Comparative Analysis of Bile-Salt Degradation in Sphingobium sp. Strain Chol11 and Pseudomonas stutzeri Strain Chol1 Reveals Functional Diversity of Proteobacterial Steroid Degradation Enzymes and Suggests a Novel Pathway for Side Chain Degradation

ABSTRACT

The reaction sequence for aerobic degradation of bile salts by environmental bacteria resembles degradation of other steroid compounds. Recent findings show that bacteria belonging to the Sphingomonadaceae use a pathway variant for bile-salt degradation. This study addresses this so-called Δ^4,6-variant by comparative analysis of unknown degradation steps in Sphingobium sp. strain Chol11 with known reactions found in Pseudomonas stutzeri Chol1. Investigations of strain Chol11 revealed an essential function of the acyl-CoA dehydrogenase (ACAD) Scd4AB for growth with bile salts. Growth of the scd4AB deletion mutant was restored with a metabolite containing a double bond within the side chain which was produced by the Δ²²-ACAD Scd1AB from P. stutzeri Chol1. Expression of scd1AB in the scd4AB deletion mutant fully restored growth with bile salts, while expression of scd4AB only enabled constricted growth in P. stutzeri Chol1 scd1A or scd1B deletion mutants. Strain Chol11 Δscd4A accumulated hydroxylated steroid metabolites which were degraded and activated with coenzyme A by the wild type. Activities of five Rieske type monooxygenases of strain Chol11 were screened by heterologous expression and compared to the B-ring cleaving KshAB_Chol1 from P. stutzeri Chol1. Three of the Chol11 enzymes catalyzed B-ring cleavage of only Δ^4,6-steroids, while KshAB_Chol1 was more versatile. Expression of a fourth KshA homolog, Nov2c228, led to production of metabolites with hydroxylations at an unknown position. These results indicate functional diversity of proteobacterial enzymes for bile-salt degradation and suggest a novel side chain degradation pathway involving an essential ACAD reaction and a steroid hydroxylation step.

IMPORTANCE This study highlights the biochemical diversity of bacterial degradation of steroid compounds in different aspects. First, it further elucidates an unexplored variant in the degradation of bile-salt side chains by sphingomonads, a group of environmental bacteria that is well-known for their broad metabolic capabilities. Moreover, it adds a so far unknown hydroxylation of steroids to the reactions Rieske monooxygenases can catalyze with steroids. Additionally, it analyzes a proteobacterial ketosteroid-9α-hydroxylase and shows that this enzyme is able to catalyze side reactions with nonnative substrates.

INTRODUCTION

While steroid biosynthesis is apparently restricted to very few prokaryotic phyla (1, 2), steroids from eukaryotic organisms can serve as energy and carbon sources for many different environmental bacteria (3, 4). Steroids have a nucleus of three C₆- and one C₅-ring, named A to D, and differ in the number and position of functional groups such as hydroxy groups and the presence of a side chain that may be attached to the C₅ ring D. Bile salts comprise steroidal compounds involved in digestion and signaling of, e.g., vertebrates, and are excreted into the environment in large amounts (5). Mammalian bile salts carry a carboxylic side chain as well as one to three hydroxy groups (e.g., cholate; I in Fig. 1) attached to the steroid nucleus. Bacterial steroid degradation is modular and can be divided into four phases (3, 4, 6, 7): (i) oxidation of the A-ring (Fig. 1A), (ii) side chain degradation (Fig. 1B), (iii) degradation of rings A and B by oxygenation (aerobic) (Fig. 1C) or hydrolysis (anaerobic; not shown [8, 9]), and (iv) hydrolytic degradation of the remaining rings C and D. While the last module is very conserved (4, 10), the other modules can differ depending on the group of bacteria. In particular, the presence of a 7-OH in some bile salts enables an only scarcely elucidated variation, called Δ^4,6-variant, of initial oxidation reactions (phase 1), which is found in many Sphingomonadaceae, which are also proposed to employ a so far unknown side chain degradation mechanism (11–13).

FIG 1.

Overview over the degradation of cholate by bacteria. (A) A-ring oxidation and diversion into the Δ^1,4- and Δ^4,6-variants by either introduction of a second double bond into the A-ring (Δ^1,4-variant, blue) or the elimination of water at C-7 by key enzyme 7α-hydroxysteroid dehydratase (Δ^4,6-variant, green). (B) Side chain degradation by P. stutzeri Chol1 (dark blue) or R. jostii RHA1 (light blue) and known side chain degradation steps of Sphingobium sp. strain Chol11 (green). (C) Section of the degradation of the steroid nucleus via the 9,10-seco pathway as found in P. stutzeri Chol1 (blue) and potential channeling of Δ^4,6-intermediates to common C- and D-ring degradation (green). Dotted lines, elucidated in this study; broken lines, suggested reactions; boldface names, experimentally verified; lightface names, bioinformatically predicted. V, HOCDA (12α-hydroxy-3-oxo-4,6-choldienoic acid); VI, 12β-DHADD (7α,12β-dihydroxy-androsta-1,4-diene-3,17-dione); VII, 7α,9α,12β-trihydroxy-androsta-1,4-diene-3,17-dione (unstable); VIII, THSATD [3,7,12-trihydroxy-9,10-seco-androsta-1,3,5(10)-triene-9,17-dione]; IX, DH-HIP (3′,7-dihydroxy-H-methyl-hexahydro-indanone-propanoate); X, HATD (12-hydroxy-androsta-1,4,6-triene-3,17-dione); XI, DHSATD [3,12β-dihydroxy-9,10-seco-androsta-1,3,5(10),6-tetraene-9,17-dione].

In most model strains, including Rhodococcus jostii RHA1, Comamonas testosteroni, and Pseudomonas stutzeri Chol1, aerobic bile-salt degradation proceeds via the so-called 9,10-seco pathway involving intermediates with a Δ^1,4-3-keto structure of the steroid skeleton, which are formed by oxidation of the steroidal A-ring (blue in Fig. 1A) (3, 14, 15). In this first phase, the hydroxy group at C-3 is oxidized to a keto group, and two double bonds at C-1 and C-4 are introduced into the A-ring, which leads to Δ^1,4-3-ketocholate (IV in Fig. 1A) for the bile salt cholate (I in Fig. 1A) (16). In the second phase, degradation of the C₅ carboxylic side chain proceeds similarly to the β-oxidation of fatty acids, and acetyl-CoA and propionyl-CoA are released consecutively (Fig. 1B) (16–18). The side chain is first activated with CoA by a CoA-ligase such as StdA1 in Pseudomonas putida DOC21 (19), and then a double bond is introduced between C-22 and C-23 by an acyl-CoA-dehydrogenase (ACAD) such as Scd1AB in P. stutzeri Chol1 (20). To this double bond, water is added by a hydratase such as ShyI to gain a β-hydroxy group (17). While acetyl-CoA is released by β-oxidation in actinobacteria such as R. jostii RHA1 (14), aldolytic cleavage similar to β-oxidation can be found in proteobacteria such as P. stutzeri Chol1 (17, 21). For this, acetyl-CoA is released by the aldolase SalI. This results in a shortened side chain with an aldehyde group, which is then oxidized to a carboxyl group by aldehyde dehydrogenase Sad (17). The resulting intermediates, such as 7,12-dihydroxy-3-oxo-pregna-1,4-diene-carboxylate (DHOPDC) (see XII in Fig. 2B), have a C₃ carboxylic side chain (16, 21, 22), which is then released as propionyl-CoA. For this, another cycle of aldolytic cleavage in actinobacteria, as well as proteobacteria, is catalyzed by consecutive CoA activation by StdA2 (19), dehydrogenation by the second ACAD, Scd2AB (20, 22), hydroxylation, and aldolytic cleavage, resulting in C₁₉-steroids called androsta-1,4-diene-3,17-diones (ADD), such as 7,12β-dihydroxy-ADD (12β-DHADD VI in Fig. 1) for cholate (I in Fig. 1) (14, 18, 23). In the third phase of degradation, rings A and B are cleaved, which occurs simultaneously or consecutively to side chain degradation in actinobacteria (24) or proteobacteria, respectively (22). First, the B-ring is cleaved by monooxygenase KshAB, which introduces a hydroxy group at C-9 (Fig. 1C). The resulting 9α-hydroxy-ADD (VII in Fig. 1) spontaneously reacts to the name-giving 9,10-seco intermediates, such as 3,7,12-trihydroxy-9,10-seco-androsta-1,3,5-triene-9,17-dione (THSATD; VIII), driven by the aromatization of the A-ring. Further degradation is achieved by oxygenation and meta-cleavage of the A-ring and hydrolytic cleavage of the A-ring residue, resulting in intermediates consisting of former rings C and D called H-methyl-hexahydro-indanone-propanoates (HIP), such as 3′,7-dihydroxy-HIP (DH-HIP; IX in Fig. 1) for cholate (blue in Fig. 1C) (3, 4). At this stage of degradation, intermediates from differently hydroxylated bile salts are channeled to one common intermediate in P. stutzeri Chol1 (20). Further degradation of HIPs proceeds via β-oxidation and hydrolytic cleavages, resulting in acetyl-CoA, succinyl-CoA, and propionyl-CoA (25, 26).

FIG 2.

Phenotype of Sphingobium sp. strain Chol11 Δscd4A. (A) Growth of strain Chol11 Δscd4A (dotted lines, purple) and WT (solid lines, green) with 1 mM cholate. (B) Growth of strain Chol11 Δscd4A (dotted lines, purple) and WT (solid lines, green) with steroids with shortened C₃-side chain (XII, 7,12-dihydroxy-3-oxo-pregna-1,4-diene-carboxylate, DHOPDC; 1 mM, triangles) or no side chain (12β-DHADD, VI in Fig. 1; 2 mM, circles). (C) Growth of strain Chol11 Δscd4A (dotted lines, purple) and WT (solid lines, green) with 7,12-dihydroxy-3-oxo-chol-1,4,22-triene-oate (DHOCTO, XIII; about 0.2 mM). (D) Complementation of strain Chol11 Δscd4A. Growth of strain Chol11 Δscd4A pBBR1MCS-5::scd4A (solid lines, purple), Δscd4A pBBR1MCS-5 (dotted lines, purple), and WT pBBR1MCS-5::scd4AB (solid lines, green) with 1 mM cholate. (E) Heterologous complementation of strain Chol11 Δscd4A with scd1AB of P. stutzeri Chol1. Growth of strain Chol11 Δscd4A pBBR1MCS-5::scd1AB_Chol1 (solid lines, turquois) and Δscd4A pBBR1MCS-5::scd1A_Chol1 (dotted lines, turquois) with 1 mM cholate. Error bars indicate standard deviations and may not be visible if too small (n = 3).

In contrast to this well-elucidated pathway, some steps of 7-hydroxy bile-salt degradation proceed differently in Sphingobium sp. strain Chol11, which uses the aforementioned Δ^4,6-variant. While the first steps of A-ring oxidation also lead to Δ⁴-3-keto intermediates such as Δ⁴-3-ketocholate (III in Fig. 1) (11, 27), in the next step, the hydroxy group at C-7 is eliminated by 7α-hydroxy steroid dehydratase Hsh2 (green in Fig. 1A) (28). This leads to Δ^4,6-3-keto intermediates such as 12-hydroxy-3-oxo-chol-4,6-dienoate (HOCDA, V) with a double bond in the B-ring. This pathway variation has also been found in other Sphingomonadaceae such as Novosphingobium tardaugens NBRC16725 and Novosphingobium aromaticivorans F199 (13). The transient accumulation of Δ^4,6-derivatives of ADDs such as 12-hydroxy-androsta-1,4,6-triene-3,17-dione (HATD; X in Fig. 1) in the supernatants of cultures of strain Chol11 (11) indicates that side chain degradation is the next step of degradation. The side chain is CoA-activated by CoA-ligase SclA in strain Chol11, and in enzyme assays, introduction of a double bond into the side chain was observed (12). However, the position of this double bond, as well as further side chain degradation, has not been elucidated yet, and strain Chol11 apparently does not harbor homologs of all proteins involved in either cycle of side chain cleavage of P. stutzeri Chol1 or R. jostii RHA1 (12). Differential proteome analyses of strain Chol11 adapted to growth with cholate versus ADDs without side chain together with bioinformatical analyses of several bile-salt-degrading Sphingomonadaceae led to the identification of a potential side chain degradation cluster (see Fig. S1 in the supplemental material) (13). The core set of genes in this cluster consists of SclA (12), two adjacent putative ACADs, a Rieske monooxygenase with similarity to KshA, and an amidase. In strain Chol11, the cluster additionally encodes an amidase that cleaves conjugated bile salts (13), two adjacent thioesterase family proteins, and putative steroid dehydrogenases, as well as transporters.

Further degradation of the steroid skeleton is also unknown, but proteomic and bioinformatical analyses strongly indicated that the degradation of the steroid nucleus also proceeds via the 9,10-seco pathway (12, 13). However, no seco-steroids with Δ^4,6-structure have so far been detected in culture supernatants of strain Chol11 growing with cholate.

The goal of this study was to elucidate bile-salt side chain degradation in Sphingobium sp. strain Chol11 by a functional analysis of the gene cluster predicted in the proteomic study. For this, deletion mutants of several genes in the putative side chain degradation gene cluster were constructed. Additionally, comparative analyses with the model organism P. stutzeri Chol1, in which bile-salt degradation is very well elucidated, were performed, e.g., by heterologous expression of candidate genes in suitable mutants of P. stutzeri Chol1 (17, 21).

RESULTS

Nov2c221/222 is involved in steroid C₅ side chain degradation.

The next plausible step of side chain degradation after CoA-activation would be the introduction of a double bond at C-22 by an ACAD, which has been observed for HOCDA-CoA in enzyme assays (12). In the side chain degradation cluster of Sphingobium sp. strain Chol11, two subunits for an ACAD (Nov2c221 and Nov2c222) are encoded by adjacent genes. This gene synteny is also known for other ACADs involved in steroid metabolism, which are heterotetramers of two ACAD subunits (29). In a reciprocal BLASTp analysis, Nov2c221 and Nov2c222 were indeed annotated as the two subunits of the HIP-ACAD, Scd3A (20) (also called ScdD1 [26]) and Scd3B (ScdD2), with 58% and 40% identity to the homologs of P. stutzeri Chol1, respectively (12). This is also reflected in a phylogenetic tree of steroid-degradation ACADs (Fig. S2), in which Nov2c221 and Nov2c222 cluster with the α- and β-subunits, respectively, of HIP-ACADs. However, due to their location and higher abundance in cholate- versus ADD-grown cells (13), a role in side chain degradation seems to be more likely, although the similarities to the respective subunits of C₅- and C₃-side chain ACADs Scd1AB and Scd2AB were much lower, with only up to 33% (Table 1).

TABLE 1.

Identities of Nov2c221 and Noc2c222 of Sphingobium sp. strain Chol11 to the subunits of different steroid-degradation ACADs of P. stutzeri Chol1^a

Parameter	Substrate and protein
	C5-side chain		C3-side chain		HIP		MOODA
	Scd1A	Scd1B	Scd2A	Scd2B	Scd3A (ScdD1)	Scd3B (ScdD2)	ORF21	ORF22
UniProt ID	K5Z6V1	K5Y9H2	A5HIK4	K5Z6U6	K5Z6U1	K5Y9G2	K5Z6X8	K5Y9J8
Identity (%)
Nov2c221	33	21	31	19	58	19	41	18
Nov2c222	20	33	20	27	19	40	22	28

^aNames in C. testosteroni CNB-2 are given in parentheses if different. MOODA, 4-methyl-5-oxo-octanedioate; linear intermediate in the degradation of the C- and D-ring.

To test the function of Nov2c221 and Nov2c222, an unmarked deletion mutant lacking the gene for the presumptive catalytically active α-subunit, Nov2c221, was constructed. The deletion mutant strain Chol11 Δnov2c221 did not grow with cholate as the only carbon source (Fig. 2A).

Growth with cholate was restored by expression of a plasmid-borne copy of nov2c221 (Fig. 2D), excluding possible downstream effects, especially in this operon-like structure. The mutant strain grew with cholate degradation intermediates with a shortened or without a side chain, namely, DHOPDC (XII in Fig. 2) and 12β-DHADD (VI in Fig. 1) (Fig. 2B). With glucose as the only carbon source, strain Chol11 Δnov2c221 grew very similarly to the wild type (not shown). Cholate transformation experiments with dense cell suspensions showed that the deletion mutant depleted cholate with a strongly decreased rate compared to the wild type (Fig. S3A; complete depletion of cholate within 30 h instead of 4 h). The deletion mutant constantly formed HOCDA (V in Fig. 1; Fig. S3B) within these 30 h, which reached 100× elevated concentrations compared to the maximum concentrations that were transiently accumulated by the wild type. These results showed that the deletion mutant harbored functional A-ring oxidation and water elimination from the B-ring as well as the degradation of the steroid ring, including a shortened side chain, but did not degrade the C₅-side chain.

In the next step, we tried to restore the growth of strain Chol11 Δnov2c221 by providing a substrate that already has a double bond in the side chain at C-22, namely, 7,12-dihydroxy-3-oxo-chol-1,4,22-triene-oate (DHOCTO; XIII in Fig. 2). Both the wild type and the deletion mutant grew with and degraded DHOCTO (Fig. 2C).

Nov2c221/222 and C₅ side chain ACAD Scd1AB can replace each other with different efficiencies in cross-complementation experiments.

As DHOCTO is formed by the ACAD Scd1AB in P. stutzeri Chol1, we tested whether cross-complementation of the individual enzymes of strains Chol1 and Chol11 was possible. For this, deletion mutants of P. stutzeri Chol1 lacking either subunit (20) were heterologously complemented with nov2c221 and nov2c222. P. stutzeri Chol1 strains Δscd1A and Δscd1B carrying plasmid-borne copies of nov2c221 and nov2c222 in combination grew with cholate as the only carbon source (Fig. 3A), while the vector control did not. The lag phase of these cross-complemented strains was strongly, but inconsistently, increased for these strains; lag phases of P. stutzeri Chol1 Δscd1A pBBR1MCS-5::nov2c221-222 varied from 24 h to up to 250 h, and the lag phases of P. stutzeri Chol1 Δscd1B pBBR1MCS-5::nov2c221-222 always were about 250 h (Fig. 3A shows representative examples). In addition, growth rates were reduced and final optical density at 600 nm (OD₆₀₀) values were diminished to 0.4 instead of 0.7 in comparison to the wild type (23). In the culture supernatant of the cross-complemented strains, DHOPDC (XII in Fig. 2) with shortened C₃-side chain and THSATD (VIII in Fig. 1) without side chain were detected as degradation intermediates (shown for P. stutzeri Cho1 Δscd1A pBBR1MCS-5::nov2c221-222 in Fig. 3B). This indicates functional two-step side chain degradation in the cross-complemented strains. Additionally, nov2c221 supposedly encoding the α-subunit was expressed in P. stutzeri Chol1 Δscd1A but did not restore growth on its own (data not shown). The other way around, scd1AB_Chol1 were expressed in strain Chol11 Δnov2c221. The strain carrying the plasmid pBBR1MCS-5::scd1AB_Chol1 grew very similarly to the wild type and, thus, without detectable lag phase with cholate as the only carbon source (Fig. 2E). In contrast to this, scd1A_Chol1 alone did not restore growth (Fig. 2E).

FIG 3.

Heterologous complementation of P. stutzeri Chol1 Δscd1A and Δscd1B with scd4AB. (A) Growth of cross-complementation strains P. stutzeri Chol1 Δscd1A pBBR1MCS-5::scd4AB (turquois circles) and Δscd1B pBBR1MCS-5::scd4AB (turquois triangles), as well as empty vector controls P. stutzeri Chol1 Δscd1A pBBR1MCS-5 (purple circles) and Δscd1B pBBR1MCS-5 (purple triangles) with 1 mM cholate. Error bars indicate standard deviations and may not be visible if too small (n = 3). (B) Transient accumulation of various intermediates in the supernatant of a P. stutzeri Chol1 Δscd1A pBBR1MCS-5::scd4AB culture with cholate at an OD₆₀₀ of about 0.13. The graph shows the MS base peak chromatogram in negative ion mode. Intermediates were identified according to retention time, UV absorbance, and mass.

We also tested whether Nov2c221 and Nov2c222 might be able to take over the function of the second ACAD pair in P. stutzeri Chol1 by heterologous expression in P. stutzeri Chol1 R1, which is a transposon mutant defective in Scd2AB (16). However, expression of nov2c221 and nov2c222 in P. stutzeri Chol1 R1 did not restore growth and degradation of cholate (data not shown).

Together these results strongly indicate that Nov2c221 and Nov2c222 form an ACAD that introduces a double bond in the C₅-side chain of cholate at C-22 and were therefore renamed Scd4A and Scd4B, respectively, for steroid-acyl-CoA-dehydrogenase.

Strain Chol11 Δscd4A transforms bile salts to hydroxylated metabolites with complete side chain that can be degraded by the wild type.

To explore the metabolic capacities of Sphingobium sp. strain Chol11 Δscd4A, the strain was incubated for prolonged time periods with different bile salts and glucose as an additional carbon source. Apart from the trihydroxy bile salt cholate (I in Fig. 1), the dihydroxy bile salts deoxycholate (XVI in Fig. 4A) and chenodeoxycholate (XV), as well as the monohydroxy bile salt lithocholate, (XIV) were used, which can all be completely degraded by the strain Chol11 wild type (WT) (28). After incubation for 2 weeks, strain Chol11 Δscd4A had transformed all bile salts to several metabolites that were not degraded upon further incubation (Fig. 4B and C). Four (P1 to P4) out of five metabolites from the transformation of lithocholate and chenodeoxycholate overlapped (Fig. 4B), as well as those from the transformation of cholate and deoxycholate (P6 to P8; Fig. 4C). In general, the masses of these metabolites indicated that the side chains of the bile salts remained complete and unaltered; none of these metabolites had an absorption shoulder around 210 nm, which is indicative of the additional double bond at C-22 (12, 22). Interestingly, mass spectra indicated that some of the metabolites were hydroxylated.

FIG 4.

Transformation of bile salts by Sphingobium sp. strain Chol11 Δscd4A. (A) Structures of the tested bile salts. (B) Accumulation of metabolites (P1 to P5) generated by biotransformation of 12-deoxy bile salts chenodeoxycholate (purple) and lithocholate (black) by strain Chol11 Δscd4A. (C) Accumulation of metabolites (P6 to P8) generated by biotransformation of 12-hydroxy bile salts cholate (black) and deoxycholate (purple) by strain Chol11 Δscd4A. Both graphs show the MS base peak chromatogram in negative ion mode. Masses are indicated for the respective deprotonated acids. Structure suggestions are based on retention time, UV absorbance, and mass. UV and mass spectra of all metabolites are shown in Fig. S5. (D) Potential pathway for transformation of chenodeoxycholate (XV) and lithocholate (XIV) to products P1 to P5 by strain Chol11 Δscd4A. 3α-HSD, 3α-hydroxysteroid dehydrogenase (e.g., Nov2c6); 5β-Δ⁴-KSTD1, 5β-Δ⁴-ketosteroid dehydrogenase Nov2c19; Hsh2, Nov2c400; Δ¹-KSTD, Δ¹-ketosteroid dehydrogenase. Locus tags of monooxygenases are given in parentheses (e.g., 228 for Nov2c228). Gray, P1 and P3 can be completely degraded by strain Chol11.

To investigate whether these metabolites were true intermediates of bile-salt degradation or dead-end products, filter-sterilized culture supernatants of lithocholate- and cholate-grown cells of strain Chol11 Δscd4A were mixed with fresh medium and used for growth experiments with strain Chol11 WT. While P1, P3, P5, P6, and P8 were completely degraded by strain Chol11 WT, P2 and P7 remained (Fig. S4A and B).

Two of the degradable steroid metabolites can be activated by CoA-ligase SclA.

To obtain further evidence that the degradable metabolites are true intermediates of the metabolic pathway, it was tested whether they could be activated by steroid CoA-ligase SclA from strain Chol11 (12). In assays with cell extract of E. coli MG1655 expressing sclA, two compounds with masses of 1,134.5 Da and 1,136.5 Da and characteristic absorption maxima at about 250 nm were formed when CoA and ATP were present (Fig. 5A). The mass of 1,136.5 Da indicates that this compound is the CoA-ester of P4 (387 Da of P4 + 767.5 Da of CoA – 18 Da H₂O, which is removed for thioester formation) (Fig. 5B), and the mass of 1,134.5 Da accordingly indicates that this compound is the CoA-ester of P3 (385 Da of P3 + 757.5 Da of CoA – 18 Da H₂O) (Fig. 5C). The UV-absorption spectra of these compounds are also indicative of CoA-esters (16). Both CoA-esters were not formed when the cell extract was omitted and when cell extracts of E. coli empty vector controls were used. This indicates that the activation is catalyzed by SclA. Minor formation of these CoA-esters in assays without either CoA or ATP could be due to residual CoA and ATP in the cell extracts.

FIG 5.

(A) CoA-activation of the metabolites P1 to P4 produced from lithocholate by Sphingobium sp. strain Chol11 Δscd4AB by cell extract of E. coli MG1655 pBBR1MCS-5::sclA after 4 h at 30°C. MgCl₂ was added in all assays. Graphs show MS chromatograms in negative ion mode of the CoA-activation assay and controls as indicated. Black and gray, base peak chromatogram; purple, extracted ion chromatogram for m/z = 566.7 in negative mode (dominant m/z value of P3-CoA, [M+H]⁻² = 566.7 Da); green, extracted ion chromatogram for m/z = 567.7 in negative mode (dominant m/z value of P3-CoA, [M+H]⁻² = 567.7 Da). Steroid compounds were assigned to structures by retention time, UV absorbance, and mass. Chromatograms are shown with offset in intensity for better overview. Masses calculated from negative mode MS measurements and absorption maxima are given. Masses are indicated for the respective deprotonated acids. P2 was not detected in these measurements, probably due to insufficient separation with the altered method. (B and C) UV and mass spectra of the CoA-activated P3 (B) and P4 (C) produced in the enzyme assays, with retention times of 11.7 min and 12.1 min, respectively.

The metabolites produced by strain Chol11 Δscd4A are mono- or dihydroxylated and have Δ^1,4-structures or cleaved B-rings.

As it was not possible to produce the unknown metabolites and dead-end products in sufficient amounts in the required purity for nuclear magnetic resonance (NMR) analyses, information on the structure could only be inferred indirectly. Strain Chol11 Δscd4A transformed both lithocholate and chenodeoxycholate to the apparently identical products P1, P3, and P4 (Fig. 4B). According to their absorption maxima at approximately 245 nm, products P1, P3, and P4 had Δ⁴- or Δ^1,4-3-keto structures of the A-ring (Fig. S5); such compounds have previously been detected in supernatants of strain Chol11 (11, 27, 28). According to their masses, P3 (385 Da) and P4 (387 Da; Fig. 4B) have an additional hydroxy group at an unknown position compared to Δ^1,4- and Δ⁴-3-ketolithocholate, respectively. For finding the position and stereochemistry of this hydroxy group, P4 was compared to Δ⁴-3-ketochenodeoxycholate, Δ⁴-3-ketohyodeoxycholate, and Δ⁴-3-ketoursodeoxycholate with hydroxy groups in the 7α-, 6α- or 7β-position, respectively (Fig. S6A). These reference compounds were produced from the parent bile salts with mutants from our strain collection as described in Materials and Methods. Although all substances had similar retention times ranging from 19 to 20 min, the retention time of P4 was most similar to that of Δ⁴-3-ketochenodeoxycholate. As Δ⁴-3-ketochenodeoxycholate is a substrate for Hsh2 (28), we tested if P4 could be transformed by Hsh2. However, no formation of the respective Δ^4,6-3-keto product was observed in enzyme assays with P4 and Hsh2 (Fig. S6B). Thus, the hydroxy group of P4 was apparently in a different position that could not be defined yet. The masses of the other two metabolites, P1 (403 Da) and P2 (401 Da), indicate two hydroxy groups compared to Δ⁴- and Δ^1,4-3-ketolithocholate, respectively. While the absorption spectrum of P1 indicates a Δ⁴-3-keto structure, the absorption spectrum of P2, which is very similar to that of THSATD (VIII in Fig. 1), indicates a 9,10-seco structure that could plausibly be caused by hydroxylation at C-9 (23); this seco-steroid with C₅-side chain was apparently not degraded by strain Chol11 WT.

P5 was not found in cultures of strain Chol11 Δscd4A grown in the presence of lithocholate but was unique to growth in the presence of chenodeoxycholate. The UV spectrum of P5 was indicative of a Δ^4,6-3-keto structure of the steroid skeleton, which agrees with the mutant strain still exhibiting 7α-dehydratase activity. As the mass of P5 indicated an additional hydroxy group, it can most likely be excluded that this hydroxy group is at C-7, but potentially, the hydroxy group is at the same position as in P3 and P4.

Cholate and deoxycholate were also transformed to identical products (Fig. 4C). Assuming that the 12-hydroxy bile salts cholate and deoxycholate are transformed like chenodeoxycholate and lithocholate, P6 (399 Da, 245 nm) and P8 (401 Da, 243 nm) probably are the 12-hydroxy derivatives of P3 and P5, respectively, according to their masses and absorption spectra (Fig. S5). The absorption maximum of P7 at 253 nm is not typical for steroid compounds, and consequently, no structure could be assigned to this compound. However, its mass indicated that it was hydroxylated twice compared to the precursor cholate.

Strain Chol11 lacks candidate genes for further degradation of the side chain as known from strain Chol1.

The next step in steroid side chain degradation during both aldolytic and thiolytic degradation in P. stutzeri Chol1 and R. jostii RHA1 is the addition of water to the double bond to gain a β-hydroxy group (17, 30). The only candidates for this hydratase reaction that could be found in genomic and proteomic analyses (12, 13), were nov2c219 and nov2c220 in the side chain degradation gene cluster (Fig. S1). However, the deletion mutant strain Chol11 Δnov2c219-220 did not differ significantly from the wild type during growth with cholate (Fig. S7A). Cross-complementation with plasmid-encoded nov2c219 and nov2c220 did not restore growth of the deletion mutant P. stutzeri Chol1 Δshy, which lacks the hydratase necessary for C₅ side chain degradation. Furthermore, deletion of the adjacent gene nov2c218, which encodes a short-chain reductase with 31% similarity to the 7α-hydroxysteroid dehydrogenase of E. coli (31), also did not cause a phenotype during growth with cholate (Fig. S7B).

Five potential steroid hydroxylating Rieske monooxygenases and homologs to KshA are encoded in the genome of strain Chol11.

Sphingobium sp. strain Chol11 Δscd4A transformed bile salts to seco-steroids with cleaved B-ring and further hydroxylated bile-salt derivatives, indicating the activity of monooxygenases for 9α-hydroxylation and for hydroxylation at other positions. Therefore, we aimed at elucidating the function of putative steroid monooxygenases of strain Chol11 as a next step. One P450 monooxygenase is encoded near one steroid-degradation gene cluster of strain Chol11 but was not detected in any cells in proteomic analyses (13). Additionally, strain Chol11 encodes five homologs of the oxygenase component KshA of the ring cleaving 9α-monooxygenase KshAB, which belong to the Rieske monooxygenases (Table 2) (13). Interestingly, Nov2c66, Nov2c407, Nov2c430, and Nov2c440 had high similarities of >40% among themselves, while Nov2c228 is more different, with <30% identity to the others (Table 2; Fig. S8). Nov2c407, Nov2c430, and Nov2c440 were produced specifically during degradation of steroids, whereas Nov2c228 was only produced during degradation of side chain bearing steroids (13). For the reductase component KshB, only one putative homolog could be found (Table 2), which had less than 20% identity to KshB from R. jostii RHA1.

TABLE 2.

KshA and KshB homologs from Sphingobium sp. strain Chol11 and P. stutzeri Chol1 compared to those of KshA1 and KshB from R. rhodochrous DSM43269^a

Name	Identity (%) to:			RefSeq ID
	KshA1DSM43269	KshAChol1	Nov2c407
KshA
Nov2c66	26	32	58	WP_097093404.1
Nov2c228	23	27	25	WP_097093548.1
Nov2c407	29	31	100	WP_097092973.1
Nov2c430	28	31	45	WP_097092990.1
Nov2c440	29	31	99	WP_097093001.1
C211_11582/KshAChol1	38	100		WP_008568691.1
	Identity (%) to:
	KshBDSM43269	KshBChol1		RefSeq ID
KshB
Novbp123	18	17		WP_013039106.1
C211_11317/KshBChol1	50	100		WP_008568639.1

^aKshA1 is the KshA homolog from R. rhodochrous DSM43269 involved in degradation of bile salts.

As the multiplicity of KshA homologs would probably necessitate multiple deletions for completely abolishing KshA activity, we conducted a heterologous expression. So far, no KshAB from Proteobacteria has been characterized, so we first aimed at constructing a kshA deletion mutant in P. stutzeri Chol1 as a heterologous expression platform.

P. stutzeri Chol1 has a single KshAB that can functionally be expressed in Escherichia coli.

In P. stutzeri Chol1, only one homologous protein for each KshA and KshB can be found. These are C211_11582 (GenBank RefSeq ID WP_008568691.1) and C211_11317 (WP_008568639.1), with 38% and 50% identity to KshA1 and KshB from Rhodococcus rhodochrous DSM43269, respectively (Table 2; Fig. S8). Both genes are located in a steroid degradation cluster (17, 20, 32).

A deletion mutant of kshA_Chol1 was constructed that showed strongly decreased growth with cholate (I in Fig. 1) (Fig. 6A) and accumulated a single metabolite that was identified as 12β-DHADD (VI) (Fig. 6B) according to its retention time, mass (316 Da), and UV spectrum (maximum at 245 nm). With chenodeoxycholate (XV in Fig. 4), deoxycholate (XVI), and lithocholate (XIV), P. stutzeri Chol1 ΔkshA had the same phenotype and accumulated the respective ADDs derived from these bile salts (Fig. S9). Growth was restored by plasmid-borne kshA_Chol1 (Fig. 7A). In the next step, kshA and kshB were coexpressed in E. coli MG1655. In cell suspensions supplied with androsta-1,4-diene-3,17-dione (ADD; XIX in Fig. S9), E. coli MG1655 pBBR1MCS-5::kshA_Chol1-kshB_Chol1 cleaved the substrate and formed the 9,10-seco-steroid 3-hydroxy-9,10-seco-androsta-1,3,5-triene-9,17-dione (HSATD; XX in Fig. 8) (Fig. 8A). These reactions were not observed in suspensions of the vector control (Fig. 8B), indicating that this is a functional platform for observing the activities of KshAB complexes (16, 17, 22). To further explore the substrate spectrum of KshAB_Chol1, a variety of metabolites with the Δ^1,4-3-keto structure, which had either no or different side chains, were tested in this platform. KshAB_Chol1 also hydroxylated and cleaved all tested Δ^1,4-metabolites without side chain, irrespective of the hydroxylation pattern of the steroid skeleton (Fig. S10).

FIG 6.

Phenotype of P. stutzeri Chol1 ΔkshA. (A) Growth of P. stutzeri Chol1 ΔkshA (dashed line, purple) and WT (solid line, blue) with 1 mM cholate. Error bars indicate standard deviations and may not be visible if too small (n = 3). (B) Accumulation of 12β-DHADD (VI in Fig. 1) as a dead-end metabolite in the supernatant of P. stutzeri Chol1 ΔkshA grown with cholate for about 30 h. The graph shows the MS base peak chromatogram in positive ion mode.

FIG 7.

Heterologous complementation of P. stutzeri Chol1 ΔkshA. (A) Maximal OD₆₀₀ reached by P. stutzeri Chol1 WT and ΔkshA complemented with the given genes on vector pBBR1MCS-5 when grown with 1 mM cholate. (B) Transformation of HOCDA (V in Fig. 1) to DHSATD (XI) by P. stutzeri Chol1 ΔkshA expressing different genes on plasmid pBBR1MCS-5. Graphs show MS base peak chromatograms in negative mode of the supernatants of cell suspensions of P. stutzeri Chol1 pBBR1MCS-5::nov2c430 (green), pBBR1MCS-5::kshA (blue), and ΔkshA pBBR1MCS-5 (empty vector control, black) supplemented with HOCDA and incubated for 6 days. (C) UV and mass spectra of DHSATD produced by P. stutzeri Chol1 ΔkshA pBBR1MCS-5::nov2c430.

FIG 8.

Transformation of steroid compounds by E. coli MG1655 expressing kshAB_Chol1. (A) Transformation of ADD (XIX) to the respective 9,10-seco-steroid 3-hydroxy-9,10-seco-androsta-1,3,5(10)-triene-9,17-dione (HSATD, XX) by E. coli MG1655 pBBR1MCS-5::kshAB_Chol1. (B) No transformation of ADD by empty vector control E. coli MG1655 pBBR1MCS-5. (C) Transformation of HATD (X in Fig. 1) and 12β-DHADD (VI) by E. coli MG1655 pBBR1MCS-5::kshAB_Chol1 to the seco-steroids THSATD (VIII) and DHSATD (XI) and the side product 1,2,12-trihydroxy-androsta-4,6-diene-3,17-dione (THADD, XXIII in Fig. S13). The figures show 3D UV chromatograms of supernatants of cell suspensions of the respective E. coli strain incubated for 4 days with ADD. Red indicates highest absorption. Steroid compounds were identified by retention time, UV absorbance, and mass.

KshAB_Chol1 is active with a metabolite from the Δ^4,6-variant and produces a dihydroxylated dead-end metabolite.

P. stutzeri Chol1 has already been shown to transform HOCDA (V in Fig. 1) to HATD (X), which is further converted to the 9,10-seco steroid DHSATD (XI in Fig. 1), indicating functional 9α-hydroxylation with substrates from the Δ^4,6-variant (11). To investigate whether this reaction is also catalyzed by KshAB in P. stutzeri Chol1, we supplied the kshA deletion mutant with HOCDA. P. stutzeri Chol1 ΔkshA with the empty vector pBBR1MCS-5 transformed HOCDA only to HATD (X in Fig. 1) (Fig. S11B), whereas the complemented mutant expressing kshA_Chol1 like the wild-type P. stutzeri Chol1 (Fig. S11A and C) formed DHSATD, which cannot be further degraded by P. stutzeri Chol1 (11). Some THSATD (VIII in Fig. 1) produced by P. stutzeri Chol1 WT and the mutant complemented with kshAB_Chol1, as well as 12β-DHADD (VI) produced by the empty vector control, were likely derived from a Δ⁴-3-ketocholate contamination in the HOCDA stock solution.

In the next step, we supplied HATD to the E. coli expression platform. E. coli pBBR1MCS-5::kshAB_Chol1 transformed HATD (X in Fig. 1) into DHSATD (XI in Fig. 1) (Fig. 8C, in contrast to a control in Fig. S12). Some THSATD (VIII in Fig. 1) found in these supernatants is most likely derived from a contamination of the HATD stock solution with 12β-DHADD (VI).

During transformation of HOCDA by P. stutzeri Chol1, we previously observed the formation of a dihydroxylated metabolite, 1,2,12-trihydroxy-androsta-4,6-diene-3,17-dione (THADD; XXIII in Fig. S11A and C, Fig. S13) (11), which was formed as a further dead-end metabolite in addition to DHSATD. In our transformation experiment with the kshA_Chol1 deletion mutant and HOCDA as substrates, no THADD was formed (Fig. S11B). When HATD was supplied to E. coli pBBR1MCS-5::kshAB_Chol1 the formation of some THADD was shown according to its retention time, mass (332 Da), and UV spectrum (maximum at 290 nm) (Fig. 8C; Fig. S13). These experiments showed that KshAB_Chol1 is not only able to catalyze 9α-hydroxylation of a substrate from the Δ^4,6-variant but can also catalyze a side reaction on the A-ring.

Three out of five KshA homologs from Sphingobium sp. strain Chol11 have B-ring cleaving activity in P. stutzeri Chol1 with HOCDA as the substrate.

After KshA_Chol1 was characterized, the respective homologues from strain Chol11 were studied. First, all five kshA homologs from Sphingobium sp. strain Chol11 were expressed in P. stutzeri Chol1 ΔkshA to check for cross-complementation. However, none of them was able to replace natural kshA_Chol1 during growth with cholate (Fig. 7A), and all cross-complemented strains accumulated 12β-DHADD as a single product from cholate.

As we proposed that Δ^4,6-intermediates are substrates for B-ring cleavage in strain Chol11, we next supplied the Δ^4,6 intermediate HOCDA (V in Fig. 1) as an alternative substrate (Fig. S11; Table 3). Strains expressing nov2c407, nov2c430, or nov2c440 produced DHSATD (XI in Fig. 1) (shown for nov2c430 in Fig. 7B and C and Fig. S11F; shown for nov2c407 and nov2c440 in Fig. S11G to J). This conversion was much less efficient than with the innate KshA homolog of P. stutzeri Chol1 (see above; Fig. S11C). Nevertheless, DHSATD formation clearly indicated steroid 9α-monooxygenase activity of Nov2c407, Nov2c430, and Nov2c440 in P. stutzeri Chol1 leading to B-ring cleavage.

TABLE 3.

Metabolites produced by transformation of HOCDA (V in Fig. 1) by P. stutzeri Chol1 ΔkshA complemented with the given genes on vector pBBR1MCS-5 after 6 days as determined by HPLC-MS and identified by retention time, mass, and UV-spectrum^a

Gene in strain Chol1 ΔkshA	Products formed from HOCDA
kshAChol1	DHSATD (XI in Fig. 1), THADD (XXIII)
nov2c066	HATD (X), 12β-DHADD (VI)
nov2c228	HATD, 12β-DHADD, other (Fig. 9)
nov2c407	HATD, 12β-DHADD, DHSATD
nov2c430	HATD, 12β-DHADD, DHSATD
nov2c440	HATD, 12β-DHADD, DHSATD
Empty vector	HATD, 12β-DHADD

^aTHADD, 1,2,12-trihydroxy-androsta-4,6-diene-3,12-dione (XXIII in Fig. S13).

These DHSATD-forming kshA homologs from strain Chol11 were also expressed in E. coli, either in combination with KshB_Chol1 or with the putative KshB homolog Novbp123 from strain Chol11. For providing a diverse set of substrates, cell suspensions were incubated with several different steroid compounds (Fig. S10). Neither hydroxylation nor ring cleavage of any of the substrates, including the presumptive physiological substrate HATD were observed for any putative KshA_Chol11 with either KshB homolog by high-pressure liquid chromatography-mass spectrometry (HPLC-MS) measurements (data not shown). Notably, KshA_Chol1 was only active with KshB from strain Chol1 and not with Novbp123 (Fig. S10).

Rieske monooxygenase Nov2c228 has steroid-hydroxylating activity in P. stutzeri Chol1.

P. stutzeri Chol1 ΔkshA with Nov2c228 did not catalyze B-ring cleavage but transformed HOCDA to four products (P9 to P11; XXI in Fig. 1), which were not produced by the empty vector control or the wild type (Fig. 9; Fig. S14). Interestingly, two of the products (P9 and P10) were assigned to structures with an additional hydroxylation at an unknown position, potentially similar to the mono-hydroxylated dead-end products of strain Chol11 Δscd4A. P10 has a Δ^1,4,6-3-keto structure as found in HATD (X in Fig. 1) according to its absorption spectrum, and its mass indicated a C₃-side chain and an additional hydroxy group at a so far unknown position. In contrast, P9 has a Δ^1,4-3-keto structure, a C₃-side chain, and an additional hydroxy group according to its absorption spectrum and mass and therefore may be derived from a Δ⁴-3-ketocholate contamination of the HOCDA stock solution. Two more products did not have additional hydroxylations but probably were 12-hydroxy-3-oxo-pregna-1,4,6-triene-carboxylate (HOPTC; XXI in Fig. 9) with a C₃ carboxylic side chain, and Δ^1,4,6-3,12-diketocholate according to their masses and absorption spectra. In the E. coli system, Nov2c228 did not show activity toward any of the tested substrates with either KshB homolog.

FIG 9.

Transformation of HOCDA (V in Fig. 1) by P. stutzeri Chol1 ΔkshA pBBR1MCS-5::nov2c228 (green) and empty vector control P. stutzeri Chol1 ΔkshA pBBR1MCS-5 (black). The graph shows MS base peak chromatograms in negative ion mode of supernatants of cell suspensions of the respective strain incubated for 6 days with HOCDA. Identification of steroid compounds and structure suggestions are based on retention time, UV absorbance, and mass. Masses calculated from negative mode MS measurements and absorption maxima are given. Masses are indicated for the respective deprotonated acids. UV and mass spectra of all steroidal metabolites are shown in Fig. S14. XXI, 12-hydroxy-3-oxo-pregna-1,4,6-triene-carboxylate (HOPTC). The substances eluting at 21 min, 21.5 min, and 24.5 min cannot be assigned to any structures and are probably not steroidal compounds according to their UV and MS spectra.

P. stutzeri Chol1 ΔkshA expressing the fifth KshA homolog, Nov2c66, did not produce any hydroxylated metabolites (Fig. S11D), and no activity of Nov2c66 was observed in the E. coli system.

DISCUSSION

Previous genomic (12) and proteomic (13) studies have already suggested that the degradation of the bile-salt side chain in Sphingobium sp. strain Chol11 proceeds differently from the reaction found in pseudomonads, comamonads, and Actinobacteria, and our study provides further functional evidence in comparative studies with strain Chol11 in contrast to P. stutzeri Chol1. Our results reveal that the ACAD reaction for introducing a Δ²² double bond catalyzed by Scd4AB is necessary for side chain degradation, which is still equivalent to the mechanism observed in the model organisms mentioned above. In contrast to all deletion mutants obtained so far in strain Chol11 (12, 28), Δscd4A did not grow with cholate anymore. However, the next steps of side chain degradation are apparently different and probably involve a hydroxylation step that might plausibly be catalyzed by the Rieske monooxygenase Noc2c228. As a further result regarding bile-salt degradation in strain Chol11, we obtained evidence that B-ring cleavage also proceeds via 9α-hydroxylation, leading to 9,10-seco structures with DHSATD (XI in Fig. 1) as intermediates because the Rieske monooxygenases Nov2c407, Nov2c430, and Nov2c440 catalyze the formation of DHSATD from HOCDA (V) in the kshA deletion mutant of strain Chol1 as a heterologous host.

The physiological role of Scd4AB as an ACAD for the introduction of a double bond at C-22 was concluded from the functional chemical complementation of the defective growth of the respective deletion mutant with DHOCTO (XIII in Fig. 2), which already had this double bond. Furthermore, heterologous expression of Scd1AB and Scd4AB in the individual ACAD deletion mutants of strain Chol11 and P. stutzeri Chol1, respectively, also enabled genetic complementation. Obviously, the complexes Scd4AB from strain Chol11 and Scd1AB from P. stutzeri Chol1 were interchangeable. However, both innate subunits were needed to form an active ACAD complex, indicating that the ACADs are too different to form promiscuous complexes. This points at a quaternary structure for Scd4AB that is similar to the α₂β₂ heteromeric structure of several other steroid ACADs for side chain degradation, such as Scd1AB, as well as for HIP degradation in steroid degradation model organisms (20, 33–37). In contrast to this common structure of steroid ACADs, the ACADs catalyzing this dehydrogenation of an activated C₅ side chain in R. jostii RHA1 during bile-salt degradation and in Mycobacterium tuberculosis during cholesterol degradation have a homodimeric structure composed of two subunits that each consist of the two ACAD domains found in the heterotetrameric ACADs (33, 38).

Furthermore, the strain Chol11 Δscd4A deletion mutant complemented with scd1AB grew with cholate without any difference from the wild type, while the strain Chol1 Δscd1A and Δscd1B deletion mutants complemented with scd4AB showed strongly extended lag phases of various lengths as well as reduced growth. These differing phenotypes could plausibly be explained by the fact that the CoA-ester of HOCDA (V in Fig. 1) with a Δ^4,6-structure of the steroid skeleton is the presumptive native substrate of Scd4AB, while Scd1AB oxidizes the CoA-ester of Δ^1,4-3-ketocholate (20). From previous studies, it is known that HOCDA can be readily converted by the side chain-degrading enzymes of strain Chol1 (11, 28). Scd4AB might be less active with steroid-acyl-CoA substrates that have a Δ^1,4-structure of the steroid skeleton and might require a Δ^4,6-substrate, which was not supplied in strain Chol1, for its full activity. Thus, the differences between Scd1AB and Scd4AB indicated by the phylogenetic analyses, which may hint at a different phylogenetic origin of this side chain degradation mechanism, were also reflected by this functional analysis.

The physiological role of the involvement of hydroxylation for side chain degradation is based on the interconnection of multiple independent results. First, strain Chol11 Δscd4A transformed cholate, deoxycholate, chenodeoxycholate, and lithocholate to several mono- and dihydroxylated steroid compounds with complete side chain. It is known from P. stutzeri Chol1 that hydroxylations may occur as side reactions when a steroid cannot be degraded further (11). Therefore, the formation of these hydroxylated steroids could be assumed to be side reactions, too. However, the fact that most of these compounds were further degraded by the strain Chol11 WT contradicts this assumption. Furthermore, the formation of CoA-esters of two of these steroids (P3 and P4) by the acyl-CoA ligase SclA, which is specific for the Δ^4,6-variant of bile-salt degradation (12), further supports the physiological relevance of these hydroxylated compounds.

For the dihydroxylated compound (P2), the position of one hydroxy group could certainly be assigned to the C-9 position of the steroid skeleton because of its UV spectrum that is characteristic for 9,10-seco steroids (11). In contrast, compounds P3 and P4 had UV spectra specific for steroids with a Δ^1,4-3-keto structure of the steroid skeleton and one hydroxy group. Comparison with authentic standards excluded that these hydroxy groups were located at carbon atoms 3, 6, 7, 9, or 12 of the steroid skeleton. Given that the hydroxylation is part of the metabolic pathway for bile-salt degradation, a monooxygenase must be involved. As the Rieske monooxygenase Nov2c228 is specifically upregulated in cells grown with cholate and deoxycholate compared to growth with 12β-DHADD without side chain (13), a specific role for side chain degradation is strongly suggested. In support of this hypothesis, Nov2c228, which is quite different from the apparent 9α-hydroxylating Rieske monooxygenases of strain Chol11, produced two metabolites that had an additional hydroxy group at an unknown position when HOCDA was supplied to the kshA deletion mutant of P. stutzeri Chol1.

The formation of DHSATD by the three Rieske monooxygenases Nov2c407, Nov2c430, and Nov2c440 plus the formation of the apparent seco-steroid P2 strongly suggest that bile-salt degradation via the Δ^4,6-variant also proceeds via the 9,10-seco pathway. Although this was assumed before, 9,10-seco-intermediates had not been detected in this pathway before.

As a prerequisite for the heterologous testing, we identified KshAB_Chol1 as a 9α-hydroxylase in Gram-negative bacteria. In contrast to previously identified KshAB monooxygenase systems from Actinobacteria (39, 40), KshAB_Chol1 only hydroxylated steroids without side chain, which is in agreement with mutants defective in side chain degradation not producing any seco-steroids (16, 17, 20, 22). Additionally, KshAB_Chol1 was shown to be involved in conversion of HATD (X in Fig. 1) to both DHSATD (XI) and THADD (XXIII in Fig. S13). Thus, KshAB_Chol1 performs the formerly known side reaction yielding THADD (11) in addition to its native function of 9α-hydroxylation. It remains unclear if both hydroxylations necessary for THADD production are catalyzed by KshAB alone. Other Rieske monooxygenases have been shown to catalyze different reactions with the same substrate, depending on binding position and also several subsequent hydroxylations on one substrate molecule upon repeated binding (41–43).

Although the 9α-hydroxylases from strain Chol11 could not be studied in the same detail as KshAB_Chol1, differences between these proteobacterial enzymes were evident. First, 9α-hydroxylase activity of KshA homologs from strain Chol11 was exclusively detected with P. stutzeri Chol1 ΔkshA as a host and HOCDA (V in Fig. 1) as the substrate, indicating that a Δ^4,6-structure was a prerequisite for this reaction. For KshAB_Chol1, in contrast, both Δ^4,6- and Δ^1,4-structures served as substrates. Second, the formation of the seco-steroid P2, which had a C₅-side chain, indicates that one of these enzymes presumably catalyzed the 9α-hydroxylation of steroids with side chain; this activity was not observed with KshAB_Chol1. Third, the 9α-hydroxylases from strain Chol11 and the fourth Rieske monooxygenase Nov2c228 did not appear to have specific reductase subunits like KshA_Chol1 has with KshB_Chol1. The only putative reductase component from strain Chol11, Novbp123, did not stimulate the activity of any of the KshA homologs from both P. stutzeri Chol1 and strain Chol11 in E. coli. Novbp123 is expressed constitutively and encoded in a cluster of membrane and electron transport proteins (13). This might indicate a different electron shuttling mechanism for KshA and additional enzymes in strain Chol11. The slight activities of Nov2c407, Nov2c430, and Nov2c440 in the P. stutzeri Chol1 background therefore must rely on another (unspecific) reductase, which is not present in E. coli MG1655.

The apparent dependence of key enzymes for steroid degradation in strain Chol11 on a Δ^4,6-substructure is in agreement with the constricted growth phenotype of the hsh2 deletion mutant of strain Chol11 (28). However, P3 and P4 detected in this study did not have a Δ⁶ double bond anymore according to their UV spectra. It might therefore be possible that after elimination of the 7-hydroxy group, the double bond is reduced similarly to the reductive dehydroxylation in Clostridium strains (5, 44) (Fig. 4D). All three homologs (Nov2c19 = 5β-Δ⁴-KSTD1, Nov2c85, and Nov2c314) of BaiH, which catalyzes this reaction in Clostridium scindens (44), had no activity toward HOCDA with NADH as electron donor when expressed in E. coli (data not shown). It is unclear if this reaction is a side reaction, which does not hinder further degradation, as these compounds are degraded by strain Chol11 WT, or a part of regular bile-salt degradation. As HATD (X in Fig. 1) can be found in supernatants of strain Chol11 cultures, the B-ring should not be reduced prior to side chain degradation during unhindered degradation.

The lack of homologs for steroid hydratases, aldolases, and thiolases, as well as the lack of a second ACAD for C₃ side chain degradation raises the question of whether strain Chol11 cleaves off the whole C₅-side chain by a so far unknown mechanism that might involve a monooxygenase reaction catalyzed by Nov2c228 (Fig. 10). A hydroxylation at C-17 is required at least at one point during cholate degradation in strain Chol11 because HATD (X in Fig. 1) is an intermediate of this pathway. Interestingly, the dead-end metabolites produced by P. stutzeri Chol1 ΔkshA expressing nov2c228 with HOCDA as the substrate had C₃ side chains. Also, the additional metabolites HOPTC (XXI in Fig. 9) and P11 accumulated, which could normally be further degraded by P. stutzeri Chol1. This could suggest that the additional hydroxylation had an inhibiting effect on the second step of side chain degradation in P. stutzeri Chol1. Although this clearly remains speculation at the current stage, a hydroxy group at C-17 might cause such inhibition because of its vicinity to the side chain. On the other hand, a hydroxy group at this position is also necessary for side chain cleavage and is inserted by Shy2 followed by aldolytic cleavage by Sal2 in strain Chol1 (20). Despite detailed studies on this reaction sequence in Actinobacteria, the stereochemistry of the C-17 hydroxy group is not known and can only be predicted so far (45–47). Therefore, an unsuited orientation of a hydroxy group at this position may prevent the aldolytic cleavage by Sal2 in strain Chol1. For elucidating side chain cleavage, a deletion mutant of Nov2c228 would be very useful, but despite numerous attempts, we have not succeeded so far with this specific gene.

FIG 10.

Proposed pathway for A-ring oxidation, B-ring cleavage, and side chain degradation during cholate degradation in Sphingobium sp. strain Chol11. Gray, structure suggestion; bold, known enzymes.

The only enzymes within the side chain degradation cluster that could have additional functions in side chain degradation are the amidase Nov2c229 and the putative hydroxysteroid dehydrogenases Nov2c226 and Nov2c231. Of these, only Nov2c229 is conserved in the putative side chain degradation gene clusters of other bile-salt-degrading sphingomonads (Sphingobium herbicidovorans MH, Novosphingobium tardaugens NBRC16725, and Novosphingobium aromaticivorans F199) (13) and therefore is an interesting candidate in our ongoing studies.

MATERIALS AND METHODS

Cultivation of bacteria.

Strains of Sphingobium sp. strain Chol11 (DSM 110934) (11), P. stutzeri Chol1 (DSM 103613) (23), and E. coli MG1655 (DSM 18039) (48) were grown in the HEPES-buffered mineral medium MB as described previously (11, 49). E. coli ST18 (DSM 22074) (50) was grown in lysogeny broth medium (LB) (51) with 50 μg ml⁻¹ 5-aminolevulinic acid. For strain maintenance and if not indicated otherwise, P. stutzeri Chol1 and strain Chol11 WT were grown with 1 mM cholate (I in Fig. 1) as the carbon source, mutants of P. stutzeri Chol1 were grown with 12 mM succinate and E. coli MG1655, and mutants of strain Chol11 were grown with 15 mM glucose. Strains containing pDM4 (52) (30 or 90 μg ml⁻¹ chloramphenicol), pEX18AP (53) (100 μg ml⁻¹ ampicillin for E. coli strains or 100 μg ml⁻¹ carbenicillin for P. stutzeri Chol1), or pBBR1MCS-5 (54) (20 μg ml⁻¹ gentamicin) were maintained on LB agar with the respective antibiotics and otherwise cultivated in the aforementioned medium with the respective antibiotics. During growth experiments and cultivation with steroids, antibiotics were omitted. Strains were maintained on agar plates prepared from the aforementioned media with 1.5% (wt/vol) Bacto agar (BD, Sparks, MD, USA).

Liquid cultures up to 5 ml were incubated in 10-ml test tubes and at 200 rpm, whereas larger cultures were incubated in 500-ml Erlenmeyer flasks without baffles. Except for strain maintenance of E. coli strains at 37°C, all strains were cultivated at 30°C.

Growth experiments were performed in 3 to 5 ml medium in 10-ml test tubes at 30°C with orbital shaking (Minitron or Ecotron; Infors HT, Einsbach, Germany). Starter cultures for growth experiments of P. stutzeri Chol1 or strain Chol11 were grown with succinate for 15 h or glucose for 20 h, respectively, and with antibiotics where appropriate. The main cultures were inoculated from starter cultures without previous washing at an OD₆₀₀ of about 0.02. Growth was monitored by measuring the OD₆₀₀ (Camspec M107; Spectronic Camspec, United Kingdom). At suitable time points, samples for HPLC-MS measurements were withdrawn.

Biotransformation experiments.

Biotransformation of different bile salts by strain Chol11 Δscd4A was determined in 5-ml or 100-ml cultures with 15 mM glucose and 1 mM respective bile salt, inoculated from precultures with a method similar to that used for growth experiments, and incubated at 30°C for 2 weeks.

For determining biotransformation of the resulting metabolites by strain Chol11, the supernatants of the aforementioned biotransformations were filtered and mixed 1:1 with fresh MB. The resulting medium was inoculated with strain Chol11 from starter cultures and incubated at 30°C for 4 days.

For tracking biotransformation of cholate by strain Chol11 cells in dense cell suspensions, 100-ml main cultures with 15 mM glucose were inoculated from starter cultures with glucose and incubated for about 40 h. Cells were harvested by centrifugation (8,000 × g, 4°C, 8 min), washed with MB without a carbon source, and resuspended with an OD₆₀₀ of about 1 in MB. Then, 5-ml aliquots of cell suspensions were prepared in 10-ml reaction tubes, and 1 mM cholate was added. Samples for HPLC-MS measurements were withdrawn directly after addition of cholate and at defined time points thereafter.

For testing the transformation of HOCDA (V in Fig. 1) by P. stutzeri Chol1 ΔkshA strains and transformation of various steroid compounds by E. coli MG1655 strains, cell suspensions with an OD₆₀₀ of about 1 were prepared from starter cultures by washing with and resuspending in MB. Next, 12 mM to 24 mM succinate was added to suspensions of P. stutzeri Chol1 strains, and 15 mM to 30 mM glucose was added to E. coli suspensions. The suspensions were incubated for 2 to 4 days for E. coli MG1655 or 6 days for P. stutzeri Chol1 at 30°C with 200 rpm orbital shaking. After incubation, cultures were directly frozen at −20°C and thawed for HPLC-MS measurements.

The ability of SclA to CoA-activate the metabolites P1 to P4 produced by strain Chol11 Δscd4A was tested in enzyme assays with cell extracts of E. coli MG1655 pBBR1MCS-5::sclA (12) or an empty vector control E. coli MG1655 pBBR1MCS-5. Cell extracts were prepared from 50-ml to 100-ml cultures in LB with gentamicin by sonication after washing and resuspending in 50 mM MOPS (morpholinepropanesulfonic acid) buffer (pH 7.8 with NaOH) as previously described (27) and stored at −20°C. The protein concentration in the cell extracts was determined with a bicinchoninic acid (BCA) assay kit (Pierce, Thermo Scientific, Rockford, IL, USA). Enzyme assays were prepared as follows: cell extract (12 mg ml⁻¹ total protein), 0.5 ml cell-free supernatant containing P1 to P4, 3 mM MgCl₂, 2 mM ATP, 2 mM Li₃CoA, and 50 mM MOPS buffer (pH 7.8 with NaOH, to a final volume of 1 ml). For controls, ATP, CoA, or cell extract was omitted. Enzyme assays were incubated at 30°C, and samples for HPLC-MS measurements were withdrawn as indicated.

For elucidating the position of the first additional hydroxy group in P3 and P4, P1 to P4 were purified from P1- to P4-containing cultures by solid-phase extraction with reversed-phase C₁₈ columns (Chromabond; Macherey-Nagel, Düren, Germany). Supernatants were acidified to pH 3 with HCl, and columns were washed with methanol and equilibrated with MilliQ water (H₂O_MQ). After the compounds were loaded on the columns, columns were washed first with H₂O_MQ and additionally with 10% methanol. The compounds were eluted with methanol, which was evaporated, and dissolved in H₂O with NaOH to gain a neutral pH. Purity and approximate concentration were determined by HPLC-MS. Enzyme assays with purified Hsh2 (28) were prepared in 50 mM MOPS (pH 7.8 with NaOH) and with approximately 2 mM steroid compounds in total. Enzyme assays were incubated at 30°C for up to 20 h, and transformation was monitored by HPLC-MS.

Cloning techniques and construction of deletion mutants.

Cloning was performed according to standard procedures and as described elsewhere (27).

Unmarked deletion mutants of strain Chol11 and P. stutzeri Chol1 were constructed as described elsewhere (17, 27) using the suicide vectors pDM4 (52) for strain Chol11 and pEX18AP (53) for P. stutzeri Chol1. Up- and downstream regions of the gene were amplified with primer pairs upfor/uprev and dnfor/dnrev (Table 4), respectively, and assembled by splicing by overlapping extension (SOE) PCR (55) using primer pair upfor/dnrev. Ligation into pDM4 or pEX18AP was checked using the primer pair MCS_for/MCS_rev or M13 primers, respectively. After transfer of the resulting plasmid into strain Chol11 or P. stutzeri Chol1 by conjugation, the respective primer pair backbone_for/backbone_rev was used to verify insertion into the genome. Second recombination was forced by cultivation on LB containing 10% sucrose. Colonies were checked for gene deletion using primer pair upfor/dnrev. After isolation of a pure culture, gene deletion was verified by PCR of genomic DNA and sequencing of the resulting fragment.

TABLE 4.

Primers used for construction of unmarked deletion mutants and plasmids for heterologous expression

Name	Sequencea	Restriction site
upfor_Chol11_nov2c218	TTTTTTTTCTAGACGTATCGCCGTGATGAGGAT	XbaI
uprev_Chol11_nov2c218	CCGCGATGGTTTGAGGAGAG
dnfor_Chol11_nov2c218	CTCTCCTCAAACCATCGCGG GGCTCTCTTCCTTTTCGTTC
dnrev_Chol11_nov2c218	TTTTTTTAAGCTTGTATCTTCCGGGCCGATCAT	HindIII
upfor_Chol11_nov2c219-220	TTTTTTTCTCGAGATCCTGCAGTCAATCGGTCG	XhoI
uprev_Chol11_nov2c219-220	GCGGGAACGCCTGAACGGAACGAAAAGGAAGAGAGCC
dnfor_Chol11_nov2c219-220	CGTTCAGGCGTTCCCGC
dnrev_Chol11_nov2c219-220	TTTTTTTTCTAGATGCCCATATCTCCGACTGGA	XbaI
upfor_Chol11_scd4A	TTTTTTTCTCGAGCATCATCAGGACGTCGGTGT	XhoI
uprev_Chol11_ scd4A	ACGGTGGACGAGGAAGATGC
dnfor_Chol11_ scd4A	GCATCTTCCTCGTCCACCGTGCCGCTTCTCCCAGGATCAT
dnrev_Chol11_ scd4A	TTTTTTTCTAGATTAGTGTGACGAGCCAGACG	XbaI
upfor_Chol1_kshA	TTTTTTTTCTAGATTCGGTGGATTGGGTCGAAG	XbaI
uprev_Chol1_kshA	GTCGTTACCCTCAGGCGTTC
dnfor_Chol1_kshA	GAACGCCTGAGGGTAACGACAGGATAACGCTGTGCCAGAC
dnrev_Chol1_kshA	TTTTTTTAAGCTTTCAATCCAAGCCTACGGACG	HindIII
expfor_Chol11_nov2c218	TTTTTTTCTCGAGATGGCGCTGTTGGATGGAC	XhoI
exprev_Chol11_nov2c218	TTTTTTTTCTAGATCACGCAAAAGCCGTCCC	XbaI
expfor_Chol11_nov2c219-220	TTTTTTTCTCGAGGTGGACGAGGAAGATGCCTTTG	XhoI
exprev_Chol11_nov2c219-220	TTTTTTTTCTAGATCAGTCCAGCAGCAGATGCA	XbaI
expfor_Chol11_nov2c221	TTTTTTTCTCGAGATGAAACTCGGATTTTCCCCGG	XhoI
exprev_Chol11_nov2c221	TTTTTTTTCTAGATCAGGCGTTCCCGCG	XbaI
expfor_Chol11_nov2c221-222	TTTTTTTCTCGAGATGGAATTCGCATTTACCGACGAACAACAG	XhoI
exprev_Chol11_nov2c221-222	TTTTTTTTCTAGATCAGGCGTTCCCGCGCG	XbaI
expfor_Chol1_scd1A	TTTTTTTCTCGAGATGAGCACTATCGAGCAGTT	XhoI
exprev_Chol1_scd1A	TTTTTTTTCTAGATTACTTGCTATCCGGCAGGC	XbaI
expfor_Chol1_scd1AB	TTTTTTTAAGCTTATGAGCACTATCGAGCAGTT	HindIII
exprev_Chol1_scd1AB	TTTTTTTTCTAGATCAGTCCAGCGTCTTGC	XbaI
expfor _Chol1_kshAChol1	TTTTTTTCTCGAGGAACGCCTGAGGGTAACGAC	XhoI
exprev_Chol1_kshAChol1	TTTTTTTGGATCCTCATTTGCGCTGGTCGACC	BamHI
expfor_Chol11_nov2c066	TTTTTTTCTCGAGCGATGGAGAGGCAGGACTCG	XhoI
exprev_Chol11_nov2c066	TTTTTTTGGATCCTCAATCGACCCCCGGCG	BamHI
expfor_Chol11_nov2c228	TTTTTTTCTCGAGAATCTGACGGAGCCTTTTCC	XhoI
exprev_Chol11_nov2v228	TTTTTTTGGATCCTTATTCCGCGGCCGC	BamHI
expfor_Chol11_nov2c407	TTTTTTTCTCGAGCGCAGTGAGAGGATCAAGCA	XhoI
exprev_Chol11_nov2c407	TTTTTTTGGATCCTTATGCCGCCTGCGTGG	BamHI
expfor_Chol11_nov2c430	TTTTTTTCTCGAGACAAGAGAGAGGATGGACGG	XhoI
exprev_Chol11_nov2c430	TTTTTTTGGATCCTCATTCGGCCGCTGC	BamHI
expfor_Chol11_nov2c440	TTTTTTTCTCGAGGCGAGTGAGAGGATCAAGCA	XhoI
exprev_Chol11_nov2c440	TTTTTTTGGATCCTTATGCCGCCTGCGTGG	BamHI
expfor_Chol1_kshBChol1	TTTTTTTGGATCCTTCGTCATGAGTATCTGAGC	BamHI
exprev_Chol1_kshBChol1	TTTTTTTTCTAGATCAGTCGGGAAAGCG	XbaI
expfor_Chol11_novbp123	TTTTTTTGGATCCCGCACACTAAGGGGATGAGG	BamHI
exprev_Chol11_novbp123	TTTTTTTTCTAGATCAGGCTTCGACCACGAG	XbaI
M13 for (–43)	AGGGTTTTCCCAGTCACGACGTT
M13 rev (–49)	GAGCGGATAACAATTTCACACAGG
pEX18AP_backbone_for	AGGAGACATGAACGATGAACA
pEX18AP_backbone_rev	TTTTTTTCCCGGGTCGGCATTTTCTTTTGCGTT
pDM4_MCS_for	ACTTAACGGCTGACATGGGA
pDM4_MCS_rev	GCGAAGTGATCTTCCGTCAC
pDM4_backbone_for	AAGATGTGGCGTGTTACGGT
pDM4_backbone_rev	AGGCTCTGGGAGGCAGAATA

^aUnderlined, restriction sites.

For expression of various genes in P. stutzeri Chol1, E. coli MG1655, and strain Chol11 strains, genes were amplified using the primer pair expfor/exprev (Table 4) and ligated into vector pBBR1MCS-5. The respective plasmids as well as pBBR1MCS-5::sclA (12) were transferred to E. coli MG1655 or ST18 by heat shock transformation. For addition of a second gene to the expression vector, the plasmid was isolated, and the second gene was added by ligation after restriction. From E. coli ST18, vectors were transferred to strains of E. coli MG1655, P. stutzeri Chol1, and strain Chol11 by conjugation as previously described (17, 28). The presence of plasmids was confirmed by colony PCR using M13 primers.

HOCDA production strain P. stutzeri Chol1 ΔstdA1 ΔkstD1 pBBR1MCS-5::hsh2 was generated by transferring plasmid pBBR1MCS-5::hsh2 (28) into P. stutzeri Chol1 ΔstdA1 ΔkstD1 (27).

Preparation of steroid compounds.

Cholate (≥99%, from ox or sheep bile), deoxycholate (≥97%), and lithocholate (≥95%) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Chenodeoxycholate (≥98%) was purchased from Carl Roth (Karlsruhe, Germany). When lithocholate was added to cultures as a substrate, it was added to MB together with 1% (wt/vol) methyl-β-cyclodextrin (TCI, Tokyo, Japan) before autoclaving for solubilization.

Steroid compounds 12β-DHADD (VI in Fig. 1), DHOPDC (XII in Fig. 2), DHOCTO (XIII in Fig. 2), and the Δ⁴-3-keto derivatives of bile salts were prepared by biotransformation as described elsewhere (17, 23, 27). All ADDs (VI in Fig. 1 and XVII, XVIII, and XIX in Fig. S9) were produced similarly to 12β-DHADD by anoxic transformation of the respective bile salt with P. stutzeri Chol1. HATD (X in Fig. 1) was produced similarly using cholate and P. stutzeri Chol1 pBBR1MCS-5::hsh2 (28). For production of HOCDA (V in Fig. 1), P. stutzeri Chol1 ΔstdA1 ΔkstD1 pBBR1MCS-5::hsh2 was incubated with cholate and succinate until the cholate was completely transformed into HOCDA. For the production of Δ⁴-3-keto bile salts, chenodeoxycholate, ursodeoxycholate, or hyodeoxycholate was supplied to P. stutzeri Chol1 ΔstdA1 ΔkstD1 (27). Dead-end products produced by strain Chol11 Δscd4A from all the aforementioned bile salts were produced by biotransformation in 100-ml cultures containing 2 mM the respective bile salt and incubated for 2 weeks. Δ⁶-HOCTO (XXII) was produced using DHOCTO as a substrate and P. stutzeri Chol1 ΔstdA1 ΔkstD1 pBBR1MCS-5::hsh2 for transformation. HOPTC (XXI in Fig. 9) was produced like DHOPDC but using HOCDA as the substrate.

ADDs, HATD, DHOPDC, DHOCTO, HOCDA, and the Δ⁴-3-keto bile salts were purified by organic extraction as previously described with dichloromethane for ADDs and HATD or ethyl acetate for the other compounds (21, 23) and resolved in MilliQ pure water (Merck, Darmstadt, Germany). For Δ⁶-HOCTO and the metabolites P1 to P8, supernatants of production cultures were sterilized by filtration and used directly in a dilution of 1:1 with fresh medium if not indicated otherwise.

The purity of all steroid compounds was assessed using HPLC-MS measurements.

The concentration of HOCDA, DHOPDC, and ADDs was determined photometrically as previously described (11, 23).

Analytical methods.

Steroid compounds were analyzed by HPLC-MS. Samples were centrifuged (>16,000 × g, ambient temperature, 5 min) to remove cells and particles directly prior to measurement. HPLC-MS measurements were performed using a Dionex Ultimate 3000 HPLC (Thermo Fisher Scientific, Waltham, MA, USA) with a UV/visible light diode array detector, coupled to an ion trap Amazon speed mass spectrometer (Bruker, Bremen, Germany) with an electrospray ion source and equipped with a reversed-phase C₁₈ column (150 by 3 mm; Eurosphere II, 100-5 C18; Knauer Wissenschaftliche Geräte, Berlin, Germany). For separation, a gradient from 90% to 10% 10 mM ammonium acetate buffer with 0.1% formic acid and acetonitrile as described in reference 27 was used. For analyzing samples of CoA-activation enzyme assays, no formic acid was added to the buffer, and measurements were performed at neutral pH.

Cholate concentrations were determined as the peak area from base peak chromatograms measured in negative mode. Intermediates were identified due to retention time, UV and MS spectra, and comparison with known compounds. Structures of unknown metabolites were proposed based on retention time and UV and MS spectra.

Bioinformatical methods.

Homology searches and similarity determinations were performed using the BLASTp algorithm (56, 57). Primer-BLAST (58) was used for generation of primers for construction of deletion mutants. Interpro (59) was used for prediction of protein domains. Alignments and phylogenetic trees were calculated using MEGA X (60) (alignment, ClustalW with standard parameters [61]; phylogeny, neighbor joining method with standard parameters and 100 bootstrap repetitions [62]) and visualized with iTOL (63).