Organ specificity in the plant circadian system is explained by different light inputs to the shoot and root clocks

Summary Circadian clocks allow the temporal compartmentalization of biological processes. In Arabidopsis, circadian rhythms display organ specificity but the underlying molecular causes have not been identified. We investigated the mechanisms responsible for the similarities and differences between the clocks of mature shoots and roots in constant conditions and in light : dark cycles. We developed an imaging system to monitor clock gene expression in shoots and light‐ or dark‐grown roots, modified a recent mathematical model of the Arabidopsis clock and used this to simulate our new data. We showed that the shoot and root circadian clocks have different rhythmic properties (period and amplitude) and respond differently to light quality. The root clock was entrained by direct exposure to low‐intensity light, even in antiphase to the illumination of shoots. Differences between the clocks were more pronounced in conditions where light was present than in constant darkness, and persisted in the presence of sucrose. We simulated the data successfully by modifying those parameters of a clock model that are related to light inputs. We conclude that differences and similarities between the shoot and root clocks can largely be explained by organ‐specific light inputs. This provides mechanistic insight into the developing field of organ‐specific clocks.


Model structure
The P2011 model, which is shown in Fig. 6 of the main text, was based on seedling data but also accurately described our shoot data. The model comprises a set of ordinary differential equations representing the interrelation of mRNA and protein levels for the genes LHY/CCA1, TOC1, PRR9, PRR7, NI, ELF3, ELF4, LUX, and GI, and the protein levels of ZTL, COP1 and protein complexes such as the evening complex (EC); CCA1 and LHY are partially redundant and treated as one entity. In this model external light is represented by the light function L whose value can be 0 (dark) or 1 (full light) depending on the light conditions simulated; brief twilights are simulated by L values between 0 and 1. Dark is represented by the parameter D = 1 -L. The value of L is not a linear description of light intensity but could represent the strength of light input.
Our data suggests that the core clock structure is similar in shoots and roots. In addition roots can perceive light, either by direct illumination or by light piping from exposed tissues. We therefore used the P2011 model including the light function to simulate the behaviour of the root clock.

Equations including light inputs
Light enters the P2011 model through the parameter L in the following equations: (1) In these equations: c i m and c i represent concentrations of mRNA and protein, respectively. The subscript "i" denotes LHY/CCA1 (L), PRR9 (P9), PRR7 (P7), the Night Inhibitor (NI), TOC1 (T), ELF3 (E3), ELF4 (E4), LUX, the Evening Complex (EC), GI (G), ZTL and its complex with GI (ZG), and the protein P; a, c, d and e exponents are Hill coefficients; n j and m j are rate constants of transcription and degradation, respectively; g j are Michaelis-Menten constants; q j are the rate constants of acute (P-dependent) light activation of transcription; p j are constants of translation, protein modification, protein complex formation and translocation between nucleus and cytoplasm.
In these equations twelve parameters are multiplied by L and are termed the "light-related parameters". They can be grouped in two categories, respectively acute and permanent effects of light on the clock: -the "P-related parameters" are also multiplied by c P , that is m 11 , n 6 , q 1 , q 2 and q 3 . They affect the clock after a dark to light transition but not in constant light (LL) when c P rapidly tends to 0; -the other light-related parameters are multiplied by L but not by c P . They all represent rate constants related to light: m 1 (degradation of CCA1/LHY mRNA), m 27 (degradation of COP1n), m 32 (degradation of the EC), p 1 (translation of CCA1/LHY mRNA), p 12 (ZTL-GI complex formation), p 15 (translocation of COP1 between nucleus and cytoplasm), and p 24 (modification of EC). They are termed the "L-related parameters" and affect the clock whenever light is present. The values of these parameters are presented in Supplementary Table S3. When L = 0, none of these parameters have an influence on clock dynamics except m 27 and m 32 , so these two parameters were not changed in the optimisation process.

Simulations
We used COPASI (version 4.8.35) to change the default (shoot) value of parameters related to light inputs (see below) and run simulations; the level of GI mRNA served as the main marker to compare experimental and theoretical data. To model the light function we used two step functions (Adams et al., 2012): step function1 to simulate light:dark cycles (LD), LL and constant dark (DD) by setting two of its parameters (the "amplitude step1" and the "offset step1") to 0 or 1 as shown in Supplementary Table S3, and step function2 to simulate skeleton photoperiods. Mutations of clock genes were simulated by decreasing the rates of gene transcription to zero. For each new set of parameters the simulations were first run for 20 d under LD (which corresponds to 3 wk old plants) before release into other condition (e.g. LL or DD).

Parameter change strategy
Our experimental data suggests that light inputs to the clock are organ-specific and may explain the different dynamics observed in shoots and roots. To test this hypothesis and to fit our root clock data we changed the value of L and light-related parameters with the following constrains (ranked in order of decreasing importance): -longer periods in light-and dark-grown roots compared to shoots under LL -lower amplitudes in dark-grown roots compared to shoots under LD -higher trough levels for evening genes in dark-grown roots compared to shoots We noticed that reducing the value of L gives a longer free-running period (FRP) in LL (Fig.  S11a). (Changing the value of L to x requires changing the parameter D to D = x-L for simulations in LD, so that D reaches the value 0 during light cycles.) This reproduces the observation that period lengthens in LL when the light intensity is reduced, consistent with Aschoff's rules. In addition the amplitude of transcript levels are reduced when L = 0.5 ( Fig  S11a), which reproduces our observations in dark-grown roots compared to shoots (Fig. 1). We therefore decided to set L to 0.5 and 1 to simulate dark-and light-grown roots respectively. Since this single change could not reproduce all our root data we then modified other light-related parameters as described below.
The concentration of the protein P (c P ) rapidly tends to 0 when light is on so the "P-related parameters" have virtually no influence on the dynamics of the clock under LL. Therefore these parameters did not have to be changed to fit the longer FRP of roots compared to shoots. However our experimental data showed that the expression of GI and PRR9, but not CCA1, was acutely induced in roots exposed to light (Fig. 1b). Therefore the value of q 1 , which represents the acute activation of CCA1/LHY mRNA transcription by light, was halved. This change also allowed a more accurate representation of the phase of CCA1/LHY expression under LD cycles (when L = 0.5) (Fig. S11b).
The "L-related parameters" (m 1 , p 1 , p 12 , p 15 and p 24 ) were changed individually and in combination. Each of these parameters was first scanned individually with COPASI by increasing or decreasing their default value up to 10-fold, and then at higher resolution over a 2-fold range. None of these single changes gave a satisfactory fit to the root data (e.g. if the period was long enough, the oscillation dampened out). Therefore combinations of parameters had to be changed simultaneously.
Starting from the default values of the P2011 model (except that the values of L and q 1 were halved as above) we decided to focus on the parameters p 1 , p 15 and p 24 for the following reasons: -The changes to L and q 1 could simulate dark-grown root data satisfactorily ( Fig. S11b) but the FRP in LL was still too short (< 26 h), and some amplitudes were too high and/or troughs were too low. -Troughs are generally higher in roots, especially for evening genes (Fig. S6B). Given the structure and equations of the P2011 model we reasoned that the trough values could be increased if the levels of the LHY protein or the EC were lower (Fig. S12a). This could be achieved by decreasing p 1 and p 15 and by increasing p 24 .
-The system was extremely sensitive to changes in m 1 but insensitive to changes in p 12 , so their values were not changed initially.
Many combinations of p 1 , p 15 and p 24 were tested as follows: -The FRP should be ~27 h, i.e. 2 h more than the shoot FRP in the model. Experimentally, these values may vary depending on the technique used and the initial conditions, but the difference between shoot and root periods is typically 2 h or more.
-Satisfactory sets of parameters were then tested in the same condition (LL) but with L = 1 to simulate light-grown roots; the FRP should still be higher than in shoots -these sets of parameters were also tested in LD to check whether the amplitude was lower in dark-grown roots compared to light-grown roots and shoots.
Our optimal combination of parameters was the following: q 1 , p 1 and p 15 were halved and p 24 was increased by 50%. This gave a longer FRP in dark-(L = 0.5) and light-grown roots (L = 1) compared to shoots and lower amplitudes in roots compared to shoots. To further increase the FRP in light-grown roots and reach a similar value as dark-grown roots, the value of m 1 was decreased by 25% for light-grown roots (i.e. with L = 1). These new parameter sets for roots also fitted experimental data that had not been used in the optimisation, such as the effects of mutation on periods (Fig. 7) and the organ-specific response to skeleton photoperiod (Fig. 8).

Figure S1: Brief overview of the optimised imaging protocol, from sowing to imaging
Seedlings were grown on 1/2 MS medium for 10-12 d (top left) before being transferred on fresh 1/2 MS medium with charcoal (top middle). Roots were then either light-or darkgrown. In the latter case, a black acrylic bar was placed horizontally under the shoots (top middle) to minimise light leakage to the roots; to keep the roots in the dark, black tape was placed around the plate (top right). After another 10-12 d of entrainment in LD (12/12) at 20º C, plants were ready for imaging (top right). The original lid (with black tape, or without if roots were light-grown) was removed. The plants were sprayed with luciferin (~6 µmol per plant) and the new lid (bottom left) was used to seal the plate. This new lid contained a black barrier and tape that distinguished two compartments (one for shoots and one for roots) to reduce light scattering between the two organs. An example of a bright field image and the corresponding luminescence are shown in the bottom middle and bottom left panels respectively. For dark-grown roots, a shutter (represented by hatched bars, bottom middle) was placed over the root compartment when the lights were on.

Figure S2: The expression of CCR2 in roots is affected by direct exposure to light
Plants carrying the CCR2:LUC+ fusion were entrained for 3-4 wk in LD before release into LL. The promoter activities in plants with light-grown (a) and dark-grown (b) roots were monitored over the last d in LD (time 0 = dawn) and in LL. From 108 h (red arrow in (b)), dark-grown roots were exposed to the same light conditions as shoots and light-grown roots. For each experiment, 1-3 clusters of 2-6 plants (organs) were imaged separately and luminescence data were normalised to the mean values between 0 and 108 h. Error bars are +SEM for 3 independent experiments. The backgrounds show d or subjective d (white bars), night (dark grey bars) and subjective night (hatched bars). Circadian periods (c) and relative amplitudes (d) were estimated between 48 and 108 h for individual time-courses (S = Shoots, R = Roots; L= Light-grown roots and D= Dark-grown roots). Relative amplitudes were normalised to the mean relative amplitude of the shoots (SL and SD). Error bars are +SEM.

Figure S3: Relative amplitudes in LD are lower in dark-grown roots than shoots
Plants were grown in hydroponic culture with dark-grown roots. Plants were entrained for 4 wk in LD before harvesting from dusk (ZT -12). mRNA levels of CCA1 (left) and TOC1 (right) were measured in shoots and roots. The experiment was repeated 3 times independently. For each experiment mRNA levels were normalised to ISU1, expressed to the highest value in shoots over the corresponding time course and then averaged. Error bars are + SD. The backgrounds show d (white bar) and night (dark grey bar).

Figure S4: The expression of GI in roots is affected by direct exposure to light
Plants carrying the GI:LUC+ fusion were entrained for 3-4 wk in LD before release into LL. The promoter activities in plants with light-grown (a) dark-grown (b) roots were monitored over the last d in LD (ZT0 = dawn) and in LL. From ZT108 (shown by the red arrow), darkgrown roots were exposed to the same light conditions as shoots and light-grown roots. The experiment was repeated 3 times independently. Fig. 1 shows normalised data from these experiments. Here the raw data of 4 clusters of 2-6 plants are shown. The backgrounds show d or subjective d (white bars), night (dark grey bar) and subjective night (hatched bars).

Figure S5: TOC1 is expressed rhythmically in Col-0 roots
Col-0 plants carrying the TOC1:LUC+ fusion were entrained for 3-4 wk in LD before release into LL. The promoter activities in plants with dark-grown roots were monitored over the last d in LD (ZT0 = dawn) and in LL. From time 108 h (red arrows), roots were exposed to the same light conditions as shoots. Conditions are as in Fig. 1 but with a different ecotype. One cluster of 3-6 plants was imaged in two independent experiments. Data from individual timecourses were normalised to their mean values between time 0 and 108 h. Bars in the backgrounds represent d or subjective d (white bars), night (dark grey bars) and subjective night (hatched bars). Mean circadian periods were estimated between 48 and 108 h as 24.14 h and 27.60 h for shoots and roots respectively.

Figure S6: The cca1/lhy double mutation affects the shoot and root clocks similarly
Experimental data (top) and simulations (bottom) for shoots and dark-grown roots in wild type and mutant plants. cca1/lhy double mutant plants carrying the CCR2:LUC+ reporter were entrained for 5 wk in LD cycles before imaging. They were then imaged for 1 d in LD and 5 d in LL. Simulations of GI mRNA were used for comparison with the experimental data of the evening gene CCR2. Shoots and dark-grown roots were simulated in the wild type with the same parameters as in Figs. 7-8, and the cca1/lhy mutant was simulated by setting the values of n 1 and q 1 to 0. Each trace was normalised to the mean over the whole timecourse. The dark grey bars in the backgrounds represent the last night before release into LL.

Figure S7: The root clock can be entrained by very low light intensity
Plants with dark-grown roots carrying the GI:LUC+ reporter were sprayed with luciferin, maintained for 2 further LD cycles (12 h light : 12 h dark) at 15 µmol.m -2 .s -1 , for 48 h in DD, over 4 LD cycles at 0.15 µmol.m -2 .s -1 (equal intensities of red and blue light) and finally in DD. Imaging commenced at the start of the second LD cycle. Some plants were decapitated 24 h before this. Periods were estimated for 96-168 h (LD) and 180-240 h (DD). Panel (a)time courses, means + SD for n=8 clusters of plants from 2 biological replicates. Panel (b)assessment of period against relative amplitude error (RAE) for the individual traces summarised in panel (a).

Figure S8: Rhythms of GI:LUC+ expression in DD without sucrose
Plants with light-grown roots carrying the GI:LUC+ reporter were entrained for 3-4 wk in LD (white light) before imaging over 24 h in LD followed by DD. Panel (a) shows representative traces for individual clusters of plants, panel (b) shows periods and RAE values estimated over 72 h in DD (48-120 h in (a)) for individual clusters of plants that were scored rhythmic in two separate experiments. Mean periods + SD were: roots, 32.4 + 2.0 h (n=6); shoots, 33.1 + 3.0 h (n=5), P>0.1 by Student's t-test.

Figure S9: Transcript levels of morning and evening genes free run in DD with similar FRPs in shoots and roots
Plants were grown in hydroponic culture with dark-grown roots. Plants were entrained for 4 wk in LD before release in DD. mRNA levels of CCA1, LHY and PRR9 (left) and GI and TOC1 (right) in shoots and roots; mRNA levels are normalised to ISU1. Time 0 is the last dusk before DD. Periods are shown in Table S5.  Simulations of GI (a) and LHY/CCA1 (b) mRNA in LL and LD respectively. The parameter L was scanned with COPASI between 0.2 and 1 in LL (a). In LD simulations are shown for L = 1 and L = 0.5 with the default value of q 1 , and L = 0.5 with q 1 halved (set to 0.6).

Figure S12: Correlation between protein levels and mRNA troughs in shoots and roots
Protein and mRNA levels were simulated using the P2011 model, with the default parameters for shoots and the modified parameters described in Fig. 7 and table S4 for dark-grown roots (a) Simulation of EC protein and GI mRNA in LL (b) Simulation of LHY/CCA1 protein and GI mRNA in LD Proteins are represented by dark green lines (shoots) or brown dashes (roots). GI mRNA is represented by light green dash-dot (shoots) or orange dots (roots). Vertical red lines and dashes represent inhibition of transcription by proteins in shoots and roots respectively.

Table S3: Periods and amplitudes of gene expression in shoots and roots by qPCR
Periods and relative amplitudes were estimated between times 24 and 84 h for the individual time-courses whose means are shown in Fig. 2 Fig. 6. Values are means + SD for n=4 clusters of plants in 2 biological replicates. Differences in period between red and blue light were assessed by Student's t-test, *, P<0.05, ** P< 0.01  Table S6: Parameter values for shoot and root clock models The default parameters of the P2011 model were used except as shown below. *, in the day L = 0.5 or 1 except for brief twilight periods at dawn and dusk.