Filter-based cultivation with air as the sole carbon and energy source
The strains were pre-incubated in liquid cultures in 100 mL serum bottles containing 10 mL of 10x diluted, EDTA-free NMS medium (DSMZ medium 921 with 10x the iron concentration and 1 µM lanthanum) and a headspace of 20% CH4 in air. The Fe-EDTA stated in the original DSMZ recipe was substituted with FeSO4. Depending on the strain, the medium was adjusted to a pH of 6.8 (Methylocapsa gorgona MG08, Methylosinus trichosporium OB3b, Methylocystis rosea SV97, Methylobacter tundripaludum SV96) or 5.8 (Methylocapsa aurea KYG, Methylocapsa acidiphila B2, and Methylocapsa palsarum NE2). The serum bottles were sealed using butyl rubber stoppers (Chromacol 20-B3P, Thermo Fisher Scientific, Waltham, Massachusetts, USA) and crimp caps, and incubated at 20°C in dark, until the strains were in an exponential growth phase. Strain purity was controlled routinely via microscopy and by confirming the absence of heterotrophic growth on agar plates with a rich medium containing tryptone, yeast extract, and glucose30. After reaching exponential growth phase, the strains were diluted with the described medium and filtered on 25 mm polycarbonate (PC) filters (Whatman 10417006, Cytiva, Massachusetts, Marlborough, USA) using a filtration manifold (EQU-FM-10X20-SET, DHI, Hørsholm, Denmark)28. The final cell density on the filters amounted approximately 20 cells per photo area (630x magnification). The filters were transferred into Petri dishes containing 10x diluted, EDTA-free NMS medium (DSMZ medium 921 with 10x the iron concentration and 1 µM lanthanum) and left floating on the medium with the cells facing upwards. The cultures were incubated at 20°C in dark, with air as the sole energy and carbon source, for at least three months until initiation of experiments. All cultivation steps were carried out under sterile conditions. Before experiments, colony formation and microcolony morphology of the strains was routinely checked via microscopy as previously described31.
Trace gas oxidation experiments
For all trace gas oxidation experiments, filter cultures of the respective strains, pre-incubated with air as the sole carbon and energy source, were transferred into 250 mL glass bottles containing 50 mL 1:10 EDTA-free NMS medium (DSMZ medium 921 with 10x the iron concentration and 1 µM lanthanum). The bottles were capped with Safety-Caps (JR-S-11011, VICI AG International, Schenkon, Switzerland). To create a defined atmosphere within the bottles, the headspace of each bottle was flushed for 10 minutes with high-purity synthetic air containing 400 p.p.m.v. CO2 (HiQ, AGA, Sweden) using a gassing manifold. The headspace pressure was adjusted to approximately 1.05 bar. Then, 1 mL 1000 p.p.m.v. CH4 in N2 (HiQ, AGA, Sweden), 1 mL 1000 p.p.m.v. H2 in N2 (HiQ, AGA, Sweden), and 1 mL 1000 p.p.m.v. CO in N2 (HiQ, AGA, Sweden) were added to the headspace using a gas tight syringe. To assess the oxidation rate of the respective strains, the cultures were incubated at 20°C and the change in CH4, H2, and CO concentrations within bottle headspaces were measured. For each measurement, 2 mL of the gas in the headspace were sampled using a gas tight syringe. The gas samples were analyzed with a gas chromatograph (ThermoScientific Trace 1300, Thermo Fisher Scientific) equipped with a sample loop, a Hayesep column (SU12875, RESTEK, Bellefonte, Pennsylvania, USA), a Molsieve 5A column (PKC17080, RESTEK), a pulsed discharge detector, and a flame ionization detector. A high-quality gas containing 2.5 p.p.m.v. CH4, 2.5 p.p.m.v. H2, and 2.5 p.p.m.v. CO in N2 served as standard (HiQ, AGA, Sweden). To determine the gas concentrations in the bottle headspaces, standard curves were created every day of measurement. The headspace pressure was measured at the end of the experiment using a manometer (LEO1, Keller, Winterthur, Switzerland). The mass of each trace gas was calculated by applying the ideal gas law and adjusted for changes in pressure caused by gas removal during measurement.
Cell quantification
For quantification, cells from filters were processed as follows: The filters were placed on 150 µL 25x SYBR green I (10463252, Thermo Fisher Scientific) with cells facing upwards and incubated for 10 minutes in dark. Next, the cells were washed twice by letting the filters float on milliQ water for 5 minutes in dark. To remove the cells from the filters, the filters were transferred into 5 mL tubes (0030122321, Eppendorf, Hamburg, Germany) containing 2 mL 10x diluted solution 1 (DSMZ medium 921), so that the filters sticked to the tube wall and the cells faced inwards63. The tubes were vortexed for 10 minutes at medium speed. After vortexing, remaining cells on the filters were washed off and collected by rinsing the filters with 1 mL solution 1 (DSMZ medium 921). The washed filters were dried in dark, and the cell removal controlled via fluorescence microscopy. The 3 mL cell suspension resulting from the cell removal was spiked with 50 µL absolute counting beads (Invitrogentm CountBrighttm Plus, Thermo Fisher Scientific). After, the cells in suspension were immediately counted using a flow cytometer (BD FACSAria III, BD Biosciences, Franklin Lakes, New Jersey, USA). The flow cytometer was set up to capture SYBR Green I (excitation = 498 nm, emission = 522 nm) in the green channel and the Invitrogentm CountBrighttm Plus Absolute Counting Beads (excitation = 350–810 nm, emission = 385–860 nm) in the blue channel. Unstained cells of M. gorgona MG08, M. palsarum NE2, M. rosea SV97, and M. aurea KYG were used to define the autofluorescence threshold in the different channels. Size beads (NPPS-4K, Spherotech, Lake Forest, Illinois, USA) were used to draw the forward scatter gate covering events between 0.58 and 4 µm. SYBR green I stained cells and absolute counting beads were counted to determine the cell number within filter cultures, using the following equation:\({x}_{filter}=\frac{{b}_{abs}}{{b}_{c}} \times {x}_{c}\)
Where \({x}_{filter}\) is the absolute cell number of the filter culture, \({b}_{abs}\) the number of counting beads added, \({b}_{c}\) the number of counting beads counted, and \({x}_{c}\) the number of SYBR green I stained cells counted.
Cell-specific oxidation rates and energy calculation
The gas-specific oxidation rates of the different strains at atmospheric CH4, H2, and CO concentrations and pseudo first order kinetics were calculated and tested as described by Tveit et al30. The oxidation rate per cell was calculated by dividing the oxidation rate of the filter culture by the corresponding cell number. The Gibbs free energy changes (\({\varDelta }_{r}G\)) for the following reactions: \({CH}_{4}+2{O}_{2}\to 2{H}_{2}O+C{O}_{2}\), \({2H}_{2}+{O}_{2}\to 2{H}_{2}O\), \(2CO+{O}_{2}\to 2C{O}_{2}\), at 20°C, 1.013 bar absolute, and atmospheric concentrations of CH4 (1.87 p.p.m.v.), H2 (0.5 p.p.m.v.), and CO (0.2 p.p.m.v.) amount to -797.4 kJ mol−1, -236.8 kJ mol−1, and − 199.9 kJ mol−1 respectively. The values are based on the values for Gibbs free energy of formation found in literature30,64 and the following equation:
\({\varDelta }_{r}G= {\varDelta }_{r}{G}^{^\circ }+RTln{Q}_{r}\)
where \({\varDelta }_{r}{G}^{^\circ }\) is the Gibbs free energy change at standard conditions, \(R\) the gas constant, \(T\) the temperature, \(ln\) the natural logarithm, and \({Q}_{r}\) the reaction quotient. The estimates of energy yield per cell were obtained from the trace gas oxidation per cell and the \({\varDelta }_{r}G\). The energy yield from trace gas oxidation per mol of biomass and hour (kJ C-mol−1 hour−1) was estimated including the strain specific carbon content and dry weight as previously shown24.
Cellular dry mass and carbon content estimations
The strain-specific cellular dry mass was determined by measuring single-cell buoyant mass distributions of the strains in H2O-based and deuterium oxide (D2O)-based solutions of phosphate saline buffer (PBS) using a suspended microchannel resonator (SMR)65,66. To do so, filter cultures of M. gorgona MG08, M. rosea SV97, M. palsarum NE2, and M. aurea KYG were incubated on 10x diluted, EDTA-free NMS medium (DSMZ medium 921 with 10x the iron concentration and 1 µM lanthanum) under an atmosphere of 1000 p.p.m.v. CH4 in air for three weeks. After, cells of filter cultures were fixed by incubating the filters on 150 µl formaldehyde in 1×PBS (4% w/v) for an hour at room temperature. The cells were washed twice by letting the filters float on water for 5 minutes. The fixed cells were harvested as described in “Cell quantification”. The resulting cell suspension was stored at 4°C until measurements of the buoyant cell mass. For the measurements, two aliquots with the same volume of each cell suspension were created. The water of the aliquots was replaced by H2O-based (1x PBS in H2O) and D2O-based (1x PBS in 9:1 D2O:H2O) solutions of known density (1.0043 g cm− 3 and 1.1033 g cm− 3 at 20°C, respectively). To do so the aliquots were dried in a vacuum concentrator at 37°C. After, the cells of one of the two aliquots were resuspended in 50 µl of the H2O-based solution. The cells of the second aliquot were resuspended in 50 µl of the D2O-based solution. The buoyant mass of the cells in the aliquots were measured with a SMR (LifeScale, Affinity Biosensors, Santa Barbara, California, USA). The precision and accuracy of the SMR was verified by creating calibration curves using NIST-certified polystyrene beads (ThermoFisher Scientific) as performed previously67.
The buoyant mass data were exported from the LifeScale instrument and further analyzed using the Python 368 packages Pandas69, Matplotlib70, and Seaborn68,71 (Supplementary Fig. 8). The dry mass of the strains was calculated as described65 using the median of the single-cell buoyant mass distributions in H2O-based and D2O-based solutions and the following equation:
$${m}_{dry}= \frac{{\rho }_{{D}_{2}O} \times {m}_{{b.H}_{2}O}-{\rho }_{{H}_{2}O} \times {m}_{{b.d}_{2}O} }{{\rho }_{{D}_{2}O}- {\rho }_{{H}_{2}O}}$$
Where \({m}_{dry}\) is the dry mass, \({\rho }_{{D}_{2}O}\) the density of the D2O-based solution, \({m}_{{b.H}_{2}O}\) the cell’s buoyant mass in the H2O-based solution, \({\rho }_{{H}_{2}O}\) the density of the H2O-based solution, and \({m}_{{b.d}_{2}O}\) the cell’s buoyant mass in the D2O-based solution.
The cellular carbon contents of the atmMOB were analyzed using an elemental analyzer coupled to an isotope ratio mass spectrometer (EA-IRMS; EA1110 coupled via a ConFlo III interface to a DeltaPLUS IRMS, Thermo Fisher Scientific). Biomass for the carbon content analysis was derived from stirred-tank bioreactor (DASbox® Mini Bioreactor System, Eppendorf) cultures. The cultures were grown in 10x diluted, EDTA-free NMS medium (DSMZ medium 921 with 10x the iron concentration and 1 µM lanthanum) at 80 rpm (marine impeller), 20°C, pH 6.8 (CO2 controlled) and a gassing rate of 0.24 vessel volumes per minute (6000 p.p.m.v. CH4 in air) using a microsparger (78530205, Eppendorf). After harvest, cultures were washed three times with milliQ water and lyophilized.
Comparative proteomics
For the atmospheric CH4 treatment, filter-cultures of M. gorgona MG08, M. rosea SV97 and M. palsarum NE2 were incubated with air as the only carbon and energy source for seven months. After, the cells on filters were harvested as described in “Cell quantification”, lyophilized, and stored at -80°C until further processing. The 1000 p.p.m.v. CH4 treatment of the respective strains was processed the same way as the atmospheric CH4 treatment with the difference that filter cultures had been exposed to approximately 1000 p.p.m.v. CH4 in air for two weeks before harvest. Samples of lyophilized cells were lysed by sonication in 20 µL buffer containing 4 M urea, 2.5% sodium deoxycholate (SDC) and 100 mM triethylammonium bicarbonate (TEAB). Samples were sonicated for 25 cycles (1 min on, 30 sec off) with maximum amplitude in a cup horn sonicator with a recirculating chiller (Cup horn: model 413C2, Qsonica, Newtwon, Connecticut, USA; Sonicator: Fisherbrandtm FB705, Thermo Fisher Scientific; Recirculating chiller: model 4905 Qsonica). Then, disulfide bridges were reduced with 1,4-dithiothreitol (DTT) at a final concentration of 5 mM and incubation at 54°C for 30 min. Cysteines were alkylated with 15 mM iodoacetamide (IAA) and incubated for 30 min at room temperature in dark. To remove excess IAA, DTT solution corresponding to a final concentration of 5 mM was added. Calcium chloride solution (final concentration of 1 mM) and 1 µg Lysyl Endopeptidase (125–05061, FUJIFILM Wako Chemicals Europe, Neuss, Germany) were added to the samples and incubated for 5 hours under gentle agitation at 37°C for enzymatic digestion. After, samples were diluted with a buffer containing 100 mM triethylammonium bicarbonate (TEAB) and 1 mM CaCl2 to lower the urea and sodium deoxycholate (SDC) concentration to 1 M and 0.65% v/v, respectively, resulting in a final sample volume of 80 µL. For digestion, 2 µg trypsin (V511A, Promega, Wisconsin, USA) were added and the samples incubated on a gently agitated shaker at 37°C for 16 hours. After digestion, SDC was precipitated by adding 50% formic acid to the sample (final concentration of 2.5% v/v). Samples were then incubated for 10 min and centrifuged at 16200 rcf for 15 min. Supernatants containing peptides were transferred to low-protein-binding tubes. The peptides were concentrated and cleaned up using DPX C18 pipette tips (DPX Technologies, XTR tips 10 mg C18AQ 300Å) on a Tecan Fluent pipetting robot (Tecan Group Ltd., Männedorf, Switzerland). Purified peptide samples were dried in a vacuum concentrator and dissolved in 12 µL 0.1% formic acid. Peptide concentrations were measured on a spectrophotometer (Nanodrop ONE, Thermo Fisher Scientific) at 205 nm. 0.25 µg peptides per sample were loaded onto a liquid chromatograph (EASY-nLC1200, Thermo Fisher Scientific) equipped with an EASY-Spray column (C18, 2µm, 100 Å, 50µm, 50 cm). Peptides were fractionated using a 5–80% acetonitrile gradient in 0.1% formic acid over 120 min at a flow rate of 300 nL min− 1. The separated peptides were analyzed using a mass spectrometer (Orbitrap Exploris 480, Thermo Fisher Scientific). Data was collected in data dependent mode using a Top40 method. Annotated genomes of the three strains downloaded from MicroScope72 served as databases for the CHIMERYS-based data search using Proteome Discoverer 3.0. Normalized abundances (scaling mode: On All Average) of proteins were further processed via Perseus73. Normal distribution of Log2 fold transformed data was visually screened using histograms. Proteins were filtered using a threshold of at least three valid values in at least one treatment (n = 4 per treatment). Proteins that passed the filtering were included in downstream analyses. Missing values were imputed from normal distribution using default settings (width = 0.3, down shift = 1.8). The Pearson correlation coefficient ranged from ρ = 0.918 to ρ = 0.989 between replicates and from 0.68 to 0.799 between replicates of the different treatments. Imputed protein abundance results of the 1.9 p.p.m.v. and 1000 p.p.m.v. treatments were used for correspondence analysis (CA). CA was conducted in R74 using the “ca” function of R package “ca”75. The top 10% of proteins that contributed most to the inertia of the CA’s first dimension were extracted using the function “get_ca_row” of the R package “factoextra”76. To map the abundances and the hierarchical EggNOG42 annotations of the top 10% proteins, the EggNOG annotations of the three strains were downloaded from MicroScope. For the in-depth analysis of trace gas oxidation, carbon assimilation, and the electron transport chain, the differences between treatments were tested using a two-sided t-test (s0 = 2) and permutation-based false discovery rate (FDR = 0.1). From the z-score normalized results, the proteins involved in trace gas oxidation, carbon assimilation via the serine cycle, and the electron transport chain were selected for further analysis using Python 368. The core carbon and energy metabolism of M. rosea SV97 and M. palsarum NE2 were reconstructed manually using MicroScope annotations, the published metabolism entries of M. gorgona MG0831, KEGG77, and protein BLAST searches78. To conduct a thorough screening for putative carbon monoxide dehydrogenase subunits a blast-searchable database was constructed out of all sequences retrieved from the NCBI Identical Protein Groups resource43 (accessed May, 30th 2023) using the search term “carbon monoxide dehydrogenase” (54076 amino acid sequence entries). Entries with an amino acid sequence length < 30 were removed from the database prior to the blastp search. The annotated genomes of M. rosea SV97 and M. palsarum NE2, downloaded from MicroScope, were used as blastp queries (default settings, -outfmt 6 including slen). After a pre-filtering removing hits with an E-value > 1 and an alignment length < 50% of the aligned database entry, the blastp output tables were manually evaluated (Supplementary table S10-11).
Specific affinity
Filter cultures of M. gorgona MG08 and M. palsarum NE2 were pre-incubated on 10x diluted, EDTA-free NMS medium (DSMZ medium 921 with 10x the iron concentration and 1 µM lanthanum) for five months with air as sole energy and carbon source. The filter cultures were transferred into 250 mL glass bottles containing 250 mL of the mentioned medium. The glass bottles were capped with Safety-Caps (JR-S-11011, VICI AG International). For each strain, the CH4 concentrations in the 50 mL headspace were adjusted to approximately 1.9, 14, 30, 70, and 175 p.p.m.v. by adding 0–2.5 mL 1000 or 5000 p.p.m.v. CH4 in N2 (HiQ, Linde, Sweden). The headspace pressure of all bottles was adjusted to approximately 1.1 bar using air. After, the strains were incubated for 48 hours at 20°C. The change in CH4 was measured at incubation start, after 24 hours and 48 hours as described above. Standard curves were created using 2.5 and 50 p.p.m.v. CH4 in N2 (HiQ, AGA, Sweden). At the end of the oxidation experiment, cells in filter cultures were quantified as described above. The change in mass of CH4 in the headspace was calculated by applying the ideal gas law and adjusted for changes in pressure caused by gas removal during measurement. The mass of dissolved CH4 at different CH4 partial pressures in the headspace was calculated by applying the Henry’s law solubility constant for CH4 at 20°C. The Michaelis-Menten CH4 oxidation kinetics were modelled using the “nls” function of the “nlstools” R package79, specifying the “michaelis” model and providing start values for Km(app) and Vmax(app). The \({a}_{A}^{0}\) was calculated by dividing Vmax by Km.
N2 fixation at atmospheric trace gas concentrations
Filter cultures of M. gorgona MG08, M. palsarum NE2, M. rosea SV97, and M. aurea KYG were pre-incubated on 10x diluted, EDTA-free, and potassium nitrate (KNO3)-free NMS medium (DSMZ medium 921 with 10x the iron concentration and 1 µM lanthanum) with air as sole energy, carbon, and nitrogen source. After three months, colony formation was checked via microscopy31. After 12 months, the activity of M. gorgona MG08 was measured as described in the section “Trace gas oxidation experiments” with the only difference that the KNO3-free medium was used.
For the detection of 15N2 fixation via NanoSIMS, filter cultures were pre-incubated on 10x diluted, EDTA-free, and potassium nitrate (KNO3)-free NMS medium with air as sole energy, carbon source, and nitrogen source. After three months, the filters were transferred into 250 ml glass bottles containing 50ml of the KNO3-free NMS medium. The bottles were capped with Safety-Caps (JR-S-11011, VICI AG International) and the headspace of each bottle was flushed for five minutes with compressed air using a gassing manifold. Afterwards, 50 ml of 98 + at% 15N2 gas (NLM-363-1-LB, Cambridge Isotopes Laboratories, Tweksbury, Massachusetts, USA) were added using a gas syringe so that the 15N-N2 in headspace atmosphere amounted approximately 23 at% of the total N2. The headspace pressure was adjusted to 1.05 bar and 0.5 mL 1000 p.p.m.v. CH4 in N2 (HiQ, AGA, Sweden), 1 mL 1000 p.p.m.v. H2 in N2 (HiQ, AGA, Sweden), and 1 mL 1000 p.p.m.v. CO in N2 (HiQ, AGA, Sweden) were added to the headspace using a gas syringe. To account for the natural abundance of 15N-N2, a control without the addition of 15N2 gas was prepared accordingly. The filter cultures were incubated at 20°C for two months. The headspace concentration of CH4, H2, and CO during incubation was measured once a week as mentioned in the “Trace gas oxidation experiments” section and replenished if the concentration of CH4, H2, and CO dropped below 1.9 p.p.m.v., 0.5 p.p.m.v., 0.2 p.p.m.v., respectively.
After two months, the cells of the filter cultures were fixed by incubating the filters on 150 µl formaldehyde in 1×PBS (4% w/v) for an hour at room temperature. The cells were washed twice by letting the filters float on water for 5 minutes. The fixed cells were harvested as described in “Cell quantification”, lyophilized, and stored at -80°C until further processing. For NanoSIMS analysis the lyophilized cells were resuspended in 20 µl MilliQ water. 10 µl of the cell suspensions were deposited on antimony-doped silicon wafer platelets (7.1 × 7.1 × 0.75 mm, Active Business Company, Germany) and dried in air. The following two samples were prepared accordingly for NanoSIMS analysis: i) The unlabeled M. gorgona MG08 cells, serving as control for the natural abundance of 15N2; and ii) the M. gorgona MG08 cells incubated with 15N2 gas and trace concentrations of CH4, H2, and CO that were analyzed to see whether the strain is capable of fixing atmospheric nitrogen during growth with air as sole energy and carbon source.
The NanoSIMS measurements were performed on a NanoSIMS 50 l (Cameca, Gennevilliers, France) at the Large-Instrument Facility for Advanced Isotope Research at the University of Vienna. Before the data acquisition, analysis areas were preconditioned in situ by rastering of a high-intensity, defocused Cs+ ion beam in the following sequence of high and extreme low ion impact energies (HE/16 keV and EXLIE/50 eV, respectively): HE at 25 pA beam current to a Cs + fluence of 5.0E14 ions cm− 2; EXLIE at 400 pA beam current to a fluence of 5.0E16 ions cm− 2; and HE at 25 pA to a fluence of 5.0E14 ions cm− 2. Data were acquired as multilayer image stacks by repeated scanning of a finely focused Cs+ primary ion beam (c. 80 nm probe size at approx. 2 pA beam current) over areas between 34 × 34 and 72 × 72 µm2 at 512 × 512-pixel image resolution and a primary ion beam dwell time of 5 ms pixel− 1.
NanoSIMS images were generated and analyzed with the OpenMIMS plugin80 in the image processing package Fiji81. All images were auto-tracked for compensation of primary ion beam and/or sample stage drift, and secondary ion signal intensities were corrected for detector dead-time and quasi-simultaneous arrival (QSA) of secondary ions, utilizing sensitivity factors (‘beta’ values) of 1.06, and 1.05 for C2−, and CN− ions, respectively. Regions of interest (ROIs) were defined in the 12C−-ion image where each ROI corresponded to an individual cell. Cells touching the border of the image were omitted from the selection.
Acquisition cycles of all three analysis areas were reduced to 27 each to improve comparability among the measurements. Regions of interest were analyzed for their 15N content by calculating the average value across acquisition cycles per analysis area, referred to by the 15N/(14N + 15N) isotope fraction designated as at% 15N, which was calculated from the 12CN− signal intensities via:
$$at\% {}^{15}N= \frac{{}^{12}C{}^{15}{N}^{-}}{{}^{12}C{}^{15}{N}^{-}+{}^{12}C{}^{14}{N}^{-}}$$
The natural abundance of 15N in cellular biomass was inferred from unlabeled cells yielding 0.369 ± 0.043 at% (mean ± 1 SD).
Plotting
Plots were created using the Python 368 packages Pandas69, Matplotlib70, and Seaborn71 and R packages ggplot282. Figures were finalized using Adobe Illustrator 2023.