It has been recognized since 1895 (1) that some gram-negative bacteria are sensitive to the lytic action of fresh serum, whereas others are highly serum resistant. In general, serum-resistant organisms are more pathogenic than serum-sensitive bacteria in animal models of infection, and serum-resistant organisms are more commonly isolated from the bloodstream of patients with gram-negative bacteremia (2). In attempts to define the basis of this important virulence factor, characteristics of the outer membrane of serum-sensitive and serum-resistant organisms have been analyzed and compared (3)(4)(5). The presence of a complete lipopolysaccharide (LPS) 1 (i.e., the smooth phenotype) is the characteristic most clearly associated with serum resistance. Rough bacteria lacking a complete LPS are almost invariably serum sensitive.The antibody and complement requirements for serum killing of bacteria also have been examined. It has been shown (6-8) that killing of gram-negative bacteria by serum requires the participation of terminal components of the complement system ((25-9). However, the mechanism of resistance of gram-negative bacteria to serum killing in the presence of adequate antibody is still unknown.Resistance to serum killing could involve the inability to form a membrane attack complex on the organism. An alternative hypothesis, however, is that a membrane attack complex that forms on the bacterial surface may be functionally impotent either because of failure to insert into the bacterial outer membrane or because the inserted complex does not cause damage to vital outer or inner membrane structures.Previous studies have examined a number of aspects of this issue. Studies have suggested that serum-sensitive and serum-resistant strains of Escherichia coli (9, 10) or Salmonella typhimurium (11) have equivalent amounts of C3 deposited. It is not resolved whether (25 is deposited on serum-resistant bacteria. Reynolds et al. (11) could not demonstrate deposition of functional C5 on serum-resistant S. typhimurium in Mg ++ saline after incubation in C6-deficient rabbit serum. On the other hand, Ogata and Levine (10) demonstrated equivalent (25 consumptior~ by strains of E. coli that varied in complement sensitivity; however, evidence for levels of cell-bound (25 was not 1 Abbreviations used in this paper: CFU, colony-forming unit; HBSS, Hanks' balanced salt solution; LPS, lipopolysaccharide; PNHS, pooled normal human serum; RT, room temperature. J. Exp. MED.

The mechanism for consumption of terminal complement components and release of bound components from the surface of serum-resistant salmonella minnesota S218 was studied. Consumption of C8 and C9 by S218 occurred through interaction with C5b67 on the bacterial surface because C8 and C9 were consumed when added to S218 organisms previously incubated in C8-deficient serum and washed to remove all C5b67 on the bacterial surface because C8 and C9 were consumed when added to S218 organisms previously incubated in C8- deficient serum and washed to remove al but cell bound C5b67. Rapid release of (125)I C5 and (125)I C7 from the membrane of S218 was dependent on binding of C8 because (125)I C5 and (125)I C7 deposition in C8D serum was stable and was twofold higher in C8D than in PNHA, and addition of purified C8 or C8 and C9 to S218 previously incubated in C8D serum caused rapid release of (125)I C5 and (125)I C7 from the organism. Analysis by sucrose density gradient ultracentrifugation of the fluid phase from the reaction of S218 and 10 percent PNHS revealed a peak consistent with SC5b-9, in which the C9:C7 ratio was 3.3:1, but the NaDOC extracted bound C5b-9 complex sedimented as a broad peak with C9:C7 of less than 1.2:1. Progressive elution of C5b67 and C5b-9 from S218 but not serum-sensitive S. minnesota Re595 was observed with incubation in buffers of increasing ionic strength. Greater than 90 percent of the bound counts of (125)I C5 or (125)I C9 were released from S218 by incubation in 0.1 percent trypsin, but only 57 percent of (125)I C9 were released by treatment of Re595 with trypsin. These results are consistent with the concept that C5b-9 forms on the surface of the serum-sensitive S. minnesota S218 in normal human serum, but the formed complex is released and is not bactericidal for S218 because it fails to insert into hydrophobic outer membrane domains.

Escherichia coli strains of group O111 were characterized with respect to sensitivity to complement killing, amount of lipopolysaccharide and O antigen-containing capsule, and distribution of O antigen. All wild-type E. coli O111 strains were resistant to complement killing in the absence of specific antibody. Presensitization of strains with antibody to whole cells (OK antibody), followed by incubation in 50% pooled normal human serum as a source of complement, subdivided wild-type strains into three types: completely resistant, partially resistant, and sensitive. Completely and partially resistant mutants were isolated by cycles of serum killing, starting with one sensitive strain. Completely resistant mutants had no O antigen-containing capsule, but had 50% more lipopolysaccharide than did the parent, and this lipopolysaccharide had 30% fewer lipid A core molecules devoid of O antigen. Partially resistant mutants still had O antigen-containing capsule, but contained 40% more lipopolysaccharide than did the parent; the extent of coverage of lipid A core with O antigen remained unchanged. No correlations were found between outer membrane protein composition and the degree of serum resistance. Since the terminal membrane attack complex (C5b-9) must stably insert into a hydrophobic membrane site to effect killing, we conclude that both increased lipid A core and increased coverage of lipid A core with O antigen preclude access of C5b-9 to lethal sites on the cell surface.

A lipopolysaccharide (LPS) mutant (FA5100) of a serum-resistant strain of Neisseria gonorrhoeae (FA19) was found to be highly sensitive to the bactericidal activity of normal human serum (NHS). Both strain FA5100 and an unrelated serum-sensitive clinical isolate (F62) were killed by NHS via the classical complement pathway since killing required C2 and Ca++. However, the fact that only strain FA5100 was sensitive to human hypogammaglobulinemic and cord serum suggested that this strain might activate the classical complement pathway in the absence of antibody. Anticomplementary concentrations of LPS from strain FA5100 inhibited the bactericidal activity of NHS against either strain FA5100 or strain F62. However, concentrations of LPS from strain FA5100 that exhibited marginal anticomplementary behavior also inhibited the killing of strain F62 by NHS. The ability of LPS from strain FA5100 to inhibit the bactericidal activity of NHS against strain FA5100 and to activate complement was reduced by treatment with mild alkali. However, alkali-treated LPS from strain FA5100 still inhibited the bactericidal activity of NHS against strain F62.

No abstract