Continuous measurements of nitrous oxide isotopomers during incubation experiments

. Nitrous oxide ( N 2 O ) is an important and strong greenhouse gas in the atmosphere and part of a feed-back loop with climate. N 2 O is produced by microbes during nitriﬁcation and denitriﬁcation in terrestrial and aquatic ecosystems. The main sinks for N 2 O are turnover by denitriﬁcation and photolysis and photo-oxidation in the stratosphere. The position of the isotope 15 N in the linear N=N=O molecule can be distinguished between the central or terminal position (isotopomers of N 2 O ). It has been demonstrated that nitrifying and denitrifying microbes have a different relative preference for the terminal 5 and central position. Therefore, measurements of the site preference in N 2 O can be used to determine the source of N 2 O i.e. nitriﬁcation or denitriﬁcation. Recent instrument development allows for continuous (on the order of days) position dependent δ 15 N measurements at N 2 O concentrations relevant for studies of atmospheric chemistry. We present results from continuous incubation experiments with denitrifying bacteria, Pseudomonas ﬂuorescens (producing and reducing N 2 O ) and Pseudomonas chlororaphis (only producing N 2 O ). The continuous position dependent measurements reveal the transient pattern ( KNO 3 to 10 N 2 O and N 2 , respectively), which can be compared to previous reported site preference (SP) values. We ﬁnd bulk isotope effects of -5.5 ‰ ± 0.9 for P. chlororaphis . For P. ﬂuorescens , the bulk isotope effect during production of N 2 O is -50.4 ‰ ± 9.3 and 8.5 ‰ ± 3.7 during N 2 O reduction. The values for P. ﬂuorescens are in line with earlier ﬁndings, whereas the values for P. chlororaphis are larger than previously published δ 15 N bulk measurements from production. The calculations of the SP isotope effect from the measurements of P. chlororaphis result in values of -6.6 ‰ ± 1.8. For P. ﬂuorescens , the calculations 15 results in SP values of -5.7 ‰ ± 5.6 during production of N 2 O and 2.3 ‰ ± 3.2 during reduction of N 2 O . In summary, we implemented continuous measurements of N 2 O isotopomers during incubation of denitrifying bacteria and believe that similar experiments will lead to a better understanding of denitrifying bacteria and N 2 O turnover in soils and sediments and ultimately hands-on knowledge on the biotic mechanisms behind greenhouse gas exchange of the Globe.


Introduction
The atmospheric concentration of nitrous oxide (N 2 O) has increased from approximately 271 ppb before the industrialization to 324 ppb in 2011 [(Ciais et al., 2013)]. This increase has resulted in (1) an enhanced radiative forcing, e.g. N 2 O has the third 5 highest contribution to the radiative forcing of the naturally occurring greenhouse gasses [(Hartmann et al., 2013)], and (2) an increased production of nitrogen oxides (NOx) in the stratosphere and thereby an increased ozone-depletion. [(Forster et al., 2007); (Kim and Craig, 1993)] Ice core records show that concentrations of N 2 O correlates with northern hemispheric temperature variations, e.g. during the last glacial-interglacial termination as well as over the rapid climate variations occurring during the glacial period, known as 10 Dansgaard-Oeschger events (D-O events). However, occasionally (e.g. D-O event 15 and 17) the N 2 O concentration increases long before the onset of the dramatic temperature change Schilt et al. (2010), providing a potential early warning for rapid climate change. Isotopomers of N 2 O provide information on the sources [ (Clark, 1999)] and may improve our understanding on why N 2 O is leading over some rapid climate change events. The stable isotopes of nitrogen are 14 N and 15 N with average isotopic abundances (mole-fraction) in the atmosphere of 0.99634 and 0.00366, respectively [(Junk and Svec, 1958)]. 15 The N 2 O molecule has an asymmetric linear structure (N=N=O) where the position of the 15 N can be discriminated. The isotopomers are named 15 N α and 15 N β or short α and β for 14 N 15 N 16 O and 15 N 14 N 16 O, respectively [(Yoshida and Toyoda, 2000)]. The two isotopomers cannot be distinguished directly by isotope ratio spectrometry, as they have the same mass. However, a distinction is possible using mid-infrared spectroscopy because the rotational and vibrational conditions are different for the two isotopomers providing regions where absorptions of the two isotopomers do not overlap. For isotopomer measure-20 ments at low N 2 O concentration (low ppm range), a joint instrument development was executed, applying cavity ring down spectroscopy (CRDS) to enable continuous measurements (on the order of days) of the isotopomer abundances and yielding values for 15 N α and 15 N β .
For isotopomers, the isotopic composition is reported as delta values; the deviation of the elemental isotope ratio R from a standard (equation 1). The standard for nitrogen is today's atmospheric composition.
The N 2 O bulk isotopic composition calculates as the average of δ 15 N α and δ 15 N β while the site preference (SP) is defined as their difference (δ 15 N SP = δ 15 N αδ 15 N β ). [ (Brenninkmeijer and Röckmann, 1999); (Park et al., 2011);(Toyoda et al., 2002)]. 30 There are multiple natural and anthropogenic sources of N 2 O. The primary anthropogenic sources of N 2 O are fertilizers including nitrogen minerals used for agriculture. The natural sources are primarily nitrification and denitrification in terrestrial and aquatic ecosystems. [ (Mosier et al., 1998); (Olivier et al., 1998)].
Denitrification is a stepwise biological reduction process in which denitrifying bacteria produce nitrogen (N 2 ). Under anaerobic conditions the denitrifying bacteria use nitrate (NO − 3 ) instead of oxygen as an electron acceptor in the respiration of organic matter. Through multiple anaerobic reactions N 2 is produced as the end product of the complete denitrifying process (reaction R1) [ (Firestone and Davidson, 1989)].
Each of these anaerobic reactions is carried out by a genuine enzyme, i.e., the production of N 2 O is caused by the reaction between nitric oxide (NO) and the enzyme nitric oxide reductase (NOR). The NOR enzyme works as a catalyst in the reduction of NO as shown in equation R2. [(Wrage et al., 2001); (Tosha and Shiro, 2013)] Reactions with different enzymes typically result in specific isotopic fractionation. The isotope fractionation of the intermediately produced N 2 O during denitrification is a consequence of multiple reaction steps i.e., the isotope fractionation is determined as product-to-substrate fractionation. Two species of denitrifying bacteria with slightly different enzymes potentially leads to different fractionation. In this study, we compared the fractionation of N 2 O by two contrasting denitrifying bacteria; Pseudomonas fluorescens producing and reducing N 2 O, and Pseudomonas chlororaphis producing but not reducing 15 N 2 O. We hypothesized that these contrasting denitrifying bacteria show differences in isotope enrichment and SP during N 2 O production and reduction.

Method
Our objective was to perform continuous position dependent δ 15 N measurements of two different bacterial cultures during incubation experiments. Using two denitrifying bacterial cultures we determine the isotope enrichment and SP during production 20 and reduction of N 2 O, respectively.

Instrumentation
Bacterial production of N 2 O was continuously measured by mid-infrared cavity ringdown spectrometry using a prototype of the Picarro G5101-i analyzer (in the following named G5101i-CIC) (Picarro, Santa Clara, California, USA Measurements are made by placing the sample delivery system of the G5101i-CIC in a closed loop with a microbial incubation glass chamber (Fig. 1). Circulation is provided by a "leak-reduced" diaphragm pump installed downstream from the analyzer.  In addition to the measuring mode, the system can be flushed with N 2 (not shown in Fig. 1). The flushing mode is used to obtain an anaerobic starting point of the incubation experiment free of N 2 O. The flushing procedure is fully automated to ensure reproducibility. The entire incubation setup is flushed with N 2 for 310 seconds at a high flow rate. The resulting 10 overpressure in the incubator is released prior to switching back to the closed loop position.

Correction of CRDS concentration dependence
Isotope measurements made with the G5101i-CIC have a N 2 O concentration dependence and need to be corrected. surfaces, and whose resulting optical transmission or reflection is periodic in wavelength. The ripples are not always constant in phase, which means that the ripples can shift spectrally, which can cause the offset to drift over time. The result is a concentration dependent error to δ of the form ± 1/concentration. Because baseline ripple effects become more dominant as N 2 O concentration decreases, the "relative" error is largest at low concentrations. We chose to fit the raw data with a cubic spline smoothing function (CSS-function) [ (Brumback and Rice, 1998)]. The best 25 fit of these CSS-functions are found using a smoothing parameter of p = 0.999 in a regression analysis. Four outliers were identified to be outside the 2σ boundary and removed from the data set (the red circles in Fig. 2). After these outliers were removed the best fit was found again and the concentration dependent correction was applied as shown with the green profiles in Fig. 2. Over the course of the experiments, no further instrumental drift was observed.  (Table 1).

Pure bacterial cultures 15
The two bacterial cultures used in this study are both gram-negative bacteria with the capability to denitrify, i.e. reduce nitrate to gaseous nitrogen. Isolates were obtained from an agricultural soil of sandy loam type (Roskilde Experimental Station) on 11 April 1983. One culture is a Pseudomonas fluorescens, bio-type D that reduces NO − 3 all the way to N 2 . The second culture, Pseudomonas chlororaphis, is only capable of reducing NO − 3 to N 2 O [ (Christensen and Bonde, 1985)], which means that the nitrous oxide reductase is absent or at least not active in this organism. The latter bacterium is contained in the American Type 20 Culture Collection with accession number ATCC 43928 [ (Christensen and Tiedje, 1988)]. The cultures were grown anoxic in 50 ml serum bottles with 1/10 tryptic soy broth (Difco) added 0.1 g KNO 3 · L −1 . After six days of growth at room temperature The bacterial culture of P. Fluorescens was cultivated for six days at a slightly lower temperature (15 • C) to assure that the cultures were in a comparable phase of potential activity when assayed for gas production/reduction activity. The six days old cultures were used in the incubation experiment in which it is 25 conditional for the denitrifying process that organic carbon is available, that the concentration of oxygen is low and that the concentration of NO − 3 is high [ (Wrage et al., 2001); (Stuart Chapin III et al., 2002)].

Bacterial incubation experiments
50 mL bacterial solution of P. chlororaphis or P. fluorescens was placed in a petri-dish in the 1000 mL incubation chamber.
Hereafter, the setup was flushed with pure N 2 (purity 99.9999 %) to ensure anaerobic conditions. To ensure no N 2 O gas

Analysis of isotope enrichment
The observed isotope changes in N 2 O during our incubation experiment can be analyzed in terms of Rayleigh fractionation. 10 Rayleigh fractionation describes the changing isotopic composition in reactant and product of a unidirectional reaction. Equation 3 gives the isotope ratio of the reactant as: where R s,0 is the initial isotope ratio of the reactant, R s is the isotope ratio of the product at time t, α p/r is the fractionation factor of the product versus the substrate, and f is the unreacted fraction of substrate at time t. The isotope enrichment calculates as 15 = α -1 from the fractionation factors in equation 3. We do not measure the isotopic composition of KNO 3 in our experiments.
However, by definition the end values of N 2 O for P. chlororaphis when all KNO 3 has reacted has to be identical to the initial value of KNO 3 .
The Rayleigh type distillation is valid for the bulk isotope ratios of N 2 O. The corresponding equation for the accumulated product is: The isotope enrichment for the bulk can be described using the isotope enrichment of the product ([ (Menyailo and Hungate, 2006); (Ostrom et al., 2007); (Mariotti et al., 1981); (Lewicka-Szczebak et al., 2014)]). bulk p,acc = (α bulk − 1) The described Rayleigh equation is not directly applicable to the isotopomers, as these are both direct products of the same 25 denitrification process from the same batch of denitrifying bacteria and nitrate. An isotopomer correction factor derives to ϕ α and ϕ β for the two isotopomers respectively. The Rayleigh equation for the accumulated product of an isotopomer is therefore: Equation 8 is valid for P. chlororaphis. For P. fluorescens, both an immediate reduction and an uptake reduction take place simultaneously with N 2 O production due to the pre-experimental cultivation leading to activation of all enzymes in the bacterial solution. Part of the freshly produced N 2 O is therefore immediately reduced to N 2 . This reduction is fractionating with 5 fractionation factor α R . The isotope imprint of the reduction on the remaining N 2 O depends on the ratio between reduction and production rate (γ). Assuming γ is constant results in the following first order approximation.
For any calculated ratio the values are given in ‰ using the delta-notation (equation 1).
The SP of N 2 O, being the difference between the δ 15 N α and δ 15 N β , is not compatible to the Rayleigh equations. Applying the assumption that δ s,0 << 1000 ‰, the isotope enrichment simplifies to p/s ≈ δ p − δ s , such that e.g., α = (δ 15 N αδ 15 N N O3 ).

Application of Rayleigh model
We determine the respective isotope enrichment during production and reduction of N 2 O for each of the bacterial strains assuming a Rayleigh type process. The Rayleigh distillation model is fitted to the δ-value and the N 2 O concentration data. As P. chlororaphis is a pure producer of N 2 O this is straight forward. For P. fluorescens the section of production is defined as 20 being from the start of the measurements until the end of net production. From the calculations of the production rates (see Fig. 3) we believe that N 2 O production continues after the point of maximum concentration. Therefore the unreacted fraction at the end of net production is iteratively found and is > 0. Other parameters that are iteratively found are the fractionation factor between α = 0 and α = 1 and the reduction to production rate γ (between 0 and 1). The fractionation factors resulting in the highest R 2 values was picked as the correct fractionation factor for each specific evolution. Production of N 2 O by P. The models are fitted using both the CDC data and the 5 minutes running mean of the CDC data. The best fit is found using an iterative approach of the R 2 value between the measured data and the Rayleigh distillation equation for the accumulated product (equation 8 and equation 10 for P. chlororaphis and P. fluorescens, respectively). Iterative calculations are performed for fractionation factor between α = 0 and α = 1 during production of N 2 O, and between α = 1 and α = 2 during reduction of N 2 O. The reduction correction parameter (γ) is iteratively determined to be between γ = 0 and γ = 1. The fractionation factors resulting in the highest R 2 values are picked as the correct fractionation factor for each specific evolution.

3 Results
The evolution of N 2 O over time from the two bacterial strains shows two very distinctive patterns with both an increasing and decreasing N 2 O concentration characteristic for P. fluorescens and an increasing N 2 O concentration followed by a stabilization characteristic for P. chlororaphis (Fig. 3A) as has previously been described by Christensen and Tiedje (1988). These distinctive characteristics are only vaguely seen in the respective dynamics of the SP for the two bacterial strains (Fig. 4A, 4B, 5A and 10 5B).

Pseudomonas chlororaphis
Paralleling the increase in N 2 O concentration (Fig. 3A), we also find an increase in δ 15 N α , δ 15 N β and δ 15 N bulk over time ( Fig. 4A and 4B). The final product of P. chlororaphis is N 2 O; this is a unidirectional transfer of nitrogen from KNO 3 to N 2 O and thereby a Rayleigh process although multiple fractionations are involved. In Fig. 4A and 4B, we plot the best fit Rayleigh 15 profile for δ 15 N α and δ 15 N β respectively.
The modeled Rayleigh distillation profiles were found to match the production of N 2 O from P. chlororaphis to a relatively high degree. The average correlation coefficients (R 2 ) between data and fitted Rayleigh curves are 0.709, 0.654, and 0.767 for δ 15 N α , δ 15 N β and δ 15 N bulk respectively. The calculations of the isotope enrichment for the fractionation of δ 15 N α give a mean value of -8.8 ‰ ± 1.4. For δ 15 N β the mean enrichment factor was found to be -2.2 ‰ ± 1.1 (Table 4). These values 20 leads to a mean SP value of -6.6 ‰ ± 1.8 and a δ 15 N bulk enrichment factor of -5.5 ‰ ± 0.9.

Pseudomonas fluorescens
Continuous measurements of the evolution of N 2 O produced and consumed by the denitrifying bacteria P. fluorescens are presented in Fig. 5A and 5B for the δ 15 N α and δ 15 N β , respectively. The correlation coefficient of the fitted apparent Rayleigh model for the production matches the continuously measured δ 15 N α data by 94.1 % on average using the R 2 method for the 25 seven replicates of P. fluorescens incubations. Equivalent R 2 average for δ 15 N β are 88.7 %, whereas the average for δ 15 N bulk are found to be 94.8 %. The R 2 found for the reduction part for the two isotopomers and the bulk is 91.3 % for δ 15 N α , 76.3 % for δ 15 N β , and 91.7 % for δ 15 N bulk on average for the seven replicates. The fractionation during both the production and the reduction are therefore following the apparent Rayleigh type distillation to a large degree. The isotope enrichment calculated using these models are therefore a good representation for the fractionation caused by the P. fluorescens bacteria on the N 2 O.

30
The resulting isotope enrichment is presented in Table 2 for the production part and in Table 3 for the reduction part together 8 Biogeosciences Discuss., doi: 10.5194/bg-2016-258, 2016 Manuscript under review for journal Biogeosciences Published: 30 June 2016 c Author(s) 2016. CC-BY 3.0 License. with the calculated isotope enrichment for the SP. During production of N 2 O, the mean enrichment for SP was found to be SP = -5.7 ‰ ± 5.6 while the bulk = -50.4 ‰ ± 9.3 for the bulk, hence there is a difference of 44.9 ‰ and 0.8 ‰ for SP and bulk , respectively, to P. chlororaphis. During reduction of N 2 O the mean enrichment for SP was found to be SP = -2.3 ‰ ± 3.2 and bulk = 8.5 ‰ ± 3.7 for the bulk.

5
The two bacteria investigated are denitrifiers, i.e. functionally similar but with P. chlororaphis lacking the ability to reduce N 2 O to N 2 . Both denitrifiers were cultivated under anaerobic conditions leading to active nitric oxide reductase (both cultures) and nitrous oxide reductase (P. chlororaphis). This leads to a pre-experimental expectation that when fed the same amount of nitrate the maximum N 2 O concentration and the N 2 O production rate should be lower for P. fluorescens than for P. chlororaphis.
In Fig. 3, an example of the N 2 O evolution by the two bacterial strains is plotted as the concentration of N 2 O and the produc-10 tion rate versus time. Starting at time zero and moving with the profiles forward in time, the concentration of N 2 O produced by P. chlororaphis has a higher production rate and reaches a higher level in concentration than that produced by P. fluorescens even though P. fluorescens received six times more nutrients than P. chlororaphis. These observations indicate that the nitrous oxide reductase consumes N 2 O from a very early stage of the N 2 O turnover, likely because the cultures were grown under anaerobic conditions leaving both N 2 O-producing and -consuming enzymes active from the beginning of the experiment. 15 N 2 O produced by the two denitrifying bacteria differs in the bulk isotope enrichment whereas the SP enrichments are averaging to similar values. From the presented experiments we have found that the difference in enrichment between P. chlororaphis and P. fluorescens on average is 44.9 ‰ and 0.8 ‰ for bulk and SP respectively (Fig. 6). We therefore find that the isotopomers produced by P. fluorescens are more depleted than those produced from P. chlororaphis, since the Rayleigh is calculated as product-to-substrate fractionation. Sutka and Ostrom (2006) conclude that a difference in the nitrite reductase does not have an 20 effect on the SP during denitrification. This conclusion is based on measurements of P. chlororaphis (ATCC 43928) and P. aureofaciens (ATCC 13985) possessing cd1-type nitrite reductase and Cu-containing nitrite reductase, respectively. We conclude the same for P. fluorescens and P. chlororaphis and therefore propose that the conclusion applies to all denitrifying bacteria.
The observed difference in the isotope enrichment during production of N 2 O could originate from a difference in the nitric oxide reductase enzymes. Nitric oxide reductase is the primary enzyme in a chain of catalytic reactions leading to the produc- Hino et al., 2010); (Hendriks et al., 2000)]. The catalytic cycle involving production of N 2 O from NO has yet to be completely understood with respect to the formation of the N-N double bond, the complexity of the structural information of nitric oxide reductase, the proton transfer pathway into nitric oxide reductase [(Tosha and Shiro, 2013)], and the very short lifetime of the intermediate states of the molecules [ (Collman et al., 2008)]. We hypothesized that the difference in the bulk observed during incubation of our two bacterial species was due to different nitric oxide reductases produced by the two species.

30
To test this hypothesis, we compared the DNA sequences of the norB and norC genes coding for the large and small subunit, [749309655], and UFB2 [836582503]) as well as two closely related denitrifying species, P. aeruginosa and P. stutzeri. Our analysis revealed 1) a very high similarity of the two genes in P. fluorescens and P. chlororaphis and 2) that the intra-species variability of the two genes was similar to the inter-species variation. This led us to reject our hypothesis and conclude that differences in nitric oxide enzymes produced by the two species were not responsible for the observed differences in the bulk.
We hypothesized that the difference in the bulk isotope enrichment between the two bacteria originates from mass-dependent 5 fractionation associated with the nitrous oxide reductase in P. fluorescens. After production of N 2 O molecules, the light molecules degas quickly out of the liquid phase and away from the nitrous oxide reductase. The heavy N 2 O molecules are slower and react with the nitrous oxide reductase leading to a depletion of the heavy N 2 O isotopes. As this is a diffusion driven process it is mass-dependent with no detectable effect on SP.

10
Numerous publications have presented experiments with both in situ measurements of denitrifying bacterial production and reduction of N 2 O during incubation of bacterial cultures and soil samples. In Fig. 6, we present a comparison between the results from this study and the results from a selection of the previously published results. The general understanding is that denitrification results in SP ≤ 10 ‰. Applying different incubation techniques on soils with different properties showed SP during production of N 2 O between -3 ‰ and 9 ‰ and between -2 ‰ and -8 ‰ during reduction 15 2014); (Well and Flessa, 2009a, b)]. During production of N 2 O in bacterial culture experiments involving P. chlororaphis (ATCC 43928) and P. aureofaciens (ATCC 13985), Sutka and Ostrom (2006) found SP values of -2.5 ‰ and 1.3 ‰ and δ 15 N bulk values of between -7 ‰ and -10 ‰. Ostrom et al. (2007) investigated bacterial reduction of N 2 O using P. stutzeri (provided by J. M. Tiedje) and P. denitrificans (ATCC 13867), and the SP resulting from this bacterial reduction of N 2 O was between -6.8 ‰ and -5 ‰. 20 Our results for N 2 O SP and bulk enrichment from P. chlororaphis are in the same range as what has been reported previously.

Conclusions
We have presented successful continuous measurements of the denitrifying bacterial process using two different strains of bacteria: P. fluorescens which is a full denitrifier, and P. chlororaphis which is a denitrifier without nitrous oxide reductase activity. Assuming a Rayleigh type fractionation, modified for isotopomers and simultaneous reduction, we have calculated the isotope enrichment during production and reduction of N 2 O. The enrichment for P. chlororaphis is in line with previous results 30 for both SP and bulk. For P. fluorescens, we find similar SP enrichment values during N 2 O production and reduction. The bulk 10 Biogeosciences Discuss., doi:10.5194/bg-2016-258, 2016 Manuscript under review for journal Biogeosciences Published: 30 June 2016 c Author(s) 2016. CC-BY 3.0 License. isotope enrichment calculated for N 2 O reduction is in line with previously presented results though for production we find an isotope depletion. We believe that, in our experiment, the bulk isotope depletion is due to mass-dependent fractionation.
Author contributions. MW and TB designed the experiments and MW carried out the measurements and analyzed data. SC prepared the bacteria prior to experiments. AP analyzed the bacterial DNA sequences. DBH and EC developed the G5101i-CIC analyzer. MW prepared the manuscript with contribution from all co-authors.

5
Acknowledgements. We thank Jan Kaiser for isotope specific gas samples used for our reference gasses, Sakae Toyoda, Naohiro Yoshida, Carina van der Veen, and Thomas Röckmann for assistance on measurements of our reference gasses. We want to thank Center for Permafrost (CENPERM DNRF100) and the Centre for Ice and Climate, funded by the Danish National Research Foundation for their support, and The Danish Agency for Science Technology and Innovation, for funding used in supporting this project. We want to thank Picarro Inc. and especially Eric Crosson, and Nabil Saad for the collaboration and guidance in the development of the N2O isotope analyzer prototype used       Mean -53.3 ± 9.8 -47.5 ± 9.7 -50.4 ± 9.3 -5.7 ± 5.6  Mean -8.8 ± 1.4 -2.2 ± 1.1 -5.5 ± 0.9 -6.5 ± 1.8