Quantification of biological nitrogen fixation by Mo-independent complementary nitrogenases in environmental samples with low nitrogen fixation activity

Biological nitrogen fixation (BNF) by canonical molybdenum and complementary vanadium and iron-only nitrogenase isoforms is the primary natural source of newly fixed nitrogen. Understanding controls on global nitrogen cycling requires knowledge of the isoform responsible for environmental BNF. The isotopic acetylene reduction assay (ISARA), which measures carbon stable isotope (13C/12C) fractionation between ethylene and acetylene in acetylene reduction assays, is one of the few methods that can quantify isoform-specific BNF fluxes. Application of classical ISARA has been challenging because environmental BNF activity is often too low to generate sufficient ethylene for isotopic analyses. Here we describe a high sensitivity method to measure ethylene δ13C by in-line coupling of ethylene preconcentration to gas chromatography-combustion-isotope ratio mass spectrometry (EPCon-GC-C-IRMS). Ethylene requirements in samples with 10% v/v acetylene are reduced from > 500 to ~ 20 ppmv (~ 2 ppmv with prior offline acetylene removal). To increase robustness by reducing calibration error, single nitrogenase-isoform Azotobacter vinelandii mutants and environmental sample assays rely on a common acetylene source for ethylene production. Application of the Low BNF activity ISARA (LISARA) method to low nitrogen-fixing activity soils, leaf litter, decayed wood, cryptogams, and termites indicates complementary BNF in most sample types, calling for additional studies of isoform-specific BNF.

. Spread Table S5.  13 C values (‰) used for background and Rayleigh corrections Data S1. Excel spreadsheet containing raw and processed  13 C data and templates for data analyses Methods S1. Description of the EPCon-GC-C-IRMS system Samples and standards are loaded into sealed, Helium (He, Airgas, catalog no. He UHP300) flushed, 20 mL autosampler vials to reach an ethylene concentration of 2.5 nmol C in each vial. Sample vials are loaded into a Leap Technologies CTC GC-PAL autosampler with a custom-built tray to accommodate 120, 20 mL vials. During analysis, the headspace of one vial is fully displaced into the EPCon by a stream of He (Airgas, catalog no. He UHP300). He is used as the carrier gas throughout the system and the flow rate decreases across each valve between the autosampler and the GC-C-IRMS (Thermo Scientific Trace GC Ultra-Isolink connected to a Thermo Scientific Delta V Plus Isotope Ratio Mass Spectrometer operated in continuous flow mode with a Conflo IV).
Using He, the sample is carried through a series of traps to remove moisture from the sample stream; a magnesium perchlorate (Mg(ClO4)2) chemical trap, Nafion™ tubing, and a cooling bath (stainless steel loop submersed in ethanol bath held at -70° C, Trap 1; Fig. 1b). The gas analytes (including ethylene and acetylene) are then trapped in a stainless-steel loop submersed in liquid nitrogen (-195° C, Trap 2) while gasses with a freezing point lower than liquid N2 (e.g., CH4) are vented. Analytes are then carried from Trap 2 through a 2 m GC column (GC 1; Supelco 1/8" x 2.1 mm stainless steel custom packed with 80/100 HayeSep N resin) held at 80° C, where ethylene and acetylene are separated. Ethylene elutes from GC 1 first and is collected in Trap 3 (1/16" diameter Silcosteel® tubing submersed in liquid N2). The timing of Valve 4 is adjusted to ensure complete recovery of the ethylene from GC 1 and the remaining acetylene is removed through a vent. The sample ethylene is then released from the trap (once it lifts out of liquid nitrogen) and is introduced into the GC-C-IRMS through a silica capillary that is connected directly to the front end of the GC column (Agilent HP-PLOT/Q capillary GC column [30 m, i.d. = 0.32 mm, f.t. = 20 μm]) in the Isolink. From here the sample follows the same flow path as with the classical direct injection method used in Zhang et al. (2016).
Detailed information on instrument settings (e.g. flowrates, timing, etc.) and optimized sample loading order can be found in Tables S1 and S2.

Source acetylene for acetylene reduction assays
The relatively high background of ethylene (~20 ppmv) in commercially available acetylene tanks precluded the use of tank acetylene in acetylene reduction assays (ARA) of low BNF activity samples, which typically yield < 100 ppmv ethylene in ARAs. To obtain a higher purity source of acetylene for ARAs, we generated acetylene by reaction of calcium carbide (CaC2, Sigma Aldrich, piece thickness < 10 mm, typically technical grade, ~80%, part no. 270296-500G) with Milli-Q® water. Reactions were performed in a fume hood in an evacuated 5 L gas tight Tedlar bag with dual Luer Lock valves (10-15 g CaC2, 50 mL water, 10-Calibrated Instruments Cali-5-bond gas tight bag [GSB-P/5] with two stopcock bases [STOPBASE-F]), in an evacuated 1.6 L Tedlar bag (5 g CaC2, 25 mL distilled and deionized water, Chemware® Tedlar® PVF Gas sampling bag, manufacturer part no. D1075018-10) , or in an evacuated 160-240 mL serum bottle sealed with a 20 mm blue butyl stopper (0.8-4 g CaC2 with 330-440 L water). Background ethylene from carbide-generated acetylene was ~1-3 ppmv as measured by GC-FID.

Ethylene concentration analyses in ARAs
Headspace ethylene concentrations in ARAs were measured by GC-FID on a Shimadzu GC-8A with a flame ionization detector (FID) and fitted with a Supelco 1/8" x 2.0 m long stainless-steel GC column custom packed with 80/100 HayeSep N resin. GC oven temperature was held at 80°C.

Azotobacter culturing
Azotobacter vinelandii mutants using Mo-only nitrogenase (strain CA70.1 1 ) or V-only nitrogenase (strain CA11.70 2 ) for nitrogen fixation were grown aerobically in a modified nitrogen fixing Burks medium 3 The pH of the medium was adjusted to 6.8 using NaOH and filter-sterilized with a 0.2 m bottle top filter prior to aseptic dispensation into pre-sterilized Nunc growth flasks (Nunc EasyFlask). Bacteria were cultured in 50 mL Nunc flasks with 75% headspace for aeration of ambient air through filter paper in the caps, at 30°C in a shaking incubator (200-250 rpm shaking, orbital platform shaker New Brunswick EXCELLA E24R). Growth phase was monitored by optical density at 620nm using spectrophotometry (Spectronic 20 Gensys Visible Spectrophotometer). Strains were revived from freezer stock on modified Burks solid medium with ammonium acetate (10 mM) then sequentially transitioned into diazotrophy through growth in modified Burks liquid medium with ammonium acetate at 10 mM (1 transfer) and at 0 mM added (2 transfers). Diazotrophy in the last transfer in medium with no added ammonium was confirmed by GC-FID.

Azotobacter calibration ARAs
In an optimal experiment using the direct scaling method with Azotobacter ARAs to quantify complementary nitrogenase contribution to AR as %VNase ( Figure 2, method 1), nine 30 mL ARA vials (3 sets of triplicates) are prepared with 10% v/v acetylene (in air) using the same stock of acetylene as for environmental samples. Two sets of triplicates are used for the conversion into ethylene via ARA using the Mo-only and V-only Azotobacter mutants, and one set is saved to measure  13 C of background ethylene (following acetylene removal with chemical precipitation). Triplicate  13 Cethylene measurements of each Azotobacter mutant are measured on the EPCon-GC-C-IRMS for each batch of acetylene generated to be used as end-member values for the %VNase scale which is applied to all samples processed with that batch of acetylene.

Methods S3. Additional environmental sample information
Laboratory maintenance of Zootermopsis termites purchased from Ward Scientific was accomplished using a habitat consisting of a plastic box with a lid that was punctured and covered in screen to allow sufficient air flow. The box was kept in a foil lined drawer to keep it dark. To maintain humidity, a cup with saturated KCl was placed next to the enclosure and all materials were wetted with MiliQ water 1-2 times per week. Enclosure materials included an ~1 cm thick layer of vermiculite. The termites were fed with small chunks of degraded wood (autoclaved), Whatman filter paper, and paper towels. Termite ARAs were done using 0.4-2.7 g (fresh weight) of termites (~20-100 individuals). Additional information on collection and ARAs on natural surface samples collected from Northeastern US forests and termites can be found in Supplementary Table S3.

Methods S4. Correction for background ethylene carried over in acetylene generated from calcium carbide
Eqn. S1  13 Cbackground EY is the 13 C value of the background ethylene in 'pure' acetylene. ntotal is the measured ethylene concentration taken during ARA (includes the concentration of the sample and the background ethylene). nsample is the measured ethylene concentration of the sample minus nbackground EY, the concentration of background ethylene due to acetylene addition. If background ethylene concentration in source acetylene was not measured, an estimate of 2 ppmv was assumed. In cases where there was not enough pure acetylene remaining in a batch to measure the  13 Cbackground EY, the average of all measured acetylene batches was used (8.4 ± 1.9‰, n = 8). The calculation of %VNase contribution to AR from  13 Cacetylene and  13 Cethylene (Eqn. S4b) followed the 13 AR method of Zhang et al., 2016 5 (Eqn. S5a). If the particular batch of acetylene used in ARAs could not be directly measured, we relied on a long-term estimate of  13 Cacetylene (14.92‰) for acetylene repeatedly generated from calcium carbide over the course of eight months by three different researchers (  CEst.source acetylene in Eqn. S5b, see Fig. S1), which has an associated uncertainty of ~1‰. Measurement of  13 Cacetylene was achieved by direct injection of samples into the GC-C-IRMS. Values of 13 Mo and 13 V used for calculations Eqn.

Methods S5. Calculations of %VNase in samples
S5b were obtained as in Method 2 (i.e. using Eqns. S4e-S4d in which  13 C for Azotobacter MoNase and VNase ARAs derive from the same IRMS runs as samples or are averages of nitrogenase specific ARAs across multiple organisms, ) × 100

Methods S6. Replicate and Uncertainty information for environmental sample analyses
Summary statistics for calculation of %VNase for environmental samples with n being the number of incubation replicates for each sample type (Fig. 3 Typical uncertainty (expressed as s.d.) for %VNase values for each scaling method (methods 1 -3 in Fig. 2) was calculated by compounding analytical or estimated longterm uncertainties of each individual term from the relevant equation in Figure 2. Similarly, the uncertainty associated with %FeNase was estimated according to the following equation (%FeNase = %VNase x ( 13 Mo-V)/( 13 Mo-Fe)) using data from Supplementary Table S4, and is found to be similar to %VNase (5-20%). All error calculations can be found in Supplementary Data S1 (Tab Error_Calculations). The highest error term in the calculation stems from VNase and FeNase uncertainty. At low values (0-30%VNase or FeNase), typical uncertainty is 6%, 11%, and 14% for methods 1, 2, 3 respectively. At high VNase and FeNase values (70-100%), uncertainty is 16%, 18%, and 20% for the three methods, respectively.
Given the estimated uncertainties, we conclude that samples with >120% in the %FeNase scale or 160% in the %VNase scale must be influenced by some sort of ethylene cycling (production or consumption), or by a leak in incubation or storage container. Figure S1. Spread of  13 Cacetylene values of different batches of acetylene. Each batch of acetylene was generated from calcium carbide, made by three different researchers, and analyzed over eight different days.  13 Cacetylene values are reported relative to lab CO2. Error bars are +/s.d. within the same day of analysis. Several data points for the same batch number represent analyses conducted over multiple days.  To do this, we took the difference between the measured value of each drift correction anchor and the long-term average of the EY-4 tank ( 13 CEY-4, long-term avg = 10.10‰ ± 0.3‰) then used a two-point regression between each pair of anchors to get a correction factor that was applied to each value measured between the pair of drift correction anchors.     Sample ID   20  Sample  Sample ID   21  Sample  Sample ID   22  Sample  Sample ID   23  Sample  Sample ID   24  Sample  Sample ID   25  Drift correction standard  Ethylene tank   26  Sample  Sample ID   27  Sample  Sample ID   28  Sample  Sample ID   29  Sample  Sample ID   30  Sample  Sample ID   31  Sample  Sample ID   32  Drift Correction  Ethylene tank   33  Sample  Sample ID   34  Sample  Sample ID   35  Sample  Sample ID   36  Sample  Sample ID   37  Sample  Sample ID   38  Sample  Sample ID  39  Drift correction standard  Ethylene tank   40  QC standard  Sample ID   41  QC standard  Sample ID   42  Sample  Sample ID   43  Sample  Sample ID   44  Sample  Sample ID   45  Sample  Sample ID Fig. S1). + (n) = the number of replicate analyses ^Azotobacter culturing and ARA information are in the Methods section of the main text and are further described in Supplementary Methods S2. Rhodopseudomonas palustris single nitrogenase strains Mo-nitrogenase only CGA753, V-nitrogenase only CGA766, and Fe-nitrogenase only CGA755 were grown in batch culture under anaerobic photo-heterotrophic conditions in defined nitrogen-fixing medium with added Mo or V as needed 5 . Anabaena variabilis (wild type ATCC 29413) and strain MZ49 (molybdate uptake modBC mutant) were grown in BG-11 medium containing added Mo or V 5 . ARAs for R. palustris and A. variabilis used acetylene from calcium carbide in sealed containers filled at most 30% by volume with exponential culture and background headspace of N2 (R. palustris) or air (A. variabilis).