An overview of the metabolic differences between Bradyrhizobium japonicum 110 bacteria and differentiated bacteroids from soybean (Glycine max) root nodules: an in vitro 13C- and 31P-nuclear magnetic resonance spectroscopy study


  • Pierre Vauclare,

    Corresponding author
    1. Extremophiles and Large Macromolecular Assemblies (ELMA) group, Institut de Biologie Structurale J.-P. Ebel, UMR 5075 CEA-CNRS-UJF-PSB, Grenoble Cedex, France
    • Département de Biologie Moléculaire Végétale (DBMV), Bâtiment Biophore, Lausanne, Switzerland
    Search for more papers by this author
  • Richard Bligny,

    1. Laboratoire de Physiologie Cellulaire Végétale, Unité Mixte de Recherche 5168, Institut de Recherche en Technologie et Sciences pour le Vivant, CEA, Grenoble cedex 9, France
    Search for more papers by this author
  • Elisabeth Gout,

    1. Laboratoire de Physiologie Cellulaire Végétale, Unité Mixte de Recherche 5168, Institut de Recherche en Technologie et Sciences pour le Vivant, CEA, Grenoble cedex 9, France
    Search for more papers by this author
  • François Widmer

    1. Département de Biologie Moléculaire Végétale (DBMV), Bâtiment Biophore, Lausanne, Switzerland
    Search for more papers by this author

Correspondence: Pierre Vauclare, Extremophiles and Large Macromolecular Assemblies (ELMA) group, Institut de Biologie Structurale J.-P. Ebel, UMR 5075 CEA-CNRS-UJF-PSB, 41, Rue Jules Horowitz, 38027 Grenoble Cedex 1, France. Tel.: +33 438789569; fax: +33 438785494; e-mail:


Bradyrhizobium japonicum is a symbiotic nitrogen-fixing soil bacteria that induce root nodules formation in legume soybean (Glycine max.). Using 13C- and 31P-nuclear magnetic resonance (NMR) spectroscopy, we have analysed the metabolite profiles of cultivated B. japonicum cells and bacteroids isolated from soybean nodules. Our results revealed some quantitative and qualitative differences between the metabolite profiles of bacteroids and their vegetative state. This includes in bacteroids a huge accumulation of soluble carbohydrates such as trehalose, glutamate, myo-inositol and homospermidine as well as Pi, nucleotide pools and intermediates of the primary carbon metabolism. Using this novel approach, these data show that most of the compounds detected in bacteroids reflect the metabolic adaptation of rhizobia to the surrounding microenvironment with its host plant cells.


With around 18 000 species, Leguminosae are considered as the largest plant family and play an important role in the earth ecosystem. By developing a nitrogen-fixing root nodules, leguminous have established a symbiotic relationship with specific soil bacteria known as rhizobia. Rhizobia are Gram-negative soil bacteria belonging to genera Rhizobium, Sinorhizobium, Mesorhizobium, Phylorhizobium, Azorhizobium and Bradyrhizobium. They have the capacity to live either as free-living bacteria or as nitrogen-fixing bacteroids inside a specialized root organ of Leguminosae called nodules to establish a reciprocal profitable metabolic exchange with its host plant (Young & Haukka, 1996; Brewin, 2004). Whereas bacteroids convert atmospheric dinitrogen (N2) into ammonia for the benefit of the host, leguminous plants in turn supply microsymbionts with photosynthate (Prell & Poole, 2006; Cooper, 2007). To reach this symbiotic relationship in nodules, plant roots have to be first colonized by rhizobia, which induce nodule organogenesis. In a second time, when rhizobia are uptaken by endocytosis into the cytoplasm of nodule-infected cell, they are enclosed within a peribacteroid membrane (PBM) forming a specialized cytoplasmic organelle-like structure termed symbiosome in which bacteria divide and differentiate into bacteroids (Whitehead & Day, 1997; Oke & Long, 1999; Perret et al., 2000; Brewin, 2004). This differentiation process from free-living bacteria into bacteroids implies a global change in gene and protein expression patterns (Sarma & Emerich, 2005, 2006; Pessi et al., 2007; Tsukada et al., 2009; Delmotte et al., 2010). In particular, Pessi et al. (2007) have observed by analysing transcriptomic data of Bradyrhizobium japonicum bacteroids that 692 genes were induced, whereas 2086 were downregulated compared with free-living cells grown in aerobic culture. This overall slowdown of the bacteroid metabolism seems to be consistent with its principal function in nodules: satisfy the requirement of host plant in ammonia via a symbiotic mechanism (for a review, see Patriarca et al., 2002). For this reason, because differentiation process affects number of gene expression in rhizobia, enhanced understanding of differentiation of free-living cells into bacteroids by studying their ‘metabolite fingerprint’ is a key analytic tool. In this report, using 13C- and 31P-nuclear magnetic resonance (NMR), we present an overview of soluble metabolites of free-living B. japonicum bacteria and of its symbiotic differentiated state to investigate principal metabolic differences.

Materials and methods

Bacterial strains and plant growth conditions

Bradyrhizobium japonicum wild-type strain 110spc4 (Regensburger & Hennecke, 1983) was grown aerobically in sterilized liquid culture [4% bacto peptone (p/v), 1 mM KH2PO4, 2 mM MgSO4·7H2O, pH 6.8] with spectinomycin (100 μg mL−1) for 6 days at 28 °C (DO600 = 0.161; 3.2 108 cells mL−1). Then, cultivated bacteria were used on the one hand to inoculate soybean seeds (Glycine max L., var Mapple arrow; Schweizer Samen AG, Thoune, Switzerland) and on the other hand to be analysed by NMR. The nodulation efficiency was improved by inoculating each soybean seeds (7 days after imbibitions) with 108–109 centrifuged B. japonicum strain diluted in fresh phosphate-buffered saline medium (0.14 M NaCl, 2.5 mM KCl, 4 mM Na2HPO4, 2 mM KH2PO4, pH 7.4). Plants were grown in a greenhouse under temperature of 20°/18.5 °C (day/night) and 16-h photoperiod. Mature nodules were collected 28–32 days of growth after inoculation. For NMR measurements, B. japonicum cells were harvested and the pellet was stored at −80 °C after washing in phosphate buffer.

Purifications of bacteroids

Bacteroids were purified using a modified method of Reibach et al. (1981). One hundred and twenty grams of mature red nodules of soybean (Glycine max) was gently ground with a mortar and a pestle in extraction buffer [0.15M NaCl, 50 mM KH2PO4, pH 7.6, 0.5% (w/v) PVP 35 000, 0.1% (w/v) BSA, 1 mM EDTA, pH 8.0, 2 mM DTT] at 4 °C. The homogenate was centrifuged at 750 g during 10 min at 4 °C and then the supernatant was layered onto a Percoll gradient (70% Percoll, 0.15M NaCl, 50 mM KH2PO4). After centrifugation at 48 000 g for 60 min at 4 °C, bacteroids were collected, diluted with 5 volumes of cold washing buffer (0.15M NaCl, 50 mM KH2PO4, 1 mM EDTA, pH 8.0) and centrifuged at 10 000 g for 15 min at 4 °C. The pellet was suspended in washing buffer, and the purified bacteroids pellet was stored at −80 °C after centrifugation at 10 000 g for 10 min at 4 °C.

In vitro NMR experiments

For perchloric acid (PCA) extracts, equal amounts (4.8 g each) of frozen B. japonicum cells and purified bacteroids were ground to a fine powder in liquid nitrogen and prepared according to the method described by Gout et al. (2000). Briefly, extracts were lyophilized, dissolved in 13% 2H2O, neutralized to pH 7.5 and analysed using an NMR spectrophotometer (AMX 400, Bruker, Bilerica, MA). Data acquisition conditions used for 13C- and 31P-NMR were as described by Vauclare et al. (2010). Peak identification was performed using NMR metabolites database of the laboratoire de Physiologie Cellulaire Végétale in Grenoble (France). The concentration of each identified peaks was estimated using referenced compounds added during PCA extraction such as maleate and methylphosphonate for 13C- and 31P-NMR spectra, respectively.

Results and discussion

To determine the metabolite profiles for cultures grown and bacteroid forms of B. japonicum, two experiments were conducted: the first one was used to analyse by 13C- and 31P-NMR aerobically grown B. japonicum cells at the growth stage used to inoculate soybean seeds and the second, the differentiated form of rhizobia in nodules. It is important to notify that all the procedures for bacteroids purification were performed at 4 °C to prevent cells' metabolic activity because the principal aim of this work consists to highlight the metabolic adaptation of B. japonicum to survive within the soybean root nodule cells. Moreover, replicate experiments with bacteroids purified from soybean exactly at the same developmental stages were very difficult to realize (particularly for the quantification of metabolites). For this reason, results analysis presented in the following sections corresponds to NMR spectra that clearly illustrate the most drastic differences between the metabolome data of purified bacteroids and of free-living B. japonicum cells (Figs 1 and 2, Table 1).

Table 1. Metabolic profile of Bradyrhizobium japonicum cells grown under aerobic conditions and of purified bacteroids. PCA extracts were prepared from 4.8 g fresh weight of free-living B. japonicum cells and of bacteroids. Metabolites were identified and quantified using maleate and methylphosphonate as internal standards for 13C- and 31P-NMR analyses, respectively
Metabolites B. japonicum
  1. a

    (PCCG?): hypothetical cyclic oligosaccharide named phosphocholine-substituted β-1,3; 1,6 cyclic glucan.

  2. Values are expressed as μmol g−1 fresh weight (FW) of free-living cells or bacteroids.

  3. Result are given as mean ± SD (n = 3).

  4. n.d., not detected.

Sucrose1.2 ± 0.23n.d.
Trehalose0.52 ± 0.1046.4 ± 1.3
Myo-inositoln.d.1.9 ± 0.38
Glutamaten.d.1.82 ± 0.36
Homospermidine0.52 ± 0.1041.82 ± 0.36
Pi8.1 ± 1.629.2 ± 5.2
Glucose 6-P0.14 ± 0.030.6 ± 0.1
Mannose 6-Pn.d.0.14 ± 0.03
Glycerol 3-Pn.d.1 ± 0.2
PGAn.d.0.18 ± 0.04
a(PCCG?)0.63 ± 0.13n.d.
GPI0.36 ± 0.070.21 ± 0.04
GPC0.1 ± 0.020.26 ± 0.05
AMP0.63 ± 0.131.6 ± 0.3
GMP0.05 ± 0.010.3 ± 0.06
CMP0.11 ± 0.020.44 ± 0.09
UMP0.12 ± 0.020.58 ± 0.1
ADP0.09 ± 0.020.66 ± 0.1
UDPn.d.0.14 ± 0.03
ATPn.d.0.11 ± 0.02
NADP+n.d.0.052 ± 0.01
NADPHn.d.0.063 ± 0.01
Figure 1.

Representative proton-decoupled 13C-NMR spectra of PCA extracts prepared from 4.8 g fresh materials of Bradyrhizobium japonicum cells grown under aerobic conditions (a) and of purified soybean bacteroids (b). In vitro spectra were recorded on a NMR spectrometer (AMX 400; Bruker, Bileria, MA) in a multinuclear probe tuned at 100.9 MHz. Spectra were the result of 3600 transients with radio frequency pulses (19 μs) at 6-s intervals and a 90° pulse angle (40). Peak assignments are as follows: Glu; Hsp; myo-inositol (i); sucrose (S); trehalose (T); not identified (n.i.).

Figure 2.

Representative 31P-NMR spectra of PCA extracts prepared from 4.8 g of fresh materials of Bradyrhizobium japonicum cells grown under aerobic conditions (a) and of purified soybean bacteroids (b). In vitro spectra were recorded on a NMR spectrometer (AMX 400; Bruker) in a multinuclear probe tuned at 162 MHz. Spectra were the result of 4096 transients with radio frequency pulses (15 μs) at 3.6-s intervals and a 70° pulse angle (40). Peak assignments are as follows: Glc-6-P; Gly-3-P; GPC; GPI; mannose 6-phosphate (Man-6-P); 3-phosphoglycerate (PGA); Pi. *: phosphorous head group of a hypothetical cyclic oligosaccharide (PCCG?).

Measurement of carbohydrate pools

One of the major differences observed in 13C-NMR spectra is the increase of trehalose by 92% in bacteroids which is the only soluble store carbohydrates detected (Fig. 1). Indeed, three independent trehalose biosynthetic pathways have been characterized in B. japonicum, and the trehalose present in soybean nodules is specifically synthesized by bacteroids (Streeter & Gomez, 2006). With a concentration of 6.4 μmol g−1 FW of bacteroids (Table 1), trehalose could be considered as the only store carbohydrates whose concentration is higher inside bacteroids than in the cytosol of soybean nodular cells (3.24 μmol g−1 FW). Moreover, if we consider that bacteroids occupy around 15% of soybean nodules volume (Lin et al., 1988), the bacteroids' trehalose pool represents approximately 23% of that measured previously in nodule samples (4.2 μmol g−1 FW) (Vauclare et al., 2010). This suggests that about 77% of the trehalose synthesized by the bacteroids was recovered in the nodule cytosol, a result in agreement with Streeter (1985). Moreover, the fact that trehalose concentration in bacteroids is twelve times higher than it is in cultivated cells testifies to the significance of this unreactive disaccharide in bacteroids. Indeed, because legume rhizobia symbioses are extremely sensitive to osmotic stress, which results in a significant decrease in azote fixation capacity, it has been demonstrated that trehalose plays a crucial role as osmoprotectant to enhance survival during root hairs' rhizobia infection but also under a large variety of abiotic stresses encountered in nodules like desiccation, salt or osmotic stress (Zahran, 1999; Brechenmacher et al., 2010; Sugawara et al., 2010). For these reasons and because bacteroids in planta can sense osmotic variations, we could suggest that the overaccumulation of osmolytes detected in bacteroids could correspond to adaptive responses against osmotic stress, probably using distinct metabolic pathways from those present in free-living rhizobia, which are less sensitive to osmotic stress (Zahran, 1999; Boscari et al., 2006). Moreover, although Hoelzle & Streeter (1990) hypothesized that the trehalose synthesis may be also induced in nodules by the microaerobic environment, further experiments must be carried out to establish a link between trehalose production by B. japonicum and low oxygen tension in soybean nodules. Among the other detected soluble carbohydrates, sucrose is present at low concentration, specifically in free-living cells (Fig. 1a, Table 1). This result may be explained by the inability of bacteroids to transport and catabolize sucrose, contrary to free-living Rhizobium (Udvardi & Day, 1997). Close to the chemical shift of trehalose, we have identified in the enlarged portion of bacteroid spectra a polyol assigned to myo-inositol, which could be used as osmoprotectant (Fig. 1b, Table 1). Because of a large difference in concentration in infected nodule cells, myo-inositol was probably imported via a passive mechanism (Vauclare et al., 2010). Nevertheless, it could be toxic for the bacteroids in case of accumulation. Indeed, mutant of Sinorhizobium freedii strain, which is unable to catabolize myo-inositol, revealed some bacteroids structural alteration as a consequence a lower capacity to reduce nitrogen (Jiang et al., 2001). For this reason, we could suggest that the absence of several myo-inositol catabolic gene in B. japonicum species (Boutte et al., 2008) could be one of the reasons why B. japonicum strains are considered as poor nitrogen fixers (Jiang et al., 2001), probably because of the excessive myo-inositol accumulation.

A multifunctional role of glutamate

Glutamate (Glu) (1.82 μmol g−1 FW) is another important compound detected in bacteroids spectra (Fig. 1b, Table 1). Considered as the major free amino acid in bacteroids, Glu accumulation results in the inhibition of α-ketoglutarate dehydrogenase by the high NADH/NAD ratio due to the O2 limitation in nodules (Salminen & Streeter, 1990; Green & Emerich, 1997). However, Glu could be partially metabolized to feed respiration but also it contributes to the osmotic adjustment (Salminen & Streeter, 1987; Fujihara & Yoneyama, 1994). Furthermore, Glu can serve as a good substrate for the synthesis of homospermidine (Hsp) via the ornithine decarboxylase pathway (Fujihara, 2009). Indeed, our analyses revealed the presence of four peaks at 47.8, 39.9, 24.9 and 23.6 ppm, corresponding to Hsp (Fig. 1). Three times more abundant in bacteroids (1.82 μmol g−1 FW) than in cultivated rhizobia (0.52 μmol g−1 FW) (Fig. 1, Table 1), this unusual polyamine may be produced specifically by bacteroids during the symbiosis because root cells of host legumes do not contain Hsp and an activity of Hsp synthase was detected in bacteroids (Abe, 1994; Fujihara et al., 1994). However, although the function of Hsp is still unclear in bacteroids, its detection in a wide variety of nitrogen-fixing rhizobia (Fujihara, 2009) suggests that it might play a role in relation to N2 fixation. One can suggest that, in addition to being an important storage form of combined nitrogen and probably a competitor of ureide pathway for nitrogen (Sarma & Emerich, 2005), Hsp might buffer pH variations, osmotic stress and the toxic effect of oxygen described in fast-growing rhizobia and Escherichia coli to optimize nitrogen fixation (Chattopadhyay et al., 2003; Fujihara, 2009).

Modifications of phosphorylated metabolites

The most abundant metabolite measured by 31P-NMR in both samples was the soluble inorganic phosphate (Pi) with concentration being 8.1 μmol g−1 FW in free-living B. japonicum cells and 29.2 μmol g−1 FW in bacteroids (Fig. 2 and Table 1). Indeed, Pi is massively imported via a phosphate carrier (ABC family of proteins), which is induced during the symbiosis. However, the reason for the presence of a large pool of Pi in B. japonicum cells and in bacteroids is quite different (Sarma & Emerich, 2005; Pessi et al., 2007). In our opinion, it is likely that free rhizobium use Pi as reserve of phosphate to survive under low-phosphorus conditions as well as during rhizosphere colonization. Concerning the high concentration of Pi in bacteroids, it may reflect a strong involvement of Pi in the metabolism of bacteroids. Indeed, nitrogen fixation consumes a large amount of energy involving phosphorylated intermediates. Nevertheless, if we calculate the adenylate energy charge [AEC = (ATP + 1/2 ADP)/(ATP + ADP + AMP)] of bacteroids, it reaches approximately 0.19 (Table 1), which is slightly lower than that reported by Tajima & Kouzai (1989) (AEC: 0.37). This result reflects typical hypoxic conditions (AEC ≤ 0.75), which may limit N2 fixation. Moreover, the relatively low ratio of NADP+/NADPH (~1.2) attests the quite low rate of oxidation of NADPH compared to its production by the NADP-malic enzyme and by the pentose phosphate pathway. Concerning the low energy charge observed in free-living B. japonicum cells (no ATP detected) (Fig. 2a, Table 1), it could be attributed to the growth phase of the cells and to the relative permeability of rhizobia membrane to protons, thus lowering the efficiency of energy metabolism (Ratcliffe et al., 1983). This was confirmed by the concentration of glucose 6-P (Glc-6-P), which was fourfold lower in B. japonicum cells (0.14 μmol g−1 FW) than in symbiotic cells (0.6 μmol g−1 FW) (Table 1). This suggests that soybean root infection did not require rhizobium cells with high energetic potential. Moreover, among identified P-compounds, phosphoglyceric acid (PGA), mannose 6-P and glycerol 3-P (Gly-3-P) were specifically detected in bacteroids spectra (Fig. 2b and Table 1). A number of previous studies on genome and protein expressions in symbiotic bacteria have reported a repression of some genes and the detection of only three enzymes involved in the glycolysis, whereas key genes encoding for enzymes of gluconeogenesis and for glycogen synthesis were specifically induced (Sarma & Emerich, 2005; Pessi et al., 2007; Delmotte et al., 2010). It could be argued that the increase in Glc-6-P and PGA observed in bacteroids could be attributed to the synthesis of glycogen because of the presence of excess of intermediate of the tricarboxylic cycle. Alternatively, the presence of Gly-3-P (or mannose 6-P) may contribute to feed gluconeogenesis pathway via Glc-6-P synthesis because a gene encoding to the glycerol 3-P dehydrogenase was especially expressed in bacteroids (Pessi et al., 2007). In addition, to explain the accumulation of Gly-3-P in bacteroids, we have to refer to Dean et al. (1999) who demonstrated that nodulin 26 (a major protein component of the PBM) facilitates glycerol fluxes between the host plant and the endosymbiont. Because nodules are very sensitive to osmotic stress (Hunt & Layzell, 1993), the huge accumulation of Gly-3-P in bacteroids could play an osmoregulatory role to optimize the symbiosis between nodule cells and rhizobia. Concerning the glycerophosphodiesters mentioned in Fig. 2, we observed a threefold increase in glycerylphosphoryl-choline (GPC) in bacteroids (Fig. 2b and Table 1). GPC could be involved in the synthesis of glycine betaine, an effective osmoprotectant present in bacteroids (Fougère & Le Rudulier, 1990; Mandon et al., 2003), but not detected in 13C-NMR spectra (Fig. 1). More likely, we can suggest that GPC could be involved in the turnover of phosphatidylcholine (PC) (Van der Rest et al., 2002). Considered as the major membrane phospholipid in the family of Rhizobiaceae, PC is required to optimize soybean host plant–B. japonicum symbiosis (Minder et al., 2001). Interestingly, in relation to GPC metabolism, we noted that contrary to soybean nodule cells (Vauclare et al., 2010), no phosphorylcholine was detected in bacteroids. This suggests that this compound was only present in eukaryotic cells, in which hydrolytic activity of phospholipids occurs (Vauclare et al., 2010). Surprisingly, in Fig. 2a, we observed an accumulation of glycerylphosphoryl-inositol (GPI), whereas GPC was virtually undetectable. As for Saccharomyces cerevisiae (Patton et al., 1995), we could suggest that the presence of GPI in free-living B. japonicum cells represents a pathway of phosphatidylinositol metabolism. In fact, Tang & Hollingsworth (1998) have demonstrated that whereas phosphatidylcholine was no longer produced, this typical plant phospholipid was specifically synthesized in B. japonicum cultivated under low-oxygen concentration, one of the factors known to induce conversion of rhizobium into bacteroids in nodules. The significance of this result is that under conditions which mimic nodules microaerobic environments, free rhizobia change their membrane compositions probably to avoid being rejected by the plant host. Finally, we detected at 2.24 ppm in free-living rhizobia spectra a huge unidentified peak that we could attributed to the phosphocholine group of a cyclic oligosaccharide in B. japonicum (Fig. 2a and Table 1). Indeed, Rolin et al. (1992) and Pfeffer et al. (1994) detected in B. japonicum, using in vivo NMR spectroscopy, a peak corresponding to the phosphocholine group of a cyclic β-glucan named phosphocholine-substituted β-1,3; 1,6 cyclic glucan (PCCG). Produced in very large quantity after 4 days of growth in standard medium at low osmotic pressure, this unusual cyclic oligosaccharide is not synthesized in high osmotic pressure, a condition encountered in the cytosol of the host plant cells (Pfeffer et al., 1994). For this reason, because no peak at 2.24 ppm was detected in bacteroids spectra, we could speculate that the signal detected in free-living cells could correspond to PCCG or to a similar cyclic oligosaccharide playing a role in the osmoadaptation of rhizobia. Further biochemical analyses will help us to confirm the identity and the function of this compound in B. japonicum.

In summary, 13C- and 31P-NMR profiles of bacteroids present quantitative and qualitative differences from that of their vegetative state, which reflect some physiological adaptations of rhizobia to the plant host environment (Fig. 3). However, we cannot exclude that this quantitative difference observed may be due to the ability of rhizobia to grow and multiply more rapidly in a complex medium than in nodules, a property that could explain the low amount of storage metabolites in free-living rhizobia. At this point, further studies could be carried out to follow metabolic profiles of B. japonicum grown with different carbon sources and at different growth phases.

Figure 3.

Schematic representation of major differences between free-living Bradyrhizobium japonicum (a) and bacteroids (b). White police in black box represent specific metabolites identified by NMR in free-living rhizobia and in its symbiotic state. Black polices were metabolites detected by NMR. Metabolites in grey are not detected by NMR. BM, bacteroid membrane;DHAP, dihydroxyacetone phosphate;Fru-6-P, fructose 6-phosphate;GAP, glyceraldehyde 3-phosphate;Glc, glucose;Man-6-P, mannose 6-phosphate;OAA, oxaloacetate;PI, phosphatidylinositol;PGA, 3-phosphoglycerate;RM, Rhizobium membrane;TCA cycle, tricarboxylic acid cycle.


We are grateful to Dr Henry-Sung Kim (Institut de Biologie Structurale, Grenoble) for his work in correction of English text in the manuscript. We also thank Josiane Bonetti for her excellent bibliographic assistance. This work was supported by grants from the University of Lausanne (Switzerland).