Contribution of the 7β-hydroxysteroid dehydrogenase from Ruminococcus gnavus N53 to ursodeoxycholic acid formation in the human colon.

Bile acid composition in the colon is determined by bile acid flow in the intestines, the population of bile acid-converting bacteria, and the properties of the responsible bacterial enzymes. Ursodeoxycholic acid (UDCA) is regarded as a chemopreventive beneficial bile acid due to its low hydrophobicity. However, it is a minor constituent of human bile acids. Here, we characterized an UDCA-producing bacterium, N53, isolated from human feces. 16S rDNA sequence analysis identified this isolate as Ruminococcus gnavus, a novel UDCA-producer. The forward reaction that produces UDCA from 7-oxo-lithocholic acid was observed to have a growth-dependent conversion rate of 90-100% after culture in GAM broth containing 1 mM 7-oxo-lithocholic acid, while the reverse reaction was undetectable. The gene encoding 7β-hydroxysteroid dehydrogenase (7β-HSDH), which facilitates the UDCA-producing reaction, was cloned and overexpressed in Escherichia coli. Characterization of the purified 7β-HSDH revealed that the kcat/Km value was about 55-fold higher for the forward reaction than for the reverse reaction, indicating that the enzyme favors the UDCA-producing reaction. As R. gnavus is a common, core bacterium of the human gut microbiota, these results suggest that this bacterium plays a pivotal role in UDCA formation in the colon.

The dynamic balance of bile acid composition in the colon is infl uenced by bile acid fl ow in the intestines, the population of bile acid-converting bacteria, and the properties of the responsible bacterial enzymes (1)(2)(3). However, our knowledge is still far from being able to predict changes in the dynamic balance because we have a limited catalog of bile acid-converting bacteria and little information about the enzymes involved in bile acid conversion. Contribution of the 7 ␤ -hydroxysteroid dehydrogenase from Ruminococcus gnavus N53 to ursodeoxycholic acid formation in the human colon Screening and identifi cation of UDCA-producing bacteria from human feces UDCA-producing bacteria were screened using methods described previously ( 32 ). Fecal samples were provided by a healthy Japanese adult male. Samples were homogenized, diluted, and plated on 1/4 GAM agar medium. Colonies on the plates were picked and cultured in 200 l GAM broth containing either 0.1 mM 7-oxo-LCA or UDCA (both as free acids) in 96-well microtiter plates for 48 h. Detection of bile acid conversion in the culture broths was carried out by TLC. Culture samples showing corresponding spots with either 7-oxo-LCA or UDCA by TLC were further analyzed by GC-MS. Identifi cation of the candidate strain by 16S rDNA sequencing and its biochemical characterization using the API 20A and Rapid ID 32A kits (bioMérieux SA, Marcy-l'Etoile, France) were performed as described previously ( 32 ). of the bacteria involved in bile acid conversion is important for host health.

MATERIALS AND METHODS
Bacterial strains, culture media, and culture conditions UDCA-producing R. gnavus N53 was isolated from human feces and characterized in our laboratory. R. gnavus ATCC 29149 T , a reference strain, and Collinsella aerofaciens ATCC 25986 T , another UDCA-producing bacterium ( 29,30 ), were obtained from the Japan Collection of Microorganisms (JCM; Tsukuba, Ibaraki, Japan). The strains were grown at 37°C in Gifu anaerobic medium broth (GAM broth; Nissui Pharmaceutical Co. Ltd., Tokyo, Japan) under anaerobic conditions (80% N 2 , 10% CO 2 , and 10% H 2 ) in an anaerobic chamber (Coy Laboratory Products Inc., Grass Lake, MI). When necessary, 15 g/l agar was added to solidify the medium. The media were kept in the anaerobic chamber at least for 24 h after autoclaving and before use. Luria-Bertani (LB) medium (plate/broth) and M9 liquid medium containing ampicillin (100 g/ml) were used to culture Escherichia coli JM109 transformants. Strains were cultured aerobically at 37°C unless otherwise specifi ed.

Chemicals
UDCA was purchased from Sigma-Aldrich Corp. (St. Louis, MO). Sodium UDCA and 7-oxo-LCA were purchased from Tokyo Chemical Industry Co. Ltd. (Tokyo, Japan). Sodium 7-oxo-LCA was prepared by neutralizing free 7-oxo-LCA with NaOH using a method described previously ( 31 ). Sodium 7-oxo-LCA was used as the substrate for measuring enzyme activity because it dissolves easily in water, while free 7-oxo-LCA does not. Epimerization reaction from CDCA to UDCA, which is catalyzed by the 7 ␣ -HSDH and 7 ␤ -HSDH enzymes. The 7 ␤ -HSDH from R. gnavus N53 preferentially catalyzes the UDCA-forming reaction (reductive reaction) as compared with the 7-oxo-LCA-forming reaction (oxidative reaction), which is denoted by the thickness of the arrow lines. product using Ligation high Ver.2 (Toyobo Co. Ltd., Osaka, Japan). The plasmids were transformed into E. coli JM109 and were purifi ed from colonies on LB plates containing ampicillin. The insert DNA was sequenced using standard primers. The resulting plasmid was named pQE30-7 ␤ -HSDH and was also used for the expression of 7 ␤ -HSDH for enzyme purifi cation.

Purifi cation of recombinant 7 ␤ -HSDH from R. gnavus N53
E. coli JM109 containing pQE30-7 ␤ -HSDH was precultured overnight in 5 ml LB broth containing ampicillin. The preculture was inoculated into 50 ml M9 medium containing ampicillin at an initial OD 660 of 0.01. 7 ␤ -HSDH gene expression was induced by adding isopropyl ␤ -D -Thiogalactoside at a fi nal concentration of 0.1 mM to exponentially growing cells at an OD 660 of 0.4, at which time the culture temperature was shifted from 37°C to 25°C. After 22 h of culture at 25°C, the cells were collected by centrifugation at 7,000 g for 10 min at 4°C. Cell-free extract for Conversion of 7-oxo-LCA into UDCA by R. gnavus N53 and other bacteria in GAM broth R. gnavus N53 and its type strain ATCC 29149 T , as well as C . aerofaciens ATCC 25986 T , another UDCA-producing bacterium ( 29,30 ), were precultured in 5 ml GAM broth in screw-capped vials at 37°C in an anaerobic chamber until stationary phase. The cultures were then inoculated into 50 ml GAM broth in 200 ml screw-cap bottles supplemented with 1 mM 7-oxo-LCA at an initial optical density at 660 nm (OD 660 ) of 0.01 and cultured under the same conditions. During culture, 5 ml of the culture broth was withdrawn periodically. Growth was measured at OD 660 using a photometer (MiniPhoto518R; TAITEC Corp., Koshigaya, Saitama, Japan), and bile acid concentration was determined by HPLC.

Measurement of 7 ␤ -HSDH activity
Cells were harvested by centrifugation at 7,000 g for 10 min. Pellets (0.2 g wet cells) were resuspended in 1 ml extraction buffer containing 0.1 M Tris, 3 mM EDTA, 1 mM dithiothreitol (DTT; Wako Pure Chemical Industries Ltd., Osaka, Japan), and 10% (v/v) glycerol (adjusted to pH 8.0 with HCl). Cell suspensions were disrupted under cooling conditions by sonication with 1 min pulses, followed by 1 min breaks repeatedly for 15 min using a Bioruptor UCD-200T (COSMO BIO Co. Ltd., Tokyo, Japan). Cell debris was removed by centrifugation (46,000 g , 4°C) for 40 min, and the supernatant was used as the crude enzyme solution. The assay mixture for the UDCA-forming reaction (forward reaction) contained, in a total volume of 1 ml, 10 mM Tris-HCl buffer (pH 6.0), 250 M NADPH, 1 mM sodium 7-oxo-LCA, and 10 l enzyme solution, while that for the 7-oxo-LCA-forming reaction (reverse reaction) contained 10 mM glycine-NaOH buffer (pH 10.0), 1 mM NADP + , 2 mM sodium UDCA, and 10 l enzyme solution. The reactions were carried out at 37°C with addition of the substrate. Reaction mixtures without the substrate served as controls. The 7 ␤ -HSDH activity was determined by the change in NADP(H) concentration by monitoring absorbance at 340 nm with a spectrophotometer (DU800; Beckman Coulter Inc., Brea, CA). Enzyme activity was calculated using a molar extinction coeffi cient of 6.22 mM Ϫ 1 ×cm Ϫ 1 for NADPH. The protein concentration was determined by the Bradford method using the Bio-Rad Protein Assay (Bio-Rad Laboratories, Hercules, CA) with BSA as the standard. Enzyme activity was expressed as micromoles of NADPH oxidized or NADP + reduced per minute per milligram of protein .

Cloning of the gene encoding 7 ␤ -HSDH
Total genomic DNA was extracted from both R. gnavus N53 and ATCC 29149 T using ISOPLANT II (Nippon Gene Co. Ltd., Tokyo, Japan) and used as the template for PCR amplifi cation. The gene putatively encoding 7 ␤ -HSDH (hypothetical protein RUMGNA_02585; accession number, ZP_02041813; GenBank database, <www.ncbi.nlm.nih.gov/Genbank>) was amplifi ed from the genomic DNA of R. gnavus strains by PCR using synthetic primers: forward (5 ′ -ttaa GCATGC ATGA CATTGA GAG AAAAATA-3 ′ ) and reverse (5 ′ -taat CTGCAG TTATTCTTGATAGAAAGATC-3 ′ ). The underlined bases denote the Sph I and Pst I restriction sites, respectively. The PCR conditions were as follows: initial 2 min denaturation at 94°C followed by 30 cycles of amplifi cation at 98°C for 10 s, 50°C for 30 s, and 68°C for 1 min, and then elongation at 68°C for 7 min. The PCR product from N53 was further purifi ed using the MinElute PCR Purifi cation Kit (QIAGEN GmbH, Hilden, Germany). The PCR product and vector pQE30 (ampicillin resistance; QIAGEN) containing an N-terminal six-His tag were double-digested with Sph I and Pst I, and then purifi ed using the MinElute Reaction Cleanup Kit (QIAGEN). The digested pQE30 was dephosphorylated and ligated with the PCR TLC. Consequently, we isolated strain N53, which gave a spot corresponding to UDCA by TLC after culture in GAM broth containing 7-oxo-LCA ( Fig. 2 ). The bile acids extracted from the culture medium were characterized by GC-MS; their retention times (RT) corresponded to 7-oxo-LCA (RT, 12.2 min) and UDCA (RT, 12.6 min), respectively. The m/z values for the fragment ions corresponding to the conversion products (213, 255, 355, 370, and 460) were identical to authentic UDCA. 16S rDNA sequencing showed that the strain N53 was closely related to R. gnavus ATCC 29149 T (99.85% sequence homology). 7 ␤ -HSDH activity has been reported for several intestinal bacteria, such as Collinsella aerofaciens 25986 T ( 29,30 ), Ruminococcus productus b-52 ( 29,30,33 ), Ruminococcus sp. PO1-3 ( 34 ), and Clostridium baratii ( 26 ), but not for R. gnavus . Biochemical tests of strain N53 and the R. gnavus type strain ATCC 29149 T with API 20 and Rapid API ID 32A showed a few differences in sugar assimilation (i.e., saccharose; supplementary Table I) and enzyme activity ( ␣ -fucosidase, leucine arylamidase, pyroglutamic acid arylamidase, and alanine arylamidase; supplementary Table II) between the two strains. From these phylogenetic and biochemical data, strain N53 was identifi ed as R. gnavus . Thus, we identifi ed the R. gnavus N53 isolate as a novel UDCA producer.

Conversion of 7-oxo-LCA to UDCA
To identify the conversion reaction catalyzed by 7 ␤ -HSDH, R. gnavus N53 was cultured in GAM broth supplemented with 7-oxo-LCA ( Fig. 3A ). The type strain ATCC 29149 T ( Fig. 3C ) and the known UDCA-producing bacterium C. aerofaciens ATCC 25986 T ( Fig. 3B ) were also cultured for comparison. The three bacterial strains nearly reached stationary phase 6 h after inoculation. All of these strains were found to convert 7-oxo-LCA to UDCA, and their conversion rates peaked (>90%) during stationary phase, suggesting that the reductive conversion reaction was growth-dependent. In contrast, when R. gnavus N53 was cultured in the presence of UDCA, 7-oxo-LCA was not detected (data not shown), indicating that the oxidative reaction with whole 7 ␤ -HSDH purifi cation was prepared using the method described above. The cell-free extract, after membrane fi ltration (Minisart, 0.20 µm; Sartorius Stedim Biotech GmbH, Goettingen, Germany), was applied to a TALON Single Step Column (Clontech Laboratories Inc., Mountain View, CA) equilibrated with equilibration/ wash buffer (50 mM sodium phosphate buffer at pH 8.0, 300 mM NaCl). Nonabsorbed materials were removed by washing the column with two column bed volumes of equilibration/wash buffer. Weakly bound proteins were removed by washing with two column bed volumes of wash-2 buffer (50 mM sodium phosphate buffer at pH 8.0, 300 mM NaCl, 7.5 mM imidazole). His-tagged 7 ␤ -HSDH protein was eluted with 1 ml elution buffer (50 mM sodium phosphate buffer at pH 8.0, 300 mM NaCl, 150 mM imidazole). The eluate was dialyzed in a dialysis tube (TOR-3K; Nippon Genetics Co. Ltd., Tokyo, Japan) with a molecular weight cutoff

Amino acid sequence analysis
Partial N-terminal amino acid sequencing was performed. The purifi ed enzyme was subjected to SDS-PAGE and electrically transferred to a polyvinylidene difl uoride membrane (Bio-Rad) using transfer buffer containing 10 mM CAPS and 10% methanol (adjusted to pH 11.0 with NaOH). Transferred protein was stained with 0.25% Coomassie Brilliant Blue R-250. The band was excised and sequenced by the Instrumental Analysis Division, Equipment Management Center, Creative Research Institution at Hokkaido University.

Screening and identifi cation of UDCA-producing bacteria from a human fecal sample
As a result of screening for UDCA-producing bacteria from human feces, 1,482 colonies were isolated; these were evaluated for their ability to convert 7-oxo-LCA to UDCA by absonum DSM 599 T was observed (GenBank accession number, AET80684; homology, 40%). Moreover, these proteins contain three amino acids, Ser-143, Tyr-156, and Lys-160, which are putative active site residues that may comprise the catalytic triad of SDRs ( 37 ) ( Fig. 4 ).

Cloning, expression, and identifi cation of the 7 ␤ -HSDH gene
To determine whether RUMGNA_02585 encodes 7 ␤ -HSDH, PCR amplifi cation was conducted using both R. gnavus ATCC 29149 T and R. gnavus N53 genomic DNA as templates with primers designed for RUMGNA_02585 of cells was unable to proceed. Measurement of the 7 ␤ -HSDH activity of the cells cultured for 12 h revealed that R. gnavus N53 showed the highest specifi c activity (0.023 mol· min Ϫ 1 ·mg -1 protein), followed by C. aerofaciens ATCC 25986 T (0.013 mol·min Ϫ 1 ·mg -1 protein), then R. gnavus ATCC 29149 T (0.007 mol·min Ϫ 1 ·mg -1 protein). These results confi rmed that R. gnavus is a novel UDCA producer that has a UDCA productivity comparable to C. aerofaciens.

Prediction of the 7 ␤ -HSDH gene in R. gnavus ATCC 29149 T genome
Several 7 ␤ -HSDH enzymes from intestinal bacteria have been purifi ed and characterized. However, the gene encoding this enzyme has not been sequenced. Recently, the genes encoding NADPH-dependent 7 ␤ -HSDH enzymes were identifi ed and cloned from an intestinal bacterium, C. aerofaciens ATCC 25986 T ( 35 ), and a soil bacterium, Cl. absonum DSM 599 T ( 36 ), by using genome information. HSDHs in general belong to the short-chain dehydrogenase/reductase superfamily (SDR) ( 37 ). From BLAST analysis using both sequence information of the 7 ␤ -HSDH gene from C. aerofaciens and an N-terminal cofactor-binding motif of SDR proteins, an open reading frame (ORF) named RUMGNA_02585 (formerly a hypothetical protein) in R . gnavus ATCC 29149 T , the whole genome of which has been sequenced, was predicted to encode the gene for 7 ␤ -HSDH. The ORF consisted of 789 nucleotides encoding 263 amino acids, including the N-terminal cofactor-binding motif (Gly-X-X-X-Gly-X-Gly) ( 38 ). Amino acid sequence alignment indicated high sequence identity of RUMGNA_02585 with 7 ␤ -HSDH COLAER_02088 from C. aerofaciens ATCC 25986 T (GenBank accession number, ZP_01773061; homology, 84%) and with short-chain dehydrogenases of various substrate specifi cities [e.g., CBL26204 from R. torques L2-14 (GenBank accession number, CBL26204; homology, 76%)], whereas relatively low homology with 7 ␤ -HSDH from Cl.  ) enzyme, and 100 mM MES (pH 6.0) in a fi nal volume of 0.5 ml. The oxidative reaction mixture contained 1 mM sodium UDCA, 1 mM NADP + , 50 l enzyme, and 100 mM glycine-NaOH (pH 10.0). Reaction mixtures without the enzyme served as controls. All reaction mixtures, after incubation at 37°C for 1 h, were analyzed by HPLC as described previously ( 32 ). Solid line, reaction in the presence of 7 ␤ -HSDH; dashed line, control reaction mixture without 7 ␤ -HSDH. Arrows indicate the conversion product of each reaction.

Characterization of 7 ␤ -HSDH from R. gnavus N53
All reactions in this section were carried out using purifi ed 7 ␤ -HSDH under the conditions described in Materials and Methods, unless otherwise stated. To identify cofactor specifi city, enzyme activities for the reduction and oxidation reactions were measured using NADPH/NADP + or NADH/NAD + . The activity observed with NADH/NAD + was less than 0.5% of that assayed using NADPH/NADP + as cofactor in both directions (data not shown). Thus, it was confi rmed that 7 ␤ -HSDH from R. gnavus N53 is an NADP(H)-dependent enzyme.
R. gnavus ATCC 29149 T . The results of DNA sequencing of the PCR products from both R . gnavus ATCC 29149 T and N53 were identical. The PCR product from R. gnavus N53 was cloned into the pQE30 expression vector. The His-tagged expressed protein purifi ed by cobalt affi nity chromatography showed high 7 ␤ -HSDH activity of 25.2 mol·min Ϫ 1 ·mg -1 protein, as assayed in the reductive reaction, with a purifi cation fold of 1,100 based on the activity of crude extract from R. gnavus N53. By SDS-PAGE, the molecular weight of the enzyme was revealed to be 30 kDa ( Fig. 5 ). Edman degradation of the purifi ed enzyme revealed a 20 N-terminal amino acid sequence, which matched the N-terminal amino acid sequence of RUMGNA_02585, according to the database. To identify the bile acid conversion products for the reductive and oxidative reactions using the purifi ed enzyme, the reaction mixtures were subjected to TLC and HPLC. Both the reduction and oxidation reaction products, which correspond  Table 2 in reference 36 for V max and k cat did not include enzyme amount information. Thus, these values are just apparent ones. the two conditions or by the possible existence of a 7 ␤ -HSDH isozyme(s) that can produce UDCA or an unknown metabolic pathway(s) yielding UDCA from 7-oxo-LCA. Under these conditions, it is interesting to note that strain R. gnavus N53 was found to have a unique 7 ␤ -HSDH that primarily catalyzed the reductive reaction (forward reaction) to yield UDCA ( Fig. 6 ).
Functional characterization of purifi ed 7 ␤ -HSDH from R. gnavus N53 showed properties similar to those of 7 ␤ -HSDHs from other bacterial strains. In terms of cofactor requirements, the enzyme showed specifi city toward NADPH/ NADP + but not NADH/NAD + . Namely, 7 ␤ -HSDHs from C. aerofaciens ATCC 25986 T ( 29,30 ), R. productus b-52 ( 29,30,33 ), Ruminococcus sp. PO1-3 ( 34 ), and Cl. baratii ( 26 ) have been shown to be NADP(H)-dependent. The optimum pH range of the enzyme was similar to already known 7 ␤ -HSDHs: an acidic range for the forward reaction producing UDCA and an alkaline range for the reverse reaction producing 7-oxo-LCA ( 33-35 ) ( Fig. 7 ). Determination of the K m value of the 7 ␤ -HSDH from R. gnavus N53 revealed a 25-fold higher affi nity toward 7-oxo-LCA than that toward UDCA, whereas those of 7 ␤ -HSDHs from other bacteria, such as C. aerofaciens ATCC 25986 T and Cl. absonum DSM 599 T , were similar toward both substrates ( Table 1 ). As a result, the k cat / K m value for the UDCA-producing reaction of 7 ␤ -HSDH from R. gnavus N53 was about 55-fold higher than that for the reverse reaction, while the ratios of the k cat / K m values for both reactions in C. aerofaciens ATCC 25986 T ( 35 ) and Cl. absonum DSM 599 T ( 36 ) were in the range 1 to 2 ( Table 1 ). These results indicate that the 7 ␤ -HSDH from R. gnavus N53 preferentially catalyzes the UDCA-producing reaction ( Fig. 1 ), refl ecting the results obtained in growth experiments in which only the UDCA-producing reaction ( Fig. 3A ), but not the reverse reaction, was detected in R. gnavus N53. In contrast, C. aerofaciens ATCC 25986 T ( 30,35 ) and Cl. absonum DSM 599 T (39)(40)(41) showed interconversion between 7-oxo-LCA and UDCA in whole-cell reactions ( Table 2 ). These results indicate that the reaction direction of 7 ␤ -HSDH is strain-specifi c.
Despite the presence of 7 ␣ -and 7 ␤ -HSDH-positive bacteria in the intestinal microbiota, UDCA is a minor bile acid component of human feces (2% of total bile acid pool) as compared with DCA (34%) and LCA (29%) ( 1,12 ). The bile acid composition of the human intestine is determined by multiple factors, including the characteristics of the bioconversion reactions, the abundances of the bacterial taxa involved in the reaction, and the availability of substrates/cofactors in the intestinal environment. Recently, R. gnavus has been identifi ed as a common core member of the human intestinal microbiota ( 28 ). Therefore, our data suggest that R. gnavus contributes to the formation of UDCA in the human intestine.
The authors thank Dr. H. Matsuura (Laboratory of Bio-organic Chemistry, Research Faculty of Agriculture, Hokkaido University) for GC-MS analysis. The authors also thank the Regional Innovation Strategy Support Program of the MEXT (Ministry of Education, Culture, Sports, Science and Technology) of the Japanese government for fi nancial support to A.Y. and S.F.
To identify the optimal reaction temperature, the reductive reaction was conducted at temperatures between 20°C and 55°C. The highest activity was observed at 37°C. Raising or lowering the temperature caused a decrease in enzyme activity. To determine thermal stability, the purifi ed enzyme was incubated at temperatures between 20°C and 60°C for 30 min in 100 mM 2-( N -Morpholino)ethanesulfonic acid (MES) buffer at pH 6.0. The residual enzyme activity measured in the reductive reaction decreased to ‫ف‬ 50% of the original value at 50°C within 30 min, and the enzyme appeared mostly inactivated at 60°C. At a pH range of 5.0-11.5, maximal enzyme activities for the reduction ( Fig. 7A ) and oxidation ( Fig. 7B ) reactions were observed at pH 6.0 and 10.0, respectively. The kinetic parameters of the enzyme in both directions were determined by Lineweaver-Burk plots. The enzyme showed relatively lower K m and higher maximum velocity values for the reductive reaction than for the oxidative reaction at each optimal pH value ( Table 1 ). These results indicate that this enzyme catalyzes reductive reactions more readily than oxidative reactions.