Heterologous expression and characterization of a 3-ketosteroid-∆1-dehydrogenase from Gordonia neofelifaecis and its utilization in the bioconversion of androst-4,9(11)-dien-3,17-dione

3-Ketosteroid-∆1-dehydrogenase (KstD), a key enzyme in microbial steroid catabolism, catalyzes the trans-axial elimination of the C1 and C2 hydrogen atoms of the A-ring from the polycyclic ring structure of 3-ketosteroids, and it was usually used to transform androst-4-ene-3,17-dione (AD) to produce androsta-1,4-diene-3,17-dione. Here, the KstD from Gordonia neofelifaecis was expressed efficiently in Escherichia coli. E. coli cells expressing KstD3gor were subjected to the investigation of dehydrogenation activity for different steroids. The results showed that KstD3gor has a clear preference for steroid substrates with 3-keto-4-ene configuration, and it exhibits higher activity towards steroid substrates carrying a small or no aliphatic side chain than towards substrates having a bulky side chain at the C-17 atom. The recombinant strain could efficiently convert androst-4,9(11)-dien-3,17-dione into androst-1,4,9(11)-trien-3,17-dione (with conversion rate of 96%). 1(2)-Dehydrogenation of androst-4,9(11)-dien-3,17-dione is one of the key steps in glucocorticoid production. To the best of our knowledge, this is the first study reporting on the conversion of androst-4,9(11)-dien-3,17-dione catalyzed by recombinant KstD; the expression system of KstD3gor reported here would have an impact in the industrial production of glucocorticoid in the future.


Introduction
Significant progress has been made over the last ten years in the use of enzymes and microorganisms for the manufacturing of complex chemical compounds and replacing multi-steps chemical syntheses. Actinobacteria are known as efficient biocatalysts of steroid bioconversion since 1913 (Tak 1942). However, the recent advances in genome sequencing and bioinformatics technologies provided tools for identification of new players in cholesterol bioconversion. Although steroids are highly resistant to biodegradation, many bacteria use them as a source of carbon and energy (García et al. 2012). Microbial transformation could be carried out under mild reaction conditions with excellent yields of products and remarkable regio-and stereo-selectivity, which is hardly available for chemical synthesis. Therefore, for producing novel steroidal drugs and generating active pharmaceutical ingredients, microbial transformation is employed as a novel, efficient and economical tool (Donova 2007;García et al. 2012;Yang et al. 2015). For example, side chains of phytosterol, a byproduct from soybeans, sugar and paper industries, can be selectively degraded by a process similar to the boxidation of fatty acids, yielding 17-ketosteroids (Wei et al. 2010). One of the products of this degradation, 9a-hydroxyandrost-4-ene-3,17-dione, and its D9-analog are considered as the most important intermediates for the synthesis of corticoids such as prednisolone, betamethasone, dexamethasone, and triamcinolone (Fokina and Donova 2003;Yuan et al. 2015). The efficiency of enzymatic processes and purity of their products have obvious advantages in comparison with multi-steps chemical syntheses of hormonal drugs. However, the development of steroid biotechnology requires further studies of microorganisms able to degrade/modify steroids as well as enzymes catalyzing these reactions on the molecular level (Yang et al. 2015).
The degradation of cholesterol or its derivatives begins with the transformation of cholesterol to cholest-4-en-3-one by a cholesterol oxidase (Shao et al. 2015). The subsequent catabolism involves elimination of the alkyl side chain followed by the opening of the rings A/B and rings C/D. A 3-ketosteroid D1-dehydrogenase (KstD) [EC 1.3.99.4], catalyzing the elimination of the hydrogen atoms of the C-1 and C-2 in the A-ring from the polycyclic ring structure of 3-ketosteroids, is a key enzyme in microbial steroid catabolism needed for the opening of the steroid B-ring (Fernández de Las Heras et al. 2012;Zhang et al. 2013). KstD is a FADdependent enzyme, the natural electron acceptor appears to be vitamin K2 (Choi et al. 1995a), and they can transfer electrons to N-methyl phenazolium sulfate. KstDs were found in various bacteria, including Mycobacterium sp., Rhodococus sp., Pseudomonas sp. and Arthrobacter sp. (Choi et al. 1995b;Molnar et al. 1995;van der Geize et al. 2000;Brzostek et al. 2005;Knol et al. 2008). They display a broad substrate spectrum. The KstD 1SQ1 (KstD 1 from Rhodococcus erythropolis SQ1) and KstD 2SQ1 enzymes (KstD 2 from Rhodococcus erythropolis SQ1) were specific for steroids with the 3-keto-4-ene structure such as 9a-hydroxy-androst-4-ene-3,17-dione (Knol et al. 2008). KstD 3SQ1 (KstD 3 from Rhodococcus erythropolis SQ1) had a clear preference for 3-ketosteroids with a saturated A-ring. The role of three KstDs from Rhodococcus ruber strain Chol-4 was studied in the steroid metabolism (Fernández de Las Heras et al. 2012).
Gordonia neofelifaecis NRRL B-59395 was initially isolated from fresh faeces of a clouded leopard (Neofelis nebulosa) for its ability to degrade cholesterol (Liu et al. 2011). In spite of a significant number of known actinobacteria that could degrade/modify steroids, only a few genomes of them have been completely sequenced to identify all sterol catabolic genes. We sequenced G. neofelifaecis NRRL B-59395 genome and found 5 putative genes encoding KstD Li et al. 2014). We also studied the substrate specificity of these KstDs in our former study . One enzyme, KstD 3gor , had the broadest spectrum of substrate specificity, exhibiting activity to progesterone, 16a, 17a-epoxyprogesterone and cholest-4-en-3-one.
In this work, we cloned KstD 3gor gene into E. coli vector, expressed the recombinant enzyme and characterized its enzyme specificity and selectivity. Our results indicated that KstD 3gor could have a possible application for the production of androst-1,4,9(11)-trien-3,17-dione in the pharmaceutical industry.

Construction and expression of KstD 3gor in E. coli
The KstD 3gor gene was amplified from previously cloned KstD 3gor construct  and cloned into the Nde I/BamH I restriction sites of pET-28a(?) vector (Novagen), giving the construct pET28a-KstD3. Recombinant DNA techniques were done according to standard protocols (Sambrook and Russell 2001). E. coli transformation was performed as described previously (Chung et al. 1989). The DNA was isolated using the Plasmid Mini Kit (Omega, USA) according to the manufacturer's instructions. DNA sequence was verified by sequencing service provided by the Beijing Genome Institute Sequence Facility (Shenzhen, China).

Expression of KstD 3gor in E. coli and preparation of clarified lysate
Recombinant KstD 3gor protein was expressed in BL21(DE3) cells. E. coli cells transformed with pET28a-KstD 3 construct were grown overnight (18 h) at 37°C in 10 mL of LB medium containing 25 lg/mL kanamycin (the same concentration was used in all cultivation procedures) in a 50-mL Erlenmeyer flask. Ten mL of this culture was inoculated into 500 mL of LB medium containing antibiotic in a 2000-mL Erlenmeyer flask and incubated at 30°C until the OD 600 reached *0.6. Then, the culture was induced with 1 mM isopropyl b-D-1thiogalactopyranoside (IPTG), and incubated with shaking (200 rpm) at 30°C for 6 h. The cells were harvested by centrifugation at 10,000 rpm for 10 min. The cell pellet was washed for three times with 10 mL of chilled (4°C) 50 mM Tris-HCl buffer (pH 7.0) (every time, centrifugation was performed at 10,000 rpm for 10 min at 4°C), resuspended in 10 mL of the same buffer and sonicated using an ultrasonic homogenizer (JY92-II, Scientz Biotechnology Co. Ltd., China) in an ice bath, with 90 cycles of 5 s on and 5 s off at 220 W. The cell lysates were centrifuged for 30 min at 18,000 rpm and 4°C. The supernatant (clarified lysate) and the washed cell pellet were used for KstD activity gel assay or biochemical assays for the measurements of bioconversion of steroid substrates.

KstD activity staining on native PAGE
The cell extract was used for analysis of KstD activity on native 12.5% PAGE (Knol et al. 2008). The KstD activity was visualized by incubating native gels in 100 mL 50 mM Tris-HCl buffer (pH 7.0) containing 3.1 mg phenazine methosulphate, 2.9 mg androst-4-en-3, 17-dione dissolved in ethanol and 41 mg nitroblue tetrazolium (NBT) dissolved in 70% dimethylformamide. The native gels were stained for several hours until the appearance of distinct purple-colored bands of the product of the reaction, formazan on the gel. Then, the staining was stopped by replacing the staining solution with 100 mL of 10% (v/v) acetic acid. -complex) was prepared by the co-evaporation method as previously reported (Manosroi et al. 2008). Briefly, 0.2 mol of methyl-b-cyclodextrin and 0.1 mol of steroid substrates were dispersed in 95% ethanol and stirred with a speed of 200 rpm for 4 h at 37°C. Then, ethanol was evaporated at 60°C on a Buchi model R-210 rotary evaporator (0.09 Mpa, with a speed of 60 rpm at 65°C) (The same parameters of the instrument were used throughout the study to concentrate extraction samples).

Bioconversion of steroid substrates by KstD 3gorexpressing E. coli cells and extracts
The bioconversion of steroid substrate was performed in 100 mL Erlenmeyer flasks with the clarified lysates or E. coli cells. To prepare E. coli cells, the cells expressing KstD 3gor were harvested by centrifugation from 4 mL fresh culture, and resuspended in 20 mL Tris-HCl buffer (pH 7.0) to the concentration of 5 9 10 7 CFU/mL. The clarified lysates from 4 mL fresh culture were prepared as described above. Then, steroid complex was added to the 20 mL of cell suspension or clarified lysate. The final concentration of each substrate was 2 g/L. The reaction mixture was incubated at 37°C with shaking at 200 rpm for 10-24 h. Steroids were extracted from the medium by adding the equivalent volume of ethyl acetate; after phase separation, the upper organic phase was analyzed by thin-layer chromatography (TLC). The TLC was performed on 0.25 mm-thick silica gel G (silica gel 254, Qingdao Haiyang Chemical Co., Ltd.) with cyclohexane/ethyl acetate (7:3, v/v) as the mobile phase. 10 lL sample of organic phase was applied on TLC plate. The products of the enzymatic reaction were visualized by spraying a mixture of sulfuric acid and methanol (1:6, v/v) on the plates and heating them at 100°C until the colors developed. The extraction procedure was repeated three times, and the steroid extract was dried by reduced pressure distillation (0.09 MPa, 65°C) and further separated by high-performance liquid chromatography (HPLC) (Shimadzu, Kyoto, Japan). To perform HPLC analysis, samples were diluted to an appropriate concentration (about 1 mg/mL) with methanol and filtered through 0.22 lm pore-size membranes. HPLC analysis was performed on an Alltima C18 column (250 9 4.6 mm, 5 lm, Alltech, USA) under the control of an HPLC system (Shimadzu, Japan) equipped with an LC-20AB HPLC pump system and SPD-20A UV detector. The HPLC was performed using methanolwater (6:4, v/v) as the mobile phase at a flow rate of 1 mL/min.
The conversion rate of sterol was calculated according to the following formula: Conversion rate % ð Þ ¼ moles of product moles of substrate converted : Purification and spectroscopic analysis of transformation products Purification of the products of bioconversion was performed as previously described (Liu et al. 2011). Briefly, 500 mL of the reaction mixture was extracted with 500 mL ethyl acetate (v/v) at room temperature for 30 min. The organic phase was collected and evaporated under reduced pressure.
The crude gum was dissolved in ethyl acetate, and applied to a silica gel column (2.5 cm 9 30 cm), eluted with acetone/ ethyl acetate (the eluent system consisted of gradient mixtures of chloroform and acetone and ethyl acetate), at a flow rate of 1 mL/min. Fractions of 10 mL were collected. The fractions containing the same steroid intermediates were pooled and concentrated in a rotary evaporator. Finally, the product was recrystallized from anhydrous alcohol and was subjected to 1 HNMR, 13 C NMR, or TOF-MS analysis.

Results
Expression of the catalytically active KstD 3gor in E. coli cells To expand our knowledge about KstD 3gor from G. neofelifaecis, the gene of this enzyme was sub-cloned into a commercial E. coli vector pET28a(?) and recombinant KstD 3gor protein was expressed in BL21(DE3) cells. The calculated molecular mass of KstD 3gor (57 kDa) corresponded to the most abundant band on the 10% SDS-PAGE in the sample with induced expression of the protein (Fig. 2a, lane 3). A significant amount of KstD 3gor was expressed as a soluble protein (Fig. 2a, lanes 3, 4). Decreasing the temperature (from 30 to 25°C) and increasing the induction time (from 4 to 6 h) led to a high level of recombinant protein expression. KstD activity staining on native PAGE confirmed the expression of catalytically active enzyme. The clarified lysates were prepared and assayed by native PAGE as described in Sect. ''Materials and methods''. The electron transfer from AD via phenazine methosulphate to NBT catalyzed by KstD 3gor resulted in the formation of a purplecolored product of the reaction, formazan (Fig. 2b, lanes 2,  3). Meanwhile, no activity was detected in the negative control sample (Fig. 2b, lane 1).

Bioconversion of steroid substrates by KstD 3gorexpressing E. coli cells and clarified lysates
The E. coli cells expressing KstD 3gor were tested in the dehydrogenation reaction of androst-4,9(11)-dien-3,17dione. The structures of the products were determined using spectroscopic techniques including 1 HNMR, 13 C NMR and TOF-MS. The TOF-MS spectra are shown in  -19). The result showed that the product reduces two mass units from androst-4,9(11)-dien-3,17-dione (284.39), suggesting that dehydrogenation reaction has occurred at the C-1 and C-2 in the A-ring.

Discussion
KstDs are widely found in actinobacteria as they play an important role in microbial steroid degradation. Different isoforms of the enzyme have been found exhibiting different substrate specificity and having different roles that could be strain-dependent (Knol et al. 2008). The genome of R. ruber strain Chol-4 contains three different KstD genes, and two of them contribute to cholesterol catabolism (Fernández de Las Heras et al. 2012). Rhodococcus jostii RHA1 KstDs display a quite different substrate specificity (Knol et al. 2008).
The sterol catabolic genes are highly conserved in G. neofelifaecis, R. jostii RHA1, and Mycobacterium tuberculosis, and mainly organized in three specific clusters. In our previous research, the substrate preference of five G. neofelifaecis KstDs was investigated preliminarily . Here, the KstD 3gor was biochemically characterized in detail.
A phylogenetic tree (Fig. 6) of the characterized KstDs revealed that they share high similarity with other bacterial KstD homolog. They clustered into at least four distinct groups (Knol et al. 2008). KstD 3gor showed 63.6 and 44.7% amino acid identity with R. ruber KstD 1 and R. erythropolis KstD 1 , correspondingly. Both KstD 3gor and R. erythropolis KstD 1 prefer 3-ketosteroids with a saturated A-ring as substrates.