Isotope discrimination provides new insight into biological nitrogen fixation


(tel +61 83138104; email

Biological N2 fixation, the assimilation of atmospheric N2 into NH3, is the province of highly specialized microorganisms, and a key entry point for atmospheric nitrogen (N) into terrestrial ecosystems (Vitousek et al., 2002). It is probably the most important biologically mediated process after photosynthesis, and is universally carried out by the enzyme nitrogenase. Collectively, N2 fixing organisms are termed diazotrophs, some of which can fix N2 in a ‘free-living’ state, while others fix N2 in loose association with plants, and a select few in highly evolved, complex symbioses on plant roots or stems. Despite its importance, physiological control of biological N2 fixation is only partially understood and quantification of N2 fixation at the field level is difficult (Unkovich et al., 2008). Further progress in quantifying N cycle fluxes in ecosystems will rely heavily on stable isotope (15N) investigations. These powerful techniques can be used at scales ranging from cell to globe (Vandover et al., 1992; Robinson, 2001; Werner & Schmidt, 2002), but require an understanding of the isotope discrimination associated with N transformations. Generally, compounds containing the lighter of two isotopes react more quickly, resulting in reaction products being isotopically lighter than the substrate, unless all of the substrate is converted to the product (Dawson & Brooks, 2001). In the case of biological N2 fixation this equates to differences in the relative abundance of the stable isotopes 15N and 14N between atmospheric N2 and the fixed NH3 produced by the nitrogenase enzyme in the diazotroph. Natural N isotope abundances (δ15N) are expressed as a parts per thousand (‰) deviation from the 15N composition of atmospheric N2 (0‰) (Mariotti, 1983) and thus one might anticipate fixed NH3 to have a negative δ15N given that the N2 fixation substrate substrate (N2) would be in unlimited supply and one would anticipate preferential reduction of 14N14N (mass 28) over 14N15N (mass 29) or 15N15N (mass 30). The aim of the present paper is to highlight uncertainties surrounding the extent of isotope fractionation associated with N2 fixation, and to provide a possible working framework for interpretation of the available data.

There appears to be only one report (Sra et al., 2004) of isotopic discrimination associated with N2 fixation by purified nitrogenase, which estimated the fixed N to be −17‰ compared to the bottled N2 supplied. This value was calculated based on the consumption of N2 and the initial and final δ15N of that N2, not on direct measurement of the fixed NH3. This value is very different from those given for whole organisms, perhaps because exposure of the enzyme to N2 in a test tube does not include the same controls on gaseous diffusion rates and dinitrogen concentration that occur within free-living diazotrophs or more complex symbiotic systems. Nevertheless the result is consistent with what one might anticipate when the substrate is in unlimited supply.

Isotope discrimination during N2 fixation in free-living diazotrophs has been investigated by growing the organisms in laboratory cultures with N2 gas as the only N source. The earliest experiments (Hoering & Ford, 1960) concluded that there was no significant isotope discrimination associated with N2 fixation across four species examined. However, later measurements gave the δ15N of N2-fixation-dependent Anabaena (Macko et al., 1987) as −2.4, −2.4 and −2.2‰ after 9, 12 or 14 d of growth, and a range of other values for N2 dependent bacteria have since been reported from 3.8 (Hoering & Ford, 1960), through −0.37 (Karl et al., 1997) to −3.7‰ (Delwiche & Steyn, 1970). Comparisons of the δ15N of Azotobacter or Anabaena dependent on molybdenum- or vanadium-associated nitrogenase (Rowell et al., 1998) suggested greater discrimination for the vanadium (−4.0‰) than molybdo-nitrogenase (−1.0‰). This would accord with vanadium-nitrogenase being less efficient at N2 fixation (lower relative substrate consumption) than the molybdo-nitrogenase. These results are not strictly direct measures of isotope discrimination associated with the operation of nitrogenase, but being whole organism analyses, also include other possible isotope discrimination effects associated with the movement of all forms of N into or out of the organisms during the incubation periods, which would include death and turnover of the microorganisms.

In N2 fixing symbioses involving legumes, the N2 fixing symbiont is housed in specialized root nodules (Sprent, 2007). Studies of isotope discrimination in nodulated legumes are conducted on intact plants grown in N-free media, totally dependent on N2 fixation for growth. Results from such studies are confusing, with both positive and negative discrimination factors reported. For example, whole plant δ15N values for N2-dependent soybean range from −2.2 (Okito et al., 2004) to + 2.6‰ (Turner & Bergersen, 1983). However, both positive and negative discrimination is not possible for the same reaction (Dawson & Brooks, 2001). A second example of conflicting data can be seen in Table 1 for nodulated legumes dependent on N2 fixation. One study shows discrimination against 14N (Turner & Bergersen, 1983) and the other none. A possible explanation lies in the difference between the two studies in the timing of sampling in relation to physiological development. Preferential losses of 14N from maturing plants (O'Deen, 1989) result in 15N enrichment of the remaining shoot N so that values for mature plants may include discrimination due to processes other than N2 fixation. The second set of data from the author's laboratory, indicated no isotope discrimination for plants harvested just before any senescence (details of plant culture and analysis methods are provided in Unkovich et al., 1994).

Table 1. δ15N (‰) of pot cultured whole legume plants totally dependent on N2 fixation from two different studies
Legume speciesTurner & Bergersen (1983)Author's laboratory
Trifolium subterraneum 2.500.09
Medicago sativa 3.560.02
Lupinus luteus 5.100.10

Another potential source of difference in apparent discrimination in nodulated legumes may lie in errors of measurement because accurate analysis for δ15N is not trivial. For example, analysis of a single sample of soybean roots sent to four different laboratories in Australia resulted in values (Table 2) over a 4.56‰ range. Such variation in δ15N between laboratories is not unique to Australia, with a range of 2‰ being observed for three laboratories elsewhere (Okito et al., 2004). It is likely that measurement errors have crept in with the switch from dual gas inlet machines to single inlet mass spectrometers coupled to combustion systems (Dawson & Brooks, 2001). Along with the associated changes in sample preparation methods, very few laboratory working standards used today have ever been calibrated directly against atmospheric N2. Whereas for the older datasets presented in the literature (Bergersen et al., 1986; Unkovich et al., 1994), laboratory standards, and indeed the actual samples (Hoering & Ford, 1960), were often measured directly against atmospheric N2, ensuring the accuracy of the relative measurement of δ15N for substrate (N2) and product (fixed N). A solution to problems arising from this in relation to measurement of N fixation by 15N natural abundance (see Unkovich et al., 2008) would be to analyse shoot material from N2 dependent plants along with the field samples, thus ensuring the correct relativities between the individual components. This is already evident in many cases in the literature.

Table 2. Variation in δ15N for a standard soybean root sample analysed in different laboratories
Laboratoryδ15N (‰)SD n
  1. The standard sample has been sent to the different laboratories on one or several occasions over a period of 6 yr.

2−3.16 1

Another potential source of error is in the use of indirect methods for estimating isotope discrimination in symbiotic systems (Doughton et al., 1992; Okito et al., 2004). These methods use 15N enriched isotope dilution techniques to estimate fractional dependence of legumes on N2 fixation when grown in the presence of NH4+ and/or NO3. Isotope discrimination is then inferred for matched plants growing without the enriched tracers (Okito et al., 2004). This approach assumes that the 15N isotope dilution technique is absolutely accurate, which it is not (Chalk & Ladha, 1999; Unkovich et al., 2008) and such indirect approaches should not be relied upon.

The author contends that there is no isotope discrimination associated with symbiotic N2 fixation in nodulated legumes. In a study of N2 dependent Lupinus luteus with 12 different strains of rhizobia (Bergersen et al., 1986), none of the whole plant δ15N values were significantly different from 0‰, and the same result was obtained for Trifolium subterraneum with 10 different rhizobia (Unkovich et al., 1994). In both cases there was significant variation in nodule and shoot δ15N dependent on the identity of the microsymbiont, but overall these results clearly showed that on a whole plant basis, there was no evidence for isotope discrimination associated with legume symbiotic N2 fixation per se. The data of Table 3 indicate no fractionation for 10 species examined, with the exception of Medicago truncatula. However, in this case the fractionation is in the wrong direction (enriched substrate) and indicates possible error in the reconstruction of the whole plant values from weighting of component shoot, root and nodule fractions. Some published values indicating significant fractionation might thus be the result of combinations of errors of measurement, cultural conditions, including the use of vermiculite which has been shown to result in isotopic exchange reactions (Bergersen et al., 1988), the timing of sampling or the use of indirect methods.

Table 3. δ15N (‰) of the shoots, roots, nodules and whole plants for pot cultured legumes totally dependent on N2 fixation
SpeciesShootRootNoduleWhole plant
  1. The whole plant δ15N is calculated from the weighted total N contributions of the shoot, root and nodules to the whole plant. With the exception of Medicago truncatula, none of the whole plant values are significantly different from 0‰ by Student's t-test.

Cicer arietinum −1.34−1.175.410.30
Lens culinaris −0.510.721.55−0.08
Lupinus albus −0.64−0.148.540.31
Lupinus luteus −0.88−0.097.850.19
Medicago littoralis −0.201.310.480.21
Medicago polymorpha 0.020.99−0.460.13
Medicago sativa −
Medicago truncatula 0.851.101.110.90
Vicia faba 0.090.361.040.21
Vicia sativa −0.710.340.22−0.37

A common observation in N2 fixing legumes is 15N enrichment of nodules and a depletion of 15N in shoots (Table 3). Suggested explanations for this have included: denitrification in nodules (Shearer et al., 1980) preferentially releasing 14N; importation from phloem of 15N enriched amino acids into nodules (Bergersen et al., 1988); export of 15N depleted ureide from nodules resulted in nodule enrichment (Shearer et al., 1982); deamination of proline or glutamine resulted in enriched soybean nodules (Kohl et al., 1989); or diffusion of NH3 from bacteroids results in discrimination (Yoneyama et al., 1991). None of these hypotheses has provided sufficient explanation for the range of observations on nodule 15N enrichment. The author supports a theory that deamination reactions in the bacteroid are highly fractionating and produce 15N-depleted NH3. Since the symbiont rhizobia are unable to assimilate NH3 (Prell & Poole, 2006), this 15N depleted NH3 will be exported from the nodule, giving rise to 15N enriched nodules and 15N depleted shoots. Evidence for production of 15N depleted NH3 in bacteroids comes from the high enrichment observed in the rhizobial polyamine homospermidine (Yoneyama et al., 1998), which is produced from the combination of two prutrescine molecules and the associated release of NH3 (Fujihara, 2009) which must be substantially depleted in 15N to support the observations of high enrichment of homospermidine. While the quantities of homospermidine produced in nodules are not sufficient to induce the nodule enrichments observed, when combined with other potential deaminating reactions in the bacteroid, deamination reactions provide a plausible explanation for observed nodule enrichment and shoot depletion in 15N in many N2 fixing legumes.

The reduction in discrimination observed from purified nitrogenase enzyme, through the free living diazotrophs to the legume symbioses implies decreasing concentrations of dinitrogen in these N2 fixing systems (Fig. 1). On a whole plant basis there is evidence both consistent with, and contrary to, isotope discrimination by nitrogenase in N2 fixing legumes. No discrimination would imply that the nitrogenase enzyme within the legume nodule is not saturated with N2, because if it were, isotope discrimination would be likely. A lack of discrimination would imply that diffusion of N2 within the N2 fixing nodule is restricted to the extent that it provides no more N2 than that immediately required by the nitrogenase enzyme, a proposition which needs to be tested. The following points lend possibility to the proposition. An effective O2 diffusion barrier is known to operate in legume nodules (Minchin et al., 2007), there is very little difference in the rate of diffusion in water of O2 and N2, and because the solubility of O2 in water (1.04 M) is about double that of N2 (0.54 M) concentrations of N2 in water at atmospheric pressure are only double those of O2 (Bergersen, 1999). Furthermore transporters (leghaemoglobins) for O2 are present, and facilitated transport across the peribacteroid membrane is likely for O2 (Strodtman & Emerich, 2009), yet there has been no suggestion of any facilitated N2 transport within nodules. Against these encouragements, a model of nodule function suggested no N2 limitation to nitrogenase in legume nodules (Bergersen, 1999), although an earlier paper by the same author (Bergersen et al., 1986) showed no N isotope fractionation. Further work is undoubtedly required to reconcile the observations indicating no isotope fractionation and models of nodule function. In earlier studies (Hoering & Moore, 1958) no discrimination during diffusion of N2 through water was observed. Regardless, N2 scarcity in nodules does not mean that legumes are N limited, because they could grow more or bigger nodules to meet total plant N demand.

Figure 1.

When the nitrogen fixing enzyme nitrogenase is purified and allowed to fix N2 in a test tube (a), the enzyme discriminates against 15 N resulting in strong isotope fractionation (−17‰) compared to atmospheric N2 (0‰). Within N2 fixing bacteria growing in laboratory cultures (b), the isotope discrimination by the nitrogenase against 15 N is reduced to −2.5‰, but when the bacteria are housed inside root nodules of legumes (c) no isotope discrimination against 15 N by nitrogenase is observed (0‰). The decreasing discrimination against 15 N from (a) to (c) could be explained by a decreasing concentration of N2 available to the enzyme, which will become less discriminating as the N2 supply is reduced.


For N2 dependent legumes, differences in the symbiont bacteria (Bergersen et al., 1986) and differences in plant age appear to influence nodule and shoot 15N, but not always whole plant δ15N. It is important that 15N isotope discrimination is reliably measured and thus more profitably exploited in N cycle studies at plant and ecosystem scales. Careful attention needs to be paid to standardization of mass spectrometric analysis, particularly since the adoption of continuous flow machines. More accurate rather than more precise measurement of 15N might permit a better understanding of the N exchanges that occur between the atmosphere and N2 fixing bacteria, and between legumes and their microsymbionts.


The author thanks Dr Mark Peoples for comments on a draft of this manuscript and Frank Minchin for advice on diffusion in legume nodules. The author is grateful to the anonymous reviewers for the time they put into their reviews and the improvements resulting from their insight.