Modification of methane oxidation pathways 1 during long-term incubations of methanic lake sediments

Anaerobic oxidation of methane (AOM) is one of the major processes limiting the release of the 12 greenhouse gas methane from natural environments. In Lake Kinneret sediments, iron-coupled AOM 13 (Fe-AOM) was suggested to play a substantial role (10-15% relative to methanogenesis) in the 14 methanic zone (>20 cm sediment depth), based on geochemical profiles and experiments on fresh 15 sediments. Apparently, the oxidation of methane is mediated by a combination of mcr gene bearing 16 archaea and aerobic bacterial methanotrophs. Here we aimed to investigate the survival of this 17 complex microbial interplay under controlled conditions. We followed the AOM process during long- 18 term (~ 18 months) anaerobic slurry experiments of these methanic sediments with two stages of 19 incubations and additions of 13 C-labeled methane, multiple electron acceptors and inhibitors. After 20 these incubation stages carbon isotope measurements in the dissolved inorganic pool still showed 21 considerable AOM (3-8% relative to methanogenesis). Specific lipid carbon isotope measurements 22 and metagenomic analyses indicate that after the prolonged incubation aerobic methanotrophic 23 bacteria were no longer involved in the oxidation process, whereas mcr gene bearing archaea were 24 most likely responsible for oxidizing the methane. Humic substances and iron oxides are likely 25 electron acceptors to support this oxidation, whereas sulfate, manganese, nitrate, and nitrite did not 26 support the AOM in these methanic sediments. Our results suggest in the natural lake sediments 27 methanotrophic bacteria are responsible for part of the methane oxidation by the reduction of 28 combined micro levels of oxygen and iron oxides in a cryptic cycle, while the rest of the methane is 29 converted by reverse methanogenesis. After long-term incubation, the latter prevails without bacterial 30 methanotropic activity and with a different iron reduction pathway. (DIC). several and different The results of these experiments were compared to batch and semi-bioreactor experiments that were set up 90 with freshly collected to follow the changes in methane oxidation pathways along the incubation period. We also calculated methane oxidation and production rates of representative pre- incubated long-term slurry experiments. Alongside the 13 C-labeled DIC measurements, we investigated the structure of the microbial population using metagenomics and lipid biomarkers to identify the potential microbial players and their over various incubation

alphaproteobacterial methanotroph, to reduce iron by methane in these unique conditions. 83 Here, we explored the role of methanotrophic activity in natural methanic lake sediments, its survival 84 outside of the natural conditions during long-term anaerobic incubations, and whether there is a shift 85 in the potential electron acceptors. To answer these questions, we diluted fresh methanic sediments 86 from Lake Kinneret with porewater from the same depth twice and amended the sediment with 13 C-87 labeled methane to follow its oxidation to dissolved inorganic carbon (DIC). These incubations were 88 then also amended with several types of potential electron acceptors and different inhibitors. The 89 results of these experiments were compared to batch and semi-bioreactor experiments that were set up 90 with freshly collected sediments to follow the changes in methane oxidation pathways along the 91 incubation period. We also calculated methane oxidation and production rates of representative pre-92 incubated long-term slurry experiments. Alongside the 13 C-labeled DIC measurements, we 93 investigated the structure of the microbial population using metagenomics and lipid biomarkers to 94 identify the potential microbial players and their dynamics over various incubation periods. 95

Study site 97
Lake Kinneret is a warm monomictic freshwater lake, located in the North of Israel. Its maximum 98 depth is ~42 m and the average depth is 24 m. The lake is thermally stratified from March until 99 December, with the hypolimnion turning anoxic starting from April. The sediment is composed 100 156 repetitive results (no activity) for numerous previous experiments. For the humic substrate experiment 157 we received natural humic substance extracted from a lake by a colleague in the University of Alaska, 158 Fairbanks. One experiment was set up without any additional electron acceptor in order to assess the 159 rate of methanogenesis in the pre-incubated slurries. 160

Semi-bioreactor experiment 161
Two semi-bioreactors ( Fig. 1) were set up with fresh sediments from the methanic zone (25 -40 cm) 162 of Lake Kinneret central station (Station A) immediately after their collection. Both reactors were 163 filled headspace-free with a slurry of a 1:4 sediment -pore water ratio. One of the bioreactors was 164 amended with 10 mM hematite. To dissolve 13 C-labeled methane in the porewater, 15 ml of 165 headspace was produced with only methane gas for 24 hours. The reactors were shaken repeatedly 166 during those hours. After 24 hours, the gas was replaced with anoxic pore-water, so that there was no 167 head-space at all. The oxidation-reduction potential was monitored continuously by a redox electrode 168 (Metrohm, Herisau, Switzerland) throughout the incubation period to verify anoxic conditions and to 169 know the redox state of the slurry in the reactor. The bioreactors were subsampled weekly to bi-170 weekly, and the sample volume (5-10 ml) was replaced immediately by preconditioned anoxic 171 (flushed with N2 gas for 15 minutes before the exchange) porewater from the methanic zone. Samples 172 were analyzed for dissolved Fe(II), CH4 and δ 13 CDIC as outlined below. Additional subsamples for 173 metagenome analysis and lipid analysis were taken at the beginning of the experiment and on day 174 151, and day 382 respectively. The purpose of the semi-bioreactors was to set up an experiment that 175 can monitor the redox state regularly, to have a closer to natural conditions, and to have another 176 indication for the processes involving methane in freshly collected sediments.

Porewater analyses 190
About 0.3 ml of filtered (0.22 um) pore-water was injected to 12 ml glass vial with He atmosphere 191 and 10 µl of H3PO4 85% to acidify all the DIC species to CO2 (g). The autosampler takes a gas sample 192 from the vials and measures the δ 13 CDIC of the sample on the GasBench interface of a DeltaV 193 Advantage Thermo Scientific isotope-ratio mass-spectrometer (IRMS) at a precision of ±0.1 ‰. It 194 should be noted that to measure δ 13 CCH4 the gas sample must be combusted before this procedure, 195 which means that the δ 13 C measured is of the DIC only. Results are reported versus the Vienna Pee 196 Figure 1: Flow diagram of experimental design. Three types of experiments were set up from sediments of the deep methanic zone (below 25 cm): A. Pre-incubated slurry experiments. Fresh sediments collected from the lake were incubated in a 1:1 ratio with porewater extracted from the same depth for 3 months. Then the slurry was divided to the experiments bottles and diluted again with fresh porewater to reach 1:3 ratio. 10 experiments were set up this way, 8 of them with different electron acceptors for 6-18 months, and 3 with different inhibitors for 12-18 months (to one experiment both electron acceptor and an inhibitor were added). B. Semibioreactor experiment. Fresh sediments collected from the lake were inserted to two bioreactors and diluted with fresh porewater to reach 1:4 ratio. The bioreactors were set up with no headspace. One of the bioreactors was amended with iron oxide (hematite). C. Slurry experiment with freshly collected sediments. The sediments were diluted with porewater to reach 1:3 ratio, and was amended with different iron oxides (Bar-Or et al., 2017). The experiment was set up for 17 months. the ferrozine method (Stookey, 1970) by a spectrophotometer at 562 nm wavelength with 198 a detection limit of 1 µmol L −1 . A gas sample was taken from the experiment bottle's 199 headspace by a gas-tight syringe and was analyzed for methane and ethylene concentrations by a 200 focus gas chromatograph (GC) equipped with a flame ionization detector (FID) with a detection limit 201 of 50 µmol L -1 . Methanogenesis rate was derived from temporal changes in methane concentration in 202 a representative pre-incubated slurry experiment (Fig. S2). The amount of methane oxidized was 203 calculated by a simple mass balance calculation according to Eq. 1 and 2: 204 Where x is the mixing fraction of two sources which compose the final DIC; the initial DIC pool and 207 the oxidized 13 C-CH4. The letter x denotes the fraction of oxidized 13 C-CH4, while 1-x denotes the 208 fraction of the initial DIC pool out of the final DIC pool. F 13 CH4 is the fraction of 13 C out of the total 209 CH4 at t0, FDI 13 Ci is the fraction of 13 C out of the total DIC at t0, and FDI 13 Cf is the fraction of 13 C out 210 of the total DIC at t-final.
[CH4]ox is the amount (concentration in pore water) of the methane oxidized 211 throughout the full incubation period, and [DIC]f is the DIC concentration at t-final. We assumed that 212 the isotopic composition of the labeled CH4 did not change significantly throughout the incubation 213 period. 214

Lipid analyses 215
A sub-set of samples was investigated for the assimilation of 13 C-labeled methane into polar lipid-216 derived fatty acids (PLFAs) and intact ether lipid-derived hydrocarbons. A total lipid extract (TLE) 217 was obtained using a modified Bligh and Dyer protocol (Sturt et al., 2004). PLFAs in the TLE were 218 Reported fatty acid isotope data are corrected for the introduction of the methyl group during 229 (1) (2) derivatization by mass balance calculation similar to eq. 1 using the measured δ 13 C value of each 230 FAME and the known isotopic composition of methanol as input parameters. 231

248
In this study we followed the progress of the methane oxidation process in long-term incubations from 249 Lake Kinneret methanic sediments. This is by quantifying the modifications between experiments 250 conducted on fresh sediments from the methanic zone (batch slurries presented by Bar-Or et al.

Geochemical trends 254
In the pre-incubated long-term experiments, similarly to the fresh incubations, there was a conversion 255 The geochemical experiments tested the potential of several electron acceptors to perform and 260 stimulate this considerable AOM process. It should be noted that the actual involvement of sulfur 261 cycling can be quantified directly by inhibiting this cycle, while the rest can be tested for their 262 potential involvement by their addition to the slurries. First, metal oxides were added. The addition of 263 hematite as an electron acceptor did not change the δ 13 CDIC increase with time (the slope) compared to 264 the methane-only controls (Fig. 2). This is in contrast with the freshly collected sediment experiments, 265 where this addition stimulated the conversion of 13 C-methane to 13 C-DIC and thus the AOM (Fig. 2). The actual involvement of sulfate was quantified directly by the addition of Na-molybdate, an 274 inhibitor of sulfate reduction and sulfur disproportionation, to the methane-only controls and to 275 slurries amended with magnetite ( Fig. 3A). This addition did not change the slope of the δ 13 CDIC 276 increase with time, clearly indicating no AOM inhibition and no role for sulfate in the AOM process.  We also amended long-term pre-incubated slurries with potential organic electron acceptors. No 289 13 CDIC enrichment was observed with the addition of AQDS (an analog for humic substrate) to slurries 290 with and without hematite (Fig. 3E). Similar trends were observed in δ 13 CDIC following the addition of 291 PCA, an analog for methanophenazines that are found in some archaeal membranes and shuttle 292 electrons (Wang and Newman, 2008) (Fig. 3F). We further tested the effect of naturally occurring 293 humic substances, using those isolated from a different natural lake. The results show that in the 294 beginning the δ 13 CDIC values did not change (Fig. 3F), while a steep increase in their Fe(II) 295 concentrations was observed (Fig. S3). However, after 20 days, the δ 13 CDIC values of these slurries 296 started to increase dramatically with a steep slope, indicating high AOM activity (Fig. 3F). We also 297 tested the addition of black coffee, as another example of a complex natural organic substance. In this 298 incubation, again, the δ 13 CDIC values decreased during the first 20 days, but then increased very 299 steeply (from 102‰ to 596‰). In those additions there is in general a mirrored trend of the dissolved 300 Geochemical analysis of δ 13 CDIC was performed also on two experiments that tested the effect of 307 inhibitors on methane metabolism. In one experiment, BES, a specific inhibitor for methanogens and 308 ANME's mcrA genes, was added, and in another experiment, acetylene, a non-specific inhibitor for 309 methanogens, was added. Both cases showed a complete inhibition of labeled 13 C-DIC production 310 following the addition, similarly to the killed control (Fig. 4). Acetylene can also inhibit nitrogen 311 cycling in some cases, however ethylene is produced then (Oremland and Capone, 1988). In our case 312 no ethylene was detected, supporting the inhibition only of methanogens' activity.  Table S2). Each data point is the average of duplicate samples that were taken from each bottle; the error bars are smaller than the symbol.

Metagenomic and lipid analyses 316
The metagenomic analysis points to the potential involvement of several archaea and bacteria in the 317 AOM observed in the pre-incubated slurries. Bona fide ANME (ANME-1), as well as various 318 methanogens and high abundance of Bathyarchaeia were present in all the samples (Table S3). 319 Known sulfate reducing bacteria, including Desulfobacterota, Desulfuromonadota and 320 Thermodesulfovibrionia, but not seep sulfate reducing bacteria, were found, and some in large read 321 abundances (Table S3). Only very few metagenomic reads mapped to Methylomirabilaceae (NC-10) 322 (<1%) and no reads mapped to Methanoperedens. The number of metagenomic reads mapped to 323 functional genes narH and narG, which encode subunits of the respiratory nitrate reductase in 324 Methanoperedens decreased with time in the pre-incubated sediments (Table S4). Very few reads 325 mapped to the nirS gene, which encodes the nitrite reductase, and its coverage did not increase over 326 time (Table S4). 327 The δ 13 C values of the archaeol-derived isoprenoid phytane showed 13 C-enrichment (between 15-27‰ 328 enrichment), and no 13 C-enrichment in the killed control, indicative of methane assimilation by 329 archaea. This signal was also found for acyclic biphytane but less pronounced (between 10 -5 ‰ 330 enrichment) (Table 1).

338
Our many porewater profiles of Lake Kinneret indicate that microbial sulfate reduction dominates the 339 anoxic hypolimnion and the surface sediments, while methanogenesis is confined to the sediments 340 The first noticeable observation from the current pre-incubated long-term slurries data is that the 349 δ 13 CDIC values of the natural amendments (only with the addition 13 C-labeled methane) increased 350 dramatically. This indicates a clear AOM signal, even after the long-term incubations and the low 351 abundance of the microbial populations. Below, we characterize this AOM process. 352

Potential electron acceptors for AOM in the long-term pre-incubated experiments 353
The pre-incubated long-term incubations data show a sharp increase in the δ 13 CDIC values of both 354 natural and hematite amendments. However, as opposed to the freshly collected sediment 355 experiments, there was no difference between the addition of hematite as the electron acceptor and the 356 natural (methane-only) amendment. This means that hematite does not have a potential to stimulate 357 the AOM activity or that there is enough natural Fe(III) in the sediments to sustain the maximum 358 potential of Fe-AOM. 359 Following this observation, we quantified the effect of other metal oxides, such as magnetite, 360 amorphous iron, and manganese oxide (Fig. 3A and B), which are present in Lake Kinneret sediments 361 AOM. However, the addition of the clay minerals appears again to encourage only organoclastic iron 371 reduction (Fig. 3F, Fig. S3). Like iron oxides, manganese oxide, did not support AOM and likely 372 encouraged organoclastic manganese reduction. Given that manganese oxides are found in very low 373 abundance in Lake Kinneret sediments (0.1 %, Table S1), their potential role in metal-AOM is likely 374 low anyway. concentrations appeared to inhibit AOM, potentially facilitating denitrification (Fig. 3D). shuttling ability and encouraging organoclastic oxidation that adds light carbon isotope (as opposed to 400 the labeled 13 C-methane) and lowers the δ 13 CDIC values without AOM at all, or by masking its signal 401 (Fig. 3E). Similar trends were observed in δ 13 CDIC following the addition of PCA, a synthetic analog 402 for methanophenazines (Fig. 3F). Yet, the addition of natural humic acids or black coffee exhibited 403 different behavior. At first, the natural humic substances promoted organoclastic iron reduction, 404 probably by shuttling electrons from organic compounds other than methane to natural iron oxides in 405 the sediments (Figs. 3F, S3). Then, perhaps when the availability of the iron oxides or the organic 406 matter decreased, humic substances were used as terminal electron acceptors for AOM, as was 407 suggested by Valenzuela et al. (2017). In that study, AOM was coupled to the reduction of humic 408 substances in the presence of inorganic electron acceptors simultaneously with methanogenesis. 409 Overall, our experiments with different electron acceptors indicate clearly that sulfate is not involved 410 in the AOM process in Lake Kinneret methanic sediments, and that nitrate, nitrite and Mn-oxides are 411 less likely. The potential electron acceptors are natural humic substrates with or without iron minerals 412 that are abundant in the sediment and preferably react with methane rather than with other organics. 413

Main microbial players in the long-term pre-incubated experiments 414
As mentioned above, the pre-incubated long-term incubations data show a sharp increase in the 415 δ 13 CDIC values of natural amendments. However, the addition of BES, a specific inhibitor for 416 methanogens and ANME's mcrA genes, stopped immediately the AOM, similarly to the killed bottles, 417 and to the fresh sediment experiments (Bar-Or et a., 2017), indicating methane oxidation by 418 methanogens or ANMEs in all stages of incubations (Fig. 3). In addition, the complete inhibition of 419 labeled DIC production following the addition of acetylene (Fig. 4) suggests the involvement of 420 methane metabolizing microorganisms, also evidenced by the enrichment in δ 13 C values of phytane 421 and biphytane (Table 1). Such a signal is generally indicative of active archaea, i.e. methanogens or 422 ANMEs in this case, which assimilate 13 C-carbon from an unknown intermediate or existing DIC. 423 The essential role of methanogens or ANMEs in the AOM in all stages of incubations suggest that this were found. It should be noted that ANME-1 was found in very low abundance (< 1.5 %) and other 428 might be involved in methane metabolism (Evens et al., 2015), the mcrA genes were not found in their 430 Lake Kinneret MAGs, thus their role in AOM is questionable. 431 On the other hand, both the metagenomic and lipid isotopic analysis suggest that the role of aerobic 432 type I methanotrophs (of the class gammaproteobacteria) in methane turnover in the long-term 433 incubations is negligible (Table S3). This contrasts with the natural sediments and fresh incubations 434 that show their presence in the sediments and their important role in oxidizing the methane. 435

Methane oxidation pathway in the long-term incubations 436
Our results indicate net methanogenesis in long-term incubations with an average rate of 2 µM day -1 437 ( Fig. S2 and Table S5), similarly to the fresh incubation experiments (Bar-Or et al., 2017). This is 438 even with the overall development of increasing δ 13 CDIC values resulting from potential methane 439 turnover ( Fig. 2 and 3). A likely explanation for this signal is an interplay between methane 440 production and oxidation, with the latter triggered by reverse methanogenesis, which is demonstrated  (Table S6). Therefore, under natural conditions, methanogens alone are unlikely to 469 produce our observed considerable amounts of DIC just by back flux. To conclude, trace levels of oxygen may fuel aerobic methane oxidizers in a cryptic cycle between 489 oxygen and iron in the natural lake methanic sediments, and they are responsible for part of the 490 methane oxidation and maybe the iron reduction. The rest of the methane is oxidized to DIC by 491 methanogens or ANME-1. The DIC production from methane turnover in the long-term experiments 492 is performed only by methanogens or ANME-1. It seems less likely that this is by back flux alone, but 493 rather by active metabolic AOM by reverse methanogenesis and an external electron acceptor.