Yersinia pestis Is Viable with Endotoxin Composed of Only Lipid A

Abstract

Lipopolysaccharide (LPS) is the major outer membrane component of gram-negative bacteria. The minimal LPS structure for viability of Escherichia coli and Salmonella enterica serovar Typhimurium is lipid A glycosylated with 3-deoxy-D-manno-octulosonic acid (Kdo) residues. Here we show that another member of the Enterobacteriaceae, Yersinia pestis, can survive without Kdo in its LPS.

Lipopolysaccharide (LPS) (endotoxin) is anchored in the outer membrane by its lipid A component, to which a conserved inner core composed of 3-deoxy-D-manno-octulosonic acid (Kdo) and heptose is attached (10). Neisseria meningitidis is viable without Kdo (15, 16) or even with an lpxA mutation that causes a loss of all detectable endotoxin (13). However, Kdo was found to be essential for growth and survival of Escherichia coli and Salmonella enterica serovar Typhimurium (1, 5, 11), leading to the suggestion that the minimal LPS structure for viability of enteric bacteria is lipid A glycosylated with Kdo residues (1, 10). Here we report that yrbH and waaA (previously called kdtA) deletion mutants of Yersinia pestis, the enteric bacterium that causes plague, are viable and synthesize LPS lacking Kdo.

Arabinose 5-phosphate isomerase (API), encoded by E. coli K-12 yrbH, catalyzes the conversion of ribulose 5-phosphate into arabinose 5-phosphate (A5P), the first committed step in the Kdo biosynthesis pathway (8). WaaA is a transferase catalyzing Kdo glycosylation of lipid A and is essential for the survival of E. coli (1). We constructed yrbH and waaA deletion mutants of Y. pestis KIM6+ by using the allelic replacement vector pCVD442 (4). Deletion of each complete open reading frame was confirmed by PCR, DNA sequencing, or Southern hybridization. LPS was visualized on sodium dodecyl sulfate-polyacrylamide gels stained with silver, which reacts with saccharide components (Fig. 1A) (14). No signal was obtained for the yrbH and waaA mutants, even when lysates from ca. 8 × 10⁷ cells were loaded on the gel. Presence of lipid A in the samples was confirmed using the Limulus amebocyte lysate assay (Sigma) for endotoxin activity (12).

FIG. 1.

LPS defects in Y. pestis yrbH and waaA mutants. (A) Silver-stained 15% Tris-HCl sodium dodecyl sulfate-polyacrylamide gel of either proteinase K-digested whole-cell lysates from ca. 8 × 10⁷ cells (lanes 1 through 8) or 2 μg LPS purified by phenol-chloroform-petroleum ether extraction from the wild type (lane 9) or the ΔyrbH mutant (lane 10). Y. pestis strains: lane 1, ΔyrbH mutant; lanes 2 through 4, ΔyrbH mutants complemented by Y. pestis yrbH, E. coli K-12 yrbH, and E. coli K1 kpsF, respectively; lane 5, wild type; lane 6, ΔwaaA mutant; lanes 7 and 8, ΔwaaA mutants complemented by Y. pestis waaA and E. coli K-12 waaA, respectively. (B) yrbH mutant is complemented by A5P supplementation. WT, wild type; Glc, glucose.

LPS from the yrbH mutant and the wild type was purified by a modified phenol-chloroform-petroleum ether extraction (7) and quantitated by Limulus amebocyte lysate assay to 1.3 × 10⁵ and 1.3 × 10⁶ endotoxin units per mg, respectively. Overloading (2 μg) of this material on a gel produced a strong signal from the wild type but none from the yrbH mutant (Fig. 1A, lanes 9 and 10). The monosaccharide and fatty acid composition of the extracts were analyzed by gas chromatography and mass spectrometry at the Complex Carbohydrate Research Center, University of Georgia. The only sugar detected in the yrbH mutant sample was glucosamine, which is part of lipid A. The wild type contained Kdo and other sugars of core oligosaccharide: heptose, glucose, galactose, and N-acetylglucosamine (17). Both samples contained the same types, and similar amounts, of fatty acids.

Y. pestis YrbH is 77% identical to E. coli K-12 YrbH, and Y. pestis WaaA is 80% identical to E. coli K-12 WaaA. Complementation experiments were performed by using both native Y. pestis genes and homologous genes from E. coli K-12 strain MG1655. DNA was amplified by PCR using a high-fidelity polymerase and cloned into the low-copy-number vector pWKS130 (18). Upstream promoter sequences were included, except in the case of E. coli yrbH, which is in the middle of an operon. For this gene, the open reading frame was fused to the Y. pestis yrbH promoter by overlap extension (6). Clones were verified by restriction digestion and DNA sequencing.

The LPS defects in the yrbH and waaA mutants were complemented by the wild-type genes of Y. pestis, confirming that the defects were due solely to the mutations (Fig. 1A). The E. coli homologues also complemented completely (Fig. 1A), showing that YrbH and WaaA perform the same functions in Y. pestis that they do in E. coli.

For further confirmation that Y. pestis YrbH is an API, we grew the yrbH mutant in LB supplemented with A5P, the product of YrbH activity. A concentration of 0.1 mM A5P (Sigma) was sufficient to restore normal LPS production, whereas no complementation occurred in a control supplemented with 1 mM glucose (Fig. 1B). As expected, A5P supplementation did not complement the waaA mutant, because WaaA functions in Kdo transfer, not synthesis.

E. coli K1 contains a protein, KpsF, that is 44% identical to Y. pestis YrbH. KpsF has been implicated in production of the polysialic acid capsule of E. coli K1 (3). A kpsF clone from E. coli K1 strain EV291 fully complemented the Y. pestis yrbH mutant (Fig. 1A, lane 4), indicating that KpsF has API activity, possibly in addition to other uncharacterized functions.

Although the yrbH and waaA mutants are viable, they have reduced growth rates, with doubling times during exponential phase in shaking LB cultures of about 50% and 60% longer than that of the wild type, respectively. Mutant cells aggregate in broth, and when shaking is halted, they immediately settle to the bottom of the culture tube. On L agar plates, the mutants produce small colonies whose cells adhere tightly to one another but only loosely to the agar, so that a whole colony can be pushed around on the surface like a hockey puck.

We have found, unexpectedly, that Kdo is not essential for growth and survival of Y. pestis. This indicates that caution is required in extending conclusions from investigation of E. coli and S. enterica to other members of the Enterobacteriaceae. In E. coli and S. enterica, two Kdo residues must be added to lipid A before acylation can be completed (2), suggesting that Kdo is required to avoid lethal underacylation of lipid A. However, the order of LPS assembly is not universal. In Pseudomonas aeruginosa, acylation precedes Kdo addition (9), and in N. meningitidis, a waaA mutant synthesizes fully acylated lipid A despite lack of Kdo glycosylation (15). Survival of Y. pestis without Kdo suggests either that this organism tolerates underacylation or that acylation of lipid A is completed prior to Kdo glycosylation.