In an attempt to obtain pure and well characterized smooth lipopolysaccharide (S-LPS) and rough lipopolysaccharide (R-LPS), smooth and rough strains of Brucella abortus were extracted by two different modifications Qf the phenolwater method. S-LPS was obtained in the phenol phase, and R-LPS was obtained in the aqueous phase. Further purification was accomplished by treatment with enzymes, detergents, NaI as a chaotropic agent to separate non-covalently bound contaminants, and by gel filtration. The degree of purity of the molecules was determined by chemical and immunological analysis and by electrophoresis in sodium dodecyl sulfate-polyacrylamide gels. Lipid identification by gas-liquid chromatography showed seven major fatty acids. Palmitic acid accounts for about 50%, stearic acid accounts for about 10%, and hydroxylated fatty acids account for less than 5% of total fatty acids. 2-Keto-3-deoxyoctonate but not heptose was detected in the sugar analysis. Protein was found to be firmly bound to S-LPS but not to R-LPS. Several investigators have shown that the crude endotoxin from smooth Brucella spp. is isolated from the phenol phase when cells are extracted with hot phenol-water (2, 3, 6, 18, 22, 23, 30). This fraction, containing 2-keto-3-deoxyoctonate (KDO), sugars, and lipids, presents both similarities and differences when compared with lipopolysaccharides (LPS) from Enterobacteriaceae. Examples of such similarities and differences include observations on toxicity, pyrogenicity, and hybrid formation with Escherichia coli LPS. Baker and Wilson (1, 2) found that endotoxin preparations of smooth Brucella abortus were toxic for mice, but were less toxic than those from E. coli. Leong et al. (30) described qualitative differences in biological activity but not structural differences between brucella endotoxins and enterobacterial endotoxins. Jones et al. (22) reported that saline-extracted rough LPS (R-LPS) from B. ovis was toxic for mice and had limulus lysate gelation activity (LLGA) comparable to that of E. coli LPS. Munoz et al. (36) showed that an endotoxin preparation of smooth B. abortus had comparable LLGA to LPS from several different strains of Salmonella and E. coli, but did not increase sensitivity of mice to histamine. Because these experiments were performed with crude brucelia endotoxin preparations rather than with purified LPS, it was not possible to establish any relationship between chemical composition and biological activity that could help to explain the differences in behavior of these LPS molecules. The work reported here describes preparation and analytical data for purified smooth LPS (S-LPS) and R-LPS from B. abortus which is essential to experiments on biological activity. These, in turn, may contribute to the understanding of the relationship of chemical composition to activity. MATERIALS AND METHODS Bacterial cultures. The bacterial strains used, their characteristics, and the conditions of culture were described previously (22). Extraction of crude LPS. The crude S-LPS (fM) was extr...

Purified lipopolysaccharide (LPS) extracted with phenol-water from smooth Brucella abortus was hydrolyzed with 1% acetic acid at 1000C. The degraded polysaccharide (AH) released gave reactions of identity with the native polysaccharide hapten (NH) in phenol-wateror trichloroacetic acid-extracted endotoxin preparations of B. abortus and with the polysaccharide (poly B) extracted by trichloroacetic acid from rough B. melitensis strain B115. Poly B was present in the soluble cytoplasmic fraction but not in the membrane fraction, of disrupted B115 cells. It could not be extracted from three rough mutants of B. abortus or from B. canis or B. ovis cells. Both AH and NH shared determinants present on smooth LPS and missing from poly B. Sugars found in purified LPS, NH, and AH included mannose, glucose, quinovosamine, glucosamine, and 2-keto-3-deoxyoctonate. Poly B contained only a trace amount of quinovosamine and no 2keto-3-deoxyoctonate detectable by the thiobarbiturate assay. Sera from some rabbits immunized with pure smooth LPS and some, but not all, cows infected with field strains of B. abortus recognized the determinants missing from poly B. A subclass-specific enzyme-linked immunoassay showed that most of the antibody in sera from infected cows which binds to smooth LPS and to NH is of the immunoglobulin Gl subclass.

Purified lipopolysaccharide (LPS) from smooth (S) and rough (R) strains of Brucella abortus and lipid A isolated from S-LPS by mild acid hydrolysis were examined in several assays of biological activity. Brucella Sand R-LPSs and Brucella lipid A activated the complement cascade. Previously reported mitogenic activation by Brucella LPSs of spleen cells from endotoxin-resistant C3H/ HeJ mice was confirmed and also produced by isolated Brucella lipid A. Mitogenicity was not inhibited by polymyxin B, and amino acid analysis showed no binding of polymyxin B to Brucella LPS under conditions in which mitogenicity of phenol-water-extracted Escherichia coli LPS was inhibited. S and R Brucella LPSs and lipid A all produced equivalent polyclonal stimulation of C3H/HeJ and C3H/HeAU spleen cells. Crude and purified LPS from S but not from R B. abortus was toxic for outbred mice, with 50% lethal doses approximately six times greater than that for E. coli LPS. Sand R-LPSs were abortifacient in pregnant outbred mice. S Brucella LPS was lethal for carrageenen-pretreated C3H/HeJ and C3H/HeAU mice, whereas only C3H/HeAU mice were killed by E. coli LPS. The data are consistent with the hypothesis that the unique fatty acid composition of Brucella lipid A is responsible for its biological activity in endotoxin-resistant C3H/HeJ mice. The participation of the protein strongly bound to the lipid A cannot be excluded, but its mode of action, if any, is different from that of the lipid A-associated protein of enterobacterial LPS.

Lipopolysaccharides (LPS) were extracted from rough strains of Brucella abortus and Brucella melitensis and from strains of the naturally occurring rough species Brucella ovis and Brucella canis. Brucella rough lipopolysaccharides (R-LPS) were readily distinguished from Brucella smooth lipopolysaccharides (S-LPS) and enterobacterial R-LPS, by their chemical, physical, and serological characteristics. B. ovis R-LPS was differentiated from B. abortus, B. melitensis, and B. canis R-LPS by its reaction of partial identity in immunodiffusion. Monospecific mouse sera against B. ovis R-LPS agglutinated only the homologous bacteria but not R cells of other species ofBrucella. B. ovis R-LPS contained more 2-keto, 3-deoxyoctonate, and glucosamine as a percentage of dry weight than any other R-LPS tested. B. abortus R-LPS was identified by the absence of an unidentified sugar present in the other R-LPS molecules, and B. melitensis R-LPS could be differentiated from B. canis R-LPS by its higher content of fatty acids. In contrast to S-LPS, all of the R-LPS studied lacked quinovosamine. In electron micrographs, Brucella R-LPS had a granular appearance, in contrast to typical lamellar structures formed by Brucella S-LPS and Escherichia coli R-LPS.

An enzyme-linked immunosorbent assay was developed to follow the bovine response, by immunoglobulin class and subclass, to defined smooth and rough lipopolysaccharides (LPS) of Brucella abortus. Binding to smooth LPS of immunoglobulin G1 (IgG1) and IgG2 in sera from Brucella-infected animals was significantly greater than binding in sera from normal uninfected animals. Competition or steric blocking among IgM, IgG1, and IgG2 for binding sites on smooth LPS was shown to occur. Binding of IgM to Brucella smooth LPS with sera from uninfected animals was elevated above the assay control levels, and attempts to eliminate this nonspecific IgM binding were not successful. The same levels of nonspecific IgM binding were also seen with Brucella rough LPS, Escherichia coli LPS, and Pseudomonas solanacearum LPS. Sera from some, but not all, Brucella-infected animals showed elevated binding of IgG1 and IgM to both E. coli LPS and Brucella rough LPS as well as to Brucella smooth LPS. This was interpreted as specific antibody. Cross-reactions between B. abortus smooth or rough LPS and E. coli LPS could not be shown by immunodiffusion.