Low endotoxic activity of lipopolysaccharides isolated from Bradyrhizobium, Mesorhizobium, and Azospirillum strains


Iwona Komaniecka, Department of Genetics and Microbiology, Maria Curie-Sklodowska University, Akademicka 19, 20-033 Lublin, Poland.
Tel: +48 81 537 59 81; fax: +48 81 537 59 59; e-mail: ikoma@hektor.umcs.lublin.pl


The endotoxic activities of lipopolysaccharides (LPS) isolated from different strains of rhizobia and rhizobacteria (Bradyrhizobium, Mesorhizobium, and Azospirillum) were compared to those of Salmonella enterica sv. Typhimurium LPS. The biological activity of all the examined preparations, measured as Limulus lysate gelation, production of tumor necrosis factor (TNF), interleukin-1β (IL-1β), and interleukin-6 (IL-6), and nitrogen oxide (NO) induction in human myelomonocytic cells (line THP-1), was considerably lower than that of the reference enterobacterial endotoxin. Among the rhizobial lipopolysaccharides, the activities of Mesorhizobium huakuii and Azospirillum lipoferum LPSs were higher than those of the LPS preparations from five strains of Bradyrhizobium. The weak endotoxic activity of the examined preparations was correlated with differences in lipid A structure compared to Salmonella.

List of Abbreviations: 




E. coli

Escherichia coli






interleukin 1β


interleukin 6


Limulus amebocyte lysate (gelation assay)




Mesorhizobium; MD-2,myeloid differentiation factor 2


nitrogen oxide


phorbol 12-myristate 13-acetate










standard deviation




human myelomonocytic cell line


toll-like receptor 4


tumor necrosis factor

Soil bacteria belonging to the rhizobium lineage are able to fix atmospheric nitrogen during symbiosis with legume plants. Bacteria from the genus Bradyrhizobium induce nitrogen-fixing nodules on the roots of cultivated (Glycine max and Glycine soya) and wild-growing legumes (1, 2). M. huakuii induces the formation of nodules on the roots of Astragalus sinicus (3). A. lipoferum represents plant-growth-promoting rhizobacteria which colonize the root surface and are not able to penetrate root cells. They live in association with roots of grasses, cereals, and other monocotyledonous plants (4, 5).

Lipopolysaccharide, as an integral component of the cell walls of Gram-negative bacteria, plays an essential role in the proper development of symbiotic relationships (6). LPS, together with Omp proteins, is responsible for the asymmetric structure and semi-permeability of outer membranes. This is important for the appropriate morphogenesis and functionality of bacteroids, endosymbiotic forms of rhizobia which perform nitrogen fixation (7). LPS may play a role in the protection of rhizobia against plant defense response mechanisms. Suppression of systemic acquired resistance or hypersensitivity reaction has been shown during infection of plant tissues by microsymbionts (8–10).

Most pathogenic bacteria possess LPSs displaying endotoxic activity against host organisms. Lipid A, the part of LPSs that anchors the whole macromolecule in the outer membrane, is the centre of endotoxicity. The fine structure of enterobacterial lipid A has been identified as a glycolipid comprised of a β-(1,6)-linked glucosaminyl disaccharide substituted by two phosphate groups at positions C-1 and C-4 and six fatty acid residues with two acyloxyacyl moieties with a distinct location (Fig. 1) (11, 15, 16).

Figure 1.

The structure of lipids A from:Salmonella enterica sv. Typhimurium (11), E. coli (11), A. lipoferum (12) M. huakuii (acc. to 13) and B. elkanii (14).

The activity of lipid A in LPSs is a result of its ability to recognize TLR4 on the surface of macrophages and endothelial cells. TLR4, acting in association with MD-2, recognizes LPS, which is extracted from the bacterial membrane and transferred to the TLR4-MD-2 complex by two accessory proteins: LPS binding protein and cluster of differentiation 14 (17,18). Activation of TLR4 receptors initiates a signaling cascade, resulting in the biosynthesis by macrophage cells of diverse mediators of inflammation (TNF, IL-1β or IL-6) (11). In the case of excessive release of cytokines, either clearing of local infection or a septic shock reaction may take place. It has been proved that the presence of phosphate groups and two acyloxyacyl moieties at distinct positions is needed for the activation of TLR4 receptors followed by the triggering of an endotoxin response in human immune cells (16, 19). Lipids A, which are significantly different from enterobacterial lipid A, are usually weakly toxic or nontoxic. This is the case with lipids A isolated from the LPSs of R. leguminosarum and R. etli (20), R. Sin-1 (21), and M. loti (22). The backbone of rhizobial lipid A is composed either of GlcpN or GlcpN3N disaccharide. Lipid A containing GlcpN can be modified by oxidation of the reducing GlcpN to 2-aminogluconate, as has been found in the LPSs of some Rhizobium species. The backbone may be substituted by phosphate, uronic acids, or other components, and is linked to an oligosaccharide core through a ketosidic bond formed by O-6 of the distal amino sugar and 3-deoxy-d-manno-oct-2-ulosonic acid residue (7). The amino groups of GlcpN3N and GlcpN, and the C-3 position of GlcpN are substituted by 3-hydroxy fatty acids. The hydroxyl groups may be further acylated either by nonpolar or (ω-1)-hydroxylated fatty acids, forming acyloxyacyl moieties (13, 14, 23–25). A comparison of the detailed structure of some rhizobial lipids A and the enterobacterial endotoxin shows that rhizobial lipids A are unusual. According to Urbanik-Sypniewska et al. (22), Vandenplas et al. (21) and Tsukushi et al. (26) some Sinorhizobium and Mesorhizobium strains possess varied endotoxic activity. Here, we report an investigation of the toxicity of lipopolysaccharides containing lipids A with unusual structures (see: 12–14).



LPS preparations were isolated from seven strains (listed in Table 1) using the hot phenol/water method as previously described (31). The LPS preparations were purified by electrodialysis and converted into a water-soluble form by triethylamine (Sigma, St Louis, MO, USA) neutralization according to Galanos and Lüderitz (32).

Table 1.  Bacterial strains used in this study
StrainHost plant and geographic originSource and references
  1. ATCC, American Type Culture Collection; CCBAU, Culture Collection of Beijing Agricultural University; IFO, Culture Collection of Institute of Fermentation, Osaka; USDA, United States Department of Agriculture.

B. japonicum USDA 110Glycine max, USAUSDA, (27)
B. elkanii USDA 76Glycine max, USAUSDA, (1)
B. liaoningense USDA 3622Glycine soya, Glycine max, ChinaUSDA, (28)
B. yuanmingense CCBAU 10071Lespendeza cuneata, ChinaCCBAU, (29)
B. sp. (Lupinus) USDA 3045Lupinus sp, United StatesUSDA
M. huakuii IFO 15243Astragalus sinicus, ChinaIFO, (3)
A. lipoferum SpBr17, (ATCC 29709)Monocotyledons, BrazilATCC, (30)
Salmonella enterica sv. Typhimurium Sigma
E. coli O55:B5 Sigma

The reference LPS preparations of Salmonella enterica sv. Typhimurium (Cat. No. 40H4000) and E. coli O55:B5 (part of the E-Toxate assay) were purchased from Sigma.

Sodium dodecylsulfate polyacrylamide gel electrophoresis of lipopolysaccharides

SDS-PAGE of the LPS preparations was performed in 12.5% acrylamide as described by Krauss et al. (33). The electropherograms were silver-stained (34).

Limulus amebocyte lysate gelation assay

LPSs from the examined strains and standard endotoxin preparations (Salmonella and E. coli) were dissolved in sterile, endotoxin-free water to obtain concentrations of from 0.1 mg/mL to 10 pg/mL, and mixed with an equal amount of LAL (E-Toxate, Sigma). After 1 hr of incubation at 37°C (in a water bath), gelation was determined by inverting the test tubes once.

Cell line culture conditions

The human myelomonocytic cell line THP-1 (from the European Collection of Cell Cultures, Cat No. 88081201) was cultured in RPMI 1640 medium supplemented with 2 mM L-glutamine, 10% FBS (Sigma), and 1% antibiotic-antimycotic solution (Sigma). The culture was maintained at 37°C in a humidified atmosphere containing 5% CO2. A mature macrophage-like state was induced by treating the THP-1 cells with PMA (Sigma).

Determination of nitrite concentration

Release of NO, measured as its end product, nitrite, was assessed using Griess reagent (35). Briefly, THP-1 cells were stimulated with the LPS preparations (0.01 μg/mL) for 24 hr. The culture supernatant (100 μL) was mixed with 100 μL of Griess reagent for 10 min, then the absorbance at 570 nm was measured using a microplate reader (Molecular Devices, Sunnyvale, CA, USA) and computer software (Softmax).

Cytokine induction

THP-1 cells were plated on 24-well tissue culture plates (Nunc, Roskilde, Denmark) at a density of 5 × 105 cells/mL (1 mL in each well) and cultured in RPMI 1640 cell culture medium supplemented with 2mM L-glutamine, 10% FBS, antibiotics, and 50 ng/mL PMA for 72 hr. Differentiated, plastic-adherent cells were washed twice with cold Dulbecco's PBS (Sigma) and incubated with a fresh culture medium without PMA. The medium was then changed every 24 hr for another 3 days. Cytokine induction was performed on the fourth day after removal of PMA. The medium was replaced by fresh RPMI 1640 medium supplemented with 2% FBS and LPSs from the examined strains or standard LPS from Salmonella enterica sv. Typhimurium. The LPSs were diluted in RPMI 1640 cell culture medium and added at concentrations of 0.01 μg/mL and 1 μg/mL. After 24 hr of incubation at 37°C in a humidified atmosphere containing 5% CO2, supernatants were collected, centrifuged, and stored at −80°C until cytokine assay.

Assay for cytokines

The concentrations of IL-1β, IL-6, and TNF in the supernatants were measured by ELISA using kits from Bender MedSystems, GmbH (Vienna, Austria) according to the manufacturer's protocols. The detection limits were 0.32 pg/mL for IL-1β, 0.92 pg/mL for IL-6, and 3.83 pg/mL for TNF.

Statistical analysis

For each experiment, the mean of three wells ± SD was expressed. Analyses were performed with GraphPad Prism 5 software. Statistical significances were determined by Student's t-test and set at P < 0.05 or P < 0.01.


Isolation and sodium dodecylsulfate polyacrylamide gel electrophoresis characteristics of lipopolysaccharide preparations

The LPS preparations were isolated using standard hot phenol/water extraction. The majority of LPSs from B. sp. (Lupinus), B. japonicum, B. yuanmingense, M. huakuii, and A. lipoferum strains were found in the water phase, whereas LPSs from B. elkanii and B. liaoningense were extracted into the phenol phase.

SDS-PAGE analysis revealed a high degree of heterogeneity for all the examined LPSs (Fig. 2). The LPS from M. huakuii (lines 1 and 2) migrated as three clusters of bands: a very intensively stained R-form LPS, an S-form, and an SR-form. A. lipoferum LPS (lines 3 and 4) was separated into two main fractions: the first one representing an R-form and the second one, high molecular weight material. Those complete LPS molecules contained approximately 20 repeating units in the O-chain, as calculated by comparison with the standard Salmonella LPS (line 7 and 14) (see also: 36). B. japonicum and B. yuanmingense LPSs (lines 5, 6 and 8, 9, respectively) were represented by complete molecules (S-form), mainly with short O-chains. The R fraction (containing only lipid A and core) was scarcely visible on the gel. In contrast, B. elkanii LPS (lines 12 and 13) occurred mainly as an R or SR form accompanied by a small amount of a ladder-like S-form containing up to 20 repeating units. LPS from B. liaoningense (lines 10 and 11) was represented mainly by an SR-form, though a small amount of the R- and the S-forms was also present.

Figure 2.

SDS-PAGE of LPS preparations from:M. huakuii IFO 15243 (line 1, 0.5 μg; line 2, 3 μg), A. lipoferum SpBr17 (line 3, 5 μg; line 4, 10 μg), B. japonicum USDA 110 (line 5, 0.5 μg; line 6, 2.5 μg) B. yuanmingense CCBAU 10071 (line 8, 0.5 μg; line 9, 2.5 μg), B. liaoningense USDA 3622 (line 10, 0.5 μg; line 11, 2.5 μg), B. elkanii USDA 76 (line 12, 0.5 μg; line 13, 3 μg), and Salmonella enterica sv. Typhimurium (lines 7 and 14, 2 μg).

Limulus activity of rhizobial lipopolysaccharide

The endotoxic properties of rhizobial LPSs were measured as their ability to gelate Limulus amebocyte lysate. For the LPSs from B. japonicum and B. yuanmingense, gelation was observed at a concentration of 0.1 μg/mL, whereas for the LPSs of B. elkanii, B. sp. (Lupinus), and B. liaoningense, the minimum LPS dose required for a positive reaction was ten times smaller (0.01 μg/mL). The LPSs from M. huakuii and A. lipoferum exhibited significantly greater endotoxic activity and gelated the amebocyte lysate at a concentration of 0.1 ng/mL. For the standard LPS preparations (Salmonella and E. coli), a positive reaction was observed at a concentration of 0.01 ng/mL.

Nitrous oxide-inducing activity of rhizobial lipopolysaccharide

Production of NO was determined in cultures of THP-1 cells which were stimulated with 1 μg/mL LPS preparations for 24 hr (Fig. 3). A significant amount of NO release was observed only for the standard LPS of Salmonella enterica bv Typhimurium (more than 300% of negative control). The amount of NO production by cells incubated with the B. sp. (Lupinus), B. elkanii, B. japonicum, M. huakuii, and A. lipoferum LPSs was just over half as much as that for Salmonella endotoxin, and exceeded by 50 to 100% the amount of spontaneous NO production by cells in the control sample.

Figure 3.

LPS induced (1 μg/mL) NO production in THP-1 cells. THP-1 cells were incubated with LPS from B. liaoningense USDA 3622, B. sp. (Lupinus) USDA 3045, B. yuanmingense CCBAU 10071, B. elkanii USDA 76, B. japonicum USDA 110, M. huakuii IFO 15243T, A. lipoferum SpBr17 and Salmonella enterica sv. Typhimurium strains for 24 hr, and the amount of NO was measured with Griess reagent. Data are expressed as mean ± S.D. from three independent experiments. *, statistically significant difference in comparison with the control, P < 0.05; **, statistically significant difference in comparison with the control, P < 0.01 (Student‘s t-test).

A statistically significant difference in NO production in comparison with the negative control (Student's t-test, P value <0.05) was noted for B. sp. (Lupinus), B. japonicum, and M. huakuii.

Cytokine-inducing activity of rhizobial lipopolysaccharides

Production of the cytokines TNF, IL-1β, and IL-6 was determined in cultures of THP-1 cells stimulated with two LPS concentrations, 0.01 and 1 μg/mL (Fig. 4). At an LPS dose of 0.01 μg/mL, the Bradyrhizobium and the Azospirillum strains induced production of very small amounts of the cytokines. In the case of the two interleukins (IL-1β and IL-6), the measured amounts were within the same range as for the control sample (spontaneous activity of THP-1 cells) and the differences were not statistically significant. However, TNF production was slightly greater than the control with one exception, B. sp. (Lupinus) LPS, which induced an extremely small amount of this cytokine. Induction of cytokine production by M. huakuii LPS at a dose of 0.01 μg/mL was a little higher, but still within a low range, when compared to the standard endotoxin. At a concentration of 1 μg/mL of LPS, cytokine production was much more diversified. Cells induced with the LPSs from B. elkanii, B. liaoningense, and B. yuanmingense produced very small amounts of cytokines, especially interleukins. Production of cytokines by THP-1 cells induced with B. sp. (Lupinus) and B. japonicum LPSs was somewhat higher, but still approximately 10–20 times lower than in the presence of Salmonella endotoxin. The LPSs isolated from M. huakuii and A. lipoferum induced significantly greater amounts of cytokines, especially TNF (see Fig. 4). Although, the amount of both interleukins (IL-1β and IL-6) released was rather high, it was still considerably lower than that found with the standard LPS of Salmonella.

Figure 4.

Cytokine production (TNF, IL-1β, and IL-6) in THP-1 cells stimulated by LPS from the strains (a) Salmonella enterica sv. Typhimurium, (b) M. huakuii IFO 15243, (c) A. lipoferum SpBr17, (d) B radyrhizobium sp. (Lupinus) USDA 3045, (e) B. japonicum USDA 110, (f) B. yuanmingense CCBAU 10071, (g) B. elkanii USDA 76, and (h) B. liaoningense USDA 3622. The concentrations of LPS were 0.01 μg/mL and 1 μg/mL. Data are expressed as mean ± S.D. from three independent experiments. *, statistically significant difference in comparison with the control, P < 0.05 (Student‘s t-test).


Minute amounts of LPS released from the surface of enteric bacteria are an early signal of infection for animal immune systems. A majority of host cells recognize traces of an endotoxin through the CD14-MD2-TLR4 protein complex. On the other hand, appearance of LPSs originating from non-enterobacterial species does not trigger a massive response from the host innate immune system (16, 37). All rhizobial LPSs have lipids A with unusual structures. Features which place these lipids A in the atypical group include the presence of very long chain fatty acids hydroxylated at penultimate positions (i.e. 27-octacosanoic acid); partial or complete absence of phosphate residues, which are replaced by uronic acid or neutral sugars; or proximal backbone amino sugar which has been oxidized to 2-aminogluconate (38).

All rhizobial lipopolysaccharides (lipids A) studied till now, with the single exception of S. meliloti (26), exhibit low endotoxic activity. Most experiments concerning the biological properties of these LPSs have been carried out on animal (mouse) models or using murine spleen leukocytes, monocytes, or a mouse leukemic monocyte macrophage cell line (RAW 264.7) (22, 26, 39). The biological properties of the LPS isolated from Sinorhizobium Sin-1 are the only ones to have been tested on a human monocytic cell line (Mono Mac 6) (21). However, in most cases, the responses of the murine immune system have been similar to, or identical with, those of the human one.

The biological activity of the LPSs examined in the present paper, measured as their ability to induce production of the cytokines TNF, IL-1β, and IL-6, and release of NO from human myelomonocytic cells (THP-1), demonstrates that the LPSs from the five Bradyrhizobium strains and from M. huakuii, and A. lipoferum exhibit significantly less endotoxic potency than Salmonella LPS. Gelation of LAL occurred at an LPS concentration of 0.1 μg/mL for B. japonicum and B. yuanmingense LPSs, and of 0.01 μg/mL in the case of B. elkanii, Bradyrhizobium sp. (Lupinus), and B. liaoningense. These results indicate that Bradyrhizobium LPSs are 1000–10,000 times weaker endotoxins than are enterobacterial LPS. For M. huakuii and A. lipoferum LPSs, gelation was observed at 0.1 ng/mL, which indicates that these endotoxins are 10 times weaker than the standard LPSs. Thus, our studies lead to the conclusion that all the examined LPSs are weak endotoxins and probably have low lethality for animals (22).

The differences between the examined strains and the standard endotoxin in biological activities of the LPS preparations were reflected in differences in the structure of lipid A, the centre of the endotoxic properties of the whole LPS molecule. The relationship between lipid A structure and its biological activity has been extensively studied, and the factors regulating the immunological activity of LPS identified. Among them, phosphate residues and the number, type, and distribution of fatty acids in lipid A are the most important (40). For proinflammatory activity, an enterobacterial lipid A that contains six fatty acids, of which two nonpolar ones are asymmetrically located creating two acyloxyacyl moieties, is required. Lipid A deprived of one fatty acid residue is about 100-fold less toxic, whereas lipid A analogues carrying only four primary fatty acids completely lack agonistic activity (16,41).

M. huakuii produces a naturally heterogenic lipid A, in particular due to the occurrence of hexa-acyl, penta-acyl, and tetra-acyl subspecies (13). The monophosphorylated subfraction of this lipid A occurs mainly as penta-acyl and hexa-acyl, containing, apart from 27-hydroxyoctacosanoic fatty acid, one eicosanoic moiety. The unphosphorylated subfraction of the lipid A is represented mainly as the hexa-acyl fraction. Thus, the presence of a large proportion of lipid A molecules with a lower degree of acylation might be a strong factor in the reduced biological activity of this LPS preparation. In addition, the presence of an unusual, very long chain hydroxylated fatty acyl (27-hydroxyoctacosanoic), which is typical of rhizobial lipids A, might affect toxicity, possibly by handicapping accommodation in the active site of the MD-2 receptor. The impaired toxicity of mesorhizobial lipid A may also result from reduced substitution by the ester-linked phosphate residue (50% of total). The C-1 position of the reducing end of the backbone in this lipid A is occupied by a galacturonic acid unit. The presence of two phosphate groups (at positions C-1 and C-4) in the lipid A greatly affects the endotoxic activity of enterobacterial LPS (40, 42). Removal of one of the phosphate groups reduces the biological activity of the enterobacterial endotoxin almost 100-fold, and monophosphoryl lipid A is a weak activator of the human innate immune response. Furthermore, the deletion of phosphate from the C-1 position in Salmonella Minnesota not only weakens the affinity of the ligand but also induces a structural rearrangement of the TLR4-MD-2-adaptor multimer receptor (16). Our results are also in agreement with the findings by Tsukushi et al. (26) and Urbanik-Sypniewska et al. (22), whose studies of the endotoxic properties of M. loti lipopolysaccharides have shown that LPSs from bacteria belonging to the genus Mesorhizobium are very weak endotoxins.

As has been described recently, A. lipoferum lipid A is completely lacking in phosphate, but contains galacturonic acid linked to the diglucosamine backbone at position C-1. This lipid A is heterogeneous in relation to the acylation pattern (12). Among the pool of lipid A molecules, at least three subfractions have been identified (penta-, tetra-, and tri-acylated lipids A). A. lipoferum lipid A does not contain any very long chain fatty acids. Thus, it seems that a lack of phosphate and a low degree of acylation play crucial roles in reduction of the toxicity of this lipid A.

The structure of lipid A isolated from B. elkanii LPS has been described in detail by Komaniecka et al. (14). This lipid A is completely lacking in any negatively charged residues. The GlcpN3N disaccharide backbone is further substituted by three mannopyranose residues, forming a pentasaccharide. Although B. elkanii lipid A is homogenous in the number of fatty acids and contains six acyl residues, two of them are unusual, being very long (ω-1)-hydroxylated secondary fatty acids, with a chain length ranging from 26 to 33 carbon atoms (14). Our current data suggest that the structure of B. japonicum lipid A is similar to that published for B. elkanii (unpublished data). Thus, it seems that bradyrhizobial lipids A are unusual high-molecular-mass molecules with weak endotoxic activity because of the presence of a large hydrophobic part, which probably blocks the active site of the TLR4 receptor and prevents it from forming the TLR4-MD-2-LPS complex.


We thank Jadwiga Dolecka for excellent SDS-PAGE analyses. We are also very grateful to Dr. Teresa Urbanik-Sypniewska for critical reading of this paper. This work was financially supported by the Polish Ministry of Science and Higher Education, grant No. 303 109 32/3593.