TEXT

Top Abstract Text References

The genus Brucella is a gram-negative, facultative, intracellular pathogen that produces disease in a large number of mammals, including humans (8). The outer membrane (OM) of Brucella, which has been implicated in the virulence of the species, is unique in many respects. It has been demonstrated, for instance, that this layer is permeable to hydrophobic permeants and is considerably more resistant to bactericidal cationic peptides than most gram-negative bacteria (9, 15, 16). In smooth pathogenic Brucella species (Brucella abortus, Brucella melitensis, and Brucella suis), the polysaccharide O chain of the lipopolysaccharide (LPS) has been implicated as a virulence factor. This proposition is based on the observation that smooth Brucella survives and replicates in animals and in cultured macrophages more efficiently than rough mutants, and on the increased susceptibility of rough strains to microbicidal cationic peptides relative to that of their smooth counterparts (14, 16, 31).

Brucella ovis and Brucella canis, pathogenic for rams and dogs, respectively, do not possess O-chain polysaccharides; their LPSs react with antibodies against core epitopes specific to the genus and are lysed by the R/C phages specific for rough Brucella strains (3, 5, 8). Due to the fact that these properties are shared by mutants derived from smooth strains, these two Brucella species have been regarded as natural rough variants (8). Although B. ovis has been observed within trophoblasts of placentomes from infected sheep (17), this species attaches in lower numbers to epithelial HeLa cells than the rough mutants B. abortus 45/20 and B. abortus RB51 (22, 27). B. abortus, moreover, is able to proliferate inside nonprofessional phagocytes and induce loss of viability of the infected cells, while B. ovis is readily destroyed within lysosomes of these cells and does not induce cellular death (22). Since the differences observed between the natural and mutant rough Brucella strains cannot be due to their roughness, we have asked whether this distinct behavior may be correlated with different OM properties displayed by these strains. For this purpose, we have compared the permeabilities of the OM, the susceptibilities to bactericidal cationic peptides, and the levels of binding of the LPS to polymyxin B for mutant rough B. abortus strains and natural rough B. ovis.

The growth conditions of the bacterial strains used, as well as the purification and characterization of their LPS molecules, have been described elsewhere (6, 10, 15, 16, 18). Briefly, smooth B. abortus S19, rough B. abortus 45/20, and rough B. ovis REO 198 (CO₂ independent) strains were originally obtained from Lois Jones (University of Wisconsin, Madison, Wis.). Attenuated rough B. abortus RB51 was obtained from Gerhardt Schurig (Virginia Polytechnic Institute and State University). Smooth Salmonella montevideo SH94 serogroup D1 is maintained as part of the collection of the Division of Clinical Bacteriology, Karolinska Institute, Huddinge, Sweden. All strains are maintained in lyophilized stocks at the Veterinary School (National University, Heredia, Costa Rica, and University of Navarra, Pamplona, Spain) and have been demonstrated to be stable strains throughout the years, without detectable phenotypic or biochemical changes (2, 9, 10, 15, 16, 18, 25). B. abortus and S. montevideo strains were propagated in tryptic soy broth, and B. ovis was propagated in the same medium with 0.5% yeast extract (Difco Laboratories, Detroit, Mich.). Bacteria were harvested (5,000 × g for 15 min at 4°C) in the exponential phase of growth. Attenuated B. abortus S19 is phenotypically a smooth strain with 90% smooth-type LPS, 10% rough-type LPS (10), and a considerable quantity of surface native hapten (NH) polysaccharide (2, 25). The biological, chemical, and physical characteristics of B. abortus S19 LPS are indistinguishable from those of preparations isolated from virulent strains (10). Attenuated B. abortus 45/20 strain contains more than 99% rough-type LPS and a small number of lipid-bound O-polysaccharide-containing molecules (4, 10, 26), which has been estimated by immunogold electron microscopy with monoclonal antibody against epitope C/Y of the O chain (9) to be from 0 to 12 molecules per cell (Fig. 1). This bacteria does not contain any detectable NH (18). B. ovis REO 198 does not possess O chain or NH as demonstrated by immunogold detection (Fig. 1) and by sodium dodecyl sulfate-polyacrylamide gel (SDS-PAGE) analysis (18, 25). The rough-type LPSs of Brucella strains demonstrate extensive immunological cross-reactivity and similar chemical and physical properties (18, 25). Cationic protein 18A, bactenecin 5, and bactenecin 7 were provided by R. Gennaro and D. Romeo (Department of Biochemistry, Biophysics and Chemistry of Macromolecules, Trieste University, Trieste, Italy) or synthesized by Chiron Mimotopes Pty. Ltd. (Victoria, Australia). The polymyxin B sulfate, dansyl-polymyxin B, melittin, poly-L-lysine, and poly-L-ornithine were purchased from Sigma Chemical Co. (St. Louis, Mo.). Lactoferricin B was provided by W. Bellamy (Morinaga Dairy Company, Higashihara, Japan). The bactericidal sensitivity assays (results expressed as either the number of viable CFU or the diameters of bactericidal halos in agar plates) and the method for adsorption of the peptides to different bacteria (reported as the reduction of bactericidal halos in agar plates) were performed as described previously (9). Fluorimetric assays of peptide-treated bacteria with N-phenyl-naphthylamine (NPN) as the fluorescent probe were performed as reported by Martínez de Tejada and Moriyón (15). Binding of polymyxin B to LPSs isolated from rough and smooth Brucella strains was estimated by fluorometric analysis. Briefly, LPS suspensions (yielding concentrations of 60 to 63 nM lipid A in 2.5 mM HEPES, pH 7.2, and prepared by sonic dispersion) were incubated with different concentrations (5.5, 8.25, and 13.75 µM) of dansyl-polymyxin B. The fluorescence was estimated at room temperature under conditions of excitation at 340 nm, with an LS-50 fluorimeter (Perkin-Elmer Ltd., Beaconsfield, England) with a slit width (for both excitation and emission) of 2.5 nm and a range of 400 to 600 nm. Bacterial cell damage was evaluated by observation of the peptide-treated bacteria on a Hitachi 1100 transmission electron microscope (Hitachi Scientific Instruments, Mountain View, Calif.) operating at 100 kV as previously described (9). Experiments were performed in quadruplicate, and the results were expressed either as the percentage in the reduction of the bactericidal activity inhibition or the lethal concentration of the peptides in micrograms per milliliter (mean ± standard deviation) with respect to the control. Both the Student t test and variance analysis were performed for statistical examination.

View larger version (73K): [in this window] [in a new window]   FIG. 1.   Immunogold detection of O chain (C/Y) and core polysaccharide epitopes (R) in the surface of rough Brucella. (A) B. ovis REO 198 stained with anti-O-chain (C/Y epitope) monoclonal antibody and anti-mouse immunoglobulin gold (5 nm) conjugate; (B) B. abortus 45/20 stained with anti-O-chain (C/Y epitope) monoclonal antibody and anti-mouse immunoglobulin gold (5 nm) conjugate; (C) B. ovis REO 198 stained with anti-R (R1 epitope) monoclonal antibody and anti-mouse immunoglobulin gold (5 nm) conjugate; (D) B. abortus 45/20 stained with anti-R (R1 epitope) monoclonal antibody and anti-mouse immunoglobulin gold (5 nm) conjugate. Bar, 0.125 µm.

Just as the reduction of CFU showed the bactericidal action of cationic peptides in various Brucella strains (9, 11, 16), the MICs showed that the rough B. abortus 45/20 and RB51 strains were more susceptible than their smooth counterparts, although B. abortus displayed more resistance than B. ovis (Table 1). The results of the adsorption experiments were consistent with those of the bactericidal assays in that B. ovis cells adsorbed more peptide than did the rough B. abortus cells (Table 2). Moreover, B. ovis LPS bound considerably more polymyxin B than did LPSs from both rough and smooth B. abortus strains (Fig. 2). As with the binding of cationic peptides by Brucella cells, LPS molecules from rough strains captured more polymyxin B than LPS from smooth B. abortus, as demonstrated by the increased fluorescence at low concentrations of dansyl-polymyxin B (Fig. 2). At higher concentrations, quenching due to turbidity of the dansyl-polymyxin B-LPS suspensions was observed for B. abortus 45/20 LPS (optical density [OD] at 400 nm = 0.723), and to a lesser extent for B. abortus RB51 (OD at 400 nm = 0.357). Little quenching was observed with B. ovis (OD at 400 nm = 0.288) and smooth B. abortus (OD at 400 nm = 0.157) LPSs. These results are consistent with micelle size and solubility of rough B. abortus LPS with respect to those of LPS from smooth strains (18).

View larger version (23K): [in this window] [in a new window]   FIG. 2.   Binding of polymyxin B by LPSs from different Brucella strains. LPS suspensions adjusted to 60 to 63 nM concentrations of lipid A were incubated with a 13.75 µM concentration of dansyl-polymyxin B, and the fluorescence was estimated at a range of 400 to 600 nm. The inserted table shows the binding of different concentrations of dansyl-polymyxin B of the different LPSs at 480 nm. Dansyl-polymyxin B alone did not produce detectable fluorescence under these conditions. RFU, relative fluorescence units.

In previous investigations it was observed that, in contrast to enterobacterial OM, the Brucella spp. OMs were not barriers to hydrophobic permeants and that bactericidal cationic peptides did not alter the kinetics during the entrance of the hydrophobic probe NPN (9, 15). Figure 3 shows that the NPN partition kinetics in the OM of B. ovis differs considerably from that in the OM of B. abortus. None of the peptides tested altered the partition of NPN in the OM of Brucella strains, although the abrupt partition of NPN in the B. ovis OM makes it difficult to evaluate the action of the peptides in this bacterium.

View larger version (13K): [in this window] [in a new window]   FIG. 3.   Fluorimetry assays of bacteria (B. ovis and B. abortus 45/20 or RB51) treated with bactenecins 5 and 7, lactoferricin B, or polymyxin B. The NPN uptake profiles of B. abortus 45/20 and B. abortus RB51 are practically the same. The arrow indicates the time of addition of NPN or NPN and the peptide. No effect was observed when NPN or NPN plus peptide was added. RFU, relative fluorescence units.

Transmission electron microscopy demonstrated that the effects induced by bactenecins 5 and 7 were more evident in S. montevideo than they were in rough Brucella (Fig. 4 and 5). The effects of bactenecins 5 and 7 on rough B. abortus and on B. ovis appeared to be similar (Fig. 4). The morphological changes observed in peptide-treated Brucella were mild and evident only at high concentrations of the bactericidal agents. These changes were characterized by detachment of the internal membrane, vacuolization, and the appearance of electron-dense bodies within the cytoplasm. Most of the time, the OM maintained its appearance. Poly-L amino acids produced a thick coating on rough B. abortus and B. ovis, being more conspicuous in the latter than in the former bacteria (Fig. 4). This phenomenon hampered the infiltration of the resin, making observation difficult. In spite of this problem, minimal morphological effects were observed in the Brucella OM treated with poly-L amino acids. Some spotted precipitation and vacuolization of the cytoplasm was evident; however, the significance of these was unclear. In contrast, poly-L amino acids produced severe damage on S. montevideo cells, characterized by severe vacuolization, formation of cytoplasmic electron-dense bodies, and distortion of the cell shape (Fig. 5). These effects were more evident in the bacterial poles. Treatment with a dose of polymyxin B lethal for S. montevideo (Fig. 5) did not affect the OM structure of the rough Brucella strains. The use of large quantities of polymyxin B resulted in the deposition of an electron-dense layer on the surface of B. ovis with little detectable cell damage (Fig. 4), an effect that may be related to the augmented binding of polymyxin B by the LPS of this rough Brucella (Fig. 2). Control experiments showed that insoluble polymyxin B in agarose was stained with uranyl and osmium salts in a pattern similar to that of the electron-dense coat observed around polymyxin B-treated B. ovis (data not shown). No morphological differences were observed between smooth and rough B. abortus control strains.

View larger version (67K): [in this window] [in a new window]   FIG. 4.   Transmission electron microscopy of peptide-treated rough Brucella. Rows 1 and 2 show B. ovis REO 198, and rows 3 and 4 show rough B. abortus 45/20. (A) Untreated controls (rows 1 and 3) and bacteria treated with 20 µg of polymyxin B (rows 2 and 4); (B) bacteria treated with 20 µg of bactenecin 5 or 20 µg of poly-L-lysine (rows 2 and 4); (C) bacteria treated with 20 µg of bactenecin 7 (rows 1 and 3) or 20 µg of poly-L-ornithine (rows 2 and 4). Bar, 0.25 µm.

View larger version (141K): [in this window] [in a new window]   FIG. 5.   Transmission electron microscopy of peptide-treated smooth Salmonella. Row 1 shows the following: untreated S. montevideo SH94 (A) and Salmonella treated with 10 µg of bactenecin 5 (B) or 10 µg of bactenecin 7 (C). Row 2 shows results of treatment with 10 µg of polymyxin B (A), 10 µg of poly-L-lysine (B), or 10 µg of poly-L-ornithine (C). Bar, 0.25 µm.

The OM structure of Brucella organisms differ from that of other pathogenic groups (6, 15, 16, 19), including several facultative intracellular bacteria such as Salmonella or Shigella spp. However, within the genus Brucella, which is a close monophyletic bacterial group (28), most of the OM components have been found to be very similar (19). Some of the structural features of the Brucella OM correlate well with the observed permeability of the Brucella OM to hydrophobic compounds (9, 15), and with the resistance of these bacteria to EDTA, Tris, some detergents (20), and oxygen-dependent and -independent killing mechanisms (13, 16, 24). Features which have been implicated in this resistance are the O chain (linked to the core lipid A) and the NH, both being characteristic of smooth Brucella strains (2, 9). Since the resistance of B. ovis and rough B. abortus strains to cationic peptides has been demonstrated to be lower than that displayed by the smooth Brucella, we concluded that the lack of any O chain in the LPS and the absence of NH are factors which contribute to the sensitivity of the OM. It is unlikely that the small amounts of O chains present on the surface of Brucella 45/20 account for this difference. Findings supporting this proposition include the almost identical NPN kinetics displayed by the rough B. abortus RB51 (which does not possess any O chain) and the B. abortus 45/20 cells, and the similar levels of binding of polymixin B of their LPSs (Fig. 2 and 3). Other OM characteristics may also participate in this susceptibility, however, since the pathogenic B. ovis is more affected by the peptides than the rough B. abortus. In this respect, it has been demonstrated that naturally occurring rough Brucella strains expose a uniform set of anionic sites on their surfaces, which are easily visualized with positively charged probes (30). Although the peptides did not have any clear effect on the uptake of NPN in the Brucella strains, the partition of this hydrophobic probe on the OM of B. ovis was strikingly different in smooth and rough B. abortus mutant strains (Fig. 3 and reference 15). The sudden uptake of NPN by the OM of B. ovis marked a definitive difference between this species and the other Brucella spp. In conclusion, the higher susceptibility of B. ovis to cationic peptides, the increased uptake of hydrophobic probes, the augmented polymyxin binding of its LPS, and the distinct structural properties of its OM correlate with the different biological behavior of this rough strain in animal hosts and culture cells (Table 3). We must be cautious, however, since other factors not necessarily related to the structure of the OM may also participate in the host preferences of B. ovis, the most differentiated species among the genus Brucella (19).