Bioconversion of 4-hydroxyestradiol by extradiol ring-cleavage dioxygenases from Novosphingobium sp. PP1Y

Four ERCDs from Novosphingobium sp. PP1Y, production, specific activity and kinetic parameters determination on catecholic substrates

The majority of known ERCDs belong to a heterogeneous protein family characterized by the presence of several subfamilies³². Each family has evolved to optimize the cleavage of a specific type of substituted catechol: 3- or 4-alkylcatechols, 3,4-dihydroxyphenylacetate, 2,3-dihydroxybiphenyl, 1,2-dihydroxynaphthalene, 3,4-dihydroxyphenanthrene.

Genome analysis of PP1Y strain³⁰ revealed the presence of 7 orfs coding for putative ERCDs (three present in double copy with a 100% identity). A thorough phylogenetic study, supported by a homology modelling analysis, allowed to hypothesize the role of the seven enzymes in the metabolism of mono and polycyclic aromatic compounds³⁰. In particular, it was hypothesized that orfs AT15671/AT31616, Mpl3065, AT15599/AT31688 and AT32663 code for, respectively, a putative (di)(methyl)catechol dioxygenase, a putative dihydroxybiphenyl dioxygenase and two dihydroxynaphtalene/dihydroxyphenanthrene dioxygenases. Herein, these enzymes will be named, respectively, PP00193, PP26077, PP00124 and PP28735.

The homology modelling/docking analysis described in D’Argenio et al.³⁰ suggested that the active sites of the mentioned ERCDs should be able to host catechols with substituents at the positions 3 and/or 4, but the dimensions of the active site pocket progressively increase in the order PP00193 < PP26077 < PP00124 < PP28735. Very intriguingly, PP00124 and PP28735 showed active site pockets large enough to host dihydroxylated polycyclic aromatic hydrocarbons with 3–4 rings (e.g. phenanthrene, anthracene, benz[a]anthracene and chrysene) and 4-OHE2 that, from a steric point of view, is similar to 1,2-dihydroxychrysene (see Supplementary Figure S1). The determination of the specific activity of the four ERCDs on a limited panel of substrates, including 4-OHE2, confirmed the hypothesized roles of these proteins in the metabolism of Polycyclic Aromatic Hydrocarbon (PAH) in strain PP1Y³⁰.

Therefore, the 4 ERCDs were selected as interesting activities to screen for estradiol bioconversion. Recombinant expression conditions were optimized in E. coli BL21(DE3) cells transformed with the corresponding pET22b( +) vectors, whose construction was previously described³⁰. Expression conditions were optimized by testing different growth temperatures, induction times and IPTG concentrations, to obtain active enzymes. To select growing conditions, we evaluated the catalytic activity of ERCDs on 2,3-dihydroxybiphenyl (2,3-DHBP) as substrate in E. coli BL21(DE3) recombinant whole cells (Supplementary Figure S2 a), over the induction phase by IPTG. As showed in Supplementary Figure S2 b, all recombinant ERCDs were expressed in the soluble fraction, except for protein PP28735, which was mainly present in the insoluble fraction, as highlighted by SDS-PAGE analysis of the soluble and insoluble fractions after cell lysis (Supplementary Figure S2 b). The expression level of proteins in soluble form (Supplementary Table S1) was estimated to be 120, 40 and 100 mg/L of culture for PP26077, PP00124 and PP00193, respectively, but only 1–2 mg/L for enzyme PP28735.

ERCDs were then purified by an anionic exchange chromatography on a Q-Sepharose FF resin. A marked instability of all ERCDs in buffers without Fe(NH₄)₂(SO₄)₂ was observed, suggesting that, in the absence of exogenous iron in the buffer, the Fe(II) in the active site easily was lost or oxidized to Fe(III) thus leading to enzyme inactivation. A protocol for a quick “batch” chromatography was therefore set up, which involved the use of a buffer containing 30% glycerol and 0.1 mM Fe(NH₄)₂(SO₄)₂. As shown in Supplementary Figure S3 a, the yield of the purification procedures ranged from 44 to 95%, based on the evaluation of active enzyme purified (final total units) compared to the amount of active enzyme in cellular lysates (initial total units). SDS-PAGE analysis of purified proteins (Supplementary Figure S3 b) revealed the presence of contaminants up to 10–20% of total proteins for ERCDs PP26077, PP00124 and PP00193, whereas purified PP28735 showed a higher presence of contaminant proteins, most likely related to the low initial amount of protein in the soluble fractions of the induced cultures.

As all Ring Cleavage Dioxygenases are endowed with a Fe(II) ion as cofactor at the active site, and the lack of incorporation of Fe(II) into the mature protein, or its oxidation to Fe(III), causes total inactivation of the dioxygenase activity, we measured the iron content by Ferene S assay. Results reported in Supplementary Table S1 highlighted that total Fe(II) was found to be 60%, 100% and 88% for PP26077, PP00124 and PP00193 proteins, respectively. Recovery yield of protein PP28735 in the purified fractions was too low to measure the iron amount. Furthermore, iron assay allowed to estimate that 25% and 7% of PP26077 and PP00193 proteins, respectively, were iron-free, whereas the remaining fractions were endowed with Fe(III), leading to inactive enzymes. It is worth noting that purified PP00124 protein retained 100% Fe(II) at the end of expression and purification procedure, thus indicating a higher enzymatic stability compared to the other ERCDs.

Then, specific activity (S.A.) and kinetic parameters of purified ERCDs on the four catecholic substrates selected, 2,3-DHBP, 3-methylcatechol (3MC), 4-methylcatechol (4MC) and catechol (CAT) were determined (“Materials and Methods”). Specific activity of the purified RCDs was measured by monitoring the formation over time of the corresponding cis-muconic semialdehydes. Results shown in Table 1 suggested that proteins PP28735, PP26077 and PP00124 were mainly active on 2,3-DHBP and showed lower values of S.A. on the other monoaromatic substrates tested. Conversely, PP00193 was more active on CAT, 3-MC and 4-MC, as expected based on the modelling results³⁰, which indicated that this enzyme has a smaller active site.

Kinetic features of the 4 ERCDs, summarized in Table 2, showed that proteins PP00124 and PP00193 were endowed with the higher activity towards all substrates, with K_M values ranging from 30 to 1 µM. As expected, PP00193 seemed the best enzyme for the conversion of monoaromatic compounds and 2,3-DHBP, showing the higher k_cat/K_M values for these substrates. Conversely, proteins PP28735 and PP26077 had a significant activity only on 2,3-DHBP, the larger substrate, while displaying K_M values higher than 450 µM on monoaromatic catechols. Indeed, these enzymes displayed k_cat/K_M values for 2,3-DHBP conversion between 80 and 200 times lower compared to 3-MC. Protein PP00124 showed a comparable activity on all substrates tested, with a higher efficiency towards 2,3-DHBP.

Bioconversion of catechol estrogens

Bioconversion of catechol estrogens using strain PP1Y ERCDs was tested. Substrates used for the bioconversions were 4-OHE2 and 2-OHE2 (Fig. 1a). These are derived from E2 hydroxylation, bearing the –OH substituents at positions 3, 4 and 2, 3 of the aromatic ring, respectively. Starting from the molecular docking analysis previously performed³⁰, 4-OHE2 and 2-OHE2 should be better accommodated in PP28735 and PP00124 active sites. Therefore, these enzymes should display the higher affinity for the selected catechol estrogens. Preliminary data obtained in our previous work³⁰ further supported this hypothesis. In this work, the S.A. of the four ERCDs was determined for 4-OHE (Table 5 in³⁰). Data showed that PP28735 and PP00124 proteins were able to catalyse 4-OHE2 cleavage into its putative cis-muconic semialdehyde, whereas PP00193 and PP26077 proteins showed negligible activity.

To select the best enzyme for 4-OHE2 conversion, we performed time-course experiments (Fig. 2) carried out with PP28735 and PP00124 proteins by spectrophotometer analyses. Reactions, performed at pH 7.5 in the presence of 100 µM 4-OHE2, led to the formation of a reaction product endowed with λ_max at 298 nm (Fig. 2). Within 4 (PP00124) and 8 (PP28735) min incubation time, no further changes in spectra were observed, suggesting a complete conversion of 4-OHE2. Based on the spectral properties of the cleavage products, it could be hypothesized that the two enzymes catalyse the same reaction. However, it is worth noting that the semialdehydes obtained from meta cleavage of catechols appear, in general, as yellow products with λ_max between about 350 and 450 nm, whereas cleavage product from 4-OHE2 absorbed in spectra UV region with λ_max at 298 nm. It is well known that the yellow-coloured forms represent the dianionic form of semialdehydes, generally the most abundant at pH 7.5 at which the reactions were performed³³. To verify the properties of the 4-OHE2 conversion products by PP28735 and PP00124 proteins, we alkalinized the reaction mixture by NaOH adding. The spectral properties in alkaline buffer reported in Supplementary Figure S4, highlighted products of yellow colour (λ_max at 417 nm) as expected for semialdehydes, thus confirming the ring cleavage reaction, and suggesting that a higher pH value is needed to obtain the dianionic yellow form.

UV–vis spectra of the time-course of 4-OHE2 enzymatic conversion. Reactions were carried out using PP28735 (a) and PP00124 (b) enzymes in 1 mL of 50 mM Tris/HCl pH 7.5 buffer containing 100 μM 4-OHE2 (dashed black lines). The reactions were started by addition of purified enzymes. Semialdehyde production was monitored by the Scanning Kinetics program on Cary 100 UV–VIS spectrophotometer in a wavelength range from 230 to 500 nm, recording the absorption for 4–8 min, at 25 °C (gray scale lines). Bold black lines represent the spectra of semialdehyde at the end of reaction at pH 7.5 (λmax 298 nm).

We also performed kinetic analyses of 4-OHE2 conversion by the ERCDs to select the best catalyst. Product formation was monitored spectrophotometrically at 298 nm, wavelength at which no absorbance was recorded for 4-OHE2 substrate. As reported in Table 3, protein PP00124 showed the best kinetic parameters for the bioconversion of 4-OHE2 with sub-micromolar KM value, which prompted the set-up of a small-scale bioconversion protocol using cell lysates of E. coli expressing PP00124 as catalyst.

We also tested the ability of ERCDs to cleave 2-OHE2, but no conversion product was observed for any of the proteins, thus suggesting a selectivity of these enzymes toward 4-OHE2.

Based on the interesting kinetic properties of PP00124 protein, we further studied the kinetic of 4-OHE2 cleavage by PP00124 containing cell extracts performing HPLC analyses.

The complete conversion of 4-OHE2 was verified spectrophotometrically. Bioconversion products obtained from the reaction were then analysed by HPLC. Results shown in Fig. 3 revealed that under our experimental conditions, the complete conversion of 4-OHE2 (retention time: 12.7 min and λ_max: 279 nm, panel 3a) into a semialdehyde product (retention time: 11.4 min and λ_max: 305 nm, panel 3b) was obtained, thus confirming the spectrophotometric analyses. On the contrary, when using 2-OHE2 as substrate (retention time: 13.4 min and λ_max: 286 nm) (Fig. 3, panels c,d) no conversion was observed at any time of the reaction, in accordance again with spectrophotometric analyses. These data confirmed the tight specificity of PP00124 towards substrates bearing substituents in positions 3, 4 of the aromatic A rings.

HPLC analysis of 4-OHE2 and 2-OHE2 bioconversion using recombinant PP00124 as catalyst. The assay was carried out in 50 mM Tris/HCl buffer pH 7.5 at 25 °C, with 100 µM and 200 µM of 4-OHE2 and 2-OHE2, respectively. Aliquots of the reaction mixtures were collected before the addition of the enzyme. Samples were acidified, diluted 100-fold, centrifuged, and analyzed in HPLC as negative controls (shown in panels (a) and (c) as blank reactions). Then, PP00124 cell lysate was added to start reactions. Product formation was followed spectrophotometrically over 15 min. Samples were then acidified, diluted 100-fold, centrifuged, and analysed by HPLC panels (b) and (d). (a) HPLC chromatogram of 4-OHE2 blank reaction and UV–vis spectrum of 4-OHE. (b) HPLC chromatogram of 4-OHE2 reaction: semialdehyde reaction product and corresponding UV–vis spectrum. (c) HPLC chromatogram of 2-OHE2 blank reaction and UV–vis spectrum. (d) HPLC chromatogram of 2-OHE2 reaction and corresponding UV–vis spectrum. All shown chromatograms were acquired at 280 nm.

High resolution mass spectrometry analysis of conversion products was performed. Aliquots of the blank and end-point reactions were collected and subjected to a liquid–liquid extraction with ethyl acetate. The obtained extracts were analyzed by LC–MS/MS in negative ion mode. Two major reaction products generated by the PP00124-catalyzed bioconversion of 4-OHE2 were detected: two co-eluent species, with molecular weights of 306.1839 (measured m/z 305.1761, Fig. 4b) and 324.1951 (measured m/z 323.1873 , Fig. 4c), respectively.

High resolution MS spectra of 4-OHE2 bioconversion products. Full mass (MS1) spectra were acquired in negative ion mode. (a) 4-OHE2 m/z spectrum (m/z 287.1652). (b) 4-OHE2 meta cleavage product m/z spectrum (m/z 323.1873). (c) 4-OHE2 cyclization product (hypothetical) m/z spectrum. (m/z 305.1761).

These data confirmed the presence of the 4-OHE2 meta cleaved semialdehyde product (theoretical M.W. 324.1937), as expected based on the catalytic mechanism of ERCDs mediating the insertion of an O₂ molecule in the substrate³⁴. This is indicated in Fig. 4 as meta cleavage product. The second reaction product could be generated from the meta cleavage product, through a spontaneous nucleophilic attack of the hydroxyl on C4 onto the carbonylic group on C5, followed by dehydration (theoretical M.W. 306.1831). This is indicated in Fig. 4c as cyclization product (hypothetical). The hypothetic structures of these compounds and the proposed reaction are reported in Fig. 5.

4-OHE2 hypothesized degradation mechanism using PP00124. The presence of the 4-OHE2 meta cleaved semialdehyde product (theoretical M.W. 324.1951), was verified by mass spectrometry. It is in line with ERCDs catalytic mechanism acting through an O₂ insertion in the substrate. We hypothesize the generation of the second product identified, through a cyclization of the meta cleavage product, with the spontaneous nucleophilic attack of the hydroxyl on C4 onto the carbonylic group on C5. The hypothetic structure of this compound (theoretical M.W. 306.1839) is reported as 4-OHE2 cyclization product (hypothetical) .

Bioconversion of 4-OHE2 using whole cells

Once the complete conversion of 100 μM 4-OHE2 using PP00124 was confirmed, we evaluated the possibility to obtain the 4-OHE2 bioconversion with whole cells as biocatalysts. To this purpose, E. coli recombinant cells expressing PP00124 were used.

After recombinant expression, cells were assayed to test recombinant protein activity on 2,3-DHBP. A specific activity of 0.7 U_2,3DHBP/OD was measured. 3.5 total Units_2,3-DHBP (5 OD₆₀₀) of PP00124 expressing E. coli cells were incubated in minimal medium containing 100 μM (28 mg/L) 4-OHE2 and a time course experiment was carried out. The experiment was also performed with 10 OD₆₀₀ cells of E. coli BL21(DE3) transformed with empty pET22b(+) plasmid as a negative control. Each time point of PP00124 cells and E. coli negative control reactions was extracted twice in ethyl acetate and analyzed by HPLC–MS/MS. No 4-OHE2 degradation was observed using E. coli cells transformed with the empty plasmid. Results obtained for PP00124 expressing cells are shown in Fig. 6. As observed in the reaction carried out using the purified enzyme, a fast decrease of 4-OHE2 (m/z 287) and formation over time of the two compounds generated by 4-OHE2 hydrolysis was evident (m/z 305 and 323). Interestingly, the kinetics accumulation of these two compounds were almost superimposable, thus suggesting that an equilibrium between these two species occurred.

E. coli whole cells bioconversion of 4-OHE2 using recombinant PP00124 as catalyst. E. coli recombinant cells expressing P00124 were incubated at 37 °C with 100 μM 4-OHE2 in minimal medium. At different timepoints, aliquots of the supernatant were collected, and acidified. A liquid–liquid extraction with ethyl acetate was performed. HPLC–MS/MS quali-quantitative analyses was carried out. Full mass (MS1) spectra were acquired in high resolution negative ion mode. Peak areas of the extracted ion chromatograms for each compound and time point, were registered. Trend overtime of 4-OHE2 (m/z 287) and its cleavage products (m/z 305 and 323) are shown. (a) 2-h time-course bioconversion. (b) detail of 20 min bioconversion. Data are reported in Area (%) expressing relative abundance of the compounds. For each compound, the higher peak area registered was set as 100% area. Samples were analysed in triplicate and data were reported as the mean of the measured areas.

The bioconversion rate performed using PP00124 expressing cells was significantly slower than the one carried out with the purified enzyme (outlined above). However, PP00124 k_cat/K_M allowed 4-OHE2 complete conversion in 15 min using 3.5 Units_2,3-DHBP in 5 OD₆₀₀. Noteworthy, to the best of our knowledge, no examples of 4-OHE2 whole cells biodegradation have been reported so far. Li and co-workers showed a complete degradation of E2 using a microbial consortium, after 3 days of incubation at 20 mg/L³⁵.