Introduction

Bacterial autoaggregation is the process whereby bacteria physically interact with each other and settle to the bottom in static liquid suspensions. This process is industrially significant since separation of biomass from culture medium is facilitated when bacteria remain aggregated [21]. Extracellularly secreted polysaccharides, i.e., cellulose and an arabinose-rich exopolysaccharide, were shown to be involved in autoaggregation of the soil-borne bacteria Agrobacterium tumefaciens and Azospirillum brasilense, respectively [2, 16].

Sinorhizobium meliloti, a symbiotic Gram-negative rhizobacterium, has the potential to produce two different exopolysaccharides: EPS I (succinoglycan) and EPS II (galactoglucan). EPS I consists of a repeated octasaccharide structure containing one galactose and seven glucose units, with succinyl, acetyl, and pyruvyl modifications [23]. The genetic determinants for EPS I biosynthesis are located in a 25-kb region containing the exo-exs genes [24]. EPS II consists of a repeated disaccharide containing an acetylated glucose and a pyruvylated galactose unit [11]. Biosynthesis of EPS II is controlled by a 32-kb cluster of exp genes [8, 29]. In two strains (Rm2011 and Rm1021) derived from Sinorhizobium meliloti SU47, EPS II synthesis is activated under phosphate limitation [18, 31], in the presence of either a mutated mucR locus [14, 30], or an intact, functional copy of the expR gene [8, 22]. The low molecular weight fractions of EPS I and EPS II (trimers of the octasaccharide and 15–20 units of the disaccharide, respectively) are involved in symbiosis, whereas the corresponding high molecular weight fractions (more than 2000 subunits of the octasaccharide unit in the case of EPS I; more than 25 disaccharide subunits in the case of EPS II) are not [3, 10].

The regulatory protein MucR is a transcriptional repressor of the exp genes involved in biosynthesis of EPS II, and an activator of EPS I biosynthesis in S. meliloti [14]. We found recently that disruption of mucR gene in strain Rm1021 biofilms increased EPS II, but reduced EPS I gene expression; however, biofilm formation in this strain was independent of EPS synthesis [27]. These results are consistent with those from previous studies of planktonic bacteria. In this study, we evaluated the possible involvement of EPS I and II in autoaggregation of planktonic cells of various S. meliloti strains.

Materials and Methods

Bacterial Strains, Media, and Growth Conditions

Various S. meliloti strains were grown at 30°C in TY medium [4] on a rotary shaker at 200 rpm. Antibiotics were used at the following final concentrations: streptomycin, 500 μg/ml; neomycin, 200 μg/ml; gentamycin, 50 μg/ml. Strains, plasmid, and phage used are listed in Table 1.

Table 1 Strains and plasmids used in this study
Full size table

Phage Transductions

The mutant alleles expA3::Tn5-233, exoY210::Tn5, and mucR31::Tn5 were transferred from Rm1021 expA::Tn5-233, Rm7210, and Rm3131, respectively, to recipient strains Rm8530 or Rm1021, using generalized transduction with phage φM12, as described by [9]. Cotransduction of the resistance markers (neomycin or gentamycin), and specific relevant phenotypes, were verified in each transductant strain, i.e., reduction or absence of fluorescence in LB plates supplemented with 0.05% Calcofluor (for the mucR and exoY mutants), dry colony phenotype (for the expA mutant). Donor and recipient strains were included as controls.

Autoaggregation Assay

Each S. meliloti strain was grown for 48 h in 3 ml TY medium supplemented with appropriate antibiotic, diluted (1/1000), and grown in 100 ml TY for additional 48 h. Bacterial suspension (5 ml) was transferred to a glass tube (10 × 70 mm) and allowed to settle for 24 h at 4°C. A 0.2 ml aliquot of the upper portion of suspension was carefully transferred to a microtiter plate, and OD600 was measured (ODfinal). A control tube was vortexed for 30 s, and OD600 was determined (ODinitial). Autoaggregation percentage was calculated as 100[1 − (ODfinal/ODinitial)]. For both homologous and heterologous autoaggregation assays, cultures were centrifuged at 4200g for 20 min prior to the settling period. For homologous assay, the resulting bacterial pellet of a given strain was gently resuspended in cell-free supernatant from independent culture of the same strain. For heterologous assay, the pellet was resuspended in cell-free supernatant from culture of a different strain.

Statistical Analysis

Autoaggregation assays were performed in quadruplicate and experiments were repeated at least three times. Mean values and standard deviations were calculated. Data were subjected to one-way Analysis of Variance (ANOVA) test, followed by comparison of multiple treatment levels with control using post hoc Fisher’s Least Significant Difference (LSD) test. Statistical analyses were performed with Infostat software version 1.0.

Results and Discussion

When a suspension of planktonic cells of strain Rm8530 (expR + derivative of Rm1021) was placed in a glass tube and left without shaking for 24 h at 4°C, most of the bacteria settled at the bottom, forming a type of aggregate, termed “floc,” bound together by extracellular mucopolysaccharides (Fig. 1a). Under standard shaking conditions, such visible autoaggregation of Rm8530 did not occur. This finding is in contrast to results from autoaggregation studies of Azospirillum, in which floc formation was evident under shaking conditions [20]. Much of the Rm8530 floc remained at the bottom of the tube following 2–3 gentle inversions, indicating a certain degree of integrity. However, the floc was easily dispersed by repeated pipetting or a vigorous shake. The autoaggregation of bacteria at the bottom of the tube was associated with decreased OD600 of the upper portion of the suspension, and autoaggregation percentage was quantified by spectrophotometry as described in “Materials and Methods”. When suspensions of wild-type strains Rm1021, Rm2011, and 102F34 were left without shaking for 24 h at 4°C, most of the bacteria remained suspended, without visible autoaggregation (Fig. 1b). Autoaggregation percentage was similar (18-20%) for the three wild-type strains, but significantly higher (95%) for Rm8530 (Fig. 1a).

Fig. 1
figure 1

Autoaggregation assay of wild-type reference strains of S. meliloti. a Quantitative assay, showing mean aggregation percentage for each strain. Each experiment was performed in four or more independent replicates, each in quadruplicate. Bars represent standard deviation, and different letters indicate significant differences (P ≤ 0.05) between mean values according to Fisher’s LSD test. b Macroscopic appearance of tubes containing bacterial suspensions, following a 24-h resting period at 4°C

Full size image

When Rm8530 cells, prior to the resting period, were washed and resuspended in fresh, sterile TY media, their autoaggregation was almost completely abolished (Fig. 2), suggesting that some extracellularly secreted and/or labile surface factor was responsible for the autoaggregation. When Rm8530 pellets were resuspended in cell-free supernatant from the same (or different) Rm8530 culture, the capacity for aggregation was not significantly decreased (Fig. 2a, b). This finding indicates that an extracellular component, rather than a centrifugation-labile surface factor, was the main determinant of autoaggregation.

Fig. 2
figure 2

Modified autoaggregation assays of strain Rm8530. a Quantitative assay. Statistical considerations as in Fig. 1. b Macroscopic appearance of tubes. 1. Autoaggregation of Rm8530 suspension (control). 2. Homologous autoaggregation assay: Rm8530 cells were resuspended in cell-free supernatant from independent culture of the same strain. 3. Autoaggregation of washed cells of Rm8530

Full size image

In contrast to Rm8530, non-autoaggregating wild-type strains Rm1021, Rm2011, and 102F34 have a modified, non-functional expR locus [22]. This suggests a link between status of expR locus (i.e., intact versus altered sequence) and autoaggregation ability in these S. meliloti strains. Presence of a functional copy of the expR regulator gene is clearly necessary for autoaggregation. ExpR controls many aspects of S. meliloti physiology, including exopolysaccharide production [12, 22]. We performed quantitative autoaggregation assays of a set of S. meliloti mutants defective in biosynthesis of various exopolysaccharides to clarify the roles of these components in autoaggregation.

Rm8530 exoY210, a mutant incapable of producing succinoglycan, showed autoaggregation percentage (93%) similar to that of parental Rm8530 (Fig. 3). Autoaggregation percentage was much lower (18%) for Rm8530 expA, a mutant defective in biosynthesis of galactoglucan. Using phage φM12, we transduced the exoY210::Tn5 mutant allele from Rm7210 to Rm8530 expA, and obtained double mutant Rm8530 expA exoY, as described in “Materials and Methods”. Degree of autoaggregation was low (similar to that of Rm8530 expA) for Rm8530 expA exoY and Rm1021, which are also unable to produce galactoglucan under normal conditions (Fig. 3). These findings suggest that galactoglucan is the extracellular factor mainly responsible for autoaggregation of S. meliloti.

Fig. 3
figure 3

Autoaggregation of mutant strains of S. meliloti defective in exopolysaccharide production. a Quantitative assay. Statistical considerations as in Fig. 1. b Macroscopic appearance of tubes

Full size image

Galactoglucan is produced in two major fractions: one has low molecular weight and is symbiotically active; the other has high molecular weight and no symbiotic function [10]. Although galactoglucan synthesis is activated in Rm1021 mucR, specific synthesis of the low molecular weight fraction does not occur. As an initial attempt to elucidate the roles of the two fractions in autoaggregation, we transduced the mucR31::Tn5 allele from Rm3131 to Rm1021, and performed quantitative autoaggregation assay of the resulting Rm1021 mucR strain, which is unable to synthesize low molecular weight EPS II fraction. Autoaggregation percentage for this mutant was much lower than that of Rm8530, and did not differ significantly from those of Rm1021 and Rm8530 expA (Fig. 3). This initial finding allowed us to conclude that high molecular weight EPS II, by itself, cannot mediate cellular interactions leading to autoaggregation.

To further substantiate our results, we performed extracellular complementation assays, in which cell pellets were resuspended in cell-free supernatant from early stationary phase bacterial cultures. Homologous and heterologous assays were conducted as described in “Materials and Methods”. Under our experimental conditions, autoaggregation of the homologous and heterologous suspensions depended only on identity of the supernatant, as shown in Table 2. As expected, cell-free supernatants from cultures of both galactoglucan-producing strains, Rm8530 and Rm8530 exoY, enhanced autoaggregation of all the S. meliloti strains tested. In contrast, there was no significant aggregation-promoting effect by cell-free supernatants from the non-galactoglucan-producing strains (Rm1021, Rm8530 expA, Rm8530 expA exoY), or from Rm1021 mucR (which does not produce low molecular weight galactoglucan). These findings suggest that the low molecular weight fraction of EPS II is responsible for the autoaggregation phenotype. Further experiments, including direct testing of purified low molecular weight galactoglucan, will be necessary to confirm this hypothesis.

Table 2 Extracellular complementation of non-autoaggregating strains with cell-free supernatant from culture of autoaggregating strains
Full size table

DNA microarray analysis has shown that expR regulates many other genes, in addition to EPS and motility genes, in S. meliloti [12]. Under our experimental conditions, and in the absence of EPS II, it appears that no additional expR-regulated factors play a role in autoaggregation, since autoaggregation percentages of Rm1021 and Rm8530 expA (or washed cells of Rm8530 and Rm8530 exoY) were similar.

Bacterial cell–cell interactions contribute to both autoaggregation and microcolony development, which are important early steps in formation of biofilms. Several factors, including EPS, are involved in these interactions [28]. Interestingly, biofilm formation and autoaggregation of planktonic cells seem to be related processes in S. meliloti, since EPS II (but not EPS I) was shown to be critical for development of highly structured biofilms. A high molecular weight EPS II producer, Rm1021 mucR strain, failed to form organized biofilms [26], consistent with the poorly autoaggregative phenotype of this mutant in our assays.

We speculate that other molecular determinants, besides EPS, may be involved in both autoaggregation of planktonic cells and biofilm formation (at least, the early steps) in S. meliloti. Inoculation of plants with EPS-producing rhizobacteria, such as Rhizobium sp. YAS34 [1] and Rhizobium sp. KYGT207 [13], modifies the aggregation of root-adhering soil and leads to improvement of plant growth. Biofilm formation, although it may provide rhizobia with an advantageous microenvironment to persist in the soil, colonize root surfaces, and establish symbiosis, is not essential for legume invasion [25]. On the other hand, host plants, as well as other soil microorganisms, may benefit from the biofilm formation capability of plant growth-promoting rhizobacteria, since EPS within biofilms improve soil structure and help maintain soil moisture [19].

Results of this study strongly suggest that expR-regulated EPS II plays a key role in promoting bacterial self-aggregation. Low molecular weight EPS II, either alone or in combination with high molecular weight fraction, may function as the polymeric extracellular matrix that agglutinates bacterial cells. Future extracellular complementation experiments, using purified low molecular weight EPS II, will help clarify this point. We are currently in the process of identifying a bacterial surface component that interacts with EPS II, and expect that results of this study will provide a more complete picture of the S. meliloti autoaggregation process.