Approximating subcellular organisation of carbohydrate metabolism during cold acclimation in different natural accessions of Arabidopsis thaliana


Author for correspondence:

Arnd G. Heyer

Tel: +49 711 685 65050



  • Accessions of Arabidopsis thaliana originating from climatically different habitats show different levels of cold acclimation when exposed to low temperatures. The central carbohydrate metabolism plays a crucial role during this acclimation. Subcellular distribution of carbohydrates over the compartments cytosol, vacuole and plastids, and putative interactions of the compartments, are analyzed in three differentially cold-tolerant accessions of Arabidopsis thaliana, originating from the Iberian Peninsula (C24), Russia (Rschew) and Scandinavia (Tenela), respectively.
  • Subcellular carbohydrate concentrations were determined by applying the nonaqueous fractionation technique. Mathematical modeling and steady-state simulation was used to analyse the metabolic homeostasis during cold exposure.
  • In all accessions, the initial response to cold exposure was a significant increase of plastidial and cytosolic sucrose concentrations. Raffinose accumulated in all cellular compartments of cold-tolerant accessions with a delay of 3 d, indicating that raffinose accumulation is a long-term component of cold acclimation. Minimal rates of metabolite transport permitting steady-state simulations of metabolite concentrations correlated with cold tolerance, indicating an important role of subcellular re-distribution of metabolites during cold acclimation.
  • A highly regulated interplay of enzymatic reactions and intracellular transport processes appears to be a prerequisite for maintaining carbohydrate homeostasis during cold exposure and allowing cold acclimation in Arabidopsis thaliana.


Many temperate herbaceous plants species, including the model plant Arabidopsis thaliana, can grow at low temperature and even survive freezing (Hurry et al., 1995; Zhen & Ungerer, 2008). Exposure to low but nonfreezing temperatures induces a multifaceted and complex process termed cold acclimation, by which plants are able to increase their cold tolerance. During cold acclimation, numerous physiological and biochemical changes occur enabling plants to tolerate temperatures several degrees lower than before cold exposure (Xin & Browse, 2000; Hannah et al., 2006). Many studies have shown that cold acclimation capacity is a multigenic trait that is influenced by a variety of factors including regulation of gene expression, enzyme activity and metabolite concentrations (Stitt & Hurry, 2002; Cook et al., 2004; Davey et al., 2009). Reprogramming of the central carbohydrate metabolism, comprising the regulation of photosynthetic activity and concentrations of soluble sugars, was shown to play a crucial role during cold acclimation (Wanner & Junttila, 1999; Stitt & Hurry, 2002). In particular, the concentrations of sucrose and raffinose were shown to correlate with cold tolerance in Arabidopsis (Klotke et al., 2004). In a cytosolic reaction, raffinose is synthesized from sucrose and myo-inositol by raffinose synthase and may then be transported into chloroplasts to function as a protectant (Schneider & Keller, 2009). However, the accumulation of soluble sugars during cold exposure is insufficient to fully explain the process of cold acclimation (Hincha et al., 1996).

In Arabidopsis thaliana grown at low temperature, a reprogramming of carbon metabolism was shown to result in a shift in partitioning of fixed carbon into sucrose rather than starch (Strand et al., 1997, 1999). Additionally, the overexpression of sucrose phosphate synthase in Arabidopsis leads to an improved photosynthetic activity and an increased flux of fixed carbon into sucrose, associated with an increase in cold tolerance compared to wild-type plants (Strand et al., 2003). Besides the finding that sucrose may act as a cryoprotectant for membrane systems in plants (Hincha et al., 2003), it may also serve as a substrate for the synthesis of other cryoprotective compounds, such as raffinose. Analyzing the cold acclimation potential of two genetically distinct Arabidopsis accessions, C24 and Columbia (Col-0), the basic cold tolerance and the capacity to cold acclimate were found to correlate with tissue raffinose concentrations (Klotke et al., 2004). Recently, it was demonstrated that raffinose specifically acts to protect the photosystems located in the thylakoid membranes of plastids from damage during freeze-thaw cycles (Knaupp et al., 2011).

The analysis of regulatory instances involved in reprogramming of the central carbohydrate metabolism is complex. First, this is due to the large number of cellular metabolite interactions, which frequently include nonlinear enzyme kinetics. In case of allosteric effectors, metabolites are part of regulatory circuits themselves, for example, by inhibiting or activating enzymes. Secondly, metabolism of photosynthetically active leaf cells is highly compartmentalized with substantial redundancy of metabolic pathways among the compartments (depicted schematically in Fig. 1). While CO2 is fixed in chloroplasts, sucrose is synthesized in the cytosol and then either exported to sinks or transported into the vacuole or chloroplast. Alternatively, it is cleaved by cytosolic invertase or consumed during synthesis of raffinose. Hexoses, which are products of sucrose cleavage, can either be transported across the tonoplast and the chloroplast envelope or they can be phosphorylated by hexokinase thereby supplying substrates for re-synthesis of sucrose. Considering metabolic cycles and the numerous metabolite transport processes across intracellular membrane systems (an overview given in (Linka & Weber, 2010)), it becomes obvious that principles of regulation allowing establishment of metabolic homeostasis are difficult to unravel. One attractive method for studying complex networks is mathematical modeling and simulation, which was already applied successfully to various biochemical networks (Ross & Arkin, 2009). A frequently used approach of mathematical modeling is the representation of biological networks by ordinary differential equations (ODEs) describing time-dependent changes in metabolite concentrations. The changes result from mass input and output, as well as enzyme-catalyzed metabolite interconversions. Considering compartment-specific metabolite concentrations, changes may also occur due to transport processes. ODE-based modeling approaches have been successfully applied to simulate various complex processes in plants, for example photosynthetic CO2 fixation (Pettersson & Ryde-Pettersson, 1988; Laisk et al., 2006), dynamics of the circadian clock in Arabidopsis (Zeilinger et al., 2006) and the diurnal regulation of central carbohydrate interconversions (Nägele et al., 2010; Henkel et al., 2011).

Figure 1.

Schematic representation of carbohydrate compartmentation in leaf cells of Arabidopsis thaliana. Reaction rates r represent central steps of metabolite interconversion. Transport processes t across the chloroplast envelope and the tonoplast are subdivided into forward (tf) and reverse (tr) reactions. Suc, Sucrose; Hex, Hexoses; Raf, Raffinose; pl, plastidial; cyt, cytosolic; vac, vacuolar; exp, exported.

Carbohydrate dynamics during cold acclimation of Arabidopsis thaliana are different for cold-sensitive vs -tolerant accessions and indicate a complex interplay of plastidial and cytosolic reactions (Nägele et al., 2011, 2012). Based on this finding, we performed a detailed analysis of the compartmental localization of carbohydrates during cold acclimation. We used nonaqueous fractionation to resolve carbohydrate concentrations in chloroplasts, cytosol and vacuole of the three differentially cold-tolerant accessions C24, Rsch and Te, where C24 represents a cold-sensitive accession, followed by the cold-tolerant accession Rsch and the most tolerant accession Te. The different levels of cold tolerance of these accessions, originating from the Iberian Peninsula (C24), Russia (Rsch) and Scandinavia (Te), were determined in previous studies (Hannah et al., 2006; Mishra et al., 2011), and the development of cold tolerance during a time course of 7 d of cold exposure was investigated in this study. Rates of net photosynthesis and activities of rate-limiting enzymes were incorporated into a steady-state modeling approach. The simulation of metabolite fluxes revealed distinct regulatory strategies for sucrose–hexose interactions and their compartmentation, allowing discrimination between cold-sensitive (C24) and –tolerant, as well as between tolerant (Rsch) and highly tolerant (Te) accessions.

Materials and Methods

Plant material

Arabidopsis thaliana (L.) Heynh (Brassicaceae), accessions C24, Rschew (Rsch) and Tenela (Te), were grown in GS90 soil and vermiculite (1 : 1) with three plants per 10 cm pot in a growth chamber at 8 h light (50 μmol m−2 s−1; 22°C) : 16 h dark (16°C) for 4 wk and then transferred to a glasshouse with a temperature of 22°C during the day (16 h) and 16°C during the night (8 h). In the glasshouse, natural light was supplemented to an intensity of at least 80 μmol m−2 s−1. The relative humidity was 70%. Plants were watered daily and fertilized with standard nitrogen-phosphorus-potassium fertilizer immediately after transfer to long-day conditions and a second time 2 wk later. At the bolting stage, that is, c. 3 wk after transfer to long-day conditions, a set of plants was harvested and another set was shifted to a growth chamber with a 16 h : 8 h light : dark regime of 4/4°C and a light intensity of 50 μmol m−2 s−1. Each sample consisted of three plant rosettes taken from three different pots in a random design before and after 1, 3 and 7 d of exposure to 4°C. Samples were taken at the mid-point of the light period. Because the aerial parts of the plants were composed exclusively of rosette leaves, metabolite and enzyme concentrations could be directly compared to CO2 exchange data. Samples were immediately frozen in liquid nitrogen and stored at −80°C.

Electrolyte leakage measurement

Cold tolerance was analyzed according to the electrolyte leakage method as described (Nägele et al., 2011). The cooling rate was set to 4°C h−1 and samples were taken at 2°C intervals over a temperature range of 0°C to −18°C. Conductivity was measured using an inoLab740 conductivity meter (WTW GmbH, Weilheim, Germany) and the multilabPilot software. The 50% lethality temperature (LT50) values were calculated as the log EC50 value of sigmoidal dose response curves fitted to the measured leakage values using Graphpad Prism 3 software (Graphpad Software Inc., La Jolla, CA, USA).

Gas exchange measurement

The exchange rates of CO2 were measured using an infrared gas analysis system (Uras 3 G, Hartmann & Braun AG, Frankfurt am Main, Germany). A whole-rosette cuvette design was used as described in (Nägele et al., 2010). Gas exchange was measured in the glasshouse and the growth chamber shortly before plant harvest. Means of raw data for gas exchange were converted to flux rates per gram DW obtained at the end of the exposure by freeze drying and weighing complete rosettes.

Carbohydrate analysis

Freeze dried leaf samples were ground to a fine powder. The homogenate was extracted twice in 400 μl of 80% ethanol at 80°C. Extracts were dried and dissolved in 500 μl of distilled water. Contents of glucose, fructose, sucrose and raffinose were analysed by high-performance anion exchange chromatography (HPAEC) using a CarboPac PA-1 column on a Dionex (Sunnyvale, CA, USA) DX-500 gradient chromatography system coupled with pulsed amperometric detection by a gold electrode. For starch extraction, pellets of the ethanol extraction were solubilized by heating them to 95°C in 0.5 N NaOH for 45 min. After acidification with 1 N CH3COOH the suspension was digested for 2 h with amyloglucosidase. The glucose content of the supernatant was then determined and used to assess the starch content of the sample.

Measurement of enzyme activities

Enzyme activities were determined in crude extracts of freeze dried leaf samples. To assess activities of soluble acid invertase and neutral invertase, c. 10 mg of dried leaf tissue were homogenized in 50 mM HEPES-KOH (pH 7.4), 5 mM MgCl2, 1 mM EDTA, 1 mM EGTA, 1 mM PMSF, 5 mM dithiothreitol (DTT), 0,1% Triton-X-100 and 10% glycerol. Suspensions were centrifuged at 2800 g for 25 min at 4°C and supernatants were desalted using a G-25 Sephadex gel filtration medium. Soluble acid invertase was assayed in 20 mM Na-Acetate buffer (pH 4.7) using 100 mM sucrose as a substrate. Neutral invertase was assayed in 20 mM HEPES-KOH (pH 7.5) using also 100 mM sucrose as substrate. The control of each assay was boiled immediately for 5 min. Reactions were incubated for 30 min at 30 and 4°C, stopped by boiling for 5 min, and the concentration of glucose was determined photometrically.

Activity of glucokinase and fructokinase was measured as described in (Wiese et al., 1999) at ambient temperature (22°C) and 4°C. Synthesized glucose-6-phosphate was converted to 6-phosphogluconolactone by glucose-6-phosphate-dehydrogenase and could be measured photometrically as a change in concentration of the reduced co-substrate NADPH. For isomerisation of fructose-6-phosphate, phosphogluco-isomerase was added.

Nonaqueous fractionation

Sub-cellular fractionation was based on the procedure described by (Iftime et al., 2011). Approximately 100 mg of freeze dried leaf homogenate were suspended in 10 ml heptane–tetrachlorethylene (ρ = 1.34 g cm−3) and repeatedly sonified on ice for 5 s with pauses of 15 s over a time course of 12 min (Branson Sonifier 250, Branson, Hannover, Germany). The sonified suspension was passed through 30-μm pore nylon gauze (Eckert, Waldkirch, Germany) and centrifuged. The sediment was suspended in heptane–tetrachlorethylene (ρ = 1.34 g cm−3) and loaded on a linear gradient of heptane–tetrachlorethylene (ρ = 1.34 g cm−3) to tetrachlorethylene (ρ = 1.6 g cm−3). After ultracentrifugation at 121 000 g for 3 h, the gradient was fractionated into nine 1-ml fractions that were divided into three sub-fractions of 0.3 ml. These were dried under vacuum. One of these fractions was used for marker enzyme determination and another for sugar analysis using HPAEC. Alkaline pyrophosphatase was measured as a marker for the plastidial compartment as described (Jelitto et al., 1992), UGPase was measured as cytosolic marker as described by (Zrenner et al., 1993). Acid phosphatase was used as marker for the vacuolar compartment (Boller & Kende, 1979).

A detailed description of the calculation process of the sub-cellular metabolite distribution is provided in Supporting Information Notes S1. The relative distribution of metabolites across the compartments as well as the total sugar concentrations are provided in Tables S1 and S2.

Mathematical modeling and simulation

A mathematical model of the central carbohydrate metabolism was developed comprising metabolite interconversions as well as transport processes, as indicated in Fig. 1. The model was based on a system of ordinary differential equations which is described in detail within the model structure given in Notes S2. Soluble carbohydrate concentrations of each accession after 0, 1, 3 and 7 d at 4°C were assumed to be at steady state (d/dt = 0). At steady state, the rate of raffinose synthesis (r5) became 0. This assumption was based on the finding that there was only minor enzymatic capacity for raffinose turnover in Arabidopsis thaliana (Nägele et al., 2012). Starch was assumed to be synthesised with a constant rate (r2) leading to the measured starch concentrations after 8 h in the light phase (14:00 h):

display math

Rates of net photosynthesis (rNPS) were calculated as the average rate of carbon uptake during the first half of the light phase (n = 8 h):

display math

(xNPSi, the integral of carbon uptake per hour).

The flux rate of carbon into sucrose synthesis (r1) was calculated as the difference between the rate of net photosynthesis and starch synthesis:

display math

At steady state, r1 becomes equivalent to the rate of sucrose export to sinks (t7).

The rate of cytosolic, plastidial and vacuolar sucrose cleavage (r3, r6, r7), catalysed by invertase, was modeled by a Michaelis–Menten enzyme kinetic including competitive product inhibition. Thus, rates of sucrose cleavage depended on substrate concentration, maximum activity of invertase and the enzyme specific substrate affinity, expressed by KM. It also depended on the concentration of the reaction product as well as the dissociation constant Ki for inhibitor binding.

The rate of hexose phosphorylation (r4) was described by the mass action rate law assuming the reaction rates to depend on substrate concentration and a rate constant.

Rates of metabolite transport were modeled by a mass action rate law as described for hexose phosphorylation.

Specific metabolite concentrations in cellular compartments were calculated assuming the volume proportions of chloroplasts, cytosol and vacuole to be similar to those described by Winter et al. (1994) for mesophyll cells in spinach leaves (Plastids: 10%; Cytosol: 5%; Vacuole: 80%). The remaining 5% of volume were assumed to consist of other cellular compartments, for example, mitochondria and nucleus. Experimentally determined metabolite concentrations, rates of net photosynthesis and enzyme activities were determined with respect to 1 g dry weight. The volume of 1 g of freeze-dried plant material was determined to be c. 10 ml which allowed for the calculation of concentrations in mM.

The identification of kinetic parameter sets enabling a steady-state simulation of metabolite concentrations (d/dt = 0, t→∞) was performed using a particle swarm pattern search method for bound constrained global optimization (Vaz & Vicente, 2007). Experimentally determined maximum rates of invertase and hexokinase (means ± SD) were set as upper and lower bounds for parameter identification.

The model was implemented in the numerical software MATLAB (R2009b) with the software packages Systems Biology Toolbox2 and the SBPD Extension Package (Schmidt & Jirstrand, 2006).


Cold acclimation and low temperature-induced changes in net photosynthesis and enzyme acitvities of hexokinase and invertase

The development of cold tolerance during 1 wk of cold exposure was assayed using the well-established electrolyte leakage method that yields the 50% lethality temperature LT50 (Fig. 2). The LT50 of C24 leaves was significantly higher at all time points of cold acclimation when compared to Rsch and Te (P < 0.01). Basic tolerance of C24 was −3.54 ± 0.1°C and increased to −5.3 ± 0.2°C during 7 d at 4°C. Basic tolerance in Rsch and Te was −4.9 ± 0.1°C and −5.3 ± 0.2°C, respectively. Te displayed an almost linear increase of cold tolerance until day 3 at 4°C (−8.0 ± 0.15°C) and gained a further 1.2°C of tolerance, reaching −9.2 ± 0.12°C after 1 wk. In Rsch, the LT50 decreased to −6.3 ± 0.3°C within 3 d and ended up at −8.1 ± 0.2°C after 7 d in the cold. During the first 24 h of cold acclimation induced by the 4°C treatment, rates of net photosynthesis (rNPS) were significantly reduced in all accessions (P < 0.05; Fig. 3). In C24, rNPS further decreased between days 1 and 3 of cold exposure (P < 0.05) and was significantly lower than in Rsch (P < 0.05). After 7 d at 4°C, there was no significant difference of rNPS between all accessions.

Figure 2.

Cold tolerance expressed as LT50 of Arabidopsis thaliana, C24 (blue line), Rsch (red line) and Te (green line) over time of exposure to 4°C. Filled circles represent means ± SD of six replicates (n = 6).

Figure 3.

Rates of net photosynthesis rNPS during cold exposure to 4°C in Arabidopsis thaliana, C24 (blue bars), Rsch (red bars) and Te (green bars). Bars represent means of measurements ± SD (n = 3).

In all accessions, the activities of hexokinase, as well as of soluble neutral and acid invertase were significantly (P < 0.001) decreased at 4°C due to thermodynamic effects on the enzyme activity (Fig. 4a–c). While hexokinase activities went up significantly between days 1 and 7 of cold exposure (P < 0.01; Fig. 4a), activities of both, neutral and acid invertase, remained low or decreased even further in the cold (Fig. 4b,c). In nonacclimated Te, maximum activities of acid invertase were significantly higher as compared to C24 and Rsch (P < 0.001; Fig. 4c).

Figure 4.

Maximum rates of cytosolic hexose phosphorylation (a), sucrose cleavage catalyzed by neutral invertase (b), and sucrose cleavage catalyzed by acid invertase (c) during cold exposure in Arabidopsis thaliana, C24 (blue bars), Rsch (red bars) and Te (green bars). Measurements of nonacclimated plants (non-acc) were performed at ambient temperature, those of cold-acclimated plants (1 d acc, 3 d acc, 7 d acc) were performed at 4°C. Bars, measurements ± SD (n = 5).

Starch content and subcellular concentration changes for soluble carbohydrates during cold exposure

The content of transitory leaf starch was not significantly altered during the first 24 h of cold exposure in all accessions and remained constant until day 7 in Rsch, while it decreased significantly (P < 0.05) in C24 and Te (Fig. 5). Light intensities at 22°C (80 μmol m−2 s−1) and 4°C (50 μmol m−2 s−1, see 'Materials and Methods') were set to restrict starch synthesis to the so-called programmed or baseline mode (Sun et al., 1999) in order to minimize differences in diurnal starch tunover at both growth regimes, because such differences would have biased model simulations for soluble sugars (Nägele et al., 2012).

Figure 5.

Starch contents at the mid of the light phase during cold exposure to 4°C in Arabidopsis thaliana, C24 (blue bars), Rsch (red bars) and Te (green bars). Bars, means of measurements ± SD (n = 5).

Plastidial sucrose content significantly increased during the first 24 h of cold exposure in all accessions (P < 0.01; Fig 6a). In Rsch, this increase was most pronounced, reaching a calculated concentration of 17.5 ± 1 mM. In the cold-tolerant accessions Te and Rsch, plastidial sucrose content declined between days 3 and 7, while in C24 it stayed constant. Cold-induced dynamics of cytosolic sucrose concentrations were similar to those in plastids except for Te, where a peak value of 51.7 ± 2.8 mM was reached after 3 d of cold exposure (Fig. 6b). Additionally, cytosolic sucrose concentrations were significantly higher than in plastids and vacuole both before and during cold exposure. Cold-induced dynamics of vacuolar sucrose concentration were most pronounced in Te, again reaching the highest concentration after 3 d at 4°C (Fig. 6c). Concentrations of free hexoses showed large fluctuations and increased in plastids of C24 and Rsch significantly during the first 24 h of cold exposure, while this increase was observed after 3 d in Te (Fig. 6d). After 7 d, C24 had the highest hexose concentrations in plastids (P < 0.01), while cytosolic hexoses were highest in Rsch (Fig. 6e). The dynamics of cytosolic hexose concentrations revealed a significant increase at day 1 in C24, while Rsch and Te showed peak values at day 3. Vacuolar hexose concentration peaked at day 1 in Rsch, at day 3 in C24 and continually increased until day 7 in Te. However, the concentrations were similar in Rsch and Te, but far lower in C24 (Fig. 6f).

Figure 6.

Compartment specific metabolite concentrations of sucrose (a–c), hexoses (d–f) and raffinose (g–i) during cold exposure in Arabidopsis thaliana, C24 (blue bars), Rsch (red bars) and Te (green bars). Bars, means ± SD (n = 5). Concentrations were calculated based on compartment specific volumes (see 'Materials and Methods').

Raffinose concentration in plastids rose significantly during the entire cold treatment in the tolerant accessions Rsch and Te (P < 0.001) but increased only slightly in C24 (Fig. 6g). While accumulation in Rsch was strongest between days 1 and 3, it was almost linear in Te. Concentrations of plastidial raffinose after 7 d at 4°C were 0.3 ± 0.07 mM in C24, 2.1 ± 0.2 mM in Rsch and 2.6 ± 0.45 mM in Te. During 7 d of cold exposure, cytosolic concentrations of raffinose rose significantly in all accessions (P < 0.01; Fig. 6h). However, concentrations were significantly higher in Rsch and Te than in C24 from the beginning of the cold exposure (P < 0.05). Vacuolar content of raffinose was also significantly elevated during cold exposure in all accessions, but accumulation was stronger in cold-tolerant accessions than in C24 (Fig. 6i).

Simulated rates of intracellular membrane transport and metabolite interconversion

Based on the experimentally determined metabolite concentrations and enzyme activities, a steady-state simulation of metabolite fluxes in C24, Rsch and Te was performed for nonacclimated plants as well as for plants exposed to 4°C for 1, 3 and 7 d. Although numerous intracellular metabolite transporters in plants have been identified, only a few have been biochemically well characterized (Linka & Weber, 2010), and thus it was impossible to comprehensively determine in vivo transport capacities at the tonoplast and the chloroplast envelope. To overcome this limitation and to minimize the number of unknown parameters in simulations, transport processes were modeled by the mass-action law – that is, the transport rate was assumed to be proportional to the concentration of the transported metabolite, disregarding possible effects of saturation, inhibition or activation. Even with this simplification, simulations yielded a multitude of solutions for the system of differential equations, because cyclic transport events cannot be restricted without exact knowledge of maximum transport activities. Therefore, we decided to identify the minimum rate constants that were necessary to allow simulation of a metabolic steady state. This identification was performed by increasing upper and lower bounds stepwise for the rate constants until every metabolite concentration could be simulated within its experimentally determined standard deviation. The limiting rate constants identified this way were then fixed, and only kinetic parameters of enzymatic metabolite interconversions were subject to further runs of parameter identification. Kinetic parameters and transport rate constants, which resulted from parameter identification, are summarized in Tables S3–S5.

Simulation of limiting metabolite transport rates indicated differences between cold-sensitive and cold-tolerant accessions (Fig. 7). While C24 did not show significant modulation of transport rates during cold exposure (Fig. 7a), rates of sucrose transport into chloroplasts and hexose transport into the vacuole were elevated in Rsch (Fig. 7b) and, to an even larger extend, in Te (Fig. 7c). Additionally, rates of sucrose transport between cytosol and vacuole were significantly elevated in Te as compared to Rsch and C24. In contrast to the elevated rates of metabolite transport in cold-tolerant accessions, rates of metabolite interconversion were found to be either lower than or similar to those in C24 (Fig. 8a–c). Most significant differences between C24 and Rsch occurred in rates of cytosolic sucrose cleavage and hexose phosphorylation, which were higher in cold exposed C24, while nonacclimated plants showed higher rates in Rsch (Fig. 8a,b). In Te, rates of metabolite interconversions were intermediate and showed a slight decrease during cold exposure (Fig. 8c). In all accessions, plastidial as well as vacuolar rates of sucrose cleavage were significantly lower than in the cytosol.

Figure 7.

Surface plots of simulated limiting rates of metabolite transport for Arabidopsis thaliana, C24 (a), Rsch (b) and Te (c). Values of transport rates, which are indicated by the color bar, were calculated to be minimal for the successful simulation of a metabolic steady state after 0, 1, 3 and 7 d of cold exposure. Nomenclature of transport rates refers to Fig. 1.

Figure 8.

Surface plots of simulated rates of metabolite interconversion for Arabidopsis thaliana, C24 (a), Rsch (b) and Te (c). The surface represents mean values of metabolite interconversions calculated in steady-state simulations (n = 50). Nomenclature of reaction rates refers to Fig. 1.


Arabidopsis accessions differ significantly in carbohydrate compartmentation during cold exposure

Many studies have emphasized the importance of primary carbohydrate metabolism during cold acclimation, and various regulatory instances significantly affecting this process have been identified (Strand et al., 1997, 1999; Klotke et al., 2004; Zuther et al., 2004; Lundmark et al., 2006; Nägele et al., 2011). Yet, due to the complexity of the cellular network of metabolic interactions and compartmentalization, still many aspects of metabolism and its regulation are not well understood. As already stated, compartmentalization of metabolism in eukaryotic cells significantly affects activity and function of enzymes as well as concentrations and regulatory impact of metabolites (Lunn, 2007; Klie et al., 2011). While compartmentalization of enzymes can be predicted computationally based on the presence of sequence motifs (Emanuelsson et al., 2000), the subcellular allocation of metabolites is difficult to assess because of redundant biochemical pathways, as for example the cytosolic and plastidic pathways of carbohydrate oxidation (Masakapalli et al., 2010), and various transport processes enabling rapid exchange of small compounds (Linka & Weber, 2010; Wingenter et al., 2010; Schneider et al., 2012). In the present study, compartmentalization of the central carbohydrate metabolism was studied to gain further insight into regulatory instances involved in the cold acclimation process in Arabidopsis thaliana.

Cold tolerance, expressed as LT50 values, showed significant differences before and during a 1-wk period of cold acclimation in three accessions, C24, Rsch and Te. Dynamics of cold acclimation revealed a fast increase in tolerance during the first 3 d, which is in accordance with previous findings (Nägele et al., 2011) and showed that Te from Scandinavia is the most tolerant accession at any time point of cold acclimation. Substantial redistribution of metabolites during the first 3 d in the cold occurred for sucrose and hexoses, while cytosolic increase of the raffinose concentration was delayed. In vitro experiments performed by Hincha et al. (2003) showed that sucrose and raffinose can protect large unilamellar liposomes from damage during drying, indicating a possible role in protecting the plasma membrane during freeze-induced dehydration. However, a lack of raffinose in a raffinose synthase mutant isolated from the cold-tolerant accession Col-0 did not alter cold tolerance determined as freeze–thaw stability of the plasma membrane (Zuther et al., 2004). In agreement with this, cytosolic raffinose content did not correlate with cold tolerance in the present study, again implying that the plasma membrane is not a target for protection by raffinose, as has recently been demonstrated (Knaupp et al., 2011).

Besides raffinose, cytosolic sucrose concentration was also not correlated with cold tolerance at day 7 of the cold exposure. However, the sucrose concentration had already rapidly increased during the first day of cold exposure, thus accompanying the significant decrease of the LT50 in all accessions. It has been hypothesized that cold acclimation is the result of a sequential accumulation and disappearance of several metabolites (Alberdi et al., 1993). In this sense, sucrose could serve as an early protectant of cellular membranes, but is later substituted by other compounds. In line with this, cytosolic sucrose was highest at day 1 in Rsch, when the increase in cold tolerance was maximal in this accession, while it peaked at day 3 in Te that displayed a larger gain in tolerance between day 1 and 3 than did Rsch. The very high concentrations of cytosolic sucrose at day 1 in Rsch and day 3 in Te, reaching concentrations of 45–50 mM, may also indicate saturation during cold acclimation. It might thus be speculated that cytosolic sucrose accumulates rapidly after cold exposure, serving as a transient cryoprotectant for cellular membranes at early stages of cold exposure, while later it becomes replaced by a metabolically less critical compound, for example, raffinose. At that stage, sucrose in the cytosol would serve as a substrate for the synthesis of other cryoprotectants or has a regulatory role in cold acclimation. In this context, it is interesting that raffinose accumulation follows the rise in cytosolic sucrose. As sucrose and galactinol are the substrates for raffinose synthesis taking place in the cytosol (Peterbauer & Richter, 2001), the increase of sucrose in the cytosol might be considered a trigger for raffinose production. Raffinose could then be transported across the chloroplast envelope, probably by an active transport mechanism as proposed by Schneider & Keller (2009), serving in the plastids as a cryoprotectant of the thylakoids (Knaupp et al., 2011). However, both cytosolic and plastidial sucrose content increased significantly faster than raffinose content, suggesting an additional transitory protective effect of sucrose for the thylakoids. This is difficult to test because of the simultaneous role of sucrose as substrate for raffinose synthesis. A likely reason for a substitution of sucrose by raffinose might be the reduced metabolic reactivity of raffinose and its marginal regulatory influence on primary carbon metabolism.

While an increase of plastidial sucrose and raffinose was also found in cold-acclimated cabbage (Santarius & Milde, 1977) and Arabidopsis, accession Columbia (Col-0) (Knaupp et al., 2011), a decrease of the cytosolic content has only been shown for Arabidopsis, accession Col-0 (Knaupp et al., 2011). The obvious difference in behaviour of Col-0 and the accessions investigated in the present study might be related to the longer time of cold exposure, that is, 14 vs 7 d, used by Knaupp and co-workers. This would imply that cold acclimation is either not saturated after 7 d, or that during a longer cold exposure, leaves develop that are in a metabolic state different from those that were shifted from warm to cold. The latter has clearly been demonstrated by Strand et al. (1999). However, it is difficult to judge whether 14 d in the cold would be long enough to bring about such an effect, because the rosette of the plants is a mixture of shifted and newly developed leaves. Also, a deviation of the laboratory strain Col-0 from the accessions used in our study cannot be excluded.

Steady-state simulations indicate differences in metabolic reprogramming

The rapid accumulation of sucrose and other metabolites observed in all accessions can be considered an immediate consequence of the cold-induced thermodynamic effects on enzyme activities and/or transport rates, which can be deduced from the experimentally determined maximum turnover rates. For example, vacuolar and cytosolic invertase activity dropped approximately five-fold in the cold in all accessions, while the reduction in hexokinase activity was even stronger directly after the temperature shift, and activity came back to only c. 20% of the initial value after 7 d (Fig. 4). However, sugar accumulation would also result from reduced consumption of assimilates in the cold when growth is stalled, or from an at least transiently reduced export of carbohydrates to sink organs, as predicted by modeling the impact of cold acclimation on photosynthesis and assimilate export to sink organs (Nägele et al., 2011). Although thermodynamic and growth effects of low temperature should be similar for all accessions, simulated reaction rates of steady-state models discriminated the cold-sensitive from the cold-tolerant accessions, the latter showing a more pronounced reduction predominantly of cytosolic sucrose cleavage and hexose phosphorylation. Reaction rates in the sensitive accession C24 were also reduced at 4°C, yet the reduction was not as strong as in Rsch and Te. While this proves that the cold response of the tolerant accessions is more distinct, simulations of reaction rates at the whole-cell level did not reveal whether differences in turnover rates between tolerant accessions and C24 would result in substantial differences at the metabolite level that could explain the differential cold tolerance.

This led us to implement intracellular redistribution of metabolites during cold acclimation in our model of steady-state simulations of carbohydrate metabolism in order to investigate whether metabolite dynamics at the compartment level could be more significant than at the whole-cell level. To this end we first identified the minimal rates of metabolite exchange between cellular compartments that would allow establishing a metabolic homeostasis, that is, a steady state, which would be compatible with the reduced enzyme activities as well as the modified metabolite concentrations observed in cold exposed plants. Certainly it would have been preferable to measure transport activities. However, the insufficient characterization of many of the transporters precluded this concept. We thus applied a modeling approach using stepwise increments of upper and lower bounds for the transport rate constants. This was done until a steady state of metabolite concentrations could be simulated within the experimentally determined ranges of enzyme activities given as the standard deviations of the measurements. We then compared the calculated limiting transport rates obtained for cold-sensitive and cold-tolerant accessions.

While simulated rates of metabolite transport did not change significantly during cold exposure in C24, modeling suggested substantially elevated rates of sucrose transport across the chloroplast envelope, as well as hexose transport across the tonoplast, in Rsch and even more so in Te. Additionally, sucrose transport across the tonoplast was modeled to be strongly elevated in cold-acclimated Te. As mentioned above, because knowledge on intracellular sugar transporters is still incomplete, especially regarding sucrose and raffinose import into plastids (Schneider & Keller, 2009), the results of the simulations depend on the assumption that bidirectional transport of hexoses, sucrose and raffinose is possible across the plastid envelope as well as the tonoplast. This provided, our findings indicate that intracellular metabolite transport could make a substantial contribution to sustaining metabolic homeostasis during cold acclimation. Given that only few metabolites, at the whole-cell level, could be correlated to cold tolerance of various accessions of Arabidopsis (Hannah et al., 2006), intracellular re-distribution might in addition also be important to achieve protection of cellular organelles such as plastids against damage during freeze–thaw cycles.

As stated above, raffinose synthesis is delayed when compared with the time course of cold acclimation, and thus sucrose transport into the plastids might be necessary to protect the thylakoids until raffinose accumulation catches up. A very interesting result of the simulations was that, after 3 d of cold exposure, the rate of sucrose transport across the chloroplast envelope and the tonoplast could even discriminate the two cold-tolerant accessions Rsch and Te. The latter, which at day 3 was c. 1.7°C more cold-tolerant than Rsch, not only displayed higher transport rates, but also had significantly higher sucrose concentrations in the cytosol and vacuole, while hexoses predominantly accumulated in Rsch. Our data indicate that an increased transport of sucrose from cytosol into the vacuole allowed for more sucrose to accumulate in mesophyll cells of Te, probably because of the low vacuolar invertase activity. In contrast, high cytosolic hexose concentrations were observed in Rsch at day 3 of the cold exposure, which were again reduced at day 7, when the cold tolerance of Rsch drew near that of Te. It is tempting to speculate that the high cytosolic hexose concentration in Rsch at day 3 could be related to the low gain in cold tolerance between days 1 and 3 in this accession. High hexose concentrations may cause feedback-inhibition of cytosolic invertase, thus inhibiting sucrose cyling, which is important for buffering primary carbohydrate metabolism against environmental disturbances (Nägele et al., 2010). In fact, cytoslic sucrose cleavage was calculated to be higher in Te than in Rsch and, thus, higher transport rates for hexoses into the vacuole must be claimed for Te. In this respect, it is interesting that the tonoplast monosaccharide transporter, TMT1, which re-directs glucose and fructose from the cytsol into the vacuole, was demonstrated to be strongly induced at low temperature in the cold-tolerant accession Col-0 (Wormit et al., 2006). This may, in fact, point to a major difference in the cold-acclimation strategies of Rsch and Te: by shuffling sucrose into the vacuole, Te reduces accumulation of hexoses in the cytosol which in turn allows for a sustained cytosolic sucrose cycling and a stabilization of primary carbohydrate metabolism against environmental changes (Nägele et al., 2010). An exchange of sucrose between cytosol and vacuole may thus be important for supporting regulation of cytosolic reactions of primary carbon fixation.

Of course, a final proof of the concept of limiting transport rates as applied in this work would rely on measuring transport activities in planta. However, under conditions of insufficient knowledge of transporters involved, the approach of mathematically simulating transport processes at limiting transporter activities may help to understand how cold-induced processes taking place at the intracellular membranes could be involved in cold acclimation. The simulations plausibly demonstrated the importance of intracellular re-distribution of carbohydrates and indicated that, besides raffinose transport into the plastids, which has already been demonstrated experimentally (Schneider & Keller, 2009), transport of sucrose into the plastids and the vacuole as well as hexose transport across the tonoplast might become limiting for cold acclimation in Arabidopsis thaliana. Taking into account that the volumes of compartments may change during the process of cold acclimation, particularly the cytosolic and vacuolar volume (Strand et al., 1999), predicted minimum rates of transport, as well as metabolite interconversion for steady-state simulations, would vary to a corresponding extent. Yet, as long as the effect on compartment volumes can be assumed to be systematic across all considered accessions, the inference derived from our model simulations would hold although absolute values of the minimal transport rates may deviate. This would not be the case if volume changes were different for the different accessions. However, investigations by us (Hannah et al., 2006) and others (Cook et al., 2004) have shown that global metabolite changes during cold acclimation, which would result from changes in cellular water content, are very similar for cold-sensitive as well as cold-tolerant accessions.

Taken together, the present study reveals a significant elevation of cytosolic, plastidial and vacuolar sucrose concentrations as an early response to cold exposure, being stronger in the cold-tolerant accessions Rsch and Te as compared to the sensitive accession C24. While C24 did not show any further alterations in subcellular carbohydrate allocation, the tolerant accessions displayed an exchange of raffinose for sucrose especially in the plastids. The substitution of sucrose by raffinose is in agreement with the protective role of raffinose for the thylakoids and may indicate the need for a replacement of sucrose by a metabolically less reactive compound in tolerant accessions. Steady-state simulations of carbohydrate metabolism suggested that intracellular carbohydrate transport processes are indispensable in establishing a low temperature compatible carbohydrate homeostasis in tolerant accessions, while in the cold-sensitive accession C24 this adjustment appears less pronounced.


We would like to thank Annika Allinger for excellent plant cultivation. We also thank the members of the Department of Plant Biotechnology at the University of Stuttgart for fruitful discussions and constructive advices, and the referees for their constructive advice during the reviewing process for this article.