INFECTION AND IMMUNITY, Apr. 1986, p. 56-62 Vol. 52, No. 10019-9567/86/040056-07$02.00/0Copyright C) 1986, American Society for Microbiology

Characterization of Murine Monoclonal Antibodies toEscherichia coli J5

KAREN M. MINER, CHRISTINE L. MANYAK, ELAINE WILLIAMS, JESSE JACKSON, MARVIN JEWELL,MAUREEN T. GAMMON, CINDY EHRENFREUND, EDWARD HAYES, LYNN T. CALLAHAN III, HANS

ZWEERINK, AND NOLAN H. SIGAL*Merck Sharp and Dohme Research Laboratories, Rahway, New Jersey 07065

Received 24 November 1984/Accepted 13 December 1985

Twenty-eight independently derived monoclonal antibodies (MAb) directed against Escherichia coli J5endotoxin were produced and characterized. Each MAb exhibited a specific titer by both radioimmunoassayand passive hemagglutination assay. Most of the MAb were of the immunoglobulin G isotype; however, severalimmunoglobulin M antibodies and one immunoglobulin A antibody were produced. When characterized fortheir capacity to cross-react with purified endotoxin preparations from several gram-negative bacteria, 22 MAbexhibited no cross-reactivity; 6 demonstrated a limited capacity to cross-react with other endotoxin prepara-tions. When characterized for their capacity to react with the intact organism instead of the purified endotoxinthe pattern of cross-reactivity was quite different. Most of the MAb were able to react with Salmonellaminnesota Re595. Eighteen were able to react with E. coli O111:B4 (the parent strain of E. coli J5), 13 MAbreacted weakly with Pseudomonas aeruginosa, and 3 reacted weakly with Klebsiella pneumonia. The data implythat MAb generated against E. coli J5 endotoxin demonstrate greater cross-reactivity when assayed against thewhole bacterium than when assayed against the corresponding purified endotoxin. We were unable todemonstrate that any of the 28 MAb could passively protect mice against lethal endotoxin challenge.

Gram-negative bacteria have become the leading agents infatal bacterial infections in hospitals. The mortality rate canbe as high as 30 to 40% (9, 20, 29). Individuals most oftenaffected are patients whose immune system is suppressed,such as cancer patients receiving chemotherapy or radiationtherapy, organ transplant patients receiving immunosup-pressive drugs, and burn patients. Antibiotic therapy isrelatively ineffective because of the ability of these organ-isms to develop resistance to the antibiotics and may worsenthe condition by contributing to the release of endotoxin intothe blood stream. It is for these reasons that passive immu-nization with antisera, or perhaps monoclonal antibodies(MAb), is being viewed as an attractive alternative.The approach to the development of an antiserum is based

on the requirement that it be protective against a broadspectrum of gram-negative organisms. Antibodies to gram-negative bacteria are directed primarily against the lipopoly-saccharide (LPS), or outermost portion, of the cell wall. LPSis composed of three major parts; the 0 polysaccharide, thecore region, and lipid A. Antisera to a wild-type organismprimarily contains O-polysaccharide antibodies. Because thestructure of the 0 polysaccharide differs widely from strainto strain, an antiserum to this portion of the molecule isusually protective against the biological effects of the immu-nizing strain but it is not protective against a wide variety ofgram-negative bacteria (11). There is much less strain vari-ation in the core region of LPS and even less in the lipid Aportion of the molecule. Bacterial mutants have been de-rived which lack the 0 polysaccharide and expose variousportions of the core region and lipid A. The two mutantsmost commonly used for producing antibodies to core anti-gens are Escherichia coli J5 (8) and Salmonella minnesotaRe595 (14).Over the last 15 years, investigators in several laboratories

have reported that antisera to these mutant strains are

* Corresponding author.

effective in the prctection of animals against a wide varietyof experimentally induced bacteremias (2, 3, 5, 7, 16, 17, 27,28). Furthermore, Ziegler et al. (29) recently published theresults of a 7-year clinical study which demonstrated thathuman antisera, prepared by vaccinating healthy men withE. coli J5, could protect bacteremic patients from deathwhen compared with preimmune serum.Our interest in this area has centered around the possibil-

ity that a MAb directed against the common regions of LPSmolecules could be effective in the protection against gram-negative sepsis. A library of murine MAb directed against E.coli J5 have been produced and characterized for theircross-reactive and cross-protective capacities.

MATERIALS AND METHODS

Endotoxins. Purified endotoxins were obtained from ListBiological Laboratories (Campbell, Calif.) and contain <1%contaminating nucleic acids and proteins. According to thesupplier, the endotoxins were prepared according tomodifications of published methods (11, 19, 27, 30). Allpurified endotoxins were solubilized with triethylamine (1jig/ml).

Bacteria. The following bacteria were used: E. coli J5(D. A. Braude, University of California, San Diego); S.minnesota Re595 (J. Rudbach, University of Montana);Streptococcus pneumoniae S III (Robert Austran, Univer-sity of Pennsylvania); E. coli O111:B4 (O. Westphal, Max-Planck Institut); S. typhimurium (D. A. O'Brien, UniformedServices University of Health Sciences); Klebsiella pneu-moniae (E. Thiele, Merck Sharp and Dohme ResearchLaboratories); and Pseudomonas aeruginosa, Fisher-Devlinimmunotype 1 (Matthew Pollack, Uniformed Services Uni-versity of Health Sciences). Each strain was grown in 100 mlof tryptic soy broth overnight at 37°C with shaking. The cellswere harvested by centrifugation and washed three times inphosphate-buffered saline (pH 7.4). The cells were sus-pended in phosphate-buffered saline and allowed to boil for

56

on Decem

ber 2, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

MURINE MAb TO E. COLI J5 57

150 min over a Bunsen burner. The cells were then washedthree times in phosphate-buffered saline and diluted to 17 to20% transmittance as measured by a Spectronic 20 (Bausch& Lomb, Inc., Rochester, N.Y.) colorimeter. A concentra-tion of 1:10,000 thimerosal was added for storage.

Cell lines. Myelomas and hybridomas were maintained inCorning tissue culture flasks in Dulbecco modified Eaglemedium (GIBCO Laboratories, Grand Island, N.Y.) con-taining 10 to 15% fetal bovine serum (Hyclone, Logan,Utah), 1% nonessential amino acids (GIBCO), and 50 ,ug ofgentamicin (GIBCO) per ml. Cell cultures were grown at37°C in a humidified incubator containing 7.5% CO2 in air.

Immunizations. Inbred 8- to 10-week-old female BALB/cmice (Jackson Laboratories, Bar Harbor, Maine) were im-munized by one of two immunization schedules. Mice wereimmunized intraperitoneally with 0.25 ml of a heat-killed E.coli J5 suspension (see above) in complete Freund adjuvanton day 0 and boosted with the same dose in incompleteFreund adjuvant on days 14 and 21. Alternatively, mice wereimmunized subcutaneously with 0.25 ml of a heat-killed E.coli J5 suspension on days 0, 2, and 21. Sera designatedimmune mouse sera were collected 1 week after the finalboost in each case and titered by a direct binding radioim-munoassay (RIA). Sera designated normal mouse sera werecollected from unimmunized mice and used as a negativecontrol.New Zealand White rabbits (2 to 3 kg) were bled 1 week

prior to immunization to obtain preimmune serum. On day 0,rabbits were immunized (intradermally at 5 to 10 sites on theback) with 1.0 ml of a heat-killed E. coli J5 suspension incomplete Freund adjuvant. Rabbits were boosted 1 monthlater with E. coli J5 in incomplete Freund adjuvant, using thesame dose and regimen.

Fusion and generation of MAb. Mice were boosted intra-venously 3 days prior to hybridization with 0.1 ml of aheat-killed suspension of E. coli J5 (see above). Spleen cellswere fused with Sp2/0 murine myeloma cells (Human Ge-netic Mutant Cell Repository, Camden, N.J.) according topublished methods (24). Positive cells were cloned at leastonce by limiting dilution or in soft agar. Hybridoma asciteswere obtained by injecting pristane (2,6,10,14-tetrameth-ylpentadecane)-primed mice with 107 cells and asepticallyremoving the ascites fluid 7 to 10 days later.RIA. A direct binding RIA was used for the initial screen-

ing and characterization of anti-endotoxin antibody. Briefly,96-well microtiter plates (Dynatech Laboratories, Inc.,Alexandria, Va.) were coated with antigen (endotoxin at 1,ug/ml or a suspension of the heat-killed organism; see above)and allowed to stand overnight at 4°C. The plates werewashed and blocked with 10% agamma horse serum(GIBCO) followed by the addition of 100 RI of tissue culturesupernatant or ascites fluid containing MAb. The plates werelabeled with radioiodinated goat anti-mouse immunoglobulin(approximately 50,000 cpm/well). Samples were always as-sayed in duplicate. Background counts were determinedwith ascites fluid containing an unrelated MAb directedagainst hepatitis B surface antigen. RIA titers were deter-mined with 10-fold dilutions of ascites fluid and were re-corded as the dilution at which the counts per minute boundwere twice the background counts bound.

Competition RIA studies were performed by preincubat-ing the MAb with either the desired endotoxin or heat-killedorganisms at various concentrations for 4 h at 37°C beforeadding the mixture to an antigen-coated plate.

Passive hemagglutination. Anti-endotoxin activity of theMAb was also determined by passive hemagglutination,

using sheep erythrocytes coated with E. coli J5 endotoxin(6). Agglutination was facilitated with rabbit anti-mouseantisera which had been previo4sly absorbed with sheeperythrocytes. Hemagglutination titers were determined withserial twofold dilutions and werie recorded as the reciprocalof the endpoint dilution.

Isotyping. The immunoglobulin class was determined byenzyme-linked immunosorbent assay, using a modificationof the Mouse Immunoglobulin Subtype Kit (no. 100-036;Boehringer Mannheim Biochemicals, Indianapolis, Ind.).Ninety-six-well enzyme immunoassay plates (Costar, Cam-bridge, Mass.) were coated with E. coli J5 endotoxin (1p,g/ml) and allowed to stand overnight at 4°C. The rest of theassay was performed as directed, except ABTS (2,2'-azino-di-[3-ethyl-benzthiazoline] sulfonate) was used as the perox-idase substrate. Color was read at 415 nm.Immunoglobulin concentration. The immunoglobulin con-

centration was determined by a direct binding RIA asdescribed above. Rabbit anti-mouse kappa light chain(Litton Bionetics, Charleston, S.C.) at 10 ,ug/ml was used asthe coating antigen. Mouse myeloma protein MOPC195-yG2b [IgH(K)] (Litton Bionetics) was used as a standard.

Passive protection assays. Endotoxin toxicity assays wereperformed in 8- to 10-week-old BALB/c or CF-1 mice(Jackson Laboratories), using five mice per experimentalgroup. Each experiment consisted of five to seven endotoxindoses with twofold dilutions between doses. In some assays,mice were challenged directly with endotoxin, and in ptherassays the mice were sensitized with P. acnes (Formalin-killed P. acnes [Paris strain] in 0.9% NaCl injected intrave-nously at 0.7 mg of protein per mouse 7 days prior to LPSchallenge [25]), D-galactosamine (20 mg per mouse adminis-tered intraperitoneally at the same time as LPS [10]), oractinomycin D (15 ,ug per mouse administered intravenouslyat the same time as LPS [4]). In all assays the endotoxin usedwas derived from E. coli J5. The number of deaths weredetermined at 24 to 72 h after challenge. (Fifty percent lethaldoses (LD50) were calculated as described by Reed andMuench (23) from the number of survivors 72 h afterchallenge.

Passive protection by heat-inactivated rabbit serum orheat-inactivated ascites fluid containing MAb was measuredin the above assays. The antibody was injected eitherintraperitoneally or intravenously 1 h before challenge withendotoxin.

RESULTSCharacterization of MAb to E. coli J5 endotoxin. Twenty-

eight stable independently derived hybridomas secretingMAb to E. coli J5 endotoxin were derived from two fusions.The animals used for the fusions were immunized by one oftwo protocols described in Materials and Methods. Ascitesfluid induced by each of the cloned hybridomas was used asa source of MAb in the studies reported here. Most of thehybridomas produced IgG antibodies; however, we obtainedseveral IgM clones and one IgA clone (Table 1). We wereunable to isotype two of our MAb with the reagents avail-able; they may be IgE. The MAb titer in ascites fluid wasassessed by both RIA and passive hemagglutination, usingpurified E. coli J5 endotoxin as test antigen. Each of theascites exhibited a specific titer by both RIA and passivehemagglutination: however, the titers do not always corre-late with one another. This may be due to the way in whichthe antibodies agglutinate and precipitate antigen or the wayin which the antigen is presented in the two assays. Theconcentration of each MAb is also presented in Table 1.

VOL. 52, 1986

on Decem

ber 2, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

58 MINER ET AL.

TABLE 1. Characterization of MAb directed against E. coliendotoxina

Titer by: Concn of AbMAb Isotype RIA PHA (MglMl)b

79.4 IgG2a 107 256 1382.1 IgG2a 105 256 3

195.5 IgG2b 102 >1,024 2290.5 IgG2b 104 > 1,024 41066.14 NTc 103 512 11074.1 IgA 102 > 1,024 11114.3 IgG2a 102 256 21121.3 IgG2a 101 NDd 41224.5 IgGl 104 >1,024 11517.2 IgG2b ND > 1,024 81519.4 IgG2b 104 > 1,024 41628.6 NT 102 512 11659.4 IgM 101 >1,024 11676.31 IgGl 104 >1,024 11680.2 IgGl 104 >1,024 81682.1 IgM 102 > 1,024 21692.6 IgM l03 >1,024 41717.5 IgM 104 256 11727.1 IgG3 103 512 81730.1 IgG3 104 512 0.51732.2 IgG2a 108 256 41733.6 IgG2a 104 64 51742.6 IgG2a 102 512 11814.5 IgGl 104 > 1,024 211823.14 IgGl 104 1,024 11840.2 IgG3 101 32 11883.4 IgG2a 106 256 31884.17 IgGl 106 > 1,024 1

a Assays are described in Materials and Methods, using ascites fluid as asource of MAb.

b Normal ascites fluid contained approximately 0.1 ,ug of immunoglobulinper ml.

c NT, Not typable.d ND, Not determined.

Normal ascites fluid obtained as a pool of ascites frompristane-primed mice contained approximately 0.1 mg ofantibody per ml. That some MAb were present in a highconcentration in the ascites fluid, yet exhibited a relatively

A. mAb 1692.17

100C

0

._

.r_c

0II

60

40

20

103 104

Concentration of LPS (ng/ml)

B. mAb 1733.6

100

C

0

._

-c

0

80

60

40

20

Concentration of LPS (ng/mI)FIG. 1. Competition binding assay, using various concentrations

of endotoxin. Ascites fluid at a 1/10 dilution was used as a source ofMAb. Illustrated is (A) non-cross-reactive MAb 1692.17 and (B)cross-reactive MAb 1733.6. Data are expressed as percent inhibitionof binding in the absence of endotoxin. LPS used were: E. coli JS(0); E. coli 0111:B4 (0); E. coli 055:BS (a); E. coli 0127:B8 (O);E. coli K235 (A); E. coli 026:B (L); S. typhimurium (*); S.minnesota (0). The other LPS, S. minnesota ReS9S, Serratiamarcescens, Y. enterocolitica, V. cholerae, K. pneumoniae, and P.aeruginosa, were all negative.

TABLE 2. Cross-reactive binding capacity of various MAb, using LPS as antigen'Monoclonal/polyclonal antibody (cpm bound)'

LPS Non-cross-reactive MAb Cross-reactive MAbIMS NMS

79.4 82.1 1224.5 1733.6 1883.4 1121.3 1114.3

E. coli J5 4,180 3,653 3,075 3,487 3,375 2,182 3,391 3,688 584E. coli 0111:B4 85 86 15 882 904 321 388 3,905 413E. coli O55:BS 123 78 0 1,221 1,165 555 646 595 391E. coli 0127:B8 53 0 0 587 578 365 457 513 518E. coli K235 206 252 11 2,683 2,438 1,502 2,409 912 455E. coli 026:B6 0 0 0 822 718 567 783 1,548 531S. typhimurium 67 35 0 1,218 1,064 486 857 472 358S. minnesota 19 25 35 811 893 482 677 614 424S. minnesota ReS95 375 213 79 12 54 331 255 803 876Serratia marcescens 194 104 52 20 34 305 262 570 443Y. enterocolitica 125 108 92 34 45 253 212 742 442V. cholerae 143 1 15 0 5 197 216 646 400K. pneumoniae 127 0 0 0 0 46 0 744 465P. aeruginosa 77 40 0 0 0 275 219 591 383

a The RIA assay is described in Materials and Methods. Ascites fluid at a 1:10 dilution was used as a source of MAb.b Nonspecific binding of an irrelevant MAb is subtracted from the direct binding data. IMS, Immune mouse serum; NMS, Normal mouse serum.

INFECT. IMMUN.

on Decem

ber 2, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

MURINE MAb TO E. COLI JS 59

TABLE 3. Correlation of direct binding RIA data with competition RIA data, using endotoxins as antigenaCross-reactive MAbb Non-cross-reactive MAbb

LPS 1733.6 1883.4 1732.2 1121.3 1114.3 1717.5 1692.17 1066.4Direct Indirect Direct Indirect Direct Indirect Direct Indirect Direct Indirect Direct Indirect Direct Indirect Direct Indirect

E. coli J5 8,623 97 9,142 95 8,054 85 7,221 90 9,518 93 6,080 96 1,907 82 3,833 940111:B4 3,435 2 3,281 3 0 0 0 0 0 8 251 13 220 2 163 1055:BS 3,625 36 4,324 59 540 0 2,340 20 3,058 21 3,858 41 0 3 138 130127:B8 1,533 7 2,117 27 147 0 852 0 834 12 2,312 19 76 23 219 0K235 7,274 84 8,322 90 4,040 39 5,558 75 7,722 81 8,656 56 0 5 229 4026:B6 308 22 450 68 82 0 127 27 256 10 238 30 98 0 81 0

S. typhimurium 3,976 25 4,204 61 711 0 2,154 24 3,390 14 3,618 26 0 0 23 2S. minnesota 3,762 9 .3,646 42 547 0 1,591 6 2,870 6 2,784 11 0 0 0 0S. minnesotaReS9S 0 0 884 12 0 0 279 0 29 12 989 8 1,185 0 333 0

S. marcescens 0 0 0 8 0 0 0 0 0 9 0 0 0 0 117 5Y. enter- 0 0 0 0 0 0 0 0 0 7 0 5 0 0 59 1

ocoliticaV. cholerae 0 0 0 0 0 0 0 0 0 13 0 8 0 0 43 1K. pneumoniae 0 0 0 7 0 0 0 0 0 11 102 7 0 2 221 0P. aeruginosa 0 0 0 4 0 0 0 0 0 0 0 1 0 0 256 6

a The RIA assays are described in Materials and Methods. Data are expressed as counts per minute bound for the direct binding RIA and as percent inhibition(at 10 ,tg of endotoxin per ml) for the indirect or competition RIA data.

b Nonspecific binding of an irrelevant MAb is subtracted from the direct binding data.

low titer by RIA and vice versa, was probably due to the spectrum activity to most gram-negative bacteria. To assessrelative affinities of the different MAb. cross-reactivity, a direct binding RIA was used, with purified

Cross-reactive binding capacity of the various MAb. Our endotoxin preparations as the coating antigen. Of the 28rationale for generating a library ofMAb to purified E. coli J5 cloned hybridomas, 22 exhibited no cross-reactivity withendotoxin was to obtain one or several MAb with broad- any of the purified endotoxin preparations other than E. coli

TABLE 4. Binding activity of various MAb, using whole bacteria (WB) or endotoxins as antigenacpm bound

MAb E. coli J5 E. coli 0111:B4 K. pneumoniae S. minnesota Re595 P. aeruginosaWB LPS WB LPS WB LPS WB LPS WB LPS

79.4 7,086 7,490 1,144 0 241 0 6,046 582 1,306 082.1 7,193 8,049 1,241 0 599 0 6,091 800 1,496 0195.5 6,146 6,251 908 0 846 0 5,608 0 2,100 0290.12 5,835 6,447 703 0 347 0 5,186 0 896 01224.5 6,230 5,119 618 0 700 0 4,994 0 579 01517.2 4,345 6,078 55 0 0 0 4,520 0 353 01519.4 3,439 6,528 0 0 0 0 2,757 137 9 01628.6 6,135 8,634 1,919 0 482 0 5,687 42 136 01114.3 7,693 8,686 5,926 401 0 0 6,853 0 1,596 01121.3 3,851 4,730 2,703 0 0 0 4,496 0 1,137 01682.1 1,901 2,943 23 0 36 0 1,855 0 1,008 01676.31 2,784 4,374 0 0 0 0 1,440 0 335 01727.1 2,191 1,272 21 0 0 0 455 0 0 01717.5 4,802 4,739 4,650 991 0 0 5,678 0 869 01732.2 6,178 6,139 1,688 0 0 0 3,839 0 361 01742.6 6,408 7,613 896 0 0 0 3,945 0 15 01823.4 2,585 4,392 473 0 0 0 1,359 0 669 01066.14 2,128 3,709 664 373 0 227 240 99 4 2381074.1 522 2,235 303 316 0 86 46 508 465 251659.4 884 310 580 147 154 42 240 16 411 391680.4 1,568 3,450 204 190 0 111 617 0 312 961692.6 1,924 3,236 922 750 459 376 1,047 845 1,492 7781730.1 1,510 2,667 1,138 13 317 383 0 320 140 01733.6 5,812 8,144 4,986 4,064 367 303 4,685 669 1,678 5081883.4 5,058 7,472 4,821 4,012 271 511 4,340 731 1,123 7801840.2 1,870 2,917 240 152 0 140 1,310 110 390 01884.17 1,068 3,199 280 75 0 0 157 0 273 01814.5 1,465 3,512 587 79 134 0 641 0 475 0

a The direct binding RIA assay is described in Materials and Methods. Ascites fluid at a 1:10 dilution was used as a source of MAb. Nonspecific binding of an ir-relevant MAb is subtracted from the data.

VOL. 52, 1986

on Decem

ber 2, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

60 MINER ET AL.

A. mAb 1883.4

Dilution of Bacterial Suspension

B. mAb 79.4

100

c

0

'-._

._-c

0-

la-3

Dilution of Bacterial SuspensionFIG. 2. Competition RIA illustrating percent inhibition by vari-

ous concentrations of bacterial suspension, using (A) cross-reactiveMAb 1883.4 and (B) non-cross-reactive MAb 79.4. Bacterial sus-pensions were adjusted to similar initial concentrations by nephel-ometry. E. coli J5 was adsorbed to the microtiter plate at 1 ,ug/ml,and a bacterial suspension of E. coli JS (O), E. coli 0111:B4 (0), P.aeruginosa (O). K. pneumoniae (A), or S. minnesota Re595 (A) wasused to compete for reactivity with the respective MAb.

J5. Six MAb exhibited a limited capacity to cross-react. Thedata for a representative group of MAb are represented inTable 2. Immune mouse sera generated by immunizing micewith heat-killed E. coli J5 bacteria showed antibody activityto E. coli J5 endotoxin and E. coli 0111:B4 endotoxin (theparent strain of E. coli J5) but exhibited little cross-reactivitywith other purified endotoxin preparations. Normal mouse

serum exhibited little reactivity with any of the purifiedendotoxin preparations. MAb 1883.4 also showed specificbinding to purified lipid A in the RIA (data not shown), butthe other MAb bound to the J5 LPS but not lipid A.To confirm our direct binding RIA results, we assessed the

cross-reactive binding capacity of several MAb by a compe-tition RIA in which purified endotoxin at various concentra-tions was preincubated with MAb before adding the mixtureto an E. coli J5 endotoxin-coated RIA plate. The results of acompetition binding study, using two different MAb, arepresented in Fig. 1. Figure 1A represents those MAb whichexhibit no cross-reactivity and Fig. 1B represents thoseMAb which are cross-reactive. In all assays the homologousendotoxin was an effective inhibitor. A correlation of thedirect binding RIA data and the competition RIA data ispresented in Table 3. With a few exceptions (most notablythe reactivity of clones 1733.6 and 1883.4 with E. coliO111:B4 endotoxin), there was good correlation between thetwo assays. One possible explanation for the binding patternelicited by MAb 1733.6 and 1883.4 is that the physical natureof drying LPS onto plates exposes determinants not nor-mally exposed when LPS is in suspension.When cross-reactivity was assessed in a direct binding

assay with the whole bacteria instead of the purifiedendotoxin, the pattern of reactivity was quite different(Table 4). As expected, all of the MAb reacted with E. coliJ5. Most of the MAb reacted with S. minnesota Re595,whereas none exhibited a high binding activity to the purifiedendotoxin from S. minnesota Re595. Eighteen MAb reactedwith E. coli O111:B4, yet only two of these (1733.6 and1883.4) were able to react with the corresponding purifiedendotoxin. Thirteen of the MAb demonstrated a low bindingactivity to P. aeruginosa and three demonstrated low bind-ing to K. pneumonia. When the cross-reactivity was as-sessed by a competition RIA, using a whole bacterialsuspension instead of purified endotoxin, good correlationbetween the direct binding assay and the competition assaywas found. Figure 2 represents a typical dose response curveand Table 5 shows the correlation.

Passive protection studies. The protective potential of ourMAb in a lethal endotoxin challenge assay was assessed.Since the LD50 of normal mice to E. coli J5 endotoxin isapproximately 150 ,ug, it was necessary to sensitize the miceto demonstrate passive protection with immune rabbit anti-sera. Administration of actinomycin D at the time of LPSchallenge lowered the LD50 to 2 ng. When 0.2 ml of immunerabbit serum was given 1 to 2 h before challenge, there wasa sixfold increase in LD50, with no significant protectiondemonstrated with preimmune sera. All MAb were assayedwith these actinomycin D-sensitized animals. However,none of the MAb protected mice in this assay system.

TABLE 5. Correlation of direct binding RIA data with competition RIA data, using whole bacterium as antigenaMAb

Antigen 1883.4 1114.3 1717.5 1692.6 1733.6 79.4 1224.5 1682.1

Direct Indirect Direct Indirect Direct Indirect Direct Indirect Direct Indirect Direct Indirect Direct Indirect Direct Indirect

E. coli 35 5,058 87 7,693 81 4,802 88 1,924 68 5,812 87 7,086 90 6,230 87 1,901 83E. coli 0111:B4 4,821 78 5,926 72 4,650 53 922 12 4,986 76 1,144 7 618 0 23 25P. aeruginosa 1,123 30 1,596 53 869 8 1,492 13 1,678 7 1,306 2 579 0 1,008 20K. pneumoniae 271 3 0 1 0 12 459 3 367 1 241 0 700 0 36 13S. minnesota Re595 4,340 90 6,853 68 5,678 88 1,047 56 4,685 91 6,046 89 4,994 92 1,855 78

a The RIA assays are described in Materials and Methods. Data are expressed as counts per minute bound for the direct binding RIA and as percent inhibition(at 1:10 dilution of bacterial suspension) for the indirect or competition RIA.

C

0

.0

0-

INFECT. IMMUN.

on Decem

ber 2, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

MURINE MAb TO E. COLI J5 61

Several MAb were assessed more than once in this assaysystem and at no time did any show any significant levelprotection over the control. MAb to hepatitis B surfaceantigen, which at times caused a threefold increase in LD50,was used as the negative control in the passive protectiveexperiments.

DISCUSSIONThere has been an increasing interest in the potential

protective activity of antibodies to gram-negative organisms.Several early studies indicated that antisera against smoothbacteria can neutralize the toxicity of homologous endotoxin(1, 11, 22, 26) but not heterologous endotoxin (11). Becauseof the vast heterogeneity found among strains of gram-negative organisms, the only therapeutically useful anti-bodies would be ones that protected against a broad spec-trum of gram-negative organisms. During the last 15 years,therefore, an emphasis has been placed on characterizingantisera to rough mutants of gram-negative bacteria whichlack the heterogeneous 0 polysaccharide and expose themore highly conserved core and lipid A regions of the LPSmolecule. The reports of cross-reactive and protective anti-sera to rough mutants are conflicting, however. Severallaboratories have reported that antisera to these roughmutants are effective in the protection of animals (2, 3, 5, 7,16, 17, 27, 28) and humans (29) against a wide variety ofexperimentally induced bacteremias. Other reports indicatethat such antisera is not protective (11, 18, 21). Greisman etal. (11) have suggested that something other than antibody inserum might be protective because, in some cases, preim-mune rabbit serum is as effective as hyperimmune sera in theprotection of mice against infection. Now, with the devel-opment of hybridoma technology, it should be possible todetermine if antibodies are protective against gram-negativeinfection. In fact, MAb directed against gram-negative orga-nisms have been reported (6, 13, 19). Two of the reportscharacterized MAb directed against smooth bacteria with anintact 0 polysaccharide portion of LPS and protection wasobserved when mice were challenged with the homologousstrain (6, 13). Mutharia et al. (19) recently reported theproduction of cross-reactive MAb to E. coli J5 endotoxin.Protection studies were not described in the report.

In the present study, 28 murine MAb directed against E.coli J5 endotoxin were characterized for their ability tocross-react with other gram-negative organisms. When theMAb were assayed in a direct binding RIA with purifiedendotoxin as the coating antigen, 22 were unable to reactwith any of the endotoxin preparations tested except E. coliJ5. Six MAb demonstrated some capacity to cross-react withother endotoxin preparations, but the cross-reactivity waslimited primarily to other types of E. coli and some strains ofSalmonella spp. The direct binding RIA pattern was con-firmed by a competitive binding study. When the MAb werecharacterized for their ability to react with the heat-killedwhole bacterium, however, many more of the MAb demon-strated cross-reactivity. Most of the MAb reacted with S.minnesota Re595 which lacks both the 0 polysaccharide andthe inner core region of the LPS molecule. Furthermore, 18MAb reacted with E. coli O111:B4, the parent strain of E.coli J5. Thirteen MAb were able to react weakly with P.aeruginosa and there was much less binding activity to K.pneumoniae. The implication of this study is that MAbgenerated against E. coli J5 endotoxin demonstrate greatercross-reactivity when assayed against the whole bacteriumthan when assayed against the corresponding purifiedendotoxin preparation. One explanation for this is that the

cross-reactive antigenic determinant(s) on the LPS moleculeis somehow modified during the purification process but ispresent and accessible on the intact bacterium. Anotherpossibility is that there are common antigens on the surfaceof gram-negative bacteria which are not part of the LPSmolecule. A third possibility is that whole bacterial cellssynthesize incomplete LPS moities in small amounts whichcould account for the cross-reactivity observed. In addition,it is known that small amounts of R core are synthesized inthe S. minnesota Re595 mutant.We assessed the protective potential of the MAb in a lethal

LPS challenge assay. Because mice are relatively resistantto challenge with LPS, an effective method of sensitizing themice was established. We found that it was necessary tolower the LD50 to approximately 2 ng of E. coli J5 endotoxinbefore passive protection with rabbit antisera could bedemonstrated. All MAb were tested in this assay with thehighly sensitized animals and no significant protection wasever observed. We also attempted several lethal infectionassays with E. coli J5 but were unsuccessful in demonstrat-ing passive protection. E. coli J5 is a relatively avirulentorganism in mice (LD50 = 109 organisms); therefore, it isdifficult to assess the protective ability of the MAb in thisassay due to the magnitude of the bacterial challenge.

In conclusion, a library of MAb directed against E. coli J5LPS have been produced. Most of these antibodies exhibitedsome cross-reactive binding potential when assayed againstthe whole bacterium but were not cross-reactive whenassayed against the purified endotoxin. We were unable todemonstrate that these MAb were protective in mice. How-ever, protection is difficult to assess because of the highnatural resistance of these animals to gram-negative organ-isms.

LITERATURE CITED1. Berczi, I. 1967. Endotoxin neutralizing effect of antisera to

Escherichia coli endotoxin. Z. Immunitatsforsch. Allerg. Klin.Immunol. 132:303-307.

2. Braude, A. I., and H. Douglas. 1972. Passive immunizationagainst the local Schwartzman reaction. J. Immunol. 108:505-512.

3. Braude, A. I., H. Douglas, and C. E. Davis. 1973. Treatment andprevention of intravascular coagulation with antiserum toendotoxin. J. Infect. Dis. 128(Suppl.):S157-164.

4. Bruins, S. C., R. Stumacher, M. A. Johns, and W. R. McCabe.1977. Immunization with R mutants of Salmonella minnesota.III. Comparison of the protective effect of immunization withlipid A and the Re mutant. Infect. Immun. 17:16-20.

5. Chedid, L., M. Parant, F. Parant, and F. Boyer. 1968. Aproposed mechanism for natural immunity to enterobacterialpathogens. J. Immunol. 100:292-301.

6. Colwell, D. E., S. M. Michalek, D. E. Briles, E. Jirillo, and R.McGhee. 1984. Monoclonal antibodies to Salmonella lipopoly-saccharide: anti-O-polysaccharide antibodies protect C3H miceagainst challenge with virulent Salmonella typhimurium. J.Immunol. 133:950-957.

7. Davis, C. E., E. J. Ziegler, and K. Arnold. 1978. Neutralizationof meningococcal endotoxin by antibody to core glycolipid. J.Exp. Med. 147:1007-1017.

8. Elbein, A. D., and E. C. Heath. 1965. The biosynthesis of cellwall lipopolysaccharide in Escherichia coli. I. The biochemicalproperties of a uridine diphosphate galactose 4-epimeraselessmutant. J. Biol. Chem. 240:1919-1925.

9. Finland, M. 1970. Changing ecology of bacterial infections asrelated to antibacterial therapy. J. Infect. Dis. 122:419-431.

10. Galanos, C., M. A. Freudenberg, and W. Reutter. 1979.Galactosamine-induced sensitization to the lethal effects ofendotoxin. Proc. Natl. Acad. Sci. USA 76:5939-5943.

VOL. 52, 1986

on Decem

ber 2, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

62 MINER ET AL.

11. Greisman, S. E., E. J. Young, and B. DuBuy. 1973. Mechanismsof endotoxin tolerance. VIII. Specificity of serum transfer. J.Immunol. 111:1349-1360.

12. Johns, M., A. Skehill, and W. R. McCabe. 1983. Immunizationwith rough mutants of Salmonella minnesota. IV. Protection byantisera to 0 and rough antigens against endotoxin. J. Infect.Dis. 147:57-67.

13. Kirkland, T. N., and E. J. Ziegler. 1984. An immunoprotectivemonoclonal antibody to lipopolysaccharide. J. Immunol. 132:2590-2592.

14. Luderitz, O., C. Galanos, H. J. Risse, E. Ruschmann, S.Schlecht, G. Schmidt, H. Schulte-Holthausen, R. Wheat, 0.Westphal, and J. Schlosshardt. 1966. Structural relationships ofSalmonella 0 and R antigens. Ann. N.Y. Acad. Sci. 133:349-374.

15. Luderitz, O., A. M. Staub, and 0. Westphal. 1966. Immuno-chemistry of 0 and R antigens of Salmonella and relatedEnterobacteriacae. Bacteriol. Rev. 30:192-255.

16. Marks, M. I., E. J. Ziegler, H. Douglas, L. B. Corbeil, and A. I.Braude. 1982. Induction of immunity against Haemophilus in-fluenzae type B infection by Escherichia coli core lipopolysac-charide. J. Clin. Invest. 69:742-749.

17. McCabe, W. R. 1972. Immunization with R mutants of S.minnesota. I. Protection against challenge with heterologousgram-negative bacili. J. Immunol. 108:601-610.

18. Mullan, W. A., P. M. Newsome, P. G. Connington, G. H.Palmer, and M. E. Wilson. 1974. Protection against gram-negative infections with antiserum to lipid A from Salmonellaminnesota Re595. Infect. Immun. 10:1195-1201.

19. Mutharia, L. M., G. Crockford, W. C. Bogard, Jr., and R. E.Hancock. 1984. Monoclonal antibodies specific for Escherichiacoli J5 lipopolysaccharide: cross-reaction with other gram-negative bacterial species. Infect. Immun. 45:631-636.

20. Myerowitz, R. L., A. A. Medeiros, and T. F. O'Brien. 1971.

Recent experience with bacillemia due to gram-negative organ-isms. J. Infect. Dis. 124:239-246.

21. Ng, A. K., C. H. Cehn, C. M. Chang, and A. Nowotny. 1976.Relationship of structure to function in bacterial endotoxins:serologically cross-reactive components and their effect onprotection of mice against some gram-negative infections. J.Gen. Microbiol. 94:107-116.

22. Nowotny, A., R. Radvany, and N. L. Neale. 1965. Neutralizationof toxic bacterial 0-antigens with 0-antibodies while maintain-ing their stimulus on non-specific resistance. Life Sci. 4:1107-1114.

23. Reed, L. J., and H. A. Muench. 1939. A simple method ofestimating fifty percent end points. Am. J. Hyg. 27:493-497.

24. Shulman, M., C. D. Wilde, and G. Kohler. 1978. A better cellline for making hybridomas secreting specific antibodies. Na-ture (London) 276:269-270.

25. Suter, E., G. E. Ullmann, and R. G. Hoffman. 1958. Sensitivityof mice to endotoxins after vaccination with BCG (BacillusCalmette-Guerin). Proc. Soc. Exp. Biol. Med. 99:167-169.

26. Tate, W. J., H. Douglas, and A. I. Braude. 1966. Protectionagainst lethality of E. coli endotoxin with 0-antiserum. Ann.N.Y. Acad. Sci. 133:746-762.

27. Ziegler, E. J., H. Douglas, J. E. Sherman, C. E. Davis, and A. I.Braude. 1973. Treatment of E. coli and Klebsiella bacteremia inagranulocytic animals with antiserum to a UDP-Gal epimerase-deficient mutant. J. Immunol. 111:433-438.

28. Ziegler, E. J., J. A. McCutchan, H. Douglas, and A. I. Braude.1975. Prevention of lethal Pseudomonas bacteremia withepimerase-deficient E. coli antiserum. Trans. Assoc. Am. Phy-sicians 88:101-108.

29. Ziegler, E. J., J. A. McCutchan, J. Fierer, M. P. Glauser, J. C.Sadoff, H. Douglas, and A. I. Braude. 1982. Treatment ofgram-negative bacteremia and shock with human antiserum to amutant of Escherichia coli. N. Engl. J. Med. 307:1225-1230.

INFECT. IMMUN.

on Decem

ber 2, 2020 by guesthttp://iai.asm

.org/D

ownloaded from