Mechanism of O-antigen distribution in lipopolysaccharide.

O-antigen units are nonuniformly distributed among lipid A-core molecules in lipopolysaccharide (LPS) from gram-negative bacteria, as revealed by polyacrylamide gel electrophoresis in sodium dodecyl sulfate; the actual distribution patterns are complex, multimodal, and strain specific. Although the basic biochemical steps involved in synthesis and polymerization of O-antigen monomers and their subsequent attachment to lipid A-core are known, the mechanism by which specific multimodal distribution patterns are attained in mature LPS has not been previously considered theoretically or experimentally. We have developed probability equations which completely describe O-antigen distribution among lipid A-core molecules in terms of the probability of finding a nascent polymer (O antigen linked to carrier lipid) of length k (Tk) and the probability that a nascent polymer of length k will be extended to k + 1 by polymerase (pk) or transferred to lipid A-core by ligase (qk). These equations were used to show that multimodal distribution patterns in mature LPS cannot be produced if all pk are equal to p and all qk are equal to q, conditions which indicate a lack of selectivity of polymerase and ligase, respectively, for nascent O-antigen chain lengths. A completely stochastic model (pk = p, qk = q) of O-antigen polymerization and transfer to lipid A-core was also inconsistent with observed effects of mutations which resulted in partial inhibition of O-antigen monomer synthesis, lipid A-core synthesis, or ligase activity. The simplest explanation compatible with experimental observations is that polymerase or ligase, or perhaps both, have specificity for certain O-antigen chain lengths during biosynthesis of LPS. Our mathematical model indicates selectively probably was associated with the polymerase reaction. Although one may argue for a multimodal distribution pattern based on a kinetic mechanism i.e., varying reaction parameters in space or in time during cell growth, such a model requires complex sensory and regulatory mechanisms to explain the mutant data and mechanisms for sequestering specific components of LPS biosynthesis to explain the distribution pattern in normal cells. We favor the simple alternative of enzyme specificity and present generalized equations which should be useful in analysis of other analogous biochemical systems.