Similarity analysis between cultivars of litchi (Litchi chinensis Sonn.) leaf by structural characteristics of polysaccharides

Abstract

Litchi leaf has been used as a traditional Chinese medicine due to its unique pharmacological actions. There are more than two hundred of cultivars distributed over the world. Application of polysaccharide structural characteristics to analyse the similarity between cultivars is a novel and interesting attempt. In this work, litchi leaves (cv. ‘Nandao seedless fruit’, ‘Dingxiang’ and ‘Ziniangxi’) were selected to extract water-soluble polysaccharides (LLPs I and II). LLP I consisted of Ara, Rha, Gal and Glc, while LLP II was of Ara, Rha and Gal. However, the relative molar percentages of each monosaccharide between cultivars were different. The glycosidic linkages were determined by gas chromatography/mass spectrometry. Gal had two linkages of →3)-Gal-(1→ and →6)-Gal-(1 →. Ara and Rha were found to be linked as →5)-Ara-(1→ and →3)-Rha-(1→, while Glc appeared to be Glc-(1→ and →6)-Glc-(1 →. LLPs I and II belonged to type II arabinogalactans. The hierarchical clustering analysis of structural characteristics of LLPs I and II showed that ‘Nandao seedless fruit’ and ‘Dingxiang’ had a good similarity. This method was an efficient way to analyse the similarity between cultivars of litchi leaf.

---

Introduction

Polysaccharides, lignins and proteins construct the backbone of plant cell walls.¹ They also work as a natural defence against pathogen attack. In recent years, a great deal of attention has been paid to polysaccharides for their unique biological properties,² and useful application in drug development.³ The health effects of plant polysaccharides in the human diet include anticancer, immune modulation, anti-bacteria and anti-cardiovascular diseases.^4,5 Some plant polysaccharides have been commercially applied in drugs and functional foods.⁶

Litchi (Litchi chinensis Sonn.) is an exotic subtropical fruit, which is mainly planted in Southeast Asia and many other regions.⁷ It is desirable for consumers over the world due to the sweet and juicy pulp with beneficial health effects.⁸ Furthermore, the litchi leaf is also an important bioactive ingredient in traditional Chinese medicines, which has been used for treating dermatosis and wounds. Bersa, Sharma and Gomes have reported that the litchi leaf shows great anti-inflammatory, analgesic and anti-pyretic activities.⁹ The bioactive substances in litchi leaf are responsible for its pharmacological properties. However, litchi leaf with different cultivars might have different pharmacological effects. Similarity is an important definition in plant biology, which is often analysed by physical traits in plant taxonomy. The cultivars with high similarity should have similar bioactive functions. Identification of polysaccharide structure can also be a potential tool to analyse the similarity between cultivars, though it has been rarely applied before. Therefore, in this work, the water-soluble polysaccharides were extracted from litchi leaf. The monosaccharide composition and glycosidic linkage were investigated by gas chromatography/mass spectrometry (GC-MS). The similarity of litchi between cultivars was also evaluated by comparing polysaccharide structural characteristics.

Results and discussion

Chemical compositions of LLPs I and II

During the extraction, the ethanol was firstly used to remove ethanol-soluble substances of litchi leaf. LLPs I and II were sequentially obtained by hot water extraction and ethanol precipitation. The water-soluble polysaccharides with high molecular weight are precipitated in aqueous ethanol at low concentration, while those having low molecular weight can only be precipitated by high-concentration ethanol. Table 1 shows their chemical compositions and relative contents. All the LLP samples were determined with moisture content from 8.3% to 10.6%. The LLP I of ‘Ziniangxi’ had the highest yield of 12.0 mg g⁻¹ (dry weight), while LLP I of ‘Nandao seedless fruit’ had the lowest yield of 1.3 mg g⁻¹ (dry weight). The LLP II yield was in a decreasing order, ‘Nandao seedless fruit’ > ‘Dingxiang’ > ‘Ziniangxi’. Though Sevag reagent was used to remove proteins during extraction, small amounts of proteins could still be observed in the final products. LLP II of three cultivars showed a lower protein content than LLP I. It was hypothesized that most water-soluble proteins in litchi leaf were readily sedimented at 40% ethanol solution. Lignins are difficult to extract as a pure substance since they are commonly bound with hemicelluloses and cellulose through covalent linkages.¹⁰ Phenylpropyl and phenylethyl units bound by ether and carbon-carbon linkages through oxidative polymerisation provide the basic structure of lignins.¹¹ In this work, a small amount of lignins were found in all the LLP samples with content in the range of 1.6%–3.0%. The polysaccharide content was expressed as galactose equivalents, which were not significantly (p > 0.05) different between LLP samples, except for LLP I of ‘Dingxiang’.

Table 1 Chemical compositions (%, W W⁻¹) and yields (MG/G) of LLPs I and II^a

Samples		Chemical composition				Yield
		Protein	Lignin	Water	Polysaccharide
a The results having the same letters in one column are not significantly (p > 0.05) different.
‘Nandao seedless fruit’	LLP I	2.3 ± 0.2ac	3.0 ± 0.1a	8.5 ± 0.3a	80.5 ± 0.8ab	1.3 ± 0.1a
	LLP II	0.5 ± 0.1b	2.1 ± 0.2b	10.6 ± 0.5b	82.4 ± 1.2b	9.9 ± 0.5b
‘Dingxiang’	LLP I	1.8 ± 0.2a	3.3 ± 0.4a	8.3 ± 0.5a	78.7 ± 1.1a	2.2 ± 0.1c
	LLP II	0.7 ± 0.1b	1.8 ± 0.2bc	9.7 ± 0.2c	81.9 ± 0.9b	8.5 ± 0.5d
‘Ziniangxi’	LLP I	2.6 ± 0.1c	2.7 ± 0.1a	8.6 ± 0.6a	81.4 ± 0.5b	12.0 ± 0.3e
	LLP II	0.6 ± 0.1b	1.6 ± 0.2c	9.9 ± 0.5bc	81.9 ± 1.1b	5.9 ± 0.3f

---

Monosaccharide compositions of LLPs I and II

Through comparing the retention time with external standards, the monosaccharide compositions of LLPs I and II were identified (Table 2). Four monosaccharides, including Ara, Rha, Glc and Gal, were found in LLPs I of three cultivars, while LLPs II of three cultivars consisted of Ara, Rha and Gal with the absence of Glc. The relative molar percentages were calculated by internal standard method. As shown in Table 2, though the monosaccharide composition of LLPs I (or II) of three cultivars were the same, the relative molar percentage of each monosaccharide between cultivars were apparently different. The sums of Ara and Gal in all the LLP samples were approximately 80%. The results indicate that Ara constructed the backbone in combination with Gal.

Table 2 The monosaccharide compositions of LLPs I and II.^a

Samples		Ara	Rha	Gal	Glc
a The results having the same letters in one column are not significantly (p > 0.05) different. b not detected.
Nandao seedless fruit	LLP I	42.6 ± 0.6a	13.3 ± 0.3a	33.2 ± 0.2a	10.9 ± 0.3a
	LLP II	62.6 ± 0.3b	10.5 ± 0.2b	26.9 ± 0.1b	—^b
Dingxiang	LLP I	52.0 ± 0.4c	11.9 ± 0.3c	30.7 ± 0.5c	5.4 ± 0.2b
	LLP II	63.3 ± 0.5b	10.5 ± 0.2b	26.2 ± 0.6b	—
Ziniangxi	LLP I	40.5 ± 0.3d	7.2 ± 0.3d	45.5 ± 0.5d	6.8 ± 0.1c
	LLP II	46.8 ± 0.6e	10.3 ± 0.2b	42.9 ± 0.3e	—

---

Water-soluble polysaccharides are an important component of plant cell walls, which play a critical role in controlling the shape of the cell and physiological functions, like intercellular communication and interaction with the environment.¹² No uronic acid was observed in the monosaccharide composition, which indicated that LLPs I and II were neutral polysaccharides. Ara and Gal were the main monosaccharides with proportions of about 80% in LLP. This suggests that LLPs I or II were arabinogalactan, which could be found in some other botanical sources. Mateos-Aparicio, Redondo-Cuenca and Villanueva-Suárez have pointed out that a polysaccharide of legume byproduct consists of arabinose (22%) and galactose (37%).¹³ 30.8% of Arabinose and 20.8% of galactose have also been detected in non-starch polysaccharides of finger millet.¹⁴ In this work, rhamnose was a minor monosaccharide constituent of both LLPs I and II. Though this monosaccharide also constructed the polysaccahride backbone in some botanical species. For example, Zhao et al. have shown that 53.8% of rhamnose occurs in a water-soluble polysaccahride fraction of Opuntia monacantha cladodes.¹⁵ However, in leaves of many plant species, rhamnose often occurs as a minor monosaccharide, like Yanang and Taxus chinensis var. mairei.^16,17 From the results obtained above, Glc was only observed in LLP I of three cultivars, and it was not detectable in three LLP II samples. This behaviour revealed that Glc only participated in the construction of LLP chain with high molecular weight.

Glycosidic linkages of LLPs I and II

The glycosidic linkages of LLPs I and II were determined by methylation method and analysed by GC-MS. As shown in Table 3, the glycosidic linkages of LLP I were in coincidence with those of LLP II. Gal had two linking forms, which were →3)-Gal-(1→ and →6)-Gal-(1→, respectively. Ara and Rha were found to be linked as →5)-Ara-(1→ and →3)-Rha-(1→, while Glc was appeared to be Glc-(1→ and →6)-Glc-(1 →. However, the relatively molar percentage of each linkage was different between LLPs I and II. →6)-Gal-(1→ had a relatively molar percentage of 30.0% for LLP I of ‘Ziniangxi’, whereas it was only 19.6% in LLP I of ‘Dingxiang’.

Table 3 Relative molar percentages of the glycosidic linkages of LLPs I AND II

Samples	‘Nandao seedless fruit’		‘Dingxiang’		‘Ziniangxi’
	LLP I	LLP II	LLP I	LLP II	LLP I	LLP II
→5)-Ara-(1→	41.8	60.6	53.8	61.5	40.8	44.5
→3)-Rha-(1→	13.0	11.3	12.6	10.5	8.6	11.3
→3)-Gal-(1→	11.2	8.6	9.1	8.2	14.6	13.5
→6)-Gal-(1→	23.5	19.5	19.6	19.8	30.0	30.7
Glc-(1→	3.3	—	2.6	—	3.1	—
→6)-Glc-(1→	7.2	—	2.3	—	2.9	—

---

The existence of →5)-Ara-(1→ indicated that Ara appeared as furanose form, and the detection of Glc-(1→ confirmed that it was located at a non-reduced terminal position. Further analysis of glycosidic linkage by GC-MS indicated that LLPs I and II had similar linkage types, but significant differences in the relative molar percentages. Moreover, no more than two positions of each monosaccharide residue were detected to be linked with other monosaccharides, which indicated that LLP I or II consisted mainly of linear chains. The glycosidic linkage of →5)-Ara-(1→ was widely distributed in neutral polylsaccharide of plant sources, such as the main chain of neutral polysaccharides from Ginkgo biloba leaf.¹⁸ As another major constituent of LLP, Gal had two linkages, →3)-Gal-(1→ and →6)-Gal-(1 →. Such forms were also observed in polysaccharides extracted from lamina and midrib of rosette chicory leaf by hot water.¹⁹ Arabinogalactan can be subdivided into two main structural types: type I, Ara in combination with →4)-Gal-(1→; type II, Ara in combination with →6)-Gal-(1→ and →3)-Gal-(1 →.²⁰ The type II arabinogalactans often occur as part of pectic substances or glycoproteins, which have been reported with good bioactivities, like immunomodulation.²¹ The results obtained in this work showed that LLPs I and II belonged to type II arabinogalactans. They might have a positive effect on the pharmacological properties of litchi leaf. Arabinogalactan is an important pectin fraction, though pectins are usually galacturonic acid-rich polysaccharides.²² However, no galacturonic acid was detected in LLP. It indicated that LLP could not be defined as pectins.

Hierarchical clustering analysis

The monosaccharide compositions of all six LLP samples were compared by hierarchical clustering analysis. The dendrogram could tell the similarity between samples. As shown in Fig. 1, LLPs II between ‘Nandao seedless fruit’ and ‘Dingxiang’ had the shortest euclidean distance. It revealed the closest similarity between them, followed by LLP I/II of ‘Ziniangxi’ and LLPs I between ‘Nandao seedless fruit’ and ‘Dingxiang’. The close similarity between LLP I (and II) of ‘Nandao seedless fruit’ and ‘Dingxiang’ confirmed the good similarity of these two cultivars. The large euclidean distance of LLP I (or II) of ‘Ziniangxi’ to that of other two cultivars suggested that ‘Ziniangxi’ had a relatively poor similarity to ‘Nandao seedless fruit’ and ‘Dingxiang’. By observing the apparent traits of litchi leaves of three cultivars, ‘Nandao seedless fruit’ and ‘Dingxiang’ had similar size and surface glossiness. Their sizes were larger than ‘Ziniangxi’, and the surface glossiness is better than ‘Ziniangxi’. These physical traits further confirmed that ‘Nandao seedless fruit’ and ‘Dingxiang’ had a good similarity.

Fig. 1 The dendrogram of the monosaccharide compositions for all the LLP samples. (1) LLP I of ‘Nandao seedless fruit’; (2) LLP II of ‘Nandao seedless fruit’; (3) LLP I of ‘Dingxiang’; (4) LLP II of ‘Dingxiang’; (5) LLP I of ‘Ziniangxi’; (6) LLP II of ‘Ziniangxi’.

Multivariate statistical techniques, like hierarchical clustering analysis, are the right tools for viewing and analysing a matrix of complex data. Hierarchical clustering analysis mathematically treats each variable as a point in the multidimensional space described by the samples.²³ When a given sample is taken as a point in the space defined by the variables, one can calculate the distance between this point and all the other points, thereby establishing a matrix that describes the proximity between all the samples investigated. This analysis is an efficient way of grouping the samples to evaluate the similarity between cultivars. Though the monosaccharide compositions of LLPs from three cultivars occurred closely, only ‘Nandao seedless fruit’ and ‘Dingxiang’ had good similarity. The hierarchical clustering analysis was proven to be an efficient tool for similarity identification between plant species.

Experimental

Plant materials

Fresh litchi (Litchi chinensis Sonn.) leaves were collected from a local orchard in Danzhou, Hainan province of China. Three cultivars, namely ‘Nandao seedless fruit’, ‘Dingxiang’ and ‘Ziniangxi’, were selected on the basis of uniformity of shape and green colour. The leaves were dried by sun. Then they were pulverized by a miller and screened through a 60-mesh iron sieve.

Chemicals

Standards of xylose (Xyl), arabinose (Ara), rhamnose (Rha), glucose (Glc), galactose (Gal), fructose (Fru), Fucose (Fuc), mannose (Man), galacturonic acid (GalA) and glucuronic acid (GlcA) were purchased from Sigma Chemical Co (St. Louis, MO, USA). All the other chemicals used were of analytical grade.

Preparation of litchi leaf polysaccharides (LLP)

The litchi leaf powder (10 g) was immersed into 100 ml of absolute ethanol. The extraction process was performed in a water bath shaker at 50 °C for 1 h. The programme was repeated twice. The extract was filtered and the residues were added into 100 ml of distilled water. The slurry was kept at 50 °C for 2 h. Then the extract was filtered and the residues were subjected to extraction twice. The filtrates were combined and concentrated to 30 ml with a rotary evaporator at 65 °C under vacuum. The proteins were removed from the extract using the Sevag reagent.²⁴ After removal of the Sevag reagent, anhydrate ethanol was added to a final concentration of 40% (v/v), then the mixture was kept in a conical flask overnight at 4 °C and centrifuged at 8000 × g for 20 min to precipitate LLP I. More anhydrate ethanol was added to the supernatant to a final concentration of 80% (v/v). It was kept overnight at 4 °C. The LLP II was obtained by centrifugation at 8000 × g for 20 min. All the samples were lyophylized.

The content of polysaccharides was determined by the phenol-sulfuric acid method.²⁵ Galactose was used to construct a standard curve. The recovery of LLP was expressed as mg of gal equivalents per gram of litchi leaf on dry weight basis. The protein content was converted from the nitrogen content (N × 6.25), which was determined by a PE-2400 series II automatic elemental analyser (PerkinElmer, Waltham, Massachusetts, USA). The lignin content was determined by the method of Cajuste and Lafuente.²⁶ The moisture content was determined by AOAC (1997).²⁷

Analysis of monosaccharide composition

LLP I or II (10 mg) was hydrolysed by 10 ml of 2 mol L⁻¹ trifluoroacetic acid at 100 °C for 4 h.²⁸ Derivatization of the released monosaccharides was then carried out by trimethylsilylation reagent according to the method of Guentas et al.²⁹ The trimethylsilylated derivatives were loaded onto a GC-2010 gas chromatography system (Shimadzu, Kyoto, Japan) equipped with a RTX-5 capillary column and a flame ionization detector. The following programme was adopted for gas chromatography analysis: injection temperature: 230 °C; detector temperature: 230 °C; column temperature programmed from 130 to 180 °C at 2 °C min⁻¹, holding for 3 min at 180 °C, then increasing to 220 °C at 10 °C min⁻¹ and finally holding for 3 min at 220 °C. Nitrogen was used as the carrier gas and maintained at 40.0 ml min⁻¹. The speed of air and hydrogen gas were 400 and 40 ml min⁻¹, respectively. A splitless mode was chosen. Inositol was used as the internal standard to quantify the monosaccharide content.

Methylation analysis

Methylation of LLP I (or II) was carried out using the method of Yang et al. with minor modification.³⁰ Five milligrams of dry LLP I (or II) were weighted precisely and dissolved in 5.0 ml of DMSO before 200 mg of NaOH was added. The mixture was then treated by ultrasonic wave (120 w, 40 KHz, 25 °C) for 5 min. After incubation for 1 h at room temperature (25 °C), 1.5 ml of methyl iodide was added for methylation. The sample was kept in the dark for 1 h before 4.0 ml of distilled water was used to decompose the remaining methyl iodide. The methylated polysaccharides were extracted by 3 × 2 ml of chloroform and dried at low pressure by a rotary evaporator (RE52AA, Yarong Instrument Co, Shanghai, China). After hydrolysis by 10 ml of 2 M trifluoroacetic acid, the hydrolysates were dissolved in 4 ml of 10% (w/w) NaOH. Twenty milligrams of NaBH₄ were added to reduce hemiacetal bond. After incubation at 40 °C for 30 min, 2 ml of glacial acetic acid was used to terminate the reduction. The sample was dried under low pressure, and then acetylated by 2 ml of acetic anhydride and 2 ml of pyridine. The reaction was kept at 100 °C for 1 h. Two millilitres of distilled water were used to decompose the remained acetic anhydride. The acetylated derivatives were extracted by 4 ml of methylene chloride. A gas chromatography/mass spectrometer (GCMS-QP 2010, Shimadzu, Kyoto, Japan) was used to analyze the glycosidic linkage. The acetylated derivatives were loaded into an RTX-5 capillary column (Shimadzu, Kyoto, Japan). The temperature program was set as follows: the initial temperature of column was 150 °C, increased to 180 °C at 10 °C min⁻¹, then to 260 °C at 15 °C min⁻¹, held for 5 min at 260 °C; injection temperature: 260 °C. The ion source of mass spectrometer was set at 260 °C. One microlitre of sample was injected.

Statistical analyses

Data were expressed as mean ± standard deviation of three replicated determinations. One way of variance analysis was applied to determine the significant difference at p < 0.05. Hierarchical cluster analysis was employed to evaluate the similarity of litchi between cultivars by comparing monosaccharide composition. Proximity matrix and agglomeration schedule was checked for statistics. The dendrogram similarity scale ranges from 0 to 25. The similarities between the analysed samples were present in the dendrogram for each LLP sample. Between-group linkage was chosen for clustering method and the interval was Euclidean distance.³¹

Conclusions

The water-soluble polysaccharides of litchi leaves (cv. ‘Nandao seedless fruit’, ‘Dingxiang’ and ‘Ziniangxi’) exhibited similar structural characteristics. LLP I consisted of Ara, Rha, Gal and Glc, while LLP II was of Ara, Rha and Gal. The glycosidic linkage analysis revealed that the occurrence of →3)-Gal-(1→ and →6)-Gal-(1→, →5)-Ara-(1→ and →3)-Rha-(1→ in LLPs and II. Moreover, Glc appeared to be Glc-(1→ and →6)-Glc-(1→ in LLP I. The hierarchical clustering analysis indicated that ‘Nandao seedless fruit’ and ‘Dingxiang’ had a good similarity. This method can also be used for similarity analysis instead of apparent traits.

Acknowledgements

The financial support provided by the National Facilities and Information Infrastructure for Science and Technology of China (Nos. 2003DIA3N022, 2004DKA30420 and 2005DKA21005) was appreciated.

References

1. R. C. Sun, X. F. Sun and J. Tomkinson, ACS Symp. Ser., 2004, 864, 2.
2. I. Schepetkin and M. Quinn, Int. Immunopharmacol., 2006, 6, 317.
3. X. Li, A. Zhou and Y. Han, Carbohydr. Polym., 2006, 66, 34.
4. A. M. Deters, C. Lengsfeld and A. Hensel, J. Ethnopharmacol., 2005, 102, 391.
5. Y. Sonoda, T. Kasahara, N. Mukaida, N. Shimizu, M. Tomoda and T. Takeda, Immunopharmacology, 1998, 38, 287.
6. A. Deters, A. Dauer, E. Schnetz, M. Fartasch and A. Hensel, Phytochemistry, 2001, 58, 949.
7. B. Yang, M. M. Zhao, J. Shi, N. Yang and Y. M. Jiang, Food Chem., 2008, 206, 685.
8. N. Rangkadilok, S. Sitthimonchai, L. Worasuttayangkurn, C. Mahidol, M. Ruchirawat and J. Satayavivad, Food Chem. Toxicol., 2007, 45, 328.
9. S. E. Besra, R. M. Sharma and A. Gomes, J. Ethnopharmacol., 1996, 54, 1.
10. B. Yang, Y. M. Jiang, R. Wang, M. M. Zhao and J. Sun, Food Chem., 2009, 112, 428.
11. L. C. da Cunha, L. Serve, F. Gadel and J. L. Blazi, Org. Geochem., 2001, 32, 305.
12. D. J. Cosgrove, Annu. Rev. Cell Dev. Biol., 1997, 13, 171.
13. I. Mateos-Aparicio, A. Redondo-Cuenca and M. J. Villanueva-Surez, Food Chem., 2010, 122, 339.
14. M. V. S. S. T. S. Rao and G. Muralikrishna, Food Chem., 2001, 72, 187.
15. M. M. Zhao, N. Yang, B. Yang, Y. M. Jiang and G. H. Zhang, Food Chem., 2007, 105, 1480.
16. J. Singthong, S. Ningsanond and S. W. Cui, Food Chem., 2009, 114, 1301.
17. M. Wu, Y. Wu, J. Zhou and Y. Pan, Food Chem., 2009, 113, 1020.
18. J. Kraus, Phytochemistry, 1991, 30, 3017.
19. X. Sun, I. G. Andrew, K. N. Joblin, P. J. Harris, A. McDonald and S. O. Hoskin, Plant Sci., 2006, 170, 18.
20. M. M. Huisman, L. P. Brüll, J. E. Thomas-Oates, J. Haverkamp, H. A. Schols and A. G. J. Voragen, Carbohydr. Res., 2001, 330, 103.
21. Q. Dong and J. N. Fang, Carbohydr. Res., 2001, 332, 109.
22. H. S. Yang, H. J. An, G. P. Feng, Y. F. Li and S. J. Lai, Eur. Food Res. Technol., 2005, 220, 587.
23. E. A. F. S. Torres, M. L. Garbelotti and J. M. M. Neto, Food Chem., 2006, 99, 622.
24. L. Navarini, R. Gilli and V. Gombac, Carbohydr. Polym., 1999, 40, 71.
25. M. Dubois, K. A. Gilles, J. K. Hamilton, P. A. Rebers and F. Smith, Anal. Chem., 1956, 28, 350.
26. J. F. Cajuste and M. T. Lafuente, Postharvest Biol. Technol., 2007, 45, 193.
27. AOAC. Maryland: AOAC International, 1997.
28. B. Erbing, P. E. Jansson, G. Widmalm and W. Nimmich, Carbohydr. Res., 1995, 273, 197.
29. L. Guentas, P. Pheulpin, P. Michaud, A. Heyraud, C. Gey, B. Courtois and J. Courtois, Carbohydr. Res., 2001, 332, 167.
30. B. Yang, Y. M. Jiang, M. M. Zhao, F. Chen, R. Wang, Y. L. Chen and D. D. Zhang, Food Chem., 2009, 115, 609.
31. E. A. F. S. Torres, M. L. Garbelotti and J. M. M. Neto, Food Chem., 2006, 99, 622.

---