Identification and Characterization of a Novel Translational Repressor of the Steroid-inducible 3α-Hydroxysteroid Dehydrogenase/Carbonyl Reductase Gene in Comamonas testosteroni*

Comamonas testosteroni 3α-hydroxysteroid dehydrogenase/carbonyl reductase (3α-HSD/CR) is a key enzyme in the degradation of steroid compounds in soil and may therefore play a significant role in the bioremediation of hormonally active compounds in the environment. The enzyme is also involved in the degradation of the steroid antibiotic fusidic acid. In addition, 3α-HSD/CR mediates the carbonyl reduction of non-steroidal aldehydes and ketones. Because the gene of 3α-HSD/CR (hsdA) is inducible by steroids, we were interested in the mode of its molecular regulation. Recently, we could identify the first molecular determinant in procaryotic steroid signaling, i.e. a repressor protein (RepA), which acts as a negative regulator by binding to upstream operator sequences of hsdA, thereby blocking hsdA transcription. In this work, we identified and cloned a second novel regulator gene that we named repB. The gene locates 932 bp downstream from hsdA on the C. testosteroni chromosome with an orientation opposite to that of hsdA. The open reading frame of repB consists of 237 bp and translates into a protein of 78 amino acids that was found to act as a repressor that regulates hsdA expression on the translational level. Northern blot analysis, UV-cross linking, gel-shift assays, and competition experiments proved that RepB binds to a 16-nucleotide sequence downstream of AUG at the 5′ end of the 3α-HSD/CR mRNA, thereby blocking hsdA translation. Testosterone, on the other hand, was shown to specifically bind to RepB, thereby yielding the release of RepB from the 3α-HSD/CR mRNA such that hsdA translation could proceed. Data bank searches with the RepB primary structure yielded a 46.2% identity to the regulator of nucleoside diphosphate kinase, a formerly unknown protein from Escherichia coli that can restore a growth defect in alginate production in Pseudomonas aeruginosa. In conclusion, the induction of hsdA by steroids in fact is a derepression where steroidal inducers bind to two repressor proteins, RepA and RepB, thereby preventing blocking of hsdA transcription and translation, respectively.

Microorganisms capable of utilizing various naturally occurring steroids, including cholesterol and the phytosterols ␤-sitosterol, stigmasterol, and campestrol, as carbon and energy sources are relatively widespread in nature (1)(2)(3)(4). Complete assimilation of these substrates is achieved through an adaptive complex metabolic pathway involving many enzymatic steps of oxidation responsible for the breakdown of the steroid nucleus (3)(4)(5)(6). Since the isolation of Comamonas testosteroni from various soil samples, some work has emerged on the steroid-inducible characteristics of this organism (7)(8)(9)(10). As a result, several steroid-metabolizing enzymes as well as steroid binding or transport activities have been described and characterized (5,(11)(12)(13)(14)(15)(16). Since the pioneering work of Talalay and co-workers (7,17), it is well known that 3␣-hydroxysteroid dehydrogenase is one of the first enzymes of the steroid-catabolic pathway and therefore plays a central role in steroid metabolism. However, the exact mechanism of regulation of this important enzyme by steroids has long been unknown.
In previous investigations, the gene encoding 3␣-hydroxysteroid dehydrogenase/carbonyl reductase (3␣-HSD/CR) 1 (hsdA) was cloned and overexpressed in Escherichia coli (18). According to its primary structure, the enzyme was classified as belonging to the short chain dehydrogenase/reductase superfamily (18). The enzyme was found to be functional as oxidoreductase toward a variety of steroid substrates including the steroid antibiotic fusidic acid (19). The enzyme also catalyzes the carbonyl reduction of non-steroidal aldehydes and ketones such as a novel insecticide (NKI 42255) and has therefore been named 3␣-HSD/CR (19). It is suggested that 3␣-HSD/CR contributes to important defense strategies of C. testosteroni against natural and synthetic toxicants in addition to its functional role in the steroid degradation pathway for carbon supply (19). The crystal structure of 3␣-HSD/CR has been solved (20), revealing that the enzyme is active as a homodimer, which shows an unusual dimerization pattern among the short chain dehydrogenase/reductases (21)(22)(23).
Interestingly, 3␣-HSD/CR from C. testosteroni has been found to be inducible by testosterone and progesterone (24), but the mechanism of steroid-dependent gene regulation in procaryotes remained obscure. Recently, we reported the first molecular determinant of steroid signaling in C. testosteroni. By investigating the cis-and trans-acting elements of hsdA ex-* This work was supported by grants from the European Community and the Deutsche Forschungsgemeinschaft (SFB 395) (to E. M.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AY395575.
‡ To whom correspondence should be addressed. Tel.: 49-431-597-3540; Fax: 49-431-597-3558; E-mail: maser@toxi.uni-kiel.de. pression in C. testosteroni, we characterized two palindromic operator domains upstream of hsdA and identified a new gene coding for a trans-acting negative regulator (repressor A, RepA) of hsdA expression (25). We proved the specific interaction among RepA, testosterone, and the operator domain and found that hsdA is under negative transcriptional control by RepA (25).
In the present work, we identified and cloned a second regulator gene, which we named repB. We show that the novel regulator protein RepB under non-induced conditions acts as a repressor by binding to the 3␣-HSD/CR mRNA, thereby blocking 3␣-HSD/CR translation. The presence of testosterone resulted in the liberation of the 3␣-HSD/CR mRNA such that translation could proceed. According to our results, the induction of hsdA by steroids in fact is a derepression where steroidal inducers bind to two repressor proteins, RepA and RepB, thereby preventing the blocking of hsdA transcription and translation, respectively.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Plasmids-Host strains E. coli HB101 (Promega) and C. testosteroni ATCC 11996 (Deutsche Sammlung fü r Mikroorganismen) were used for cloning and gene expression. Subcloning of fragments was carried out in plasmids pBBR1MCS-2 (containing the kanamycin resistance gene, a gift from Peterson and co-workers (26)) and pUC18 (containing the ampicillin resistance gene and obtained from Invitrogen). Plasmid pK18 containing the kanamycin resistance gene was a gift from Ciba Pharmaceuticals, Inc., Department of Biotechnology (Basel, Switzerland). The plasmid copy numbers determined were 80 copies of pK18 and pUC18 and 5 copies of pBBR1MCS-2/cell in E. coli. For overexpression and purification of RepB, E. coli strain BL21(DE3)pLysS together with plasmid pET15b from Novagen was used. For 3␣-HSD/CR mRNA preparation, plasmid pSPT18 from Roche Applied Science was employed. The tac promoter (280 bp) was obtained by BamHI digestion from plasmid pHA10, which was a gift from H. Arai et. al. (27).
Growth Media and Growth Conditions-Bacterial cells were grown in a shaker (180 rpm) in Standard I Nutrient broth medium (Merck) or LB medium at 37°C (E. coli) or at 30°C (C. testosteroni). Growth media contained 60 g/ml ampicillin and 30 g/ml kanamycin.
Restriction Enzymes and Other Reagents-Restriction enzymes, T4 ligase, S1 nuclease, and shrimp alkaline phosphatase were obtained from Roche Applied Science, New England Biolabs, MBI, Promega, and Amersham Biosciences, respectively, and used according to the manufacturers' instructions. Ampicillin and kanamycin were from AGS (Heidelberg, Germany).
DNA Manipulations-Recombinant DNA work was carried out following standard techniques according to Sambrook et al. (28). The fragments cloned in this work are shown in Fig. 1. All of the primers and oligonucleotides were prepared by MWG (Ebersberg, Germany).
Transformation of Bacteria-All of the constructs were verified by restriction enzyme analysis of the resulting band patterns. Plasmids were purified with the Tip-100 kit from Qiagen. Ligated constructs were transferred into competent E. coli or C. testosteroni cells prepared by the calcium chloride or electroporation method. Double plasmid cotransformations were performed by exploiting the kanamycin resistance gene of pBBR1MCS-2 or pK18 and the ampicillin resistance gene of pUC18. In these cases, both antibiotics were added to the culture medium. Plasmid isolation and agarose gel electrophoresis were performed to prove successful double transformations.
Generation and Subcloning of hsdA Gene Regulatory Elements-To localize the coding region and to determine the orientation of the novel repressor gene, restriction fragments were prepared as shown in Fig. 1. A 5.257-kb EcoRI fragment of C. testosteroni chromosomal DNA was subcloned into pBBR1MCS-2 with pairwise XmaI and EcoRI or AvrII and EcoRI digestions, resulting in the generation of pBBX1 and pB-BAX7. Corresponding fragments with AvrII and EcoRI or NdeI and EcoRI pairwise digestions were subcloned into pUC18 to produce pAX1, p6N4, and p1N3. Digestion of p6N4 with PstI yielded p6NP1. Kpn2I and NdeI were used to subclone the repA gene fragment into pK18. For silencing repressor RepA, a frameshift mutation was performed by insertion of an additional base at the translational start point of repB to yield pRM4. The generation of pKAN10 was performed as described previously (25).
Subcloning of the repB Repressor Gene-Subcloning of the translational repressor gene repB and the preparation of respective plasmids is shown in Fig. 1. Standard PCR was used to prepare a repB-containing fragment. For transforming C. testosteroni cells with repB, plasmids pBB6N5 and pBB6NP1 were generated from pBBR1MCS-2. The cultures were incubated in 3 ml of kanamycin Standard I Nutrient medium with 27 g/ml testosterone at 30°C overnight. The fragment was cloned into pET15b (Novagen) with XhoI and BamHI to yield pETrepB. The repB sequence was checked by MWG. For preparing mRNA of 3␣-HSD/CR, the fragment was digested with AvrII and EcoRI and was cloned into pSPT18 (Roche Applied Science) to yield pSPT8 (Fig. 1).
Protein Extraction-The probes for 3␣-HSD/CR ELISA detection were prepared by lysozyme digestion of 3 ml of bacterial culture and subsequent centrifugation at 13,000 ϫ g for 10 s. The pellet was washed three times with 1 ml of phosphate-buffered saline and resuspended in 200 l of phosphate-buffered saline with 1 g/ml lysozyme. To complete cell lysis, the suspension was frozen (Ϫ20°C, three times). Finally, the samples were centrifuged again at 13,000 ϫ g for 20 min. The supernatant was diluted to 1 mg/ml protein and used for ELISA and protein determinations. Protein solutions for gel-shift mobility assays were prepared as follows. 2 ml of culture were centrifuged at 13,000 ϫ g for 10 s, and the resulting pellet was washed with 1 ml of TEN (0.01 M Tris-HCl, 0.001 M EDTA, 0.1 M NaCl, pH 8.0) buffer (three times). After resuspending the cell pellet in 100 l of TEN buffer, the samples were frozen (Ϫ20°C, three times) and finally centrifuged (13,000 ϫ g, 20 min). The supernatant was used for gel-shift mobility assays.
Activity Determination of 3␣-HSD/CR by HPLC-3␣-HSD/CR enzyme activity was assessed by HPLC as described previously (16). Protein concentration was measured by the method of Lowry et al. (29).
ELISA of 3␣-HSD/CR-To quantify 3␣-HSD/CR protein expression, an ELISA was established and respective antibodies were generated (25). Rabbit antibodies directed against 3␣-HSD/CR from C. testosteroni were prepared according to standard methods. ELISA plates were coated with protein extracts containing 3␣-HSD/CR in coating buffer. After washing, antibodies against 3␣-HSD/CR were added in a 1:10,000 dilution. The further procedure corresponded to that of the chloramphenicol acetyltransferase ELISA kit from Roche Diagnostics. Comparison of the signals detected by ELISA and 3␣-HSD/CR activity measured by HPLC (18,25) always showed a good correlation.
Overexpression and Purification of Recombinant RepB-The bacterial overexpression of RepB was performed in E. coli strain BL21(DE3)pLysS (Novagen), and the recombinant was protein purified by its His tag sequence. Cells transformed with plasmid pETrepB were grown at 37°C in a shaker (250 rpm), and maintenance of plasmids was ensured by adding 100 g/ml ampicillin to the culture media. 8 ml of a culture grown overnight was used to inoculate 400 ml of fresh medium. At an A 600 of 0.6, expression was induced by the addition of isopropyl-␤-D-thiogalactoside to a final concentration of 1 mM. After 3 h, cells were sedimented and washed once in 50 mM Tris⅐HCl, 100 mM sodium chloride, pH 8.0. The cell pellet was either stored at Ϫ80°C for further usage or directly suspended in 15 ml of ice-cold buffer A (20 mM sodium phosphate, 500 mM sodium chloride, 10 mM imidazole, 6 M urea, pH 7.4). Cells were lysed by sonication (4 ϫ 10 s, ice cooling), and the resulting mixture was centrifuged (1 h, Beckman 50.2Ti rotor, 40,000 rpm, 4°C). 2 ml of the supernatant were applied to a 1 ml of nickel-agarose affinity column (HisTrap TM Pharmacia Biotech). After a 10-ml wash with buffer A, a step gradient of 25, 60, and 100% buffer B (20 mM sodium phosphate, 500 mM sodium chloride, 500 mM imidazole, 6 M urea, pH 7.4) at 5 ml each was applied (Ä KTA TM Purifier, Amersham Biosciences). RepB eluted from the column in the 60% step corresponding to an imidazole concentration of 316 mM. Samples containing pure protein as assessed by SDS-PAGE were pooled and dialyzed against phosphatebuffered saline (50 mM sodium phosphate, 150 mM sodium chloride, pH 7.4).
Messenger RNA Mobility Shift Assays-For the mRNA mobility shift assays, digoxigenin-11-UTP-labeled 3␣-HSD/CR mRNA was prepared using the SP6/T7 transcription kit according to the manufacturer's instruction (Roche Diagnostics). 8 g of ClaI-linearized pSPT8 was used as DNA template (Fig. 1). The linear DNA (2.4 pM) was incubated with 4 l of each ATP, CTP, and GTP as well as with 2.6 l of UTP, 1.4 l of digoxigenin-11-UTP, 4 l of buffer, 2 l of RNase inhibitor, 1 l of Sp6 RNA polymerase, and 9 l of H 2 O from the SP6 transcription kit (Roche Applied Science) at 37°C for 2 h. Two units of DNase I (RNase-free) were added and incubated at 37°C for 15 min. The mixture was heated at 65°C for 5 min, and 4 l of 0.2 M EDTA were added to stop the reaction. The resulting mRNA was purified twice by ethanol precipitation.
For the mRNA mobility shift assays, an amount of 15 ng of labeled mRNA/lane was incubated at 37°C for 10 min with 1 g of purified RepB in 1ϫ binding buffer (from Roche Kit). After the incubation, mixtures were separated by 4% SDS-PAGE and blotted on nylon mem-branes and the labeled RNA was visualized by chemiluminescence using the digoxigenin luminescence detection kit (Roche Diagnostics).
In Vitro Binding Assays and UV-cross-linking-Binding assays with undigested digoxigenin-labeled 3␣-HSD/CR mRNA and RepB consisted of 15 ng of RNA and 1 g of purified RepB in a volume of 20 l of 1ϫ binding buffer (from Roche kit). After a 10-min incubation at 37°C, the mixtures were separated by 4% SDS-PAGE and blotted on nylon membranes and labeled RNA was visualized by chemiluminescence using the digoxigenin luminescent detection kit.
Binding assays for UV-cross-linking contained 32 P-labeled 3␣-HSD/CR mRNA prepared by using the SP6/T7 transcription kit (Roche Diagnostics) and 1 g of purified RepB. The reaction mixtures were incubated for 5 min at 37°C in 1ϫ binding buffer (from Roche kit). Different amounts of RNase T1 were then added and incubated at 37°C for 15 min. The binding reactions were transferred into a 96-well plate and irradiated at 254 nm for 15 min at a distance of 3 cm. Residual RNA was digested by the addition of RNase A (100 g/ml) and incubation at 37°C for 15 min. The reaction mixtures were supplemented with SDSloading buffer, boiled for 3 min, and separated by SDS-PAGE.
Homology Search and Sequence Alignments-The BLAST program (32) was used to screen protein and DNA databases for proteins that shared sequence similarities. Amino acid sequences were aligned using the multiple sequence alignment option of ClustalX, version 1.8.

RESULTS
Identification of a Gene Coding for a Negative Regulator of 3␣-HSD/CR Expression-For the identification of the gene product of repB as a repressor of hsdA expression, C. testosteroni cells were transformed with plasmids pBB6N5 and pBB6NP1 (Fig. 1). Compared with the control vector pBBR1MCS-2, a strong repression of hsdA expression occurred (Fig. 2). For further characterization of RepB, co-transformation of E. coli was performed with plasmids containing the hsdA gene (pBBX1, pBBAX7) and the repB gene (p6N4, p6NP1, p1N3) (Fig. 1). Plasmids pBBX1 and pBBAX7 are descendants of the vector pBBR1MCS-2, which contains a kanamycin resistance gene, whereas p6N4, p6NP1, and p1N3 are descendants of the vector pUC18, which contains an ampicillin resistance gene. Supplementing the growth medium with both antibiotics ensured successful double transformations.
To determine the orientation of repB, the inserts were cloned in both orientations under the lacZ promoter as shown in Fig.  1. Plasmids p6N4 and p6NP1 are designed such that the lacZ promoter is located upstream and, in p1N3, is located downstream of the repB gene. As can be seen in Fig. 3, the amount of 3␣-HSD/CR expressed is significantly lower in cells co-transformed with p6N4 relative to the control (pUC18). 3␣-HSD/CR expression is even lower in cells co-transformed with p6NP1. In p1N3 co-transformed cells, 3␣-HSD/CR expression is higher but still lower than in the respective control experiments. These results demonstrate that the gene product of repB represses hsdA expression. Because inhibition is strongest when the lacZ promoter is located upstream of repB, its orientation is opposite to that of hsdA (Fig. 1).
Decreasing the length between the transcriptional start site of repB and the tac or lacZ promoter reduces the amount of 3␣-HSD/CR expression with plasmids pBB6NP1 (Fig. 2) and  (Fig. 3), respectively. To conclude, the open reading frame of repB consists of 237 bp (Fig. 4) and translates into a protein of 78 amino acids with a calculated molecular mass of 8.8 kDa.
RepB Does Not Repress hsdA Transcription-RepB does not decrease the amount of 3␣-HSD/CR mRNA produced in C. testosteroni cells double transformed with plasmids carrying the genetic information for both genes. In Northern blots, a strong signal of 3␣-HSD/CR mRNA can be detected, whereas a significantly weaker band is found in cells expressing RepA (Fig. 5). This indicates that RepB does not act on the level of hsdA transcription.
RepB Acts Independently from RepA-To provide evidence that the translational repressor RepB acts independently from the transcriptional repressor RepA, the gene of the latter (repA) was knocked out in pRM4 by the insertion of an additional base at the translational start point of repA (Fig. 1). This causes a frameshift mutation in the repA sequence and consequently a complete loss of the RepA-repressing activity. After transformation of E. coli cells with pRM4, 3␣-HSD/CR expression increased compared with cells transformed with pKAN10, which harbors a functional repA gene (Fig. 6A) (25). According to this repA silencing in pRM4, the decreased hsdA expression of cells co-transformed with pRM4 and p6NP1 therefore exclusively represents the activity of RepB but not that of RepA.
Bacterial Overexpression and Purification of RepB-For fur-

FIG. 3. Expression of 3␣-HSD/CR in E. coli cells double trans-
formed with the indicated plasmids. pBBX1 and pBBAX7 carry the genetic information for 3␣-HSD/CR, whereas p6N4, p6NP1, and p1N3 carry the genetic information for RepB under the lacZ promoter. Plasmid pUC18 served as negative control. The observed expression of 3␣-HSD/CR is always lower when RepB is expressed (see "Results" for details). The overall lower expression with plasmids pBBX1 compared with plasmids pBBAX7 may indicate that in the latter fragment the operator Op2 is not fully active for the corresponding hsdA repression. ther characterization, we overexpressed the His-tagged RepB in E. coli and purified the recombinant protein to homogeneity. After inducing E. coli BL21(DE3)pLysS cells harboring the expression construct pETrepB, a strong band of the expected size (11.4 kDa for the His-tagged protein) was readily observed (data not shown). The resulting protein was packed in inclusion bodies and, for its purification, had to be solubilized using 6 M urea as detergent. By taking advantage of the N-terminal His tag of the recombinant protein, one-step purification to homogeneity by metal chelate affinity chromatography was achieved. After removal of urea, the protein proved to be active and to bind to 3␣-HSD/CR mRNA; therefore, thrombin cleavage of the His tag was unnecessary.
Testosterone Binding to RepB Causes an Induction of 3␣-HSD/CR Expression-Importantly, hsdA repression by RepB is reversed by the addition of testosterone to the culture medium, providing first evidence that RepB binds testosterone (Fig. 6A). Purified RepB did not bind testosterone. However, the addition of increasing amounts of purified RepB to an extract containing 20 g of E. coli HB101 protein resulted in a concentration-dependent binding of labeled testosterone. This fact reveals, on the one side, the specificity of RepB to bind testosterone but, on the other side, indicates the necessity of an unknown cofactor, which is obviously present in the E. coli extract (Fig. 6B). The ability of RepB to bind testosterone was further substantiated by mixing [ 3 H]testosterone with protein preparations of E. coli HB101 cells expressing RepB after transformation with p6NP1. The addition of competing, unlabeled testosterone gradually reduces the amount of bound, labeled testosterone until background levels are reached (Fig. 6C).
Messenger RNA Mobility Shift Assay and UV-cross-linking Studies-A 122-bp fragment of digoxigenin-labeled 3␣-HSD/CR mRNA synthesized by the transcription of ClaIdigested pSPT8 was used for mobility shift assays (Fig. 7). The addition of purified recombinant RepB generates a marked shift in mobility. No mobility shift was detected with RepA or empty pET15b plasmid as negative control. The specific binding of RepB to the 3␣-HSD/CR mRNA was further substantiated by UV-cross-linking studies (Fig. 8). The addition of both RNase T1 and RNase A to the protein-mRNA complexes resulted in a strong signal of protected RNA covalently attached to RepB. Digestion of mRNA by RNase T1 of non-cross-linked RepB-RNA mixtures generated a weaker FIG. 6. RepB acts independently from RepA and binds testosterone. A, as shown in Fig. 1, pRM4 contains both hsdA and repA, the latter overlapping with hsdA but silenced by a frameshift mutation. Compared with pKAN10 bearing a functional repA gene, no repression occurs with pRM4. Repression of hsdA after co-transformation of E. coli with pRM4 and p6NP1 is therefore the result of repB only. Repression with pRM4 is reversed by the addition of testosterone, which indicates that RepB binds to and is inactivated by testosterone. B, testosterone binding of RepB. Increasing amounts of purified RepB (0 -100 ng) to an extract containing 20 g of E. coli HB101 protein lead a concentrationdependent binding of labeled testosterone. C, competition with unlabeled testosterone. 20 g of protein isolated from E. coli cells expressing RepB through transformation with p6NP1 binds 3 H-labeled testosterone (lane 5). The amount of RepB in this preparation corresponds to 50 ng (cf. lane 5 in B). The addition of increasing amounts of unlabeled competing testosterone (lanes 1-4) gradually reduces protein bound testosterone until background levels (lane 6) are reached.  A, determination of the binding region of RepB on the 3␣-HSD/CR mRNA. The binding region of RepB was localized within nucleotides 17-28 downstream from the AUG start on the 3␣-HSD/CR mRNA. An amount of 15 ng of digoxigenin-labeled 3␣-HSD/CR mRNA was incubated with 1 g of RepB and increasing amounts of specific competitor DNA (antisense OL 1-4 in Fig. 9). Lane 1 contains 3␣-HSD/CR mRNA and RepB without any competing OL where the complete shifted complex relative to the control (without RepB, lane 6) is observed. The addition of OL 2 and OL 4 (lanes 3 and 5, respectively) but not that of OL 1 and OL 3 (lanes 2 and 4, respectively) competes away the bound protein, indicating that the binding region of RepB locates in the region of OL 2 and OL 4 binding overlap. The different mobility of the bands in lane 3 and 5 is attributed to differences in size of OL 2 (23 nucleotide) and OL 4 (16 nucleotide) (cf. Fig. 9). B, concentration dependence of OL 4 competition. Decreasing amounts of OL 4 (lanes 2-4) led to an attenuation of OL 4 binding competition, which results in a gradually increasing complex of shifted 3␣-HSD/CR-RepB. Concentrations of OL 4 are given as X-fold excess to 20 pmol of 3␣-HSD/CR mRNA. Lane 1, no shift due to lack of RepB; lane 5, complete shift due to lack of competing OL 4.
with least free energy of 3␣-HSD/CR mRNA was calculated using the mfold program. The predicted secondary structure of the first 134 nucleotides of the 3␣-HSD/CR mRNA with minimal free energy is shown in Fig. 9 with the sequence beginning at the transcription start point and ending 103 nucleotides after the initiating AUG. Several base pairing regions could be observed.
Four antisense DNA oligonucleotides complementary to different regions of the 3␣-HSD/CR mRNA as indicated in Fig. 9 were used as competitors for RepB binding in mobility shift assays (Fig. 10). Only oligonucleotides numbers 2 and 4 (OL 2, OL 4) prevented the binding of RepB to the mRNA as can be seen from the failure of the mobility shift (Fig. 10A). This finding indicates the specific RepB binding to a region between nucleotide positions 17 and 28 beginning from the AUG start on the mRNA structure as being depicted in Fig. 9. The specificity of RepB binding to this particular 3␣-HSD/CR mRNA region is further substantiated in Fig. 10B where OL 4 is shown to compete with RepB binding in a concentration-dependent manner. As expected, no shift in mobility is detected without RepB.
Conservation of the RepB Sequence-The C. testosteroni repB sequence was used in BLAST searches of the Swiss Protein and TrEMBL peptide sequence data bases for homologues. The searches revealed a 46.2% identity to the regulator of nucleoside diphosphate kinase (Rnk), a formerly unknown protein from E. coli that can restore a growth defect in alginate production in Pseudomonas aeruginosa (33,34), and a 75% identity to a putative enzyme-regulator fusion protein from Bordetella species (42). Further Rnks are found in both Grampositive and Gram-negative bacteria and, notably, include many pathogenic species (data not shown). It remains to be determined whether Rnks have a role as translational regulators or even function in steroid signaling.
A Scheme on the Negative Regulation of hsdA Expression-A model for hsdA-negative transcriptional and translational regulation is proposed in Fig. 11. The transcriptional repressor RepA is encoded by repA, and the translational repressor RepB is encoded by repB. Both genes are transcribed in an opposite orientation to hsdA. Whereas RepA binds to operator sequences (Op1 and Op2) upstream of hsdA and blocks hsdA transcription, RepB binds to the 3␣-HSD/CR mRNA and blocks hsdA translation. In the presence of appropriate steroids, how-ever, these bind to both RepA and RepB, thereby enabling hsdA transcription and translation, respectively. Hence, the induction of hsdA expression by steroids in fact is a derepression on the transcriptional and translational level. DISCUSSION Although the mechanism of steroid signaling in vertebrates is relatively known (35), very little data exist on the molecular mechanism of steroid signaling in procaryotes. The remarkable ability of C. testosteroni to respond to steroids as signal molecules for the expression of enzymes to break up the steroid nucleus as carbon source has been in the focus of much research.
3␣-HSD/CR is a particularly important enzyme to study because it is a major determinant in the degradation pathway of steroid substrates in C. testosteroni (17). In addition, 3␣-HSD/CR provides the enzymatic basis for the resistance of C. testosteroni against fungal steroid antibiotics like fusidic acid and contributes to important defense strategies against synthetic insecticides (19).
It is interesting to note that testosterone in C. testosteroni simultaneously induced the expression of additional enzymes that, on the one hand, could act "downstream" from 3␣-HSD/CR in the steroid degradation pathway and, on the other hand, are known to function in mineralization pathways of polycyclic aromatic hydrocarbons (10,36). Despite its importance, until recently nothing was known on the molecular basis of steroid-dependent gene regulation in C. testosteroni. Knowledge on this mechanism could not only provide an insight into the adaptation of C. testosteroni to the natural steroid pool but also elucidate the ability of this organism to mineralize environmentally harmful polycyclic aromatic hydrocarbons in polluted soils (8).
In the course of our work to resolve the steroid-dependent regulation of hsdA, we have previously reported on the identification of a gene, repA, which encodes a repressor of hsdA expression (25). The corresponding protein, RepA, binds to specific operator regions upstream of hsdA, thereby preventing hsdA transcription.
In this report, we describe the identification and features of a novel gene (repB) in the C. testosteroni genome, the protein product of which (RepB) does also act as a negative regulator of hsdA expression. The repB gene comprises an open reading FIG. 11. A model on the transcriptional and translational regulation of hsdA in C. testosteroni. In the absence of inducing steroids, RepA blocks transcription of hsdA and RepB blocks translation of the 3␣-HSD/CR mRNA. In the presence of appropriate steroids, however, these bind to both negative regulators, leading to a dissociation of RepA from the operator region and to a release of RepB from the 3␣-HSD/CR mRNA. Hence, the induction of hsdA expression by steroids in fact is a derepression of hsdA transcription and translation. frame of 237 bp and was located 932 nucleotides downstream of hsdA on a 5.2-kb EcoRI fragment of C. testosteroni chromosomal DNA. Interestingly, similar to repA, the repB gene is transcribed in the opposite direction to that of hsdA. However, in contrast to the transcriptional repressor RepA, RepB was found to act as a repressor of hsdA translation. Translational regulation is a key level in regulating gene expression. Procaryotes use translational repression by a variety of mechanisms to determine levels of expression of a variety of essential proteins. Translational regulation has been shown in different systems to involve cis-acting mRNA sequences that form secondary or tertiary structures sequestering the ribosome binding site. The binding of trans-acting proteins and/or antisense RNA to target mRNA can allosterically control translation (reviewed in Ref. 37). As being revealed from our results on hsdA regulation, RepB acts as a trans-acting factor that modulates hsdA expression by repressing 3␣-HSD/CR mRNA translation. The occurrence of cis-acting mRNA elements in the 3␣-HSD/CR mRNA structure that could repress hsdA translation was additionally excluded by temperature shift experiments of the bacterial cultures from 30 to 42°C (data not shown) (37).
Data bank searches with the RepB primary structure yielded a 46.2% identity to the regulator of nucleoside diphosphate kinase (Rnk) from E. coli. Nucleoside diphosphate kinase (Ndk, EC 2.7.4.6) catalyzes the exchange of a ␥-phosphate between nucleoside triphosphates and diphosphates via a ping-pong mechanism with a high energy phosphohistidine intermediate (38). Ndk of E. coli was first described in 1953 (39) and purified in 1984 (40). Several isoforms have been identified until now that comprise a large family of proteins found in procaryotes and eucaryotes with a high level of sequence and structure homology throughout the whole family (38). Ndks were originally proposed to be pivotal as housekeeping enzymes producing nucleoside triphosphates as precursors for RNA, DNA, and polysaccharide synthesis. During the last decade, a large body of evidence has been collected regarding their involvement in the regulation of gene expression during development, tumorigenesis, and tumor metastasis (reviewed in Ref. 41).
As yet, very little is known regarding the regulation of ndk. The E. coli rnk gene encodes a 14.9-kDa protein (34) and has been identified as one of two genes that can restore alginate synthesis in P. aeruginosa (33). All of the rnk genes identified so far do not show any significant relationship to other known proteins. Moreover, a comparison of the predicted amino acid sequence of rnk against protein data bases failed to shed any light on its function as is the case with repB, the newly described gene presented here. However, the ability of RepB to regulate the levels of 3␣-HSD/CR in C. testosteroni on the translational level may provide important clues regarding the regulation of Ndk. Therefore, it would be interesting to find out whether Ndk regulators such as Rnk or RepB are also conserved in other organisms.
Combined, our finding on the role of the repA and repB genes in the regulation of hsdA provides an important insight into the molecular adaptation of C. testosteroni to steroid substrates. The observation that RepA and RepB regulate the levels of 3␣-HSD/CR both on the transcriptional and translational levels, respectively, suggest an efficient regulatory system that can respond to changes in the carbon resources available, i.e. from organic acid or amino acid substrates to steroids. It re-mains to be seen whether proteins such as RepA and RepB are active as regulators in the expression of steroid-metabolizing enzymes in other bacteria. As the understanding of the machinery and regulation of steroid-degradative enzymes becomes clearer in C. testosteroni, a more complete picture of steroid signaling in procaryotes will emerge. To this point, no relationships with the steroid/steroid-receptor system known from vertebrates could be inferred in procaryotes.