Search

Objectives: Endotoxin is one of the principal mediators of Gram-negative septic shock. Pretreatment with monophosphoryl lipid A, a hydrolyzed derivative of endotoxin from Salmonella minnesota R595, induces endotoxin tolerance and nonspecific resistance to infection in experimental animals. The present clinical trial was undertaken to test the response to monophosphoryl lipid A in humans and the ability of monophosphoryl lipid A to attenuate the response of normal human volunteers to U.S. Reference Ec-5 endotoxin., Design: Prospective, double-blind, randomized, controlled trial., Setting: Clinical research center., Patients: Forty-four healthy volunteers., Interventions: In part 1 of the study, 29 volunteers were randomized in varying ratios to receive vehicle control or monophosphoryl lipid A intravenously in a double-blind dose escalation trial. In part 2 of the study, 12 volunteers were randomized to receive either monophosphoryl lipid A (20 micrograms/kg) or vehicle control and, 24 hrs later, all 12 volunteers were challenged with U.S. Reference Ec-5 endotoxin (20 units/kg intravenous, bolus injection). Systemic response to endotoxin challenge was evaluated and compared between the monophosphoryl lipid A and vehicle control-pretreated subjects., Measurements and Main Results: In part 1 of the study, subjective effects and increases in cytokine levels were not observed until a dose of 10 micrograms/kg of monophosphoryl lipid A was administered. Six volunteers receiving a maximum dose of 20 micrograms/kg experienced mild-to-moderate symptoms that did not require therapy. Moderate increases in temperature, heart rate, and tumor necrosis factor (TNF)-alpha, interleukin (IL)-6, and IL-8 release were observed. IL-1 alpha and IL-1 beta were not detected but a significant increase in IL-1 receptor antagonist was observed. In part 2 of the study, monophosphoryl lipid A pretreatment reduced the number of volunteers who experienced one or more subjective complaints after endotoxin administration (3/6 vs. 6/6; p = .09). The febrile response and tachycardic response to endotoxin were significantly reduced by pretreatment with monophosphoryl lipid A. Monophosphoryl lipid A-pretreated volunteers demonstrated significantly reduced concentrations of TNF-alpha after endotoxin challenge, as compared with subjects treated with vehicle control (84 +/- 76 vs. 244 +/- 128 pg/mL; p < .05). IL-6 concentrations (100 +/- 91 vs. 268 +/- 171 pg/ml; p < .05) and IL-8 concentrations (136 +/- 86 vs. 632 +/- 323 pg/mL; p < .05) elicited by endotoxin challenge were also significantly reduced by monophosphoryl lipid A pretreatment., Conclusions: Data indicate that monophosphoryl lipid A, in a dose 10,000 times that of endotoxin, used in experimental pyrogenicity trials, is well tolerated in human volunteers. Pretreatment of normal human volunteers with monophosphoryl lipid A attenuated the systemic response to bacterial endotoxin. These data support further clinical testing of monophosphoryl lipid A for the prevention or amelioration of the severe sequelae of sepsis.

Monophosphoryl lipid A (MLA), derived either from Salmonella minnesota or Salmonella typhimurium, was tested for its ability to alter Candida albicans mannan (MAN)-specific suppression. Since we showed previously that naive mice injected intravenously (i.v.) with MAN developed suppressor T cells capable of down-regulating delayed-type hypersensitivity when transferred to immunized recipients, MLA was tested for its ability to influence suppressor activity in the donors of suppressor cells. T-lymphocyte-enriched suspensions from donor mice treated with MLA, especially that derived from S. typhimurium, 2 or 3 days after the injection of MAN lost the ability to suppress delayed-type hypersensitivity when transferred to immunized mice. Transferable suppressor activity was reduced but not always completely abrogated when donor animals were treated with MLA 1 day following the administration of MAN. In several experiments, S. minnesota MLA also abrogated activity, but it was not effective in other transfer experiments. In a different type of experiment, MLA was given to immunized mice which had been suppressed directly with MAN. Mice were immunized, either by the introduction of C. albicans intragastrically followed by inoculation intradermally (i.d.) or by two i.d. inoculations, and MAN-specific suppressor cells were induced in such animals by the i.v. injection of MAN 1 day before the first or second i.d. inoculation in animals given intragastric plus i.d. inoculations and those given two i.d. inoculations, respectively. MLA was administered to such mice prior to the i.v. injection of MAN, on the same day, or 1 to 4 days thereafter. S. typhimurium MLA, especially when given to mice 2 days following the administration of MAN, caused a partial abrogation of suppressor activity. Overall, however, MLA, at 5 to 100 micrograms, had variable and minimal effects on suppressor activity in immunized mice suppressed by the i.v. administration of MAN. In summary, MLA is clearly capable of abrogating MAN-induced suppression when given to nonimmunized animals in which MAN-specific suppressor cells had been induced, but its efficacy in immunized animals suppressed by the i.v. administration of MAN was marginal.

Concurrent injection of monophosphoryl lipid A (MPL) in saline or as an oil-in-water emulsion enhanced both the primary and secondary serum antibody responses to the capsular polysaccharide (CP) components of seven conjugates: the enhanced responses were Ag-specific. In contrast, MPL did not enhance the serum antibody response to five of the six unconjugated CP. MPL and trehalose dimycolate injected concurrently with the unconjugated Vi CP of Salmonella typhi (Vi) enhanced the serum antibody response to that Ag. MPL further enhanced the Vi antibody levels when injected with conjugates of this CP. The serum antibody responses to Pseudomonas aeruginosa exotoxin A, used as the carrier protein for the Staphylococcus aureus types 5 and 8 conjugates, were also enhanced by MPL. MPL in oil-in-water emulsion was generally more effective than when administered in saline.

Glycogen-induced guinea pig polymorphonuclear leukocytes (PMNs) were employed in experiments designed to establish the optimal conditions for casein-stimulated chemotaxis. Subsequently, it was shown that guinea pig PMNs, in a serum-free medium, were susceptible to inhibition of migration by three distinct types of cell-directed inhibitors of chemotaxis found in normal or diseased human sera. Comparison of inhibition of migration of both guinea pig and human peripheral PMNs showed that guinea pig and human PMNs were equally susceptible to inhibition by sera from trauma victims; this inhibitor had a molecular weight of about 10,000 d. Human PMNs were slightly more susceptible than were guinea pig PMNs to inhibition of migration by a approximately 110,000-d inhibitor found in normal human sera. On the other hand, guinea pig PMNs were somewhat more susceptible to inhibition of migration by a approximately 400,000-d inhibitor of chemotaxis that was analogous to the inhibitor found in anergic serum. This information shows that guinea pig PMNs, in a serum-free medium, may be substituted for human cells in quantitative assays for these human serum factors.

The monoclonal antibodies produced by a variety of hybridomas making antibody specific for E. coli 0113 lipopolysaccharide (LPS) were purified by affinity chromatography and their fine specificity studied. All reacted specifically with the polysaccharide moiety of LPS from E. coli 0113 and from Neisseria lactamica; two reacted with LPS from Pseudomonas aeruginosa and one reacted with LPS from Klebsiella pneumoniae. Polyclonal and monoclonal syngeneic and semi-syngeneic anti-idiotypic antisera were produced to study the idiotypy of LPS-specific monoclonal antibodies which express a complex cross-reactive idiotype (IdX) as well as individual idiotypes. E. coli 0113 LPS-specific antibodies produced by BALB/c mice express this IdX and the kinetics for its expression was examined using mice either primed or hyperimmunized with LPS; idiotypic maturation was observed, but we were unable to detect an auto-anti-idiotypic antibody response. This IdX was expressed on E. coli 0113 LPS-specific antibodies from all strains of mice examined, indicating that its expression is not restricted by genes linked to the IgCH locus.

Immunological memory to the lipopolysaccharide of Escherichia coli O113 was generated in strains of inbred mice given a single subimmunogenic dose of either E. coli O113 lipopolysaccharide or the native protoplasmic polysaccharide of E. coli O113. Such memory, which only involved antibody of the immunoglobulin M class, developed in a cyclic manner that was characteristic for the strain of mice used. It involved cell proliferation as well as differentiation and persisted for at least 25 days after priming with a single injection of a subimmunogenic dose of E. coli O113 lipopolysaccharide.

RIIIS/J mice lack an autosomal dominant gene(s) that has a decisive influence in determining the capacity to make an IgM antibody response to several representative helper T cell-independent antigens; these mice appear to possess functionally active helper T cells because of their ability to make a reasonably good antibody response to SRBC, a helper T cell-dependent antigen. The genetic defect involved cannot be attributed to the absence of B cells capable of giving a mitogenic response to bacterial LPS.

Treatment with nontoxic monophosphoryl lipid A (MPL), which was derived from a polysaccharide-deficient, heptoseless Re mutant of Salmonella typhimurium, was found to inactivate suppressor T-cell activity, as evidenced by a decrease in the degree of low-dose immunological paralysis expressed and an increase in the magnitude of the antibody response to type III pneumococcal polysaccharide. The effects produced, which could not be attributed to the polyclonal activation of immune B cells by MPL, were dependent upon the dose of MPL used, as well as the time when MPL was given relative to low-dose priming or immunization with type III pneumococcal polysaccharide. Neither amplifier nor helper T-cell activity was decreased by treatment with the same, or larger, doses of MPL. The significance of these findings to the use of MPL as an immunological adjuvant or an immunomodulating agent is discussed.

Functional separation of the two signals involved in stimulating immunological responses was achieved through the judicious use of two natural bacterial antigens. Native protoplasmic polysaccharide (NPP) extracted from Escherichia coli was immunochemically identical to the lipopolysaccharide (LPS) extracted from the same organism. However, NPP was not endotoxic, not mitogenic, did not fix complement, and was immunologically independent of T cells. The NPP, which appeared to contain only the antigenic signal, could induce a primary antibody response in mice and could sensitize mice for a secondary response. However, the antigenic signal contained in NPP was insufficient to trigger a secondary response in mice primed with either NPP or LPS. LPS, containing both the antigenic and second signals, was required to trigger a secondary response in primed mice.

Treatment with nontoxic monophosphoryl lipid A increased the magnitude of the immunoglobulin M (IgM) antibody response to type III pneumococcal polysaccharide in young (2- to 4-week-old) mice. This was accompanied by the appearance of significant numbers of IgG1- and IgG3- secreting antibody-forming cells in 4-week-old mice. These findings indicate that monophosphoryl lipid A can be used as an adjuvant to improve the immunogenicity of poorly immunogenic antigens in young, immunologically immature animals.

Treatment with bacterial lipopolysaccharide elicits the appearance of greater numbers of background antigen-specific plaque-forming cells (PFC) in the spleens of mice previously exposed or primed to subimmunogenic amounts of various non-cross-reacting antigens so as to generate detectable immunological memory. These findings suggest that treatment with lipopolysaccharide results in the activation of increased numbers of antigen-primed or memory B cells in mice previously exposed to antigen.

The effect of bacterial lipopolysaccharide (LPS) on the capacity of mice to mount a specific T-cell response to a protein antigen was examined. Inoculation of mice with LPS and ovalbumin (OA) resulted in enhancement of the T-cell proliferative response to OA. This enhancement was manifested in vitro by an increase in magnitude and by a more rapid appearance of the response after challenge with OA. This enhancement was also shown because the latent periods for the antigen-specific responses were reduced to 3 days after inoculation with antigen plus LPS, as compared with 5 days after inoculation with antigen alone. Various mouse strains, including the C3H/HeJ and C57BL/10ScN strains, responded to the adjuvant action of LPS at the T-cell level. Results suggest that LPS exerts this adjuvant effect by facilitating clonal expansion of antigen-reactive T cells during priming.

Mice given a single optimally immunogenic dose (20 micrograms) of bacterial LPS make an antibody response that is characterized by cyclic (oscillatory) patterns. This kinetic pattern is dose as well as mouse strain dependent. LPS can induce a secondary response in BALB/c mice that is also cyclic, although it differs in amplitude as well as periodicity from a primary response. To rule out the involvement of B cell mitogenic activity of LPS in the expression of such cyclic patterns, the responses produced by both C3H/HeN and C3H/HeJ mice to a single injection of LPS were compared; the latter do not give B cell proliferative responses to LPS and also have been reported to give low LPS-specific responses in vivo. Both strains gave an oscillatory kinetic pattern. By contrast, BALB/c nu/nu mice given an optimal immunogenic dose of LPS (2 micrograms) do not present a cyclic immune response. This indicates that mature T cells are required for the generation of an oscillatory anti-LPS response.

Treatment with trehalose dimycolate (TDM) increases the magnitude of the immunoglobulin M (IgM) antibody response of mice to type III pneumococcal polysaccharide (SSS-III). Such enhancement is demonstrable over a wide range of immunizing doses and does not require thymus-derived (T) cells to be elicited. Although young adult mice immunized with SSS-III do not usually make anti-SSS-III antibodies of the IgG1 and IgG3 classes, antibodies of one or both isotypes were produced after immunization and treatment with TDM and/or monophosphoryl lipid A (MPL); the additive nature of the effect produced by both TDM and MPL suggests that the two immunomodulators act by different mechanisms. TDM and MPL have different effects on the induction and expression of low-dose immunological paralysis, a form of unresponsiveness known to be mediated by suppressor T cells. The relevance of these findings to the modes of action of TDM and MPL is discussed.

The quantitative and qualitative nature of the antibody responses of euthymic (normal, RML) and athymic (nude) mice injected with dinitrophenylated (DNP) lipopolysaccharide (LPS) was evaluated. Antibody responses to both the haptenic (DNP) and carrier (LPS) determinants were measured. On a quantitative basis, RML and nude mice stimulated with DNP-LPS produced only primary anti-DNP responses, whereas both primary and secondary anti-LPS responses were elicited by this conjugate. The failure of DNP-LPS to trigger secondary anti-DNP responses was not dependent on the amount of DNP-LPS given in the primary or secondary doses and could not be overcome by repeated injections of DNP-LPS. Also, the anti-DNP responses of RML mice injected repeatedly with DNP-LPS were restricted to immunoglobulin M antibodies whereas both immunoglobulin M and G anti-LPS responses were elicited. Nude mice also produced immunoglobulin G antibodies to the LPS determinants. These data showed a dissociation of the anti-hapten and anti-carrier antibody responses and suggested that different immunological signals were functioning in the respective anti-DNP and anti-LPS responses.

The capacity to precipitate quantitatively with homologous antiserum of native protoplasmic polysaccharide from Escherichia coli was not reduced by incubation in human plasma; it was reduced with the immunochemically related lipopolysaccharide.

A radioimmunoassay has been developed for the serodiagnosis of gonorrhoea. When 382 human sera from acutely infected and well documented control subjects were assayed by this procedure, 90% were diagnosed correctly, 5% were diagnosed incorrectly, and in 5% the results were indecisive.

Carrageenan can potentiate the lethal effects of endotoxin by 100- to more than 3,000-fold.

The antibody responses of outbred normal mice and nude mice injected with alkaline detoxified lipopolysaccharide (Alk-LPS) were measured. In some cases the antibody against lipopolysaccharide (LPS) and native protoplasmic polysaccharide (NPP). The kinetics of the primary responses to Alk-LPS and NPP were similar, whereas LPS stimulated a more rapid appearance of antibodies in the primary responses. Alk-LPS stimulated only primary antibody responses in both types of mice and sensitized nude mice for secondary responses which could be triggered with LPS. However, secondary antibody responses could not be triggered in normal mice primed with Alk-LPS. These data suggested that, on a functional basis, Alk-LPS possessed the specific antigenic signal associated with LPS antigens but lack the second nonspecific mitogenic signal dependent on the lipid A portion of LPS.

Studies conducted with F1 and F2 progeny of crosses between strains of inbred mice that differ greatly in their capacity to make an antibody response to type III pneumococcal polysaccharide, dextran B-1355, and lipopolysaccharide from Escherichia coli 0113 have shown that multiple genes influence the magnitude of the antibody response to these antigens. Other studies with hybrids derived from crosses between C3H/HeJ, CBA/N, and RIIIS/J mice have indicated that the genetic defects characteristic of these strains of mice are dissimilar and unlinked and that autosomal, as well as X-linked, genes control serum immunoglobulin M in unimmunized mice.

CBA/N mice, which possess an X-linked immunodeficiency (xid), produce a convincing antibody response to lipopolysaccharide derived from Escherichia coli 0113 (LPS 0113), a thymus-independent antigen. The antibody response produced was shown to be specific for the O-polysaccharide moiety of LPS 0113, rather than lipid A or lipid-A-associated protein. The relevance of this finding to the nature of the genetic defect of xid-mice is discussed.

Mice of the C3H/HeJ strain, which were unresponsive to the biological effects of bacterial lipopolysaccharide (LPS), could not be induced to make specific secondary immunological responses to LPS; they responded to two doses of LPS with a primary response. This lack of secondary responsiveness by C3H/HeJ mice was due to a defect in a single, autosomal, dominant gene. Thus, further evidence was provided that an intact second immunological signal and responsiveness thereto were required to trigger secondary antibody responses in primed animals.

An in vitro proliferative system was used to assess the capacity of lipopolysaccharide (LPS) to induce a specific T-cell response in mice. Sensitized T cells were generated in vivo by subcutaneous inoculation with LPS. These T cells, which were purified on nylon wool columns, were stimulated to proliferate in vitro by LPS. Results of several lines of experimentation confirmed that the responding cells were T cells. Additional experiments indicated that sensitized T cells could distinguish between LPS and ovalbumin. Finally, it was found that genetic responsiveness to the biological effects of lipid A was required for full elicitation of LPS-induced T-cell proliferation. The data were interpreted to indicate that LPS interacted with, and stimulated, antigen-specific murine T cells.

A large pool of refined endotoxin was prepared from Escherichia coli O113 by extraction with hot acqueous phenol. It was characterized chemically and biologically and will be available for a reference standard designated as reference endotoxin EC.

Treatment with nontoxic monophosphoryl lipid A (MPL) derived from a polysaccharide-deficient, heptoseless Re mutant of either Salmonella typhimurium or Salmonella minnesota R595 enhanced the immunoglobulin M (IgM) anti-type III pneumococcal polysaccharide (SSS-III) antibody response of C3H/HeSnJ mice. Such an adjuvant effect was not observed in lipopolysaccharide-nonresponder C3H/HeJ mice. Nevertheless, C3H/HeJ spleen cells produced a weak mitogenic response to both preparations of MPL in vitro, and C3H/HeJ mice showed a significant increase in serum IgM levels without an increase in numbers of splenic IgM-secreting plaque-forming cells after in vivo treatment with MPL. A significant increase in serum IgG3 levels was accompanied by a transient decrease in serum IgG1 levels in C3H/HeSnJ mice given MPL; such non-antigen-specific polyclonal effects were not observed in C3H/HeJ or in athymic nu/nu mice. Since the enhanced antibody response to SSS-III has been attributed to the inactivation of suppressor T cells by MPL and since suppressor-T-cell activity is demonstrable in both C3H/HeSnJ and C3H/HeJ mice, these findings imply that (i) the suppressor T cells of C3H/HeJ mice are refractory to inactivation by MPL and (ii) some of the polyclonal and mitogenic effects produced in C3H/HeJ mice are due to the direct action of MPL on B lymphocytes.

A single dose of bacterial lipopolysaccharide, administered 21 days after mice had been treated with the macrophage toxins carrageenan, or microparticulate crystalline silica, resulted in a secondarytype antibody response. The phenomenon of pseudosecondary responsiveness was investigated to determine the mechanism for its generation. The results showed that generating a pseudosecondary response was T cell dependent; but except for this, it followed the same kinetic and genetic patterns as a true, two dose of antigens, secondary response. It was concluded that pseudosecondary responsiveness resulted from priming the mice by the consequential effects of the macrophage toxins; thereafter, secondary responsiveness to LPS was generated in a normal manner.

Antigenic competition between bovine gamma globulin (BGG) and endotoxin from a smooth strain (S-ET) and a rough (R-ET) heptoseless mutant strain of Salmonella minnesota was studied in mice. Both endotoxins acted as adjuvants for enhancing the antibody response to BGG. However, other work showed that the R-ET had minimal antigenicity, and it was used as a control for the competition studies. Antigenic competition between BGG and endotoxin as expressed by a suppression of the antibody response to BGG could not be demonstrated when varying adjuvant doses of S-ET or R-ET were injected simultaneously with a small constant dose of BGG into normal mice. However, mice presensitized with S-ET several weeks before immunization with the S-ET and BGG combination produced anti-BGG levels which were four to eightfold lower than in normal mice. Nearly complete suppression of the anti-BGG response could be obtained in presensitized mice by reducing the BGG dose 10-fold or by increasing the adjuvant dose of endotoxin. Mice pretreated with R-ET and challenged with BGG plus S-ET or R-ET showed no depression of the anti-BGG response. These and other experiments confirmed the immunological basis of the competitive effect.

Anacker, R. L. (Rocky Mountain Laboratory, Hamilton, Mont.), W. D. Bickel, W. T. Haskins, K. C. Milner, E. Ribi, and J. A. Rudbach. Frequency of occurrence of native hapten among enterobacterial species. J. Bacteriol. 91:1427-1433. 1966.-Smooth cultures of representative Enterobacteriaceae were screened for the presence of native hapten, a substance previously extracted with trichloroacetic acid from the protoplasmic fraction of one strain each of Escherichia coli O111:B4 and O113. Trichloroacetic acid extracts of protoplasmic fractions of the cells were analyzed for chemical composition, for constituent sugars by paper chromatography, for immunochemical relationship to endotoxin purified by gel filtration, for sedimentation behavior, and for pyrogenicity in rabbits and lethal toxicity in chick embryos. Extracts from two of three additional strains of E. coli O113, all five additional strains of E. coli O111:B4, and one strain each of E. coli O26:B6 and O55:B5 were similar to previously described native hapten in chemical composition, sedimentation properties (S(20,w), 3.7 to 5.2), biological potency (usually less than 0.1% that of corresponding endotoxin), and immunochemical relationship to endotoxin. Extracts of one strain each of E. coli O127:B8, Serratia marcescens, Citrobacter freundii, Salmonella enteritidis, and of two lipopolysaccharide-deficient mutants of S. enteritidis differed from typical native hapten. The biosynthetic relationship of native hapten to endotoxin has not yet been revealed.

Rudbach, Jon A. (The University of Michigan, Ann Arbor), and Arthur G. Johnson. Alteration of endotoxin activity after complexing with plasma proteins. J. Bacteriol. 92:892-898. 1966.-A substantial decrease in the ability of endotoxin to be precipitated by homologous antiserum and to cause fever occurred after incubation with human plasma or human plasma, Cohn fraction IV-1. The endotoxin, thus altered, also displayed decreased lethality for rabbits. These alterations in endotoxin activity could be restored when the endotoxin-plasma protein mixture was treated with a proteolytic enzyme, and the endotoxin was precipitated with ethyl alcohol. Inactivation of the antigenic and toxigenic properties of the endotoxin molecule by plasma is discussed as resulting from complexing with plasma proteins rather than from enzymatic degradation.

Protoplasm separated from disrupted cells of gram-negative bacteria was extracted with hot phenol-water or was precipitated with ethyl alcohol after digestion with Pronase. These methods recovered about 10 times more endotoxin than was detectable in the untreated protoplasm. Inactivation of endotoxin by protoplasm also occurred in vitro when the endotoxin was first dissociated into subunits before reaction with protoplasm. Despite this increased yield from another source, the major proportion of endotoxin was still found in the cell walls.

Treatment with alkali is one of several methods for removing fatty acids from bacterial endotoxins and, in the process, detoxifying the material. Saponification of fatty acid esters is the major detectable chemical change produced by alkali; however, kinetic studies of mild alkaline hydrolysis of endotoxin failed to correlate rates of detoxification with rates of loss of ketodeoxyoctonates, heptose, O-acetyl groups, or fatty acid esters. The alterations occurring during the critical stages of hydrolysis apparently changed the essential chemical conformation of endotoxic particles before cleavage of a significant amount of material took place. The rates of both saponification and detoxification were markedly increased by carrying out the reaction in media of ethyl alcohol or dimethylsulfoxide instead of water.

Ribi, E. (Rocky Mountain Laboratory, Hamilton, Mont.), R. L. Anacker, R. Brown, W. T. Haskins, B. Malmgren, K. C. Milner, and J. A. Rudbach. Reaction of endotoxin and surfactants. I. Physical and biological properties of endotoxin treated with sodium desoxycholate. J. Bacteriol. 92:1493-1509. 1966.-Endotoxins from three species of gram-negative bacteria were shown to be dissociated by the bile salt sodium deoxycholate (NaD) into nontoxic subunits with molecular weights of about 20,000. When the bile salt was removed by dialysis, the subunits reaggregated in an orderly manner to form a relatively uniform population of biologically active endotoxin particles with average molecular weights of 500,000 to 1,000,000. If a small amount of human plasma was added to the dissociated endotoxin before removal of the NaD, reassociation apparently did not occur and the preparation remained nonpyrogenic. However, the plasma protein could subsequently be removed from the endotoxin subunits, and reaggregation to the toxic form would then occur. The studies on the physical nature of endotoxin performed with biophysical solution techniques were supplemented and confirmed by direct examination of the endotoxin polymers by electron microscopy. The results of these studies were consonant with the theory that the biologically active endotoxic elements are composed of micellar aggregates of linear lipopolysaccharide subunits.

Endotoxin dissociated into subunits by sodium deoxycholate treatment exhibited diminished capacity to kill chick embryos, protect mice against the lethal effects of infection with Salmonella typhi, evoke hemorrhagic necrosis in skin inoculated with epinephrine, prepare for and provoke the dermal Shwartzman reaction, and induce pyrogenic tolerance. Surfactant-treated material which had been allowed to reaggregate displayed activity equivalent to that of untreated material. These findings were consistent with the working hypothesis that a macromolecular complex of critical size is required in order for endotoxin to elicit its characteristic effects in the host.