Similarities and differences of innate immune responses elicited by smooth and rough LPS

Abstract

The lipopolysaccharide is the major component of Gram-negative bacteria outer membrane. LPS comprises three covalently linked regions: the lipid A, the rough core oligosaccharide, and the O-antigenic side chain determining serotype specificity. Wild-type LPS (sLPS) contains the O-antigenic side chain and is referred to as smooth. Rough LPS (rLPS) does not contain the O-side chain. Most wt bacteria and especially wt Enterobacteriaceae express prevalently the sLPS form although some truncated rLPS molecules always reach the external membrane. The two sLPS and rLPS forms are used almost indistinctly to study the effects on innate immune cells. Nevertheless, there is evidence that their mechanism of action may be different. For instance, while sLPS requires CD14 for the initiation of both MyD88-dependent and independent signal transduction pathways at least at low doses, rLPS leads to MyD88-dependent responses in the absence of CD14 even at low doses. Here we have identified additional differences in the signaling capacity of the two LPS species in the mouse. We have found that rLPS, diversely from sLPS, is capable of activating in dendritic cells (DCs) the Ca²⁺/calcineurin and NFAT pathway in a CD14-independent manner, moreover it is also capable per se of activating the inflammasome and eliciting IL-1β secretion independent of the presence of additional stimuli required instead for sLPS. The ability of rLPS of activating the inflammasome in vitro has as a direct consequence a higher efficiency of rLPS-exposed DCs in activating natural killer (NK) cells compared to sLPS-exposed DCs. However, diversely from possible predictions, we found that the different efficiencies of the two LPS species in eliciting innate responses are almost nullified in vivo. Therefore, sLPS and rLPS induce nearly similar in vivo innate responses but with different mechanisms of signaling.

Introduction

In 1884, Hans Christian Gram, developed a staining procedure, the Gram staining, for bacteria classification. Based on this method, almost all bacteria can be divided into two large groups depending on the structural differences of their cell wall; the Gram-positive and Gram-negative bacteria. Gram-positive bacteria retain the crystal violet dye of the Gram staining thanks to the presence of high amount of peptidoglycan in their cell wall. In contrast, Gram-negative bacteria do not retain crystal violet dye since they have a relatively thin cell wall consisting of few layers of peptidoglycan surrounded by a second lipid membrane. A major component of the outer membrane of Gram-negative bacteria is the lipopolysaccharide (LPS), a complex molecule indispensable for the maintenance of the structural and functional integrity of the membrane itself [1]. For this reason, the general structure of LPS is conserved among all Gram-negative bacteria. LPS is composed of three major parts, the lipid A, the core region and the O-chain [2]. The lipid A moiety is highly hydrophobic and it is largely responsible for the endotoxic activity of the whole LPS molecule. This moiety is inserted into the external face of the bacterial outer membrane. The core region is a conserved polysaccharidic structure that can be subdivided into inner and outer core. The inner core is proximal to the lipid A and it contains unusual sugars, such as Kdo and heptose, which are absolutely required for bacterial viability and therefore are well conserved among all LPS species. The variability of the different LPS species is contained in the outer core, which typically consists of common hexose sugars.

The hydrophilic O-chain is the outer region of the LPS molecule. The O-chain is a highly variable region composed of repeating saccharidic units formed by up to eight glycosyl residues that differ between bacterial strains in terms of sugars, sequences, linkages and substitutions used. Additionally, these forming units can be repeated up to 50 times and a single organism will produce a wide range of these lengths due to incomplete synthesis of the chain. Some Gram-negative bacteria; especially members of the Enterobacteriaceae, such as E. coli and Salmonella Thyphimurium, carry mutations in the genes involved in the synthesis and attachment of the O-chain and do not express it at all [3]. These mutants are called “rough” because of the morphology of the colonies they form in a plate that is different from that observed for wild-type, “smooth” bacteria. Thus, the truncated form of LPS is called rough (rLPS), while the wild type form, containing the O-chain, is called smooth (sLPS).

Although sLPS and rLPS share the same receptor complex (consisting of the Toll like receptor 4, TLR4, and MD2 subunits and the glycosyl-phosphatidylinositol-anchored CD14 molecule [4]) there is evidence that their mechanism of action may be different. For instance, while sLPS requires CD14 for the initiation of both MyD88-dependent and independent signal transduction pathways at least at low doses [5], rLPS leads to MyD88-dependent responses in the absence of CD14 even at low doses. This has led to the assumption that rLPS activates a broader range of cells (CD14-positive and low/negative cells) and with a higher efficiency compared to sLPS [6].

Since the two forms of LPS are used almost indistinctly to study the reactions of the innate immune system (the most common form of rLPS used is composed of the lipid A and 3 Kdo), it is important to define to what extent the consequences of sLPS and rLPS exposure are similar or dissimilar to avoid possible confusion between common and LPS specie-specific responses.

Here we provide further evidence of the diverse mechanisms through which sLPS and rLPS may activate pro-inflammatory innate responses. We show that the ability of rLPS to function in a CD14-independent manner is not limited to the activation of the MyD88-dependent pathway but is also extended to the activation of the Ca²⁺/calcineurin and NFAT pathway that leads to IL-2 production in conventional dendritic cells (DCs). We also show that, unlike sLPS, rLPS is capable per se of activating the inflammasome and inducing interleukin (IL)-1β secretion by DCs. Nevertheless, although these observations could support the prediction that rLPS could induce more potent innate responses with respect to sLPS, an in vivo comparison revealed that the two LPS species elicit almost comparable responses, with sLPS being slightly more efficient.