During the past several years, research in our laboratory has focused on the cyclic β-glucans of bacteria within the Sinorhizobium and Bradyrhizobium genera (8, 9). In Sinorhizobium species, the cyclic β-glucans are linked solely by β-1,2 glycosidic bonds, whereas in Bradyrhizobium species, these molecules are linked by both β-1,3 and β-1,6 glycosidic bonds. The cyclic β-glucans are localized within the periplasmic compartment of these bacteria, but they are also released into the medium. Indeed, we have previously shown that Bradyrhizobium japonicum cultures excrete relatively high levels of cyclic β-1,6-β-1,3-glucans into the culture medium, with levels approaching up to sevenfold higher than levels associated with cells (22). As part of our ongoing research efforts, we recently developed a radiolabel screening strategy to isolate mutants of B. japonicum impaired for the synthesis of cyclic β-1,6-β-1,3-glucan (H. A. Louch and K. J. Miller, Abstr. 95th Gen. Meet. Am. Soc. Microbiol. 1995, abstr. N-197, p. 366, 1995). During the course of screening putative cyclic β-glucan mutants of B. japonicum USDA 110, a novel, low-molecular-weight form of EPS was identified which copurified with extracellular cyclic β-1,6-β-1,3-glucans. The purification and structural analysis of this low-molecular-weight form of EPS are described in the present study.

Identification of B. japonicum mutant M1E7. Tn5 mutagenesis of B. japonicum USDA 110 was performed using a biparental mating procedure, essentially as described by Hom and coworkers (17). Tn5 mutants were plated onto GMS (32) containing (per ml) 100 μg of streptomycin, 100 μg of kanamycin, and 50 μg of trimethoprim. Approximately 2,100 Tn5 mutants of B. japonicum USDA 110 were isolated, and each was screened for defects in cyclic β-1,6-β-1,3-glucan biosynthesis as described below.

Each B. japonicum USDA 110 Tn5 mutant was inoculated into 5 ml of GMS medium. Cultures were grown to an optical density at 650 nm of between 0.3 and 0.6 in a 30°C rotary shaker (7 to 8 days on average), at which time radiolabeled glucose (either [6-³H]glucose or [¹⁴C]glucose) was added to a final concentration of 100 μM and at a specific activity of 0.5 μCi/ml. Cultures were incubated for 3 to 6 h in the presence of radiolabel. After incubation, cells were pelleted by centrifugation (12,000 × g for 5 min) and washed twice with water (1 ml), and cyclic β-1,6-β-1,3-glucans were extracted with 160 μl of 70% ethanol at 70°C for 30 min. The level of radiolabeled cyclic β-1,6-β-1,3-glucans within each ethanol extract was determined by adsorption onto C₁₈ silica gel resin (Supelco, Bellefonte, Pa.) followed by selective elution using 30% methanol. This screening strategy using C₁₈ silica gel resin was based on an earlier report by Rolin and coworkers (28) that revealed that cyclic β-1,6-β-1,3-glucans could be bound to C₁₈ silica gel resin and selectively eluted using 30% methanol. Of approximately 2,100 mutants screened, 1 mutant, referred to as mutant M1E7, was found to contain extremely low levels of radiolabel in the 30% methanol eluent (i.e., 4% of the level produced by the parent strain, USDA 110). Based on this finding, mutant M1E7 was selected for further study.

Analysis of extracellular low-molecular-weight polysaccharides from B. japonicum cultures. B. japonicum strains were cultured in 500 ml of YM medium (23) at 30°C until reaching an optical density at 650 nm of 0.6. Cells were harvested (13,000 × g for 10 min) and washed with 25 ml of YM salts buffer at pH 7.0 (23), and culture supernatants were frozen. After thawing, culture supernatants were concentrated 25-fold by rotary evaporation. Next, high-molecular-weight EPS was precipitated from concentrated supernatants by adding 3 volumes of ice-cold ethanol as described by Breedveld and coworkers (10). High-molecular-weight EPS was then removed from concentrated supernatants by centrifugation (12,000 × g for 10 min). Low-molecular-weight, ethanol-soluble polysaccharides were then purified from concentrated supernatants using gel permeation chromatography as described below.

Concentrated supernatants containing ethanol-soluble, extracellular low-molecular-weight polysaccharides were concentrated under vacuum. Samples were applied to a Sephadex G-25 column (1 by 52 cm) which was eluted at room temperature with 0.15 M ammonium acetate (pH 7.0) containing 7% propanol (vol/vol) at a rate of 15 ml/h. Fractions (1 ml) were collected and assayed for carbohydrate content (12). Material eluting in the position expected for cyclic β-1,6-β-1,3-glucan was pooled, concentrated, and subsequently desalted using a Sephadex G-15 column (1 by 49 cm). The Sephadex G-15 column was eluted at room temperature with 7% propanol (vol/vol) at a rate of 15 ml/h. Fractions (1 ml) were collected and assayed for carbohydrate content. Material eluting in the position expected for cyclic β-1,6-β-1,3-glucan was pooled and subsequently analyzed by thin-layer chromatography (TLC) using aluminum-backed Silica Gel 60 plates (EM Industries, Gibbstown, N.J.) and a butanol-ethanol-water (5:5:4) solvent system. Samples were visualized on TLC plates by charring at 170°C for 20 min after spraying with 5% sulfuric acid in methanol (vol/vol) (7).

Identification of a novel, low-molecular-weight form of EPS within B. japonicum cultures. TLC is routinely performed in our laboratory to monitor the purification of cell-associated cyclic β-1,6-β-1,3-glucans from B. japonicum cultures; however, it can also be used to monitor the purity of cyclic β-1,6-β-1,3-glucans obtained from the supernatants of B. japonicum cultures. When the low-molecular-weight, extracellular polysaccharides of B. japonicum mutant M1E7 were examined by TLC, a major spot, migrating with a substantially higher R_f value than the cyclic β-1,6-β-1,3-glucans, was detected (data not shown). Indeed, this material was a major contaminant within cyclic β-1,6-β-1,3-glucan preparations and represented approximately 75% (glucose equivalent) of the total carbohydrate present within the low-molecular-weight fraction isolated from culture supernatants of mutant M1E7. Further analysis revealed that this contaminant could be bound to DEAE-cellulose at pH 8.4 and subsequently eluted using a buffer containing 200 mM KCl, indicative of anionic character. This is in contrast to the cyclic β-1,6-β-1,3-glucans, which are uncharged and do not bind to DEAE-cellulose under these conditions.

Additional characterization of the extracellular anionic contaminant material isolated from mutant M1E7 was performed using negative-ion fast atom bombardment mass spectometry (FABMS). FABMS analysis revealed a mass spectrum distinctly different from that obtained for the cyclic β-1,6-β-1, 3-glucans. For example, it has previously been shown that a typical mass spectrum for purified cyclic β-1,6-β-1,3-glucan contains predominant molecular ion species ([M-H)⁻] at m/z values of 1,619, 1,781, 1,943, and 2,005 (23), which correspond to unsubstituted cyclic glucans containing 10 to 13 glucose residues. However, when the anionic, low-molecular-weight extracellular polysaccharide material obtained from mutant M1E7 was examined by negative-ion FABMS, the analysis revealed a very different spectrum, although the predominant molecular ion species had m/z values in the same range as the cyclic β-1,6-β-1,3-glucans (Fig. 1). This result confirmed that the anionic low-molecular-weight polysaccharide material is very similar in size to the cyclic β-1,6-β-1,3-glucans, consistent with the fact that these materials copurify on Sephadex G-25 and Sephadex G-15.

FIG. 1 Negative-ion FABMS analysis of the extracellular low-molecular-weight, anionic polysaccharide isolated from B. japonicum mutant M1E7 culture supernatants. The major molecular ion species had m/z values of 1,791, 1,806, 1,833, 1,848, and 1,862. Negative-ion (more ...)

A compositional analysis of the extracellular low-molecular-weight polysaccharide isolated from the B. japonicum M1E7 mutant was performed using gas chromatography linked to electron impact mass spectrometry. As shown in Table 1, the material was found to contain glucose, mannose, galacturonic acid, and galactose, in a 2:1:1:1 ratio. This is the same composition previously reported for the high-molecular-weight EPS of B. japonicum strains (18, 24, 25, 27). Based on these results, additional analyses were performed on the low-molecular-weight anionic polysaccharide material obtained from culture supernatants of the wild-type parent strain, USDA 110. These results are also shown in Table 1 and reveal that both mutant M1E7 and the wild-type parent strain, USDA 110, produce and excrete a low-molecular-weight form of EPS.

Further structural examination of the putative low-molecular-weight EPS was performed at the Complex Carbohydrate Research Center at the University of Georgia (Athens) using methylation and gas chromatography-mass spectrometry analysis (as described by Ciucanu and Kerek [11]), as well as 1-D ¹H nuclear magnetic resonance (NMR) analysis. Methylation analysis revealed the presence of terminally linked galactose, 3-linked glucose, 3-linked mannose, and 3,6-linked glucose. Unfortunately, methylation analysis could not be used to identify uronic acid residues; thus, galacturonic acid could not be detected. However, NMR analysis revealed the presence of an O-methyl group, an O-acetyl group, and resonances consistent with a 4-O-acetylated galacturonic acid residue (Fig. 2). Thus, the combined results of the methylation and NMR analyses are fully consistent with the structure of the pentasaccharide repeating unit previously reported for B. japonicum EPS (24, 25). The NMR spectra previously published for high-molecular-weight EPS of B. japonicum strains is in good agreement with our spectrum (25, 27). Based on the apparent size of this low-molecular-weight polysaccharide, as indicated by gel permeation chromatography and negative-ion FABMS, it may be concluded that this material corresponds to a dimeric form of the pentasaccharide repeating unit of the B. japonicum EPS.

FIG. 2 1-D 1H NMR spectrum of the extracellular, low-molecular-weight, anionic polysaccharide from B. japonicum USDA 110 culture supernatants. The resonances are identified as follows: 1 = H4 of the alpha-linked 4-O-methyl galacturonic acid; 2 = H1 of the alpha-linked (more ...)

Concluding remarks. To our knowledge, this is the first study to demonstrate the presence of a low-molecular-weight form of EPS in B. japonicum cultures, although it should be noted that Becker and coworkers (6) previously provided preliminary evidence for a low-molecular-weight form of EPS in B. japonicum 110spc4 cultures. The finding of a low-molecular-weight form of EPS in B. japonicum cultures was not surprising, since low-molecular-weight forms of EPS have previously been identified in Sinorhizobium meliloti cultures (2, 14). In fact, the low-molecular-weight forms of EPS produced by S. meliloti have been shown to promote nodulation of alfalfa by this symbiont (2, 14). In future studies, it will be of interest to examine whether or not the low-molecular-weight form of the B. japonicum EPS influences the nodulation of the soybean host, particularly since studies have indicated that effective symbioses are possible with mutants that do not synthesize the high-molecular-weight form of EPS (20, 26).

Our discovery of a low-molecular-weight form of B. japonicum EPS was initially made with B. japonicum mutant M1E7, a Tn5 mutant of USDA 110. This mutant was selected for further study because radiolabeling studies suggested that it was impaired for cyclic β-1,6-β-1,3-glucan biosynthesis. Curiously, analyses of nonradiolabeled, large-scale cultures of M1E7 revealed that the levels of cell-associated cyclic β-1, 6-β-1,3-glucans are similar to the levels found in wild-type USDA 110 cultures (data not shown). However, these analyses have also revealed that mutant M1E7 produces approximately twofold higher levels of extracellular, low-molecular-weight polysaccharides when compared to the wild-type USDA 110 strain. Therefore, it is possible that the low level of radiolabeled glucose incorporated into the cyclic β-1,6-β-1,3-glucan pool of M1E7 during the radiolabel screening procedure was a result of increased extracellular polysaccharide production by this mutant.

Further analysis of mutant M1E7 revealed another curious result. Specifically, the Tn5 insertion was found to lie within the clpA gene (the B. japonicum USDA 110 clpA sequence as well as the location of Tn5 within the clpA gene of mutant M1E7 are described within GenBank accession no. AF254897). ClpA is the regulatory subunit of the ClpAP protease, an ATP-dependent protease involved in the turnover of abnormal proteins (16, 29, 31). The results of the present study suggest a possible role for the ClpAP protease in the regulation of bacterial polysaccharide production. In this regard, it is interesting to note that the Lon protease has previously been shown to have a regulatory role in capsular polysaccharide biosynthesis in Escherichia coli (15, 30).

This research was supported by National Science Foundation grant MCB-9505706.