INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Bile acids are produced de novo in the liver from cholesterol. The steroid nucleus is conjugated with an amide bond at the carboxyl C-24 position to one of two amino acids, glycine or taurine (1, 20). These conjugates are secreted via the common bile duct into the duodenum (22). Because of their amphipathic nature, the conjugates form spontaneous micelles that trap dietary cholesterol and fats and facilitate their absorption by the intestinal epithelium. The bile acids are then actively transported by a sodium-dependent transporter (27) through the epithelium and into the bloodstream. Their return to the liver completes the enterohepatic cycle (1).

In humans, 130 to 650 mg of bile acids per day elude absorption through the intestinal epithelium (1). With bile acids flowing in such large amounts through the digestive tract, bacterial members of the autochthonous gastrointestinal microbiota have evolved the ability to alter these compounds (1, 20). Isolates of indigenous bacteria of numerous genera and species produce enzymes that alter the side chains on the steroid ring system of bile acids. Such enzymes are involved in transformation reactions such as 7- and 7-dehydroxylation, 6-dehydrogenation, and desulfation (1). Some of the best described enzymes, however, are bile salt hydrolases (BSH), which catalyze hydrolysis of the amino acid taurine or glycine from the C-24 position of conjugated bile salts (for example, choloylglycine hydrolase, EC 3.5.1.24). This reaction, deconjugation, may be required before side chain transformations can occur (2).

BSH activity has been detected in several bacterial genera of the autochthonous gastrointestinal microbiota of animals including mice, rats, humans, chickens, and swine (1, 19). Enzymes with the activity have been purified and characterized from Bacteroides vulgatus VI-31 (9), Bacteroides fragilis subsp. fragilis (21), Bifidobacterium longum BB536 (8), Clostridium perfringens MCV 815 (7), and Lactobacillus sp. strain 100-100 (hereafter referred to as Lactobacillus johnsonii 100-100) (11, 20). Genes encoding the enzymes in Lactobacillus plantarum 80 and in C. perfringens 13 have been cloned and sequenced (3, 4). Except for L. johnsonii 100-100, only one protein has been purified in each case. For L. johnsonii, four proteins with the activity (A, B, C, and D) were purified from intracellular fractions of stationary-phase anaerobic cultures (12, 20). These BSH enzymes exist as heterotrimers composed of two antigenically distinct subunits, designated and . The four possible subunit combinations, ₃, ₂, ₂, and ₃, coincide with the four observed isozymes expressed by the bacterium. The ₃ trimer was determined to be catalytically more active than the ₃ trimer, with V_max values of 17 and 2.4 µmol of cholic acid formed per min per mg of protein, respectively (12).

The BSH activity of L. johnsonii 100-100 differs in another way from that of other reported genera. When conjugated bile salts are added to suspensions of stationary-phase cells, the activity increases as much as three- to fivefold within 20 min (11, 20). This increase is not due to induction of either the enzymatic proteins or the transport proteins. Rather, it is due to synthesis of a soluble extracellular molecule of a relatively small molecular size (12 to 25 kDa) that enhances BSH activity not only in L. johnsonii but also in certain other Lactobacillus isolates. The molecule (called BSH enhancing factor, or EF) is protease and heat resistant, partially partitions into organic solvents, and has not been shown to bind bile acids. Only the sulfhydryl group inhibitor N-ethylmaleimide inhibits its function (13).

Because of the unique features of the BSH system of L. johnsonii, we have a long-term goal of learning how the enzymes in this complex system are regulated. To achieve this end, we have cloned from strain 100-100 and sequenced an approximately 3,000-bp genomic DNA sequence that heterologously expresses BSH activity in Escherichia coli cells. Along with a BSH gene, we are reporting two other predicted open reading frames (ORFs) on the fragment, which we implicate in cellular uptake of conjugated bile salts as part of a possible BSH operon.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains, plasmid vectors, and growth conditions. L. johnsonii 100-100 was maintained frozen at 80°C (11, 20). It was grown anaerobically in MRS broth or agar medium (Becton Dickinson) as previously reported (11, 20). E. coli HB101 (Promega) and E. coli DH5 (Gibco BRL) were used as host cell strains for cloning and sequencing and for conjugated bile salt uptake assays, respectively. A plasmid secretion vector, pINIIIA3 (14), was provided by B. Lampson (Energy Biosystems, The Woodlands, Tex.). The pINIIIA3 replicon was constructed from pBR322 and contains an ampicillin resistance gene and an EcoRI cloning site. The E. coli strains were grown aerobically at 37°C in Luria-Bertani (LB) medium (1% tryptone, 0.5% yeast extract [Difco Laboratories], 0.5% NaCl [Fisher], pH ~7, with 1 M NaOH). Plasmid vectors pUC18 and pUC19 (Gibco BRL) were used for manual and automated fluorescent DNA sequencing. The pSportI (Gibco BRL) cloning vector was used to express the putative transport gene(s) in assays of uptake of bile salts.

BSH assays. BSH activity was detected in L. johnsonii 100-100 by two methods. In the first, the cells were plated on MRS-TDCA (MRS agar medium containing 0.5% taurodeoxycholic acid [Sigma]), and the activity was detected when deoxycholic acid precipitated around the colonies in the medium (5). In the second, the bacterium was grown in MRS medium overnight to stationary phase. [24-¹⁴C]taurocholic acid (New England Nuclear) was added to a 1:10 dilution of the culture in 5 mM sodium acetate buffer (pH 5.0) and incubated for 10 min at 37°C. The reaction was stopped by adding 1 ml of both ethyl acetate and 6 M HCl. [24-¹⁴C]cholic acid released from the conjugate by BSH activity partitioned into the ethyl acetate. The ethyl acetate solution was placed in scintillation vials to which Ecoscint A scintillation cocktail (National Diagnostics) was added. The amount of [¹⁴C]cholic acid produced was estimated with a Beckman LS 7000 liquid scintillation counter (11). The latter assay is referred to below as the isotopic BSH assay. The same assays were used to detect the activity in E. coli strains containing chimeric plasmids. In this case, however, the MRS-TDCA medium contained 100 µg of ampicillin/ml (referred to as MRS-TDCA-AMP) to select for the plasmid chimeras. As previously reported, L. johnsonii expressed BSH activity when it was growing at 37°C on agar medium or in liquid medium incubated anaerobically in an atmosphere enriched with CO₂ (11, 20). E. coli strains containing chimeric plasmids also expressed the activity when they were growing on agar media or in liquid media incubated anaerobically (BBL GasPak system with anaerobic system envelope containing palladium catalyst) at 37°C.

Recombinant DNA methods. Total chromosomal DNA was extracted from cells of L. johnsonii (Wizard Genomic DNA Purification Kit; Promega) and digested with EcoRI (New England Biolabs) according to the manufacturer's instructions. The EcoRI-digested pINIIIA3 vector was treated with calf intestinal alkaline phosphatase (New England Biolabs). The digested genomic fragments were ligated with phage T4 DNA ligase (New England Biolabs) into pINIIIA3 according to standard protocols (Protocols and Applications Guide; Promega). The vector chimeras were transformed into E. coli HB101 by a standard protocol (Promega). The transformants were plated onto LB agar medium containing 100 µg of ampicillin/ml. Colonies growing on the medium were either replicated or picked with sterile pipette tips to MRS-TDCA-AMP plates, which were then incubated anaerobically. Colonies growing on that medium and containing cells expressing BSH activity were detected within 72 h by a white precipitate of deoxycholic acid in the surrounding medium (5). Such colonies were picked, plasmid DNA was extracted (Wizard Plus Midipreps DNA Purification Kit; Promega), and the DNA was electrophoresed in 1% agarose gels with 0.5 µg of ethidium bromide/ml by standard methods (25).

DNA sequencing methods. Nested deletions of cloned L. johnsonii DNA were prepared with exonuclease III (New England Biolabs) by standard procedures (Protocols and Applications Guide; Promega). The entire sequence and the truncated fragments were ligated with T4 DNA ligase into predigested and dephosphorylated pUC18 and pUC19 vectors. The chimeric plasmids were transformed into E. coli DH5. Single-stranded DNA prepared from the chimeras was sequenced in both directions manually by the dideoxy chain termination method, as facilitated by the Sequenase Kit (Promega) and reconfirmed with ABI Prism automated fluorescent sequencing (Molecular Biology Resource Facility, University of Tennessee, Knoxville). -³⁵S-dATP (New England Nuclear), commercial universal forward and reverse primers (Promega), and other synthesized oligonucleotide walking primers (Gibco BRL) were used.

Extension of genomic DNA sequence. Thermal asymmetric interlaced PCR (TAIL-PCR) (10) was used to amplify and extend the genomic DNA sequence 3' of the cloned fragment (see Fig. 1). A nested set of three oligonucleotide primers, 5'-GCTACTCTTCTGGAAGCAAGACTTACTAC-3' (DN1-1), 5'-CTACTGTAATTTTGAAGATGATTTTGAA-3' (DN1-2), and 5'-AAAGACTTATAAACTAGACGATCACAC-3' (DN1-3; Gibco BRL), to the 3' end of the clone were prepared. A degenerate primer, 5'-(G/C)TTG(A/T/G/C)TA(G/C)T(A/T/G/C)CT(A/T/G/C)TGC-3' (AD2), provided by Gary Stacey (University of Tennessee, Knoxville), was also used. Three successive high- and low-stringency PCR amplifications were performed with a genomic template from strain 100-100. Each successive PCR used a different nested sequence-specific primer, the same degenerate primer, and the template from the previous reaction according to guidelines established by Liu and Whittier (10). The DNA product from the third TAIL-PCR was purified (Wizard PCR Preps DNA Purification System; Promega) and sequenced by automated fluorescent sequencing using DN1-3 as the primer. The sequence was reconfirmed by directly amplifying genomic DNA with the DN1-1 primer and another sequence-specific primer, 5'-CTTTACTAAAGTAAATCAAATAGTTAGAGGCTGGA-3' (DN1-4; GIBCO BRL), engineered to the 3' end of the new sequence.

DNA and amino acid sequence analysis. DNA sequence was analyzed for predicted start and stop codons with MacVector. ORFs were translated with MacVector into predicted amino acid sequences. The basic local alignment search tool (BLAST [NIH website]) was used to compare amino acid sequences of putative ORFs with published sequences. Kyte-Doolittle hydropathy plots were obtained from both Genepro, version 5.00, and the Wisconsin Package (Genetics Computer Group).

Preparation of BSH EF. Taurocholic acid solubilized in 1 ml of sodium acetate buffer (pH 5.0) (final culture concentration, 0.4 mM) was added to 20 ml of an MRS culture of L. johnsonii grown anaerobically at 37°C for 20 to 24 h. The cultures were then incubated for 30 min to obtain EF⁺ supernatant solution. For EF supernatant solution, an equal volume of the acetate buffer was incubated with 20 ml of culture. The cells were pelleted at 5,000 × g for 10 min; the supernatant solutions were harvested and stored at 4°C (13).

Isotopic conjugated bile acid uptake and specificity assays. A bile acid uptake assay was adapted from the transport assay of Mallonee and Hylemon (16). E. coli DH5 p2000 and E. coli DH5 pSportI cells were grown overnight to stationary phase (Klett reading, ~220) with shaking at 37°C in LB medium containing 100 µg of ampicillin per ml. The cultures were diluted 50% with LB medium and were incubated for 45 min in order to allow the cells to recover in the fresh medium. The cells were induced with 0.5 mM IPTG (final culture concentration) and incubated for an additional 90 min. Cell densities were normalized with LB medium to a Klett reading of approximately 150. For both control (pSportI) and test (p2000) cultures, 8 ml of cells were harvested by centrifugation at 3,000 × g for 10 min at room temperature. Cell pellets were resuspended and washed in 5 ml of cold 50 mM Tris-HCl (pH 7.5) and centrifuged again at 3,000 × g for 10 min. Washed cell pellets were then resuspended in 600 µl of appropriate cold EF⁺ or EF solutions and placed on ice. Three 200-µl portions of each culture were transferred into 1.5-ml Eppendorf tubes and preincubated in a 37°C water bath for 7 to 8 min. Fifty nanocuries of either [24-¹⁴C]taurocholic acid (0.020 mCi/ml; specific activity, 51 mCi/mmol) (New England Nuclear) or [24-¹⁴C]cholic acid (0.1 mCi/ml; specific activity, 54.5 mCi/mmol) (American Radiolabeled Chemicals, Inc.) was then added to each Eppendorf tube. The tubes were incubated at 37°C for 4 min, after which 1 ml of ice-cold 100 mM LiCl-100 mM potassium phosphate (pH 7.0) was added to each. The samples were immediately centrifuged at 3,000 × g for 5 min. The supernatant solution was quickly removed from the cell pellet, and the 100 mM LiCl-100 mM potassium phosphate wash and pelleting was repeated once again. Each cell pellet was then digested with 1 ml of formamide (Fisher Biotech) in a 65°C water bath for 1 to 2 h. The digested cell solution was placed into a scintillation vial containing 10 ml of scintillation cocktail (Ecoscint A; National Diagnostics) and 5 ml of ethanol. The samples were mixed by inverting the vial several times until it was clear. Radioactivity was quantitated in a Beckman LS 7000 liquid scintillation counter. Cholic acid uptake by DH5 cells harboring either pSportI or p2000 was also measured by an alternative procedure adapted from the work of Thanassi et al. (24). The procedure described above was followed up to and including incubation with [24-¹⁴C]cholic acid. Thereafter, however, each of the 200-µl aliquots was layered on top of 150 µl of silicon oil (105 µl of fluid no. 550 and 45 µl of fluid no. 510) (50 centistokes [cst]; Dow Corning Corp.) in 1.5-ml Eppendorf tubes. The samples were centrifuged at 12,000 × g for 2 min in order to pellet the cells and then frozen in a dry-ice-ethanol bath. The tips of the Eppendorf tubes containing the cell pellets were cut into scintillation vials. The pellets were resuspended in 0.3 ml of water to which 10 ml of scintillation cocktail was added and mixed. The radioactivity was quantitated as described above. The nonparametric Wilcoxon rank sum test was used as a statistical measure to compare the isotopic data.

Nucleotide sequence accession numbers. The sequences of ORF1, ORF2, and ORF3 have been registered with GenBank under accession no. AF008584, U90067, and U90066, respectively.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Identification of Lactobacillus sp. strain 100-100 as L. johnsonii 100-100. Lactobacilli have been found to colonize the keratinized, nonsecreting, squamous epithelium in the stomachs of rodents (19). A rat stomach isolate was choosen for high levels of BSH expression. This isolate had been previously identified as a Lactobacillus sp. and given the strain designation 100-100 (11, 20). It was identified to the species level by Martin Kullen and Todd Klaenhammer (North Carolina State University, Raleigh). Genomic DNA sequencing of the V1 region of its 16S rRNA gene identified it as L. johnsonii. The sequence, approximately 75 bp from the 5' end of the rRNA gene, identically matched the same sequence of GAGCGAG CTTGCCTAGATGATTTTAGTGCTTGCACTAAATGAA ACT in the type strain of L. johnsonii, VPI 7960.

Cloning of the BSH gene. Genomic DNA from L. johnsonii cut with EcoRI, ligated into pINIIIA3, and transformed into E. coli HB101 yielded two ampicillin-resistant colonies in which the cells expressed BSH activity on MRS-TDCA-AMP agar medium. The activity expressed by the cells in both of the colonies was stable upon transfer to fresh medium. Cells from each of the colonies yielded a plasmid chimera containing an approximately 3,000-bp insert (Fig. 1). The chimeras were designated pIN-BSH1 and pIN-BSH2. The insert in pIN-BSH2 proved to be more stable than that in pIN-BSH1 upon repeated transfer on MRS-TDCA-AMP. Therefore, pIN-BSH2 was selected for further study.

View larger version (13K): [in this window] [in a new window]   FIG. 1.   BSH-positive genomic clone pIN-BSH2 with predicted ORFs.

Sequencing and extension of the 2,977-bp insert in pIN-BSH2. Nested deletions in pUC18 and pUC19 of the 2,977-bp insert (Fig. 2) of lactobacillus DNA in pIN-BSH2 were sequenced manually and with an ABI Prism automated fluorescent sequencer. The sequence was predicted by MacVector to contain one complete ORF (designated ORF2) and two partial ORFs (ORF1 and ORF3) of 1,353, 651, and 927 nucleotides, respectively. The predicted ORFs were arranged in a unidirectional, tandem manner with minimal intergenic sequences of 27 and 19 nucleotides (Fig. 1). DNA sequences centered 8 to 13 nucleotides 5' of ORF2 and ORF3 (AAAGAAGGTAA and TAAGGAGGTTT, respectively) closely matched the Shine-Dalgarno 16S rRNA sequence of TAAGGAGGTGA. The partial ORF3 DNA sequence was extended and completed. A DNA product of approximately 550 bp was amplified by TAIL-PCR (Fig. 1). Once it was sequenced, by using DN1-1 as the primer, the amplified fragment extended the known genomic sequence 458 nucleotides 3' of the EcoRI-cloned DNA. A stop codon for ORF3 was predicted by MacVector, resulting in a complete coding sequence of 948 bp. MacVector did not predict any ORFs 3' of the stop codon for ORF3.

View larger version (13K): [in this window] [in a new window]   FIG. 2.   Exonuclease III 5' deletion mutants of the 2,977-bp clone indicate that BSH activity is lost when the deletions disrupt the predicted start site of ORF3. BSH activity was assayed with MRS-TDCA-AMP. BSH+ colonies were identified by precipitates of deoxycholate in the medium (5).

Analysis of ORFs. BLAST analysis of the predicted amino acid sequence of ORF3 revealed numerous proteins with high levels of similarity. The highest levels (approximately 70% similarity to the published consensus sequence [4]) were found for the products of the two BSH genes: the BSH from C. perfringens 13 (4) and the choloylglycine hydrolase gene from L. plantarum 80 (3). In addition, three conserved amino acid motifs, DXXNEXGL, GVXTNXP, and GXGXXGLPGD at residues 72, 174, and 217, respectively, were conserved within the two cloned hydrolases as well as within ORF3. BLAST analysis of ORF2 yielded many sequences with similarity. Those with the highest levels of similarity were several monoamine, acetylcholine, and multidrug resistance transporters of the major facilitator superfamily (MFS). BLAST analysis of the incomplete ORF1 yielded only one similar nucleotide sequence: that of ORF2. The nucleotide sequence of the entire partial ORF1 is 72% identical to the carboxyl half of ORF2 (Fig. 3). A Kyte-Doolittle hydropathy plot of ORF2 predicted 12 transmembrane domains arranged in a 6-plus-6 pattern (Fig. 4). This arrangement is characteristic of the MFS and the ATP-binding cassette (ABC) superfamily of transporters (17, 18). The hydropathy plot for the carboxyl half of ORF1 revealed six transmembrane domains (Fig. 4).

View larger version (49K): [in this window] [in a new window]   FIG. 3.   Predicted amino acid sequences of ORF1 and ORF2 (see Fig. 1). Identical residues are indicated by black highlighting. Grey shading indicates conservative substitutions.

View larger version (27K): [in this window] [in a new window]   FIG. 4.   Hydrophobicity plots of ORF1 and ORF2 (see Fig. 1). The hydrophobicity model of ORF2 predicts 12 transmembrane domains arranged in a 6-plus-6 motif characteristic of transport proteins in the MFS. The carboxyl end of the partial ORF1 indicates a similar structure. The plots shown were prepared with Genepro, version 5.00.

Expression of BSH activity in E. coli HB101/pIN-BSH2 and in E. coli DH5/pUC18 and E. coli DH5/pUC19 nested deletion constructs. E. coli HB101 cells containing pIN-BSH2 but not pINIIIA3 expressed BSH activity not only while growing on MRS-TDCA-AMP plates but also in a radioisotopic assay after being grown in LB-AMP broth (Fig. 5). Cells containing nested deletion constructs, which were made with exonuclease III, ligated into pUC18 and pUC19 for sequencing, and transformed into E. coli DH5, were plated on MRS-TDCA-AMP plates to monitor functional BSH activity. Once the exonuclease deletions compromised the predicted start site of ORF3, as with constructs pIN-BSH22050 through pIN-BSH22895 (Fig. 2), no deoxycholic acid precipitate, as an indicator of BSH activity, was produced in the medium. Therefore, ORF1 and ORF2 were not required for functional BSH activity in E. coli HB101. It should be noted that ORF3 in pIN-BSH2 lacks the 7 carboxyl-terminal amino acid residues of BSH. Thus, these residues are not required for BSH activity. However, all further deletion mutants from the 3' end of the clone resulted in loss of BSH activity.

View larger version (24K): [in this window] [in a new window]   FIG. 5.   BSH activity of E. coli HB101 cells harboring pIN-BSH2 as assayed by liquid scintillation counting of [24-14C]cholic acid released from [24-14C]taurocholic acid. E. coli HB101 cells containing pIN-BSH2 that had been heated to 85°C for 5 min were used as a control.

Identification of ORF1 and ORF2. A hypothesis that ORF2 is involved in the uptake of conjugated bile salts was tested with a construct from which the predicted BSH had been deleted. Cells harboring the clone, p2000, were assayed for the ability to take up [24-¹⁴C]taurocholic acid (Table 1). The uptake assays were performed on stationary-phase cultures grown aerobically at 37°C to a Klett reading of approximately 150. Cultures harboring either pSportI or p2000 and not treated with IPTG or EF took up [24-¹⁴C]taurocholic acid in equivalent amounts. By contrast, the cultures harboring p2000 and induced with IPTG took up significantly more of the bile salt than cultures harboring pSportI similarly induced. Likewise, when not induced with IPTG, cells containing p2000 exposed to EF demonstrated more uptake of the bile salt than cells containing pSportI. The highest levels of uptake were detected when cells containing p2000 were both induced with IPTG and suspended in EF⁺ supernatant solution (Table 1). We conclude from these findings that ORF2 and possibly also ORF1 encode proteins involved in the transport of bile salts.

View this table: [in this window] [in a new window]   TABLE 1.   Uptake of [24-14C]taurocholic acid by E. coli DH5 culturesa containing cloned genomic DNA from L. johnsonii 100-100

Role of EF. EF appeared to function as an enhancer of the transport of bile salts facilitated by ORF2. However, cells containing pSportI, which lacked the putative transporters, exhibited higher levels of uptake of [24-¹⁴C]taurocholic acid when exposed to EF (Table 1). Therefore, EF appears to facilitate the uptake of bile salts nonspecifically, at least in E. coli.

Specificity of uptake of bile salts. EF is produced and BSH activity is increased three- to fivefold within 20 min after 0.4 mM (final culture concentration) conjugated bile salts such as taurocholic acid (but not deconjugated acids) are added to overnight cultures of L. johnsonii 100-100 (11, 13). This finding was confirmed and extended in assays of BSH activity in cells of strain 100-100 exposed to either conjugated or unconjugated bile salts or free amino acids. Only the conjugated bile salts enhanced BSH activity. Both conjugates tested, taurocholic acid and taurochenodeoxycholic acid, increased the BSH activity of strain 100-100 more than threefold (5,234.0 ± 234.2 and 5,145.1 ± 850.6 pg of cholic acid formed/min/ml of culture, respectively) over control values (1,428.0 ± 424.6 pg of cholic acid formed/min/ml of culture) obtained with cultures incubated in medium free of bile salts. None of the unconjugated bile salts and free amino acids tested enhanced the BSH activity of strain 100-100 (data not shown). To test whether this specificity extended to the uptake of bile salts in E. coli DH5 cells containing p2000, we induced such cells with IPTG and exposed them to EF. These cells exhibited two- to threefold-higher levels of uptake of [24-¹⁴C]taurocholic acid than control cells containing pSportI that had been induced with IPTG but not exposed to EF. By contrast, induced cells from the same culture containing p2000 and exposed to EF exhibited lower levels of uptake of the deconjugated [24-¹⁴C]cholic acid than the control cells when uptake was assayed with wash buffer (Table 2). However, when the cells were centrifuged in silicon oil, levels of uptake by cells containing p2000 were similar to those of cells containing pSportI (Table 2).

View this table: [in this window] [in a new window]   TABLE 2.   Specificity of bile salt uptake by E. coli DH5 culturesa containing cloned genomic DNA from L. johnsonii 100-100

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Our evidence supports a hypothesis that the 2,977-bp genomic clone from L. johnsonii 100-100 contains genes encoding two bile acid-related functions that are heterologously expressed in E. coli cells: BSH activity and transport of conjugated bile acids. Nucleotide sequence analysis of the cloned DNA predicted three clustered ORFs. 5' Deletion mutants of the genomic clone lost BSH activity when the predicted start site of ORF3 was compromised. In addition, BLAST analysis revealed that the predicted amino acid sequence of ORF3 had the greatest similarity to the predicted sequences of the two BSH genes previously reported (3, 4). The molecular weight of 34,767 calculated for the protein product from ORF3 more closely approximates the estimated molecular weight of BSH  (38,000) than that of BSH  (42,000) (12). Furthermore, the N-terminal 25-amino-acid sequence of BSH  does not match the predicted N-terminal amino acid sequence of BSH  (12). We tentatively conclude, therefore, that ORF3 encodes peptide . The gene encoding peptide  (12) has yet to be cloned but may lie within close proximity to this group of functionally related genes.

The similarities of the predicted amino acid sequences of the incomplete ORF1 and the complete ORF2 to each other and to transporter proteins, as well as the proximity of these genes to the BSH gene, indicated that they may function as bile acid transporters. The p2000 construct containing the partial ORF1 and the complete ORF2 increased the uptake of conjugated bile salts in E. coli cells, especially when they were supplemented with EF. We conclude from these findings that ORF2, and possibly ORF1, may encode a conjugated bile acid transporter.

In addition to enhancing the uptake of conjugated bile salts in E. coli transformed with p2000, EF also enhanced the uptake of the conjugates in cells transformed with the vector pSportI. EF may act as a general enhancing factor mimicking the binding proteins of ABC transport systems in gram-negative bacteria (18). EF is produced when conjugated but not deconjugated bile salts are present in the growth medium. Since EF also enhances the uptake of conjugated bile salts, we tested whether the putative transporter would transport only such salts. Therefore, the p2000 construct was tested for the ability to take up either taurocholic or cholic acid.

As consistently observed (Table 1), E. coli cells transformed with p2000 and treated with EF took up more conjugated bile salts than cells containing the vector. However, such cells took up no more [24-¹⁴C]cholic acid than cells transformed with pSportI (Table 2). When silicon oil centrifugation was used, the amount of deconjugate taken up was over sixfold higher than the amount of conjugate. The physical properties of deconjugated and conjugated bile salts differ significantly. The deconjugate, cholic acid, has a pK_a value close to 7 and consequently may diffuse readily across a bacterial membrane. When we assayed the uptake of cholic acid without using silicon oil centrifugation and washed the cells twice with 100 mM LiCl-100 mM potassium phosphate, much of the deconjugate may have diffused out from the cells. Recent work by Thanassi et al. indicates that the E. coli genome encodes a deconjugated bile acid exporter (23). The capacity of E. coli to export deconjugated bile acids suggests that these molecules enter the cells passively. In contrast, conjugated bile acids, with their low pK_a values, cannot enter bacterial cells passively and consequently require specific transporters.

Our data from BSH expression and bile acid uptake experiments indicate that the three clustered ORFs of the 2,977-bp clone are functionally related. We suggest that this clone is part of a coordinately regulated operon. The genes encoding two phenotypes have yet to be cloned: the peptide of BSH  and the compound with EF activity. In addition, there may be a gene encoding a deconjugated bile acid exporter. The 7-dehydroxylation (Bai) pathway of a Eubacterium sp. (15, 26) consists of several genes arranged in an operon. That pathway is involved, however, in one of several known bile acid side chain transformations (1), while the putative BSH operon in L. johnsonii is involved in intracellular deconjugation of conjugated bile salts.

The BaiG gene of the Bai pathway was the first bile acid transporter reported in a prokaryote (16). Its sequence similarity to other transport genes and proposed membrane topology suggest that it may be part of the MFS of transport proteins (16). However, the BaiG gene is a deconjugated bile acid transporter, while ORF2 is a transporter of conjugated bile acids. In addition, ORF2 is enhanced in function by EF. ORF2, like BaiG, exhibits homology to multidrug resistance MFS transporters; unlike BaiG, ORF2 also exhibits homology to vesicular monoamine transporters.

Bacterial members of the autochthonous microbiota may have evolved BSH-related functions in order to obtain energy sources under anaerobic conditions (1) and therefore to be able to outcompete allochthonous species in habitats in the digestive tract. Alternatively or in addition, the functions may operate to protect the bacteria from bile acid toxicity. Since high concentrations of bile acids are toxic to bacteria (6), tight regulation of conjugated bile acid transport by the bacterium may exist to protect it from toxic levels of the substrate. Identifying the genes involved and genetic regulation of the BSH capacity in L. johnsonii 100-100 may help explain this function in the ecology of the gut.