Bacterial SBP56 identified as a Cu-dependent methanethiol oxidase widely distributed in the biosphere

Oxidation of methanethiol (MT) is a significant step in the sulfur cycle. MT is an intermediate of metabolism of globally significant organosulfur compounds including dimethylsulfoniopropionate (DMSP) and dimethylsulfide (DMS), which have key roles in marine carbon and sulfur cycling. In aerobic bacteria, MT is degraded by a MT oxidase (MTO). The enzymatic and genetic basis of MT oxidation have remained poorly characterized. Here, we identify for the first time the MTO enzyme and its encoding gene (mtoX) in the DMS-degrading bacterium Hyphomicrobium sp. VS. We show that MTO is a homotetrameric metalloenzyme that requires Cu for enzyme activity. MTO is predicted to be a soluble periplasmic enzyme and a member of a distinct clade of the Selenium-binding protein (SBP56) family for which no function has been reported. Genes orthologous to mtoX exist in many bacteria able to degrade DMS, other one-carbon compounds or DMSP, notably in the marine model organism Ruegeria pomeroyi DSS-3, a member of the Rhodobacteraceae family that is abundant in marine environments. Marker exchange mutagenesis of mtoX disrupted the ability of R. pomeroyi to metabolize MT confirming its function in this DMSP-degrading bacterium. In R. pomeroyi, transcription of mtoX was enhanced by DMSP, methylmercaptopropionate and MT. Rates of MT degradation increased after pre-incubation of the wild-type strain with MT. The detection of mtoX orthologs in diverse bacteria, environmental samples and its abundance in a range of metagenomic data sets point to this enzyme being widely distributed in the environment and having a key role in global sulfur cycling.


Protein purification
An initial protocol for purification of methanethiol oxidase from Hyphomicrobium sp. VS was developed as shown below in Supplementary Table  S1. This resulted in a protein preparation that revealed a single dominant polypeptide in SDS-PAGE analysis with a molecular weight of approximately 46 kDa (Kertens-Mennink, Pol, Op den Camp, unpublished results). The activity of methanethiol oxidase in this preparation was enriched 36-fold in this fraction. The above purification procedure (Supplementary Table S1) resulted in an active enzyme preparation showing a single dominant polypeptide on SDS-PAGE of a molecular weight of ~45-46 kDa (result not shown). The final purification protocol adopted for MTO was as described in Materials and Methods of the main text, consisted of sequential steps of anion exchange chromatography on a MonoQ 10/100 column followed by size exclusion chromatography using a Superdex-75 column. This also resulted in a protein preparation that was dominated by a 46 kDa polypeptide. The fraction with MTO activity eluting from the anion-chromatography run again showed a single dominant polypeptide of 46 kDa, which was up to 50-fold enriched in specific activity in neighbouring fractions compared to the cleared supernatant (Supplementary Table S2). This demonstrated that the enzyme activity responsible for MT degradation in Hyphomicrobium sp. VS was successfully enriched. Table S2. Modified purification scheme, comparison of specific enzyme activity in active fraction recovered after MonoQ anion exchange chromatography with initial soluble enzyme extract after removal of cell membranes

Electron Paramagnetic Resonance Spectroscopy -Results
In resting and oxidized samples we observed EPR signals at temperatures of 7 and 13 K (Supplementary Fig S4). The resting oxidized MTO EPR signals did not have well-resolved Cu(II) EPR signals from an isolated mono-nuclear Cu(II) site(s), suggesting that the Cu-ions are coupled/interacting between Cu ions and/or with a radical in some redox state ( Supplementary Fig S4). We could observe trace amounts of a radical with a very narrow signal centered around 2.0 (as it is a first derivative of a peak it has both positive and negative features). These broad signals were probably from two magnetically-interacting Cu(II) centres, with positive signals before 2.05, somewhat similar to CuA in cytochrome c oxidase or nitrous-oxide reductase, both binuclear copper centres, as well as Cu model complex possibly also without bridging sulfur (Antholine et al 1992, Monzani et al 1998, Solomon et al 1996, Kaim et al, 2013. These could be species from a Cu(II)Cu(I)/ Cu(1.5)Cu(1.5) cluster. The oxidized samples also had a signal around g=4.8 (see Supplementary Fig S5 at 7 and 13 K) indicating spin-coupled or spin-interacting species at this high g-value. The EPR spectra (Supplementary Figures S4 and S5) have several features of a Cu protein that cannot come from a single Cu site as the positive signals at g = 2.46 lack other resolved hyperfine interactions from a single isolated Cu (it is also a very weak Cu(II) at higher g-values visible with 4 hyperfine couplings resolved). The g= 2.39 and 2.24 could be linked to the 2.46 signal but all these three features did not look like normal hyperfines but could possibly indicate that sulfur may be a ligand. The g=2.053 is a zero crossing signal (like the radical) with both positive and negative signal and might be linked to any of the 2.46, 2.39 or 2.24 signals, but at this point of time it is unclear which. The last negative feature with g= 1.905 has unusually low g-value that might be seen in spin-interacting signals (Andersson et al 2003).
In summary, the changes in features in the EPR spectra indicate changes in coordination of Cu when substrate binds, which could indicate direct interaction of the substrate with the Cu centre. Although at this point the exact nature of the Cu environment and status cannot be fully resolved, it is likely to be a binuclear site, as the data do not support a single atom Cu centre.

X-ray Spectroscopy Analysis of Methanethiol Oxidase
Detailed Methods X-ray absorption spectra were obtained in fluorescent mode on station B18 of the Diamond Light Source (Didcot, UK). This uses the technique of quick EXAFS (QuEXAFS), where the monochromator rotates at a constant rate during data acquisition. The fluorescence was detected using a nine element germanium solid state detector. Data were obtained at the Cu K edge for a variety of samples and standards. All data were obtained with the samples at 77K in a cryostat. To minimize radiation damage the beam was rastered across the sample, which was moved between each scan. Each scan took about 20 minutes to acquire.
The fluorescence signals were merged and normalized using the Athena program. The output from this program is used in studies of the absorption edge shape and the chemical shifts. The background was then subtracted using the Pyspline program to produce the oscillatory EXAFS spectrum. This was analysed using the EXCURV program. This employs the rapid curved wave theory, including multiple scattering when necessary (Gurman et al. 1986). Scattering properties of the atoms were calculated ab initio within the program. The structure surrounding the Cu atom is described in terms of shells of atoms: a shell is a set of atoms of the same chemical type at the same average distance. The structural parameters fitted are the number and type of atom within a shell, their average distance from the Cu atom and the mean square deviation in this distance (the Debye-Waller factor).
Copper metal (foil), CuO and CuS were used as reference samples. The coppercontaining enzyme tyrosinase (Sigma Aldrich, Gillingham, UK) was used as additional reference. Five samples of purified MTO were analysed: as-isolated enzyme (MTO1 and MTO2); enzyme treated with the oxidising agent sodium hexachloroiridate (2 mM) (MTO3); enzyme treated with the substrate methanethiol (MTO4); enzyme treated with the reducing agent sodium dithionite (1 mM) (MTO5).
In fitting the EXAFS from the tyrosinase standard sample it became apparent that the spectrum was contaminated by a copper metal signal. This probably arose from fluorescence from the cryostat window mounts which are hit by scattered primary X-rays, and will also occur in all other samples. Fitting the EXAFS spectrum enabled us to determine the proportion of copper metal signal in the data. This was then subtracted from the fluorescence data to obtain corrected spectra for use in the analysis. The nearest neighbour (Cu-N) structural data obtained from the EXAFS fits were the same in the original and corrected data. However, the presence of this contaminating signal means that it is not possible to determine whether there is any Cu-S coordination in any of the samples (Cu-Cu at 2.51 Å, Cu-S at about 2.65 Å). Edge spectra corrected in this way were used for the edge analysis.
In studies of the edge structure the main interest is in the energy of the edge, which gives the chemical shift. The position of the edge is usually defined by the half way point (E50). This could not be used here this because of the structure on the edge, which was interpreted as arising from transitions to the unoccupied Cu 4p level. Instead, the 70% absorption point (E70) was used as a proxy for the position of the ionization threshold and the 30% absorption point (E30) as a proxy for the energy of the 3p level. Increasing ionization of the copper atom will increase the energy of the ionization threshold and decrease the energy of the 4p level relative to this because of changes in electron screening.

EXAFS Results
EXAFS analysis ( Figure S6) of the as-isolated MTO (samples MTO 1 and 2) and the hexachloroiridate-oxidised samples gave very similar data. These data show Cu coordinated by four light atoms (which are presumed to be nitrogen based upon the availability of histidines as ligands) with a 1.99 Å Cu-N distance. This bond distance is comparable to a three coordinate Cu-N bond distance of 1.94 Å obtained from a sample of the model copper enzyme tyrosinase that was analysed via EXAFS in parallel with the MTO samples, and a Cu-N distance of 1.97-2.05 Å in the X-ray crystal structure of peptidylglycine α-hydroxylating monooxygenase (Chauhan et al., 2014; PDB accession number 1YI9) and in the complex between copper and the peptide Gly-Leu-Tyr moiety with a 4 N square pyramid and a Cu-N distance of 1.92 Å (van der Helm and Franks, 1968). The hexachloroiridate treated sample (MTO 3) showed the presence of three nitrogen ligands at 1.98 Å and has at least 1-2 light atom ligands (N or O).
The sample treated with the substrate (methanethiol) (MTO 4) showed 2-3 ligands at a Cu-N bond distance of 2.04 Å. The dithionite-reduced sample (MTO 5) gave structural data that were somewhat unclear, with about four nitrogen ligands at a Cu-N bond distance of about 2.05 Å.
The position and shapes of the Cu-K absorption edge suggest that the oxidation state of the Cu in the as-isolated (MTO 1 and 2) and hexachloroiridate-treated (MTO 3) samples is between 1 and 2. The samples treated with substrate (MTO 4) and sodium dithionite (MTO 5) are somewhat more reduced than the other samples. This conclusion is in line with the increased Cu-N bond length in the last two samples (and illustrated by the correlation shown in Figure S7). The results are also consistent with a redox enzyme in which copper undergoes changes in coordination and oxidation state during turnover. Whilst the corrected data do not give any information about whether the substrate sulphur atom becomes ligated to the copper, they do indicate a reduction in the number of light-atom (presumably nitrogen) ligands and a decrease in the oxidation state of copper upon binding of the substrate.  Figure S6. Normalised X-ray spectra of the MTO and tyrosinase samples, corrected to remove the signal due to copper metal contamination as detailed in the text.