Xylem parenchyma cells (XPCs) of boreal hardwood species adapt to sub-freezing temperatures by deep supercooling to maintain a liquid state of intracellular water near −40 °C. Our previous study found that crude xylem extracts from such tree species exhibited anti-ice nucleation activity to promote supercooling of water. In the present study, thus, we attempted to identify the causative substances of supercooling. Crude xylem extracts from katsura tree (Cercidiphyllum japonicum), of which XPCs exhibited deep supercooling to −40 °C, were prepared by methanol extraction. The crude extracts were purified by liquid–liquid extraction and then by silica gel column chromatography. Although all the fractions obtained after each purification step exhibited some levels of anti-ice nucleation activity, only the most active fraction was retained to proceed to the subsequent level of purification. High-performance liquid chromatography (HPLC) analysis of a fraction with the highest level of activity revealed four peaks with high levels of anti-ice nucleation activity in the range of 2.8–9.0 °C. Ultraviolet (UV), mass and nuclear magnetic resonance (NMR) spectra revealed that these four peaks corresponded to quercetin-3-O-β-glucoside (Q3G), kaempferol-7-O-β-glucoside (K7G), 8-methoxykaempferol-3-O-β-glucoside (8MK3G) and kaempferol-3-O-β-glucoside (K3G). Microscopic observations confirmed the presence of flavonoids in cytoplasms of XPCs. These results suggest that diverse kinds of anti-ice nucleation substances, including flavonol glycosides, may have important roles in deep supercooling of XPCs.
Trees that inhabit cold regions have developed the highest level of freezing resistance within the plant kingdom (Sakai & Larcher 1987). Among tissue cells in a tree, furthermore, the freezing adaptation mechanisms of the cells differ depending upon the type of tissues. While cells in cortical and cambial tissues adapt to sub-freezing temperatures by extracellular freezing, which is the most common freezing adaptation mechanism in herbaceous plant cells (Sakai & Larcher 1987), parenchyma cells of xylem tissues adapt to sub-freezing temperatures by so-called deep supercooling (Quamme, Weiser & Stushnoff 1973; Quamme, Chen & Gusta 1982; Ashworth et al. 1988; Malone & Ashworth 1991; Fujikawa & Kuroda 2000; Kuroda et al. 2003). These two adaptation mechanisms are similar in regard to avoid lethal intracellular freezing at sub-freezing temperatures when apoplast water has frozen, but the two avoidance mechanisms against intracellular freezing are essentially different.
In adaptation by extracellular freezing, when extracellular water has frozen at near sub-zero temperatures, water in living cells temporarily supercools. The difference between vapor pressures of extracellular ice and intracellular water results in dehydration of intracellular water to extracellular ice in direct parallel to temperature reductions (Steponkus 1984). Because of such equilibrium dehydration, cells are able to avoid lethal intracellular freezing. Extracellular freezing thus causes shrinkage and deformation of cells not only because of dehydration but also because of growth of large extracellular ice crystals (Pearce 1988; Fujikawa 1994). On the other hand, in adaptation by deep supercooling, cells are not dehydrated when apoplast water has frozen. Thus, xylem parenchyma cells (XPCs) adapt by supercooling to maintain a liquid state of intracellular water. XPCs of boreal trees are able to maintain this metastable supercooling state for a long period exceeding a few weeks at low sub-freezing temperatures, and such supercooling is therefore called deep supercooling to distinguish this supercooling from temporal supercooling (Quamme 1991). Deep supercooling XPCs are not shrunken or deformed even at low sub-freezing temperatures and retain their original morphology (Fujikawa, Kuroda & Fukazawa 1994).
Until recently, the mechanism of deep supercooling in XPCs has been explained only by physical isolation of water in XPCs (Ashworth & Abeles 1984), based on in vitro experiments on supercooling of small isolated water droplets (Fletcher 1970; MacKenzie 1977). Water in XPCs is isolated from the effects of extracellular ice because of the presence of specific cell walls that allow neither dehydration nor penetration of extracellular ice (George & Burke 1977; Quamme et al. 1982; George 1983; Ashworth & Abeles 1984). Thus, it has been suggested that XPCs as isolated water droplets can supercool to homogeneous ice nucleation temperature at about −40 °C and sometimes to lower temperatures by freezing temperature depression that results from high concentrations of solutions in the cells because of incomplete dehydration during supercooling (Gusta, Tyler & Chen 1983; Kuroda et al. 2003).
The isolation of protoplasts in XPCs from the effects of extracellular ice due to the presence of specific cell walls is undoubtedly a prerequisite for supercooling in XPCs, and it is also suggested that change of cell wall properties may be responsible for change of supercooling capability in XPCs (Wisniewski, Davis & Schaffer 1991). Recent studies, however, have indicated that the supercooling capability of XPCs is significantly changed by the release of intracellular substances and have suggested that a variety of intracellular substances may have important roles in supercooling of XPCs (Kasuga et al. 2006; Kasuga, Arakawa & Fujikawa 2007a).
Our most recent study has shown that crude ethanol extracts from the xylem of six boreal hardwood species, Japanese white birch (Betula platyphylla var. japonica), Japanese chestnut (Castanea crenata), katsura tree (Cercidiphyllum japonicum), Siebold's beech (Fagus crenata), mulberry (Morus bombycis) and Japanese rowan (Sorbus commixta), in which their XPCs adapt to sub-freezing temperatures by deep supercooling, exhibited anti-ice nucleation activity that promoted supercooling of water (Kasuga et al. 2007b). Among these six boreal hardwood species, crude xylem extracts of katsura tree exhibit the highest anti-ice nucleation activity. In the present study, therefore, we tried to identify anti-ice nucleation substances that might play an important role in the deep supercooling of XPCs from the crude xylem extracts of katsura tree.
MATERIALS AND METHODS
Samples were harvested from approximately 20-year-old katsura trees (Cercidiphyllum japonicum Sieb. et Zucc.) growing in the campus of Hokkaido University. Woody trunks, which had not developed heartwood, were harvested in December 2005 for providing crude xylem extracts. Four-year-old twigs from adult katsura trees were harvested in February 2008 for microscopic study.
Extraction of crude xylem extracts
After the bark and cambium had been removed from the discs of woody trunks with a knife, fresh xylem tissues were crushed into small chips using a band saw (BSW-200; Ryowa Co., Tokyo, Japan). Approximately 3.7 kg of fresh xylem chips was soaked in 6 L of methanol at room temperature for 1–2 weeks. Then, the xylem chips were removed by filtration on a filter paper (qualitative filter paper no. 2; Tokyo Roshi Kaisha, Tokyo, Japan). Residual xylem chips were soaked again in 6 L of methanol for 3 d, and the second methanol extracts were collected as described earlier. These procedures were duplicated. After methanol extracts had been concentrated by a rotary evaporator, the concentrated methanol solubles (94 g) were redissolved in 300 mL of MilliQ water (Millipore, Billerica, MA, USA). The resultant suspensions were centrifuged at 14 000 g for 10 min at room temperature, and the supernatants were used as crude xylem extracts.
Measurement of anti-ice nucleation activity
Anti-ice nucleation activity was estimated by ice nucleation (freezing) temperature of solutions. The ice nucleation temperatures of the solutions were analysed by a droplet-freezing assay described by Vali (1971) with some modifications. The extracts, glucose or synthetic flavonol glycosides, which were purchased from Extrasynthèse (Genay, France), were diluted in phosphate buffer solution (50 mm potassium phosphate, pH 7.0) containing 2 mg mL−1 of ultraviolet (UV)-sterilized and lyophilized Erwinia ananas (Wako Pure Chemical Industries, Osaka, Japan). Small droplets (2 µL) of these solutions were placed onto the surface of a copper plate coated with mineral oil (Nacalai Tesque, Kyoto, Japan). The copper plate was floated on a coolant in an alcohol bath (F26; Julabo Labortechnik, Seelback, Germany) maintained at 0 °C and immediately cooled at a rate of 0.2 °C min−1 to −30 °C. The frozen droplets with 0.5 °C temperature decrements were counted by naked eye during the cooling process. In each experiment, 120 droplets in total from three separate examinations were used (Vali 1971). The temperature required for freezing 50% of the droplets was indicated as ice nucleation temperature of 50% (INT50) of the droplets.
Fractionation of crude extracts and isolation of anti-ice nucleation substances
Aqueous solutions of crude xylem extracts (approximately 300 mL) were extracted with 600 mL of ethyl acetate (EtOAc) three times. Then, the EtOAc layer was dried over MgSO4 and concentrated by a rotary evaporator. The resulting EtOAc solubles (15 g) were subjected to silica gel column chromatography (750 g of Wakogel C-100, Wako Pure Chemical Industries) and were eluted with n-hexane : 2-propanol : water by decreasing the proportion of n-hexane (from 60:10:1 to 30:10:1, v/v/v). Among 20 fractions (each 500 mL) that were prepared by silica gel chromatography, the most active fraction (fraction J: 780 mg with n-hexane : 2-propanol : water = 40:10:1) was subjected to preparative high-performance liquid chromatography (HPLC). Preparative HPLC was performed using an octadecylsilane (ODS)-C18 column (Wakosil-II 5C18HG, Wako Pure Chemical Industries), and the column was eluted with methanol : water (2:3, v/v) at a flow rate of 1 mL min−1 at room temperature. Eluates were detected at UV 210 nm by a UV detector (875-UV; Jasco Corporation, Tokyo, Japan), and six fractions were collected on the basis of the HPLC profile. Eventually, 35, 10, 23, 96, 8 and 50 mg of isolates were obtained from fractions 1–6, respectively.
UV, mass and nuclear magnetic resonance (NMR) spectrometry
The UV spectra of the isolates in methanol were recorded on an Ultrospec 3300pro (GE, Buckinghamshire, UK). Fast atom bombardment (FAB) mass and high-resolution-FAB (HR-FAB) mass spectra in a negative mode were obtained with a JMS-SX102A (JEOL, Tokyo, Japan). The NMR spectroscopy, including 1H and 13C NMR spectroscopy, [1H, 1H] correlation spectroscopy (COSY), [13C, 1H] heteronuclear single-quantum coherence (HSQC) spectroscopy and [13C, 1H] heteronuclear multiple-bond correlation (HMBC) spectroscopy, was performed on an AMX-500 (Bruker, Karlsruhe, Germany) using chloroform-d (CDCl3) and benzene-d6 (C6D6). Chemical shifts are given in δ values (ppm) relative to trimethylsilane (TMS) as an internal reference.
Chemical modification of compounds and acidic hydrolysis
For acetylation of the chemicals, approximately 10 mg of each compound dissolved in 200 µL of methanol, 1 mL of acetic anhydride and 2 mL of pyridine was added to a test tube, and then the test tube was heated at 70 °C for 90 min. After completion of the reaction, the mixture was concentrated together with a 10-fold volume of toluene under low pressure. The obtained remnant was dissolved in a small volume of EtOAc and was dispensed onto a silica gel thin-layer chromatography (TLC) plate (silica gel 60 F254, 0.5 mm in thickness; Merck, Darmstadt, Germany). The TLC was developed in chloroform : methanol (50:1, v/v), and the main spot visualized by UV 254 nm radiation was scraped from the TLC plate. Acetylated compounds were extracted from the scraped silica gel powders by EtOAc.
Acidic hydrolysis of compound 3 was performed by heating a mixture of 0.5 mL ethanol, 0.5 mL of 2 N sulfuric acid and 5 mg of compound 3 at 80 °C for 3 h. After the reaction had been quenched by addition of 10 mL of MilliQ water, the mixture was extracted with EtOAc three times. The resulting EtOAc layer was dried under vacuum. The remaining aqueous layer was also dried after neutralization with NaOH. The obtained hydrolysed products were used for NMR and TLC analyses.
Radial sections of fresh xylem tissues were cut to a thickness of 60 µm by a steel blade mounted on a sliding microtome (Yamato Kohki, Saitama, Japan). The sections were stained with 0.5% (w/v) of diphenylboric acid 2-aminoethyl ester (DPBA) in a solution consisting of 10 mm potassium phosphate (pH 6.0), 10% (w/v) sucrose and 2% (v/v) dimethyl sulfoxide at room temperature for 15 min. Subcellular localization of flavonoids was made by epifluorescent microscopy. Blue light (450–490 nm)-excised fluorescence from 515 to 565 nm was observed by an Axiophot (Carl Zeiss, Oberkochen, Germany). Observation was also carried out by a confocal laser scanning microscope (LSM-410A, Carl Zeiss) using a 488 nm blue laser for excitation and 515–560 nm fluorescence for emission.
Anti-ice nucleation activity in crude xylem extracts from katsura tree
Crude xylem extracts were obtained by methanol extraction from xylem of katsura tree, and anti-ice nucleation activity of the extracts was determined by a droplet-freezing assay (Fig. 1). The ice nucleation temperature (INT50) of water droplets, which were made of buffer solution including E. ananas, was reduced from −5.2 °C without addition of xylem extracts to −6.8 °C with addition of xylem extracts at 1 mg mL−1 (reduction of 1.6 °C). In contrast, when glucose was added instead of xylem extracts to the droplet solution at the same concentration, the INT50 of water droplets was not significantly reduced at such a low concentration (Fig. 1). Thus, these results suggest that crude xylem extracts from katsura tree contain anti-ice nucleation substances, as already shown in our previous study using ethanol extracts (Kasuga et al. 2007b).
The effect of crude xylem extracts on INT50 was not changed by treatment of the extracts with boiling or by treatment with protease followed by boiling (data not shown). These results suggest that anti-ice nucleation substances in crude methanol extracts from xylem are not proteinous components.
Anti-ice nucleation activity in EtOAc and water fractions
The crude xylem extracts were separated by liquid–liquid extraction using EtOAc–water, and anti-ice nucleation activities of both fractions were measured. While the INT50 of water droplets containing glucose at 1 mg mL−1 was −5.2 °C, the addition of the same concentrations of extracts in the EtOAc fraction and in the water fraction reduced INT50 to −7.5 °C (reduction of 2.3 °C) and to −6.1 °C (reduction of 0.9 °C), respectively (Fig. 2). The results suggest that the two fractions contained different types of anti-ice nucleation substances. Because the EtOAc fraction exhibited a higher level of anti-ice nucleation activity than that of the water fraction, however, the EtOAc fraction was chosen for further purification.
Anti-ice nucleation activity in fractions separated by silica gel column chromatography
The EtOAc fraction was separated by silica gel column chromatography into 20 fractions (fractions A–T). Because fractions A–C were insoluble in water, these three fractions were not used for further examination. All of the remaining 17 fractions (D–T) showed distinct anti-ice nucleation activity in comparison to that of water droplets containing glucose at the same concentration (Fig. 3). Among the peaks with anti-ice nucleation activity, we can identify two distinct groups; one group was a sharp peak around fraction J, and the other group was a broad peak extending from fraction N to fraction R (Fig. 3). These results also indicate the presence of diverse kinds of anti-ice nucleation substances in extracts from the xylem. In order to find more effective anti-ice nucleation substances, we chose fraction J, which showed the highest level of activity, for further purification.
Anti-ice nucleation activity in peaks of HPLC
The preparative HPLC profile of fraction J showed many peaks (Fig. 4). Depending on the elution time, minor peaks that were eluted during the first 13 min were collected together (fraction 1), and the other five peaks (fractions 2–6) were separately isolated, and the anti-ice nucleation activity of each fraction was measured (Fig. 5). In this series of experiments, the control solution did not include glucose, because addition of glucose at a low concentration did not cause a significant difference in INT50 to that of the solution without glucose (Figs 1 & 2; see also Fig. 8in which similarity of supercooling capability in the presence and absence of glucose is shown). INT50 of the control solution in this series of experiments was −4.8 °C, which is a slightly higher temperature than that of the control solution in the previous experiments (−5.2 °C). Such minor variation is usual in the measurement of the metastable supercooling temperature of water. However, our repeated measurements clearly showed the effect of these extracts to promote supercooling of water, and Fig. 5 was chosen for the typical example.
Among fractions 1–6 (F1–F6 in Fig. 5), the addition of 1 mg mL−1 of fraction 1 isolates (F1 in Fig. 5) or fraction 5 isolates (F5 in Fig. 5) did not cause a noticeable reduction in INT50 in water droplets. In contrast, isolates from the other four fractions, fractions 2–4 (F2–F4 in Fig. 5) and fraction 6 (F6 in Fig. 5), showed clear anti-ice nucleation activity by addition of 1 mg mL−1 of these isolates. The magnitudes of reduction in INT50 caused by the addition of 1 mg mL−1 of isolates from each active fraction were 2.8 °C (from −4.8 to −7.6 °C) in fraction 2 (F2 in Fig. 5), 9.0 °C (from −4.8 to −13.8 °C) in fraction 3 (F3 in Fig. 5), 3.4 °C (from −4.8 to −8.2 °C) in fraction 4 (F4 in Fig. 5) and 4.0 °C (from −4.8 to −8.8 °C) in fraction 6 (F6 in Fig. 5). The magnitudes of reduction in INT50 were lowered by the addition of lower concentrations of these isolates but in a non-linear manner. The anti-ice nucleation activities became undetectable at 0.5 mg mL−1 of isolates from fraction 4 and at 0.1 mg mL−1 of isolates from fractions 2 and 6. On the other hand, isolates from fraction 3, which was the most effective fraction, retained a high level of anti-ice nucleation activity even at 0.1 mg mL−1 (magnitude of reduction in INT50: 8.2 °C). The activity of isolates from fraction 3 decreased at concentrations lower than 0.1 mg mL−1 but still existed at 0.05 mg mL−1 and became hardly detectable at 0.01 mg mL−1. These four fractions (F2–F4 and F6 in Fig. 5) with anti-ice nucleation activity, which may correspond to each compound, were named compounds 2, 3, 4 and 6, respectively.
Chemical structure elucidation of anti-ice nucleation compounds
In order to determine the chemical structure of compounds 2, 3, 4 and 6, UV, mass and NMR analyses were carried out. The UV spectra of compound 3 showed UVλmax (in methanol) at 267 and 367 nm, which is typical of flavonols (Mabry, Markham & Thomas 1970). The HR-FAB mass spectrum of compound 3 in a negative mode gave [M-H]- ion peak at m/z 447.0942 showing the formula of C21H19O11 (calculated 447.0927). The negative FAB mass spectrum of compound 3 showed [M-H-162]- fragmentation peaks at m/z 285, with 152% relative intensity from its [M-H]- ion peak. This fragmentation indicated a characteristic loss of a hexose sugar. Therefore, compound 3 was thought to be a flavonol monoglycoside.
We subsequently acetylated compound 3 in acetic unhydride/pyridine to give 3a, and the chemical structure of this derivative was analysed by one- and two-dimensional NMR spectroscopic analyses (1H NMR, 13C NMR, DEPT, 1H-1H COSY, HSQC and HMBC) ( Tables 1 & 2, Fig. 6). The 1H NMR spectrum of 3a in CDCl3 exhibited seven acetyl groups (-OCOCH3 × 7). Furthermore, a 1′,4′-disubstituted B-ring and a 5,7-distributed A-ring of the flavonol moiety were also visible (Table 1). 1H NMR coupling sequences of 3a, thus, indicate that the unmodified 3 is a hexose-conjugated flavonol. In fact, acid hydrolysis resulted in yielding kaempferol (3,5,7,4′-tetrahydroxyflavone) and d-glucose, both of which agreed with the authentic compounds in 1H NMR spectra and TLC analyses. The glycosylated position on kaempferol was determined by the HMBC analysis of 3a in C6D6. Its β-anomeric proton signal at δ 4.77 (d, J = 7.6 Hz, H-1″) was correlated with the C-7 carbon signal (δC 160.36) (Fig. 6). This cross peak undoubtedly evidenced that the glycosylated position of the flavonol was at C-7. In consequence, compound 3 was identified as kaempferol-7-O-β-glucoside (K7G, Fig. 7).
Table 1. 1H nuclear magnetic resonance (NMR) data for acetylated compounds 2 (2a), 3 (3a), 4 (4a) and 6 (6a)
2a (in CDCl3)
3a (in CDCl3)
3a (in C6D6)
4a (in CDCl3)
6a (in CDCl3)
1H NMR spectra were measured at 400 MHz. Chemical shifts are given in δ values (ppm) relative to trimethylsilane (TMS) as an internal reference. Proton signals were assigned on the basis of 1H-1H COSY, HSQC and HMBC.
Table 2. 13C nuclear magnetic resonance (NMR) data for acetylated compounds 3 (3a), 4 (4a) and 6 (6a)
3a (in CDCl3)
3a (in C6D6)
4a (in CDCl3)
6a (in CDCl3)
13C NMR spectra were measured at 100 MHz. Chemical shifts are given in δ values (ppm) relative to trimethylsilane (TMS) as an internal reference. Carbon signals were assigned on the basis of 1H-1H COSY, HSQC and HMBC.
With the same process of derivatization and spectroscopic analyses on compound 3, the chemical structures of other active compounds were also elucidated. Compound 2, possessing one more hydroxyl group (C21H19O12 for [M-H]- ion, found 463.0893), showed a 1′,3′,4′-trisubstituted B-ring agreeable to quercetin. In the HMBC spectrum, the β-anomeric proton signal of 2a (acetylated 2) at δ 5.60 exhibited a cross coupling with an olefinic carbon signal at approximate δC 137 (Fig. 6), characteristic of C-3 carbon on flavonol skeleton. Thus, compound 2 was characterized as quercetin-3-O-β-glucoside (Q3G, Fig. 7).
Compounds 4 and 6 had formulae of C22H21O12 (found 477.1038) and C21H19O11 (found 447.0958) for [M-H]- ion, respectively. In the same way, 4a (acetylated 4) and 6a (acetylated 6) showed cross couplings between their β-anomeric proton signals and C-3 olefinic carbons to have their 3-O-glucoside moiety (Fig. 6). Aglycone of 6 was identical to kaempferol, so the compound 6 was kaempferol-3-O-β-glucoside (K3G) which is a positional isomer of compound 3 (Fig. 7). On the other hand, aglycone of 4 had an additional methoxy group (at δ 3.97 in 4a) to be substituted to either C-6 or C-8 (Table 1). Cross couplings of sole aromatic proton signal at δ 6.79 (1H, s) with aromatic carbon signals assigned substitution of A-ring on 4a as 5,7-dihydroxy-8-methoxy moiety (Fig. 6). Therefore, compound 4 was identified as 8-methoxykaempferol-3-O-β-glucoside (8MK3G, Fig. 7) Thus, the chemical structure of all the active compounds were elucidated including stereochemistry.
Anti-Ice nucleation activity of synthetic flavonol glycosides
The anti-ice nucleation activities of synthetic Q3G, K7G and K3G were also investigated for comparison to the activity with compounds 2, 3 and 6, which were isolated from xylem extracts of the katsura tree. The synthetic Q3G, K7G and K3G may correspond to compounds 2, 3 and 6, respectively. The differences in INT50 between water droplets containing glucose at 1 mg mL−1 and water droplets containing synthetic Q3G, K7G and K3G at the same concentration were 3.0, 9.1 and 3.7 °C, respectively. The synthetic flavonol glycosides exhibited almost the same level of anti-ice nucleation activities as those shown by compounds 2, 3 and 6 (compare F2, F3 and F6 in Fig. 5 with Fig. 8). This result also confirmed that the anti-ice nucleation activities in the peaks (Fig. 5), which were separated by preparative HPLC (Fig. 4), originated from these flavonol glycosides, not from any other contaminations.
Localization of flavonoids in xylem tissue
Because our starting materials as crude xylem extracts may include not only extracts from XPCs but also extracts from all other xylem tissues, the localization of flavonoids in XPCs was confirmed in order to find the relationship with deep supercooling of XPCs. For this purpose, thick radial sections of xylem tissues were prepared, stained with DPBA and examined by epifluorescent microscopy or by confocal laser scanning microscopy. The presence of flavones and flavonols is evidenced by yellow-green fluorescence under UV light (Hutzler et al. 1998; Polster et al. 2006) or under blue light (Hutzler et al. 1998; Markham et al. 2000; Markham, Gould & Ryan 2001) after DPBA staining. In xylem tissues without staining, autofluorescence with a yellow-green colour was hardly observed (Fig. 9a,b). In contrast, DPBA staining resulted in bright fluorescence with a yellow-green colour in the protoplasts of XPCs but weak fluorescence in the cell walls (Fig. 9c,d). In the protoplasts of XPCs stained by DPBA, the entire cytoplasms showed bright fluorescence, while the center of the cytoplasms, which may correspond to central vacuoles, showed dark fluorescence (Fig. 9e,f). These results indicate significant accumulation of flavonoids in the cytoplasms of deep supercooling XPCs.
In the present study, we tried to identify anti-ice nucleation substances that might exist in deep supercooling XPCs and might be involved in deep supercooling of XPCs in trees. While the presence of anti-ice nucleation substances in deep supercooling XPCs has never been considered in previous studies, our recent study showed a strong possibility of the existence of such substances in several boreal hardwood species that contained deep supercooling XPCs (Kasuga et al. 2007b). Substances known as anti-ice nucleation substances can reduce the freezing point (Tf) of water by supercooling via a non-colligative effect (Holt 2003). Anti-ice nucleation substances can promote the supercooling capability of water at very low concentrations. The change in supercooling capability caused by such substances is not linear depending on solute concentration (Fig. 5). On the other hand, all of the substances (solutes), including glucose as a general solute that was used in this study as a reference substance, reduce the equilibrium melting point (Tm) linearly depending on solute concentration by a colligative effect. Furthermore, general solutes depress Tf of aqueous solution by about twofold more than the level of depression of equilibrium Tm (MacKenzie 1977). General solutes at 100 mosmol kg−1 depress Tm to −0.186 °C by a colligative effect (Sweeney & Beuchat 1993) and consequently reduce Tf to −0.372 °C. Glucose at a concentration of 1 mg mL−1, which was used in this study as a reference, corresponds to 5.6 mosmol kg−1 and, on the other hand, xylem extracts at a concentration of 1 mg mL−1 have far less osmolarity than that of glucose because of their higher molecular weight. In our system, thus, the effect of concentration of added substances to supercooling capability of water droplets can be neglected. Actually, the results of the present study indicate that supercooling capability was not significantly changed by addition of glucose at 1 mg mL−1 (Figs 1, 2 & 8).
The present study suggested that the xylem of katsura tree, which has been shown to have XPCs that undergo deep supercooling to −40 °C during winter (Kasuga et al. 2007b), is likely to contain many kinds of anti-ice nucleation substances other than those identified. The level of anti-ice nucleation activity of crude methanol extracts from xylem was 1.6 °C (Fig. 1). Separation of the crude xylem extracts by liquid–liquid extraction into two fractions provided higher anti-ice nucleation activity in the EtOAc fraction (level of anti-ice nucleation activity, 2.3 °C), but the water fraction also showed the activity (level of anti-ice nucleation activity, 0.9 °C) (Fig. 2), suggesting the presence of different types of anti-ice nucleation substances in the water fraction. Further separation of the EtOAc fraction by a silica gel column into 17 water-soluble fractions also showed that all of these 17 fractions contained high levels of anti-ice nucleation activity in the range of 1.4–5.8 °C (Fig. 3). The results also suggest the existence of further diverse kinds of anti-ice nucleation substances. Furthermore, only one fraction among the 17 fractions contained four kinds of compounds that exhibited high levels of anti-ice nucleation activity (Fig. 5). The possibility that the xylem of katsura tree with deep supercooling XPCs contains diverse kinds of numerous anti-ice nucleation substances is surprising, considering the small number of anti-ice nucleation substances so far reported (Table 3).
In order to identify anti-ice nucleation substances with higher levels of activity, we chose the fraction with the highest level of activity through each process of purification. We finally obtained one fraction (J) with the highest activity level separated by silica gel column chromatography, and the fraction was analysed by preparative HPLC (Fig. 4). The HPLC profile showed numerous peaks, but only four peaks showed anti-ice nucleation activity (Fig. 5). By analyses of these four peaks with UV, mass and NMR spectra (Tables 1 & 2, Fig. 6), we finally identified four kinds of flavonol glycosides, Q3G, K7G, 8MK3G and K3G, as novel anti-ice nucleation substances (Fig. 7, Table 3). This is the first study in which anti-ice nucleation substances from tree xylem have been identified and also the first study in which flavonol glycosides with anti-ice nucleation activity have been discovered. Compared with the anti-ice nucleation activity levels of previously reported substances, the anti-ice nucleation activity levels of flavonol glycosides from katsura tree are very high, and the level of activity by K7G (9.0 °C) is the highest among all previously reported anti-ice nucleation substances (Table 3). Among the previously reported anti-ice nucleation substances of which chemical structures have been determined, polyglycerol shows the highest level of activity (6.6 °C) at 1% (w/w) (Holt 2003), although the concentration of polyglycerol added to the medium in that study was much higher than the concentration used in the present study. Furthermore, among all previously reported substances, crude water extracts from seeds of trees and supernatant liquid from germinating legume seeds showed the highest levels of anti-ice nucleation activity in the range of 2.6–8.1 °C, although the causative substances are not identified and the concentrations of such crude extracts that were used for measurement were not indicated (Caple et al. 1983).
Although we identified the existence of flavonol glycosides with high levels of anti-ice nucleation activity in crude extracts obtained from the entire xylem tissue, we also confirmed that flavonoids existed in protoplasts of deep supercooling XPCs by histochemical observation using DPBA staining (Fig. 9). The accumulation of flavonoids in protoplasts including cytoplasms, vacuoles and nuclei has been shown by DPBA staining in epidermal and cortex cells in root tips of Arabidopsis (Saslowsky, Warek & Winkel 2005), in epidermal cells of broad bean leaves and of rye leaves (Hutzler et al. 1998), and in cells in needles of coniferous species, including Abies lasiocarpa, Cedrus deodara, Cedrus libani, Juniperus communis, Picea abies, Picea orientalis and Pseudotsuga menziesii (Polster et al. 2006). The results of the present study, however, indicate specific accumulation of flavonoids in cytoplasms except for vacuoles (Fig. 9f). The localization of flavoids in cytoplasms except for vacuoles has also been reported in epidermal cells in needles of Norway spruce (P. abies) (Hutzler et al. 1998) and in inner petal regions of Eustoma grandiflorum, Lathyrus chrisanthus and Dianthus caryophyllus after DPBA staining (Markham et al. 2001).
From the present results, it is assumed that the presence of flavonol glycosides with high levels of anti-ice nucleation activity in XPCs have an important role in the deep supercooling capability of XPCs. In support of this assumption, it has been reported that the amounts of anthocyanins increase in xylem tissue of European beech (Fagus sylvatica) during winter (Schmucker 1947), when the deep supercooling capability of XPCs becomes maximum. It has also been reported that expression of the WXL5 gene, which encodes flavonol-3-O-glucosyltransferase, in the xylem of larch (Larix kaempferi) increases significantly during winter, when the supercooling capability of XPCs becomes maximum, and decreases or disappears during summer or by deacclimation of winter twigs in parallel with significant reduction in supercooling capability of XPCs (Takata et al. 2007). Furthermore, the expression level of the WXL5 gene in larch is much higher in xylem with deep supercooling XPCs than in cortical tissues with extracellular freezing CPCs during winter (Takata et al. 2007). Additionally, it is also noted that recent study reports close association between increased accumulation of major flavonols and increased freezing tolerance by cold acclimation of leaf cells in Arabidopsis (Korn et al. 2008), although these cells adapt to sub-freezing temperatures by extracellular freezing.
Recent studies have reported that the supercooling capability of XPCs is closely associated with change in intracellular concentration of soluble sugars in Japanese white birch (B. platyphylla var. japonica) (Kasuga et al. 2007a), change in expression of proteins in xylem of peach (Prunus persica) and red osier dogwood (Cornus sericea) (Arora, Wisniewski & Scorza 1992; Arora & Wisniewski 1996; Sarnighausen, Karlson & Ashworth 2002), and change in gene expression in the xylem of larch (L. kaempferi) (Takata et al. 2007). Therefore, it is suggested that high supercooling capability in XPCs may be achieved by combination of these changes including accumulation of anti-ice nucleation substances.
Currently, together with the attempt to identify many other kinds of anti-ice nucleation substances that may exist in crude xylem extracts from katsura tree, we are analyzing seasonal changes in the accumulation of anti-ice nucleation substances, including flavonol glycosides, in XPCs as well as their tissue-specific accumulation in order to clarify the relationships of these anti-ice nucleation substances with the deep supercooling capability of XPCs.
The authors wish to thank Dr E. Fukushi and Mr K. Watanabe (GC-MS and NMR Laboratory, Graduate School of Agriculture, Hokkaido University) for mass and NMR measurements. The authors also thank Mr T. Ito (Electric Microscope Laboratory, Graduate School of Agriculture, Hokkaido University) for technical assistance in microscopic observation. This work was supported by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (19880002 to J.K.), a Grant-in-Aid for Scientific Research from the Ministry of Education, Sports, Culture, Science and Technology of Japan (17380101 to S.F.) and a Special Science Grant from Sekisui Integrated Research Co. Ltd. (to S.F.).