Results

Metabolism of TFDG in SPF Mice and GF Mice
We have identified TF, TF3G, TF3′G, and GA as the major fecal metabolites of TFDG in mice and hypothesized that these compounds are the product of microbial enzymatic activities [11]. To test this hypothesis, fecal samples were collected from SPF and GF mice treated with 200 mg/kg TFDG via oral gavage and analyzed by HPLC coupled with electrochemical detector (ECD) (Figure 2). Compared with the samples collected from control mice, 4 major metabolites (M2−M5) were observed in fecal samples collected from TFDG treated SPF mice (Figures 2A). The four metabolites showed the same retention times as those of the authentic standards of GA, TF, TF3G, and TF3′G (Figure 2A). The structures of the four metabolites were confirmed by LC/MS analysis (data not shown). Whereas, none of those metabolites were detected in fecal samples collected from TFDG treated GF mice (Figure 2B), indicating that those metabolites are indeed the microbial metabolites of TFDG in mice.
Figure 2  HPLC-ECD chromatograms of fecal samples collected from TFDG (treated, 200 mg/kg, oral gavage) or DMSO (control) treated special pathogen free (SPF) mice (A) and germ-free (GF) mice (B).
TFDG: theaflavin 3,3′-digallate.

Metabolism of Theaflavins by Human Microbiota
To investigate the metabolism of theaflavins by human microbiota, TFDG, TF3G and TF3′G were incubated with fecal slurries collected from three healthy subjects. Samples were collected as a function of time (0, 6, 12, 24, 48, and 72 h) and analyzed by HPLC-ECD and LC/MS for characterization of the derived microbial metabolites. Figure 3 shows the representative HPLC chromatograms of TFDG incubated with human fecal slurries. TFDG was degraded progressively with time increasing. GA, TF, TF3G and TF3′G (M2–M5) were identified as the metabolites of TFDG by comparing their retention time and tandem mass data with those of the authentic standards (data not shown). In addition, a new peak (M1, RT: 6.5 min) appeared at the time point of 12 h in all three samples. This new metabolite had a molecular weight of 126 as determined by the mass ion at m/z 125 [M−H]−, which is the same as that of pyrogallol (PG) (m/z 125 in negative mode) (data not shown). Further tandem mass analysis indicated that the mass fragments of M1 was almost identical to those of the authentic PG (Figure 4D), suggesting that M1 is PG. To further confirm this, we incubated GA with human fecal slurries and analyzed those samples using HPLC-ECD (Figures 4A–4C). Our results clearly indicated that PG is the microbial metabolite of GA (Figure 1). Moreover, interindividual differences were observed on the metabolism rate of GA to PG among the three human subjects (Figures 4A−4C). GA was almost completely degraded to PG after 48 h incubation with fecal slurries collected from subject C and very little GA was degraded to PG even after 72 h incubation with fecal slurries collected from subject B (Figures 4A−4C). This phenomenon was also observed in the incubation of TFDG with fecal slurries (Figure 3).
Figure 3  HPLC-ECD chromatograms of microbial metabolites of TFDG after incubation with human fecal bacteria (A–C).
A, B and C represent the three human volunteers, respectively. TFDG: theaflavin 3,3′-digallate.
Figure 4  HPLC-ECD chromatograms of microbial metabolites of GA after incubation with human fecal bacteria (A–C); and MS/MS (negative ion) spectra of M1 and authentic PG (D).
A, B and C represent the three human volunteers, respectively. GA: gallic acid; and PG: pyrogallol. We hypothesized that TFDG can be metabolized to TF3G and TF3′G and then both TF3G and TF3′G can be further degraded to TF by gut microbiota. To test this hypothesis, TF3G and TF3′G were incubated with human fecal slurries for up to 72 h. The samples were analyzed by HPLC-ECD as well as LC/MS. Figures 5 and 6 showed that both TF3G and TF3′G could be metabolized to TF and GA, and GA was further metabolized to PG by gut microbiota.
Figure 5  HPLC-ECD chromatograms of microbial metabolites of TF3G after incubation with human fecal bacteria (A–C).
A, B and C represent the three human volunteers, respectively. TF3G: theaflavin 3-digallate.
Figure 6  HPLC-ECD chromatograms of microbial metabolites of TF3′G after incubation with human fecal bacteria (A–C).
A, B and C represent the three human volunteers, respectively. TF3′G: theaflavin 3′-gallate.

Effects of Specific Bacteria on the Metabolism of TFDG
It has been reported that Lactobacillus plantarum exhibited galloyl-esterase and decarboxylase activities which allowed hydrolysis of the grape seed polyphenols and leads to the formation of gallic acid and pyrogallol, respectively [18]. In addition, different kinds of esterases from Bacillus subtilis have been isolated and demonstrated to hydrolyze various esters [19]–[21]. Therefore, we selected these two bacteria to assess their impact on TFDG metabolism. Lactobacillus plantarum 299v and Bacillus subtilis were incubated with TFDG and samples were collected as a function of time (up to 72 h) and analyzed by HPLC-ECD and LC/MS. Figure 7 shows the representative HPLC chromatograms of TFDG incubated with Lactobacillus plantarum 299v (Figure 7A) and Bacillus subtilis (Figure 7B). TFDG was degraded progressively with time increasing. PG, GA, TF, TF3G and TF3′G (M1–M5) were identified as the metabolites of TFDG by comparing their retention time and tandem mass data with those of the authentic standards (data not shown).
Figure 7  HPLC-ECD chromatograms of microbial metabolites of TFDG after incubation with Lactobacillus plantarum 299v (A) and Bacillus subtilis (B).
TFDG: theaflavin 3,3′-digallate.