Proanthocyanidins (PA) are a class of dietary polyphenols. They are polymers of flavan-3-ols present in a wide variety of plant-based foodstuffs, such as berries, cocoa or certain nuts(Reference Serrano, Puupponen-Pimiä and Dauer1). Several supplementation studies both in animals and in human subjects using PA-rich products have shown that PA play a preventive role against several conditions including CVD(Reference Zern, Wood and Greene2–Reference Bladé, Arola and Salvadó4) or diabetes(Reference Serrano, Puupponen-Pimiä and Dauer1, Reference Montagut, Bladé and Blay5, Reference Yokozawa, Kim and Cho6). Similarly, a recent epidemiological study showed an inverse association between the intake of polymeric PA and the risk of colorectal cancer(Reference Rossi, Bosetti and Negri7). To advance knowledge of the possible effects of PA on human health, it is important to characterise their metabolic fate in detail, that is to say, the degradation they may suffer once ingested and the distribution of their metabolites through the different organs and fluids.
Most studies of PA, including those that address their metabolism, assume that the PA in foodstuffs correspond exclusively to the supernatants obtained after extracting the food with acetone; the most common procedure for their analysis(Reference Gu, Kelm and Hammerstone8). However, these PA would actually only correspond to the extractable proanthocyanidins (EPA), which represent only a fraction of dietary PA. Recent work has emphasised that a considerable proportion of PA, the non-extractable proanthocyanidins (NEPA), remains in the residue from such extractions(Reference Arranz, Saura-Calixto and Shaha9) and may very well play a more significant functional role than EPA. NEPA are associated with other components of the food matrix, mainly dietary fibre, and in fact constitute a part of it, according to current definitions of dietary fibre(Reference Goñi, Díaz-Rubio and Pérez-Jiménez10, Reference Report11). To date, there is no common method for determining NEPA. They are usually determined by destructive spectrophotometric methods and only a few recent papers have attempted structurally meaningful analysis(Reference Hellström and Mattila12, Reference White, Howard and Prior13); therefore, structural evidence regarding the composition of NEPA as essential constituents of many foodstuffs is still scarce. The currently available evidence suggests that NEPA may be more abundant than EPA in much food(Reference Hellström and Mattila12–Reference Pérez-Jiménez, Arranz and Saura-Calixto14) and, therefore, that significant amounts of NEPA are ingested daily.
Over the last decade, several studies have addressed the metabolism of PA. Although initial studies emphasised fairly poor intestinal absorption that was limited to dimers(Reference Abia and Fry15–Reference Tsang, Auger and Mullen17), later observations indicate that once intact PA reach the colon they are widely transformed by the colonic microbiota into small phenolic acids(Reference Baba, Osakabe and Natsume18–Reference Urpí-Sardá, Garrido and Monagas21). These metabolites are absorbed, and then transformed in the liver, and the resulting conjugates are transferred to the bloodstream. A recent study in which [14C]procyanidin B2, a labelled PA dimer, was administered to rats, reported bioavailability of around 80 %, based on total urinary 14C(Reference Stoupi, Williamson and Drynan20).
Nevertheless, these papers have mostly addressed the bioavailability of dimers or trimers, while the most abundant PA in food are polymers(Reference Neveu, Pérez-Jiménez and Vos22). Indeed, some recent studies have suggested that PA are also depolymerised into (epi)catechin (EC) units before cleavage into smaller species and further metabolism(Reference Urpí-Sardá, Garrido and Monagas21, Reference Rios, Gonthier and Rémésy23–Reference Touriño, Pérez-Jiménez and Mateos-Martín25). These studies suggest that phenolics that are bioavailable after the ingestion of PA-rich foodstuffs must have come from NEPA(Reference Touriño, Fuguet and Vinardell24–Reference Saura-Calixto, Pérez-Jiménez and Touriño26), but this has not been proved as the specific metabolism of NEPA has never been reported.
Grape antioxidant dietary fibre (GADF) is a food product obtained from red grapes that is rich in dietary fibre and polyphenols(Reference Pérez-Jiménez, Serrano and Tabernero27). Besides extractable polyphenols, including EPA(Reference Touriño, Fuguet and Jáuregui28), GADF contains a significant amount (14·8 %) of NEPA(Reference Pérez-Jiménez, Serrano and Tabernero27) and was used in the studies which suggest that non-extractable polyphenols are an important source of metabolites that are bioavailable in rats(Reference Touriño, Fuguet and Vinardell24–Reference Saura-Calixto, Pérez-Jiménez and Touriño26). To study the contribution of NEPA to the pool of phenolic metabolites from fruit and vegetables, we considered using a NEPA-rich fraction from GADF.
Our objective was to evaluate the fate of NEPA in rats 24 h after ingestion of a preparation free from any extractable polyphenols. NEPA metabolites, including hepatic and microbially derived metabolites, were analysed in urine and faeces using liquid chromatography coupled to a mass spectrometer equipped with an electrospray ionisation (ESI) chamber and a triple quadrupole mass analyser for tandem analysis (HPLC–ESI–QqQ–MS/MS).
Experimental methods
Reagents and samples
GADF was obtained from red grapes (Cencibel variety, harvested in 2005 in La Mancha region of Spain) by a patented procedure(Reference Saura-Calixto and Larraruri García29). The NEPA content of GADF has previously been reported to be 14·8 g/100 g of dry weight(Reference Pérez-Jiménez, Serrano and Tabernero27). To obtain an EPA-free (and therefore NEPA-rich) fraction, GADF (4 g) was defatted with hexane (3 × 40 ml), air-dried overnight and the residue was extracted with methanol–water (50:50, v/v, 200 ml) and then with acetone–water–acetic acid (70:29·5:0·5, by vol., 200 ml) once each at room temperature. The supernatant was decanted and the residue, including the NEPA-rich fraction, was vacuum filtered and lyophilised. The actual NEPA content of this residue is around 25 % according to published information(Reference Saura-Calixto, Pérez-Jiménez and Touriño26).
Standards of EC ( ≥ 97 %), 3- and 4-hydroxyphenylacetic acid ( ≥ 98 %), 3,4-dihydroxyphenylacetic acid ( ≥ 98 %), 3- and 4-hydroxybenzoic acid ( ≥ 97 %), vanillic acid ( ≥ 97 %), caffeic acid ( ≥ 95 %), 3,4-di-hydroxyphenylpropionic acid ( ≥ 98 %), 4-hydroxyphenylpropionic acid (>98 %), protocatechuic acid ( ≥ 97 %), ferulic acid ( ≥ 99 %), isoferulic acid ( ≥ 97 %), p-coumaric acid ( ≥ 98 %), m-coumaric acid ( ≥ 97 %) and taxifolin ( ≥ 85 %) were obtained from Sigma Chemical (St Louis, MO, USA). Methanol (analytical grade), phosphoric acid ( ≥ 85 %) and acetic acid were purchased from Panreac (Castellar del Vallès, Barcelona, Spain). Acetonitrile (HPLC grade) and formic acid (analytical grade) were obtained from Merck (Darmstadt, Germany). Water was purified by a Milli-Q plus system from Millipore (Bedford, MA, USA) to a resistivity of 18·2 mΩ/cm.
Animal experiments
The study was carried out on female Sprague–Dawley rats (n 10, body weight 233 (sd 9·3) g, 12 weeks of age) provided by Harlan Interfauna Ibérica SL (Barcelona, Spain). The animals were fed with a polyphenol-free diet (TD94048), also purchased from Harlan Interfauna Ibérica SL, and they were maintained in plastic cages at room temperature (22 ± 2°C) and 55 (sd 10) % relative humidity, with a 12 h light–12 h dark cycle for 1 week, in accordance with European Union regulations. After food deprivation for 12 h with free access to water, a group of animals (n 5) was administered a suspension of NEPA from GADF in tap water (1 g NEPA-rich fraction/10 ml, 1·6 g NEPA-rich fraction/kg body weight) by oral gavage, while a control group (n 5) was administered tap water (16 ml/kg body weight). The animals were then placed in metabolism cages and urine and faeces were collected over 24 h and stored at − 80°C until extraction and analysis. These experimental protocols were approved by the Experimental Animal Ethical Research Committee of the CSIC in accordance with the current regulations for the use and handling of experimental animals.
Sample preparation
The biological samples were prepared according to previously described procedures for the extraction of phenolic metabolites(Reference Urpí-Sardá, Garrido and Monagas21, Reference Touriño, Fuguet and Vinardell24, Reference Touriño, Pérez-Jiménez and Mateos-Martín25). Briefly, urine samples were concentrated via a nitrogen stream at room temperature and then resuspended in 1 ml of acid water (addition of phosphoric acid to reach pH 3). Taxifolin (100 μl of a 50 parts per million (ppm) solution) was added as an internal standard, to obtain a final concentration of 5 ppm. Then the samples were subjected to solid phase extraction in Oasis HLB (60 mg) cartridges from Waters Corporation (Mildford, MA, USA). The cartridges were activated with methanol (1 ml) and acid water (2 ml) and the samples loaded. To remove interfering components, the samples were washed with acid water (9 ml) and then the phenolic compounds were eluted with methanol (1 ml).
Faeces (0·5 g) were defatted with hexane (10 ml) and the residue was extracted with methanol–water–phosphoric acid (8:1·9:0·1, by vol., 10 ml) and concentrated down to 1 ml by nitrogen stream at room temperature. Taxifolin (100 μl of a 50 ppm solution, final concentration 5 ppm) was added to each sample as an internal standard.
Extracts from both urine and faeces were filtered through a polytetrafluoroethylene 0·45-μm membrane from Waters Corporation into amber vials for HPLC–MS/MS analysis.
HPLC–electrospray ionisation–MS/MS analysis
A Quatro LC from Waters Corporation triple quadrupole mass spectrometer with an electrospray source was used in negative mode to obtain MS and MS/MS data. Liquid chromatography separations were performed using an Alliance 2695 system from Waters Corporation equipped with a Phenomenex (Torrance, CA, USA) Luna C18 (50 × 2·1 mm internal diameter) 3·5 μm particle size column and a Phenomenex Securityguard C18 (4 × 3 mm internal diameter) column. Gradient elution was performed with a binary system consisting of (A) 0·1 % aqueous formic acid and (B) 0·1 % formic acid in CH3CN. An increasing linear gradient (v/v) was used (t (min), %B): 0, 8; 10, 23; 15, 50; 20, 50; 21, 100, followed by a re-equilibration step.
Metabolites were detected in multiple reaction monitoring (MRM) experiments, and their identity was confirmed by product ion scan experiments. The dwell time for the MRM experiments was 100 ms and the cycle time for all the experiments was 2 s. Cone energy and collision energy in MRM mode were optimised for each group of metabolites: 30 V and 10 eV for taxifolin, 30 V and 15 eV for microbially derived phenolic metabolite derivatives, 30 V and 20 eV for EC, and 40 V and 20 eV for EC derivatives.
Results
(Epi)catechin and its phase II metabolites
Free EC (MRM transition 289 → 245) and a signal corresponding to a dimer (577 → 289) were detected in faeces by liquid chromatography–ESI–MS/MS. The product ion spectrum of this dimer provided characteristic fragments at m/z 425 and m/z 405 originated by cleavage of the C-ring of one of the EC units through a retro Diels–Alder reaction and consecutive loss of water. A fragment at m/z 451, caused by heterolytic ring fission, was also observed.
A total of ten EC conjugates derived from the activity of phase-II enzymes in both the intestinal tract and liver were detected in urine samples (Table 1). All these compounds were either not detected in the control group or detected at concentrations that were at least 10-fold lower. The metabolites were initially identified by previously reported MRM transitions corresponding to the main fragments(Reference Touriño, Fuguet and Vinardell24, Reference Wishart, Tzur and Knox30) and their identity was confirmed by a second MRM transition and/or by product ion scan experiments.
Table 1 (−)-(Epi)catechin (EC) and conjugated metabolites in urine and faeces from rats fed a non-extractable proanthocyanidin-rich fraction from grape antioxidant dietary fibre
MRM, multiple reaction monitoring; PC, procyanidins; GlcA, glucuronide; Sulf, sulphate; GSH, glutathione; Me: methyl group.
The EC metabolites included three glucuronidated forms (465 → 289), two sulphated forms (369 → 289), a monoconjugated metabolite with glutathione (594 → 289), two methylated and glucuronidated forms (479 → 303), a di-glucuronidated form (641 → 289) and a tri-conjugated metabolite (397 → 289) corroborated by MS/MS experiments. All these derivatives were detected in urine samples and the di-glucuronidated conjugate was also detected in faeces.
Fig. 1 shows the product ion spectrum of glucuronide (GlcA)-EC-3 (m/z 465). Fragments at m/z 289 and m/z 245, corresponding to the loss of the conjugate moiety and the respective cleavage of CO2 from the free EC unit, were observed; as were two fragments at m/z 175 and 113, from the fragmentation of the GlcA moiety. The MS/MS spectra of other conjugates showed characteristic fragments corresponding to a B-ring retro Diels-Alder fission of the EC, such as the fragment at m/z 137 for Me-GlcA-EC, which corresponds to a B-ring fragment with attached methyl group and GlcA moieties, indicating that the conjugation was located on the B-ring.
Fig. 1 HPLC–electrospray ionization-MS/MS product ion scan spectrum of glucuronide-(epi)catechin (m/z 465). cps, Counts per second; amu, atomic mass unit; max., maximum.
Microbially derived proanthocyanidin metabolites
A total of twenty microbially derived PA metabolites were identified in urine from rats fed the NEPA-rich fraction; two of them were also detected in faeces (Table 2). All these metabolites were either not detected in the control group or detected at concentrations that were at least 10-fold lower. Microbially derived PA metabolites detected in urine included valerolactones, phenylvaleric acids, phenylpropionic acids, phenylacetic acids, benzoic acid, cinnamic acids and lignans. 3,4-Dihydroxyphenylacetic acid, detected by MRM transition (167 → 123), has been reported to be as a specific metabolite of polymeric PA, since it has never been identified in studies performed with pure monomeric flavan-3-ols(Reference Das31). 3,4-Dihydroxyphenylacetic acid may be metabolised to 3-hydroxyphenylacetic acid and 4-hydroxyphenylacetic acid (151 → 107), which were also detected. Similarly, 3,4-dihydroxyphenylpropionic acid, 3-hydroxyphenylpropionic acid and 4-hydroxyphenylpropionic acid were detected in urine. Both 3,4-dihydroxyphenylacetic acid and 3,4-dihydroxyphenylpropionic acid were absorbed and later conjugated in the liver as shown by the detection of the derivatives GlcA-3 or 4-hydroxyphenylacetic acid (327 → 151), Sulf-3,4-dihydroxyphenylpropionic acid (261 → 181) and Sulf- 3 or 4-hydroxyphenylpropionic acid (245 → 165). Conjugated forms of phenylvaleric and hippuric acid were detected in urine.
Table 2 Detection of microbially derived proanthocyanidin metabolites in urine and faeces from rats fed a non-extractable proanthocyanidin-rich fraction from grape antioxidant dietary fibre
MRM, multiple reaction monitoring; Sulf, sulphate; GlcA, glucuronide; Me, methyl group.
Two microbially derived phenolic metabolites were identified in the faeces from rats fed the NEPA-rich fraction: 4-hydroxyphenylpropionic acid (165 → 121) and 3,4-dihydroxyphenylpropionic acid (181 → 137).
Phenolic acids generated fragments corresponding to the successive loss of two CO2 molecules. Similarly, MS/MS spectra of sulphated forms showed signals corresponding to the loss of sulphate and CO2. These fragments confirm the identity of some of the microbially derived metabolites; others were confirmed by the use of standards, i.e. 4-hydroxybenzoic acid (Fig. 2).
Fig. 2 HPLC–electrospray ionisation-MS profile corresponding to the detection by multiple reaction monitoring of 4-hydroxybenzoic acid (transition 137 → 93): (a) urine from rats fed non-extractable proanthocyanidins from grape antioxidant dietary fibre, (b) 4-hydroxybenzoic acid standard, (c) urine from rats fed water.
Discussion
Several studies have addressed the metabolism of dietary oligomeric PA, mostly dimers. That work provides quite a clear picture of the different steps in the metabolism of PA dimers in laboratory animals and human subjects(Reference Baba, Osakabe and Natsume18–Reference Urpí-Sardá, Garrido and Monagas21). This process comprises the absorption of monomers and small oligomers (dimers) of PA in the small intestine and the absorption of microbially derived metabolites in the large intestine, after direct fermentation of the oligomers by microbiota without prior depolymerisation into EC. The absorbed metabolites may be conjugated in the liver, mostly resulting in GlcA, sulphates and methyl derivatives which pass to the bloodstream and eventually reach other tissues. Finally, the metabolites are excreted in urine and the fraction of PA that is not absorbed is excreted in faeces.
The metabolic fate of larger PA polymers is believed to follow the same pattern as that of dimers and trimers: essentially, direct cleavage of the EC units into smaller phenolic acids by the intestinal microbiota. By examining the metabolic fate of GADF, we have recently suggested that polymeric PA undergo depolymerisation into EC units during their transit along the intestinal tract(Reference Touriño, Fuguet and Vinardell24, Reference Touriño, Pérez-Jiménez and Mateos-Martín25). This is important because it implies that the polymers may gradually release EC moieties during the postprandial period. Using a NEPA-rich fraction, devoid of EC monomers and extractable oligomers, we show here that this is indeed occurring. The faeces of rats fed with NEPA contained monomeric and dimeric EC and their urine contained 10 phase II EC metabolites. These results clearly show that this fraction of dietary fibre generates bioavailable derivatives of EC. The di-glucuronidated EC derivative detected in faeces further demonstrates that those monomers that are released from NEPA efficiently enough to reach the liver suffer conjugation and are transported back to the intestine via bile. Our evidence for intestinal depolymerisation is consistent with a recent observation by Jové et al. (Reference Jové, Serrano and Ortega32) who report a 600 % recovery of free EC in the caecal content of rats after providing them with a single dose of PA-rich almond extract. Moreover, the more than twenty EC-derived smaller metabolites detected in urine are consistent with the previous description of microbial fermentation and absorption. Our results also corroborate NEPA as PA, since direct evidence of the structure of these insoluble polymers is scarce. In fact, the residue after extraction with 70 % acetone is commonly not considered to be a source of polyphenols. Our results agree with those which report that the residues of the common extraction with 70 % acetone contain significant amounts of PA(Reference Hellström and Mattila12–Reference Pérez-Jiménez, Arranz and Saura-Calixto14).
The transformation of NEPA (Fig. 3) differs in part from the process suggested for the transformation of small EPA, previously described. In the case of NEPA, a proportion of the larger PA polymers appears to be depolymerised during their transit along the intestinal tract, resulting in delivery of EC monomers and possibly oligomers. The posterior degradation by the intestinal microbiota into small units may also differ between NEPA and EPA. As NEPA are associated with the food matrix in foodstuffs, particularly with other insoluble polymers constitutive of dietary fibre, their conversion may be slow compared to that of EPA. This deferred release would make NEPA metabolites bioavailable for particularly long times after intake and may result in them having health effects for a long time. This may explain the previously reported delay in the increase of plasma antioxidant capacity after the intake of GADF by human subjects compared to that observed after the intake of food rich in EPA, such as red wine(Reference Pérez-Jiménez, Serrano and Tabernero33). In addition, the metabolites detected in faeces prove that putatively active species remain in contact with the colonic epithelium for at least 24 h after ingestion. Indeed, the intake of PA, and particularly PA with a high degree of polymerisation, has been associated with a reduced risk of colorectal cancer(Reference Rossi, Bosetti and Negri7) and our results suggest that food sources of NEPA could provide such putative cancer-preventative PA.
Fig. 3 Metabolism of non-extractable proanthocyanidins (NEPA).
Previous nutritional studies have considered the extractable fraction of PA as the only source of dietary polyphenols. We show here that NEPA should be taken into account as most of the food in these studies contains significant amounts of NEPA. Further work is needed, both on the systematic analysis of NEPA in foodstuffs and on the metabolism of NEPA from different food sources, in order to unravel the contribution of this fraction of dietary PA to the health-promoting effects of fruit and vegetables.
In conclusion, we show here that NEPA are a source of polymeric PA that are progressively depolymerised during their transit along the intestinal tract into EC monomers and dimers, and later metabolised by the intestinal microbiota into smaller units. As a result, EC, phenolic acids and their phase II metabolites are in contact with the intestinal tract and bioavailable for at least 24 h after ingestion.
Acknowledgements
The present work was supported by the Spanish Ministry of Education and Science (AGL2009-12374-C03-03/ALI). J. P.-J. thanks the Spanish Ministry of Science and Innovation for granting her a Sara Borrell postdoctoral contract (CD09/00068). GADF was a generous gift from Professor Fulgencio Saura-Calixto, ICTAN-CSIC. None of the authors had any conflict of interest. J. L. T., J. P.-J. and M. L. M.-M. designed the research. M. L. M.-M. and J. P.-J. carried out the experimental work. M. L. M.-M., J. P.-J. and E. F. analysed the data. M. L. M.-M. and J. P.-J. wrote the first version of the manuscript. All the authors contributed to writing the manuscript and approved the final version. Language revision by Christopher Evans is appreciated.
