Plant fructans stabilize phosphatidylcholine liposomes during freeze-drying


D. K. Hincha, Max-Planck-Institut für Molekulare Pflanzenphysiologie, D-14424 Potsdam, Germany. Fax:  + 49 331 5678250; Tel.: + 49 331 5678253; E-mail:


Fructans have been implicated as protective agents in the drought and freezing tolerance of many plant species. A direct proof of their ability to stabilize biological structures under stress conditions, however, is still lacking. Here we show that inulins (linear fructose polymers) isolated from chicory roots and dahlia tubers stabilize egg phosphatidylcholine large unilamellar vesicles during freeze-drying, while another polysaccharide, hydroxyethyl starch, was completely ineffective. Liposome stability was assessed after rehydration by measuring retention of the soluble fluorescent dye carboxyfluorescein and bilayer fusion. Inulin was an especially effective stabilizer in combination with glucose. Analysis by HPLC showed that the commercial inulin preparations used in our study contained no low molecular mass sugars that could be responsible for the observed stabilizing effect of the fructans. Fourier transform infrared spectroscopy showed a reduction of the gel to liquid-crystalline phase transition temperature of dry egg PtdCho by more than 20 °C in the presence of inulin. A direct interaction of inulin with the phospholipid in the dry state was also indicated by dramatic differences in the phosphate asymmetric stretch region of the infrared spectrum between samples with and without the polysaccharide.




N-(lissamine rhodamine B sulfonyl)-dioleoylphosphatidylethanolamine


gel to liquid-crystalline lipid phase transition temperature

Environmental stresses such as drought and cold severely limit the productivity of crop plants and the geographical distribution of species under natural conditions. It has been widely recognized that cellular membranes are the primary targets of injury during freezing and desiccation [1–3]. Stress tolerant plants employ a variety of strategies to survive under adverse environmental conditions, including the synthesis of protective compounds such as specific proteins [4,5], and low molecular mass sugars [3,4,6].

The fructan (polyfructose)-family of polysaccharides is another group of compounds that have been implicated in plant stress tolerance [7,8]. While there is good evidence for a functional role of specific proteins [5] and several mono-and disaccharides [1] in cellular stress tolerance, the evidence for a functional role of fructans is still circumstantial. In several species, fructans are either accumulated or modified in chain length during cold acclimatization or desiccation [7,9]. Additional evidence for a role of fructans in plant stress tolerance was provided by the fact that transgenic tobacco plants that accumulate low levels of a bacterial fructan have been shown to be slightly more tolerant of osmotic stress than wild-type plants [10]. Two recent reports point to the possibility that fructans may directly stabilize membranes under stress conditions. Ozaki and Hayashi [11] showed that a cyclic bacterial fructan (cycloinulohexaose) protected phosphatidylcholine liposomes during freezing and freeze-drying. Demel et al. [12] showed that both inulin and levan type fructans interact with lipid monolayers. While these data present no direct evidence for the ability of the linear plant fructans to stabilize bilayer membranes during freezing and drying, they prompted us to investigate the ability of some of these polymers to protect unilamellar liposomes during freeze-drying.

In addition to the obvious interest this has for elucidating a possible functional role for fructans in plant stress tolerance, it also seemed interesting because the polysaccharides investigated so far (hydroxyethyl starch and dextran) were completely ineffective in stabilizing membranes during freeze-drying [13]. The available evidence suggests that this is due to the inability of the polymers to interact with membrane lipids in the dry state and thereby depress tm of the dry membranes [14–16]. This has been attributed to the large size of the polymers, which would sterically prevent them from interacting with membrane lipids. The largest saccharides able to interact with membranes were shown to be trisaccharides [14,17].

Here we present evidence that inulins of a mean degree of polymerization of 15 from chicory and dahlia stabilize PtdCho liposomes during freeze-drying, and that stabilization is mediated by a direct interaction of the polysaccharides with the membrane lipids. We also show that hydroxyethyl starch is completely ineffective under the same experimental conditions.

Materials and methods


Egg PtdCho was purchased from Avanti Polar Lipids (Alabaster, AL, USA). Carboxyfluorescein was obtained from Molecular Probes (Eugene, OR, USA) and was purified according to the procedure described by Weinstein et al. [18]. N-(7-nitro-2,1,3-benzoxadiazol-4-yl)-phosphatidylethanolamine (NBD-PtdEtn) and N-(lissamine rhodamine B sulfonyl)-dioleoylphosphatidylethanolamine (Rh-PtdEtn) were purchased from Molecular Probes. Inulins from dahlia tubers and chicory roots were purchased from Sigma, Glc was from Pfanstiehl Carbohydrates (Waukegan, IL, USA). Hydroxyethyl starch (Mr = 200 000 [14]) was a gift of B. Spargo (Naval Research Laboratory, Washington, DC, USA) and was purified by extensive dialysis against distilled water.

Analysis of the purity of inulins

Inulin preparations from chicory roots and dahlia tubers (Sigma) were analyzed by HPLC using a CarboPac PA-100 anion exchange column on the Dionex DX-300 gradient chromatography system (Dionex, Sunnyvale, CA, USA) coupled with pulsed amperometric detection by a gold electrode. The column was equilibrated in 0.15 m NaOH and was eluted with a linear gradient of 1 m NaAc in 0.15 m NaOH as described in detail in recent publications [19,20].

Preparation of liposomes

Egg PtdCho was dried from chloroform under a stream of N2 and stored under vacuum overnight to remove traces of solvent. Liposomes were prepared from hydrated lipids using a hand-held extruder [21] (Avestin, Ottawa, Canada) with two layers of polycarbonate membranes (Poretics, Livermore, CA, USA) with 100-nm pores.

Freeze-drying experiments

Liposomes (20 µL) were mixed with an equal volume of concentrated solutions of solutes to a final lipid concentration of 5 mg·mL−1 in 1.5-mL microcentrifuge tubes. The samples were frozen by direct immersion in liquid N2 and the tubes were placed in a freeze-drier (Virtis FreezeMobile) for at least 16 h. Damage to the liposomes was determined after rehydration with 1 mL of Tes/EDTA/NaCl buffer (10 mm Tes, 0.1 mm EDTA, pH 7.4, 50 mm NaCl) either as leakage of the soluble marker carboxyfluorescein, or as membrane fusion. The figures show the means ± SD from three parallel samples. Where no error bars are visible on the results graphs, they were smaller than the symbols.

Leakage and fusion measurements

For leakage experiments, 10 mg of lipid were hydrated in 0.5 mL of 100 mm carboxyfluorescein. After extrusion, the vesicles were passed through a column (0.5 × 10 cm) of Sephadex G-50 (Pharmacia) equilibrated in distilled water to remove the carboxyfluorescein not entrapped by the vesicles. The eluted samples had a lipid concentration of approximately 10 mg·mL−1. For leakage measurements, 20 µL of sample were diluted in a cuvette in 3 mL of Tes/EDTA/NaCl. Measurements were made in a Perkin-Elmer LS-5 fluorometer at an excitation wavelength of 460 nm and an emission wavelenght of 550 nm. Fluorescence of carboxyfluorescein is strongly quenched at the high concentration inside the vesicles and is increased when carboxyfluorescein is released into the medium. The total carboxyfluorescein content of the vesicles (0% retention value) was determined after lysis of the membranes with 50 µL of 1% Triton X-100. The 100% retention values were determined with freshly prepared liposomes prior to lyophilization. For resonance energy transfer measurements [22], two liposome samples were prepared in water. One contained 0.5 mol% each of NBD-PtdEtn and Rh-PtdEtn, and the other contained only unlabeled lipids. After extrusion, liposomes were combined at a ratio of 1 : 9 (labeled/unlabeled), resulting in a lipid concentration of 10 mg·mL−1. Membrane fusion was measured by resonance energy transfer [22] as described in detail in recent publications [23,24].

FTIR spectroscopy

Spectra were obtained from freeze-dried samples loaded between CaF2 windows in a dry box purged with dry air at 0–1% relative humidity [16], with a Perkin-Elmer 1750 optical bench, using a personal computer equipped with the spectrum 2000 software. The temperature was controlled by a Peltier device and monitored with a fine thermocouple fixed on the FTIR window [25]. The peak frequencies of the CH2 symmetric stretch region (3000–2800 cm−1) were estimated by eye after baseline flattening and normalization of absorbance, using the interactive flat and abex routines, respectively [16]. tm was estimated by eye as the midpoints of the lipid melting curves [15] (see below). The peaks from the phosphate asymmetric stretch vibrations of different samples were compared after normalization of absorbance (abex) in the unflattened 1300–1200 cm−1 region.


Previous studies showed that mono- and disaccharides were the most effective sugars for membrane stabilization, while larger oligosaccharides and polysaccharides were increasingly less effective [14]. In order to assay plant inulins for their ability to stabilize liposomes during freeze-drying, it was therefore important to first investigate their possible contamination with low molecular mass sugars. We have used analytical HPLC to determine the chain lengths of commercial preparations of inulins from chicory roots and dahlia tubers (Fig. 1). For comparison, Fig. 1 also shows the elution diagram of glucose. It can be clearly seen that neither of the inulins contained detectable amounts of low molecular mass sugars. The first peak at a retention time of 24 min corresponds to a degree of polymerization of 9 and the vast majority of both inulins has a degree of polymerization between 10 and 30, which corresponds to an appoximate molecular mass of between 1600 and 5000.

Figure 1.

HPLC analysis of the degree of polymerization of the inulin preparations from chicory roots and dahlia tubers used in this study. For comparison, the elution diagram of Glc is shown. Increased retention times indicate longer chain lengths of the saccharides.

Figures 2 and 3 show that these polysaccharides had a stabilizing effect during freeze-drying on large unilamellar liposomes made from egg PtdCho. While the vesicles retained only between 0 and 5% of the encapsulated fluorescent dye carboxyfluorescein after rehydration when they were originally suspended in water alone, retention increased to about 20–25% in the presence of inulin. When the samples contained 1 mg·mL−1 Glc in addition to the inulins, retention was further increased with increasing inulin concentration, to more than 30% in the presence of 10 mg·mL−1 chicory inulin. This corresponds to a mass ratio of polymer to lipid of 2 : 1. Unfortunately, higher inulin concentrations could not be used because of the low solubility of both polymers. Glc alone at 1 mg·mL−1 did not increase retention (Figs 2 and 3) and at a mass ratio of 2 : 1 (10 mg·mL−1) retention reached only a value of 12.3%. Inulin was much more effective than Glc and led to a measurable increase in retention at concentrations between 0.2 and 0.4 mg·mL−1 (Fig. 3). At the very low inulin concentrations (below 2 mg·mL−1), the presence of 1 mg·mL−1 Glc actually slightly reduced carboxyfluorescein retention (Figs 2 and 3). This is probably due to the fact that in the frozen and dry states the effective concentrations of the inulins around the membranes are reduced by the presence of Glc, thereby limiting the protective effect of the polymers. Only at higher inulin concentrations can Glc increase the effectiveness of the fructans. Hydroxyethyl starch at the concentrations used in this study had no protective effect on liposomes during freeze-drying, either in the presence or absence of additional Glc (Fig. 2). The protective effect of fructans was not limited to liposomes suspended in pure water, which are hypertonic with respect to the external medium. Under near-isotonic conditions (50 mm NaCl outside, 100 mm carboxyfluorescein inside, 10 mm Tes, pH 7.4, on both sides) we observed a similar degree of protection by the inulins as reported in Fig. 2 (data not shown).

Figure 2.

Protection of large unilamellar liposomes by different polymers during freeze-drying. Liposomes were prepared from egg PtdCho and were loaded with the soluble fluorescent marker carboxyfluorescein (CF). Retention of CF in the vesicles was determined after freeze-drying and rehydration. Prior to freeze-drying, different solutes were added to the solution surrounding the vesicles: ○, inulin from dahlia tubers; □, inulin from chicory roots; ▵ hydroxyethyl starch (HES). Solid symbols denote samples that contained 1 mg Glc per mL in addition to the indicated amounts of the respective polymers.

Figure 3.

Protection of large unilamellar liposomes by inulin from chicory roots during freeze-drying. Retention of carboxyfluorescein (CF) in the vesicles was determined after freeze-drying and rehydration. Solid symbols denote samples that contained 1 mg Glc per mL in addition to the indicated amounts of inulin.

The interplay between the polymers and Glc in membrane stabilization was investigated further by freeze-drying liposomes with two fixed amounts (5 and 10 mg·mL−1) of either chicory inulin or hydroxyethyl starch, and a series of different Glc concentrations yielding the Glc/polymer mass ratios indicated in Fig. 4. Vesicle stability after rehydration was measured as carboxyfluorescein retention (Fig. 4A) or membrane fusion (Fig. 4B). The experiments with the chicory inulin showed a clear optimum in the Glc/ polymer ratio at around 0.4, which was more pronounced at the higher inulin concentration. The maximum retention values reached approximately 45%, but declined to about 20%, the value obtained in the absence of Glc, at the highest Glc concentrations. Hydroxyethyl starch, on the other hand, provided no protection. The effect of Glc in combination with hydroxyethyl starch was actually slightly lower than the effect of Glc alone (10% versus 13% retention in the presence of 10 mg·mL−1 Glc ± hydroxyethyl starch). Neither polymer by itself provided any protection against vesicle membrane fusion during freeze-drying (Fig. 4B). However, in combination with Glc, the chicory inulin reduced fusion in a concentration-dependent manner, while hydroxyethyl starch was again ineffective. In contrast to carboxyfluorescein retention, the reduction in vesicle fusion did not show an optimal Glc/inulin ratio. Instead, fusion decreased steadily with increasing ratio.

Figure 4.

Protection of egg PtdCho liposomes during freeze-drying by mixtures of Glc, and inulin from chicory roots (solid symbols) or hydroxyethyl starch (open symbols). Samples contained either 5 mg·mL−1 polymer (□,▪), or 10 mg·mL−1 polymer (○,●) and different concentrations of Glc giving the indicated Glc/polymer mass ratios. The intactness of the liposomes after rehydration was assessed either as carboxyfluorescein (CF) retention (A) or as vesicle fusion (B).

The stability of liposomes during freeze-drying, as measured by carboxyfluorescein leakage, is mainly the function of two factors, vesicle fusion and gel to liquid-crystalline phase transitions of the membrane lipids (reviewed in [26]). We have therefore used FTIR spectroscopy to measure phase-transition temperatures of freeze-dried liposomes. For this purpose we have monitored the frequency of the CH2 stretching mode around 2850 cm−1, which increases with increasing temperature as the chains melt [16]. Figure 5 shows data from such temperature scans and the arrows in Fig. 5A indicate the tm derived from these measurements. Egg PtdCho liposomes dried from pure water showed a tm of 29 °C. This tm suggests that the samples were not completely dehydrated (1–2 mol H2O per mol lipid), as the tm of anhydrous egg PtdCho is approximately 40 °C [15]. We did not determine the tm of samples freeze-dried in the presence of hydroxyethyl starch, as this information has already been published [15]. However, our data clearly indicate that tm was not reduced by the presence of hydroxyethyl starch. The addition of chicory inulin, both at a inulin/lipid mass ratio of 1 : 1 or 2 : 1 (corresponding to 5 or 10 mg·mL−1 inulin in the leakage and fusion experiments) reduced tm by more than 20 °C. Where Glc was also present, this had no additional effect on tm (Fig. 5B).

Figure 5.

Effects of different solutes on phase transitions in freeze-dried egg PtdCho liposomes, as measured with is taken as the midpoint of each melting curve, as indicated by the arrows in (A). Samples in (A) contained: ○, egg PtdCho without additional solutes (tm = 29 °C); ●, egg PtdCho and chicory inulin at a 1 : 1 mass ratio (tm = 8 °C); □, egg PtdCho and chicory inulin at a 1 : 2 mass ratio (tm = 7 °C); ▪, egg PtdCho and hydroxyethyl starch at a 1 : 2 mass ratio. All samples in (B) contained egg PtdCho and chicory inulin at a 1 : 2 mass ratio. In addition, some samples contained Glc at a Glc/inulin mass ratio of 0.5 (▵) or 1 (○). The addition of Glc had no significant influence on the tm of egg PtdCho freeze-dried in the presence of inulin.

The strong depression of tm in dry egg PtdCho in the presence of inulin suggests a direct interaction between the membrane lipids and the polysaccharide. This is quite surprising considering the high molecular mass of inulin and earlier findings with other polymers (see introduction). We therefore used the phosphate asymmetric stretch region around 1240 cm−1, to obtain additional evidence for such an interaction. Our measurements showed an absorbance peak centered at 1242.0 cm−1 for egg PtdCho liposomes freeze-dried from pure water (Fig. 6A). This peak was only slightly affected when egg PtdCho was freeze-dried in the presence of hydroxyethyl starch, in agreement with previous results [15]. In the presence of inulin, however, the phosphate peak was split into two, centered at 1240.5 cm−1 and 1220.6 cm−1, respectively. This indicates a direct hydrogen bonding interaction between the phosphate groups of egg PtdCho and the inulin hydroxyls. There was no significant difference between the two lipid/inulin ratios investigated (Fig. 6A). The presence of Glc in the dried samples had a similar effect as inulin (Fig. 6B). The two phosphate peaks showed maxima at 1238.8 cm−1 and 1221.2 cm−1, but were less well separated. Addition of Glc and inulin led to a further separation of the two phosphate peaks. With increased Glc concentration, the lower-frequency peak was shifted further from 1220.9 cm−1 (egg PtdCho/inulin/Glc, 1 : 2 : 1) to 1217.2 cm−1 (1 : 2 : 2).

Figure 6.

FTIR spectra of the phosphate asymmetric stretch in egg PtdCho (EPC) liposomes freeze-dried in the presence of the indicated solutes at the lipid to solute mass ratios indicated in parentheses. All spectra were obtained at 20 °C.


The results presented in this paper provide the first experimental evidence that plant fructans can stabilize membranes during freezing and desiccation stress. This makes a functional role of fructans in plant stress tolerance more likely. However, conclusive evidence for this hypothesis can only be obtained by analyzing the stress responses of transgenic plants that accumulate sufficient amounts of these molecules.

Our data also provide some insight into the physical mechanism of membrane stabilization in the dry state by inulin. Leakage of a soluble marker from the liposomes was reduced (Figs 2–4) probably due to a depression of tm in the dry lipid (Fig. 5). When egg PtdCho was freeze-dried from pure water, tm was 29 °C. Therefore, at room temperature (approximately 23 °C) the dry lipid was in the gel state. Because fully hydrated egg PtdCho has a tm of −5 °C [13], the membranes go through a phase transition during rehydration. It has been shown in several studies that this can lead to solute leakage both in model and biological systems (reviewed in [26]). When egg PtdCho was freeze-dried in the presence of inulin, tm was only 7 °C. Consequently, the lipid was in the liquid-crystalline state at room temperature and remained so during rehydration.

This is the first time this effect has been observed for a polysaccharide, as both hydroxyethyl starch (Fig. 5;[15]), and dextran [13] have no effect on tm. A strong depression of tm in dry lipids has, however, been shown for mono-and disaccharides (reviewed in [26]). An analysis of the inulins used in our experiments shows that they did not contain appreciable amounts of low molecular mass sugars (Fig. 1). Therefore, the effects we observed with the inulins were not due to contaminations, but must be ascribed to the polysaccharides. On the basis of the FTIR spectra in the region between 1800 and 1500 cm−1 (data not shown), where H2O vibrations would be visible, we could also rule out the possibility that the fructans retained additional water in the dried samples, which might have reduced leakage after freeze-drying. On first consideration, it seems difficult to reconcile our findings with previous results that suggested that steric hindrance prevents interactions between polymers such as hydroxyethyl starch and dextran, and dry lipids [13,15]. The inulins seem to behave more like low molecular mass sugars (cf [15]) than like polymers. It is possible that the difference between inulin and the other polymers resides in the fact that inulin is a linear molecule with little higher order structure, while hydroxyethyl starch and dextran are highly branched and are possibly more compact particles than the inulins. Thus, the inulins may lack the higher order structures that prevent interactions of other polymers with bilayers. In addition, hydroxyethyl starch and dextran are much larger molecules than the inulins used in this study.

tm is depressed in the dry lipids because inulin is able to hydrogen bond to the lipid headgroups. This can be inferred from the FTIR spectra of the phosphate groups (Fig. 6). In the fully hydrated state, the P = O peak in egg PtdCho has a maximum at 1230 cm−1[15]. When the water of hydration is, at least largely, removed during freeze-drying, this frequency increases to 1242 cm−1 (Fig. 6A). In agreement with the phase-transition data and earlier measurements, [15,16] hydroxyethyl starch had no significant influence on the phosphate vibration. However, with inulin, Glc, and the inulin/Glc mixtures, two peaks were observed. While one of these peaks was shifted to a lower frequency (approximately 1220 cm−1) the other peak was roughly in the same position (approximately 1240 cm−1) as the peak seen for pure dry egg PtdCho (Fig. 6A,B). This split into two peaks probably results from our experimental conditions, where the solutes were only present on the outside of the membranes. Therefore, only the outer monolayer was able to interact with the solutes. In addition, inhomogeneous interactions of the inulins with the outer monolayer may have contributed to the presence of two spectral components. If the solutes were also encapsulated within the vesicles, we would expect only one, low frequency peak and probably a higher degree of retention.

Although inulin depressed tm, it had by itself only a limited effect on retention, which was strongly increased by the addition of Glc (Figs 2 and 4A). This can, at least in part, be explained by the fact that inulin did not prevent fusion, while mixtures of inulin and Glc were quite effective (Fig. 4B). It is unclear at this point why higher ratios of Glc/inulin led to reduced retention, as they further decreased fusion and had no adverse effects on tm (Fig. 5B).

Hydroxyethyl starch had no effect on vesicle fusion during freeze-drying (Fig. 4B). This was surprising, as hydroxyethyl starch provided very effective protection against vesicle fusion during air-drying [15]. It has been shown that dextran had similar effects on membrane fusion during freeze-drying as hydroxyethyl starch [13]. Dextran actually led to an increase in fusion at low concentrations, similar to those used here for hydroxyethyl starch, and only reduced fusion significantly at concentrations several-fold higher. Whether hydroxyethyl starch would have protective effects at much higher concentrations remains to be determined. Unfortunately, similar experiments with inulin are not possible, due to its low solubility.


We would like to thank Drs L. M. Crowe and W. F. Wolkers for many helpful discussions and suggestions during the course of this project. Financial support was provided by the Deutsche Forschungsgemeinschaft through a Heisenberg stipend and a travel grant to D. K. H. and a research grant to A.G.H., and grant R01HL57810-01 from the National Institutes of Health (USA) to J. H. C.