Fluxes in central carbohydrate metabolism of source leaves in a fructan-storing C₃ grass: rapid turnover and futile cycling of sucrose in continuous light under contrasted nitrogen nutrition status

Abstract

This work assessed the central carbohydrate metabolism of actively photosynthesizing leaf blades of a C3 grass (Lolium perenne L.). The study used dynamic ¹³C labelling of plants growing in continuous light with contrasting supplies of nitrogen (‘low N’ and ‘high N’) and mathematical analysis of the tracer data with a four-pool compartmental model to estimate rates of: (i) sucrose synthesis from current assimilation; (ii) sucrose export/use; (iii) sucrose hydrolysis (to glucose and fructose) and resynthesis; and (iv) fructan synthesis and sucrose resynthesis from fructan metabolism. The contents of sucrose, fructan, glucose, and fructose were almost constant in both treatments. Labelling demonstrated that all carbohydrate pools were turned over. This indicated a system in metabolic steady state with equal rates of synthesis and degradation/consumption of the individual pools. Fructan content was enhanced by nitrogen deficiency (55 and 26% of dry mass at low and high N, respectively). Sucrose content was lower in nitrogen-deficient leaves (2.7 versus 6.7%). Glucose and fructose contents were always low (<1.5%). Interconversions between sucrose, glucose, and fructose were rapid (with half-lives of individual pools ranging between 0.3 and 0.8 h). Futile cycling of sucrose through sucrose hydrolysis (67 and 56% of sucrose at low and high N, respectively) and fructan metabolism (19 and 20%, respectively) was substantial but seemed to have no detrimental effect on the relative growth rate and carbon-use efficiency of these plants. The main effect of nitrogen deficiency on carbohydrate metabolism was to increase the half-life of the fructan pool from 27 to 62 h and to effectively double its size.

¹³C, central carbohydrate metabolism, carbon fluxes, compartmental analysis, fructan, futile cycling, sucrose, store, tracer, turnover

Introduction

This study was concerned with sucrose as a key branch point in the central carbohydrate metabolism of photosynthetically active, exporting leaf blades of perennial ryegrass (Lolium perenne L.), a fructan-storing species. In particular, the fluxes of sucrose in distinct, interdependent processes were assessed: fructan synthesis, sucrose hydrolysis, sucrose storage in vacuoles, sucrose export/use by sinks, and sucrose synthesis from current assimilation and from the products of its hydrolysis or from fructan metabolism (sucrose cycling). This analysis was performed with plants growing in continuous light with either a high or a low nitrogen fertilizer supply rate.

Fructans are the main long-term storage carbohydrate in the vegetative parts of many important economic plant species, including wheat, barley, rye, oat, and C3 forage grasses (Pollock and Cairns, 1991; Vijn and Smeekens, 1999). Fructans serve vital roles in seed filling and recovery from biotic and abiotic stresses, providing carbohydrate substrates for growth and respiration when provision by current assimilation is less than demand by sinks (Schnyder, 1993; Morvan-Bertrand et al., 2001; Yang and Zhang, 2006; Valluru and Van den Ende, 2008). Fructans are stored in vacuoles and are synthesized from sucrose by fructosyltransferases (FTs) and degraded by fructan exohydrolases (FEHs) (Wagner et al., 1983; Frehner et al., 1984; Darwen and John, 1989). The transfer of the fructosyl residue of sucrose to acceptor sucrose or fructan molecules is exergonic and essentially irreversible (Wagner and Wiemken, 1987; Duchateau et al., 1995). Fructans of L. perenne are structurally diverse, with this diversity resulting from the cooperation of several FTs (Pavis et al., 2001a,b; Chalmers et al., 2005; Lasseur et al., 2006, 2011). Degradation by FEHs involves the successive cleavage of terminal fructose residues of fructans. Most, if not all, of the released fructose is used in resynthesis of sucrose for export (Pollock and Cairns, 1991). However, resynthesized sucrose may be utilized again in fructan synthesis if fructan synthesis and degradation occur simultaneously inside the same cell. Moreover, the glucose released during fructan synthesis is probably used in resynthesis of sucrose for fructan synthesis. Evidence for such futile cycling of carbon through the fructan store has been put forward repeatedly (e.g. Pollock, 1982; Wagner et al., 1986; Simpson and Bonnett, 1993; Morvan-Bertrand et al., 1999, Amiard et al., 2003), but its quantitative importance for sucrose partitioning (and cycling) has not been assessed. As fructan turnover in leaf blades can be high (Borland and Farrar, 1988; Farrar, 1989), sucrose cycling could be substantial.

Carbohydrates can also be stored in the form of sucrose in vacuoles (Kaiser et al., 1982). In L. perenne and other fructan-storing grasses, vacuolar sucrose may even replace transient starch as the diurnal storage compound in leaves or other photosynthetic organs/tissues (Wilson and Bailey, 1971, Riens et al., 1994; Isopp et al., 2000; Gebbing, 2003). This sucrose store serves night-time export of autotrophic tissue. Fructan and sucrose storage follow divergent distribution patterns between organs and tissues: sucrose storage predominates in the photosynthetically active organs, and fructan storage is most active in heterotrophic tissue (Volenec, 1986; Schnyder and Nelson, 1987; Pollock and Cairns, 1991; Riens et al., 1994; Morvan-Bertrand et al., 2001; Gebbing, 2003). This contrast could result from a fundamental trade-off, a biochemical switch, between sucrose utilization in vacuolar sucrose storage and fructan synthesis in the same compartment. Yet, according to Martinoia et al. (2007), plants can have several types of vacuoles with distinct functions, even within one cell, theoretically providing a basis for the coexistence of vacuolar sucrose storage and fructan metabolism in different compartments in the same cell.

Hydrolysis by invertase is another, potentially strong, sink for sucrose in source leaves (Huber, 1989). Guerrand et al. (1996) found very high extractable soluble acid invertase activity in mature photosynthesizing leaf blades of L. perenne induced to accumulate fructan. Similarly, mature photosynthetically active leaves of the fructan-storing grass Lolium temulentum contain intracellular invertase activity that is sufficient to hydrolyse the total sucrose content of the leaf in 0.5 h (Kingston-Smith et al., 1999). Vargas et al. (2007) found an upregulation of alkaline invertase in wheat leaves following exposure to cold or osmotic stress, conditions that are also conducive to fructan accumulation. Notably, some FEHs are also capable of hydrolysing sucrose (Kingston-Smith et al., 1999; Tamura et al., 2011). However, it is often not clear how much of the tissue sucrose is accessible by invertase in source leaves, as different forms of invertase are distributed between cellular compartments, including apoplast, cytoplasm, and vacuole (Kingston-Smith et al., 1999; Cairns and Gallagher, 2004; Vargas et al., 2007; Vargas and Salerno, 2010), and the activity may be regulated (Roitsch and González, 2004). In particular, Cairns and Gallagher (2004) found no sucrose hydrolysis in treatments blocking sucrose resynthesis in excised, pre-illuminated and non-fructan-containing leaves of L. temulentum. From this, they suggested that soluble acid invertase and sucrose did not share the same compartment in these leaves.

Nitrogen nutrition is known to affect carbohydrate metabolism in fundamental ways (Stitt and Krapp, 1999). Nitrogen deficiency decreases sink strength and growth rate (Kavanová et al., 2008), increases carbohydrate accumulation in vegetative tissue (Lehmeier et al., 2010a), and increases the expression of carbohydrate storage-related genes (Ruuska et al., 2008) and their activities (Wang and Tillberg, 1996; Wang et al., 2000; Morcuende et al., 2004). From these relationships, one might expect that nitrogen nutrition will also affect the turnover of leaf carbohydrate pools and cycling of sucrose through fructan and other stores, but, to the best of our knowledge, these effects have not been explored.

Accordingly, this work addressed the following question: how does the nitrogen fertilizer supply affect central carbohydrate metabolism in source leaves of a fructan-storing species, L. perenne, in continuous light? In particular, how does it affect: (i) sucrose synthesis from current assimilation; (ii) sucrose export to heterotrophic tissue; (iii) sucrose hydrolysis and resynthesis; and (iv) fructan synthesis and sucrose resynthesis from fructan metabolism products? Lastly, is vacuolar sucrose storage disabled/suppressed by continuous light, as the requirement for buffering day/night cycles in assimilation would not exist in continuous light? In addressing these questions, this study also investigated the possible implication of vacuolar compartmentation on sucrose cycling, an issue recently raised and discussed by Kruger et al. (2007). The present analysis employed dynamic labelling (as described by Ratcliffe and Shachar-Hill, 2006) with ¹³CO₂/¹²CO₂ over intervals of 1 h to 8 d (Schnyder et al., 2003), measurement of the quantity and tracer time course in sucrose, fructan, glucose, and fructose (Morvan-Bertrand et al., 1999; Gebbing and Schnyder, 2001), and compartmental modelling using similar principles to those used previously to characterize the substrate supply system of leaf growth and respiration in the same species (Lattanzi et al., 2005; Lehmeier et al., 2008, 2010a,b).

Materials and methods

Plant material and growth conditions

Details of plant material and growth conditions for the present experiments have been described previously (Lehmeier et al., 2010b). Briefly, the experiment was performed in four plant growth chambers (Conviron E15, Winnipeg, Canada). Plants were grown singly in pots (350 mm high, 50 mm diameter) filled with washed quartz sand and arranged at a density of 378 m⁻².

All plants were grown in continuous light, with 275 μmol m⁻² s⁻¹ photosynthetic photon flux density at canopy height. Relative humidity was controlled at 85% and temperature at 20 °C at the level of the leaf growth zone. Every 3 h, the plants received a nutrient solution with either 1.0 mM nitrate (‘low N’) or 7.5 mM nitrate (‘high N’) as the sole nitrogen source.

Control of CO₂ concentration and isotope composition

The growth chambers formed part of the ¹³CO₂/¹²CO₂ labelling and gas exchange mesocosms described by Schnyder et al. (2003). Briefly, CO₂-free air and CO₂ of known carbon isotope composition (δ¹³C, in ‰, with δ¹³C=[(¹³C/¹²C_sample)/(¹³C/¹²C_{VPDB standard})–1]×1000) were mixed and supplied to the growth chambers. Each chamber had its own mixing system. The concentration of CO₂ (360 μl l⁻¹) and its δ¹³C (δ¹³C_air) in the chamber air was monitored with an infrared gas analyser (Li-6262, Li-Cor, Lincoln, NE, USA) and a continuous-flow isotope-ratio mass spectrometer (CF-IRMS, Delta Plus, Finnigan MAT, Bremen, Germany). At high N, one chamber was kept at δ¹³C_air –28.8‰ [±0.2 standard deviation (SD)] and the other at –1.7‰ (±0.2). At low N, δ¹³C_air was –3.6‰ (±0.2) in one chamber and –30.9‰ (±0.3) in the other. The stability of the isotopic composition and concentration of CO₂ (±3 μl l⁻¹ SD) in chamber air was achieved by small periodic adjustments of air flow and CO₂ concentration in the mixing systems serving the chambers. All chambers were equipped with air locks, which permitted access to plants for sampling etc. with minimal disturbance of the concentration and isotope composition of CO₂ (Lehmeier et al., 2008).

Labelling and sampling

When plants had three tillers (about 3 and 6 weeks after sowing at high and low N), labelling was initiated by randomly swapping selected individual plants between chambers with different δ¹³C_air (i.e. ¹³C-enriched CO₂→¹³C-depleted CO₂ and vice versa) within the same nitrogen fertilizer supply. At this stage, the plants in the two treatments had a very similar size, minimizing eventual effects of nutrient levels on size-dependent carbon relations. In both treatments, plants were labelled for periods of 1, 2, 3, 4, 8, or 16 h, or for 1, 2, 4, 8, or 16 d. Four replicate plants were used for each labelling interval.

At the end of the labelling period, plants were removed from the stands. In each plant, the blade (lamina) of the youngest fully expanded leaf was sampled from all mature tillers. Blade area was measured on a subsample of 16–20 replicates. Thickness was estimated as blade volume per area, with volume estimated from fresh mass (Arredondo and Schnyder, 2003). Mature tillers were defined as having at least three fully expanded leaves. Samples from one plant were combined, weighed, frozen in liquid nitrogen, freeze dried for 72 h at –80 °C, weighed again, ground to flour mesh quality with a ball mill, and stored at –30 °C until carbohydrate extraction. Non-labelled control plants were sampled in the same way in both chambers of both treatments in parallel with labelled plants. The controls were grown continuously in the same chamber, i.e. in the presence of a constant δ¹³C_air.

Carbon and nitrogen elemental analysis

Carbon and nitrogen elemental contents of bulk biomass samples were measured using the protocols described by Lehmeier et al. (2008).

Carbohydrate extraction, separation, quantification, and ¹³C analysis

Soluble carbohydrates were extracted from 100 mg of freeze-dried ground tissue by addition of 2 ml of cold (4 °C) 100 mM sodium phosphate buffer (pH 7.4). Samples were mixed by vortexing three times for 30 s each at 4 °C before centrifugation at 3200 g for 10 min at 4 °C. The pellets were extracted once again with 2 ml of 100 mM sodium phosphate buffer using the same protocol. The pellets were preserved for a later water extraction. The two supernatants were combined, 6 ml of 96% acetone was added for protein precipitation, and the tubes were incubated overnight at –80 °C. After warming, the tubes were centrifuged for 10 min at 10 000 g and the supernatants were preserved. The pellets from the sodium phosphate extraction were extracted twice with 2 ml of pure water at 60 °C for 30 min and the samples centrifuged at 3200 g for 10 min at 4 °C. The supernatants were pooled with the acetone supernatants and the extracts dried under reduced pressure and redissolved in 1 ml of water. Aliquots were passed through a column containing cation exchange resin (Dowex 50 W, H⁺-form, Sigma, St Louis, MO, USA) and anion exchange resin (Dowex 1X8, Formate form, Sigma, St Louis, MO, USA) to remove charged compounds (Smouter and Simpson, 1991). The columns were eluted with water, and the samples were again concentrated under reduced pressure and redissolved in water. Glucose, fructose, sucrose, and fructan were separated by HPLC on a cation exchange column (Sugar-PAK, 300 mm×6.5 mm, Millipore Waters, Milford, MA, USA) as given in Guerrand et al. (1996). After separation, the sucrose, fructan, glucose, and fructose were collected separately and concentrated under reduced pressure. To make sure that the sucrose was not hydrolysed by residual invertase activity during the sodium phosphate buffer extraction, relative sucrose, glucose, and fructose levels were compared with the same samples (six independent replicates) extracted using a classical ethanol/water extraction protocol. In this classical protocol, 20 mg of freeze-dried ground tissue were extracted with 80% ethanol at 80 °C for 15 min, followed by two extractions with pure water at 60 °C for 15 min. Samples were centrifuged between each extraction at 10 000 g for 10 min and the three supernatants were pooled. The pooled supernatants were dried and desalted as described above. For all samples, the two protocols (sodium phosphate buffer and ethanol/water extraction) gave the same relative proportions of sucrose, glucose, and fructose (data not shown), indicating that no sucrose hydrolysis occurred during the sodium phosphate buffer extraction.

Aliquots of the aqueous solution of separated sugars containing approximately 2 mg of carbohydrate were dried on Chromosorb, packed in tin cups and the δ¹³C of the samples was determined by continuous-flow isotope ratio mass spectrometry (Delta Plus, Finnigan MAT, Bremen, Germany) after combustion in an elemental analyser (NA 1110, Carlo Erba Instruments, Milan, Italy). Each sample was measured against a laboratory CO₂ working standard, which was calibrated previously against an IAEA secondary standard (IAEA-CH6, accuracy of calibration 0.06‰ SD). Finely ground wheat flour served as an internal laboratory standard and was run regularly every tenth sample to determine the precision of the isotope analysis; the SD in δ¹³C of sample repeats was 0.15‰.

The extraction and purification scheme caused some loss of carbohydrates during the purification steps. To correct for these losses, we used a procedure that eliminate the abovementioned requirement of removing charged compounds and provided virtually full recovery (Schnyder and de Visser, 1999). For this, 10 mg of ground dry material was weighed into 2.2 ml capped Eppendorf tubes and 2 ml of water was added. The tubes were sealed immediately and transferred to a 95 °C water bath for 10 min, shaken for 30 min, and centrifuged at 10 000 g for 10 min, and the supernatants were removed with a pipette. Comparison with two additional extractions demonstrated that ≥98% of the total water-soluble carbohydrates were dissolved in the first extract. Water-soluble carbohydrates were analysed with a continuous-flow system: an aliquot of the extract was hydrolysed in 0.1 mM sulphuric acid for 25 min at 92 °C. The reducing power of the hydrolysed carbohydrates was detected photometrically at 425 nm after reduction of a potassium ferricyanide solution. Analysis of the reference sucrose, fructan, glucose, and fructose (all analytical grade from Merck, Darmstadt, Germany) yielded response factors for the individual carbohydrates, which were proportional to the amount of hexose residues present in 1 g of substrate.

The fraction of tracer in carbohydrates

Tracer incorporation in carbohydrates was followed via the time course of the fraction of unlabelled carbon (f_unlab) in sucrose, fructan, glucose, and fructose of control plants and plants transferred to and kept inside a labelling chamber for a distinct period of time (see above). The f_unlab fraction represented the fraction of carbon in carbohydrates that was derived from assimilation in the chamber of origin, prior to transfer to the labelling chamber. In the labelling chamber, f_unlab was gradually replaced by a labelled fraction (f_lab), which resulted from assimilation after the transfer (f_unlab+f_lab=1). Note that, except for the isotopic composition of CO₂ in the chambers, the conditions in the chamber of origin and in the labelling chamber were identical (see above). Details of the labelling data evaluation are given in de Visser et al. (1997) and Schnyder and de Visser (1999). The f_unlab fraction was obtained from mass balance considerations as f_unlab=(δ¹³C_s–δ¹³C_lab)/(δ¹³C_unlab–δ¹³C_lab), with δ¹³C_s designating the isotopic composition of a certain carbohydrate sample (sucrose, glucose, fructose, or fructan, as appropriate), and δ¹³C_unlab and δ¹³C_lab denoting the δ¹³C of the same carbohydrate extracted from control plants grown continuously in the chamber of origin (unlabelled) and in the labelling chamber, respectively. Importantly, the direction of the transfer of labelled plants (from ¹³C-enriched CO₂ to ¹³C-depleted CO₂ or vice versa) had no effect on f_unlab, as is expected for artefact-free labelling experiments and appropriate consideration of isotope discrimination effects during labelling.

Compartmental analysis of tracer time courses

The time course of f_unlab in the different carbohydrates is a reflection of their turnover by labelled carbon. This time course is a function of the metabolic pathways that include these carbohydrates, the compartmentation of the carbohydrates, the size of the different carbohydrate pools, and the magnitude of the fluxes among them. These functional relationships can be represented in compartmental models.

In compartmental modelling terminology, a ‘pool’ is defined as a population of molecules that exhibit the same proportion of labelled carbon atoms; thus, a pool represents an entity with uniform isotopic composition (Lattanzi et al., 2005, Lehmeier et al., 2008). This means that, for instance, compartmental modelling per se can not distinguish cytosolic and vacuolar sucrose pools if these pools exhibit identical tracer kinetics.

The compartmental model chosen in the present study assumed that: (i) the system is in a metabolic steady state; (ii) fluxes obey first-order kinetics; and (iii) pools are homogeneous and well mixed, so that the flux(es) out of and into the pool, as well as the pool’s size and half-life, are constant. Growth of plants occurred in a constant environment with continuous light, which was conducive to steady conditions of growth and related metabolism. Isotope discrimination in carbohydrate interconversions was accounted for by collection and analysis of control plants (see above). The fact that labelling with ¹³C-depleted and ¹³C-enriched CO₂ produced identical results proved that this approach prevented artefacts related to isotope discrimination. For other assumptions of the compartmental model, such as the constancy of carbohydrate pool sizes with labelling duration and plant age, see Results and Discussion.

The biologically realistic model that best reflected the tracer data in our experiments (‘constrained four-pool model’ in Fig. 1; see also Supplemental Table S2 in JXB online) is described by the following set of differential equations (implemented in the software ModelMaker; Cherwell Scientific, UK). The change in the amount of unlabelled carbon in pool i with time was given by the sum of fluxes of unlabelled carbon into that pool minus the fluxes of unlabelled carbon leaving the pool:

(1)

(2)

(3)

(4)

with t the time since the onset of labelling (labelling duration). f_{unlab_Qi}(t) is the fraction of unlabelled carbon in pool i at time t, F_In equals the flux into the system, and f_{unlab_In} the fraction of unlabelled carbon in that flux [f_{unlab_In}(t)=1 for t <0 and f_{unlab_In}(t)=0 for t ≥0]. Q_i is the size of pool i, and k_ij the rate constant for the flux from pool i to pool j (with j=0 denoting the flux leaving the system).

Fig. 1.

Four-pool compartmental model of central carbohydrate metabolism in source leaves of L. perenne. Suc, sucrose; Glc, glucose; Fru, fructose; F_In, tracer flux into the system. Q_i represents the size of the carbohydrate pool I, and k_ij is the rate constant for the flux from pool i to pool j. Thus, k₁₀ denotes the export of sucrose from the system, k₁₂ denotes the FT-catalysed transfer of the fructosyl residue of sucrose to a fructan (or sucrose) acceptor molecule, k₁₃ denotes glucose production by FT plus invertase(-like) activities, k₁₄ denotes fructose production by invertase(-like) activity, k₂₄ denotes cleavage of fructose from fructan by FEH, k₃₁ denotes glucose use in sucrose resynthesis, and k₄₁ denotes fructose use in sucrose resynthesis. For mathematical details of the model, including the differential equations describing the fluxes, see Materials and methods.

For a system in metabolic steady state, pool sizes are constant with time, and described by:

(5)

(6)

(7)

(8)

The half-lives reflect the turnover of pools. In a metabolic steady state, they are determined by the rate constants for efflux alone. Thus, the half-life (T_1/2) of the different pools were obtained as:

(9)

(10)

(11)

(12)

The rate constants were optimized such that f_{unlab_Qi}(t) derived from Equations (1)–(4) fitted the observed tracer kinetics of sucrose (Q₁), fructan (Q₂), glucose (Q₃), and fructose (Q₄) best. This was done under the constraints that: (i) the pool size ratios Q₁/Q₂ were given by the observed ratios of carbohydrate content; and (ii) identical amounts of glucose and fructose were utilized for sucrose regeneration, conserving the stoichiometric ratio of 1:1. Interconversions between glucose and fructose were not considered (see Compartmental modelling of carbohydrate metabolism in Results). The mean residence time (τ) of carbon in the system was obtained as τ=(Q₁ + Q₂ + Q₃ + Q₄)/F₁₀.

A number of biologically plausible, alternative variants of pool models were also implemented in ModelMaker using the same principles and used to fit the observed tracer kinetics by optimizing half-lives of the pools.

Results

Leaf parameters

Nitrogen fertilizer supply affected key leaf parameters in typical ways (Table 1). Plants were sampled at equal size, and thus total dry mass per leaf was similar for the low and high N supply. Nitrogen fertilizer greatly increased leaf area and nitrogen content per unit dry mass, but it decreased leaf thickness and tissue density (i.e. dry mass per fresh mass). Nitrogen content per unit leaf area was unchanged because of the compensating effects of tissue density and leaf thickness on nitrogen content per unit dry mass.

Table 1.

Leaf parameters: area, thickness, fresh and dry mass, and nitrogen content per unit biomass and per unit leaf area

Parameter	Low N		High N
	Mean	SE	Mean	SE
Area per leaf (cm2)	2.9a	±0.3	7.8b	±0.7
Leaf thickness (mm)	0.57a	±0.04	0.30b	±0.01
Fresh mass per leaf (mg)	150a	±13	226b	±19
Dry mass per leaf (mg)	54a	±5	64a	±6
Nitrogen per dry mass (mg g−1)	9.3a	±0.6	24.5b	±1.7
Nitrogen per area (g m−2)	2.0a	±0.2	2.0a	±0.1

Plants were grown with 1 mM nitrate (low N) or 7.5 mM nitrate (high N) in the nutrient solution. Results are shown as means ±SE of 16–20 replications. Numbers in a row with different superscript letters indicate a statistically significant difference at P <0.05.

Carbohydrate concentration

Nitrogen fertilizer supply also affected carbohydrate concentration: it decreased fructan concentration to approximately one-third and it doubled sucrose concentration (Table 2). Conversely, in both treatments, the concentration of fructose and glucose was low and not affected by nitrogen fertilizer supply. Fructan was by far the largest component of the total water-soluble carbohydrates in both treatments: 91% at low N and 74% at high N. The contents (and concentrations) of the individual carbohydrates did not change during labelling and with plant age, except for the fructan concentration at high N, which was constant with labelling time but increased with plant age at a relative rate of 10% d⁻¹ (data not shown).

Table 2.

Mean concentration of glucose, fructose, sucrose and fructan in the blade (lamina) of the youngest fully expanded leaf of perennial ryegrass.

	Low N		High N
	Mean	SE	Mean	SE
Sucrose	10.6a	±1.8	19.9b	±3.2
Fructan	217a	±36	78b	±13
Glucose	4.8a	±1.3	4.5a	±0.7
Fructose	5.5a	±1.7	3.6a	±0.6

Plants were grown with 1 mM nitrate (low N) or 7.5 mM nitrate (high N) in the nutrient solution. Results are shown as mg g⁻¹ fresh mass (means ±SE) of two to seven replicates. Numbers in a row with different superscript letters indicate a statistically significant difference at P <0.05.

Tracer kinetics

The tracer kinetics observed in the individual carbohydrates are shown in Fig. 2. In each treatment, the tracer kinetics of sucrose, glucose, and fructose followed a double exponential function (r²≥0.98). Single exponential functions exhibited a systematic lack of fit, and higher-order exponential fits did not significantly improve the fit. This observation is consistent with the idea that each of these sugars was supplied by two sources with distinct labelling kinetics: rapidly labelled precursors from current photosynthesis and labelled carbon released during fructan turnover.

Fig. 2.

The fraction of unlabelled carbon (f_unlab) in fructan (filled squares), sucrose (open circles), glucose (filled triangles), and fructose (open triangles) with labelling duration at low N (upper panel) and high N supply (lower panel). The insets expand the first 24 h of the dynamic labelling. The curves represent the fits of the constrained four-pool model shown in Fig. 1. Carbohydrates were extracted from the blade of the youngest fully expanded leaf of perennial ryegrass grown with either 1 mM nitrate (low N) or 7.5 mM nitrate (high N) in the nutrient solution. Plants were labelled during normal growth in swards kept in ¹³CO₂/¹²CO₂ mesocosms. In both treatments, swards grew in continuous light with a photosynthetic photon flux density of 275 μmol m⁻² s⁻¹ at 20 °C air temperature and 85% relative humidity.

In contrast with sucrose, glucose and fructose, the tracer kinetics of fructan was well fitted by a single exponential function; a double exponential function did not improve the fit. This meant that the fructan store functioned as one well-mixed pool that was fed by a single source. These relationships were the same in both treatments, except that the tracer incorporation into the fructan pool occurred more slowly in the low N than in the high N treatment.

Compartmental modelling of carbohydrate metabolism

The compartmental model:

Carbohydrate metabolism was modelled based on knowledge of carbohydrate metabolism in source leaves of C3 grasses and the findings presented above. This dictated the following model features.

1. The system was in metabolic steady state, as was supported by the (near) constancy of pool sizes. The steady state meant that: (i) fluxes between the individual carbohydrate pools were constant; and (ii) production/synthesis rate of a given carbohydrate was balanced by an equal degradation/consumption rate; for instance, the rate of carbohydrate use in fructan synthesis equalled the rate of carbohydrate release during concurrent fructan (exo)hydrolysis/breakdown.

2. Pools were well mixed. This was supported by the high r² (r²≥0.98) of the single and double exponential fits to the tracer data. In particular, the tracer kinetics in the fructan pool showed that the tracer was almost instantaneously and homogeneously mixed within and between fructan molecules during labelling.

3. Tracer entered the modelled carbohydrate system via sucrose.

4. Sucrose was used in three types of processes: (i) export from the leaf; (ii) fructan synthesis, which involved the transfer of the fructosyl residue of a donor sucrose molecule onto a sucrose or fructan molecule, and the release of one glucose molecule, dictating a 1:1 stoichiometry of fructose transfer onto fructan and glucose production during fructan synthesis. (To reduce the complexity of the model, and to enhance its statistical robustness, we ignore the use of sucrose as the starter molecule in fructan synthesis. The bias associated with this approximation is small for fructan molecules with an average degree of polymerization of 20 in leaf blades of L. perenne.); (iii) sucrose hydrolysis process(es): the hydrolysis of sucrose to glucose and fructose. Invertase-like activity was dictated by the rapid appearance of label in fructose. This can not be attributed to fructan hydrolysis, because the tracer kinetics of fructan did not show such a fast component (Fig. 2). Equimolar amounts of fructose and glucose are produced by sucrose hydrolysis.

5. Fructan hydrolysis produces fructose. Minor amounts of glucose and sucrose are also released. The total of the latter is approximately 10% of the total degraded fructan in L. perenne [see 4(ii) above]. Glucose and sucrose release during fructan hydrolysis was ignored for the sake of model simplicity and robustness.

6. Fructan turnover in the steady state yields equimolar amounts of glucose and fructose: one glucose molecule is released during fructosyl transfer in the FT-catalysed fructan synthesis step, and one fructose is released by FEH in the concurrent degradation step.

7. All fructose and glucose produced by invertase, FT, and FEH activities are used in resynthesis of sucrose.

These features dictated a four-pool model with a distinct topology (the ‘constrained four-pool model’ in Fig. 1) as the simplest biologically meaningful representation of the central carbohydrate metabolism of a source leaf. The differential equations describing the different fluxes are given in Materials and methods. To account for metabolic steady state and stoichiometry according to points 4(ii), 4(iii) and 6 above, the model was constrained such that identical amounts of glucose and fructose were utilized for sucrose regeneration (the direct fluxes from glucose to sucrose and from fructose to sucrose were equal: k₃₁*Q₃=k₄₁*Q₄, conserving the stoichiometric ratio of 1:1). In the model, the transfer of fructosyl residues to fructan (by FT), fructose production from fructan (by FEH), and fructose production from sucrose (by invertase) are single enzyme-catalysed steps. Glucose production results from both invertase and FT activity. Resynthesis of sucrose from hexoses is catalysed by several enzymes.

Other biologically plausible four-pool models were tested. Of these, the ‘constrained four-pool model’ (shown in Fig. 1) best fitted the entire data with respect to statistical criteria and model simplicity. For instance, a ‘free four-pool model’ (which did not contain the 1:1 stoichiometric constraint for glucose and fructose use in sucrose resynthesis and allowed for fructose↔glucose interconverting reactions) yielded non-significant estimates of the rate constants for the interconverting fluxes (Supplementary Table S2). This indicated that model realism, as reflected in the tracer data, did not require these interconverting reactions. With respect to the Bayesian information criterion (Motulsky and Christopoulos, 2004), the (stoichiometrically) ‘constrained four-pool model’ performed slightly better than the analogous stoichiometrically unconstrained four-pool model (SupplementaryTable S2), as it yielded a similar fit quality (r² and residual sum of squares) with fewer fitting parameters. As the constrained four-pool model is biologically more realistic (sucrose is indeed resynthesized from equimolar amounts of fructose and glucose) and accords with steady-state conditions (see points 6 and 7 above), we prefer this model as the most parsimonious reflection of system properties given by the tracer kinetics. In addition, the fact that the 1:1 stoichiometric constraint on glucose and fructose formation from sucrose (hydrolysis) fitted the data well may be taken to indicate that sucrose synthase was probably not active in these source tissues. A five-pool model derived from the constrained four-pool model, which separated sucrose into cytosolic and vacuolar compartments, did not provide significant parameter estimates for the two compartments. Thus, our tracer data were unable to resolve a cytosolic and a vacuolar sucrose pool. For this reason, and arguments provided in the Discussion, we did not explore further any aspects of compartmentation.

Results of the constrained four-pool model:

For each nitrogen treatment, the above model optimized three rate constants (k₁₀, k₁₂, and k₁₄) such that the f_unlab kinetics fitted the carbohydrate data (a total of 44 data points, 11 data points in each of the four carbohydrate pools) best (minimum least squares). The optimization yielded a fit between the model and the data with r²=0.97 in both treatments.

The relative uncertainty (SE/estimate) for k₁₀ was 5–6% in both N treatments, 11–12% for k₁₂, and 35–42% for k₁₄ (Supplementary Table S2). These uncertainties translated into similar errors in the corresponding fluxes (F₁₀, F_12, and F₁₄), as the carbohydrate contents exhibited only small variability (Table 2).

The half-life of sucrose was short in both treatments but significantly shorter at low N than at high N (0.3 versus 0.8 h; P <0.05) (Table 3). The half-lives of glucose and fructose were short (0.3–0.4 h) and did not differ between the two hexoses and between the treatments (P >0.05). Conversely, the fructan pool had a very long half-life. Beyond that, its half-life at low N was 2.3 times that at high N (62 versus 27 h, P <0.05).

Table 3.

Parameters of the constrained four-pool compartmental model of central carbohydrate metabolism in the blade of the youngest fully expanded leaf of L. perenne

	Low N		High N
	Mean	SE	Mean	SE
Half-life
Sucrose (T1/2 Q1, h)	0.30a	±0.09	0.75b	±0.17
Fructan (T1/2 Q2, h)	62a	±7	27b	±3
Glucose (T1/2 Q3, h)	0.27a	±0.09	0.39a	±0.10
Fructose (T1/2 Q4, h)	0.34a	±0.11	0.31a	±0.08
Flux
F10 (g m−2 h−1)	0.99a	±0.11	0.73b	±0.09
F12 (g m−2 h−1)	1.25a	±0.18	0.59b	±0.09
F13 (g m−2 h−1)	5.8a	±2.0	2.3a	±0.6
F14 (g m−2 h−1)	4.5a	±1.9	1.7a	±0.6
F24 (g m−2 h−1)	1.25a	±0.17	0.59b	±0.09
F31 (g m−2 h−1)	5.8a	±2.4	2.3a	±0.6
F41 (g m−2 h−1)	5.8a	±2.4	2.3a	±0.6
Sucrose partitioning*
To export	0.15a	±0.04	0.24a	±0.05
To fructan synthesis	0.19a	±0.06	0.20a	±0.05
To hydrolysis	0.67a	±0.34	0.56a	±0.23
Residence time of CHO in the system (h)
	124a	±7	42b	±6
Goodness of model fit (r2)	0.97		0.97

Parameters were optimized as described in Materials and methods, based on the tracer kinetics data shown in Fig. 2 (0–384 h at low N and 0–192 h at high N). Fluxes are expressed as g carbohydrate m⁻² leaf area h⁻¹, and calculated by the equations given in Materials and methods, using the carbohydrate contents per unit leaf area as shown in Supplementary Table S1 (available in JXB online) and the rate constants of the constrained four-pool model (Supplementary Table S2). The results of the optimized rate constants and the corresponding fluxes are given in bold. Numbers in a row with different superscript letters indicate a statistically significant difference at P <0.05.

*

Fraction of all fluxes leaving the sucrose pool.

The simulation also demonstrated that the two phases in the tracer kinetics of sucrose, glucose, and fructose originated from the feeding by two sources, current photosynthesis (i.e. tracer uptake), and carbohydrate recycling from fructan. Implicitly, this meant that the second (slow) phase observed in our tracer data was not determined by sucrose compartmentation (such as distribution of these sugars between the cytosol and vacuole) but resulted from fructan metabolism. Together with the short half-life of sucrose and the results of the five-pool model, the data indicated that the bulk of the sucrose must have resided in the cytosol.

The export flux of sucrose (F₁₀, expressed as g m⁻² h⁻¹) was slightly greater at low N than at high N (Table 3). However, the relationship between the treatments changed when fluxes were expressed on a different basis, that is, when the ‘per unit leaf area’ basis was replaced by ‘per leaf’, ‘per fresh mass’, or ‘per dry mass’ (compare the fluxes in Table 3 with the leaf characteristics in Table 1). This was true for all fluxes in the systems. For instance, the rate of sucrose export per unit leaf dry mass (g g⁻¹ h⁻¹) was significantly higher at high N relative to low N. This highlighted a problem that exists in comparisons of physiological characteristics between treatments (and possibly species/varieties and mutants). The complication did not exist when fluxes were expressed as a fraction of total sucrose use (partitioning) or total sucrose regeneration (Table 3).

Partitioning of sucrose efflux between export (approx. 20%), fructan synthesis (approx. 20%) and hydrolysis (approx. 60%) did not differ between the treatments (Table 3). Implicitly, the fractional contributions of fructan and sucrose hydrolysates to recycling of sucrose did not differ. On average for the two treatments, the rate of sucrose recycling from the products of sucrose hydrolysis was three times that of fructose release in fructan breakdown. However, the residence time of carbohydrates in the system at low N was 3.0 times that at high N. This effect was primarily due to the much longer half-life of the fructan pool at low N.

Discussion

On the approach

This analysis provided evidence of very substantial sucrose cycling related to fructan turnover and sucrose hydrolysis by invertase-like activity in source leaves of L. perenne in continuous light, and a strong effect of nitrogen status on the half-life of the fructan pool in these leaves. By contrast, it founds no evidence for substantial (vacuolar) storage of sucrose or hexoses. These interpretations derive from dynamic ¹³CO₂/¹²CO₂ labelling over intervals of 1 h to 16 d, quantification of tracer kinetics in sucrose, glucose, fructose, and fructan, and compartmental modelling of the tracer data. The compartmental model (Fig. 1) reflected the key features of central carbohydrate metabolism in source leaves of a fructan-storing species. Except for sucrose resynthesis from hexoses, which is catalysed in several enzymatic steps, and glucose production resulting from both invertase and FT activities, the different fluxes in the system represent single enzyme-catalysed reactions (considering different FTs as a single enzymatic activity insofar as they all transfer fructosyl residues into the fructan pool).

Simplifying approximations underlying the analysis had little (if any) effect on the conclusions of the work. By and large, the system was in a metabolic steady state. The only exception to this was the fructan pool at high N, which increased at approximately 10% per day; that is, the rate of fructosyl transfer into fructan would have been approximately 20% higher than the concurrent rate of fructose cleavage from fructan. This could introduce a relative error in fructan turnover-related sucrose cycling of approximately 20%. This bias would not alter conclusions regarding the relative importance of fructan turnover for sucrose cycling. Also, the bias did not occur in the other N treatment. In the model, we did not consider uses of hexoses in reactions other than sucrose resynthesis in these fully developed, actively photosynthesizing leaves (similar to Nägele et al., 2010). Although there may have been some consumption in metabolism, this would have a relatively small effect on the estimate of cycling. Furthermore, the conclusions of the work are not altered by the stoichiometric constraints in the final four-pool model proposed here (Fig. 1), or by neglecting interconverting reactions between the two hexose pools (the constrained four-pool model did not consider exchange fluxes between the fructose and glucose pool) and the eventual presence of sucrose synthase activity, or by neglecting intracellular compartmentation.

Suppression of significant vacuolar sucrose storage in continuous light

All evidence from the present work suggests that the bulk of the sucrose must have been contained in the cytosol. In both treatments there was no evidence for significant vacuolar sucrose storage. If a substantial amount of sucrose was stored in the vacuole, then the tracer data should have revealed it by tri-phasic tracer kinetics, where the three phases represented sucrose sourced by: (i) current assimilation; (ii) resynthesis from fructan breakdown; and (iii) storage in the vacuole. Phases (i) and (ii) were clearly present, while phase (iii) was not. Rapid consumption of vacuolar sucrose in fructan synthesis may have depleted the vacuolar pool. In addition, sucrose hydrolysis (inside or outside the vacuole) may have limited its accumulation in the vacuole. Where it occurs, vacuolar sucrose storage (in sucrose- and fructan-storing species) is typically associated with elevated tissue sucrose contents and a half-life of the vacuolar pool in the order of 1 d (Farrar, 1989). This was not evident in the present work. Huber (1989) compared sucrose storage and acid invertase activity in mature source leaves of a range of species, and concluded that absence of vacuolar sucrose accumulation was related to high (presumably vacuolar) acid invertase activity. The absence of significant vacuolar sucrose storage and the presence of a high concentration of fructans may be related to the substrate function of sucrose in fructan synthesis, and a biochemical switch between fructan and sucrose storage.

Vacuolar storage of hexoses also seemed minor. Vacuolar hexoses must have included glucose liberated during fructan synthesis and fructose released during fructan breakdown. A long residence time of these sugars in the vacuole would have increased the concentration of these sugars in the tissue and increased the half-life of the respective pools. Such an effect was not observed; the half-life of fructose and glucose was very short, forcing the conclusion that these sugars were rapidly exported from the vacuole.

The absence of significant vacuolar sucrose and hexose storage was indicated by the results of both treatments, showing that the feature was independent of nitrogen status. By contrast, continuous light may well have been a causal factor. Source leaves (and other photosynthesizing tissue) maintained in day/night cycles often show higher sucrose concentrations and diurnal variations, which are typically interpreted in terms of diurnal vacuolar storage (Jenner and Rathjen, 1972; Riens et al., 1994; Gebbing and Schnyder, 2001; Gebbing, 2003). The same organs store very little fructan in day/night regimes, again pointing to a biochemical trade-off between vacuolar sucrose and fructan storage.

Sucrose cycling

Only about 20% of sucrose synthesis originated from current assimilation, where ‘current assimilation’ means the direct route CO₂→triose-P→sugar-P→sucrose. In a steady state, this corresponds to the export flux of sucrose. Approximately 80% derived from cycling – often referred to as ‘futile’ cycling – of sucrose within source leaves before export. The smaller part (approx. 20%) of the cycling was connected to fructan turnover and the greater part (approx. 60%) was related to recovery of glucose and fructose produced by sucrose hydrolysing and FT activities. These relationships were very similar in the two treatments.

These rates of sucrose cycling were markedly higher than estimates of others with different leaf (or cotyledon) systems. Using simulations with a mathematical model of central carbohydrate metabolism, and data of diurnal variations of CO₂ exchange and carbohydrates, Nägele et al. (2010) estimated a recycling flux of 10–25% of the total sucrose synthesis in source leaves of wild-type Arabidopsis during the light period. Huber (1989) estimated a recycling rate of 17–30%, based on the relationship between hexose accumulation rate in girdled soybean and tobacco leaves and export rates of ungirdled leaves. On the basis of ¹⁴C pulse–chase experiments and balance sheets of carbohydrates, similar recycling was estimated by Geigenberger and Stitt (1991) for cotyledons of intact Ricinus communis seedlings. Significant cycling has also been reported for many heterotrophic systems (see discussion by Kruger et al., 2007). Conversely, Cairns and Gallagher (2004) found no evidence for sucrose cycling in excised leaves of L. temulentum in which sucrose resynthesis was abolished by anoxia or darkness or by treatments with mannose or vanadate. However, these leaves were in a very different metabolic state compared with the present work as they were quickly accumulating sucrose before treatments and did not contain fructan.

Taken together, our results and those of Cairns and Gallagher (2004) suggest that sucrose futile cycle occurrence and magnitude can be highly variable. The mode of storage (fructan versus starch) could be one reason for the high cycling observed in the present study compared with those in the abovementioned studies with starch- or sucrose-storing species or fructan-storing species under a condition where they do not accumulate fructans. However, cycling related to fructan turnover was much smaller than that related to recovery of hexoses produced by sucrose hydrolysis. This was true despite the high content and continuous turnover of the fructan store. One may wonder if the substantial cycling was linked to some specific effect of continuous light on carbohydrate metabolism of source leaves (see below). For instance, Geigenberger and Stitt (1991) observed a strong stimulation of cycling in detached (relative to intact) cotyledons of R. communis, suggesting that sink limitation enhanced cycling in the source tissue. In the present continuous-light study, leaf carbohydrate concentration was very high, suggesting that sink limitation may have been a factor. In addition, cycling tended to be somewhat higher at low N than at high N, further lending support to a role for sink limitation. Under conditions of reduced sink demand, sucrose futile cycling might therefore increase to stabilize the sucrose pool in source organs.

The costs of cycling seem to be relatively small: 1 ATP plus 0.5 inorganic pyrophosphates per hexose molecule are used if sucrose is hydrolysed by alkaline invertase (Dancer et al., 1990). This compares with a yield of 30–36 ATP per hexose molecule respired (Dancer et al, 1990; Geigenberger and Stitt, 1991). An estimation of the eventual extra costs of ‘continuous-light’-related cycling can be made via a comparison with similar plants growing in a more ‘normal’ day/night regime under otherwise identical conditions. Lehmeier et al. (2010b) compared the same cultivar of L. perenne growing in a 16 h day/8 h night regime and in continuous light, with the same daily photosynthetic photon flux density and the same environmental conditions and nutrient solution in both regimes. They found no difference in photosynthetic carbon use efficiency and specific growth rate of plants between the light treatments. Actually, specific respiration rate per unit structural dry mass (total plant mass minus water-soluble carbohydrates) was practically the same for plants in continuous light and in the day/night regime. On this basis, the eventual extra costs of cycling in continuous light might be compensated by beneficial effects of cycling that remain to be identified.

A substantial sucrose futile cycle implies a high flow through hexokinase. There are several studies in favour of a negative impact of a high flow through hexokinase on the expression of photosynthetically relevant genes (Krapp et al., 1993; Moore et al., 1998) leading to a decrease in photosynthesis activity. Under continuous light conditions, this downregulation of photosynthesis (to reach a level similar to that in ‘normal’ plants; Lehmeier et al., 2010b) might be necessary to adjust the source offer to the sink demand at the whole-plant level. An alternative beneficial aspect of sucrose futile cycling might be related to the protection of leaves against oxidative stress. In general, sucrose futile cycling can involve soluble acid invertase or alkaline invertase. Antisense suppression of soluble acid invertase in muskmelon (Cucumis melo) has revealed destructive effects on chloroplast ultrastructure (Yu et al., 2008) that point to a possible involvement of sucrose cycling in preventing reactive oxygen species accumulation in chloroplasts. Similarly, the upregulation of alkaline invertase in wheat leaves following exposure to cold or osmotic stress conditions (Vargas et al., 2007) could also be interpreted as a mechanism against oxidative stress, as many forms of environmental stress including low temperatures are associated with oxidative stress (Prasad et al., 1994). The very substantial sucrose futile cycle that occurs in fructan-storing leaves under continuous light might then be essential to regulate the photosynthetic activity (putatively via a hexokinase sensing mechanism) and to protect leaf structures against oxidative stress. It is then feasible that the discrepancy between the findings of Cairns and Gallagher (2004) in L. temulentum leaves and the present work on L. perenne leaves is related to the contrasting environmental scenarios used in the two studies: fructan-free, excised leaves from plants pre-adapted to low light in day/night cycles with a short (8 h) photoperiod (Cairns and Gallagher, 2004) versus intact, fructan-storing leaves from plants growing in continuous light (present work). Clearly, more work is needed to unravel the physiological role of sucrose futile cycling.

Nitrogen deficiency slows fructan turnover

Nitrogen nutrition had a strong effect on one particular aspect of carbohydrate metabolism: deficiency increased the half-life of the fructan pool by 2.3-fold relative to high N and increased fructan content (26.4% of dry mass at low N and 55% at high N). This was in contrast to the lack of nitrogen deficiency effects on most aspects of central carbohydrate metabolism of source leaves: half-lives and contents of hexoses were very similar, and sucrose was partitioned quite equally in the two treatments. Thus, the control of turnover was the dominant factor determining the dry matter-based fructan content of these source leaves.

The actual mechanism underlying this nitrogen response is not well understood. Interestingly, partitioning of sucrose to fructan was not related to sucrose concentration: sucrose concentration per unit fresh weight was two times higher at high N, but partitioning to fructan synthesis was not increased. This may indicate a relative limitation by FT activity at high N. By contrast, the lower turnover at low N could be due to greater limitation by FEH. To the best of our knowledge, the effects of nitrogen nutrition on FTs and FEH in source leaves have only been studied by Wang and Tillberg (1996) and Wang et al. (2000). These studies followed FEH and FT activity under transient conditions, following imposition of nitrogen starvation or resupply. Starvation was followed by fructan accumulation and a concurrent increase in FT activity (sucrose:sucrose fructosyltransferase), and resupply led to mobilization concurrent with a decrease in the activity of FTs. In both cases, in vitro FEH activity did not change appreciably.

In conclusion, this work proves the usefulness of dynamic ¹³CO₂ labelling of plants in conjunction with analysis of tracer incorporation in major carbohydrate fractions and compartmental modelling to unravel and quantify fundamental features of central carbohydrate metabolism of photosynthetically active tissue, particularly partitioning of fluxes at branch points and cycling issues. Notably, the approach permits studies of central carbohydrate metabolism with intact plants growing in undisturbed conditions, providing quantitative data of the involvement of stores in metabolism (Lattanzi et al., 2005; Lehmeier et al., 2008, 2010a,b). Importantly, up to 16 d of dynamic labelling were necessary to accurately characterize the different carbon sources of central metabolites such as sucrose, glucose, and fructose because a longer-term pool (vacuolar fructan) contributed substantially to carbon fluxes. It should be interesting to combine this approach with other tools, such as subcellular fractionation and phloem sap sampling (e.g. Winter et al., 1992) or parallel fluxomics studies based on feeding of intramolecular position-labelled intermediates of carbohydrate metabolism (Tcherkez et al., 2009) or other experimental and mathematic methods providing compartment- and process-specific flux information (Ratcliffe and Shachar-Hill, 2006; Allen et al., 2007; Libourel and Shachar-Hill, 2008; Allen et al., 2009; Kruger et al., 2011; Zamboni, 2011).

Abbreviations

Abbreviations
 
• FEH

  fructan exohydrolase

 
• FT

  fructosyltransferase

 
• SD

  standard deviation

The members of the Lehrstuhl für Grünlandlehre at Technische Universität München and the Plant Ecophysiology Group at Université de Caen are thanked for continuous support. Expert technical assistance by Anja Schmidt and Wolfgang Feneis is gratefully acknowledged. This work was supported by the Deutsche Forschungsgemeinschaft (SFB 607) and a PROCOPE grant to M.-P.P and H.S.

Author notes

*

These authors contributed equally to this work.

†

Current address: Bavarian State Research Center for Agriculture, Vöttinger Strasse 38, D-85354 Freising, Bavaria

‡

Current address: Department for Ecology and Evolutionary Biology, Kansas Biological Survey, University of Kansas, 66047 Lawrence, KS, USA