Differential phosphorylation of the N-terminal extension regulates phytochrome B signaling

Phytochrome B (phyB) is an excellent light quality and quantity sensor that can detect subtle changes in the light environment. The relative amounts of the biologically active photoreceptor (phyB Pfr) are determined by the light conditions and light independent thermal relaxation of Pfr into the inactive phyB Pr, termed thermal reversion. Little is known about the regulation of thermal reversion and how it affects plants’ light sensitivity. In this study we identiﬁed several serine/threonine residues on the N-terminal extension (NTE) of Arabidopsis thaliana phyB that are differentially phosphorylated in response to light and temperature, and examined transgenic plants expressing nonphosphorylatable and phosphomimic phyB mutants. (cid:1) The NTE of phyB is essential for thermal stability of the Pfr form, and phosphorylation of S86 particularly enhances the thermal reversion rate of the phyB Pfr – Pr heterodimer in vivo . We demonstrate that S86 phosphorylation is especially critical for phyB signaling compared with phosphorylation of the more N-terminal residues. Interestingly, S86 phosphorylation is reduced in light, paralleled by a progressive Pfr stabilization under prolonged irradiation. (cid:1) By investigating other phytochromes (phyD and phyE) we provide evidence that accelera-tion of thermal reversion by phosphorylation represents a general mechanism for attenuating phytochrome signaling.


Introduction
Plants use photoreceptors to constantly monitor ambient light conditions in order to adjust their growth and development in an ever-changing environment. Red and far-red light is detected by the phytochrome (phy) family of sensory photoreceptors, which in Arabidopsis thaliana comprises five members (phyA-E) with different but also partially overlapping functions (S anchez-Lamas et al., 2016). Phytochromes are synthesized in the red light-absorbing form (Pr) that is, upon exposure to red light, photoconverted into the biologically active far-red light-absorbing form (Pfr) (Rockwell et al., 2006). Light absorption by Pfr in turn induces photoconversion to Pr. The Pfr form is thermally unstable and reverts back into Pr via light-independent thermal reversion, and thus photoconversion and thermal reversion determine the steady-state amount of the active Pfr form. PhyB, the dominant phy in light-grown plants, is a potent light quality and quantity sensor and gradually controls photomorphogenic development (Klose et al., 2015). Upon light exposure, the activated phytochromes translocate into the nucleus where they can localize to subnuclear structures called photobodies (PBs) (Yamaguchi et al., 1999;Kircher et al., 2002). In the nucleus, Pfr interacts specifically with multiple signaling molecules and induces massive transcriptional changes related to the initiation of photomorphogenic development (Quail, 2002).
Phytochromes are composed of an N-terminal photosensory module (PSM) and a C-terminal output module (OPM) connected by a flexible hinge region (Fig. 1a). The PSM consists of an N-terminal extension (NTE), a Per/Arnt/Sim (PAS) domain of unknown function, a cyclic guanosine monophosphate (cGMP) phosphodiesterase/adenylyl cyclase/FhlA (GAF) domain that binds a bilin chromophore and a phytochrome-specific PHY domain that is crucial for the stability of the Pfr conformer (Rockwell et al., 2006;Burgie et al., 2014). The OPM contains two PAS-related domains (PAS-A, PAS-B) and a Histidine Kinase Related Domain (HKRD) and is required for phytochrome dimerization (Nagatani, 2010;Qiu et al., 2017).
Phytochromes function as dimers and the Pfr-Pfr homodimer in the nucleus was proposed to be the active conformer of phyB (Klose et al., 2015). Thermal reversion occurs in two steps: a slower reversion from Pfr-Pfr to the Pfr-Pr heterodimer (k r2 ) and a much faster reversion from Pfr-Pr to the Pr-Pr homodimer (k r1 ) (Klose et al., 2015). Both reversion rates display strong temperature dependency in a physiological temperature range, enabling phyB to act as temperature sensor (Jung et al., 2016;Legris et al., 2016). In strong light, Pr-to-Pfr photoconversion is dominant and phyB Pfr-Pfr accumulates to high concentrations that decay slowly after transfer to darkness, enabling, for example,

Research
New Phytologist night length measurement. In low light conditions, thermal reversion, particularly k r1 , becomes increasingly important: it competes with the Pr-to-Pfr photoconversion in determining the amount of Pfr established in a fluence rate-dependent manner. High temperature accelerates k r1 and k r2 , leading to a decrease of the Pfr concentrations (Jung et al., 2016;Legris et al., 2016).
Previous studies performed in oat found that phytochromes are phosphoproteins acting as autophosphorylating serine/threonine kinases (McMichael & Lagarias, 1990;Lapko et al., 1997;Yeh & Lagarias, 1998;Lapko et al., 1999) and that the kinase domain of phyA is critical for ATP binding and efficient light signaling (Shin et al., 2016). Our knowledge about phosphorylation of Arabidopsis phytochromes is rather limited. It was demonstrated, that Arabidopsis phyA, phyB and phyD autophosphorylate and have kinase activity towards their interaction partner PHYTOCHROME INTERACTING FACTOR 3 (PIF3) in vitro (Shin et al., 2016). Recent reports also showed that Arabidopsis phyA is phosphorylated in planta (Zhang et al., 2018;Zhou et al., 2018).
A number of phosphorylated residues in phyB were identified in vivo. Whereas phosphorylation of S86 was reported to accelerate thermal reversion, phosphorylation of Y104 was proposed to inhibit binding to PIF3; thus it was suggested that phosphorylation of phyB negatively regulates phyB signaling Nito et al., 2013). Several additional evolutionarily conserved amino acids (S84, T89-91, S106 and Y113) were found to be phosphorylated in a light-dependent manner which locate in the phosphorylation cluster of signaling modulation (PCSM) (Nito et al., 2013). These data suggest that phosphorylation could be a mechanism to modulate phyB signaling.
In this study we examined dynamic phosphorylation at the NTE of phyB in response to light and temperature in planta by LC-MS/MS. We investigated the functional role of specific phosphosites using transgenic plants expressing nonphosphorylated and phosphomimic mutants and found that phosphorylation of S86 in phyB's NTE severely alters phyB-mediated red light sensitivity by reducing the amount of physiologically active Pfr. Our data revealed that regulation of thermal reversion by dynamic phosphorylation pattern is particularly important under limiting light conditions where the effect of thermal reversion on red light sensitivity is strong. By investigating phyD and phyE phosphorylation, we provide further evidence that phosphorylation of the PCSM represents a general mechanism for attenuating phytochrome signaling via accelerating thermal reversion.
The final constructs have been verified by sequencing and transformed into Arabidopsis (Clough & Bent, 1998). Homozygous T3 progenies with expression levels comparable to those of the wild-type phyB-expressing lines or the phyE-YFP and phyD-YFP phosphomutants to the corresponding phyE-YFP and phyD-YFP levels were selected for further experiments. Additional independent mutant transgenic lines were also tested and are shown in Figs S1, S4 and S7 (see later), obtaining similar results to those presented in the main text.

Plant sample collection for LC-MS/MS analysis of phyB
Plants were grown on Murashige & Skoog medium (Sigma) containing 3% sucrose for 10 d under an 8 h : 16 h white light : dark regime. Seedlings were collected at the end of the day (EOD) and at the end of the night (EON). The sample preparation and MS analysis are based on Klement et al. (2019) and described in detail in the Methods S1.

Plant growth conditions and hypocotyl growth assays
Arabidopsis seeds were sown in Petri dishes on four layers of wet filter paper and stratified for 72 h at 4°C, before they were irradiated with WL for 4 h at 22°C to induce germination and transferred to the dark for 18 h at 22°C. For hypocotyl length measurements seedlings were irradiated with continuous red light (LED; 660 nm) under various temperature conditions for 4 d. Seedlings were placed on agar plates and scanned with a flatbed scanner (Epson, Suwa, Japan). Hypocotyl length was determined using METAMORPH software (Universal Imaging, Downingtown, PA, USA). Relative hypocotyl length was calculated as the ratio of the hypocotyl length of light-grown and the corresponding dark-grown seedlings. The experimental procedure for measurement of hypocotyl growth rates is described in the Methods S1.

In vivo spectroscopy
For measuring thermal reversion kinetics, 4-d-old etiolated Arabidopsis seedlings were irradiated with saturating red light (50 µmol m À2 s À1 for 20 min at 22°C) to establish the photoequilibrium of 87% Pfr/Ptot (Ptot, total amount of phytochrome), transferred to darkness and kept at either 17, 22 or 27°C. As the 35S:PHYB[S86D]-YFP line failed to reach the photoequilibrium under this light treatment, we irradiated seedlings for 1 h with 200 µmol m À2 s À1 on ice in order to increase the Pfr/Ptot value, and transferred them to prewarmed plates in darkness at respective temperatures. Pfr/Ptot was measured using a dual-wavelength ratiospectrophotometer (Klose, 2019) at indicated time points during dark incubation. For steady-state Pfr/ Ptot measurements, seedlings were irradiated with red light for 1 h unless otherwise stated and immediately transferred to ice water to minimize Pfr loss during sample handling before measurement. For steady-state Pfr/Ptot measurements in continuous red light, seedlings were grown on ½ MS-agar plates containing 5 µM norflurazon. The herbicide norflurazon (SAN 9789) effectively inhibits carotenoid and Chl accumulation without affecting the phytochrome system (Jabben & Deitzer, 1978;Frosch et al., 1979;Jabben & Deitzer, 1979).
Calculation of the thermal reversion rates k r1 and k r2 The thermal reversion rates k r2 of each phyB variant and temperature combination were calculated from the measured thermal reversion kinetics using single exponential decay functions. As the calculated k r2 exhibited an exponential temperature dependency, k r2 was extrapolated for the additional temperatures (4, 12 and 32°C) and used for calculating k r1 respectively. The thermal reversion rate k r1 was calculated based on the three-state-dimer model (Klose et al., 2015) using the following equation:
Dynamic phosphorylation at specific serine residues in the NTE of phyB in response to light and temperature In order to investigate whether phyB phosphorylation changes in response to light and temperature, we grew seedlings expressing phyB-GFP for 10 d in short-day conditions at 17, 22 or 27°C and harvested them at EOD or at EON. For quantitative analyses the tryptic digest of the immunoprecipitated phyB-GFP protein was directly analyzed by LC-MS/MS without phosphopeptide enrichment. Two of the detected phosphorylated phyB fragments exhibited dynamic changes in their phosphorylation status depending on the light and temperature conditions. We observed that the relative phosphopeptide signal of the fragment containing phosphorylated S23/S24/S25/T27 was elevated at the EOD compared with the EON and decreased with temperature rise from 17 to 27°C (Fig. 1b). By contrast, phosphorylation of S86 was higher at EON compared with EOD and elevated temperatures increased phosphorylation in light and darkness (Fig. 1b). These two fragments showed opposite phosphorylation patterns in response to light and temperature.

Impact of temperature and S86 phosphorylation on red light sensitivity
To investigate how S86 phosphorylation modulates light and temperature signaling of phyB, we measured fluence rate response curves for the inhibition of hypocotyl elongation in red light at different ambient temperatures (17, 22 and 27°C). The serine residue at position 86 was substituted with alanine to obtain a nonphosphorylatable mutant phyB[S86A] or with a negatively charged aspartate to mimic a constitutively phosphorylated residue phyB[S86D], as described previously for transgenic lines overexpressing phyB by the 35S promoter . It is well established that phyB signaling is dose-dependent (Wagner et al., 1991)

Research
New Phytologist hyposensitive phenotype in red light compared with phyB, whereas phyB[S86A] was hypersensitive under all tested temperatures (Figs 2a-c, S1), although phyB[S86D] was considerably more highly expressed (Figs 2d, S1). The red light responsiveness of all three genotypes was progressively reduced with increasing temperatures. Interestingly, the differences in red light sensitivity between the examined lines were obvious at fluence rates below 5 µmol m À2 s À1 but they exhibited similar responses at higher fluence rates (Fig. 2a-c).

Impact of temperature and S86 phosphorylation on Pfr steady-state concentrations in light
The phyB activity depends on the amount of phyB in the Pfr-Pfr homodimer conformation (Klose et al., 2015). Whereas thermal reversion from the Pfr-Pfr to Pfr-Pr occurs at a slow rate, k r2 , the thermal reversion rate, k r1 , from Pfr-Pr to Pr-Pr is much faster. In limiting light conditions and high temperatures, thermal reversion, especially k r1 , becomes increasingly important for the irradiance and temperature dependence of phyB activity (Sellaro et al., 2019). It has been reported that the phosphomimic phyB [S86D] mutant has accelerated thermal reversion after transfer from light to darkness, and this is mainly a result of the slow reversion rate, k r2 , that has a minor impact on Pfr concentrations in light and thus cannot directly account for the hyposensitivity of phyB[S86D] seedlings in red light .
Here, we wanted to investigate the extent to which the fast reversion rate k r1 is affected by S86 phosphorylation. Therefore, we determined the Pfr concentration relative to the total phytochrome amount (Pfr/Ptot) in steady-state conditions for phyB, phyB[S86A] and phyB[S86D] in different light intensities and temperatures by in vivo spectroscopy. As physiological phyB levels are too low for detection, we used lines overexpressing phyB, phyB[S86A] or phyB[S86D] fused to YFP by the 35S promoter in a phyA-211 phyB-9 double mutant background. Steadystate Pfr/Ptot ratios for phyB measured in vivo showed strong fluence rate dependence in the range 0.5-5 µmol m À2 s À1 of red light (Fig. 2e). PhyB[S86A] can establish Pfr/Ptot values comparable to WT phyB in two-to three-fold lower red fluence rates, which is consistent with its hypersensitive phenotype. By contrast, much higher red light intensities had to be applied to reach equivalent Pfr/Ptot ratios for phyB [S86D], which is in agreement with the strongly impaired red light sensitivity of the phyB [S86D] mutant. Even 100 µmol m À2 s À1 red light was not sufficient to establish the photoequilibrium of 87% Pfr/Ptot, indicating that the phyB[S86D] Pfr-Pr form is thermally highly unstable. We detected a strong temperature dependence of the steady-state Pfr/Ptot values under nonsaturating light conditions for all three different genotypes showing reduced relative Pfr concentrations at elevated temperatures (Fig. 2f).
To calculate the thermal reversion rates k r1 and k r2 we additionally obtained temperature-dependent thermal reversion kinetics for phyB, phyB[S86A] and phyB [S86D]. Consistent with previous findings  the S86D mutation showed accelerated thermal reversion kinetics, whereas S86A exhibited slower thermal reversion kinetics at all tested temperatures (Fig. S2). In addition, all lines showed a strong temperature dependence, displaying faster thermal reversion at higher temperatures. Thermal reversion is usually efficiently suppressed at low temperatures, but even at 4°C the phyB[S86D] mutant exhibited fast and complete reversion comparable to the kinetics measured in the WT at 27°C (Fig. S2). We calculated the thermal reversion rates k r1 and k r2 (Fig. 2g-i): k r1 was reduced two-to three-fold for phyB[S86A] compared with phyB, but about 50-fold increased for phyB [S86D], which correlates with the measured fluence rate dependence of the relative Pfr/Ptot values (Fig. 2e,h). Interestingly, k r1 of all three phyB versions showed equal temperature dependence (Fig. 2h), indicating that S86 phosphorylation is not the only mechanism affecting k r1 and that temperature dependency of thermal reversion represents instead an intrinsic property of the chromophore.

PhyB Pfr is stabilized under prolonged irradiation
We noticed that the fluence rate range of the phyB-mediated physiological response was much broader compared with the one of the measured steady-state Pfr/Ptot ratios. Transgenic seedlings expressing WT phyB responded strongly to light below 1 µmol m À2 s À1 , whereas the measured Pfr/Ptot values were < 20% (Fig. 2b,e). Also the phyB[S86D] mutant responded to red light below 10 µmol m À2 s À1 but possesses hardly any detectable Pfr (Fig. 2b,e). The reason for this discrepancy could be a consequence of the differential experimental conditions used: hypocotyl growth inhibition was monitored after 4 d of growth in continuous light, whereas Pfr/Ptot was measured in etiolated seedlings after 1 h red light treatment. Thus we determined steady-state Pfr/Ptot values in seedlings grown for up to 3 d in continuous red light. As Chl interferes with the in vivo spectroscopic measurements, seedlings were grown on medium containing 5 µM norflurazon to bleach the chloroplasts. After 1 h red light treatment, etiolated seedlings grown on norflurazon established Pfr/Ptot values comparable to seedlings grown in standard conditions, indicating that the norflurazon treatment does not affect Pfr/Ptot (Figs 2e, 3a-c). However, after 3 d of irradiation, Pfr/Ptot ratios were considerably higher compared with 1 h or 1 d, indicating a progressive Pfr stabilization in light (Fig. 3a,b). Seedlings expressing phyB or phyB[S86D] that received a Pfr-reverting far-red light pulse after 3 d of red irradiation and subsequently were irradiated with 1 h red light established the same high Pfr/Ptot values, demonstrating that the steady state is established within 1 h of light treatment (Fig. 3c). Furthermore, the Pfr stabilization of phyB[S86D] was less pronounced compared with phyB, indicating that this phenomenon is also regulated, at least partially, by phosphorylation.
Pfr/Ptot ratios measured under prolonged irradiation nicely matched the fluence rate response curves at lower fluence rates. After 3 d of irradiation, full photoequilibrium was established in plants expressing phyB[S86A] at 1 µmol m À2 s À1 (Fig. 3a) while their red light response reached a plateau (Fig. 2b), which was less pronounced for WT phyB that still reaches > 60% Pfr/Ptot at 1 µmol m À2 s À1 (Fig. 2b) similar hypocotyl growth inhibition at c. 10 µmol m À2 s À1 but seedlings only accumulated < 40% Pfr/Ptot (Fig. 3b); however, the response could have been compensated by the higher phyB [S86D] expression level (Fig. 2d). This indicates that the irradiance-sensitive physiological response at higher fluence rates (> 10 µmol m À2 s À1 ) is independent of S86 phosphorylation and  The accumulation of phyB in PBs is strictly Pfr-dependent and can be used to monitor the Pfr content (Klose et al., 2015;Legris et al., 2016). Our microscopic analyses revealed that the PB formation is correlated with the measured Pfr/Ptot values, confirming the stabilization of Pfr under prolonged irradiation (Fig. S3). Although WT phyB does not form detectable PBs after 6 h of light treatment, they could be well observed after 3 d of continuous 1 µmol m À2 s À1 irradiation. This light fluence could not induce PB formation of the phyB[S86D] mutant even after 3 d of irradiation, but 10 µmol m À2 s À1 of red light illumination for 3 d was necessary to detect the appearance of PBs containing phyB[S86D] (Fig. S3).

PhyB S86 phosphorylation affects growth rate under simulated natural growth conditions
Under natural conditions, a major function of phyB is to sense shade signals arising from competing neighbors, characterized by low red : far-red (R : FR) ratio (Smith, 2000). Shade directly alters the photoconversion rates, favoring the formation of Pfr-Pr heterodimers. In turn, temperature affects the Pfr-Pr thermal reversion rate k r1 , a major determinant of light sensitivity (Klose et al., 2015;Sellaro et al., 2019). As k r1 strongly depends on S86 phosphorylation (Fig. 2h), we investigated the response to shade at different temperatures in de-etiolated seedlings expressing YFP-fused phyB, phyB[S86A] and phyB [S86D]. The seedlings were grown under diurnal white-light cycles (10 h : 14 h light : dark, 22°C) for 3 d and then transferred to 12 different combinations of shade and temperature during the photoperiod of the fourth day to measure the growth rate of the hypocotyl during that period. The experimental setup is optimal to determine effects on k r1 , which is important for growth responses during the day. The Col WT was included under the same conditions and its growth rate was used as a biologically meaningful quantification of the integrated impact of the shade and temperature combinations. We observed the lowest growth rate under white light (R : FR = 1.0) at the lowest temperature (17°C) and the highest growth rates under deep shade (R : FR = 0.1) combined with the highest temperature (28°C) for all tested lines (Fig. 4a,b). As expected, when plotted against the growth rate of the Col WT, the regression line corresponding to the PHYB: PHYB-YFP was close to the 1 : 1 line and that corresponding to New Phytologist the 35S:PHYB-YFP (phyB overexpressor) was below the 1 : 1 line, owing to stronger phyB-mediated growth inhibition (Fig. 4a,b). In both cases, the slope of the regression line belonging to phyB[S86D] expressors was significantly steeper and that of the phyB[S86A]-expressing plants was less pronounced than their respective transgenic control expressing phyB (Fig. 4a,b). These results indicate that the degree of S86 phosphorylation significantly affects the function of phyB as a sensor of neighbor and temperature cues. Furthermore, the distortion caused by genetic modification of the S86 phosphorylation status was particularly large when the plants were exposed to the combination of increased degrees of shade and high temperatures.

Phosphorylation of serine residues in the NTE of phyD and phyE
PhyD and phyE are evolutionarily related to phyB, and thus we decided to examine the phosphorylation pattern of phyD and of phyE in planta. Interestingly, our LC-MS/MS analyses revealed phosphorylation of only single serine residues in the NTE of phyD (S79 or S82) and phyE (S53) ( Fig. 5a; Notes S2). Serine residues in close proximity to the identified phosphorylation sites, phyD S88 and phyE S50, are homologous to the conserved S86 of phyB (Fig. 5a). To analyze the effect of phosphorylation of these serines, we generated and examined phyD-YFP or phyE-YFP overexpression lines in phyABD or phyABE triple mutant background, respectively.
The nonphosphorylatable phyD mutants phyD[S82A] and phyD[S88A] displayed a hypersensitive red light response compared with phyD (Figs 5b-d, S4), although expression levels of these lines were lower (Fig. S5). The phosphomimic phyD [S88D], by contrast, exhibited a reduced red light responsiveness (Figs 5b-d, S4). Along with these results we noticed that the Sto-A mutants of phyD have stronger preference to localize in PBs than the WT or the S-to-D mutant counterparts, indicating that these structures contribute to signaling (Fig. S5). The red light responses of all lines were gradually reduced with increasing temperature, indicating that thermal reversion of phyD could be responsible for light and temperature dependence of the response. Astonishingly, the amounts of photoreversible phyD we detected in the in vivo spectroscopic assay were too low to allow precise Pfr estimation, despite the fact that we were using strong phyD overexpressor lines. This suggests that phyD could be highly thermally unstable in the Pfr form and hence circumvent detection in our system. Nevertheless, preventing phosphorylation at the NTE enhanced red light sensitivity of phyD, indicating slightly enhanced Pfr thermal stability.
The phyE overexpression line showed a mild hypocotyl growth inhibition in red light that was not fluence rate-dependent (Figs 5e-g, S4), in good agreement with published data ( Ad am et al., 2013). PhyE Pfr proved to be highly thermally stable, without showing detectable thermal reversion within 4 h of darkness (Fig. 5h) and hence accumulated high Pfr concentrations close to the photoequilibrium already at very low red light intensities (0.1 µmol m À2 s À1 ) where phyB did not show any detectable Pfr (Fig. 5i), which explains the lack of fluence rate dependency. The nonphosphorylatable phyE[S50A] mutant was also thermally stable (Fig. 5h) and exhibited a physiological response similar to phyE ( Fig. 5e-g). By contrast, the phosphomimic phyE[S50D] mutant was almost blind to red light and showed almost complete thermal reversion within 4 h ( Fig. 5e-g). It is interesting to note that phyE and the nonphosphorylatable mutant versions have relatively shorter hypocotyls at higher temperature, which could indicate higher physiological activity or higher stability of , we found that phyE does not form PBs after extended red irradiation and thus we speculate that they are not required for phyE signaling (Fig. S6). Taken together, these data indicate that phosphorylation at the NTE could be a general mechanism to attenuate light sensitivity of phytochromes by accelerating thermal reversion.

The NTE is essential for Pfr thermal stability
The NTE of phyB has been shown to be important for Pfr thermal stability in vitro (Burgie et al., 2014;Burgie et al., 2017), but the physiological relevance of phyB without NTE in Arabidopsis has not yet been examined. An Arabidopsis line expressing truncated phyB lacking the N-terminal 89-amino-acid phyB[dN89] fused to YFP displayed strongly impaired red light responsiveness as manifested in hyposensitivity for hypocotyl growth inhibition, despite the fact that the expression level of phyB[dN89] was considerably higher than that of the control line (Figs 6a,b, S7). The hyposensitive phenotype of phyB[dN89] was much more severe than that of the phyB[S86D] mutant and correlated with further reduced steady-state Pfr/Ptot values (Fig. 6c). This indicates that thermal stability of the Pfr-Pr heterodimer is dramatically compromised in phyB[dN89].

Multiple phosphorylated residues within the NTE influence phyB signaling
We investigated whether phosphorylation of serine residues in the NTE of phyB (Fig. 1a) also plays a role in red light sensitivity via the modulation of thermal reversion in vivo using lines expressing multiple nonphosphorylatable (S to A) or phosphomimic (S to D) amino acid substitutions at S3 and S23-25 positions. The S3/23-25A and S3/23-25D quadruple mutations were combined with the S86A or S86D mutations to obtain phyB[S3/ 23-25/86A] and phyB[S3/23-25/86D] quintuple mutants to test whether balancing the phosphorylation status along the NTE is important for phyB activity. Although the phyB[S3/23-25A] mutant had a WT-like response in red light corresponding to a WT-like Pfr/Ptot value, the hypersensitive phyB[S86A] mutant phenotype was suppressed in the phyB[S3/23-25/86A] mutant that still had higher Pfr/Ptot compared with phyB (Fig. 6d,f). The phosphomimic phyB[S3/23-25D] mutant was hyposensitive in red light compared with WT phyB but had normal Pfr/Ptot. The phyB[S3/23-25/86D] mutant had reduced light responsiveness comparable to phyB[S86D], reflected by a strongly decreased Pfr/Ptot (Fig. 6d,f). Taken together, these data indicate that phosphorylation at S86 plays a dominant role in regulating phyB thermal reversion, but phosphorylation at S23-25 affects red light signaling, presumably by other pathways independent of thermal reversion.
The phosphorylation of Y104 was shown to affect light sensitivity dramatically (Nito et al., 2013), but we wanted to know whether it affects the thermal reversion of phyB. We found that the phosphomimic phyB[Y104E] shows accelerated thermal reversion kinetics and reduced steady-state Pfr/Ptot, whereas nonphosphorylatable phyB[Y104F] had no effect on thermal reversion in vivo (Fig. S8).

The phyB D453R mutation enhances red light sensitivity through reduced thermal reversion
It has been shown that the kinase activity and the integrity of the ATP binding residue in the N-terminal photosensory domain of oat phyA are necessary for phyA function and the oat phyA [D422R] mutant exhibited strong defects in ATP binding and kinase activity (Shin et al., 2016). Although the ATP-binding site and kinase activity of Arabidopsis phyB remain to be identified in planta, we tested whether the equivalent residue D453 at the Nterminal domain of phyB is important for phosphorylation and signaling. Arabidopsis seedlings expressing phyB[D453R] fused to YFP at physiological levels were hypersensitive in red light (Figs 7a, S7). Light sensitivity was even further enhanced compared with phyB[S86A], although expression levels were lower than in the control lines (Fig. 7a,b). Consistent with that, phyB [D453R] established higher Pfr/Ptot under nonsaturating red light irradiation (1 µmol m À2 s À1 ), indicating a very slow thermal reversion (Fig. 7c). Our quantitative MS analyses demonstrate that the fragment containing residues S23, S24, S25, T27 was hyperphosphorylated in phyB[D453R] compared with phyB, whereas S86 phosphorylation remained unchanged (Fig. 7d). As the D453R mutation did not abolish phyB phosphorylation at the NTE, we concluded that D453 is not essential for the proposed autophosphorylation activity of phyB, but it seems to have a specific effect on S23-S25/T27 phosphorylation. Alternatively, D453R mutation could affect thermal reversion of phyB independently of NTE phosphorylation.
It is interesting that the phyB[G564E] mutant, having the mutation also in the PHY domain, has extremely slow thermal reversion (phyB-401) (Kretsch et al., 2000;Ad am et al., 2011). This phyB molecule did not exhibit hyperphosphorylation; instead its phosphorylation pattern at S86 and S23-25/T27 was no longer light-dependent (Fig. 7e). It is plausible that the mutant cannot distinguish between night and day as a result of highly stable Pfr form.

Discussion
PhyB is an excellent light quality and quantity sensor that can detect even subtle changes in light conditions. Thermal reversion is an intrinsic property of the phyB molecule that can be extensively modulated by intramolecular interactions and external factors (Viczi an et al., 2017). Hence, manipulating the thermal reversion rate represents an efficient mechanism to change red light sensitivity of the system. In this context, our results corroborate the findings of  nicely and, in addition, demonstrate that the phosphomimic phyB[S86D] mutant alters phyB-mediated red light sensitivity by reducing physiologically active Pfr concentrations as a result of strongly accelerated thermal reversion of the Pfr-Pr heterodimer (k r1 ). The fast reversion rate, k r1 , of the Pfr-Pr heterodimer is the most critical parameter for the light sensitivity of phyB-mediated  (Klose et al., 2015). Whereas k r2 , the thermal reversion rate of the Pfr-Pfr homodimers, is only increased two-to threefold in the phyB[S86D] mutant, k r1 is increased about 50 times, causing strong reduction of the Pfr/Ptot ratios in light (Fig. 2ei).
The S86 phosphorylation status changes dynamically in vivo in response to light and temperature, providing evidence for the capability of plants to modulate thermal reversion and consequently phyB activity via dynamic phosphorylation. High temperature and darkness, two conditions that reduce phyB activity, also enhanced S86 phosphorylation (Fig. 1c). However, this is not always reflected by the hypocotyl phenotypes: the difference in hypocotyl growth inhibition between WT phyB and phyB [S86A] was not increasing with temperature (Fig. 2a-c), although S86 phosphorylation was higher in WT phyB (Fig. 1c). This might be a result of the fact that only a small fraction of   phyB is phosphorylated, which is not enough to cause a visible phenotype under the conditions used. Alternatively, S86 dephosphorylation in continuous light could compensate for the temperature-induced phosphorylation. In addition, the phosphorylation status of phyB could be balanced at multiple sites apart from S86 as shown for the S23-25/T27 fragment, which is phosphorylated in an opposite manner.
Whereas S86 is dephosphorylated in light, Pfr is progressively stabilized under continuous irradiation (Figs 1c, 3a-c). As there is no evidence that the photochemical reactions are affected by continuous irradiation, we conclude that Pfr stabilization in light is caused by a progressive reduction of the thermal reversion rate k r1 . Our data show that only a small percentage of the phyB pool is phosphorylated at a certain time point, and thus we believe that it is unlikely that S86 dephosphorylation is the only mechanism accounting for the observed Pfr stabilization. It was proposed that Pfr is stabilized through interaction with other proteins and is protected from thermal reversion within PBs (Sweere et al., 2001;Rausenberger et al., 2010;Klose et al., 2015;Enderle et al., 2017). Among these proteins, PHOTOPERIODIC CONTROL

Research
New Phytologist OF HYPOCOTYL 1 (PCH1) and its homolog PCHL (PCH1like), which accumulate in light and colocalize with phyB in PBs, were shown to inhibit phyB thermal reversion upon phyB binding (Huang et al., 2016;Enderle et al., 2017). Furthermore, a very recent study demonstrated that PCH1 stabilizes phyB Pfr in vitro and that PCH1 is an essential structural component of phyB PBs (Huang et al., 2019). We found that the phyB molecules, mutated at phosphorylated residues and exhibiting thermal reversion phenotype, show no impaired binding to PCH1 and PCHL (Fig. S9). This result suggests that the role of PCH1 and PCHL in the regulation of phyB thermal reversion is not based on their binding ability to differently phosphorylated phyB molecules and both pathways act independently in the regulation of Pfr stability.
It remains to be determined how S86 phosphorylation enhances phyB thermal reversion mechanistically. In general, the NTE of phyB is assumed to be important for thermal stability of the Pfr form. Removal of the NTE from the phyB protein accelerated thermal reversion of phyB PSM fragments in vitro and abolished phyB localization to PBs in Arabidopsis (Chen et al., 2005;Burgie et al., 2014;Burgie et al., 2017). Here, we provide photobiological and physiological evidence that NTE deletion severely enhances thermal reversion of phyB in planta, leading to very low Pfr concentrations even in high light, thus strongly reducing phyB activity (Fig. 6a-c). Until recently there was no structural information about phyB NTE available, as the published crystal structure of Arabidopsis phyB PSM lacks the NTE (Burgie et al., 2014). Lately, a state-dependent interaction between the chromophore and the NTE in phyB was demonstrated by Raman spectroscopy (Vel azquez Escobar et al., 2017). Deletion of the NTE affected the chromophore and its surrounding hydrogen bonding network, particularly in the Pfr state, which could potentially affect the thermal reversion kinetics, and it was also revealed that the NTE undergoes light-dependent structural changes particularly in the S84-K88 region (Horsten et al., 2016). Phosphorylation of residues in this region could lead to steric hindrance of the Pfr form, explaining the increased thermal reversion of the phosphomimic phyB[S86D] mutant of phyB. Although the effect of NTE deletion on thermal reversion of phyB is very pronounced, we cannot exclude the possibility that mutations in this region or NTE deletion might affect the photochemical properties of phyB and in that way contribute to the reduced red light sensitivity of the phyB[dN89] mutant.
As phosphorylation of PCSM residues was previously shown to regulate phyB signaling negatively (Nito et al., 2013), it was proposed that phosphorylations at the PCSM motif promote rapid thermal reversion in planta, although direct experimental evidence was only available for S86 . In contrast to the phosphomimic S86D mutant, which only partially impaired phyB signaling, a phosphomimic phyB [Y104E] was unable to complement the phyB mutant phenotype (Nito et al., 2013). Here we demonstrate that phyB[Y104E] does indeed have strongly accelerated thermal reversion, but we observed that the nonphosphorylated phyB[Y104F] mutant exhibited a WT-like thermal reversion kinetics in vivo (Fig. S8). Interestingly it was found that a phyB[Y104A] mutant also has accelerated thermal reversion in vitro (Burgie et al., 2014). Structural analyses revealed that an a-helix formed by residues Y104-R110 connects the NTE with the PAS domain and sterically shields the chromophore with Y104 directly adjoining the chromophore (Burgie et al., 2014;Horsten et al., 2016). Several studies demonstrate that glutamate is not always an effective mimetic for phosphotyrosine, as it has little chemical and structural similarity (Honda et al., 2011;Chen & Cole, 2015); thus additional studies are needed to draw conclusions from tyrosine-phosphomimics.
The packing model between the NTE and the PSM predicts an intimate interaction between the PCSM and the light-sensing knot region which resembles the putative PIF binding site (Kikis et al., 2009;Horsten et al., 2016). Therefore, it is conceivable that the mechanism for inactivation of phyB signaling by phosphorylation involves blocking of the PIF binding capability of phyB. The Y104E mutation completely abolished Pfr-dependent PIF3 binding in an in vitro assay (Nito et al., 2013) and the Nterminal fragment carrying S86D mutation only had a weakened interaction with PIF3 in yeast that could be compensated by using higher red light intensity . As both phosphomimic mutants have failed to accumulate to high Pfr concentrations in light, it is also possible that the phosphorylation affects PIF3 binding indirectly via reducing the amount of phyB Pfr molecules available for binding.
The large NTE is unique to phyB and its paralog phyD. Whereas phosphorylation at the PCSM promotes rapid thermal reversion, dynamic phosphorylation at the more N-terminal residues S23-25/T27 did not affect thermal reversion of phyB in vivo but could modulate phyB activity by other, as yet unknown mechanisms (Fig. 6d-f). In line with this, sequential deletion of the N-terminal 50 amino acids of phyB, which are more or less absent in phyA, phyC and phyE, was shown to have little impact on thermal reversion in vitro (Burgie et al., 2017). Taken together, our data show that phosphorylation also occurs in the PCSM of phyD and phyE in vivo, that it affects phyDand phyE-mediated red light sensitivity and accelerates phyE thermal reversion (Fig. 5). This provides further evidence that the mechanism by which phosphorylation in the PCSM region inactivates red light signaling is common for all light-stable phytochromes.
We detected different phyB phosphorylation patterns in response to light and temperature, implying the activity of specific kinases and phosphatases. Several phosphatases were shown to interact with phytochromes (Kim et al., 2002;Ryu et al., 2005;Phee et al., 2008) and furthermore phytochromes have been shown to act as autophosphorylating serine/threonine kinases. In oat phyA, autophosphorylation sites were identified residing in the NTE and a residue critical for ATP binding in the photosensory domain is necessary for kinase activity (Han et al., 2010;Shin et al., 2016). Our data show that these findings obtained for oat phyA are not directly conferrable to phyB. Mutating the corresponding residue in phyB, which is critical for ATP binding in phyA, did not abolish phosphorylation at the NTE but rather induced hyperphosphorylation at S23-25/T27. It suggests that this position is not essential for ATP binding, autophosphorylation or to activate other kinases to phosphorylate phyB. Considering the different modes of action for phyA and phyB, it is not surprising that specific phosphorylation or dephosphorylation could be regulated differentially or could have different effects on phytochrome activity and signaling.
The control of phytochrome phosphorylation status represents a vital mechanism for fine-tuning the light responsiveness mediated by phytochromes. Whereas photochemical reactions are dominant under strong light, the thermal reversion rate k r1 becomes increasingly important for phyB Pfr steady-state concentrations as irradiance decreases, for example, under canopy shade, in cloudy days and/or at the extremes of the natural photoperiod (Klose et al., 2015;Sellaro et al., 2019). The impact of thermal reversion is also enhanced by high temperatures (Jung et al., 2016;Legris et al., 2016). Thus the phosphorylation status becomes physiologically most relevant in plants exposed to deep shade and high temperatures (Fig. 4). Therefore, the genetic modification of the S86 residue could offer a biotechnological target to adjust phyB sensitivity in a context of climate change without altering the ability of the photoreceptor to perceive the R : FR ratio.
Although the thermal reversion process has been known for decades, we are just starting to understand how it affects signaling and how it is modulated in planta to regulate plants' responsiveness to environmental stimuli. It is now understood that thermal reversion is regulated by a number of molecular mechanisms, including intra-as well as intermolecular interactions and posttranslational modifications, but it is not clear how the different mechanisms are integrated. Further investigations are needed to reveal how these pathways are integrated in the regulation of Pfr thermal stability and red light sensitivity in Arabidopsis.

Supporting Information
Additional Supporting Information may be found online in the Supporting Information section at the end of the article.         Methods S1 Detailed description of the experimental procedures used in this study.
Notes S1 MS/MS protein spectra of phosphorylated phyB peptides.
Notes S2 MS/MS protein spectra of phosphorylated phyD and phyE peptides.

Table S1
Transgenic Arabidopsis lines used in this study.

Table S2
List of oligonucleotides used for cloning.
New Phytologist is an electronic (online-only) journal owned by the New Phytologist Trust, a not-for-profit organization dedicated to the promotion of plant science, facilitating projects from symposia to free access for our Tansley reviews and Tansley insights.
Regular papers, Letters, Research reviews, Rapid reports and both Modelling/Theory and Methods papers are encouraged. We are committed to rapid processing, from online submission through to publication 'as ready' via Early View -our average time to decision is <26 days. There are no page or colour charges and a PDF version will be provided for each article.
The journal is available online at Wiley Online Library. Visit www.newphytologist.com to search the articles and register for table of contents email alerts.
If you have any questions, do get in touch with Central Office (np-centraloffice@lancaster.ac.uk) or, if it is more convenient, our USA Office (np-usaoffice@lancaster.ac.uk) For submission instructions, subscription and all the latest information visit www.newphytologist.com New Phytologist (2019) Ó 2019 The Authors New Phytologist Ó 2019 New Phytologist Trust www.newphytologist.com