**4. Discussion**

*N. tardaugens*, ARI-1 strain was isolated based on its ability to metabolize E2 [24]. Here we show that in addition this strain has a prominent capability to metabolize other steroids. Particularly, we have demonstrated that this strain is able to degrade steroids that can also be considered as toxic EDCs, such as TES, TES-Ac, DHEA and other C-19 steroids, like AD and ADD (Figure 2).

The recent reduction of the number of contigs in a previous work [29] allowed us to identify the genes that could be responsible for the degradation of C-19 steroids in *N. tardaugens*. A comparison with the genes described for other C-19 degradation pathways, particularly with those described for the degradation of TES in *C. testosteroni*, was carried out. A significant finding was that ARI-1 contains most of the predicted genes in a very compact cluster compared with *C. testosteroni* that has at least two main clusters to encode the complete pathway (Figure 3) [26]. The same cluster structure as that of *N. tardaugens* has been described in other estrogen degrader strains such as *Sphingomonas* sp. KC8 strain and *Altererythrobacter estronivorus* [18]. In the particular case of the KC8 strain, those genes are being expressed in the E2 and TES grown cells, suggesting that strain KC8 uses the same gene products to degrade the C/D rings of both C18 and C19 compounds [18]. Since many of the genes of the SD cluster putatively responsible for CD-ring degradation of TES (C19) in *N. tardaugens* do not have any other homolog in the genome, it seems very likely that this cluster is also responsible for the degradation of C18 (E2) compounds. This is in accordance with recent work where a grea<sup>t</sup> level of conservation of predicted CD-ring degradation genes was found among di fferent genera of steroid-metabolizing proteobacteria [47,48].

In *N. tardaugens*, only few genes required for C-19 degradation are located outside the SD cluster. We speculate that the genes *EGO55\_02230* (*hsd*), *EGO55\_13510* (*kstD*) or *EGO55\_13445* (*kshA*), that are required to metabolize C19 compounds, but that are not included in the SD cluster, might be involved in the aerobic degradation of other steroid compounds, so its location outside the cluster suggests that they could be playing a more global role.

It is interesting that *kshB* (*EGO55\_04915*) (*ORF17* in *C. testosteroni*) is not present in the SD cluster and no homologous gene to *tesA1* is found in the genome of *N. tardaugens*. KshB and TesA1 are the flavin reductase components of the flavin-dependent two-component monooxygenases KshA and TesA2, respectively. It has been demonstrated that these type of reductase components act in trans and are not highly specific and does not require a particular interaction [49,50]. However, it cannot be discarded that the reduced flavin required by KshA and TesA2 should be provided by unspecific reductases in *N. tardaugens*.

There are some genes included within the SD cluster of *N. tardaugens* that are not found in the TES cluster of *C. testosteroni*. This is the case of *EGO55\_13755*, *EGO55\_13720* and *EGO\_13715*, which encode a nuclear transport factor 2 family protein, a lipid transfer protein and a benzoylsuccinyl-CoA thiolase, respectively. The precise role of these enzymes in the SD cluster remains unknown. On the contrary, there are some genes located within the TES cluster of *C. testosteroni* that have been located outside of the SD cluster. This is the case of *EGO55\_13615* gene that is homologous to tesI gene from *C. testosteroni* encoding a ketosteroid Δ4-dehydrogenase, involved in epiandrosterone degradation [51]. In addition, the *ORF25* and *ORF26* genes of TES cluster encoding a 6-aminohexanoate-cyclic-dimer hydrolase are homologous to the *EGO55\_02680* and *EGO55\_04710* of *N. tardaugens*, respectively, that are located far from the SD cluster. These genes are also absent in the SD cluster of other estrogen-degrading strains like *Sphingomonas* sp. KC8 and *Altererythrobacter estronivorus* [18]. The role of these genes in TES catabolism in *C. testosteroni* is not clear since their disruption does not impair TES degradation [26].

Remarkably, we have detected a large number of genes dispersed in the genome of *N. tardaugens* that are homologous to those contained in the SD cluster and are expressed at significant levels in the presence of TES. This is the case of *EGO55\_15045* (*tesD* homologous), *EGO55\_02915* (*tesF* homologous), *EGO55\_02335* (*ksi* homologous) and *EGO55\_13440* (*tesA2* homologous), *EGO55\_09255* and *EGO55\_20150* (*ORF23* homologous), and *EGO55\_01175* (*kstD* homologous) (Figure 1, Table S5). Therefore, we cannot discard the possibility that these genes could be involved in the metabolism of C-19 compounds in *N. tardaugens* enhancing the versatility and robustness of steroid metabolism.

In spite of the fact that the expression of other C-19 clusters, as occurs in *C. testosteroni*, are regulated by a specific regulator, in ARI-1 the data obtained from the transcriptomic analyses showed only a slight di fferential expression of the genes from the SD cluster in the presence of TES (Table S6). The catabolic enzymes for TES degradation described so far were not constitutively expressed but, rather, were significantly induced by their respective substrates [39,52]. For instance, in *C. testosteroni* the LuxR-type transcription activator TeiR regulates the transcription of genes involved in the initial enzymatic steps of TES degradation [39]. Moreover, the *teiR* deletion mutant was not able to use TES as a carbon source. TesR from *C. testosteroni* TA441, almost identical to that of TeiR from *C. testosteroni* ATCC11996, was shown to be necessary for induction of the steroid degradation gene clusters, *tesB* to *tesR*, *tesA1* to *tesG*, and *tesA2* to *ORF18* in *C. testosteroni* TA441 [53]. Interestingly, it has been reported that *tesR*-like regulatory genes are only present in *C. testosteroni* strains but are not found in other testosterone-degrading bacteria [26]. The in silico analysis showed that neither a *teiR* homologue nor other putative regulatory genes are present in the vicinity of the SD cluster in ARI-1. This observation is consistent with the transcriptomic analysis revealing a significant basal expression of the SD cluster in the absence of TES, which might be explained because this pathway could be fundamental for the survival of the ARI-1 strain in the specific niche where it was isolated.

There are other genes that appear to be induced by pleiotropic induction e ffects in the presence of TES and which are not directly related to the degradative C-19 pathway but rather to central metabolic processes (e.g., stress processes, requirements of cofactor synthesis, etc.). For instance, the production of propionyl-CoA as a presumable metabolite of TES degradation or the requirements of CoA to mineralize the TES rings can promote the generation of multiple stresses and pleiotropic di fferential expression signals when compared with the transcriptome of pyruvate metabolism [54,55]. According to the metabolic scheme shown in Figure 1, TES would be converted into succinyl-CoA (1 mol), pyruvate (1 mol), acetyl-CoA (3 mol) and propionyl-CoA (2 mol). The analysis of the genome revealed that *N. tardaugens* does not have a methylcitrate cycle and thus, it metabolizes propionyl-CoA by the methylmalonyl-CoA pathway to synthesize succinyl-CoA. The finding that the methyl-malonyl-CoA pathway is highly upregulated under TES growth conditions, is a solid evidence that propionyl-CoA is formed.

3β/17β-HSDs are essential enzymes in the biosynthesis of all classes of mammalian steroids. They catalyze the interconversion of alcohol and carbonyl functions stereospecifically in defined positions using oxidized or reduced NAD(H) or NADP(H) as co-substrates [56]. It is also well known that 3β/17β-HSDs are involved in the first catabolic step in TES degradation and, particularly, the 3β/17β-HSD of *C. testosteroni* has been extensively studied [20,46,57,58]. It catalyzes the reversible reduction/dehydrogenation of the oxo/β-hydroxy groups at positions C3 and C17 of steroids, including hormones and isobile acids. The dual positional specificity of this 3β/17β-HSD has been explained after resolving its 3D structure [46]. Kinetic studies revealed an ordered mechanism and suggested a single catalytic site accommodating both the 3β and 17β activities [46]. *N. tardaugens* has 16 genes that might encode putative 3β/17β-HSD homologous to the enzyme described in *C. testosteroni* (Table S5). The transcriptomic data obtained in the presence of TES allowed us to postulate specific candidates that could be involved in the metabolism of TES. Furthermore, after exploring the genetic environment of the 16 candidates, we identified the *EGO55\_02235*-*EGO55\_02230* tandem genes that code for two

putative 3β/17β-HSD isoenzymes (Hsd60 and Hsd70) as part of a putative four-genes operon together with the *EGO55\_02225* gene coding for a putative esterase and the *EGO55\_02240* gene encoding a putative permease of the major facilitator superfamily. Taking into account all these reasons we cloned and tested the activity of the two putative HSDs of this operon demonstrating that they have a differential activity on C-19 steroids (Figure 6). Our results showed that isoenzyme Hsd60 is more efficient performing the C17-OH dehydrogenation than Hsd70 (Figure 6a,b), which in turn performed more efficiently the dehydrogenation of C3-OH (Figure 6c,d). The substrate specificity of Hsd60 makes this enzyme a very interesting candidate for the development of biocatalysts for TES production using mycobacterial strains that accumulate AD as described [59]. The alignment of the two proteins showed a variable region at the carboxyl-terminal domain (196*–*230 residues) that only share 42% amino acid sequence identity, in contrast with the 80% of the whole polypeptide chain. It has been described that the steroid binding pocket in 3<sup>α</sup>, 20β-HSD from Streptomyces hydrogenans is formed by the carboxyl-terminal 60 residues [60]. Interestingly, the sequence alignment of 3<sup>α</sup>, 20β-HSD with Hsd60 and Hsd70 showed that the variable region detected for both isoenzymes align perfectly with the sequence responsible for steroid binding at 3<sup>α</sup>, 20β-HSD (Figure S3). The lower similarity of this region is consistent with the variable substrate specificity observed between the two isoenzymes.

It is well known that many steroids are currently used as pharmaceutical drugs in their esterified forms. The esterification of natural or synthetic androgens or anabolic steroids renders metabolic resistant prohormones with improved oral bioavailability, increased lipophilicity, and extended elimination half-life. This is for instance the case of TES that can be esterified with acetate (TES-Ac) that was first described in 1936 and was one of the first androgen esters to be synthesized and used as an anabolic steroid. To introduce steroid esters in the C-19 degradation pathway they should be firstly hydrolyzed by an esterase. On the other hand, the uptake of these lipophilic xenobiotic compounds might require a specific transport system to cross the two membranes of a Gram-negative bacterium. Then, the production of an esterase together with two HSD enzymes and the assistance of a transport protein appear to constitute an efficient system to handle the degradation of these pharmaceutical compounds that frequently contaminate the municipal waste waters from where *N. tardaugens* was isolated [7]. In agreemen<sup>t</sup> with this hypothesis, we have demonstrated that *N. tardaugens* is able to catabolize TES-Ac (Figure 2).

The fact that *N. tardaugens* can degrade both C3-OH and C17-OH steroids (Figure 2) supports the existence of both HSD activities. This finding does not rule out that some of the other HSD homologous enzymes encoded in the genome of *N. tardaugens* can fulfil these activities and additional experiments should be carried out to demonstrate a specific implication of this operon in the metabolism of steroids. Nevertheless, this result opens a new and interesting scenario to study in more detail the degradation of xenobiotic steroid esters of EDCs.

Several examples of bacteria capable of using both cholesterol and C-19 steroids as the only carbon and energy source have been described (e.g., *Rhodococcus erythropolis* strain SQ1 [61], *Rhodococcus ruber* strain Chol-4 [62], *Rhodococcus rhodochrous* DSM 43269 [63], *Gordonia neofelifaecis* NRRL B-59395 [64]). Using the genes of the SD cluster of *N. tardaugens* as a template, we screened the presence of homologous genes in other steroid metabolizing bacteria (Table S8). Apart from *C. testosteroni* TA441, we observed homologous genes in *Sphingomonas* sp. KC8, *Pseudomonas* sp. Chol1, *Rhodococcus jostii* RHA1, *Mycobacterium tuberculosis* H37Rv, *Mycobacterium smegmatis* mc2155, *Altererythrobacter estronivorus* MH-B5 and *Sterolibacterium denitrificans* Chol-1S(T) genomes. In spite of their large phylogenetic distances, a significant identity was observed with some Actinobacteria, e.g., RHA1, H37Rv and mc2155 strains (Table S8). This suggests a grea<sup>t</sup> level of conservation of key degradative enzymes among bacteria adapted to metabolize different steroidal compound.

Genetic manipulation of the ARI-1 strain will pave the way to unravel in more detail not only TES degradative pathway but also the other pathways involved in the degradation of E2 and several steroids that are used as carbon and energy sources by this bacterium. The data presented will enable to build upon the knowledge on metabolic pathways and the biotransformation capabilities of this

Gram-negative bacterium that could become a new, model system in the steroid field. Moreover, *N. tardaugens* NBRC 16725 (strain ARI-1) might cover a wide spectrum of steroid biotransformation reactions and their improvement may lead to promising alternative biotechnological processes.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4425/10/11/871/s1, Figure S1: 12.5% SDS-polyacrylamide gel electrophoresis of the overproduction of Hsd70-Hsd60, Hsd60 and Hsd70 proteins in the soluble fraction of the crude extract of *E. coli* BL21(DE3) strains.Twenty-five μg of total protein of each sample where loaded, Figure S2: Bacterial growth (Log2OD600) of *N. tardaugens* NBRC 16725 (red), *N. tardaugens* Δ*fadD3* (yellow), *N. tardaugens* Δ*tesD* (purple) and *N. tardaugens* Δ*hsd* (green) strains when cultured in M63 minimal medium containing 1.89 mM TES, Figure S3: Protein alignment of the 3<sup>α</sup>,20β-HSD from *Streptomycens hydrogenans* with HSD60 and HSD70. The key amino acids of the active site are in red. The variable sequence is shown in a red box. This sequence is involved in substrate binding in the 3<sup>α</sup>, 20β-HSD from *S. hydrogenans*, Table S1: Isolated bacteria able to degrade E2 and/or TES, Table S2: Bacterial strains and plasmids used in this study, Table S3: Primers used in this study, Table S4: Statistical results of whole genome sequencing and mapping of all transcripts, Table S5: Genes found in *N. tardaugens* genome homologous to genes involved in testosterone degradation in *C. testosteroni* TA441. Percentage of identity (ID %) and fold change (FC) increase in expression levels when *N. tardaugens* grows in testosterone are shown, Table S6: Gene expression analysis (RNA-seq) of *N. tardaugens* grown in testosterone condition compared to pyruvate. Genes located in the SD cluster (blue) and those involved in methylmalonyl-CoA pathway (green) and cofactor B12 biosynthesis pathway (orange) are highlighted, Table S7: Methylmalonyl-CoA degradation cluster in *N. tardaugens* NBRC 16725. Genes homologous to those described as involved in the pathway are highlighted in green, Table S8: Steroid degradation genes in the putative testosterone degradation pathway of *N. tardaugens* (accession CP034179). Homologous genes found in the genomes of *C. testosteroni* TA441 (accession LC010134), *Sphingomonas* sp. KC8 (accession CP016306), *Pseudomonas* sp. Chol1 (accession AMSL00000000), *R. jostii* RHA1 (accession CP000431), *M. tuberculosis* H37Rv (accession AL123456.3), *M. smegmatis* mc2 155 (accession CP000480), *A. estronivorus* MH-B5 (accession NZ\_JRQQ00000000) and *S. denitrificans* Chol (accession LT837803) are listed and the percentage identity is shown. A cut-off value of 39 % identity was used.

**Author Contributions:** J.I. carried out molecular genetic studies, in silico analysis, fermentation experiments and helped to draft the manuscript. E.D. and J.L.G. conceived the study. J.L.G. and B.G. coordinated the study and drafted the manuscript. All authors read, reviewed and approved the final manuscript.

**Funding:** This research was funded by Ramón Areces Foundation.

**Acknowledgments:** The bioinformatics support of the Bioinformatics and Biostatistics Service of CIB particularly Guillermo Padilla Alonso is greatly appreciated. The technical work of A. Valencia is greatly appreciated. We acknowledge support of the publication fee by the CSIC Open Access Publication Support Initiative through its Unit of Information Resources for Research (URICI). We also acknowledge the financial support provided by the Ramón Areces Foundation.

**Conflicts of Interest:** The authors declare no conflicts of interest.
