Arabidopsis was grown in a 12, 8, 4 or 3 h photoperiod to investigate how metabolism and growth adjust to a decreased carbon supply. There was a progressive increase in the rate of starch synthesis, decrease in the rate of starch degradation, decrease of malate and fumarate, decrease of the protein content and decrease of the relative growth rate. Carbohydrate and amino acids levels at the end of the night did not change. Activities of enzymes involved in photosynthesis, starch and sucrose synthesis and inorganic nitrogen assimilation remained high, whereas five of eight enzymes from glycolysis and organic acid metabolism showed a significant decrease of activity on a protein basis. Glutamate dehydrogenase activity increased. In a 2 h photoperiod, the total protein content and most enzyme activities decreased strongly, starch synthesis was inhibited, and sugars and amino acids levels rose at the end of the night and growth was completely inhibited. The rate of starch degradation correlated with the protein content and the relative growth rate across all the photoperiod treatments. It is discussed how a close coordination of starch turnover, the protein content and growth allows Arabidopsis to avoid carbon starvation, even in very short photoperiods.
The adjustment of starch turnover to a decreased C supply will have to be accompanied by a decrease in the rate of C utilization for respiration and growth (Smith & Stitt 2007; Stitt et al. 2007). Changes of the C supply lead to marked changes in the expression of thousands of genes (Price et al. 2004; Li et al. 2006; Osuna et al. 2007). Many C-regulated genes show marked changes of expression during the night or the first hours of an extended night (Thimm et al. 2004; Bläsing et al. 2005; Usadel et al. 2008). Modelling of global transcriptional responses confirmed that these diurnal changes are caused by C signalling (Usadel et al. 2008). They affect genes that encode enzymes of primary metabolism and many genes that are required for protein synthesis and cellular growth (Gibon et al. 2004b, 2006; Usadel et al. 2008). However, most enzyme activities and the overall protein level do not show diurnal changes (Gibon et al. 2004b). Further, although the large diurnal changes of sugars in the starchless pgm mutant result in accentuated diurnal changes of transcripts, the encoded enzyme activities do not show larger diurnal changes; instead, they shift towards the values found in wild-type plants after several days of darkness (Gibon et al. 2004b, 2006). This led us to propose that diurnal changes of transcripts are integrated over time as changes of enzyme activities. The time delay between the changes of transcripts and the encoded enzymes would allow information to be extracted from the ‘noise’ of diurnal changes, and facilitate gradual adaptation to sustained changes in the environment. This proposal implies that measurements of metabolic parameters like enzyme activities could provide a useful tool to study how metabolism adjusts to changes in the diurnal supply of C.
The following experiments were carried out to test the capacity of Arabidopsis to grow in extremely short-day conditions, and to investigate the accompanying changes in starch turnover, protein content and the activities of 25 enzymes from central metabolism. The results show that Arabidopsis continues to grow in extremely short photoperiods, albeit at a strongly reduced rate. This is accompanied by large and coordinated changes in C allocation, the protein content and central metabolism.
MATERIAL AND METHODS
Arabidopsis thaliana var Col-0 was grown in soil [GS 90 soil mixed with vermiculite in a ratio 2:1 (v/v)]. After germination the seedlings were grown for 1 week in a 16/8 h light (250 µE m−2 s−1, 20 °C)/dark (6 °C) regime, for 1 week in an 8 h light (160 µE m−2 s−1, 20 °C)/16 h dark (16 °C) regime, and then replanted with one seedling per pot and transferred for 1 week to growth cabinets with an 8/16 h light–dark cycle (160 µE m−2 s−1, 20 °C throughout the day/night cycle). The plants were then distributed into small growth cabinets with a 12, 8, 4, 3 or 2 h photoperiod (all with a light intensity of 160 µE m−2 s−1 and 20 °C throughout the day/night cycle). The plants were harvested 17 d later. Typically, at least five separate samples, each containing three rosettes, were harvested.
Chemicals and enzymes
Inorganic compounds were purchased from Merck (Darmstadt, Germany), organic compounds from Sigma (Taufkirchen, Germany), except ethanol (Merck) and NADH (Roche), and enzymes from Roche (Mannheim, Germany), except phosphoglycerate kinase (Sigma). Glycerokinase is required for several of the enzyme assays. As commercial sources of glycerokinase are not longer available, yeast glycerokinase was overexpressed in Escherichia coli as a His6-tagged fusion protein by cloning the coding region of the Pichia farinosa GUT1 gene (kindly provided by Prof. Cândida Lucas, Universidade do Minho, Portugal) between the NcoI and EcoRI sites of expression plasmid pETM11 (Günter Stier, EMBL, Heidelberg, Germany). The protein was expressed in E. coli strain Rosetta™ (Novagen, Darmstadt, Germany) carrying the pLysS/RARE plasmid, and purified by immobilized metal affinity chromatography on Talon™ (Co2+) metal affinity resin (Clontech Laboratories Inc., Mountain View, CA, USA) as described in Lunn et al. (2006) for the E. coli trehalose-6-phosphate synthase. The purified glycerokinase had a specific activity of 50 µmol min−1 mg−1protein, and was stable for at least 6 months when stored at 4 °C as a suspension in 80% (v/v) saturated (NH4)2SO4.
Extraction and assay of enzymes
In all cases, the entire sample was powdered under liquid nitrogen and stored at −80 °C until its use. Aliquots of ∼20 mg fresh weight (FW) were weighed out at −180 °C, and were extracted by vigorous shaking with 1 mL of extraction buffer, leading to an initial ∼50-fold (w/v) dilution. The composition of the extraction buffer was 20% (v/v) glycerol, 0.25% (w/v) bovine serum albumin, 1% (v/v) Triton-X100, 50 mm HEPES/KOH pH 7.5, 10 mm MgCl2, 1 mm EDTA, 1 mm EGTA, 1 mm benzamidine, 1 mmε-aminocaproic acid, 1 mm phenylmethylsulfonylfluoride, 10 µm leupeptin and 0.5 mm dithiothreitol. Phenylmethylsulfonylfluoride was added just prior to extraction. Enyzme activities were assayed using a robotized platform as described in Gibon et al. (2004b, 2006) and Sulpice et al. (2007). The assay for cytosolic and plastidial phosphoglucose isomerases (cPGI and pPGI, respectively) was adapted from Kruckeberg et al. (1989) by scaling down to the microplate format, and validated using A. thaliana RNAi lines with no pPGI activity (unpublished results).
Extraction and assay of metabolites and protein content
Sucrose, glucose, fructose were determined in ethanolic extracts as in Jelitto et al. (1992); starch was determined as in Hendriks et al. (2003); protein content as in Bradford (1976); and malate and fumarate as in Nunes-Nesi et al. (2007). Assays were prepared in 96 well microplates using a Janus pipetting robot (Perkin-Elmer, Zaventem, Belgium). The absorbances were read at 340, 570 or 595 nm in a Synergy, an ELX-800 or an ELX-808 microplate reader (Bio-Tek, Bad Friedrichshall, Germany).
Standard procedures were carried out using functions of the Excel program. Linear regression analysis was performed in R software (R Development Core Team 2006). Hierarchical clustering using Euclidian distance and average linkage (Eisen et al. 1998) was performed on data expressed as log2 ratios to the respective controls (data from 12 h day/12 h night samples) using the MultiExperiment Viewer software (http://www.tm4.org/mev.html, Dana-Farber Cancer Institute, Boston, MA, USA).
After germination and seedling establishment in standard growth conditions, 21-day-old plants were transferred to a series of different photoperiod regimes, that is, 12 h light/12 h dark, 8 h light/16 h dark, 4 h light/20 h dark, 3 h light/21 h dark or 2 h light/22 h dark. Plant FW was approximately 0.12 g at the time of transfer. The plants were harvested 17 d later. Plant FW at harvest in the five regimes was 0.83, 0.46, 0.26, 0.19 and 0.14 g, respectively. The average relative growth rate (RGR; Fig. 1a) was calculated assuming that growth rates were constant between the time of transfer and harvest. This was checked for the 12 h photoperiod treatment in separate experiments (Tschoep et al. 2009). Growth decreased progressively as the light period was shortened, but was still possible even in the 3 h light/21 h dark photoperiod. There was effectively no growth in the 2 h light/22 h dark regime (Fig. 1); by 40 d after transfer, >50% of the plants were dead (data not shown).
Starch, sucrose, reducing sugars, total amino acids, organic acids and total protein (Fig. 2) were measured at the end of the day and at the end of the night. The results will be presented first for photoperiods between 12 and 3 h, and then for the extreme 2 h photoperiod. They will be compared with the published data from experiments in which Col-0 was transferred at the end of the night to continuous darkness for 2, 3 and 7 d, and in which the starchless pgm mutant was grown in a 12 h light/12 h dark regime (Gibon et al. 2004b, 2006).
Levels of carbohydrates, amino acids and organic acids
Short photoperiods led to lower levels of carbohydrates at the end of the day (see Fig. 2). This was mainly caused by a decrease of starch. The amount of starch synthesized during the light period is equivalent to the amount degraded during the night, and will be referred to as the diurnal starch turnover. Diurnal starch turnover decreased from 31 µmol hexose equivalents gFW−1 in a 12 h photoperiod to 27, 20 and 15 µmol hexose equivalents gFW−1 in 8, 4 and 3 h photoperiods, respectively (see Fig. 2). Diurnal starch turnover was normalized on the length of the light and dark period to calculate the rates of starch synthesis (µmol hexose equivalents incorporated into starch gFW−1 per hour light period) and degradation (µmol hexose equivalents mobilized from starch gFW−1 per hour dark period). To do this, we assumed that synthesis and degradation occur linearly during the day and night, respectively (see Gibon et al. 2004b). When the photoperiod was decreased from 12 to 8, 4 and 3 h, the average rate of starch synthesis increased from 2.6 to 3.3, 4.9 and 4.9 µmol hexose equivalents per gFW per hour, and the average rate of starch degradation decreased from 2.6 to 1.7, 1.0 and 0.7 µmol hexose equivalents per gFW per hour, respectively (Fig. 1b).
Despite the lower level of starch at the end of the day and the decreased rate of starch breakdown during the night, there was no evidence for a major shortfall of carbohydrates at the end of the night in very short day regimes (Fig. 2, see also Table 1). The levels of starch and reducing sugars at the end of the night were similar to those in a 12 h light/12 h dark regime. Sucrose showed a significant (Table 1) approximately twofold decrease at the end of the night (Fig. 2). The levels of starch, sucrose and reducing sugars at the end of the night in short photoperiods were much higher than after several days of darkness, or at the end of the night in pgm (Fig. 1).
Table 1. Correlation analysis of the response of carbohydrates, total amino acids and organic acids to a shortening of the photoperiod from 12 to 8, 4 and 3 h
Fresh weight (FW) basis
Metabolite levels were expressed on an FW or a protein basis. The analysis was carried out with the values at the end of the night, except for organic acids, where the values at the end of the day are also analysed. Each analysis is based on five samples for each of the four photoperiod treatments (n = 20 in total). Parameters that show significant changes (P < 0.005) are indicated by bold face.
Starch (end of night)
Sucrose (end of night)
Glucose (end of night)
Fructose (end of night)
Amino acids (end of night)
Malate (end of night)
Fumarate (end of night)
Malate (end of day)
Fumarate (end of day)
Amino acid levels increased slightly during the light period in a 12 h photoperiod (see also Gibon et al. 2006). Amino acid levels were not affected by shortening the light period to 8, 4 or 3 h (Fig. 2, Table 1). This contrasts with pgm where there is a small increase of amino acids, and extended darkness where there is a threefold increase of amino acids (Fig. 2, see also Gibon et al. 2006).
Malate and fumarate can accumulate to high levels in leaves (Chia et al. 2000; Fahnenstich et al. 2007). When Arabidopsis was grown in a 12 h photoperiod, their diurnal turnover was equivalent to approximately 20–30% of that of starch (Fig. 2b,c). Short photoperiods led to a two- to threefold decrease of organic acids at the end of the day, decreased diurnal turnover (from 11.9 to 7.8, 5.7 and 4.2 µmol gFW−1), an unaltered or slightly increased rate of accumulation in the light (0.99, 0.98, 1.42 and 1.4 µmol gFW−1 h−1), a much slower rate of mobilization during the night (from 0.99 to 0.48, 0.28 and 0.20 µmol gFW−1 h−1), and a threefold decrease of the level at the end of the night (Fig. 2, Supporting Information Table S1).
Metabolite levels showed a qualitatively different response in the extreme 2 h light/22 h dark photoperiod regime, in which (see earlier discussion) growth was not possible. Starch synthesis was severely inhibited, leading to strong reduction of the starch content at the end of the light period. Organic acid levels at the end of the day decreased another twofold, compared with the 3 h photoperiod. Unexpectedly, the levels of starch, sugars, amino acids and organic acids at the end of the night were similar to or higher than those in a 3 h photoperiod.
Total protein decreased by 10–15% as the photoperiod was decreased from 12 to 3 h (Fig. 2). This decrease was highly significant (P = 0.00012, Table 2). It was not caused by N deficiency; amino acids remained unaltered in short photoperiods (see earlier discussion), and nitrate remained unaltered on a FW basis (R = −0.13, P = 0.47) and increased significantly on a protein basis (R = 0.62, P = 0.0016). There was also a significant decrease of chlorophyll (Table 2, see Supporting Information Table S1 for the original data). In the 2 h photoperiod, protein levels decreased by a further 25% compared with the 3 h photoperiod, and by 35–40% compared with the 12 h photoperiod.
Table 2. Correlation analysis of the response of protein, chlorophyll and enzyme activities to a shortening of the photoperiod from 12 to 8, 4 and 3 h
Fresh weight (FW) basis
Protein was expressed on an FW basis, and chlorophyll and enzyme activities on an FW or a protein basis before the analysis. For all parameters except ADP-glucose pyrophosphorylase (AGPase) and NR, the values at the end of the night and the end of the day were pooled (n = 10 for each of the four photoperiod treatments, 40 in total). For AGPase and NR, the data for the end of the night and the end of the day were analysed separately (n = 20 in total). Parameters that show significant changes (P < 0.005) are indicated by bold face.
Protein decreases by about 30% when Col-0 is darkened for several days (Fig. 2; see also Thimm et al. 2004; Usadel et al. 2008). In pgm, protein levels are slightly lower than in wild-type Col-0 (Fig. 2). A more extensive analysis of many separate experiments confirmed that pgm contains lower levels of protein than wild-type plants (Hannemann et al. 2009).
Relation between the RGR, starch turnover, protein content and organic acid levels
Diurnal starch turnover showed a curvilinear relation to the RGR (Fig. 3a). The rate of starch synthesis was unrelated to the RGR (data not shown). The rate of starch degradation was tightly correlated with RGR (Pearson's R2 = 0.99; Fig. 3b) and, though less strongly, the protein content (R2 = 0.86; Fig. 3c). Protein content also correlated with RGR (R2 = 0.80; Fig. 3d). The average rate of organic acid mobilization correlated strongly with the rate of starch degradation (R2 = 0.96) and the RGR (R2 = 0.98. not shown), and more weakly with the protein content (R2 = 0.73, not shown).
Changes of enzyme activities
Enzyme activities were assayed at the end of the light period and the end of the night in a 12, 8, 4, 3 and 2 h photoperiod (see Supporting Information Fig. S1 and Supporting Information Table S1 for the original data). Regression analysis was performed on the combined 12, 8, 4 and 3 h photoperiod data set to identify enzymes that show a significant change of activity (Table 2). The extreme 2 h photoperiod treatment was omitted from the regression analysis. Of the 25 enzymes investigated, 18 showed a significant (P < 0.005) decrease in activity on a FW basis as the photoperiod was decreased to 3 h. Most others showed a small but non-significant decrease. Glutamate dehydrogenase (GDH) showed a highly significant increase in activity.
Most enzymes had similar activities at the end of the night and the end of the day. Marked diurnal changes in activity were found for ADP-glucose pyrophosphorylase (AGPase) and nitrate reductase (NR). Regression analyses were therefore performed separately on the data sets for the end of the night and the end of the day. Short photoperiods led to a significant decrease of AGPase activity at the end of the day but not at the end of the night. Short photoperiods led to a significant decrease of NR activity at the end of the night but not at the end of the day.
Comparison of the changes of enzyme activities and the overall protein content
In many cases, the decrease of the enzyme activity resembled the decrease of the total protein content (Table 2; compare also Fig. 2a and Supporting Information Fig. S1). To identify enzymes that show changes in activity that are not just because of the changes in the total protein content, we recalculated the activities on a protein basis (Fig. 4), and performed regression analysis on the combined 12, 8, 4 and 3 h photoperiod data set (Table 2).
There was no significant change in the activity of ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco; Fig. 4a) and four other Calvin cycle enzymes (Fig. 4b–e) when the photoperiod was shortened from 12 to 8, 4 and 3 h. There was a decrease of Rubisco, aldolase and transketolase activity on a protein basis in the extreme 2 h photoperiod. In the pathway of starch synthesis, pPGI activity did not show any significant changes (Fig. 4f). AGPase remained high at the end of the night, but showed a strong and significant decrease at the end of the light period (Fig. 4g). The diurnal changes of AGPase activity became increasingly marked as the photoperiod was shortened. Cytosolic PGI (cPGI), phosphoglucomutase (PGM) and sucrose phosphate synthase are required for sucrose synthesis. Their activities remained high in very short day treatments (Fig. 4h–k). Figure 4l–o shows enzymes from glycolysis including ATP-phosphofructokinase (PFK), pyrophosphate (PPi)-dependent phosphofructokinase (PFP), pyruvate kinase (PK) and phosphoenolpyruvate carboxylase (PEPCase), and Fig. 4p–s shows citrate synthase (CS), NAD-isocitrate dehydrogenase (ICDH), fumarase and NAD-malate dehydrogenase (MDH), which are involved in organic acid metabolism. There was a marked and significant decrease of PEPCase activity (Fig. 4o, P = 0.0007, see Table 2), and a weaker but significant (see Table 2) decrease of PFK (Fig. 4m), PK (Fig. 4n), ICDH (Fig. 4q) and fumarase (Fig. 4r) activity in short photoperiods, whereas PFP (Fig. 4l), CS (Fig. 4p) and NAD-MDH (Fig. 4s) activities did not show any significant change. In nitrogen metabolism, NR activity did not show any significant changes at the end of the night but increased significantly at the end of the day in short photoperiods (Fig. 4t). As for AGPase, the diurnal changes become more marked in short photoperiods. In the extreme 2 h photoperiod, NR activity remained low at the end of the day. Glutamine synthase (GS) and ferredoxin-dependent glutamate synthase (Fd-GOGAT) activity remained unaltered or increased in short days, except for the 2 h treatment (Fig. 4u,v). GDH activity increased progressively and significantly in short photoperiods (Fig. 4w).
Overall, these measurements show that when activity is expressed on a protein basis, the activities of enzymes that are required for photosynthesis, sucrose and starch synthesis and inorganic nitrogen assimilation remain high whereas several enzymes from glycolysis and organic acid metabolism still show a significant decrease of their activity in short photoperiods.
Comparison of enzyme activities in short photoperiods with those found in an extended night or the starchless pgm mutant
The data set for enzyme activities in different photoperiods was combined with the published data sets for the responses in prolonged darkness and in the starchless pgm mutant (Gibon et al. 2004b), and subjected to correlation and clustering analysis. Whereas prolonged darkness leads to a sustained C depletion, pgm alternates every day between high and very low sugar (see Introduction). The combined data set included a total of 17 treatments; the end of the light period and the end of the night for the five photoperiod treatments with Col-0, the end of the light period and the end of the night for pgm, and 48, 72 and 148 h of prolonged darkness with Col-0. It contained 16 enzymes, which had been measured in all of these treatments. The enzyme activities were expressed on a FW basis, because normalization on protein leads to loss of the activity changes of many of the enzymes. Activities were normalized on the average of the activity at the end of the light period and the end of the night in wild-type Col-0 in a 12 h photoperiod treatment from the same experiment.
Table 3 summarizes Pearson's regression coefficients (R) between the different treatments. As expected, the profiles in wild-type Col-0 in a 12 h photoperiod differ from the profiles in prolonged darkness and in pgm. When the photoperiod is shortened to 8, 4 or 3 h, the wild-type Col-0 profile at the end of the light period becomes increasingly similar to that in pgm at the end of the light period (R = 0.41–0.78, P < 0.01). When the profiles are compared at the end of the night the similarity is weaker and non-significant (R = 0.14–0 56). The similarity to the profile in prolonged darkness is also weaker and non-significant (R < 0.16 and <0.36 for profiles from the photoperiod treatments at the end of the light period and the end of the dark period, respectively). When the photoperiod is shortened to 2 h, the similarity to pgm increases further (R = 0.41–0.78, P < 0.01 at the end of the light period and P < 0.05 at the end of the night). A significant correlation also appears between the profiles at the end of the night and after 148 h prolonged darkness (R = 0.56, P < 0.01).
Table 3. Regression coefficients between the enzyme activities in very short days, and enzyme activities after 6 d extended darkness, or in starchless pgm mutants
148 h XN
48 h XN
Pearson regression coefficients (R) and P values (**, less than 0.01; *, less than 0.05; both indicated in bold face) were calculated with the responses of 16 enzymes to the various treatments. All enzyme activities were log2 normalized on the level in wild-type Col-0 at the end of the night in a 12 h light/12 h dark cycle.
D, dark; EL, end of the light period; EN, end of the night period; L, light; XN, extended night.
Figure 5 shows an unbiased two-way clustering of these 17 experimental treatments against the profile for the 16 enzyme activities and the total protein content. The protein content and enzyme activities are shown as a heat map. The photoperiod treatments separate along the axis from right to left, according to the length of the photoperiod. This presumably reflects the magnitude of the response of the enzyme activities, which becomes progressively larger as the photoperiod is shortened. The end of the night and the end of the day treatments group together for a given photoperiod. This shows that the long-term adjustments of enzyme activities to the photoperiod are larger than the diurnal changes in a given photoperiod. The only exceptions are the 3 and 4 h photoperiods, where the samples group according to the time of harvest, probably because of the more accentuated changes of AGPase and NR activities across the day–night cycle. The profiles of wild-type plants grown under photoperiods ranging from 8 to 3 h are closer to the pgm profiles than to the profiles of plants after an extension of the night. The largest linkage distance was found for the 2 h photoperiod treatment.
Figure 5 also provides an overview of the responses of individual parameters. Total protein decreases in all three treatments. GDH activity rises in all of the treatments; this is summarized in Fig. 6, which emphasizes that the increase in activity in the short photoperiod treatments is much smaller than in pgm or prolonged darkness. Most other enzyme activities decrease in all three treatments (see also Supporting Information Fig. S3). Rubisco, SPS, PEPC, PK and fumarase activity show a larger increase in very short photoperiods and in prolonged darkness than in pgm. In specific cases, there is a qualitative difference between the treatments. Invertase activity does not increase in short photoperiods, even though it rises in pgm and prolonged darkness. AGPase activity decreases in the short photoperiod and pgm treatments, but rises slightly in prolonged darkness. As already noted, AGPase activity decreases in the light and peaks at the end of the night (see also Gibon et al. 2004a,b). NR shows only small changes in short photoperiods and pgm, but a large decrease in prolonged darkness. NR activity typically increases in the first 2 h of the light period in a 12 h photoperiod (Gibon et al. 2004b). The high AGPase and low NR activity in prolonged darkness may be caused by the absence of a light phase in this treatment.
The experiments presented in this article investigate: (1) whether Arabidopsis is able to adjust to a large decrease in the C supply and grow in extremely short photoperiods; (2) if changes in starch turnover, the protein content and central metabolism contribute to the adjustment to low C; and (3) whether the response to short photoperiods resembles the responses when C deficiency is imposed by two other treatments; a severe sustained C starvation after darkening wild-type Col-0 for several days, and a recurring transient C limitation in the starchless pgm mutant.
Adjustment of growth, starch turnover and the protein concentration to very short photoperiods
Wild-type Arabidopsis grows in extremely short photoperiods, down to a 3 h light/21 h dark cycle, showing that it adjusts flexibly to very large changes in the C supply. One contributory factor is a major modification of starch turnover. In short photoperiods, starch is synthesized more rapidly in the light and is degraded more slowly during the night. This resembles the response previously documented in less extreme photoperiods (see Introduction). As a result, the amount of starch and the levels of sugars and amino acids at the end of the dark period are largely independent of the length of the light period.
Short photoperiods also lead to a significant 10–15% decrease of the rosette protein concentration. This decrease is unlikely to be caused by a direct lack of amino acids. The activities of enzymes involved in the assimilation of inorganic nitrogen remained high or increased (see below for more discussion), and nitrate and amino acid levels remained high in short photoperiods. A similar decrease of the protein concentration in short photoperiods has been seen in many other independent experiments, where the photoperiod was varied between 2 and 20 h (Hannemann et al. 2009).
Expression profiling in wild-type Col-0 has revealed that diurnal changes of C lead to coordinated changes in the levels of transcripts for hundreds of genes that are involved in protein synthesis and targeting, with transcripts starting to decrease early in the night (Bläsing et al. 2005; Usadel et al. 2008). Further, polysome loading and, by implication, the rate of protein synthesis decreases during the night (Kumar, Piques & Stitt, unpublished data). The simplest explanation for the decreased protein content in short day conditions is that C regulates the rate of protein synthesis. However, it is possible that light-signalling also plays a role. Very short days (2–4 h) lead to petiole elongation (data not shown), which is reminiscent of the responses to low light. Kozuka et al. (2005) showed that petiole extension and leaf blade expansion is regulated by an interaction between sugars and phytochrome and cryptochrome-dependent light signalling, whereas Nozue & Maloof (2006) have proposed that hypocotyl growth is regulated by an interaction between C signalling, light signalling and diurnal responses.
Our analyses reveal a striking positive correlation between the rate of starch breakdown and the RGR (R2 = 0.97, Fig. 3e). This relation has been confirmed in independent experiments (Hannemann et al. 2009). This result implies that starch turnover and growth are tightly coordinated. More studies are needed to explore the causal reasons for this tight correlation. One contributing factor may be the decrease of the protein content in short photoperiods. There was a correlation between the protein content and the rate of starch breakdown (R2 = 0.86, Fig. 3c). This correlation has been confirmed in a meta-analysis of data from 27 experiments (R = 0.55, P = 4.4 × 10−15, Hannemann et al. 2009). A lower protein content will decrease the requirement for C in two ways. Firstly, a decrease in the protein content, or more generally the nitrogen content, is typically accompanied by a near-linear decrease in the rate of respiration (James 1953; Ryan 1991; van der Werf et al. 1992). This will decrease the amount of starch that is required to support respiration during the night. Secondly, a decrease in the protein content may decrease construction costs, that is, the amount of C that is needed to produce a unit amount of biomass. The assimilation of inorganic nitrogen into amino acids and the subsequent conversion of amino acids to protein are energetically expensive processes, with both requiring about five ATPs per amino acid incorporated into protein. Modelling and experiments that use cycloheximide to inhibit protein synthesis indicate that protein synthesis is responsible for 20–70% of the total respiratory costs in plant tissues (Penning de Vries 1975; van der Werf et al. 1992; Bouma et al. 1994; Zagdanska 1995; Noguchi, Nakajima & Terashima 2001; Hachiya, Treashima & Nochuchi 2007). The observation that plant growth often occurs during the night (Schurr, Walter & Rascher 2006; Wiese et al. 2007) underlines the potential importance of a link between the C supply from starch degradation and construction costs. This is likely to become increasingly important in very short photoperiods.
Rasse & Tocquin (2006) recently modelled the relation between photosynthesis, starch turnover, sugar levels, respiration and growth. Their model assumes that starch is produced at a baseline rate when photosynthate is low, and does not address the question of how plants adjust to changes in the C supply. Our and (see Introduction) earlier studies show that starch synthesis is increased when less C is available. Our study also shows that the changes in starch turnover are accompanied by coordinated changes in the protein content and the rate of growth. It will be important in the future to model these additional interactions, and quantitatively explore their impact on C balance and growth.
Adjustment of central metabolism to very short photoperiods
Enzyme activities and metabolite levels were measured to investigate whether changes in central metabolism contribute to the adjustment to short photoperiods. Shortening the photoperiod from 12 to 3 h led to a relatively small decrease in the activities of enzymes that are required for C fixation and starch and sucrose synthesis. The decrease was largely caused by a decrease of the overall protein content. This response will minimize the inhibition of photosynthesis and carbohydrate synthesis, and maintain C gain in short photoperiods. The activities of enzymes involved in nitrogen assimilation including NR, GS and Fd-GOGAT also remained high. Indeed, NR activity at the end of the light period actually increased in short photoperiods. Amino acid levels also remained high in short photoperiods. This highlights the ability of Arabidopsis to maintain a metabolic balance in short days. There is a marked contrast to tobacco, where growth in an 8 h photoperiod led to a strong inhibition of NR activity, a marked decrease of amino acids and a large decrease of leaf protein (Matt et al. 1998).
AGPase showed marked diurnal responses, with activity increasing during the night and decreasing in the light period. These diurnal changes were absent in a 12 h photoperiod, and became larger as the photoperiod became shorter. Gibon et al. (2004a,b) also reported that diurnal changes of AGPase activity are absent in Col-0 in a 20 h, small in a 12 h photoperiod and marked in a 7 h photoperiod, whereas in pgm diurnal changes are already seen in a 12 h photoperiod, indicating that they are a response to change of C rather to the light period per se. The diurnal changes of AGPase activity are caused by the changes of AGPase protein (Gibon et al. 2004b). High activity of AGPase at the end of the night might contribute to the high rate of starch synthesis in the following light period (see earlier discussion). The mechanisms that lead to these diurnal changes are unknown.
Eight of the enzymes that we investigated are involved in glycolysis and respiration. Of these, five (ATP-phosphofructokinase, pyruvate kinase, PEPCase, isocitrate dehydrogenase and fumarase) decreased more strongly than the total protein content in short photoperiods. The decrease was especially marked for PEPCase, which is required for net production of organic acids. In agreement, organic acids decreased much more strongly than carbohydrates and amino acids in short photoperiods.
Low levels of organic acids might serve to decrease C utilization for respiration at night. Respiration of malate requires the concerted action of NADH-MDH and malic enzyme. The latter generates pyruvate, which is converted to acetyl-CoA and combines with the oxaloacetate generated by NADH-MDH. Fahnenstich et al. (2007, 2008) recently showed that over-expression of NADP-malic enzyme in Arabidopsis leads to lower levels of malate and fumarate. This was associated with faster senescence in continuous darkness, and pale green leaves and increased accumulation of starch when the plants were grown in low irradiance and short day conditions, supporting the idea that organic acids provide an important respiratory substrate when carbohydrates are low. Fahnenstich et al. (2008) proposed that starch synthesis is stimulated when less C is stored in organic acids. In our experiments, short photoperiods led to a twofold increase in the rate of starch accumulation in the light but did not markedly affect the rate of malate and fumarate accumulation. On the other hand, the rates of starch degradation and organic acid utilization during the night correlated with each other and with RGR. This indicates that although allocation to these two C stores may be differentially regulated, their utilization is tightly coordinated.
Response to a very short 2 h photoperiod
Arabidopsis Col-0 was unable to grow in a 2 h photoperiod, but did survive for several weeks. There were several marked changes in metabolism, compared with a 3 h photoperiod. Firstly, there was a further and marked decrease in total protein. Secondly, there was a marked decrease of NR, GS and Fd-GOGAT activity. These three enzymes are required for inorganic nitrogen assimilation. The reduction of NR was in part because activity no longer rose during the light period, and is reminiscent of the inhibition of NR activity in tobacco in an 8 h photoperiod (see earlier discussion). Thirdly, starch synthesis was inhibited. The reasons for this unexpected inhibition are unclear, as similar AGPase activities were found at the end of the night in a 3 and a 2 h photoperiod. Finally, and most unexpectedly, the levels of starch, sugars and amino acids at the end of the night in a 2 h photoperiod were similar to those at the end of the night in longer photoperiods. These results indicate that the inability to grow in a 2 h photoperiod is not a simple consequence of an acute lack of sugars or amino acids. Rather, it appears to be because of an inhibition of the utilization of these central metabolites for growth. Further studies will be needed to reveal the precise reason for this inhibition.
Comparison of the response of enzyme activities to short photoperiods or prolonged darkness in wild-type Col-0 and to repeated transient periods of C starvation in pgm
There were several similarities between the enzyme activity profiles in short photoperiods and the profiles in prolonged darkness or the starchless pgm mutant. All the three treatments led to a decrease of the overall protein content, lower activities on a FW basis of many enzymes from photosynthetic C metabolism, nitrogen assimilation, glycolysis and respiration, and an increase of GDH activity. Regression analysis and two-way clustering indicated that the enzyme profile in 8, 4 or 3 h photoperiods resembles the profile in pgm. This is consistent with the proposal (see Introduction) that diurnal changes of transcripts are integrated over time as changes of the protein content and enzyme activities.
All the three C-depletion treatments led to an increase of GDH activity, indicating that high GDH activity is a good metabolic marker for C deprivation. GDH operates in the direction of ammonium release (Miflin & Habash 2002), and recycles glutamate to 2-oxoglutarate during amino acid catabolism (Melo-Oliveira, Oliveira & Coruzzi 1996; Lam et al. 2001). As already noted, carbohydrate levels at the end of the night in short photoperiods were higher than at the end of the night in pgm, or in prolonged darkness. This indicates that adjustments of starch turnover and metabolism allow Arabidopsis to avoid severe C depletion when it is grown in short photoperiods. In agreement, GDH activity did not increase as much in short photoperiods as in prolonged darkness or in pgm. The increase of GDH activity in low-C conditions may be partly caused by transcriptional regulation. GDH is encoded by a small family of three genes. Two of these (GDH1, GDH2) are strongly induced by C starvation in seedlings and rosettes (Price et al. 2004; Thimm et al. 2004; Osuna et al. 2007; Usadel et al. 2008). In rosettes, transfer to prolonged darkness leads to a rapid increase of GDH1 and GDH2 transcripts, whereas GDH activity increases gradually for several days (Gibon et al. 2004b). In pgm, GDH1 and GDH2 transcripts peak transiently in the night, whereas GDH activity is stable and about twice that in wild-type Col-0.
In conclusion, Arabidopsis Col-0 adjusts to increasingly short photoperiods by a progressive inhibition of growth, stimulation of starch synthesis, inhibition of starch degradation and decrease of the overall protein content. There is a striking correlation between the rate of starch breakdown and the rate of growth, and a significant correlation between both of these parameters and the overall protein concentration. The activities of enzymes that are involved in respiration decrease, relative to enzymes that are involved in photosynthesis, starch and sucrose synthesis and nitrogen assimilation. This allows C levels to be maintained at the end of the night that are similar to those in longer photoperiods, and are much higher than in acute or regularly recurring C starvation. These results highlight the ability of Arabidopsis to restrict C utilization and gauge growth to the C supply, and point to the importance of regulatory mechanisms that regulate starch turnover and the protein concentration as important components of this response.
This research was supported by the Max Planck Society and by the German Ministry for Research and Technology, in the framework of the German Plant Genomics programme GABI (0312277A, 0313110, 0313123) and GoFORSYS (http://www.goforsys.de/). We are grateful to Kristin Retzlaff for validating the PGI assay. The writing of the paper was greatly helped by perceptive comments from two anonymous referees.