Common Enterobacterial Antigen

Transcription

1 INFECTION AND IMMUNITY, Mar. 1973, p Vol. 7, No. 3 Copyright 1973 American Society for Microbiology Printed in U.S.A. Common Enterobacterial Antigen II. Effect of Immunization on Challenge with Heterologous Bacilli WILLIAM R. McCABE AND ANNA GREELY Departments of Medicine and Microbiology, University Hospital, Boston University School of Medicine, Boston, Massachusetts Rteceived for publication 8 September 1972 Studies were carried out evaluating the protective activity of immunization with common enterobacterial antigen (CA) against challenge with heterologous gram-negative bacilli. Active immunization of mice with Escherichia coli 0:14 elicited titers of antibody to CA of 1: 640 or greater but completely failed to enhance resistance to challenge by mouse virulent strains of Klebsiella pneumoniae or E. coli. Similarly, two lots, 324 and 422, of rabbit antisera to CA failed to afford passive protection to mice challenged with K. pneumoniae or E. coli. A third lot, 166, of rabbit antisera to CA did passively protect mice. The protective activity of antisera 166 was demonstrated to reflect its content of antibody to another crossreactive antigen, Re determinant, of gram-negative bacilli which has previously been shown to protect against infections with heterologous bacilli rather than any protective effect of antibody to CA. These studies failed to demonstrate any protective activity of antibody to CA against challenge with heterologous CA containiing gram-negative bacilli. Most studies of immunity to infection have been concerned primarily with type-specific immune mechanisms. Little attention has been directed toward investigation of any potential protective effects mediated by antibodies to shared, cross-reactive antigens. Historically, cross-reactive antigens have proven important in the development of practical immunizing procedures as evidenced by Jenner's classical demonstration that cowpox protected against small-pox infection. More recently, prior infection with influenza A has been shown to protect mice against infection with influenza A2 (4, 13). Similarly, studies by one of the authors have demonstrated that both active and passive immunization with a rough mutant, Re 595, of Salmonella minnesota afforded significant protection to mice against challenge with heterologous smooth gram-negative bacilli (9). The chemical similarity of the core region of the lipopolysaccharides of most gram-negative bacilli results in their sharing cross-reactive antigens (8). One of the cross-reactive antigens shared by Enterobacteriaceae is the common enterobacterial antigen (CA) first described by Kunin (5-7). These studies demonstrated that, although most enterobacteria contain CA, only a few straiins of Escherichia coli (0:14, 0:56, 0:124, and 0:144) were capable of inducing an antibody response to CA. Although other investigators have elicited an antibody response to CA using semipurified ethanol-soluble fractions from other gram-negative bacilli (14), highly purified CA prepared by both Kunin and in this laboratory have not proven to be immunogenic (5; M. Johns et al., J. Immunol., in press). Antibodies to CA have been demonstrated to opsonize heterologous gramnegative bacilli for phagocytosis (1). In addition, equivocal prolongation of survival of Salmonellachallenged mice after immunzation with CA has suggested that antibody to CA might have protective activity in experimental animals (3). These observations, plus previous demonstrations by the author of protective activity of antibody to other cross-reactive antigens (determinants of Re mutants) of gram-negative bacilli in both experimental animals and man (9, 11), prompted the present studies to determine whether immunization with CA protected experimental animals against challenge with heterologous gramnegative bacilli. MATERIALS AND METHODS Immunizing agents. E. coli 0:14 was supplied through the courtesy of Calvin Kunin. An additional strain of E. coli 0:14 and E. coli 0:56, 0:124, and 0:144 were provided from the Center for Disease Control by John Winn. Immunization of mice or rabbits with these strains of E. coli induces an antibody response to CA as well as the 0-specific antigen of each strain. 386

2 VOL. 7, 1973 IMMUNIZATION WITH CA 387 A strain of Pseudomonas aeruginosa isolated in the Bacteriology Laboratory, University Hospital, from a blood culture was used to provide a nonspecific endotoxin stimulus for control animals as previously described (9). Immunization of mice and rabbits with this strain of P. aeruginosa, which does not contain CA, did not elicit an antibody response to CA or the Re determinant of S. minnesota. Infecting agents. Two species of mouse virulent gram-negative bacilli were used as challenge agents. The Caroli strain of K. pneumoniae type II was obtained from Louis Chedid of the Pasteur Institute. The second challenge strain, a smooth, nontypable E. coli 107, had been isolated from a blood culture of a patient with bacteremia. The characteristics of the infections produced by these bacteria are described in an earlier publication (9). Both of these challenge strains were shown to contain CA (see below). Immunization of mice. Mice were immunized with E. coli 0:14, 0:56, 0:124, and 0:144 to induce antibodies to CA. Single-colony isolates of each strain were inoculated into 10 ml of brain heart infusion (BHI) broth. After overnight incubation at 37 C, this was spread onto BHI agar plates to prepare a lawn. These plates were incubated overnight and the bacterial growth was harvested in 10 ml of 0.9% saline. The bacteria were washed three times in 0.9% saline, resuspended in 10 ml of saline, and boiled for 2.5 h. This antigen preparation was diluted 1:10 with 0.9% saline before immunization (42 X 108 bacteria/ml). Twenty- to 22-g male, CF1 mice were injected intraperitoneally with 0.2 ml of the E. coli 0:14, 0:56, 0:124, and 0:144 antigen suspensions for 3 consecutive days of each week for 2 weeks and then once weekly for 2 weeks. Five days later, lots of 10 mice were bled from the retro-orbital plexus, the blood was pooled, the serum was separated, and antibody titers were determined by indirect hemagglutination. If antibody titers to CA were less than 1:640, weekly injections of antigen were continued until titers of antibody to CA of 1:640 or greater were attained. After final injection of antigen, at least 1 week was allowed to elapse before challenge to minimize any nonspecific protective effect of endotoxin. Immunization of rabbits. Antisera to CA was prepared by immunization of rabbits with E. coli 0:14, 0:56, 0:124, and 0:144 using methods similar to those described for the preparation of antigens for immunization of mice. The bacterial growth was harvested in 10 ml of 0.9% saline, washed three times in saline, resuspended in 10 ml of saline, and killed by the addition of 0.5% Formalin. This material was resuspended in saline to 80%o transmission at 550 nm in a Coleman spectrophotometer prior to rabbit immunization. Groups of three rabbits were immunized by intravenous injection of 0.5, 1.0, 2.0, 4.0, and 4.0 ml at 4- to 5-day intervals. Animals were bled for antibody determination 7 days after the last injection of antigen. If satisfactory titers of antibody had been attained, animals were anesthetized and exsanguinated. After separation of serum, Merthiolate was added to a concentration of 1:10,000 as a preservative, and the serum was stored at -70 C until used. Hemagglutination reaction for CA. Titrations of antibody to CA were performed by modification of the indirect hemagglutination technique of Neter (12) using erythrocytes coated with extracts of a bacillus previously shown to contain large quantities of CA, Proteus rettgeri 6572 (11). P. rettgeri 6572 was grown overnight at 37 C oni BHI agar and harvested in 10 ml of saline. The bacterial suspeusion was boiled for 2.5 h and centrifuged at 6,000 rpm at 4 C, and the supernatant was removed. One milliliter of 95% ethyl alcohol was added to the supernatant. A 1:10 dilution of this crude antigen preparation was incubated with an equal volume of a 2.5% suspeusion of human group O erythrocytes in phosphate-buffered saline (PBS) for 1 h at 37 C in a shaker water bath. The sensitized erythrocytes were washed three times in PBS and resuspended at a 0.5% concentration for immediate use in the hemagglutination test. Serum was serially diluted in PBS in test tubes (13 by 100 mm) at dilutions ranging from 1:10 to 1:163,840, and an equal volume of sensitized erythrocytes was added. Tubes were incubated for 1 h at 37 C, refrigerated overnight at 4 C, and read by pattern and for agglutination after shaking. Hemmagglutination inhibition. Hemagglutination-inhibition studies were performed using the methods described for indirect hemagglutination except that 0.2 ml of boiled extracts of Enterobacter aerogenee, another CA containing strain (11), was added to one series of antiserum dilutions while 0.2 ml of PBS was added to a duplicate set as controls. Antiserum and the inhibiting antigen extract were incubated for 1 h at 37 C prior to addition of sensitized erythrocyes. Antibody titers to CA were considered to be those titers in serum from mice or rabbits immunized with E. coli 0:14 which agglutinated erythrocytes coated with extracts of P. rettgeri and which was completely blocked by extracts of E. aerogenes. Confirmation that hemagglutination inhibition with E. aerogenes accurately reflected inhibition of antibodies to CA was obtained by demonstration of identical results when highly purified CA (M. Johns, et al., J. Immunol., in press) was used for inhibition. CA content of challenge strains. Content of CA of the two challenge strains, K. pneumoniae type II and E. coli 107, was determined by hemagglutination inhibition as described above. The maximal dilution of each organism capable of inhibiting agglutination of erythrocytes coated by P. rettgeri 6572 by 4 to 8 hemagglutinating units of E. coli 0:14 antiserum was considered to represent their CA content and was reported as hemagglutination-inhibiting units. 0-specific and Re antibody. Antibody titers to both challenge strains were determined in serum from immunized mice and rabbits by indirect

3 388 McCABE AND GREELY I NFECT. IMMUNITY hemagglutination of erythrocytes coated with extracts of each organism. Prior inhibition, with E. aerogenes extracts, was carried out as described above to distinguish type-specific antibody from antibody to CA. In addition, since previous investigations have demonstrated that antibodies to the determitlants of Re mutants protect against heterologous gram-negative bacilli, antibody titers to Re were measured in the serum of all immuniized animals by previously described techniques (9). Mercaptoethanol treatment. Treatment with 2-mercaptoethanol was used to distinguish 19s from 7s antibody to CA. A 0.05-ml amount of 1 M 2-mercaptoethanol was added to serial dilutions of serum and incubated at 37 C for 1 h. Sensitized erythrocytes were then added and incubated for 1 h. Concomitant titrations of each serum specimen, untreated with 2-mercaptoethanol, were run for comparison. Adsorption with Re determinants and CA. Adsorption of rabbit antiserum to CA was carried out using the Re 595 mutant of S. minnesota. Bacteria were grown overnight in meat extract broth, removed by centrifugation, washed three times in saline, and killed and dried with acetone. To prevent release of endotoxin and toxicity of the Readsorbed serum, dried bacteria were hydrolyzed in 1% acetic acid at 100 C for 2 h prior to adsorption of CA antiserum. Preliminary investigation revealed that 1 mg of purified CA per ml would remove all the CA antibody from antiserum 166. Six milliliters of 166 antiserum was added to 6 mg of CA dissolved in 0.2 ml of PBS. The mixture was held at room temperature for 1 h and then at 4 C for 3 days with daily shaking to resuspend the precipitate. The precipitate was then removed by centrifugation and the supernatant was tested for CA antibody content prior to injection into mice. Experimental infections. Mice were infected by intravenous injection of the challenge organism into the tail vein. For most experiments, animals were challenged with 100 mean lethal dose units (LDw) of either infecting organism. Maintenance of virulence of the two challenge strains and the characteristics of the experimental infections have been described previously (9). RESULTS Active immunization. Interperitoneal immunization with E. coli 0:14 induced rises in antibody titers to CA from less than 1:20 to 1:640 or greater in mice. In contrast, immunization with E. coli 0:56, 0:124, and 0:144 failed to induce antibody titers to CA in excess of 1:80 to 1:160 even when protracted courses of immunization were given, and studies of animals immunized with these strains were not pursued further. Antibody titers to CA showed only a slight increase ( -1 :80) within the first 3 weeks after immunization with E. coli 0:14. Titers reached levels of 1:320 to 1:640 during the 4th week and 1:1,280 to 1:2,560 in the 5th week of immunization. Mouse antibody to CA appeared to be composed of approximately equal portions of 7s and 19s antibody since treatment with 2-mercaptoethanol consistently produced only a 1-tube (twofold) reduction in antibody titer to CA. Studies of serum from mice with CA titers of 1: 640 or greater after immunization failed to demonstrate any rise in 0-specific antibody to either of the challenge strains or the determinants of Re mutants. Once titers of 1: 640 or greater to CA had been achieved, at least 1 week was allowed to elapse after the last injection of antigen before the immunized mnice and control animals were challenged with 100 LD10 units of K. pneumoniae (106) or E. coli 107 (109) to obviate any nonspecific protective effects of endotoxin. The results of these studies are illustrated in Table 1. Since no significant differences in fatality rates were observed between controls injected with saline or P. aeruginosa, both groups of controls were combined for analysis. Survival rates in animals immunized with CA prior to challenge with K pneumoniae were 23%, whereas 30% of control animals survived. Slightly higher survival rates (49%) were observed in CA-immunized mice than in controls challenged with E. coli 107 (31%), but this difference was not statistically significant TABLE 1. Effect of immunization with common antigen (Escherichia coli 0:14) on intravenous challenge with Klebsiella pneumoniae or E. coli 107 a Common antigen. b Number of survivors per number of animals challenged (percent survivors).

4 VOI,. 7, 1973 (X2 = 3.6; P > 0.05). Additional analysis also failed to demonstrate any relation between antibody titers and survival rates in mice with titers to CA ranging from 1:640 to 1:5,120. Animals with higher titers of antibody to CA did not have greater survival rates after challenge with either K. pneumoniae or E. coli than animals with lower titers. Since the dose of the two challenge strains (100 LD,o) which were used, represented a relatively stringent challenge, additional studies were undertaken to determine whether immunization with CA induced lesser degrees of protection. Table 2 demonstrates the number of bacilli of each of the challenge strains required to produce and LDo in controls, mice immunized with CA, and animals who had received type-specific immunization. The LD,o of K. pneumoniae in control animals was 2.6 X 10' bacilli which was identical with the number of bacilli (2.6 X 10') required to produce an LD,o in CA-immunized mice. In contrast, the number of K. pneumoniae required to produce an LD5o was increased by 1,000-fold in mice who were immunized against the challenge strain. Immunization with CA failed to increase the LD50 (1.3 X 104) in comparison with controls (1.1 X 104), whereas type-specific immunization induced at least a 10-fold increase in protection against E. coli 107. Passive immunization. The ability of passively transferred rabbit antiserum to CA to protect against challenge with. heterologous gramnegative bacilli was also evaluated. Antibody titers to CA in rabbits immunized with E. coli IMMUNIZATION WITH CA 0:56, 0:124, and 0:144 were considerably lower than those in rabbits immunized with E. coli 0:14, similar to the results observed in mice. Because of the low titers of CA in rabbits immunized with E. coli 0: 56, 0:124, and 0: 144, the protective activity of the antisera was not studied. Mice were injected intravenously with 0.3 ml of saline, normal rabbit serum, or serum from rabbits immunized with CA (E. coli 0:14) and challenged 60 minutes later by the injection of 10' K. pneumoniae or 10' E. coli 107 intravenously. The results observed with three different antisera (166, 324, 422), and the titers of antibody to CA and other antigens of these antisera are shown in Table 3. Titers of antibody to CA varied from 1:10,240 to 1:163,840 in the three sera studied. The demonstration that agglutination of erythrocytes coated with extracts of P. rettgeri 6572 was completely inhibited by extracts of an unrelated species, E. aerogenes, as well as purified CA, indicated that these were antibodies to CA. Antibody titers ranging from 1:160 to 1:5,120 to the challenge strains were also observed. However, the ability of extracts of E. aerogenes and purified CA to completely inhibit agglutination of erythrocytes sensitized with extracts of the challenge strains demonstrated that such antibody was directed against CA present in the two challenge strains rather than 0-specific antibody. Measurement of CA content in extracts of the two challenge strains revealed >1,000 hemagglutinationinhibiting units per ml of K. pneumoniae extracts and 320 units/ml of E. coli 107 extracts. One of the CA antisera, 166, also contained antibody TABLE 2. Comparison of type-specific immunization and immunization with commorn antigen (CA) Challenge strain and dose Immunizing agent Saline coi 0:14) Klebiela pneumoniae K. pneumoniae 2.0 X 106 0/10a 0/10 10/ X 10' 0/10 0/10 10/ X 104 4/10 4/10 10/ X 10' 6/10 6/10 10/ X 102 7/10 7/10 10/10 LD5ob 2.6 X 10' 2.6 X 103 >2.0 X 106 E. coli 107 (E. coli 107) 3.0 X 108 0/10 0/10 8/ X 10' 1/10 2/10 10/ X /10 10/10 10/10 LD&o 1.1 X X 107 >3.0 X 10l a Number of survivors 72 h after challenge per total number challenged. b Mean lethal dose. 389

5 390 McCABE AND GREELY INFECT. IMMUNITY TABLE 3. Effect of passive transfer of serum from rabbits immunized with common antigen (CA) on lethal Klebsiella pneumoniae and Escherschia coli infections of mice Antibody titers Determination To CA To challenge strain To Re Survivors/no. of animals Unni- Inhibited Uni Inhibited challenged.uninhib- b by E. itd ie aerogenes aerogenes (K. pneumioniae challenge) Controls. Saline 15/47 (32%) NRSb < 1: 20 < 1: 20 < 1: 20 23/69 (33%) Serum from rabbits immunized with CA 166 1:10,240 < 1:20 1:640 < 1:20 1:160 64/69 (81%) 324 1:40,960 <1:20 1:150 <1:20 <1:20 19/37 (51%) 422 1:163,840 < 1:20 1:1280 <1:20 <1:20 3/30 (10%) (E. coli challenge) Controls Saline 4/40 (10%) NRS <1:20 <1:20 4/40 (10%) Serum from rabbits immunized with CA 166 1:10,240 < 1:20 1:320 < 1:20 1:160 27/30 (90%) 324 1:40,960 < 1:20 1:160 < 1:20 < 1:20 2/40 (05%) 422 1:163,840 < 1:20 1:5,120 < 1: 20 < 1:20 0/40 (00%) a Enterobacter aerogenes. b NRS = Nonimmune rabbit serum. against determinants of Re mutants. Antibody to Re was not inhibited by extracts containing CA. In contrast to mouse antiserum to CA, rabbit antiserum to CA was composed primarily of 198 antibody. Treatment of rabbit antiserum with 2-mercaptoethanol produced a 32- to 256- fold reduction in antibody titer to CA. Studies were carried out to measure persistence of passively injected rabbit antiserum to CA. Serum obtained from four mice, 1 h after passive transfer of 0.3 ml of 324 antiserum, revealed that all had titers of antibody to CA of 1:2,560. Titers of 1:1,280 and 1:320 were present 6 h and 24 h, respectively, after passive transfer of rabbit antiserum 324. Survival rates in mice passively immunized by intravenous injection of 0.3 ml of 324 or 422 antiserum to CA and subsequently challenged with K. pneumoniae or E. coli were not significantly greater than among controls given saline or normal rabbit serum (Table 3). In contrast, survival rates in mice given antiserum 166 and challenged with both K. pneumoniae (81%) and E. coli (90%) were significantly greater (X2 = 70; P < 0.001) than in control animals (10-33%). The demonstration that only one of the three antisera to CA afforded significant protection against challenge with heterologous gram-negative bacilli and that this antiserum contained the lowest titers of antibody to CA suggested that antibody to CA might not be responsible for the protection observed. Since previous investigations have demonstrated that antibody to Re protects against challenge with heterologous bacilli and since only CA antiserum 166 contained antibodies to Re (Table 3), this possibility was evaluated further by determination of the protective activity of 166 antiserum before and after adsorption with Re. Antiserum 166 was adsorbed overnight at 4 C with 100 mg of dead Res59 mutant (previously demonstrated not to contain CA) of S. minnesota per ml of serum. Such adsorption reduced antibody titers to Re from 1:160 to less than 1:20 but did not affect antibody titers to CA. After centrifugation for removal of bacteria, 0.3 ml of adsorbed or unadsorbed 166 antiserum was injected intravenously into each animal of groups of 40 mice 1 h prior to challenge with K. pneumoniae or E. coli. The results of these studies are summarized in Table 4. Although unadsorbed CA antiserum 166 afforded significant protection against challenge

6 VoL. 7, 1973 IMMUNIZATION WITH CA 391 TABLE 4. Effect of passive transfer of rabbit antiserum to common antigen (CA; 166) before and after adsorption with Re Passive transfer of Challenge strain No. of survivors/no. of animals challenged Saline Escherichia coli 107 3/20 NRSG E. coli 107 2/20 CA antiserum 166 (unadsorbed) E. coli /201 X2 = 17; P <0.001 CA antiserum 166 (adsorbed E. coli 107 2/20' with Re) Saline Klebsiella pneumoniae 4/20 NRSa K. pneumoniae 5/20 CA antiserum 166 (unadsorbed) K. pneumoniae 14/201 X2 = 12; P <0.001 CA antiserum 166 (adsorbed K. pneumoniae 3/20 with Re) a Nonimmune rabbit serum. with both K. pneumoniae and E. coli, this protective activity was obliterated by adsorption with Re. Thus, the protective activity of CA antiserum 166 appeared to reflect its content only of antibody to Re rather than antibody to CA Ȧdditional studies were carried out with antiserum 166 after adsorption with purified CA. These studies were limited by the availability of only a small amount of 166 serum remaining after previous studies. Adsorption with purified CA reduced antibody titers from 1:10,240 to less than 1:20 but failed to inhibit its protective activity. Survival rates, 5 of 10 (50%), were signiificantly greater (X2 = 4.03; P < 0.05) in mice injected with 0.3 ml of 166 antiserum adsorbed with CA prior to challenge. In contrast, only 3 of 20 (15%) mice injected with normal rabbit serum survived challenge with E. coli 107. Thus, the protective activity of CA antiserum 166 failed to reflect its content of antibody to CA and reflected, instead, the antibody activity to another crossreacting antigen, determinants of Re mutants, of gram-negative bacilli. DISCUSSION The failure to demonstrate any protective effect of active or passive immunization with CA against lethal infections with either K. pneumoniae or E. coli contrasts with earlier studies by the authors which have shown that immunization with another cross-reactive antigen, determinants of Re mutants, afforded significant protection against challenge with these same strains. The present investigations provided additional confirmation of the protective activity of antibody to Re. Only one of the three rabbit antisera to CA, 166, protected mice against challenge with K. pneumoniae and E. coli, but this antiserum was also the only one of the three which contained antibody against Re. Adsorption of antiserum 166 with purified CA failed to remove its protective activity, whereas adsorption of the same antiserum with killed Re mutants did not affect antibody titers to CA but removed antibody reactive against Re and abolished the protective activity of this serum. This observation clearly emphasizes the problems of interpretation of protective effects observed after immunization with whole bacilli or impure antigen preparations which may produce antibodies against a variety of determinants. If prior studies had not demonstrated that the determinants of Re mutants are shared by most gram-negative bacilli and that antibody to Re protected against challenge with heterologous smooth gram-negative bacilli, the protection observed with the 166 CA antiserum might have been attributed to antibody to CA. Similarly, previous reports of possible protective activity induced by immunization with CA (2, 3) might also have reflected antibody directed against antigens other than CA. The failure to demonstrate protection against challenge with heterologous bacilli in mice actively and passively immunized with CA is consistent with observations in man. Studies of 206 episodes of gram-negative bacteremia demonstrated that bacteremia often developed despite the presence of high titers of antibody to CA and that high titers of CA did not ameliorate the severity of bacteremia. However, despite the lack of protective activity of antibody to CA in man, antibody to another shared, cross-reactive antigen (Re) did appear to be associated with lesser severity and enhanced survival in human gramnegative bacteremia analogous to observations in animals (10).

7 392 McCABE AND GREELY INFECT. IMMUNITY ACKNOWLEDGMENTS These studies were supported by U.S. Public Health Service research grants AI , AI , AI-11,116-01, and training grant 5-TO1-AI from the National Institute of Allergy and Infectious Diseases and grant 5 P01-HE from The National Heart Institute. LITERATURE CITED 1. Domingue, G. J., and E. Neter Opsonizing and bacterial activities of antibodies against common antigen of Enterobacteriaceae. J. Bacteriol. 91: Domingue, G., A. Salki, C. Rountree, and W. Little Prevention of experimental hematogenous and retrograde pyelonephritis by antibodies against enterobacterial common antigen. Infect. Immunity 2: Gorzynski, E. A., J. L. Ambrus, and E. Neter Effect of common enterobacterial antiserum on experimental Salmonella typhimurium infection of mice. Proc. Soc. Exp. Biol. Med. 137: Henle, W., and F. S. Lief The broadening of antibody spectra following multiple exposures to influenza viruses. Amer. Rev. Resp. Dis. 88: Kunin, C. M Separation, characterization and biological significance of a common antigen in Enterobacteriaceae. J. Exp. Med. 118: Kunin, C. M., and M. V. Beard Serological studies of 0 antigens of E8cherichia coli by means of the hemagglutination tests. J. Bacteriol. 85: Kunin, C. M., M. V. Beard, and N. E. Halmagyi Evidence for a common hapten associated with endotoxin fractions of E. coli and other Enterobacteriaceae. Proc. Soc. Exp. Biol. Med. 111: Ltlderitz, O., A. M. Staub, and 0. Westphal Immunochemistry of 0 and R antigens of SalmoneUa and related Enterobacteriaceae. Bacteriol. Rev. 30: McCabe, W. R Immunization with R mutants of S. minne8ota. I. Protection against challenge with heterologous gram-negative bacilli. J. Immunol. 108: McCabe, W. R., M. Johns, and T. DiGenio Common enterobacterial antigen. III. Initial titers and antibody response in bacteremia caused by gram-negative bacilli. Infect. Immunity 7: McCabe, W. R., B. E. Kreger, and M. Johns Type specific and cross-reactive antibodies in gram-negative bacteremia. N. Engl. J. Med. 287: Neter, E., L. F. Bertram, D. A. Zak, M. R. Murdock, and C. E. Arbesman Studies on hemagglutination and hemolysis by Escherichia coli antisera. J. Exp. Med. 96: Schulman, J. L., and E. D. Kilbourne Induction of partial specific heterotypic immunity in mice by a single infection with Influenza A virus. J. Bacteriol. 89: Suzuki, T., E. A. Gorzynski, and E. Neter Separation by ethanol of common and somatic antigens of Enterobacteriaceae. J. Bacteriol. 88: Downloaded from on July 22, 2018 by guest