The 09-specific mannan of Eschericlzia coli was synthesized in vitro from GDP-['4C]mannose by membranes which were obtained from a phosphomannose isomerase-less mutant of E. coli 09:K29-: H -. Subsequent treatment of the membranes with dilute acid liberated a neutral prodfict, whereas with aqueous phenol a charged product was obtained. Chromatography on DEAE-cellulose, incubation with alkaline phosphatase and microdetermination showed that the charged mannan was substituted with one phosphate per chain.The neutral 14C-labelled product of the incubation in vitro was reduced with sodium b~ro [~H]hydride. After total acid hydrolysis, the radioactive material was chromatographed on paper in the presence of borate. It was found that [3H]glucitol, but no [3H, 14C]mannitol was present. When the neutral product, which was obtained after incubation of 14C-prelabelled membranes with nonradioactive GDP-mannose, was hydrolyzed with and without prior reduction with non-radioactive sodium borohydride, in subsequent paper chromatography, ['4C]glucitol or [14C]glucose was found. The glucose was also converted enzymatically to gluconic acid, which was identified by paper electrophoresis. These results show that in the neutral 09-specific mannan glucose is at the reducing end and they indicate that the mannan chain grows at the non-reducing end. This is discussed with respect to the overall mechanism of the biosynthesis of the 0 9 antigen. The 0 9 antigen of Eschevichia coli is a lipopolysaccharide which has an a-mannan as the O-specific polysaccharide moiety (for definitions and further information see [1,2] . Thus, the synthesis of the 0 9 antigen differs from that of the other enterobacterial 0 antigens hitherto described [3,4]. Bray and Robbins [S] have shown that the O-specific polysaccharide from Salmonella anaturn, which is assembled from undecaprenyl-pyrophosphoryl oligosaccharides grows at the reducing end. We were interested in the mode of chain elongation of the 0 9 mannan and its linkage to the cytoplasmic membrane during biosynthesis. Therefore we have analyzed the material synthesized in vitro and found that according to the method of liberation from the membrane, the mannan is either neutral or charged. The charge proved to be due to phosphate substitution. Using reduction with sodium borohydride, we could show that the reducing end of the neutral mannan is not mannose, but glucose. These results are described in the present communication and they are discussed with respect to the direction of chain elongation and possible overall mechanisms of the 0 9 mannan biosynthesis. Amersham-Buchler (BraunschweigWenden). Alkaline phosphatase and glucose oxidase were from Boehringer (Mannheim).
MATERIALS AND METHODS
Chemicals and Enzymes
Bacteria and CultivationAs described previously [7] E. coli F860, a mutant of E. coli F397 ( 0 9 : K29Y: H-) lacking phosphoman-