Production of 11α‐hydroxysteroids from sterols in a single fermentation step by Mycolicibacterium smegmatis

In this work we have rationally designed new recombinant bacteria able to produce 11alpha‐hydroxylated steroids in a single fermentation step using sterols as feedstock. The key elements to achieve this goal were the hydroxylase system from R. oryzae and the actinobacteria Mycocilibacterium smegmatis.


Introduction
Steroids represent one of the most widely used drugs for multiple clinical purposes (e.g. anti-inflammatory, immunosuppressive, anti-allergic, anti-cancer) (Fern andez-Cabez on et al., 2018). The oxidation state of steroid rings and the presence of different attached functional groups determine the specific physiological function of each steroid (Lednicer, 2011), and therefore, structural modifications of steroids, such as hydroxylation, dehydrogenation or esterification, highly affect their biological activity (Donova and Egorova, 2012;Szaleniec et al., 2018). Among them, hydroxylation is one of the most important steroid modifications introduced by the pharmaceutical industry, since it introduces deep changes not only in their clinical activities, but also in their physicochemical properties (e.g. solubility, adsorption).
Fungi have been reported to carry out hydroxylations at almost all stereogenic centres of the steroid molecule (Kristan and Ri zner, 2012), and they have been widely applied at industrial scale for the production of hydroxylated steroids with a broad range of biological activities (Donova, 2017). This oxyfunctionalization of steroids is mainly catalysed by cytochromes P450 (CYPs) acting as monooxygenases by inserting a single oxygen atom into a non-activated C-H bond of the substrate with a concomitant reduction of other oxygen atom to water (Bernhardt, 2006). Fungal hydroxylations are usually carried out by two-component enzyme systems consisting of a CYP monooxygenase and a NAD(P)H CYP-reductase (CPR) (Cre snar and Petri c, 2011; Kristan and Ri zner, 2012).
In particular, 11a-hydroxylated steroid synthons are one of the most important commercially steroid intermediates used for the production of contraceptive drugs and glucocorticoids. Important fungal species for steroid 11a-hydroxylation reactions include Rhizopus nigricans, Aspergillus orchraceus, Aspergillus niger and Rhizopus oryzae Kollerov et al., 2020). In this sense, the 11a-hydroxylating system of R. oryzae has been used for the production of hydroxyprogesterone (Fernandes et al., 2003;Petric et al., 2010). This fungal system consists in the CYP509C12, one of the 48 CYPs encoded in its genome, and its redox partner RoCPR1, a NAD(P)H-dependent CRP. By expressing CYP509C12 in yeast, it has been demonstrated that this CYP hydroxylates predominantly at 11a and 6b positions of steroids (Petric et al., 2010), although 11a-hydroxylation of steroids by fungi is the synthetic procedure currently used by the steroid industry (Petric et al., 2010). The current method for obtaining 11a-OH-AD/ADD at industrial level is a two-step one-pot bioconversion from phytosterols based on the specific biochemical activities of two microbial strains without separation and purification of the intermediate product (Dovbnya et al., 2017). However, the large number of CYPs contained in the fungal strains often generates during the in vivo hydroxylation a multiplicity of oxygenated steroid by-products, which not only are difficult to separate, but also contribute to reduce the bioconversion yields (Fernandes et al., 2003). Therefore, the design of alternative and more specific hydroxylating microbial cell factories created by recombinant DNA technologies has been considered an industrial challenge (Donova, 2017;Fern andez-Cabez on et al., 2018).
In this work, we have tested the possibility of rationally design new recombinant bacteria able to produce 11ahydroxylated steroids in a single fermentation step using sterols as feedstock. To this aim, we have designed a synthetic operon containing the 11a-hydroxylating enzymes from R. oryzae that was cloned and expressed into two previously engineered mutant strains of M. smegmatis created to produce AD or ADD from sterols. The expression of the synthetic operon in these modified bacterial chassis has allowed us to produce for the first time 11a-hydroxylated compounds directly from sterols in a single fermentation step.

Production of 11aOH-ADD in M. smegmatis
To test the possibility of producing 11a-hydroxylated steroids from natural sterols (CHO and PHYTO) in a single step in M. smegmatis, we designed the synthetic bacterial operon FUN (RoCPR59830-CYP509C12), harbouring the CYP 11a-hydroxylase from R. oryzae and its redox partner CRP (Fig. S1). The synthetic operon was cloned into the mycobacterial replicative plasmid pMV261, creating the pMVFUN plasmid, that was transformed into the M. smegmatis MS6039 mutant strain. This mutant strain has been previously engineered and accumulates ADD from CHO or PHYTO (Table 1) (Garc ıa et al., 2012;Garc ıa et al., 2017). The expression of the enzymes RoCPR59830 and CYP509C12 in the MS6039 (pMVFUN) recombinant strain was determined by SDS-PAGE (Fig. S2A) and the ability to transform sterols into 11aOH-ADD was analysed by HPLC-MS along the growth curve using sterols (CHO or PHYTO) as feedstock.
First, MS6039 (pMVFUN) and MS6039 (pMV261) strains were grown in the biotransformation medium in the presence of CHO. During the exponential phase, the recombinant strain MS6039 (pMVFUN) showed a slight delay in growth compared to the control strain MS6039 (pMV261), but both cultures reached similar biomass at the stationary phase ( Fig. 2A). After 24 h, 11aOH-ADD was detected in the culture supernatant of MS6039 (pMVFUN) and the maximum bioconversion of 99.6 AE 0.3 % was obtained after 60 h of growth with a production yield of 65.8 AE 3.9 % for φ11aOH-ADD/CHO (Fig. 2B). Interestingly, the HPLC-UV/DAD-MS monitoring allowed us to identify and quantify small amounts of some by-products in the culture medium, such as ADD, 11aOH-AD and trace amounts of 1,4-HBC (Fig. 2B). Production yields for the main by-products were 25.4 AE 4.3% for ADD/CHO and 8.6 AE 0.5% for 11aOH-  Fig. S3A and B). We have detected one additional compound eluting at 4.92 min ( Fig. S3A and B). Its m/z of 345 coincides with the molecular mass of 1,4-HBC increased by 16, suggesting that it could correspond to 11aOH-1,4-HBC. As expected, MS6039 (pMV261) control strain did not produce 11aOH-ADD from CHO ( Fig. S3A) and ADD was detected as the main biotransformation product with a conversion rate of 96.2 AE 5.9% and a production yield of 99.3 AE 0.3% for ADD/CHO.
Taking into account that PHYTO are used in the steroid industry as the preferred low-cost raw material to produce steroid synthons, we tested it as feedstock to produce 11aOH-ADD in the MS6039 (pMVFUN) recombinant strain. To this aim, MS6039 (pMVFUN) and MS6039 (pMV261) strains were grown in the biotransformation medium in the presence of PHYTO and monitored by HPLC DAD-MS as performed for CHO cultures (Fig. 2C). The MS6039 (pMVFUN) strain successfully achieved the transformation of PHYTO into 11aOH-ADD. The conversion rate was 67.5 AE 0.3%, and the 11aOH-ADD production yield 11aOH-ADD/PHYTO was 33.3 AE 0.2% ( Fig. 2D and Fig. S4). Some by-products as ADD, 11aOH-AD, 1,4-HBC and 11aOH-1,4-HBC were detected when PHYTO was used as feedstock (Fig. S4).
The yields for these by-products were 59.5 AE 0.4% for ADD/PHYTO and 3.4 AE 0.1% for 11aOH-AD/PHYTO. Derived compounds 11aOH-1,4-HBC and 1,4-HBC could not be quantified because they were present at very low concentrations.
As expected, MS6039 (pMV261) control strain only produced ADD from PHYTO with a conversion rate of 67.5 AE 0.3% and a transformation yield of 95.7 AE 0.9% for ADD/PHYTO (Fig. S4).

Production of 11aOH-AD in M. smegmatis
To achieve the conversion of sterols into 11aOH-AD in a single fermentation step, we used as chassis the M. smegmatis MS6039-5941 mutant strain that was engineered to produce AD from CHO or PHYTO (Table 1) (Garc ıa et al., 2012;Garc ıa et al., 2017). The MS6039-5941 strain was transformed with the pMVFUN plasmid as described above. The production of CYP509C12 and RoCPR59830 proteins in the recombinant MS6039-5941 (pMVFUN) strain was confirmed by SDS-PAGE (Fig. S2B), and the ability to transform sterols (CHO and PHYTO) into 11aOH-AD was analysed by HPLC-MS.
It is worth to mention that the control strain MS6039-5941 (pMV261) did not produce 11aOH-AD (Fig. S5) and, as expected, AD is the main product, having a yield of 76.7 AE 3.7% with a conversion of 99.1 AE 0.2%. Curiously, a significant amount of 4-HBC was detected in the control strain (4-HBC/CHO = 18.3 AE 2.9 %) when compared to the strain carrying the FUN operon. A small amount of ADD was also detected in the control strain (ADD/CHO = 4.9 AE 0.9%).

Discussion
In the pharmaceutical sector, steroid hydroxylation plays an important role to produce new functionalized steroids,  because it usually introduces deep changes in their physicochemical and pharmaceutical properties. In particular, the 11a-or 11b-hydroxylation of steroids are essential functionalization steps to develop commercially important intermediates to synthetize glucocorticoids and contraceptive drugs. Current methods of production of hydroxylated steroids mainly rely on biotransformations using wild-type fungal whole cells that harbour these enzymatic activities. The production of the hydroxylated steroids is carried out in at least two fermentation steps exhibiting in most cases some drawbacks such as low selectivity and reduced conversion yield. Therefore, the design of alternative fermentation processes by using recombinant DNA technologies has been proposed in recent years. In this sense, several fungal hydroxylases have been successfully expressed in yeasts (Petric et al., 2010;Hull et al., 2017;Lu et al., 2018) demonstrating their potential biotechnological applications. However, to the best to our knowledge these recombinant yeasts have not been implemented at industrial scale yet.
In this work, we have advanced one-step forward in the direction of creating alternative processes to produce the 11-hydroxylated steroids directly from sterols in a single fermentation step. To fulfil this aim, we have used a rationally designed synthetic operon (named FUN operon) carrying the 11a-hydroxylating system from R. oryzae and cloned them in M. smegmatis generating new recombinant bacterial strains able to produce 11-hydroxylated derivatives in a single fermentation directly from natural sterols such as CHO and PHYTO.
M. smegmatis is a model microorganism that has been used for the study of steroid catabolism due to some strengths: (i) it is a fast-growing bacterium capable to grow in natural sterols (CHO and PHYTO) as a sole carbon and energy source; (ii) its genome sequence has been accessible since 2006; (iii) it can be manipulated genetically with many genetic tools; (iv) it is a robust chassis resistant to stressing industrial production conditions; and (v) it has a very effective transport system for sterols. Hence, M. smegmatis has been proposed and used before as an efficient platform to produce steroidal drugs (Fern andez-Cabez on et al., 2017; Gal an et al., 2017).
One of the most important achievements of this work is the demonstration that the synthetic fungal operon was fully functional in M. smegmatis. Only few fungal CYPs have been successfully produced in its active form in bacteria so far (Barnes et al., 1991;Gonz alez and Korzekwa, 1995;Hannemann et al., 2006;Felpeto-Santero et al., 2019). In our case, we achieved significant levels of CYP and CRP expression as determined by SDS-PAGE (Fig. S1). Apparently, depending on the substrate and the mutant strain used, the production of the hydroxylating system generates some cellular stress in the host as deduced by comparing the corresponding growth curves (Fig. 2).
The detection of the 11a-hydroxylated steroids in the culture medium of the recombinant bacteria at high yields demonstrated that the fungal 11a-hydroxylating enzymes have been integrated in the sterol metabolism in M. smegmatis strains creating a new expanded pathway. Such an efficient integration was in fact a surprising result, because one likely outcome could be the hydroxylation of different metabolic intermediates that might block some of the multiple reactions required to accumulate the desired compound due to the specificity of the sterol catabolic enzymes. However, MS6039 (pMVFUN) and MS6039-5941 (pMVFUN) recombinant strains successfully accumulated 11aOH-AD and 11aOH-ADD, respectively, using CHO and PHYTO as precursors. In this sense, the detection of 11aOH derivatives of AD, ADD, 4-HBC and 1,4-HBC suggests, on the one hand that the enzymes of R. oryzae are not highly specific for a particular steroid. However, on the other hand, this also means that although several hydroxylated intermediates can be produced, none of them blocks the pathway, and fortunately, can be finally funnelled to the final products 11aOH-AD or 11aOH-ADD, depending on the mutant host used. Therefore, we can conclude that the enzymes of the M. smegmatis sterol degradation pathway can also recognize as substrates the 11aOH derivatives of their own natural substrates.
It is worth to mention that when the 11a-hydroxylating system of R. oryzae was expressed in yeast, it was able to hydroxylate steroids, both at 11a and 6b positions, with good yields (Petric et al., 2010). However, in our case the main products obtained were hydroxylated at 11a position. We have detected trace amounts of a compound of m/z 319 compatible with the dihydroxylated 11a-hydroxylation of steroids by CYP450 in bacteria 2519 derivative of 4-HBC in the biotransformation of CHO by the MS6039-5941 (pMVFUN) strain. Although more experiments are required to confirm the presence of small amounts of 6b-OH derivatives in M. smegmatis, this finding suggests that 6b-hydroxylation is not relevant in our recombinant strain. This interesting result suggests that, most probably, the specificity of CYP enzymes depends on their metabolic environment (location, pH, substrates, other reductases, etc.), and the hydroxyl group destination (11a or 6b position) can be favoured in one selective direction. This characteristic has been also previously observed in the case of heterologous expression of the same R. oryzae hydroxylating system in yeast where the distribution between 11a-and 6b-hydroxylated derivatives, and the appearance of non-identified monohydroxysteroids depended on the substrate that was used (Petric et al., 2010). Although the whole process has to be further optimized at industrial scale to improve the yield of the 11ahydroxylated compounds and to reduce the amount and number of by-products, our results open a new avenue for searching effective biocatalysts for producing hydroxylated steroids from sterols in a single fermentation step. In addition, they reinforce the assumption that engineered M. smegmatis strains represent a new generation of biocatalysts with a great potential to be applied for industrial processes. Based on the increased knowledge on the steroid metabolism in M. smegmatis, we uphold this bacterium as an exceptional bacterial chassis to implement a la carte metabolic engineering strategies based on synthetic biology for the industrial production of other valuable pharmaceutical steroids directly from sterols.

Strains, oligonucleotides and culture growth
The strains, plasmids and oligonucleotides used in this study are listed in Table 1. Escherichia coli DH10B strain was used as a host for cloning. It was grown in rich LB medium at 37°C in an orbital shaker at 200 rpm. LB agar plates were used for solid media. Gentamicin (10 lg ml À1 ), ampicillin (100 lg ml À1 ) or kanamycin (50 lg ml À1 ) were used for plasmid selection and maintenance in this strain. M. smegmatis mc 2 155 was cultured on Bacto Middlebrook 7H9 (7H9, Difco) supplemented with Middlebrook ADC Enrichement (ADC, Difco) (10 % v/v), glycerol (0.2 % v/v) (Sigma) at 37°C in an orbital shaker at 200 r.p.m. Tween 80 % (0.05 % v/v) (Sigma) was added to M. smegmatis cultures to avoid cell aggregation. Antibiotics were used where indicated at the following concentrations: kanamycin (20 µg ml -1 ). Cell grow was monitored following OD 600nm .

Design and construction of the bacterial synthetic FUN (CYP509C12-RoCRP1) operon
To achieve the heterologous production of the 11a-hydroxylase activity from R. oryzae RA 99-880 (Fungal Genetics Stock Center, FGSC, University of Missouri; Petric et al., 2010) in bacteria, we designed the synthetic operon FUN that encodes the cytochrome CYP509C12 (EIE80372) with 11a-hydroxylase activity and its natural redox partner RoCPR1 cytochrome reductase (EIE89541). The codon usage was manually optimized for M. smegmatis, keeping a nucleotide identity percentage of approximately 83% for both genes. A consensus Shine-Dalgarno sequence of 6 bp (AAAGGGAG) was added upstream the respective start codons as well as some restriction sites to facilitate cloning (Fig. S1). Alanine was also included as the second amino acid of the resulting proteins to increase protein translation (Bivona et al., 2010). The operon engineered was chemically synthesized by ATG:biosynthetics GmbH and initially cloned into the pGH vector yielding plasmid pGH-FUN that was used to transform E. coli DH10B cells and check the sequence. Plasmid pGH-FUN was digested with BamHI-EcoRI to release the fragment containing the FUN operon that was further cloned under the control of the P hps constitutive promoter, into the shuttle E. coli/Mycobacterium vector pMV261 (Stover et al., 1991) yielding pMVFUN. Plasmids pMV261 (empty vector, control plasmid) and pMVFUN were individually transformed into the M. smegmatis mutants by electroporation.

Analysis of the FUN operon expression by SDS-PAGE analysis
The recombinant strains MS6039 and MS6039-5941 carrying plasmid pMVFUN (Table 1) were grown in biotransformation media and conditions during 24 h. Cells were harvested by centrifugation (15 min, 5000 9 g, 4°C ) thawed on ice, washed twice with NaCl 0.9 % (w/v) and resuspended in 0.5 ml of Tris-HCl 50 mM pH 7.5. Cells were disrupted by sonication using a Branson sonicator 150 (6-8 pulses of 1 min at 90% power, with 30 s of cooling on ice between each). Cell debris was removed by centrifugation at 14,000 9 g for 15 min at 4°C . Soluble and insoluble fractions were analysed by SDS-PAGE to check FUN operon expression.

Bioinformatic tools
DNA and protein sequences from R. oryzae genome were obtained from Broad Institute Server (http://www.b road.mit.edu/annotation/genome/rhizopusoryzae/Multi Home.html) and National Center for Biotechnology Information (NCBI) Database (http://www.ncbi.nlm.nih.gov/ge ne/). Nucleotide and amino acid sequences were compared to National Center for Biotechnology Information (NCBI) database using BLAST algorisms (http://www.ncb i.nlm.nih.gov/cgibin/Entrez/genom_table_cgi); to align and compared to local database, Local-BLAST (BioEdit) and HMMER algorithm at European Bioinformatics Institute (EBI) (http://www.ebi.ac.uk/Tools/hmmer/) were used. Multiple alignments of protein were carried out with the MUSCLE server programme at the European Bioinformatics Institute (EBI) (http://www.ebi.ac.uk/Tools/msa/ muscle/).

Steroids extraction
Aliquots of 10 µl of 5 mM TES in 10 % (v/v) tyloxapol were added to each 0.2 ml sample taken from the biotransformation experiments prior to extraction with chloroform, as an internal standard (ISTD). The samples were extracted using two volumes of chloroform. The aqueous fraction was discarded, and the chloroform fraction was dried at 60°C using a Thermoblock and then dissolved in 0.5 ml of acetonitrile. Each sample was subjected to chromatographic analysis by HPLC-UV/DAD-MS (25 µl).

HPLC-UV/DAD-MS analysis
Experiments were carried out using a DAD detector and a LXQ Ion Trap Mass Spectrometer, equipped with an atmospheric pressure chemical ionization source, electrospray ionization source and interfaced to a Surveyor Plus LC system (all from Thermo Electron, San Jose, CA, USA). Data were acquired with a Surveyor Autosampler and MS Pump and analysed with the Xcalibur Software (from Thermo Fisher Scientific, San Jose, CA, USA). High-purity nitrogen was used as nebulizer, sheath and auxiliary gas. MS analysis was performed both in full scan and in selected ion monitoring (SIM) mode by scanning all the daughter ions of the products in positive ionization mode. The quantification was performed from parent mass of compounds, and the specificity was obtained by following the specific fragmentations of all compounds.