Non-enzymatic methane formation by aerobic organisms

Methane (CH4) is the most abundant organic compound in the atmosphere, largely originating from anthropogenic and natural biogenic sources1. Traditionally, biogenic CH4 has been regarded as the nal product of the anoxic decomposition of organic matter by methanogenic Archaea. However, plants2–4, fungi5, algae6,7 and cyanobacteria8 have recently been shown to produce CH4 in the presence of oxygen. While methanogens produce CH4 enzymatically during anaerobic energy metabolism9, the requirements and pathways for CH4 production by “non-methanogenic” cells are poorly understood. Here we demonstrate that CH4 formation by Bacillus subtilis is triggered by free iron species, enhanced by oxidative stress and restricted to metabolically active life-cycle stages. We also show that other model organisms from Bacteria and Eukarya including a human cell line release CH4 and respond to inducers of oxidative stress by enhanced CH4 formation. Our results imply that all living cells possess a common mechanism of CH4 formation without the need for specic enzymes. We propose that CH4 formation is a conserved feature of living systems which is coupled to metabolic activity and the concomitant generation of reactive oxygen species. Our ndings open new perspectives for our understanding of environmental CH4 cycling, oxidative stress responses and the search for extraterrestrial life.

we demonstrate that CH4 formation by Bacillus subtilis is triggered by free iron species, enhanced by oxidative stress and restricted to metabolically active life-cycle stages. We also show that other model organisms from Bacteria and Eukarya including a human cell line release CH4 and respond to inducers of oxidative stress by enhanced CH4 formation. Our results imply that all living cells possess a common mechanism of CH4 formation without the need for speci c enzymes. We propose that CH4 formation is a conserved feature of living systems which is coupled to metabolic activity and the concomitant generation of reactive oxygen species. Our ndings open new perspectives for our understanding of environmental CH4 cycling, oxidative stress responses and the search for extraterrestrial life.

Background
Methane is a highly potent greenhouse gas that affects Earth's climate. Around 70 % of all emissions to the atmosphere derive from biogenic sources 10 . Biological CH 4 formation has long been considered to occur only under strictly anoxic conditions in organisms belonging to the domain Archaea. To generate the cellular fuel ATP, methanogenic archaea convert simple compounds, such as CO 2 , H 2 or acetate, into CH 4 . This process of methanogenesis depends on reactions that are catalysed by unique sets of enzymes and co-enzymes 9 . In addition, small amounts of CH 4 can be formed via 'mini-methanogenesis' in several sulphate-reducing bacteria which contain the enzyme carbon monoxide dehydrogenase 11,12 . However, during the past 15 years, evidence has been accumulating that other organisms produce CH 4 under aerobic conditions. These include both multicellular organisms, such as plants [2][3][4] and saprotrophic fungi 5 , and unicellular organisms, including marine and freshwater algae 6,13 and cyanobacteria 8 . These organisms generate energy via photosynthesis and/or respiration, and it is unclear why and how they release CH 4 . Multiple marine and freshwater bacteria harbouring the C-P lyase pathway have been reported to generate CH 4 from methylphosphonate [14][15][16][17][18] . Several bacteria and archaea have also been shown to possess alternative nitrogenases 19,20 or nitrogenase-like reductases 21 , which can produce CH 4 and other hydrocarbons. In addition -as we will demonstrate here -living systems can form CH 4 without the need for speci c enzymes, and such pathways could drive CH 4 formation in all cells.
In plants, CH 4 formation is enhanced by UV irradiation 3,22,23 , inhibition of cytochrome c oxidase 24 and environmental stressors 4,25 . This raises the question whether reactive oxygen species (ROS, e.g., hydrogen peroxide H 2 O 2 , hydroxyl radicals ·OH and superoxide radicals O 2 -) might be involved in the formation of CH 4 . In highly oxidizing environments generated in vitro by a chemical model system containing iron (II/III), hydrogen peroxide (H 2 O 2 ) and the radical scavenger ascorbic acid, CH 4 is readily formed from organosulphur compounds at ambient temperatures 26 . Under Fenton-type conditions, nonheme oxo-iron(IV) ([Fe IV =O] 2+ ) oxidizes methyl sulphides to sulphoxides, which then results in selective formation of methyl radicals by sulphoxide demethylation, and ultimately leads to CH 4 and methanol formation depending on oxygen availability 26, 27 . Moreover, ROS such as ·OH and H 2 O 2 react with methyl sulphoxides to produce methyl radicals or peroxomethyl radicals in the presence of oxygen 28 , subsequently resulting in generation of CH 4 or oxidized C 1 species.
Fenton chemistry takes place in living cells, as iron is an essential trace element 29 and H 2 O 2 is produced during metabolism 30 . Hydrogen peroxide and Fe 2+ react to either ferric iron (Fe 3+ ), OHand ·OH radicals, or alternatively [Fe IV =O] 2+ and water 31 . This therefore raises the possibility that non-enzymatic CH 4 formation occurs in cells under oxic conditions at ambient temperatures. Suitable methyl donors for CH 4 formation could, in principle, be derived from a wide spectrum of molecules containing sulphur-and/or nitrogen-bonded methyl groups, which are endogenously produced during cellular metabolism and exogenously secreted during natural product formation. Such methylated sulphur compounds include methionine, dimethylsulphoniopropionate (DMSP), dimethyl sulphide (DMS) and dimethyl sulphoxide (DMSO), which are ubiquitous in the environment 32 . Thus, we tested the hypothesis that a non-enzymatic pathway of CH 4 formation exists in all cells, which is based on interactions between ROS, iron and methyl donors (Fig. 1A).

Formation of methane by Bacillus subtilis
We rst explored CH 4 formation in the Gram-positive model bacterium Bacillus subtilis which, to the best of our knowledge, is not known to release CH 4 . To detect authentic CH 4 formation, we performed stable carbon isotope labelling experiments, using sterile media as baseline controls, and measured CH 4 in the headspace of atmospherically sealed bacterial cultures (Fig. 1B). We rst investigated whether CH 4 could be derived from glucose, an abundant carbon source that is readily taken up and metabolized by B.
subtilis. Indeed, small but signi cant amounts of CH 4 were formed in bacterial cultures from [ 13 C 6 ]glucose (0.21 nmol, p < 0.001). Moreover, the corresponding 13 C-isotope values (Δδ 13 C; difference between culture and sterile media) were highly enriched (79 ± 1 ‰). This indicates that glucose metabolism in B. subtilis results in at least one precursor compound that contributes to the formation of CH 4 . We next investigated whether B. subtilis could utilize methyl sulphide and methyl sulphoxide as exogenous precursor carbon compounds for CH 4 formation. In fact, addition of either 13 C-labelled DMS or DMSO to bacterial cultures resulted in a clear shift in the isotopic value, demonstrating that the methyl groups of each sulphur compound were converted into CH 4 by reduction. Addition of DMSO increased CH 4 formation by 8-fold (1.85 nmol, p < 0.001) relative to glucose. Formation of CH 4 containing 13 Clabelled carbon derived from DMS was also observed. Although in the latter case CH 4 amounts were low, and varied from experiment to experiment, Δδ 13 C values were correlated with the amount of CH 4 produced. Taken together, these experiments provide unambiguous evidence for CH 4 formation from both exogenous and endogenous carbon compounds by B. subtilis.

Free iron and oxidative stress promote CH 4 formation
We further tested our hypothesis by varying the availability of suitable methyl donors, free iron and the level of oxidative stress, respectively (Fig. 1C). To minimize the effects of biomass variation on CH 4 formation, we used the same stationary-phase culture in a minimal medium and systematically varied the reaction conditions for CH 4 formation by supplying the substrate DMSO, additional free iron in the form of Fe 2+ (as FeSO 4 ) and hypochlorous acid (HOCl) to induce oxidative stress in all possible combinations.
The optical density (OD 600nm ) of stationary cultures changes only marginally if at all (±1%), which implies that the observed variation in CH 4 can be attributed to increased CH 4 formation by B. subtilis.
Under these conditions, external supply of DMSO (S) was required to stimulate signi cant CH 4 formation, as the amounts of CH 4 produced in its absence were close to atmospheric levels ( Fig. 1C). In low-iron environments, addition of DMSO resulted in a small but signi cant quantity of CH 4 (0.45 nmol ± 0.03 nmol). Supplementation with HOCl to stimulate oxidative stress (S+O) increased this yield by ~40%. Moreover, an iron-rich environment (S+F) boosted CH 4 formation from DMSO by 17-fold, while the additional imposition of HOCl-induced oxidative stress (S+F+O) resulted in a 35-fold increase in CH 4 levels relative to substrate alone (S). In contrast, heat-inactivated B. subtilis released only marginal amounts of CH 4 (0.038 ± 0.006 nmol, p = 0.007) in conditioned media containing DMSO, Fe 2+ and HOCl, suggesting that viable but not dead biomass promotes formation of CH 4 . In accordance with our hypothesis, these data show that CH 4 formation from DMSO in stationary-phase B. subtilis cells is enhanced by both free iron and oxidative stress. CH 4 formation is restricted to metabolically active life-cycle stages According to our model, the generation of H 2 O 2 /ROS by living organisms is a prerequisite for CH 4 formation in the presence of suitable methyl donors. B. subtilis is a spore-forming organism that cycles between a dormant endospore state and a vegetative cell that grows and divides. Dormant spores are considered to be metabolically inactive, but resume metabolism upon germination 33 . DMSO readily enters dormant Bacillus spores and does not interfere with spore revival 34 . We thus asked whether the ability of B. subtilis to generate CH 4 from DMSO is restricted to its metabolically active life-cycle stages ( Fig. 2A). To investigate this, we performed stable isotope labelling experiments on dormant and germinated B. subtilis spores using 2 H-labelled DMSO (Fig. 2B). The use of 2 H-labelled compounds and measurements of δ 2 H values increases the detectability of CH 4 formation by orders of magnitude in comparison to 13 C labelling. Despite the exquisite sensitivity of the assay, CH 4 release from dormant spores after 10 h of incubation with DMSO was within the margin of error (0.02 nmol, p = 0.066), as were the stable hydrogen isotope signatures (Δδ 2 H = 10 ± 9 ‰). In contrast, when spores were germinated by supplying the nutrient germinant mixture AGFK, CH 4 formation was clearly detected (0.1 nmol, P = 0.019).
Moreover, the stable hydrogen isotope signatures increased by three orders of magnitude, unambiguously con rming the conversion of the 2 H-labelled methyl group of 2 H-DMSO into CH 4 (Fig. 2B). Note that, under our experimental conditions, AGFK-induced germination is evidenced by the loss of spore refractivity; however, these spores did not grow out (Fig. 2B, insets). These data therefore suggest that germinated and thus metabolically active -but not dormant -spores can generate CH 4 from DMSO, in accordance with our hypothesis. CH 4 accumulates during growth under oxic conditions According to our model, CH 4 is mainly formed under oxic conditions. Since B. subtilis is a facultative anaerobe 35 , we followed the dynamics of CH 4 accumulation during growth in complex media while monitoring the development of oxygen levels in the sample headspace (Fig. 2C). Methane levels increased during the exponential growth phase when oxygen was at atmospheric levels. Coincident with oxygen depletion, population growth slowed down rapidly, while CH 4 levels continued to increase for another ~ 2.5 h into stationary phase. Little additional CH 4 was formed during later cultivation stages when oxygen levels were low and the OD 600nm of the culture gradually declined. This indicates that B.
subtilis forms CH 4 under oxic conditions, in agreement with our model. All microorganisms formed CH 4 , and levels were enhanced under oxidative stress induced by the addition of HOCl (Fig. 3B). Cellular CH 4 formation by bacteria and fungi exceeded the respective (sterile) media controls by factors from ~1.1 minimum (non-stressed S. cerevisiae) to ~423 maximum (HOCl-stressed A. niger). We also observed CH 4 formation in HEK293T cells (Fig. 3C)

Discussion
In summary, our study strongly suggests that CH 4 is formed as a universal by-product of life, in addition to known CH 4 generation mechanisms by (mini-)methanogenesis 9,11 , by alternative nitrogenases 19,20 and other enzymes 14,21 (e.g. C-P lyase). All organisms investigated here were not previously known to produce CH 4 . Our results clearly demonstrate continuous formation of CH 4 , which is substantially enhanced upon supplementation with free iron and induction of oxidative stress, in accordance with the proposed underlying mechanism. The use of position-speci c 13  Formation of CH 4 as a consequence of fundamental biological processes implies that living biomass in all of Earth's environmental compartments continuously releases a background amount of CH 4 . Although the CH 4 production by the investigated organisms is relatively small, if compared to methanogenic Archaea living in anaerobic environments, CH 4 release from aerobic organisms into the atmosphere might be of relevance to the biogeochemical cycles of CH 4 . This has recently been proposed for cyanobacteria 8 and marine and freshwater algae 6, 13 and their contribution to the abundance of CH 4 in oxygen-rich surface waters, commonly known as the "methane paradox" 18 . Note that it is not possible to quantitatively compare the amounts of CH 4 formed by the different organisms (e.g. on a dry weight basis) examined in our study. Determination of CH 4 production rates under eld-like conditions requires an enormous experimental and analytical effort and must therefore await further studies.
Secondly, our results imply that the CH 4 formed in cells may be an integral part of their responses to changes in their oxidative status. Non-enzymatic formation of CH 4 was shown here to be coupled to cell metabolism. Moreover, in all investigated cells CH 4 release substantially increased under conditions that promote oxidative stress. Since many factors can cause secondary changes in oxidative stress in cells 37 , these ndings provide an explanation for elevated CH 4 formation under stress, which has previously been observed in plants 4,22,24,38 , animals 39,40 and phototrophic organisms 4,8,41,42 . We therefore propose that CH 4 might serve as a read-out for stress and hypoxia 43 . For example, monitoring of variations in the CH 4 content in human breath has revealed age-and disease-dependent changes in cell metabolism 44,45 . Furthermore, it is conceivable that cells have evolved means to utilize CH 4 as a signalling molecule to trigger adaptive stress responses 46,47 . This in turn could provide a rationale for the use of CH 4 as a therapeutic gas 43,48 . In this context, it has been shown that supplementation of CH 4 in animals has antiin ammatory effects 49,50 and strengthens defences against organ dysfunction 51   Quanti cation of CH 4 (GC-FID) CH 4 was quanti ed by gas-chromatography ame-ionization detection (GC-FID). Aliquots (6 mL) of the sample headspace were retrieved using a gas-tight syringe (20 mL BD Luer-Lok TM , Becton Dickinson, Madrid, Spain) with a needle (Large Hub RN, Hamilton, Bonaduz, Switzerland) and injected into a gas chromatograph [a high-grade steel tube (column length 2 m, inner diameter 3.175 mm) packed with Molecular Sieve 5A 60/80 mesh from Supelco] connected to a ame-ionization detector (Shimadzu GC-14B). Levels were quanti ed by comparison of the measured CH 4 peak area to that of a reference standard containing 2192 parts per billion by volume (measured three times). Unless indicated otherwise, CH 4 formation was calculated by subtracting the level of CH 4 of sterile medium from that of the inoculated culture. δ 13 C stable isotope measurements (GC-C-IRMS) δ 13 C values of CH 4 were determined by gas chromatography-combustion-isotope ratio mass spectrometry (GC-C-IRMS). Aliquots of sampled headspace gas were transferred to an evacuated sample loop (40 mL) and a cryogenic pre-concentration unit in order to trap CH 4 . CH 4 was trapped on HayeSep D, separated from interfering compounds by GC and transferred to a gas chromatography-combustionisotope ratio mass spectrometer (GC-C-IRMS). The system consists of a cryogenic pre-concentration unit that is directly connected to a HP 6890N GC (He ow rate: 1.8 mL min -1 ; Agilent Technologies, Santa Clara, USA) tted with a GS-Carbonplot capillary column (30 m * 0.32 mm i.d., d f 1.5 µm; Agilent Technologies) and a PoraPlot capillary column (25 m * 0.25 mm (i.d.), d f 8 µm; Varian, Lake Forest, USA). The GC ow was coupled via a press-t connector to a combustion reactor comprised of an oxidation reactor (ceramic tube (Al 2 O 3 ), length 320 mm, inner diameter 0.5 mm, with oxygen-activated Cu/Ni/Pt wires inside; reactor temperature 960°C) and a GC Combustion III Interface (ThermoQuest Finnigan) to decompose CH 4 into CO 2 . 13 C/ 12 C ratios were determined with a Delta PLUS XL mass spectrometer (ThermoQuest Finnigan, Bremen, Germany). High-purity CO 2 (carbon dioxide 4.5, Messer Griesheim, Frankfurt, Germany) was used as the working monitoring gas. 13 C/ 12 C ratios (δ 13 C values) are expressed in the conventional δ notation in per mil versus VPDB, calculated as: see formula 1 in the supplementary les. δ 13 C values were corrected using three reference standards of high-purity CH 4 with δ 13 C values of -54.5 ± 0.2 ‰ (Isometric Instruments, Victoria, Canada), -66.5 ± 0.2 ‰ (Isometric Instruments) and -42.3 ± 0.2 ‰ (in-house), calibrated against International Atomic Energy Agency and NIST reference substances. Finally, the isotope difference (Δδ 13 C) between the mean δ 13 C values of sample and nonsupplemented media was calculated as Δδ 13 C = δ 13 C sample -δ 13 C media . δ 2 H stable isotope measurements (GC-TC-IRMS) δ 2 H values of CH 4 were determined via gas chromatography-temperature conversion-isotope ratio mass spectrometry (GC-TC-IRMS). Here, the same analytical set-up was applied as for δ 13 C stable isotope measurements (changed He ow rate: 0.6 mL min -1 ) but instead of combustion to CO 2 and H 2 O, CH 4 was thermolytically converted (at 1450 °C) to hydrogen and carbon. After IRMS measurements, the obtained δ 2 H values were corrected by using two reference standards of high-purity CH 4 with δ 2 H values of -149.9 ± 0.2 ‰ (T-iso2, Isometric Instruments) and -190.6 ± 0.2 ‰ (in-house). All 2 H/ 1 H ratios (δ 2 H values) are expressed in the conventional δ notation in per mil versus VSMOW, calculated as: See formula 2 in the supplementary les.
Finally, the isotope difference (Δδ 2 H) between the mean δ 2 H values of sample and non-supplemented media was calculated as Δδ 2 H = δ 2 H sample -δ 2 H media .

Oxygen measurements (GC-BID)
The atmospheric oxygen content was determined via gas chromatography-barrier discharge ionization detection (GC-BID) using a GC-2010 Plus (Shimadzu, Japan). The GC-2010 (Shimadzu, Japan) comprises a stainless-steel ShinCarbon ST packed column (80/100 mesh; length: 2 m; diameter: 0.53 mm). The GC was operated with an injection temperature of 230 °C and an injection volume of 50 µL. Helium 6.0 (ALPHAGAZ, Air Liquide, France) served as the carrier gas with a ow rate of 50 mL min -1 . The temperature of the GC oven was held at 30 °C for 6.5 min, increased to 75 °C at a rate of 10 °C/min and then rising to 180 °C at a rate of 30 °C/min. The oxygen content of the sample was determined by comparing the obtained O 2 peak with a sample of atmospheric air (O 2 content 20.95%).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.