Epigallocatechin

Isolation and characterization of rat intestinal bacteria involved in biotransformation of (−)‑epigallocatechin

Akiko Takagaki · Yuko Kato · Fumio Nanjo
Received: 15 October 2013 / Revised: 6 June 2014 / Accepted: 9 June 2014
© Springer-Verlag Berlin Heidelberg 2014

Abstract

Two intestinal bacterial strains MT4s-5 and MT42 involved in the degradation of ( )-epigallocatechin (EGC) were isolated from rat feces. Strain MT4s-5 was tentatively identified as Adlercreutzia equolifaciens. This strain con- verted EGC into not only 1-(3, 4, 5-trihydroxyphenyl)-3-(2, 4, 6-trihydroxyphenyl)propan-2-ol (1), but also 1-(3, 5-dihy-droxyphenyl)-3-(2, 4, 6-trihydroxyphenyl)propan-2-ol (2), and 4′-dehydroxylated EGC (7). Type strain (JCM 9979) of Eggerthella lenta was also found to convert EGC into 1. Strain MT42 was identified as Flavonifractor plautii and con- verted 1 into 4-hydroxy-5-(3, 4, 5-trihydroxyphenyl)valeric acid (3) and 5-(3, 4, 5-trihydroxyphenyl)-γ-valerolactone (4) simultaneously. Strain MT42 also converted 2 into 4-hydroxy-5-(3, 5-dihydroxyphenyl)valeric acid (5), and 5-(3, 5-dihydroxyphenyl)-γ-valerolactone (6). Furthermore, F. plautii strains ATCC 29863 and ATCC 49531 were found to catalyze the same reactions as strain MT42. Interestingly, formation of 2 from EGC by strain MT4s-5 occurred rapidly in the presence of hydrogen supplied by syntrophic bacteria. Strain JCM 9979 also formed 2 in the presence of the hydro- gen or formate. Strain MT4s-5 converted 1, 3, and 4 to 2, 5, and 6, respectively, and the conversion was stimulated by hydrogen, whereas strain JCM 9979 could catalyze the con- version only in the presence of hydrogen or formate. On the basis of the above results together with previous reports, the principal metabolic pathway of EGC and EGCg by catechin- degrading bacteria in gut tract is proposed.

Introduction

It is well recognized that tea catechins have various physi- ological functions (Johnson et al. 2012) including antioxi- dative (Higdon and Frei 2003), blood cholesterol lower- ing, blood sugar level lowering (Kuroda and Hara 2004) and cancer preventive activities (Yang and Wang 2010). Many studies have shown that ( )-epigallocatechin gal- late (EGCg), the most abundant catechin in green tea, has strong physiological activity. Accordingly, it is considered important to elucidate the mechanisms responsible for their biological activities, and thereby, a number of stud- ies concerning the absorption, distribution, and excretion of tea catechins have been conducted (Lee et al. 1995; Kida et al. 2000; Kohri et al. 2001; Sang et al. 2008). In previous studies, we determined that after oral administration of ( )-[4-3H] EGCg to rats, the amount of intact EGCg absorbed in the body is very low (0.26 % of its dose), whereas the absorption rate of its degradation products by intestinal bacteria is 32.1 % of the dose (Kohri et al. 2001). The above observations suggest that elucidation of the met- abolic pathway of EGCg in the intestinal tract is important for further understanding of its physiological function.

We have investigated the biotransformation of EGCg by rat intestinal microflora and have proposed the metabolic pathway of EGCg in the gut tract (Takagaki and Nanjo 2010). EGCg is hydrolyzed to produce ( )-epigallocat- echin (EGC) and gallic acid in the first step of microbial metabolism. Then, EGC undergoes ring-opening between the 1 and 2 positions to form 1-(3, 4, 5-trihydroxyphenyl)-3-(2, 4, 6-trihydroxyphenyl)propan-2-ol (1) followed by the dehydroxylation of metabolite 1 to 1-(3, 5-dihy- droxyphenyl)-3-(2, 4, 6-trihydroxyphenyl)propan-2-ol (2). Subsequently, the metabolites 1 and 2 undergo deg- radation of the phloroglucinol moiety to produce, respec- tively, 4-hydroxy-5-(3, 4, 5-trihydroxyphenyl)valeric acid (3) and 4-hydroxy-5-(3, 5-dihydroxyphenyl)valeric acid (5) as major products. At the same time, 5-(3, 4, 5-trihydroxyphenyl)-γ-valerolactone (4) from 1 and 5-(3, 5-dihydroxyphenyl)-γ-valerolactone (6) from 2 are pro- duced. Full formulas of compounds 1–6 are illustrated in Fig. 7. A part of the above metabolites is further con- verted to their corresponding hydroxyphenyl valeric acids and hydroxyphenyl propionic acids. With respect to cat- echin metabolites including the above EGCg metabolites, we have recently shown that the metabolites had radical scavenging activities against ABTS radical cation and their activities were estimated to be stronger than or nearly equal to 6-hydroxy-2, 5, 7, 8-tetramethylchroman-2-carboxylic acid (Trolox) (Takagaki et al. 2011).

On the other hand, there is limited information available on identification of intestinal bacteria directly involved in the degradation of catechins. We have previously reported that Enterobacter aerogenes, Raoultella planticola, Kleb- siella pneumoniae subsp. pneumoniae and Bifidobacte- rium longum subsp. infantis were capable of hydrolyz- ing EGCg to EGC and gallic acid (Sakamoto et al. 2009; Takagaki and Nanjo 2010). Wang et al. (2001) have found that Eggerthella sp. SDG-2 (accession No. EF413638, formerly Eubacterium sp.), which was identified by Jin et al. (2007), catalyzed the C ring-cleaving and subsequent 4′-dehydroxylation of ( )-epicatechin (EC) and EGC. Jin and Hattori (2012) reported that Eggerthella sp. CAT-1 (Accession No. JF798636), which is closely similar to strain SDG-2 (99.8 % 16S rRNA gene sequence identity), was capable of catalyzing the ring-cleaving and subsequent 4′-dehydroxylation of EC and ( )-catechin. Kutschera et al. (2011) clearly demonstrated that Eggerthella lenta rK3 catalyzed the ring-opening of EC and ( )-cat- echin and the resulting 1-(3, 4-dihydroxyphenyl)-3-(2, 4, 6-trihydroxyphenyl)propan-2-ols were degraded to 4-hydroxy-5-(3, 4-dihydroxyphenyl)valeric acids and their
corresponding γ-valerolactones by Flavonifractor plautii aK2. These findings revealed the platform for the metabolic route of catechins by intestinal bacteria.In this report, we describe the biotransformation of EGC by intestinal bacteria (strains MT4s-5 and MT42) isolated from rat feces and their related bacterial strains.Furthermore, since it was found that some bacterial strains were responsible for catabolism of EGC as syntrophic bac- teria, we refer to the role of these bacteria.

Materials and methods

Chemicals, medium, and bacteria

( )-Epigallocatechin was obtained from Sigma-Aldrich Japan (Tokyo). EGCg metabolites, 1-(3, 4, 5-trihydroxy- phenyl)-3-(2, 4, 6-trihydroxyphenyl)propan-2-ol (1), 1-(3, 5-dihydroxyphenyl)-3-(2, 4, 6-trihydroxyphenyl)propan-2-ol (2), 4-hydroxy-5-(3, 4, 5-trihydroxyphenyl)valeric acid (3), 5-(3, 4, 5-trihydroxyphenyl)-γ-valerolactone (4), 4-hydroxy-5-(3, 5-dihydroxyphenyl)valeric acid (5), 5-(3, 5-dihydroxyphenyl)-γ-valerolactone (6), and (2R, 3R)-fla- van-3, 3′, 5, 5′, 7-pentol [4′-dehydroxylated EGC (7)], were prepared according to the methods previously reported (Takagaki and Nanjo 2010) and were used as reference standards for identification and quantification. All other chemicals were available products of analytical or HPLC grade. E. lenta JCM 9979 was purchased from Riken BioResource Center (Ibaragi, Japan). Escherichia coli NBRC 3301 (K-12) was obtained from Biological Resource Center, NITE (Chiba, Japan). F. plautii ATCC 29863 (for- merly Eubacterium plautii) and F. plautii ATCC 49531 (formerly Clostridium orbisciendens) were purchased from American Type Culture Collection (ATCC, Manassas, VA, USA). GAM broth and GAM agar were obtained from Nis- sui Pharmaceutical Co. Ltd (Tokyo, Japan). All cultures in this study were carried out under anaerobic condition at 37 °C by using an Anaero Pack (anaerobic cultivation) system with a Standing-Pouch and an AnaeroPack-Anaero (Mitsubishi Gas Chemical Company, Inc., Tokyo, Japan) unless otherwise stated.

Isolation of EGC-degrading bacteria

Fresh rat fecal sample (2 g) was homogenized in 5 ml of GAM broth and a loopful of the resulting homogenate was streaked on GAM agar plate. The plate was incubated at 37 °C for 72 h. Anaerobic bacterial populations located on several areas of the GAM agar plate were separately transferred to fresh GAM broth (3 ml) containing 1 mM EGC which was prepared by adding EGC solution previ- ously filtrated with a sterilized membrane filter (DISMIC- 25cs, cellulose acetate, 0.2 μm, ADVANTEC Toyo, Tokyo, Japan). After incubation for 48 h at 37 °C, a portion of each culture was centrifuged at 12,000 g for 10 min and the resulting supernatants were analyzed by an LCQ Deca XPplus LC/MS system as described later to select the bacterial mixtures capable of cleaving EGC. The selected bacterial mixtures were separately re-streaked on the GAM agar plates again, and the plates were incubated at 37 °C for 72 h. Several bacterial populations grown on each agar plate were picked up and then transferred separately to GAM broth containing 1 mM EGC. After incubation for 48 h at 37 °C and then centrifugation to remove bac- terial cells, the supernatants were analyzed by the LC/MS system. The bacterial cultures capable of degrading EGC were selected. After repeating the above procedures five to six times, the selected bacterial cultures possessing EGC- degrading ability were diluted by factors of 10−2–10−7 with sterile water and each of the diluted culture solutions was spread onto the GAM agar plates. After incubation for 48 h at 37 °C, two bacterial strains (MT4s-3 and MT4s-5) which formed a single colony were picked up and were separately cultured in GAM broth containing 1 mM EGC for 48 h at 37 °C. After centrifugation, metabolites in each supernatant were analyzed by the LC/MS system.

We further attempted to isolate intestinal bacterium which is capable of cleaving 1-(3, 4, 5-trihydroxyphe- nyl)-3-(2, 4, 6-trihydroxyphenyl)propan-2-ol (1). Fresh fecal sample (2 g) was suspended in 5 ml of distilled water, and 0.1 ml each of the suspension was placed in ten or more test tubes containing GAM broth (2 ml) supplemented 1 mM metabolite 1. After incubation at 37 °C for 48 h, 1 ml of each of the cultures was withdrawn and centrifuged at 12,000 g for 10 min. The supernatants were applied to the LC/MS system to examine the converting ability of 1. The cultures showing converting abilities were selected, and 0.1 ml each of the selected cultures was further subcultured in several fresh GAM broths (2 ml) containing metabolite 1 and the cultures showing conversion abilities were selected again by the LC/MS analysis. After repeating the above procedure five to six times, a bacterial culture which main- tained the converting ability of 1 was finally selected. Then, the 10−1 to the 10−7 dilute solutions (0.1 ml each) of the culture were spread onto GAM agar plates, and the plates were incubated at 37 °C for 48 h. More than 100 individual colonies were picked up and examined for the degradation ability of 1 by LC/MS analysis.

Isolation of syntrophic bacteria working with strain MT4s-5

Fresh rat fecal sample (2 g) was homogenized in 5 ml of GAM broth. The homogenate was diluted serially from 10−1 to 10−9 with the same medium. Each suspension (50 μl) from the 10−5 to the 10−9 dilutions was put on petri dishes. GAM broth containing 0.5 % agar, which had been previously autoclaved and cooled to achieve a temperature of around 40 °C, was poured into the dishes and then mixed gently. The dishes were anaerobically incubated at 37 °C for 72 h. Single colonies in the dishes were each transferred to 3 ml of GAM broth containing 1 mM EGC, and then preculture (50 μl) of isolate MT4s-5 in GAM broth was added to the medium. After incubation at 37 °C for 48 h, cultures were centrifuged at 12,000 g for 10 min and the supernatants were analyzed by the LC/MS system to exam- ine the conversion of EGC to metabolite 2.

Determination of 16S rRNA gene sequences of isolates

16S rRNA gene sequences were determined using a MicroSEQ 16S rRNA Full Gene PCR and Sequencing Kit according to the protocol of the manufacturer (Life Tech- nologies, Carlsbad, CA USA). PCR products were purified using a QuickStep 2 PCR Purification Kit (Edge BioSys- tems, MD USA) and AutoSeq G-50 (GE Healthcare UK Ltd.). 16S rRNA sequences of the PCR products were analyzed by an ABI PRISM 3100 Genetic Analyzer with Sequencing Analysis Software version 5.1.1 (Life Tech- nologies). Primer sequences were removed from the target nucleotide sequences, and the trimmed sequences were subjected to identity searches and closest relatives identi- fied using a BLASTN search algorithm of the GenBank database (Altschul et al. 1990) or a MicroSEQ ID Software version 2.1.1. (Life Technologies). Sequences of isolates and their related species were aligned by a CLUSTAL W program (Thompson et al. 1994), and gaps and unidentified base positions were removed using a BioEdit (Hall 1999) software package.

Degradation experiments of bacterial isolates and their related bacterial strains

Strain MT4s-5 and E. lenta JCM 9979 were separately precultured in GAM broth (5 ml) at 37 °C for 48 h. Each preculture (0.5 ml) was then inoculated into 5 ml of fresh GAM broth containing 1 mM EGC and was incubated at 37 °C. GAM broth containing EGC was also incubated without bacterium under the same condition as a control. After incubation for 24, 48, and 72 h, aliquots (0.5 ml) of the incubation mixture were withdrawn in an anaerobic glove box under CO2 atmosphere for the LC/MS/MS anal- ysis as described below.

After incubation of strain MT42 in GAM broth (5 ml) at 37 °C for 24 h, aliquots (0.5 ml) of the culture were inoculated into 5 ml of fresh GAM broth containing 1 mM metabolite 1 and the culture was incubated at 37 °C. A por- tion (0.5 ml) of the culture was taken out after every 24 h of incubation for the LC/MS/MS analysis. Degradation experiments for EGC and metabolite 2 were performed in the same way as the experiment for metabolite 1. In the case of F. plautii strains ATCC 29863 and ATCC 49531, degrada- tion experiments for EGC, 1 and 2 were also carried out according to the same procedure.

Degradation of EGC by isolate MT4s-5 and Eggerthella lenta JCM 9979 in the presence of syntrophic bacteria

In this study, isolates MT4s-3, MT01 and MT12, and E. coli NRBC 3301 were used as syntrophic bacteria. Strains MT4s-5 and MT4s-3 were individually precultured with GAM broth (5 ml) at 37 °C for 48 h. The preculture (0.5 ml each) of both strains was inoculated into GAM broth (5 ml) containing 1 mM EGC and the mixed culture was carried out at 37 °C. After incubation for 24, 48, and 72 h, aliquots (0.5 ml) of the incubation mixture were withdrawn for the LC/MS/MS analysis. Co-culture of strains MT4s-5 with NRBC 3301, MT4s-5 with MT01, MT4s-5 with MT12, E. lenta JCM 9979 with MT4s-3, JCM 9979 with NRBC 3301, JCM 9979 with MT01, and JCM 9979 with MT12, respectively, was also conducted in the same way as above. To examine what factor(s) generated by syntrophic bacteria are necessary for 4′-dehydroxylation by strains MT4s-5 and JCM 9979, a culture vessel (PAL Corpora- tion, Tokyo, Japan) separated into two partitions A and B by a membrane filter was used. This method was designed by Ohno et al. (1999). The vessel consists of two parts, each of which has a cotton-plugged inlet and the assem- bled vessel has a capacity of 114 ml of media in total. A membrane filter (Supor 200, PES, 0.2 μm, PAL Corporation, Tokyo, Japan) with a diameter of 47 mm was placed between the two partitions A and B. GAM broth (50 ml) was poured in each partition of the vessel. Then, EGC solu- tion sterilized by membrane filter was aseptically added to the GAM broth in both partitions to make the medium con- taining 1 mM EGC. Preculture (15 ml) of strain MT4s-5 in GAM broth was inoculated into the medium in partition A and preculture (1 ml) of strain MT4s-3 into the medium in partition B. The vessel was anaerobically incubated at 37 °C for 24 h with gentle shaking. Aliquots (0.5 ml) of the culture in partition A were withdrawn for LC/MS/MS analysis. The same experiments were conducted with the following strain pairs: MT4s-5 and E. coli NRBC 3301, JCM 9979 and MT4s-3, and JCM 9979 and NRBC 3301, respectively.

Furthermore, factors necessary for dehydroxylation reaction by isolate MT4s-5 and E. lenta JCM 9979 were investigated. Preculture (0.5 ml) of strain MT4s-5 of GAM broth at 37 °C for 24 h was inoculated into GAM broth (5 ml) containing 1 mM EGC and 3 mM sodium formate and was incubated anaerobically at 37 °C. After 24 and 48 h incubation, aliquots (0.5 ml) of the incubation mix- ture were withdrawn for LC/MS/MS analysis. In the case of the hydrogen test, GAM broth (5 ml) containing 1 mM EGC was individually inoculated with preculture (1 ml) of strain MT4s-5 and then hydrogen gas was aseptically bub- bled into the incubation mixture for 10 s at a flow rate of about 50 ml/min. The resultant culture in a test tube was packed in a Standing-Pouch (w 220 L 320 mm) with AnaeroPack-Anaero (Mitsubishi Gas Chemical Company, Inc.). Air in the pouch was roughly removed by hand, and hydrogen gas was injected to the pouch for 30 s at a flow rate of about 800 ml/min. Cultivation, sampling, and the LC/MS/MS analysis were performed in the same manner as above. Further, the experiments with strain JCM 9979 were also conducted in the same way.

Dehydroxylation by isolate MT4s-5 and Eggerthella lenta JCM 9979

Preculture of strain MT4s-5 and E. lenta JCM9979 was separately inoculated in GAM broth (5 ml) containing 1 mM each of metabolites 1, 3, and 4, and the cultures were performed anaerobically at 37 °C. At 24 h intervals, a por- tion (0.5 ml) of each incubation mixture was taken out for LC/MS/MS analysis. Dehydroxylation experiments of the above metabolites were carried out individually in the pres- ence of syntrophic bacteria, hydrogen, and formate.

Preparation of purified metabolites for biotransformation experiments

Metabolites used for biotransformation experiments were purified according to the method described in our previ- ous paper (Takagaki and Nanjo 2010). For the production of metabolite 1, preculture (100 ml) of E. lenta JCM 9979 in GAM broth was poured into fresh GAM broth (500 ml) containing 3 mM EGC. The cultivation mixture was incu- bated at 37 °C with gentle shaking until elimination of EGC. A mixed culture of E. lenta JCM 9979 (100 ml of preculture) and strain MT4s-3 (20 ml of preculture) in 500 ml of GAM broth containing 3 mM EGC at 37 °C for 48 h was incubated to produce metabolite 2. Metabolites 3 and 4 were produced by a mixed culture of E. lenta JCM 9979 (100 ml of preculture) and strain MT42 (20 ml of pre- culture) in GAM broth (500 ml) containing 3 mM EGC at 37 °C for 48 h. Each of the above cultures was centrifuged to remove bacterial cells. After pH adjustment of the super- natant containing either metabolite 1 or 2 to about 3.0–4.0 with 5 M HCl, each supernatant was extracted three times with 400 ml of ethyl acetate. The ethyl acetate fraction containing either 1 or 2 was evaporated to dryness and was then dissolved in a small volume of 5 % aqueous methanol. In the case of the supernatant containing both metabolites 3 and 4, the supernatant was adjusted to around pH 1.5–2.0 with 5 M phosphoric acid and was extracted three times with 400 ml of ethyl acetate/n-butanol (1/1, v/v). The ethyl acetate/n-butanol fraction was extracted three times with 200 ml of 50 mM aqueous sodium carbonate solution con- taining 0.1 % sodium ascorbate. In this extraction process, 3 moved into the aqueous fraction, whereas 4 remained in the organic fraction. The organic fraction was concentrated to dryness and dissolved in a small volume of 5 % aque- ous methanol. The aqueous methanol solutions were each purified to preparative HPLC as reported in our previous paper (Takagaki and Nanjo 2010). Finally, metabolites 1 (165 mg), 2 (210 mg), and 4 (67 mg) were obtained. The sodium carbonate fraction containing 3 was adjusted to pH 7 with 5 M HCl and was evaporated to about 10–20 ml. The concentrate was adjusted to pH 2.0 with 2 M HCl, and the resulting solution was immediately subjected to preparative HPLC under the conditions as described in our previous paper (Takagaki and Nanjo 2010). The effluent containing 3 was collected, concentrated to remove organic solvents, and applied to a Diaion SK1B ion exchange column [Na+ form, 65 10 mm (i.d.)]. The column was eluted with 25 ml of distilled water. The eluate was concentrated and freeze-dried to obtain 3 (65 mg).

LC/MS and LC/MS/MS analyses

LC/MS for screening and structural analysis was per- formed by a Surveyor HPLC and an LCQ Deca XPplus system (Thermo Fisher Scientific K. K., Yokohama, Japan) as described in our paper (Takagaki and Nanjo 2010). For LC/MS/MS analysis, samples were prepared as follows. To each cultivation mixture (0.5 ml) was added 0.05 ml of 2 M HCl and the resulting mixture was centri- fuged at 12,000 g for 10 min at 4 °C. A portion (0.2 ml) of each resulting supernatant was diluted with 0.8 ml of 0.5 % aqueous acetic acid and was used as sample solu- tion. LC/MS/MS analysis for quantitation of metabolites was performed using a model Agilent 1100 series LC sys- tem (Agilent Technologies, Tokyo, Japan) coupled with a 3200 QTRAP LC/MS/MS system (AB SCIEX, MA, USA).

HPLC was carried out on a 100 × 2 mm (i.d.), 3 μm, CAP-CELLPAK C18 MG column (Shiseido Co. Ltd, Tokyo, Japan), and the solvent system consisted of gradient system with solvent A (water/acetonitrile/acetic acid 500/12.5/0.5, v/v/v) and solvent B
(water/acetonitrile/methanol/ace- tic acid 100/100/100/0.3, v/v/v/v). Gradient elution was performed at 0.2 ml/min with the following concentration: 4 min hold at 100 % solvent A, 5-min linear gradient in sol- vent B from 0 to 10 %, 9-min hold at 10 % B, 3-min linear gradient from 10 to 25 % B, 3-min linear gradient from 25 to 100 % B, 0.1-min linear gradient in solvent A from 0 to 100 %, and 6-min hold at 100 % A. The mass detector was equipped with turbo-ion spray (electrospray ionization, ESI) source and operated in multiple reactions monitoring under negative ion mode. For all the mass scan modes, ion spray voltage was maintained at 4,000 V, curtain gas was set to ten (arbitrary units), the collision gas was five (arbi- trary units), and capillary temperature was set at 550 °C. The optimized instrument setting of each metabolite was listed in Table 1. The LC/MS/MS system was controlled using an Analyte version 1.6.1 software. The data were acquired and processed by the same software. Standard solutions were prepared as follows: Each reference stand- ard (metabolites 1, 2, 3, 4, 5, 6, and 7) weighed accurately was dissolved in 0.1 % aqueous acetic acid containing 3 % methanol, and each solution was diluted with 0.5 % aque- ous acetic acid at five concentration levels in the range from 50 to 500 μM. To the resulting solutions (0.2 ml) were sep- arately added 0.2 ml of GAM broth and 0.6 ml of 0.5 % aqueous acetic acid. The standard solutions thus prepared were analyzed by the LC/MS/MS system, and calibration curves were obtained by plotting the MS intensity of each reference standard against the concentration.

Formate and hydrogen analysis

Syntrophic bacteria were inoculated individually into 5 ml of GAM broth and incubated under anaerobic con- dition for 24 h at 37 °C. After centrifugation (12,000 g for 5 min) to remove the bacterial cells, the supernatants were applied to a G1600A Capillary electrophoresis (CE) system (Agilent Technologies Japan, Ltd, Tokyo, Japan). Separation of organic acids was performed using a fused silica capillary tube (i.d. 75 μm, effective length 72 cm, total length 80.5 cm, Agilent Technologies) at a con- stant temperature (20 °C).
The electrode was calibrated using an Organic acids buffer (pH 5.6) for CE (Agilent Technologies). Sample solutions were injected at 50 mbar for 2.0 s. The electrophoretic separation process was con- ducted under 25 kV and monitored by photodiode array detector at signal 350 nm/bandwidth 20 nm and reference 270 nm/bandwidth 10 nm.

GAM broth (10 ml) in sterilized glass vessel (25 ml capacity) was inoculated with each preculture of the syn- trophic bacteria as mentioned above. After incubation at 37 °C for 2 h, each vessel was sealed up in a glove box filled with CO2 atmosphere and was further incubated anaerobically for 24 h. Headspace gas (2 ml) in each vessel was analyzed by a GC-14B system (Shimadzu corporation, Kyoto, Japan) at SEIKAN Co. Ltd. (Shizuoka, Japan) to determine hydrogen generated.

Other analytical methods

NMR analysis was taken on a Bruker Ultrashield 400 plus system (1H, 400 MHz; 13C, 100 MHz: Bruker BioSpin K. K., Yokohama, Japan). All samples were dissolved in meth- anol-d4 (Kanto Chemical, Tokyo, Japan). Chemical shifts were referenced to tetramethylsilane (TMS) at 0 ppm. Spe- cific rotations ([α]20) of metabolites were measured by a P-1020 Polarimeter (JASCO Corporation, Tokyo, Japan), and the solvents and sample concentrations used were described in the text.

Results

Isolation and identification of EGC-degrading bacteria

To clarify intestinal bacteria responsible for the catabo- lism of EGCg and EGC, we made an attempt to isolate bacteria capable of degrading EGC from rat feces. A bac- terial strain MT4s-5 was isolated as an EGC-degrading bacterium. This isolate catalyzed the conversion of EGC into metabolites 1 [1-(3, 4, 5-trihydroxyphenyl)-3-(2, 4, 6-trihydroxyphenyl)propan-2-ol], 2 [1-(3, 5-dihydroxy-
phenyl)-3-(2, 4, 6-trihydroxyphenyl)propan-2-ol], and 7 [4′-dehydroxylated EGC]. The 16S rRNA gene sequence of the strain was determined and compared with corre- sponding bacterial sequences in databases. The sequence of strain MT4s-5 (1,460 bp, accession number AB 693938) was highly homologous not only to A. equolifaciens strains FJC-B9 (AB306661, 99.9 % identity) and FJC-A10 (AB306660, 99.8 % identity) (Maruo et al. 2008) but also to Asaccharobacter celatus do03 (AB266102, 99.9 % iden- tity) (Minamida et al. 2008) and a human intestinal bacte- rium SNU Julong 732 (AY310748, 99.8 % identity). The above bacteria which were highly homologous to strain MT4s-5 were reported to convert daidzein to equol (Maruo et al. 2008; Minamida et al. 2008). Strain MT4s-5 was also found to have the ability to convert daidzein to equol via dihydrodaidzein (Ito et al. 2008). The 16S rRNA gene sequence of A. equolifaciens FJC-B9 was also shown to be highly homologous (99.9 % identity) to that of A. cela- tus do03, and hence, isolate MT4s-5 was determined to be either A. equolifaciens or A. celatus. Therefore, we could not give a single name to isolate MT4s-5. However, since A. equolifaciens gen. nov., sp. nov., (Maruo et al. 2008) has been proposed slightly earlier than A. celatus gen. nov., sp. nov., (Minamida et al. 2008), we tentatively named strain MT4s-5 A. equolifaciens. It was already reported that Egg- erthella sp. SDG-2 (Wang et al. 2001), Eggerthella sp. CAT-1 (Jin and Hattori 2012), and E. lenta rK3 (Kutschera et al. 2011) had similar ability to strain MT4s-5, but the 16S rRNA gene sequence of strain MT4s-5 showed only 91.8, 92.3, and 91.3 % identities to that of Eggerthella sp. SDG-2 (Jin et al. 2007), Eggerthella sp. CAT-1 (Jin and Hattori 2012), and E. lenta rK3 (Kutschera et al. 2011), respectively.

Adlercreutzia equolifaciens MT4s-5 could catalyze C ring cleavage between the 1 and 2 positions of EGC to yield metabolite 1, but there was no further degradation of 1 except for its dehydroxylation at 4′ position. Then, we screened bacteria capable of degrading 1 from rat feces and finally isolated a strain MT42. 16S rRNA gene sequence of strain MT42 was shown to be almost identical to that of
F. plautii strains such as AIP 165.06 (EU541436, 99.9 % identity), CCUG 28093 (AY724678, 99.7 % identity), aK2 (HQ455040, 99.5 % identity) and DSM 6740 (Y18187, 99.5 % identity). Consequently, strain MT42 was identified as F. plautii (AB693937). Among F. plautii strains, F. plau- tii aK2 and F. plautii DSM 6740 have been reported to pos- sess similar degrading ability to F. plautii MT42 (Kutschera et al. 2011).

Biotransformation of EGC by Adlercreutzia equolifaciens MT4s-5 and Eggerthella lenta JCM 9979 and identification of their products

Adlercreutzia equolifaciens MT4s-5 was anaerobically cultured in GAM broth containing 1 mM EGC at 37 °C. A portion (1 ml) of the culture was withdrawn every 24 h, and then the bacterial cells were removed by centrifuga- tion. The resulting supernatant was analyzed by both LC/ MS for structural identification and LC/MS/MS for quan- tification as indicated in “Materials and Methods” section. Metabolites 1, 2, and 7 were detected, and their forma- tions with incubation time and their MS and MS/MS data are illustrated in Fig. 1. The major metabolite 1 showed almost the same in HPLC retention time, UV spectra, MS and MS/MS data, and 1H- and 13C-NMR spectra as those of the reference standard 1-(3, 4, 5-trihydroxyphenyl)-3-(2,4, 6-trihydroxyphenyl)propan-2-ol and was therefore con- firmed to be the same compound. Similarly, other metabo- lites 2 and 7 were also determined in the same manner as above. Finally, 2 and 7 were determined to be 1-(3, 5-dihy- droxyphenyl)-3-(2, 4, 6-trihydroxyphenyl)propan-2-ol and 4′-dehydroxylated EGC [(2R, 3R)-flavan-3, 3′, 5, 5′, 7-pen- tol], respectively. Optical rotation value of 7 was observed to be [α]20 51.4° (c 0.49, methanol).

Fig. 1 Biotransformation of EGC by A. equolifaciens MT4s-5 and MS and MS/MS spectra of EGC and its metabolites. EGC (open circle), metabolites 1 (filled circle), 2 (open triangle) and 7 (filled triangle), and EGC control in the absence of bacterium (open diamond).

Wang et al. (2001) have reported that Eggerthella sp. SDG-2, which was identified by Jin et al. (2007), catalyzed C ring-opening reaction of EGC, followed by 4′-dehy- droxylation to give 1-(3, 5-dihydroxyphenyl)-3-(2, 4, 6-trihydroxyphenyl)propan-2-ol (2). We purchased E. lenta JCM 9979 ( ATCC 25559) as a bacterial strain closely related to strain SDG-2 (99.9 % 16S rRNA gene sequence identity), and examined its C ring-opening ability on EGC.E. lenta JCM 9979 was shown to convert EGC to 1 (Fig. 2) as well as A. equolifaciens MT4s-5 and Eggerthella sp. SDG-2 (Wang et al. 2001). However, E. lenta JCM 9979 dif- fered in the formation of metabolite 2 from A. equolifaciens MT4s-5 and Eggerthella sp. SDG-2, in which the two strains MT4s-5 and SDG-2 could catalyze the 4′-dehydroxylation to form metabolite 2, but E. lenta JCM 9979 could not.

Fig. 2 Biotransformation of EGC by E. lenta JCM 9979. EGC (open circle), Metabolite 1 (filled circle).

Analyzed by the LC/MS system. F. plautii MT42 produced metabolites 3 and 4 from 1. Similarly, F. plautii MT42 converted 2 into metabolites 5 and 6. We confirmed their structural features by comparison of their HPLC reten- tion times, and UV spectra, MS, MS/MS and NMR data with those of reference standards. Finally, metabolites 3 and 4 were determined to be 4-hydroxy-5-(3, 4, 5-trihy- droxyphenyl)valeric acid and 5-(3, 4, 5-trihydroxyphenyl)- γ-valerolactone, respectively. Structures of metabolites 5 and 6 were also determined in the same way and were con- firmed to be 4-hydroxy-5-(3, 5-dihydroxyphenyl)valeric acid and 5-(3, 5-dihydroxyphenyl)- γ-valerolactone, respectively. Furthermore, metabolite formation from 1, 2, Biotransformation of EGC by Adlercreutzia equolifaciens MT4s-5 and Eggerthella lenta JCM 9979 in the presence of syntrophic bacteria During the isolation process of EGC-degrading bac- teria from rat feces, bacterial mixture including iso- late MT4s-5 was observed to quickly convert EGC into metabolite 2. Two kinds of bacteria were eventually iso- lated from the above bacterial mixture. One was identi- fied to be A. equolifaciens MT4s-5 which could convert EGC into metabolite 1 as mentioned above (Fig. 1). Another one was identified as E. coli MT4s-3 because 16S rRNA gene sequence of strain MT4s-3 showed 99.8 % identity to that of E. coli ATCC 11775 (type strain). However, E. coli MT4s-3 showed no conversion ability of EGC (data not shown). Interestingly, it was observed that although the bacterial mixture, including at least strains MT4s-5 and MT4s-3, rapidly converted EGC into metabolite 2 as mentioned above, the formation of 2 by A. equolifaciens MT4s-5 alone was observed to be very slow. From these observations, it was predicted that
A. equolifaciens MT4s-5 would exert its full potential ability to convert EGC to 2 in coexistence with E. coli MT4s-3. To clarify this supposition, co-cultivation of A. equolifaciens MT4s-5 with E. coli MT4s-3 was carried out in GAM broth containing 1 mM EGC and metabolite formation was examined. E. coli MT4s-3 was found to facilitate the formation of 2 by A. equolifaciens MT4s-5 as illustrated in Fig. 4a. As well as E. coli MT4s-3, E. coli NRBC 3301 also facilitated the formation of 2 by A. equolifaciens MT4s-5 (Fig. 4a). The results revealed that E. coli strains stimulated the production of 2 from 1 by 4′-dehydroxylation reaction catalyzed by A. equolifa- ciens MT4s-5, and therefore functioned as syntrophic bacteria.

Although E. lenta JCM 9979 was found to catalyze only the conversion of EGC to 1 (Fig. 2), the formation of 2 was examined in co-cultivation with E. coli strains MT4s-3 or NBRC 3301. Interestingly, E. lenta JCM 9979 was found to have the ability to catalyze the conversion of EGC into 2 in the presence of E. coli strains (Fig. 4b). The results indicated that E. lenta JCM 9977 had the potential ability to catalyze 4′-dehydroxylation of 1 and its ability could be exerted in coexistence of syntrophic bacteria. Thus, syn- trophic bacteria are considered to play an important role for the catabolism of EGC.

Then, we further screened syntrophic bacteria involved in the conversion of EGC to 2. Finally, two bacterial strains were isolated from rat feces and identified to be Butyrici- monas synergistica MT01 ( JCM 15148 CCUG 56610) and Butyricimonas virosa MT12 ( JCM 15149 CCUG 56611) as reported in our previous paper (Sakamoto et al. 2009). A. equolifaciens MT4s-5 and E. lenta JCM 9979 were co-cultured with each strain of MT01 and MT12 in GAM broth containing EGC and the formation of 2 was examined. Similar results as shown in Fig. 4 were observed in the experiments (data not shown), and hence, B. syner- gistica MT01 and B. virosa MT12 were confirmed to serve as syntrophic bacteria. It is understood that B. synergistica MT01 and B. virosa MT12 could not catalyze the biotrans- formation of EGC and metabolite 1 at all as well as E. coli strains MT4s-3 and NBRC 3301.

Biotransformation of EGC to 2 by Adlercreutzia equolifaciens MT4s-5 and Eggerthella lenta JCM 9979 in the presence of hydrogen or formate

Subsequently, we investigated the factors necessary for A. equolifaciens MT4s-5 and E. lenta JCM 9979 to con- vert metabolite 1 into 2. A. equolifaciens MT4s-5 and each of the syntrophic bacterial strains (B. synergis- tica MT01, B. virosa MT12 and E. coli MT4s-3) were cultured with a culture vessel physically separated into two partitions A and B by a membrane filter (0.2 μm) as described in “Materials and Methods” section. The same experiments were also conducted with E. lenta JCM 9979. After centrifugation of the culture, the result- ing supernatant was analyzed by the LC/MS/MS system.

A. equolifaciens MT4s-5 and E. lenta JCM 9979 rapidly produced 2 but not 1 from EGC (data not shown), sug- gesting that the two strains require some substance(s) which is produced by the syntrophic bacteria. We further attempted to determine the substances produced by the syntrophic bacteria. Krumholz and Bryant (1986) have already demonstrated that Eubacterium oxidoreducens G41 degraded pyrogallol in the presence of supernatant fluids of either anaerobically glucose-grown G44 strain or E. coli, and H2 or formate replaced the supernatant fluids from strain G44 or E. coli. They have also indi- cated that strain G41 catabolized gallate and phloroglu- cinol in the presence of formate or hydrogen. In addi- tion, it is well known that exocellular electron transfer plays an important role in anaerobic microbial commu- nities and interspecies hydrogen transfer among anaero- bic microorganisms is the driving force for degradation of organic compounds (Stams et al. 2006). Accordingly, A. equolifaciens MT4s-5 and E. lenta JCM 9979 were individually cultured in GAM broth containing 1 mM EGC in the presence of either hydrogen or formate as an electron donor. As a result, it was found that A. equolifa- ciens MT4s-5 could convert EGC into metabolite 2 rap- idly in the presence of hydrogen and E. lenta JCM 9979 could catalyze the conversion in the presence of either hydrogen or formate (data not shown). The observations suggested that syntrophic bacteria supplied hydrogen and/or formate to A. equolifaciens MT4s-5 and E. lenta JCM 9979. We further examined whether or not syn- trophic bacteria, B. synergistica MT01, B. virosa MT12, and E. coli strains MT4s-3 and NBRC 3301, actually produced hydrogen and/or formate. The experimental results showed that hydrogen was produced by all the four syntrophic bacteria and formate by E. coli strains (Table 2). Therefore, it was concluded that A. equolifa- ciens MT4s-5 and E. lenta JCM 9979 convert EGC into 1 and subsequently convert metabolite 1 into 2 by utiliz- ing hydrogen and/or formate.

We further investigated whether or not A. equolifaciens MT4s-5 and E. lenta JCM 9979 could cause p-dehydrox- ylation of pyrogallol moiety of metabolites 1, 3, and 4. A. equolifaciens MT4s-5 alone was found to catalyze the p-dehydroxylation of 1, 3, and 4 to produce 2, 5, and 6, respectively (Fig. 5a). The dehydroxylation reaction was also observed to be promoted by the presence of B. syn- ergistica MT01 (Fig. 5b) or hydrogen (Fig. 5c). Similar results were obtained in the presence of other syntrophic bacteria, B. virosa MT12 and E. coli strains MT4s-3 and NBRC 3301 (data not shown), but formate did not show the acceleration effects (Fig. 5d). These results demonstrated that A. equolifaciens MT4s-5 had the ability to catalyze the p-dehydroxylation reaction of the metabolites and required hydrogen supplied by syntrophic bacteria for stimulating the reaction. On the other hand, E. lenta JCM 9979 alone could not catalyze the p-dehydroxylation of 1, 3, and 4 at all (Fig. 6a). However, E. lenta JCM 9979 rapidly cata- lyzed the p-dehydroxylation of these metabolites in the presence of B. virosa MT12 (Fig. 6b). Similar results were also obtained with B. synergistica MT01 and E. coli strains MT4s-3 and NBRC 3301 (data not shown). The dehydrox- ylation readily progressed in the presence of either hydro- gen or formate (Fig. 6c, d). The results clearly showed that E. lenta JCM 9979 required at least either hydrogen or formate for catalyzing the dehydroxylation reaction of 1, 3, and 4.

Discussion

In this study, we isolated two bacterial strains, A. equolif- aciens MT4s-5 and F. plautii MT42, from rat feces. A. equolifaciens MT4s-5 catalyzed C ring cleavage of EGC to produce its corresponding 3-diphenylpropan-2-ol (1) as well as Eggerthella sp. SDG-2 (Wang et al. 2001). The ring cleavage was also found to be catalyzed by E. lenta JCM 9979 which is commercially available. However,and E. lenta JCM 9979. These strains could not catalyze the ring-cleaving reaction of 7 (data not shown), suggesting that the 4′-hydroxyl group of EGC is important for its ring cleavage caused by A. equolifaciens MT4s-5 and E. lenta JCM 9979. Wang et al. (2001) already reported that the presence of free 4′-hydroxyl group of catechins seems to be necessary for C ring cleavage of EGC since 4′-methylated EGC did not undergo biotransformation by Eggerthella sp. SDG-2.

Metabolite 1 was further converted into 3 and 4 simul- taneously by F. plautii MT42. Commercially available F. plautii strains ATCC 29863 and ATCC 49531 were also found to convert 1 into 3 and 4. The three F. plautii strains could also convert metabolite 2 into 5 and 6. In addition to this, F. plautii ATCC 49531 is capable of degrading the phloroglucinol moiety of 1-(3, 4-dihydroxyphenyl)-3-(2, 4, 6-trihydroxyphenyl)propan-2-ol derived from EC and ( )-catechin because strain ATCC 49531 is the same strain as DMS 6740 reported by Kutschera et al. (2011). These observations may suggest that F. plautii strains have the capability to degrade the phloroglucinol moiety (A ring) regardless of the number and positions of hydroxyl groups in the B ring of 3-diphenylpropan-2-ols. Accordingly, it is most likely that F. plautii ak2 degrades metabolites 1 and 2. From a wider perspective, it may be anticipated that F. plautii strains generally have the ability to degrade metabo- lites having phloroglucinol moiety.

We revealed for the first time that 4′-dehydroxylation followed after the C ring cleavage of EGC by A. equolifa- ciens MT4s-5 was accelerated remarkably by co-cultivation with syntrophic bacteria such as B. synergistica MT01 (JCM 15148), B. virosa MT12 (JCM 15149), and E. coli strains MT4s-3 and NBRC 3301, and that the dehydroxyla- tion by E. lenta JCM 9979 proceeded only in a mixed cul- ture of the syntrophic bacteria. We further found that the dehydroxylation by A. equolifaciens MT4s-5 was accel- erated in the presence of hydrogen and the reaction by E. lenta JCM 9979 proceeded only in the presence of either hydrogen or formate. Such syntrophic relationships are well recognized in anaerobic microbial communities and the transfer of hydrogen is termed interspecies hydrogen/ formate transfer (Stams and Plugge 2009). In this study, interspecies hydrogen/formate transfer was thought to play an important role in the catabolic processes of EGC that are caused by EGC-degrading bacteria in partnership with hydrogen/formate-producing bacteria. In relation to this, Decroos et al. (2005) have reported that the equol produc- tion from isoflavone daidzein by an equol-producing bacte- rial consortium (EPC4 containing at least 4 different bacte- ria) was stimulated to a large extent by hydrogen. Recently, Bolca and Verstraete (2010) have reported that in the co- culture of EPC4 and a methanogenic or a sulfate-reduc- ing bacterium, EPC4 showed significant suppression of methane or hydrogen sulphide production in the presence of soy germ powder as a source of daidzein. From these results, they suggested the beneficial health effects of soy consumption may be mainly attributable to equol produc- tion along with the reduction of the methane and hydro- gen sulphide which are negatively associated with health. In this study, we have found that hydrogen was utilized by the EGC-degrading bacteria, A. equolifaciens MT4s-5 and E. lenta JCM 9979. From these observations, it may be expected that the EGC-degrading bacteria reduce methano- genesis and sulphidogenesis in the presence of catechins. In fact, it has been reported that human fecal sulfide signifi- cantly decreased 21 days after administration of catechin mixture in elderly residents under enteral feeding (Goto et al. 1998).

Furthermore, we investigated whether or not A. equolifa- ciens MT4s-5 was capable of catalyzing 4′-dehydroxylation in the B ring of metabolite 1. Surprisingly, A. equolifaciens MT4s-5 easily produced 2 when 1 was used as the starting substrate, although when EGC was the starting substrate, strain MT4s-5 showed little production of 2. On the other hand, A. equolifaciens MT4s-5 smoothly catalyzed both the C ring cleavage of EGC and subsequent dehydroxylation of 1 to yield 2 in the presence of syntrophic bacteria or hydro- gen. We cannot explain clearly the reason for the above phenomena, but the following is one possible explanation.

A. equolifaciens MT4s-5 catalyzes both the C ring cleavage of EGC and the dehydroxylation of 1 by utilizing the same cofactor, e.g., NADH, NADPH, FADH2, coenzyme F420, which is accumulated in its bacterial cell. However, since the cofactor is consumed in the ring cleavage of the first step when EGC is used as the substrate, the dehydroxylation at the next step is strongly suppressed or ceases due to the lack of the cofactor. On the other hand, the ring-cleaving of EGC and subsequent dehydroxylation progressed smoothly in the presence of syntrophic bacteria or hydrogen because A. equolifaciens MT4s-5 regenerates the consumed cofac- tor by utilizing hydrogen supplied by syntrophic bacteria. When metabolite 1 which already underwent the ring cleav- age is used as the starting material, A. equolifaciens MT4s-5 catalyzes the conversion of 1 into 2 readily. This is because there is a sufficient amount of cofactor remaining in the bacterial cell. The above explanation is applicable to the dehydroxylation of metabolites 3 and 4 by A. equolifaciens MT4s-5. Recently, Thawornkuno et al. (2009) reported that a crude enzyme preparation (culture supernatant) from A. celatus do03 (JCM 14811), which is closely similar (99.9 %) in 16S rRNA gene sequence to A. equolifaciens MT4s-5, catalyzed the reduction of daidzein to dihydrodaid- zein in the presence of NADPH and another crude enzyme preparation (cell debris) caused reductive conversion of dihydrodaidzein into equol by utilizing the same coenzyme. These findings showed that the above two type of enzymes required the same coenzyme (NADPH) in their reaction. Accordingly, it could be that the two enzymatic reactions (C ring cleavage of EGC and subsequent dehydroxylation) requiring the same cofactor in A. equolifaciens MT4s-5 exist. In the case of E. lenta, JCM 9979 catalyzed the ring cleavage of EGC, but not the subsequent dehydroxylation of
1. The dehydroxylation did not proceed at all, even when metabolite 1 was used as the starting material. On the other hand, the dehydroxylation took place only in the presence of either hydrogen or formate supplied by syntrophic bac- teria. An explanation for why there is no dehydroxylation reaction by E. lenta JCM 9979 alone may be that this strain catalyzes the ring cleavage of EGC with a cofactor existing in its bacterial cell, but cannot catalyze the dehydroxylation with the same cofactor. An explanation for the quick pro- gress of the ring cleavage and subsequent dehydroxylation by E. lenta JCM 9979 in the presence of the hydrogen or the formate may be that another cofactor which is required for the dehydroxylation is generated using the above electron donors. This hypothesis is justified by the fact that the dehy- droxylation of metabolites 3 and 4 by E. lenta JCM 9979 takes place only in the presence of hydrogen and/or formate. However, further study on enzymes which catalyze the ring cleavage and dehydroxylation will be required to clarify the above phenomena.

Fig. 7 Proposed biotransformation of EGC and EGCg by intestinal bacteria involved in their catabolism. aQuotation from Takagaki and Nanjo (2010), bquotation from Kutschera et al. (2011), cE. coli MT4s-

On the basis of the findings here, together with previous reports, the major metabolic pathway of EGC and EGCg by catechin-degrading bacteria is proposed as illustrated in Fig. 7. EGCg is hydrolyzed at first to EGC and gallic acid by enteric bacteria which may have tannase activity (Takagaki and Nanjo 2010). Then, EGC undergoes C ring cleavage to produce 1 by A. equolifaciens MT4s-5, E. lenta JCM 9979, Eggerthella sp. SDG-2 (Wang et al. 2001), and probably E. lenta rK3 (Kutschera et al. 2011) and Eggerthella sp. CAT-1 (Jin and Hattori 2012). Eggerthella sp. SDG-2 alone is capa- ble of converting EGC into 2 via 1. A. equolifaciens MT4s-5 and E. lenta JCM 9979 also catalyze the above conversion rapidly in the presence of hydrogen and/or formate produced by syntrophic bacteria such as E. coli strains MT4s-3 and NBRC 3301, B. synergistica MT01, and B. virosa MT12. Subsequently, metabolite 1 formed undergoes the decom- position of its phloroglucinol moiety to yield metabolites 3 and 4 simultaneously by F. plautii strains MT42, ATCC 29863, ATCC 49531, and most probably aK2 (Kutschera et al. 2011). Similarly, metabolite 2 is converted simultane- ously into 5 and 6 by the above F. plautii strains. In addi- tion, metabolites 1, 3, and 4 undergo the p-dehydroxylation of pyrogallol moiety in their structure to form 2, 5, and 6, respectively, by A. equolifaciens MT4s-5 both in the pres- ence and absence of hydrogen, although the dehydroxylation by A. equolifaciens MT4s-5 is facilitated in the presence of the hydrogen. On the other hand, the dehydroxylation of 1, 3, and 4 by E. lenta JCM 9979 takes place only in the pres- ence of the hydrogen and/or formate. Also, A. equolifaciens MT4s-5 possesses the ability to catalyze the 4′-dehydroxyla- tion in the B ring of EGC to produce metabolite 7.

In this study, we demonstrated that EGC, which was produced from hydrolysis of EGCg by enteric bacteria (Sakamoto et al. 2009; Takagaki and Nanjo 2010), was cat- abolized by the combination of A. equolifaciens MT4s-5 or E. lenta JCM 9979 possessing C ring cleavage abilities and
F. plautii strains possessing phloroglucinol moiety-decom- posing ability. In addition, we found that A. equolifaciens MT4s-5 facilitated and E. lenta JCM 9979 caused p-dehy- droxylation of EGC metabolites possessing the pyrogallol moiety by utilizing hydrogen/formate formed by syntrophic bacteria. Thus, this report revealed for the first time that syntrophic bacteria play an important role in the catabolic processes of EGC, especially dehydroxylation processes. The findings explain to some extent the reason why a variety of metabolites are produced from EGC. However,further enzymological approaches will be needed to gain a better understanding of the degradation mechanism of cat- echins by intestinal bacteria.

Acknowledgments We thank Matsumoto K. in our laboratories for analysis of 16S rRNA gene sequences of bacterial isolates. And we acknowledge the assistance of Andrea K. Suzuki (Mitsui Norin Co. Ltd.) in the review of the manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.