Introduction

Some epidemiological studies have demonstrated that the intake of soy foods containing isoflavonoid phytoestrogens may reduce the risk of some hormone-dependent diseases, such as not only postmenopausal symptoms, but also certain cancers (breast, prostate, and colon) and cardiovascular diseases [1–4]. With the publicity given to the benefit of soy isoflavonoids for human health, the consumption of dietary supplements derived from soybeans has increased significantly.

When soy foods are consumed, the soy isoflavonoids (isoflavone glucosides) are metabolized into their aglycones by intestinal bacteria at first (Fig. 1). Then the metabolites are absorbed through the gastrointestinal membrane and are brought to the liver. Finally, they are further metabolized in the liver to yield various conjugated forms [5–7]. We have already reported about the structures of soy isoflavonoid metabolites excreted in human urine together with their estrogenic activities [8, 9]. In the first step, intestinal bacteria play an important role for the metabolism of soy isoflavone glucosides. They are greatly responsible for catabolism and scission of the isoflavonoids [10]. The reductive products of soy isoflavonoids, for example, equol and O-desmethylangolensin (DMA), showed stronger estrogenic activity than soy isoflavone aglycone (daidzein) itself [9, 11]. Therefore, it seems to be very important for elucidation of the soy isoflavonoid metabolism by intestinal bacteria.

Fig. 1
figure 1

Metabolic pathway of soy isoflavonoids by intestinal bacteria

Full size image

Herein, we report the microbial transformation of soy isoflavone glucosides using 29 bacterial strains (14 strains of facultative anaerobe, 13 strains of obligate anaerobe, and 2 strains of aerobe) isolated from the human gastrointestinal tract.

Materials and methods

Bacterial strains

Bifidobacterium adolescentis JCM1275, Bifidobacterium bifidum JCM1254, Bifidobacterium longum JCM1217, Bifidobacterium breve JCM7016, Bifidobacterium pseudocatenulatum JCM1200, Bifidobacterium pseudolongum subsp. pseudolongum JCM1205, Lactobacillus acidophilus JCM1132, Lactobacillus gasseri JCM1131, Lactobacillus reuteri JCM1112, Lactobacillus brevis JCM1059, Lactobacillus paracasei subsp. Paracasei JCM1109, Lactobacillus ermentum JCM1560, Streptococcus bovis JCM7891, Enterococcus faecalis JCM5803, Bacteroides vulgatus JCM5826, Bacteroides distasonis JCM5825, Bacteroides thetaiotaomicron JCM5827, Bacteroides ovatus JCM5824, Bacteroides fragilis JCM11019, Mitsuokella multacida JCM2054, Clostridium celatum JCM1394, Clostridium ramosum JCM1298, Clostridium spiroforme JCM1432, Peptostreptococcus productus JCM1471, Peptostreptococcus hydrogenalis JCM7635, Fusobacterium nucleatum JCM6328, Prevotella veroralis JCM6290, Klebsiella pneumoniae JCM1662, and Escherichia coli JCM5491 were purchased from the Japan Collection of Microorganisms (RIKEN BioResource Center, Ibaraki, Japan).

Culture conditions

Each bacterial strain was maintained in general anaerobic medium (GAM) semisolid. It was inoculated into GAM broth (Nissui Co., Tokyo, Japan), then was continuously cultured at 37°C in the anaerobic chamber. The air was replaced by mixed gas (N2 80%, CO2 10%, H2 10%) together with AnaeroPack Kenki (Mitsubishi Gas Chemical Co. Inc., Tokyo, Japan) for anaerobic conditions. Each bacterial strain (1 ml) precultured for 24 h in GAM broth was inoculated into 10 ml of GAM broth containing 2 mg soy isoflavones (genistin and daidzin). The culture medium was incubated at 37°C for 24 h. The cultured medium was extracted three times with ethyl acetate, then was evaporated to dryness under reduced pressure. The residue was analyzed by HPLC.

HPLC analysis of metabolism

The HPLC system consisted of CCPM-II pump, UV-8020 UV detector (Tosoh, Tokyo, Japan). HPLC conditions were as follows: column: C18 reversed-phase column (Cadenza CD-C18 3 μm, 150 mm × 4.6 mm i.d., Imtakt Co. Inc., Kyoto, Japan), solvent A: H2O with 0.05% trifluoroacetic acid (TFA), solvent B: H2O:CH3CN (4:6) with 0.05% TFA, column temperature: 40°C, flow rate: 1 ml/min. Elution was done by the following process: solvent A: solvent B (7:3, 0 min) → solvent A: solvent B (35:65, 20 min). All of the samples were detected at 254 nm.

Time course experiments of metabolism

Some bacterial strains (0.1 ml) precultured for 24 h in GAM broth were inoculated into 1.1 ml of GAM broth containing 0.1 mg soy isoflavonoid (daidzin, genistin). The culture medium was incubated under anaerobic conditions in a similar manner as above for designated times (0, 3, 6, 12, and 24 h). The growth of bacteria was evaluated by turbidity (absorbance at 595 nm). Each cultured medium was purified over Oasis HLB cartridges (Waters, Milford, MA); the H2O–MeOH (19:1) washing step was followed by 100% MeOH. The MeOH fr. was evaporated (under N2 stream at 40°C), and the residue was taken up by 500 μl of H2O–MeOH (1:9). The sample solution was analyzed by HPLC. HPLC conditions were as follows: column: C18 reversed-phase column (Cadenza CD-C18 3 μm, 150 mm × 3.0 mm i.d., Imtakt), flow rate: 0.75 ml/min. The elution processes for daidzin was [solvent A: solvent B (75:25, 0 min) → solvent A: solvent B (7:93, 10 min)], for genistin was [solvent A: solvent B (70:30, 0 min)  → solvent A: solvent B (0:100, 10 min) → solvent A: solvent B (0:100, 11 min)]. All other HPLC conditions were the same as above.

Results

Six strains of Bifidobacterium genus and Lactobacillus genus along with Enterococcus faecalis and Streptococcus bovis were used as a typical facultative anaerobe, which was called lactic acid bacteria (Table 1). Almost all strains metabolized soy isoflavone glucosides to their aglycones, i.e., from genistin and daidzin to genistein and daidzein, respectively. However, one Bifidobacterium (B. bifidum) and three Lactobacillus (L. brevis, L. ermentum and L. reuteri) strains could not appreciably convert isoflavone glucosides to their aglycones. The ratio of metabolism from glucoside to aglycone was different in each strain of facultative anaerobe. Especially Lactobacillus gasseri showed 100% productive rate from genistin to genistein, although that from daidzin to daidzein was 60%. This strain seems to recognize the structural difference between genistin and daidzin.

Table 1 Microbial metabolism of soy isoflavone glucosides by intestinal bacterial strains
Full size table

Five strains of Bacteroides genus, three strains of Clostridium genus, and two strains of Peptostreptococcus genus along with Fusobacterium nucleatum, Mitsuokella multacida and Prevotella veroralis were used as a typical obligate anaerobes (Table 1). Contrary to the facultative anaerobe, some of obligate anaerobes (B. vulgatus, C. spiroforme, F. nucleatum, and Peptostreptococcus hydrogenalis) did not metabolize soy isoflavone glucosides at all. In addition, M. multacida and Prevotella veroralis could not appreciably convert into aglycones. Contrary to Lactbacillus gasseri of the facultative anaerobes, Bacteroides thetaiotaomicron showed a higher productive rate to daidzein than to genistein. Although the total yield of aglycones was not good, Clostridium celatum produced more daidzein than genistein.

Klebsiella pneumoniae and Escherichia coli were used as typical aerobes. Both the aerobic bacteria hardly metabolized soy isoflavone glucosides.

Bifidobacterium adolescentis, Lactobacillus acidophilus, Bacteroides distasonis, Clostridium ramosum, and Peptostreptococcus productus were used for the time course experiments (Fig. 2) of soy isoflavone metabolism since these strains converted to aglycones in good yield. The order of growth speed was Bifidobacterium adolescentis > Clostridium ramosum >> Peptostreptococcus productus  >>  Bacteroides distasonis > Lactobacillus acidophilus (Fig. 2c), although the saturated time for growth of each bacterial strain was different. In the hydrolysis, Bifidobacterium adolescentis and Clostridium ramosum metabolized both glucosides within 6 h (Figs. 2a, b ). On the other hand, Lactobacillus acidophilus and Bacteroides distasonis metabolized very slowly (within 24 h). Peptostreptococcus productus showed moderate activity. Therefore, it seemed that the bacterial growth speed showed good correlation to the metabolizing speed of both glucosides.

Fig. 2
figure 2

Time course experiments of soy isoflavone metabolism by several bacterial strains. (a) Daidzin, (b) genistin, (c) bacterial growth,* Bifidobacterium adolescentis (open square); Lactobacillus acidophilus (filled triangle); Bacteroides distasonis (open triangle); Clostridium ramosum (filled diamond); Peptostreptococcus productus (open circle). *The growth of bacteria was evaluated by turbidity (absorbance at 595 nm)

Full size image

Discussion

These results indicated that the metabolism from soy isoflavone glucosides to their aglycones was performed by various bacterial strains except for aerobes (Table 1). However, Hur et al. [12] reported that Escherichia coli of aerobes metabolized daidzin and genistin into their aglycones. It might be caused by the difference between strain levels, even if the bacteria had the same scientific name. The difference between facultative anaerobes and obligate anaerobes was not largely observed. Many microbial enzymes were found to possess broad substrate specificity, yet they catalyze reactions with a high degree of regio- and stereospecificity [13]. Some of the bacterial strains (Lactbacillus gasseri, Bacteroides thetaiotaomicron, and Clostridium celatum) showed specific hydrolysis for the glucosidic unit of daidzin rather than that of genistin (Table 1). However, no other reaction, such as reduction or scission, was observed in this experiment. Namely, dihydrodaidzein, equol, O-DMA, dihydrogenistein, 5-hydroxyequol, and 6-hydroxy-O-DMA were not detected in the metabolic studies. Therefore, the metabolic enzyme in these bacterial strains might have the nature of glucosidase only. In the time course experiments of soy isoflavone metabolism (Fig. 2), Bifidobacterium adolescentis of facultative anaerobes and Clostridium ramosum of obligate anaerobes metabolized more rapidly than the other strains, whereas the metabolizing speed of Lactobacillus acidophilus of facultative anaerobes and Bacteroides distasonis of obligate anaerobes was very slow. The order of metabolizing speed would be parallel to that of the growth speed of each bacterial strain(Fig. 2c).

It was known that the Bifidobacterium, Lactobacillus, and Streptococcus genus colonized in the upper gastrointestinal tract, for example, the oral cavity, stomach, and duodenum. Therefore, soy isoflavone glucosides would be metabolized in the early stage after administration. The greater part of soy isoflavone aglycones could be absorbed before further reductive reaction, since the reductive product appeared later than isoflavone aglycone in plasma [14, 15]. Consequently, long-term incubation in consideration of enterohepatic circulation may be needed to obtain the reductive products of soy isoflavone aglycones [16, 17]. Moreover, since the flora constitution of intestinal bacteria was different in each individual [18, 19], combination culture using some bacterial strains [20] might be usuful to obtain the reductive products.