Reassessment of major products of N2 fixation by bacteroids from soybean root nodules

Youzhong Li1, Richard Parsons2, David A. Day3 and Fraser J. Bergersen1

School of Biochemistry and Molecular Biology, Faculty of Science, The Australian National University, Canberra, ACT 0200, Australia1
School of Life Sciences, University of Dundee, Dundee DD1 4HN, UK2
Biochemistry Department, The University of Western Australia, Nedlands, WA 6907, Australia3

Author for correspondence: Fraser J. Bergersen. Tel: +61 2 61254291. Fax: +61 2 61250313. e-mail: fraser.bergersen{at}anu.edu.au


   ABSTRACT
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS AND DISCUSSION
REFERENCES
 
NH3/ was the principal product from soybean bacteroids, prepared by various procedures, when assayed in solution in a flow chamber under N2 fixation conditions. In addition, small quantities of alanine were produced (reaching 20% of NH3/ under some conditions). Some 15N was assimilated by bacteroids purified from soybean root nodules on Percoll density gradients and shaken with 15N2 and 0·008 atm O2. Under these conditions, accounted for 93% of the 15N fixed into the soluble fraction. This fraction contained no measurable [15N]alanine. Neither these bacteroids nor those prepared by the previously used differential centrifugation method, when incubated with exogenous alanine under non-N2-fixing conditions, gave rise to NH3 from alanine. Therefore, contamination of bacteroid preparations with enzymes of plant cytosolic origin and capable of producing NH3 from alanine cannot explain the failure to detect [15N]alanine [as reported elsewhere: Waters, J. K., Hughes, B. L., II, Purcell, L. C., Gerhardt, K. O., Mawhinney, T. P. & Emerich, D. W. (1998). Proc Natl Acad Sci USA 95, 12038–12042]. Cell-free extracts of the bacteroids as used in the 15N experiments contained alanine dehydrogenase and were able to produce alanine from pyruvate and . Other experiments with alanine dehydrogenase in extracts of cultured rhizobia and bacteroids are reported and discussed in relation to the 15N experiments. Possible reasons for the differences between laboratories regarding the role of alanine are discussed. It is concluded that NH3 is the principal soluble product of N2 fixation by suspensions of soybean bacteroids ex planta and that should continue to be considered the principal product of N2 fixation which is assimilated in vivo in soybean nodules.

Keywords: alanine, alanine dehydrogenase, ammonia, nitrogen fixation, 15N balance


   INTRODUCTION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS AND DISCUSSION
REFERENCES
 
For many years it has been accepted that the fixation of atmospheric N2 by the root nodules of legumes involved the production of NH3 by the symbiotically diazotrophic bacteria (bacteroids) within the cells of the central tissue of the nodules. It was recognized that, due to a pK of 9·26 for the NH3 equilibrium, at physiological pH only a very small proportion of this nitrogen existed as NH3. The ion would not diffuse across biological membranes, giving rise to concerns about the mechanism by which fixed N2 left bacteroids (e.g. Kahn et al., 1985 ). Subsequently, an channel was identified on the peribacteroid membrane (PBM) (Tyerman et al., 1995 ) and the accepted interpretation persisted. That is, the acidic nature of the symbiosome space acts as an acid trap to form ions, thereby creating a gradient for diffusion of NH3 out of the bacteroids, with subsequently transported to the plant via the PBM channel.

The basis for this interpretation was substantially due to the following experimental observations. (1) NH3 was the earliest 15N-labelled product of N2 fixation when detached root nodules of soybean (Glycine max Merr.; Bergersen, 1965 ) or serradella (Ornithopus sativa L.; Kennedy, 1966a , b ) were incubated for short periods of time in atmospheres containing 15N2. (2) With anaerobically prepared soybean nodule bacteroids (the symbiotic form of Bradyrhizobium japonicum) in microaerobic, shaken assays with 15N2 in the gas phase, the principal product was 15NH3 (Bergersen & Turner, 1967 ). (3) Later, 15NH3 was the principal product of 15N2 fixation by soybean bacteroids in an open, flow-reaction system in which a well-stirred suspension of soybean bacteroids was perfused with solutions containing dissolved O2 and 15N2; no 15N-labelled amino acids were detected. The long-accepted view was challenged by Waters et al. (1998) who found that alanine, not NH3, was the principal 15N-labelled product when N2-fixing soybean bacteroids, purified on anaerobic sucrose density gradients, were supplied with malate and shaken in a gas mixture containing 15N2 and 0·008 atm O2. These authors concluded also that the earlier results arose from the use of bacteroid preparations that were contaminated with cytosolic enzymes from the host tissue, which released NH3 from the primary product, alanine. Allaway et al. (2000) subsequently showed that both [15N]alanine and 15NH3 were produced by bacteroids prepared from nodules of Pisum sp. on anaerobic Percoll density gradients, but that the proportion of these products was altered by the conditions applied. When conditions were optimized for N2 fixation, NH3 was the first and major product formed, but alanine was formed at high bacteroid densities when NH3 accumulated (Allaway et al., 2000 ). These authors also used a mutant strain of Rhizobium leguminosarum, defective in alanine catabolism, to produce N2-fixing nodules on peas. Such plants fix N2, but grow more slowly than plants nodulated by the wild-type, although the basis for this was not determined. Bacteroids from these nodules produced only NH3 in shaken assays. It was concluded that alanine was a secondary product of N2-fixing bacteroids; alanine amino-nitrogen arose from NH3, derived from N2 fixation, which accumulated in the experimental system used. More recently, Atkins & Thumfort (2001) have reported that assimilation of 15N2 by nodulated roots of cowpea produced 15N-labelled products which showed no evidence that alanine was a precursor of the amide groups of glutamine or of the purine ring of ureides (the major N2-fixation product translocated from the nodules).

In the present paper we report experiments with density-gradient-purified soybean bacteroids, which were undertaken to verify the results of Waters et al. (1998) . Verification was not achieved. Instead, we found that NH3 was the sole significant 15N-labelled product of 15N2 fixation accumulated during 30 min in shaken assays with 0·008–0·01 atm O2. Alanine, although sometimes found in low concentrations in flow chamber reactions, was not labelled with 15N in shaken, closed-system experiments. We conclude also that these and earlier results (Bergersen & Turner, 1967 , 1990 ) were not due to contamination with host cytosolic enzymes as suggested by Waters et al. (1998) and that NH3 is the principal product of N2 fixation in soybean.


   METHODS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS AND DISCUSSION
REFERENCES
 
Bacteria.
Bradyrhizobium japonicum strain USDA110c is the strain used in this laboratory for many years. Strain USDA110de was kindly supplied by D. W. Emerich, University of Missouri, USA, and strain SU1014/1 (CB1809) by the SUNFix culture collection, University of Sydney, Australia. Rhizobium leguminosarum bv. viciae strains 3841 and RU1327 (aldA-) were gifts from P. S. Poole, University of Reading. Cultures were maintained on yeast extract/mannitol agar (Dalton 1980 ) and on Brown & Dilworth’s defined liquid medium (Dalton 1980 ), but with 10 mM and succinate, respectively, as nitrogen and carbon sources (Allaway et al., 2000 ) when rhizobia were grown for preparation of cell-free extracts for determination of alanine dehydrogenase activity.

Nodules and bacteroid suspensions.
Nodules aged about 35 days were picked from roots of soybean (Glycine max Merr. cv. Stevens, and sometimes for comparison, cv. Williams) inoculated at planting with B. japonicum strain USDA110c and grown in pots of sand, supplied twice weekly with McKnight’s nutrient solution free of combined nitrogen (Gibson, 1980 ). The nodules were washed in tap water, drained and blotted dry with paper towels and used immediately for preparation of bacteroid suspensions: (i) by the standard method of anaerobic homogenization in phosphate buffer (0·1 M, pH 7·4) containing sucrose (0·2 M), Mg2+ (2 mM), granular PVP (Polyclar) and filtration through Mira-cloth in a closed system under flowing argon [bacteroids were obtained by differential centrifugation and washed as described by Bergersen & Turner (1990) ]; or (ii) by homogenization in the same solution under argon in an anaerobic glove box, followed by separation of bacteroids by anaerobic centrifugation on Percoll (Pharmacia) density gradients and washing, essentially as described by Udvardi et al. (1988 ; cf. Allaway et al., 2000 ). Finally, the bacteroids were dispersed in reaction solution and an aliquot was saved for determination of dry weight after centrifugation and washing with distilled water.

Experimental systems.
Experiments in closed systems were based on those described by Waters et al. (1998) and Allaway et al. (2000) , but some of the buffer constituents and concentrations were different. N2 fixation experiments were run in conical flasks (100 ml) closed with Suba Seals. Reaction solutions (9·0 ml per flask) contained 50 mM MOPS/KOH buffer, pH 7·4, 2 mM DL-malate and 0·2 M sucrose. After inserting the Suba Seals, the flask contents were degassed under vacuum for 10 min with periodic agitation on a manifold fitted with hypodermic needle connections and an Hg manometer (Turner & Gibson, 1980 ). The flasks were flushed twice with Ar and then filled with gas mixtures containing 0·008–0·02 atm O2 and N2 to 1 atm, from a screw-piston reservoir (Turner & Gibson, 1980 ). They were brought to the reaction temperature (26 °C) before injecting 1·0 ml bacteroid suspension containing 3–6 mg (dry wt) bacteroids and shaking at 100 or 150 r.p.m. in a rotary shaker. Reactions were terminated by removing the seals, admitting air to inactivate nitrogenase, immediately chilling on ice and then centrifuging at 0 °C and 10000 g to separate bacteroids from the reaction solution. Products of N2 fixation were measured (below) in the supernatant reaction solution.

Experiments in the flow chamber system (Bergersen & Turner, 1990 ) were performed as described by Li et al. (2001) , but samples of effluent were analysed for NH3, alanine and other amino acids as described below.

Detection of alanine-degrading enzymes adhering to bacteroids.
The presence of enzymes able to deaminate alanine and which adhered to bacteroids was sought in bacteroids prepared by differential centrifugation or by density gradient purification. Dehydrogenases were detected by following the reduction of NAD+ spectrophotometrically at 340 nm in assays containing 100 mM K-CAPS buffer (pH 10·0), alanine (4 mM), NAD+ (3 mM) and 3 mg (dry wt) bacteroids. Also, any alanine-dependent production of was sought in similar aerobic (non-N2-fixing) assays devoid of exogenous NAD+.

Endogenous alanine dehydrogenase.
Alanine dehydrogenase activities were determined in cell-free extracts prepared by disruption of suspensions in a French press. Liquid cultures of various strains of rhizobia were harvested in mid-exponential phase by centrifugation and resuspended in breakage buffer. Bacteroids were prepared from soybean nodules as described above and resuspended in breakage medium (TES/NaOH buffer, pH 7·5, containing 50 mM KCl, 5 mM MgSO4 and 5 mM DTT). The homogenates were clarified by centrifugation at 35000 g at 5 °C for 1 h to yield crude soluble extracts. Alanine dehydrogenase activity was estimated at 25 °C according to Allaway et al. (2000) and Smith & Emerich (1993) . The apparent Km values for pyruvate and were determined.

15N experiments.
15N2 gas was prepared from (15NH4)2SO4 (57·8 atoms% 15N; Isotec, Miamisburg, OH, USA) by oxidation with alkaline hypobromite (Bergersen, 1980 ) and purified by storage over alkaline potassium permanganate solution (to remove oxides of nitrogen) and then over dilute H2SO4 (to remove any residual NH3) (Burris, 1976 ). Contamination of this gas with O2 was measured electrometrically as described below and its concentration adjusted to the desired level by addition of Ar and air after transfer at measured pressure to the screw-piston reservoir (see above). Typically, reaction gas mixtures used contained 0·008 atm O2, 0·5 atm Ar and 0·49 atm N2 (54·6 atoms% 15N) [At the elevation of Canberra, 1 atm is 705–715 mm Hg (94–95·3 kPa).] For experiments in shaken flasks (see above), a blank, unsealed control flask, containing all reaction components except 15N2, was placed on ice at zero time and thereafter treated in the same manner as the reaction flasks. Reactions were terminated as above after shaking for 30 min. The flask contents were analysed for N2 fixation products (see below).

Chemicals and analytical methods.
Except where indicated, all chemicals were of analytical grade purchased from Sigma. NH3 in reaction solutions and in fractions after Kjeldahl digestion and recovery by steam distillation (Bergersen, 1980 ) or by diffusion at pH 11 onto filter paper strips (Whatman No. 1; 15x4 mm) soaked in 0·5 M H2SO4, was determined colorimetrically by the Chaney–Marbach method (Bergersen, 1980 ). Alanine and other amino acids were determined by HPLC analysis after derivitization at the Nucleic Acid and Applied Protein Chemistry Unit, Department of Plant Science, Waite Campus, University of Adelaide, Australia, or at the Ecosystems Research Group, Department of Botany, University of Western Australia.

The concentration of O2 contaminating the 15N2 used in N2 fixation experiments was determined in a sample (20 ml) of the gas collected in a gas-tight syringe and flushed through the previously Ar-flushed gas space (volume 5 ml) above 1 ml of stirred distilled water in the chamber of a pre-calibrated O2 electrode (Rank Bros.). The chamber was sealed before electrode currents were recorded and the zero checked after measurements by addition of a few crystals of sodium dithionite to the water layer.

In the 15N experiments, samples (1·0 ml) of the reaction flask contents were taken for analysis of total nitrogen by Kjeldahl digestion and distillation (Bergersen, 1980 ). The rest of the reaction mixture was separated into the supernatant (soluble) and bacteroid fractions by centrifugation at 10000 g for 10 min. The small quantities of free NH3 in the soluble fraction, being insufficient for 15N analysis, were recovered by diffusion onto paper strips (below) and after drying were supplemented with carrier nitrogen (50 µg) as (NH4)2SO4 of accurately known natural nitrogen abundance. For these analyses, the atoms% 15N before addition of carrier nitrogen (x) was calculated as:

(1)

in which a =µg NH3 nitrogen in the sample before adding carrier, b=µg carrier nitrogen added, c=atoms% 15N measured in the sample plus carrier and d=atoms% 15N of the carrier nitrogen.

Bacteroid total nitrogen, soluble total nitrogen and the total nitrogen of the residue of the soluble fraction after removing NH3 nitrogen, were recovered by Kjeldahl digestion and distillation. Thus the distribution of all 15N incorporated from 15N2 could be determined. NH3 in distillates and in samples of supernatants was recovered by diffusion for >24 h onto strips of filter paper soaked in 0·5 M H2SO4 (see above; Bergersen, 1980 ). The filter paper strips were dried in tin capsules and submitted to total nitrogen and 15N analysis using an ANCA SL stable isotope analytical system (Europa Scientific) in the CSIRO Division of Plant Industry, Canberra, Australia. In addition to these analyses, the presence of [15N]alanine in the soluble fraction was sought by purifying any amino acids present by using chromatography on Sephadex (Pharmacia) SP-25 (Redgwell, 1980 ), lyophilizing eluates and submitting them to GC-MS analysis after derivatization with N-(tert-butyldimethylsilyl)-N-methyltrifluoroacetamide (Allaway et al., 2000 ). The amount of 15N incorporated into alanine was assessed from the ratios of the mass peaks at m/e 260 and 261 (the principal fragments of the derivative of alanine), as follows:

(2)

in which Rs and Rc are, respectively, the ratios mass 261/mass 260 for the mass spectral peaks of the sample (s) and the control (c). The calculation assumes that, apart from nitrogen, the isotopic ratios for all elements present in the fragment are the same in sample and control.


   RESULTS AND DISCUSSION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS AND DISCUSSION
REFERENCES
 
15N experiments
There were two similar 15N2 experiments with Percoll-gradient-purified bacteroids shaken in stoppered flasks using conditions designed to be close to those reported to produce [15N]alanine (Allaway et al., 2000 ; Waters et al., 1998 ). These experiments were conducted 6 months apart; the results of the first (bacteroids from summer-grown nodules) are presented in Table 1. The data for analysis of the total nitrogen of the experimental system indicated that digestion of the morpholino-moiety of the MOPS buffer may have been incomplete. However, this was a constant error and did not affect calculation of the nitrogen-weighted 15N balance in which a matrix of determinations of 15N-labelled fractions and no 15N2 controls was used. Also, as noted elsewhere (Bergersen & Turner, 1967 ; Waters et al., 1998 ) there was a substantial background of endogenous NH3 in these reaction mixtures (perhaps including NH3 adsorbed on the untreated filter paper strips onto which the 15NH3 diffused). These factors contributed to the relatively low 15N enrichment of soluble NH3 (1·2 atoms% excess; Table 1) compared with the enrichment of the 15N2 supplied (54·6 atoms% 15N). The analysis accounted for 79·2 and 75·4%, respectively, of total nitrogen and 15N that was fixed. About 25% of the 15N that was fixed was incorporated in the bacteroids (similar to previous reports; Bergersen & Turner, 1967 , 1990 ). The soluble fraction after removal of the bacteroids contained >50% of the fixed 15N, 93% of which was 15NH3 with enrichment of 1·2 atoms% excess. The residue of the soluble fraction, after removal of NH3, where any amino acids would have been located, contained only 6% of the total 15N excess.


View this table:
[in this window]
[in a new window]
 
Table 1. Incorporation of 15N from 15N2 (54·6 atoms% 15N) by USDA110c bacteroids in 30 min at 26 °C in flasks (100 ml) containing 0·008 atm O2, shaken at 100 r.p.m.

 
The second experiment, using winter-grown nodules, produced an almost identical distribution and enrichment of fixed 15N, but the total amount of 15N fixed was 25% lower, perhaps reflecting lower rates of fixation during winter.

In both experiments alanine was present in the soluble fraction, but the GC-MS analysis revealed no significant excess of 15N in flasks containing bacteroids exposed to 15N2 for 30 min, compared with control flasks (Table 2). Values of R for the standard (Sigma) alanine were lower than those for the control (no 15N2) alanine. This may have been due to different isotopic composition of any of the elements in the Sigma alanine, but the m/e signals were free of signals from other chemicals which, at very low concentration, may alter the m/e signals of control or enriched experimental alanine. Additionally, the data for pure alanine were obtained at slightly higher concentration than for control and experimental alanine. In all cases, alanine was the most significant compound derivitized. It should be noted that had the standard alanine (Sigma; as used by Waters et al., 1998 ) been used as the control at natural 15N abundance, the alanine from the reactions would appear to have been enriched with 15N (0·88 atoms% excess).


View this table:
[in this window]
[in a new window]
 
Table 2. Data for GC-MS analysis of the N-(tert-butyldimethylsilyl)-N-methyltrifluoroacetamide derivatives of alanine from standard alanine (Sigma), control (purified from supernatants of bacteroid reactions in assay flasks in air, stopped in ice at zero time), and purified from soluble fractions of 30 min reactions with 15N2

 
Allaway et al. (2000) used bacteroids from pea root nodules to show that alanine was synthesized from the soluble pool of NH3/ in shaken assays. In our experiments, although NH3/ had by far the greatest atoms% excess 15N of any fraction (Table 1), it was only 1·2 atoms% excess. This may have contributed to our failure to measure significant [15N]alanine by the relatively insensitive GC-MS method (Table 2). Nevertheless, Table 1 shows that there was little room in the 15N balance of the soluble fraction for [15N]alanine. In this experiment, an increment of 2·0±0·8 µg NH3 nitrogen was measured between zero time and 30 min samples. Therefore it was possible to calculate that newly fixed NH3/ had 22·8 (95% confidence limits 1·8–43) atoms% excess 15N. This rather imprecise estimate reflects the analytical difficulty of determining small increments of NH3 nitrogen.

Flow chamber experiments
Waters et al. (1998) stressed the need to use gradient-purified bacteroids which earlier work had shown to be free of host cytosolic enzyme activity and that the critical pO2 for alanine production by density-gradient-purified bacteroids in their shaken, closed system was 0·008 atm. Therefore, we sought alanine production in gradient-purified bacteroids and in bacteroids prepared by differential centrifugation, in flow chamber reactions at low steady-state concentrations of dissolved O2 ([O2 free]). Examples from these experiments are presented in Table 3. It has been shown that N2 fixation into NH3 by bacteroids in the flow chamber system depends on respiration rates (Bergersen, 1997 ). Sucrose-density-gradient-purified bacteroids, as used by Waters et al. (1998) , generally had lower rates of respiration and NH3 production, but yielded no more alanine than bacteroids prepared by other methods (data not presented). Percoll-gradient-purified bacteroids were more active metabolically than sucrose-gradient-purified bacteroids and produced alanine at up to 19% of rates of NH3 production (Table 3), whilst bacteroids prepared by the differential centrifugation method, although producing more NH3 per unit of respiration (Table 3), produced little alanine at rates unrelated to respiration. N2 fixation into NH3 by these bacteroids in shaken assays continued for 30–40 min, declining gradually thereafter (Fig. 1), but in the flow chamber, steady rates were sustained easily for several successive periods of >20 min. Allaway et al. (2000) suggested that failure to detect significant alanine in flow chamber effluents (Bergersen & Turner, 1990 ) could have been due to continuous removal of NH3 from the flow chamber, thus preventing NH3 reaching a concentration needed for significant alanine formation.


View this table:
[in this window]
[in a new window]
 
Table 3. Examples of bacteroid activities during steady states in flow chamber experiments

 


View larger version (12K):
[in this window]
[in a new window]
 
Fig. 1. Production of NH3 and alanine by USDA110c bacteroids in microaerobic shaken assays. (a) Bacteroids prepared by differential centrifugation, with 0·02 atm O2 in the gas phase. (b) Bacteroids prepared on a Percoll density gradient, with 0·008 atm O2 in the gas phase. {bullet}, NH3; {blacktriangleup}, alanine.

 
Alanine in closed shaken bacteroid systems
As noted above, some alanine is present in the soluble fractions from such experiments. However, it is unlikely that alanine production was directly due to N2 fixation because there was little difference between the concentrations of alanine in 1 h assays with Percoll-gradient-purified bacteroids in air (in which nitrogenase activity would have been destroyed) and at a pO2 of 0·008 atm at which production of alanine by soybean bacteroids has been reported to be optimal (Waters et al., 1998 ). In this experiment, alanine comprised, respectively, 72 and 90% of the total of 10 amino acids detected (data not presented). Alanine was always produced at much lower rates than NH3 (Fig. 1). When up to 5 mM was supplied in such experiments, the rates of alanine production doubled, but remained much lower than NH3 production with no added (data not shown). We often found that bacteroid suspensions contained significant amounts of alanine (and NH3; Fig. 1) upon isolation. We suggest that this arises during the isolation procedure. Experiments with extracts from bacteroids prepared by a Percoll density gradient and other methods (see below) indicated that alanine dehydrogenase was present internally in our bacteroid preparations, but produced only limited amounts of alanine in the medium from exogenous malate and . This suggests that B. japonicum USDA110c, maintained in our laboratory, has little potential for alanine formation under N2-fixing conditions.

Alanine dehydrogenase activity
An unpurified, soluble extract of the bacteroid preparation, as used in the 15N experiment (Table 1), contained endogenous alanine dehydrogenase activity [>500 nmol NADH min-1 (mg protein)-1 in the aminating direction], with an apparent Km of 0·9 and 7 mM for pyruvate and , respectively (Table 4). This activity was greater than the activity of extracts of the same strain grown in succinate medium. These results indicated clearly that the enzyme activity had not been lost in the symbiotic state or during the isolation of bacteroids, and the failure to detect substantial alanine excretion from N2-fixing bacteroids did not result from such loss.


View this table:
[in this window]
[in a new window]
 
Table 4. Activities of alanine dehydrogenase in the amination direction in cell-free extracts of strains of B. japonicum grown in succinate broth and in bacteroids of strain USDA110c prepared from fresh nodules on cv. Stevens or cv. Williams

 
Alanine dehydrogenase activities in extracts of cultured B. japonicum strains
This work was undertaken to test whether the failure to detect [15N]alanine as a product of 15N2 fixation (Table 1; Bergersen & Turner, 1990 ) may have been due to the absence of alanine dehydrogenase activity from the bacteria used, either constitutively or due to differences in symbiotic expression. Therefore, a comparison was made of cultures from a variety of sources, some of which had known alanine dehydrogenase phenotypes, grown in succinate medium (Allaway et al., 2000 ), which is known to promote expression of aldA in R. leguminosarum (P. S. Poole, personal communication). Also, a comparison was made between bacteroids from nodules on cv. Williams and cv. Stevens, in case there may have been a difference between aldA expression in the two varieties when nodulated by the same strain of B. japonicum (cf. Stripf & Werner, 1978 ). The results are presented in Table 4. It is clear that in all strains in which alanine dehydrogenase is known to be present and in the strains used in this work (USDA110c) or previously [USDA110de (Waters et al., 1998 ); CB1809 (Bergersen & Turner, 1990 )] alanine dehydrogenase was present. The aldA mutant strain (RU1327; Allaway et al., 2000 ) had negligible activity. The USDA110c bacteroids from nodules of cv. Stevens had high AldA activity when freshly prepared (although those from winter-grown plants had lower activity, causing differences between experiments; Table 4). After storage at -70 °C for several months, bacteroids prepared as for the 15N experiment (Table 1) yielded cell-free extracts with alanine dehydrogenase activity and kinetic values (data not shown) similar to those in Table 4. Bacteroids prepared from USDA110c nodules on cv. Stevens and cv. Williams both had active alanine dehydrogenase (Table 4). These results show that it is most unlikely that the data in Table 1 or of Bergersen & Turner (1990) were due to defects in expression of aldA or to lack of alanine dehydrogenase activity in the bacteroids.

The need for density-gradient bacteroid preparation
Waters et al. (1998) reported that bacteroids prepared by a differential centrifugation method (as used by Bergersen & Turner, 1967 , 1990 ) were contaminated with enzymes originating from the host cells of the nodules. These contaminants may have degraded alanine to produce NH3, thus preventing identification of alanine as a major product of N2 fixation. It seemed to us that Waters et al. (1998) may not have used exactly the same procedure as the other workers and so we tested bacteroid preparations, as used in the present paper, for the presence of such contamination. In neither Percoll-gradient-purified bacteroids nor bacteroids prepared by our fractional centrifugation method was alanine-dependent reduction of NAD+ detected (data not shown). There was a slow production of NH3 (<5 nmol NH3 mg-1 h-1) from bacteroids prepared by fractional centrifugation, but this was not stimulated in the presence of alanine concentrations up to 10 mM. Therefore, contamination of bacteroids with enzymes degrading alanine was not the cause of the failure to detect alanine as a major product of N2 fixation in the present work (Tables 1, 2) or in previous work (Bergersen & Turner, 1967 , 1990 ). The finding suggests that the density gradient methods, which take longer and result in bacteroids with impaired metabolic properties (Table 3), may not be necessary.

Conclusion
We conclude that NH3 is the principal product of N2 fixation by bacteroids from soybean nodules under a range of different conditions ex planta, including those used by Waters et al. (1998) . The reason for this difference between the two laboratories could not be determined from our experiments, but the possibility remains that sucrose-density-gradient-purified bacteroids (Waters et al., 1998 ), in closed shaken assays, differ from those purified on Percoll density gradients.


   ACKNOWLEDGEMENTS
 
The authors thank Richard Phillips and members of the analytical services group of the CSIRO Division of Plant Industry, Canberra, Australia, for the 15N analyses, and J. Lahnstein (Nucleic Acid and Applied Protein Chemistry Unit, Department of Plant Science, Waite Campus, University of Adelaide, Australia) and C. Warren (Department of Botany, University of Western Australia) for amino acid analyses. Gifts of strains of B. japonicum were received from D. W. Emerich (University of Missouri, Columbia, MO, USA) and I. R. Kennedy (University of Sydney, NSW, Australia) and R. leguminosarum from P. S. Poole (University of Reading). Seeds of the soybean cv. Williams were a gift from Dr P. Lawrence (Australian Tropical Crops Genetic Resource Centre, Biloela, Qld, Australia). The research was supported by a grant from the Australian Research Council to D.A.D. and F.J.B. Youzhong Li acknowledges the award of an International Postgraduate Research Scholarship and an ANU PhD scholarship. F.J.B. thanks the Australian National University for support as a Visiting Fellow.


   REFERENCES
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS AND DISCUSSION
REFERENCES
 
Allaway, D., Lodwig, E. M., Crompton, L. A., Wood, M., Parsons, R., Wheeler, T. R. & Poole, P. S. (2000). Identification of alanine dehydrogenase and its role in mixed secretion of ammonium and alanine by pea bacteroids. Mol Microbiol 36, 509-515.

Atkins, C. A. & Thumfort, P. P. (2001). Assimilation of fixed N in a ureide-forming symbiosis. In Proceedings of the 13th International Conference on Nitrogen Fixation, Hamilton, Ontario, July 2001. Dordrecht: Kluwer.

Bergersen, F. J. (1965). Ammonia – an early stable product of nitrogen fixation by soybean root nodules. Aust J Biol Sci 18, 1-9.

Bergersen, F. J. (1980). Measurements of nitrogen fixation by direct means. In Methods for Evaluating Biological Nitrogen Fixation , pp. 65-110. Edited by F. J. Bergersen. Chichester:Wiley.

Bergersen, F. J. (1997). Physiological and biochemical aspects of nitrogen fixation by bacteroids in soybean nodule cells. Soil Biol Biochem 29, 875-880.

Bergersen, F. J. & Turner, G. L. (1967). Nitrogen fixation by the bacteroid fraction of breis of soybean root nodules. Biochim Biophys Acta 141, 507-515.[Medline]

Bergersen, F. J. & Turner, G. L. (1990). Bacteroids from soybean root nodules: respiration and N2-fixation in flow-chamber reactions with oxyleghaemoglobin. Proc R Soc Lond B 238, 295-320.

Burris, R. H. (1976). Nitrogen fixation by blue-green algae of the Lizard Island area of the Great Barrier Reef. Aust J Plant Physiol 3, 41-51.

Dalton, H. (1980). The cultivation of diazotrophic microorganisms. In Methods for Evaluating Biological Nitrogen Fixation , pp. 13-64. Edited by F. J. Bergersen. Chichester:Wiley.

Gibson, A. H. (1980). Methods for legume in glasshouse and controlled environment cabinets. In Methods for Evaluating Biological Nitrogen Fixation , pp. 140-184. Edited by F. J. Bergersen. Chichester:Wiley.

Kahn, M. L., Kraus, J. & Somerville, J. E. (1985). A model of nutrient exchange in the Rhizobium – legume symbiosis. In Nitrogen Fixation Research Progress , pp. 193-199. Edited by H. J. Evans, P. J. Bottomely & W. E. Newton. Dordrecht:Nijhoff.

Kennedy, I. R. (1966a). Primary products of symbiotic nitrogen fixation I. Short-term exposures of serradella nodules to 15N2. Biochim Biophys Acta 130, 285-294.[Medline]

Kennedy, I. R. (1966b). Primary products of symbiotic nitrogen fixation II. Pulse-labelling of serradella nodules with 15N2. Biochim Biophys Acta 130, 295-303.[Medline]

Li, Y., Green, L. S., Holtzappffel, R., Day, D. A. & Bergersen, F. J. (2001). Supply of O2 regulates demand for O2 and uptake of malate by N2-fixing bacteroids from soybean nodules. Microbiology 147, 663-670.[Abstract/Free Full Text]

Redgwell, R. J. (1980). Fractionation of plant extracts using ion-exchange Sephadex. Anal Biochem 107, 44-50.[Medline]

Smith, M. T. & Emerich, D. W. (1993). Alanine dehydrogenase from soybean nodule bacteroids, kinetics, mechanism and pH studies. J Biol Chem 268, 10746-10753.[Abstract/Free Full Text]

Stripf, R. & Werner, D. (1978). Differentiation of Rhizobium japonicum, II. Enzymatic activities in bacteroids and plant cytoplasm during the development of nodules of Glycine max. Z Naturforsch 33C, 373-381.

Turner, G. L. & Gibson, A. H. (1980). Measurement of nitrogen fixation by indirect means. In Methods for Evaluating Biological Nitrogen Fixation , pp. 111-138. Edited by F. J. Bergersen. Chichester:Wiley.

Tyerman, S. D., Whitehead, L. F. & Day, D. A. (1995). A channel-like transporter for on the symbiotic interface of N2-fixing plants. Nature 378, 629-632.

Udvardi, M. K., Salom, C. L. & Day, D. A. (1988). Transport of L-glutamate across the bacteroid membrane but not the peribacteroid membrane from soybean root nodules. Mol Plant–Microbe Interact 1, 250-254.

Waters, J. K., Hughes, B. L.II, Purcell, L. C., Gerhardt, K. O., Mawhinney, T. P. & Emerich, D. W. (1998). Alanine, not ammonia, is excreted from N2-fixing soybean nodule bacteroids. Proc Natl Acad Sci USA 95, 12038-12042.[Abstract/Free Full Text]

Received 31 December 2001; revised 10 February 2002; accepted 18 February 2002.