Reversed-phase liquid chromatographic analysis of hydrophobic interaction between proanthocyanidins and a C₈-alkyl compound in aqueous solution

Abstract

Structural and physicochemical properties of oligomeric flavan-3-ols (proanthocyanidins) in aqueous solution were investigated by spectrometric and reversed-phase (RP) HPLC analyses. Circular dichroism and fluorescence spectra of (–)-epicatechin (EC) oligomers linked through C-4 to C-8 interflavan bonds showed that EC oligomers larger than dimers formed a stable secondary structure in water. These EC oligomers are water-soluble hydrophilic compounds, whereas the oligomers were strongly retained by a C₈-alkyl stationary phase under conventional RP-HPLC conditions. In a further C₈-HPLC study, the hydrophobic interaction between EC oligomers and 1-octanesulfonic acid sodium salt (OSA Na) added to the mobile phase was quantitatively evaluated based on the relationship between the logarithm of the retention factor of the solute and the OSA Na concentration in the mobile phase. The strength values of the hydrophobic interaction of EC oligomers larger than dimers were the highest of 22 tested polyphenolic standards.

Key words:

• proanthocyanidins
• circular dichroism
• reversed-phase liquid chromatography
• hydrophobic interaction
• 1-octanesulfonic acid

Polyphenolic catechins are naturally occurring flavan-3-ols that are present in a wide variety of dietary plants. The greater part of the catechins, containing their isomers and gallates, is constituted by monomer units with the proanthocyanidin (PA) structure. Plant PAs are well known as natural antioxidants and important functional food factors in the prevention of human cancers and other diseases.^1–3) Plant PAs are generally a mixture of oligomers consisting chains of flavan-3-ol units linked most commonly through C₄–C₈ or C₆ interflavan bonds (type-B PAs). The condensation of monomeric flavan-3-ol units leads to the formation of a compact geometric PA structure that is helical in aqueous solution.^4–6) Based on secondary structure, the hydrophobicity of PA significantly decreases with an increase in its degree of polymerization (DP), and its water solubility significantly increases.^7,8) These properties reveal that the molecular surface of PA is covered by a large number of phenolic hydroxyls and that these hydroxyls are hydrated in aqueous solution. On the other hand, in a chromatographic column with organic solvents, plant PAs are strongly retained to polar column packings such as bare silica,^9–12) diol-silica,¹³⁾ amide-silica,⁸⁾ Sephadex LH-20,^14,15) and Toyopearl HW-40^16–18) by hydrogen bonding between the phenolic hydroxyls of the solute PAs and proton donors and acceptors on the packing materials. Thus, under most suitable normal-phase HPLC (or hydrophilic interaction chromatography: HILIC) conditions, plant PAs up to decamers have been separated well according to DP.^8,10,12,13)

However, this excellent DP-based separation of plant PAs cannot be reproduced by conventional reversed-phase (RP) HPLC using an aliphatic C₁₈ or C₈ alkyl column because hydrogen bonding works poorly as a means of retention on alkyl columns with an aqueous mobile phase. Even though RP-HPLC is a useful method for stereoisomeric separation of monomeric flavan-3-ols and small DP oligomers such as dimers and trimers,^16,19–22) the elution order of these small compounds is not based on DP. Furthermore, when higher DP oligomers coexist in a sample solution, a broad unresolved hump derived from the higher DP oligomers overlaps the separation profiles of the smaller oligomers; therefore, the separation profile becomes too complex to assign each peak. The main retention mechanisms in common RP-HPLC are as follows: (1) solute partitioning between the alkyl stationary phase and aqueous mobile phase and (2) hydrophobic interaction between the solute in the mobile phase and the alkyl surface of the stationary phase; however, details of retention behavior and the mechanism for the separation of PAs in RP-HPLC remain unknown. In RP-HPLC separation of PAs, the most important observation is that “very hydrophilic PAs having high DP” are strongly retained on aliphatic alkyl stationary phases, as are comparatively hydrophobic monomers (the chromatographic data will be shown in the later Result section of this study). This suggests that the retention mechanism of PAs in RP-HPLC may be different from that of monomeric flavan-3-ols. Furthermore, the unique retention behavior of the hydrophilic PAs in RP-HPLC may be attributed to their unique secondary structure in aqueous solution.

Based on the above speculation, we first examined the secondary structure of (–)-epicatechin (EC) oligomers (up to pentamers) associated with an increase of their DP by circular dichroism and fluorescence spectrometric analyses. Next, the retention behavior of the EC oligomers in C₈-HPLC, especially the hydrophobic interaction between the oligomers and a C₈-alkyl compound added to aqueous mobile phase, was investigated in detail.

Materials and methods

Reagents

Sodium chloride, potassium nitrate, and 1-octanesulfonic acid sodium salt (OSA·Na) were purchased from Wako Pure Chemical (Osaka, Japan). HPLC-grade acetonitrile and trifluoroacetic acid (TFA) were from Kanto Chemical (Tokyo, Japan). Ultrapure water was produced from distilled water by a Puric-ZII system (Organo; Tokyo, Japan).

Polyphenolic compounds

Six types of monomeric flavan-3-ols and five types of oligomeric flavan-3-ols from dimers to pentamers (see Fig. 1) were obtained from commercial sources or isolated from a PA-rich fraction obtained from immature apples (apple PA mixture).^11,16) Monomeric 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. Dimeric compounds labeled as 2a (procyanidin B1: EC-(4β→8)-C) and 2b (procyanidin B2: EC-(4β→8)-EC) were purchased from Funakoshi (Tokyo, Japan). Oligomeric compounds labeled as 3a (procyanidin C1: EC-(4β→8)-EC-(4β→8)-EC), 4a (EC-(4β→8)-EC-(4β→8)-EC-(4β→8)-EC), and 5a (EC-(4β→8)-EC-(4β→8)-EC-(4β→8)-EC-(4β→8)-EC) were isolated from a methyl acetate extract of the apple PA mixture by preparative silica HPLC and preparative ODS HPLC.^11,23) Other polyphenolic compounds were purchased from the following manufacturers or distributors: Caffeic acid, chlorogenic acid, and naringenin were from Sigma-Aldrich Japan (Tokyo, Japan); gallic acid, p-hydroxybenzoic acid, and 3,4-dihydroxybenzoic acid were from Tokyo Kasei (Tokyo, Japan); quercetin, hyperoside (quercetin-3-galactoside), and myricetin were from Funakoshi; and rutin (quercetin-3-glucopyranosyl-rhamnoside) and p-coumaric acid were from Wako.

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

Spectrometric analyses

A J-720 circular dichroism (CD) spectrophotometer (Jasco; Tokyo, Japan) and an RF-5300PC spectrofluorometer (Shimadzu; Kyoto, Japan) were used for analysis of EC (1c) and EC oligomers (2b, 3a, 4a, and 5a). Each test compound was dissolved in water (1 mg/mL), and each solution was diluted to 50 μg/mL for CD analysis or 1 μg/mL for fluorescence (FL) analysis. A CD spectrum of each sample (50 μg/mL) was obtained under the following conditions: optical path length: 2 mm; temperature: 25 °C; scan range: 200–300 nm; sensitivity: 200 mdeg, response: 8 s; band width: 2 nm; resolution: 0.1 nm/data point, scan speed: 20 nm/min; accumulation: 1. The intensity of the CD signal at a certain wavelength was expressed as molar ellipticity ([Ɵ] × 10⁻³ deg cm² dmol⁻¹) based on the molar concentration in terms of the EC monomer unit (MW: 291) in each sample containing EC or EC oligomers with different DP. A FL emission spectrum of each sample (1 μg/mL) was obtained under the following measurement conditions: optical path length: 10 mm; excitation wavelength: 281 nm; scan range: 281–400 nm; sensitivity: high; resolution: 1 nm/data point; scan speed: medium.

RP-HPLC

Monomeric and oligomeric flavan-3-ols were separated by RP-HPLC on a Hitachi (Tokyo, Japan) HPLC system, which consisted of an L-2130 pump, a Reodyne 7725i injector, and an L-2400 UV detector. Chromatograms were recorded on a personal computer using D-2000 Elite chromatography data station software (Hitachi). A Chromolith Performance RP-8e analytical column (4.6 mm i.d. × 100 mm; Merck; Darmstadt, Germany) was used for the separation. The retention behavior of the monomeric and oligomeric flavan-3-ols was examined in isocratic elution mode using the following two mobile phase solvents: (A) acetonitrile–water (10:90, v/v) containing 0.1% TFA and (B) solvent A containing OSA·Na (1, 2, 5, 10, or 20 mM). The OSA·Na was added to the mobile phase as hydrophobic interaction competitor (with the C₈-alkyl stationary phase), not as ion-pair reagent. The RP-8e (C₈-alkyl) column was initially conditioned with the mobile phase solvent at a flow rate of 2.0 mL/min. A 10-μL aliquot of sample (50 μg/mL in mobile phase) was injected and eluted at a flow rate of 2.0 mL/min. The absorbance of the eluate was monitored at 280 nm. The retention factor (k) of each solute was calculated using the equation

where t_R is the retention time (in min) of the solute and t₀ is the elution time of an unretained compound (potassium nitrate).

Results and discussion

Spectral and structural characteristics of EC oligomers associated with an increase in DP

First, we investigated structural characteristics of PA oligomers by simple spectrometric analyses. Fig. 2 shows the UV and CD spectra in water of monomeric EC (1c) and its oligomers (2b, 3a, 4a, 5a). Their UV spectra (200–300 nm) did not differ much, whereas the CD spectra of oligomers larger than dimers (3a, 4a, 5a) differed significantly in shape from the monomer (1c) and dimer (2b). The oligomers (3a, 4a, 5a) gave strong positive CD signals in the wavelength range from 210 to 250 nm due to asymmetric C-4 carbons on interflavan bonds of their oligomeric structures; the signal shapes and intensities of the tetramer (4a) and the pentamer (5a) appeared similar. In the case of CD of polypeptide and protein, the far-UV CD is generally reflective of the backbone conformation, and different secondary structures in polypeptides and proteins give characteristic far-UV CD spectra.²⁴⁾ Then, the observed strong positive CD signals of EC oligomers (3a, 4a, 5a) would be reflective of their secondary structures.

Fig. 2. CD, UV, and FL (emission spectra excited at 281 nm) spectra of EC (1c) and EC oligomers (2b, 3a, 4a, 5a) in water.

Note: The concentration of each sample was 50 μg/mL for CD and UV analyses, and 1 μg/mL for FL analysis.

Fig. 2 also shows their FL emission spectra on excitation at their λ_max wavelength (281 nm), resulting from aromatic rings in their structures. Monomeric EC (1c) gave a strong FL spectrum (maximum emission wavelength: 310 nm), whereas the FL spectra of oligomers (2b, 3a, 4a, 5a) were quenched, and the quenched FL intensities of the tetramer (4a) and the pentamer (5a) were nearly the same. The observed FL quenching of the oligomers is assumed to be due to a conformational change from the planar configuration of monomeric EC to non-planar one of its oligomeric structure.

On the whole, both the CD and FL spectra in Fig. 2 strongly suggest that EC oligomers larger than dimers (3a, 4a, 5a) form a stable secondary structure in water.

Elution behavior and retention mechanism of monomeric and oligomeric flavan-3-ols in C₈-HPLC

We next investigated the elution behavior of several monomeric and oligomeric flavan-3-ols in RP-HPLC. To compare the values of the retention factor k and P_o/w (P_o/w: partition coefficient in 1-octanol/water two-phase system) for the solutes, the length of the alkyl chain of the separation column was chosen as C₈ to match the length of 1-octanol. A typical separation profile of eleven of monomeric (1a–f) and oligomeric (2a, 2b, 3a, 4a, 5a) flavan-3-ols is shown in Fig. 3. Under isocratic elution in acetonitrile–water (10:90, v/v) containing 0.1% TFA, almost all solutes except 1d and 2a were resolved within 10 min. However, the elution order of the solutes was not based on their DP. Furthermore, the elution order and the retention times of the solutes in Fig. 3 (under the TFA acidic pH condition) were no different from those under the neutral pH 7 condition without TFA (data not shown), because the solute flavan-3-ols are all neutral phenolic compounds in which pK_a and P_o/w values are no different between the TFA acidic and neutral pH 7 conditions. After the calculation of k for each peak in the chromatogram of Fig. 3, we examined the relationship between log k and log P_o/w of the eleven flavan-3-ols (log P_o/w values at pH 7.4 were obtained from our previous data.)⁸⁾ A scatter plot between log k and log P_o/w values (Fig. 4) shows the following two relationships: (A) one among six hydrophobic monomers (1a–1f) and (B) another among five EC species (1c, 2b, 3a, 4a, 5a) from monomer to pentamer. The relationship among the six monomers can be expressed as the linear regression formula y = 1.1356 x − 0.1569 (r²= 0.8457), with log k (= x) values positively correlated with log P_o/w (= y) values. The relationship reveals that separation of the monomeric flavan-3-ols was based on a simple RP partitioning mechanism between the C₈-alkyl stationary phase and aqueous mobile phase. On the other hand, the relationship among EC and its oligomers can be expressed as y = −3.3331x + 0.793 (r²= 0.7293), with log k values negatively correlated with log P_o/w values. The relationship reveals that the hydrophilic EC oligomers (with negative values for log P_o/w) are more strongly retained by the C₈-alkyl column according to their DP. It is unusual for such hydrophilic compounds to partition to a C₈-alkyl stationary phase from an aqueous mobile phase. We therefore predicted that the oligomers were retained by adsorption based on a hydrophobic interaction between the C₈-alkyl surface of the column and hydrophobic regions in the secondary structure of the solute oligomer when in an aqueous mobile phase. To test the prediction, we modified the mobile phase composition of the C₈-HPLC and investigated the elution behavior of the flavan-3-ols under the modified conditions.

Fig. 3. Merged chromatogram of eleven flavan-3-ol standards separated by RP C₈-HPLC in isocratic elution mode.

Notes: Column: Chromolith Performance RP-8e (4.6 mm I.D. × 100 mm; Merck); mobile phase: acetonitrile–aqueous 0.1% TFA (10:90, v/v); flow rate: 2.0 mL/min; detection: absorbance at 280 nm; injection volume: 10 μL; concentration of each standard: 50 μg/mL. The number of each peak corresponds to the number of each standard shown in Fig. 1.

Fig. 4. Relationship between log k values of eleven standards in C₈-HPLC and their log P_o/w values.

Notes: The inset regression equation (A) was obtained from six monomeric flavan-3-ols (1a–1f). The regression equation (B) was for EC (1c) with respect to four EC oligomers (2b, 3a, 4a, 5a) linked with (4β→8) bond

Quantitative evaluation of hydrophobic interaction between flavan-3-ols and C₈-alkyl moiety of OSA·Na added to aqueous mobile phase

It is difficult to quantify the strength of hydrophobic interaction between a solute and an alkyl stationary phase of a column in RP-HPLC, because it is not easy to change and estimate the amount of alkyl groups on a silica surface in an LC column. Instead of attempting direct quantification, we dissolved small amounts of 1-octanesulfonic acid sodium salt (OSA·Na) in the aqueous mobile phase used in C₈-HPLC, and indirectly evaluated the hydrophobic interaction between the solute flavan-3-ols and the C₈-alkyl moiety of the OSA·Na in the mobile phase. OSA·Na has been used as a water-soluble ion-pair reagent for polar cationic solutes in conventional RP-HPLC. In this study, it was used as a hydrophobic interaction competitor with the C₈-alkyl stationary phase for nonionic flavan-3-ols in a mobile aqueous phase containing TFA.

Fig. 5 compares the C₈-HPLC profiles of EC oligomers and related flavan-3-ols under the two conditions, with or without 10 mM OSA·Na in the mobile phase. When OSA·Na was added (Fig. 5(B)), the retention time of monomeric EC (1c) scarcely changed while those of EC oligomers, in particular 3a, 4a, and 5a, were significantly decreased. The decrease in the retention times of the EC oligomers means that the OSA·Na added in the mobile phase formed a complex with the oligomers by hydrophobic interaction and interfered with the retention of the oligomers by the C₈-alkyl stationary phase. We expected that any decrease in the retention time of a solute would depend on the OSA·Na concentration. Therefore, further C₈-HPLC experiments were carried out under various OSA·Na concentrations (0–20 mM), and the k values of the same flavan-3-ols were calculated from the resulting chromatograms.

Fig. 5. Comparison of C₈-HPLC profiles of EC oligomers and related flavan-3-ol standards in different mobile phases.

Notes: A: acetonitrile–aqueous 0.1% TFA (10:90, v/v); B: solvent A containing 10 mM OSA·Na. Other chromatographic conditions were identical to those described in Fig. 3 caption.

Fig. 6 shows scatter plots for the concentration of OSA·Na added to the mobile phase vs. the log k value of the flavan-3-ols in C₈-HPLC for each mobile phase. The log k value of all solutes tended to decrease with increasing OSA·Na concentration in the mobile phase. The OSA·Na concentrations (=x) were negatively correlated with the log k (=y) values of each solute, and all relationships could be expressed as a simple linear regression. For compound 1c: y = −0.0042x + 0.4439 (r²= 0.9154), 1e: y = −0.0126x + 1.0524 (r²= 0.9944), 2a: y = −0.012x + 0.0192 (r²= 0.9796), 2b: y = −0.0124x + 0.3261 (r²= 0.9926), 3a: y = −0.0215x + 0.5999 (r²= 0.9899), 4a: y = −0.0285x + 0.6944 (r²= 0.9816), 5a: y = −0.0329x + 0.8059 (r²= 0.9154). Each linear regression could be expressed as general expression: y = −Sx + a, and each flavan-3-ol had a different value for the slope (S). A compound with a large S value must more strongly interact hydrophobically with the OSA·Na in the mobile phase. Therefore, we concluded that the S value would be an appropriate index value for the strength of hydrophobic interaction between a solute flavan-3-ol and the C₈-alkyl moiety of OSA·Na. The S value of EC monomer (1a: 4.2 × 10⁻³) was smallest, the values of 1e, 2a, and 2b (12.6 × 10⁻³, 12.0 × 10⁻³, 12.4 × 10⁻³, respectively) were almost the same, and the values of EC oligomers increased with increasing DP (3a: 21.5 × 10⁻³, 4a: 28.5 × 10⁻³, 5a: 32.9 × 10⁻³) (Fig. 6). Thus, the hydrophobic interaction between OSA·Na and EC pentamer (5a) was the strongest of the seven flavan-3-ols.

Fig. 6. Relationship between the concentration of OSA·Na added to mobile phase (solvent A) and log k values of seven standards in C₈-HPLC in each mobile phase.

In addition to these flavan-3-ols, we measured the S values of four monomeric flavan-3-ols, six phenolic acids, and five flavonols. Each S value was calculated from each scatter plot (as shown in Fig. 6) for the concentration of OSA·Na added to the mobile phase vs. the log k of each solute in C₈-HPLC for each mobile phase. The comparison data of their S values is shown in Fig. 7. Of all the polyphenolic compounds tested, the compounds most strongly interacting with the C₈-alkyl moiety of OSA·Na were oligomeric PAs larger than dimers (i.e. 3a, 4a, and 5a).

Fig. 7. Comparison of S values of twenty-two polyphenolic standards.

Structural and physiological characteristics of PAs

LC is not only a useful tool for separation, but also for interaction analysis, because the retention behavior of a solute is especially sensitive to intermolecular forces acting in the separation column. In our previous HILIC study,⁸⁾ water-soluble hydrophilic PAs were strongly retained by an amide-silica column in an organic mobile phase due to hydrogen bonding. On the other hand, in the present RP-HPLC study, PAs were also retained by an aliphatic C₈-alkyl column in an aqueous mobile phase by hydrophobic interaction. In summary, PA has a dual nature, i.e. hydrophilicity and hydrophobicity. Presumably, the amphiphilic nature of PA is caused by its high-order oligomeric structure linked through C₄–C₈ interflavan bonds, suggested by the CD and FL data shown in Fig. 2. We speculate that higher DP PAs have a helical structure comprising a hydrophilic surface covered by a large number of hydroxyl groups and an internal hydrophobic cavity including stacked rings. Based on these structural characteristics, the PA would interact with biogenic substances such as proteins, peptides, carbohydrates, lipids, and oligonucleotides, resulting in positive (or sometime negative) physiological functions. Under aqueous physiological conditions (for example, in the digestive tract, cell fluid, or blood), nonspecific hydrophobic interaction may predominate between PA and biogenic substances, whereas under physiological conditions resembling organic solvents (for example, in lipid bilayer membranes or in large receptors or enzymes), specific hydrogen bonding may predominate. To clarify such interaction mechanisms in vivo, detailed high-order structural analyses of oligomeric PA are needed.

Author contributions

A.Y. and Y.S. designed the research plan, S.T. and A.Y. performed the experiments, S.T. and A.Y. analyzed the data, A.Y. and Y.S. supervised the research, and A.Y. wrote the manuscript.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

Abbreviations: PA, proanthocyanidin; DP, degree of polymerization; RP, reversed-phase; HPLC, high-performance liquid chromatography; OSA·Na, 1-octanesulfonic acid sodium salt; EC, (–)-epicatechin; CD, circular dichroism; FL, fluorescence.