Circadian control of root elongation and C partitioning in Arabidopsis thaliana

Authors


J. Fisahn. Fax: +49 331 567 8236; e-mail: fisahn@mpimp-golm.mpg.de

ABSTRACT

Plants grow in a light/dark cycle. We have investigated how growth is buffered against the resulting changes in the carbon supply. Growth of primary roots of Arabidopsis seedlings was monitored using time-resolved video imaging. The average daily rate of growth is increased in longer light periods or by addition of sugars. It responds slowly over days when the conditions are changed. The momentary rate of growth exhibits a robust diel oscillation with a minimum 8–9 h after dawn and a maximum towards the end of the night. Analyses with starch metabolism mutants show that starch turnover is required to maintain growth at night. A carbon shortfall leads to an inhibition of growth, which is not immediately reversed when carbon becomes available again. The diel oscillation persists in continuous light and is strongly modified in clock mutants. Central clock functions that depend on CCA1/LHY are required to set an appropriate rate of starch degradation and maintain a supply of carbon to support growth through to dawn, whereas ELF3 acts to decrease growth in the light period and promote growth in the night. Thus, while the overall growth rate depends on the carbon supply, the clock orchestrates diurnal carbon allocation and growth.

Abbreviations
CCA/LHY

circadian clock associated late elongated hypocotyls

ELF3

early flowering 3

ELF4

early flowering 4

SEX

starch excess

PGM

phosphoglucomutase

INTRODUCTION

Plants grow autotrophically using light, CO2, nutrients and water that they acquire from the abiotic environment. As they need to maximize capture of these resources, plants are unavoidably exposed to changes in the environment. One of the most pervasive environmental changes is the daily alternation between light and dark (Geiger, Servaites & Fuchs 2000; Nozue & Maloof 2006; Smith & Stitt 2007). In the light, plants have a positive carbon (C) balance and can use carbohydrates that are delivered by photosynthesis for growth. In the dark, growth depends on resources that have been stored in preceding light periods. In many species, starch is accumulated in the light and remobilized at night (Geiger & Servaites 1994; Geiger et al. 2000; Smith & Stitt 2007; Stitt, Usadel & Lunn 2010). Sugars and organic acids (Chia et al. 2000; Zell et al. 2010) also contribute to transient C storage.

The rate of starch degradation in leaves at night is essentially linear, with about 95% of the starch being utilized by dawn (Fondy & Geiger 1985; Geiger & Servaites 1994; Matt et al. 1998; Smith et al. 2004; Gibon et al. 2004a; Graf et al. 2010). A correspondence between the time taken to degrade starch reserves and the length of the night is important to optimize growth in C-limiting conditions. Growth will be decreased if a significant fraction of the daily photosynthate remains as starch at the end of the night, rather than being invested in new leaf and root biomass (Rasse & Toquin 2006). Premature exhaustion of starch also carries a growth penalty. Mutant plants that are impaired in the synthesis or degradation of starch have strongly reduced biomass, except in continuous light or very long days (Caspar, Huber & Somerville 1985; Gibon et al. 2004a, 2009). This is partly due to inhibition of growth during the night (Wiese et al. 2007). In addition, in the first hours of the light period almost all the new photosynthate accumulates as sugars, indicating there is a delay until growth resumes (Gibon et al. 2004b).

Plants adjust their starch turnover to changes in the amount of photosynthate that is fixed, or the length of the night. Thus, when plants are grown in lower light intensities, lower CO2 concentrations or shorter light periods, they allocate more of their current photosynthate to starch in the light and degrade the starch more slowly at night (Stitt, Bulpin & Rees 1978; Chatterton & Silvius 1979, 1980, 1981; Mullen & Koller 1988; Lorenzen & Ewing 1992; Matt et al. 2001; Gibon et al. 2004a). As a result, a small amount of starch remains at the end of the night. Most remarkably, the rate of starch degradation in Arabidopsis adjusts immediately to a sudden and unexpected early or late onset of night. There is an immediate increase in the rate of starch degradation when the light period is suddenly lengthened and, as a consequence, the night is shortened (Lu, Gehan & Sharkey 2005) and there is an immediate decrease in the rate of starch breakdown after exposure to a premature dusk and, as a consequence, a longer night (Lu et al. 2005; Graf et al. 2010). These observations imply that the rate of starch breakdown is gauged to the anticipated length of the night.

Like other organisms, plants contain sophisticated biological clocks (Harmer, Panda & Kay 2001; Schaffer et al. 2001; Michael & McClung 2003; Schultz & Kay 2003; Webb 2003; de Montaigu, Tóth & Coupland 2010). Current clock models (Locke et al. 2006; Zeilinger et al. 2006; Nakamichi et al. 2010) contain a dawn loop, in which expression of the two Myb-related transcription factors LHY and CCA1 is modulated by PRR5, PRR7 and PRR9, and a dusk loop including the PRR family member TOC1, which is modulated by GI and LUX. LHY and CCA1 act to repress TOC1, and TOC1 acts to induce CCA1 and LHY via a process that is not fully understood but includes CCA1 HIKING EXPEDITION (CHE), a member of the TCP family that acts as a transcriptional repressor of CCA1 (Pruneda-Paz et al. 2009). The multiple PRR genes of Arabidopsis uncouple events in the late night from light-driven responses in the day, increasing the flexibility of rhythmic regulation (Pokhilko et al. 2010). Based on an analysis of temperature responses, it was recently proposed that ELF3 is also part of the core circadian clock (Thines & Harmon 2010). Earlier studies attributed ELF3 a role in the regulation of light inputs into the clock (McWatters et al. 2000; Covington et al. 2001; Liu et al. 2001; Yu et al. 2008). It has also recently been proposed that ELF4 integrates the morning and evening loops of the clock (Kolmos et al. 2009).

Most studies of the clock have been performed in leaves or hypocotyls. They emphasize the importance of interactions between light and the clock in light/dark cycles in entraining the clock (Nozue et al. 2007; de Montaigu et al. 2010). On the other hand, in a recent comparison of circadian rhythm in gene expression in shoots and roots, James et al. (2008) reported that the evening loop is absent in roots in continuous light, and that the root clock is resynchronized with the shoot clocks in light/dark cycles by a photosynthesis-related signal, which can be overridden by including sucrose in the medium in which the roots are growing.

The clock plays a major role in the regulation of processes that are linked to day length, such as floral induction (Hayama & Coupland 2003; de Montaigu et al. 2010). It also regulates the expression of thousands of genes for metabolism and growth (Harmer et al. 2000; Schaffer et al. 2001). Transcripts for different sets of genes peak at different times in a free-running 24 h cycle (Harmer et al. 2000, 2001; Covington et al. 2008) leading to the proposal that circadian regulation anticipates diurnal changes.

The importance of the clock for the regulation of growth in light/dark cycles is underlined by the finding (Dodd et al. 2005; Graf et al. 2010) that growth is decreased when the lengths of the circadian and diurnal cycle differ. Dodd et al. (2005) reported that the lower growth rates in mismatched conditions correlated with reduced chlorophyll and lower rates of photosynthesis. Graf et al. (2010) showed that starch was exhausted about 24 h after the last dawn, irrespective of the actual duration of the light/dark cycle. Thus, in a 14 h light/14 h dark cycle, starch was exhausted about 4 h before the actual dawn. Furthermore, the lhy/cca1 double mutant exhausted its starch about 20 h into the diel cycle (Graf et al. 2010), which corresponds to dawn as anticipated by the fast-running circadian clock in this mutant (Alabadi et al. 2002; Ding et al. 2007). Graf et al. (2010) concluded that the reduced growth of wild-type plants in 28 h days and lhy/cca1 in 24 h days is due to the inappropriately high rate of starch degradation, leading to a period of C starvation at the end of night. This would be analogous to the growth inhibition seen in starchless mutants, except that the recurring daily period of C starvation is due to an inappropriate regulation of diel starch turnover, rather than a block in the pathways of starch synthesis and degradation.

Changes in the rate of starch mobilization, on their own, will not allow a plant to avoid periods of C starvation at the end of the night. It will also be necessary to decrease the rate of C utilization. This could, in principle, occur via coordinate regulation of starch breakdown and C utilization, or via a very sensitive regulation of the rate of C utilization in response to small changes in supply of sucrose and other metabolites that are synthesized from starch. Gibon et al. (2009) observed a strong positive correlation between the rate of starch degradation and the relative growth rate when Arabidopsis Col-0 was grown in a range of different photoperiods. However, this study only provided information about the average rate of growth. Highly time-resolved measurements of growth will be required to test whether inappropriate timing of starch breakdown leads to a time-of-day-dependent inhibition of growth, that is, at the end of the night.

It is known that plant growth is highly rhythmic with respect to the time of day. In general, growth peaks at around dawn in leaves of dicot species (Schmundt et al. 1998; Walter & Schurr 2000, 2005; Wiese et al. 2007) and the middle of the day in monocot species (Watts 1974; Acevedo et al. 1979; Seneweera et al. 1995). In 12/12 light–dark cycles, the Arabidopsis accessions Ler and Col-0 display a maximum soon after dawn, a subsequent decrease of growth during the day, and a minimum early in the dark period (Wiese et al. 2007). Hypocotyl growth in wild-type Arabidopsis growing in a light/dark cycle exhibits an even sharper maximum at the end of the night, followed by a decrease in the light (Nozue et al. 2007). Evidence for the involvement of the circadian clock in the control of hypocotyl growth in Arabidopsis seedlings was provided by the observation that the growth oscillations in continuous light are modified in lines with constitutive overexpression of CCA1 and in the elf3 mutant (Nozue et al. 2007).

The following experiments investigate diel growth rhythms in wild-type Arabidopsis, in mutants in starch turnover, and in a set of clock mutants that have previously been shown to be affected in starch turnover or diel growth rhythms. We focus, for several reasons, on primary root growth in Arabidopsis seedlings. First, roots are completely dependent on the shoot for the provision of C. This allows complicating effects of light on the local generation of energy or sugars to be excluded. It is also possible to complement changes in endogenous C provision by supplying sugars in the root medium. Second, local changes in water potential are likely to be less marked than in leaves. Third, root growth is essentially a unidirectional process and in seedlings occurs at one site, the primary root tip. This should make it easier to detect small and transient changes in the growth rate.

Root extension growth can be measured by using digital calipers to determine root tip displacement, by marking root tip position on a transparent surface, or by capturing a series of time lapse records. Commercial software such as WINRHIZO (Arsenault et al. 1995), OPTIMAS analysis software (Media Cybernetics, http://www.mediacy.com) or IMAGE J (Abramoff, Magelhaes & Ram 2004) has been introduced to assess root length. Recent studies have also quantified differences in root architecture (Armengaud et al. 2009). Using these approaches, it has been shown that there are rapid adaptations of growth in responses to modulation in light intensities and photoperiods (Aguirrezabal, Deleens & Tardieu 1994; Muller, Stosser & Tardieu 1998; Nagel, Schurr & Walter 2006). There are different growth zones at the root tip, which are differentially affected by different treatments (Walter, Feil & Schurr 2003; Walter & Schurr 2005). However, there is little information on how roots respond to diurnal stimuli or changes in the carbohydrate supply.

We have developed a platform for high throughput analysis of Arabidopsis root growth kinetics (Yazdanbakhsh & Fisahn 2009, 2010). Seedlings are grown on agar plates in a custom-designed controlled climate chamber, in which the roots are illuminated by infrared diodes and are oriented and fixed in the focal plane of a charge-coupled device (CCD) camera that captures changes in the tip position of the primary root in time-lapse videos, and quantified to estimate the growth rate. We use this method to show that Arabidopsis wild-type seedlings exhibit robust oscillations in their root elongation rate in different photoperiods and in continuous light, and investigate the contribution of changes in the C supply and the clock to these diel rhythms.

MATERIALS AND METHODS

Plant material and growth conditions

Measurements were performed on Arabidopsis thaliana wild-type Col-0 and Ws-2 as well as on mutants in starch metabolism (pgm, Caspar et al. 1985; sex1, Caspar et al. 1991) and in circadian clock mutants in ELF3 (Covington et al. 2001), ELF4 (Doyle et al. 2002) and CCA1/LHY. The elf3 and elf4 mutants were in Col-0 background (Yu et al. 2001; Kolmos & Davis 2007). The double mutant cca1/lhy originated from Ws-2 accessions (Mizoguchi et al. 2002).

Seeds of all lines were surface-sterilized for 20 min with 10% sodium hypochlorite solution containing 0.1% surfactant (Triton X-100, Sigma-Aldrich, Munich, Germany), rinsed several times with sterile water and plated on the surface of solid nutrient agar (7.0% m/v) supplemented with half-strength Murashige–Skoog medium (Murashige & Skoog 1962; M02 555, pH 5.6; Duchefa, Haarlem, Netherlands). After 4 d stratification Petri dishes were placed vertically in the phytotron (21 °C constant day and night temperature, 100 µmol m−2 s−1 photon flux density, 12–12 h light–dark). At day 9, seedlings that had developed roots of >1 cm length were selected and distributed equally in a row 3 cm from the top on the surface of solid medium filled in 120 × 120 mm rectangular Petri dishes, with 15–25 seedlings per plate. After two further days in the phytotron, Petri dishes were used for measurement.

Image acquisition and root elongation analysis

Root growth kinetics was collected as described previously (Yazdanbakhsh & Fisahn 2009, 2010). In brief, a custom designed phytochamber housed the central measuring head of the plant root monitor (PlaRoM). The actinic photon flux density in the chamber at the surface of the leaves of the seedlings was 90 µmol m−2 s−1. Temperature was controlled by a cooling device providing 0.5 °C accuracy. The PlaRoM imaging platform screens the surface of two Petri dishes and captures time lapse records of the seedlings growing on them (Yazdanbakhsh & Fisahn 2009). Image stacks were collected by a CCD camera (Panasonic Colour CCTV Camera, WV-CP210/G, Matsushita Communication Industrial Co. Ltd., Yokohama, Japan) mounted on the video port of the microscope. To monitor the seedlings regardless of actinic light requirements, an infra red light source (Infra-Red Illuminator CE-7710, Jenn Huey Enterprise Co., Ltd., Taipei, Taiwan) provided measuring light. Screening of the Petri dish and capturing of time lapse records was controlled by the PlaRoM imaging software application (Yazdanbakhsh & Fisahn 2010). The magnification of the microscopes was set such that the video stream covered a 4.58 × 3.33 mm area of the surface of the Petri dish, and allowed images to be captured with a resolution of (5.96 µm × 5.78 µm) pixel−1. The root extension profiling software application analysed the time lapse records and provided the growth velocity profiles and custom specified visualization of root extension profiles (Yazdanbakhsh & Fisahn 2010).

Metabolite analysis

Shoots and roots of seedlings were separated and immediately frozen in liquid nitrogen. Following homogenization and sub-aliquoting at −70 °C, starch, sucrose, glucose, fructose and total amino acids were measured in roots and shoots as described in Cross et al. (2006).

RESULTS

Diel changes in root elongation in different photoperiods

Root growth kinetics of wild-type Col-0 seedlings were investigated in 8/16, 12/12 and 16/8 light/dark cycles (Fig. 1). The measurements were carried out for three to five consecutive days with 11- to 15-day-old plants that had been in that light/dark cycle from germination onwards. The average growth rate over a 24 h cycle was highest in a 16/8 (121 ± 2.3 µm h−1), intermediate in a 12/12 (78 ± 2.3 µm h−1) and lowest in an 8/16 (33 ± 1.1 µm h−1) light/dark cycle (Supporting Information Table S1).

Figure 1.

Diurnal changes in root extension growth rates in A. thaliana (Col-0) growing in different photoperiods. Plants were grown in a given photoperiod from germination onwards. Root extension rates of 11- to 15-day-old plants were monitored for three to five consecutive days. The result shows the mean absolute elongation rates for 71, 39 and 50 seedlings growing on 5, 5 and 3 separate plates in the 16 h/8 h, 12 h/12 h and 8 h/16 h light/dark photoperiod treatments, shown as blue squares, red dots and pink triangles, respectively. Black symbols denote the dark period. Error bars indicate the SE. Average elongation rates, as well as the timing and growth rate values of maxima and minima are listed in Supporting Information Table S1.

There were marked diurnal changes in all three light/dark cycles. The relative magnitude of the changes was smallest in the 16 (maximum and minimum rates of 157 ± 4.6 and 96 ± 3.4 µm h−1, n = 291), intermediate in 12 (123 ± 5.0 and 64 ± 3.0 µm h−1, n = 101) and largest in 8 h light periods (51 ± 2.3 and 21 ± 1.6 µm h−1, n = 239). At first glance, the diurnal growth profiles vary between the three light/dark cycles, but closer inspection indicates some common patterns. Firstly, after illumination, extension rates rose to a transient maximum after 1.5–2 h. Secondly, after darkening, growth was depressed for 1–2 h. Thirdly, these rapid transient changes are superimposed on a more gradual oscillation, whose timing is largely independent of the length of the light period. Extension rates decrease to a minimum at 9–11 h after dawn, and gradually recover during the remaining 13–15 h (Fig. 1).

We next investigated whether the rhythms in root elongation persist in absence of a light–dark cycle. To do this, wild-type Col-0 seedlings were entrained in light/dark cycles and then transferred to continuous darkness (Fig. 2) or continuous light (Fig. 3).

Figure 2.

Root extension growth of A. thaliana (Col-0) seedlings in an extended night. Plants were grown in a 16 h photoperiod from germination onwards. Root extension rates of seedlings were monitored between 13 and 18 d. At time 0, when the seedlings were 15 d old, the night was extended for 4 d. Dark grey areas indicate subjective night; light grey areas indicate subjective day. Lower trace: Relaxation kinetics in the absence of exogenous sucrose (n = 8 seedlings). Upper trace: Relaxation kinetics in the presence of 1% extracellular sucrose (n = 11 seedlings). The results show the mean absolute elongation rates of each group. Error bars indicate the SE. The black arrow indicates the point of inflection of root growth in the subjective day.

Figure 3.

Oscillations in the rate of root extension growth of A. thaliana (Col-0) after transfer to continuous illumination. Subjective nights are marked by light grey shading; dark periods by dark grey. (a) Direct transfer from 12/12 h diurnal cycles to continuous light. Plants were grown in 12 h photoperiod from germination onwards. Root extension rates of 19-day-old seedlings were monitored during a 12/12 h light cycle followed by 5 d of continuous illumination (n = 9). (b, c) Recovery of root elongation growth in continuous light after 4 d of darkness. Prior to darkness plants were entrained to photoperiods of 16 h [(b), n = 10, 18 d] or 12 h [(c), n = 11, 15 d] light from germination onwards. They were then darkened for 4 d, before transfer to continuous light. The plots exhibit the mean absolute elongation rate values and the error bars indicate the SE.

Root elongation rates in continuous darkness

To investigate the response after transfer to continuous darkness (Fig. 2), seedlings were grown in a 16/8 h light/dark cycle to maximize the length of the period in which the responses could be compared. Seedlings were transferred to continuous darkness at the end of the night. One group of seedlings was grown without sucrose in the medium, and another was supplied with 1% sucrose to support growth in the long dark treatment (n = 16, n = 22, respectively). During the last light–dark cycle, the pattern of root extension in seedlings grown without sucrose resembled that in 16 h photoperiod treatment of Fig. 1; root elongation showed a transient maximum early after light on, followed by a gradual decline for the next 7–8 h and a gradual recovery that started in the last part of the light phase and continued throughout the night, but was interrupted by a transient inhibition after darkening. Inclusion of 1% sucrose in the medium led to a ca. 50% increase in the growth rate, but the diurnal changes resembled those in the absence of sucrose.

After transfer to continuous darkness, the transient increase seen after illumination was abolished. Root extension was maintained for 1–2 h into the subjective light period, and was then decreased (Fig. 2). In the absence of sucrose, elongation was rapidly inhibited, decreasing by >50% during the initial 10 h, and being completely inhibited by 24–28 h. In the presence of exogenous sucrose, the rate of root extension stabilized and started to recover 8–9 h after the anticipated dawn (Fig. 2, arrow), that is, at the time at which growth would start to recover in a light/dark cycle, and rose to a weak maximum in the middle of the subjective night. On the following days, growth continued to decline. This gradual decline may occur because 1% exogenous sucrose does not fully compensate for the lack of photosynthesis in prolonged darkness, or may also be due to additional signals that are derived from light and are needed to promote growth. Thus, an oscillation with a period of about 24 h appears to be superimposed on the gradual decline.

Root elongation rates in free-running continuous light

To investigate the response after transfer to free-running continuous light (Fig. 3) seedlings were grown without exogenous sucrose in a 12/12 light/dark cycle, and transferred to continuous light at the start of the light period (Fig. 3a). During the last light–dark cycle, root growth showed a transient maximum in the first 1–2 h of the light period, followed by a decline until 9–10 h into the light period, and a gradual recovery during the night (as already seen in Figs 1 & 2). After transfer to continuous light the slow oscillations persisted, with a maximum at the subjective dawn, and a minimum towards the end of the subjective day (Fig. 3a). There was a trend to the minima being shifted forwards slightly. These oscillations were superimposed on a gradual increase in the rate of root extension, which rose after 3 d of continuous light to a value that was almost double that seen in a 12/12 light/dark cycle. The transients after illumination or darkening were abolished in a free-running cycle.

In a second set of experiments, plants growing in the absence of exogenous sucrose were entrained to photoperiods of 16/8 (long day) or 12/12, and then dark-adapted for 4 d. Growth completely ceased under these conditions. They were then re-illuminated in continuous light (Fig. 3b, c). Root growth recovered slowly, with almost no growth in the first 24 h, and a gradual rise over the next 5 d. Oscillations were superimposed on this gradual recovery, with a maximum at, or just after, the anticipated dawn, and a minimum before the anticipated dusk (Fig. 3b, c). The sharp transients after illumination or after transfer to darkness were again absent.

The results in Figs 1–3 point to several conclusions. Firstly, in our growth conditions, the overall rate of root growth is limited by C. The average rate per 24 h cycle can be increased by providing exogenous sucrose, by extending the photoperiod, or by transfer to continuous light. Secondly, the use of C is tightly regulated to maintain reserves until the end of the 24 h diel cycle and to avoid acute limitation of growth by C. Across a wide range of photoperiod treatments, root extension rates increase through to the end of the night. Further, plants retain enough reserves to support a further 1–2 h growth when the night is extended. Thirdly, root extension growth shows strong diel changes, which can be divided into two components: (1) a rapid and transient stimulation at the start of the light and inhibition at the start of the night period, which depend on the occurrence of a light–dark transition; and (2) slower oscillations whose timing is largely independent of the length of the light period. These slow oscillations are maintained in continued light and, to a certain extent, in continuous darkness.

Further experiments were performed using mutants to investigate the importance of diurnal C allocation and (see further discussion) the circadian clock for the regulation of diurnal root extension within each 24 h cycle.

Root elongation kinetics in a starchless and a SEX mutant

Wild-type plants accumulate starch in the light and degrade it at night to provide C to support metabolism and growth in the dark. Two mutants were investigated in which the supply of C at night was disturbed in different ways. The pgm mutant lacks plastidic PGM and is unable to synthesize starch (Caspar et al. 1985). As a result, this mutant becomes acutely C-limited 3–4 h into the night (Gibon et al. 2004a; Usadel et al. 2008). The sex1 mutant lacks a glucan water dikinase (GWD), which is required to phosphorylate starch, and is impaired in starch degradation. This mutant contains large amounts of starch, which is only broken down slowly during the night (Caspar et al. 1991; Yu et al. 2001). The mutants and the corresponding Col-0 wild type were grown in a 16 h/8 h light/dark cycle with no added sucrose (open symbols), or in the presence of 1% sucrose (filled symbols; Fig. 4).

Figure 4.

Diurnal root elongation rates in Col-0, pgm and sex1 in presence and absence of external sucrose. Twelve-day-old seedlings of Col-0, pgm and sex1 growing in 16 h photoperiod were transferred to two Petri dishes filled with solid agar medium, one of which contained additionally 1% sucrose. Three days later, root elongation was monitored for 5 d and the 24 h averaged root elongation pattern of each genotype/sucrose supply was calculated (mathematical equations in Yazdanbakhsh & Fisahn 2010). Col-0 is represented by red circles (18 individuals, n = 97, 24 individuals, n = 121). Empty symbols represent seedlings in absence of sucrose and filled symbols represent seedlings in the presence of 1% sucrose, pgm is shown by blue triangles (20 individuals, n = 102, 16 individuals, n = 84) and sex1 by green squares (6 individuals, n = 36, 4 individuals, n = 20, respectively). Plots exhibit the mean hourly absolute elongation rate values and error bars indicate the SE.

Col-0 wild-type plants showed similar diel changes of growth to those seen in Figs 1–3, with a gradual decline in the later part of the light period, and a gradual recovery during the night. The transient peak at the start of the light period was less marked than in other experiments, possibly due to the lower temporal resolution of the growth measurements in this experiment. The diel response was strongly modified in pgm and sex1. Both mutants showed a strong inhibition of growth during the night and a gradual recovery of growth during the light period. The lag until extension growth was re-established after illumination was longer for pgm (3–4 h) than sex1 (within 1 h). Like wild-type plants, both mutants showed a slight minimum during the light period at 8–10 h, followed by a weak but sustained increase during the rest of the light period.

Inclusion of 1% sucrose in the medium led to a stimulation of root extension in wild-type plants (see also Fig. 2), pgm and sex1. Although addition of sucrose did not alter the diel changes of root growth in wild-type plants (see also Fig. 2), it led to a marked change in the mutants. In particular, the inhibition of extension growth during the night was completely reversed in sex1, and almost completely reversed in pgm. These results show that the inhibition of root extension growth in the night in these mutants is due to a lack of sugars. This is presumably the consequence of the absence of starch (pgm) or the slower rate of starch breakdown (sex1).

Together, these results show that starch turnover provides a reserve of C to fuel growth in the night and, more generally, that correct allocation of C is important to maintain root extension growth through the entire 24 h cycle. They also show that a shortfall of C in the night leads to an impairment of root growth, including an acute inhibition of growth in the night and a delay before growth resumes in the following light period. This lag is already visible after a few hours of C depletion (Fig. 4). It extends to one or more days when plants are exposed to C starvation for more than 24 h (Fig. 3).

Root elongation kinetics in circadian clock mutants

The robust rhythms in root growth kinetics in light/dark cycles (Fig. 1) and free running conditions (Figs 2 & 3) indicate that the circadian clock is involved in control of root growth. We therefore analysed the growth kinetics of three Arabidopsis mutants that are known to be affected in circadian clock function: the double mutant cca1/lhy (circadian clock associated/late elongated hypocotyl; Mizoguchi et al. 2002), elf3 (Covington et al. 2001) and elf4 (Doyle et al. 2002) (Figs 5 & 6). The mutants and the corresponding wild-type lines were grown in a 12/12 light/dark cycle, with zero (open symbols) or 1% (closed symbols) sucrose in the medium.

Figure 5.

Diurnal growth rates of Arabidopsis thaliana Ws-2 wild type and circadian clock double mutant cca1/lhy in presence and absence of external sucrose. Seventeen-day-old seedlings growing in 12 h photoperiod from germination were transferred to two Petri dishes filled with solid agar medium, one of which contains additional 1% sucrose. Three days after transfer to new plates, root elongation was monitored for 4 d. The 24 h averaged root elongation pattern of each genotype/sucrose condition was calculated (based on Yazdanbakhsh & Fisahn 2010). Empty symbols represent seedlings in absence of sucrose and filled symbols represent seedlings in the presence of 1% sucrose, Ws-2 is represented by red circles (10 individuals, n = 40, 6 individuals, n = 24) and cca1/lhy by green squares (6 individuals, n = 24, 5 individuals, n = 20, respectively), empty symbols represent seedlings in absence of sucrose and filled symbols represent seedlings in the presence of 1% sucrose. Plots exhibit the mean hourly absolute elongation rate values together with the SE.

Figure 6.

Diurnal growth rates of Arabidopsis thaliana Col-0 and circadian clock mutants elf3 and elf4 in presence and absence of external sucrose. Seventeen-day-old seedlings growing in 12 h photoperiod from germination were transferred to two Petri dishes filled with solid agar medium, one of which contains 1% sucrose. Three days after transfer to new plates, root elongation was monitored for 4 d. The 24 h averaged root elongation pattern of each genotype/sucrose condition was calculated (based on Yazdanbakhsh & Fisahn 2010). Empty symbols represent seedlings in absence of sucrose and filled symbols represent seedlings in the presence of 1% sucrose, Col-0 is represented by red circles (eight individuals, n = 32, nine individuals, n = 36), elf3 by cyan triangles (seven individuals, n = 28, nine individuals, n = 36) and elf4 by blue squares (10 individuals, n = 40, 7 individuals, n = 28, respectively). Plots exhibit the mean hourly absolute elongation rate values together with the SE.

The cca1/lhy double mutant was in a Ws-2 background. Ws-2 wild-type plants (Fig. 5) exhibited similar diurnal root extension growth rhythms to Col-0 wild type. As in Col-0, these diurnal changes were maintained in the presence of 1% sucrose, although the relative amplitude was smaller. The cca1/lhy double mutant showed a strongly modified diurnal rhythm (Fig. 5). Root growth rates were similar to those of wild-type Ws-2 in the later part of the light period, but remained steady rather than rising in the first part of the night, and declined almost twofold in the second part of the night. After illumination in the morning, the rate of root elongation was initially lower than in wild-type Ws-2, but rose to a similar rate after 2–3 h in the light. As a result, growth rates in cca1/lhy were at a minimum at the end of the night, rather than towards the end of the day as in wild-type plants. The inhibition of growth in cca1/lhy in the second part of the night was largely relieved when 1% sucrose was included in the medium. This indicates that low C limits root growth of cca1/lhy at the end of the night.

The elf3 and elf4 mutants are in a Col-0 background. The diurnal response for wild-type Col-0 (Fig. 6) resembled that seen in Figs 1–3. As already seen (Fig. 2), inclusion of 1% sucrose in the growth medium led to a higher overall growth rate (here twofold higher) but did not alter the diurnal rhythm in wild-type Col-0. elf3 and elf4 showed increased rates of root extension in the second part of the light period, and lower rates of growth in the second part of the dark period (Fig. 6). These changes were especially marked for elf3, which showed a twofold higher growth rate in the light period, and a twofold inhibition during the night. As a result, extension growth rates in elf3 were highest at the end of the light period, and lowest at the end of the night. This resembles the pattern in cca1/lhy. In contrast to cca1/lhy, including 1% sucrose in the medium did not relieve the inhibition of root growth at the end of the night in elf3 and elf4. This indicates that the decline of root extension growth rates during the night in elf3 and elf4 is not primarily due to a lack of carbohydrate at the end of the night.

Diurnal changes of starch and sugars in shoots and roots of wild-type Ws-2 and Col-0 and cca1/lhy and elf3

Graf et al. (2010) reported that cca1/lhy mutants show a modified starch turnover phenotype, with starch being exhausted by ZT20-22 (i.e. 2–4 h before dawn), corresponding to the time at which dawn is anticipated in cca1/lhy in a light/dark cycle. Their experiments were carried out with 23-day-old plantlets growing on soil. We investigated whether the decrease in root growth rate towards the end of the night in cca1/lhy seedlings might also be due to a premature exhaustion of starch. We carried out similar measurements in elf3 because this mutant showed a similarly strong inhibition of root extension during the night that, in contrast to cca1/lhy, was not relieved by adding exogenous sucrose (see earlier discussion). The mutants and the corresponding wild-type lines were grown in a 12 h/12 h light/dark cycle in the absence of exogenous sucrose. Four replicate samples were harvested at dawn, after 2 and 12 h light, and after 10 and 12 h darkness (i.e. 22 and 24 h after dawn). On the day of harvest, the night was also extended for a further 4 h, to allow a further set of plants to be harvested 28 h after the previous dawn. The plants were separated into shoot and roots, and then analyzed for carbohydrates, amino acids, nitrate and protein content (Fig. 7, see Supporting Information Table S2 for a summary of all the measurements).

Figure 7.

Diurnal turnover of starch (a) and sucrose (b) in Col-0, elf3, Ws-2, and cca1/lhy. Plants were grown in 12 h photoperiod for 17 d before harvest. The results show the mean and standard deviation of four biological replicates of each genotype. Samples harvested during the dark period are denoted by a grey background.

In wild-type Col-0 and Ws-2, starch increased during the light period and decreased at night, with about 17 and 20% of the initial starch content remaining at 22 h, 8.7 and 11.6% at 24 h, and almost none (4.7 and 7.0%) after an extension of the night for another 4 h (Fig. 7a). Leaf sugars rose slightly (Col-0) or about two fold (Ws-2) during the light period and decreased during the night. Leaf sugars decreased by a further two- to threefold when the night was extended by 4 h. Root sugars rose twofold in wild-type Col-0 and Ws-2 in the light, and decreased during the night, with a further twofold decrease when the night was extended for 4 h. Starch levels in the root were negligible (<0.25 and <0.1 µmol hexose equivalents/g FW in Ws-2 and Col-0, respectively, Supporting Information Table S2). The absolute levels of starch and sugars in these young wild-type seedlings are lower than in older plants (see, e.g. Gibon et al. 2004a; Bläsing et al. 2005; Gibon et al. 2009; Graf et al. 2010). Nevertheless, the diurnal changes of carbohydrates resemble those reported for older plants, with a large ca. 10-fold increase of starch in the day and almost complete re-mobilization of starch during the night, smaller (twofold or less) changes of sugars during the light/dark cycle, and complete exhaustion of starch and a decrease of sugars when the night is extended. These diurnal changes of sugars are consistent with the observation that root extension growth is maintained until the end of the night, but is inhibited when the night is extended by 3–4 h or more.

The diurnal changes of carbohydrates were modified in a different manner in the two clock mutants. The cca1/lhy mutant accumulated marginally (although not significantly) higher levels of starch at the end of the day than Ws-2 wild-type (Fig. 7a). During the night, starch was degraded more quickly, with only 5.4 and 4.4% remaining after 22 and 24 h, respectively (see following for more data: Fig. 11). This resembles the levels in wild-type plants after a 4 h extension of the night. Leaf sugars in cca1/lhy resembled those in Ws-2 wild type at the end of the light period, but were strongly depleted at 22 h and 24 h when they resembled those seen in wild-type plants after a 4 h extension of the night (Fig. 7b). The depletion of sugars in the last hours of the night in cca1/lhy included a large decrease of sucrose and a smaller decrease of fructose, whereas glucose was unaltered. Root sugars in cca1/lhy were also depleted in the last hours of the night, falling to levels similar to those seen after a 4 h extension of the night in wild-type plants (Supporting Information Table S2). These results extend the finding that cca1/lhy exhausts its starch prematurely (Graf et al. 2010) to young seedlings. Indeed, while in Graf et al. (2010), the double mutant accumulated marginally less starch than Ws-2 in the light, in the present study cca1/lhy accumulated the same amount or marginally more starch than wild-type Ws-2. As a result, cca1/lhy shows not only changes in the timing of starch breakdown but also a higher absolute rate of starch breakdown than wild-type Ws-2. Further, we now demonstrate that the premature exhaustion of starch in cca1/lhy is accompanied by premature depletion of sugars in the shoot and in the root. This is consistent with the notion that an inappropriate timing of starch degradation is contributing to the decline in root growth at the end of the night in cca1/lhy.

Figure 11.

Diurnal turnover of starch (a) and sucrose (b) in wild type (Ws-2) and cca1/lhy double mutants in a 12 h light cycle and a 4 h advanced transition to dark. Pre-dusk is indicated by a dark grey bar; the light grey bar denotes the regular 12 h night. The result shows the average of four biological replicates and standard deviation. Samples taken in a 12 h light/dark cycle are marked in green while the premature night samples are in red. (a) Starch content in shoots of Ws-2 (empty squares) or cca1/lhy (filled squares) in a regular 12 h cycle (green) and the first night of transition to 8 h photoperiod (red). (b) Diurnal changes of sucrose levels in shoots of Ws-2 (empty squares) and cca1/lhy (filled squares) seedlings growing in a regular 12 h cycle (green) and the first night of transition to an 8 h photoperiod (red).

In contrast, elf3 exhibited higher levels of starch than the Col-0 wild type at all times of the diurnal cycle (Fig. 7a). The relative difference was especially large (>twofold) at the end of the night. Leaf sugar levels were increased in elf3 compared with Col-0 wild type at most times in the diurnal cycle, with an especially large increase at the end of the night (mainly due to glucose and fructose, see Supporting Information Table S2) and the beginning of the day (due to higher levels of sucrose, glucose and fructose, see Supporting Information Table S2).

We also investigated the diurnal changes of total amino acids in Ws-2 wildtype, cca1/lhy, Col-0 wild type and elf3 (Supporting Information Table S2). As previously seen (Bläsing et al. 2005; Gibon et al. 2006), total amino acids increased slightly in the light and decreased slighly in the night. Total amino acids were marginally lower in cca1/lhy than in Ws-2 wild type throughout the entire diurnal cycle. They were marginally higher in elf3 than Col-0 wild type at most times in the diurnal cycle. However, there was no evidence that either of these mutations led to a major depletion of amino acids.

Root elongation kinetics of clock mutants in free running conditions

To further investigate the effect of mutations in CCA1/LHY, ELF3 and ELF4 on root elongation, we investigated the response after transfer from a 12/12 light–dark regime to continuous illumination (Fig. 8). The diurnal growth kinetics of the wild-types Col-0, Ws-2 and the three clock mutants in the entraining light/dark cycle in Fig. 8 resembles that already shown in Figs 5 and 6; wild-type Col-0 and Ws-2 showed a rise in growth throughout the night, cca1/lhy rose to a peak in the middle of the night, followed by a decline to a minimum at the end of the night, elf4 showed a rise in the first part of the night and slight decline towards the end of the night, while elf 3 showed a maximum at the end of the light period and a decline throughout the entire night.

Figure 8.

Root growth rates of Col-0, Ws-2, elf3, elf4 and cca1/lhy in free running conditions of continuous illumination. Seedlings were grown in 12 h photoperiod from germination. Root elongation of 18-day-old seedlings was monitored during the last cycle of a 12 h photoperiod and the following 5 d of continuous illumination. The plots show the 2 h averaged absolute elongation rate of each genotype (Col-0: red circles, n = 9, Ws-2: red downward triangles, n = 8, elf3: cyan triangles, n = 9, elf4: blue squares, n = 7, cca1/lhy: green diamonds, n = 5). Shaded areas represent subjective nights. The x-axis indicates time in continuous light. Error bars indicate the SE.

Transfer of Col-0 (see also Fig. 3a) or Ws-2 wild type to continuous light results in a gradual doubling of the absolute extension rate, with superimposed strong oscillations that have a maximum at the subjective dawn and a minimum towards the end of the subjective day. The mutants showed contrasting responses in continuous light (Fig. 8). The cca1/lhy double mutant displayed arrythmic root elongation kinetics during the first 48 h of continuous illumination. After 48 h continuous illumination cca1/lhy recovered a weak oscillatory growth phenotype, with a maximum at or just after the subjective dawn and a minimum towards the end of the subjective light period. However, individuals lost synchrony during this prolonged light treatment, as is revealed by the large error bars. Averaged growth rates in cca1/lhy were similar to wild type in the entraining light/dark treatment, but lower than wild-type plants in continuous light. In elf3, the depression of root extension during the subjective light period was abolished. The slow oscillations in continuous light were almost completely absent, and growth rates increased progressively for 36 h to a new and fairly stable plateau, which was about 60% higher than the average growth rate in wild-type Col-0 in continuous light. These results indicate that a clock-mediated restriction of growth in the light is abolished in elf3. In elf4, the timing of the maximum moved forward slightly to the middle of the subjective night. The responses of root elongation in free-running cycles have similarities to published changes in other circadian rhythm phenotypes in these three clock mutants (see Discussion).

These results indicated that the clock regulates root growth in two different ways. One mechanism involves ELF3, is independent of starch turnover and restricts root extension in the subjective light period and promotes it in the subjective night. The other involves CCA1/LHY and impacts on root growth by regulating the turnover of starch and the supply of sugars for growth in the night.

Oscillations in root elongation kinetics of starch mutants in free running conditions

To provide evidence that the slow oscillator is itself independent of starch turnover, we investigated root growth kinetics of pgm and sex1 in free running conditions in continuous illumination (Fig. 9). Prior to continuous light exposure, seedlings were forced to cease growth completely by treatment in continuous darkness for 4 d. As already seen, there was a delay of about 36 h until growth resumed, and growth rates continued to rise until at least 120 h after re-illumination. Once growth resumed, wildtype Col-0, pgm and sex1 exhibited synchronized oscillations in root elongation rate, with maximal extension growth 2–4 h after the anticipated dawn and minima 8–10 h later. Thus, continuous light induces wild-type-like oscillations of root extension rates in starchless and starch degradation mutants.

Figure 9.

Root growth recovery in Col-0, pgm and sex1 seedlings in continuous light after 4 d of continuous darkness. Col-0 (n = 9), pgm (n = 4) and sex1 (n = 5) seedlings were grown in a 12 h photoperiod from seed stage. Prior to growth measurement, seedlings were kept in continuous dark for 4 d, which resulted in complete cessation of root elongation. Subsequently, the 18-day-old seedlings were exposed to continuous illumination and root elongation was monitored for 128 h. Light grey bars denote subjective nights. Dark grey indicates the 4 d prolonged night. Each trace represents the 2 h mean absolute elongation rates together with the SE.

Adjustment of root growth, starch turnover and sugar levels to a sudden early dusk

Finally, we investigated the response to a sudden imposition of an early dusk. Graf et al. (2010) showed that sudden premature darkening of 23-day-old Arabidopsis plants leads to an immediate decrease in the rate of starch breakdown, compared with the rate in plants that were darkened at the end of the light period. As a result, starch reserves last until the end of the night. They argued that the clock acts as a timer, and that this information is used by an unknown mechanism to adjust the rate of starch breakdown to the anticipated length of the night. However, a decreased rate of starch breakdown, on its own, would not allow a plant to avoid a period of C starvation at the end of the night. This will also require a concomitant decrease in the rate of C consumption (see Introduction). Our experimental set-up allowed us to investigate the kinetics of root extension growth after subjecting plants to a sudden early dusk and directly test if the decrease in the rate of starch breakdown in the first night after a premature dusk is accompanied by an immediate decrease in the rate of root growth.

Two separate experiments were performed: one with Col-0 wild type (Fig. 10a) and one with Ws-2 wild type and the cca1/lhy double mutant (Fig. 10b). In both experiments, plants were grown in a 12 h/12 h light/dark cycle for 13 d, and then subjected to an early night after 8 h in the light. These plants experienced a 16 h (instead of a 12 h) dark period, using reserves that had been built up in 8 h (rather than 12 h) of light. The early dusk was retained for the following 2 d. Root extension growth was monitored, starting a full day before the early dusk was imposed, and continuing for 2 d after the transition.

Figure 10.

Adjustment of root growth rates to an unexpected pre-dusk. (a) Root elongation of Col-0 seedlings growing in a 12 h photoperiod was monitored for 1 d in a 12 h photoperiod followed by 3 d of 8 h illumination. Transition to 8 h cycles was performed on 15-day-old seedlings (n = 11). (b) Kinetics of root elongation in Ws-2 wild type and cca1/lhy double mutants during 1 day of a 12 h photoperiod followed by 3 d of 8 h illumination (n = 5, n = 6, respectively). Seedlings were grown in 12 h cycles for 16 d prior to the experiment. Dark grey shadings denote regular night time; light grey indicates darkness, which is applied earlier. The results exhibit the mean hourly averaged absolute elongation rates of each genotype. Error bars indicate the SE.

Root extension growth kinetics of Col-0 and Ws-2 in a 12/12 light/dark cycle resembled those in previous experiments (Fig. 10a & b). In both wild types, an early unanticipated dusk led to an almost immediate decrease in the rate of root extension, compared with roots in a 12 h photoperiod (Fig. 10a & b). The precise kinetics are difficult to resolve in the first 2–3 h after darkening, because darkening anyway leads to a transient decrease in the rate of root extension (see earlier discussion), and because the transition after 8 h illumination is close to the inflection point in the slow oscillation in extension rates. However, a clear additional inhibition of extension growth can be seen by 3–4 h after the premature dusk. Whereas the rate of root growth rises strongly during the night in a 12/12 cycle, it only increases slightly in the early dusk treatment. In the last 3–4 h of the night, growth declines in Col-0 but remains stable in Ws-2. Compared with the rate at the end of a normal night, the rate of root extension at the end of the first extended night is decreased >threefold in Col-0 and approximately twofold in Ws-2.

On the day after the premature dusk, the normal diurnal pattern was re-established, with a peak 2–3 h into the light period, a decline later in the light, and a gradual recovery to a maximum at the end of the night. This pattern was retained on the second day after the transfer. The absolute rates are about twofold lower than in 12/12 light dark cycles; this resembles the growth responses in plants that were grown from germination in a 12/12 or a 8/16 light/dark cycle (Fig. 1; Supporting Information Table S1).

The cca1/lhy doublemutant showed a different response to a premature dusk (Fig. 10b). The transient inhibition after darkening is followed by an increase in the rate of growth, similar to that seen in a normal night. Growth is then strongly inhibited in the last hours of the night. Further, cca1/lhy does not adjust to the long night, but instead continues to show a strong inhibition of growth in the last part of the night for at least the next two nights.

Ws-2 wild-type and cca1/lhy plants were harvested to measure shoot carbohydrate levels during a 12/12 light dark cycle, and on the first night after the premature dusk (Fig. 11, see Supporting Information Table S3 for the original data). Starch was degraded in a near-linear manner in wild-type Ws-2 (Fig. 11a, see also Fig. 7a). Premature darkening led to a halving of the rate of starch degradation, with the result that starch reserves lasted through the lengthened night (Fig. 11a). There was also an immediate decrease in the levels of sucrose (Fig. 11b) and glucose (Supporting Information Table S3). In cca1/lhy, starch accumulated slightly more rapidly in the light and was degraded more rapidly at night in a 12/12 light/dark cycle (Fig. 11a, see also Fig. 7a). When cca1/lhy was exposed to an early night, the rate of starch degradation remained similar to that in a 12/12 cycle, and was >twofold higher than in wild-type Ws-2 after a premature dusk. As a result, after a premature dusk, cca1/lhy depleted its starch several hours before the end of the night. Furthermore, sucrose levels in cca1/lhy were higher than in wild-type Ws-2 in the first part of the night, and lower in the last part of the night (Fig. 11b).

These results show that in wild-type Arabidopsis the rate of starch degradation adjusts immediately to an early dusk, resulting in a lower rate of starch degradation and conservation of starch reserves until the end of the night (see also Graf et al. 2010). Crucially, the decrease in the rate of starch degradation is accompanied by a rapid decrease in the rate of root extension growth. Further, cca1/lhy is unable to adjust the rates of starch degradation or root growth to this sudden challenge.

DISCUSSION

Arabidopsis exhibits a strong diel rhythm for root extension growth

Arabidopsis exhibits a diel rhythm in primary root extension growth in a range of different photoperiods, and in the absence and presence of external sucrose. This extends previous reports of diel rhythms in Arabidopsis root elongation (Yazdanbakhsh & Fisahn 2009, 2010). There appear to be at least two components: (1) a gradual oscillation in which growth decreases in the first part of the 24 h cycle and then recovers; and (2) transient changes after illumination and darkening. These components were observed repeatedly in independent experiments.

The gradual inhibition of root elongation during the major part of the light period, and gradual recovery of extension growth during the remainder of the 24 h diel period was seen in a range of photoperiods (Figs 1, 2, 3a, 4–6, 8 & 10). This gradual oscillation still occurs in the presence of exogenous sugar (Figs 2 & 4–6) and is retained in free running conditions (Figs 2, 3b–c, 8 & 9), indicating that it is not due to changes in the C supply but is, rather, driven by the clock (see below for further discussion).

The sharp transient stimulation of growth 1–2 h after illumination and the transient inhibition of growth 1–2 h after darkening were also seen in all photoperiods (Figs 1, 2, 3a, 4–6, 8 & 10). The decline after darkening was less marked in short days, possibly because of the lower absolute elongation rates in this photoperiod. These transient changes were absent in free running conditions (Figs 2, 3, 8 & 9), indicating they are a direct response to illumination and darkening. They were in general still seen in the presence of external sucrose, indicating they are not just due to a temporary excess and shortfall of sugars after illumination and darkening, respectively.

Although earlier studies often had less temporal resolution, they provide evidence for qualitatively similar diel rhythms in root growth. Head (1965) reported a significant increase in the average rate of root growth in the night compared with the day in cherry (Prunus avium, variety Early Rivers). Schmidt et al. (2010) recently reported similar patterns of root growth in a study of the jasmonic acid response of Arabidopsis growing in a 12 h photoperiod.

Leaves of dicotyledonous species often exhibit maximal growth activity at dawn (Schmundt et al. 1998; Walter & Schurr 2000, 2005). A similar qualitative pattern was observed with detached leaf discs (Biskup et al. 2009). The Arabidopsis accessions Ler and Col-0 showed a maximum in leaf relative growth rates soon after dawn, with a subsequent decrease of growth during the day and a minimum early in the night in 12–12 h light–dark cycles (Wiese et al. 2007). Hypocotyl growth of wild-type Arabidopsis also exhibits a decline in the light and a maximum at the end of the night (Nozue et al. 2007). Taken together, roots, leaves (Wiese et al. 2007) and hypocotyls (Nozue et al. 2007) of Arabidopsis display a qualitatively similar diel growth pattern, with increasing growth rates in the night, although the precise timing of the rise depends on the organ.

The average diel growth rate is gauged to the daily C supply

Root growth is known to be highly dependent on C availability (Aguirrezabal et al. 1994; Muller et al. 1998; Freixes et al. 2002). The photoperiod and sucrose addition treatments used in Figs 1–3 imposed varying degrees of C limitation on the plants and resulted in a fourfold difference in the average diel root growth rate.

The response of root growth to the C supply may include a fairly rapid and a slow component. When plants are transferred from a 12 h light/12 h dark cycle to continuous light (Fig. 3a, see also Fig. 8) there was an enhancement of root growth in the first subjective night. However, there was a further and marked increase in the average diel growth rate for the next 48–60 h. When plants were darkened for several days and then transferred to continuous light (Figs 3b–c & 9) the growth rate was very low in the first 24 h and rose gradually over the next 3–4 d. In both cases, the gradual increase in the average diel growth rate was superimposed on a circadian rhythm, in which growth decreased in the subjective light period and increased in the subjective night.

Thus, these results show that the average diel growth rate depends on the C supply, and indicate that it is regulated by a mechanism that requires several days to be reset to allow faster growth after a change in the conditions. While it is known that large numbers of genes that are required for growth are transcriptionally repressed during the night and activated in the light period (Bläsing et al. 2005; Usadel et al. 2008) and many of these diurnal changes are at least partly due to changes in the sugar supply (Usadel et al. 2008) it is not yet known how this is translated into an appropriate rate of growth. One possible general explanation for the slow response of growth after a change in the conditions may be that several days are required until changes in transcript levels lead to a change in the levels of the encoded proteins (Gibon et al. 2004a; Piques et al. 2009).

The timing of starch turnover and root growth is regulated to avoid depletion of C and inhibition of growth at the end of the night

The gradual decline in root growth in the first part of the 24 h cycle and the recovery to a stable maximum in the second part of the 24 h cycle is seen in all photoperiods and sugar treatments. This implies that the diel timing of growth is regulated to avoid premature depletion of C at the end of the night, and that this balance is achieved across a wide range of C supplies and overall growth rates. Indeed, the relative changes in diel growth were larger in short than long days (Fig. 1), and were larger in the absence than the presence of exogenous sucrose (Figs 2 & 4–6). Our time-resolved measurements of root growth imply that sufficient C remains at the end of the night to support about 1–2 h of root growth (Fig. 2). Similarly, Wiese et al. (2007) concluded that leaf growth is not limited by the C supply at any time during the diel cycle in wild-type Arabidopsis growing in a 12 h light/12 h dark cycle.

Avoidance of C starvation during the diel cycle will require appropriate allocation of C to storage pools in the light and remobilization of these pools in the dark. As discussed in the Introduction, starch is essential for growth in a light/dark cycle and starch turnover provides a supply of C through to the end of the night. Premature exhaustion of starch in pgm and sex1 leads to a rapid and strong inhibition of root growth in the night, which can be prevented if sugars are supplied in the medium (Fig. 4). These results underline the importance of starch turnover in maintaining a supply of C for growth throughout the diel cycle.

The nocturnal inhibition of growth in starch turnover mutants is not reversed immediately after re-illumination, especially in pgm, confirming conclusions drawn from a whole plant C balance analysis (Gibon et al. 2004a). It has been shown previously that the exhaustion of starch is accompanied by transcriptional changes of hundreds of genes indicative of C starvation, and the appearance of low metabolites indicative of the onset of catabolism and autophagy (Bläsing et al. 2005; Gibon et al. 2006; Usadel et al. 2008). Our results imply that this cessation of growth and switch to catabolism cannot be immediately reversed. This is especially apparent after a longer period of starvation, when one or more days are required to re-establish growth (Figs 3b–c & 9).

It is intriguing that while photosynthetic C gain occurs in the light period, the slow diel oscillation results in a high rate of root growth in the second part of the night. Nozue et al. (2007) observed a similar or even more extreme phenomenon in hypocotyl growth, which increases strongly towards the end of the night. They suggested possible reasons for this diel response; it may allow plants to buffer against acute changes in light by only responding to extended periods of darkness, or time their growth to coincide with maximum water availability. Our results point to a further possibility: lower rates of growth during the first part of the diel cycle and a gradual recovery during the night will favor the accumulation of starch reserves during the light period, and shield against premature use of C reserves during earlier phases of the 24 h cycle. This would minimize the risk of a combination of excessive growth early in the 24 h cycle and later unexpected events that lead to premature exhaustion of C before the end of the night.

The clock contributes to the regulation of root extension growth

Several lines of evidence point to a central role for the circadian clock in the diel regulation of root growth. First, the slow oscillation is maintained in free-running continuous light (Fig. 3b & c). It is also retained in continuous darkness (Fig. 2), although less strongly and only when external sucrose is provided. To our knowledge, this is the first observation of rhythmic root growth in continuous light. The slow rhythm in free-running continuous light was independent of any possible input from starch metabolism (Fig. 9). Secondly, the gradual oscillation was profoundly modified in the clock mutants elf3, elf4 and cca1/lhy. These mutants all exhibited increased growth in the first part of the 24 h cycle, and less growth in the second part of the cycle. These mutants also modified the response after transfer to continuous light, but in differing ways. The circadian oscillation of root extension showed irregular arrhythmic responses in cca1/lhy, was largely abolished in elf3, and showed a shortened period and somewhat weaker depression in the subjective light period in elf4.

Operation of the clock may vary in different parts of the plant (James et al. 2008). Our results for roots differ in some respects from those of James et al. (2008), who reported that the evening loop was absent in roots, and that the root clock is entrained by a photosynthesis related signal from the shoots, which is blocked by sucrose in the growth medium. Our data show that root growth is rhythmic in continuous light on sucrose, indicating that the root clock can remain entrained even when sucrose is present in the medium, and point to a role for ELF3 and ELF4 in the clock regulation of root growth. However, it should be noted that in contrast to the experiments of James et al. (2008) roots in our experiments were exposed to actinic illumination during the light period.

The clock-dependent root growth rhythms have similarities to other circadian rhythm phenotypes in Arabidopsis seedlings. Arabidopsis leaves open during the day and close at night, and this rhythm was maintained in continuous light (Hicks et al. 1996). Germinating wild-type Arabidopsis seedlings exhibit circadian rhythms in hypocotyl elongation, with a peak at or just after dawn in light/dark cycles (Nozue et al. 2007), and at subjective dusk in continuous light (Dowson-Day & Millar 1999; Nozue et al. 2007). cca1/lhy mutants have a short period in light/dark cycles, and are not able to sustain rhythmicity for expression of the clock-regulated gene CCR2 or cotyledon movements after transfer to continuous light (Alabadi et al. 2002). The irregular arrhythmic response of root growth in continuous light in cca1/lhy (Fig. 8) matches this pattern. ELF3 is required to sustain rhythmicity in long photoperiods and continuous light by inhibiting phototransduction at dusk (Doyle et al. 2002). elf3 mutants show an immediate loss of circadian rhythms in CAB expression (Dowson-Day & Millar 1999), leaf movement (Hicks et al. 1996) and hypocotyl elongation (Nozue et al. 2007) after transfer to continuous light. The immediate loss of rhythmicity of root growth (Fig. 8) fits this pattern. ELF4 is also important for the maintenance and accuracy of circadian rhythms under continuous light, but shows a slower loss of rhythmicity of CAB expression and leaf movement, often associated with changes in periodicity (Doyle et al. 2002). Similarly, root extension shows a shortened period and smaller oscillations in elf4 compared with wild-type plants.

Clock contributes to the matching of C supply and growth during the diel cycle

The results in Figs 5–8 are consistent with a scenario in which the clock acts via at least two different, but complementary, mechanisms to regulate diel root growth in Arabidopsis in C-limiting growth conditions (Fig. 12).

Figure 12.

Schematic model of how clock outputs affect root growth. For explanation, see text.

One loop involves CCA1/LHY, and acts to regulate the timing and duration of starch turnover. It has already been reported that cca1/lhy mutants exhaust their starch prematurely, resulting in a period of C starvation in the last hours of the night (Graf et al. 2010). The present experiments confirm this finding for seedlings, and extend it by showing that the premature exhaustion of starch is accompanied by a temporally-delimited inhibition of root growth in the last hours of the night that is reversed by supplying sucrose in the medium. Our results imply that plants are able to measure the amount of starch, and integrate this with information about the duration until the anticipated dawn to set the rate of starch breakdown and an appropriate rate of C consumption for growth, and that malfunctioning of this regulatory network results in an almost immediate inhibition of growth. CCA1 and LHY encode central clock components. Their loss alters the periodicity of the clock in light/dark cycles. However, the timing of starch degradation is still linked to the clock in cca1/lhy (Graf et al. 2010), showing that CCA1 and LHY are not directly involved in the mechanism that integrates clock signals into the starch degradation pathway. Further studies will be required to identify the clock output pathway that regulates starch degradation and growth (Fig. 12).

The second loop involves ELF3, and acts to depress root growth in the first part of the 24 h diel cycle and promote it in the second part of the 24 h diel cycle. This regulatory loop acts in a conservative manner to shift root growth towards the end of the 24 h cycle. Its action depends on the clock, and is independent of changes in C allocation. Thus, whereas wild-type Arabidopsis shows a depression of root growth in the subjective light period when it is grown in free-running continuous light, this restriction is abolished in elf3 (Fig. 8). The decline of growth in the night is not, or only very weakly, reversed by including sucrose in the medium, and elf3 mutants contain enhanced levels of starch and sugars in their shoots and roots at the end of the night.

As already discussed, ELF3 also regulates hypocotyl growth. Whereas hypocotyl growth is inhibited in the light and the inhibition is maintained for most of the night in wild-type seedlings, the inhibition is relieved early in the night in elf3 mutants (Nozue et al. 2007). This has led to the suggestion that the light-inhibition of hypocotyl growth is buffered by the clock in an ELF3-dependent manner. ELF3 expression and protein in hypocotyls is maximal during major parts of the night (Liu et al. 2001; James et al. 2008), when it suppresses light inputs into the clock (McWatters et al. 2000; Covington et al. 2001). Gating of hypocotyl extension towards the late night and early morning is regulated by coincidence of clock-regulated changes in PIF transcripts with increased stability of PIF4 and PIF5 protein in the dark (Nozue et al. 2007). Our results indicate that ELF3 serves an analogous function in roots, by gating growth to the root clock. However, instead of suppressing growth in the early and middle night, ELF3 acts to decrease root growth in the real or the subjective light period and promote it at night. The diurnal changes of ELF3 transcript are much weaker in the roots than the shoots (James et al. 2008) with the result that considerable amounts of ELF3 transcript are present in the light period in roots. James et al. (2008) suggested that the root clock is entrained by sucrose or some other photosynthesis-related metabolite that is transferred from the shoot to the root. This resembles mammals, where dietary restriction can reset the clock in peripheral tissues like the liver without affecting the central clock in the suprachiasmatic nucleus (Damiola et al. 2000; Hara et al. 2001; Stokkan et al. 2001). One possible function for ELF3 in roots might be to interact with metabolic signals deriving from the shoot. The observation that circadian rhythms in root growth are retained when sucrose is added in the growth medium of elf3 supports this hypothesis.

The role of ELF4 is less clear. ELF4 was originally implicated in light sensing, because elf4 seedlings are hyposensitive to red light and are modified for red-light repression of hypocotyl elongation. The elf4 mutant has strongly decreased CCA1 and LHY expression (Doyle et al. 2002; Khanna, Kikis & Quail 2003; Kikis, Khana & Quail 2005), and elevated TOC1 expression (McWatters et al. 2007) in light/dark cycles, leading to the proposal that ELF4 is required for the feedback loop that controls rhythmicity of CCA1, LHY and TOC1 expression, and acts at night to promote CCA1/LHY expression and thus indirectly repress TOC1 (McWatters et al. 2007). A reduction in CCA1 transcript levels would provide one explanation for the observed reduction in root growth rate at the end of the dark period in the elf4 mutant (Fig. 6). However, whereas exogenous sucrose strongly reverted the growth inhibition at the end of the night in cca1/lhy, it did not revert the inhibition in elf4, indicating that the ELF4 might act in additional ways. ELF4 encodes a coiled-coil-like dimer that could act as an interaction platform (Kolmos et al. 2009). It was recently proposed that ELF4 has multiple inputs, at PRR9/PRR7 and GI/LUX expression, respectively, and functions to repress light-induced expression of these entry points, integrating the morning and evening loops of the clock (Kolmos & Davis 2007; McWatters et al. 2007; Kolmos et al. 2009).

Much of the previous research on ELF3 and ELF4 has been driven by interest in interactions between the clock and light signaling. In view of the differing impact of the elf3 mutation on the light responsiveness of growth in the hypocotyl and root, it may be useful to extend studies of these genes to other environmental or physiological inputs. There is evidence that hormones including gibberellic acid may contribute to the gating of hypocotyl growth by promoting degradation of PIF4 protein (de Lucas et al. 2008; Feng et al. 2008; de Montaigu et al. 2010). In this context, it is also noteworthy that ELF3 is one of the most sugar-responsive of the clock associated genes (Usadel et al. 2008). Indeed there is a growing realization that there is a close interaction between the clock and metabolism, both on the input and the output side (Yin et al. 2007; Hatanaka et al. 2010; Yang 2010).

Clock components are also involved in rapid coordinated adjustments of starch turnover and root growth to the sudden changes in the environment

While most experiments occur in controlled and reproducible conditions, plants in the field are exposed to erratic and unforeseeable changes in the conditions. Graf et al. (2010) showed that wild-type Arabidopsis growing in a 12 h photoperiod can adjust the rate of starch degradation to avoid a period of C starvation at the end of the night when suddenly exposed to a 4 h premature dusk, but not a 6 h premature dusk. Adjustment to a 4 h premature dusk requires a halving of the rate of starch breakdown. As discussed in the Introduction, a decrease in the rate of starch breakdown, on its own, will not secure the plant against C starvation; this will also require a corresponding decrease in the rate at which C is consumed. As the use of C for maintenance is probably fairly inflexible, it will be necessary to decrease the rate of growth. A sudden premature transfer of wild-type seedlings into the dark indeed led to a decrease in the rate of root growth in the first night (Fig. 11). This raises a series of questions relating to the nature of the molecular mechanism that mediates this coordinated change in the rate of starch degradation in leaves and the rate of growth at the primary root tip. Intriguingly, this response was lost in cca1/lhy, indicating that the clock is not only required to set an appropriate rate of starch breakdown, but also to adjust the rate of growth at night.

In conclusion, root growth in Arabidopsis seedlings is regulated to gauge the overall rate of growth to the C supply and avoid transient periods of C starvation. The clock drives strong diel rhythms in the rate of growth via two complementary regulatory loops, one involving ELF3 that leads to a decline in growth in the first part and a recovery in the second part of the diel cycle, and another involving CCA1 and LHY that allows the rate of starch breakdown to be adjusted to the anticipated length of the night (Fig. 12). The approach used so far to measure root growth activity monitors the physical size of the roots, and will largely reflect cell expansion, which is at least partly due to vacuolar expansion. It will be important to develop methods to monitor cellular growth processes like protein synthesis and cell wall synthesis, as well as cell division, in order to achieve a deeper understanding of how metabolism and growth are regulated and integrated.

ACKNOWLEDGMENTS

This work was supported by the Max Planck Society and the European Commission (FP7 Collaborative Project TiMet). Technical help from Beatrice Encke and Nicole Krohn for metabolite determinations is greatly acknowledged.

Ancillary