Introduction

Both nitrogen (N) and oxygen (O) isotopes of fixed nitrogen compounds (i.e., ammonia, nitrite, and nitrate) are fractionated during their microbial production and consumption processes comprising the global marine N cycle. Thus, measurements of nitrogen (15N/14N) and oxygen (18O/16O) isotope ratios of fixed nitrogen compounds have long been used as invaluable biogeochemical stable isotopic tracers to estimate the global marine N budget [1, 2]. The dual N and O isotope analyses provide complementary signatures of co-occurring N transformation processes that could not be revealed by N isotope measurement alone [3]. By convention, these stable isotope ratios are expressed in delta notation (δ15N and δ18O) in per mille (‰) versus atmospheric N2 (air) and Vienna Standard Mean Ocean Water (VSMOW): δ15N = ([15N/14N]sample/[15N/14N]air − 1) × 1000 and δ18O = ([18O/16O]sample/[18O/16O]VSMOW − 1) × 1000, respectively. To quantitatively assess the impacts of these processes on dissolved nitrogen species, the degree of isotope fractionation is quantified by the kinetic isotope effect, ε (‰) = [(kL/kH) – 1] × 1000, where kL/kH is the ratio of the reaction rate constants between the light (kL) and heavy (kH) isotopically substituted substrates. The kinetic N and O isotope effects (15ε and 18ε, respectively) for key microbial processes provide the basis for interpretation of natural abundance N isotopic distributions in the ocean, freshwater, terrestrial, and groundwater ecosystems.

The kinetic N and O isotope effects associated with microbial ammonium oxidation [4,5,6,7,8], nitrite oxidation [9, 10], dissimilatory nitrite reduction to nitric oxide [11], dissimilatory nitrate reduction (denitrification) [12, 13], and assimilatory nitrate reduction [14, 15] have been determined for laboratory cultures of corresponding bacteria (Fig. 1). These isotope effects vary depending on the microbial species [12], enzymes [11], growth conditions, and/or rates of reaction [6]. Furthermore, the coupled N and O isotope measurements of NOx (nitrate and/or nitrite) have been performed in the North Pacific margin (e.g. [16,17,18]), Hadal oceans [19], and the Peru oxygen-deficient zone (ODZ) [20,21,22] to assess the regional N cycles. However, there is still much uncertainty with regard to the N cycling processes, especially in ODZs because the isotope effects are not well known for all of the relevant processes.

Fig. 1
figure 1

Marine microbial nitrogen cycle with the reported N and O isotope effects (15ε and 18ε, respectively) of key transformation processes. The O isotope effects of anammox have not been determined yet

Anaerobic ammonium oxidation (anammox) and denitrification are the two major sinks of fixed nitrogen (N) in the ocean. It has been estimated that these microbial processes together remove 230–450 Tg N yr−1 from the global ocean [23, 24], which is thought to occur mainly in oxygen-deficient water columns and sediments. Furthermore, anammox bacteria also contribute to re-oxidation of nitrite to nitrate (i.e., recycling N), because they fix CO2 into biomass with reducing equivalents generated from oxidation of nitrite to nitrate [25]. Nitrate production by anammox bacteria significantly influences the nitrite and nitrate N and O isotope effects in freshwater and marine systems, which, however, has been overlooked so far [26]. Despite the importance of anammox bacteria in the global N cycle [27,28,29,30,31], the N isotope effect (15ε) associated with anammox metabolism has been determined for only one freshwater anammox strain, “Ca. Kuenenia stuttgartiensis” [32]. The kinetic O isotope effect (18ε) of anammox metabolism has not yet been determined. Consequently, their impacts on the distributions of N and O isotopes in the natural environmets could not be addressed.

Five candidatus genera, ‘Ca. Brocadia’ [33], ‘Ca. Kuenenia’ [34], ‘Ca. Scalindua’ [35], ‘Ca. Anammoxoglobus’ [36] and ‘Ca. Jettenia’ [37], and about 20 candidatus species have been tentatively identified to date. The genus ‘Ca. Scalindua’ is halotolerant and the most abundant anammox bacteria found in marine environments [38]. We therefore hypothesized that the nitrite and nitrate N and O isotope effects induced by different genera of anammox bacteria might be different, since they are phylogenetically diverse and might possess different enzymes (e.g., nitrite reductase) and consequently different metabolic pathways [39]. Especially, the determination of isotope effects of marine anammox species, ‘Ca. Scalindua’, is essential to better understand the natural abundance of stable isotope ratios (δ18O and δ15N) in the ocean.

Here, we analyzed the N and O isotope effects (15ε and 18ε) of nitrite and nitrate associated with anammox metabolism by three anammox species: ‘Ca. S. japonica’, ‘Ca. J. caeni’, and ‘Ca. B. sinica’ in continuous enrichment cultures. We found that species-dependent N and O isotope effects (15ε and 18ε), which could provide significant insights into the relative contribution of anammox bacteria to the fixed N loss and nitrite re-oxidation (recycling N) in various natural environments.

Materials and methods

Continuous culture experiments

Free-living planktonic cultures of three anammox bacteria species were enriched and cultivated in 3 L membrane bioreactors (MBRs) equipped with a hollow fiber membrane module (pore size 0.1 μm, polyethylene) as previously described [38, 40, 41] (Fig. S1). The pH was not controlled but was always between 7.9–8.0 for “Ca. S. japonica”, 8.6–8.9 for “Ca. J. caeni”, and 7.3–7.4 for “Ca. B. sinica”, respectively. The temperature was controlled at 25 °C for “Ca. S. japonica”, 30 °C for “Ca. J. caeni”, and 37 °C for “Ca. B. sinica”, respectively. The details of reactor operation and culture preparation are given in Fig. S1. Once the MBRs have reached a steady state (the concentrations of nitrogen compounds (NH4+, NO2, and NO3) in the effluents stabilized after about 2-month operation), three or four sub-samples of MBR influent and effluent (permeate) were taken. After sampling, pH was measured and immediately filtered using 0.2-μm cellulose acetate filter (Advantec) for concentration and isotopic measurements of NH4+, NO2, and NO3.

Sample preparation for isotopic analysis

After filtration, immediately samples were adjusted to pH 2 by adding 2 M H2SO4 solution and then stored at −20 °C until analysis for N isotope of NH4+ to prevent NH4+ from volatilizing. To analyze N and O isotope of NO2, after filtration, immediately sample solution was adjusted to pH 12 by adding 2 M NaOH solution and stored at −20 °C until analysis to prevent O isotope exchange between NO2 and H2O during sample storage. To analyze N and O isotope of NO3, after filtration, if the concentration ratio of NO2/NO3 was over 5%, NO2 in the sample solution was immediately removed by adding sulfamic acid (H3NSO3), because NO2 interferes with NO3 isotope analysis. The concentration of NO2 was measured with naphthylethylenediamine method [42] to confirm NO2 was completely removed. We confirmed that NO2 was completely removed in all samples. Then, sample solution was adjusted to pH 8.5–9 by adding 2 M NaOH solution and stored at −20 °C until analysis.

Chemical analyses

The concentration of NH4+ was measured by the indophenol blue method [42] with a multi-label plate reader (ARVO MX 1420-01J; PerkinElmer; Waltham, MA, USA). The NO2 concentration was measured by the naphthylethylenediamine method [42]. The concentration of NO3 was measured using ion chromatographs (IC-2010, TOSOH; Tokyo, Japan) equipped with a TSKgel IC-Anion HS column (TOSOH; Tokyo, Japan).

Isotope ratio analyses

NH4+ nitrogen isotope analyses were performed by using the ammonium diffusion method [43, 44] and subsequently measured by a EA-IRMS (Flash EA1112, ConFlo IV interface, Delta plus Advantage; ThremoFinnigan). International and internal NH4+ isotopic standards, USGS25 (δ15N = −30.4‰), USGS26 (δ15N = 53.7‰), and IAEA-N-2 (δ15N = 20.3‰) were used for the calibration. Replicate analyses yielded respective precision of 0.3‰ for δ15NNH4+.

NO2 nitrogen and oxygen isotope ratios were measured by chemical conversion of NO2 to nitrous oxide (N2O) with the azide method [45]. All samples and standards were exactly adjusted to same pH (pH = 12) and salinity (0.5 M NaCl) as mentioned above. The N2O was then analyzed in duplicate using a GC–IRMS (SerCon) with in-house calibrated NO2 isotopic standards that were calibrated against N-23, N-7373, and N-10219 [46]; NO2-1 (δ15N = −66.9‰, δ18O = 27.0‰), NO2-2 (δ15N = −64.8‰, δ18O = 1.1‰), NO2-3 (δ15N = −67.3‰, δ18O = 20.3‰), NO2-4 (δ15N = −68.4‰, δ18O = 14.9‰), NO2-5 (δ15N = 0.2‰, δ18O = 20.2‰). Replicate analyses yielded respective precisions of 0.3‰ for δ15NNO2− and 0.5‰ for δ18ONO2-, respectively.

NO3 nitrogen and oxygen isotope ratios were measured by microbial conversion of NO3 to N2O with the denitrifier method [47, 48]. N2O was analyzed in triplicate using a GC–IRMS (SerCon) with international NO3 isotopic standards; IAEAN3 (δ15N = 4.7‰, δ18O = 25.6‰), USGS32 (δ15N = 180‰, δ18O = 25.7‰), USGS34 (δ15N = −1.8‰, δ18O = −27.9‰), and USGS35 (δ18O = 57.5‰). Replicate analyses yielded respective precisions of 0.2‰ for δ15NNO3− and 1.0‰ for δ18ONO3−, respectively.

Calculations of isotope effects

Anammox bacteria oxidize NH4+ directly to N2 gas with NO2 as the terminal electron acceptor in the absence of oxygen, and NO2 is concomitantly oxidized to NO3 as shown in the following stoichiometric equation [49]:

$$\begin{array}{l}1{\mathrm{NH}}_4^ + + 1.146{\mathrm{NO}}_2^ - + 0.071{\mathrm{HCO}}_3^ - + 0.057{\mathrm{H}}^ + \\ \to 0.986{\mathrm{N}}_2 + 0.161{\mathrm{NO}}_3^ - + 0.071{\mathrm{CH}}_{1.74}{\mathrm{O}}_{0.31}{\mathrm{N}}_{0.20} + 2.002{\mathrm{H}}_2{\mathrm{O}}\end{array}$$

The continuous MBR system is considered as an open system with balanced input and two or three output fluxes (product(s) and residual substrate) [50]. The N isotope effects were analyzed for the following redox reaction: (1) NH4+ oxidation to N2, (2) NO2 reduction to N2, and (3) NO2 oxidation to NO3.

N isotope effect of NH4 + oxidation to N2

Ammonium (NH4+) was continuously fed into the MBR where NH4+ oxidation to N2 (NH4+ →  N2) occurs with fractionation (15ε1), and unused NH4+ exits without further fractionation (i.e., 15ε2 = 0) (Fig. 2a). The isotope compositions of product N2 (δproduct) and residual NH4+ (δRS) at steady state can be given simply by subtracting ε fractionations from the isotope composition of intermediate pool (δp) [50].

$$\delta_{{\mathrm{product}}} = \delta _{\mathrm{p}} - {\, }^{15}\varepsilon _{1}$$
(1)
$${\mathrm{\delta }}_{{\mathrm{RS}}} = {\mathrm{\delta }}_{\mathrm{p}} - {\, }^{15}{\mathrm{\varepsilon }}_2 = {\mathrm{\delta }}_{\mathrm{p}}\;\;\left( {{\mathrm{when}}\, {\, }^{15}{\mathrm{\varepsilon }}_2 = 0} \right)$$
(2)
Fig. 2
figure 2

Diagram of a steady-state box model. a a substrate (NH4+) entering to a MBR and a product (N2) is formed and unused substrate (NH4+) exits without further fractionation. b A substrate (NO2) entering to a MBR and two products (N2 and NO3) are formed and unused substrate (NO2) exits without further fractionation

The steady-state isotope mass balance can be described as follows:

$$\begin{array}{*{20}{c}} {\delta _{{\mathrm{IN}}} = f\delta _{{\mathrm{product}}} + \left( {1-f} \right)\delta _{{\mathrm{RS}}}} \\ {\delta _{{\mathrm{IN}}} = f\left( {\delta _{\mathrm{p}} - {\, }^{15}\varepsilon _{1}} \right) + \left( {1-f} \right)\left( {\delta _{\mathrm{p}} - {\, }^{15}\varepsilon _{2}} \right)} \\ {\delta _{\mathrm{p}} = \delta _{{\mathrm{IN}}} + f^{15}\varepsilon _{1}} \end{array}$$
(3)

where f is the fraction of NH4+ consumed at steady state.

Therefore, from Eqs. (2) and (3), the kinetic isotope effects associated with NH4+ oxidation to N2 (15ε1) can be determined as follows, when δIN, δRS, and f are measured experimentally:

$$\begin{array}{*{20}{c}} {{}^{{\mathrm{15}}}{\mathrm{\varepsilon }}_{\mathrm{1}} = \left( {{\mathrm{\delta }}_{{\mathrm{RS}}} - {\mathrm{\delta }}_{{\mathrm{IN}}}} \right){\mathrm{/f}}} \\ {f = \left( {{\mathrm{C}}_{{\mathrm{in - NH4 + }}} - {\mathrm{C}}_{{\mathrm{out - NH4 + }}}} \right){\mathrm{/C}}_{{\mathrm{in - NH4 + }}}} \end{array}$$
(4)

where Cin-NH4+ and Cout-NH4+ are the NH4+ concentrations of MBR influent and effluent, respectively.

N isotope effects of NO2 reduction and oxidation

Nitrite (NO2) is also continuously fed together with NH4+ into the MBR where NO2 reduction to N2 and NO2 oxidation to NO3 concomitantly occur with fractionations (defined as 15ε3 and 15ε5, respectively), and unused residual NO2 exits the MBR without further fractionation (i.e., 15ε4 = 0) (Fig. 2b). The isotope compositions of product N2 (δproduct1), product NO3 (δproduct2), and residual NO2RS) at steady state can be given simply by subtracting ε fractionations from the isotope composition of intermediate pool (δp):

$${\mathrm{\delta }}_{{\mathrm{product1}}} = {\mathrm{\delta }}_{\mathrm{p}}-{}^{{\mathrm{15}}}{\mathrm{\varepsilon }}_{\mathrm{3}}$$
(5)
$$\delta_{\mathrm{RS}} = \delta_{\mathrm{p}}-{ }^{15}\varepsilon_{4} = \delta_{\mathrm{p}}\left({\, }^{15}\varepsilon_{4} = 0\right)$$
(6)
$$\delta _{{\mathrm{product2}}} = \delta _{\mathrm{p}}- {}^{15}\varepsilon _{5}$$
(7)

Now it is assumed that input flux is 1 and output flux is divided into α, β, and γ (Fig. 2b)

$${\mathrm{1}} = {\mathrm{\alpha }} + {\mathrm{\beta }} + {\mathrm{\gamma }}$$
(8)

where α is the fraction of NO2 converted to N2. It is represented by NH4+ consumption, because NH4+ and NO2 react at 1:1 ratio to form N2 (excluding NO2 oxidation to NO3 from the overall anammox reaction).

$${\mathrm{\alpha }} = \left( {{\mathrm{C}}_{{\mathrm{in - NH4}} + } - {\mathrm{C}}_{{\mathrm{out - NH4}} + }} \right)/{\mathrm{C}}_{{\mathrm{in - NO2 - }}}$$
(9)

β is the fraction of residual NO2 (unused) at steady state, which is experimentally determined.

$${\mathrm{\beta }} = {\mathrm{C}}_{{\mathrm{out - NO2 - }}}/{\mathrm{C}}_{{\mathrm{in - NO2 - }}}$$
(10)

γ is the fraction of NO2 oxidized to NO3. Although γ could be determined from the NO3 concentration in MBR effluent at steady state, the influence of heterotrophic denitrification and dissimilatory nitrate reduction to ammonium (DNRA) by anammox bacteria cannot be excluded. Therefore, γ was determined as follows:

$${\mathrm{\gamma }} = {\mathrm{1}} - {\mathrm{\alpha }} - {\mathrm{\beta }}$$
(11)

Applying Eqs. (5), (6), and (7) yields the following steady-state isotope mass balance:

$$\delta _{{\mathrm{IN}}} = \alpha \, \delta _{{\mathrm{product}}1} + \beta \, \delta _{{\mathrm{RS}}} + \gamma \, \delta _{{\mathrm{product}}2}$$
(12)
$${\mathrm{\delta }}_{{\mathrm{IN}}} = {\mathrm{\alpha }}\left( {{\mathrm{\delta }}_{\mathrm{p}}-{}^{{\mathrm{15}}}{\mathrm{\varepsilon }}_{\mathrm{3}}} \right) + {\mathrm{\beta }}\,{\mathrm{\delta }}_{\mathrm{p}} + {\mathrm{\gamma }}\left( {{\mathrm{\delta }}_{\mathrm{p}}-{}^{{\mathrm{15}}}{\mathrm{\varepsilon }}_{\mathrm{5}}} \right)\\ \\ = \left( {{\mathrm{\alpha }} + {\mathrm{\beta }} + {\mathrm{\gamma }}} \right){\mathrm{\delta }}_{\mathrm{p}}-{\mathrm{\alpha }}^{15}{\mathrm{\varepsilon }}_3-{\mathrm{\gamma }}^{15}{\mathrm{\varepsilon }}_{5}$$
(13)
$$\hskip 15pt = {\mathrm{\delta }}_{\mathrm{p}}-{\mathrm{\alpha }}^{{\mathrm{15}}}{\mathrm{\varepsilon }}_{\mathrm{3}}-{\mathrm{\gamma }}^{{\mathrm{15}}}{\mathrm{\varepsilon }}_{\mathrm{5}} = {\mathrm{\delta }}_{{\mathrm{RS}}}-{\mathrm{\alpha }}^{{\mathrm{15}}}{\mathrm{\varepsilon }}_{\mathrm{3}}-{\mathrm{\gamma }}^{{\mathrm{15}}}{\mathrm{\varepsilon }}_{\mathrm{5}}$$
(14)

Therefore, 15ε3 and 15ε5 can be calculated as follows, when δIN, δRS, and δproduct2 are measured experimentally:

$$^{15}{\mathrm{\varepsilon }}_3 = \left( {{\mathrm{\delta }}_{{\mathrm{RS}}} - {\mathrm{\delta }}_{{\mathrm{IN}}} - {\mathrm{\gamma }}^{15}{\mathrm{\varepsilon }}_5} \right){\mathrm{/\alpha }}$$
(15)
$$^{{\mathrm{15}}}{\mathrm{\varepsilon }}_{\mathrm{5}} = {\mathrm{\delta }}_{{\mathrm{RS}}} - {\mathrm{\delta }}_{{\mathrm{product2}}}$$
(16)

δproduct2 in the Eqs. (7) and (16) is not simply δout-NO3- because the influent media contained 0.84–1.21 mmol-N/L of NO3, which was originated from the university ground water (Table S1). Thus, the background isotope ratio of NO3 should be considered.

Isotope mass balance

$${\mathrm{\delta }}_{{\mathrm{out - NO3 - }}} \times {\mathrm{C}}_{{\mathrm{out - NO3 - }}} = \,\,{\mathrm{\delta }}_{{\mathrm{in - NO3 - }}} \times {\mathrm{C}}_{{\mathrm{in - NO3 - }}} \\ + {\mathrm{\delta }}_{{\mathrm{produced - NO3 - }}} \times {\mathrm{C}}_{{\mathrm{produced - NO3 - }}}$$
(17)

Mass balance

$${\mathrm{C}}_{{\mathrm{out - NO3 - }}} = {\mathrm{C}}_{{\mathrm{in - NO3 - }}} + {\mathrm{C}}_{{\mathrm{produced - NO3 - }}}$$
(18)

Applying Eqs. (17) and (18) yields the following equation for δproduced-NO3-:

$$ {\mathrm{\delta }}_{{\mathrm{produced - NO3 - }}} = {\mathrm{\delta }}_{{\mathrm{product2}}} =\\ \ \ \left( {{\mathrm{\delta }}_{{\mathrm{out - NO3 - }}} \times {\mathrm{C}}_{{\mathrm{out - NO3 - }}} - {\mathrm{\delta }}_{{\mathrm{in - NO3 - }}} \times {\mathrm{C}}_{{\mathrm{in - NO3 - }}}} \right)\\ \ /{\mathrm{C}}_{{\mathrm{produced - NO3 - }}}$$
(19)

O isotope effect of nitrite oxidation

For calculation of oxygen isotope effect for nitrite oxidation (NO2 → NO3), the produced NO3 contains three oxygen atoms: two oxygen atoms come from NO2 and one oxygen atom comes from H2O. The incorporation of an oxygen atom from water should be considered (18ε6 in Fig. 2b). Thus, for oxygen isotopes, Eq. (7) is rewritten as

$${\mathrm{\delta }}_{{\mathrm{product2}}} = {\mathrm{2}}/{\mathrm{3}}\left( {{\mathrm{\delta }}_{{\mathrm{RS}}} - ^{{\mathrm{18}}}{\mathrm{\varepsilon }}_{\mathrm{5}}} \right) + {\mathrm{1/3}}\left( {{\mathrm{\delta }}_{{\mathrm{H2O}}} - ^{{\mathrm{18}}}{\mathrm{\varepsilon }}_{\mathrm{6}}} \right)$$
(20)
$${\mathrm{\delta }}_{{\mathrm{product2}}} = {\mathrm{2}}/{\mathrm{3\delta }}_{{\mathrm{RS}}} + {\mathrm{1/3\delta }}_{{\mathrm{H2O}}} - \left( {{\mathrm{2/3}}^{{\mathrm{18}}}{\mathrm{\varepsilon }}_{\mathrm{5}} + {\mathrm{1/3}}^{{\mathrm{18}}}{\mathrm{\varepsilon }}_{\mathrm{6}}} \right)$$
(21)

18ε6 is the isotope effect for water incorporation. Since there are two unknown values (18ε5 and 18ε6) in the Eq. (21), 18ε5 and 18ε6 cannot be calculated independently. Thus, the term (2/3 18ε5 + 1/3 18ε6) in Eq. (21) is denoted as a combined oxygen isotope effect 18ENO2−→NO3−.

$$ ^{{\mathrm{18}}}{\mathrm{E}}_{{\mathrm{NO2}} \to {\mathrm{NO3 - }}} =\\ \ \left( {{\mathrm{2/3}}^{{\mathrm{18}}}{\mathrm{\varepsilon }}_{\mathrm{5}} + {\mathrm{1/3}}^{{\mathrm{18}}}{\mathrm{\varepsilon }}_{\mathrm{6}}} \right) = {\mathrm{2/3\delta }}_{{\mathrm{RS}}} + {\mathrm{1/3\delta }}_{{\mathrm{H2O}}} - {\mathrm{\delta }}_{{\mathrm{product2}}}$$
(22)

This combined oxygen isotope effect can be calculated from the obtained data set in this study.

Results

Reactor performance

After the MBRs have reached a steady state, three or four sub-samples of MBR influent and effluent (permeate) were taken and analyzed for the concentrations and N and O isotope ratios of NH4+, NO2, and NO3, respectively. The steady-state concentrations of NH4+, NO2, and NO3 in the MBR effluents were very stable during the entire experiments (Tables S1 and 1). NO2 was almost completely consumed, whereas 1.3–3.1 mM NH4+ remained in all the MBR effluents and a small amount of NO3 was produced. The average stoichiometric ratios of consumed NO2 and consumed NH4+ (∆NO2/∆NH4+, ranging from 1.11 ± 0.5 to 1.28 ± 0.5) and produced NO3 and consumed NH4+ (∆NO3/∆NH4+, ranging from 0.10 ± 0.01 to 0.2 ± 0.03) (Table 1) agreed with the previously observed stoichiometry of anammox process (i.e. 1.15 and 0.16 for ∆NO2/∆NH4+ and ∆NO3/∆NH4+, respectively) [49], suggesting that the anammox process being responsible for transformation of nitrogen compounds occurring in all three MBRs.

Table 1 Summary of MBR performance and N and O isotope analyses

Although influent NH4+ and NO2 concentrations were different (~10 mM for “Ca. S. japonica”, 10 mM for “Ca. J. caeni” and 16 mM for “Ca. B. sinica”), anammox activity in each MBR culture was consistent as demonstrated by similar volumetric NH4+ consumption, NO2 consumption, and NO3 production rates (Table 1). Furthermore, there were no significant difference in specific NH4+ consumption rates (ranging from 1.14 ± 0.06 to 1.27 ± 0.06 mg-N mg-protein−1 h−1), NO2 consumption rates (ranging from 1.41 ± 0.05 to 1.48 ± 0.04 mg-N mg-protein−1 h−1), and NO3 production rates (ranging from 0.11 ± 0.01 to 0.24 ± 0.03 mg-N mg-protein−1 h−1) among three MBR cultures.

The ratio of anammox bacteria to the cell culture (degree of enrichment cultures) was determined by fluorescent in situ hybridization (FISH); 94.4 ± 6.5% for “Ca. S. japonica”, 86.1 ± 4.9% for “Ca. J. caeni”, and 96.9 ± 2.6% for “Ca. B. sinica”, respectively.

Nitrogen isotope effects (15ε)

The culture media with an equimolar amount of NH4+15NIN = −7.0 ± 0.3‰ to −2.4 ± 0.2‰) and NO215NIN = −3.7 ± 0.1‰ to −2.6 ± 0.7‰ and δ18OIN = 6.0 ± 0.5‰ to 8.1 ± 0.7‰) were continuously fed to the individual MBR cultures (Table 1). N isotope effects of NH4+ oxidation to N2 (15εNH4→N2), NO2 reduction to N2 (15εNO2→N2), and NO2 oxidation to NO3 (15εNO2→NO3) were calculated for all three species using the equations described in “Materials and methods” (Fig. 3 and Table S1). In the case of NH4+ oxidation to N2, the values of 15εNH4→N2 were consistent among three species (30.9 ± 0.2‰ – 32.7 ± 0.7‰). In contrast, there were significant differences in the 15εNO2→N2 values (NO2 reduction to N2). “Ca. J. caeni” showed the largest value (29.5 ± 3.9‰), whereas “Ca. B. sinica” yielded the smallest value (5.9 ± 4.5‰). The value of “Ca. S. japonica” was 15εNO2→N2 = 19.9 ± 1.7‰, which was close to the previously reported value of “Ca. K. stuttgartiensis” (16.0 ± 4.5‰) [32]. In the case of NO2 oxidation to NO3, all three species showed strong inverse kinetic isotope effects (15εNO2→NO3 < 0); −30.1 ± 3.0‰ for “Ca. S. japonica”, −45.3 ± 4.2‰ for “Ca. J. caeni”, and −31.5 ± 4.0‰ for “Ca. B. sinica”, respectively (Fig. 3 and Table S1), which consists with the previously reported value of “Ca. K. stuttgartiensis” (−31.1 ± 3.9‰) [32].

Fig. 3
figure 3

Summary of N and O isotope effects induced by different anammox species

Oxygen isotope effect (18ε)

δ18OH2O of ground water which was used for medium preparation was determined to be −11.12‰ ± 0.2‰ (n = 3) and remained stable during an entire experimental period. Both δ18ONO2− and δ18ONO3− in MBR influents were consistent among three anammox cultures, respectively (Table 1). Under steady-state conditions, similar values of δ18ONO2− were determined in all three MBR effluents; 6.4 ± 0.2‰ for “Ca. S. japonica”, 5.2 ± 0.3‰ for “Ca. J. caeni”, and 4.9 ± 1.4‰ for “Ca. B. sinica”, respectively (Table 1). In contrast, the “Ca. S. japonica” MBR yielded a higher δ18ONO3- produced value (12.7 ± 0.8‰) as compared with those of “Ca. J. caeni” (1.7 ± 0.8‰) and “Ca. B. sinica” (1.0 ± 0.4‰). Based on these δ18O data, the combined O isotope effect during NO2 oxidation to NO3 was calculated for all three species using the Eq. (22) (18ENO2→NO3 = (2/3 18ε5 + 1/3 18ε6) = 2/3δRS + 1/3δH2O−δproduct2) (Table S1 and Fig. 3). All three species showed inverse kinetic isotope effects; −12.1 ± 0.8‰ for “Ca. S. japonica”, −1.9 ± 0.8‰ for “Ca. J. caeni”, and −1.5 ± 1.2‰ for “Ca. B. sinica”, respectively.

Discussion

Continuous culture method

In this study, the N and O isotope effects induced by anammox bacteria were measured using continuous MBR anammox-enrichment cultures. There are some advantages of use of the continuous culture system over batch system. A steady-state fractionation model is basically simpler than a batch model such as the Rayleigh model, in which the isotope effect (ε) values can be directly determined from the isotopic compositions of reactants in the influent and products in the effluent at steady state (Fig. 2). Nearly identical results of the N and O isotope effects were obtained from the different sampling campaigns for “Ca. B. sinica” (Table S2), which indicates the high reproducibility of continuous steady-state culturing systems for isotope effect analyses (Table S3). The precision of 15ε and 18ε as measured in the steady-state continuous systems compares favorably with that of batch culture experiments [7].

In addition, growth conditions (i.e., pH, concentrations of reactants and products, and so on) vary over time, which may also significantly affect the isotope fractionations. However, oxygen isotope exchange between NO2 and H2O cannot be evaluated in the continuous culture experiment alone, and thus batch culture experiments must be conducted in parallel.

Species-level differences

Among five tentatively proposed candidatus genera of anammox bacteria, one putative marine strain (‘Ca. Scalindua japonica’) and two freshwater strains (‘Ca. Brocadia sinica’ and ‘Ca. Jettenia caeni’) were examined for N and O isotope effects of anammox metabolism in this study. The results revealed species-dependent isotope effects of NO2 reduction to N2. During anammox metabolism (NO2 + NH4+ → N2), the following three enzymatic reactions occur; (i) NO2 reduction to nitric oxide (NO) or hydroxylamine (NH2OH) [51], (ii) hydrazine (N2H4) formation from NO or NH2OH and NH4+, and (iii) N2H4 oxidation to N2 [25]. Hydrazine synthesis is considered to be the rate-limiting step in this reaction sequence due to three-electron reduction reaction [25].

For the conversion of NH4+ to N2, the N isotope effects (15εNH4→N2) of all three species are consistent (30.9–32.7‰), which also do not significantly differ from the range of 15εNH4→N2 reported previously for “Ca. K. stuttgartiensis” (23.5–29.1‰) [32]. This is probably because this reaction is mediated through the same enzymes, such as hydrazine synthase (hzs) and hydrazine dehydrogenase (hdh) in all anammox bacteria species (Fig. 4).

Fig. 4
figure 4

Proposed species-level difference in N isotope effects of nitrite (15εNO2→N2) induced by different anammox bacteria species

On the other hand, for the conversion of NO2 to N2, significant variations of the N isotope effects (15εNO2→N2) were found among the three species: 15εNO2→N2 = 19.9 ± 1.7‰ for “Ca. S. japonica”, 15εNO2→N2 = 29.5 ± 3.9‰ for “Ca. J. caeni”, and 15εNO2→N2 = 5.9 ± 4.5‰ for “Ca. B. sinica”, respectively (Fig. 3 and Table S1.). The previously reported 15εNO2→N2 values of “Ca. K. stuttgartiensis” ranged between 11.9‰ and 18.9‰ (average = 16.0 ± 4.5‰) [32].

Different degree of N isotopic fractionation could be imparted by different nitrite reductase (Fig. 4). It has been reported that “Ca. S. japonica” and “Ca. K. stuttgartiensis” possess a cytochrome cd1(iron, Fe)-type NO-forming nitrite reductase (Fe-NIR) [52, 53], whereas “Ca. J. caeni” has a copper (Cu)-containing NO-forming nitrite reductase (Cu-NIR) [40]. “Ca. B. sinica”, however, does not possess canonical nitrite reductase genes (neither Fe-NIR nor Cu-NIR) and reduces NO2 to NH2OH, instead of NO [51]. Interestingly, both “Ca. S. japonica” and “Ca. K. stuttgartiensis” yielded similar 15εNO2→N2 values (15εNO2→N2 = 19.9 ± 1.7‰ and 16.0 ± 4.5‰), while “Ca. J. caeni” yielded higher values (15εNO2→N2 = 29.5 ± 3.9‰). Furthermore, “Ca. B. sinica” yielded distinctively lower 15εNO2→N2 values (5.9 ± 4.5‰) (Fig. 4). The different N and O isotope effects between Fe-NIR and Cu-NIR were also demonstrated for nitrite reduction by denitrifying bacteria [11]. This difference was explained by the difference in NO2 and enzyme-binding mechanism: the Cu-NIR binds to both O atoms of NO2 whereas the Fe-NIR binds to the N atom, resulting in a smaller N isotope effect for Fe-NIR [11].

However, the recent literatures have reported that Fe-NirS was not hardly expressed at the transcriptional level in both “Ca. K. stuttgartiensis” [54] and Scalindua-related single amplified genomes from ODZs [55, 56]. In addition, Cu-NirK expression was not identified in “Ca. J. caeni” [40]. It has been postulated recently that the reduction of NO2 to NO could be catalyzed by a HAO-like octahaem oxidoreductase in the case of “Ca. K. stuttgartiensis” [54]. The highly similar protein was also identified in the ODZ SAGs [56]. “Ca. B. sinica” that has neither NirS nor NirK also possesses the identical HAO-like octahaem oxidoreductase [51]. However, nitrite reductase that actually works in individual anammox species is not identified yet. To confirm this enzyme-level differences in 15εNO2→N2, true nitrite reductase must be identified.

Three anammox bacterial species were cultured at different pH and temperature in this study; pH 7.9–8.0 and 25 °C for “Ca. S. japonica”, pH 8.6–8.9 and 30 °C for “Ca. J. caeni”, and pH 7.3–7.4 and 37 °C for “Ca. B. sinica”, respectively. However, this probably does not cause variations in 15εNO2→N2. To our best knowledge, there were no studies that show a dependence of the organism-level N isotope effect on either pH or temperature so far, although temperature and pH influence the O isotope effect of NO2 due to isotopic exchange with H2O [8, 57]. Taken together, a difference in nitrite reductase would most likely cause the differences in NO2 reduction isotope effects (15εNO2→N2) in this study.

For the oxidation of NO2 to NO3, all three anammox species exhibited pronounced inverse N isotope effects (−45.3 ± 4.2‰ to −30.1 ± 3.0‰), which agreed with the previously reported value for “Ca. K. stuttgartiensis” (−31.1‰ ± 3.9‰) [32], but exceeded the values for nitrite-oxidizing bacteria (NOB) (−9.1‰ to −20.6‰) [10]. Anammox bacteria can reverse this enzymatic reaction, namely they can reduce NO3 back to NO2 [58, 59]. It is thus hypothesized that this reversible reaction would promote isotope exchange between NO2 and NO3 and lead to more pronounced isotope effects as observed in sulfur metabolism [60]. Intriguingly, it has been also speculated that environmental stresses cause a significant N isotope exchange between NO2 and NO3 (−60.5 ± 1.0‰) in a “Ca. K. stuttgartiensis” batch culture [32]. However, since this phenomenon was not ubiquitously observed, it still remained unclear whether this was caused by cell lysis during cultivation and/or sample preparation. In the present study, all samples for isotope effect measurements were collected from steady-state continuous anammox-enrichment cultures grown under physiological anoxic conditions. Thus, environmental stresses could be minimized, and active biomass dominated in all cultures (>98% were active cells as determined Live/Dead staining, data not shown). Since isotope exchange between NO2 and NO3 is indeed an interesting and important phenomenon for interpretation of the N and O isotope effects, it must be addressed in the future.

It should be also noted that N and O isotope effects are influenced even by subcellular localization and amino acid sequences of enzymes (Fig. S2). For example, membrane-bound cytoplasmic and periplasmic Nxr of Nitrobacter and Nitrospira yielded significantly different 15εNO2−→NO3− and 18εNO2−→NO3− [10]. However, since the actual subcellular localization and amino acid sequences of enzymes in anammox bacteria are not fully understood currently, their influences need to be further investigated.

O isotope effects of NO2 and NO3

We could report only the combined O isotope effects for NO2 oxidation to NO3 (18ENO2−→NO3− = 2/3 18εNO2−→NO3− + 1/3 18εH2O, Eq. (22)) by anammox bacteria in this study. Since both NO2 reduction to N2 and oxidation to NO3 are simultaneously occurring in anammox process (Fig. S3B), the δ18O values of NO2 represents a superimposed signal of the two processes. In addition, the δ18ONO2 value is affected by abiotic O isotope exchange between NO2 and H2O (18εeq in Fig. S3). A water-derived O atom is also incorporated into NO3 during NO2 oxidation to NO3 [61]. Therefore, the δ18ONO3 value of the MBR effluent is directly related to the culture medium (water) δ18OH2O through both O isotope equilibration of NO2 (18εeq) and incorporation of a water-derived O atom (18εH2O) [26]. Thus, O isotope effect for NO2 oxidation to NO3 (18εNO2−→NO3−) and O isotope effect for water incorporation (18εH2O) cannot be determined separately in the continuous culture experiment alone in this study. These parameters could be species dependent and thus should be separately determined for individual anammox bacteria species for better understanding of N and O isotope systematics and nitrogen cycling in natural environments.

The combined O isotope effects for NO2 oxidation (18ENO2−→NO3−) showed inverse kinetic isotope effects; 18ENO2 → NO3: −12.1 ± 0.8‰ for “Ca. S. japonica”, −1.9 ± 0.8‰ for “Ca. J. caeni” and −1.5 ± 1.2‰ for “Ca. B. sinica”, respectively (Fig. 3 and Table S1). According to the Eq. (22) for the combined O isotope effect (18ENO2−→NO3− = 2/3 18εNO2−→NO3− + 1/3 18εH2O), 18εNO2−→NO3− by anammox bacteria can be estimated when O isotope effect for water incorporation (18εH2O) was assumed to be 14‰ as assumed for NOB previously [18, 26]: −25.2‰ for “Ca. S. japonica”, −9.9‰ for “Ca. J. caeni”, and −9.3‰ for “Ca. B. sinica”, respectively. Aerobic nitrite oxidizing bacteria (NOB) also yielded inverse kinetic isotope effects of NO2 oxidation (18εNO2→NO3) ranging from −1.3 ± 0.4‰ to −8.2 ± 2.5‰ [9]. It should be noted again that we report a combined O isotope effect (Eq. (22)) whereas Buchwald and Casciotti [9] report the kinetic isotope effect on NO2 alone (equivalent to 18ε5 in Eq. (20)).

The rate of abiotic O isotope exchange between NO2 and H2O is rapid relative to the biological NO2 turnover rate and dependent on temperature and pH [8, 57]. The rate is faster at lower pH and higher temperature. Three anammox bacterial species were cultured at different pH and temperatures in this study, which might cause variations in δ18ONO2 and δ18ONO3 values of MBR effluent and consequently the combined O isotope effects for NO2 oxidation (18ENO2→NO3) (Table S1).

Application to ecological studies

Although natural abundance N and O isotope ratios of nitrate (δ15NNO3 and δ18ONO3) have been used as an invaluable tool to identify the source and to determine the biogeochemical transformation processes [62], the isotope balances of oceanic NO3 are still poorly constrained at present. This is partly because NO3 can be produced during anammox (Fig. 1), which has been overlooked and led to divergent interpretation of δ15NNO3 in freshwater and marine systems. Therefore, the contribution of anammox bacteria to NO2 reoxidation to NO3 (recycling N) in the environments is currently one of the most prominent research topics.

A numerical NO3 isotope dynamics model was developed to evaluate the relative contribution of anammox to NO3 production in the marine and freshwater systems [26]. In this model, since O isotope effect of NO2 oxidation to NO3− (18εNO2−→NO3−) by anammox bacteria was not available, the 18εNO2−→NO3− values of NOB (−7.0‰ to −3.0‰) were used instead. According to this model, the inverse N and O isotope effects significantly influence the δ15N and δ18O of produced NO3 and consequently the corresponding Δδ18O:Δδ15N trajectories. More negative (i.e., lower) 15εNO2−→NO3− value pushes up δ15NNO3− value, thereby lowering the Δδ18O:Δδ15N trajectories. In contrast, more negative 18εNO2−→NO3− value pushes up δ18NNO3− value, thereby lifting up the Δδ18O:Δδ15N trajectories. It should be noted that the lower 18ENO2−→NO3− value (i.e., more negative) was yielded for a marine species “Ca. S. japonica” than other two freshwater species (Fig. 3), suggesting that the higher Δδ18O:Δδ15N trajectories could be expected in marine systems than in freshwater systems. This can partly explain the widely observed Δδ18O:Δδ15N trajectories in freshwater systems (<1) and in marine systems (≥1).

Nitrite is an important branch compound between N loss by denitrification and anammox, and N retention by NO2 reoxidation to NO3. Natural abundance N and O isotopes of nitrite (δ15NNO2 and δ18ONO2) also provided an additional diagnostic to estimate the relative contribution of anammox to the NO3 pool [26]. Nitrite δ15N and δ18O measurements have been used to evaluate what oxidative and reductive NO2 transformation processes are occurring and to what extent in ODZs [3].

Anammox bacteria in ODZs do not affiliate with the genus Scalindua, but with distinct clusters that are clearly separated from the sediment species [56]. Thus, dual N and O isotope effects of NO2 reoxidation should be further explored for other marine water and sediment anammox species for better model simulations for the oceanic N budget.