A newborn mouse model for the study of intestinal pathogenesis of shigellosis.

Shigella infection is characterized by the induction of acute inflammation, which is responsible for the massive tissue destruction of the intestinal mucosa. A murine model would be a valuable tool for gaining a better understanding of the physiopathology of shigellosis and the host immune response to Shigella infection, but adult mice do not develop disease upon oral inoculation. We therefore attempted to develop a model of infection in newborn mice. Four-day-old mice inoculated with 50 microl of 5 x 10(9) invasive wild-type Shigella flexneri 5a were susceptible to bacterial infection, but mice inoculated with the non-invasive strain BS176 were not. Histologically, 4-day-old mice infected with the invasive strain presented intestinal lesions and inflammation similar to those described in patients with shigellosis. Moreover, cytokine and chemokine responses consistent with inflammation were observed. Lower bacterial inocula induced less severe intestinal damage. In contrast, 5-day-old mice inoculated with either the invasive or the non-invasive strain were not infected. We have thus established a mouse model that is suitable for the study of the pathogenesis of intestinal Shigella infection.

Down-modulation of L-selectin by lipopolysaccharide is not required for lipopolysaccharide-induced expression of CD14 in mouse bone marrow granulocytes.

We established in previous studies that a constitutive lipopolysaccharide (LPS) receptor of low affinity is present on mouse bone marrow granulocytes (BMG). This yet-unidentified receptor is involved in the LPS-induced expression of a second LPS receptor, CD14. Because it has been claimed that L-selectin (CD62L) is a low-affinity LPS receptor in mature granulocytes (polymorphonuclear leukocytes), it may be asked whether this molecule could be the constitutive LPS receptor in BMG. We show in this study that L-selectin is constitutively present on BMG and is down-regulated after exposure of the cells to LPS. A phorbol ester induced a down-regulation of CD62L and blocked the LPS-induced expression of CD14. However, a metalloproteinase inhibitor (BB-3103) blocked the former but not the latter effect of PMA. We also observed an absence of cross-reactivity between LPS and a CD62L ligand (fucoidan) in binding studies with radiolabeled derivatives of the two agents. Furthermore, BMG from L-selectin-deficient mice expressed normal levels of CD14 in response to LPS. Taken together, these results demonstrate that in BMG, L-selectin is not the constitutive LPS receptor required for the LPS-induced expression of CD14.

Down-modulation through protein kinase C-alpha of lipopolysaccharide-induced expression of membrane CD14 in mouse bone marrow granulocytes.

We have previously shown that stimulation of mouse bone marrow granulocytes (BMC) by lipopolysaccharide (LPS) induces the expression of CD14. We found here that phorbol 12-myristate 13-acetate (PMA) blocks this LPS effect. The aim of this study was to investigate the mechanism by which PMA can block the LPS signaling pathway in BMC. The unmodified binding of a radiolabeled LPS in PMA-treated cells indicated that the PMA effect was not the consequence of a shedding or an internalization of the LPS receptor, but was rather due to a biochemical event that follows the interaction of LPS with its receptor. The observations that a selective activator of protein kinase C (PKC)-alpha (sapintoxin D) mimics the PMA effect, whereas a selective PKC-alpha inhibitor (Ro-320432) antagonizes this effect, suggest a regulatory role of PKC-alpha in the LPS signaling pathway in mouse BMC.

The lipid A region of lipopolysaccharides from Rhizobiaceae activates bone marrow granulocytes from lipopolysaccharide-hyporesponsive C3H/HeJ and C57BL/10ScCr mice.

We established in previous studies that the binding of Salmonella lipopolysaccharide (LPS) to constitutive receptors of low affinity triggers the expression of the inducible LPS-binding molecule CD14 in bone marrow cells (BMC) of C3H/HeOU mice, but not in BMC from C3H/HeJ mice. We show in this study that BMC from C3H/HeJ and C57BL/10ScCr mice do not express CD14 after exposure to LPSs from Salmonella enterica and Bordetella pertussis, but do express this marker when treated with several LPSs from Rhizobiaceae, or their lipid A fragments. This shows that the constitutive LPS receptor in BMC from C3H/HeJ and C57BL/10ScCr mice is fully able to trigger a complete signalling cascade. Results of cross-inhibition of the binding of radiolabelled LPS indicated that active LPSs (from R. species Sin-1 and R. galegae) and inactive LPSs (from S. enterica and B. pertussis) bind to the same site of the constitutive LPS receptor of C3H/HeJ cells. Furthermore, binding of R. species Sin-1 LPS, and signalling induced by this LPS, were both inhibited by pre-exposure of C3H/HeJ cells to B. pertussis lipid A. This correlation between binding and signalling suggests that in C3H/HeJ cells, the constitutive receptor, which recognizes a large panel of LPSs from different origins, appears selectively unable to be activated by some particular LPSs, such as those of Enterobacteria and Bordetella.

Protein phosphorylation pathways involved during lipopolysaccharide-induced expression of CD14 in mouse bone marrow granulocytes.

Lipopolysaccharide (LPS) of Gram-negative bacteria interacts with a CD14-independent receptor of mouse bone marrow granulocytes (BMC), and triggers in these cells the expression of CD14, an inducible type of LPS receptor (iLpsR). This particular response of BMC to LPS required the activation of protein tyrosine kinase and p38 MAP kinase. The inhibition of the LPS effect by the MEK inhibitor PD-98059 suggested that the ERK pathway was also involved. Unexpectedly, protein kinase C, myosin light chain kinase, cAMP-, cGMP-, and Ca(2+)/calmodulin-dependent kinases, as well as ecto-protein kinases, were not required for iLpsR expression. However, other yet unidentified serine/threonine protein kinase(s) were implied since the BMC response to LPS was markedly reduced after exposure to three inhibitors of such kinases (K-252a, H-7, and KT-5823). The atypical kinase requirements observed in this study may be due either to a novel signaling LPS receptor complex present in BMC, or to the particular events involved in CD14 biosynthesis.

Exogenous cyclic AMP, cholera toxin, and endotoxin induce expression of the lipopolysaccharide receptor CD14 in murine bone marrow cells: role of purinoreceptors.

Little is known about the mechanisms of lipopolysaccharide (LPS) signaling in immature cells that do not express the LPS receptor CD14 yet. Bone marrow granulocytes do not constitutively express CD14 but can be stimulated by low doses of LPS in the absence of serum and then express an inducible form of LPS receptor (iLpsR). We show that in addition to LPS, cholera toxin (CT) and various cyclic AMP (cAMP) analogs can also induce the expression of iLpsR, which was identified as CD14. Induction was independent of intracellular cAMP. The hypothesis that cAMP analogs act via a cell surface receptor was suggested by the unresponsiveness of trypsin-treated cells to these inducers and by the specific binding of [(3)H]cAMP to the cells. This binding was not inhibited by LPS or CT but was inhibited by various purine derivatives. However, the receptor involved is not a conventional purinoreceptor since both an agonist and an antagonist of such receptors were able to induce iLpsR expression. The results suggest that cAMP analogs and other purine derivatives induce iLpsR after interaction with an unconventional, trypsin-sensitive, purinoreceptor distinct from LPS and CT receptors.

Lipopolysaccharide and the glycoside ring of staurosporine induce CD14 expression on bone marrow granulocytes by different mechanisms.

We established previously that lipopolysaccharide (LPS) can induce the expression of LPS-binding sites on bone marrow cells (BMC). We now report that staurosporine (STP), a glycosylated indolocarbazole alkaloid with potent inhibitory activity for various protein kinases, can induce the same effect. With both agents, the newly expressed LPS receptor was found to be CD14. The STP-induced effect was independent of its protein kinase inhibitory activity because several other protein kinase inhibitors, such as the indolocarbazole K-252a, the bisindolylmaleimide RO-31-8220, the perylenequinone calphostin C, and the isoquinolinesulfonamide H7, did not induce CD14 expression. The observation that the STP analog K-252a with an identical polyaromatic aglycon moiety was inactive yet the analog UCN-01 with an identical glycoside ring was active suggests that the induction of CD14 expression is triggered by the sugar moiety of STP. Three lines of evidence show that the mechanism of CD14 expression induced by STP differs from that induced by LPS: (i) unlike LPS, STP can stimulate BMC from LPS-unresponsive C3H/HeJ mice, (ii) LPS and STP effects are additive at a saturating dose of LPS, and (iii) the protein kinase inhibitor K-252a inhibits the LPS-induced but not STP-induced stimulation. Therefore, our findings show that both a protein kinase-dependent (LPS-induced) and a protein kinase-independent (STP-induced) mechanism can lead to the expression of the LPS receptor CD14 on BMC. We also found that the STP-induced stimulation of BMC is modulated by cyclosporin A, vinblastine, and verapamil. This observation may suggest that the inducible effect of STP could be initiated by its interaction with P-glycoprotein, a membrane pump with drug efflux function that plays a critical role in the multidrug resistance of cancer cells.

Functional lipopolysaccharide receptors of low affinity are constitutively expressed on mouse bone marrow cells.

Although lipopolysaccharide (LPS)-induced overproduction of cytokines, involved in the pathogenesis of septic shock, occupies the spotlight of endotoxin research, another LPS effect, the differentiation of various cell types including haematopoietic bone marrow cells (BMC), which is probably related to its radioprotective activity, deserves equal attention. We have previously established that nanomolar concentrations of LPS trigger in human BMC the expression of CD14 by an induction mechanism independent of CD14 or any other molecule anchored to the cell membrane by a glycosyl phosphatidylinositol glycolipid. We now show that this LPS-induced stimulation is triggered by the binding of a small number of LPS molecules (13,000 molecules/cell) to constitutive LPS receptors of low affinity (Kd = 480 nM). This interaction, which was inhibited by a synthetic LPS antagonist, appeared specific, reversible, saturable, time- and temperature-dependent, but was independent of divalent cations, and was inhibited by serum. Exposure of BMC to LPS did not induce a down-modulation of these receptors, but enhanced their sensitivity to trypsin degradation. Inhibition of LPS binding following different treatments correlated with inhibition of BMC stimulation, thus suggesting that the sparse constitutive receptors of low affinity are efficient signalling receptors for LPS.

Inability of the Francisella tularensis lipopolysaccharide to mimic or to antagonize the induction of cell activation by endotoxins.

We studied the ability of the lipopolysaccharide (LPS) extracted from a vaccine strain of Francisella tularensis (LPS-Ft) to mimic LPSs from other gram-negative bacteria for activation of various murine cell types or to antagonize the effects of other LPSs. We found that activation of macrophages for the production of tumor necrosis factor alpha and NO, of pre-B lymphocytes for the expression of surface immunoglobulins, and of bone marrow cells for the expression of LPS-binding sites was either undetectable with LPS-Ft or required concentrations 100 to 1,000 times higher than for standard LPSs. Preexposure of macrophages to LPS-Ft also failed to trigger down-regulation of tumor necrosis factor alpha (desensitization) or up-regulation of NO responses to an endotoxin challenge. In contrast to other atypical LPSs, LPS-Ft was also unable to antagonize any of the endotoxin-induced cellular responses mentioned above, suggesting that this LPS does not interact with LPS receptors.

Variation of LPS-binding capacity, epitope expression, and shedding of membrane-bound CD14 during differentiation of human monocytes.

The myeloid differentiation Ag CD14 is considered to play a critical role in the binding of LPS to monocytes. To determine if differences in LPS-binding capacities of cells could reflect a variability of CD14 molecules, we analyzed the interactions of various reagents with these molecules in human blood monocytes and in promyelocytic (HL60) and monocytic (THP-1) cell lines. The expression of CD14 epitopes was analyzed with the fluorescent anti-CD14 mAbs My4 and LeuM3. Expression of LPS-binding sites (LPS+ molecules) was detected with LPS-FITC. THP-1 cells stimulated with calcitriol (VitD3), as well as the majority of blood monocytes (50-90%) were My4+/LPS+. However, untreated THP-1 cells, and a substantial population (10-50%) of human monocytes from healthy donors, were My4+/LPS-, thus suggesting the existence of CD14 isoforms with different LPS-binding capacities. In line with this assumption, monocytes stimulated with PMA selectively shed LeuM3+ molecules, but almost no My4+ and LPS+ constituents. Analysis of monocytes after treatment with phosphatidylinositol-specific phospholipase C indicated that among CD14 molecules with LPS-binding capacity, some are susceptible and others are resistant to the enzyme, each type being mainly expressed by a different monocyte subset. Studies of uninduced and chemically induced THP-1 cells showed that wheat-germ agglutinin blocked the binding of My4 to constitutive, but not to chemically inducible CD14. The overall results suggest the existence of at least three different forms of CD14, which may reflect different stages of cell maturation.

Phosphatidylinositol-anchored molecules and inducible lipopolysaccharide binding sites of human and mouse bone marrow cells.

We have previously established that lipopolysaccharide (LPS) induces the expression of new specific LPS-binding sites (LpsR) in mouse bone marrow cells (BMC). We now show that exposure of human BMC to LPS elicits the production of both CD14 molecules (detectable with monoclonal antibody My4) and LpsR (detectable with fluorescein isothiocyanate-LPS). Pretreatment of stimulated human BMC with My4 inhibited the binding of fluorescein isothiocyanate-LPS. The stimulation of human BMC, but not mouse BMC, required the presence of serum. Other characteristics of mouse and human BMC examined were very similar. Their inducible LpsR interacted with the lipid moieties of LPS and Leishmania donovani lipophosphoglycan and with a soluble preparation of peptidoglycan. Moreover, mouse and human LpsR were susceptible to treatment with a phosphatidylinositol-specific phospholipase C (PI-PLC), thus suggesting that both are PI-anchored CD14 molecules. Neither LpsR appeared able to interact with a synthetic LPS antagonist (compound PPDm2) structurally related to the lipid region of LPS. However, PPDm2 blocked LPS-induced expression of LpsR in both BMC. Furthermore, in both species, pretreatment of BMC with PI-PLC did not prevent the cells from expressing LpsR in response to LPS. The results support the hypothesis that the elicited LpsR of mouse and human BMC is an inducible form of CD14, whereas the putative "signaling LPS receptor" of these cells is not CD14 or any other PI-anchored molecule.

Lipopolysaccharide and tumor necrosis factor-alpha induce lipopolysaccharide receptor expression on bone marrow cells by different mechanisms.

Two cell types, monocytes/macrophages and neutrophilic granulocytes, play a prominent role in the pathogenic effects of endotoxins during Gram-negative infections. We previously established that nanomolar concentrations of LPS induce the expression of new specific LPS-binding sites (LpsR) in bone marrow granulocytes of LPS-responsive mice. To examine this induction process further, we asked whether it, like other LPS activities, can be mediated by TNF-alpha. We report that exogenous rTNF-alpha can induce LpsR expression in bone marrow cells (BMC) from both LPS-responsive (C3H/HeOU) and LPS-hyporesponsive (C3H/HeJ) mice. In BMC, LPS elicited a down-regulation of the TNF-alpha receptors (TNF-R) without direct binding to TNF-R. On the other hand, taxol, a microtubule stabilizer that has shown LPS mimetic activity in macrophages, was unable to elicit LpsR expression or induce TNF-R down-regulation in BMC. Thus, unlike the LPS-signaling receptor of macrophages, that of BMC is apparently not functionally associated with microtubules. The LPS-induced expression of LpsR was inhibited only partially with an anti-TNF-alpha serum, and with dexamethasone, suggesting that an autocrine activity of endogenous TNF-alpha cannot alone account for the LPS effect. Comparative analyses also indicated that dexamethasone inhibited the LPS-induced increase of LpsR, but enhanced the number of TNF-induced LpsR+ BMC. Furthermore, the synthetic lipid PPDm2 (a 1,4-bisphosphorylated and N,N-diacylated derivative of 2,3-diamino-2,3-dideoxy-D-glucose) inhibited LPS-induced, but not TNF-induced, expression of LpsR. These data show that in BMC, LPS and TNF-alpha induce LpsR expression by different mechanisms.

A synthetic analog of the 3-deoxy-D-manno-2-octulosonic acid disaccharide moiety of rough-type endotoxins does not bind to mouse peritoneal macrophages and human monocytes.

Strong evidence supports the concept that lipid A is the main biologically active region of endotoxins and is recognized by specific binding sites of different cell types. However, receptors for carbohydrates are also present on mononuclear phagocytes, and it has been suggested that one of these lectin-like proteins may be specific for the 3-deoxy-D-manno-2-octolosonic acid (Kdo) residues of endotoxins. To reexamine this hypothesis, we prepared a 125I-labeled conjugate consisting of a synthetic Kdo-2,4-Kdo disaccharide covalently linked to bovine serum albumin (125I-Kdo2-BSA). The Kdo disaccharide residues of this radiolabeled conjugate were fully accessible to a monoclonal antibody which reacts specifically with this epitope. However, 125I-Kdo2-BSA did not exhibit any detectable specific binding on thioglycolate-elicited mouse peritoneal macrophages or on human monocytes. Furthermore, the specific binding of biotin-labeled lipopolysaccharide derivatives to mouse macrophages and human monocytes was not inhibited by a soluble synthetic Kdo-2,4-Kdo-polyacrylamide copolymer or by a synthetic glycolipid consisting of an alpha-Kdo residue glycosidically linked to O-6 of allyl-4-O-phosphoryl-N-3-hydroxytetradecanoyl-beta-D-glucosaminide. These results indicate that binding sites specific for Kdo are not present (or not accessible) on the surface of mouse macrophages and human monocytes.

Endotoxin-induced expression of endotoxin binding sites on murine bone marrow cells.

A variety of binding sites for endotoxin (LPS) have been identified on leukocytes. However, the sequence of expression of these receptors, and their interrelations, are poorly understood. In this report, we show that in LPS-responsive hosts, interaction of nanomolar concentrations of LPS with bone marrow cells induces the expression of new specific LPS-binding sites. Cells from LPS-nonresponsive (C3H/HeJ) mice do not express these receptors after LPS treatment. Experimental differences in the conditions allowing the induction and the detection of these binding sites (influence of Leishmania lipophosphoglycan, role of serum), support the hypothesis that interaction of LPS with primary receptors on bone marrow cells triggers the expression of secondary LPS receptors, and that the two types of receptors have distinct fine specificities.

Differential effects of a monoclonal antibody recognizing 3-deoxy-D-manno-2-octulosonic acid on endotoxin-induced activation of pre-B cells and macrophages.

A monoclonal antibody (E1), which reacts with the 3-deoxy-D-manno-2-octulosonic acid region of lipopolysaccharides belonging to the rough Re chemotype (LPS-Re) was used to analyze in vivo and in vitro effects of endotoxin. E1 inhibited LPS-Re-induced activation of Z0Z/3 pre-B cells (expression of surface immunoglobulins) and mature splenic B cells (DNA synthesis) and blocked the binding of biotin-labeled LPS-Re to mouse peritoneal macrophages. However, other effects elicited by LPS-Re, such as the production of interleukin 6 and tumor necrosis factor alpha by macrophages, and the acute lethality in galactosamine-sensitized mice, were not inhibited by E1. The results suggest that the specificity of the LPS receptor which triggers the production of cytokines in macrophages is different from that of the LPS receptor which induces activation of pre-B and B lymphocytes, and is also different from that of the major LPS-binding sites of macrophages. This is consistent with the hypothesis that the large majority of LPS-binding sites of macrophages do not trigger cytokine production, and that the small number of "signaling LPS receptors" of macrophages have fine specificities which are different from those of B cells.

Nontoxic forms of lipopolysaccharides, poorly cleared in vivo, are present in variable amounts in phenol-extracted endotoxins.

A radiolabeled lipopolysaccharide (LPS) from Salmonella choleraesuis was as toxic for galactosamine-treated mice as the unlabeled preparation. After intravenous injection, a large proportion (37%) of the labeled material remained in the circulation. This circulating form of LPS was not toxic, even when reextracted with phenol, and remained in the circulation when reinjected into mice. Furthermore, toxic and nontoxic constituents were directly isolated in vitro by hydrophobic gel chromatography of the radiolabeled LPS preparation. The nontoxic form, of lower molecular weight according to its electrophoretic migration, did not react with an anti-lipid A monoclonal antibody but reacted with polymyxin B and was as active on pre-B and B cells as the unfractionated LPS preparation. The results of this study suggest that some LPS constituents of phenol-extracted endotoxin preparations are not toxic for galactosamine-sensitized mice and are not trapped by appropriate LPS-binding sites in the liver as efficiently as are the toxic forms.

Preparation and binding specificity of a monoclonal antibody recognizing 3-deoxy-D-manno-2-octulosonic acid (Kdo) in lipopolysaccharides of Re chemotype.

A mouse monoclonal antibody (MAb E1) was raised against the lipopolysaccharide (LPS) of the Re mutant R595 of Salmonella minnesota. This IgG3 antibody (MAb E1), unstable at low pH and low ionic strength, was purified by chromatography on QAE Sepharose A50. The binding specificity of MAb E1 was characterized by direct and inhibition enzyme immunoassays, using natural LPSs from different strains and chemotypes, and synthetic analogs of LPS substructure of the 3-deoxy-D-manno-2-octulosonic acid (Kdo) and Lipid A regions. Among various LPSs, MAb E1 reacted exclusively with those of Re-chemotype. It recognized alpha-Kdo- monosaccharide and disaccharide structures present as non-reducing side chains in various Re-type LPSs and synthetic antigens. The antibody did not react with Lipid A or various lipids, and the presence of the lipid region was not necessary for the reaction. The recognition of the epitope was not reduced by the presence of a substituent at O-8 of one of the two Kdo units present in the Re LPS from Proteus mirabilis, but the reaction was inhibited by phosphorylation of O-4 of Kdo, by the proximity of core (heptose) or Lipid A (acylated glucosamine) residues, or by certain LPS-LPS interactions.

New synthetic analogs of lipid A as lipopolysaccharide agonists or antagonists of B lymphocyte activation.

We have studied the ability of synthetic analogs of lipid A to mimic lipopolysaccharide (LPS) for activation of 70Z/3 pre-B cells (expression of surface Igs) or to antagonize this effect. The results indicate that the presence of glucosamine (mono- or disaccharide) as a 'backbone' for the attachment of fatty acids is not necessary for activation of cells of the B lineage. Phosphate groups are not necessary either. Other structural features such as the configuration of particular asymmetric carbons, and the distance between an anionic group and an N-acyl chain, seem to be much more critical parameters for activation of B cells. Among the synthetic lipids which were unable to activate 70Z/3 cells, one compound, consisting of N,N-acylated and bisphosphorylated 2,3-dideoxy-2,3-diamino-D-glucose, behaved as a specific LPS antagonist and blocked also the activation triggered by the other synthetic inducers. The influence of the synthetic lipids on the entry of mature mouse B lymphocytes into the G1A phase of the cell cycle (cell enlargement) was also investigated. A high correlation was observed between the potency to activate pre-B cells and the ability to induce blast formation in mature B cells.

The splenic pool of mouse stem cells: in vitro differentiation and expression of the BP-1 alloantigen on cells of the lymphoid and myeloid lineages.

The relative paucity of data about the development of the stem cell pool present in the spleen prompted this study. During in vitro cultures of B-enriched lymphocytes from mouse spleens and in the presence of a culture supernatant of WEHI-3 cells (WEHI-SUP), a population of cells expressing the BP-1 antigen appears progressively, reaches an optimal size 8 days after initiation of the culture, and disappears on day 28. In 8-day-old cultures, a minor population of cells bearing both BP-1 and B220 can be detected. The growth of this cell population, with characteristics of the B lymphoid lineage (pro-B), is strictly dependent on the presence of WEHI-SUP in the medium. After 2 weeks of culture, the BP-1 antigen is expressed on a cell population, which is essentially constituted of B220-, polynuclear cells. The BP-1 antigen, which is considered as characteristic of early cells of the B lymphoid lineage, can therefore also be expressed on cells of the myeloid lineage. The injection of BP-1+ or B220+ cells in irradiated mice can hardly reconstitute their B cell pool, whereas BP-1- and B220- cells are much more efficient in vivo progenitors of this cell lineage.

Effects of lipopolysaccharide on macrophages analyzed with anti-lipid A monoclonal antibodies and polymyxin B.

Six monoclonal antibodies (mAb) to the lipid A region of bacterial lipopolysaccharide (LPS), obtained from mice immunized with lipid A-coated Bordetella pertussis cells (mAb 3.E8, 2.21, 2.37, 2.41) or with lipid A covalently coupled to keyhole limpet hemocyanin (mAb R1 and R7), were examined for their potential to inhibit in vitro activities of LPS on macrophages. mAb R7 was inactive in vitro, but the five other mAb inhibited efficiently some in vitro activities of LPS. mAb R1, 2.21 and 3.E8 reduced the LPS-induced secretion of tumor necrosis factor (TNF) and interleukin 1 (IL 1) by macrophages, but did not modify the binding of LPS to macrophages. On the other hand, mAb 2.37 and 2.41 reduced LPS binding to macrophages and subsequent IL1 secretion, but did not modify TNF production. This is in agreement with our previous finding that IL1 and TNF productions can be selectively triggered by synthetic analogs of lipid A substructures (Lasfargues and Chaby, Cell. Immunol. 1988. 115: 165). The pattern of in vitro inhibition of LPS activities (LPS binding to macrophages and production of TNF and membrane IL 1) by polymyxin B was different from those of the two groups of anti-lipid A mAb mentioned above. These observations suggest the presence on lipid A of four functionally distinct substructures.