The current scheme for typing strains of Serratia marcescens by their heat-stable antigens is undergoing major reconstruction as a result of chemical studies allied with serological reappraisal .
L* Evidence has accumulated that many of the existing 24 0 serogroups are actually defined by acidic, microcapsular polymers, which can camouflage structural variations in lipopolysaccharide side-chains (the conventional 0 antigens). For example, an acidic glucomannan3 apparently accounts for the prevalence4 of "014" strains, the lipopolysaccharides of which differ in their neutral polymeric fractions (at least 4 polymers are represented, each of which also occurs in lipopolysaccharides from strains of other 0 serogroups)'.
In a recent survey of clinical isolates of S. marcescens*, 2 new 0 antigens were identified, one of which (S1254) was present in 13.5% of the strains tested and, apparently, was related to the 04 antigen. During our studies of the surface polysaccharides of S. matcescens, we have isolated both a partially acetylated glucorhamnan6 and an acidic galactomannan7 from the 04 reference strain, and it was therefore of interest to characterise the 51254 antigen.
Lipopolysaccharide was isolated (yield, 36%) from cell walls of strain S1254 by the aqueous phenol method. Mild acid hydrolysis (aqueous 1% acetic acid, 2.25 h, lOO'>, followed by chromatography (Sephadex G-50) of the water-soluble products, gave a polymeric fraction (yield, 43%). Most of the fraction (61%) was eluted with water from a column of DEAE-Sepharose CL-6B, and the remainder with 0. IM NaCl; there was no evidence for the presence of an acidic polymer. Both subfractions had the same sugar composition and gave identical n.m.r. spectra, so further studies were confined to the material eluted with water.
The major components of the polymer were L-rhamnose and D-glucose (ratio of peak areas in g.1.c. of the alditol acetates, 1 .OO: 1.15); small proportions of aldoheptoses were also present. The n.m.r. spectra of the polymer confirmed that the repeating unit was a disaccharide of pyranosyl residues, and showed the absence of 0-acetyl groups. In the 'H-n.m.r. spectrum, there were anomeric signals (each 1 H) at 6 5.07 (J,,* 3.8 Hz) and 4.95 (J1,2 1.6 Hz), pointing to a-pyranosyl residues, and a methyl signal at S 1.29 (J5,6 6.2 OOOS-6215/91/SO3.50