, organs, or developmental stages remains largely unknown. Finally, the links between different modes of phytochrome signaling are unclear. For example, does phytochrome-mediated alternative promoter usage depend on an initial transcriptional change or are these two independent modes of phytochrome action? Also, how do phytochromes regulate more long-term changes of the transcriptional landscape? Does this occur through epigenetic changes96,193,194, and how is this linked to early transcriptional changes?

Light and phytochromes affect numerous facets of plant development but for practical reasons only a limited number of phenotypes have been investigated in detail (e.g. seed germination, control of hypocotyl elongation). In some instances lessons learned from simple systems yield precious information to understand phytochrome responses in other organs. However, given that there is ample evidence for organ-specific phytochrome responses, focusing on other phenotypes such as leaf development will certainly yield interesting new insight. Such studies will also benefit from a better integration of cellular, developmental, and photo-biology. This includes understanding how phytochromes through the regulation of several classes of transcription factors regulate expression at the tissue and cellular levels. In addition, studying other species than Arabidopsis will broaden our knowledge of organ-specific phytochrome functions. In this sense, the improvement in CRISPR gene editing will be instrumental.

Despite the fundamental importance of using simple light conditions to understand phytochrome signaling155,191, experiments performed in more natural conditions may reveal important new roles of these photoreceptors and identify important factors controlling their activity. Most of our knowledge about Arabidopsis phytochromes is based on lab experiments with artificial light that does not have the same spectrum as sunlight and is typically about 10 times lower than what a plant would experience on a sunny day. In addition, light and temperature are mostly treated as binary parameter (on/off, high/low) without the gradual changes observed in nature. Moreover, strong and rapid fluctuations in light intensity that are common in nature are not captured in most experimental setups, with notable exceptions195. Using more realistic conditions led to new insights as exemplified by a study showing that in strong high R/FR phyA plays a role in de-etiolation, although phyA was typically regarded as not being important in high light196. Similarly, photoperiodic regulation of flowering time is quite different in standard lab conditions versus outside-grown plants197. This study suggests that the model explaining day-length-regulated flowering requires adjustments and shows that by better simulating natural light (R/FR) and temperature fluctuations data generated in the lab mimic outdoor experiments much better197. We hypothesize that in the natural context the role of other phytochromes (phyC, phyD, and phyE) and some phytochrome signaling elements may become more apparent. Moreover, using more realistic light environments is important to understand the crosstalk between phytochromes and other photosensory receptors.

In addition to more realistic environments more extensive use of genetic resources would be very informative. Although many of the phytochrome signaling elements are deeply conserved in land plants (e.g. phytochrome, PIF, COP1, SPA, HY5, and FHY1)103, there is substantial evidence for differences in light responses across Arabidopsis accessions that in some cases can be traced to the phytochrome or phytochrome-signaling elements198,199. Combined with the sequence of more than a 1000 accessions and increasingly detailed information about the collection site, this information is a fantastic resource to investigate adaptation to local light environments.