A Novel Steroid-Coenzyme A Ligase from Novosphingobium sp. Strain Chol11 Is Essential for an Alternative Degradation Pathway for Bile Salts

ABSTRACT

Bile salts such as cholate are steroid compounds with a C₅ carboxylic side chain and occur ubiquitously in vertebrates. Upon their excretion into soils and waters, bile salts can serve as growth substrates for diverse bacteria. Novosphingobium sp. strain Chol11 degrades 7-hydroxy bile salts via 3-keto-7-deoxy-Δ^4,6 metabolites by the dehydration of the 7-hydroxyl group catalyzed by the 7α-hydroxysteroid dehydratase Hsh2. This reaction has not been observed in the well-studied 9-10-seco degradation pathway used by other steroid-degrading bacteria indicating that strain Chol11 uses an alternative pathway. A reciprocal BLASTp analysis showed that known side chain degradation genes from other cholate-degrading bacteria (Pseudomonas stutzeri Chol1, Comamonas testosteroni CNB-2, and Rhodococcus jostii RHA1) were not found in the genome of strain Chol11. The characterization of a transposon mutant of strain Chol11 showing altered growth with cholate identified a novel steroid-24-oyl–coenzyme A ligase named SclA. The unmarked deletion of sclA resulted in a strong growth rate decrease with cholate, while growth with steroids with C₃ side chains or without side chains was not affected. Intermediates with a 7-deoxy-3-keto-Δ^4,6 structure, such as 3,12-dioxo-4,6-choldienoic acid (DOCDA), were shown to be likely physiological substrates of SclA. Furthermore, a novel coenzyme A (CoA)-dependent DOCDA degradation metabolite with an additional double bond in the side chain was identified. These results support the hypothesis that Novosphingobium sp. strain Chol11 harbors an alternative pathway for cholate degradation, in which side chain degradation is initiated by the CoA ligase SclA and proceeds via reaction steps catalyzed by so-far-unknown enzymes different from those of other steroid-degrading bacteria.

IMPORTANCE This study provides further evidence of the diversity of metabolic pathways for the degradation of steroid compounds in environmental bacteria. The knowledge about these pathways contributes to the understanding of the CO₂-releasing part of the global C cycle. Furthermore, it is useful for investigating the fate of pharmaceutical steroids in the environment, some of which may act as endocrine disruptors.

INTRODUCTION

Bile salts are a subclass of steroids, which occur in all vertebrates where they act as emulsifiers of lipophilic nutrients in the intestine and affect lipid and energy metabolism (1, 2). In addition, many further signaling functions of bile salts are currently being discovered with regard to, e.g., microbiome-host interactions (3). Considerable amounts of bile salts are released into the environment, including approximately 400 to 800 mg bile salts released per human per day via urine and feces (4). Additionally, some vertebrates release bile salts as pheromones (5, 6). In soils and waters, bile salts are a carbon- and energy-rich substrate for heterotrophic bacteria. Accordingly, diverse Gram-negative and Gram-positive bacteria, which can grow with bile salts as carbon and energy sources under aerobic conditions, have been isolated and investigated in recent years (7, 8). Using the trihydroxy bile salt cholate (Fig. 1, compound I) as a model compound, two pathways for initiating bile salt degradation have been identified. The first degradation pathway is extensively being investigated with Pseudomonas stutzeri strain Chol1 and Rhodococcus jostii strain RHA1 as well as with different strains of Comamonas testosteroni (9–12) and can be separated into four distinct reaction sequences: partial oxidation of the A ring, stepwise degradation of the carboxylic side chain, cleavage of the B ring between C-9 and C-10, and further degradation of the resulting 9,10-seco steroid. This degradation pathway proceeds via intermediates with a 3-keto-Δ^1,4-diene structure of the steroid skeleton, which is formed by the initiating oxidative reactions at the A ring.

FIG 1.

Section of the proposed cholate (I) degradation pathways in P. stutzeri Chol1 (blue) and Novosphingobium sp. strain Chol11 (yellow). In both strains, cholate is sequentially oxidized to 3-ketocholate (II) and Δ⁴-3-ketocholate (III). In P. stutzeri Chol1, this intermediate is further degraded via Δ^1,4-3-ketocholate (IV), CoA ester of 7α,12α-dihydroxy-3-oxochola-1,4,(22E)-triene-24-oate ([DHOCTO] V), CoA ester of THOCDO (VI), 7α,12α-dihydroxy-3-oxopregna-1,4-diene-20S-carbaldehyde, ([DHOPDCA] VII), 7α,12α-dihydroxy-3-oxopregna-1,4-diene-20-carboxylate ([DHOPDC] VIII), 7α,12β-dihydroxy-androsta-1,4-diene-3,17-dione ([12β-DHADD] IX), and 3,7,12-trihydroxy-9,10-secoandrosta-1,3,5(10)-triene-9,17-dione ([THSATD] X). In Novosphingobium sp. strain Chol11, Δ⁴-3-ketocholate (III) is degraded via 12α-hydroxy-3-oxo-4,6-choldienoic acid ([HOCDA] XI), 3,12-dioxo-4,6-choldienoic acid ([DOCDA] XII), and 12β-hydroxy-androsta-1,4,6-triene-3,17-dione ([HATD] XIII). 3,12β-Dihydroxy-9,10-secoandrosta-1,3,5(10),6-tetraene-9,17-dione ([DHSATD] XIV) is the next expected intermediate, which cannot be metabolized by P. stutzeri Chol1. In Novosphingobium sp. strain Chol11, 12β-DHADD can be converted into HATD by Hsh2.

In Proteobacteria, the side chain is usually completely degraded before the cleavage of the B ring (9, 13). In Actinobacteria, these reactions can occur concomitantly (14). Side chain degradation is initiated by coenzyme A (CoA) activation catalyzed by a steroid-24-oyl-CoA ligase in the model organisms P. stutzeri Chol1, Pseudomonas putida strain DOC21, and C. testosteroni strain KF-1 (15). In strain DOC21, this enzyme (StdA1_DOC21) also catalyzes the activation of C₅ side chain degradation products with an oxidized A ring. Further degradation of the side chain proceeds by stepwise removal of an acetyl-CoA and a propionyl-CoA residue by modified β-oxidation reactions (9, 16). For this, the side chain is dehydrogenated at the Δ²² position and the resulting double bond is hydrated. Biochemical and genetic evidences in Proteobacteria suggest that acetyl-CoA is subsequently removed from the β-hydroxyacyl-CoA ester by an aldolytic cleavage reaction, leading to the formation of an aldehyde intermediate with a shortened C₃ side chain (17). The aldehyde function is subsequently oxidized to the corresponding acid, leading to the formation of 3-oxopregna-1,4-diene-20-carboxylate derivatives (OPDCs) with free carboxyl groups. In strain DOC21, a second steroid acyl-CoA ligase, StdA2_DOC21, was shown to specifically activate the C₃ side chain (15). The remaining side chain is removed via a mechanism similar to the first reaction cycle, ending with another aldolytic cleavage reaction splitting off propionyl-CoA (9, 15, 16). Homologs of StdA1 and StdA2 are also present in P. stutzeri Chol1 and C. testosteroni KF-1. In R. jostii RHA1, most side chain degradation reactions are thought to be similar to the progression in Proteobacteria, as it harbors two CoA ligases, CasG and CasI, which catalyze the CoA activation of cholate degradation intermediates with C₅ and a C₃ side chains, respectively (18). However, the exact reaction sequence leading to the cleavage of acetyl-CoA from the side chain is currently unknown in R. jostii.

The complete removal of the side chain results in the formation of C₁₉-androstadienediones (ADDs), which are further transformed into the aforementioned 9,10-seco steroids (7, 12). This reaction is catalyzed by a monooxygenase, which hydroxylates the C-9 atom, leading to the opening of the B ring and the simultaneous aromatization of the A ring. Further degradation of this 9,10-seco steroid proceeds via opening of the aromatic A ring followed by hydrolytic cleavage of the resulting open rings (12). These reactions yield 2-hydroxyhexa-2,4-dienoic acid and derivatives of H-methylhexahydroindanone-propanoates (HIPs), which consist of the remaining C and D rings plus a propionyl side chain derived from the former B ring (9). Further HIP degradation reactions have only recently been discovered (19).

In the actinobacterium Dietzia sp. strain Chol2 and the Alphaproteobacteria Sphingomonas sp. strain Chol10 and Novosphingobium sp. strain Chol11, an alternative degradation pathway has been detected (8). In these bacteria, cholate degradation proceeds via novel 3-keto-Δ^4,6-diene metabolites with double bonds in the B rings, namely, 12α-hydroxy-3-oxo-4,6-choldienoic acid (HOCDA; XI in Fig. 1) and 3,12-dioxo-4,6-choldienoic acid (DOCDA; XII). To our knowledge, these metabolites are not formed during bile salt degradation in any of the above-described model organisms, P. stutzeri. Chol1, P. putida DOC21, C. testosteroni, and R. jostii RHA1. The hydroxysteroid 7α-dehydratase Hsh2 has recently been identified as the key enzyme for formation of these compounds from cholate and other 7-hydroxy bile salts in Novosphingobium sp. strain Chol11 (20). After the oxidation of the 3-hydroxy group and the introduction of a double bond at the Δ⁴ position, the hydroxyl group at C-7 is removed by Hsh2, resulting in the formation of HOCDA (XI), which is further degraded to 12β-hydroxy-androsta-1,4,6-triene-3,17-dione (HATD; XIII). Dehydration of the 7-hydroxy group has been shown to also proceed during the degradation of the 7β-hydroxy bile salt ursodeoxycholate (20). Our studies suggest that this dehydration reaction is an obligatory step during the degradation of 7-hydroxy steroids in strain Chol11.

In this study, we aimed at further exploring this alternative bile salt degradation pathway. For this, we applied bioinformatics analysis of the draft genome of strain Chol11 as well as transposon and directed mutagenesis to identify further genes, proteins, and reaction steps involved in this pathway.

RESULTS

In silico genome analysis of Novosphingobium sp. strain Chol11.

The genome of Novosphingobium sp. strain Chol11 (EMBL database [EBI] accession no. OBMU01000001 to OBMU01000010) comprises two chromosomes (chromosome 1 [2.54 Mb] and chromosome 2 [0.86 Mb]) and two plasmids (pSa [0.13 Mb] and pSb [0.13 Mb]) containing 3,532 putative open reading frames. For identifying potential steroid degradation proteins, we analyzed the in silico proteome of strain Chol11 with hidden Markov models (HMMs) and BLASTp. This analysis identified 50 potential steroid degradation proteins encoded in the genome of strain Chol11, including 16 hits for putative steroid side chain degradation proteins, 14 hits for A/B ring degradation, 17 hits for C/D ring degradation, and three hits for the degradation of 2-hydroxyhexa-2,4-dienoic acid (see Table S1 in the supplemental material). Eight of those proteins are encoded on chromosome 1 (2.54-Mb chromosome) and 40 are encoded on chromosome 2 of strain Chol11. While HMM hits for putative steroid degradation genes on chromosome 1 are scattered throughout the chromosome, three potential steroid degradation gene clusters can be identified on chromosome 2 (see Fig. S1). Cluster 1 contains four potential steroid degradation genes encoding homologs of the A/B ring degradation proteins KstD, KshA, and HsaD and a potential side chain degradation protein. Cluster 2 contains 20 potential steroid degradation genes, including two sets of neighboring genes encoding the potential A/B ring degradation proteins HsaA to -D and the potential 2-hydroxyhexa-2,4-dienoic acid degradation proteins HsaE to -G. These hits are accompanied by five genes presumably encoding side chain degradation proteins and four genes presumably encoding C/D degradation proteins, as well as three genes encoding homologs of KshA. Cluster 3 contains nine potential steroid degradation genes, including one gene encoding a homolog of KstD, as well as three genes encoding homologs of putative side chain degradation proteins and five genes encoding C/D degradation proteins.

The reciprocal BLASTp analysis was performed using characterized and hypothetical steroid degradation proteins of P. stutzeri strain Chol1, R. jostii RHA1, and C. testosteroni strain CNB-2 as query sequences and the proteome of strain Chol11 as the subject (Fig. 2). In general, Chol11 has more homologs to steroid degradation proteins from the Gram-negative strains Chol1 and CNB-2 than to proteins from the Gram-positive strain RHA1. While multiple homologs to proteins encoded in the cholesterol degradation gene cluster in strain RHA1 exist, only minor sequence similarities to proteins in the cholate degradation gene cluster in RHA1 can be found.

FIG 2.

Reciprocal BLASTp analysis of steroid degradation proteins of Novosphingobium sp. strain Chol11. The heat map shows BLAST similarities to characterized and hypothetical steroid degradation proteins from Pseudomonas stutzeri Chol1, Rhodococcus jostii RHA1, and Comamonas testosteroni CNB-2. Characterized side chain degradation proteins are marked in boldface font.

The proteome of strain Chol11 contains homologs of most key enzymes of steroid ring degradation, such as KstD, KshA, HsaC, and HsaD, as well as homologs of most C/D ring degradation proteins (Fig. 2). Strikingly, the proteome of Chol11 does not contain reciprocal BLAST hits of most known steroid side chain degradation proteins from strain Chol1 or RHA1. In particular, there are no homologous proteins for Scd1AB, Shy1, Sal1, or Sad, which catalyze the release of acetyl-CoA from the C₅ side chain of cholate in Chol1, or homologs of Scd2AB or Sal2, which catalyze part of the subsequent release of propionyl-CoA from the steroid skeleton. These findings suggest that the steroid side chain degradation in strain Chol11 may proceed via so-far-unknown reaction steps.

Characterization of the transposon mutant strain Chol11 Tn50KL.

To identify potential new side chain degradation genes in strain Col11, we subjected the wild type to random mutagenesis by the insertion of the transposon mini-Tn5 Km1. Three of 5,000 transposon mutants had altered growth phenotypes when growing with cholate. Transposon mutant strain Chol11 Tn50KL grew only poorly with cholate as the sole carbon and energy source (see Fig. S2) and was therefore further characterized. High-performance liquid chromatography-mass spectrometry (HPLC-MS) analyses revealed that HOCDA (XI) and DOCDA (XII) transiently accumulated in the culture supernatants of strain Chol11 Tn50KL (Fig. 3, top). After 10 days of incubation, two unknown intermediates with absorption maxima at 290 nm and with molecular masses at 414 and 430 Da remained in culture supernatants (P1 and P2, respectively) (Fig. 3, bottom; see also Fig. S3A and B). Growth experiments with extracted supernatants showed that these compounds were not degraded by the wild-type strain Chol11 (not shown).

FIG 3.

HPLC chromatograms of a supernatant of the transposon mutant Chol11 Tn50KL during growth with 1 mM cholate. HOCDA and DOCDA were detected after 51 h of incubation (top). Two novel intermediates, P1 and P2, were detected in the supernatant after 240 h of incubation. Analysis wavelength, 290 nm.

Plasmid sequencing of kanamycin-resistant Escherichia coli clones of a Chol11 Tn50KL clone library revealed that the transposon was inserted into the gene nov2c230, which was renamed sclA (see below). Automatic annotation suggested that sclA codes for an AMP-dependent synthetase and ligase with a molecular mass of approximately 63 kDa.

Characterization of the mutant strain Chol11 ΔsclA.

In the next step, we constructed an unmarked sclA gene deletion mutant to rule out potential polar effects of the transposon in strain Chol11 Tn50KL. The growth rate of strain Chol11 ΔsclA with cholate as the substrate was 5-fold lower than that of wild-type cells (μ_wild-type = 0.34 ± 0.02 h⁻¹ and μ_{Chol11 ΔsclA} = 0.07 ± 0.01 h⁻¹) (Fig. 4A). Similar to results for strain Chol11 Tn50KL, HPLC-MS analyses showed that Δ⁴-3-ketocholate (III), HOCDA (XI), and DOCDA (XII) accumulated transiently in culture supernatants, indicating that the initial A and B ring oxidation reactions of cholate degradation were not affected in the deletion mutant. At the end of the exponential growth phase after approximately 10 days, cholate was no longer detectable. Also the dead-end intermediates P1 and P2 accumulated in culture supernatants. So far, P1 and P2 have not been identified.

FIG 4.

(A) Growth of Novosphingobium sp. strain Chol11 (filled diamonds) and mutant strain Chol11 ΔsclA (open circles) with 1 mM cholate. (B) Growth of Novosphingobium sp. strain Chol11 (filled symbols) and mutant Chol11 ΔsclA (open symbols) with 1 mM DHOPDC (circles) or with 1 mM DHADD (diamonds). Error bars indicate standard deviations (n = 3).

Strain Chol11 ΔsclA grew with deoxycholate with a significantly lower growth rate than the wild-type strain Chol11 and did not reach a similar maximum cell density (see Fig. S4). This indicates that SclA is required for metabolizing both cholate and deoxycholate. In addition, we performed growth experiments with two cholate degradation intermediates from P. stutzeri Chol1, namely, 7α,12α-dihydroxy-3-oxopregna-1,4-diene-20-carboxylate (DHOPDC; VIII in Fig. 1) with a C₃ side chain and 7α,12β-dihydroxy-androsta-1,4-diene-3,17-dione (12β-DHADD; IX) without any side chain (Fig. 4B). The growth phenotype of the deletion mutant with DHOPDC (VIII) and 12β-DHADD (IX) resembled that of the wild type, indicating that the degradation of these substrates was not affected in Chol11 ΔsclA.

As sclA was annotated as a putative CoA ligase, we analyzed CoA ligase activities in cell extracts of wild-type and mutant strains with cholate and its degradation intermediates in the presence of CoA, ATP, and Mg²⁺ using HPLC-MS. No CoA ester was formed by the wild type or mutant cell extracts with cholate or DHOPDC (VIII) as the substrates. In contrast, CoA esters of 3-ketocholate (II), Δ⁴-3-ketocholate (III), and DOCDA (XII) were formed within 15 min of incubation in wild-type cell extracts (see Fig. S5). After 40 min of incubation, approximately 45% of DOCDA (XII) was transformed into DOCDA-CoA in these assays. Further incubation led to a decrease of DOCDA-CoA accompanied by a concomitant increase of free DOCDA (XII), indicating hydrolysis of the thioester bond (see Fig. S6). Without the addition of ATP, DOCDA-CoA was not formed (data not shown), suggesting that these reactions were catalyzed by an ATP-dependent CoA ligase. In cell extracts of strain Chol11 ΔsclA, the formation of DOCDA-CoA was not observed (Fig. S5). These results strongly suggest that sclA encodes a CoA ligase responsible for side chain activation of cholate degradation intermediates with C₅ side chains and oxidized A rings. In the extracts of cells grown with glucose, CoA activation of DOCDA was also measured (see Fig. S7), indicating that the presence of cholate is not required for the expression of this CoA ligase.

Several attempts to complement the sclA deletion mutant were made, including the expression of an intact copy of sclA under the control of the lac promoter of Escherichia coli or a putative promoter found in the upstream region of sclA. However, all attempts failed.

Characterization of SclA.

To confirm the proposed CoA ligase activity of SclA, we produced a His-tagged recombinant SclA protein in E. coli. The corresponding protein band of the His-tagged recombinant protein on SDS-PAGE was consistent with the calculated molecular mass of 63 kDa for SclA (see Fig. S8). The substrate spectrum of the purified recombinant SclA protein was characterized using cholate and cholate degradation intermediates known from P. stutzeri Chol1 (21) and Novosphingobium sp. strain Chol11 (20). For this purpose, SclA was incubated with cholate (I), 3-ketocholate (II), Δ⁴-3-ketocholate (III), DOCDA (XII), 7α,12α,dihydroxy-3-oxochola-1,4,(22E)-triene-24-oate (DHOCTO; V), 7α,12α,22-trihydroxy-3-oxochola-1,4-diene-24-oate (THOCDO; VI), and DHOPDC (VIII) as the substrates in the presence of CoA, ATP, and Mg²⁺. The reaction mixtures were analyzed for substrate consumption and CoA ester formation by HPLC-MS after 45 min (see Fig. S9). In accordance with CoA ligase activity in cell extracts, the highest concentrations of CoA esters were measured with DOCDA (XII) as the substrate (Fig. 5). Simultaneously, AMP formation and ATP depletion were observed, suggesting an AMP-forming CoA ligase (see Fig. S10). Compared to the amount of DOCDA-CoA formed after 45 min, less than 2% of cholyl-CoA was formed with cholate as the substrate and approximately 32% and 83% of the respective CoA esters of 3-ketocholate (II) and Δ⁴-3-ketocholate (III). In contrast, no CoA ester formation was observed with DHOPDC (VIII). CoA ester formation with DHOCTO was lower than 5%. Δ^1,4-3-Ketocholate (IV) was also activated with CoA; however, this substrate preparation also contained a considerable amount of Δ⁴-3-ketocholate (III), which makes it difficult to make a clear statement on the activation level for this substrate. These results showed that SclA predominantly activates C₅ acyl side chains of steroids with a 3-keto-Δ⁴-monoene or 3-keto-Δ^4,6-diene structure of the steroid skeleton. On the basis of the results of genetic and biochemical analyses, we named this gene sclA for steroid CoA ligase 1. In vector control strains of E. coli, no CoA ligase activity was observed with DOCDA as the substrate in cell extracts excluding any CoA ligase activity derived from the host strain used for overexpression.

FIG 5.

Analysis of substrate spectra of SclA. Purified recombinant SclA was incubated with cholate (a [I in Fig. 1]), deoxycholate (b), 3-ketocholate (c [II in Fig. 1]), Δ⁴-3-ketocholate (d [III in Fig. 1]), DOCDA (e [XII in Fig. 1]), DHOCTO (f [V in Fig. 1]), and DHOPDC (g [VIII in Fig. 1]), and CoA ester formation after 45 min of incubation was analyzed by HPLC-MS. The amount of DOCDA-CoA formed after 45 min was set to 100%, and CoA ester formation with other substrates was compared with the concentration of DOCDA-CoA. Integrated peak areas originated from extracted ion chromatograms in negative ion mode of MS (for cholyl-CoA, m/z 1,156.7; for deoxycholyl-CoA, m/z 1,140.7; for 3-ketocholyl-CoA, m/z 1,154.7; for Δ⁴-3-ketocholyl-CoA, m/z 1,152.7; for DOCDA-CoA, m/z 1,132.8; for DHOCTO-CoA, m/z 1,148.7; for DHOPDC, m/z 1,122.7) (see Fig. S11 in the supplemental material for mass and UV spectra) (n = 3).

Analysis of side chain degradation in Novosphingobium sp. strain Chol11.

In our previous studies, we identified HATD (XIII) as a cholate degradation intermediate of Novosphingobium sp. strain Chol11, from which the carboxylic side chain had been completely removed (8). However, we did not detect any intermediates with a modified side chain in culture supernatants of this strain during growth with cholate or other bile salts. As the discovery of SclA suggested that the degradation of DOCDA (XII) is initiated by CoA activation, we performed enzyme assays with cell extracts of strain Chol11 to investigate further degradation of DOCDA-CoA. For this purpose, the above-mentioned CoA activation assays with DOCDA (XII; 384 Da) were performed in the presence of the artificial electron carrier phenazine methosulfate (PMS). Under these conditions, a novel intermediate eluting very closely to DOCDA (XII) was detected in reaction mixtures after 300 min of incubation (Fig. 6A, top and middle). This compound (P3) had an absorption maximum at approximately 205 nm in addition to the maximum at 290 nm, and its molecular mass (382 Da) was 2 Da lighter (see Fig. S11A) than that of the initial substrate DOCDA (XII). Control assays revealed that the formation of this compound depended on the presence of both CoA and PMS. In the presence of PMS and the absence of CoA, another unknown compound (P4) was detected instead of the P3 (Fig. 6A, bottom). This intermediate also had a molecular mass of 382 Da. However, its absorption spectra differed clearly from the first one. It had a UV spectrum with features characteristic of the Δ^1,4,6-triene intermediate HATD (XIII), exhibiting absorption maxima at 226, 255, and 305 nm (Fig. S11B) (8). When PMS was replaced with NAD⁺ in the presence and absence of CoA, none of these compounds were formed. The molecular masses of P3 and P4, as well as the dependency of their formation on PMS, indicate that both of these compounds have one C=C double bond more than the substrate DOCDA (XII). Supplying cell extracts with recombinant SclA increased the formation of P3 (see Fig. S12A and B). Under these conditions, a novel intermediate was detected in reaction mixtures after 300 min of incubation (see Fig. S12B). Control assays revealed that the formation of this compound depended on the presence of both CoA (Fig. S12C) and PMS. This compound (P5) had absorption maxima at approximately 215, 260, and 300 nm and its molecular mass (380 Da) was 4 Da lighter (see Fig. S13) than that of the initial substrate DOCDA (XII). When extracts of cholate-grown cells were replaced with extracts of glucose-grown cells, P4 was the main product in the reaction mixture after 300 of incubation (see Fig. S14). In contrast, P3 and P5 were not detected under these conditions (Fig. S14). The absorption maximum at approximately 205 nm (see Fig. S15) suggests that P3 has an additional double bond in the C₅ side chain. The fact that the formation of this double bond required CoA activation of the substrate also supports this suggestion. As in P. stutzeri Chol1, an acyl-CoA dehydrogenase (ACAD) might catalyze the α,β-dehydrogenation of the side chain. The exact position of this potential double bond is unknown, but assuming a β-oxidation-like pathway, it would be expected between C-22 and C-23. Accordingly, P3 was proposed to be 3,12-dioxo-4,6,(22E)-choltrienoic acid (Δ²²-DOCTRA). In agreement with the suggested cholate degradation sequence in strain Chol11, the compound P4 presumably has an additional double bond at the Δ¹ position and was thus identified as 3,12-dioxo-1,4,6-choltrienoic acid (Δ¹-DOCTRA). The high similarity between the absorption spectra of this compound and those of the Δ^1,4,6-compound HATD (XIII) supports this suggestion.

FIG 6.

(A) Transformation of DOCDA (top) in cell extracts of Novosphingobium sp. strain Chol11. In the presence of PMS, CoA, ATP, and Mg²⁺, DOCDA was converted into P3 (middle). In the presence of PMS, DOCDA was converted into P4 (bottom). P3 and P4 are predicted to be Δ²²-DOCTRA [3,12-dioxo-4,6,(22E)-choltrienoic acid] and Δ¹-DOCTRA (3,12-dioxo-1,4,6-choltrienoic acid), respectively. HPLC-MS data are displayed as basic peak chromatograms in negative mode of MS. (B) Chemical structures of DOCDA, Δ¹-DOCTRA, Δ²²-DOCTRA, and DOCTTRA [3,12-dioxo-1,4,6,(22E)-choltetraenoic acid].

In the next step, enzyme assays were performed with purified P4 as the substrate. In the presence of CoA, ATP, and Mg²⁺, most of P4 was activated by SclA with CoA within 15 min (see Fig. S16A and C). When CoA activation assays were performed with cell extracts supplied with recombinant SclA and PMS, small amounts of P5 were detected in the reaction mixture (Fig. S16B), indicating that P4 is the precursor of P5. Control assays revealed that the formation of this compound depended on the presence of CoA (Fig. S16B). On the basis of mass and UV spectra, P5 was proposed to be 3,12-dioxo-1,4,6,(22E)-choltetraenoic acid (DOCTTRA) (Fig. 6B).

DISCUSSION

Novosphingobium sp. strain Chol11 is proposed to have an alternative initial degradation route for 7-hydroxysteroids such as the bile salt cholate (compound I in Fig. 1). This pathway is characterized by the activity of the 7-hydroxysteroid dehydratase Hsh2 as the key enzyme and 3-keto-7-deoxy-Δ^4,6 steroids as key intermediates (20). In this study, we identified SclA from strain Chol11 as a novel steroid CoA ligase, which preferentially activates the C₅ acyl side chain of bile salt degradation intermediates with a 3-keto-7-deoxy-Δ^4,6 structure of the steroid skeleton. In agreement with this function, an sclA deletion mutant had a strong phenotype during growth with cholate. These findings support further the suggested cholate degradation pathway, which proceeds via the 3-keto-Δ^4,6-diene-7-deoxy intermediates HOCDA (XI) and DOCDA (XII) in strain Chol11.

SclA comprises conserved residues for CoA, ATP, and AMP binding sites, indicating an AMP-forming acyl-CoA ligase, which is in agreement with experimental results. Moreover, it contains a facl-like domain, which is characteristic for these enzymes and is also found in the steroid acyl-CoA ligases StdA1 from P. putida DOC21 and CasG from R. jostii RHA1. SclA shares 38% sequence identity with StdA1 and 32% sequence identity with CasG. The substrate specificity of CoA ligases involved in steroid side chain degradation appears to be determined by the length of the side chain (15, 18). While the aforementioned StdA1 and CasG only activate side chains with five carbons, StdA2 and CasI catalyze the CoA activation of side chains with three carbons in P. putida DOC21 and R. jostii RHA1, respectively. In contrast to the influence of the side chain, it is so far unknown to what extent modifications on the steroid skeleton affect the activity of CoA ligases. CasI exhibits different affinities to metabolites with varied configurations of the steroid skeleton (18). StdA1_DOC21 is known to be able to activate both cholate and 3-ketocholate (II) with CoA (15). In cell extracts of P. stutzeri Chol1, Δ^1/4-3-ketocholate and Δ^1,4-3-ketocholate are thioesterified with CoA, probably by StdA1_Chol1, although the in vitro oxidation of cholate to Δ^1,4-3-ketocholate does not require a thioesterification step (22). Our data strongly suggest that a 3-keto-7-deoxy-Δ^4,6 steroid skeleton is the preferred substructure for SclA, indicating that DOCDA (XII) is the physiological substrate of this enzyme in strain Chol11. First, with purified SclA, the largest amounts of CoA thioesters were formed with DOCDA (XII), while only very small amounts of cholyl-CoA and 3-ketocholyl-CoA were formed. Second, CoA ligase assays with cell extracts of strain Chol11 revealed no cholyl-CoA formation, while DOCDA (XII) was activated; this is in agreement with the observation that CoA activation is not required for the initial A and B ring-modifying reactions of cholate metabolism, which form DOCDA (XII) from cholate (20). However, it must be noted that the amounts of CoA ester formed from Δ⁴-3-ketocholate and DOCDA differed only slightly from each other. Therefore, it is possible that the dehydration reaction catalyzed by Hsh2 can also occur after the activation of Δ⁴-3-ketocholate with CoA. In support of this possibility, it has been shown that BaiE, which is a bile acid 7α-dehydratase from Clostridium scindens and shares 38% sequence identity with Hsh2, can use the CoA ester of Δ⁴-3-ketocholate as the substrate (23). According to the crystal structure analyses of BaiE, the CoA moiety does not bind to the active site pocket of the enzyme.

The lack of SclA resulted in a dramatic growth rate decrease and caused the accumulation of two novel but yet unidentified dead-end products. This phenotype supports the hypothesis that a CoA activation step catalyzed by SclA is an essential part of the degradation sequence for bile salts in strain Chol11. However, similar to the deletion of hsh2 (20), the deletion of sclA strongly impaired but did not abolish the growth of strain Chol11 with cholate. This phenotype is in contrast to that of P. stutzeri Chol1, in which the deletion of genes involved in initial reactions of bile salt degradation results in a complete lack of growth (9, 21, 22), and supports the notion that strain Chol11 has a broad metabolic repertoire. Our bioinformatics analysis revealed that the genome of strain Chol11 encodes several potential isoenzymes for some steroid degradation reactions. For example, five putative homologs of KshA, which could act as isoenzymes, are encoded on chromosome 2 of strain Chol11. However, we did not find any gene that could potentially compensate for the loss of sclA. Nevertheless, we cannot rule out the possibility that genes for the so-far-unknown steroid CoA ligases or CoA transferases are among those encoding hypothetical proteins. Alternatively, the loss of SclA might be compensated for by unspecific activities of other CoA ligases or CoA transferases toward substrates of SclA. The reduced growth rate of the ΔsclA mutant supports the second possibility, as unspecific enzymes might plausibly compensate for the loss of sclA, allowing continued cholate metabolism though at a significantly reduced rate.

Despite different complementation strategies, the phenotype of wild-type cells could not be restored in mutant strain Chol11 ΔsclA. A sequencing analysis of regions neighboring the deleted gene ruled out a frameshift. A potential reason for the lack of a complementing effect of plasmid-encoded SclA could be that its expression in strain Chol11 but not in E. coli has some structural requirements, which can only be provided when the gene is in its original genomic context. Nevertheless, the assigned function of SclA was consistent with enzyme activities in cell extracts and with the purified recombinant protein, as well as with the phenotype of the mutant.

Activity of some AMP-dependent acyl-CoA synthetases is proposed to be regulated by the acetylation of a lysine residue, which inhibits the formation of an adenylate intermediate but not the thioesterification of preadenylated intermediates (24). Interestingly, the characterized steroid-24-oyl-CoA ligases, namely, StdA1 and CasG, as well as the putative CoA ligase from P. stutzeri Chol1, encoded by C211_RS11125, contain an acetylation site lysine residue within a highly conserved PX₄GK motif, which is required for this posttranslational modification (25). The acetylation of these CoA ligases may be an important regulatory mechanism for bile salt metabolism in these strains, because it is the first step of the carboxylic side chain degradation. In contrast to these steroid CoA ligases, the PX₄GK motif is not conserved in SclA (see Fig. S17), indicating potential differences in the regulation of bile salt degradation.

It is currently unknown how the side chain of DOCDA is further degraded. In cell extracts, we detected two dehydrogenation reactions acting on DOCDA. While the first one was CoA dependent, the second one was CoA independent. The next plausible step in the side chain degradation of DOCDA-CoA would be the formation of an enoyl-CoA compound catalyzed by an acyl-CoA dehydrogenase. This possibility is supported by the CoA-dependent dehydrogenation of DOCDA (XII) into a compound, whose spectroscopic properties are in agreement with Δ²²-DOCTRA (P3) (Fig. 6B). In other cholate-degrading Proteobacteria, e.g., P. stutzeri Chol1, P. putida DOC21, and C. testosteroni strain TA441, the side chains are completely removed before the B rings can be opened (9, 13, 15). The fact that Chol11 forms the C₁₉ steroid HATD (XIII) as a cholate degradation intermediate, which has no side chain but an intact ring system, suggests that the side chain is also completely removed before the opening of the steroid skeleton in strain Chol11. As such, it may be possible that the deletion of sclA forces the mutant strain to metabolize the steroid skeleton before metabolizing the C₅ side chain. Apart from the aforementioned lower growth rate, the ΔsclA mutant also reached a lower final optical density, indicating a lower molar growth yield. In agreement with that, the accumulation of the dead-end metabolites P1 and P2 in the supernatant of the sclA mutant indicates that the degradation is incomplete. Thus, SclA appears to have an essential function for the optimal growth of strain Chol11 with bile salts.

The CoA-independent dehydrogenation was likely to be catalyzed by a Δ¹-ketosteroid dehydrogenase, which converts DOCDA (XII) into Δ¹-DOCTRA (P4) (Fig. 6B). It is unknown whether Δ¹ and side chain dehydrogenation concomitantly occur in strain Chol11. However, the in vitro conversion of Δ¹-DOCTRA into DOCTTRA (P5) (Fig. 6B) in the presence of CoA indicates that the degradation of side chains can also occur after the introduction of a double bond at the Δ¹ position.

Despite the observation of Δ²²-DOCTRA and DOCTTRA (Fig. 6B) as plausible degradation intermediates, there are neither physiological nor genomic hints of how the side chain is further degraded. A common characteristic of sphingomonads is the complex arrangement of genes involved in degradation pathways, which are frequently scattered in several clusters (26, 27). In agreement with this complex localization, HMMs and reciprocal BLASTp analyses identified three putative steroid degradation gene clusters on chromosome 2 of strain Chol11 encoding several key enzymes presumably mediating steroid ring degradation. Moreover, several candidate genes involved in steroid metabolism were also found outside these cluster regions. However, homologs of known key enzymes involved in side chain degradation were not found. In P. stutzeri Chol1, the removal of acetyl-CoA from the C₅ side chain occurs via aldolytic cleavage (17). The resulting aldehyde intermediate is oxidized and subsequently activated with CoA. There are no aldolase (sal), aldehyde dehydrogenase (sad), or steroid-22-oyl-CoA ligase (sdtA2) homologs present in the genome of strain Chol11. Nevertheless, the removal of the side chain may be very similar to that in P. stutzeri strain Chol1, but the involved enzymes may not resemble those known from P. stutzeri Chol1 or other well-studied strains. In agreement with this, we could not find homologs for the characteristic heteromeric acyl-CoA dehydrogenases, which are responsible for desaturation of carboxylic side chains of steroids in other bacteria, such as P. stutzeri Chol1 (9) and Mycobacterium tuberculosis (28, 29), indicating that the formation of Δ²²-DOCTRA must have been catalyzed by a so-far-unknown enzyme system.

Another explanation for the lack of known genes for side chain degradation in strain Chol11 might be that the actual reactions involved in this process are different from the known reactions for side chain removal in other bacteria. For example, a thiolytic cleavage of acetyl-CoA catalyzed by a β-keto-thiolase would directly form a CoA-activated C₃ side chain intermediate and would not require an additional steroid-22-oyl-CoA ligase. Accordingly, we did not observe any CoA ligase activity toward DHOPDC (VIII) harboring a C₃ carboxyl side chain in cell extracts of cholate- or DHOPDC-grown strain Chol11.

In conclusion, this study further supports the hypothesis that strain Chol11 has an alternative pathway for the degradation of 7-hydroxy bile salts and adds further evidence of the diversity of metabolic pathways for the degradation of steroid compounds in environmental bacteria. The knowledge about these pathways might be useful for exploring the fate of pharmaceutical steroids in the environment, some of which may act as endocrine disruptors (30).

MATERIALS AND METHODS

Bacterial strains, growth media, and growth experiments.

Novosphingobium sp. strain Chol11 was grown in the HEPES-buffered mineral medium B (MB) (31) with 1 mM cholate as described previously (8). Mutant strain Chol11 ΔsclA was grown in the same medium with 15 mM glucose or 1 mM cholate. Mutant strain Chol11 ΔsclA with plasmid pBBR1MCS-5, pBBR1MCS-5::sclA, or pBBR1MCS-5::sclA+P was grown in the same medium with 15 mM glucose in the presence of 60 μg · ml⁻¹ gentamicin. Strain Chol11 Tn50KL was initially grown in phosphate-buffered mineral medium (MMChol) (32) with 15 mM propionate and 75 μg · ml⁻¹ kanamycin; later, MMChol1 was replaced by MB for all further experiments with strain Chol11 and its mutants. Escherichia coli was grown in lysogeny broth (LB) at 37°C and 200 rpm. LB was supplemented with gentamicin (30 μg · ml⁻¹) for ST18 strains carrying pBBR1MCS-5 and pBBR1MCS-5::sclA, with chloramphenicol (30 μg · ml⁻¹) for strain ST18(pDM4::230UpDown), and with kanamycin (50 μg · ml⁻¹) for E. coli Tuner(pET28b::sclA) and E. coli S17-1 λpir pUT(mini-Tn5Km1) (33–35). Growth media for E. coli ST18 strains, which are auxotrophic for 5-aminolevulinic acid, were additionally supplemented with 50 μg · ml⁻¹ of this substance (36). Bacto agar (1.5% [wt/vol]; BD, Sparks, USA) and growth substrate were added to the aforementioned media for preparing respective solid media, on which each strain was maintained and transferred weekly. Growth substrates were 1 mM cholate (for strain Chol1), 15 mM propionate (for strain Chol11 Tn50KL), and 15 mM glucose (for strain Chol11 ΔsclA).

All growth experiments were performed in 10-ml test tubes containing 3 to 5 ml of the respective media at 30°C with orbital shaking at 200 rpm (Minitron; Infors HT, Einsbach, Germany). Growth was followed by measuring the optical density at 600 nm (OD₆₀₀) with a test tube photometer (Camspec M107; Spectronic Camspec, United Kingdom). For precultures of strain Chol11, test tubes containing 3 to 5 ml medium with 1 mM cholate were seeded with the respective strains from agar plates. Precultures of strain Chol11 Tn50KL were grown with 15 mM propionate in the presence of 75 μg · ml⁻¹ kanamycin. Precultures of strain Chol11 ΔsclA were grown with 15 mM glucose in the presence of 1 mM cholate. Precultures of strain Chol11 ΔsclA with plasmid pBBR1MCS-5, pBBR1MCS-5::sclA, or pBBR1MCS-5::sclA+P were grown with 15 mM glucose in the presence of 60 μg · ml⁻¹ gentamicin. All main cultures were inoculated from precultures in late exponential phase to an OD₆₀₀ of 0.01 to 0.02 and contained 1 mM bile salt (cholate or deoxycholate) or 1 mM cholate degradation intermediates DHOPDC (VIII) or 12β-DHADD (IX) as carbon and energy sources.

Transposon mutagenesis and localization of transposon in genome.

For transposon mutagenesis, the suicide vector pUT(mini-Tn5Km1) (34) was mobilized into Novosphingobium sp. strain Chol11 by biparental mating with donor strain E. coli S17-1 λpir. For this, main cultures of strains ST18 and Chol11 were grown as described above. In late exponential phase, 1 × 10⁹ cells of strain S17-1 λpir and 3 × 10⁹ cells of strain Chol11 were harvested by centrifugation at 7,300 × g for 8 min, washed with 500 μl LB, and resuspended in 50 μl LB. Suspended cells were mixed by gentle pipetting and spread onto LB agar plates. After incubating at 30°C for 4 h, cells were resuspended in 2 ml of 0.9% NaCl, spread onto LB agar plates containing 75 μg · ml⁻¹ kanamycin, and incubated at 30°C for 7 days.

For localizing the transposon in the chromosome of the transposon mutant strain Chol11 Tn50KL, a genomic library of this mutant strain was established. For this, genomic DNA of strain Chol11 Tn50KL was purified with the Puregene tissue core kit B (Qiagen) and partially digested with SalI (FasDigest; Thermo Fisher) at 37°C for 2 h. The resulting fragments were ligated into the SalI restriction site of the E. coli shuttle vector pNV18Sm (37) using standard methods. The resulting plasmids were transformed into E. coli DH5α. For the selection of plasmids containing genomic DNA fragments with the transposon, transformed cells were spread onto LB agar plates containing 50 μg · ml⁻¹ kanamycin and incubated at 37°C overnight. Kanamycin-resistant colonies were analyzed by PCR using primer pair C/D (Table 1), which is specific for the mini-Tn5 transposon. Plasmids from a positive colony were purified and used as the template DNA for sequencing the respective genomic fragments of strain Chol11. For this, primer pairs A/B (specific for pNV18Sm) and C/D (specific for the mini-Tn5 transposon) were used. Sequencing was performed by Eurofins Genomics (Ebersberg, Germany). For sequencing the remaining parts of the genomic fragment of Chol11, the primer pair E/F was used. The obtained sequences were compared with those from genomic DNA of strain Chol11 to localize the transposon.

TABLE 1.

Oligonucleotides used in this study

Designation	Sequence	Description
A	5′-AGGGTTTTCCCAGTCACGACGTT	M13-f
B	5′-GAGCGGATAACAATTTCACACAGG	M13-rev
C	5′-CATTACGCTGACTTGACGGGAC	Tn5 mini-f
D	5′-ATCTTGTGCAATGTAACATCAGAG	Tn5 mini-r
E	5′-TATGGCATGACCGAGACATCG	gen-outward-f
F	5′-AGAGGGTGCGTTACATCGAT	gen-outward-r
G	5′-CCATGCGGAACTCTCCTGTGACTCTATTCCGCCGCCCGC	nov2c230-up-f
H	5′-TTTTTTCTAGAGCCCCGGAATGATGGCATTG	nov2c230-up-r
I	5′-TTTTTTCTCGAGACCTTGATGGCGTTTTCACG	nov2c230-down-f
J	5′-ACAGGAGAGTTCCGCATGG	nov2c230-down-r
K	5′-TTTTTTAAGCTTGATCAGCGGACGAAAGGGAT	sclA+promoter-f
L	5′-TTTTTTCCATGGCAATCACGTTGTTGGCCCAG	sclA+promoter-r
M	5′-TTTTTTTCTAGATGTTCAGTTGTTGCCGGCCA	sclA-f
N	5′-TTTTTTGAATTCCGGCGGAATAGAGTCGTGTT	sclA-r
O	5′-TTTTTTCCATGGTGTTCAAGCAGAACGGCGAT	nov2c230-overex-f
P	5′-TTTTTTAAGCTTGTTGTTGCCGGCCAGTTCAG	nov2c230-overex-r

Construction of the unmarked deletion mutant Chol11 ΔsclA.

To construct an unmarked deletion of sclA, two PCR products spanning parts of the up- and downstream regions of the gene were obtained from genomic DNA of strain Chol11 using the primer pairs G/H and I/J (Table 1). The resulting fragments were used as the templates for a second splicing-by-overlap extension PCR (SOE-PCR) (38) with the primer pair H/I. The product of this second PCR was digested with XhoI and XbaI and ligated into the corresponding sites of the suicide vector pDM4 (34). The resulting plasmid was mobilized into strain Chol11 as described previously (20). Transconjugants were streaked twice onto LB agar plates containing 50 μg · ml⁻¹ chloramphenicol before they were transferred onto LB agar plates containing 7% sucrose for the selection of vector excision by a second crossover. Colonies were screened for gene deletion by PCR using primer pair H/I. Positive colonies were streaked twice onto LB agar plates containing 7% sucrose. Finally, the genomic DNA of mutant strain Chol1 ΔsclA was isolated and used as the template for PCR amplification with primer pair H/I, and the resulting DNA fragment was sequenced to confirm the deletion.

For complementation of mutant strain Chol11 ΔsclA, fragments containing the open reading frame (ORF) of sclA with or without a putative promoter region were amplified using primer pairs K/L and M/N, respectively, and genomic DNA from strain Chol11 as the template. The resulting fragments (sclA+P and sclA) were digested with XbaI and EcoRI or HindIII and NcoI, respectively, and ligated into the corresponding sites of pBBR1MCS-5. The resulting plasmids pBBR1MCS-5::sclA and pBBR1MCS-5::sclA+P were introduced into strain Chol11 ΔsclA by biparental mating as described above. Transconjugants were selected on solid minimal medium B with 15 mM glucose and 60 μg · ml⁻¹ gentamicin.

Overexpression of SclA.

A fragment containing the ORF of sclA was amplified using primer pair O/P and genomic DNA of Chol11 as the template, and was digested with NcoI and HindIII and ligated into the corresponding sites of the pET28b expression vector. The resulting plasmid pET28b::sclA was transformed into E. coli Tuner(DE3), and plasmid-harboring strains were selected on LB agar plates with 50 μg · ml⁻¹ kanamycin. To produce polyhistidine (His)-tagged SclA, three 500-ml cultures with LB and kanamycin (20 μg · ml⁻¹) were inoculated with precultures of strain Tuner(pET28b::sclA) grown in LB with kanamycin (20 μg · ml⁻¹) at 37°C and 200 rpm and were subsequently incubated at 25°C and 160 rpm. To induce protein expression, 4 mM isopropyl-β-d-thiogalactopyranoside (IPTG; Carl Roth, Karlsruhe, Germany) was added after 2 h of incubation. At an OD₆₀₀ of approximately 3 (after 16 of incubation), the cultures were harvested by centrifugation at 7,300 × g for 10 min at 4°C and washed with 20 mM Tris buffer (pH 8) containing 150 mM NaCl.

After resuspending the pellets in the same buffer, cells were disrupted by sonication (Hielscher, Teltow, Germany) on ice twice for 3 min (amplitude, 60%; pulse cycle, 0.6) followed by centrifugation at 17,000 × g for 30 min at 4°C. Supernatants were filtered through a 0.2-μm syringe filter. Cell-free lysates were loaded onto 10-ml Ni²⁺-nitrilotriacetate agarose columns equilibrated with 20 mM Tris buffer (pH 8.0) containing 150 mM NaCl and 10 mM imidazole.

His-tagged proteins were eluted with 20 to 30% elution buffer containing 20 mM Tris buffer (pH 8.0), 150 mM NaCl, and 400 mM imidazole. Protein-containing fractions were pooled, washed, and concentrated in 50 mM 3-morpholinopropane-1-sulfonate (MOPS) buffer (pH 7.8) using a centrifugal concentrator (Vivaspin 20 protein concentrator; polyethersulfone [PES] membrane; molecular weight cutoff [MWCO], 10,000; Sartorius Stedim Biotech, Göttingen, Germany). The purified proteins were supplemented with 10% glycerol and stored at −80°C.

Preparation of cell extracts and protein determination.

For preparing cell extracts, 100- to 500-ml cultures (in Erlenmeyer flasks with baffles) of strains Chol11 and the mutant Chol11 ΔsclA were grown with 1 mM cholate and with 15 mM glucose in the presence of 1 mM cholate, respectively, at 30°C and 130 rpm. Cultures were harvested in mid-exponential growth by centrifugation at 8,800 × g for 10 min at 4°C. The resulting pellets were washed by resuspending in 15 ml 10 mM MOPS buffer (pH 7.8) and were finally resuspended in 1.5 ml 50 mM MOPS (pH 7.8). Cells were disrupted by ultrasonication on ice for 10 min (amplitude, 60%; pulse cycle, 0.5) with intermittent incubations on ice every 4 min. Subsequently, cell debris was removed by centrifugation at 15,000 × g for 30 min at 4°C.

Protein concentrations from cell extracts and purified proteins were determined using the bicinchoninic acid (BCA) assay (Pierce, Thermo Scientific, Rockford, IL, USA) using bovine serum albumin as the standard. The concentrations of purified proteins were determined by absorption measurements at 280 nm using a Nanophotometer (Implen, Munich, Germany).

Enzyme assays.

The reaction mixtures for the enzyme assays were buffered with 50 mM MOPS (pH 7.8) and incubated on a shaker at 30°C and 450 rpm. Samples were withdrawn at defined time intervals and analyzed immediately by HPLC-MS. For CoA activation assays with cell extracts, the reaction mixtures were supplemented with 0.5 to 1 mM cholate (I in Fig. 1), 3-ketocholate (II), Δ⁴ 3-ketocholate (III), DOCDA (XII), or DHOPDC (VIII) as the substrate. For the assays with purified recombinant SclA, the reaction mixtures were supplemented with deoxycholate or DHOCTO (V) as the substrate in addition to the above-mentioned substrates. The reaction mixtures were also supplied with 2 mM CoA, 2 mM ATP, and 3 mM Mg²⁺ and started by the addition of cell extracts or recombinant SclA to final concentrations of 0.7 to 1.0 and 0.3 mg protein · ml⁻¹, respectively. The amount of DOCDA-CoA formed after 45 min was set to 100%, and CoA ester formation with other substrates was compared to this value. For the investigation of side chain degradation, cell extracts of Novosphingobium sp. strain Chol11 were incubated with 1 mM DOCDA (XII) in the presence of 2 mM CoA, 2 mM ATP, 3 mM Mg²⁺, 2 mM NAD⁺, and 0.25 mM phenazine methosulfate (PMS).

For the evaluation of the HPLC-MS results, the base peak chromatograms or extracted ion chromatograms at defined mass ranges were analyzed.

Preparation of steroid compounds.

The steroid compounds DHOCTO (V), DHOPDC (VIII), and 12β-DHADD (IX) were produced with P. stutzeri Chol1 or its mutants as described previously (17). To produce Δ⁴-3-ketocholate (III), cell extracts of Chol11 Δhsh2 (∼0.7 mg protein · ml⁻¹ final concentration in the reaction mixture) were incubated with 2.5 mM 3-ketocholate (II) (Steraloids, RI, USA) in the presence of 0.25 mM PMS at 30°C for 5 h. Subsequently, the reaction mixture was incubated at 99°C for 30 min to inactivate enzymes, was acidified to pH 3.5 with 25% HCl, and was extracted three times with ethyl acetate. After drying over MgSO₄, the solvent was evaporated under vacuum at 80°C and the extract was dried in a drying oven at 80°C for ca. 14 h. Finally, Δ⁴-3-ketocholate (III) was dissolved in 50 mM MOPS-buffer (pH 7.8) and stored at −20°C until use. To produce DOCDA, cell extracts of Novosphingobium sp. strain Chol11 (1 mg protein · ml⁻¹ final concentration in reaction mixtures) were incubated with 2 mM cholate in the presence of 5 mM NAD⁺ and 0.25 mM PMS overnight. The purification of DOCDA (XII) was achieved by following the same procedure described above for Δ⁴-3-ketocholate (III). To produce P4 (Δ¹-DOCTRA), cell extracts of Novosphingobium sp. strain Chol11 (1 mg protein · ml⁻¹ final concentration in reaction mixtures) were incubated with 0.4 mM DOCDA in the presence of 0.1 mM PMS and 2 mM NAD⁺ overnight. After the organic extraction as described above, extracts were fractionated using a semipreparative HPLC as described below. After pooling the collected fractions containing P4, purification was achieved by organic extraction as described above. The concentrations of purified metabolites were measured with a double-beam photometer (UV-2600; Shimadzu, Kyoto, Japan). For estimating the concentrations of purified Δ⁴-3-ketocholate (III) and Δ^1,4-3-ketocholate (IV), a molar extinction coefficient of 14.7 cm⁻¹ · mM⁻¹ (λ_{245 nm}) was used (17). For DOCDA (XII) and P4, the molar extinction coefficients of 21.1 cm⁻¹ · mM⁻¹ (λ_{290 nm}) and 13.22 cm⁻¹ · mM⁻¹ (λ_{300 nm}), respectively, were used (39). The purity of metabolites was assessed by HPLC-MS analysis.

HPLC-MS analysis.

All steroid compounds and culture supernatants were analyzed with an HPLC-MS system consisting of a Dionex Ultimate 3000 HPLC (Thermo Fisher Scientific) with a UV-visible light diode array detector and an ion trap mass spectrometer (Amazon speed; Bruker, Bremen, Germany) with an electrospray ion source (ESI).

For analyzing bile salts and their degradation intermediates and for the fractionation of extracts containing P4, a reversed-phase C₁₈ column (150 mm by 3 mm, Eurosphere II, 100-5 C₁₈; Knauer) at 25°C was used. Ammonium acetate buffer (10 mM, pH 6.7, eluent A) and acetonitrile (eluent B) were used as eluents with a flow rate of 0.3 ml · min⁻¹. For the detection of cholate and its degradation intermediates, a gradient method was used starting with 10% eluent B for 2 min, increasing to 48% eluent B within 25 min, and returning to 10% eluent B within 1 min, followed by an equilibration of 5 min at a flow rate of 0.4 ml · min⁻¹. For analyzing ATP and AMP in reaction mixtures, a polar reversed-phase C₁₈ column (250 mm by 4.6 mm by 4 μm; Phenomenex Synergi) was used. For the detection of ATP and AMP, a gradient method with a flow rate of 0.45 ml · min⁻¹ was used starting with 0% eluent B for 5 min, increasing to 70% eluent B within 10 min, keeping the concentration of eluent B at 70% for 5 min, and returning to 10% eluent B for 1 min, followed by an equilibration of 6 min.

Ionization of samples was performed at alternating the ionization mode of ESI with the following settings: capillary voltage, 4,000 V; plate offset, 500 V; nebulizer pressure, 22.5 lb/in²; dry gas flow, 12 liters · min⁻¹; dry gas temperature, 200°C. MS was operated in ultrascan mode in a scan range of 50 to 1,000 Da. For the evaluation of measurements, base peak chromatograms (BPC) or extracted ion chromatograms (EIC) with defined masses or mass ranges were used.

Bioinformatics analyses.

Protein sequences were downloaded from the National Center for Biotechnology Information (NCBI) databases. Multiple sequence alignments were performed using the Clustal Omega software (version 2.3). To identify potential steroid degradation proteins in strain Chol11, we used a set of 67 recently published hidden Markov models (HMMs) (https://github.com/MohnLab/mohn_lab_steroid_degradation_hmm_analysis_2015 [40]) representing 25 steroid degradation proteins. The proteome of strain Chol11 was searched using the program hmmsearch from the HMMER software (v3.1b1 [http://hmmer.org]) applying a maximum E value of 10⁻²⁵ and a minimum coverage of 30%. The R package genoPlotR (v 0.8.4 [41]) was used for analyzing the localization of predicted steroid degradation genes in the genome of strain Chol11.

To compare potential steroid degradation proteins in strain Chol11 with characterized and hypothetical steroid degradation proteins encoded in the steroid degradation gene clusters of P. stutzeri Chol1 (42), C. testosteroni CNB-2 (43), and R. jostii RHA1 (44), we performed a reciprocal BLASTp analysis using the program BackBLAST (v1.0 [44]). BLASTp analysis was performed using a maximum E value of 10⁻³⁰ and a minimum identity of 25%, leaving all other BLAST settings at the default.