Large hydrogen isotope fractionations distinguish nitrogenase-derived methane from other sources

Nitrogenase is the main source of natural fixed nitrogen for the biosphere. Two forms of this metalloenzyme, the vanadium (V) and iron (Fe)-only nitrogenases, were recently found to reduce small amounts of carbon dioxide into the potent greenhouse gas methane. Here we report carbon and hydrogen stable isotopic compositions and fractionations of methane generated by V- and Fe-only nitrogenases in the metabolically versatile nitrogen fixer Rhodopseudomonas palustris. The stable carbon isotope fractionation imparted by both forms of alternative nitrogenase are within the range observed for hydrogenotrophic methanogenesis (13αCO2/CH4 = 1.051 ± 0.002 for V-nitrogenase and 1.055 ± 0.001 for Fe-only nitrogenase, mean ± SE). In contrast, the hydrogen isotope fractionations (2αH2O/CH4 = 2.071 ± 0.014 for V-nitrogenase and 2.078 ± 0.018 for Fe-only nitrogenase) are the largest of any known biogenic or geogenic pathway. The large 2αH2O/CH4 shows that the reaction pathway nitrogenases use to form methane strongly discriminates against 2H, and that 2αH2O/CH4 distinguishes nitrogenase-derived methane from all other known biotic and abiotic sources. These findings on nitrogenase-derived methane will help constrain carbon and nitrogen flows in microbial communities and the role of the alternative nitrogenases in global biogeochemical cycles. Importance All forms of life require nitrogen for growth. Many different kinds of microbes living in diverse environments make inert nitrogen gas from the atmosphere bioavailable using a special protein, nitrogenase. Nitrogenase has a wide substrate range, and in addition to producing bioavailable nitrogen, some forms of nitrogenase also produce small amounts of the greenhouse gas methane. This is different from other microbes that produce methane to generate energy. Until now, there was no good way to determine when microbes with nitrogenases are making methane in nature. Here, we developed an isotopic fingerprint that allows scientists to distinguish methane from microbes making it for energy versus those making it as a byproduct of nitrogen acquisition. With this new fingerprint, it will be possible to improve our understanding of the relationship between methane production and nitrogen acquisition in nature.

small amounts of carbon dioxide into the potent greenhouse gas methane. Here we report carbon 23 and hydrogen stable isotopic compositions and fractionations of methane generated by V-and 24 Fe-only nitrogenases in the metabolically versatile nitrogen fixer Rhodopseudomonas palustris. 25 The stable carbon isotope fractionation imparted by both forms of alternative nitrogenase are 26 within the range observed for hydrogenotrophic methanogenesis ( 13  CO2/CH4 = 1.051 ± 0.002 for 27 V-nitrogenase and 1.055 ± 0.001 for Fe-only nitrogenase, mean ± SE). In contrast, the hydrogen 28 isotope fractionations ( 2  H2O/CH4 = 2.071 ± 0.014 for V-nitrogenase and 2.078 ± 0.018 for Fe-29 only nitrogenase) are the largest of any known biogenic or geogenic pathway. The large 30 2  H2O/CH4 shows that the reaction pathway nitrogenases use to form methane strongly 31 discriminates against 2 H, and that 2  H2O/CH4 distinguishes nitrogenase-derived methane from all 32 other known biotic and abiotic sources. These findings on nitrogenase-derived methane will help 33 constrain carbon and nitrogen flows in microbial communities and the role of the alternative 34 nitrogenases in global biogeochemical cycles. 35 36 Importance 37 All forms of life require nitrogen for growth. Many different kinds of microbes living in diverse 38 environments make inert nitrogen gas from the atmosphere bioavailable using a special protein, 39 nitrogenase. Nitrogenase has a wide substrate range, and in addition to producing bioavailable 40 nitrogen, some forms of nitrogenase also produce small amounts of the greenhouse gas methane. 41 This is different from other microbes that produce methane to generate energy. Until now, there 42

Introduction 49
Microorganisms produce over half of global methane emissions (1). Fermentative and 50 hydrogenotrophic methanogens are the most significant microbial producers of this potent 51 greenhouse gas (1, 2). Their metabolic pathways occur exclusively within anaerobic Archaea and 52 involve multiple enzymes working together in series, including the obligatory methyl-coenzyme 53 M reductase (mcr) enzyme. Its primary function is for catabolism, with methane production 54 thought to occur only after other more favorable electron acceptors, like oxygen, nitrate, or 55 sulfate have been consumed (3-5). Over the past decade, it has been recognized that minor 56 additional contributions of methane derive from the demethylation of organophosphonates (c.f. 57 (6-8)) and from inefficient Wood-Ljungdahl pathway carbon fixation (9). Most recently, it was 58 discovered that some forms of the metalloenzyme nitrogenase also reduce carbon dioxide 59 straight into methane (10). Nitrogenases are the only biological source of newly fixed nitrogen to 60 the biosphere, and prior to industrial reduction of dinitrogen, were the primary source of nitrogen 61 to life on Earth (11, 12). The discovery of biological methane production by nitrogenase expands 62 the known range of organisms and environments in which methane production is possible. 63 Nitrogenase is known primarily for its ability to reduce inert dinitrogen (N 2 ) gas into 64 ammonia, a process known as nitrogen fixation. This biological nitrogen source plays a critical 65 role in ecosystem fertility. Nitrogenase is generally considered a promiscuous enzyme because it 66 can reduce a variety of carbon containing compounds in addition to N 2 (13-17). For example, the 67 iron (Fe)-only nitrogenase isoform can convert carbon monoxide into hydrocarbon chains, a 68 reaction which may have been important for early forms of life (15). In addition, all forms of 69 nitrogenase reduce acetylene to ethylene (18-21), which is the basis for the most commonly used 70 acetylene reduction method to measure nitrogen fixation rates in the laboratory and field (22-71 24). The recent discovery that some forms of nitrogenase can reduce carbon dioxide to methane 72 (10) is significant because, unlike acetylene and carbon monoxide, carbon dioxide is ubiquitous 73 in nature. 74 The vanadium (V-) and Fe-only nitrogenase isoforms, which were shown to produce the 75 most byproduct methane of the various nitrogen isoforms (10), are found in both the bacterial 76 and archaeal domains and are widespread in nature (25-30). In addition, certain artificial 77 mutations near the active site of the molybdenum (Mo)-nitrogenase enabled this more common 78 isoform to produce methane (31, 32 nitrogenase strain producing over an order of magnitude more methane than the V-nitrogenase 104 strain (Fig. 1). For the Fe-nitrogenase strain, methane production per cell was higher later during 105 growth. We measured the carbon and hydrogen isotopic compositions of methane and 106 fractionations relative to carbon dioxide (CO 2 /CH 4 ) and water (H 2 O/CH 4 ), as produced by the V-107 and Fe-only nitrogenases across a range of cell densities (OD 660 ~ 0.3 to 1.3, from early log to 108 stationary phase), temperatures (14 to 30°C), carbon substrates (succinate and acetate), and 109 growth medium pH (from 6.2 to 6.8 at inoculation). 110

143
We observed only small changes in nitrogenase fractionation across a large range of 144 temperatures, cell densities, and carbon substrates (<0.02 for 13  CO2/CH4 and < 0.38 for 2  H2O/CH4 ; 145 Figs. 4, 5) relative to the variability observed for other methane production pathways. 146 Fractionation increased by ~ 0.012 as temperature decreased from 30 to 14°C for 13  CO2/CH4 (p = 147 10 -5 ) and by ~ 0.160 for 2  H2O/CH4 (p = 0.03). In contrast, the form of growth substrate (succinate 148 or acetate) did not alter 2  H2O/CH4 (p = 0.96) and only had a small impact of ~0.005 on 13  CO2/CH4 149 (p = 0.006). This is compatible with the recent observation that electron availability has only a 150 minor impact on CH 4 production by a mutant Mo-nitrogenase isoform (44). Acidification of the 151 growth medium by ~0.5 pH units also did not alter fractionation, though we note that there was 152 only one biological replicate for the acidified treatment (Table 1). Despite order of magnitude 153 differences in the rate of methane production by V-and Fe-only nitrogenase ( Fig. 1), they have 154 indistinguishable fractionation factors associated with methane production (p = 0.4 for 13  CO2/CH4 155 and 0.9 for 2  H2O/CH4 ; Table 1). This suggests there is no rate effect on fractionation and that the 156 V-and Fe-only nitrogenases share a common mechanism for CO 2 reduction to methane. 157

158
The greatest source of variability in fractionation ( 13  CO2/CH4 ~ 0.01; 2  H2O/CH4 ~ 0.25 range) 159 appears to be due to cell density, growth phase (Figs. 5C, G), or substrate (CO 2 ) concentration 160 (Figs. 5D, H). These variables are strongly correlated due to dissolved inorganic carbon (DIC) 161 production throughout growth (Fig. 5I)  The methane isotopic composition at harvest integrates the isotopic composition of methane 168 produced throughout growth. Therefore, the fractionation measured at stationary phase is altered 169 by the change observed in substrate CO 2 isotopic composition during exponential phase (Fig.  170 5J). However, using the observed shift in medium CO 2 isotopic composition to estimate the 171 effect on the fractionation measured at stationary phase, we find that the change in substrate 172 isotopic composition could account for at most half (~0.005) of the total (~0.01) shift observed in 173 13  CO2/CH4 with cell density (see S.I.). We note that it is possible that the isotopic composition of 174 intracellular CO 2 is somewhat different from the bulk composition due to localized production, 175 consumption, and depletion, given the competing reactions of CO 2 production during organic 176 substrate assimilation and re-fixation by Rubisco during photoheterotrophic growth of R. 177 palustris (41, 46, 47). 178

179
We observed changes in fractionation correlated with temperature, growth phase and dissolved 180 inorganic carbon (DIC) concentration but not with organic carbon substrate or total methane 181 production rate. Though the variability in fractionation during methane production by 182 nitrogenase is interesting from a mechanistic perspective, the range of measured hydrogen 183 isotope fractionation does not overlap with, and is readily distinguishable from, the range 184 observed for other methane production pathways (Fig. 4). This is consistent with the observation 185 that N 2 and acetylene (C 2 H 2 ) fractionations by a single nitrogenase isoform are also remarkably 186 constant across different organisms, metabolisms and environmental conditions (26, 33). 187 188

Hydrogen concentration does not influence methane isotope fractionation by nitrogenase 189
Hydrogen (H 2 ) is an obligatory product of nitrogen fixation and, in our experiments, is generated 190 simultaneously with the production of methane from carbon dioxide (48, 49). We explored 191 whether its buildup could affect methane isotope fractionation by nitrogenase, as has been 192 suggested for mcr-based methanogenesis (2, 42, 50-58). Two lines of evidence show that the 193 presence of H 2 does not alter the isotopic composition of methane produced by nitrogenase. 194 Firstly, for Fe-only nitrogenase cultures (grown on succinate at 19C in serum vials), the 195 hydrogen isotope fractionations were indistinguishable in cultures in which the headspace 196 contained 2-3% H 2 at inoculation ( 2  H2O/CH4 = 2.068  0.033, n = 3) and in cultures that were 197 flushed with 100% N 2 prior to inoculation ( 2  H2O/CH4 = 2.046  0.016, n = 4, p = 0.57; S.I. 198 Table). These data show that exogenous H 2 did not influence the isotopic composition of the 199 product methane. This result is expected given that the strains used in our experiments lack a 200 functional uptake hydrogenase (59) and that nitrogenase itself does not catalyze isotope 201 exchange between water and H 2 (60). (This is a significant distinction from the hydrogenation of 202 D 2 , forming HD, which nitrogenase can catalyze in the presence of N 2 ). We note that abiotic 203 hydrogen isotopic equilibration between H 2 -H 2 O, CH 4 -H 2 and CH 4 -H 2 O is likely too slow to be 204 important at the timescales (~weeks) and temperatures ( 30C) of relevance to our experiments 205 (36, 61-63). This finding is consistent with other reports that the source of protons for CO 206 reduction by nitrogenase is water, not hydrogen gas (16). 207

208
The second line of evidence demonstrating that the hydrogen concentration does not influence 209 nitrogenase methane isotope fractionation is based on comparing the fractionations observed in 210 different growth containers and for the different strains. For a given growth container and strain, 211 cell density and hydrogen concentration are correlated ( Fig. 6A; also see S.I. Discussion). 212 However, their respective effects on fractionation can be disentangled by comparing data from 213 the Balch tubes (10 mL medium : 17 mL headspace) and serum vials (180 mL medium : 60 mL 214 headspace). As seen in Figure 6, hydrogen and carbon isotope fractionations in cultures with 10 215 to 20% H 2 in their headspace at harvest overlap with those of cultures with 20 to 50% H 2 in their 216 headspace at harvest (p > 0.5; Figs. 6C, E). We conclude that fractionation during methane 217 production by nitrogenase is not sensitive to hydrogen concentration over the large range (10 to 218 50%) tested here. This is compatible with findings that CO 2 reduction by Mo-nitrogenase is not 219 competitively inhibited by H 2 and does not proceed through the same reversible re (reductive 220 elimination of H 2 ) step as N 2 reduction (64). The lack of hydrogen partial pressure dependency 221 on fractionation contrasts with some modes of mcr-based methanogenesis. 222 223

Mechanistic implications for nitrogenase 224
Determining whether isotope effects are due to equilibrium or kinetic fractionation and under 225 what conditions they are fully expressed can help elucidate the mechanism, intermediates, and 226 reversibility of a reaction. At 20C, the equilibrium hydrogen isotope fractionation predicted 227 between methane and water is only 2  H2O/CH4 ~ 1.019 (65). This is much smaller than the 228 fractionation observed for nitrogenase, suggesting that kinetic, rather than equilibrium, isotope 229 effects are responsible for the large hydrogen isotope fractionation observed here. This 230 conclusion is consistent with the finding that fractionation of CO 2 reduction by nitrogenase is 231 larger at colder temperatures (Fig. 5B, F), which is generally incompatible with an equilibrium 232 isotope effect (66). These results lead us to attribute the fractionation observed here to a kinetic 233 isotope effect (KIE) in which CH 4 methane production by V-and Fe-only nitrogenase is roughly 234 twice as fast as CH 3 D methane production (1.820  2  H2O/CH4 = KIE  2.199). We suggest this 235 new value can help yield insight into the mechanism of CO 2 reduction by nitrogenase. 236

237
The mechanism of CO 2 reduction by nitrogenase is a subject of much study because of its 238 potential industrial application as a renewable fuel source (65, 67, 68 and references therein). 239 The observation that the hydrogen KIE during methane production is ~2 represents a new 240 experimental constraint for these studies. Previously, the KIE for H 2 production in the absence of 241 N 2 (i.e. E4 to E2 state, where E2 is an intermediate state in the sequential reduction of the active 242 site to prepare for N 2 binding at E4) by the Mo-nitrogenase was used as a tool to determine the 243 mechanism of H 2 loss during activation of the cofactor, a catalytically inefficient reaction that 244 competes with N 2 reduction (69). Khadka and colleagues (69) demonstrated, experimentally and 245 computationally, that the KIE of ~2.7 is due to preference for 1 H during protonation of the 246 bridging Fe-hydrides by highly acidic, protonated cofactor thiols. The KIE of ~2 observed here is 247 lower than the KIE measured for H 2 production. This hints that (1) the preference for 1 H might 248 be somewhat lower for V-and Fe-only nitrogenase compared to the Mo-nitrogenase (e.g., (70-249 72) and references therein for examples of the effect that the cofactor and amino acid 250 environment have on protonation and substrate selectivity). Another possibility (2) is that, 251 because the mechanism of CO 2 reduction by nitrogenase likely involves the migratory insertion 252 of cofactor bound CO 2 into the Fe-hydride bond (64), the preference for 1 H is lower for the 253 bridging Fe-hydrides, which do not exchange with solvent at the timescales relevant to the 254 reaction, compared to protonated thiols, which do (69). It is also possible that proton tunneling, 255 which is generally thought to have a very large kinetic isotope effect (but also see 73, 74) and 256 has been proposed to occur in nitrogenase (75)  Given the ubiquity of CO 2 in cells and in the environment, it is likely that some CH 4 production 274 is occurring whenever V-and Fe-nitrogenase are active. To determine the extent to which stable 275 isotopes can attribute methane production to alternative nitrogenase activity in environments 276 with multiple sources, we developed a simple isotopic mixing model (Fig. 7). The model 277 calculates the net 2  H2O/CH4 and  2 H of the mixed methane pool given the local water isotopic 278 composition and the relative rates of methane production from traditional methanogenesis 279 pathways and nitrogenase activity, assuming that all hydrogen for acetoclastic methanogenesis 280 ultimately derive from environmental water. Hydrogen isotopic compositions as low as The alternative V-and Fe-only nitrogenases are important enzymes in the global nitrogen cycle. 320 The curious observation that these enzymes produce methane as a minor byproduct of nitrogen 321 fixation led us investigate how its isotopic compostion compares to other natural methane 322 sources. Here we show that the natural abundance deuterium to hydrogen ratio of methane 323 derived from nitrogenase is significantly lower than methane from all other known processes, 324 with  2 H as low as −550‰. This result provides new experimental constraints on the mechanism 325 of the nitrogenase enzyme and demonstrates that significantly depleted hydrogen stable isotopic 326 composition constitute a passive biosignature of V-and Fe-only nitrogenase-derived methane. 327 This isotopic fingerprint offers a means to probe the contribution of alternative nitrogen fixation 328 and nitrogenase methane emissions on Earth and beyond.  Table). The constant hydrogen isotope fractionation observed for Fe-only nitrogenase over a 361 >500‰ range in  2 H suggests that the analytical methods employed are robust (Fig. 3). Samples 362 for  13 C analysis of CO 2 were collected in the same manner as those for methane. Samples for 363  13 C of DIC were collected in He-flushed vials that contained 1 mL of concentrated HPLC grade 364 phosphoric acid (85%; Fisher Chemical). At the UC Davis Stable Isotope Facility, the  2 H CH4 , 365  13 C CH4 ,  13 C CO2 and  13 C DIC samples were measured on a Delta V Plus IRMS (Thermo 366 Scientific, Bremen, Germany) coupled to a Gas Bench II system. Water 2 H samples were 367 collected by filtering growth medium (0.22m) at the end of the experiment and storing at 368 −20C. For analysis, samples were thawed and 1.4 -1.5 mL were aliquoted into 2 mL soda glass 369 vials (Thermo Scientific, National C4010-1W with C4010-40A caps) and shipped on ice or at 370 room temperature overnight to the UC Davis Stable Isotope Facility, where they were measured 371 on a Laser Water Isotope Analyzer V2 (Los Gatos Research, Inc.). Biomass and substrate  13 C 372 were measured in the Zhang stable isotope laboratory at Princeton as described previously (41) 373 on a Vario ISOTOPE select (Elementar Isoprime). The standard deviation of standard material 374 replicates were < 1‰ for  2 H H2O , < 2‰ for  2 H CH4 , < 0.2‰ for  13 C CH4 (> 10 ppm), < 0.2‰ 375 for  13 C CO2 and 13 C DIC , and < 0.1‰ for  13 C biomass . 376 377 Isotope Calculations. Hydrogen and carbon isotopes are expressed using delta notation relative 378 to Vienna Standard Mean Ocean Water (VSMOW) and Vienna Pee Dee Belemnite (VPDB), 379 respectively. Apparent CO 2 -CH 4 and water-CH 4 isotope fractionation factors were calculated as 380 substrate over product using the equations: 381 13  CO2/CH4 = 13 R CO2 / 13 R CH4 = ( 13 C CO2 +1000)/( 13 C CH4 +1000) 382 2  H2O/CH4 = 2 R H2O / 2 R CH4 = ( 2 H H2O +1000)/( 2 H CH4 +1000) 383 In this manuscript, errors represent the standard error of multiple biological replicates. can be used as a biosignature for alternative nitrogenase activity, we developed a mixing model 388 that calculates the fractionation and isotopic composition of methane produced by multiple 389 sources (Fig. 7). We used the following parameters: 2  Nase = 2.07;  2 H H2O = −40‰ vs. VSMOW 390 as representative of the mid-latitudes and −150‰ vs. SVMOW as representative of northern 391 latitudes; and k = methane produced by nitrogenase : total methane produced by nitrogenase and 392 mcr-based anaerobic methanogenesis. For fermentative methanogenesis, the model assumes that 393 all protons ultimately derive from local water. The observed fractionation and isotopic 394 composition were calculated using the equations: 395 2 F Nase/mcrCH4 = 2 R Nase/mcrCH4 /(1+ 2 R Nase/mcrCH4 ) 396 2 F CH4 = k* 2 F NaseCH4 + (1-k)* 2 F mcrCH4 397 20 398 Data Availability. Individual datapoints are available in the S.I. Table. In addition to the S.I. 399  other natural methane sources due to its more depleted hydrogen isotopic composition. 677 Individual datapoints from this study are shown as diamonds (, n = 31). The observed range for 678 fermentative (green), hydrogenotrophic (blue) and geological (red) methane sources were taken 679 from (90), though we note that these boundaries are not absolute (e.g. (36)). headspace hydrogen concentrations. This is also apparent in histograms C and E which show 731 that the distribution of isotope fractionation is the same (p > 0.5) for cultures whose headspace 732 hydrogen concentration at harvest was between 10 and 20% (green) or 20 and 50% H 2 (blue). 733 734