Rational design of green fluorescent protein mutants as biosensor for bacterial endotoxin

Enhanced green fluorescent protein (EGFP) was selected as a signalling scaffold protein for design of a fluorescent biosensor for bacterial endotoxin [or lipopolysaccharide (LPS)]. Virtual mutagenesis was utilized to model EGFP variants containing binding sites for LPS and lipid A (LA), the bioactive component of LPS. Cationic amphipathic sequences of five alternating basic and hydrophobic residues were introduced to beta-sheets located on the surface of EGFP barrel, in the vicinity of the chromophore. Computational methods were employed to predict binding affinity of Escherichia coli LA, to the models of virtual EGFP mutants. DNA mutant constructs of five predicted best binding EGFP variants were expressed in COS-1 cells. The EGFP-mutant proteins exhibited differential expression and variable degrees of fluorescence yield at 508 nm. The EGFP mutants showed a range of LA binding affinities that corresponded to the computational predictions. LPS/LA binding to the mutants caused concentration-dependent fluorescence quenching. The EGFP mutant, G10 bearing LPS/LA amphipathic binding motif in the vicinity of the chromophore (YLSTQ(200-204)-->KLKTK) captured LA with a dissociation constant of 8.5 microm. G10 yielded the highest attenuation of fluorescence intensity in the presence of LPS/LA and demonstrated capability in fluorescence-mediated quantitative detection of LPS in endotoxin-contaminated samples. Thus, the EGFP mutant can form the basis of a novel fluorescent biosensor for bacterial endotoxin.     

Interaction between phage G13 and its oligosaccharide receptor studied by equilibrium dialysis

The reversible binding of phage G13, a phi X174-like single-strand DNA phage, to a 3H-labelled nonasaccharide from the lipopolysaccharide of its natural host Escherichia coli C was studied with equilibrium dialysis. The binding constant (Ka) was determined to 1.3 x 10(7) M-1 in Scatchard and Lineweaver-Burk plots. Approximately one saccharide bound per G13 phage particle which suggests that only one of the 12 spikes in each G13 virion was engaged in the phage/receptor saccharide interaction. Equilibrium dialysis inhibition experiments with saccharides from lipopolysaccharides of an isogenic series of Salmonella typhimurium mutants showed that hepta- and pentasaccharides from two G13-sensitive bacteria, i.e., with efficiencies of plating of 0.1-1.0 compared to E. coli C, were efficient inhibitors with Ka-values greater than or equal to 1.2 x 10(7) M-1. The octa- and hexasaccharides from two G13 resistant strains, with efficiency of plating less than or equal to x 10(-4), were either greater than 1000-fold or greater than 15-fold less efficient as inhibitors with Ka-values less than or equal to 8.8 x 10(5) M-1. The results show that phage G13 binds in a specific and reversible way to penta-, hepta-, and nonasaccharides from G13 sensitive bacteria with the specificity residing in the hexose and heptose region of the core lipopolysaccharide.     

Structural and physicochemical requirements of endotoxins for the activation of arachidonic acid metabolism in mouse peritoneal macrophages in vitro

Lipopolysaccharides of different wild-type and mutant gram-negative bacteria, as well as synthetic and bacterial free lipid A, were studied for their ability to activate arachidonic acid metabolism in mouse peritoneal macrophages in vitro. It was found that lipopolysaccharides of deep-rough mutants of Salmonella minnesota and Escherichia coli (Re to Rc chemotypes) stimulated macrophages to release significant amounts of leukotriene C4 (LTC4) and prostaglandin E2 (PGE2). Lipopolysaccharides of wild-type strains (S. abortus equi, S. friedenau) only induced PGE2 and not LTC4 formation. Unexpectedly, free bacterial and synthetic E. coli lipid A were only weak inducers of LTC4 and PGE2 production. Deacylated Re-mutant lipopolysaccharide preparations were inactive. However, co-incubation of macrophages with both deacylated lipopolysaccharide and lipid A lead to the release of significant amounts of LTC4 and PGE2, similar to those obtained with Re-mutant lipopolysaccharide. The significance of the lipid A portion of lipopolysaccharide for the induction of LTC4 was indicated by demonstrating that peritoneal macrophages of endotoxin-low-responder mice or of mice rendered tolerant to endotoxin did not respond with the release of arachidonic acid metabolites on stimulation with Re-mutant lipopolysaccharide and that polymyxin B prevented the Re-lipopolysaccharide-induced LTC4 and PGE2 release. Physical measurements showed that the phase-transition temperatures of both free lipid A and S-form lipopolysaccharide were above 37 degrees C while those of R-mutant lipopolysaccharides were significantly lower (30-35 degrees C). Thus, with the materials investigated, an inverse relationship between the phase-transition temperature and the capacity to elicit LTC4 production was revealed.     

Accelerated Adsorption of Bacteriophage T5 to Escherichia coli F, Resulting from Reversible Tail Fiber-Lipopolysaccharide Binding

A dual specificity for phage T5 adsorption to Escherichia coli cells is shown. The tail fiber-containing phages T5(+) and mutant hd-3 adsorbed rapidly to E. coli F (1.2 x 10(-9) ml min(-1)), whereas the adsorption rate of the tail fiber-less mutants hd-1, hd-2, and hd-4 was low (7 x 10(-11) ml min(-1)). The differences in adsorption rates were due to the particular lipopolysaccharide structure of E. coli F. Phage T4-resistant mutants of E. coli F with an altered lipopolysaccharide structure exhibited similar low adsorption for all phage strains with and without tail fibers. The same held true for E. coli K-12 and B which also differ from E. coli F in their lipopolysaccharide structures. Only the tail fiber-containing phages reversibly bound to isolated lipopolysaccharides of E. coli F. Infection by all phage strains strictly depended on the tonA-coded protein in the outer membrane of E. coli. We assume that the reversible preadsorption by the tail fibers to lipopolysaccharide accelerates infection which occurs via the highly specific irreversible binding of the phage tail to the tonA-coded protein receptor. The difference between rapid and slow adsorption was also revealed by the competition between ferrichrome and T5 for binding to their common tonA-coded receptor in tonB strains of E. coli. Whereas binding of T5(+) to E. coli K-12 and of the tail-fiber-less mutant hd-2 to E. coli F and K-12 was inhibited 50% by about 0.01 muM ferrichrome, adsorption of T5 to E. coli F was inhibited only 40% by even 1,000-fold higher ferrichrome concentrations.     

Polymyxin B-horseradish peroxidase conjugates as tools in endotoxin research.

The peptide antibiotic Polymyxin B (PMB) binds to bacterial endotoxin (lipopolysaccharide, LPS). We prepared covalent conjugates of PMB and horseradish peroxidase (HRP) by periodation of HRP-linked oligosaccharides followed by direct condensation with PMB. In addition we prepared monoclonal antibodies (Mabs) to PMB. The PMB-HRP conjugates and anti-PMB Mabs were used to study in ELISA the binding of PMB to LPS from Escherichia coli, Klebsiella pneumoniae, and Pseudomonas aeruginosa. In addition, PMB-HRP was used to quantify lipid A in ELISA, and to stain gram-negative bacteria histochemically. For the study of PMB-LPS interaction, PMB-HRP proved to be superior to the anti-PMB Mabs. PMB-HRP conjugates are useful general probes to detect or measure lipid A and LPS of various species using very simple methods and to stain bacteria, and they may obviate the need for many specific antisera. Thus, PMB-HRP conjugates are useful probes for endotoxin research.     

Lipopolysaccharide Layer Protection of Gram-Negative Bacteria Against Inhibition by Long-Chain Fatty Acids

Growth, amino acid transport, and oxygen consumption of Escherichia coli and Salmonella typhimurium are inhibited by short-chain (C(2)-C(6)) but not by medium or long-chain fatty acids (C(10)-C(18)) at concentrations at which these processes are completely inhibited in Bacillus subtilis. The resistance of gram-negative organisms is not correlated with their ability to metabolize fatty acids, since an E. coli mutant unable to transport oleic acid is still resistant. However, mutants of both E. coli and S. typhimurium in which the lipopolysaccharide layer does not contain the residues beyond the 2-keto-3-deoxyoctonate core are inhibited by medium (C(10)) but not by long-chain (C(18)) fatty acids. Furthermore, removal of a portion of the lipopolysaccharide layer by ethylenediaminetetraacetate treatment renders the organisms sensitive to medium and partially sensitive to long-chain fatty acids. The intact lipopolysaccharide layer of gram-negative organisms apparently screens the cells against medium and long-chain fatty acids and prevents their accumulation on the inner cell membrane (site of amino acid transport) at inhibitory concentrations. These results are relevant to the use of antimicrobial food additives, and they allow the characterization of gram-positive versus gram-negative bacteria and their lipopolysaccharide mutants.     

Cell wall receptor for bacteriophage Mu G(+).

The invertible G segment in phage Mu DNA controls the host range of the phage. Depending on the orientation of the G segment, two types of phage particles, G(+) and G(-), are produced which recognize different cell surface receptors. The receptor for Mu G(+) was located in the lipopolysaccharide (LPS) of gram-negative bacteria. The analysis of different LPS core types and of mutants that were made resistant to Mu G(+) shows that the primary receptor site on Escherichia coli K-12 lies in the GlcNAc beta 1 . . . 6Glc alpha 1-2Glc alpha 1-part at the outer end of the LPS. Mu shares this receptor site in E. coli K-12 with the unrelated single-stranded DNA phage St-1. Phage D108, which is related to Mu, and phages P1 and P7, which are unrelated to Mu but contain a homologous invertible DNA segment, have different receptor requirements. Since they also bind to terminal glucose in a different configuration, they adsorb to and infect E. coli K-12 strains with an incomplete LPS core.     

Extragenic Suppressors of Growth Defects in msbB Salmonella

Lipid A, a potent endotoxin which can cause septic shock, anchors lipopolysaccharide (LPS) into the outer leaflet of the outer membrane of gram-negative bacteria. MsbB acylates (KDO)(2)-(lauroyl)-lipid IV-A with myristate during lipid A biosynthesis. Reports of knockouts of the msbB gene describe effects on virulence but describe no evidence of growth defects in Escherichia coli K-12 or Salmonella. Our data confirm the general lack of growth defects in msbB E. coli K-12. In contrast, msbB Salmonella enterica serovar Typhimurium exhibits marked sensitivity to galactose-MacConkey and 6 mM EGTA media. At 37 degrees C in Luria-Bertani (LB) broth, msbB Salmonella cells elongate, form bulges, and grow slowly. msbB Salmonella grow well on LB-no salt (LB-0) agar; however, under specific shaking conditions in LB-0 broth, many msbB Salmonella cells lyse during exponential growth and a fraction of the cells form filaments. msbB Salmonella grow with a near-wild-type growth rate in MSB (LB-0 containing Mg(2+) and Ca(2+)) broth (23 to 42 degrees C). Extragenic compensatory mutations, which partially suppress the growth defects, spontaneously occur at high frequency, and mutants can be isolated on media selective for faster growing derivatives. One of the suppressor mutations maps at 19.8 centisomes and is a recessive IS10 insertional mutation in somA, a gene of unknown function which corresponds to ybjX in E. coli. In addition, random Tn10 mutagenesis carried out in an unsuppressed msbB strain produced a set of Tn10 inserts, not in msbB or somA, that correlate with different suppressor phenotypes. Thus, insertional mutations, in somA and other genes, can suppress the msbB phenotype.     

Polysaccharides cellulose, poly-beta-1,6-n-acetyl-D-glucosamine, and colanic acid are required for optimal binding of Escherichia coli O157:H7 strains to alfalfa sprouts and K-12 strains to plastic but not for binding to epithelial cells

When Escherichia coli O157:H7 bacteria are added to alfalfa sprouts growing in water, the bacteria bind tightly to the sprouts. In contrast, laboratory K-12 strains of E. coli do not bind to sprouts under similar conditions. The roles of E. coli O157:H7 lipopolysaccharide (LPS), capsular polysaccharide, and exopolysaccharides in binding to sprouts were examined. An LPS mutant had no effect on the binding of the pathogenic strain. Cellulose synthase mutants showed a significant reduction in binding; colanic acid mutants were more severely reduced, and binding by poly-beta-1,6-N-acetylglucosamine (PGA) mutants was barely detectable. The addition of a plasmid carrying a cellulose synthase gene to K-12 strains allowed them to bind to sprouts. A plasmid carrying the Bps biosynthesis genes had only a marginal effect on the binding of K-12 bacteria. However, the introduction of the same plasmid allowed Sinorhizobium meliloti and a nonbinding mutant of Agrobacterium tumefaciens to bind to tomato root segments. These results suggest that although multiple redundant protein adhesins are involved in the binding of E. coli O157:H7 to sprouts, the polysaccharides required for binding are not redundant and each polysaccharide may play a distinct role. PGA, colanic acid, and cellulose were also required for biofilm formation by a K-12 strain on plastic, but not for the binding of E. coli O157:H7 to mammalian cells.     

Endotoxin in germfree, gnotobiotic, or conventional-flora Sprague-Dawley rats.

The Limulus assay for bacterial endotoxin was performed on serum and (or) plasma from animals monoassociated with Clostridium species, Staphylococcus aureus, Escherichia coli, Proteus mirabilis, Enterobacter agglomerans, Bacteroides fragilis, Klebsiella pneumoniae, or Candida albicans. Plasma from animals monoassociated with the gram-negative bacteria or C. albicans consistently showed a positive Limulus test while conventional-flora controls, germfree rats, and gnotobiotic animals monoassociated with gram-positive bacteria or E. agglomerans were negative. Germfree and conventional rats were injected (intraperitoneal (i.p.)) with Salmonella typhosa lipopolysaccharide (LPS). Although no endotoxin was detectable in either group prior to the injection, by 1 h post injection endotoxin was in the plasma of all groups. The germfree rats appeared to clear the LPS quicker than their conventional-flora counterparts. Generally, LPS-injected rats (conventional and germfree) showed clumping and decreased number of platelets, a decrease in their lymphocyte counts, and increased polymorphonuclear leukocyte (PMN) counts.     

A method to measure the interaction of Rac/Cdc42 with their binding partners using fluorescence resonance energy transfer between mutants of green fluorescent protein

Enhanced blue fluorescent protein (EBFP) and enhanced green fluorescent protein (EGFP) mutants of GFP in close proximity to one another can act as a fluorescence resonance energy transfer (FRET) pair. Unstructured amino acid linkers of varying length were inserted between EBFP and EGFP, revealing that linkers even as long as 50 amino acids can be accommodated and still allow FRET to occur. This led to the development of a novel biosensor for Rac/Cdc42 binding to their effector proteins based on the insertion of amino acids 75-118 of p21-activated kinase (PAK) between the GFP mutants. We demonstrate that this protein construct allows significant FRET between EBFP and EGFP and retains the ability to bind to Rac in its GTP-bound form with a binding affinity similar to the uncomplexed PAK fragment, and furthermore, on binding to Rac or Cdc42 a marked change in FRET takes place. This forms the basis for a simple, sensitive, and rapid method to measure binding of Rac/Cdc42 to their effector proteins. Since the signal is dependent upon the interaction with active GTP-bound forms it acts as a biosensor for the activation of Rac/Cdc42. It has the potential for use in live cells and for identifying localization of Rac/Cdc42 within subcellular compartments.     

E. coli virulence factor hemolysin induces neutrophil apoptosis and necrosis/lysis in vitro and necrosis/lysis and lung injury in a rat pneumonia model

Enteric gram-negative bacilli, such as Escherichia coli are the most common cause of nosocomial pneumonia. In this study a wild-type extraintestinal pathogenic strain of E. coli (ExPEC)(CP9) and isogenic derivatives deficient in hemolysin (Hly) and cytotoxic necrotizing factor (CNF) were assessed in vitro and in a rat model of gram-negative pneumonia to test the hypothesis that these virulence factors induce neutrophil apoptosis and/or necrosis/lysis. As ascertained by in vitro caspase-3/7 and LDH activities and neutrophil morphology, Hly mediated neutrophil apoptosis at lower E. coli titers (1 x 10(5-6) cfu) and necrosis/lysis at higher titers (> or =1 x 10(7) cfu). Data suggest that CNF promotes apoptosis but not necrosis or lysis. We also demonstrate that annexin V/7-amino-actinomycin D staining was an unreliable assessment of apoptosis using live E. coli. The use of caspase-3/7 and LDH activities and neutrophil morphology supported the notion that necrosis, not apoptosis, was the primary mechanism by which neutrophils were affected in our in vivo gram-negative pneumonia model using live E. coli. In addition, in vivo studies demonstrated that Hly mediates lung injury. Neutrophil necrosis was not observed when animals were challenged with purified lipopolysaccharide, demonstrating the importance of using live bacteria. These findings establish that Hly contributes to ExPEC virulence by mediating neutrophil toxicity, with necrosis/lysis being the dominant effect of Hly on neutrophils in vivo and by lung injury. Whether Hly-mediated lung injury is due to neutrophil necrosis, a direct effect of Hly, or both is unclear.     

Endotoxin of Neisseria meningitidis composed only of intact lipid A: inactivation of the meningococcal 3-deoxy-D-manno-octulosonic acid transferase

Lipopolysaccharide, lipooligosaccharide (LOS), or endotoxin is important in bacterial survival and the pathogenesis of gram-negative bacteria. A necessary step in endotoxin biosynthesis is 3-deoxy-D-manno-octulosonic acid (Kdo) glycosylation of lipid A, catalyzed by the Kdo transferase KdtA (WaaA). In enteric gram-negative bacteria, this step is essential for survival. A nonpolar kdtA::aphA-3 mutation was created in Neisseria meningitidis via allelic exchange, and the mutant was viable. Detailed structural analysis demonstrated that the endotoxin of the kdtA::aphA-3 mutant was composed of fully acylated lipid A with variable phosphorylation but without Kdo glycosylation. In contrast to what happens in other gram-negative bacteria, tetra-acylated lipid IV(A) did not accumulate. The LOS structure of the kdtA::aphA-3 mutant was restored to the wild-type structure by complementation with kdtA from N. meningitidis or Escherichia coli. The expression of a fully acylated, unglycosylated lipid A indicates that lipid A biosynthesis in N. meningitidis can proceed without the addition of Kdo and that KdtA is not essential for survival of the meningococcus.     

Removal of endotoxin during purification of poly(3-hydroxybutyrate) from gram-negative bacteria.

Poly(3-hydroxybutyrate) (PHB) was produced by cultivating several gram-negative bacteria, including Ralstonia eutropha, Alcaligenes latus, and recombinant Escherichia coli. PHB was recovered from these bacteria by two different methods, and the endotoxin levels were determined. When PHB was recovered by the chloroform extraction method, the endotoxin level was less than 10 endotoxin units (EU) per g of PHB irrespective of the bacterial strains employed and the PHB content in the cell. The NaOH digestion method, which was particularly effective for the recovery of PHB from recombinant E. coli, was also examined for endotoxin removal. The endotoxin level present in PHB recovered by 0.2 N NaOH digestion for 1 h at 30 degrees C was higher than 10(4) EU/g of PHB. Increasing the digestion time or NaOH concentration reduced the endotoxin level to less than 1 EU/g of PHB. It was concluded that PHB with a low endotoxin level, which can be used for various biomedical applications, could be produced by chloroform extraction. Furthermore, PHB with a much lower endotoxin level could be produced from recombinant E. coli by simple NaOH digestion.     

Density gradient enrichment of Escherichia coli lpxL mutants

We previously described enrichment of conditional Escherichia coli msbA mutants defective in lipopolysaccharide export using Ludox density gradients (Doerrler WT (2007) Appl Environ Microbiol 73; 7992-7996). Here, we use this approach to isolate and characterize temperature-sensitive lpxL mutants. LpxL is a late acyltransferase of the pathway of lipid A biosynthesis (The Raetz Pathway). Sequencing the lpxL gene from the mutants revealed the presence of both missense and nonsense mutations. The missense mutations include several in close proximity to the enzyme's active site or conserved residues (E137K, H132Y, G168D). These data demonstrate that Ludox gradients can be used to efficiently isolate conditional E. coli mutants with defects in lipopolysaccharide biosynthesis and provide insight into the enzymatic mechanism of LpxL.     

Bacteria, molds, and toxins in water-damaged building materials.

Microbial toxins and eukaryotic cell toxicity from indoor building materials heavily colonized by fungi and bacteria were analyzed. The dominant colonizers at water-damaged sites of the building were Stachybotrys chartarum (10(3) to 10(5) visible conidia cm-2), Penicillium and Aspergillus species (10(4) CFU mg-1), gram-negative bacteria (10(4) CFU mg-1), and mycobacteria (10(3) CFU mg-1). The mycobacterial isolates were most similar to M. komossense, with 98% similarity of the complete 16S rDNA sequence. Limulus assay of water extracts prepared from a water-damaged gypsum liner revealed high contents of gram-negative endotoxin (17 ng mg-1 of E. coli lipopolysaccharide equivalents) and beta-D-glucan (210 ng mg-1 of curdlan equivalents). High-performance liquid chromatography analysis of the methanol extracts showed that the water-damaged gypsum liner also contained satratoxin (17 ng mg-1). This methanol-extracted substance was 200 times more toxic to rabbit skin and fetus feline lung cells than extract of gypsum liner sampled from a non-water-damaged site. The same extract contained toxin(s) that paralyzed the motility of boar spermatozoa at extremely low concentrations; the 50% effective concentration was 0.3 microgram of dry solids per ml. This toxicity was not explainable by the amount of bacterial endotoxin, beta-D-glucan, or satratoxin present in the same extract. The novel in vitro toxicity test that utilized boar spermatozoa as described in this article is convenient to perform and reproducible and was a useful tool for detecting toxins of microbial origin toward eukaryotic cells not detectable in building materials by the other methods.     

The cell wall receptor for bacteriophage Mu G(−) in Erwinia and Escherichia coli C

Abstract Adsorption of bacteriophage Mu with its invertible DNA segment in the G(−) orientation requires a terminal glucose residue for binding to the core lipopolysaccharide (LPS) of Gram-negative bacteria. Analysis of a Mu-resistant mutant shows that the receptor for Mu G(−) in Erwinia B374 is a Glc-β1,6-Glc disaccharide. A spontaneously occurring host-range mutant, Mu G(−)h101, grows on Escherichia coli C. The loss of the terminal β1,3-linked glucose from the LPS of E. coli C leads to resistance to the phage Mu. These mutants are also resistant to phage P1 and D108 which have largely homologous G segments. This shows that Mu G(+) and G(−) phage particles differ with respect to their cell-wall receptors in the type of glycosidic linkage of a terminal glucose residue: α1, 2 for G(+) and β1,6 for G(−).     

Immune islet killing mechanisms associated with insulin-dependent diabetes: in vitro expression of cellular and antibody-mediated islet cell cytotoxicity in humans

In previous work we have shown that some bacteria can bind to human lymphocytes and can be used to identify lymphocyte subpopulations in conventionally stained blood smears. These bacteria are of different species or genera, which makes it difficult to study the binding mechanism. Also, the main marker for B cells, Brucella melitensis, is of very small size and highly pathogenic. Here we show that B cells as well as some of the T cell subpopulations can be identified by different mutants obtained from a strain of an Escherichia coli. Two procedures were used to generate mutants. First, E. coli-YS57 (pro-his-trp-) was mutagenized with N-methyl-N'-nitro-N-nitrosoguanidine and the binding to mouse spleen cells was used as a selective pressure. Second, phage-resistant mutants of E. coli-YS57 were obtained and tested for the ability to bind to lymphocytes. Out of 10 strains selected by the former procedure, 5 bound to a significant number of human lymphocytes. All four phage-resistant mutants bound to human lymphocytes. Out of the total of nine mutants that bound to lymphocytes, six bound consistently, i.e., they bound to similar percentages of peripheral blood lymphocytes from different normal donors. One phage-resistant mutant, E. coli USC-106, bound only to B cells. The subpopulations of lymphocytes identified by the mutants were essentially the same as those identified by different species or genera of bacteria. We concluded that E. coli mutants can be obtained that identify human lymphocyte subpopulations and that one of these mutants recognizes B cells; these mutants may be used to study the nature of the receptors for bacteria on lymphocytes, which appear to have a lectin-like nature.     

Monoclonal antibodies against the capsular K antigen of Escherichia coli (O9:K30(A):H12): characterisation and use in analysis of K antigen organisation on the cell surface

Monoclonal antibodies were produced against the capsular antigen of Escherichia coli serotype K(A)30, using a mouse hybridoma system. The antibodies also recognised the chemically identical capsular polysaccharide produced by Klebsiella K20. Chemical modification of the K30 polysaccharide indicated that the glucuronic acid residues found in the E. coli K30 capsular antigen were important in the epitope recognised by these antibodies. Use of the antibodies as molecular probes revealed the presence of two discrete forms of the K30 antigen. One form was comprised of high molecular weight polysaccharide, present as a surface capsular layer. The second form of the antigen was of low molecular weight and was associated with lipopolysaccharide fractions from cell surface polysaccharide extracts. Separation of lipopolysaccharide fractions using gel chromatography in the presence of detergent showed that the low molecular weight K-antigenic fraction comigrated with a lipopolysaccharide lipid A core fraction present in encapsulated E. coli K30 bacteria but absent in acapsular mutants.