Tannin fractions of almonds, walnuts and hazelnuts kernels, obtained using Sephadex LH-20 column chromatography, were characterised by colour reaction with vanillin/HCl reagent enabling determination of condensed tannins content and by the BSA precipitation method allowing evaluation of the protein precipitation capacity of tannins. The results of these assays are shown in Table 1. The content of condensed tannins was 1,261 and 776 mg CE/g for the hazelnut and almond tannin fractions, respectively, whereas the result of this assay for fraction of walnut was several times lower. On the other hand, protein precipitation capacities of hazelnut and walnut tannin fractions were comparable: 940 and 873 mg TAE/g, but a more than tenfold lower value was noted for the almond tannin fraction. The content of condensed tannins in the fraction isolated from hazelnut kernels was more than twice as low as the value reported by Alasalvar et al. [10] for a fraction isolated from hazelnut skins. Gu et al. [24] determined proanthocyanidins present in nuts seeds, among others, using HPLC-MS/MS analysis. They noted that total proanthocyanidin contents for nuts included in the present study decreased in the order hazelnuts > almonds ≫ walnuts.

The obtained results indicated that tannin fraction from walnuts was characterised by its low proanthocyanidin content, but simultaneously suggested that it contained hydrolysable tannins, which do not react with vanillin/HCl reagent but are able to precipitate BSA. The presence of hydrolysable tannins (mainly ellagitannins) in walnuts was reported by Fukuda et al. [13] and Ito et al. [14]. High molecular weight hydrolysable tannins are insoluble and/or covalently bound to the cell wall. However, Li et al. [6] reported, that more than half of ellagitannins (determined as ellagic acid liberated after acidic hydrolysis) present in walnuts seeds was not bound and are thus extractable with 80% (v/v) methanol.

The tannin fraction from almonds comprised constituents which react with vanillin/HCl reagent, but precipitate BSA to a small extent, since they are probably mainly condensed tannins with low degrees of polymerisation. Dimers of condensed tannins are less effective precipitating agents [25]. Also oligomers containing three subunits are not always precipitated by proteins [26]. The low degree of polymerisation of the proanthocyanidins of the almond fractions was confirmed by SE-HPLC analysis (Figure 1a). In the chromatogram of this fraction a peak with a retention time (t_r) of about 57 min was predominant. This peak was related to the molecular weight of dimer-procyanidin B₂ and higher (Figure 1d). Oligomers of molecular weights larger than that of procyanidin B₂, but smaller than tannic acid were present also in almond fraction. On the other hand, in the chromatogram of the hazelnut fraction (Figure 1b) a highest peak with shorter t_r of 55 min was noted. The UV spectra of the discussed compounds are similar for both analysed fractions, with maxima at 281 nm (Figure 2a and 2b).

In the SE-HPLC chromatogram of the walnut tannin fraction two separated peaks are present (Figure 1c). The peak with retention time of 58 min, corresponding to z molecular weight of about 600 is predominant. The peak with shorter t_r of 56 min originated from constituents of molecular weight larger than procyanidin B₂, but smaller than that of tannic acid. UV spectra recorded at those retention times did not reveal any maxima and were characterised by shoulders only (Figure 2c and 2d). Then, hydrolysable tannins, which are predominant in the tannin fraction of walnuts, are molecules of molecular weights corresponding to strictinin (ellagitanin) or trigalloyl derivatives (gallotannin) and larger. To our knowledge, there is no literature data about the precipitating ability of low molecular weight ellagitannins with proteins. In case of gallotannins the smallest esters of glucose able to precipitate with BSA are molecules with at least three galloyl groups [27]. It seems that the smallest gallotannin able to precipitate BSA has a molecular weight of 636.5 (β-1,3,6-tri-O-galloyl-d-glucose). On the other hand, the condensed tannins molecules with molecular weight about 600 do not precipitate proteins [25]. This provides an explanation for the high protein precipitation capacity of the tannin fraction from walnuts, which differentiates them from almond tannins, despite the comparatively low molecular weights of the constituents of both fractions.

Chromatograms of SE-HPLC revealed that all the studied nut tannin fractions were characterised by the presence of constituents with lower molecular weights than the tannin fractions from buckwheat seeds and groats isolated and separated under analogous conditions [28].

The Cu(II) and Zn(II) chelating capacity of tannins from the examined nuts was determined by the assay with tetramethylmurexide (TMM). “Free” metal ions, which were not bound by tannins, were complexed with TMM. TMM solution showed absorption maximum at 530 nm, and complex formed of TMM with Cu(II) or Zn(II) at 482 nm and 462 nm, respectively. The ratio of absorbance measured at both wavelengths allows one to estimate the amount of Cu(II) or Zn(II) complexed with TMM. Knowing the total quantity of metal ions added to the reaction mixture, the % of Cu(II) or Zn(II) bound by the test tannin fraction can be calculated by difference. Copper ions were effectively chelated by the tannin constituents of all three analysed samples (Figure 3). At a 0.2 mg/assay addition level, the walnut tannins complexed almost 100% Cu(II). The same quantity of hazelnut and almond tannins bound only 72.3% and 54.5% Cu(II), respectively. Above 90% Cu(II) was chelated by hazelnut and almond tannins at the addition levels of 0.6 and 0.8 mg/assay, respectively. The greater addition of tannins tested (1 mg/assay) bound 93.5% and 94.3% copper ions. The same level of chelating of copper ions was noted for tannin fractions of buckwheat seeds [28]. Wong et al. [29] tested Cu(II) chelating ability of extracts of 25 edible tropical plants and established that they complexed from 40 to 95% of copper ions added.

The capacity to chelate Zn(II) was quite varied for the different nut tannins (Figure 4). Tannin fraction from almonds bound zinc ions much more effectively than the other two. At the lowest addition level applied - 1 mg/assay, the % of Zn(II) chelated amounted to 28.4%, and at a 5 mg/assay addition level tannins bound as much as 84% Zn(II). However, the value was only 8.7% for walnut tannins at the highest level of sample addition. Hazelnut tannins virtually did not complex Zn(II) ions; the greatest addition of ligand bound less than 1% of zinc ions. For comparison, buckwheat seeds and groats tannins complexed Zn(II) ions much weaker than the analysed fraction from almonds, but several percent stronger than walnut tannins [28], whereas 0.1% solution of instant coffee chelated 100% Zn(II) at pH 6.0 and about 75% at pH 5.0 [30].

The TMM method should not be used to evaluate the iron ion chelation capacity of tannins and plant extracts rich in tannin constituents [31]. Therefore, the method with ferrozine was employed for examination of Fe(II) chelating capacity of the tannin fractions of the tested nuts. Ferrozine forms with bivalent iron ions stable, colourful complexes with a high extinction coefficient at 562 nm. Figure 5 depicts the Fe(II) chelation capacity of the tannin fractions of almonds, hazelnuts and walnuts. The addition of 0.5 to 2.5 mg of tannin fraction per assay caused the increase of the amount of bound Fe(II). The largest increase was noted for almond tannins, then walnuts tannins and the least for hazelnuts. A 2.5 mg sample of tannins gave complexation of 90.1, 63.4 and 52.2% of the total quantity of Fe(II) added to the reaction mixture, respectively. The good Fe (II) chelating properties of almond tannins confirmed the results obtained by Wijeratne et al. [4]. They revealed that 200 ppm of extract from whole seed and brown skin of almonds complexed 97% and 98%, respectively from 400 μM of ferrous ions. On the other hand buckwheat seeds and groats tannins, obtained in the same manner as in the present study, chelated slightly less Fe(II), 53% and 24%, respectively [28].

In order to compare the ability of tannins fractions to chelate copper, iron and zinc ions, the amount of μmol of Fe(II), Cu(II) and Zn(II) was complexed by a constant amount (0.5 mg) of nut fraction tested was estimated (Table 2). Tannin fractions from nuts chelated copper ions the most effectively. The highest amount of Cu(II) was chelated by tannins from walnuts (25.0 μmol), and the lowest by tannins from almonds (20.5 μmol). Iron ions were complexed much weaker than copper ions. The ability of tannin fractions to chelate iron ions was in the range from 1.8 to 3.2 μmol Fe(II). Zn (II) was bound by tannins from walnuts and hazelnuts the least effectively. Only the almonds fraction chelated more zinc than iron ions. The chelating effectiveness of tested tannin fractions decreased in the following order: Cu(II) > Fe(II) > Zn(II), what was in line with other researchers’ results. Mira et al. [32] compared the degree of complexation of Cu(II) and Fe(II) by flavones and noted, that their ability to chelate copper ions was higher. Kumamoto et al. [33] determined acid dissociation constants of free and metal complexed four catechins. Binding with Cu(II) and Fe(II) caused much greater decrease in acid dissociation constant of catechins than complexation with Zn(II). Zinc ions were found to be precipitated by tannins to a much weaker extent than copper ions [34].

A number of studies on chelation of metal ions, especially copper and iron, by isolated phenolic compounds i.e., phenolic acids, flavonoids, anthocyanins have established that suitably oriented functional groups in the structure of ligand are essential for formation of metal ion–phenolic compound complexes [32,35–37]. Phenolic compounds with a single OH group on the aromatic ring do not bind copper and iron ions [36,38]. The presence of a catechol group (o-dihydroxyphenyl) or galloyl group (trihydroxyphenyl) is essential to complex metal ions. At the same time, iron ions preferentially bind to three phenol groups of the gallate moiety, and o-dihydroxyphenyl groups play a crucial role in complex formation with copper [35,38,39]. When the chemical structure of tannins is considered, it could be presumed that the tannin fraction from walnuts, containing mainly hydrolysable tannins, would better chelate Fe(II) ions, than tannin fractions from almonds and hazelnuts, rich in proanthocyanidins. In turn Cu(II) should be much better bound by the numerous catechol groups of condensed tannins of almonds and hazelnuts. Such simple relationships were not noted in the case of tannin fractions tested, probably due to much more complex structure of their constituents. For example it is known, that glycosylation of OH group of phenol prevents metal from binding [40]. Almonds and hazelnuts proanthocyanidins contain (epi)catechin glycosides as subunits [24], what can diminish their ability to complex Cu(II). On the other hand, the presence of methoxyl and hydroxyl groups in the ortho position increases Cu(II) chelation capacity [37]. It was also noted that structure does not influence complexing of iron ions [38].

The raw seeds of hazelnuts, walnuts and almonds with their skins on were acquired at local market in Olsztyn. Poland.

Sephadex LH-20, vanillin, tannic acid, gallic acid, (+)-catechin, BSA (bovine serum albumin), hexamine (hexamethylenetetramine), TMM (tetramethylmurexide ammonium salt), ferrozine (3-(2-pyridyl)-5,6-diphenyl-1,2,4-triazine-4’,4”-disulfonic acid) were purchased from Sigma-Aldrich Co. Ltd. (Pozna, Poland). Procyanidin B₂ was obtained from Extrasynthese S.A. (Genay Cedex, France). HPLC solvents were acquired from Merck Co. (Darmstadt, Germany). Potassium chloride (KCl), ferrous chloride tetrahydrate (FeCl₂ × 4H₂O), cupric sulfate pentahydrate (CuSO₄ × 5H₂O), zinc chloride (ZnCl₂), acetone, ethanol and other reagents, all analytical grade, were obtained from P.O.Ch. Co. (Gliwice, Poland).

Ground and defatted with hexane hazelnuts, walnuts and almonds were extracted with 80% acetone (v/v) according to procedure of Amarowicz et al. [41]. Lyophilised extracts (2 g) were dissolved in 20 mL of ethanol and applied into a column (5 × 40 cm) packed with Sephadex LH-20 gel. At first low molecular weight phenolic compounds were eluted with ethanol (1 L). Part of lower molecular hydrolysable tannins as well as procyanidins, including most momoners and some dimmers and trimers could have been coeluted in this step. Then 600 mL of acetone:water (1:1, v/v) was used to elute tannins [42]. Acetone was removed using rotary evaporator, and aqueous residue was lyophilised.

The content of condensed tannins in fractions was determined according to a modified vanillin assay [43]. To 1 mL of fractions dissolved in methanol (at a concentration of 1 mg/mL), 5 mL of vanillin/HCl reagent [0.5 g vanillin in 4% hydrochloric acid in methanol (v/v)] was added. Absorbance at 500 nm using Beckman DU 7500 diode array spectrophotometer (Beckman Instruments Inc., USA) was recorded after 20 min of samples standing in darkness. Parallel control samples (without vanillin) were prepared. Results were calculated using the standard curve for (+)-catechin (0.2–0.8 mg/mL, r² = 0.997) and expressed as mg catechin equivalents (CE) per gram of tannin fraction.

The bovine serum albumin (BSA) precipitation ability of samples was examined using method developed by Hagerman and Butler [44]. Briefly, to 2 mL of BSA (concentration of 1 mg/mL in 0.2 M acetic buffer, pH 5.0 with 0.17 M NaCl) 1 mL of tannin fractions in water (1 mg/mL) was added. Samples were allowed to stand for 15 min at room temperature and then were centrifuged at 5,000 g for 15 min. The supernatant was removed, and pellet was dissolved in 4 mL of water solution of 1% SDS and 5% triethanolamine. Then 1 mL of 0.01 M FeCl₃ in 0.01 M HCl was added. The samples were left for 30 min at room temperature and the absorbance at 510 nm was measured. Results were calculated using the standard curve for tannic acid (0.2–1.0 mg/mL, r² = 0.996) and expressed as mg tannic acid equivalents (TAE) per gram of tannin fraction.

A SE-HPLC method was employed to examine the molecular weight distribution of nut tannin fractions obtained from Sephadex LH-20 separation. The samples were dissolved in 45% acetonitrile (v/v) with 0.1% TFA (v/v) at concentration 2 mg/mL and analysed using a Shimadzu HPLC system (Shimadzu Co., Japan) consisting of LC-10AD_Vp pumps, SCL-10A_Vp system controller and UV-VIS SPD-M10A_Vp photo-diode array detector. The TSK-GEL^® G2000SW_XL column (7.86 × 300 mm, 5 μm, Tosoh Co., Japan) was employed. The mobile phase of 45% acetonitrile with 0.1% TFA was delivered at a rate of 0.2 mL/min. The detection was monitored at 280 nm and the injection volume was 20 μL. Gallic acid, procyanidin B₂ and tannic acid with molecular weights 170, 578 and 1701, respectively, were used as standards.

In order to determine Cu(II) and Zn(II) chelation capacity, tannin fractions were dissolved in 0.01 M hexamine/HCl buffer with addition of 0.01 M KCl, pH 5.0 [30]. The samples were diluted in the range 0.2–1.0 mg/mL or 1–5 mg/mL to determine bound Cu(II) and Zn(II), respectively. CuSO₄ × 5H₂O or ZnCl₂ were dissolved in the same buffer at concentration of 0.25 mM and 0.8 mM, respectively. Tannin solution (1 mL) was mixed with 1 mL of salt solution, then 0.1 mL of tetramethylmurexide at concentration of 1 mM was added. Absorbance was recorded at 482 nm [Cu(II)] or 462 nm [Zn(II)] and 530 nm [Cu(II) and Zn(II)] and the ratio of A₄₈₂/A₅₃₀ [Cu(II)] or A₄₆₂/A₅₃₀ [Zn(II)] was calculated. Control samples were prepared in the same way: redistilled water was added instead of TMM reagents. A standard curve of absorbance ratio vs. metal ions concentration was prepared; Cu(II) in the range of 0.025–0.125 mM and Zn(II) from 0.2 to 2.0 mM. The percentage of bound metal ions was calculated.

Fe(II) chelation activity of tannin fractions from hazelnuts, walnuts and almonds was examined using ferrozine [45]. Tannin fractions were dissolved in water at a concentration in the range of 0.2–1.0 mg/mL. Solutions (2.5 mL) were mixed with 0.25 mL of 0.4 mM FeCl₂ × 4H₂O, then 0.5 mL of 5 mM ferrozine was added. Then reaction mixture was left for 10 min at room temperature and absorbance at 562 nm was measured. Control samples were prepared in the same way, but water was added instead of ferrozine solution. The percentage of Fe(II) bound was calculated.