Arabidopsis seedlings were subjected to 2 days of carbon starvation, and then resupplied with 15 mm sucrose. The transcriptional and metabolic response was analyzed using ATH1 arrays, real-time quantitative (q)RT-PCR analysis of >2000 transcription regulators, robotized assays of enzymes from central metabolism and metabolite profiling. Sucrose led within 30 min to greater than threefold changes of the transcript levels for >100 genes, including 20 transcription regulators, 15 ubiquitin-targeting proteins, four trehalose phosphate synthases, autophagy protein 8e, several glutaredoxins and many genes of unknown function. Most of these genes respond to changes of endogenous sugars in Arabidopsis rosettes, making them excellent candidates for upstream components of sugar signaling pathways. Some respond during diurnal cycles, consistent with them acting in signaling pathways that balance the supply and utilization of carbon in normal growth conditions. By 3 h, transcript levels change for >1700 genes. This includes a coordinated induction of genes involved in carbohydrate synthesis, glycolysis, respiration, amino acid and nucleotide synthesis, DNA, RNA and protein synthesis and protein folding, and repression of genes involved in amino acid and lipid catabolism, photosynthesis and chloroplast protein synthesis and folding. The changes of transcripts are followed by a delayed activation of central metabolic pathways and growth processes, which use intermediates from these pathways. Sucrose and reducing sugars accumulate during the first 3–8 h, and starch for 24 h, showing that there is a delay until carbon utilization for growth recommences. Gradual changes of enzyme activities and metabolites are found for many metabolic pathways, including glycolysis, nitrate assimilation, the shikimate pathway and myoinositol, proline and fatty acid metabolism. After 3–8 h, there is a decrease of amino acids, followed by a gradual increase of protein.
Evidence is emerging that changes in endogenous sugars exert a major impact on gene expression. In the light, CO2 fixation in leaves drives sucrose synthesis and export to support growth and storage. At night there is a negative C balance (Gibon et al., 2004a; Matt et al., 2001; Smith et al., 2004). To buffer these daily fluctuations, some C is accumulated as starch in the light, and remobilized at night. Changes in sugars make a large contribution to diurnal changes of transcript levels, which themselves affect 30–50% of the genes expressed in an Arabidopsis rosette (Bläsing et al., 2005). In a regular light/dark cycle, starch reserves last until the end of the night (Geiger and Servaites, 1994; Gibon et al., 2004a; Smith et al., 2004). A few hours more darkness leads to exhaustion of starch, a collapse of sugars and major changes of transcripts (Thimm et al., 2004). In the starchless pgm mutant, sugars accumulate to very high levels in the light and are depleted at night (Caspar et al., 1985; Gibon et al., 2004a). The low levels of sugars at night drive accentuated diurnal changes of transcripts for >4000 genes (Bläsing et al., 2005).
Meta-analysis of yeast transcriptomics and proteomics data (Greenbaum et al., 2003) has revealed that changes of transcripts do not always lead to changes of protein. Gibon et al. (2004b)) developed a robotized platform to measure 23 enzyme activities in Arabidopsis rosettes in optimized conditions, where activity reflects the protein level. Changes of activity in diurnal cycles were small, delayed and not predictable from transcripts. After transfer to extended darkness, changes of enzyme activities required several days, whereas transcripts changed in hours. To generalize this conclusion, Gibon et al. (2006) investigated metabolite profiles; these integrate the activities of hundreds of enzymes. Most metabolites undergo small diurnal changes, and change slowly in an extended dark treatment. Diurnal changes of transcripts may often be important for medium-term responses, rather than alterations of metabolism during a single diurnal cycle. The repeated transient depletion of sugars and changes of transcripts each night in pgm lead to a systematic shift of enzyme activities and metabolites towards those found in wild-type plants after several days of darkness (Gibon et al., 2004b, 2006).
It is obvious that C depletion inhibits growth. It is less clearly appreciated that there is a lag before growth starts again when C becomes available. When wild-type plants are re-illuminated after a 6-h prolongation of the night, sugars and starch accumulate for 4–6 h in the leaves and roots (Gibon et al., 2004a). The pgm mutant grows like wild-type plants in continuous light, but growth is inhibited increasingly strongly as the photoperiod is shortened (Caspar et al., 1985; Gibon et al., 2004a). It was originally proposed that this is due to wasteful respiration of sugars at the start of the night (Caspar et al., 1985). Recent studies show that low sugar inhibits growth in the last part of the night, and this inhibition is only gradually reversed in the light period (Gibon et al., 2004a).
Plants adjust their C allocation and growth to the C supply. When less C is available, due for example to short-day conditions or low light, growth decreases, a larger proportion of the photosynthate accumulates as starch, and starch is broken down more slowly (Chatterton and Silvius, 1979, 1980, 1981; Matt et al., 1998). As a result, starch lasts until the end of the night and an acute limitation of metabolism and growth by C is avoided.
Transcriptional responses to changes of endogenous sugars could provide clues about signaling pathways that balance C allocation and growth with the C supply. Their interpretation requires information from simpler systems, where sugars are directly manipulated. Bläsing et al. (2005) used two data sets; addition of 100 mm glucose for 3 h to C-starved seedlings, and comparison of rosettes that had been illuminated for 4 h at ambient or low [CO2]. Data sets are also available for the response after adding 167 mm glucose to C-starved seedlings for 3 h (Price et al., 2004) or 2, 4 and 6 h (Li et al., 2006), and after darkening plants for 6 h (Thimm et al., 2004) or days (Contento et al., 2004). These data sets have shortcomings. First, the addition experiments used high sugar concentrations and glucose rather than sucrose, which is the transported form in plants. Second, some lack specificity. Price et al. (2004) varied light and nitrogen as well as C, prolonged dark treatments are complicated by light signaling and the circadian clock, and changing [CO2] affects photorespiration and, possibly, water relations. Third, many of the changes after 2–3 h may be secondary. Fourth, information about metabolism is required to place changes of expression in a functional context.
This paper analyses short- and medium-term changes of transcripts, enzyme activities and metabolites after adding 15 mm sucrose to C-starved Arabidopsis seedlings, grown in continuous light and excess nutrient to minimize changes due to other inputs. The aims are to characterize the dynamics of transcript levels and metabolism during the recovery from C starvation, to identify genes whose expression responds rapidly to sucrose, and to analyze how these genes respond in more physiological systems and identify putative upstream components of signaling pathways that may regulate C utilization during the diurnal cycle.
Physiological and metabolic responses to C-deprivation and sucrose readdition
Arabidopsis seedlings were grown in liquid culture on sucrose-supplemented full nutrient medium (FN) in continuous low light (Scheible et al., 2004). After 7 days, some cultures were transferred to fresh FN and others to C-deficient conditions (CSt). By day nine, FN and CSt seedlings had cotyledons and first leaves (Figure 1a). C starved seedlings grew more slowly (Figure 1b) and contained less carbohydrate (Figure 1c–f). The FN seedlings and some batches of C-starved seedlings were harvested. Others received 15 mm sucrose (or mannitol as an osmotic control) and were harvested after 12 min, 30 min, 3 h, 8 h or 24 h. Internal sucrose (Figure 1c) increased five- and tenfold after 12 and 30 min, and peaked after 3 h. After a delay, glucose and fructose peaked at 8 h (Figure 1d,e) while starch accumulated until 24 h (Figure 1f). Carbohydrates levels were higher after adding sucrose than in FN.
Transcript profiling was carried out with FN seedlings, CSt seedlings and seedlings harvested 30 min and 3 h after adding sucrose or mannitol. Two experiments were performed at a 2-month interval. Pair-wise plots revealed good agreement between biological replicates (Figure S1). For C-starved seedlings resupplied with 15 mm sucrose for 30 min, the R2 regression coefficient was 0.972. Of the 22 750 probe sets, 99.7% yielded gene expression ratios between 0.5 and 2, and only 0.004% gave ratios >5 or <0.2. Similar replication was obtained in other treatments, except for 3 h after adding sucrose when there were large changes whose magnitude varied between replicates. There was nevertheless strong qualitative agreement for individual genes.
Signals in full nutrition (FN) and 30 min (30′S) and 3 h (3hS) after adding sucrose were expressed relative to the signal in CSt seedlings and the ratios averaged for the two experiments. Carbon starvation led to widespread changes of expression (Figure 1g) with 1782, 544, 128 and 35 genes showing >two-, >three-, >five- and >tenfold changes of their transcripts compared with FN. Sucrose readdition led to marked changes after 30 min, with 363, 110, 41 and 10 transcripts showing >two-, >three-, >five- and >tenfold changes. By 3 h, 4100, 1747, 656 and 199 transcripts showed >two-, >three-, >five- and >tenfold changes. Many overshoot two- to fourfold compared with FN (Figure S2). There were negligible changes in the osmotic controls (Figure 1g and Figure S1n,o), and none that were conserved at the gene level across replicates or times (Table S1).
The data and averaged ratios are deposited as Excel spreadsheets (Tables S1 and S2). These also summarize data from experiments in which 100 mm glucose was added to C-starved seedlings (Bläsing et al., 2005) or endogenous sugar levels were altered by illuminating rosettes for 4 h with 350 or 50 p.p.m. [CO2] (Bläsing et al., 2005), comparing the start and end of the light period in wild-type Arabidopsis or pgm (Bläsing et al., 2005) and extending the night for 6 h (Thimm et al., 2004).
Responses of genes assigned to different functional categories
The current MapMan ontology assigns genes to about 800 hierarchical categories (Thimm et al., 2004; Usadel et al., 2005, http://www.gabi.rzpd.de/projects/MapMan/; TAIR6_MappingFileaffy_22k_0406; Table S3). Table 2 shows the proportion of genes in each major category whose transcripts change ≥threefold. Carbon starvation affects a large proportion of genes in C and nutrient metabolism, and less in categories related to the regulation of transcription, signaling, development and cell organization. A different picture emerges 30 min after adding sucrose, when categories related to regulation (regulation of transcription, protein degradation), stress, xenobiotic degradation and redox are over-represented. The pattern after 3 h is similar to C-replete vs. C-starved seedlings, except that the changes are larger.
Table 2. Assignment of sucrose regulated genes to different functional categories
No. genes in the category
% whose expression changes by ≥threefold
FN v/CSt (n = 553)
30′S/CSt (n = 115)
3hS/CSt (n = 1661)
Genes shown were grouped according the functional categories defined in MapMan (Usadel et al., 2005; version TAIR6_MappingFileaffy_22k_0406, http://www.gabi.rzpd.de/projects/MapMan/). For this analysis, only genes were included that were called ‘present’ in both biological replicates for at least one condition, and showed ≥threefold changes in transcript levels in a given comparison (see Table heading for number for each treatment). The raw data from which this Table is computed is deposited as Table S2, where the calculated ratios are provided for all individual genes, with the genes organized according to the MapMan ontologies. Over-represented categories are indicated in bold face. CSt is C starved, 30′S and 3hS are 30 min and 3 h, respectively, after adding 15 mm sucrose, and FN is full nutrient medium containing sucrose.
Oxidative pentose phosphate
Amino acid metabolism
Major CHO metabolism
Minor CHO metabolism
Co-factor and vitamins
TCA/org. acid metabolism
Mito. electron transport
The Wilcoxon test was used to identify categories whose members show a coordinated response, which is statistically significant compared with all other genes. The P-values were calculated for the difference between C-starved seedlings (the common reference) and seedlings 30 min and 3 h after adding sucrose, and C-replete seedlings. The P-values were assigned a positive or negative value, depending on whether the average signal increased or decreased. The full analysis is available in Table S4. Some major features are visualized in Figure 2. A deeper blue or red shading depicts an increasingly significant induction or repression, respectively, of the genes in a given category. Addition of sucrose leads to a highly significant repression of genes assigned to the photosynthesis. Genes assigned to glycolysis, the tricarboxylic acid (TCA) cycle, mitochondrial electron transport and ATP synthesis and cell wall precursor synthesis are induced, while genes assigned to sucrose and starch breakdown, gluconeogenesis and lipid breakdown are repressed. Genes assigned to amino acid and nucleotide synthesis are induced, and amino acid catabolism repressed. There is a highly significant induction of genes assigned to DNA synthesis/chromatin structure, RNA processing, protein synthesis and targeting. While genes assigned to cytoplasmic ribosomal proteins are induced, genes encoding plastid ribosomal proteins are repressed. Genes assigned to protein degradation [ubiquitin (UBQ) targeting, autophagy, proteases] are repressed. A few categories show highly significant changes within 30 min, including putative trehalose-6-phosphate synthase/phosphatase (TPS/TPP) proteins, transcription regulators, UBQ-mediated protein degradation and autophagy. For most other categories, the P-values after 30 min are poor, or much less significant than at 3 h. Genes in categories related to cell wall polymer synthesis and cell wall proteins are repressed in C starvation and do not revert within 3 h.
Temporal responses of genes involved in signaling
Figure 3 displays the temporal responses of genes in categories related to signaling that show a >twofold change in one of the treatments. Many transcription regulators (TRs) and ubiquitin/26S proteosome pathway components respond in 30 min (Figure 3a,b). Most genes assigned to hormone metabolism and sensing, receptor kinases, protein kinases and phosphatases, G proteins, Ca-binding proteins and phosphatidyl inositol signaling show small changes at 30 min and large ones at 3 h (Figure 3c–f).
Many TRs are expressed at low levels and are not reliably detected on ATH1 arrays (Czechowski et al., 2004; Scheible et al., 2004). Real-time RT-PCR (Czechowski et al., 2004) was used to confirm the ATH1 arrays and identify further sugar-responsive TRs. Cycle threshold (Ct) numbers and amplification efficiencies are provided in Table S5, the results and a comparison with published data in Table S6 and Figure S3, and selected results in Table 1. Over 50 TRs undergo a ≥fivefold change of expression after C starvation and/or sucrose readdition (Table 1). Of these, 30 are identified by ATH1 arrays and RT-PCR, 26 only by RT-PCR and four only by ATH1 arrays. Of the genes identified by RT-PCR but not ATH1 arrays, four are absent from the ATH1 arrays, three showed ratios 3 ≤ x < 5 and 0.2 < x ≤ 0.33 (‘weak response’) and 19 show ratios <threefold in the ATH1 arrays. Most of the latter are expressed at low levels (MASC called 14 ‘absent’ in all, three ‘absent’ in some treatments and only two ‘present’ in all treatments). The ATH1 array identifies three sucrose-responsive (≥fivefold) genes that respond weakly (3 ≤ x < 5) with real-time RT-PCR. At5g61590 shows a contradictory result, being slightly induced in FN by RT-PCR (<threefold) and slightly repressed on ATH1 arrays. blasting indicated the ATH1 probe set is specific. Contento et al. (2004) also reported repression of At5g61590 in FN compared with C starvation. Over 20 TRs were defined as rapidly responding, using a >threefold change after 30 min as a filtering criterion. They are identified by giving the AGI in bold face in Table 1, and include six members of the AP2/EREBP family, three of the bZIP family, five of the MYB family, three NAC domain proteins, one member of the WRKY, Aux/IAA and C2C2(Zn) CO-like families and one heat shock factor.
Table 1. Sucrose-regulated TR genes identified by real-time RT-PCR or Affymetrix Gene Chips®
The transcript level in full nutrition (FN) and 30 min (30’S and 3 h (3hS) after adding 15 mm sucrose is given as a ratio to the transcript level in C-starved (CSt) seedlings. Use of real-time RT-PCR platform for sensitive expression analysis of around 1500 TF genes and Affymetrix Gene Chip® allowed the identification of 50 TF genes the expression of which changed strongly in response to altered sugar nutrition (≥fivefold changes). The absence (A) or presence (P) of transcripts, as determined by Affymetrix MAS5 software, in the conditions compared is indicated, as well as genes not represented (NR). Transcription regulators that change >threefold in 30 min are flagged by giving the AGI identifier in bold type in the left hand column. For a fuller analysis see Table S6.
Over 1300 genes contribute to protein degradation via the ubiquitin/26S proteosome (Vierstra, 2003). Genes in this category that show rapid changes of expression (Figure 3b) are listed in Table 3. A set of SCF-F box RING finger and BTB family members are induced in C starvation and strongly repressed (>threefold) after sucrose addition. Relatively few E2 or E3 proteins are induced by sucrose. A full list of genes that show responses is given in Table S7.
Table 3. Genes assigned to functional categories relating to protein degradation, which change within 30 min of adding sucrose to C-starved seedlings
*The probeset for At2g44130 may be non-specific.
For abbreviations see Table 1 and for a fuller analysis see Table S7.
Autophagy, APG 8e
Over 300 receptor kinases, soluble protein kinases and phosphatases, MAP kinase pathway components, calcium-binding proteins and G-proteins show alterations of their transcript levels in C-starved seedlings, which are reversed by sucrose. Few show large changes after 30 min (Figure 3c–e, Table 4). Over 100 genes assigned to hormone metabolism and signaling (Usadel et al., 2005) show >twofold changes (Figure 3f). Some occur in 30 min (Table 4), including two tandem ethylene response element-binding proteins. Genes in these categories that show slower responses are listed in Table S8.
Table 4. Selected genes assigned to functional categories relating to signaling, which show large changes between C-starved and C-replete conditions
For abbreviations see Table 1, and for a fuller analysis see Table S8.
Ethylene-responsive transcriptional co-activator
Auxin-responsive family protein
Allene oxide cyclase,
At5lectin protein kinase
Calmodulin-like calcium-binding protein
Leucine-rich repeat protein kinase
Protein kinase like protein
AKIN beta1 protein kinase
Putative jasmonic acid regulatory protein
Similar to auxin downregulated protein ARG10
Ethylene-responsive element binding factor 5.
Genes involved in central metabolism
The response of genes assigned to central metabolism (Figure 4) provides background information for interpreting changes of enzyme activities and metabolites. In this display, the signal for each individual gene is expressed relative to the signal in C-starved seedlings, converted to a log2 scale and displayed on a color scale using MapMan software (Thimm et al., 2004; Usadel et al., 2005). Transcripts that increase or decrease are blue or red, transcripts that change <twofold are white, and transcripts called ‘not present’ are gray. The data can be explored using files provided in the Supplementary Material (Table S3) and downloadable software (http://www.gabi.rzpd.de/projects/MapMan/; Thimm et al., 2004; Usadel et al., 2005). Other illustrative screen shots are provided as Figures S4–S6.
Carbon starvation leads to widespread changes of transcripts (Figure 4a). A few were reversed 30 min after adding sucrose (Figure 4b and Table 5). Rapid responses include repression of TPS8, TPS9, TPS10 and TPS11, BAM9 (β-amylase), a glycerophosphoryldiesterase, a small set of lipases, genes involved in myoinositol (MIOX1, myoinositol oxidase 1) and proline (POX) degradation and induction of GPT2 (envelope Glc6P-transporter), APT2 (chloroplast Pi transporter) and IPS1 (inositol-3-phosphate synthase; Table 5).
Table 5. Genes assigned to central metabolism that respond rapidly to sucrose readdition
For abbreviations see Table 1. A broad overview of the response of transcripts in central metabolism is provided in Figure 4.
Most changes revert by 3 h (Figure 4c). Transcripts increase for genes assigned to starch synthesis (GPT1, GPT2, APL4), sucrose breakdown, glycolysis, the TCA cycle, mitochondrial electron transport, fermentation, nitrate and ammonium uptake (see Figure S4), ammonium and sulfate assimilation, amino acid and nucleotide synthesis, cell wall precursor synthesis and (see Figure S4) UDPGal transport into the endoplasmic reticulum. Transcripts decrease for genes assigned to amino acid degradation, GDH1, GDH2 and ASN1 (implicated in a pathway that recycles nitrogen into asparagine during C deprivation; Lam et al., 1998; Melo-Oliveira et al., 1996), the light reactions, the Calvin cycle and TPT1 (triose-phosphate export; Flügge, 1999).
Activities of enzymes in central carbon and nitrogen metabolism
Enzyme activities were measured with automated microplate-based assays (Gibon et al., 2004b; see Table S9 for data). Figure 5(a) depicts activities during C depletion and sucrose readdition, relative to FN. Two days of C depletion leads to a significant decrease of activity for enzymes in carbohydrate degradation (fructokinase, glucokinase), glycolysis [phosphofructokinase (PFP)], respiration (PFP, fumarase), the oxidative pentose phosphate (OPP) pathway [glucose-6-phosphate dehydrogenase (Glc6PDH)] and amino acid synthesis (aspartate aminotransferase (AspAT), shikimate dehydrogenase (Shik-DH)]. A further day of C starvation leads to a significant decrease of phosphoenolpyruvate carboxylase (PEPCase), sucrose-phosphate synthase (SPS) and nitrate reductase (NR) activity. Many changes revert in the first 24 h after adding sucrose (glucokinase, fructokinase, fumarase, AspAT, Glc6PDH, Shik-DH). Carbon starvation leads to increased activity of glutamate dehydrogenase (GDH; amino acid catabolism, see above), ADP-glucose pyrophosphorylase (AGPase; starch synthesis), glutamine synthetase (GS) and NADP-dependent glyceraldehyde 3-phosphate dehydrogenase (NADP-GAPDH; Calvin cycle). This is not reversed 24 h after adding sucrose.
Figure 5(b) compares the responses of enzymes and transcripts during C depletion. Most enzymes are encoded by small gene families. The transcript level for each member is indicated by the length of the bar, and the change during C-starvation by a color code. The change of activity is depicted using the same color scale. In many cases, the changes of activity reflect changes of transcripts [glucokinase, fructokinase, PFP, PEPCase, NADP-dependent isocitrate dehydrogenase (NADP-ICDH), fumarase, AspAT, glutamate dehydrogenase (GLDH), NADP-GAPDH and, with the exception of one of five genes, Glc6PDH]. However, activity changes less than transcripts. The increase of AGPase activity is accompanied by increased levels of AGS and APL1, while transcripts for less strongly expressed AGL family members decrease. Some enzymes show changes of transcript and no change of activity [glycerokinase, alanine aminotransferase (AlaAT), NR] or changes of activity but not of transcript (Shik-DH). Sometimes there is a discrepancy; transcript for cytosolic fructose-1,6-bisphosphatase (FBPase) drops markedly but activity is unaltered, transcripts for invertase rise but activity decreases slightly, and transcripts for SPS are unaltered or increase whereas activity falls. In three cases [ferredoxin-dependent glutamine-2-oxoglutarate aminotransferase (Fd-GOGAT), glutamine synthetase, pyruvate kinase], comparison is impossible because transcripts for individual family members show disparate responses. After readdition of sucrose, enzyme activity changes more slowly than transcripts; while transcripts recover or overshot by 3 h (see above) enzymes recover gradually over 24 h (Figure 5a).
Figure 6(a) summarizes the response of 70 metabolites which were present in sufficient abundance to enable accurate relative quantification with gas chromatography-mass spectroscopy (GC-MS) and liquid chromatography-mass spectroscopy (LC-MS; see Table S10 for the original data). In C-starved seedlings, carbohydrates (see also Figure 1), uridine diphosphoglucose (UDP-glucose) and organic acids (pyruvate, succinate, fumarate, malate) are low. Many other C-containing metabolites decrease, including myoinositol, raffinose, glycerate and fatty acids (especially C16:2 and C18:2). Central amino acids (Gln, Glu, Asp, Ala) and shikimate (an intermediate in aromatic amino acid synthesis) decrease, indicating that N metabolism is inhibited. Methionine decreases, indicating that S assimilation is inhibited. Most other minor amino acids increase (Ileu, Leu, Val, Phe, Tyr, Trp, Arg), indicating that proteolysis has commenced. Isopentyl pyrophosphate (an intermediate of terpenoid biosynthesis) decreases, while 2,3-dimethyl-5-phytylquinol (an intermediate of tocopherol biosynthesis) and α- and γ-tocopherol increase.
The response after adding sucrose provides insights into how quickly biosynthetic metabolism is re-established. Sucrose rises in the first 3 h, and sugars and starch more slowly (see above). Uridinediphosphoglucose and pyruvate recover by 3 h, indicating that glycolysis has been activated. The transient increase of Gln, Asp and Ala between 75 min and 3 h indicates that N-assimilation is stimulated. Shikimate rises after 3 h, indicating that aromatic amino acid synthesis starts a little later. The increase of Met indicates that S assimilation has been activated. Glutamine, Asp, Ala and most minor amino acids (Val, Ileu, Leu, Arg, Thr, Tyr, Trp, Phe) decrease from 3 h onwards, indicating that protein synthesis is starting (see below for more data). Most organic acids are present at high levels, and are located mainly in the vacuole (Gerhardt et al., 1987). They are used as an alternative C source during starvation, together with raffinose, myoinositol, glycerate and fatty acids (see above). These metabolites do not recover until 24 h or later. Tocopherol and isopentyl pyrophosphate do not revert in the first 24 h.
Figure 6(b) compares the changes of selected metabolites and with enzymes, which are directly or indirectly involved in their synthesis. All parameters are normalized on FN. Reducing sugars increase, even though acid invertase remains unaltered and glucokinase and fructokinase (see Figure 5a) increase. Starch changes independently of AGPase activity. The gradual recovery of organic acids occurs in a similar time frame to the recovery of two glycolytic (PFP, PEPCase) and one TCA cycle (fumarase) enzyme. The major central amino acids (Gln, Glu, Asp, Ala) were summed and compared with enzymes involved in nitrate (NR) and ammonium (GOGAT) assimilation and central amino acid metabolism (AspAT, AlaAT). There is reasonable agreement, especially with NR. There is good agreement between the recovery of Shik-DH activity and its product, shikimate.
Genes involved in cellular growth
Protein decreases 40% on a fresh weight basis in C-starved seedlings. Protein starts to recover 8 h after adding sucrose, but after 24 h is still lower than in C-replete seedlings (Figure 7). The changes of expression of genes for RNA and protein synthesis noted in Figure 2 are explored in more detail in Figure 8. Carbon starvation leads to a weak but widespread repression of genes for RNA and protein synthesis, especially cytosolic ribosomal proteins (Figure 8a). After adding sucrose there are no marked changes after 30 min (Figure 8b) but widespread changes by 3 h (Figure 8c). This includes induction of genes implicated in nucleolus function (AT5g55920, At5g56950, At4g26600, At3g57150, Atg312860; Table S2), cytosolic ribosomal proteins, amino acid activation and translation initiation (including ELF-4A) and elongation. The TIGR5 annotations of genes assigned to the unfolded protein response (UPR) by Martinez and Chrispeels (2003) were inspected to identify genes involved in protein folding (see Table S11). These genes are weakly suppressed in C-starved seedlings, unaltered 30 min after sucrose addition, and coordinately induced 3 h after sucrose addition (Figure 8). Genes for plastid ribosomal proteins are repressed after adding sucrose (Figure 8c), paralleling the repression of genes assigned to photosynthesis (Figures 2 and 4c).
Several genes involved in autophagy are induced in C starvation and repressed by sucrose readdition including APG8e, which targets autophagic vesicles to the vacuole (Doelling et al., 2002). APG8e is repressed 30 min after adding sucrose (Table 1). Carbon starvation induces several proteases including four putative Ser carboxypeptidases (At3g10450 and the tandem repeat At2g22920, At2g22980, At22990), two Cys proteases (At3g43960, At4g35350), three subtilisin-like proteases (At1g32940, At1g32970, At4g30020), one Asp protease (At3g12700) and several other proteases (At1g06430, At1g20380, At1g24140, At1g50250, At1g75460, At2g38860, At5g26860, At5g47040;Table S2).
Many genes involved in DNA synthesis are weakly repressed in C-starved seedlings, unaffected 30 min after adding sucrose, and induced 3 h after adding sucrose (Figure S5; see also Li et al., 2006). Class 2a, 2b, 3, 4 and 5 histones were induced, while class 1 histone (At2g18050) was strongly repressed by sucrose. Sugar-dependent changes in histone expression were noted by Davies et al. (1996). Sets of genes whose transcripts increase in the G1, S, G2 or M phase in synchronized Arabidopsis cultures were identified by Menges et al. (2002, 2003; see Table S11). The average change of transcript levels was calculated for each gene set. There were no systematic changes 30 min after adding sucrose to C-starved seedlings, but the average signal for the S-phase gene set doubled after 3 h (see Figure S6).
Identification of rapidly responding genes
Genes whose transcripts respond rapidly are good candidates for upstream transcriptional signaling components. In total, 165 genes show a >1.4-fold (log2 scale) change 30 min after adding sucrose (Table S12). Many of these changes were transient; most show partial reversion by 3 h, and little or no difference between FN and C-starved seedlings (Figure 9a). Genes that respond slowly after adding sucrose (a < 1.4-fold change after 30 min, but a >2.5-fold change by 3 h, log2 scale) usually show larger differences between FN and C-starved seedlings (Figure 9b). Figure 9(c) compares the number of genes that change by >log2(1.4) 30 min and 3 h after adding sucrose, and shows the overlap with genes that change by >log2(1.4) after 2 h in Li et al. (2006). Most of the genes whose transcripts respond within 30 min are present in the 3 h set from our study and/or the 2 h set from Li et al. (2006). However, they are masked because about tenfold more genes change after 2–3 h. The genes in our 30-min data set are spread across several of the clusters defined from 2-, 4- and 6-h responses in Li et al. (2006), representing <20% and often <10% of the genes in a particular cluster (data not shown).
Responses of rapidly responding genes to changes of endogenous sugars in rosettes
To learn if these rapidly responding genes respond to small changes of endogenous sugar levels, their response was inspected in four experiments that modify endogenous sugars in Arabidopsis rosettes: (i) Comparison of the end of the night and the end of the day in wild-type plants (Bläsing et al., 2005). (ii) Comparison of the end of the night and the end of the day in the starchless pgm mutant; at the end of the night sugars are lower and at the end of the day they are tenfold higher than in wild-type plants (Bläsing et al., 2005; Gibon et al., 2004a). (iii) Comparison of plants that were illuminated for 4 h with 50 or 350 p.p.m. [CO2] (Bläsing et al., 2005); this treatment started at the end of the night, so sugars in the 50 p.p.m. treatment are very low and in the 350 p.p.m. treatment resemble those found in the light in wild-type plants. (iv) The difference between the end of the night and a 6-h extension of the night (Thimm et al., 2004); sugars are quite low at the end of the night, and are exhausted when the night is extended.
The scatter plots in Figure 10(a–d) compare the response 30 min after adding sucrose to C-starved seedlings with the response in each of the rosette treatments. Agreement is strongest with the pgm light period (Figure 10c, R2 = 0.66) and the [CO2] treatments (Figure 10a, R2 = 0.50), weaker with the wild-type light period treatment (Figure 10b, R2 = 0.31) and weakest for the extended night treatment (Figure 10d, R2 = 0.25). In general, agreement is better for repressed than induced genes. Table 6 summarizes the results for a set of sucrose-repressed gene including TRs, components of the UBQ targeting system, members of the TPS family, APG8e and thioredoxins and glutaredoxins. A complete list for all 165 genes is provided in Table S12. In an analogous analysis using all the genes on the ATH1 array, agreement was again best for the pgm light period and [CO2] treatments (see Figure S7).
Table 6. Genes that are rapidly repressed after addition of sucrose to seedlings, and which show robust responses to changes of sugars across a wide range of treatments in which exogenous sugars are added, or the endogenous pools of sugars change
AGI locus and description
The table shows from left to right the changes of transcript levels 30 min (30′S/CS) and 3 h (3hS/CS) after addition of 15 mm sucrose to C-starved seedlings, the difference between seedlings in full nutrition medium and C-starved seedlings min (FN/Ct; data from this paper), the response 3 h after adding 100 mm glucose to C-starved seedlings (Glc3h/CSt; Bläsing et al., 2005), the difference between rosettes illuminated at the end of the night for 4 h at 350 or 50 p.p.m. [CO2] (350/50 CO2; Bläsing et al., 2005), the difference between the start and end of the light period in wild-type plants (WT ED/EN), the difference between the start and end of the light period in pgm (pgm ED/EN; Bläsing et al., 2005) and the difference between the end of the night and a 6-h extension of the night in wild-type plants (6-h ExN/EN; Thimm et al., 2004). To aid inspection, the changes of transcript levels in the various treatments are color coded. The color coding is reversed for the extended night treatment, because in this case sugars are decreasing rather than rising. Abbreviations are as in Tables 1–5; EN, end of night; ED, end of day; ExN, extended night. See Table S12 for a full analysis, and genes that are rapidly induced by sucrose.
At4g37610 TAZ/BTB/POZ domain-containing protein
At1g80440 Kelch repeat-−containing F-box family protein
At3g48360 Speckle-type POZ protein
At1g23390 Kelch repeat-containing F-box family protein
At1g69490 No apical meristem (NAM) family protein|
At2g25900 Zinc finger (CCCH-type) family protein
At4g03510 Zinc finger (C3HC4-type RING finger) family
At5g49450 bZIP family transcription factor
At3g59940 Kelch repeat-containing F-box family protein
At1g70290 TPS8, trehalose-6-phosphate synthase
At3g23880 F-box family protein
At5g41080 Glycerophosphoryl diester phosphodiesterase y
At1g13260 DNA-binding protein RAV1
At3g61060 F-box family protein/lectin-related|
At2g44130 Kelch repeat-containing F-box family protein
At5g61440 Thioredoxin family protein
At4g15700 Glutaredoxin family protein
At2g45170 Autophagy 8e (APG8e)
At1g03850 Glutaredoxin family protein
At4g36040 DnaJ-like protein DnaJ-like protein
At5g22920 Zinc finger (C3HC4-type RING finger) family
At5g61590 Ethylene-responsive element binding EREBP5
At2g17880 Putative DnaJ protein
At1g68520 Zinc finger (B-box type) family protein
At5g13080 WRKY DNA-binding protein
At5g18670 Beta-amylase 9 (BAM9)
At1g76410 Zinc finger (C3HC4-type RING finger) family
At4g28270 Zinc finger (C3HC4-type RING finger) family
At2g32150 Haloacid dehalogenase-like hydrolase family
At2g18300 Basic helix-loop-helix (bHLH) family protein
At3g15500 Putative jasmonic acid regulatory protein
At1g72200 Zinc finger (C3HC4-type RING finger) family
At5g61600 DNA-binding protein EREBP-4
At5g17300 myb family transcription factor
At4g15680 Glutaredoxin family protein
At4g11360 Zinc finger (C3HC4-type RING finger; RHA1b)
At5g21170 5′-AMP-activated protein kinase beta-2 subunit
Inspection of Table 6 (see also Table S12) reveals that some genes show changes during the light/dark cycle in wild-type plants which are as large as those in pgm. Others show smaller changes in wild type than pgm, but tend to have large changes in the extended night treatment. This is confirmed by Figure 10(e), which compares the difference between the response in pgm and wild-type plants with the magnitude of the change in an extended night treatment for all 165 genes. This indicates that rapidly responding genes respond to different ranges of sugar; for some the response saturates in the range found in the undisturbed diurnal cycle, whereas others start to respond in this range but require a further depletion of sugars to achieve the full response. This explains the good agreement between the 30-min response and the pgm light period response (this covers the full range), the weaker agreement with the wild-type light period response (this covers the mid part of the range) and the poor agreement with the extended night response (this covers only the lowest part of the range). A similar conclusion is supported by inspecting the diurnal changes of the transcripts these 165 genes (Figure 10f,g). Most show strong diurnal changes in pgm. Sucrose-repressed genes show a strong decrease within the first 4 h of the light period, remain low for the remainder of the light period and rise after 4-h dark. Sucrose-induced genes show a more complex response; many are induced 4 h into the light period, and are partially or completely repressed later in the light period (Figure 10f). Many of these genes also show a transient response to sucrose (see Figure 9), indicting they may be under feedback regulation. In wild-type plants, some but not all show marked changes during the light/dark cycle.
To compile a robust list of genes that rapidly respond to exogenous sucrose and also respond to endogenous sugars, the 165 genes were filtered by requiring that they show >1.4-fold changes in at least one of the treatments that alter endogenous sugar levels in leaves. This is a conservative list, which may exclude genes with very transient responses. Figure 10(g) shows the distribution of the remaining 121 genes among major functional categories. The largest group is unknown or unassigned genes, followed by TRs, UBQ-targeting components and four of the 11 genes annotated as TPS.
Comparison of sucrose and glucose addition
In our experiments, seedlings were kept in low light and high nitrate throughout, and starved for 2 days, before adding 15 mm sucrose or (Bläsing et al., 2005) 100 mm glucose for 3 h. Li et al. (2006) starved seedlings for 1 day in weak light, before adding 167 mm glucose for 2 h. Price et al. (2004) starved Arabidopsis seedlings for 1 day in the dark and absence of nitrogen, before adding 167 mm glucose for 3 h in the dark in the presence or absence of 40 mm nitrate. The medium-term response to glucose is similar (R2 = 0.83) in Bläsing et al. (2005) and Li et al. (2006), and more divergent between Bläsing et al. (2005) and Price et al. (2004) (R2 = 0 69), possibly due to the divergent experimental design (Figures S8a and S9a). There is weaker agreement between sucrose and glucose; comparison of the 3-h response to sucrose in the present paper with the responses to glucose in Bläsing et al. (2005); Li et al. (2006) and Price et al. (2004) gave R2 values of 0.65, 0.64 and 0.49, respectively (Figures S8b,c and S9b). Comparison of our sucrose data set with Price et al. (2004) reveals >100 genes with opposite responses, using a filter of ≥twofold change (Table S13). Comparison with Bläsing et al. (2005) reveals that many transcripts are altered strongly by one and weakly by the other sugar (Figure S8c). This might reflect gradual interconversion of sucrose and reducing sugars (Figure 1).
Price et al. (2004) and Li et al. (2006) reported that glucose induces a subset of heat-shock proteins (HSPs) and stress-related genes. A similar response is seen to sucrose and changes of endogenous sugars (Figure S8d, Table S14). Another set of stress-related genes is repressed by adding sugar and high endogenous sugar (Table S14, Figure S8d); this includes several DnaJ-like proteins (HSPs that are co-chaperones in protein translocation into plastids; Bukau and Horwich, 1998).
Biosynthetic metabolism is gradually re-established after adding sucrose
Carbohydrate depletion in Arabidopsis seedlings leads to an inhibition of nutrient assimilation, biosynthesis and growth, and stimulates the catabolism of a wide range of alternative C substrates including organic acids, raffinose, myoinositol, lipid and protein. This resembles the response in other species (Aubert et al., 1996; Brouquisse et al., 1991; Devaux et al., 2003; Yu, 1999). When sucrose is added, these changes revert gradually. Uridine diphosphoglucose (UDPGlc), pyruvate and the central amino acids recover by 3 h, revealing that glycolysis and N assimilation have been activated. Shikimate rises from 3 h onwards, indicating that there is a slightly longer delay before amino acid synthesis starts. Diagnostic genes for the S-phase of the cell cycle are induced by 3 h. Myoinositol, malate and fumarate do not recover until 24 h, and succinate, raffinose, lipids and protein recover even more slowly. Minor amino acids accumulate in C-starved seedlings, as a result of protein breakdown (see above). They decrease after 3–8 h, indicating that protein synthesis has commenced. Protein starts to rise by 8 h, but does not fully recover in the first 24 h. The delay in re-establishing biosynthesis and growth results in a temporary imbalance between the supply and utilization of C, leading to a large transient accumulation of sucrose, reducing sugars and starch. A similar response is seen when wild-type Arabidopsis is re-illuminated after a 6-h extension of the night, or at the start of the light period in the starchless pgm mutant (Gibon et al., 2004a).
Identification of rapidly responding regulatory genes
Carbon starvation modifies the levels of thousands of transcripts. Almost all of these changes are reversed 3 h after adding sucrose. Many transcripts overshoot; this may be related to the transient accumulation of carbohydrates. A similarly dramatic response was found 2–3 h after adding glucose (Li et al., 2006; Price et al., 2004). This highlights the wide range of metabolic and cellular processes that are modulated by sugars, but complicates dissection of the signaling pathways.
A simpler picture emerges 30 min after adding sucrose. First, tenfold fewer genes show marked changes. Second, many of the changes are transient; some are partly reversed after 3 h, and almost all are partly or completely reversed in C-replete seedlings. Third, many of the genes that respond after 30 min are good candidates for upstream signaling components. Price et al. (2004) noted a small over-representation of TRs 3 h after adding glucose to C-starved seedlings. This over-representation is far more evident 30 min after adding 15 mm sucrose, and extends to other classes of regulatory genes including the UBQ/26S proteosome pathway, redox components and trehalose metabolism.
The Arabidopsis genome encodes >2000 TRs (Davuluri et al., 2003; Riechmann, 2002). Transcription regulators initiate widespread changes in gene expression. Our analyses identify and validate >20 rapidly responding sucrose-responsive TRs. Many were not previously known to be sugar responsive, especially those discovered with the quantitative (q)RT-PCR platform. Some are already implicated in sugar signaling. AtbZIP11 is expressed in vascular tissue, induced by light and sucrose, and translationally repressed by high sucrose (Rook et al., 1998). Sucrose acts on a highly conserved upstream open reading frame (uORF) coding for 42 amino acids in the 5′UTR (Wiese et al., 2004, 2005). The sucrose-induced elongation factor 3h (ELF3H) is implicated in the translational regulation of AtbZIP11 (Kim et al., 2004). The rapid response of an Aux/IAA (Abel et al., 1995) family member is indicative of early cross-talk with auxin. There were slower but large changes of (At1g66390/MYB90/PAP2), which is implicated in the regulation of anthocyanin and flavonoid biosynthesis (Borevitz et al., 2000). This TR is strongly repressed after adding nitrate to N-deprived seedlings (Scheible et al., 2004).
Over 1300 genes contribute to protein degradation via the 26S proteosome (Vierstra, 2003) including about 42 E2-UBQ conjugases and about 1200 E3-UBQ ligases. This pathway initiates widespread downstream changes by degrading key target proteins. It is implicated in hormone, light and nutrient signaling (Dharmasiri et al., 2005; Guo and Ecker, 2003; Itoh et al., 2003; Vierstra, 2003; Yang et al., 2005). Sucrose addition leads to a rapid decrease of transcripts for 15 putative E3-ligases, including three TAZ/BTB/POZ proteins. The latter merge the functional properties of SKP1 and F-box proteins into a single polypeptide (Xu et al., 2003) and interact with cullins to form multisubunit E3-UBQ ligase complexes (Figueroa et al., 2005). Our results indicate a set of unknown proteins required for metabolism or growth are degraded by E3-ligases in C-starved plants. One intriguing possibility is that the targeted proteins may be encoded by genes that are rapidly induced by sucrose. Price et al. (2004) reported that part of the downstream response of gene expression to glucose requires protein synthesis. This might be related to the need to synthesize the targets of sucrose-repressed E3-ligases. Several of the sucrose-regulated E3-ligases (At1g80440, At1g22500, At1g76410, At3g59940, At5g22920, At3g48360) respond in an opposite manner after nitrate addition (Scheible et al., 2004).
The rapid response of several genes in trehalose metabolism (TPS8, -9, -10, -11) prompts the hypothesis that trehalose 6-phosphate (Tre6P) is a sugar signal. Recent studies show Tre6P increases 40-fold after adding sucrose to C-starved Arabidopsis seedlings (Lunn et al., 2006), and implicate Tre6P in the post-translational activation of AGPase and starch synthesis (Kolbe et al., 2005; Lunn et al., 2006). Further studies are needed to discover whether Tre6P has other targets. This appears likely, in view of the embyro-lethal phenotype of tps1 (Eastmond et al., 2002), and the complex phenotypes of TPS- and TPP-overexpressing transformants (Schluepmann et al., 2003). It is unclear how sucrose leads to this increase of Tre6P. Of the 11 annotated Arabidopsis TPS genes, only TPS1 has been demonstrated to possess catalytic activity, and transcript for this gene respond to sugar. The genes that respond to sugars are annotated as TPS, but the ORF includes a TPP domain, and analysis of the TPS domain reveals changes of amino acids in the catalytic domain (J. Lunn, Max Planck Institute for Molecular Plant Physiology, Golm, Germany, data not shown).
Sucrose addition alters the levels of transcripts for many receptors and signaling components, indicating that signaling pathways are rewired in response to changes in the C status. A few respond rapidly. At5g21170 (AtAKINβ1) encodes a subunit of a protein kinase that is implicated in sucrose signaling (Boisson et al., 2003). Several genes involved in ABA biosynthesis (At4g19170, At5g67030; Iuchi et al., 2000; Qin and Zeevaart, 1999), two ABI3-interacting proteins (At5g02810, At5g61380) and two GRAM domain proteins (At5g08350, At5g23350; Kaczorowski and Quail, 2003; Kang et al., 2002; Más et al., 2003) are rapidly repressed, adding to the evidence for cross-talk between sugars and ABA (Koch, 2004; Smeekens, 2000). However, most do not show large responses until 3 h, indicating that they represent downstream responses. Over 60 sucrose-responsive protein kinases and two protein phosphatase 2C homologs (At2g25620, At5g02760) respond to sucrose readdition after 3 h. At5g02760 is also induced by nitrate (Scheible et al., 2004). Several receptor kinases (Shiu and Bleecker, 2001) were induced, including a tandem repeat between At1g51790 and At1g51890, where five were induced >threefold and three >15-fold. Three wall-associated kinases (WAKs; At1g21250/WAK1, At1g21270/WAK2, At1g79680) were repressed by C depletion and induced by sucrose. Wall-associated kinases bind to pectin and may modulate cell expansion (Anderson et al., 2001). WAK1 is upregulated in nitrogen deficiency (Scheible et al., 2004).
Another striking result was the strong induction of several proteases and APG8e in C-starved seedlings, and the rapid repression of APG8e after adding sucrose. Vacuole-dependent proteolysis has been reported during prolonged C starvation (Aubert et al., 1996; Brouquisse et al., 1991; James et al., 1996). Our results indicate that APG8e is a very early target of sugar signaling.
Coordinated but delayed regulation of central carbon and nitrogen metabolism
Carbon starvation leads to decreased levels of transcripts for many genes involved in biosynthetic pathways, and increased levels for genes involved in catabolism (see also Contento et al., 2004; Thimm et al., 2004). This is accompanied a decrease of the activity of several enzymes involved in carbohydrate degradation, glycolysis, respiration, the OPP pathway and amino acid synthesis. There are also cases where changes of transcripts do not affect enzyme activities, or enzymes change independently of transcripts.
After adding sucrose, transcripts and enzyme activities recover gradually. Most transcripts assigned to metabolism show only small changes after 30 min. By 3 h there is a coordinated induction of genes assigned to glycolysis, the TCA cycle, mitochondrial electron transport and ATP synthesis, amino acid synthesis, nucleotide metabolism and cell wall precursor synthesis, and decrease of transcripts for sucrose and starch breakdown, gluconeogenesis and amino acid catabolism. Li et al. (2006) reported that genes involved in starch synthesis, amino acid biosynthesis and secondary metabolism do not peak until 4–6 h after glucose addition. While transcripts are fully recovered or overshoot at 3 h, enzyme activities recover more gradually. The gradual recovery of enzyme activity may contribute to the reactivation of metabolism. There was a reasonable agreement between the kinetics of the recovery of three enzymes involved in glycolysis and respiration (PFP, PEPCase, fumarase) and the levels of organic acids that are produced by glycolysis, between the recovery of NR activity and the levels of central amino acids that are produced by nitrate assimilation, and between Shik-DH and the level of its immediate product, shikimate. In other cases, metabolites changed independently of enzyme activities; for example, the accumulation of reducing sugars and starch. Starch accumulation may be triggered by changes of Tre6P and post-transcriptional regulation of AGPase (see above).
Coordinated expression of genes for cellular growth processes and central metabolism
Sucrose induces hundreds of genes involved in growth processes, including DNA synthesis, RNA synthesis and processing, cytosolic protein synthesis and protein folding (see also Li et al., 2006; Price et al., 2004). One of the most strongly and rapidly induced is the α-subunit of elongation factor 1B (At5g12110). However, most genes in these categories are induced after a delay (see also Li et al., 2006). Amino acids decrease from 3 h onwards after adding sucrose. A similar response is seen after adding nitrate (Scheible et al., 2004) or phosphate (Scheible et al., in press). These results imply that protein synthesis is activated even more strongly than amino acid synthesis. They underline that there is a coordinated activation of central metabolism and growth processes, which utilize metabolites produced in central metabolism.
Coordinated changes of metabolites and transcripts in carbon-scavenging pathways
Comparison of the transcript and metabolite profiles provides evidence that sugars transcriptionally regulate metabolic pathways that scavenge alternative C sources. The repression of IPS1 and IPS2 and induction of MIOX2 and MIOX4 during C starvation and the rapid reversal of these changes after adding sucrose matches the decrease of myoinositol in C-starved seedlings and its gradual recovery after adding sucrose. POX transcript rises and proline falls in C starvation, and these changes are reversed after sucrose addition. Genes involved in lipid catabolism are induced in C-starved seedlings and repressed after sucrose addition, mirroring the decrease of fatty acids in C-starved seedlings and their gradual recovery after sucrose addition. The decrease was especially marked for C16:2 and C18:2, which are mainly in 1-18:2-2-16:2-monogalactosyldiacylglycerol, a major membrane lipid in the chloroplasts (Somerville and Browse, 1996). Dieuaide et al. (1993) reported that prolonged starvation promotes lipid catabolism. Our results show transcripts for genes involved in lipid breakdown and lipid levels responds rapidly to changes in the C status.
Coordinated repression of genes for photosynthesis and chloroplast biogenesis
There is considerable interest in the possibility that high sugar represses genes involved in photosynthesis (Paul and Foyer, 2001; Sheen, 1990; Stitt and Krapp, 1999). Price et al. (2004) and Li et al. (2006) reported that genes involved in photosynthesis are repressed by glucose. These changes are significant but small when FN seedlings are compared with C-starved seedlings, but larger 3 h after sucrose addition when there is widespread repression of genes involved in photosynthesis, plastid protein synthesis and plastid protein folding. This may reflect the transiently high levels of sugars due to the imbalance between C supply and utilization at this time.
Relevance to changes triggered by endogenous changes of sugars
Analysis of the response 30 min after adding 15 mm sucrose to C-starved seedlings identified >150 genes that respond rapidly to sugar, including about 20 TRs and about 15 components of the UBQ/26S proteosome pathway and further ideal candidates for upstream components of signaling pathways (see above). To confirm that these genes are involved in physiological responses to sugars, we inspected their response in four different treatments from Bläsing et al. (2005), which alter the endogenous sugar level in different ways and ranges. Most of them responded to endogenous changes of sugars. Agreement was especially good for genes that are repressed by sucrose. The rapidity of the response after adding sucrose implies that their transcripts are either short lived, or that their degradation is triggered by sugars. For genes that are induced by sucrose, a subset responded to endogenous sugars in these four treatments. Some show transient changes after illumination of darkening, which are reversed 4–8 h later, pointing to the presence of feedback loops that modulate expression of these upstream components. The available data sets for diurnal responses have a temporal resolution of about 4 h. A higher time resolution may be needed to detect responses of genes with very transient responses to sugars.
Transcriptional responses to sugars involve multiple signaling pathways (Moore et al., 2003; Price et al., 2004; Rolland et al., 2006) and affect thousands of genes (Contento et al., 2004; Li et al., 2006; Price et al., 2004; Thimm et al., 2004). Addition of sugar to C-starved seedlings does not allow genes to be identified that respond to different levels of sugar, because seedlings shift so rapidly from C starvation to C excess. Integration of our robust list of rapidly responding genes with information about responses to endogenous sugars in diurnal cycles and extended darkness will aid dissection of these complex sugar responses. For some of these genes, the response saturates in the range found in an undisturbed light/dark cycle, while others respond when sugars fall to lower levels, for example in the starchless pgm mutant, or an extended night. Genes whose transcripts show changes in the wild-type light/dark cycle that are almost as large as those in pgm include a Myb factor, three C3H4 RING family proteins, two TAZ/BTB/POZ proteins, three Kelch F-box proteins, TPS8, TPS9, TPS10 and TPS11, APG8e and two glutaredoxins.
In conclusion, C starvation leads to major changes of metabolism and growth, which are only gradually reversed after adding sucrose back to C-starved seedlings. This highlights the importance of avoiding an acute C limitation. To do this, plants will have to sense and respond to falling sugars before they fall into the range where there is an acute limitation of metabolism and growth. Analysis of gene expression after resupplying sucrose to C-starved seedlings has identified a set of >150 genes that show large transcriptional responses within 30 min of resupplying sucrose. Many of these show large changes of their transcript levels during an undisturbed diurnal cycle in wild-type plants, highlighting them as good candidates for upstream components of signaling pathways that adjust metabolism and growth to the C supply before an acute C starvation develops.
Plant growth media and conditions
Wild-type Col-0 Arabidopsis seedlings (100–120) were grown in sterile liquid culture (250-ml Erlenmeyer glass flasks) on orbital shakers with constant, uniform fluorescent light (∼50 μE in the flask) and temperature (22°C), in 30 ml of sterile FN medium, with the same macronutrients and microelements as in Scheible et al. (2004). The shaker speed was low (30 r.p.m.) during the first 3 days, and was then increased to 80 r.p.m. Care was taken to prevent the seedlings from significant clumping. After 7 days, media were changed in all flasks. Some received 30 ml of FN medium and others received medium that contained all the components in FN medium except sucrose. The oxygen concentration in the medium was about 22%, showing that the seedlings performed photosynthesis and maintaining a small positive C balance at the whole plant level. Internal nitrate was >40 mm in all treatments (not shown), showing the plants were always nitrogen replete.
Sucrose addition and seedling harvest
On day 9 FN cultures and some of the CSt cultures were harvested. At the same time all the other flasks of CSt cultures were opened, and either reclosed without addition or after adding 450 μl 1 m sucrose (15 mm), 450 μl water (control) or 450 μl 1 m mannitol (15 mm, osmotic control). The added liquid was allowed to disperse without changing the shaking speed. Groups of CSt flasks that either received or not sucrose or mannitol were harvested after 12 min, 30 min, 75 min, 3 h, 8 h and 24 h. Seedlings grown on full nutrient medium were also harvested at 24 h. Plant materials from each flask were quickly (<10 sec for the entire procedure) blotted on tissue paper, washed twice in an excess of desalted water, blotted on tissue paper and frozen in liquid N2. Material from three flasks was pooled per sample. Materials were stored in liquid N2 until pulverization using a mortar and pestle at −180°C, and then sub-aliquoted and the powder kept at −80°C until further use.
Ribonucleic acid preparation, array hybridization, data analysis and MapMan display
The ATH1 analysis was performed on two biological replicates from experiments at a 2-month interval for seedlings from FN, Cst and 30 min and 3 h after adding 15 mm sucrose or 15 mm mannitol as an osmotic control. The RNA preparation, array hybridization and data analysis were carried out as in Scheible et al. (2004). The normalized signal intensities for all 22 750 ATH1 probe sets from the 12 arrays are given in Table S1. The raw Affymetrix signals (CEL files) were processed using RMA (log scale robust multi-array analysis) software (Bolstad et al., 2003). It is based on the quantile normalization method and has better precision than MicroArray Suite 5.0 (Affymetrix, Santa Clara, CA, USA) and dCHIP (http://www.biosun1.harvard.edu/complab/dchip/), especially for low values (Irizarry et al., 2003).
The averaged signals for a given treatment were expressed relative to those in CSt seedlings of the same experiment, the ratios averaged, converted to a log2 scale. The data were visualized and figures produced using MapMan software (Thimm et al., 2004) and the update (Usadel et al., 2005) at http://www.gabi.rzpd.de/projects/MapMan/. The overview figures in the present article were prepared using mapping file version TAIR6_MappingFileaffy_22k_0406.
Real-time RT-PCR analysis
Sequences of RT-PCR primers for transcription factor genes, real-time PCR conditions, data analysis and procedures for cDNA synthesis were exactly as in Czechowski et al. (2004).
Leaf tissue was stored in liquid nitrogen and extracts for enzyme activities were prepared and enzyme activities were measured exactly as in Gibon et al. (2004b)).
Sucrose, glucose, fructose and starch were measured in the soluble and residual fractions of an ethanol–water extract (Scheible et al., 1997a) as described in Stitt et al. (1989). Amino acids were determined in the ethanol/water extracts by HPLC (Geigenberger et al., 1996). Frozen material was used for extraction of phosphorylated metabolites with perchloric acid and assayed as in Stitt et al. (1989). The analysis of metabolites by GC-MS was performed as described by Roessner et al. (2000). Liquid chromatography-MS, analyses were performed using an Agilent 1100 capillary LC system (Agilent Technologies, Santa Clara, CA, USA) coupled with an Applied Biosystems/MDS SCIEX API 4000 triple quadrupole mass spectrometer (Niessen, 2003; Applied Biosystems, Foster City, CA, USA). After reversed phase HPLC-separation detection and quantification was performed in the multiple reaction monitoring (MRM) mode (Gergov et al., 2003).
On request, all novel materials described in this publication will be made available in a timely manner for non-commercial research purposes, subject to requisite permission from any third-party owners of all or parts of the material. Obtaining permission is the responsibility of the requestor.
We are grateful to Florian Wagner and his team at RZPD Berlin (German Resource Center for Genome Research, Berlin) for providing expert Affymetrix array service, including all steps from total RNA to data acquisition. The support of Ralf Looser and the Bioanalytics Technical Center at metanomics in performing the metabolic analyses is gratefully acknowledged. The work was supported by the Max-Planck-Society and the BMBF-funded project GABI Verbund Arabidopsis III ‘Gauntlets, C and Nutrient Signaling: Test Systems, and Metabolite and Transcript Profiles’ (0312277A).