Retention behavior of oligomeric proanthocyanidins in hydrophilic interaction chromatography

Journal of Chromatography A, 1143 (2007) 153–161

Retention behavior of oligomeric proanthocyanidins in hydrophilic interaction chromatography Akio Yanagida a,∗ , Hirokazu Murao a , Mayumi Ohnishi-Kameyama b , Yutaka Yamakawa a , Atsushi Shoji a , Motoyuki Tagashira c , Tomomasa Kanda c , Heisaburo Shindo a , Yoichi Shibusawa a a

Division of Structural Biology and Analytical Science, School of Pharmacy, Tokyo University of Pharmacy and Life Science, 1432-1 Horinouchi, Hachioji, Tokyo 192-0392, Japan b Mass Analysis Laboratory, National Food Research Institute, 2-1-2 Kannon-dai, Tsukuba, Ibaraki 305-8642, Japan c Fundamental Research Laboratory, Asahi Breweries Ltd., 1-21 Midori 1-chome, Moriya, Ibaraki 302-0106, Japan Received 28 July 2006; received in revised form 27 December 2006; accepted 2 January 2007 Available online 16 January 2007

Abstract A novel method was developed for the separation of proanthocyanidins (PAs; oligomeric flavan-3-ols) by hydrophilic interaction chromatography (HILIC) using an amide–silica column eluting with an aqueous acetonitrile mobile phase. The best separation was achieved with a linear gradient elution of acetonitrile–water at ratios of 9:1 to 5:5 (v/v) for 60 min at a flow rate of 1.0 ml/min. Under these HPLC conditions, a mixture of natural oligomeric PAs (from apple) was separated according to degree of polymerization (DP) up to decamers. The DP of each separated oligomer was confirmed by LC/electrospray ionization MS. In further HILIC separation studies of 15 different flavan-3-ol and oligomeric PA (up to pentamer) standards with an isocratic elution of acetonitrile–water (84:16), a high correlation was observed between the logarithm of retention factors (log k) and the number of hydroxyl groups in their structures. The coefficient of this correlation (r2 = 0.9501) was larger than the coefficient (r2 = 0.7949) obtained from the correlation between log k and log Po/w values. These data reveal that two effects, i.e. hydrogen bonding between the carbamoyl terminal on the column and the hydroxyl group of solute oligomer and hydrophilicity based on the high-order structure of oligomeric PAs, corporately contribute to the separation, but the hydrogen bonding effect is predominant in our HILIC separation mode. © 2007 Elsevier B.V. All rights reserved. Keywords: Proanthocyanidin; Hydrophilic interaction chromatography; Amide-80; Hydrogen bonding; Normal-phase HPLC; log P

1. Introduction Polyphenolic catechins are naturally occurring flavan-3-ols that are present in a wide variety of foods and plants. Because the general structure of flavan-3-ol (see, Fig. 1) has two centers of asymmetry at C-2 and C-3 on the heterocyclic ring, two pairs of diastereoisomers are theoretically present. In addition to these isomers, analogous compounds having a galloyl ester (gallate) at the C-3 hydroxyl group of each flavan-3-ol isomer have also been identified. In most plants, these isomeric flavan-3-ols and their gallates constitute the monomer units of the oligomeric proanthocyanidins (PAs). Plant PAs are a mixture of oligomers

∗

Corresponding author. Tel.: +81 42 676 4546; fax: +81 42 676 4542. E-mail address: [email protected] (A. Yanagida).

0021-9673/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2007.01.004

consisting of chains of flavan-3-ol units linked most commonly through C-4 to C-8 (or C-6) interflavan bonds (type-B PA), but unusual double-linked hetero-oligomers (type-A PA) exist also [1]. Monomeric flavan-3-ols and oligomeric PAs are well-known natural antioxidants, and several recent studies have reported that the PAs in edible fruits and seeds also play an important role as functional food factors in the prevention of human diseases and cancers [2–4]. The greater part of these activities depends on structure, particularly degree of polymerization (DP). To elucidate the mechanisms of the physiological functions of PAs, and to provide these oligomers for subsequent in vivo studies, suitable separation methods for PAs according to DP are therefore necessary. The chromatographic separation of PAs is complicated by the enormous variety of similar isomeric oligomers in plant or food sources. HPLC using an ODS column eluting with

154

A. Yanagida et al. / J. Chromatogr. A 1143 (2007) 153–161

an aqueous acetonitrile (or methanol) mobile phase was previously the general method for stereoisomeric separation of flavan-3-ols and small DP oligomers, such as dimers and trimers [5–8]. However, the elution order of these monomers and oligomers by ODS–HPLC was not based on DP, and broad unresolved peaks derived from higher polymerized PAs overlapped the separation profiles of these monomers and small DP oligomers [8]. For more efficient separation of more highly polymerized PAs, several useful methods based on other separation mechanisms have recently been established. Size-exclusion chromatography (SEC) eluting with dimethylformamide [9–12] or with another aqueous-organic mobile phase [8,13,14], and high-speed counter-current chromatography (HSCCC) using immiscible two-phase solvent systems [15,16] enable partial separation of oligomeric PAs according to their DP. When compared with SEC and HSCCC methods, HPLC using a bare silica column and elution with an organic mobile phase shows more efficient separation performance. On silica HPLC under the most suitable conditions, oligomeric PAs from different plant sources were clearly separated up to decamers with a gradient elution of dichloromethane-methanol solvent mixture (containing a small volume of acidic water) [17–21], and PAs from apple were also separated up to pentamers with a gradient elution of hexane–acetone solvent mixture [22,23]. However, these organic mobile phases for silica HPLC are associated with some drawbacks. In the case of the mobile phase containing dichloromethane, the use of halogenated solvents possesses a health hazard. In the case of the hexane–acetone mobile phase, oligomeric PAs with high DP are scarcely eluted from the silica column because of their low solubility in the organic mobile phase and strong adsorption to the silica stationary phase. Safer and water compatible mobile phases are therefore necessary for these silica HPLC methods. In addition to these practical drawbacks, the separation and retention mechanisms of oligomeric PAs in this mode have not been suf-

ficiently elucidated. PAs separation using the above-described silica HPLC conditions is usually described as “normal-phase HPLC” in previous papers (including our previous works) [17–23]. However, no one has confirmed whether PAs separation was actually based on hydrophilicity of solute oligomers. In addition, the difference in hydrophilicity (or hydrophobicity) of oligomeric PAs of varying DP remains uncertain. The present paper describes a novel separation method for natural PAs by hydrophilic interaction chromatography (HILIC) using a silica-based stationary phase bonded with acrylamide (TSKgel Amide-80) and an aqueous acetonitrile mobile phase. This method is superior to traditional silica HPLC methods in separation performance and convenience. Mechanistic study was also performed on the measurement of hydrophobicity (i.e. log Po/w ) of several PA standards and on their retention behavior under our HILIC conditions. 2. Experimental 2.1. Apple PA mixture The method for preparation of a PA-rich fraction from immature apples (apple PA mixture) was as described elsewhere [8,22,23]. The matrix-assisted laser desorption/ionization (MALDI) time-of-flight (TOF) MS spectrum of this fraction indicated a mixture of monomeric flavan-3-ols (1a and 1c in Fig. 1) and oligomeric PAs ranging in size from dimer (2mer) to pentadecamer (15-mer) [24]. The constituents of apple PA mixture were divided into two fractions by the following methyl acetate extraction [22]. The greater part of monomers and oligomeric PAs (less than hexamers) were selectively extracted with methyl acetate while the non-extractive constituents containing highly polymerized PAs remained in the residue. The extract and residue were applied to the HILIC separation system described in Sections 2.4 and 2.5. Furthermore, a portion

Fig. 1. Structures of monomeric flavan-3-ols and oligomeric PAs used as experimental standards.

A. Yanagida et al. / J. Chromatogr. A 1143 (2007) 153–161

of the former extract was used as the starting material for the chromatographic isolation of PA oligomers (3b, 3c, 4a and 5a in Fig. 1) as experimental standards, described in Section 2.2. 2.2. Monomeric flavan-3-ols and PA oligomers as experimental standards

155

single piece of PTFE tubing (50 m × 1.0 mm i.d. × 2.0 mm o.d.; total capacity, about 40 ml) directly onto a holder hub to yield eight coiled layers (26 turns in each layer) between a pair of flanges (β = 0.5–0.6). The revolution speed of the apparatus was regulated at 1000 rpm with a speed control unit. The coiled column rotates around its own axis as it synchronously revolves around the central axis, producing an efficient mixing of the two phases while retaining a sufficient amount of stationary phase solvent in the coiled column.

Six types of monomeric flavan-3-ols and nine types of PA oligomers from dimers to pentamers (see Fig. 1) were obtained from commercial sources or isolated from the methyl acetate extract of apple PA mixture. Compounds labeled as 1a ((+)-catechin: C), 1b ((+)-gallocatechin: GC), 1c ((−)-epicatechin: EC), 1d ((−)-epigallocatechin: EGC), 1e ((−)-epicatechin gallate; ECg) and 1f ((−)-epigallocatechin gallate: EGCg) were purchased from Wako (Osaka, Japan). Compounds 2a (procyanidin B1: EC-(4β → 8)-C), 2b (procyanidin B2: EC-(4β → 8)-EC) and 3a (procyanidin C1: EC-(4β → 8)-EC-(4β → 8)-EC) were from Funakoshi (Tokyo, Japan). Compounds 2c (procyanidin B3: C-(4α → 8)-C) and 2d (prodelphinidin B3: GC-(4α → 8)-C) were from Sigma–Aldrich Japan (Tokyo, Japan). PA trimers (3b and c), tetramer (4a) and pentamer (5a) were isolated from the methyl acetate extract of apple PA mixture by preparative silica and ODS HPLC. The chromatographic conditions of these preparative HPLCs were as described elsewhere [22,23]. The MALDITOF-MS spectra of the isolated oligomers demonstrated that compounds 3b and 3c were PA trimers ([M+H]+ ion peak at m/z 868), 4a was a PA tetramer ([M+H]+ ion peak at m/z 1156) and 5a was a PA pentamer ([M+H]+ ion peak at m/z 1444). The chemical structures of 3b, 3c and 4a were assigned as EC-(4β → 6)-EC-(4β → 8)-EC, EC-(4β → 8)-EC-(4β → 6)EC and EC-(4β → 8)-EC-(4β → 8)-EC-(4β → 8)-EC, respectively, based on the retention times of each oligomer in analytical ODS HPLC, which closely matched those in a previous report [23]. The chemical structure of compound 5a was determined by acid-degradation treatment in the presence of the nucleophile phloroglucinol [23,25]. The main degradation products (free EC and EC-(4β → 2)-phloroglucinol adduct) indicated that the structure of the compound 5 was a linearly linked EC pentamer through C-4 to C-8 interflavan bonds, such as EC-(4β → 8)-EC(4β → 8)-EC-(4β → 8)-EC-(4β → 8)-EC.

where the tR is the retention time (min) of solute and t0 is the elution time of the non-retained compound (anthracene) through the column.

2.3. HPLC and HSCCC apparatus

2.5. LC (HILIC)/MS conditions

The HPLC system consisted of a Model L-7100 pump (Hitachi, Tokyo, Japan), a Rheodyne 7166 injector and a Model L-7455 diode-array detector (Hitachi). Chromatograms were recorded on a personal computer using a Model D-7000 chromatography data station software (Hitachi). For HILIC separation, a TSKgel Amide-80 column (Tosoh, Tokyo, Japan) was connected to the above Hitachi HPLC system. For HSCCC experiments, a type-J coil-planet centrifuge (J-CPC; manufactured by Renesas Eastern Japan Semiconductor, Tokyo, Japan) [15,16,26] was connected to the HPLC system. The J-CPC symmetrically holds a multilayer coiled separation column and a counter-weight at a distance of 10 cm from the central axis of the centrifuge. A separation column was fabricated by winding a

LC/MS experiments were performed on an LCQ classic mass spectrometer (Thermo Electron Corp., MA, USA) equipped with an Agilent HPLC system (Agilent Technologies Japan, Tokyo, Japan) attached to a semi-micro TSKgel Amide-80 column (250 mm × 2.0 mm i.d., 5 ␮m). Mobile phases A and B were mixtures of acetonitrile–water at volume ratios of (A) 90:10 and (B) 50:50. Lyophilized powder of the residue from apple PA mixture was dissolved in solvent A (10 mg/ml) as a sample solution. A portion of this solution (2 ␮l) was injected and eluted with a linear gradient program from 100% A (0% B) to 0% A (100% B) for 30 min at a flow rate of 0.3 ml/min at 40 ◦ C. Electrospray ionization (ESI) mass spectra were recorded in negative-ion mode with a spray voltage of 4.5 kV, a capillary voltage of −18 V and

2.4. HILIC conditions Separation of PA oligomers in the methyl acetate extract and residue of the apple PA mixture was performed by HPLC using a conventional TSKgel Amide-80 column (250 mm × 4.6 mm i.d., 5 ␮m particle size) in HILIC mode. Mobile phases A and B were mixtures of acetonitrile–water at volume ratios of (A) 90:10 and (B) 50:50, respectively. The Amide-80 column was initially conditioned with solvent A at a flow rate of 1.0 ml/min. The lyophilized powder from 10 mg of methyl acetate extract or residue of apple PA mixture was dissolved in 1 ml of solvent A (final conc.: 10 mg/ml). A portion of this sample solution (20 ␮l) was injected and eluted in accordance with a linear gradient program from 100% A (0% B) to 0% A (100% B) for 60 min at a flow rate of 1.0 ml/min at ambient temperature. The absorbance of effluent was monitored using a diode-array detector. Furthermore, the retention behavior of 15 different flavan-3ols (1a–f) and PA oligomers (2a–d, 3a–c, 4a, 5a) were examined under an isocratic elution mode of acetonitrile–water (84:16, v/v) using the same size of amide column. A 10-␮l aliquot of each sample solution (100 ␮g/ml in solvent A) or mixed sample solution (100 ␮g each/ml) was injected and eluted at a flow rate of 1.0 ml/min. The retention factor (k) of each compound in HILIC mode was calculated in accordance with the following Eq. (1): k=

(tR − t0 ) t0

(1)

156

A. Yanagida et al. / J. Chromatogr. A 1143 (2007) 153–161

a capillary temperature of 220 ◦ C. Full scan mode spectra were recorded between m/z 260 and 2000. The mass range of zoom scan mode was set to include the [M−nH]n− of each oligomeric PA within 10 ␮m. 2.6. Measurement of log Po/w of oligomeric PAs by HSCCC An HSCCC system attached to a J-CPC was used to measure the partition coefficients of the oligomeric PA standards in an immiscible two-phase solvent system. The two-phase system comprised 1-octanol/50 mM potassium phosphate buffer (pH 7.4) (1:1, v/v) and was applied to HSCCC for direct determination of the octanol–water partition coefficients (Po/w ) of the monomeric flavan-3-ols and the oligomeric PAs having Po/w values larger than about 0.1 (i.e. log Po/w > −1). HSCCC was carried out in reversed-phase partition mode (stationary phase is upper octanol-rich phase), as reported previously with slight modification [26]. The lower aqueous mobile phase was eluted at a flow rate of 2 ml/min through the PTFE coiled column filled with the upper octanol phase while the column was rotated at 1000 rpm. After hydrodynamic equilibrium between the two phases was established in the coiled column and the baseline was stabilized, 100 ␮l of solution containing a 1.0 mg of sample and 0.25 mg of potassium nitrate as a void volume marker was injected and eluted with the lower aqueous mobile phase. The retention volume of solute (VR ) in partition chromatography, such as CCC, is conventionally expressed by the following equation [26,27]: VR = Vm + KVs

(2)

where Vm and Vs are the mobile phase and stationary phase volumes in the coiled column. Under the above HSCCC conditions using the octanol–water two-phase system, the partition coefficient K of a solute is equal to its Po/w . Furthermore, the Vm is also equal to the elution volume of potassium nitrate, which is scarcely retained in the octanol-rich stationary phase of the column, and the sum of the Vm and Vs is equal to the PTFE column volume (VC ) from the injector to the detector. Accordingly, Eq. (2) can be rewritten as follows: Po/w =

(VR − Vm ) (VR − Vm ) = Vs (VC − Vm )

(3)

In practice, the accuracy and repeatability of the measured Po/w value from Eq. (3) significantly decreases with the VR value of the solute. Thus, an accurate Po/w of polar PA oligomers with estimated values less than 0.1 (i.e. log Po/w < −1) may be indirectly calculated from their partition coefficients (K) in another polar two-phase solvent system comprising t-butyl methyl ether (BME)–acetonitrile (ACN)–water (2:2:3, v/v/v) under almost identical HSCCC conditions as described above. In this case, the HSCCC was carried out in reversed-phase partition mode with an upper organic stationary phase and a lower aqueous mobile phase, and the K value of solute in this two-phase system was determined from its VR value according to Eq. (2). The log Po/w values of the high-polar PA oligomers, such as 2a, 2d, 3a, 3b, 3c, 4a and 5a were indirectly calculated from the

following correlation model: log Po/w = 2.7134 log K − 0.6614

(4)

This model was obtained from a linear least-square fitting of the plot between log Po/w and log k values of other mild-polar monomeric flavan-3-ols and oligomeric PAs (n = 6, r2 = 0.9633). 2.7. Calculation of log Po/w of oligomeric PAs using personal computer software A program module called PrologP 6.0, which is a part of the Pallas 3.0 software package (CompuDrug International, San Francisco, CA, USA), was used for the prediction of log Po/w values of oligomeric PA standards as neutral species. The predicted log Po/w values were determined from data in two different fragment databases (CDR and ATOMIC5) in the software package according to the following equation: log Pcombined = 0.733 log PATOMICS + 0.267 log PCDR 3. Results and discussion 3.1. HILIC separation of apple PA mixture The term “hydrophilic interaction chromatography (HILIC)” was proposed by Alpert to describe a variant of normal-phase liquid chromatography (NPLC) in 1990 [28]. As with NPLC, polar compounds, such as carbohydrates, oligosaccharides, peptides and nucleotides, are more strongly retained in HILIC, but the non-aqueous organic mobile phase in traditional NPLC is replaced with a more polar mobile phase containing water as the stronger solvent [28–32]. Because the low-solubility of polar highly polymerized PA oligomers in non-aqueous organic mobile phases was a severe drawback in our previous NPLC study [22], we used HILIC with aqueous solvents to improve PAs separation. In typical HILIC studies, bare silica [31,32], ionic aminopropyl (or cyanopropyl)–silica [31,32] or neutral amide–silica [30–32] columns are used as stationary phases for the separation of highly polar compounds. In this study, a silicabased column bonded with acrylamide including non-ionic carbamoyl terminal, TSKgel Amide-80 (Tosoh) [30], was used for PAs separation because PA oligomers are neutral polyphenolic compounds without ionic groups in their structures. Initially, HILIC separation of the oligomeric constituents in apple PA mixture was carried out with a conventional Amide80 column (250 mm × 4.6 mm i.d.) with an aqueous acetonitrile mobile phase. The best separation was achieved with a linear gradient of acetonitrile–water from a volume ratio of 9:1 to 5:5 (v/v) for 60 min at a flow rate of 1.0 ml/min. The chromatograms of the methyl acetate extract and residue of apple PA mixture are shown in Fig. 2A and B, respectively. The main peaks in the both chromatograms are numbered based on their elution order. The separation profile of the methyl acetate extract of apple PA mixture (Fig. 2A) was very similar to our previous results obtained by silica HPLC using an organic hexane–acetone mobile phase [22]. The DPs of oligomeric PAs in the separated peak fractions

A. Yanagida et al. / J. Chromatogr. A 1143 (2007) 153–161

Fig. 2. HILIC separation profiles of apple PA mixture after methyl acetate extraction. (A) Chromatogram of oligomeric PAs in methyl acetate extract (10 mg/ml). (B) Chromatogram of oligomeric PAs in extraction residue (10 mg/ml). Column: conventional TSKgel Amide-80 (250 mm × 4.6 mm i.d., 5 ␮m); solvent A, acetonitrile–water (9:1); solvent B, acetonitrile–water (5:5); linear gradient program from (100% A, 0% B) to (0% A, 100% B) for 60 min; flow rate, 1 ml/min; injection volume, 20 ␮l. Separated peaks on both chromatograms are numbered in accordance with their elution order (up to 10). Two marked (* ) peaks are from impurities (caffeoyl quinic acid and phloretin glycoside).

were characterized by off-line MALDI-TOF-MS analysis, and the MS results demonstrated that the oligomeric PAs up to pentamers in the extract were eluted based on DP (MS spectra are not shown). When compared with the Fig. 2A profile, the separation profile of the residue from apple PA mixture (Fig. 2B) showed more separated peaks (up to 10) because this sample contained highly polymerized oligomeric PAs larger than pentamers [22]. To confirm the DP-based separation in HILIC mode, a further separation experiment was carried out with an LC/ESI–MS system. To optimize the LC conditions for on-line ESI-MS detection, the Amide-80 column was replace with a semi-micro column (2.0 mm i.d.), and the flow rate (0.3 ml/min) and injection volume (2 ␮l) were reduced, while other LC conditions remained identical to those shown in Fig. 2B. The LC/MS results of the residue from apple PA mixture are shown in Fig. 3A–C). When compared with the Fig. 2B profile, the chromatographic resolution of Fig. 3A was relatively low. However, the mass chromatograms of oligomeric PAs from trimers to decamers (Fig. 3B) revealed that the retention time of each peak gradually increased with DP. These data confirm that DP-based separation of oligomeric PAs occurred in HILIC mode. Thus, under the best chromatographic conditions shown in Fig. 2B, the number of separated peaks (up to 10) must directly correspond to the DP of each oligomer (up to decamers). 3.2. Retention behavior of oligomeric PA standards in HILIC and their hydrophobicity Since the first article has reported by Regaud et al. [17], several PAs separation methods using silica HPLC with organic

157

mobile phases have been published. Furthermore, Kelm et al. recently reported the first HILIC separation of Cacao procyanidins using a diol stationary phase column, and a binary gradient of acidified acetonitrile and methanol–water [33]. Although the separation methods were described as “normal-phase HPLC” in these reports, including the recent Kelm paper, the separation and retention mechanism of oligomeric PAs have not been elucidated, as mentioned in Introduction section. To clarify the true mechanism of PAs separation in the above LC modes, several isolated oligomers, whose structures have been completely elucidated, are required. In the present study, we thus prepared 15 different monomeric flavan-3-ols and oligomeric PAs up to pentamers as standard solutes and investigated their retention behavior under our HILIC conditions. In practice, the HILIC studies of these standards were carried out under an isocratic elution mode using acetonitrie–water (84:16) for calculation of retention factor (k) of each solute. Fig. 4 shows the HILIC elution profile of the 15 standards listed in Fig. 1 under the isocratic elution mode. The k value of each standard was calculated from each retention time according to Eq. (1) and the values are listed in Table 1 (elution behavior based on k values is discussed in detail in next Section 3.3). The chromatogram confirmed the tendency to separate these standard oligomers based on DP. However, among oligomers having the same DP, retention times did not perfectly match. We initially believed that variation in retention times among same DP oligomers was attributable to differences in hydrophobicity. In general, the hydrophobicity of drugs and industrial chemicals are evaluated as based on the logarithm of the octanol–water partition coefficient, log Po/w . Thus, we measured log Po/w values for the 15 different standards by HSCCC using an octanol–water two-phase solvent system [26]. A scatter diagram of the DP values of 15 standards against the log Po/w values is shown in Fig. 5. The diagram revealed a tendency for log Po/w values to decrease significantly with increasing DP values. A minus log Po/w value indicates that the compound is highly polar (or hydrophilic). Because polar compounds are strongly retained by water-enriched stationary phase in general normal-phase partition chromatography, the differences in polarity among these standards (Fig. 5) actually contribute to the DP-based separation shown in Fig. 4. However, on detailed observation, the log Po/w values of different compounds eluted at almost same time in Fig. 4 were not identical in Fig. 5 (for example, 1c, 1a and 1e, or 2c, 2b and 2a). This inconsistency could not be explained by the normal-phase partitioning mechanism based on hydrophilicity of solute PA. Therefore, we believed that another important separation factor (or interaction) contributes to the separation of PAs in the HILIC mode employed herein. 3.3. Retention mechanism of oligomeric PAs in HILIC Based on the results in Figs. 4 and 5 for about 15 different standards, their structural parameters (DP and number of hydroxyl groups) and their chromatographic parameters (log Po/w , tR , k and log k in HILIC) are listed in Table 1. In this Table, the predicted log Po/w values (log Pcalc ) calculated by the PrologP module in the Pallas software are also listed, together

158

A. Yanagida et al. / J. Chromatogr. A 1143 (2007) 153–161

Fig. 3. LC/ESI–MS profiles of extraction residue from apple PA mixture. A: HILIC chromatogram of oligomeric constituents in extraction residue from apple PA mixture detected at UV 230 nm. Peaks at 3.0 and 7.3 min were assigned to monomeric catechins and dimeric PAs, respectively. Marked (* ) peak is from impurity. B: mass chromatograms correspond to m/z values of oligomeric PAs from trimers to decamers. The m/z values of [M−H]− ions were selected for trimers at 865, tetramers at 1153, pentamers at 1441 and hexamers at 1729. The m/z values of [M−2H]2− ions were selected for heptamers at 1008, octamers at 1152, nonamers at 1296 and decamers at 1440. C: ESI mass spectra (on zoom scan mode) corresponding to the peaks on mass chromatograms shown in B.

A. Yanagida et al. / J. Chromatogr. A 1143 (2007) 153–161

159

Table 1 Comparison of structural and chromatographic parameters of 15 standard flavan-3-ols and oligomeric PAs Compounds

1a 1b 1c 1d 1e 1f 2a 2b 2c 2d 3a 3b 3c 4a 5a a b c

C GC EC EGC ECg EGCg EC(4β–8)C EC(4β–8)EC C(4α–8)C GC(4α–8)C EC(4β–8)EC(4β–8)EC EC(4β–8)EC(4β–6)EC EC(4β–6)EC(4β–8)EC EC(4β–8)EC(4β–8)EC(4β–8)EC EC(4␤–8)EC(4β–8)EC(4β–8)EC(4β–8)EC

Structural parameters

Chromatographic parameters

Degree of polymerization

log Po/w

1 1 1 1 1 1 2 2 2 2 3 3 3 4 5

Number of OH groups

5 6 5 6 7 8 10 10 10 11 15 15 15 20 25

log Pmeas a

log Pcalc b

|Δ|c

0.31 −0.31 0.10 −0.53 1.10 0.50 −1.35 −0.90 −0.91 −1.52 −1.32 −1.49 −1.13 −1.79 −2.21

0.86 0.43 0.86 0.43 2.10 1.67 1.04 1.04 1.04 0.58 1.22 1.18 1.18 1.34 1.50

(0.55) (0.74) (0.76) (0.96) (1.00) (1.17) (2.39) (1.94) (1.95) (2.10) (2.54) (2.67) (2.31) (3.13) (3.71)

tR

k

log k

3.71 4.59 3.68 4.69 3.74 4.55 6.40 6.10 5.80 8.01 10.94 8.66 8.70 21.36 44.93

0.40 0.74 0.39 0.77 0.42 0.72 1.43 1.31 1.20 2.03 3.14 2.28 2.30 7.10 16.0

−0.39 −0.13 −0.41 −0.11 −0.38 −0.14 0.15 0.12 0.08 0.31 0.50 0.36 0.36 0.85 1.20

Measured by HSCCC. Calculated by PrologP 6.0. Absolute value of difference between measured and calculated log P s.

with the measured values (log Pmeas ) by HSCCC. Interestingly, significant differences between the log Pmeas and log Pcalc values were observed on comparison of data for all standard compounds, and this difference (|Δ|) increased with DP. Because the prediction of log P by PrologP is only based on the fragment database of the planar structure of test compounds, the calculated results do not reflect high-order structural information regarding stereoisomeric and/or conformational changes. Paradoxically, the observed differences between the log Pmeas and log Pcalc values of oligomeric PAs must be due to the formation of high-order structures. This suggests that condensation of monomeric flavan3-ol units linked through C4 to C8 (or C6) interflavan bonds leads to the formation of unique three-dimensional structures. Actually, some recent NMR [34] and CD [35] studies revealed

Fig. 4. HILIC separation profile of 15 different standards of flavan-3-ols and oligomeric PAs under isocratic elution mode. Column: conventional TSKgel Amide-80 (250 mm × 4.6 mm i.d., 5 ␮m); mobile phase: acetonitrile–water (84:16); flow rate: 1 ml/min; injection volume: 10 ␮l. The numbers of standard compounds were identical to those used in Fig. 1.

that oligomeric PAs in polar solution formed a compact geometric structure with helical nature. We also believe that the high polarity of oligomeric PAs may be due to its helical structure comprising a hydrophilic surface covered by a large number of hydroxyl groups and an internal hydrophobic cavity including stacked rings. Furthermore, based on the data listed in the Table 1, we examined the relationship between the log k values of 15 standards in our HILIC and their log Pmeas values. As shown in Fig. 6, the relationship between log k and log Po/w was expressed as the regression formula “y = −1.8177x − 0.4767”. This relationship reveals that hydrophilicity (i.e. log Po/w values) of standard oligomers actually contributes to the DP-based separation. However, as mentioned in Section 3.2, the square of the correlation coefficient of this relationship is not very high (r2 = 0.7949). To elucidate another separation factor in our HILIC mode, we examined the relationship between the log k values of 15 standards and

Fig. 5. Scatter diagram of degree of polymerization (DP) of the 15 standards against log Po/w values measured by HSCCC.

160

A. Yanagida et al. / J. Chromatogr. A 1143 (2007) 153–161

force. If such speculation is correct, the previous PAs separation mode on silica columns should be referred to as “adsorption”. On the other hand, in our PAs separation using an amide–silica column and in Kelm’s recent report using a diol–silica column [33] with an aqueous mobile phase, solute hydrophilicity based on the high-order structure of PA is not negligible as an important separation factor. In these LC cases, “hydrophilic interaction chromatography (HILIC)” proposed by Alpert [28] is the most appropriate term to include mixed mode separations relying on adsorption and normal-phase partitioning. References Fig. 6. Relationship between log k of 15 standards in HILIC and log Po/w values measured by HSCCC. Inset regression formula was calculated from the linear least-square fit of all data (n = 15).

the number of hydroxyl groups ( OH) in their structures. A high correlation was observed, as shown in Fig. 7. The relationship was expressed as the regression formula “y = 12.341x + 9.2541”, and the square of the correlation coefficient (r2 = 0.9501) was larger than the above value (0.7949) obtained from the correlation between log k and log Po/w values. This high correlation between log k and number of OH groups in the solute standards reveals that a hydrogen bonding effect between a carbamoyl terminal on the Amide-80 column and the OH group of the solute strongly contributes to PA separation in HILIC mode. Based on our overall results, particularly those in Figs. 6 and 7, two effects, i.e. hydrogen bonding and solute hydrophilicity, corporately contribute to the present DP-based PAs separation, but the hydrogen bonding effect predominates in HILIC mode using an aqueous acetonitrile mobile phase. Hydrogen bond formation between two different compounds is generally enhanced under hydrophobic conditions more than under aqueous conditions. Therefore, in previous LC reports using a bare silica column with non-aqueous organic mobile phases [17–23,36], the very strong hydrogen bonding between a silanol group on silica-beads and the hydroxyl group of solute PA might have had an effect as a primary retention (and separation)

Fig. 7. Relationship between log k of 15 standards in HILIC and the number of hydroxyl groups ( OH) in their structures. Inset regression formula was calculated from the linear least-square fit of all data (n = 15).

[1] J.-J. Macheix, A. Fleuriet, J. Billot, Fruit Phenolics, CRC Press, Boca Raton, FL, 1990, p. 84. [2] C. Santos-Buelga, A. Scalbert, J. Sci. Food Agric. 80 (2000) 1094. [3] P. Cos, T. De Bruyne, N. Hermans, S. Apers, D. Vanden Berghe, A.J. Vlietibck, Curr. Med. Chem. 11 (2004) 1345. [4] G.R. Beecher, Pharm. Biol. 42 (Suppl.) (2004) 2. [5] A.G.H. Lea, J. Chromatogr. 238 (1982) 253. [6] E. Delage, G. Bohuon, A. Baron, J.-F. Drilleau, J. Chromatogr. 555 (1991) 125. [7] G.E. Rohr, B. Meier, O. Sticher, J. Chromatogr. A 835 (1999) 59. [8] A. Yanagida, T. Kanda, T. Shoji, M. Ohnishi-Kameyama, T. Nagata, J. Chromatogr. A 855 (1999) 181. [9] Y.S. Bae, L.Y. Foo, J.J. Karchecy, Holzforschung 48 (1994) 4. [10] M. L´opez-Serrano, A.R. Barcel´o, J. Chromatogr. A 919 (2001) 267. [11] J.A. Kennedy, A.W. Taylor, J. Chromatogr. A 995 (2003) 99. [12] M. Kurumatani, R. Fujita, M. Tagashira, T. Shoji, T. Kanda, M. Ikeda, A. Shoji, A. Yanagida, Y. Shibusawa, H. Shindo, Y. Ito, J. Liq. Chromatogr. Rel. Technol. 28 (2005) 1971. [13] A. Yanagida, T. Shoji, T. Kanda, Biosci. Biotechnol. Biochem. 66 (2002) 1972. [14] C.Le. Bourvellec, M. Picot, C.M.G.C. Renard, Anal. Chim. Acta 563 (2006) 33. [15] Y. Shibusawa, A. Yanagida, M. Isozaki, H. Shindo, Y. Ito, J. Chromatogr. A 915 (2001) 253. [16] Y. Shibusawa, A. Yanagida, H. Shindo, Y. Ito, J. Liq. Chromatogr. Rel. Technol. 26 (2003) 1609. [17] J. Rigaud, M.T. Escribano-Bailon, C. Prieur, J.-M. Souquet, V. Cheynier, J. Chromatogr. A 654 (1993) 255. [18] J.F. Hammerstone, S.A. Lazarus, A.E. Mitchell, R. Rucker, H.H. Schmitz, J. Agric. Food Chem. 47 (1999) 490. [19] M. Natsume, N. Osakabe, M. Yamagishi, T. Takizawa, T. Nakamura, H. Miyatake, T. Hatano, T. Yoshida, Biosci. Biotechnol. Biochem. 64 (2000) 2581. [20] J.A. Kennedy, A.L. Waterhouse, J. Chromatogr. A 866 (2000) 25. [21] L. Gu, M. Kelm, J.F. Hammerstone, G. Beecher, D. Cunningham, S. Vannozzi, R.L. Prior, J. Agric. Food Chem. 50 (2002) 4852. [22] A. Yanagida, T. Kanda, T. Takahashi, A. Kamimura, T. Hamazono, S. Honda, J. Chromatogr. A 890 (2000) 251. [23] T. Shoji, M. Mutsuga, T. Nakamura, T. Kanda, H. Akiyama, Y. Goda, J. Agric. Food Chem. 51 (2003) 3806. [24] M. Ohnishi-Kameyama, A. Yanagida, T. Kanda, T. Nagata, Rapid Commun. Mass Spectrom. 11 (1997) 31. [25] L.Y. Foo, J.J. Karchesy, Phytochemistry 30 (1991) 667. [26] Y. Shibusawa, A. Shoji, A. Yanagida, H. Shindo, M. Tagashira, M. Ikeda, Y. Ito, J. Liq. Chromatogr. Rel. Technol. 28 (2005) 2819. [27] W.D. Conway, Countercurrent Chromatography: Apparatus, Theory and Applications, VCH, New York, 1990, p. 195. [28] A.J. Alpert, J. Chromatogr. 499 (1990) 177. [29] A.J. Alpert, M. Shukla, A.K. Shukla, L.R. Zieske, S.W. Yuen, M.A.J. Ferguson, A. Mehlert, M. Pauly, R. Orlando, J. Chromatogr. A 676 (1994) 191. [30] T. Yoshida, Anal. Chem. 69 (1997) 3038.

A. Yanagida et al. / J. Chromatogr. A 1143 (2007) 153–161 [31] B.A. Olsen, J. Chromatogr. A 913 (2001) 113. [32] Y. Guo, S. Gaiki, J. Chromatogr. A 1074 (2005) 71. [33] M.A. Kelm, J.C. Johnson, R.J. Robbins, J.F. Hammerstone, H.H. Schmitz, J. Agric. Food Chem. 54 (2006) 1571. [34] I. Tarascou, K. Barathieu, C. Simon, M.-A. Ducasse, Y. Andr´e, E. Fouquet, E.J. Dufourc, V. de Freitas, M. Laguerre, I. Pianet, Magn. Reson. Chem. 44 (2006) 868.

161

[35] A. Muranaka, K. Yoshida, T. Shoji, N. Moriichi, S. Masumoto, T. Kanda, Y. Ohtake, N. Kobayashi, Org. Lett. 8 (2006) 2447. [36] T. Shoji, S. Masumoto, N. Moriichi, T. Kanda, Y. Ohtake, J. Chromatogr. A 1102 (2006) 206.