Background: Because olive oil is an important component of the Mediterranean diet, it is necessary to establish unequivocal identification of the major potential antioxidant phenolic compounds it contains.
Methods: The major phenolic antioxidants in extra virgin olive oil were isolated and purified. Structural analysis was conducted using several spectroscopic techniques, including mass spectrometry and nuclear magnetic resonance (NMR). In particular, detailed 1H and 13C NMR data are presented, and several assignment errors in the literature are corrected.
Results: The data show for the first time that the lignans (+)-1-acetoxypinoresinol and (+)-pinoresinol are major components of the phenolic fraction of olive oils. These lignans, which are potent antioxidants, are absent in seed oils and virtually absent in refined virgin oils but are present at concentrations of up to 100 mg/kg (mean ± SE, 41.53 ± 3.93 mg/kg; range, 0.65–99.97 mg/kg) in extra virgin oils. As with the simple phenols and secoiridoids, there is considerable interoil variation in lignan concentrations. Foods containing high amounts of lignan precursors have been found to be protective against breast, colon, and prostate cancer.
Conclusion: Lignans, as natural components of the diet, may be important modulators of cancer chemopreventive activity.
The Mediterranean diet with its high content of fruits, vegetables, fiber, fish, and olive oil represents a healthy and disease-preventive diet (1). Its protective effect appears quite broad in that it not only has a chemoprotective effect against cancer (especially of the colorectum and breast) but also significantly reduces mortality from heart disease (2)(3).
Recent data suggest that the components of dietary olive oil may have a greater role in disease prevention than previously thought (4)(5)(6)(7)(8). Olive oil can be consumed in the natural unrefined state, known as extra virgin oil quality (VOQ),1 or as a refined product. The refined product (Fig. 1⇓ ) is made either from virgin oil and called refined virgin oil (RVO) or from solvent-extracted oil (9) and called refined husk oil (RHO).
Virgin and refined oils differ little in fatty acid composition (Table 1⇓ ). Monounsaturated oleic acid (C18:1) is the main component, but there are nutritionally relevant contributions from saturated palmitic acid (C16:0) and the essential polyunsaturated linoleic acid (C18:2), whereas only a small amount of linolenic acid (C18:3) is present. The proportion of linoleic acid, which is especially prone to undesirable oxidation by several processes, tends to be higher in oils from the more southern and warmer regions of the Mediterranean compared with oils from other areas.
VOQ, RVO, and RHO differ slightly in composition if the minor components, which account for 1% of VOQ and ∼3% of RHO, are considered (10). The higher concentrations of minor components in RHO compared with VOQ and RVO are chiefly attributable to fatty acid derivatives of long-chain alcohols (waxes) and to triterpenic alcohols. Many of these compounds contribute to the classical flavor of the oil, and some are detrimental to the taste, imparting bitterness or a burning sensation (11). The high stability (shelf life) of olive oil is partly attributable to its high content of oleic acid, which is less prone to oxidation than linoleic and linolenic acids. In addition, simple and complex phenolic compounds (tocopherols), including vitamin E, provide a strong stabilizing effect (12). Because refining causes a partial loss of antioxidants and their preservative actions, RVO and RHO usually are blended with a portion of VOQ to restore some of the desirable flavor and storage characteristics.
As part of our investigations into the chemopreventive effects of the Mediterranean diet, we considered it important to obtain a detailed profile of the major phenolic components in olive oil. To this end, we have studied a range of olive oils in comparison with seed oils to evaluate not only their compositions but also the antioxidative capacity of the major phenolic components (13). We found that olive oils (especially extra virgin) contain, in addition to the previously identified simple phenols and secoiridoids (SIDs), two compounds belonging to the lignan class of phenolics. Here we describe the isolation, purification, and structural elucidation of these lignans and related phenolics (Figs. 2⇓ and 3⇓ ),using various analytical techniques.
Materials and Methods
Authentic samples of the following compounds were purchased and used as references for structure elucidation: tyrosol (Sigma-Aldrich), and oleuropein glucoside (OLG; Extrasynthese Z.I. Lyon Nord). Acetic acid and methanol were obtained from E. Merck.
(3,4-Dihydroxyphenyl)ethanol (hydroxytyrosol; VII) was either synthesized according to a modification (the hydroxyl groups were not protected by acetylation) of the procedure of Papadopoulos and Boskau (14) or else obtained by acid hydrolysis of OLG. Acid hydrolysis was achieved by dissolving 100 mg of the glucoside in 1 L of 0.5 mol/L H2SO4 and incubating at 37 °C for 3 h. The hydrolysate was extracted twice with 250 mL of ethyl acetate and dried over anhydrous magnesium sulfate; the solvent was removed under reduced pressure. Hydroxytyrosol was purified by semipreparative HPLC to give an oily residue as the chromatographically and spectroscopically pure product.
extraction of oil
An olive oil (olio di oliva extravergine “ligustro” prodotto e confezionato dalla carapelli firenze s.p.a. via b. cellini 75 tavarnelle val di pesa) known to contain considerable quantities of linked phenols (13) was chosen for the isolation of sufficient quantities of these substances for detailed analytical and spectroscopic analyses.
The oil (500 g) was extracted three times with methanol (100 mL) for 1 h with vigorous shaking (5000 rpm). The methanol extracts were pooled, and the solvent was removed under reduced pressure at 35 °C. The residue was resuspended in acetonitrile (50 mL) and extracted three times with hexane (50 mL) to remove lipid components. The hexane extracts were discarded, and the acetonitrile solution was dried over anhydrous magnesium sulfate. The acetonitrile was removed under reduced pressure (35 °C), and the dried residue was resuspended in methanol (1.0 mL).
preparative thin-layer chromatography
Preparative thin-layer chromatography (PLC) was conducted on glass plates (20 × 20 cm) precoated with silica gel 50 F254 (Machery-Nagel). The extract (100 μL) was applied to the origins of 10 plates, and these were developed in the mobile phase ethyl acetate-isooctane-acetic acid (45:45:10, by volume). Bands on the thin-layer chromatography plates were detected under ultraviolet (UV) light (254 nm), and where relevant were scraped from the plates, pooled, and extracted three times with ethyl acetate. Solvent was removed under reduced pressure, and the extracts were resuspended in methanol (500 μL) for semipreparative HPLC.
Analytical HPLC was conducted on a Hewlett-Packard (HP) 1090 liquid chromatograph under computer control (HP Chemstation operating under Microsoft® WindowsTM) using a C18 (Latex), reversed-phase 25-cm (5 μm) column (4 mm i.d.) as described previously (13)(15)(16)(17); semipreparative HPLC was conducted on a similar column (10 mm i.d.).
gas chromatography–mass spectrometry
Gas chromatography–mass spectrometry (GC-MS) analyses were performed using a HP 5970 quadrupole mass spectrometer coupled to a HP 5890 gas chromatograph as described previously (13).
electrospray ionization-mass spectrometry
Electrospray ionization (ESI) mass spectra were recorded on a Finnigan MAT TSQ 7000 mass spectrometer using methanol solutions and both the positive- and negative-ion modes. Molecular masses were determined by analysis of the spectra in terms of the detected molecular and bimolecular ions and ion clusters including sodium.
nuclear magnetic resonance spectroscopy
Nuclear magnetic resonance (NMR) spectra were recorded in CDCl3 or CD3OD solvent on Bruker AC-250 and AM-500 spectrometers (Bruker Analytik) at 1H frequencies of 250.133 MHz (compounds IV and VI) and 500.135 MHz (compounds I, V, and VII–X), respectively, or 13C frequencies of 62.896 and 125.759 MHz, using conventional one-dimensional Fourier transform techniques (one-dimensional 1H, 13C with broadband 1H decoupling, 13C DEPT) for all compounds. Detailed analysis of zero-filled, resolution-enhanced spectra (peak picking, integration, and multiplet analysis) was performed using Bruker’s WIN-NMR software for PCs. 1H and 13C chemical shifts (δ) are reported in ppm relative to the internal standard tetramethylsilane (TMS; δ = 0) in CDCl3 (δ = 7.26 for 1H, 77.02 for 13C) or relative to residual CHD2OD (δ = 3.30 ppm for 1H) or CD3OD (49.0 for 13C), whereas H,H scalar couplings (J) are given in Hz. Most signal assignments could be made directly on the basis of the 1H multiplet patterns, the carbon multiplicities (DEPT), and chemical shift predictions made using the empirical increment rules of SpecTool (Ver. 2.1; Chemical Concepts) and the 13C data bank (22000 measured spectra) in Bruker’s WIN-SpecEdit software, which uses hierarchically ordered spherical description of environment (HOSE) codes for carbon atom environments (substructures) to predict shifts from measured data. To resolve remaining assignment ambiguities, 1H-coupled 13C NMR spectra were obtained from compounds I, V, and VII, and the necessary selective 1H decoupling experiments were performed. Finally, for the lignan VIII, two-dimensional NMR experiments [long-range H,H-correlation spectroscopy, 1H-rotating frame Overhauser enhancement spectroscopy (ROESY), and one-bond C,H correlation] were performed using conventional pulse sequences.
For acid hydrolysis of the crude extract and pure products, a 10-μL aliquot was transferred to a 15-mL glass centrifuge tube. Methanol was removed under a stream of nitrogen, and the residue was resuspended in 0.5 mol/L H2SO4 (100 μL) and incubated at 37 °C for 3 h on a shaking water bath. After incubation, the hydrolysates were extracted twice with ethyl acetate (2 mL). The pooled ethyl acetate extracts were dried under a stream of nitrogen and resuspended in methanol (100 μL). These solutions were used for analytical HPLC and, after derivatization with bis(trimethylsilyl)trifluoroacetamide, for GC-MS analysis.
Alkaline hydrolysis was conducted in a similar manner using 100 g/L KOH in methanol instead of 0.5 mol/L H2SO4. After incubation, the hydrolysates were diluted with distilled water (100 μL) and acidified to pH 2.0 with concentrated HCl and extracted as described above.
For large-scale alkaline hydrolysis, a delipidated extract of the oil (500 g) was resuspended in 100 g/L KOH in methanol (5.0 mL) and, after incubation and acidification, was extracted three times with ethyl acetate (25 mL). After removal of the solvent under reduced pressure, the hydrolyzed products were separated into crude fractions by PLC before purification by semipreparative HPLC.
isolation and purification of olive oil components
Hydroxytyrosol (VII) was synthesized from 3,4-dihydroxyphenyl acetic acid by the method of Papadopoulos and Boskau (14). HPLC (Fig. 4⇓ A) of the raw product mixture showed that the purity (yield) of hydroxytyrosol was ∼50%, and this was confirmed by GC-MS (Table 2⇓ ). The yield of hydroxytyrosol from the acid hydrolysis of OLG was essentially quantitative (Fig. 4B⇓ ); therefore, this route is more convenient for the synthesis of this phenolic compound, which is not available commercially. A portion of the product (20 mg) was purified by preparative HPLC and analyzed by 1H and 13C NMR (see below).
Extraction of phenolic compounds.
The HPLC chromatogram of the large-scale methanol extract of an extra virgin olive oil (500 g) is shown in Fig. 5⇓ . In the crude extract, seven major HPLC peaks (Fig. 5⇓ , peaks 1–4, 6, and 7) corresponded to the known compounds hydroxytyrosol (VII); tyrosol (VI); SID V and SID IV, which are dialdehydes related to oleuropein and ligstroside, respectively, but lacking the carboxymethyl group at C4 (as shown in Fig. 2⇑ ); oleuropein (III); and the aglycone of ligstroside (II). Peak 5 in the HPLC chromatogram contained the lignans whose structures are shown in Fig. 3⇑ .
Other simple phenols that have been reported for extracts of olive oil (10 g) could not be detected routinely by HPLC, even when 100-g samples were extracted. However, during the preparative isolation of the major components by PLC, extraction of minor bands on the thin-layer chromatography plates and subsequent GC-MS analysis revealed that vanillic acid, p-hydroxybenzoic acid, p-coumaric acid, and o-coumaric acid were present at very low concentrations, but their definitive quantification was not attempted. These findings are in agreement with the data of Angerosa et al. (18) but are contrary to that of Montedoro et al. (15).
Acid hydrolysis of the methanol extract changed the HPLC profile dramatically (Fig. 6⇓ a). Both SIDs IV and V disappeared with concomitant increases in the concentrations of tyrosol and hydroxytyrosol, respectively. This is consistent with the hydrolysis of the ester group linking the elenolic acid and phenol moieties of the SIDs. HPLC peak 5 was unaffected by acid hydrolysis, indicating that the corresponding compound(s) did not possess the SID structure.
The HPLC chromatogram (Fig. 6B⇑ ) after alkaline hydrolysis of the methanol extract yielded only tyrosol from SID IV (Fig. 6B⇑ , peak 2) because hydroxytyrosol from V is destroyed in strong alkaline solution. A new peak (Fig. 6B⇑ , peak 5b) appeared in the spectrum, whereas peak 5 (Fig. 6A⇑ ) was appreciably diminished (Fig. 6B⇑ , peak 5a). These results suggested that two compounds were co-chromatographing as peak 5 (confirmed by NMR), both resistant to acid hydrolysis and one resistant to alkaline hydrolysis.
PLC of the crude methanol extract yielded two fractions, and analytical HPLC showed that fraction 1 contained predominantly tyrosol (VI) and SID IV, whereas fraction 2 contained predominantly hydroxytyrosol (VII), SID V, and the lignans. This crude separation of IV (HPLC peak 4) from the lignans (HPLC peak 5) by PLC greatly facilitated subsequent purification by HPLC.
Purification of the PLC extracts by semipreparative HPLC gave the pure SIDs IV (10.0 mg) and V (9.0 mg). GC-MS and NMR analyses showed that the lignans (8.0 mg, corresponding to HPLC peak 5 in Fig. 5⇑ ) were a mixture of a major and a minor compound, in agreement with the alkaline hydrolysis data. Analytical HPLC of PLC fraction 2 indicated that the minor component (∼17.5%) was concentrated in the trailing edge of the lignan HPLC peak (chromatogram not shown). This facilitated further cycles of purification, giving the major component (VIII) in pure form but reduced yield (2.0 mg).
Alkaline hydrolysis of the methanol extract followed by PLC and preparative HPLC led to the isolation of the minor component (HPLC peak 5a, IX, 2.0 mg) and the alkaline hydrolysis product of the major component (HPLC peak 5b, X, 4.0 mg).
These lignans were found to be absent in seed oils and virtually absent in refined virgin olive oils, but they could be isolated at concentrations of up to 100 mg/kg (mean ± SE, 41.53 ± 3.93; range, 0.65–99.97 mg/kg; n = 18) in extra virgin oils. As for the simple phenols and SIDs, we found considerable interoil variation in lignan concentrations.
structure elucidation by ms and nmr
Simple phenols and SIDs.
The structures of the simple phenols and SIDs are shown in Fig. 2⇑ , and mass spectra data (peak lists) for the phenolic olive oil components of interest are summarized in Tables 2⇑ and 3⇓ . The GC-MS spectra (not shown) obtained from TMS derivatives of tyrosol (VI), hydroxytyrosol (VII), and SIDs IV and V are identical with those published previously (18)(19). The molecular masses were further confirmed by ESI-MS data after taking into account possible cluster formation.
The 1H and 13C NMR data for hydroxytyrosol and tyrosol, SIDs IV and V, and for OLG (I) are given in Tables 4–6⇓ ⇓ ⇓ . Signal assignments were made using the numbering scheme shown in Fig. 2⇑ . Montedoro et al. (17) have also published 1H and 13C NMR data for these compounds, and their 13C chemical shifts agree very closely with our values for all compounds. However, in their study, Montedoro et al. (17) gave an incorrect 1H chemical shift for H8′ of tyrosol, which should be equivalent by symmetry with H4′, and they interchanged the assignments for C4′,8′ and C5′,7′ for tyrosol and the related SID IV. This latter error is clearly revealed by chemical shift predictions, which give 129–131 and 115–116 ppm for C4′ and C5′, respectively (observed: 130.86 and 116.13 for tyrosol; 130.2 and 115.3 for p-hydroxytoluene). Finally, the assignments for C2′ and C6 in SID IV remained ambiguous based on chemical shift predictions. Therefore, we confirmed the assignments made by Montedoro et al. (17) by performing selective 1H decoupling experiments (see below).
For hydroxytyrosol (VII) two pairs of 13C assignments were ambiguous at the level of chemical shifts alone. The quaternary carbons C5′ and C6′ exhibited shifts near 146 ppm with a shift difference of 1.5 ppm. Shift predictions by increments rules suggested that C5′ should have the larger shift value, whereas Montedoro et al. (17) assigned C6′ to the larger shift without providing convincing proof. Similarly, the protonated carbons C4′ and C7′ gave resonances near 117 ppm with a shift difference of only 0.8 ppm. Shift predictions generally suggested that C7′ should have the larger shift, as assigned by Montedoro et al. (17). To resolve these ambiguities, we measured the fully 1H-coupled 13C spectrum of hydroxytyrosol and performed appropriate selective 1H decoupling experiments. The long-range CH coupling patterns in the aromatic ring are unique for each site, following the rule 3J > 2J > 4J. Thus, the assignments given in Table 6⇑ are unambiguous (footnotes to Table 6⇑ for diagnostic coupling constants), and correct the earlier erroneous assignments (17).
For SID V and OLG (I), fully coupled and selectively decoupled 13C NMR spectra were also obtained to resolve all ambiguities involving the hydroxytyrosol ring carbons, the olefin carbon C8 and the methylene carbons C2′ and C6. The results are summarized in Table 6⇑ and its footnotes.
Our 1H NMR analysis agrees with most of the results and assignments obtained by Montedoro et al. (17) at lower magnetic field strength. There are some differences in the absolute values of chemical shifts for the aromatic protons, but the order of shifts and their assignments agree with our results. However, at a higher field and with resolution enhancement we were able to resolve a variety of long-range couplings and the following nonequivalent methylene protons: H1′a,b in OLG and the SIDs; H2′a,b in SID V; H4a,b in SIDs IV and V; H6a,b in OLG and the SIDs. The assignments given by Montedoro et al. (17) for H4a,b and H6a,b and their geminal couplings are incorrect.
[+]-1-Acetoxypinoresinol (VIII). The UV spectrum indicated the presence of a phenolic compound [λmax, nm (log ε): 230 (3.94), 279 (3.37)]. The ESI mass spectrum (Table 3⇑ ) afforded a signal for [M+H]+ at m/z 417, consistent with the molecular formula C22H24O8 (exact mass, 416.147). Formation of the trimethylsilyl derivative (GC-MS spectrum in Fig. 8⇓ A) led to an increase of the molecular mass by 144 units, indicating the presence of two free hydroxyl groups (Table 2⇑ ). Whereas acidic hydrolysis exerted no effect, alkaline hydrolysis led to a product (X) with similar UV characteristics [λmax, nm (log ε): 232 (4.16), 280 (3.73)] and a molecular mass of 374 (Table 3⇑ ), consistent with the replacement of one acetate group by a hydroxyl. This was corroborated by the GC-MS spectrum of the trimethylsilyl derivative, which gave a molecular mass of 518 (Fig. 9⇓ A) after short-term (15 min) and 590 (Fig. 9B⇓ ) after overnight derivatization (Table 2⇑ ).
The 500 MHz 1H NMR spectrum of VIII (Fig. 7A⇑ ) showed seven aliphatic protons at δ = 3.33–5.06 and singlets at δ = 1.703, 3.891, and 3.923, indicating the presence of an acetate ester and two aromatic methoxy groups. A total of six aromatic protons were detected in the range δ = 6.85–6.97. Three multiplets could be resolved, which showed the characteristic pattern of ortho and meta coupling constants (Table 8⇑ ) for the three protons of 1,3,4 tri-substituted benzene rings (H2″, H5″, H6″ in Table 7⇑ ) with an additional long-range coupling from H6″ to the aliphatic proton H6. For the second ring system, the multiplet for H2′ was resolved, but H5′ and H6′ had nearly identical chemical shifts and gave a nonclassical pattern of three closely spaced resonances. The chemical shifts for H5′ and H6′ were assigned via a two-dimensional C,H correlation experiment. The discrimination between the “prime” ring system attached to C2 and the “double prime” system attached to C6 was made easily via an H,H-correlation spectroscopy experiment performed with an evolution delay that detected the long-range couplings between H2′ and H2 and between H6 and H2″,5″,6″. At the same time, each 3-OMe group could be assigned via long-range coupling to the corresponding ring proton at the C2′ or C2″ position. On the basis of nuclear Overhauser effects observed via a two-dimensional ROESY experiment, we tentatively assign H2, H4b, H6, and H8b to pseudo-axial positions above the plane of Fig. 3⇑ .
Deacetylation of VIII by alkaline hydrolysis to give X produced upfield shifts for C1 and for the proton signals at 5.06, 4.44, 4.24, and 3.33 ppm; a substantial downfield shift was observed for one of the aromatic protons (H2′), and substantial changes in carbon shifts were observed only for the prime ring system. Putting all of this information together, we deduced that these compounds were lignans with a 2,6-diaryl-3,7-dioxabicyclo[3.3.0]octane skeleton with the prime aryl ring attached to C2 and the double prime ring attached to C6. The chemical shifts for the prime aryl ring and the protons at C2,5,8 are particularly sensitive to the substituent at C1.
With the C,H correlation data for VIII, unequivocal assignments for all protons and protonated carbons were obtained. Quaternary aryl carbons C1,3,4 were assigned via shift predictions. The only ambiguity remaining is the discrimination between C3′ and C3″ and between C4′ and C4″. In Table 9⇑ , we have made the assignments based on the assumption that there will be minimal perturbations in the chemical shifts for C3″,C4″ for the series VIII–X, whereas larger effects are expected for C3′,C4′.
On the basis of the above data, the structures of VIII and X were elucidated as (+)-1-acetoxypinoresinol and (+)-1-hydroxypinoresinol, respectively. 13C NMR data (15 MHz) for these compounds in deuterated dimethyl sulfoxide (DMSO-d6) have been reported earlier by Tsukamoto et al. (20), and substantial absolute differences in chemical shifts (solvent effects) in comparison with our data are apparent. Nevertheless, our assignments in terms of the order of chemical shifts agree completely with the earlier work for VIII and differ only for C3′, C3″, and C5′,C5″ in X. Agrawal and Pathak (21) have tabulated 13C reference data for a large number of lignans, including IX and X, whereby the chemical shifts of the carbons of the 3,7-dioxabicyclo[3.3.0]octane moiety are found to be diagnostic for its stereochemistry and substitution pattern. Our data for X show excellent agreement with the data for structures with the stereochemistry as shown in Fig. 3⇑ .
[+]-Pinoresinol (IX). The UV spectrum [λmax, nm (log ε): 231 (4.18), 281 (3.82)] indicated the presence of a compound that was similar to both VIII and X. The ESI mass spectrum (Table 3⇑ ) gave a molecular ion signal [M+H]+ at m/z 359, consistent with the formula C20H22O6. Trimethylsilylation yielded a molecular mass of 502 by GC-MS (Table 2⇑ and Fig. 8B⇑ ), indicating the presence of two free hydroxyl groups. The compound was stable to both alkaline and acid treatment. The NMR of IX (Tables 7–9⇑ ⇑ ⇑ ) revealed the twofold symmetry of this compound. The close similarity of the spectra to those of VIII and X and the reference data of Agrawal and Pathak (21) indicated that IX is a symmetrical lignan with the 2,6-diaryl-3,7-dioxabicyclo[3.3.0]octane skeleton and stereochemistry shown in Fig. 3⇑ . Thus, we concluded that compound IX was (+)-pinoresinol, which has a C2 axis perpendicular to the plane of Fig. 3⇑ .
The results of this study show for the first time that the lignans (+)-1-acetoxypinoresinol (VΙΙΙ) and (+)-pinoresinol (X) are major components of the phenolic fraction of olive oils. In extra virgin oils, the quantity of lignans can be up to 100 mg/kg, but as with the simple phenols and SIDs, considerable interoil variation exists. The reasons for this probably lie in differences in the production areas, climate, olive varieties, and oil production techniques.
(+)-Pinoresinol is a common component of the lignan fraction of several plants such as Forsythia species (22) and Sesamum indicum seeds (23), whereas (+)-1-acetoxypinoresinol and (+)-1-hydroxypinoresinol (ΙX) and their respective glucosides have been detected (20)(24) in the bark of the olive tree (Olea europaea L.). The question of how these lignans become a major component of the phenolic fraction of olive oil has yet to be answered. Recent studies (Owen et al., manuscripts in preparation) have shown that they are not present in the pericarp of the olive drupes, nor in the leaves and twigs that may be present in the mulch prepared for olive pressing. However, preliminary studies have shown that (+)-pinoresinol is a major component of the phenolic fraction of the olive piths (Owen et al., manuscripts in preparation), and further information will be forthcoming from our ongoing studies.
Clearly, lignans constitute an important contribution to the phenolic fraction of olive oil and therefore may play a major role in the health promoting effects of the Mediterranean diet. The evidence for the chemopreventive effect of lignans is compelling. They are a group of compounds found only in plants, and a range of different classes have been described. Interest in lignans as possible chemopreventive agents began ∼20 years ago when it was discovered that certain classes could be identified and quantified in biological material, especially human urine. The major lignans identified in humans are now termed enterolactone and enterodiol, and they are described as resulting from bacterial transformation in the large bowel of matairesinol and secoisolariciresinol, respectively (25)(26)(27).
Lignans have been shown to inhibit skin, breast, colon, and lung cancer cell growth (28)(29) and to exert antiestrogenic effects. This is not surprising, considering the structural similarity between the lignans and the synthetic antiestrogen tamoxifen (30)(31). Foods containing high amounts of lignan precursors have been found to be protective against breast cancer, in particular. Consumption of flaxseed, a concentrated source of lignans (32), has been shown to inhibit mammary tumor promotion (33) and development of early markers of mammary carcinogenesis (34). Several studies have shown that the concentration of lignans in the urine of breast cancer patients is lower than in the urine of omnivorous and vegetarian women and of women with low risk of breast cancer (35)(36)(37).
Other properties attributed to lignans are the inhibition of cyclic AMP phosphodiesterase (38) and fatty acid Δ-5 desaturase activity (39). Lignans are also potent antioxidants in vitro (13) and inhibit lipid peroxidation in vivo (40). Therefore, it appears that by reducing oxidative stress they may be important dietary modulators of cancer chemopreventive activity.
We thank G. Schwebel-Schilling for measuring the NMR spectra at 250 MHz, and G. Erben for ESI-MS spectra.
1 VOQ, RVO, and RHO.
1 Each TMS group replaces a hydroxyl proton and adds 72 to the original molecular mass.
2 TMS reaction performed overnight.
1 ND, not detected.
1 Chemical shifts at 30 °C are in ppm relative to TMS = 0 in CDCl3 or relative to CHD2OD = 3.30; basic multiplet type is given in parentheses; tyrosol and SID IV were measured at 250 MHz, the other compounds at 500 MHz.
2 Shifts for the glucose moiety: H1, 4.796 (d); H2, 3.315 (dd); H3, 3.411 (dd); H4,H5, unresolved at 3.33 ppm; H6, (a) 3.876 (dd), (b) 3.668 (dd).
3 t, triplet; dt, doublet-triplet; dd, doublet-doublet; tt, triplet-triplet; ddt, doublet-doublet-triplet; d, doublet; ddq, doublet-doublet-quartet; ddd, doublet-doublet-doublet; m, multiplet; q, quartet; s, singlet.
1 See footnote a in Table 4⇑ ; except where noted, couplings in Hz were determined by first-order analysis; the signs of geminal and vicinal couplings are assumed to be negative and positive, respectively; the signs of long-range couplings have not been determined.
2 Resolved couplings for the glucose moiety: 3J1,2 = 7.70; 3J2,3 = 9.2; 3J3,4 = 8.8; 2J6a,6b = −11.95; 3J5,6a = 1.92; 3J5,6b = 5.57.
3 Observed pseudo-triplet splitting for H1′ was 7.2 Hz; data in Table 5⇑ were calculated by AA′XX′ spin system analysis of the H1′ multiplet.
4 Observed dd splittings for H1′ were 6.99 and 7.49 Hz; data in Table 5⇑ were calculated by AA′XX′ spin system analysis of the H1′ multiplet.
5 The couplings for H4′–H8′ were obtained by AA′XX′ spin system analysis; the observed pseudo-doublet splitting for H4′,8′ and H5′,7′ was 8.69 Hz for tyrosol and 8.64 for SID IV.
1 Chemical shifts at 30 °C are relative to TMS = 0 (CDCl3 = 77.014) or CD3OD = 49.0; multiplicities in parentheses; tyrosol and SID IV were measured at 62.9 MHz, the other compounds at 125.76 MHz.
2 Shifts for glucose moiety, C1–C6; 100.92, 74.76, 77.94, 71.47, 78.41, 62.73.
3 t, triplet; s, singlet; d, doublet; q, quartet.
4 Assignments according to Montedoro et al. (17) were unequivocally confirmed in this work.
5 Assignments here are based on reference data from analogous compounds; erroneously reversed in Montedoro et al. (17).
6 In Montedoro et al. (17), the assignments for C4′,C7′ are reversed; our assignments were unequivocally confirmed by long-range 1H couplings for C4′: 3J(H2′) = 5.2; 3J(H8′) = 7.4; 4J(H7′) = 1.2.
7 In Montedoro et al. (17), the assignments for C5′,C6′ are reversed; our assignments were unequivocally confirmed by 1H-coupled spectra and selective decouplings; e.g., the observed diagnostic long-range 1H couplings are 2J(H4′) = 3.2; 3J(H7′) = 7.1; 4J(H8′) = 1.2 for C′5; and 2J(H7′) = 3.0; 3J(H4′) = 6.7; 3J(H8′) = 9.7 for C′6.
8 The numbering scheme for C8,C9 is the reverse of that used by Montedoro et al. (17).
1 Chemical shifts at 500 MHz and 30 °C in ppm relative to TMS = 0 in CDCl3; multiplet type in parentheses.
2 m, multiplet; d, doublet; br, broadened; s, singlet; dd, doublet-doublet; ddd, doublet-doublet-doublet; dt, doublet-triplet; NR, multiplet not resolved; dm, doublet-multiplet.
1 Couplings in Hz at 500 MHz and 30 °C; the 2′,5′,6′ spin system was highly second order and could not be analyzed directly.
1 All measurements at 125.76 MHz (11.7 T) and 30 °C; chemical shift reference is TMS = 0 (CDCl3 = 77.02) or CD3OD = 49.0.
2 d, doublet; s, singlet; t, triplet; q, quartet.
3 Assigned by chemical shift predictions.
4 Assigned assuming that C3″,C4″ will show minimal shift perturbations across the series VIII–X.
↵1 Nonstandard abbreviations: VOQ, extra virgin olive oil; RVO, refined virgin olive oil; RHO, refined husk olive oil; SID, secoiridoid; OLG, oleuropein glucoside; PLC, preparative thin-layer chromatography; UV, ultraviolet; HP, Hewlett-Packard; GC, gas chromatography; MS, mass spectrometry; ESI, electrospray ionization; NMR, nuclear magnetic resonance; and TMS, tetramethylsilane.
- © 2000 The American Association for Clinical Chemistry