1122
DOI 10.1002/mnfr.201300441
Mol. Nutr. Food Res. 2014, 58, 1122–1131
RESEARCH ARTICLE
In vitro transformation of chlorogenic acid by human gut microbiota Francisco Tomas-Barberan1 , Roc´ıo Garc´ıa-Villalba1 , Andrea Quartieri2 , Stefano Raimondi2 , Alberto Amaretti2 , Alan Leonardi2 and Maddalena Rossi2 1 2
Department of Food Science and Technology, CEBAS-CSIC, Murcia, Spain Department of Life Sciences, University of Modena and Reggio Emilia, Modena, Italy
Scope: Chlorogenic acid (3-O-caffeoyl-quinic acid, C-QA), the caffeic ester of quinic acid, is one of the most abundant phenolic acids in Western diet. The majority of C-QA escapes absorption in the small intestine and reaches the colon, where the resident microbiota transforms it into several metabolites. C-QA conversion by the gut microbiota from nine subjects was compared to evaluate the variability of bacterial metabolism. It was investigated whether a potentially probiotic Bifidobacterium strain, capable of C-QA hydrolysis, could affect C-QA fate. Methods and results: Bioconversion experiments exploiting the microbiota from diverse subjects revealed that C-QA was metabolized through a succession of hydrogenation, dexydroxylation and ester hydrolysis, occurring in different order among the subjects. Transformation may proceed also through quinic acid residue breakdown, since caffeoyl-glycerol intermediates were identified (HPLC-MS/MS, Q-TOF). All the pathways converged on 3-(3-hydroxyphenyl)propanoic acid, which was transformed to hydroxyphenyl-ethanol and/or phenylacetic acid in few subjects. A strain of Bifidobacterium animalis able to hydrolyze C-QA was added to microbiota cultures. It affected microbial composition but not to such an extent that C-QA metabolism was modified. Conclusion: A picture of the variability of microbiota C-QA transformations among subjects is provided. The transformation route through caffeoyl-glycerol intermediates is described for the first time.
Received: June 18, 2013 Revised: October 10, 2013 Accepted: October 12, 2013
Keywords: Bifidobacterium / Bioconversion / Chlorogenic acid / Intestinal microbiota / Probiotic
1
Introduction
Chlorogenic acid (3-O-caffeoyl-quinic acid, C-QA) and its positional isomers 4-O-caffeoyl-quinic acid and 5-O-caffeoylquinic acid are hydroxycinnamic esters of D-quinic acid (QA). Correspondence: Professor Maddalena Rossi, Department of Life Sciences, University of Modena and Reggio Emilia, via Campi 183, 41125 Modena, Italy E-mail: [email protected] Abbreviations: CA, caffeic acid; C-G, caffeoyl-glycerol; CQA, 3-O-caffeoyl-quinic acid (chlorogenic acid); DHPPA, 3-(3,4dihydroxyphenyl)-propanoic acid (dihydrocaffeic acid); DHPPG, 3-(3,4-dihydroxyphenyl)-propanoyl-glycerol (dihydrocaffeoylglycerol); DHPP-QA, 3-O-(3-(3,4-dihydroxyphenyl)-propanoyl)quinic acid (dihydrochlorogenic acid); FISH, fluorescence in situ hybridization; HCi-QA, 3-O-(3-hydroxycinnamoyl)-quinic acid (dehydroxy-chlorogenic acid); HPE, 2-(3-hydroxyphenyl)-ethanol; HPPA, 3-(3-hydroxyphenyl)-propanoic acid; PAA, phenyl-acetic acid; QA, quinic acid; Q-TOF, quadrupole TOF; UV-Vis, ultravioletvisible C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
They occur in many different plant-based food products and are particularly abundant in coffee, where they account for the main fraction of phenolic acids. Coffee is their major source in Western diet, so that habitual coffee drinkers can ingest up to 1 g/d of C-QA [1–3]. The importance of these compounds relies on several health effects, due to their antioxidant properties [4–7]. Free phenolic acids, such as caffeic acid (CA), are almost completely absorbed in the small intestine, while only approximately the 30% of C-QA is assimilated in this tract, as demonstrated by measuring its recovery in ileostomy effluents [8–14]. Only a minor extent of absorbed C-QA remains unaltered, since human enzymes perform methylation, sulphation, and glucuronidation [15–17]. Conjugated metabolites are present in plasma and are excreted in urine, but are also found in the ileal fluid [12, 18, 19]. The major part of C-QA escapes absorption in the small intestine and reaches the colon, this site playing a pivotal role in C-QA metabolism and absorption. C-QA is hydrolyzed and transformed by the resident microbiota within the www.mnf-journal.com
1123
Mol. Nutr. Food Res. 2014, 58, 1122–1131
colonic lumen, then the microbial metabolites may be absorbed and modified by human enzymes, originating a greater variety of metabolites occurring in plasma and urine of human and animals [19–21]. Thus, the fate of C-QA and the bioavailability of its metabolites depend largely on microbiota composition and activity. Microbial metabolism includes C-QA hydrolysis to QA and CA and it is responsible for their subsequent transformation into several metabolites, including m-coumaric acid and hydroxylated derivatives of phenylpropionic, benzoic, and hippuric acids [22–24]. Hippuric acid largely originates from the transformation of the QA moiety, due to bacterial aromatization to benzoic acid followed by glycine conjugation in liver and kidney, while all the other metabolites originate from bacterial reductive transformations of the CA moiety [22–25]. Although in vivo and in vitro studies provided scientific evidence for the occurrence of bacterial transformation of C-QA, the interindividual differences of gut microbiota and their potential impact of phytochemical metabolism have not been described yet. It is expected that the bioavailability of C-QA metabolites largely depends on microbiota composition and activity. The present study investigated the bacterial metabolism of C-QA by the gut microbiota of nine volunteers, in order to provide a picture of the variability of bacterial transformations among subjects. A continuous culture fermentation simulating the gut microbiota was performed to examine the time course of bacterial transformation (i.e. ester hydrolysis, hydrogenation reactions, and further modifications).
2
Materials and methods
2.1 Chemicals, bacterial strains, fecal samples, and culture conditions Methanol, formic acid, and acetonitrile of LC-MS grade were supplied from Panreac (Barcelona, Spain). Water was deionized by using a Milli-Q-system (Millipore, Bedford, MA, USA). C-QA, CA, 3-(3,4-dihydroxyphenyl)-propanoic acid (DHPPA, dihydrocaffeic acid), and 3-(3-hydroxyphenyl)propanoic acid (HPPA) were purchased from Sigma Aldrich (St. Louis, MO, USA). For the preparation of the calibration curve, stock solutions at concentration of 2000 mg/L for each compound were prepared in methanol. Besides, a 75 mM stock solution of C-QA in DMSO was prepared for the incubation experiments. Fecal inocula for resting cells bioconversions and continuous cultures were prepared from the fresh feces of nine healthy volunteers (six men and three women) who had not been treated with prebiotics and/or probiotics for 1 month, and antibiotics for at least 3 months. Fecal samples were suspended 10% w/v in PBS-Cys (PBS buffer containing 0.5 g/L L-cysteine · HCl, pH adjusted to 6.5). To perform experiments with reproducible inocula, fecal slurries were supplemented with 100 g/L glycerol in order to minimize loss of cell vitality, and stored at −20⬚C until use. Preparation was performed C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
in anaerobic cabinet (Anaerobic System, Forma Scientific, Marietta, OH, USA) under a 85% N2 , 10% CO2 , 5% H2 atmosphere. B. animalis subsp. lactis WC0432 was obtained from our own collection (Department of Life Sciences, University of Modena and Reggio Emilia). It was cultured anaerobically at 37⬚C in Lactobacilli MRS broth (BD Difco, Sparks, USA) containing 0.5 g/L L-cysteine · HCl. To prepare the supplement for continuous fecal fermentations, the strain was cultured for 16 h in 200 mL of MRS-Cys. Then, biomass was harvested (centrifugation at 6000 × g, 10 min, 4⬚C), washed, tenfold concentrated with sterile MRS-Cys supplemented with 100 g/L glycerol, and stored at −20⬚C. Two days before inoculating the bioreactor, aliquots were thawed to evaluate viable cells count on MRS-Cys agar plates.
2.2 Bioconversion with resting cells and cultures of gut microbiota For bioconversion experiments with resting cells, 1 mL of fecal suspension was anaerobically diluted with 4 mL of PBSCys, and supplemented with 500 M C-QA. The sample was incubated at 37⬚C in anaerobic conditions for 24 h and analyzed for bioconversion yield and products. Blank controls containing the same nine fecal samples incubated for 24 h in PBS buffer were prepared. As control, 500 M C-QA was incubated in PBS buffer for 24 h. The microbiota from one volunteer was used to perform bioconversion experiments with growing cultures. Singlestage continuous fermentations were carried out in 500 mL bioreactors (Sixfors V3.01, Infors, Bottmingen, Swiss) containing 250 mL of medium. The medium, based on that developed by Macfarlane [26] and modified by Walker [27], was fed with a flow rate equating to one turnover per day, yielding a dilution rate of 0.042/h. At the beginning of the process, the bioreactor was inoculated with 2.5 mL of 10% w/v fecal suspension and was added with 1.5 mM C-QA. Two of the four vessels were inoculated also with B. animalis subsp. lactis WC 0432, in order to reach a final concentration of 6.0 × 107 cells/mL. The culture was kept at 37⬚C under gentle agitation; anaerobic conditions were maintained by keeping the medium under a stream of CO2 ; a pH controller delivered 4 M NaOH to maintain the pH at 6.5. Samples were collected periodically and processed differently depending on the intended use: analysis of bioconversion and fermentation products, examination of microbiota composition, and bifidobacteria enumeration and isolation.
2.3 Analysis of C-QA and its metabolites Each sample of 1.5 mL collected after incubation was lyophilized before storing at −20⬚C. Prior to analysis, samples were dissolved in 1 mL of MeOH/H2 O (50/50, v/v), vortexed, centrifuged (13 000 × g, 10 min, 4⬚C), and filtered through www.mnf-journal.com
1124
F. Tomas-Barberan et al.
a 0.22 m polyvinylidene fluoride filter (Millipore, Billerica, MA, USA). The samples were analyzed using an Agilent 1100 HPLC system equipped with vacuum degasser (G1322), autosampler (G1313A), binary pump (G1322A), and ultraviolet-visible DAD (UV-Vis DAD) (Agilent Technologies, Waldbronn, Germany). The HPLC system was coupled in series to a model Esquire 6000 IT mass spectrometer, equipped with an electrospray interface (ESI) (Agilent Technologies). Separation was carried out on a RP LiChroCART C-18 column (Merck, Darmstadt, Germany) (250 × 4 mm, 4.5 m particle size), operating at room temperature and a flow rate of 1 mL/min. The mobile phase was composed of 1% (v/v in water) formic acid (phase A) and acetonitrile (phase B), mixed with a gradient program that allowed 5% phase B to reach 30% at 30 min, 65% at 40 min, 95% at 45 min, 5% at 48 min, and 5% till 55 min. A volume of 10 L of sample was injected. The separated compounds were monitored in sequence first with DAD (280, 260, and 320 nm) and then with a MS detector. Nitrogen was used as drying gas and nebulizing gas in the MS detector and according to the flow (1 mL/min), the ESI parameters were chosen: nebulizer pressure was set at 65 psi, dry gas flow 11 L/min, and dry gas temperature 350⬚C. MS data were acquired in the negative ionization mode. The capillary voltage was set at 4 kV. Mass scan and daughter MS/MS spectra were measured in the m/z range of 100–800 and target mass was 300. The ion target of the IT was adjusted at 20 000 with a maximum accumulation time of 200 ms. Three mass scan repetitions were performed to obtain average MS spectra. Compound stability was set at 75%. Compounds were identified with their UV-Vis spectra, molecular weight, and, whenever possible, chromatographic comparison with authentic standards. Besides, to confirm the identification of the metabolites, samples were analyzed by Agilent 1290 Infinity UPLC system coupled to a quadrupole TOF (Q-TOF) mass spectrometer (6550 Accurate-Mass QTOF, Agilent Technologies) using an electrospray interface with jet stream technology . The chromatographic separation was developed under the same conditions described in HPLCUV-Vis-IT analyses. The optimal conditions of the electrospray interface were as follows: gas temperature 300⬚C, drying gas 11 L/min, nebulizer 65 psi, sheath gas temperature 400⬚C, sheath gas flow 12 L/min. Spectra were acquired in the m/z range of 100–1100, in a negative mode, and with an acquisition rate of 1.5 spectra/s in MS and 2 spectra/s in MS/MS, maintaining a mass resolution over 50 000 for the mass range used. Internal mass calibration by simultaneous acquisition of reference ions and mass drift compensation was used for obtaining low mass errors. Q-TOF MS provides valuable information about the elemental composition of compounds based on the accurate mass and the isotopic pattern. Data were processed using the Mass Hunter Qualitative Analysis software (version B.06.00). Once identified, C-QA and its metabolites were quantified in UV using external calibration curves with appropriate C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Mol. Nutr. Food Res. 2014, 58, 1122–1131
standards. The calibration curves of C-QA, CA, DHPPA, and HPPA were prepared in methanol and were linear from LOQ to 3000 M with r2 = 0.9977 – 0.9994. The LOD and LOQ were obtained by injecting successively more diluted standard solutions and were calculated based on an S/N of 3 for the LOD and of 10 for the LOQ. The results showed LODs of 0.30, 0.35, 1.31, and 2.16 M and LOQs of 1.0, 1.2, 4.3, and 7.2 M for C-QA, CA, DHPPA, and HPPA, respectively. The method repeatability was evaluated by injecting a 100 M solution of a mixture of standards three times in the same day (intraday repeatability) and in three different days (interday repeatability). The results expressed as the RSD in all cases were within the acceptable limit (≤5%). C-QA and CA were quantified at 320 nm with their own standards. DHPPA and HPPA were quantified at 280 nm with their own standards. Dehydroxy-chlorogenic acid (HCiQA) was quantified at 320 nm with the calibration curve of C-QA; dihydrochlorogenic acid (DHPP-QA) was quantified at 280 nm with the calibration curve of DHPPA; 2-(3hydroxyphenyl)-ethanol (HPE) was quantified at 280 nm with the calibration curve of HPPA. DHPP-G and phenyl-acetic acid (PAA) could not be accurately quantified since their signals overlapped with other compounds.
2.4 Analysis of fermentation products Culture samples were centrifuged (13 000 × g, 5 min, 4⬚C) to remove solid particles, then supernatants were filtered (0.22 m cellulose acetate filter) and stored at −20⬚C until analyzed. Fermentation products (formic, acetic, lactic, propionic, butyric, succinic acids, and ethanol) were analyzed using HPLC device (Agilent technologies) equipped with refractive index detector and Aminex HPX-87 h ion exclusion column. Isocratic elution was carried out at with 0.005 M H2 SO4 at 0.6 mL/min.
2.5 Fluorescence in situ hybridization (FISH) Eubacteria, Bacteroidetes, bifidobacteria, Enterobacteriaceae, cluster XIVa Clostridiales, and Faecalibacterium prausnitzii were enumerated using the probes Eub338, Bac303, Bif164, Enterobac D, Erec482, and Fpra645 [27] in a FISH protocol [28]. Culture samples were diluted 1:4 with 40 g/L paraformaldehyde and incubated overnight at 4⬚C for cells fixation. Fixed cells were washed with PBS and dehydrated with PBS-ethanol 1:1 solution for 1 h at 4⬚C. To perform hybridization, 10 L of cell suspension, 1 L of the specific FITC-labeled probe, and 100 L of hybridization buffer (20 mM TRIS-HCl, 0.9 M NaCl, 0.1% SDS) were mixed and incubated at the temperature specific for each probe for 16 h [27]. A proper amount of the cell suspension was diluted in 4 mL of washing buffer (20 mM TRIS-HCl, 0.9 M NaCl), and maintained at room temperature for 10 min before being filtered onto 0.2 m polycarbonate filters (Millipore, www.mnf-journal.com
1125
Mol. Nutr. Food Res. 2014, 58, 1122–1131
Figure 1. Human microbiota metabolism of C-QA. 1, C-QA; 2, CA; 3, DHPPA; 4, HPPA; 5, HCi-QA; 6, DHPP-QA; 7, C-G; 8, DHPP-G; 9, HPE; 10, PAA. DeOH, dehydroxylation; Est, ester hydrolysis; Hyd, hydrogenation.
Ettenleur, Netherlands) [29]. Filters were mounted on microscope slides with Vectashield (Vector Labs, Burlingame, CA, USA). The slides were evaluated with a fluorescence microscope (Eclipse 80i, Nikon Instruments) equipped with mercury arc lamp, FITC-specific filter, and digital camera. Depending on the number of fluorescent cells, 30–100 microscopic fields were counted and averaged in each slide. Each sample was enumerated in triplicate.
2.7 Statistical analysis All values are means of three separate experiments. Comparison was carried out according to Student’s t-test. Differences were considered statistically significant for p < 0.05.
3
Results
3.1 In vitro metabolism of C-QA by resting cells of human fecal microbiota 2.6 RAPD-PCR tracing of B. animalis subsp. lactis WC 0432
3.1.1 Identification of phenolic acid metabolites
Culture samples were serially diluted in Wilkins-Chalgren anaerobe broth (Oxoid) in the anaerobic cabinet, then plated on RB medium to count and isolate bifidobacteria [30]. The attribution of colonies to the genus Bifidobacterium was confirmed using specific PCR reaction [31]. A total of 200 bifidobacterial isolates from each sample were subjected to extraction of genomic DNA, using Instagene matrix (Bio-Rad, La Jolla, CA, USA). RAPD-PCR was carried out in a 15 L reaction: 10X Dream Taq Buffer (including MgCl2 2mM), 1.5 L; dNTPs mixture 0.10 mM, 0,15 L; 2 M M13-RAPD primer (5 -CCGCAGCCAA-3 ), 3.75 L; genomic DNA, 3 L; PCR water, 5.25 L. DNA amplification was performed with the following temperature profile: 94⬚C for 4 min (1 cycle), 94⬚C for 1 min, 34⬚C for 1 min, 72⬚C for 2 min (45 cycles); 72⬚C for 7 min (1 cycle). The PCR products were electrophoresed in a 2% agarose gel (25 × 25 cm) for 4 h at a constant voltage in TAE buffer (40 mM Tris-acetate, 1 mM EDTA, pH 8.0). The RAPD-PCR profiles were visualized under UV light after staining with ethidium bromide, followed by digital image capturing. The resulting fingerprints were analyzed by the Gene Directory 2.0 (Syngene, UK) software package. The similarity among digitized profiles was calculated and a dendrogram was derived with an unweighted pair-group method using arithmetic means.
The substrate and nine metabolites were observed after the incubation of C-QA with the resting cells of human gut microbiota from nine different volunteers (Fig. 1). None of these compounds were detected in the blank samples prepared incubating fecal samples in PBS buffer or C-QA in PBS buffer. The metabolites were separated with HPLC and were identified using the UV spectra and the information provided by IT MS/MS and Q-TOF (Fig. 2; Table 1). Q-TOF analysis allowed the determination of the exact mass and molecular formula for the different metabolites (Table 1). Compounds 1, 2, 3, 4, and 10 were identified as C-QA, CA, DHPPA, HPPA, and PAA as confirmed by comparisons with authentic markers. Different isomers of C-QA appeared in the samples with the following elution under the analytical conditions used: 3-caffeoyl-quinic acid (neochlorogenic acid) at 9.28 min, the most abundant one was identified as 5-caffeoyl-quinic acid (chlorogenic acid) with the same retention time of the standard (12.20 min) and 4-caffeoyl quinic acid (kryptochlorogenic acid) at 13.48 min. These isomers were also detected in the blank sample when C-QA was incubated without fecal samples, indicating a possible interconversion of this compound in its isomers during the incubation at 37⬚C. In the other cases, the mass spectra and the observed losses allowed the identification of the intermediates.
C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.mnf-journal.com
1126
F. Tomas-Barberan et al.
Mol. Nutr. Food Res. 2014, 58, 1122–1131
Compound 5 was identified as HCi-QA, since it showed a UV spectrum similar to that of C-QA, consistent with a double bond in the phenylpropanoyl residue, and a mass spectrum 16 units lower than C-QA, consistent with the loss of a hydroxyl group. The position of dehydroxylation was determined according to the other metabolites identified, indicating that the hydroxyl at 4-position of the phenyl propanoic residue was removed first. Compound 6 was identified as DHPP-QA, since it showed a m/z value of 355.1037, two mass units higher than C-QA, and its UV spectrum presented a maximum at 282 nm, clearly different from that of C-QA (maximum at 326 nm), suggesting that the double bond was lost by hydrogenation, in agreement with the MS fragments that confirmed this structure. In addition, compounds 7 and 8 with m/z values of 253.0715 and 255.0872, respectively, were also detected. Compound 7 showed a UV spectrum as a phenylpropenoid very similar to CA, while the UV spectrum of compound 8 was similar to that of DHPPA (3). The mass and molecular formula of 7, compared with that of C-QA, indicated the loss of a fragment consistent with the removal of the carboxyl and the hydroxyl present in carbon 1 of the QA residue, leaving a glycerol residue linked to the CA moiety. Besides, the fragmentation pattern of this compound showed a fragment at m/z 179, after the loss of m/z 74, which represents a glycerol unit, and other fragments at m/z 161 and 135 typical of CA. As a whole, compound 7 was identified as caffeoyl-glycerol (C-G). The mass of compound 8 was two mass units higher than compound 7, confirming the hydrogenation of the double bond, and showed fragments at m/z 181 (loss of m/z 74), 163, and 137, typical of DHPPA. This compound was identified as dihydrocaffeoyl glycerol (DHPPG). In addition, to confirm that both compounds were not produced chemically, CA and DHPPA were incubated with the medium (PBS-Cys) in the presence of glycerol. None of the glycerol derivatives were observed after incubation for 24 h at 37⬚C without gut microbiota, suggesting that they were microbial degradation products of C-QA.
Figure 2. Representative HPLC chromatograms of the metabolites produced by the microbiota. UV detection at different wavelengths (280 nm and 320 nm) highlights the most representative metabolites from volunteers 3 (A), 5 (B), and 9 (C). Compounds are numbered as in Fig. 1 and Table 1. Compound 10 (PAA) would be better observed at 260 nm (it seems to overlap with an unidentified compound).
Table 1. Identification of the metabolites originated from C-QA bioconversion with the resting cells of the microbiota from different volunteers
Compound identification 1 2 3 4 5 6 7 8 9 10
C-QAa) CA DHPPA HPPA HCi-QAa) DHPP-QAa) C-G DHPP-G HPE PAA
RT min
M/Z experimental
Score
Error ppm
Molecular formula
MS/MS fragments
max nm
9.28/12.20/13.48/15.45 13.95 12.48 19.56 13.46/14.26/15.26 8.57/11.23/11.93 14.70 12.52 27.95 25.74
353.0879 179.0347 181.0506 165.0557 337.0927 355.1037 253.0715 255.0872 137.0608 135.0456
94.04 99.58 99.55 99.91 98.54 98.59 99.25 98.85 99.79 98.57
−0.35 1.3 −0.35 0.66 0.94 −0.71 0.93 0.91 0.01 −3.37
C16 H18 O9 C9 H8 O4 C9 H10 O4 C9 H10 O3 C16 H18 O8 C16 H20 O9 C12 H14 O6 C12 H16 O6 C8 H10 O2 C8 H8 O2
191 161, 135 163,137,121 121 179, 135 191, 181, 137 179, 161, 135 181,163, 137 123, 105 –
326, 300 sh 324 282 274 325 282 325 280 282 260, 300 sh
a) Different isomers of these compounds appear in the chromatogram. C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.mnf-journal.com
1127
Mol. Nutr. Food Res. 2014, 58, 1122–1131
Table 2. Metabolites (M) resulting from the bioconversion of 500 M C-QA with the resting cells of different volunteers microbiota (V1–V9). Mean values of three replicates with RSD in all cases ≤ 6%
Compound
V1
V2
V3
V4
V5
V6
V7
V8
V9
1 2 3 4 5 6 7 9
0 0 364.26 23.19 0 0 0 0
0 15.35 138.17 31.23 1.40 0 1.24 0
0 0 190.45 36.13 0 78.75 0 0
0 4.35 325.76 38.69 0 0 0 0
2.22 3.21 301.21 14.59 2.76 0 2.60 0
1.52 0.67 305.24 40.06 0.34 0 1.31 0
0 0.57 198.49 99.29 0 0 0 0
0 1.07 321.50 38.52 0 0 0 0
0 8.31 141.62 14.59 2.58 0 1.22 169.63
Compound 8 (DHPP-G) was detected in all the samples and compound 10 (PAA) was detected in volunteers 2, 5, and 9 but it was not possible to quantify their content since their signals overlapped with other compounds.
Compound 9 was identified as HPE by its exact mass and formula and comparison with the other metabolites obtained. All the metabolites identified are shown in the proposed metabolism for C-QA in Fig. 1. A compound with m/z 191 that could account for QA released after the hydrolysis of chlorogenic acid was also detected in the MS chromatogram and identified with an authentic standard. QA, however, was not quantified as the analytical conditions used over this study were optimized for the determination of chlorogenic acid and the catabolites of phenolic nature and not for organic hydroxy acids such as QA, so its determination was not possible as QA eluted with the solvent front. 3.1.2 Quantitative estimates of C-QA and metabolites The metabolites produced by the microbiota from the different volunteers were quantified in the supernatant of resting cells incubated for 24 h (Table 2). In most of the samples, C-QA (1) disappeared and traces were found in two sole samples. Only DHPPA (3) and HPPA (4) were common metabolites produced by the microbiota of all the volunteers, the former being the major transformation product in most of the samples. HCi-QA (5) and C-G (7) were produced by the microbiota of four of the nine volunteers (V2, V5, V6, and V9), while PAA (10) was found only in three volunteers (V2, V5, and V9). DHPP-QA (6) and HPE (9) were detected only in volunteers in V3 and V9, respectively. In this last sample, HPE was found in a high amount, being one of the major metabolites. DHPP-G (8) was detected in all the samples by UPLC-Q-TOF but it was not possible to quantify its content since the signal at 280 nm overlapped with that of the DHPPA. 3.2 In vitro metabolism of C-QA by growing microbiota cultures 3.2.1 C-QA bioconversion Single-stage continuous fermentations of the colonic microbiota from subject V3 were carried out in order to perform C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
bioconversion experiments of C-QA with growing microbiota cultures. Cultures were inoculated with a bacterial concentration of approximately 1.3 × 10+9 cells/mL and were given 1.5 mM C-QA. To establish whether the fate of C-QA in microbiota cultures may be affected by a potentially probiotic strain which had been demonstrated to be efficient in hydrolyzing C-QA to CA [32], microbiota cultures were supplemented with 6.0 × 10+7 cells/mL of B. animalis subsp. lactis WC 0432 and were compared with the nonsupplemented ones (hereinafter named PMC and MC cultures, respectively). The time course of single representative PMC and MC processes are presented in Fig. 3. C-QA quickly disappeared regardless of the presence of B. animalis subsp. lactis WC 0432, being only the 0.2% and 1.5% after 8 h in PMC and MC cultures, respectively. C-QA hydrogenation proceeded faster than ester hydrolysis. The major transformation product was DHPP-QA, which peaked after 6 and 8 h in PMC and MC, respectively, while CA never accumulated above 1%. Later, DHPP-QA was hydrolyzed to DHPPA, which was further dehydroxylated to HPPA. In PMC and MC, the kinetics of these reactions were different. In MC, C-QA hydrogenation to DHPP-QA was delayed, C-QA being not transformed during the first 4 h (Fig. 3B). Conversely, in PMC, reduction of C-QA began earlier, and it was very efficient between 4 and 6 h (Fig. 3A). DHPPA, resulting from DHPP-QA hydrolysis, accumulated in greater amount in PMC cultures and was dehydroxylated more lately to HPPA, compared to MC cultures, providing a hint that dehydroxylation was the limiting step.
3.2.2 Evolution of fecal microbiota and fermentation products In continuous fermentations of the colonic microbiota (PMC and MC), bifidobacteria, Bacteroidetes, cluster XIVa Clostridiales, Enterobacteriaceae, F. prausnitzii, and total eubacteria were quantified every 4 h by FISH (Fig. 4A and B). Steadystate conditions were achieved in 16 h for total eubacteria and specific bacterial groups as well. The counts of total eubacteria increased up to 2.5 × 1010 and 3.1 × 10 10 cells/mL (in PMC and MC, respectively, p > 0.05) in the first 16 h. At the www.mnf-journal.com
1128
F. Tomas-Barberan et al.
Mol. Nutr. Food Res. 2014, 58, 1122–1131
significant difference in bifidobacterial counts at the steady state. At each time-point, bifidobacteria were isolated from PMC cultures, and then the isolates that were positive to Bifidobacterium-specific PCR were characterized by RAPDPCR fingerprinting in order to trace the B. animalis subsp. lactis WC0432. At the beginning of the process, B. animalis subsp. lactis WC0432 accounted for the 70% of bifidobacterial isolates, then decreased to the 6.7% after 4 h and stabilized at the 2.0% at the steady state, representing approximately the 0.2% of total eubacterial population. Formate, acetate, lactate, propionate, butyrate, succinate, and ethanol originated by microbiota metabolism during continuous cultivation and reached steady-state values after approximately 16 h (Fig. 4C and D). Formate and succinate accumulated at higher levels in MC than in PMC cultures during the exponential phase (6–10 h), then decreased to similar concentrations. At the steady state, the cultures differed in the levels of butyrate, which was higher in PMC than in MC (p < 0.05), and acetate, which was lower in PMC than in MC (p < 0.05). The other fermentation products exhibited similar steady-state concentrations in PMC and MC processes (p > 0.05). Lactate was found in low concentration, being always 0.05). Bacteroidetes grew with the highest growth rate, because they increased by approximately 2.7 magnitude orders, and became the dominant bacterial group in both PMC and MC (1.3 × 1010 and 2.0 × 1010 cells/mL, respectively, p > 0.05). At the steady state, Enterobacteriaceae were also similar in PMC and MC cultures (3.1 × 108 and 2.0 × 108 cells/mL, respectively, p > 0.05), even if these values evolved with different kinetics. Bifidobacteria increased by approximately 0.9 magnitude orders and were 1.9 × 108 and 3.5 × 108 cells/mL in PMC and MC cultures, respectively (p > 0.05). Compared to MC cultures, the supplementation with 6.0 × 107 cells/mL of B. animalis subsp. lactis WC0432 in PMC cultures did not determine a C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Discussion
Information from metagenome sequencing analysis provides evidence that considerable differences in gut microbiota composition occur among subjects, where this diversity is responsible of different metabolic activities [33, 34]. Consistently, studies dealing with bioavailability of phytochemical metabolites demonstrated that different transformations are largely attributable to the diverse bacterial populations [35,36]. C-QA, which is a major phenolic acid in the Western diet, is first hydrolyzed to CA, which may be subject of hydrogenases, and dehydroxylases, yielding derivatives of phenylpropanoic acid [22–24]. Information about interindividual diversity in C-QA metabolism is scarce. This study aimed to compare C-QA biotransformation by different human colonic microbiota, through resting cells experiments carried out with nine different colonic populations. In order to replicate experiments with reproducible inocula, bioconversion were performed with freeze-preserved microbiota, which have been demonstrated to yield the same bioconversion pattern of polyphenols, compared to fresh samples [37]. The diverse bacterial communities were able to hydrogenate, dehydroxylate C-QA, or break the QA moiety also before hydrolyzing its ester bond. However, all the transformation pathways converged on HPPA, the final metabolite in most of the samples. Conversion of C-QA into HPPA requires the removal or the breakdown of the QA moiety, the reduction of the propenoic double bond, and the dehydroxylation of the aromatic ring in position 4. Based on the intermediates that have been identified by HPLC-UV/Vis-MS/MS www.mnf-journal.com
Mol. Nutr. Food Res. 2014, 58, 1122–1131
1129
Figure 4. Time course of bacterial composition and fermentation products in continuous cultures of intestinal microbiota in the presence (A, C) or the absence (B, D) of B. animalis subsp. lactis WC 0432.Bacteria symbols (panels A and B): eubacteria, 䊉; Bacteroidetes, 䊊; bifidobacteria, ; Enterobacteriaceae, 䉭; cluster XIVa Clostridiales, ; Faecalibacterium prausnitzii, . Fermentation products symbols (panels C and D): lactate, 䊉; succinate, 䊊; formiate, ; acetate, 䉭; propionate, ; ethanol, ; butyrate, u. The results of a representative experiment performed in triplicate are shown.
and Q-TOF, these reactions can be performed in different sequences, resulting in diverse metabolic networks. The dehydroxylation seems to be the limiting step, because DHPPA accumulated in large amounts in most of the samples before being transformed into HPPA. Furthermore, HCi-QA was found, if any, only in low concentration, indicating that ester hydrolysis and double bond hydrogenation are more efficient and that C-QA is preferably transformed through the reactions yielding DHPPA. The hydrogenation and the ester hydrolysis and/or the breakdown of the QA moiety are necessary to obtain DHPPA. Based on the intermediates, the reactions occurred in diverse sequences and combinations in the different samples. Interestingly, the metabolism by gut microbiota of other polyphenols such as catechins, rutin, and naringin [38–40] leads to DHPPA, which was similarly subjected to paradehydroxylation into hydroxyphenyl-caroboxylic acids. Some of the compounds identified in this study, i.e. caffeoyl-glycerol metabolites C-G and DHPP-G, are described for the first time as microbiota metabolites of C-QA (Fig. 1). The presence of C-G and DHPP-G disclosed a novel degradation pathway of C-QA, where the cyclohexane ring was broken, leaving a glycerol residue esterified with CA. The C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
degradation of the QA residue before breaking the ester link, resulting in the caffeoyl-glycerol metabolites, is proposed here for the first time. HPPA was the bioconversion end product of most subjects, which did not proceed toward the formation of phenylpropanoic acid, unlike previous studies [23, 24]. Only three subjects (V2, V5, and V9) were capable to further transform HPPA into PAA, highlighting the presence of decarboxylating and dehydroxylating enzymatic activities. In one out of the three subjects (V9), HPE was also found in high amount, being one of the major metabolites of C-QA. It remains to be established whether HPE is an intermediate in HPPA conversion toward PAA, or whether the metabolic route diverges to yield two different products. Whenever the metabolic network did not stop and HPPA was subjected with great efficiency to further transformations, PAA and/or HPE accumulate in high amounts. Continuous cultures of the microbiota from V3 were carried out in presence of 1.5 mM C-QA. The choice of this microbial population was due to the fact that its resting cells did not perform the bioconversion beyond HPPA, and that it was the sole sample accumulating DHPP-QA. Then, we expected that the supplementation of the Bifidobacterium strain able to www.mnf-journal.com
1130
F. Tomas-Barberan et al.
hydrolyze the C-QA into CA could modify the flux toward this latter metabolite, highlighting the competition between the two different pathways. The substrate was rapidly consumed, while DHPP-QA, DHPPA, and HPPA were produced in succession. Neither the dehydroxylation nor the breakdown of QA residue took place as the first step and CA was detected only in minor amount (
DOI 10.1002/mnfr.201300441
Mol. Nutr. Food Res. 2014, 58, 1122–1131
RESEARCH ARTICLE
In vitro transformation of chlorogenic acid by human gut microbiota Francisco Tomas-Barberan1 , Roc´ıo Garc´ıa-Villalba1 , Andrea Quartieri2 , Stefano Raimondi2 , Alberto Amaretti2 , Alan Leonardi2 and Maddalena Rossi2 1 2
Department of Food Science and Technology, CEBAS-CSIC, Murcia, Spain Department of Life Sciences, University of Modena and Reggio Emilia, Modena, Italy
Scope: Chlorogenic acid (3-O-caffeoyl-quinic acid, C-QA), the caffeic ester of quinic acid, is one of the most abundant phenolic acids in Western diet. The majority of C-QA escapes absorption in the small intestine and reaches the colon, where the resident microbiota transforms it into several metabolites. C-QA conversion by the gut microbiota from nine subjects was compared to evaluate the variability of bacterial metabolism. It was investigated whether a potentially probiotic Bifidobacterium strain, capable of C-QA hydrolysis, could affect C-QA fate. Methods and results: Bioconversion experiments exploiting the microbiota from diverse subjects revealed that C-QA was metabolized through a succession of hydrogenation, dexydroxylation and ester hydrolysis, occurring in different order among the subjects. Transformation may proceed also through quinic acid residue breakdown, since caffeoyl-glycerol intermediates were identified (HPLC-MS/MS, Q-TOF). All the pathways converged on 3-(3-hydroxyphenyl)propanoic acid, which was transformed to hydroxyphenyl-ethanol and/or phenylacetic acid in few subjects. A strain of Bifidobacterium animalis able to hydrolyze C-QA was added to microbiota cultures. It affected microbial composition but not to such an extent that C-QA metabolism was modified. Conclusion: A picture of the variability of microbiota C-QA transformations among subjects is provided. The transformation route through caffeoyl-glycerol intermediates is described for the first time.
Received: June 18, 2013 Revised: October 10, 2013 Accepted: October 12, 2013
Keywords: Bifidobacterium / Bioconversion / Chlorogenic acid / Intestinal microbiota / Probiotic
1
Introduction
Chlorogenic acid (3-O-caffeoyl-quinic acid, C-QA) and its positional isomers 4-O-caffeoyl-quinic acid and 5-O-caffeoylquinic acid are hydroxycinnamic esters of D-quinic acid (QA). Correspondence: Professor Maddalena Rossi, Department of Life Sciences, University of Modena and Reggio Emilia, via Campi 183, 41125 Modena, Italy E-mail: [email protected] Abbreviations: CA, caffeic acid; C-G, caffeoyl-glycerol; CQA, 3-O-caffeoyl-quinic acid (chlorogenic acid); DHPPA, 3-(3,4dihydroxyphenyl)-propanoic acid (dihydrocaffeic acid); DHPPG, 3-(3,4-dihydroxyphenyl)-propanoyl-glycerol (dihydrocaffeoylglycerol); DHPP-QA, 3-O-(3-(3,4-dihydroxyphenyl)-propanoyl)quinic acid (dihydrochlorogenic acid); FISH, fluorescence in situ hybridization; HCi-QA, 3-O-(3-hydroxycinnamoyl)-quinic acid (dehydroxy-chlorogenic acid); HPE, 2-(3-hydroxyphenyl)-ethanol; HPPA, 3-(3-hydroxyphenyl)-propanoic acid; PAA, phenyl-acetic acid; QA, quinic acid; Q-TOF, quadrupole TOF; UV-Vis, ultravioletvisible C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
They occur in many different plant-based food products and are particularly abundant in coffee, where they account for the main fraction of phenolic acids. Coffee is their major source in Western diet, so that habitual coffee drinkers can ingest up to 1 g/d of C-QA [1–3]. The importance of these compounds relies on several health effects, due to their antioxidant properties [4–7]. Free phenolic acids, such as caffeic acid (CA), are almost completely absorbed in the small intestine, while only approximately the 30% of C-QA is assimilated in this tract, as demonstrated by measuring its recovery in ileostomy effluents [8–14]. Only a minor extent of absorbed C-QA remains unaltered, since human enzymes perform methylation, sulphation, and glucuronidation [15–17]. Conjugated metabolites are present in plasma and are excreted in urine, but are also found in the ileal fluid [12, 18, 19]. The major part of C-QA escapes absorption in the small intestine and reaches the colon, this site playing a pivotal role in C-QA metabolism and absorption. C-QA is hydrolyzed and transformed by the resident microbiota within the www.mnf-journal.com
1123
Mol. Nutr. Food Res. 2014, 58, 1122–1131
colonic lumen, then the microbial metabolites may be absorbed and modified by human enzymes, originating a greater variety of metabolites occurring in plasma and urine of human and animals [19–21]. Thus, the fate of C-QA and the bioavailability of its metabolites depend largely on microbiota composition and activity. Microbial metabolism includes C-QA hydrolysis to QA and CA and it is responsible for their subsequent transformation into several metabolites, including m-coumaric acid and hydroxylated derivatives of phenylpropionic, benzoic, and hippuric acids [22–24]. Hippuric acid largely originates from the transformation of the QA moiety, due to bacterial aromatization to benzoic acid followed by glycine conjugation in liver and kidney, while all the other metabolites originate from bacterial reductive transformations of the CA moiety [22–25]. Although in vivo and in vitro studies provided scientific evidence for the occurrence of bacterial transformation of C-QA, the interindividual differences of gut microbiota and their potential impact of phytochemical metabolism have not been described yet. It is expected that the bioavailability of C-QA metabolites largely depends on microbiota composition and activity. The present study investigated the bacterial metabolism of C-QA by the gut microbiota of nine volunteers, in order to provide a picture of the variability of bacterial transformations among subjects. A continuous culture fermentation simulating the gut microbiota was performed to examine the time course of bacterial transformation (i.e. ester hydrolysis, hydrogenation reactions, and further modifications).
2
Materials and methods
2.1 Chemicals, bacterial strains, fecal samples, and culture conditions Methanol, formic acid, and acetonitrile of LC-MS grade were supplied from Panreac (Barcelona, Spain). Water was deionized by using a Milli-Q-system (Millipore, Bedford, MA, USA). C-QA, CA, 3-(3,4-dihydroxyphenyl)-propanoic acid (DHPPA, dihydrocaffeic acid), and 3-(3-hydroxyphenyl)propanoic acid (HPPA) were purchased from Sigma Aldrich (St. Louis, MO, USA). For the preparation of the calibration curve, stock solutions at concentration of 2000 mg/L for each compound were prepared in methanol. Besides, a 75 mM stock solution of C-QA in DMSO was prepared for the incubation experiments. Fecal inocula for resting cells bioconversions and continuous cultures were prepared from the fresh feces of nine healthy volunteers (six men and three women) who had not been treated with prebiotics and/or probiotics for 1 month, and antibiotics for at least 3 months. Fecal samples were suspended 10% w/v in PBS-Cys (PBS buffer containing 0.5 g/L L-cysteine · HCl, pH adjusted to 6.5). To perform experiments with reproducible inocula, fecal slurries were supplemented with 100 g/L glycerol in order to minimize loss of cell vitality, and stored at −20⬚C until use. Preparation was performed C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
in anaerobic cabinet (Anaerobic System, Forma Scientific, Marietta, OH, USA) under a 85% N2 , 10% CO2 , 5% H2 atmosphere. B. animalis subsp. lactis WC0432 was obtained from our own collection (Department of Life Sciences, University of Modena and Reggio Emilia). It was cultured anaerobically at 37⬚C in Lactobacilli MRS broth (BD Difco, Sparks, USA) containing 0.5 g/L L-cysteine · HCl. To prepare the supplement for continuous fecal fermentations, the strain was cultured for 16 h in 200 mL of MRS-Cys. Then, biomass was harvested (centrifugation at 6000 × g, 10 min, 4⬚C), washed, tenfold concentrated with sterile MRS-Cys supplemented with 100 g/L glycerol, and stored at −20⬚C. Two days before inoculating the bioreactor, aliquots were thawed to evaluate viable cells count on MRS-Cys agar plates.
2.2 Bioconversion with resting cells and cultures of gut microbiota For bioconversion experiments with resting cells, 1 mL of fecal suspension was anaerobically diluted with 4 mL of PBSCys, and supplemented with 500 M C-QA. The sample was incubated at 37⬚C in anaerobic conditions for 24 h and analyzed for bioconversion yield and products. Blank controls containing the same nine fecal samples incubated for 24 h in PBS buffer were prepared. As control, 500 M C-QA was incubated in PBS buffer for 24 h. The microbiota from one volunteer was used to perform bioconversion experiments with growing cultures. Singlestage continuous fermentations were carried out in 500 mL bioreactors (Sixfors V3.01, Infors, Bottmingen, Swiss) containing 250 mL of medium. The medium, based on that developed by Macfarlane [26] and modified by Walker [27], was fed with a flow rate equating to one turnover per day, yielding a dilution rate of 0.042/h. At the beginning of the process, the bioreactor was inoculated with 2.5 mL of 10% w/v fecal suspension and was added with 1.5 mM C-QA. Two of the four vessels were inoculated also with B. animalis subsp. lactis WC 0432, in order to reach a final concentration of 6.0 × 107 cells/mL. The culture was kept at 37⬚C under gentle agitation; anaerobic conditions were maintained by keeping the medium under a stream of CO2 ; a pH controller delivered 4 M NaOH to maintain the pH at 6.5. Samples were collected periodically and processed differently depending on the intended use: analysis of bioconversion and fermentation products, examination of microbiota composition, and bifidobacteria enumeration and isolation.
2.3 Analysis of C-QA and its metabolites Each sample of 1.5 mL collected after incubation was lyophilized before storing at −20⬚C. Prior to analysis, samples were dissolved in 1 mL of MeOH/H2 O (50/50, v/v), vortexed, centrifuged (13 000 × g, 10 min, 4⬚C), and filtered through www.mnf-journal.com
1124
F. Tomas-Barberan et al.
a 0.22 m polyvinylidene fluoride filter (Millipore, Billerica, MA, USA). The samples were analyzed using an Agilent 1100 HPLC system equipped with vacuum degasser (G1322), autosampler (G1313A), binary pump (G1322A), and ultraviolet-visible DAD (UV-Vis DAD) (Agilent Technologies, Waldbronn, Germany). The HPLC system was coupled in series to a model Esquire 6000 IT mass spectrometer, equipped with an electrospray interface (ESI) (Agilent Technologies). Separation was carried out on a RP LiChroCART C-18 column (Merck, Darmstadt, Germany) (250 × 4 mm, 4.5 m particle size), operating at room temperature and a flow rate of 1 mL/min. The mobile phase was composed of 1% (v/v in water) formic acid (phase A) and acetonitrile (phase B), mixed with a gradient program that allowed 5% phase B to reach 30% at 30 min, 65% at 40 min, 95% at 45 min, 5% at 48 min, and 5% till 55 min. A volume of 10 L of sample was injected. The separated compounds were monitored in sequence first with DAD (280, 260, and 320 nm) and then with a MS detector. Nitrogen was used as drying gas and nebulizing gas in the MS detector and according to the flow (1 mL/min), the ESI parameters were chosen: nebulizer pressure was set at 65 psi, dry gas flow 11 L/min, and dry gas temperature 350⬚C. MS data were acquired in the negative ionization mode. The capillary voltage was set at 4 kV. Mass scan and daughter MS/MS spectra were measured in the m/z range of 100–800 and target mass was 300. The ion target of the IT was adjusted at 20 000 with a maximum accumulation time of 200 ms. Three mass scan repetitions were performed to obtain average MS spectra. Compound stability was set at 75%. Compounds were identified with their UV-Vis spectra, molecular weight, and, whenever possible, chromatographic comparison with authentic standards. Besides, to confirm the identification of the metabolites, samples were analyzed by Agilent 1290 Infinity UPLC system coupled to a quadrupole TOF (Q-TOF) mass spectrometer (6550 Accurate-Mass QTOF, Agilent Technologies) using an electrospray interface with jet stream technology . The chromatographic separation was developed under the same conditions described in HPLCUV-Vis-IT analyses. The optimal conditions of the electrospray interface were as follows: gas temperature 300⬚C, drying gas 11 L/min, nebulizer 65 psi, sheath gas temperature 400⬚C, sheath gas flow 12 L/min. Spectra were acquired in the m/z range of 100–1100, in a negative mode, and with an acquisition rate of 1.5 spectra/s in MS and 2 spectra/s in MS/MS, maintaining a mass resolution over 50 000 for the mass range used. Internal mass calibration by simultaneous acquisition of reference ions and mass drift compensation was used for obtaining low mass errors. Q-TOF MS provides valuable information about the elemental composition of compounds based on the accurate mass and the isotopic pattern. Data were processed using the Mass Hunter Qualitative Analysis software (version B.06.00). Once identified, C-QA and its metabolites were quantified in UV using external calibration curves with appropriate C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Mol. Nutr. Food Res. 2014, 58, 1122–1131
standards. The calibration curves of C-QA, CA, DHPPA, and HPPA were prepared in methanol and were linear from LOQ to 3000 M with r2 = 0.9977 – 0.9994. The LOD and LOQ were obtained by injecting successively more diluted standard solutions and were calculated based on an S/N of 3 for the LOD and of 10 for the LOQ. The results showed LODs of 0.30, 0.35, 1.31, and 2.16 M and LOQs of 1.0, 1.2, 4.3, and 7.2 M for C-QA, CA, DHPPA, and HPPA, respectively. The method repeatability was evaluated by injecting a 100 M solution of a mixture of standards three times in the same day (intraday repeatability) and in three different days (interday repeatability). The results expressed as the RSD in all cases were within the acceptable limit (≤5%). C-QA and CA were quantified at 320 nm with their own standards. DHPPA and HPPA were quantified at 280 nm with their own standards. Dehydroxy-chlorogenic acid (HCiQA) was quantified at 320 nm with the calibration curve of C-QA; dihydrochlorogenic acid (DHPP-QA) was quantified at 280 nm with the calibration curve of DHPPA; 2-(3hydroxyphenyl)-ethanol (HPE) was quantified at 280 nm with the calibration curve of HPPA. DHPP-G and phenyl-acetic acid (PAA) could not be accurately quantified since their signals overlapped with other compounds.
2.4 Analysis of fermentation products Culture samples were centrifuged (13 000 × g, 5 min, 4⬚C) to remove solid particles, then supernatants were filtered (0.22 m cellulose acetate filter) and stored at −20⬚C until analyzed. Fermentation products (formic, acetic, lactic, propionic, butyric, succinic acids, and ethanol) were analyzed using HPLC device (Agilent technologies) equipped with refractive index detector and Aminex HPX-87 h ion exclusion column. Isocratic elution was carried out at with 0.005 M H2 SO4 at 0.6 mL/min.
2.5 Fluorescence in situ hybridization (FISH) Eubacteria, Bacteroidetes, bifidobacteria, Enterobacteriaceae, cluster XIVa Clostridiales, and Faecalibacterium prausnitzii were enumerated using the probes Eub338, Bac303, Bif164, Enterobac D, Erec482, and Fpra645 [27] in a FISH protocol [28]. Culture samples were diluted 1:4 with 40 g/L paraformaldehyde and incubated overnight at 4⬚C for cells fixation. Fixed cells were washed with PBS and dehydrated with PBS-ethanol 1:1 solution for 1 h at 4⬚C. To perform hybridization, 10 L of cell suspension, 1 L of the specific FITC-labeled probe, and 100 L of hybridization buffer (20 mM TRIS-HCl, 0.9 M NaCl, 0.1% SDS) were mixed and incubated at the temperature specific for each probe for 16 h [27]. A proper amount of the cell suspension was diluted in 4 mL of washing buffer (20 mM TRIS-HCl, 0.9 M NaCl), and maintained at room temperature for 10 min before being filtered onto 0.2 m polycarbonate filters (Millipore, www.mnf-journal.com
1125
Mol. Nutr. Food Res. 2014, 58, 1122–1131
Figure 1. Human microbiota metabolism of C-QA. 1, C-QA; 2, CA; 3, DHPPA; 4, HPPA; 5, HCi-QA; 6, DHPP-QA; 7, C-G; 8, DHPP-G; 9, HPE; 10, PAA. DeOH, dehydroxylation; Est, ester hydrolysis; Hyd, hydrogenation.
Ettenleur, Netherlands) [29]. Filters were mounted on microscope slides with Vectashield (Vector Labs, Burlingame, CA, USA). The slides were evaluated with a fluorescence microscope (Eclipse 80i, Nikon Instruments) equipped with mercury arc lamp, FITC-specific filter, and digital camera. Depending on the number of fluorescent cells, 30–100 microscopic fields were counted and averaged in each slide. Each sample was enumerated in triplicate.
2.7 Statistical analysis All values are means of three separate experiments. Comparison was carried out according to Student’s t-test. Differences were considered statistically significant for p < 0.05.
3
Results
3.1 In vitro metabolism of C-QA by resting cells of human fecal microbiota 2.6 RAPD-PCR tracing of B. animalis subsp. lactis WC 0432
3.1.1 Identification of phenolic acid metabolites
Culture samples were serially diluted in Wilkins-Chalgren anaerobe broth (Oxoid) in the anaerobic cabinet, then plated on RB medium to count and isolate bifidobacteria [30]. The attribution of colonies to the genus Bifidobacterium was confirmed using specific PCR reaction [31]. A total of 200 bifidobacterial isolates from each sample were subjected to extraction of genomic DNA, using Instagene matrix (Bio-Rad, La Jolla, CA, USA). RAPD-PCR was carried out in a 15 L reaction: 10X Dream Taq Buffer (including MgCl2 2mM), 1.5 L; dNTPs mixture 0.10 mM, 0,15 L; 2 M M13-RAPD primer (5 -CCGCAGCCAA-3 ), 3.75 L; genomic DNA, 3 L; PCR water, 5.25 L. DNA amplification was performed with the following temperature profile: 94⬚C for 4 min (1 cycle), 94⬚C for 1 min, 34⬚C for 1 min, 72⬚C for 2 min (45 cycles); 72⬚C for 7 min (1 cycle). The PCR products were electrophoresed in a 2% agarose gel (25 × 25 cm) for 4 h at a constant voltage in TAE buffer (40 mM Tris-acetate, 1 mM EDTA, pH 8.0). The RAPD-PCR profiles were visualized under UV light after staining with ethidium bromide, followed by digital image capturing. The resulting fingerprints were analyzed by the Gene Directory 2.0 (Syngene, UK) software package. The similarity among digitized profiles was calculated and a dendrogram was derived with an unweighted pair-group method using arithmetic means.
The substrate and nine metabolites were observed after the incubation of C-QA with the resting cells of human gut microbiota from nine different volunteers (Fig. 1). None of these compounds were detected in the blank samples prepared incubating fecal samples in PBS buffer or C-QA in PBS buffer. The metabolites were separated with HPLC and were identified using the UV spectra and the information provided by IT MS/MS and Q-TOF (Fig. 2; Table 1). Q-TOF analysis allowed the determination of the exact mass and molecular formula for the different metabolites (Table 1). Compounds 1, 2, 3, 4, and 10 were identified as C-QA, CA, DHPPA, HPPA, and PAA as confirmed by comparisons with authentic markers. Different isomers of C-QA appeared in the samples with the following elution under the analytical conditions used: 3-caffeoyl-quinic acid (neochlorogenic acid) at 9.28 min, the most abundant one was identified as 5-caffeoyl-quinic acid (chlorogenic acid) with the same retention time of the standard (12.20 min) and 4-caffeoyl quinic acid (kryptochlorogenic acid) at 13.48 min. These isomers were also detected in the blank sample when C-QA was incubated without fecal samples, indicating a possible interconversion of this compound in its isomers during the incubation at 37⬚C. In the other cases, the mass spectra and the observed losses allowed the identification of the intermediates.
C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.mnf-journal.com
1126
F. Tomas-Barberan et al.
Mol. Nutr. Food Res. 2014, 58, 1122–1131
Compound 5 was identified as HCi-QA, since it showed a UV spectrum similar to that of C-QA, consistent with a double bond in the phenylpropanoyl residue, and a mass spectrum 16 units lower than C-QA, consistent with the loss of a hydroxyl group. The position of dehydroxylation was determined according to the other metabolites identified, indicating that the hydroxyl at 4-position of the phenyl propanoic residue was removed first. Compound 6 was identified as DHPP-QA, since it showed a m/z value of 355.1037, two mass units higher than C-QA, and its UV spectrum presented a maximum at 282 nm, clearly different from that of C-QA (maximum at 326 nm), suggesting that the double bond was lost by hydrogenation, in agreement with the MS fragments that confirmed this structure. In addition, compounds 7 and 8 with m/z values of 253.0715 and 255.0872, respectively, were also detected. Compound 7 showed a UV spectrum as a phenylpropenoid very similar to CA, while the UV spectrum of compound 8 was similar to that of DHPPA (3). The mass and molecular formula of 7, compared with that of C-QA, indicated the loss of a fragment consistent with the removal of the carboxyl and the hydroxyl present in carbon 1 of the QA residue, leaving a glycerol residue linked to the CA moiety. Besides, the fragmentation pattern of this compound showed a fragment at m/z 179, after the loss of m/z 74, which represents a glycerol unit, and other fragments at m/z 161 and 135 typical of CA. As a whole, compound 7 was identified as caffeoyl-glycerol (C-G). The mass of compound 8 was two mass units higher than compound 7, confirming the hydrogenation of the double bond, and showed fragments at m/z 181 (loss of m/z 74), 163, and 137, typical of DHPPA. This compound was identified as dihydrocaffeoyl glycerol (DHPPG). In addition, to confirm that both compounds were not produced chemically, CA and DHPPA were incubated with the medium (PBS-Cys) in the presence of glycerol. None of the glycerol derivatives were observed after incubation for 24 h at 37⬚C without gut microbiota, suggesting that they were microbial degradation products of C-QA.
Figure 2. Representative HPLC chromatograms of the metabolites produced by the microbiota. UV detection at different wavelengths (280 nm and 320 nm) highlights the most representative metabolites from volunteers 3 (A), 5 (B), and 9 (C). Compounds are numbered as in Fig. 1 and Table 1. Compound 10 (PAA) would be better observed at 260 nm (it seems to overlap with an unidentified compound).
Table 1. Identification of the metabolites originated from C-QA bioconversion with the resting cells of the microbiota from different volunteers
Compound identification 1 2 3 4 5 6 7 8 9 10
C-QAa) CA DHPPA HPPA HCi-QAa) DHPP-QAa) C-G DHPP-G HPE PAA
RT min
M/Z experimental
Score
Error ppm
Molecular formula
MS/MS fragments
max nm
9.28/12.20/13.48/15.45 13.95 12.48 19.56 13.46/14.26/15.26 8.57/11.23/11.93 14.70 12.52 27.95 25.74
353.0879 179.0347 181.0506 165.0557 337.0927 355.1037 253.0715 255.0872 137.0608 135.0456
94.04 99.58 99.55 99.91 98.54 98.59 99.25 98.85 99.79 98.57
−0.35 1.3 −0.35 0.66 0.94 −0.71 0.93 0.91 0.01 −3.37
C16 H18 O9 C9 H8 O4 C9 H10 O4 C9 H10 O3 C16 H18 O8 C16 H20 O9 C12 H14 O6 C12 H16 O6 C8 H10 O2 C8 H8 O2
191 161, 135 163,137,121 121 179, 135 191, 181, 137 179, 161, 135 181,163, 137 123, 105 –
326, 300 sh 324 282 274 325 282 325 280 282 260, 300 sh
a) Different isomers of these compounds appear in the chromatogram. C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.mnf-journal.com
1127
Mol. Nutr. Food Res. 2014, 58, 1122–1131
Table 2. Metabolites (M) resulting from the bioconversion of 500 M C-QA with the resting cells of different volunteers microbiota (V1–V9). Mean values of three replicates with RSD in all cases ≤ 6%
Compound
V1
V2
V3
V4
V5
V6
V7
V8
V9
1 2 3 4 5 6 7 9
0 0 364.26 23.19 0 0 0 0
0 15.35 138.17 31.23 1.40 0 1.24 0
0 0 190.45 36.13 0 78.75 0 0
0 4.35 325.76 38.69 0 0 0 0
2.22 3.21 301.21 14.59 2.76 0 2.60 0
1.52 0.67 305.24 40.06 0.34 0 1.31 0
0 0.57 198.49 99.29 0 0 0 0
0 1.07 321.50 38.52 0 0 0 0
0 8.31 141.62 14.59 2.58 0 1.22 169.63
Compound 8 (DHPP-G) was detected in all the samples and compound 10 (PAA) was detected in volunteers 2, 5, and 9 but it was not possible to quantify their content since their signals overlapped with other compounds.
Compound 9 was identified as HPE by its exact mass and formula and comparison with the other metabolites obtained. All the metabolites identified are shown in the proposed metabolism for C-QA in Fig. 1. A compound with m/z 191 that could account for QA released after the hydrolysis of chlorogenic acid was also detected in the MS chromatogram and identified with an authentic standard. QA, however, was not quantified as the analytical conditions used over this study were optimized for the determination of chlorogenic acid and the catabolites of phenolic nature and not for organic hydroxy acids such as QA, so its determination was not possible as QA eluted with the solvent front. 3.1.2 Quantitative estimates of C-QA and metabolites The metabolites produced by the microbiota from the different volunteers were quantified in the supernatant of resting cells incubated for 24 h (Table 2). In most of the samples, C-QA (1) disappeared and traces were found in two sole samples. Only DHPPA (3) and HPPA (4) were common metabolites produced by the microbiota of all the volunteers, the former being the major transformation product in most of the samples. HCi-QA (5) and C-G (7) were produced by the microbiota of four of the nine volunteers (V2, V5, V6, and V9), while PAA (10) was found only in three volunteers (V2, V5, and V9). DHPP-QA (6) and HPE (9) were detected only in volunteers in V3 and V9, respectively. In this last sample, HPE was found in a high amount, being one of the major metabolites. DHPP-G (8) was detected in all the samples by UPLC-Q-TOF but it was not possible to quantify its content since the signal at 280 nm overlapped with that of the DHPPA. 3.2 In vitro metabolism of C-QA by growing microbiota cultures 3.2.1 C-QA bioconversion Single-stage continuous fermentations of the colonic microbiota from subject V3 were carried out in order to perform C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
bioconversion experiments of C-QA with growing microbiota cultures. Cultures were inoculated with a bacterial concentration of approximately 1.3 × 10+9 cells/mL and were given 1.5 mM C-QA. To establish whether the fate of C-QA in microbiota cultures may be affected by a potentially probiotic strain which had been demonstrated to be efficient in hydrolyzing C-QA to CA [32], microbiota cultures were supplemented with 6.0 × 10+7 cells/mL of B. animalis subsp. lactis WC 0432 and were compared with the nonsupplemented ones (hereinafter named PMC and MC cultures, respectively). The time course of single representative PMC and MC processes are presented in Fig. 3. C-QA quickly disappeared regardless of the presence of B. animalis subsp. lactis WC 0432, being only the 0.2% and 1.5% after 8 h in PMC and MC cultures, respectively. C-QA hydrogenation proceeded faster than ester hydrolysis. The major transformation product was DHPP-QA, which peaked after 6 and 8 h in PMC and MC, respectively, while CA never accumulated above 1%. Later, DHPP-QA was hydrolyzed to DHPPA, which was further dehydroxylated to HPPA. In PMC and MC, the kinetics of these reactions were different. In MC, C-QA hydrogenation to DHPP-QA was delayed, C-QA being not transformed during the first 4 h (Fig. 3B). Conversely, in PMC, reduction of C-QA began earlier, and it was very efficient between 4 and 6 h (Fig. 3A). DHPPA, resulting from DHPP-QA hydrolysis, accumulated in greater amount in PMC cultures and was dehydroxylated more lately to HPPA, compared to MC cultures, providing a hint that dehydroxylation was the limiting step.
3.2.2 Evolution of fecal microbiota and fermentation products In continuous fermentations of the colonic microbiota (PMC and MC), bifidobacteria, Bacteroidetes, cluster XIVa Clostridiales, Enterobacteriaceae, F. prausnitzii, and total eubacteria were quantified every 4 h by FISH (Fig. 4A and B). Steadystate conditions were achieved in 16 h for total eubacteria and specific bacterial groups as well. The counts of total eubacteria increased up to 2.5 × 1010 and 3.1 × 10 10 cells/mL (in PMC and MC, respectively, p > 0.05) in the first 16 h. At the www.mnf-journal.com
1128
F. Tomas-Barberan et al.
Mol. Nutr. Food Res. 2014, 58, 1122–1131
significant difference in bifidobacterial counts at the steady state. At each time-point, bifidobacteria were isolated from PMC cultures, and then the isolates that were positive to Bifidobacterium-specific PCR were characterized by RAPDPCR fingerprinting in order to trace the B. animalis subsp. lactis WC0432. At the beginning of the process, B. animalis subsp. lactis WC0432 accounted for the 70% of bifidobacterial isolates, then decreased to the 6.7% after 4 h and stabilized at the 2.0% at the steady state, representing approximately the 0.2% of total eubacterial population. Formate, acetate, lactate, propionate, butyrate, succinate, and ethanol originated by microbiota metabolism during continuous cultivation and reached steady-state values after approximately 16 h (Fig. 4C and D). Formate and succinate accumulated at higher levels in MC than in PMC cultures during the exponential phase (6–10 h), then decreased to similar concentrations. At the steady state, the cultures differed in the levels of butyrate, which was higher in PMC than in MC (p < 0.05), and acetate, which was lower in PMC than in MC (p < 0.05). The other fermentation products exhibited similar steady-state concentrations in PMC and MC processes (p > 0.05). Lactate was found in low concentration, being always 0.05). Bacteroidetes grew with the highest growth rate, because they increased by approximately 2.7 magnitude orders, and became the dominant bacterial group in both PMC and MC (1.3 × 1010 and 2.0 × 1010 cells/mL, respectively, p > 0.05). At the steady state, Enterobacteriaceae were also similar in PMC and MC cultures (3.1 × 108 and 2.0 × 108 cells/mL, respectively, p > 0.05), even if these values evolved with different kinetics. Bifidobacteria increased by approximately 0.9 magnitude orders and were 1.9 × 108 and 3.5 × 108 cells/mL in PMC and MC cultures, respectively (p > 0.05). Compared to MC cultures, the supplementation with 6.0 × 107 cells/mL of B. animalis subsp. lactis WC0432 in PMC cultures did not determine a C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Discussion
Information from metagenome sequencing analysis provides evidence that considerable differences in gut microbiota composition occur among subjects, where this diversity is responsible of different metabolic activities [33, 34]. Consistently, studies dealing with bioavailability of phytochemical metabolites demonstrated that different transformations are largely attributable to the diverse bacterial populations [35,36]. C-QA, which is a major phenolic acid in the Western diet, is first hydrolyzed to CA, which may be subject of hydrogenases, and dehydroxylases, yielding derivatives of phenylpropanoic acid [22–24]. Information about interindividual diversity in C-QA metabolism is scarce. This study aimed to compare C-QA biotransformation by different human colonic microbiota, through resting cells experiments carried out with nine different colonic populations. In order to replicate experiments with reproducible inocula, bioconversion were performed with freeze-preserved microbiota, which have been demonstrated to yield the same bioconversion pattern of polyphenols, compared to fresh samples [37]. The diverse bacterial communities were able to hydrogenate, dehydroxylate C-QA, or break the QA moiety also before hydrolyzing its ester bond. However, all the transformation pathways converged on HPPA, the final metabolite in most of the samples. Conversion of C-QA into HPPA requires the removal or the breakdown of the QA moiety, the reduction of the propenoic double bond, and the dehydroxylation of the aromatic ring in position 4. Based on the intermediates that have been identified by HPLC-UV/Vis-MS/MS www.mnf-journal.com
Mol. Nutr. Food Res. 2014, 58, 1122–1131
1129
Figure 4. Time course of bacterial composition and fermentation products in continuous cultures of intestinal microbiota in the presence (A, C) or the absence (B, D) of B. animalis subsp. lactis WC 0432.Bacteria symbols (panels A and B): eubacteria, 䊉; Bacteroidetes, 䊊; bifidobacteria, ; Enterobacteriaceae, 䉭; cluster XIVa Clostridiales, ; Faecalibacterium prausnitzii, . Fermentation products symbols (panels C and D): lactate, 䊉; succinate, 䊊; formiate, ; acetate, 䉭; propionate, ; ethanol, ; butyrate, u. The results of a representative experiment performed in triplicate are shown.
and Q-TOF, these reactions can be performed in different sequences, resulting in diverse metabolic networks. The dehydroxylation seems to be the limiting step, because DHPPA accumulated in large amounts in most of the samples before being transformed into HPPA. Furthermore, HCi-QA was found, if any, only in low concentration, indicating that ester hydrolysis and double bond hydrogenation are more efficient and that C-QA is preferably transformed through the reactions yielding DHPPA. The hydrogenation and the ester hydrolysis and/or the breakdown of the QA moiety are necessary to obtain DHPPA. Based on the intermediates, the reactions occurred in diverse sequences and combinations in the different samples. Interestingly, the metabolism by gut microbiota of other polyphenols such as catechins, rutin, and naringin [38–40] leads to DHPPA, which was similarly subjected to paradehydroxylation into hydroxyphenyl-caroboxylic acids. Some of the compounds identified in this study, i.e. caffeoyl-glycerol metabolites C-G and DHPP-G, are described for the first time as microbiota metabolites of C-QA (Fig. 1). The presence of C-G and DHPP-G disclosed a novel degradation pathway of C-QA, where the cyclohexane ring was broken, leaving a glycerol residue esterified with CA. The C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
degradation of the QA residue before breaking the ester link, resulting in the caffeoyl-glycerol metabolites, is proposed here for the first time. HPPA was the bioconversion end product of most subjects, which did not proceed toward the formation of phenylpropanoic acid, unlike previous studies [23, 24]. Only three subjects (V2, V5, and V9) were capable to further transform HPPA into PAA, highlighting the presence of decarboxylating and dehydroxylating enzymatic activities. In one out of the three subjects (V9), HPE was also found in high amount, being one of the major metabolites of C-QA. It remains to be established whether HPE is an intermediate in HPPA conversion toward PAA, or whether the metabolic route diverges to yield two different products. Whenever the metabolic network did not stop and HPPA was subjected with great efficiency to further transformations, PAA and/or HPE accumulate in high amounts. Continuous cultures of the microbiota from V3 were carried out in presence of 1.5 mM C-QA. The choice of this microbial population was due to the fact that its resting cells did not perform the bioconversion beyond HPPA, and that it was the sole sample accumulating DHPP-QA. Then, we expected that the supplementation of the Bifidobacterium strain able to www.mnf-journal.com
1130
F. Tomas-Barberan et al.
hydrolyze the C-QA into CA could modify the flux toward this latter metabolite, highlighting the competition between the two different pathways. The substrate was rapidly consumed, while DHPP-QA, DHPPA, and HPPA were produced in succession. Neither the dehydroxylation nor the breakdown of QA residue took place as the first step and CA was detected only in minor amount (
