Channeling C1 Metabolism toward S-Adenosylmethionine-Dependent Conversion of Estrogens to Androgens in Estrogen-Degrading Bacteria

Estrogens comprise a group of related hormones occurring in predominantly female vertebrates, with endocrine disrupting and carcinogenic potential. Microbial biodegradation of estrogens is essential for their elimination from surface waters and wastewater. Aerobic bacteria employ oxygenases for the initial cleavage of the aromatic ring A. In contrast, anaerobic degradation of estrogens is initiated by methyl transfer-dependent conversion into androgens involving a putative cobalamin-dependent methyltransferase system. The methyl donor for this unprecedented reaction and its stoichiometric regeneration have remained unknown. With the biomass obtained from large-scale fermentation of an estrogen-degrading denitrifying bacterium, we identified S-adenosyl-methionine (SAM) as the methyl donor for the cobalamin-mediated methyl transfer to estrogens. To continuously supply C1 units to initiate estrogen degradation, genes for SAM regeneration from estradiol-derived catabolites are highly upregulated. Data presented here shed light into biochemical processes involved in the globally important microbial degradation of estrogens.

IMPORTANCE Estrogens comprise a group of related hormones occurring in predominantly female vertebrates, with endocrine disrupting and carcinogenic potential. Microbial biodegradation of estrogens is essential for their elimination from surface waters and wastewater. Aerobic bacteria employ oxygenases for the initial cleavage of the aromatic ring A. In contrast, anaerobic degradation of estrogens is initiated by methyl transfer-dependent conversion into androgens involving a putative cobalamin-dependent methyltransferase system. The methyl donor for this unprecedented reaction and its stoichiometric regeneration have remained unknown. With the biomass obtained from large-scale fermentation of an estrogen-degrading Co(II) state, which can be reactivated by ATP- (19)(20)(21) or SAM-dependent (22) electron transfer. In the vicinity of the emtABCD genes, a cluster of three further genes was assumed to play a role in ATP-dependent reductive regeneration of the cob(I)alamin state. It included a putative RACE-like protein (reductive activator of corrinoid enzyme) (19)(20)(21)23). The 1-dehydrotestosterone (DHT) formed from estradiol is subsequently degraded by oxygen-independent reactions of the 2,3-seco-pathway, as evidenced by the identification of the corresponding genes in the genome of strain DHT3 (15). This pathway was established for anaerobic cholesterol and androgen degradation (24)(25)(26).
The milestone discovery of methyl transfer-dependent retroconversion of estrogens into androgens gave rise to a number of fundamental questions. In particular, the source of the methyl group and its continuous regeneration for initiating oxygenindependent estrogen degradation have remained unclear. Further, the proposed unprecedented methyl transfer from methylcobalamin to C10 of estrogen, coupled to dearomatization, is thermodynamically unfavorable. As understood so far, methyl 17␤-estradiol degradation in denitrifying bacteria initiated by cobalamin-mediated methyl transfer resulting in the retroconversion of 17␤-estradiol to 1-dehydrotestosterone that may be reduced to 1-testosterone or oxidized to androsta-1,4-diene-3,17-dione (ADD); likewise, estrone is methylated to ADD. All androgen intermediates are converted to androst-1-ene-3,17-dione (1-AD), the common intermediate of the 2,3-seco pathway. EmtAB, the proposed subunits of estrogen methyl transferase. transfer from methylcobalamin usually requires strong nucleophiles as methyl acceptors, such as thiols (L-homocysteine/coenzyme M), R-NH 2 (tetrahydrofolate, THF), or Ni(I) (A-cluster in acetyl-CoA synthase) (22,(27)(28)(29). Here, we identified SAM as the stoichiometric methyl group donor for cobalamin-mediated estrogen methylation; its continuous regeneration is sustained by the upregulation of C1 metabolism and the channeling of its metabolites toward estrogen catabolism.

RESULTS
Large-scale cultivation of Denitratisoma oestradiolicum for quantitative in vitro assays. In the previous study by Wang et al. (15), the conversion of estrogens to androgens in strain DHT3 was demonstrated in whole-cell suspensions and cell extracts. In assays with the latter, the limiting biomass available allowed solely qualitative endpoint determination of product formation after 1.5 h of incubation. The observed ATP dependence of estrogen methylation by cell extracts was assigned to the reductive reconstitution of an active cob(I)amide from an inactive cob(II)amide, probably catalyzed by a RACE-like protein (19,20). The observed slightly stimulating, but nonstoichiometric effect of SAM on estrogen methylation was explained by catalytic reconstitution of methylcob(III)amide as described for methionine synthase (30). The proposed product of estradiol methylation, DHT, was not found in in vitro assays. Instead, androgen products with fully reduced ring A were identified, which, however, are not intermediates of the proposed 2,3-seco pathway (24,25). In summary, the nature of the product, the source of the stoichiometric methyl donor, and the cob(II)alamin activation system of estrogen methylation were at issue.
To address these open questions by quantitative biochemical approaches, D. oestradiolicum was chosen as a model organism for anaerobic estrogen degradation. In preliminary cultivation trials with estradiol (2 mM) and nitrate (15 mM) as carbon and energy sources, the bacterium reached higher densities compared to strain DHT3 (15). After large-scale cultivation in a 200-liter fermenter with a doubling time of 16 h in the early exponential phase, 122 g of cells (wet mass) was obtained, with 30 Ϯ 5 mg protein per g wet cell mass (for the growth curve see Fig. S1 in the supplemental material). This served as biomass for the biochemical assays.
SAM-dependent methylation of estrogens to androgens. We tested the effect of various potential cosubstrates involved in methyl transfer and cob(I)alamin reconstitution by cell extracts comprising SAM (0.5 mM), N 5 -methyl-THF (0.5 mM), and methylcobalamin (0.5 mM) as potential methyl donors, and Ti(III)-citrate (1.5 mM), NAD(P)H (0.5 mM), and ATP (1 mM) as components involved in the reduction of the accidentally formed inactive cob(II)alamin. Strictly depending on SAM and Ti(III)-citrate, the timeand cell extract-dependent conversion of estradiol (50 M) to DHT and a number of downstream products was observed by ultraperformance liquid chromatography (UPLC) analyses at a rate of 0.15 Ϯ 0.01 nmol min Ϫ1 mg Ϫ1 (Fig. S2 in the supplemental material). The formation of estrone, androsta-1,4-diene-3-one (ADD), and androst-1-en-3,17-dione (1-AD) can be rationalized by the presence of catalytic amounts of NAD ϩ in the cell extracts that serves as electron acceptor for the oxidation of the C-17 hydroxyl groups of estradiol and DHT to estrone and ADD, respectively. The NADH formed in this reaction is then used for the reduction of ADD to 1-AD and DHT to 1-testosterone (1-T), respectively, which regenerates NAD ϩ (Fig. 1C). To minimize the formation of downstream products, estradiol was replaced by estrone as the substrate that was converted at a virtually identical rate. In these assays, ADD was identified as the main product and 1-AD as the minor second product, whereas formation of further downstream products was negligible ( Fig. 2A and B). Substitution of the artificial electron donor Ti(III)-citrate by the substrates of a putative RACE component, ATP and NAD(P)H, resulted in only a marginal methyl transfer rate. When SAM was replaced by the alternative methyl donors N 5 -methyl-THF or methylcobalamin, virtually no conversion of estradiol was observed. However, when ATP was added in combination with N 5 -methyl-THF, maximally 10% of the activity was recovered. These results suggest that SAM functions as the methyl donor for estradiol methylation and that a low-potential electron donor is essential for keeping the cofactor of estradiol methyltransferase (EMT) in the active cob(I)alamin state. The slight stimulating effect of ATP in assays with N 5 -methyl-THF suggests that the latter may serve as methyl donor for SAM synthesis via the cobalamin-containing methionine synthase and ATP-dependent methionine adenosyltransferase, providing that catalytic amounts of SAM or precursors of it, such as L-methionine, L-homocysteine, L-serine, etc., were present in the extracts used.

SAM-and B12-Dependent Estrogen Methylation
In previous in vitro assays with strain DHT3, androgens with a fully reduced ring A in place of DHT were identified (15). The latter is an intermediate of the testosterone catabolic 2,3-seco pathway (24,25), whereas androgens with a fully reduced ring A are not. They most likely arose as side products formed by an unspecific dehydrogenase. Such side reactions were not observed in the in vitro assays with extracts from D. oestradiolicum grown with estradiol.
Kinetic parameters, stoichiometry, and inhibition of estrogen methylation. The K m -values of the methyltransferase for estradiol and estrone were 35 Ϯ 4 and 14 Ϯ 3 M, respectively (for Michaelis-Menten curves see Fig. S3 in the supplemental material). For determining the stoichiometry of methyl transfer from SAM to estrone, cell extracts were subjected to ammonium sulfate precipitation to remove nonproteinaceous molecules that may interfere with the determination of products. Using the solubilized and washed methyltransferase activity containing 70% ammonium sulfate fraction, the conversion of SAM to S-adenosyl-L-homocysteine (SAH) and estrone to ADD was followed with 0.8 mol SAH per mol ADD formed (Fig. S4 in the supplemental material). As SAM-dependent methyltransferases are typically inhibited by the product SAH (31), we tested the effect of SAH on SAM-dependent estradiol methylation. Even in the presence of 200 M SAH, no significant inhibition was observed (Fig. 3A); SAH degradation by other enzymes was negligible in the course of the reaction. Instead, the product DHT inhibited estradiol methylation in a dose-dependent manner (Fig. 3B).
Partial methyl transfer activities and the reverse reaction. Cell-free extracts catalyzed the reverse demethylation of DHT to estradiol (Fig. 4A). By increasing the amount of cell extract 4-fold, the amount of estradiol formed from DHT increased likewise from 2.3 Ϯ 0.4 M to 10.4 Ϯ 0.3 M. This result suggests a stoichiometric methyl transfer from DHT to an abundant proteinaceous acceptor, most likely the corrinoid protein of EMT, with 2.8 Ϯ 0.3 nmol methyl group transferred per mg protein in the cell extract. Addition of SAH did not increase the amount of estradiol formed from DHT further, indicating that the proposed methylcobalamin intermediate formed from DHT is not competent for SAH methylation. To test whether the proposed methylcobalamin intermediate is competent for methyl transfer to estradiol (forward reaction), we incubated extract with either SAM or DHT in the absence of estradiol, and washed the SAM-or DHT-methylated extract by ammonium sulfate precipitation to remove excess of SAM/DHT or other potential low-molecular methyl donors. Using these extracts, virtually no methyl transfer to estradiol was observed. In a control experiment, this extract catalyzed SAM-dependent estrogen methylation at the expected rate (Fig. 4B). Hence, the proposed methylcob(III)alamin intermediate was incompetent for methyl transfer to estradiol in the absence of SAM. In summary, we conclude that exergonic SAM-dependent methylation of the cob(I)alamin cofactor of EMT drives the unfavorable methyl transfer from the methylcob(III)alamin intermediate to estradiol forward. High-quality circular genome of D. oestradiolicum DSM 16959. In the recent study by Chen et al. (32), a draft genome of D. oestradiolicum DSM 16959 was generated based on short read Illumina sequencing. Though the authors produced a high-quality sequence for strain DHT3 by leveraging long read sequencing using the nanopore technology, their sequence for strain DSM 16959 was left highly fragmented into 67 contigs (65 scaffolds). To provide a basis for extensive gene neighborhood investigations and to support high quality proteogenomic analyses, we generated a high-quality genomic sequence for D. oestradiolicum DSM 16959 through a hybrid approach combining both long and short reads. A single contig corresponding to the main bacterial chromosome was obtained, encompassing 4168563 bases (62% GC content) and harboring 3,986 coding sequences. The sequence of a 124849-base plasmid (63% GC content), unnoticed in the earlier draft assembly and encompassing 187 coding sequences, was reconstructed. Both sequences are available from the ENA/NCBI/DDBJ databases under accession numbers LR778301 and LR778302, respectively (study identifier PRJEB37013). Beyond gene neighborhood analyses, the availability of a fully contiguous genome sequence enables comparative genomics investigations, e.g., the identification of close genomes based on genome-level gene syntenies (Table S1 in the supplemental material). In particular, a direct comparison of the DHT3 and DSM 16959 genomes in terms of conserved gene synteny groups (syntons) highlighted significant differences in genome organization between the two strains ( Fig. S5 in the supplemental material), consistent with the noticeable difference in genome length (around 500 kb).
Proteogenomic analysis of anaerobic estrogen degradation. To identify estrogeninduced genes in D. oestradiolicum involved in the initial methyl transfer reaction, methyl donor regeneration, and sterane skeleton degradation, the proteome of D. oestradiolicum cells grown with estradiol was compared with that from cells grown with acetate (Fig. 5, for original data, see Table S2 in the supplemental material). Proteome analyses of biological triplicates for each growth substrate revealed high reliability of the data obtained (Fig. 5A). Genes involved in estrogen catabolism were categorized according to their function in estrogen activation (Fig. 5B), rings AB, and rings CD degradation (Fig. S6 in the supplemental material). They were assigned to these functions based on their higher abundance in estrogen-versus acetate-grown D.
oestradiolicum, and on clear similarities with experimentally verified gene products (up to 98% amino acid sequence similarity). The best hits with previously identified genes from betaproteobacterial, gammaproteobacterial, or actinobacterial steroid degraders are summarized in Table S3 in the supplemental material.
In accordance with previous transcriptome analyses (15), the EmtABCD gene products (DENOEST_v1_1167-1170), and a RACE-like protein (DENOEST_v1_1164) were clearly more abundant in estradiol-versus acetate-grown cells (log 2 Ͼ 25) (Fig. 5). In addition, a number of gene products involved in SAM regeneration were identified at higher abundances in estradiol-versus acetate-grown cells, including adenosyl-L-homocysteinase, adenosine kinase, L-serine hydroxymethyltransferase, N 5 ,N 10 -methylene-THF reductase, methionine synthase, and methionine adenosyltransferase (log 2 Ͼ 25, Tables S3 and S4 in the supplemental material). The genes are located in the vicinity of the emtABCD genes and share up to 98% amino acid sequence identity to homologous genes in the genome of strain DHT3. The identification of these estrogen-induced genes escaped detection during previous transcriptome analyses in strain DHT3. This finding may be explained by the fact that we compared cells grown with estradiol versus acetate and not estradiol versus testosterone. Second copies of genes involved in regeneration of SAM from L-serine and L-methionine were identified in the genome of D. oestradiolicum and were equally abundant in cells grown with estradiol and acetate (Table S4 in the supplemental material), suggesting that these second copies are involved in housekeeping C1 metabolism. In agreement with the transcriptome analyses in DHT3 (15), the abundance of gene products involved in vitamin B 12 uptake and modification was increased by estradiol in D. oestradiolicum, confirming that all anaerobic estradiol degraders known so far depend on B 12 in the culture medium. These species upregulate genes involved in uptake and processing B 12 -derived precursors to the active cobamide cofactor.
Based on similarities with corresponding genes from other cholesterol or androgen degraders (Table S2 and S3 in the supplemental material), gene products putatively involved in sterane skeleton degradation were identified at higher abundances in D. oestradiolicum cells grown with estradiol versus acetate (Fig. S6 in the supplemental material). The corresponding ␤-oxidation genes are organized in two clusters, with each cluster being assigned to the degradation of rings AB and CD, respectively. The gene products putatively involved in rings AB degradation show high amino acid sequence similarities to those putatively involved in ␤-oxidation-like reactions from anaerobic steroid degraders, but only minor sequence similarities to those from aerobic degraders ( Table S3 in the supplemental material). In contrast, gene products putatively involved in rings CD degradation are also highly similar to genes that have been identified in aerobic steroid degradation (13). These finding are consistent with the general assumption that the degradation of rings AB differs in aerobic and anaerobic steroid-degrading bacteria, whereas rings CD degradation proceeds via similar enzymes (24)(25)(26).
Methyl-13 C-L-methionine as methyl donor for estrogen and cobalamine methylation. Results from differential proteome analyses suggest that SAM used for estradiol methylation is regenerated via L-methionine that is formed by L-methionine synthase from L-homocysteine and N 5 -methyl-THF (Fig. 5C). L-Serine is assumed to serve as the major source for loading THF with a C1 unit by L-serine hydroxymethyltransferase. Based on these findings, we designed in vitro assays with cell extracts in which SAM was replaced by either (i) L-methionine/ATP via L-methionine adenosyltransferase, or (ii) by L-serine/ATP/THF/NADH via L-methionine adenosyltransferase, SAM-and B12-Dependent Estrogen Methylation ® N 5 ,N 10 -methylene-THF reductase, and L-serine hydroxymethyltransferase. We determined the time-and cell extract-dependent conversion of estrone in the absence of SAM with both setups, at rates of 0.14 and 0.07 nmol min Ϫ1 mg Ϫ1 , respectively (Fig. 6A). Ultraperformance liquid chromatography-high-resolution mass spectrometry (UPLC-HRMS) analysis confirmed that with L-serine and L-methionine the same product was formed as with SAM as donor (m/z ϭ 285.19, Fig. 6B). To further confirm methyl group transfer from L-methionine to estrone, we used methyl-13 C-L-methonine as the methyl donor and analyzed the product by UPLC-HRMS. The shift to m/z ϭ 286.19 is indicative for the transfer of the 13 C-methyl group to estradiol (Fig. 6B).
The possibility to follow methyl group transfer from L-methionine to estrone motivated us to use a similar setup for the identification of the anticipated methylcob(III) alamin intermediate. For this purpose, L-methionine and ATP were incubated with 25% (vol/vol) (NH 4 ) 2 SO 4 precipitated cell extracts in the absence of estrone. UPLC-HRMS analysis identified coeluting compounds with m/z ϭ 672.8 and 664.79, indicative of methyl(III)cobalamin and its demethylated form, respectively, as evidenced by comparison with an authentic methyl(III)cobalamin standard (double positively charged ions, Fig. 6C). In assays with methyl-13 C-L-methionine, the m/z peak at 664.79 was unaffected, whereas the methyl(III)cobalamin peak shifted by m/z ϭ 0.5 to 673.3 as expected for double positively charged 13 C-methyl(III)cobalamin ([M ϩ 1ϩH] 2ϩ ). In control experiments without ATP, virtually no 13 C-methyl(III)cobalamin was detected. Taken together, these finding corroborate the methyl(III)cobalamin intermediate of estrogen methyltransferase during SAM-dependent estrogen methylation.

DISCUSSION
This work identified SAM as the previously unknown stoichiometric methyl donor for the conversion of estrogens into androgens by a cobalamin-dependent methyltransferase in anaerobic bacteria. The combination of SAM and cobalamin cofactors has so far only been described for a few radical SAM enzymes involved in radical-based methyl transfer reactions (33). In these enzymes, one SAM abstracts a hydrogen atom from the substrate, and the radical formed then serves as methyl radical acceptor from methylcob(III)alamin. The latter is regenerated from the cob(II)alamin intermediate at the expense of a second SAM and an electron provided by an external donor. However, there is no evidence for the involvement of a radical SAM enzyme in anaerobic estradiol degradation. Instead, estradiol methylation is likely to occur via a nonradical mechanism with the deprotonated ring A phenolate serving as a suitable nucleophilic methyl cation acceptor (15). As genes encoding classical, nonradical SAM-dependent methyltransferases are absent in the estradiol-induced gene clusters, there remains little doubt that anaerobic estrogen degradation is initiated by an unorthodox, cobalamin-dependent SAM:estrogen methyltransferase, encoded by the estrogen-induced emtABCD genes.
Methylation of carbon atoms usually depends on SAM as methyl donor either via radical or methyl cation transfer mechanisms (34,35), but have never been reported for methylcobalamin-or methyl-THF-dependent methyltransferases. Our results suggest two partial reactions for the proposed SAM:estrogen methyltransferase system: (i) exergonic methyl transfer from SAM to cob(I)alamin, and (ii) endergonic methyl transfer from the methylcob(III)alamin intermediate to estradiol, with the first exergonic partial reaction pushing the second endergonic one forward (Fig. 7). A question rises; why there is no direct transfer from SAM to estradiol catalyzed by a conventional cobalaminindependent methyltransferase? It may be simply explained by the evolution of a methyl transfer system in which SAM-and cobalamin-dependent methyl transfer components of different origin have merged to EMT. However, one may also speculate that separation of SAM demethylation and estradiol methylation into two partial reactions may be beneficial for efficient estrogen methylation. E.g., unlike most of the known SAM-dependent methyltransferases, EMT is not inhibited by the product SAH, which appears to be important for unidirectional initiation of a catabolic pathway that provides the cell with carbon and energy. The use of a noncanonical SAM-and Our results indicate that estradiol methylation strictly depends on both SAM and the artificial low-potential electron donor Ti(III)-citrate. The latter most likely keeps the EMT cofactor in the active cob(I)alamin state. NAD(P)H together with ATP, the assumed cosubstrates of a putative estradiol-induced RACE component, could not substitute for Ti(III)-citrate. Thus, the natural electron donor system for regenerating the active cob(I)alamin state needs to be studied in future work. The cofactor dependence identified in this work is in contrast to qualitative results reported in Wang et al. (15), where estradiol methylation essentially depended on ATP (22). This observed ATP dependence may be explained by L-methionine adenosyltransferasecatalyzed SAM synthesis from traces of SAM precursors present in the extracts used. The availability of higher biomass in the study presented here allowed for the conversion of much higher substrate amounts, which could be required for quantitative in vitro studies with protein fractions essentially free of interfering metabolites and side reactions.
It is obvious that standard cellular C1 metabolism, usually involved in biosynthetic and regulatory functions, can hardly cope with the stoichiometric provision of a C1 unit  for estradiol catabolism (one methyl group per estradiol). To solve this problem, second copies of the genes involved in C1 metabolism from L-serine to SAM are present in the genomes of estrogen-degrading anaerobic bacteria. These genes are strongly upregulated during growth with estradiol to guarantee continuous stoichiometric SAM regeneration from catabolites formed during estradiol degradation via L-serine and L-methionine. Thus, oxygen-independent growth with estradiol does not only depend on a specific methyltransferase system and a B 12 -processing machinery, but also on an efficient channeling of C1 units to the initial methyl transfer reaction. In this light, it is important to note that the reported B 12 dependence of estradiol-degrading anaerobes (15) has to be linked to two highly upregulated cobalamin-dependent enzymes: estradiol methyltransferase and L-methionine synthase involved in SAM regeneration. The optimized channeling of the cellular C1 pool toward SAM regeneration represents an attractive model for in vivo solutions for SAM-dependent methyl transfer reactions of biotechnological interest, e.g., in the synthesis of pharmaceuticals. There are currently considerable efforts to develop SAM regeneration systems for biocatalytic alkylation reactions (36,37).
This work elucidated the cofactor requirement and regeneration of an unprecedented catabolic SAM-and cobalamin-dependent methyl transfer system involved in the microbial degradation of a globally important pollutant. The work also illustrates how the establishment of a catabolic pathway not only depends on the evolution/ recruitment of proteins involved in uptake and catabolic and regulatory processes, but can also entail the redirection of existing anabolic reactions toward catabolic pathways for regeneration of a crucial cofactor.
E. coli BL21(DE3) was cultivated in 2ϫ YT medium (18 g liter Ϫ1 tryptone, 10 g liter Ϫ1 yeast extract, and 5 g liter Ϫ1 NaCl) supplemented with 100 g ml Ϫ1 ampicillin. Induction of recombinant proteins was carried out in the early exponential phase of optical density at 578 nm (OD 578 ) 0.4 to 0.8 with 200 ng ml Ϫ1 anhydrotetracycline for 20 h at 20°C; cells were harvested in the late exponential phase.
Enzyme assays. The conversion of estradiol/estrone to DHT/androsta-1,4-diene-3-one (ADD) was determined anaerobically in a glove box (95% N 2 , 5% H 2 , by vol) at 30°C in the dark as follows: All enzyme assays were stopped by adding two volumes of 2-propanol (for analysis of steroid compounds) or 0.5% formic acid (analysis of SAH). After 2-fold centrifugation at 10,000 ϫ g, the supernatant was applied to an Acquity H-class UPLC system (Waters) for analysis.
Liquid chromatography-electrospray ionization-mass spectrometry analyses of steroid compounds. Steroid metabolites were analyzed by an Acquity I-class UPLC system (Waters) using an Acquity UPLC CSH C18 (1.7 m, 2.1 ϫ 100 mm) column coupled to a Synapt G2-Si HRMS ESI-Q-TOF device (Waters). The acetonitrile gradient was as described above. Samples were measured in positive mode with a capillary voltage of 3 kV, 100°C source temperature, 450°C desolvation temperature, 750 liter min Ϫ1 N 2 desolvation gas flow, and 30 liter min Ϫ1 N 2 cone gas flow. LC-MS metabolites were evaluated using MassLynx (Waters). Metabolites were identified by their retention times, UV-vis spectra, by comparison with authentic standards, and m/z values.
MS data were analyzed by the Proteome Discoverer software (v.2.3, Thermo Fisher Scientific, Waltham, MA, USA) using SEQUEST HT as search engine against the protein-coding sequences of D. oestradiolicum. Search settings were set to trypsin (Full); maximum missed cleavage: 2; precursor mass tolerance: 10 ppm; fragment mass tolerance: 0.02 Da; two missed cleavage sites; dynamic modifications oxidation (Met); static modifications carbamidomethylation (Cys). Only peptides that passed the FDR thresholds set in the Percolator node of Ͻ1% FDR q value, and were rank 1 peptides, were considered for protein identification.
Genome sequencing and assembly. Genomic DNA of D. oestradiolicum was extracted using the GenElute bacterial genomic DNA kit (Sigma-Aldrich), and sequencing was carried out by combining short read (Illumina) and long read (Oxford Nanopore) technologies. For nanopore sequencing, library preparation was performed starting with 1 g of DNA, following the 1D Native barcoding genomic DNA (with EXP-NBD103 and SQK-LSK109) protocol (Oxford Nanopore). The library was sequenced using a nanopore R9.4.1 flow cell (Oxford Nanopore) and the MinION device, with the MinKNOW v.1.14.1 and Albacore v.2.3.1 softwares. For Illumina sequencing, 250 ng of DNA was sonicated to a 100 to 1,000 bp size range using the E220 Covaris Focused-ultrasonicator instrument (Covaris, Inc.). The fragments were endrepaired, 3=-adenylated, and NEXTflex HT Barcodes (Bioo Scientific Corporation) were added using the NEBNext DNA modules products (New England BioLabs). After two consecutive clean-ups with 1ϫAMPure XP, the ligated product was amplified by 12 PCR cycles using the Kapa Hifi Hotstart NGS library amplification kit (Kapa Biosystems), followed by purification with 0.6ϫAMPure XP. After libraryprofile analysis conducted with an Agilent 2100 Bioanalyzer (Agilent Technologies) and qPCR quantification (MxPro, Agilent Technologies), the library was sequenced on an Illumina HiSeq 2500 platform (2 ϫ 250 bp; Illumina, Inc.). Reads were trimmed, and low-quality nucleotides (Q Ͻ 20), sequencing adaptors, and primer sequences were discarded, as well as reads shorter than 30 nucleotides after trimming.
The long reads generated with the nanopore technology were first corrected using the canu software error correction module (39) and half of the Illumina reads. The corrected nanopore reads were then assembled using the smart de novo engine (Jue Ruan, unpublished, https://github.com/ruanjue/ smartdenovo, yielding a single 4.17-Mbp unitig corresponding to the main chromosome, along with the sequence of a 124-Kbp plasmid. Both sequences were subjected to three rounds of error correction (polishing) using the pilon software (40). Both the main chromosome and the plasmid sequences were uploaded to the Microscope platform (41) for annotation, comparative genomic analyses, and human interaction.
Data availability. Genome sequences are available from the ENA/NCBI/DDBJ databases under accession numbers LR778301 and LR778302, respectively (study identifier PRJEB37013).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.