Structure of fructans in roots and leaf tissues of Lolium perenne


Author for correspondence: M. P. Prud’homme Tel: +33 2 31 56 56 46 Fax: +33 2 31 56 53 60


  • • An analysis of fructan structures, to increase the understanding of biosynthetic pathways and enzymology of fructan synthesis in root and leaf tissues of Lolium perenne is reported.
  • • Fructan extracted from stubble of L. perenne plants was analyzed by high performance anion exchange chromatography and pulsed amperometric detection (HPAEC-PAD) using a new desalting technique. Structures of fructan isomers, separated up to DP16 (DP, degree of polymerization), were established by chromatographic elution times or by GC-MS.
  • • Fructans of DP8 belonged essentially to three series: inulin series, inulin neoseries and the levan neoseries, which is/are different in glucose (terminal or internal) and linked fructose residues. High DP fructans (DP > 8) comprised 75% molecules with an internal glucose residue. They had some branch points although 1 and 6 kestotetraose could not be detected and the β(2–6) linked fructose residues were 70 times more abundant than the β(2–1) linked fructose residues. Roots, sheaths, leaf blades and elongating leaves accumulated similar fructans although amounts of both low and high, and types of low, DP fructans, differed.
  • • It is proposed that fructans in L. perenne are synthesized via four enzymes: 1-SST (1-sucrose-sucrosefructosyl transferase), 1-FFT (1-fructan-fructanfructosyl transferase), 6G-FT (6-glucose-fructosyl transferase) and 6-FFT (6-fructan-fructanfructosyl transferase) or 6-SFT (6-sucrose-fructanfructosyl transferase).


Fructans are polymers of fructose that are based on sucrose. Fructans from different sources exhibit different degrees of polymerization (DP) and linkages between adjacent fructose residues. The fructans in the Asteraceae form a linear series of molecules (inulins) consisting almost entirely of β(2–1) linked fructose residues connected to a terminal sucrose moiety (Edelman & Jefford, 1968). The Poaceae contain fructans of a more complex structure with β(2–6) and β(2–1) fructose linkages. Fructans with both linkages in substantial amounts, such as those that occur in wheat and barley (Bancal et al., 1992; Simmen et al., 1993) are called graminans. Fructans containing primarily or exclusively β(2–6) linkages, such as those found in Dactylis glomerata (Chatterton et al., 1993c) or Poa ampla (Chatterton & Harrison, 1997) are called phleins or levans (Waterhouse & Chatterton, 1993). The glucose residue can be internal to the molecule (e.g. many of the fructan molecules from Avena (Livingston et al., 1993), and Lolium species, (Sims et al., 1992)) or can be the terminal residue (e.g. fructan from Triticum aestivum, (Bancal et al., 1992), Dactylis, (Chatterton et al., 1993a) and Bromus tectorum, (Chatterton et al., 1993b)). In some species, fructans are branched, and thus possess fructose residues, which are linked to three other fructose residues (e.g. T. aestivum (Bancal et al., 1992)).

Although grasses synthesize a wide range of fructan sizes and structures, each species appears to accumulate a characteristic suite of fructans (Chatterton et al., 1990). In the Asteraceae, fructans of the inulin type can be synthesized in vitro from sucrose by the concerted action of 1-SST (1-sucrose-sucrosefructosyl transferase) and 1-FFT (1-fructan-fructanfructosyl transferase) (Koops & Jonker, 1996). In onion, which contains fructans based on 6G-kestotriose with β(2–1) linkages between fructose residues, an additional enzyme activity is required to account for the production of 6G-kestotriose. This enzyme, the 6G-FT (6-glucose-fructosyl transferase), has been characterized by Shiomi (1989) and cloned by Vijn et al. (1997). In wheat and barley, where fructans are mainly issued from 1 and 6-kestotetraose (bifurcose) and contain predominantly β(2–6) fructose linkages, three enzymes are thought to be capable of synthesizing all the fructans, namely 1-SST, 1-FFT and 6-SFT (6-sucrose-fructanfructosyl transferase) (Duchateau et al., 1995; Sprenger et al., 1995). In Poa ampla, a specific 6-SST (6-sucrose-sucrosefructosyl transferase) would be necessary to allow the production of 6-kestotriose, which occurs in the absence of significant amounts of either 1-kestotriose or bifurcose (Chatterton & Harrison, 1997).

Lolium species accumulate high molecular mass fructans with mainly β(2–6) linkages. These fructans are mostly linear, but can also be branched and contain either an internal or terminal glucose residue ( Tomasic et al., 1978 ; Sims et al., 1992 ; Bonnett et al., 1994 ). However, oligofructans (DP < 5) found in Lolium temulentum and Lolium rigidum contain mainly β(2–1) fructose linkages with β(2–6) linked oligosaccharides present in smaller proportions ( Sims et al., 1992 ; St. John et al., 1997 ). Fructan synthesizing enzymes from Lolium have never been purified to homogeneity or cloned ( St. John et al., 1997 ). Previous studies suggest that mature leaf sheaths of L. perenne may represent a desirable tissue from which to purify 1-SST activity ( Guerrand et al., 1996 ). A recent assessment of the relative contributions of insoluble and soluble fructosyltransferase activities to fructan synthesis provides evidence that a minor fraction (2–5%) of the trisaccharides are synthesized by the insoluble pathway ( Guerrand et al., 1999 ).

Analysis of fructan structures is a prerequisite for understanding the biosynthetic pathways and the enzymology of fructan synthesis. This study reports: the separation and purification by high performance anion exchange chromatography and pulsed amperometric detection (HPAEC-PAD) of the fructan isomers extracted from stubble of L. perenne; the linkage analysis of fructo-oligosaccharides up to DP8, using GC-mass spectrometry of partially methylated alditol acetates; the structure analysis of fructans in roots and leaf tissues of L. perenne; and (4) the pattern of fructan degradation and accumulation after defoliation.

Materials and Methods

Plant material

Seeds of Lolium perenne var. Bravo were germinated in 9 L pots and grown in a controlled environment on a nutrient solution as previously described by Gonzalez et al. (1989) and Prud’homme et al. (1992). The nutrient solution was aerated continuously and replaced every week. Plants were grown with day : night temperatures of 22 : 18°C, a 16-h photoperiod and a photosynthetic photon flux density of 500 mmol photons m−2 s−1.

To obtain for fructan structure analyzes, stubble of L. perenne was enriched in fructans according to the method used by Smouter & Simpson (1991) and St. John et al. (1993). After the dark period, 8-week-old plants were placed in a nutrient solution at 5°C. The apical meristems were covered by nutrient solution and illuminated continuously at a photosynthetic photon flux density of 500 mmol photons m−2 s−1 for up to 72 h.

The fructan time-course study in roots and leaf tissues of L. perenne during regrowth after defoliation was effected on plants grown as described in Morvan-Bertrand et al. (1999). Briefly, plants were defoliated at 4 cm aboveground and transferred to a growth cabinet (E15, Conviron, Winnipeg, Canada) with a 16-h photoperiod (300 mmol photons m−2 s−1 at the top of the foliage) and a day : night temperature of 23 : 18°C. Plants were grown in 3.5 l pots (3 plants per pot) in the growth cabinet. The nutrient solution was replaced every 5 d. Triplicate pots containing three plants each were sampled at 0, 2, 6 and 14 d after defoliation. Roots were separated from shoots and frozen in liquid nitrogen. Shoots were placed on ice during dissection. The surrounding old sheaths were harvested, as well as the bases of elongating leaves (from 0 to 4 cm above the ground level) and their corresponding top parts as soon as they appeared above the cutting level. All plant tissues were frozen in liquid nitrogen, freeze-dried, ground to a fine powder and stored at 5°C in a desiccator until subsequent analyzes.

Extraction of water-soluble carbohydrates

Frozen tissues were placed in boiling 80% ethanol and extracted under reflux for 1 h. The ethanol filtrate was evaporated to dryness under vacuum and the residue was dissolved in water. This fraction contained fructose, glucose, sucrose and fructans of low DP. Fructans of higher DP were extracted from the initial insoluble residue with boiling water for 1 h (Prud’homme et al., 1992). Extraction in boiling water had apparently not resulted in hydrolysis of fructans since extraction in water at 60°C gave the same results (Morvan-Bertrand, 1998). Aliquots of carbohydrate extracts were passed through a column containing cation exchange resin (Amberlite IR 120, H + – form, Serva, Germany), anion exchange resin (Amberlite IRA 416, formate form, Serva, Germany) and C18 modified silica (Waters, MA, USA) to remove charged compounds, proteins, lipids and pigments (Smouter & Simpson, 1991). The columns were eluted with water. Samples were concentrated under vacuum, dissolved in water and component sugars were separated by high-performance anion exchange chromatography and pulsed amperometric detection (HPAEC-PAD DX-300, Dionex, Sunnyvale, CA, USA) on an analytical CarboPac PA100 column (4 × 250 mm) using a sodium acetate gradient (1 ml min−1) in 150 mM NaOH. The elution program consisted of 25 mM sodium acetate (0–6 min), 50 mM (6–12 min), 150 mM (12–18 min), 250 mM (18–19 min), 275 mM (19–30 min).

For fructan structure analysis, the ethanol or water concentrated extract was depigmented by overnight contact with polyvinylpolypyrrolidone (10%) followed by centrifugation. Carbohydrates contained in the supernatant were then cleared of charged compounds, proteins and lipids, as already described. Purified carbohydrates were freeze-dried and stored at 5°C until analyzed.

Purification of individual fructans

Individual fructan oligosaccharides contained in the 80% ethanol extract (isomers 1–11) of both elongating leaf bases and sheaths of mature leaves, were separated by HPAEC-PAD on a preparative CarboPac PA1 (9 × 250 mm), using a sodium acetate gradient in 100 mM NaOH (5 ml min−1). The elution program consisted of sodium acetate at the following concentrations: 5 mM (0–6 min); 10 mM (6–12 min); 45 mM (12–18 min); 60 mM (18–19 min); 65 mM (19–30 min); and 80 mM (30–40 min).

Individual fructans contained in the water extract (isomers 12–14), were also isolated by HPAEC-PAD on a preparative CarboPac PA1 column. However, the elution program was different and consisted of sodium acetate at the following concentrations: 25 mM (0–6 min); 50 mM (6–12 min); 150 mM (12–18 min); 250 mM (18–19 min); 275 mM (19–30 min); and 450 mM (30–60 min) in 150 mM NaOH.

Purified isomers were collected from the preparative column and pooled from multiple runs.

Deionization of extracts

Samples were deionized by passing through two cation self-regenerating suppressors (CSRS, Dionex, Sunnyvale, CA, USA) and one anion self-regenerating suppressor (ASRS, Dionex) set in a series. Assays of ion suppression were run with the elution solvent alone. Two different columns were used: the CarboPac PA1 and the CarboPac PA100. A sodium acetate gradient in 150 mM NaOH was applied at a rate of 2.5 ml min−1 and 1.0 ml min−1 for CarboPac PA1 and CarboPac PA100, respectively. A discontinuous gradient consisting of 30, 120, 210, 390, 480 and 570 mM sodium acetate in 150 mM NaOH was applied on the columns. Fractions of 3 ml (CarboPac PA100) or 5 ml (CarboPac PA1) were collected and passed successively through two CSRS and one ASRS suppressors with a Dionex DXP pump at a flow rate of 2 mL min−1. Osmolarity was determined by a milliosmometer (Digital osmometer, Roebling, Berlin, Germany), calibrated with ultrapure water and a standard solution of sodium sulfate prepared at 300 milliosmol kg−1.

Analysis of fructan structure

The degree of polymerization of each fructan oligomer was established from the glucose to fructose ratio following complete hydrolysis (in 5 M HCl at 50°C for 7–30 min depending on the DP), assuming one glucose residue per molecule. Fructan oligomer structures (DP3-DP8) were either deduced from their retention times on HPAEC-PAD by comparison with known standards, or characterized by determining the structures of their partial hydrolysates (in 5 M HCl at 50°C for 20–60 s depending on the DP) (Chatterton et al., 1993b), and/or by chemical analysis using a mass spectrometer. In that case, lyophilized materials were per-O-methylated with the Li + methylsulfinylmethanide ion. Per-O-methylated fructan oligomers were hydrolyzed in 1 M TFA containing 1 mmol of myo-inositol at 70°C for 30 min. Ter-butylalcohol was added and the acid of the solution was evaporated in a stream of N2. Carbohydrates were reduced with sodium borodeuteride and acetylated as described by Carpita et al. (1989). Derivatives were separated in a 30 m × 0.2 mm silica WCOT capillary column (Supelco, Bellefonte, PA, USA) temperature-programmed from 180°C to 240°C at 2°C min−1. A Finnigan/MAT 9610 gas chromatograph was coupled with a Finnigan/MAT 4021 quadrupole mass spectrometer. Spectra were obtained at 70 eV and a source temperature of 250°C. Derivatives were detected and quantified as described by Carpita & Shea (1989).


Removal of ions

HPAEC is currently the most powerful tool for the separation of fructans and more generally in analysis of polysaccharide isomers. However, this technique is limited for two reasons. First, the PAD detector has a different response factor characteristic for each individual compound. As such, the instrument must be calibrated individually for each peak detected, otherwise the technique is only qualitative. Second, the high concentrations of sodium acetate needed to elute the carbohydrates are problematic for further analysis. Fig. 1 shows the effects of two cations (CSRS) followed by one anion (ASRS) suppressors on ion removal from the elution gradient. Sodium acetate is converted into sodium hydroxide in the CSRS; sodium hydroxide is converted into water in the ASRS. The volumes of the fraction eluted from the preparative CarboPac PA1 and the analytical CarboPac PA100 columns were established by the quantity needed for fructan isomer determinations and varied with the type of the column. The solvent osmolarity was measured at the detector outlet after elution through the column and above the set of suppressors (Fig. 1). Efficiency of ion removal was better when the osmolarity of the solvent was low. When sodium acetate in 150 mM NaOH was < 210 mM, ion elimination was > 98% and did not vary with the quantity of ions initially present in the fraction.

Figure 1.

Effect of ion self-regenerating suppressors (two CSRS followed by one ASRS) on ion elimination expressed as (a) osmolarity (milliesmol kg −1 ) and (b) percent removal. Fractions of 3 ml and 5 ml eluted from analytical CarboPac PA100 (m, open circles) and preparative CarboPac PA1 (l, closed circles) columns, respectively, were subjected to the three suppressors set in a series. Percent removal was calculated from the initial osmolarity measured at the detector outlet and that present after passage over the suppressors. The insert shows osmolarity at the various concentrations of sodium acetate.

Using one CSRS instead of two resulted in a lower ion elimination efficiency. Indeed, osmolarity in 3 ml (eluted from CarboPac PA100) and 5 ml (CarboPac PA1) of 570 mM sodium acetate in 150 mM NaOH was 28 and 39 milliosmol kg−1, respectively, with one CSRS and fell to 9 and 19 milliosmol kg−1, respectively, with two cation suppressors.

The pH values varied from 4.0 to 3.0 and acidity increased concomitantly with the rise in the initial sodium acetate concentration. Because acidity of the solvent may result in hydrolysis of polyfructosyl isomers (Chatterton et al., 1993b), we tested fructan integrity following desalting. A fructan isomer (I) that eluted in 45 mM sodium acetate-100 mM NaOH was collected, deionized as described above and applied again to the CarboPac PA1 column (Fig. 2). Inasmuch as only isomer I was detected, hydrolysis did not occur during the desalting process. The same results were obtained with isomers collected in higher initial concentrations of sodium acetate. Consequently, this new desalting procedure did not alter the integrity of the fructans that could then be collected with a 70% yield (Fig. 2).

Figure 2.

Effect of ion suppression on fructan integrity. (a) Chromatogram of water-soluble carbohydrates extracted in 80% ethanol from stubble of Lolium perenne . The CarboPac PA1 column was equilibrated with 100 mM NaOH and then eluted using the following concentrations of sodium acetate: 5 mM (0–6 min), 10 mM (6–12 min), 45 mM (12–18 min), 60 mM (18–19 min), 65 mM (19–30 min) and 80 mM (30–40 min). (b) Chromatogram of isomer I, following deionization with two CSRS and one ASRS suppressor. nA, nanoamper.

Structures of fructans

Fructans were extracted from both elongating leaf bases and sheaths of mature leaves of 8-wk-old plants that had been subjected to low root temperatures and continuous illumination for 72 h before harvest. Fructan isomers from both ethanol and water extracts were resolved by HPAEC-PAD (Fig. 3). Although the particular NaOAc gradient used did not separate glucose from fructose, or completely resolve high DP fructans (isomers no. 14), isomers separated sufficiently well to provide a useful general profile. In order to be identified, however, fructans were first separated on the basis of their differential solubility in ethanol and water.

Figure 3.

An high performance anion exchange chromatography and pulsed amperometric detection (HPAEC-PAD) profile of total water-soluble carbohydrates extracted from both elongating leaf bases and sheaths of mature leaves of Lolium perenne . The CarboPac PA100 column was equilibrated with 150 mM NaOH and then eluted with the following concentrations of sodium acetate: 25 mM (0–6 min), 50 mM (6–12 min), 150 mM (12–18 min), 250 mM (18–19 min) and 275 mM (19–30 min). Abbreviations: F, fructose; G, glucose; L, loliose; R, raffinose; S, sucrose. An identifying number is assigned to the main peaks. Peak nos 1, 2, 3, 4, 5, 6, 7, 9, 10, 11, 13, 14 were identified by mass spectrometry ( Table 1 ) and/or partial hydrolysis ( Table 2 ). The peak, designated by an asterisk, contained both 1,1 & 6G,1-kestohexaose and 1 & 6G,1,1-kestohexaose. Peak nos 8 and 12 correspond to fructan, but their structure could not be solved. nA, nanoamper.

As reported earlier by Chatterton et al. (1990) and Chatterton et al. (1993a), raffinose and loliose (α-D-Gal-(1–6)-α-D-Glu-(1 → 2)-β-D-Fru) were found in significant amounts. 1-Kestotriose (peak no. 1) and 6G-kestotriose (peak no. 4) were the dominant trisaccharides, whilst 6-kestotriose (peak no. 3) was detected in very small amounts (Fig. 3).

The first DP4 isomer eluted from the anion exchange column (peak no. 5) contained one terminal glucosyl unit, one terminal fructose and two 1-linked fructoses (Table 1). It was thus identified as 1,1-kestotetraose (Table 2). The second DP4 isomer to elute (peak no. 6) contained one internal glucose, one 6-linked fructose and two terminal fructoses (Table 1). Partial hydrolysis produced both 1-kestotriose and 6G-kestotriose (Table 2). We concluded the chemical structure to be 1 and 6G-kestotetraose. Methylation analysis of the third DP4 oligomer (peak no. 7) yielded one internal glucose, one 6-linked fructose and two terminal fructoses. The only trisaccharide formed upon partial hydrolysis was 6G-kestotriose. Thus, the structure of this DP4 isomer is 6G,6-kestotetraose. Two other DP4 isomers, bifurcose (1 and 6-kestotetraose) and 6,6-kestotetraose, commonly found in wheat (Bancal et al., 1992), Poa and Dactylis (Chatterton et al., 1993c; Chatterton & Harrison, 1997) could not be found. Externally added bifurcose eluted between 1,1-kestotetraose (peak no. 5) and 1 and 6G-kestotetraose (peak no. 6) while exogeneous 6,6-kestotetraose eluted before 6G,6-kestotetraose (peak no. 7) at retention times where no isomers of Lolium were detected.

Table 1.  Types of glucose and fructose moieties obtained from some DP3 to DP15 fructans from Lolium perenne
Isomer no.DPt-Fru (terminal)t-Glu (terminal)6-Glu (internal)1-Fru (1-linked)6-Fru (6-linked)1,6-Fru (branched)
  1. These data refer to fructans upon hydrolysis following methylation (proportion expressed relatively to the amount of t-Fru (isomer nos 1–14) or in mol percentage (isomer no. 15)). Isomer numbers refer to designations in Fig. 1.*The proportion of t-Fru was sometimes slightly less than the expected value, possibly due to the apparent fructose degradation during methylation.

 1DP 31*11
 3DP 3111
 4DP 321
 5DP 4112
 6DP 4211
 7DP 4211
 9DP 5212
10DP 5212
11DP 6213
13DP 7115
14DP 8215
≥15DP > 88.00%0.26%0.88%1.22%86.0%3.45%
Table 2.  Products of the partial hydrolysis of some Lolium perenne oligosaccharides and identity of the main fructan isomers
Isomer no.Products of mild hydrolysisIdentity
  1. Isomer numbers refer to designations in Fig. 3 and Table 1.

 1– 1-kestotriose
 2glucose, fructose and galactoseloliose
 61-kestotriose and 6G-kestotriose1 & 6G-kestotetraose
 9no. 5, no. 6, 1-kestotriose and 6G-kestotriose1,1 & 6G-kestopentaose
10no. 7 and 6G-kestotriose6G,6,6-kestopentaose
11no. 5, no. 6, no. 9, 1-kestotriose and 6G-kestotriose1,1,1 & 6G-kestohexaose
14no. 5, no. 6, no. 9, no. 11, 1-kestotriose and 6G-kestotriose1,1,1,1,1, & 6G-kestooctaose

The first DP5 isomer to elute (peak no. 9) contained one internal glucose, two 1-linked fructoses and two terminal fructoses (Table 1) and yielded 1 and 6G-kestotetraose and 1,1-kestotetraose when partially hydrolyzed (Table 2). These data are consistent with the structure of 1,1 and 6G-kestopentaose. The second isomer of DP5 was identified (peak no. 10) as 6G,6,6-kestopentaose because it gave rise to two 6-linked fructoses and no. 1-linked fructose after methylation analysis (Table 1) and produced 6G,6-kestotetraose following mild hydrolysis (Table 2). The other DP5 isomers (peak no. 8 and the minor peak eluting after peak no. 11 at 11.4 min) were not pure enough to be clearly characterized. They coeluted, however, with 1 and 6G,1-kestopentaose and 6 and 6G,6-kestopentaose, respectively.

Only one of DP6 (peak no. 11), DP7 (peak no. 13) and DP8 (peak no. 14) isomers could be identified. Based on their methylation analysis (Table 1) and on their partial hydrolysis products (Table 2), the DP6 and DP8 fructans were characterized as members of the 1,1 and 6G series, while the DP7 isomer belonged to the inulin group (Fig. 3; Tables 1 and 2). Another DP7 isomer (peak no. 12) was collected, but its identity was not established. Its methylation analysis yielded internal glucose and 6-linked fructoses. Another DP6, corresponding to a minor peak that eluted at 12.7 min (peak *, Fig. 3) separated into three peaks when analyzed in the chromatographic conditions described previously by Chatterton et al. (1993b). Among the three peaks, two coeluted with known standards, namely the 1 and 6G,1,1-kestohexaose and the 1,1 and 6G,1-kestohexaose.

A flow chart showing the structural relationships among DP3 to DP8 fructans is presented in Fig. 4. Fructans from Lolium belonged essentially to three series: the inulin series, with a terminal glucose residue and β(2–1) linked fructose residues; the inulin neoseries with an internal glucose residue and β(2–1) linked fructose residues; and the levan neoseries with an internal glucose residue and β(2–6) linked fructose residues.

Figure 4.

A flow diagram of the fructan oligomers of DP 3–8 found in Lolium perenne , showing structural relationships. Numbers refer to designations in Fig. 3 , Table 1 and Table 2 .

The profile of the large fructans (DP > 8, peaks no. 15) indicate the presence of multiple series with one being dominant (Fig. 3). Peaks of the dominant series, nos 15–22 (DP9 to DP16, respectively), all contained internal glucose, 6-linked fructose residues and branched points. Fructans of DP > 8 were pooled and the resulting mixed fructans were analyzed using the same methods as employed for purified oligosaccharides (Table 2). In addition to a high proportion of 6-linked fructose residues (86%), a majority of internal glucose (0.88%) and few branch points (3.45%), their methylation analysis yielded 1-linked fructose residues (1.22%) and terminal glucose (0.26%). Therefore, the fructans purified from L. perenne contained predominantly internal glucose and 2,6-linked fructose. However, there was also either a high number of molecules with few (one or two) 2,1-linked fructose residues or a small number of molecules with a high proportion of 2,1-linked fructose residues. All together, these results suggest that the dominant series was composed of 6G,6,6-fructans and the minor series of 1,1 and 6G and/or inulin isomers. An argument in favor of the last hypothesis is that inulin polymers extracted from Jerusalem artichoke coeluted with some of the minor peaks of L. perenne (data not shown).

Fructan pattern in roots, mature leaves and elongating leaves

Stubble is a heterogeneous plant compartment that includes both fully expanded leaf material (leaf sheaths), as well as basal immature parts of expanding leaves (elongating leaf bases). These two compartments were separated, fructans were extracted and their patterns were compared with the complement of fructans in leaf blades and in roots. For each tissue, the same amount of fructans (50 µg) was applied to the chromatographic anion exchange column (Fig. 5).

Figure 5.

Chromatograms of water-soluble carbohydrates in roots, sheaths, elongating leaf bases and blades of 8-wk-old Lolium perenne . 50 mg of fructans from each tissue were applied to the CarboPac PA100 column. Numbers refer to designations in Fig. 3 , Table 1 and Table 2. nA, nanoamper.

When chromatographed together, fructans from leaf sheaths coeluted with fructans from bases of elongating leaves (data not shown) suggesting that these two plant parts generated the same polymers. Fructans differed, however, in their relative abundance. The proportion of low DP fructans (DP < 6, peaks nos 1–12) was more prominent in bases of elongating leaf bases than in leaf sheaths, whereas the opposite was true for fructans of DP > 6. Differences also exist amongst oligosaccharides. For instance, the three DP4 isomers (peaks nos 5–7) as well as the DP5 (peak no. 9) were relatively more abundant in leaf bases than in sheaths.

Fructans synthesized in roots were essentially low DP (DP3 to DP7). They coeluted with the oligosaccharides from leaf sheaths and elongating leaves bases.

The fructan profile of mature leaf blades was different from that in leaf sheaths and elongating leaf bases. The isomers no. 4 (6G-kestotriose), no. 6 (1 and 6G-kestotetraose) and no. 8 (unidentified) were predominant in leaf blades. Furthermore, extracts from leaf blades contained additional peaks that eluted between no. 10 and no. 12 (Fig. 5). However, the superposition of fructan pattern from leaf blades with the fructan profile of the other leaf tissues indicates that these fructans were not only present in mature leaf blades, but were also in the other plant parts, but in much smaller amounts (data not shown).

Effect of defoliation on fructan pattern in Lolium plant parts

During the initial phase of regrowth following defoliation, fructan concentrations decreased in roots, leaf sheaths and bases of elongating leaves (Morvan-Bertrand et al., 1999). In roots, only three carbohydrates remained on the second day of regrowth: peak no. 2 (loliose), no. 6 (1 and 6G-kestotetraose) and no. 8 (unidentified) (Fig. 6). These compounds slightly decreased between the second and the sixth day of regrowth. On the 14th day after defoliation fructan oligomers began to appear in root tissues. Fructans of DP4 (peaks nos 5–7) were the first synthesized.

Figure 6.

Chromatograms of water-soluble carbohydrates in roots, old sheaths, bases of cut elongating leaves and tops of cut elongating leaves during 14 d after defoliation (DO, D2, D6, D14) of 8-wk-old Lolium perenne . Carbohydrates were extracted from roots (25 mg dry weight), old sheath 1.2 mg (DO, D2) or 18 mg (D6, D14), bases of cut elongating leaves (3.7 mg) and from tops of cut elongating leaves (6.2 mg).

In sheaths that remained after defoliation (so called ‘old sheaths’ in order to distinguish them from newly synthesized sheaths), fructans of high DP (no. 13) were hydrolyzed without any transient accumulation of low DP fructans. By the 6th day of regrowth all high DP fructans had been hydrolyzed while the three oligosaccharides that persisted in roots (peaks nos 2, 6 and 8) were also predominant in old sheaths.

Given that most of the constituent cells of leaf bases were renewed every 2 d (Morvan-Bertrand et al., 1999), fructans present on the second, the sixth or the 14th day of regrowth did not correspond to products of hydrolysis but should rather be regarded as newly synthesized. The peak that eluted between peak no. 7 and peak no. 8 became predominant on the 6th day of regrowth, and was identified as 6G,1-kestotetraose.

A comparison of fructans in bases of cut elongating leaves at the onset of the experiment with those in tops of cut elongating leaves on the second day of regrowth showed that all the fructans present before defoliation in bases were not present in tops on the second day. This suggests that a high proportion of the fructans was hydrolyzed during the first 2 d of regrowth. The two predominant oligosaccharides (peak nos 6 and 8) that remained on the second day of regrowth were the same as those found in roots on the second and sixth day as well as in sheaths on the sixth day. Fructans synthesized in tops of cut elongating leaves on the sixth and the 14th days after defoliation were similar to the fructans accumulated in mature leaf blades (Fig. 5).


Fructan structures in Lolium species

The results of the linkage analysis reported here for L. perenne indicate similarities with but differences from, the structures of fructans reported in other members of Lolium. In common with L. rigidum and L. temulentum, 1-kestotriose and 6G-kestotriose were the most abundant trisaccharides in L. perenne. 6-kestotriose was present in much lower amounts (Smouter & Simpson, 1991; Sims et al., 1992).

The DP4 fructan molecules of L. rigidum contained exclusively 2,1-linked fructose residues, 6G,1-kestotetraose and 1 and 6G-kestotetraose being present in higher concentrations than 1,1-kestotetraose (St. John et al., 1997). The two latter DP4 molecules have also been identified in L. perenne (present study) and L. temulentum (Sims et al., 1992). Additionally, these two species contained a DP4 fructan molecule with a 2,6-linked fructose residue attached to the 6G-kestotriose (6G,6-kestotetraose). The 6G,1-kestotetraose has been reported in the three species.

Amongst the DP5 fructan isomers, 1,1 and 6G-kestopentaose and 1 and 6G,1-kestopentaose were common to the three L. species. The 6G,6,6-kestopentaose identified in L. perenne was also found in L. temulentum, but not in L. rigidum. However, about half of the DP5 fructan molecules in L. rigidum were not identified (St. John et al., 1997). These unknown isomers contained almost as much 2,6-linked as 2,1-linked fructose residues. It is possible therefore that the 2,6-linked fructose residues belong to the 6G,6,6-kestopentaose.

Fructans of DP > 8 from L. perenne were similar to the mean size of both fraction C described by Sims et al. (1992) in L. temulentum and the fructan mixture analyzed by Bonnett et al. (1994) in L. rigidum. The majority (76%, 66%, respectively) of high molecular weight fructans from both L. perenne and L. rigidum contained an internal glucose. By contrast, L. temulentum exhibits a progressive increase in the proportion of linear fructans with a terminal glucose residue, as the mean size of the fructans increase. In the fraction containing the highest mean DP (30) fructans, the amount of internal glucose was too small to be quantified. Consequently, high molecular weight fructan molecules in L. temulentum are thought to contain exclusively terminal glucose residues. Marked differences also exist in the relative abundance of 2,6-and 2,1-fructose-fructose linkages among the three Lolium species. There is a higher proportion of 2,6-linked fructose in high molecular weight fructan from L. rigidum (83 : 1, 2,6-linked fructose : 2,1-linked fructose) and L. perenne (72 : 1) than from L. temulentum (40 : 1).

Bonnett et al. (1997 ) proposed that branch points could be used as taxonomic markers to distinguish the Triticodae from the Poodae, since they appeared to be characteristic of the fructan from the Triticodae but absent from or rare in fructan from the Poodae to which Lolium species belong. Indeed, in L. rigidum and L. temulentum , 2,1,6-linked fructose residues were found in very low abundance (2% of all the residues), indicating less than one branch point per molecule. In L. perenne , however, branch points represented up to 3.45% of total residues, which correspond to three to four branch points per molecule, assuming that fructan isomers of DP > 8 all contained the same amount of 2,1,6-linked fructose residues. In Triticum , a member of the Supertribe Triticodae, fructans with a mean DP of nine contained an average of three branch points per molecule ( Carpita et al., 1989 ; Praznik et al., 1992 ). Therefore, the relative abundance of branch points might not be a useful taxonomic marker to distinguish the Triticodae ( Triticum ) and the Poodae ( Lolium ). The 1 and 6-kestotetraose (bifurcose), a branched isomer of DP4, is common to the members of the Triticodae studied so far: ( Triticum ( Carpita et al., 1989 ); Hordeum ( Simmen et al., 1993 ); and Bromus ( Chatterton et al., 1993b )). However, it has never been identified in Lolium species, including L. perenne . Unfortunately, this particular fructan cannot be used either as a taxonomic marker to separate the two Supertribes since it has been reported in Dactylis , a member of the Poodae ( Chatterton et al., 1993c ).

It is unclear whether differences in fructan structures are a result of species differences or of the conditions under which plant material are grown. For example, Bancal et al. (1992) obtained a very different proportion of each fructan isomer from excised wheat leaves induced to accumulate fructan than from field-grown stems and sheaths. Sims et al. (1992) used excised leaves of L. temulentum induced to synthesize fructans by continuous illumination for 24 h whereas Bonnett et al. (1994) extracted fructans from 10-d-old seedlings of L. rigidum induced to accumulate fructan by cooling the meristems and exposing the shoots to continuous high light for 4 d. Interestingly, these fructan-inducible conditions applied to 8-wk-old plants of L. perenne led to a complement of high molecular weight fructan similar to the one reported for L. rigidum. These results suggest then that differences in fructan composition reported in L. species may be related to the conditions under which fructans were accumulated. Recently, Chatterton & Harrison (1997) showed that fructan metabolism in leaves of Poa ampla was altered by a 5°C change in the day/night growth temperature. Poa ampla plants grown in the colder environment contained polymers with 2,6 linkages but no. 2,1-linked fructans. Considerable variation in the abundance of fructan isomers may also exist among different tissues of the same plant. The outer leaf sheath of tall fescue (Festuca arundinacea) contains significantly more fructan DP > 6 and significantly less fructan DP3–6 than the expanding leaves (Housley & Volenec, 1988). Similar results are reported here for L. perenne. Mature leaf sheaths were characterized by high DP fructans whereas elongating leaf bases contained more low DP fructans. These differences could reflect differences between tissues, that is in the nature of the enzymes present or in their regulation. Differences could also be the consequence of the length of time during which fructan have accumulated. Clearly, fructan in bases of elongating leaves have accumulated for only a few days (Morvan-Bertrand et al., 1999), a time presumably insufficient to allow the synthesis of significant amounts of high DP fructan. In contrast, fructan in leaf sheaths that have accumulated over a long period were essentially of high molecular weight. An argument in favor of this hypothesis is that in tall fescue, 3-d-old leaf sheaths contained fructans with a similar average DP as leaf bases (Volenec, 1986).

Fructan synthesis pathway in Lolium

Lolium species produce a complex of fructan structures that belong to three different series: the inulin series, containing exclusively 2,1-linked fructose residues attached to the fructose of the initial sucrose; the inulin neoseries, with an internal glucose residue and β(2–1) linked fructose residues; and the levan neoseries, with an internal glucose residue and β(2–6) linked fructose residues. Grass fructans are synthesized de novo from sucrose, with sucrose as the sole substrate ( Cairns et al., 1999 ) but the enzymatic mechanism is still a matter of debate because all the enzymes or genes of the pathway have not yet been purified. However, based on the current knowledge obtained for different species and assuming that in grasses multiple fructosyl transferases operate ( Vijn & Smeekens, 1999 ), a set of four enzymes would be necessary to account for the synthesis of the three fructan types found in Lolium . Both 1-SST and 1-FFT activities have been measured and the corresponding enzymes have been partially purified from Lolium ( St. John et al., 1997 ) and from other grasses ( Jeong & Housley, 1992 ; Simmen et al., 1993 ). These enzymes could be responsible for the synthesis of the inulin type fructans. 6G-kestotriose is likely the product of the 6G-FT cloned by Vijn et al. (1997 ) in onion. As a majority of fructans from Lolium are based on the 6G-kestotriose, we suggest that most of the flux of C from sucrose to 1-kestotriose is probably mediated by this particular enzyme. Based on the fact that the β(2–6) linkages prevail in grasses, Duchateau et al. (1995 ) suggested that the 6-SFT that catalyzes the transfer of a fructosyl residue from sucrose to a fructan in a β(2–6) linkage, may be the key enzyme for the formation of the (2–6)-linked fructans. Nevertheless, in barley from which 6-SFT has been purified, the main product of the enzyme is bifurcose, a branched DP4 fructan which has not been found in Lolium species. Consequently, if 6-SFT also occurs in Lolium , its affinity for 1-kestotriose must be lower than the affinity of 6G-FT for the same trisaccharide. A lower affinity for 1-kestotriose would result in the synthesis of 6G-kestotriose instead of bifurcose. Furthermore, the affinity of 6-SFT for 6G-kestotriose or for β(2–6)-linked neokestose type fructans must be higher than for 6-kestotriose because no. 6,6-kestotetraose has been found in Lolium . In contrast barley does synthesize 6,6-kestotetraose ( Duchateau et al., 1995 ). Another alternative would imply a 6-FFT (6-fructan-fructanfructosyl transferase) activity which specifically forms β(2–6) fructose linkages without using sucrose as the fructosyl donor.

Fructan hydrolysis in Lolium

Three carbohydrates – (loliose (peak no. 2), 1 and 6G-kestotetraose (peak no. 6), and an unidentified fructan (peak no. 8)) – appear to resist hydrolysis when all others are being mobilized. The reason for this is unknown, but could be attributed either to the specificity of the hydrolytic enzymes involved or to the roles of these carbohydrates in plants: source of C, metabolic intermediate or phloemic carrier of C.

According to the recent studies of Bonnett & Simpson (1993, 1995) and Marx et al. (1997), several isoforms of FEH exist in Lolium species. Some hydrolyzed β(2–1) linked fructans faster than β(2–6) linked fructans, while others exhibited β(2–6) specific activity. The present work showed that in roots and old sheaths of L. perenne, β(2–6) fructans were degraded in the same proportion as β(2–1) fructans after defoliation. Consequently, both FEH isoforms were probably induced by defoliation.

In old sheaths, the amount of high DP fructans decreased without the accumulation of low DP fructans. This result was unexpected, since oligomeric products were released from FEH enzymes purified from oat or barley after one catalytic cleavage (Henson & Livingston, 1996, 1998). This result has been used as evidence for a multichain rather than a single-chain mechanism of hydrolysis. Our results suggest that catalysis in L. perenne occurs either by single-chain attack and/or that the affinity of FEH enzymes is greatest for small fructans. This is, in fact, in accordance with the results obtained by Bonnett & Simpson (1993) on L. rigidum. The higher affinity of FEH enzymes for small molecules may preclude the transient accumulation of smaller partial hydrolysis products.


We thank N. C. Carpita (Purdue University, USA) for molecular mass analysis by GC-MS and Dionex (Jouy en Josas, France) for providing the ion self-regenerating suppressors.