Lipopolysaccharides (LPS) were extracted from seven Bacteroides strains by three different techniques: the phenol-water (PW), phenol-chloroform-petroleum (PCP) and Triton-Mg2+ methods. The strains selected included two different B. fragilis strains, one of which was grown in two different media. Yields varied between the strains, growth media and extraction technique, but generally the highest yield by weight was from the PCP method and the lowest from the PW method. The PW method was selected for the greatest amounts of carbohydrate and KDO, and the PCP method for the least. Phosphorus levels were more uniform among all extraction methods. Protein contamination was found in all Bacteroides LPS extracts, with extremely low levels in PW-LPS and the highest levels in material extracted by the PCP and Triton-Mg2+ techniques. No protein contamination could be detected after proteinase K treatment. After silver staining LPS PAGE profiles showed ladder patterns characteristic of smooth LPS for B. vulgatus, B. thetaiotaomicron and the control Escherichia coli O18:K- strains, whereas the other Bacteroides strains showed mainly rough and low M(r) material only. The PCP method did not select for high M(r) material in the B. fragilis strains; otherwise the LPS profiles for all extraction methods were identical. The biological activities of native and sodium salt form LPS were investigated on a weight for weight basis and compared to that of E. coli O18:K- PW-LPS. Amongst the LPS from Bacteroides strains, those prepared by the PW method were found to have a significantly higher activity in a galactosamine mouse lethality model, in induction of TNF and the Limulus amoebocyte lysate (LAL) assay, than LPS extracted by the PCP or Triton-Mg2+ methods. LPS from Bacteroides strains extracted by the PCP method had consistently low activity in all assays. Comparing PW-LPS from Bacteroides strains with that from E. coli O18:K- in the galactosamine mouse model, the E. coli O18:K- LPS was c. 5000-fold more active than the most active bacteroides LPS. However, in the LAL assay native PW-LPS from both the B. fragilis strains, and B. caccae had higher activities (up to 30-fold) than E. coli O18:K- LPS, with the PW-LPS from the other Bacteroides spp. being up to 15-fold less active than the E. coli O18:K- PW-LPS. In the TNF induction assay, E. coli O18:K- PW-LPS was 4-50-fold more active than bacteroides PW-LPS. In the LAL assay and galactosamine mouse model, native LPS had more activity (c. two-fold) than sodium salt form LPS. There was no clear difference in activity between native and sodium salt form LPS in the TNF induction assay. The results for the LAL and TNF induction assay were re-evaluated relative to KDO concentration. In the TNF induction assay, previously low activities seen on a weight for weight basis were due in part to less KDO being present. However, LAL activity for PCP-LPS was still low after re-evaluation relative to KDO concentration. The molecular basis for the differences in biological activity of bacteroides LPS in relation to extraction methods and chemical composition is not yet understood.
