2.4.2. RNA-seq

RNA-seq was done in Macrogen Korea. Total RNA integrity was checked using an Agilent Technologies 2100 Bioanalyzer. Ribosomal RNA was removed from the total RNA with Ribo-Zero rRNA Removal Kit to later construct a 100 bp paired-end library using TruSeq RNA Sample Prep Kit v2 that was quality-checked in an Agilent Technologies 2100 Bioanalyzer using a DNA 1000 chip. Library sequencing was performed in a HiSeq 3000 4000 (Illumina) using TruSeq 3000 4000 SBS Kit v3 as reagent. Bioinformatic analysis was performed by the Bioinformatics and Biostatistics service of the Centre for Biological Research (CIB-CSIC). Raw read data quality was checked using FastQC and trimmed with Trimmomatic. Trimmed reads were mapped against the genome sequence of *N. tardaugens* (accession number CP034179) using Bowtie2 and expression quantification was done using HTSeq-count. An average of 50 million raw sequencing reads (approximately 6.7 billion base pairs; average 1300× genome coverage per sample) were generated from samples from two independent experiments in the presence of pyruvate or TES, each with three biological replicates. After trimming the raw sequence reads, an average 48.7 million high-quality clean reads were mapped to the *N. tardaugens* reference genome and between 98.4% and 82.5% were uniquely mapped (Table S4). Differential expression analysis was done using Deseq2 and GO-term enrichment analysis was performed with GOSeq. The dissimilarity matrix shown in the heatmap was obtained with the euclidean distance and the cluster analysis was performed with the Ward's minimum variance method. Bioinformatic analysis software was used with default settings. Raw read data obtained from the three replicates of the transcriptome of the strain grown in pyruvate and TES have been deposited in the Sequence Read Archive (SRA) database of the National Centre for Biotechnology Information (NCBI) under accession numbers SRR9027780, SRR9027781, SRR9027779 (Bioproject PRJNA541800) and SRR9027897, SRR9027898, SRR9027896 (Bioproject PRJNA541801), respectively.

#### *2.5. Isolation of a Rifampicin Resistant Phenotype of N. tardaugens*

*N. tardaugens* wt strain, sensitive to rifampicin, was cultured in NB medium supplemented with rifampicin up to stationary phase (≈48 h). Cells where then plated on NB plates supplemented with the antibiotic and one single colony was picked, grown and used as *N. tardaugens* Rfr, suitable for conjugation.

#### *2.6. Construction of N. tardaugens Knockout Strains*

The knockout strains were constructed by double homologous recombination using the suicide vector pK18mobsacB [33]. *N. tardaugens* genomic DNA was used as template to amplify two fragments of ≈700 bp containing the upstream and downstream regions of the gene to delete. The fragments were digested with the appropriate restriction enzymes and cloned in the unique sites of the plasmid. The ligation product was transformed into *E. coli* DH10B competent cells and once recombinant candidates were PCR-checked, the cloned region was confirmed by sequencing. The plasmids were transformed by triparental conjugation [34] into *N. tardaugens* Rfr as recipient strain using *E. coli* HB101 (pRK600) [35] as helper and *E. coli* DH10B, harbouring the corresponding vector, as donor. The strains resulting of plasmid integration in this first recombination event were selected in NB agar plates containing kanamycin and rifampicin and screened by PCR. Selected candidates were grown up to stationary phase (≈48 h) in NB medium and then plated in NB supplemented with 5% sucrose. The clones that are resistant to sucrose and sensitive to kanamycin were checked by PCR and the amplicon was sequenced to confirm the second cross-over event.

#### *2.7. Heterologous Production of the Putative 3*β/*17*β*-Hydroxysteroid Dehydrogenase from N. tardaugens*

To overproduce the putative 3β/17β-hydroxysteroid dehydrogenase (3β/17β-HSD), genes *EGO55\_02235*-*EGO55\_02230* (*hsd70-hsd60*) were cloned together or separately into the expression vector pET-29a. The DNA fragments containing *EGO55\_02235-EGO55\_02230, EGO55\_02235* and *EGO55\_02230* were amplified with primers 5NdeIhsdTandemEcolif-3BamHIhsdTandemEcolir, 5NdeIhsd70SUBf-3XhoIhsd70SUBr, 5NdeIhsd60SUBf-3XhoIhsd60SUBr, respectively (Table S3), using *N. tardaugens* genomic DNA as template. The DNA fragments were digested with the corresponding restriction enzymes *Nde*I-*Bam*HI, *Nde*I-*Xho*I and *Nde*I-*Xho*I respectively, and then ligated into the pET29a vector yielding pET29Hsd70-Hsd60, pET29Hsd70 and pET29Hsd60 plasmids, respectively. Electrocompetent cells of *E. coli* BL21 (DE3) were transformed with plasmids pET29Hsd70-Hsd60, pET29Hsd70 and pET29Hsd60. The resulting strains *E. coli* BL21 (DE3) (pET29Hsd70-Hsd60), *E. coli* BL21 (DE3) (pET29Hsd70) and *E. coli* BL21 (DE3) (pET29Hsd60) were grown in 50 mL LB containing kanamycin up to an OD600 of 0.5–0.8 and gene expression was induced with 0.2 mM IPTG (isopropyl-β-D-thiogalactopyranoside). After 3 h of induction, the cells were harvested, washed with 0.85% saline solution and resuspended in 20 mM Tris-HCl (pH 8.0). Cells were lysed by sonication using a Branson sonicator applying three cycles of 30-s bursts at maximum power alternated with 30 s cooling in ice. Soluble fraction was separated by centrifugation in a Sorvall Linx 6000 SS-34 rotor (15 min at 4 ◦C and 14000 rpm) and protein concentration was calculated using the Bradford method [36]. The overproduction of Hsd70 and Hsd60 proteins in the soluble fraction of the crude extract was checked by SDS-polyacrylamide gel electrophoresis (Figure S1).

#### *2.8. Enzymatic Assay of 3*β/*17*β*-HSD Activity*

Enzymatic assays of 3β/17β-HSD activity were performed in 500 μL total reaction volume. Reaction mixture consisted of 0.5 mg protein of the crude extracts obtained from the recombinant strains, 500 μM of the steroidal substrate dissolved in CDX (3.5 mM final concentration of CDX), 1 mM NAD+ and 50 mM sodium phosphate buffer (pH 7.0). Reaction assays were started by adding NAD+ and stopped by adding 2 volumes of chloroform.

#### *2.9. Organic Phase Extraction and Thin Layer Chromatography (TLC) analysis*

The presence of steroidal compounds in culture media and enzymatic assay mixtures was determined after organic solvent extraction by TLC analysis. Two volumes of chloroform were added and the mixture was vortexed for 30 s and centrifuged for 1 min at 13,000 rpm in an Eppendorf microcentrifuge. The organic phase was extracted and dried. The dried sample was dissolved in 100 μL of acetonitrile and analysed by thin layer chromatography (TLC). For TLC analysis, 10 μL of the standards and the samples dissolved in acetonitrile were spotted in silica gel plates (TLC Silicagel 60 F254, Merck Millipore) and n-hexane: ethyl acetate (10:4 v/v) was used as developing system. Steroid products were visualized by UV and revealed by spraying 20% (v/v) sulphuric acid and heating at 120 ◦C.

#### *2.10. In Silico Analyses*

Gene product prediction was done with Rapid Annotations using Subsystems Technology (RAST) [37]. Homologous genes search in di fferent bacteria was performed by using the Standard Protein Basic Local Alignment Search Tool (BLASTp) [38].
