Brucella melitensis 16M produces a mannan and other extracellular matrix components typical of a biofilm

Authors


  • Editor: Gianfranco Donelli

Correspondence: Jean-Jacques Letesson, Unité de Recherche en Biologie Moléculaire, Facultés Universitaires Notre-Dame de la Paix, 61, rue de Bruxelles, Namur 5000, Belgium. Tel.: +32 81 72 44 02; fax: +32 81 72 44 20; e-mail: jean-jacques.letesson@fundp.ac.be

Abstract

Mutations in the Brucella melitensis quorum-sensing (QS) system are involved in the formation of clumps containing an exopolysaccharide. Here, we show that the overexpression of a gene called aiiD in B. melitensis gives rise to a similar clumping phenotype. The AiiD enzyme degrades AHL molecules and leads therefore to a QS-deficient strain. We demonstrated the presence of exopolysaccharide and DNA, two classical components of extracellular matrices, in clumps produced by this strain. We also observed that the production of outer membrane vesicles is strongly increased in the aiiD-overexpressing strain. Moreover, this strain allowed us to purify the exopolysaccharide and to obtain its composition and the first structural information on the complex exopolysaccharide produced by B. melitensis 16M, which was found to have a molecular weight of about 16 kDa and to be composed of glucosamine, glucose and mostly mannose. In addition, we found the presence of 2- and/or 6-substituted mannosyl residues, which provide the first insights into the linkages involved in this polymer. We used a classical biofilm attachment assay and an HeLa cell infection model to demonstrate that the clumping strain is more adherent to polystyrene plates and to HeLa cell surfaces than the wild-type one. Taken together, these data reinforce the evidence that B. melitensis could form biofilms in its lifecycle.

Introduction

Brucella melitensis is an alpha-2 proteobacterium responsible for brucellosis in small ruminants and Malta fever in humans (Smith & Ficht, 1990; Boschiroli et al., 2001). This worldwide zoonosis causes severe economic losses in endemic regions. The virulence of this facultative intracellular Gram-negative pathogen depends on its survival and replication in both professional and nonprofessional host phagocytes (Detilleux et al., 1990; Pizarro-Cerda et al., 1998), in which it diverts the phago-lysosomal trafficking to reach its intracellular replication niche derived from the endoplasmic reticulum (Starr et al., 2008). During infection, B. melitensis is exposed to diverse environmental and host stresses and thus has to adapt continuously through perception of external and internal signals and the regulation of gene expression. Among such regulation systems, stringent response, signal transduction through two-component systems and Quorum Sensing (QS) have been described previously (reviewed by Letesson & De Bolle, 2004).

Brucella melitensis is the first intracellular pathogen in which a QS system was described. Although no acyl-homoserine lactone (AHL) synthase has been found as yet, this bacterium produces two AHLs detectable in culture supernatants: a dodecanoyl-homoserine lactone (C12-HSL) and a putative 3-oxo-dodecanoyl-homoserinelactone (3-oxo-C12-HSL) (Taminiau et al., 2002), and possesses two LuxR-type regulators, called VjbR and BabR (Delrue et al., 2005). We demonstrated previously that QS, through VjbR, is a major regulatory system of important cell surface structures of Brucella (Delrue et al., 2005; Uzureau et al., 2007). Moreover, we showed that vjbR-deficient strains, all unresponsive to C12-HSL, display a clumping phenotype in liquid culture and that these aggregates contain an unknown exopolysaccharide(s) (Uzureau et al., 2007).

Clumping development is a complex process that is initiated when bacteria attach to a surface using exopolysaccharide polymers or other adhesins and develop into microcolonies. Bacteria can undergo an additional maturation step in which they develop as complex three-dimensional (3D) structures called biofilms (O'Toole et al., 2000). These structures are classically defined as matrix-enclosed bacterial populations adherent to each other and/or to surfaces or interfaces (Costerton et al., 1995). The biofilm development process requires complex cellular regulatory mechanisms in which QS is often involved (Davies et al., 1998; Hammer & Bassler, 2003; Rice et al., 2005). Aggregates of bacteria not attached to a surface are commonly termed flocs or clumps and have many of the characteristics of a biofilm (Hall-Stoodley et al., 2004). Because bacterial clumping is one of the initial steps of biofilm formation, the clumping phenotype in B. melitensis 16M described previously was the first evidence that this alphaproteobacterium could form biofilms during its lifecycle.

Biofilm or clump formation constitutes the natural behavior of numerous environmental and pathogenic bacteria. The most distinctive feature of these aggregative structures is the extracellular matrix that plays a structural role, benefiting the bacterium by enabling attachment to surfaces, improving nutrient acquisition or providing protection from environmental stresses and host defenses (Sutherland, 2001; Branda et al., 2005).

Matrix polymers of bacterial biofilms are predominantly exopolysaccharide, whose compositions vary between strains and can be affected by the growth conditions and the age of the biofilm (Sutherland, 2001). In addition to exopolysaccharide, the matrices generally contain nucleic acids, proteins, lipids and outer membrane vesicles (OMVs) in the case of Gram-negative bacteria (Tsuneda et al., 2003; Schooling & Beveridge, 2006). Because of the increasingly recognized role of these matrices in the host–pathogen or symbiont interactions, it becomes crucial to characterize the matrix of the clumps produced by the QS-deficient strain of B. melitensis.

In this work, we use as a clumping strain a B. melitensis 16M strain overexpressing aiiD (an AHL-acylase that destroys the QS signal molecules) called MG210. The characterization of the clumps produced by this strain allowed us to demonstrate the presence of exopolysaccharide(s), DNA and OMVs, three classical components of extracellular matrices. Moreover, here, we provide the first structural information on the complex exopolysaccharide produced by B. melitensis 16M since we found that its molecular weight is about 16 kDa and that it is composed of glucosamine, glucose and mostly mannose. In addition, we found the presence of 2- and/or 6- substituted mannosyl residues, which provides the first insights into the linkages involved in this polymer. We demonstrate that the MG210 strain displays increased adherence properties both on polystyrene and on HeLa cell surfaces. Taken together, our data reinforce the evidences that B. melitensis could form biofilms in its lifecycle.

Materials and methods

Bacterial strains and culture conditions

All the strains and plasmids used in this study are listed in Table 1. Brucella strains were grown with shaking at 37 °C in 2YT medium (10% yeast extract, 10 g L−1 tryptone, 5 g L−1 NaCl) containing appropriate antibiotics from an initial OD600 nm of 0.05. The Escherichia coli DH10B (Gibco BRL) and S17-1 strains were grown in Luria–Bertani medium with appropriate antibiotics. Chloramphanicol and nalidixic acid were used at 20 and 25 μg mL−1, respectively. For exopolysaccharide purifications, Brucella were grown in RPMI 1640 medium supplemented with 10 g L−1 of d-xylose and appropriate antibiotics.

Table 1.   Strains and plasmids used in this study
 Relevant genotype or descriptionReference or source
Strain
 B. melitensis
  16MWild type, NalRSangari & Aguero (1991)
  MG20016M carrying pRH001This study
  MG21016M carrying pMG003, clumping strainThis study
 E. coli
  DH10BCloning strainInvitrogen Life-Technologies
  S17-1RP4-2, Tc∷Mu, Km-Tn7, for plasmid mobilizationSimon et al. (1983)
Plasmid
 pDONR201BP cloning vectorInvitrogen Life-Technologies
 pMG001pDONR201 carrying aiiD from B. suisThis study vs. Julien
 pMG002pDONR201 carrying aiiD from B. melitensisThis study vs. Julien
 pMR10-catBroad-host-range cloning vector, RK2 OriV CmR KanRA. A. Bourniquel
 pRH001pMR10 derivative gateway destination vector, medium copy number, CmRHallez et al. (2007)
 pMG003pRH001 derivative, Plac-controlled synthesis of AiiDB.suisThis study
 pMG004pRH001 derivative, Plac-controlled synthesis of AiiDB.melitensisThis study
 pBBR1-MCS1Broad-host-range cloning vector, CmRKovach et al. (1994)
 pRH002pBBR1-MCS1 derivative gateway destination vector, high copy number, rep, CmRHallez et al. (2007)
 pMG005pRH002 derivative, Plac-controlled synthesis of AiiDB.suisThis study
 pMG006pRH002 derivative, Plac-controlled synthesis of AiiDB.melitensisThis study

Molecular techniques

Plasmid construction

DNA manipulations were performed according to standard techniques (Ausubel et al., 1991). Restriction enzymes were purchased from Roche, and primers were purchased from Invitrogen.

Derivatives of the replicative plasmids pRH001 and pRH002 (Hallez et al., 2007) harboring aiiDsuis or aiiDmelitensis were constructed using the Gateway technique (Invitrogen). The destination vectors pRH001 and pRH002 harbor a chloramphenicol resistance (cat) marker and the toxic cassette ccdB. This group of genes is flanked by attR1 and attR2 recombination sites. The wild-type allele corresponding to the total AiiD protein of Brucella suis (amino acids 1–761) was amplified with primers AiiD-B1 (5′-ATGAACGTCGCGAGTGCC-3′) and AiiD-B2 (5′-AAGATGGCTGCATAATC-3′). The wild-type allele corresponding to the total AiiD protein of B. melitensis (amino acids 1–782) was amplified with primers AiiD-B3 (5′-ATGAACGTCGCGAGTGCC-3′) and AiiD-B4 (5′-AAGATGCCTGCATAATCAGG-3′). Brucella melitensis 16M genomic DNA was used as the template for all amplifications. The resulting PCR products (aiiDsuis and aiiDmelitensis, respectively) were cloned into pDONR201 (Invitrogen Life Technologies) by the BP reaction as described previously (Dricot et al., 2004). The resulting entry clones pMG001 and pMG002 were confirmed by PCR using primers AiiD-B1 and AiiD-B2 and primers AiiD-B3 and AiiD-B4, respectively.

Entry clones containing aiiD alleles were used together with the destination vectors pRH001 and pRH002 during Gateway LR reactions as described previously (Dricot et al., 2004). The resulting vectors pMG003, pMG004, pMG005 and pMG006 were transferred into the B. melitensis wild-type strain by mating.

Mating

Matings were performed by mixing 200 μL of E. coli S17-1 donor cell liquid culture (overnight culture) and 1 mL of the B. melitensis NalR recipient strain (overnight culture). Cells were centrifuged for 2 min at 4500 g and washed two times with 2YT. The pellets were resuspended in 10 μL of 2YT and spotted on a 2YT plate for 4 h. Bacteria were then transferred onto a 2YT plate containing Cm and Nal. After 3 days of incubation at 37 °C, the exconjugates were replicated on a 2YT plate containing Cm.

ConA-FITC labeling

For confocal microscopy, 0.1 mL of ConA-FITC (1 mg mL−1) was added to 0.2 mL of PFA-fixed cells. One microliter of propidium iodide (10 mM) was added for visualizing bacteria. After incubation for 30 min in the dark, cells were washed in phosphate-buffered saline (PBS) (pH 8.5), resuspended in 100 μL of the same buffer and examined immediately using a Leica SP-1 confocal laser-scanning microscope.

Purification of the exopolysaccharide

After bacterial growth, bacteria were shaken. Trichloroacetic acid was added to the culture to a final concentration of 4% (w/v) and stirred for 2 h at room temperature. Cells and precipitated proteins were removed by centrifugation (35 min, 22 000 g, 4 °C). The supernatant was collected and filtered through a Stericup filter (0.22 μm; Millipore). To precipitate exopolysaccharide, two volumes of cold ethanol 95% was gradually added to the filtered supernatant and incubated at 4 °C for 2 days. The exopolysaccharide was collected by centrifugation (30 min, 15 000 g, 4 °C) and dissolved in milliQ water. The aqueous solution of the exopolysaccharide was dialyzed (15 min, 2000 g three times) using the Centricon method (Amicon Ultra, Millipore; MW cut off 5 kDa). To remove free lipopolysaccharide and MVs-associated lipopolysaccharide, the exopolysaccharide sample was heated to 66 °C and gently mixed with one volume of hot phenol (66 °C). This sample was incubated 15 min at 66 °C before being centrifuged (30 min, 6500 g, 4 °C). The aqueous phase containing exopolysaccharide was extensively dialyzed (Millipore; MW cut off 1 kDa) against water for two consecutive days at 4 °C with two changes of water per day and the exopolysaccharide solution was subsequently lyophilized.

Quantification of exopolysaccharide

Quantification of exopolysaccharide was carried out using the anthrone colorimetric protocol (Morris, 1948). Briefly, 800 μL of anthrone solution [0.2 g anthrone (Sigma) in 100 mL of pure sulfuric acid] was added to 400 μL of exopolysaccharide samples. Samples were vortexed and incubated for 10 min at 37 °C. The absorbance was determined at 620 nm in a spectrophotometer. The total sugar concentration was determined by reporting the exopolysaccharide sample optical density on a reference curve realized with a glucose concentration range from 0 to 200 mg L−1.

Exopolysaccharide hydrolysis and HPLC

Before HPLC analysis, exopolysaccharide polymers were hydrolyzed into monomers by adding 1 mL of TFA 4 M to 1 mL of exopolysaccharide sample. The reaction was carried out for 2 h at 120 °C and TFA was removed by SpeedVac. The final exopolysaccharide sample was resuspended in 1 mL of dH2O.

NMR spectroscopy

1D and 2D NMR spectra of the exopolysaccharide in D2O (1 mg in 0.5 mL) were recorded at 70 °C on a Bruker AVANCE III 700 MHz spectrometer and on a Bruker AVANCE 500 MHz spectrometer, both equipped with 5 mm TCI Z-Gradient CryoProbes. 1H chemical shifts were referenced to internal TSP (δH 0.00) and 13C chemical shifts were referenced to external dioxane in D2O (δC 67.40). The 1D 1H,1H-TOCSY experiments were carried out with five different mixing times between 10 and 120 ms. The 1H,13C-HMBC experiment was performed with a 65-ms delay for the evolution of long-range couplings. Data processing was performed using vendor-supplied software. Measurement of the translational diffusion coefficient of the exopolysaccharide was carried out as described previously (Eklund et al., 2005).

Lipopolysaccharide detection

We used 50 mM Tris-HCl pH 7.5 and borate–10%NaCl (in some animals, up to 10% NaCl is necessary for the IgG to precipitate with the Brucella O-chain or NH, and borate buffers often help in the diffusion of polysaccharides). Exopolysaccharide was used at 5 mg mL−1 and tested with a pool of cattle sera that yields good precipitin bands with S Brucella polysaccharides (as a reference, with this pool of sera, B. melitensis lipopolysaccharide precipitates at about 1 mg mL−1, the O-PS down to 100 μg mL−1, and the pure NH down to 5 μg mL−1). Other sera were also tested from rabbits infected with B. melitensis (109 CFU intravenously) bled 3 months later, and from a rabbit infected with B. abortus 544 bled 6 months later. We also tested the exopolysaccharide in double-gel diffusion with a serum from a rabbit hyperimmunized with B. melitensis 115 (rough) that yields several precipitin lines with soluble proteins.

DNAseI assay

Brucella melitensis were grown for 20 h in 2YT medium at 37 °C. Cultures were then supplemented with 50 μg mL−1 DNaseI (Roche), incubated at 37 °C for different times and examined immediately by an agarose pad at appropriate times.

Microscopy

For DIC imaging, cell populations of B. melitensis strains were placed on a microscope slide that was layered with a pad of 1% agarose containing PBS (agarose pads) (Jacobs et al., 1999).

Samples were observed on a Nikon E1000 microscope through a differential interference contrast (DIC) × 100 objective with a Hamamatsu Orca-ER LCD camera. Images were taken and processed with Simple PCI (Hamamatsu).

Dot-blot analysis

Brucella were grown for 20 h in 2YT medium at 37 °C. Cultures were adjusted at the same OD600 nm before centrifugation to separate the supernatants from the cell pellets. Bacteria were concentrated 10-fold in PBS and inactivated for 1 h at 80 °C. Equivalent OD600 nm for each extract was used for serial twofold dilutions. Two microliters of each dilution was applied to a nitrocellulose membrane (Hybond; Amersham). OMP immunodetection was performed with the following monoclonal antibodies (MAbs) (ref.): anti-lipopolysaccharide (A76/12G12/F12) anti-Omp16 MAb (A68/08C03/G03) at 1/100 anti-Omp25 MAb (A68/4B10/F5) at 1/100, anti-Omp31 MAb (A59/10F9/G10) at 1/10 and anti-Omp36 Mab (A68/25G5/A5) at 1/100. Horseradish peroxidase-conjugated goat antimouse antibodies (Amersham) were used at 1/5000 along with the ECL system (Amersham) to develop blots for chemoluminescence before visualization on film. Dot blots using MAbs specific for Omp16 (PAL lipoprotein) were used as internal loading controls.

TEM

Brucella melitensis were grown for 20 h in 2YT medium at 37 °C. Bacteria pellets were fixed overnight in a solution containing 2.5% glutaraldehyde and 0.1 M phosphate buffer (pH 7.4). After fixation, cells were washed, postfixed with 1% osmium tetroxide for 1 h, washed again and subjected to serial dehydration with ethanol. Samples were embedded in resin, thin-sectioned and stained with uranyl acetate and Reynold's lead citrate. Finally, the samples were examined using a TEM (Technai 10; Philips) at the Unité Interfacultaire de Microscopie Electronique (University of Namur, Belgium).

Attachment assay on polystyrene surfaces

The surface attachment assay was performed using the crystal violet method, as described previously (O'Toole et al., 1999): 200 μL cultures were grown overnight in a 96-well polystyrene plate in 2YT medium. Plates were incubated at 37 °C for 20 h with agitation. Biofilm formation was assayed by the ability of cells to adhere to the polystyrene wells. The liquid medium was removed and the attached cells were washed with sterile PBS (pH 7.4). The attached bacteria were visualized by staining with 0.05% solution of crystal violet (GRAM'S solution; Merck) for 2 min at room temperature, followed by rinsing with water and air drying. Quantification of surface-attached bacteria was achieved by dissolving crystal violet in 200 μL of 100% ethanol. The ethanol was transferred and the volume was brought to 1 mL with dH2O and the absorbance was determined at 596 nm in a spectrophotometer (Genesys).

Infection of HeLa cells

The infections of HeLa cells were performed as described previously (Delrue et al., 2001). A ΔvjbR mutant was used as a negative control for replication defects during the cellular infection. Each infection was perfomed in triplicate.

SEM

The adherence of Brucella strains to monolayers of HeLa cells was performed on glass cover slips according to the protocol described in Castaňeda-Roldán et al. (2004). Plates were centrifuged for 10 min at 200 g at room temperature in a Jouan centrifuge and placed in a 5% CO2 atmosphere at 37 °C. After 1, 6, 24, 30 and 48 h of infection, wells were washed three times with PBS and incubated for 20 min with 4% paraformaldehyde to fix cells and bacteria. Then, samples were dehydrated for 2 × 5 min in 25%, 50%, 75%, 95% and 100% ethanol at room temperature. They were finally prepared by critical-point drying, mounted on an aluminum stub and covered with a thin layer of gold (20–30 nm). Examinations were carried out using a scanning electron microscope (XL-20; Philips, Eindhoven, the Netherlands) at the Unité Interfacultaire de Microscopie Electronique (University of Namur, Belgium).

Results

Brucella melitensis 16M overexpressing the AHL-acylase gene displays a clumping phenotype and produces exopolysaccharide(s)

Recently, a bioinformatic screen of Brucella genomes was carried out to find AHL-acylase homolog(s). One gene encoding a protein with 24.8% identity to AiiD, the AHL-acylase from Ralstonia sp. strain XJ12B, was identified and called aiiD (Lin et al., 2003). It has been shown that AiiD from Brucella is a functionally secreted Quorum-Quenching enzyme displaying a broad-range AHL-acylase activity (J. Lemaire, unpublished data).

We observed that a B. melitensis 16M strain (MG210) overexpressing aiiD exhibits a strong clumping phenotype in liquid culture. As the MG210 strain reached a high density in broth culture, bacteria aggregated and formed a pellicle-like structure that settled to the bottom of the culture tube. A similar phenotype was already described in B. melitensis vjbR-defective strains unresponsive to AHL (Uzureau et al., 2007). Because in these QS mutants, the clumps contain exopolysaccharide labeled by the Concanavalin A (ConA) lectin (Uzureau et al., 2007), which is specific for α-mannopyranosyl and α-glucopyranosyl residues (Naismith & Field, 1996), we wondered whether the strain MG210 could produce a similar exopolysaccharide. To this end, we attempted to label exopolysaccharide using ConA-FITC. Propidium iodide was used to counterstain bacteria in red.

As shown in Fig. 1, strain MG210 produced a ConA-FITC-labeled matrix not observed in the wild-type strain. This result shows that the MG210 aiiD-overexpressing strain is also able to produce exopolysaccharide containing α-mannopyranosyl and/or α-glucopyranosyl residues, like B. melitensis vjbR-defective alleles did (Uzureau et al., 2007). Although all these QS mutants display a similar phenotype, the MG210 strain formed larger and more stable clumps than the previously described strains. Thus, we focused our further characterization on the clumping phenotype of this MG210 strain.

Figure 1.

 Interactions between FITC-labeled ConA lectin and clumps formed by Brucella melitensis aiiD overexpressing strain MG210 visualized by confocal laser scanning microscopy. Culture samples of (a) MG200 (wild type bearing the control vector pRH001) and (b) MG210 (wild type bearing the aiiD overexpressing pMG003 plasmid) were stained with ConA-FITC (green) and propidium iodide (red). Scale bars: 8 μm.

Brucella melitensis exopolysaccharide is a mannan structure

We were interested in solving the nature of B. melitensis exopolysaccharide(s). Exopolysaccharide was extracted from MG210 cultures as described in Materials and methods. We first tested the purity of the exopolysaccharide preparation. To this end, we carried out a dot-blot analysis using specific MAbs (Cloeckaert et al., 1990) to compare the abundance of the lipopolysaccharide O-chain and two outer membrane proteins (OMPs) described on the OMVs formed by Brucella (Omp25 and Omp31) (Gamazo & Moriyon, 1987; Boigegrain et al., 2004) in exopolysaccharide samples taken before the first dialysis step in the phenol phase of the lipopolysaccharide removal step and in the final exopolysaccharide sample.

As shown in Fig. 2, we did not detect either the lipopolysaccharide O-chain or OMPs in the final exopolysaccharide preparation, showing that this sample is not contaminated with free lipopolysaccharide or OMVs. The phenol-based lipopolysaccharide removal step was nevertheless required because the lipopolysaccharide O-chain was detected in the phenol phase (Fig. 2, lane 3).

Figure 2.

 Dot blot analysis of lipopolysaccharide, Omp25 and Omp31 amounts in exopolysaccharide samples. Samples were tested at three steps of the exopolysaccharide purification protocol: before the first dialysis step (lane 1), final exopolysaccharide sample (lane 2) and phenol phase sample from the phenol–lipopolysaccharide removal step (lane 3). Two microliters of serial twofold dilution of each sample were spotted on a nitrocellulose membrane and dilutions were subjected to dot blot analysis using different MAbs. Data are representative of at least three experiments.

The absence of smooth lipopolysaccharide in the final exopolysaccharide sample was confirmed by double gel immunodiffusion against various immune sera. Neither sera from naturally infected cows nor sera from rabbit infected with B. melitensis 16M or Brucella abortus 544 yielded precipitin bands for the exopolysaccharide sample, indicating that the preparation was free from smooth lipopolysaccharide, lipopolysaccharide O-chain or even native hapten (NH) (data not shown). In addition, as sera from rabbit hyperimmunized by rough B. melitensis B115 also failed to show precipitin bands, the exopolysaccharide should almost be devoid of soluble contaminating Brucella protein (data not shown).

We then attempted to characterize the nature of the purified B. melitensis exopolysaccharide using two complementary approaches. We chose (1) to analyze the monomer composition by HPLC and (2) we appreciated the exopolysaccharide structure by nuclear magnetic resonance (NMR). (1) The purified exopolysaccharide was hydrolyzed with trifluoroacetic acid (TFA) and the resulting monomers were identified by HPLC. Three significant peaks corresponding in increasing quantity to glucosamine, glucose and mannose, respectively, were detected (Fig. 3). Traces of galactose could also be detected.

Figure 3.

 Determination of the monomeric components of Brucella melitensis exopolysaccharide by HPLC. Exopolysaccharides were purified, hydrolyzed into monomers and analyzed by HPLC. (a) HPLC chromatograms of the hydrolyzed exopolysaccharide (blue) and corresponding standards (black). (b) HPLC chromatograms of the hydrolyzed exopolysaccharide (blue) and the hydrolyzed exopolysaccharide supplemented with 10 mg L−1 mannose (black). (c) HPLC chromatograms of the hydrolyzed exopolysaccharide (blue) and the hydrolyzed exopolysaccharide supplemented with 10 mg L−1 xylose (black). The arrow indicates the apparition of the xylose peak. Standards were injected at 10 mg L−1 and Fructose was used as internal standard. Fru, fructose; Gal, galactose; Gln, glucosamine; Glu, glucose; Man, mannose; Xyl, xylose.

Because mannose and xylose present very close retention times and because xylose was present at 10 g L−1 in the initial medium, we undertook a second analysis to certify the nature of the monomer represented by the fourth peak. To this end, we mixed the hydrolyzed exopolysaccharide with either mannose (Fig. 3b) or xylose (Fig. 3c) standard in a 3 : 1 proportion. In both cases, the profiles obtained were compared with the hydrolyzed exopolysaccharide profile. As shown in Fig. 3b, the addition of mannose to the exopolysaccharide sample induced an increase in the fourth (mannose) peak. Conversely, the addition of xylose to the exopolysaccharide sample resulted in the appearance of a supplementary shoulder on the mannose peak (Fig. 3c). Taken together, these results demonstrate that the B. melitensis exopolysaccharide is composed of traces of galactose, glucosamine, glucose and mostly mannose. (ii) NMR analyses were carried out knowing that B. melitensis exopolysaccharide contains mannose : glucose : glucosamine in the relative ratio 89 : 10 : 1 obtained from the HPLC data. The 1H NMR spectrum was highly complex and showed that the material was quite heterogeneous. Major resonances from anomeric protons were observed between 4.5 and 5.3 p.p.m. The complexity of the material therefore required the use of 2D NMR techniques, in particular 1H,13C-HSQC and 1H,13C-HMBC experiments. The anomeric region of the 1H,13C-HSQC spectrum of the exopolysaccharide is shown in Fig. 4 and reveals nine major and three minor cross-peaks.

Figure 4.

 The anomeric region of the 1H, 13C-HSQC NMR spectrum of the exopolysaccharide from MG210 strain (see text for more details).

The 1H,13C-coupled version of this experiment was used to obtain one-bond 1H,13C-coupling constants that contain information about the anomeric configuration. Thus, the 13C anomeric resonances with chemical shifts <103 p.p.m. all had 1JC,H values >170 Hz, indicating α-anomeric configurations. Major cross-peaks were present at δHC 4.92/100.3, 5.07/102.9, 5.08/102.9, 5.08/99.1, 5.11/99.2, 5.16/102.9, 5.18/102.9 and 5.28/101.4; two minor cross-peaks were observed at δHC 5.05/99.3 and 5.46/96.9. The residue having its anomeric proton resonating at 4.53 p.p.m. had 3JH1,H2=8.0 Hz and its anomeric carbon observed at 103.5 p.p.m. showed 1JC,H≈160 Hz, indicative of the β-anomeric configuration. A series of 1D 1H,1H-TOCSY experiments starting from the anomeric proton of this residue revealed the complete spin system of a hexose residue, viz., δH 4.53 (H1), 3.36 (H2), 3.52 (H3), 3.47 (H4), 3.64 (H5), 3.87 (H6a) and 4.22 (H6b), which according to its chemical shifts, should be a glucosyl residue substituted at O6 (Jansson et al., 1994). The 1H,13C-HMBC spectrum revealed a trans-glycosidic correlation between H1 and C6 at 69.8 p.p.m. and an intraresidue one between C1 and H2, indicating that the material contains a chain of →6)-β-d-Glcp-(1→residues. In the 1H,13C-HSQC spectrum, a minor cross-peak was also present at δHC 4.36/103.9. The 1H,13C-HMBC spectrum revealed correlations at δHC 4.36/57.8 and 103.9/3.57, consistent with a 1H,13C-HSQC cross-peak at δHC 3.57/57.8. These results suggest the presence of an aminosugar, such as N-acetylglucosamine, which could be the primer from which the exopolysaccharide biosynthesis is started. The residues having their anomeric 13C chemical shifts <103 p.p.m. are consequently suggested to originate from mannosyl residues. Aided by the computer program CASPER (Jansson et al., 2006), which is used for the prediction of 1H and 13C NMR chemical shifts and for the structural analysis of oligo- and polysaccharides, further analysis was carried out. The chemical shifts of the anomeric 1H,13C-HSQC cross-peaks were in accord with different combinations of 2- and/or 6-substituted mannosyl residues. This conclusion was corroborated by correlations in the 1H,13C-HMBC spectrum at, inter alia, δHC 4.92/66.6, 5.07/79.4, 5.08/66.5, 5.08/79.0, 5.11/66.6, 5.11/79.4, 5.16/78.7 and 5.28/79.3. Thus, the major structural part is reminiscent of mannan structures present in oligo- and polysaccharides of bacterial and other origins (Briken et al., 2004; Lee et al., 2005; Omarsdottir et al., 2006; Prieto et al., 2007). In addition, the translational diffusion of the exopolysaccharide material was carried out and resulted in Dt=6.8 × 10−11 m2 s−1. Subsequently, the molecular mass can be calculated (Viel et al., 2003) to be 16 kDa, which corresponds to ∼100 sugar residues for the exopolysaccharide material.

Taken together, these results show that B. melitensis exopolysaccharide is a new mannose-rich polymeric structure.

Brucella melitensis clumps contain DNA

Besides exopolysaccharide, extracellular matrices often contain DNA, which may contribute to the structural integrity of biofilms (Whitchurch et al., 2002; Steinberger & Holden, 2005). To test whether Brucella's clumps include DNA, culture samples were incubated in the presence of DNAseI and the enzyme effect was observed under a microscope. Two hours after DNAseI incubation (Fig. 5b), clumps appeared to be digested by the nuclease while culture samples incubated with the enzyme buffer did not (Fig. 5a). This effect was increased after 24 h of incubation (Fig. 5c). Brucella melitensis wild-type strain or bearing a control vector (MG200 strain), used as negative aggregation controls, showed no effect of DNAseI treatment. These results demonstrate that DNA is a component of the extracellular matrix of B. melitensis aggregates and contributes significantly to their structure.

Figure 5.

 Effect of DNAseI treatment on clumps. DIC micrographs of the clumping strain in the presence of the DNAseI buffer (a) and after 2 h (b) or 24 h (c) DNAseI treatment. Scale bars: 3 μm.

Brucella melitensis clumping strain overproduces OMVs

Because a recent study showed that OMVs are classical components of biofilm matrices (Schooling & Beveridge, 2006), we wondered whether our MG210 clumping strain could overproduce OMVs. We tested this hypothesis using transmission electron microscopy (TEM). We analyzed the abundance of OMVs' structure in culture samples from MG210 and the wild-type strain collected in the stationary growth phase. Compared with the wild-type strain, we observed that the production of OMV-like structures was strongly increased in the clumping strain (Fig. 6a and b). Moreover, we took a set of minimum 20 TEM pictures for each strain on which we counted both the number of OMVs-like structures and the amount of bacteria to obtain quantitative data. Counting was performed in triplicate for each strain. As shown in Fig. 6c, we counted a mean of 73 OMVs per 100 bacteria in the aggregative strain, but only four OMVs per 100 bacteria in the wild-type strain. These data indicate that OMVs could be a component of the matrix of the clumps formed by B. melitensis as described for other biofilm matrices.

Figure 6.

 OMVs production in Brucella melitensis wild type and MG210 strains. (a) and (b) are micrographs of thin sections through embedded bacterial pellets obtained from culture samples of the wild-type strain (a) and the clumping MG210 strain (b). Arrows indicate some of the OMVs present in inter cellular spaces, as well as blebbing from the surfaces of cells. Scale bars: 1 μm. (c) Comparative OMVs counts per 100 counted bacteria in the B. melitensis wild-type strain and the MG210 clumping strain. Counting was made in triplicate for each strain. The data are representative of three experiments and error bars indicate SDs.

To confirm this hypothesis, we compared the abundance of two major OMPs of the OMVs formed by Brucella (Omp25 and Omp31) (Gamazo & Moriyon, 1987; Boigegrain et al., 2004) in B. melitensis wild-type and MG210 strains by dot-blot analysis using specific MAbs (Cloeckaert et al., 1990). Omp16 (PAL lipoprotein) was used as an internal loading control. Dot blotting was carried out with B. melitensis culture supernatants (containing the OMVs fraction) (Fig. 7) from stationary-phase cultures. OD600 nm were used to normalize all samples.

Figure 7.

 Dot blot analysis of OMP amounts in the supernatant cultures of Brucella melitensis 16M wildtype strain and in aiiD overexpressing clumping strain (MG210), respectively. Cells were grown 20 h in 2YT medium, OD at 600 nm were used to calibrate samples. Two microliters of serial twofold dilution of supernatants were spotted on a nitrocellulose membrane and dilutions were subjected to dot blot analysis using different MAbs. Data are representative of at least three experiments.

As shown in Fig. 7, the abundance of both tested OMPs of B. melitensis' OMVs is strongly increased in MG210 supernatants compared with the control strain. Omp16 presented almost the same relative abundance in the two strains tested. Here again, these results are in agreement with our previous observations showing that QS through VjbR is involved in the control of outer membrane composition (Uzureau et al., 2007). Altogether, these results suggest that the outer membrane composition is disturbed in the clumping strain and that OMVs could be overproduced in this strain.

The MG210 clumping strain presents increased adhesion properties

Because exopolysaccharide and extracellular matrices are responsible for the adhesive properties of bacteria (Quintero & Weiner, 1995), we compared the adherence abilities of the MG210 clumping and wild-type strains in a classical adherence assay. We tested the ability of the wild type, the wild type carrying the vector control (planktonic strains) and the exopolysaccharide-producing strains to attach to a 96-well polystyrene microtiter plate. In this assay, bacteria were inoculated in 2YT and grown at 37 °C overnight in this 96-well plate. Then, cells were removed, wells were rinsed and adherent bacteria were detected by crystal violet staining (see Materials and methods).

As shown in Fig. 8, the MG210 strain showed an adherence on polystyrene wells twofold stronger than both control strains. None of the strains showed significant differences in the growth rate that could potentially account for differences in adherent bacteria accumulation (data not shown). This result shows that the clumping strain possesses an increased ability to adhere to polystyrene surfaces. The direct involvement of the exopolysaccharide in surface adherence is still to be demonstrated.

Figure 8.

 Surface attachment by wild type, MG200 (wild type bearing the control vector pRH001) and MG210 (wild type bearing the aiiD overexpressing pMG003 plasmid) strains. Quantification of the microtiter plate biofilm formation by crystal violet solubilization in ethanol and spectrophotometric determination of Abs at 596 nm. Data are the average of five independent samples, and error bars indicating SDs are shown.

Finally, we compared the adhesion of the B. melitensis wild-type strain and B. melitensis MG210 strain to cells, a biotic surface that Brucella spp. encounter during their infectious cycle. HeLa cells were infected with an equal quantity of bacteria of the wild type or the MG210 clumping strain. After 1, 24 and 48 h of infection, cells were observed by scanning electron microscopy (SEM) and the number of intracellular bacteria was evaluated. Here again, while no difference in either internalization or intracellular replication could be found between both strains (Fig. 9), we observed that as early as 1 h postinfection, the AHL-acylase overexpressing strain is strongly adherent to HeLa cells compared with the parental one: several clumps from different sizes are observable both on coverslips and on the surface of the cells in the MG210-aggregating strain (Fig. 10).

Figure 9.

 Intracellular replication of Brucella melitensis 16M wild type, MG200 (wild type bearing the control vector pRH001) and MG210 (wild type bearing the aiiD overexpressing pMG003 plasmid) strains in HeLa cells. Data are the averages of log number of CFU per well±SD (n=3) from a representative experiment.

Figure 10.

 Scanning electron micrographs showing that Brucella melitensis MG210 displays increased adhesion properties to HeLa cells as compared with B. melitensis wild type. Pictures were taken after 24 h (a and c) and 48 h (b and d) of infection. A zoom on a part of each picture showing bacteria is presented in the upper right corner of each quadrant. Scale bars: 10 μm.

Discussion

This work provides the first insights into the composition and the preliminary structure of the exopolysaccharide overproduced in B. melitensis strains affected in the AHL communication system. These strains exhibit a clumping phenotype not only because exopolysaccharide is overproduced but also because the aggregates contain extracellular DNA (eDNA). In addition to exopolysaccharide and eDNA, the clumping strain was shown to overproduce OMVs. The aggregative strain was also demonstrated to possess increased adherence properties both to polystyrene and to HeLa cells compared with the wild-type strain.

We reported previously that the QS-dependent regulator VjbR is a major regulator of important cell surface structures of B. melitensis (type IV secretion system, flagella, OMPs, exopolysaccharide) and that mutations in this regulator lead to clumping in liquid culture (Uzureau et al., 2007). Here, we show that the overexpression of the newly described AHL-acylase aiiD of Brucella (J. Lemaire, unpublished data) leads to a similar or an even stronger clumping phenotype. This observation is not unexpected because both types of strains are unresponsive to AHLs: the vjbR-defective strains [both the vjbR(D82A) and the vjbR1-180) alleles] are unable to bind C12-HSL (Uzureau et al., 2007) and the aiiD-overexpressing strain degrades all the synthesized C12-HSL, leading to constantly unbound VjbR regulators. These related strains produce at least one exopolysaccharide with d-mannose or d-glucose residues as demonstrated by the ConA-FITC labeling of the clumps (Uzureau et al., 2007 and Fig. 1).

Exopolysaccharide production and aggregate formation is a classical feature in several Alphaproteobacteria and Brucella does not seem to be an exception to the rule. For example, the plant pathogen Agrobacterium tumefaciens has been shown to produce an exopolysaccharide called succinoglycan (Stredansky & Conti, 1999) and Sinorhizobium meliloti has been reported to produce succinoglycan and galactoglucan, both required for its full virulence (Leigh et al., 1985; Glazebrook & Walker, 1989). The B. melitensis exopolysaccharide we have characterized in this paper is mainly composed of a combination of 2- and/or 6- substituted mannosyl residues with minor amounts of glucose, glucosamine and maybe galactose that build up chains of around 100 sugars. Mannose seems to be a privileged sugar in Brucella extracellular oligo- or polysaccharidic structures as the core of the lipopolysaccharide contains mannose (Velasco et al., 2000) and the O-chain of the lipopolysaccharide is a homo-polymer of 4,6-dideoxy-4-formamido-d-mannose (N-formylperosamine) (Perry & Bundle, 1990). In B. melitensis biovar 1, the N-formylperosamine homopolymer is composed of repeating blocks of five sugar residues, four α-(1→2)-linked and one α-(1→3)-linked (Aragón et al., 1996).

Biofilms of several bacterial species have been shown previously to contain eDNA (Whitchurch et al., 2002; Vilain et al., 2009). DNAse treatment of B. melitensis clumps led to their efficient dissociation, demonstrating the involvement of eDNA in the aggregates. The origin of the eDNA remains to be determined. eDNA can result from a lysis of a subpopulation of cells in the aggregate (Allesen-Holm et al., 2006) or can be actively released from living cells (Dillard & Seifert, 2001; Renelli et al., 2004).

Brucella melitensis exopolysaccharide is probably not the only surface structure involved in clumping because the outer membrane composition showed strong differences between the wild-type and the MG210 clumping strains. The production of Omp25 and Omp31 is increased in the later strain. This result correlates with our previous observations showing that VjbR regulates the production of these OMPs (Uzureau et al., 2007). This alteration of the outer membrane composition is probably linked to our TEM observations, revealing that OMVs-like structures are strongly overproduced in the MG210 clumping strain. Several roles for OMVs have been reported including involvement in DNA and QS-pheromone transport in P. aeruginosa (Renelli et al., 2004; Mashburn & Whiteley, 2005). Whether Brucella OMVs could play such a role and be directly involved in the matrix production remains to be explored. Together with exopolysaccharide and eDNA, these OMVs are the third structural element, classically described in extracellular biofilm matrices, that we have identified in B. melitensis clumps.

In addition to promoting adhesion of bacteria to neighboring cells, the sticky matrix components also contribute to surface adhesiveness. Therefore, it is not surprising that the clumping strain MG210 presents better adhesion properties than the wild-type strain both on polystyrene and on HeLa cells (Figs 8 and 10). The exact nature of the initial adhesin and the stepwise process leading to cell aggregation remain to be determined.

As we discussed in our previous publication (Uzureau et al., 2007), the ability of B. melitensis to form biofilm-like structures could have several advantages in its life cycle. If we consider that B. melitensis is a facultative intracellular pathogen able to survive for months outside the host on inert surfaces (Spink, 1956), we could easily imagine a protective role for the exopolysaccharide against desiccation and other environmental stresses encountered, as described in Nostoc commune (Tamaru et al., 2005) or Campylobacter jejuni (Joshua et al., 2006). Nevertheless, as the genome and the molecular infectious strategies of Brucella spp. are very close to those of S. meliloti and considering the role of the exopolysaccharide in S. meliloti, we hypothesize a role for Brucella clumping and/or exopolysaccharide production during its infectious cycle in the host. When aggregated Brucella spp. enter in contact with their host, exopolysaccharide could offer them protection against the extracellular immune system (as described for Streptococci (Marques et al., 1992) and help them to adhere to host cells (such as Neisseria gonorrhoeae; Greiner et al., 2005). In this regard, the adhesion we observed on HeLa cells with the MG210 strain is somehow reminiscent of the localized bacterial microcolonies of B. abortus adherent to epithelial cells depicted recently (Castaňeda-Roldán et al., 2004). The exopolysaccharide could also be involved in the earliest steps of the host trafficking as described for succinoglycan in S. meliloti (reviewed in Fraysse et al., 2003). Finally, considering the variety of eukaryotic proteins dedicated to ‘mannose’ recognition (Ip et al., 2009), the expression of an exopolysaccharide rich in mannose could well not be coincidental and could be part of the Brucella stealthy strategies to deviate the host immune response. Microbial mannans are well-known immunomodulators (Gilleron et al., 2005; Dinadayala et al., 2006). In addition, given that biofilm formation is at the root of many persistent and chronic infectious diseases (Costerton et al., 1999), the chronicity of brucellosis could be linked to the biofilm-like formation ability of B. melitensis.

Although we demonstrated that MG210 and wild-type strains do not behave in a different way either in a cellular model (Fig. 9) or in a mouse model of infection (data not shown), we cannot exclude a role for B. melitensis exopolysaccharide in vivo as mice were infected intraperitoneally, which does not reflect the natural entry route of Brucella. Moreover, among all the possible signals and regulatory pathways involved in biofilm formation, we only demonstrated a role for the QS and the AHLs in B. melitensis clumping. Other signals also probably need to be taken into account, and their discovery will help to identify the situations triggering the wild-type strain to produce exopolysaccharide and form clumps. The identification of the genes involved in the biosynthesis of B. melitensis exopolysaccharide, together with the environmental signals to which they respond in the intricate regulatory processes leading to the clumping phenotype, will help to determine the precise role of the exopolysaccharide. When looking to the B. melitensis 16M genome, several candidates involved in exopolysaccharide biosynthesis have emerged and their potential role in exopolysaccharide synthesis is actually under characterization.

Acknowledgements

We are grateful to C. Didembourg for helpful technical assistance and advices. We thank the past and present members of the Brucella team of the URBM for fruitful discussions. We also thank the Unité de Recherche en Biologie Cellulaire, the Unité Interfacultaire de Microscopie Electronique and the Unité de Recherche en Biologie Végétale (University of Namur, Belgium) for their welcome and help with use of the confocal microscope and lyophilization, the transmission and scanning electron microscopes and the HPLC, respectively.

M.G., A.M. and S.U. hold a specialization grant from the Fonds pour la Formation à la Recherche dans l'Industrie et l'Agriculture (FRIA).

This work was supported by grants from the Swedish Research Council (VR), The Knut and Alice Wallenberg Foundation and Magn. Bergvalls Stiftelse.

Ancillary