Metabolism of Daidzein and Genistein by Gut Bacteria of the Class Coriobacteriia

The intake of isoflavones is presumed to be associated with health benefits in humans, but also potential adverse effects of isoflavones are controversially discussed. Isoflavones can be metabolized by gut bacteria leading to modulation of the bioactivity, such as estrogenic effects. Especially bacterial strains of the Eggerthellaceae, a well-known bacterial family of the human gut microbiota, are able to convert the isoflavone daidzein into equol. In addition, metabolization of genistein is also described for strains of the Eggerthellaceae. The aim of this study was to identify and investigate gut bacterial strains of the family Eggerthellaceae as well as the narrowly related family Coriobacteriaceae which are able to metabolize daidzein and genistein. This study provides a comprehensive, polyphasic approach comprising in silico analysis of the equol gene cluster, detection of genes associated with the daidzein, and genistein metabolism via PCR and fermentation of these isoflavones. The in silico search for protein sequences that are associated with daidzein metabolism identified sequences with high similarity values in already well-known equol-producing strains. Furthermore, protein sequences that are presumed to be associated with daidzein and genistein metabolism were detected in the two type strains ‘Hugonella massiliensis’ and Senegalimassilia faecalis which were not yet described to metabolize these isoflavones. An alignment of these protein sequences showed that the equol gene cluster is highly conserved. In addition, PCR amplification supported the presence of genes associated with daidzein and genistein metabolism. Furthermore, the metabolism of daidzein and genistein was investigated in fermentations of pure bacterial cultures under strictly anaerobic conditions and proofed the metabolism of daidzein and genistein by the strains ‘Hugonella massiliensis’ DSM 101782T and Senegalimassilia faecalis KGMB04484T.


Introduction
Daidzein and genistein, two isoflavones with a very similar molecular structure, are well-known to be present in soy and soy-based food [1]. Daidzein and genistein belong to the phytoestrogens as they are "biologically active phenolic compounds derived from plants" with "structures similar to the principal mammalian estrogen" [2]. Several health benefits associated with the consumption of isoflavones are discussed, such as the alleviation of menopausal symptoms, prevention of osteoporosis, and improvement of cancer prognosis [3]. Besides beneficial effects, potential adverse effects of isoflavone intake are also controversially discussed [4]. These effects are often associated with their estrogenic activity. For isoflavones the endogenous transformation in humans is well known and they can be metabolized by endogenous phase I and phase II enzymes [4]. Besides that, isoflavones can be metabolized by gut bacteria leading to modulation of their bioactivity. High inter-individual variations in this microbial metabolism are observed in humans. For example, only about one-third of the human population can convert daidzein to equol, an important metabolite showing the highest affinity to estrogen receptors (ER) among all known isoflavones [4]. Especially strains of the family Coriobacteriaceae are able to convert food polyphenols like daidzein [5]. However, there is still limited knowledge of bacteria that are involved in this conversion [6]. In the intestine, daidzein can be reduced by bacterial metabolism either to equol via dihydrodaidzein (DH-daidzein) and tetrahydrodaidzein (TH-daidzein) and/or to O-desmethylangolensin (O-DMA) also via DH-daidzein [6,7] (Figure 1). Similar to the metabolization of daidzein, genistein can be converted by bacterial action to 6 -hydroxy-O-desmethylangolensin (6 -OH-O-DMA) [4]. The formation of 5-hydroxyl-equol (5-OH-equol) from genistein is described for the type strains of Adlercreutzia mucosicola, Slackia equolifaciens and Slackia isoflavoniconvertens [8][9][10]. However, to the best of our knowledge, 5-OH-equol has not yet been detected in human biofluids, so evidence for its formation in vivo is still lacking. It is well-known that not every bacterial strain that is able to metabolize e.g., daidzein can conduct each single metabolization step from daidzein to equol, which is probably the end product of the metabolic transformation. Some strains only convert daidzein to intermediate metabolites or are able to further metabolize intermediate metabolites [6]. they can be metabolized by endogenous phase I and phase II enzymes [4]. Besides that, isoflavones can be metabolized by gut bacteria leading to modulation of their bioactivity. High inter-individual variations in this microbial metabolism are observed in humans. For example, only about one-third of the human population can convert daidzein to equol, an important metabolite showing the highest affinity to estrogen receptors (ER) among all known isoflavones [4]. Especially strains of the family Coriobacteriaceae are able to convert food polyphenols like daidzein [5]. However, there is still limited knowledge of bacteria that are involved in this conversion [6]. In the intestine, daidzein can be reduced by bacterial metabolism either to equol via dihydrodaidzein (DH-daidzein) and tetrahydrodaidzein (TH-daidzein) and/or to O-desmethylangolensin (O-DMA) also via DH-daidzein [6,7] (Figure 1). Similar to the metabolization of daidzein, genistein can be converted by bacterial action to 6′-hydroxy-O-desmethylangolensin (6′-OH-O-DMA) [4]. The formation of 5-hydroxyl-equol (5-OH-equol) from genistein is described for the type strains of Adlercreutzia mucosicola, Slackia equolifaciens and Slackia isoflavoniconvertens [8][9][10]. However, to the best of our knowledge, 5-OH-equol has not yet been detected in human biofluids, so evidence for its formation in vivo is still lacking. It is well-known that not every bacterial strain that is able to metabolize e.g., daidzein can conduct each single metabolization step from daidzein to equol, which is probably the end product of the metabolic transformation. Some strains only convert daidzein to intermediate metabolites or are able to further metabolize intermediate metabolites [6]. Since especially strains of the Coriobacteriaceae and Eggerthellaceae are associated with the metabolization of daidzein and genistein, it has to be noted that taxonomic changes within the parent class Coriobacteriia occurred [11,12]. The following Eggerthellaceae strains are described to metabolize daidzein and/or genistein: Adlercreutzia equolifaciens subsp. celatus DSM 18785 T [13], A. equolifaciens subsp. equolifaciens DSM 19450 T [14], Adlercreutzia mucosicola DSM 19490 T [15], Eggerthella sp. YY7918 [16], Slackia equolifaciens DSM 24851 T [17], Slackia isoflavoniconvertens DSM 22006 T [8], Slackia sp. AUH-JLC159 [18], and Slackia sp. NATTS [19]. The metabolization of daidzein to O-DMA is described for Slackia exigua DSM 15923 T [8,14]. The enzymes that are involved in daidzein and genistein metabolism were identified and characterized in five bacterial strains, i.e., Lactococcus garvieae 20-92 [20][21][22], S. isoflavoniconvertens DSM 22006 T [23], Slackia sp. NATTS [24], A. equolifaciens DSM 19450 T [25] and Eggerthella sp. YY7918 [16]. The following three genes are described to be necessary for equol production: I) daidzein reductase (DZNR), II) dihydrodaidzein reductase (DHDR), and III) tetrahydrodaidzein reductase (THDR). A fourth enzyme (dihydrodaidzein racemase, DDRC) is described to increase equol production [22,23]. The  [6,7]. Daidzein (DAI); genistein (GEN); dihydrodaidzein (DH-DAI); dihydrogenistein (DH-GEN), tetrahydrodaidzein (TH-DAI); tetrahydrogenistein (TH-GEN); 5-hydroxyl-equol (5-OH-equol); O-desmethylangolensin (O-DMA); 6 -hydroxy-O-desmethylangolensin (6 -OH-O-DMA); daidzein reductase (DZNR), dihydrodaidzein reductase (DHDR); tetrahydrodaidzein reductase (THDR).
LC-DAD and LC-MS analyses of fermentation samples. Sample clean-up was performed according to [30] with some modifications. Samples were prepared on ice whenever this was possible. Briefly, supernatants from fermentations were thawed on ice and shortly vortexed. A sample volume of 400 µL was extracted three times with each 500 µL extraction solvent (ethyl acetate/isopropanol/1-butanol, 90/5/5, v/v/v). The combined organic layers were dried under a constant stream of nitrogen. Samples were dissolved in 20 µL DSMO followed by the addition of 180 µL dissolvent (0.1% aqueous formic acid/acetonitrile/methanol, 90/5/5, v/v/v). Afterwards, samples were vortexed (5 s) and centrifuged (5 min, 4 • C, 23,100× g). A volume of 180 µL of the supernatant was transferred to a vial and stored at 4 • C until further analysis. Samples were analyzed on a Prominence HPLC system (Shimadzu Europa GmbH, Duisburg, Germany) consisting of a controller (CBM-20A), degasser (DGU-20A3), two pumps (LC-20AD), a column oven (CTO-20AC), and an autosampler (SIL-20AC HT) coupled with an SPD-M20A diode array detector (DAD). Chromatography was carried out on a Cortecs C18 column (3.0 × 150 mm, 2.7 µm particle size; Waters GmbH, Eschborn, Germany) equipped with a pre-column (Cortecs C18, 2.1 × 5 mm, 2.7 µm particle size; Waters GmbH, Eschborn, Germany). Concerning the analysis of daidzein and its metabolites, 0.1% aqueous formic acid and methanol/acetonitrile (1/1, v/v) were used as eluents A and B, respectively. A flow rate of 0.7 mL/min was adjusted and the following gradient elution profile was used: 0.0-0.5 min isocratic with 25.5% Concerning the analysis of genistein and its metabolites, 0.1% aqueous formic acid and acetonitrile were used as eluents A and B, respectively. A flow rate of 0.7 mL/min was adjusted and the following gradient elution profile was used: 0.0-0.5 min isocratic with 23% B, 0.5-7.5 min from 23-30% B, 7.5-8.0 min from 30-99% B, 8.0-10.5 min isocratic with 99% B, 10.5-11 min from 99-23% B, and 11.0-16.0 min isocratic with initial conditions. For both methods, the column oven was set to 40 • C and the injection volume was 10 µL. The DAD recorded data from 200 to 600 nm with a sampling rate of 6.25 Hz, and a trace at 275 nm was used to monitor the analytes. The identity of each analyte was confirmed by the retention time and the UV spectrum. The system was controlled by the software LC solution 1.24 (Shimadzu Europa GmbH, Duisburg, Germany). The analysis method to quantify daidzein, DH-daidzein, (R, S)-equol and O-DMA as well as genistein, DH-genistein, and 6 -OH-O-DMA were validated and the parameters accuracy, precision, recovery, the limit of detection (LOD), limit of quantitation (LOQ), and linearity were determined. The results of the validation are given in Supplemental Data S13. For daidzein and genistein metabolization, S. equolifaciens DSM 24851 T [17] was used as a positive control. In addition, unknown metabolites of genistein were identified by accurate LC-MS (QToF) analysis: Samples were measured using a Triple TOF 5600 mass spectrometer (AB Sciex) linked to a 1290 Infinity LC system (Agilent). The LC-DAD-MS system was controlled by the software Analyst TF (version 1.8, AB Sciex, Darmstadt, Germany). Chromatography was performed as described above. Samples were measured both in the negative and in the positive mode. The DuoSpray source operated in electrospray ionization (ESI) mode using the following source parameters: Curtain gas 45 psi, ion spray voltage −4500 V and +5500 V, respectively, ion source gas-170 psi, ion source gas-260 psi, and ion source gas-2 temperature 650 • C. The declustering potential was adjusted to −100 V and +100 V, respectively. The MS full scans were recorded from m/z 100 to 1000 with an accumulation time of 100 ms and a collision energy voltage of −10 V and +10 V, respectively. The MS/MS spectra (product ion) were recorded from m/z 50 to 1000 in the high sensitivity mode with an accumulation time of 40 ms, a collision energy voltage of −35 V, and +35 V and a collision energy spread of 15 V. Nitrogen was used as collision gas. Data were analyzed using the software PeakView 2.2.0 and FormulaFinder 2.2.0 (AB Sciex, Darmstadt, Germany).

Results and Discussion
In silico approach to identify the equol gene cluster. The annotated draft genome sequence (NZ_QIBZ00000000.1) of the equol and 5-OH-equol-producing strain S. isoflavoniconvertens DSM 22006 T which was sequenced by our group previously [28] was used as a reference for the in silico approach. A comparison of the respective section on contig 17 (NZ_QIBZ01000017) of our annotated draft genome of S. isoflavoniconvertens DSM 22006 T and the sequence of the equol gene cluster (JQ358709) generated by Sanger sequencing [23] showed an identity of 100% of the nucleotide sequences.
The aim of this in silico study was to identify strains of the Eggerthellaceae and Coriobacteriaceae which harbor genes involved in daidzein and genistein metabolism. As a comparison of amino acids sequences rather than nucleotide sequences has tremendous advantages [31], the sequences of proteins of S. isoflavoniconvertens DSM 22006 T involved in daidzein metabolism, namely DZNR (AFV15453), DHDR (AFV15451), THDR (AFV15450), and putative DDRC (AFV15447), were used for BLASTp homology search. This search showed proteins with high similarity values in the following well-known equol-producing strains: A. equolifaciens subsp. equolifaciens DSM 19450 T [14], A. equolifaciens subsp. celatus DSM 18785 T [13,32], A. mucosicola DSM 19490 T [9], S. equolifaciens DSM 24851 T [17], Lactococcus garvieae 20-92 [33], Slackia sp. AUH-JLC159 [18], Slackia sp. NATTS [19], Eggerthella sp. YY7918 [34]. In addition to S. isoflavoniconvertens DSM 22006 T , the occurrence of the equol gene cluster was already described for A. equolifaciens subsp. equolifaciens DSM 19450 T [35], Lactococcus garvieae 20-92 [33], Slackia sp. AUH-JLC159 [18], Slackia sp. NATTS [19], Eggerthella sp. YY7918 [34] and discussed in [6]. Furthermore, this BLASTp search showed protein sequences with high similarity to DZNR, DHDR, THDR, and putative DDRC in the type strains 'Hugonella massiliensis' DSM 101782 T [36] and Senegalimassilia faecalis KGMB04844 T [37]. This suggests that these strains might be capable of metabolizing daidzein and genistein. All identified protein sequences were compared by performing a multiple alignment followed by cluster analysis using the unweighted pair group method with arithmetic mean (UPGMA) (Figure 2). These comparisons showed lower similarity values within DZNR (≥57.8%) and DDRC (≥64.3%) than within DHDR (≥89.0%) and THDR (≥80.7%). Comparable findings of the similarities of DZNR, DDRC, DHDR, and THDR were described by [23]. equol-producing strains: A. equolifaciens subsp. equolifaciens DSM 19450 T [14], A. equolifaciens subsp. celatus DSM 18785 T [13,32], A. mucosicola DSM 19490 T [9], S. equolifaciens DSM 24851 T [17], Lactococcus garvieae 20-92 [33], Slackia sp. AUH-JLC159 [18], Slackia sp. NATTS [19], Eggerthella sp. YY7918 [34]. In addition to S. isoflavoniconvertens DSM 22006 T , the occurrence of the equol gene cluster was already described for A. equolifaciens subsp. equolifaciens DSM 19450 T [35], Lactococcus garvieae 20-92 [33], Slackia sp. AUH-JLC159 [18], Slackia sp. NATTS [19], Eggerthella sp. YY7918 [34] and discussed in [6]. Furthermore, this BLASTp search showed protein sequences with high similarity to DZNR, DHDR, THDR, and putative DDRC in the type strains 'Hugonella massiliensis' DSM 101782 T [36] and Senegalimassilia faecalis KGMB04844 T [37]. This suggests that these strains might be capable of metabolizing daidzein and genistein. All identified protein sequences were compared by performing a multiple alignment followed by cluster analysis using the unweighted pair group method with arithmetic mean (UPGMA) (Figure 2). These comparisons showed lower similarity values within DZNR (≥57.8%) and DDRC (≥64.3%) than within DHDR (≥89.0%) and THDR (≥80.7%). Comparable findings of the similarities of DZNR, DDRC, DHDR, and THDR were described by [23].   sequences of the above-listed strains were aligned and compared (Figure 3). Detailed information about each annotated nucleotide sequence is given in Supplemental Tables S1-S12. The genes of the equol gene cluster, the position and orientation of genes as well as the upstream and downstream genes are conserved. Noticeably, S. equolifaciens DSM 24851 T and Eggerthella sp. YY7918 showed both the same insertion mutation of a NAD kinase (DMP06_RS06160 and EGYY_15680) and an interchange mutation of the genes hydG and EGYY_15670 with hydE and EGYY_15660, respectively. of strains of the Eggerthellaceae and Coriobacteriaceae. Clustering was performed by multiple-alignment and UPGMA (unweighted pair group method with arithmetic mean) in BioNumerics 8.0. Green labeled strains are not yet described in the literature to be associated with daidzein or genistein metabolism, to the best of our knowledge. Alignment and comparison of the equol gene cluster. The annotated genes of the equol gene cluster of S. isoflavoniconvertens DSM 22006 T and the respective homologue sequences of the above-listed strains were aligned and compared (Figure 3). Detailed information about each annotated nucleotide sequence is given in Supplemental Tables S1-S12. The genes of the equol gene cluster, the position and orientation of genes as well as the upstream and downstream genes are conserved. Noticeably, S. equolifaciens DSM 24851 T and Eggerthella sp. YY7918 showed both the same insertion mutation of a NAD kinase (DMP06_RS06160 and EGYY_15680) and an interchange mutation of the genes hydG and EGYY_15670 with hydE and EGYY_15660, respectively.   The complete equol gene cluster was identified both in the already known equolproducing strains as well as in S. faecalis KGMB04844 T and 'H. massiliensis' DSM 101782 T which supported the hypothesis that these strains could be able to metabolize daidzein and genistein.
It is not yet clear if the ability of daidzein and genistein metabolization is family, species, or strain dependent [6]. Interestingly, three type strains of the genus Adlercreutzia, i.e., A. equolifaciens subsp. equolifaciens DSM 19450 T , A. equolifaciens subsp. celatus DSM 18785 T , and A. mucosicola DSM 19490 T harbored the equol gene cluster, whereas the other three type strains of this genus, i.e., A. caecimuris DSM 21839 T , A. muris DSM 29508 T and A. rubneri ResAG-91 T did not. Our results are in good agreement with the study of [35], which described the presence of the complete equol operon in the genomes of A. equolifaciens subsp. equolifaciens DSM 19450 T and A. equolifaciens subsp. celatus DSM 18785 T , and the absence of this operon in the genome of strain ResAG-91 T . The incapacity of strain A. caecimuris DSM 21839 T to metabolize daidzein was already described by [38]. In addition, A. rubneri ResAG-91 T did neither metabolize daidzein nor genistein which was investigated within this study (see below). [35] described that strain IPLA37004, which belongs to the well-known equol-producing species A. equolifaciens, did not produce equol and suggested that this was caused by a deletion in the equol operon. However, results of digital DNA-DNA hybridization using TYGS [39] showed that the similarity of the genome sequence of strain IPLA37004 (GCA_009874275.1) to related type strains of the genus Adlercreutzia is below the 70% threshold level for species delineation (results not shown) and therefore, the non-equol-producing strain IPLA37004 represents a potentially new species of Adlercreutzia. PCR method to detect dzr, ddr, and tdr genes. To test the presence of nucleotide sequences coding for dzr, ddr, and tdr via PCR, primers designed by [29] were used for a total of 29 Eggerthellaceae and Coriobacteriaceae strains. The results are presented in Figure 4. The amplicon sizes of 203 bp (dzr), 205 bp (ddr), and 112 bp (tdr) as proposed by [29] were confirmed in our study. As already described by [29], the primer for dzr amplification did not amplify with the DNA of S. equolifaciens DSM 24851 T as a template. Therefore, an amplicon of dzr could only be detected in the positive control S. isoflavoniconvertens DSM 22006 T . The primers for amplification of ddr (the gene coding for DHDR) led to products of the expected amplicon size in all previously described equol-producing strains:   Clustering was performed based on multiple alignment and unweighted pair group method with arithmetic mean (UPGMA). Results of a PCR with primers amplifying for dzr, ddr and tdr [29] are shown next to the cluster.

Investigations of the metabolism of daidzein and genistein. The same 29
Eggerthellaceae and Coriobacteriaceae strains were screened for their ability to metabolize daidzein and genistein. For quantification of daidzein, genistein, and their derived metabolites, validated extraction-and LC-DAD-methods were used.
The concentrations at the time points 0, 24, 48, and 72 h of daidzein, genistein, and their derived metabolites in fermentation samples inoculated with strains that metabolized at least either daidzein or genistein are shown in Tables 1 and 2, respectively. In the control samples without bacteria but supplemented with isoflavones, the initial concentration of 78.4 μM remained stable during the entire course of the fermentation both for daidzein and genistein, and no metabolites were observed. As expected, the metabolization of daidzein to DH-daidzein was conducted by A. equolifaciens subsp. celatus DSM 18785 T , A. equolifaciens subsp. equolifaciens DSM 19450 T , A. mucosicola DSM 19490 T , S. equolifaciens DSM 24851 T , and S. isoflavoniconvertens DSM 22006 T . In addition, 'H. massiliensis' DSM 101782 T and S. faecalis KGMB 04484 T proved DH-daidzein production. DH-daidzein was detected after a period of 24 h in all fermentation samples inoculated with the abovelisted strains. Furthermore, all strains that metabolized daidzein to DH-daidzein also metabolized DH-daidzein to equol within 24 h except S. faecalis KGMB 04484 T , for which no equol production was observed at no time points. It was unexpected that no equol-production for S. faecalis KGMB 04484 T was observed since this strain possesses the equol gene Clustering was performed based on multiple alignment and unweighted pair group method with arithmetic mean (UPGMA). Results of a PCR with primers amplifying for dzr, ddr and tdr [29] are shown next to the cluster.
The concentrations at the time points 0, 24, 48, and 72 h of daidzein, genistein, and their derived metabolites in fermentation samples inoculated with strains that metabolized at least either daidzein or genistein are shown in Tables 1 and 2, respectively. In the control samples without bacteria but supplemented with isoflavones, the initial concentration of 78.4 µM remained stable during the entire course of the fermentation both for daidzein and genistein, and no metabolites were observed. As expected, the metabolization of daidzein to DH-daidzein was conducted by A. equolifaciens subsp. celatus DSM 18785 T , A. equolifaciens subsp. equolifaciens DSM 19450 T , A. mucosicola DSM 19490 T , S. equolifaciens DSM 24851 T , and S. isoflavoniconvertens DSM 22006 T . In addition, 'H. massiliensis' DSM 101782 T and S. faecalis KGMB 04484 T proved DH-daidzein production. DH-daidzein was detected after a period of 24 h in all fermentation samples inoculated with the abovelisted strains. Furthermore, all strains that metabolized daidzein to DH-daidzein also metabolized DH-daidzein to equol within 24 h except S. faecalis KGMB 04484 T , for which no equol production was observed at no time points. It was unexpected that no equolproduction for S. faecalis KGMB 04484 T was observed since this strain possesses the equol gene cluster. The presence of DH-daidzein in the fermentations inoculated with S. faecalis KGMB 04484 T was also very low. Therefore, it cannot be ruled out that S. faecalis KGMB 04484 T is capable to produce equol under different incubation conditions.   (Tables 1 and 2). There were no hints for artifacts, however, these results cannot be explained and need further investigations.
Under the conditions used in this study, S. exigua DSM 15923 T was the only strain that metabolized daidzein to O-DMA which was already reported by [8,14]. However, it has to be noted that in samples of 'H. massiliensis' DSM 101782 T small peaks around the retention time of O-DMA were detected from 24 h onwards. However, the formation of O-DMA could not be shown with certainty. Even if the detected peaks represent O-DMA, the produced amounts were around the LOD and thus less than 0.1% of the daidzein added at the beginning of the incubations (data not shown). Further analyses using more sensitive methods should be conducted to elucidate the presence of O-DMA in fermentation samples of 'H. massiliensis' DSM 101782 T . In addition, our study showed that S. exigua DSM 15923 T is also capable to metabolize genistein to 6 -OH-O-DMA which was not found by [8]. It has to be noted that under the conditions used in this study, no other investigated strain produced 6 -OH-O-DMA in detectable amounts. Interestingly, neither DH-daidzein nor DH-genistein was detected in fermentation samples inoculated with S. exigua DSM 15923 T . This leads to the assumption that O-DMA and 6 -OH-O-DMA is formed without the production of the intermediate compounds DH-daidzein and DH-genistein, respectively. This result is in line as no gene sequences homolog to DZNR could be found in the publicly available genomes of S. exigua. A recent study using lactic acid bacterial and bifidobacterial strains showed also that genistein was metabolized to 6 -OH-O-DMA without detecting DH-genistein, although daidzein was transformed to O-DMA and TH-daidzein alongside the production of DH-daidzein [40].
Interestingly, three strains metabolized genistein to an unknown metabolite which was present after 24 h of sampling and thereafter: A. equolifaciens subsp. celatus DSM 18785 T , A. mucosicola DSM 19490 T and 'H. massiliensis' DSM 101782 T converted genistein into DH-genistein and an unknown metabolite at a retention time of 3.3 min. In addition, fermentation samples inoculated with A. equolifaciens subsp. equolifaciens DSM 19450 T showed the same unknown metabolite from 48 h on. The UV spectra of this unknown metabolite exhibited a maximum absorption at 275 nm and showed similarities with the UV spectrum of the daidzein metabolite equol with a maximum absorption at 280 nm (data not shown). Thus, this metabolite is very likely to be 5-OH-equol. Calibration curves of the reference compound equol were used to estimate the amount of this metabolite. The production of 5-OH-equol was highest in strain 'H. massiliensis' DSM 101782 T , followed by Conclusively, the unknown metabolite peak in fermentation samples of genistein inoculated with A. mucosicola DSM 19490 T and 'H. massiliensis' DSM 101782 T was putatively identified as 5-OH-equol. Due to the lack of a reference standard for 5-OH-equol, the final confirmation of the identity as well as the quantitation in the samples should be conducted in further studies. The genistein fermentation samples of A. equolifaciens subsp. celatus DSM 18785 T and A. equolifaciens subsp. equolifaciens DSM 19450 T were not measured by LC-MS within this study. Nevertheless, the results of the LC-DAD analysis (retention time and UV spectra) led to the assumption that the unknown peak in fermentation samples of A. equolifaciens subsp. celatus DSM 18785 T and of A. equolifaciens subsp. equolifaciens DSM 19450 T represents also 5-OH-equol. It must be noted that the type strain of S. isoflavoniconvertens DSM 22006 T was described to be capable of 5-OH-equol production in the original strain description [8], although this metabolite could not be detected in genistein fermentation samples inoculated with this strain under comparable conditions in this study, e.g., medium, gas atmosphere, temperature, concentration of isoflavone supplementation.

Conclusions
Amino acid sequences comprising genes coding for daidzein reductase (DZNR), dihydrodaidzein reductase (DHDR), tetrahydrodaidzein reductase (THDR) and dihydrodaidzein racemase (DDRC) were successfully used to search for similar protein sequences within draft genome sequences of Eggerthellaceae and Coriobacteriaceae strains. Homolog genes of the equol gene cluster were detected and aligned in already described equolproducing strains. Furthermore, this cluster was newly detected in strains of the Eggerthellaceae and Coriobacteriaceae that have so far not been associated with equol production. The presence of genes of the equol gene cluster was confirmed via PCR amplification. In addition, the metabolism of daidzein and genistein was investigated using pure cultures of Eggerthellaceae and Coriobacteriaceae strains. In conclusion, this study led to the first description of the human gut bacterial strains 'Hugonella massiliensis' DSM 101782 T and Senegalimassilia faecalis KGMB 04484 T as capable of metabolizing daidzein and genistein.