Lake Ørn is a shallow mesotrophic Lake with iron-rich sediment located in Jutland, Denmark. It has an area of 0.43 km², an average water depth of 4 m and a maximum water depth of 10.5 m (Skovgaard, 2004). The stream Funder Å supplies the lake with a high input of iron in form of ochre (∼45 g Fe m^-2 year^-1 from 1989 to 2002; Skovgaard, 2004). Stratification is usually restricted to water depths greater than ∼5 m due to a high water flow and wind exposure.

Sediment sampling took place in April 2012 at the 4.5 m deep site near the center of the lake, for which the biogeochemistry was described previously (Norði et al., 2013; Weber et al., 2016). The water temperature was 8°C and the water column was oxygenated. Using a hand-operated Kajak sampler, three sediment cores (length = 30 cm, inner diameter = 7.5 cm) were retrieved for depth profiles of sulfate, methane, density/porosity and for RNA extraction. We dissected the cores obtained for depth profile characterization in 2 cm depth intervals under constant nitrogen flow within 2 days after sampling and subsampled for the different parameters as described below. To minimize a possible change of the biogeochemical depth profiles during storage, we air-bubbled the individual cores ∼2 cm above the sediment surface while keeping all cores in a tank filled with surface water from Lake Ørn.

Additionally, 30 cores were taken for RNA stable isotope probing incubation experiments and further treated as described in the following section.

RNA-SIP experiments were carried out in both core and slurry incubations. Both slurries and cores were subjected to five different treatments: addition of (1) ¹³C-bicarbonate (¹³C_DIC), (2) ¹²C-bicarbonate (¹²C_DIC), (3) ¹³C-methane (¹³C_CH4), (4) ¹²C-methane (¹²C_CH4), and (5) no addition as control. For core incubations, 24 cores were transferred immediately after sampling to a 5°C tempered water tank (total volume ∼0.2 m³), which was filled with surface water from Lake Ørn. All cores were kept at these conditions for the entire incubation time and were individually and continuously oxygenated by gentle air bubbling ∼1 cm above the sediment. The core liners contained holes (diameter = 0.2 cm), which were billowy winding around the liners (horizontal distance = 1 cm, vertical distance = 2 cm) and were double sealed with aquarium silicon (Dana Lim 579, ISO 11600) and poly-coated cloth tape (Tesa). ¹³C_CH4 and ¹²C_CH4 (both 99 atom-%), ¹³C_DIC and ¹²C_DIC (both 99 atom-%) were injected anoxically into the cores in 1 cm steps to 18 cm depth for methane and 25 cm depth for sodium bicarbonate, by means of a gas-tight syringe. Methane was added in increasing amounts with increasing sediment depth at days 1, 4, 7, and 15 corresponding to final added concentrations of ∼40 μmol L^-1 (0–5 cm); ∼80 μmol L^-1 (5–10 cm), ∼120 μmol L^-1 (10–15 cm), and ∼260 μmol L^-1 (15–20 cm), which corresponded to ∼50% of the initial methane concentrations and up to ∼80% of ¹³CH₄ at the last time point. ¹³C_DIC and ¹²C_DIC was supplied from an anoxic stock solution (1 mol L^-1) at days 1 and 14 to added concentrations of ∼1.4 mmol L^-1 (0–5 cm), ∼1.7 mmol L^-1 (5–15 cm), and ∼2 mmol L^-1 (15–25 cm). δ¹³C_DIC measurements confirmed a range of ∼7 to ∼25 atom-% ¹³C_DIC in the DIC pool of the entire depth, whereas in the layer 10–15 cm, ¹³C_DIC was in a range of ∼13 to 21 atom-%.

The cores were incubated for 30 days in total and sampled after 3, 6, 14, and 30 days. For sampling, one control core, one ¹²C core, and one ¹³C core were sliced under constant nitrogen flow into 5 cm intervals to a sampling depth of 25 cm. At each time point, subsamples were taken to determine porosity, concentrations of sulfate and methane, to quantify rates of microbial sulfate reduction (SR) and anaerobic oxidation of methane (AOM), and for RNA extraction for RNA-SIP analysis.

For slurry SIP incubations, sediment from the depth interval 10–15 cm was extracted from six sediment cores under constant nitrogen flow, transferred into a glass bottle (Schott) and sealed with a butyl rubber stopper. The bottle was transferred into a nitrogen-filled glove bag and sediment was mixed 1:1 with N₂-purged medium (50 μmol L^-1 Na₂SO₄ and 3 mmol L^-1 NaHCO₃). Slurry was then distributed into 20 serum bottles (60 mL) that were closed with butyl rubber stoppers (kept under helium atmosphere prior to usage; De Brabandere et al., 2012) without headspace. By means of a gas-tight syringe, ¹²C_CH4 was added at day 0 to all bottles through the butyl rubber stopper to a final concentration of 120 μmol L^-1, simulating concentrations similar to the core incubations. One-third of the incubations received additional 60 μmol L^{-1 12}C_CH4 and one-third received additional 60 μmol L^{-1 13}C_CH4, which raised the total methane concentration to 180 μmol L^-1 in these incubations. In the remaining 1/3 of the slurries, no further methane was added as they were used as controls. At days 3, 6, and 27 additional 60 μmol L^{-1 12}C_CH4 or ¹³C_CH4 was added to the ¹²C_CH4 and ¹³C_CH4 incubations, respectively. For DIC slurry experiments, 1.3 mmol L^{-1 12}C_DIC or ¹³C_DIC was added to ¹²C_DIC and ¹³C_DIC incubations, respectively, at days 1 and 27. δ¹³C_DIC measurements confirmed a range of ∼15 to ∼25 ¹³C_DIC atom-% in the DIC pool. All bottles were stored in darkness at 5°C and carefully inverted every few days. In total we had 20 slurries, which were successively sampled after 3, 6, 27, and 72 days. For sampling, one bottle from each treatment was opened in a nitrogen-flushed glove bag and sampled for measurements of the same biogeochemical, molecular and rate measurements as described for the cores.

All cores and slurries were processed under constant nitrogen flow and at 4°C and subsamples were taken either by cut-off syringes or a spatula.

For porewater analysis, 15 mL centrifugation vials were filled completely, capped, centrifuged at 4°C at 4000 g for 10 min (Centrifuge 5810; Eppendorf) and the supernatant was filtered anoxically (glove bag filled with nitrogen) through a syringe filter (pore size = 0.22 μm; syringe, needle and filter flushed and stored in helium prior to usage). For analysis of sulfate, 1 ml porewater each was mixed with 20 μL 20% zinc acetate (ZnAc) and stored at -20°C. Sulfate concentrations were quantified via ion chromatography (Dionex ICS-1500, detector: DS6 Heated Conductivity Cell), using ultrapure water (“type 1”) containing 4.5 mmol L^-1 Na₂CO₃ and 1.4 mmol L^-1 NaHCO₃, as eluent. The detection limit for sulfate was 1 μmol L^-1. For DIC concentration and δ¹³C measurements, porewater was stored in 1.8 ml septum vials on 10 μL saturated mercury chloride (HgCl₂) at 4°C. For analysis 300 μL sample was transferred into a 12 mL acid washed and helium-flushed Exetainer (Labco, High Wycombe, UK) and 50 μL of 85% phosphoric acid was added to drive CO₂ into the headspace over an equilibration phase of 15 h. The ¹³C atom-% values were determined by IRMS, using the standards NBS 18-Calcite and IAEA-LSVEC Lithium Carbonate and an in-house bicarbonate standard (average standard deviation between samples = 0.1%).

To determine methane concentrations, 2 cm³ of sediment or slurry was subsampled directly into a 20 mL serum vial, prefilled with 5 mL 2.5% sodium hydroxide (NaOH) that was closed and shaken immediately and stored upside down for at least 24 h to drive the methane into the headspace. Methane concentrations were quantified by injecting 50–100 μL of the headspace into a gas chromatograph with an FID detector (Perkin Elmer). The porosity/density of sediment or slurry was determined by subsampling 1 cm³ sediment or slurry with a cut-off pipette tip into a pre-weighted centrifugation vial and the water content/porosity from the weight loss from after drying at 90°C for 48 h.

For SR and AOM rates, 5 cm³ sediment or slurry was filled into glass tubes in triplicates or duplicates, respectively, closed with plunger and septum (stored under helium 6 weeks prior to sampling) and pre-incubated for 1 h before injection of radiotracers (Jørgensen, 1978).

For SR rates, 20 μL carrier-free ³⁵SO₄^2- tracer (100 kBq) sample^-1 was injected through the septum, the tubes inverted and the samples incubated at 5°C in darkness for 24 h. The incubations were stopped by transferring the samples into 10 mL of 20% ZnAc. Controls (“killed controls”) were taken in parallel, in which sediment/slurry was transferred into 10 mL of 20% ZnAc after 24 h and thereafter tracer was added. All samples and controls were frozen immediately and stored at -20°C until analysis. After thawing, supernatants were separated from solid phases by centrifuging the samples at 500 g for 5 min and decanting. Supernatants were subsampled for determination of ³⁵S in sulfate and reduced sulfur compounds were extracted via the cold chromium distillation for ³⁵S determination (Kallmeyer et al., 2004; Røy et al., 2014).

For AOM rates, ¹⁴CH₄ (dissolved in water; 20 μl; ∼0.5 kBq; supplied from American Radiolabeled Chemicals, cleaned with hopcalite and sodium hydroxide) was injected through the septum into the sediment or slurry, the tubes inverted and samples incubated at 5°C in darkness for 20 h. In order to stop incubations the samples were transferred into glass vials prefilled with 15 ml 2.5% (w/v) sodium hydroxide and immediately closed with butyl rubber stoppers, shaken and stored upside down. Methane concentrations were determined as described above and activities of ¹⁴CH₄ and ¹⁴CO₂ were quantified (Treude et al., 2005), using phenylethylamine and sodium hydroxide (1:1) as CO₂ trap. Duplicates of “killed controls” were taken with each time point, in which tracer was injected after termination of microbial activity.

Radioactivity was measured on a QuantaSmart-4.00 Scintillation counter (count time: 10 min; coincidence time: 18 nsec; delay before burst: 75 nsec). The detection limits for the SR and AOM rate measurements were determined by subtracting the mean plus three times the standard deviation of the killed controls from the product pools, ³⁵S sulfides and ¹⁴CO₂, respectively.

For each time point and incubation treatment, ∼5 cm³ of sediment or slurry was frozen immediately after sampling in liquid nitrogen and stored at -80°C until further processing. RNA was extracted using the PowerSoil Total RNA Isolation Kit (MoBio) and further purified using the AllPrep DNA/RNA Mini Kit (Quiagen). RNA concentrations were quantified with the Quant-iT-RiboGreen RNA Assay Kit (Life Technologies).

Caesium trifluoroacetate density gradient ultracentrifugation was applied as described previously (Whiteley et al., 2007). Briefly, extracted and purified RNA (∼600 ng in 24 μL nuclease-free ultrapure water) was mixed with a carefully adjusted density (1.8 ± 0.04 g mL^-1) gradient medium containing CsTFA (4.64 mL sample^-1), nuclease-free ultrapure water (0.936 mL sample^-1) and formamide (0.198 mL sample^-1). From this mixture, 5.1 mL was transferred into ultracentrifugation tubes (Quick-Seal, 5.2 mL, 13 × 51 mm; polyallomer, Beckmann Coulter) and closed by heat sealing. The tubes were centrifuged for 72 h (Beckmann Ultracentrifuge Optima XPN 80k RPM; Vertical rotor VTi 65,2) at 38400 rpm, 20°C with maximum acceleration and deceleration. After ultracentrifugation, the gradient was fractionated into 20 fractions by driving the gradient out at the bottom of the tube at a controlled constant flow (syringe pump; 750 μL min^-1) of nuclease-free ultrapure water into the top of the tube. From each fraction, 60 μL was used for density measurement (Refractometer DR 301-95, Krüss) and weight determination (microbalance Satorius Extend) and the rest was precipitated with two volumes of ice-cold isopropanol (molecular range, >99%) and 4 μL glycogen (5 mg mL^-1) for at least 30 min at -20°C. After precipitation, the fractions were centrifuged for 30 min at 4°C (139000 rpm) and the supernatant decanted. The residual pellet was washed with 500 μL 70% ethanol (molecular range, >99%) and centrifuged again as before. The supernatant was removed and the pellet resuspended in 30 μL nuclease-free ultrapure water. The RNA concentrations of all fractions were determined using the Quant-iT-RiboGreen RNA Assay Kit (Life Technologies).

For determining the ¹³C-enriched fractions in the CsTFA sample gradients, we aligned a ¹³C-enriched Escherichia coli gradient (“¹³C_acetate E. coli standard”, Supplementary Figure S4) and compared the locations of the ¹³C-peaks. To obtain the ¹³C_acetate E. coli standard, we incubated E. coli over 4 days in minimal medium (pH = 7) containing sodium hydrogen phosphate (Na₂HPO₄, 6 g L^-1), potassium dihydrogen phosphate (KH₂PO₄, 3 g L^-1), ammonium chloride (NH₄Cl, 1 g L^-1), sodium chloride (NaCL, 0.5 g L^-1), magnesium sulfate (MgSO₄, 0.12 g L^-1), calcium chloride (CaCl₂, 0.01 g L^-1) and ¹³C-sodium acetate (¹³CH₃¹³CO₂Na, 99 atom-%, 0.25 g L^-1) as sole carbon source. After incubation, RNA was extracted and purified, CsTFA density gradient centrifugation was carried out and RNA concentrations were quantified for 20 fractions as described for the sediment/slurry samples.

RNA of selected fractions was transcribed into complementary DNA (cDNA) using the Sensiscript RT Kit (Quiagen) with the primer U1492R (5′–GGYTACCTTGTTACGACTT–3′), following the manufacturer’s instructions.

For PCR amplification, for each sample a 50 μL PCR reaction mix containing ∼10 ng template cDNA, 20 pmol of each primer (A958R: 5′–YCCGGCGTTGAMTCCAATT–3′ and A20F 5′–TCYGGTTGATCCTGCCRG–3′), 1.25 U Taq polymerase (Fermentas), 40 nmol deoxynucleotides and 125 nmol MgCl₂ in 1x Taq reaction buffer (Fermentas) was prepared on ice and cycled in the PCR cycler with the following program: 2 min at 95°C, followed by 39 cycles of 1 min at 95°C, 2 min at 55°C, 4 min at 72°C, 4 min and finally 10 min at 72°C.

For T-RFLP analyses, cDNA was amplified by PCR as described above with the primer pair A958R and A20F-FAM (5′–TCYGGTTGATCCTGCCRG–3′). Thereafter, 10 μL of each amplicon DNA sample was digested with 5 U BsuRI (HaeIII) restriction enzyme (Fermentas) at 37°C for 8 h and thereafter purified with GeneJet PCR purification Kit (Fermentas Life Technologies). Terminal restriction fragment length polymorphism (T-RFLP) analysis was performed with an Applied Biosystems ABI3730XL sequencer (capillary electrophoresis) at Uppsala Genome Center (Sweden).

16S rRNA clone libraries were carried out for one representative gradient fraction of the ¹³C and ¹²C peak, each from the successful ¹³C_CH4 and ¹³C_DIC incubation and from the T₀ slurry and T₁ core control. Hereby, 7 μL cDNA amplicon of each sample was blunted and ligated with a pJET 1.2/blunt Cloning vector using the CloneJET PCR Cloning Kit (Thermo-Scientific) and transformed into chemically competent E. coli cells (One Shot TOP 10F′ from Invitrogen) according to the manufacturer’s instructions. The cells were spread on pre-warmed LB plates containing ampicillin and incubated overnight at 37°C. Thereafter, clones were screened using a colony PCR with primers PJF (5′–CGACTCACTATAGGGAGAGCGGC–3′) and pJR (5′–AAGAACATCGATTTTCCATGGCAG–3′). The archaeal clones were sent to Macrogen Europe (Netherlands) for sequencing.

Accurate T-RF sizes were determined using Peak ScannerTM (Applied Biosystems) on the obtained T-RFLP data from Uppsala Genome Center. Thereafter, the online source T-RFLP analysis Expedited¹ was used for noise filtration (noise filtration factor 1.0; peak area) and T-RF binning. All OTUs, which had a relative abundance < 1% were removed manually afterwards. Sequences obtained from the 16S rRNA libraries were scanned using the program “4 peaks” and low quality ends, vector and primer sequences were removed manually. Chimeras were detected by the online tool Bellerophon (Huber et al., 2004), evaluated manually and removed. Sequences, which passed the quality control, were imported into the SSU Reference 115 SILVA database for phylogenetic reconstructions (Pruesse et al., 2007) using the ARB software package (Ludwig et al., 2004). All 208 sequences have been deposited at GenBank under the accession numbers KX463063 – KX463269 (Supplementary Table S3).

The anoxic sediment column contained sulfate until a depth of ∼15 cm with a maximum concentration of ∼200 μmol L^-1 at the surface (Figure 1B), similar as previously reported (Norði et al., 2013; Weber et al., 2016). The methane content increased with depth and reached concentrations of ∼900 μmol L^-1 (Figure 1A). AOM activity was indicated at 5–15 cm depth by the concave shape of the methane profile, and thereby occurred within the depth interval of sulfate consumption and iron reduction, as indicated by the distributions of sulfate and soluble Fe²⁺ (Figure 1), and consistent with previous direct rate measurements using radiotracer incubations and with analyses of particulate iron speciation (Norði et al., 2013; Weber et al., 2016).

Sediment and slurries were incubated with ¹³C_CH4 or ¹³C_DIC, their ¹²C analogs or without addition (controls) for maximum 30 and 72 days, respectively. As described in more detail below, ¹³C-labeling of RNA was achieved in the ¹³C_CH4 core and ¹³C_DIC slurry incubations. We therefore focus on the biogeochemical results from these incubations here (Figures 2, 3), while the results of the ¹³C_DIC core and ¹³C_CH4 slurry incubations are displayed in Supplementary Figures S1, S2. At incubation start, sulfate concentrations at 10–15 cm were higher in the slurries (∼37 μmol L^-1) than in the cores (10–15 μmol L^-1), due to the addition of sulfate-containing medium (Figures 2B, 3B and Supplementary Figures S1B, S2B). All slurry incubations showed a decrease in sulfate concentrations in all treatments with a mean rate of ∼0.3 ± 0.2 μmol L^-1 d^-1. In the core incubations, sulfate concentrations remained at 10–15 μmol L^-1 or tended to decrease but only dropped below detection at the last time point in the ¹²C_DIC and ¹³C_DIC incubations. Methane concentrations increased in the control slurry incubation with ∼4 ± 3 μmol L^-1 d^-1, indicating net methanogenesis. A similar rate of increase in the control core incubation together with the decrease in sulfate indicated an upward shift of the redox zonation during the incubation while the increase in methane in the ¹²C_CH4 and ¹³C_CH4 cores and slurries reflected the repeated injection of methane to these. In the ¹²C_DIC- and ¹³C_DIC-amended core and slurry incubations, methane concentrations did not show a clear trend over time. Radiotracer incubations in control and ¹³C_CH4 incubations confirmed SR and AOM activity in cores and slurries at all time points (Figures 2D,E, 3D,E and Supplementary Figures S2D,E; see also additional discussion in Supplementary). On average, SR showed rates of 11 ± 3 μmol L^-1 d^-1 and 12 ± 5 μmol L^-1 d^-1 in control and ¹³C_CH4-amended cores, respectively, and average AOM rates of 3 ± 1 μmol L^-1 d^-1 in both (Figures 2D,E). In slurries, SR and AOM rates were similar to the rates in the core incubations, with an average SR rate of 9 ± 4 μmol L^-1 d^-1 in both control and ¹³C_CH4-amended slurries and AOM rates of 5 ± 3 and 4 ± 2 μmol L^-1 d^-1 in control and ¹³C_CH4-amended slurries, respectively (Figures 3D,E).

As the addition of ¹³C_DIC, ¹²C_DIC, ¹³C_CH4 or ¹²C_CH4 did not impact the biogeochemistry of the incubations nor SR and AOM rates substantially, we rule out a major impact on the microbial community due to amendments in comparison to the control incubations. Therefore, we interpret the control, ¹²C_CH4 and ¹³C_CH4 core incubations and the control, ¹²C_DIC, ¹³C_DIC slurry incubations, respectively, as biogeochemical replicates.

The initial composition of the active part of the archaeal community was assessed by two 16S rRNA clone libraries originating from the slurry and from a core before ¹³C-addition. In total, 76 clones were retrieved, from which ∼40 % were assigned to the phylum Euryarchaeota, ∼8% to Miscellaneous Euryarchaeotic Group and Aenigmarchaeota and ∼52% to the phylum Crenarchaeota (Table 1 and Supplementary Table S1). Within the Euryarchaeota, the two largest groups were Thermoplasmatales with 19 clones and the Methanosarcinales cluster AMNE-2d with nine clones, whereas Methanosaeta showed minor contributions. Almost all sequences that fell into the phylum Crenarchaeota were assigned to Miscellaneous Crenarchaeotic Group.

Sufficient amounts of ¹³C-labeled RNA for clear ¹³C- and ¹²C-fraction separation during CsTFA centrifugation and subsequent reverse transcription were obtained after 30 days in the ¹³C_CH4 core incubation and after 73 days in the ¹³C_DIC slurry incubation, while no RNA was detected in, and no cDNA could be obtained by reverse transcription of the high density gradient fractions sampled earlier during the incubations. The cDNA generated from the ¹²C and ¹³C fractions of the density gradient was amplified and resulted in archaeal products that were used to construct 16S rRNA clone libraries and perform T-RFLP analysis.

The archaeal 16S rRNA clone library constructed from the ¹³C fraction of the ¹³C_CH4 incubation was strongly enriched in sequences affiliated closely to ANME-2d, which made up 82% of the total (22 of 27; Figure 4, Table 1 and Supplementary Table S1). Of the remainder, three sequences were closely related to the aceticlastic methanogen Methanosaeta concilii that also falls into Methanosarcinales (Barber et al., 2011), while one was affiliated to Marine Benthic Group B in the phylum Crenarchaeota and was most closely related to clones from other lakes including AOM-active sediment from Lake Cadagno in Switzerland (Schubert et al., 2011). Most ANME-2d clones fell within a single OTU (Supplementary Table S1) and all were also closely related to AAA sequences retrieved from Lake Cadagno sediment (Schubert et al., 2011), sequences retrieved from a freshwater gas source (Timmers et al., 2016) and from the AOM zone of Antarctic cold seep sediments (Niemann et al., 2009). Furthermore, Lake Ørn ANME-2d sequences clustered with sequences derived from denitrifying methane-oxidizing bioreactors (Raghoebarsing et al., 2006; Hu et al., 2009) and with the freshwater nitrate-reducing anaerobic methane oxidizer ‘Candidatus Methanoperedens nitroreducens’ (Haroon et al., 2013).

A 16S rRNA clone library from the ¹³C fraction of the ¹³C_DIC slurry incubation yielded 29 sequences of which ∼83% fell into the phylum Euryarchaeota and ∼17% fell into the phylum Crenarchaeota (Figure 4, Table 1 and Supplementary Table S1). The majority of the euryarchaeotic sequences again fell into Methanosarcinales and therein mainly into the group ANME-2d, together with the ANME-2d sequences retrieved from the ¹³C peak of the ¹³C_CH4 incubation, as described above. A smaller amount of sequences from the ¹³C peak with ¹³C_DIC fell into the genus Methanosaeta and clustered together with the sequences from heavy fraction of the ¹³C_CH4 incubation and the aceticlastic methanogen Methanosaeta concilii (Barber et al., 2011). Furthermore, a few sequences of the ¹³C fraction of the ¹³C_DIC slurry incubation affiliated with Thermoplasmatales and the methanogenic genus Methanolinea. All ¹³C-labeled crenarchaeotic sequences from the ¹³C_DIC incubation fell into Miscellaneous Crenarchaeotic Group.

Comparison of 16S rRNA from the ¹³C peak fractions to that from the ¹²C peak of the ¹³C_CH4 and ¹³C_DIC incubations further confirmed the selective ¹³C labeling of the cluster ANME-2d. The ¹²C-peak clone libraries contained 40 and 36 clones from the ¹³C_CH4 and ¹³C_DIC incubation, respectively. Similar to the initial communities, about half of the total sequences clustered in the Miscellaneous Crenarchaeotic Group. In total ∼20% of the sequences clustered in Euryarchaeota and were assigned to Thermoplasmatales, except one sequence that was assigned to Methanosaeta and one sequence that was assigned to ANME-2d in the ¹³C_CH4 incubation. The rest of the sequences were assigned to Miscellaneous Euryarchaeotic Group and Aenigmarchaeota.

Transcription, PCR amplification, and terminal restriction fragment length polymorphism analyses were carried out for the same samples that were used to construct 16S rRNA clone libraries with the aim to obtain a representation of the entire archaeal community (Supplementary Figure S3). Overall, a large part of the archaeal community could be taxonomically identified based on the predicted T-RF length both for incubation start and after incubation with the ¹³C-labeled substrates. At incubation start, 69 and 77% of the OTUs in core and slurry, respectively, could be aligned with the OTUs obtained from the 16S rRNA clone libraries. In the ¹³C_CH4 core incubation, ∼74% of the OTUs retrieved from the ¹³C peak and in the ¹³C_DIC slurry incubation ∼69% of the OTUs retrieved from the ¹³C peak could be taxonomically identified. The Supplementary section provides a detailed discussion about the T-RFLP data (Supplementary Table S2) and an overview over the archaea identified in Lake Ørn sediment in the present study.

Anaerobic consumption of methane in Lake Ørn sediment was indicated in the depth interval of 5–15 cm by the concave increase of methane concentrations with depth and confirmed by radiotracer incubations and stable isotope analyses in previous reports (Norði et al., 2013; Weber et al., 2016). Nitrate can be excluded as electron acceptor for AOM as it was depleted within the upper 2 cm of the sediment (Norði et al., 2013). As potential electron acceptors for AOM in Lake Ørn, both sulfate and poorly crystalline ferric iron were previously discussed (Norði et al., 2013; Weber et al., 2016). However, elevated sulfate reduction rates in the AOM zone and a methane-dependent alteration of δ³⁴S signatures in slurry incubations directly implicated sulfate as electron acceptor for AOM while a coupling to iron might be indirect via an iron(III)-dependent cryptic sulfur cycle (Weber et al., 2016).

We chose to focus on the depth interval with the highest in situ AOM activities (10–15 cm, Figures 2E, 3E; see also Norði et al., 2013; Weber et al., 2016) in both cores and slurries with the aim of exploiting the different advantages of each of these two incubation types. Generally, in core incubations, the microbial communities are maintained in the natural biogeochemical gradients and largely protected from mechanical disturbances. However, the quantification of biogeochemical processes in a defined interval is rather difficult as porewater and gaseous parameters are strongly impacted by diffusion along concentration gradients, and core-to-core variability adds noise to the temporal changes observed. Higher controllability can be achieved in slurry incubations where the addition of medium can also support the enrichment of a preferred part of the microbial community. Disadvantages of slurring are physical disturbances of the microbial community and the blockage of diffusion of products and substrates such as, e.g., sulfate due to the closed system.

Anaerobic oxidation of methane and SR were active in all incubations and their rates as obtained from radiotracer incubations were comparable between cores and slurries (Figures 2D,E, 3D,E). The net consumption of sulfate in all slurry incubations and an overall tendency of sulfate to decrease in the core incubations further confirmed microbial reduction of sulfate. Still, radiotracer-based rates exceeded the rate of sulfate depletion substantially, indicating active reoxidation of sulfide as previously discussed for this site (Norði et al., 2013; Weber et al., 2016). Similarly, a net accumulation of methane occurred in most incubations and suggested a co-occurrence of AOM and methanogenesis in the layer of sulfate depletion. Based on the similarity to the previous analyses of AOM in Lake Ørn we conclude that AOM in our incubations was likely coupled to sulfate reduction, at least in part, but also that a direct coupling to iron reduction cannot be excluded.

In the sediment core incubations, ¹³C was assimilated into 16S rRNA after ¹³C_CH4 addition, giving rise to a high-density peak in the density gradient, while radiotracer incubations confirmed AOM activity. Far most of the ¹³C-labeled sequences fell into the euryarchaeotic order Methanosarcinales and therein grouped tightly in the cluster ANME-2d (Figure 4 and Table 1). Sequences falling into ANME-2d were also part of the active initial archaeal community, 12% of clones, whereas strains from other ANME groups were not detected. The presence of ANME-2d in the 16S rRNA clone libraries of the untreated sediment and the strong enrichment of this group in the high-density fraction obtained from ¹³C labeling, indicated that ANME-2d carry out AOM in Lake Ørn sediment. This was further supported by the lack of detectable RNA in the high-density fractions of the parallel incubation without ¹³C source (Lueders et al., 2004).

Members of this cluster were previously observed in AOM-active habitats, but due to lacking evidence of their methane-oxidizing capabilities, they were first grouped with uncultured methanogens in a clade named GOM Arc 1 (Lloyd et al., 2006). Subsequently, the identification of 16S rRNA sequences associated with nitrate-dependent AOM archaea in a bioreactor enrichment (Raghoebarsing et al., 2006) resulted in identification of the cluster named AOM-Associated Archaea (AAA) as a subgroup of GOM ARC 1 (Knittel and Boetius, 2009). The group AAA was renamed ANME-2d after the genetic reconstruction of ‘Cand. M. nitroreducens’, with the capability to reduce nitrate while oxidizing methane via the reverse methanogenesis pathway (Haroon et al., 2013).

In Lake Ørn, nitrate can be excluded as electron acceptor for AOM (Norði et al., 2013), suggesting that the detected ANME-2d strains most likely coupled methane oxidation to sulfate and/or iron reduction. Ferric iron-dependent AOM was recently reported in an enriched ANME-2d/AAA strain (Ettwig et al., 2016). However, a coupling of ANME-2d to sulfur cycling cannot be conclusively excluded as this group was also detected in a range of marine sulfur-cycling environments including, e.g., the Logatchev hydrothermal vent system (Voordeckers et al., 2008), deep-sea marine sediments (e.g., Inagaki et al., 2006), and Antarctic cold seep sediments (Niemann et al., 2009). Further ANME-2d sequences were reported from freshwater systems characterized by active sulfur-cycling at lower sulfate concentrations as, e.g., in sediment of the alpine Lake Cadagno (originally described as AAA; Schubert et al., 2011), in freshwater wells (Flynn et al., 2013), and in freshwater gas sources (Timmers et al., 2016).

Although we cannot link the members of ANME-2d identified in Lake Ørn to either sulfate or ferric iron reduction with certainty, the present study is, to our knowledge, the first to experimentally associate ANME-2d activity to AOM in a natural environment by following the incorporation of ¹³C_CH4 into biomass, and to demonstrate that this activity is possible in the absence of nitrate.

In the ¹³C_CH4 incubation, a small number of sequences clustering with Methanosaeta, Marine Benthic Group, and Miscellaneous Crenarchaeotic Group were detected in the high-density fraction (Figure 4, Table 1 and Supplementary Table S1). As these groups were not clearly enriched in the high-density fraction relative to the original community, they may represent the carryover of small amounts of unlabeled RNA to the ¹³C peak, which is typical for the method and possibly due to interaction with ¹³C-labeled RNA (Lueders et al., 2004). The same consideration applies to the minor groups detected in the heavy fraction of the ¹³C_DIC incubation (Figure 4, Table 1 and Supplementary Table S1). Like previous RNA-SIP studies (Hori et al., 2007; Gutierrez et al., 2015), we therefore exclude these minor groups from both treatments from further discussion.

Reverse methanogenesis as the pathway for anaerobic methane oxidation in ANME strains including ANME-2d is known on a detailed enzymatic level (Hallam et al., 2004; Haroon et al., 2013; Arshad et al., 2015). Furthermore, as shown in the present study, members of the ANME-2d cluster can assimilate CH₄ and incorporate it into biomass. However, from our ¹³C_DIC slurry, we obtained ¹³C-enriched 16S rRNA sequences affiliated to ANME-2d and therein clustering with the sequences obtained in the ¹³C_CH4 core incubation. This finding suggests that ANME-2d strains are either capable of a mixed assimilation of CH₄ and DIC and/or that cross-feeding of ¹³C-labeled metabolites could potentially have occurred.

Cross-feeding of labeled substrates is a general risk in stable isotope probing incubations (Dumont and Murrell, 2005 and references therein). In the present study, cross-feeding could potentially have occurred in two directions: (1) in the ¹³C_CH4 incubations, AOM activity could have lead to the build-up of ¹³C_DIC, which could have been assimilated by autotrophic microorganisms, e.g., by hydrogenotrophic methanogens, being falsely understood as methanotrophs. On the other hand, (2) in the ¹³C_DIC incubations, methanogens could have built up significant amounts of ¹³C_CH4 and ¹³C_CH4-assimilating anaerobic methane oxidizers could have been falsely interpreted as autotrophs. We exclude case (1) because the highest accumulation of ¹³C in the DIC pool achieved during the incubation would be insufficient to explain our observations. Thus, with an average AOM rate of 3 nmol cm^-3 day^-1 in the 10–15 cm layer of the ¹³C_CH4 incubated core (Figure 2), over the incubation period of 30 days, the amount of ¹³C_DIC generated by AOM would have reached only ∼3 atom-% of the total DIC pool (∼3 mmol L^-1, data not shown) at the end of incubation, while 20 atom-% labeling is generally needed to shift a RNA sequence from the ¹²C to the ¹³C fraction in the CsTFA gradient (Manefield et al., 2011). A similar estimate can be performed to evaluate (2) cross-feeding in the ¹³C_DIC incubations: as there was no significant net change in methane concentrations in the ¹³C_DIC slurry incubations, one can assume that methanogenic activity may have approximately equalled the activity of anaerobic methanotrophs. As a consequence, over an incubation time of 72 days, with an average rate of 5 nmol cm^-3 day^-1 and the DIC pool being ¹³C enriched by ∼25%, methanogens could have built up >100 μmol L^-1 of ¹³C_CH4 which would have been sufficient for cross-feeding under the prevailing methane conditions. Therefore, it appears evident that ANME-2d in Lake Ørn directly assimilated ¹³C_CH4, but the apparent ¹³C_DIC assimilation by the same strain may be a result of ¹³CH₄ incorporation after methanogenesis from ¹³C_DIC.

Mixed assimilation of CH₄ and CO₂ has been reported for ANME-2a and -2c strains retrieved from marine environments, but with a preferred usage of CO₂ over CH₄ for biomass production (Wegener et al., 2008; Milucka et al., 2012). Thus, ANME-2 strains of a long-term enrichment culture incorporated both CO₂ and CH₄ into their biomass, but with a 25-times higher rate for CO₂ in comparison to CH₄ (Milucka et al., 2012). A complete decoupling of methane oxidation from the assimilatory system and autotrophy instead of mixotrophy was proposed for a group of ANME-1 strains from hydrothermal sediments in Guaymas Basin (Kellermann et al., 2012). It remains unclear if the preference of the carbon source for assimilation is strain-specific or/and impacted by environmental conditions. Our observation of incorporation of ¹³C into RNA of ANME-2d in both ¹³C_CH4 and ¹³C_DIC incubations suggests that carbon flow in the ANME-2d cells may differ from that in the previously studied ANME-1, -2a, and -2c strains. However, the difference in labeling pattern between cores and slurries might further indicate a metabolic shift of the AOM-mediating community caused by the different types of incubation, as the ¹³C sources were added in comparable amounts in slurry and core incubations and samples were treated consistently. The previous studies were all carried out in slurry incubations or enrichment cultures (Wegener et al., 2008; Kellermann et al., 2012; Milucka et al., 2012). Their results were in agreement with the results of our slurry incubations that only showed ¹³C-enriched RNA with ¹³C_DIC-amendment and therefore also indicated preferred assimilation of CO₂. We conclude that, members of ANME-2d may have the potential for mixed assimilation of CH₄ and CO₂, but further investigations are required to identify if such a metabolism would be strain-specific or a physiological response to changing biogeochemical environments.