Enterobacterial Common Antigen · minant structure of the common antigen is probably covalently linked to LPS. The "hap-tenic" formofthe commonantigen is probably present in all common

BAcTrhaxowIcAL Rzviws, Sept. 1976, p. 591-632Copyright 0 1976 American Society for Microbiology

Vol. 40, No.3Printed in U.S.A.

Enterobacterial Common AntigenP. HEiENA MAKELA* mn H. MAYER

Central Public Health Laboratory, Helsinki, Finland,* and Max-Planck-Institut fur Immunbiologie,Freiburg, i. Br., Germany

INTRODUCTION .............. ................................ 591DISCOVERY AND DEFINITIONS ............................................. 591SEROLOGICAL METHODS.............................................. 592Hemagglutination (HA), Hemolysis, and Hemagglutination Inhibition (HAI) ... 592Other Serological Methods.............................................. 593

CHEMISTRY .............................................. 594Studies on Immunogenic Strains ............................................. 594Studies on Nonimmunogenic Strains ......................................... 595Chemical Identity of ECA.............................................. 600

GENETIC DETERMINATION OF ECA ......................................... 602The rMe Genes .............................................. 602Role of the rib Region ..................... ........................ 604Mutants of a New, "rif," Type ............................................. 605Genetics of ECA in Immunogenic Strains .................. .................. 606Conclusions .............................................. 606

IMMUNOGENICITY ................. ............................. 607Immunogenic Strains ............................................ 607Interference by LPS ............... ............................. 610Importance of the Form of Aggregation ...................................... 611Immunological Tolerance?............................................. 613

ECA IN THE BACTERIAL CELL ............................................ 614CLINICAL IMPLICATIONS............................................. 614Immediate Biological Effects of ECA ........................................ 614Is ECA a Virulence Factor? ............................................ 615Prevalence of Antibodies to ECA............................................. 615ECA and anti-ECA Antibodies in Relation to Disease .......... ................ 616Anti-ECA in Experimental Infection ......................................... 619Antigenic Similarities Between ECA and Animal Tissues ........ .............. 620

OTHER CROSS-REACTIVITIES AND RELATED ANTIGENS ...... ............ 620Various "Common" Antigens ................................................ 620ManNUA-Containing Bacterial Antigens ................ ..................... 622

DISTRIBUTION OF ECA: TAXONOMIC IMPLICATIONS ....... ............... 622SUMMARY ............ ..................................... 623LITERATURE CITED ................ a 624

INTRODUCTIONThe 0, K, and H antigens ofEnterobacteria-

ceae and several other gram-negative bacteriahave been serologically recognized since earlyin this century (93, 205). Their location in thebacterial cell, serological properties, biologicaleffects, and clinical implications have been ex-tensively characterized; more recently theirchemical structure, biosynthesis, and geneticdetermination have been the subject of muchresearch (82, 84, 88, 89, 109, 129, 150, 163, 190).It is, then, rather surprising that another anti-genic component, the enterobacterial "commonantigen" or "Kunin antigen," was not discov-ered until 1962 (102) and has so far receivedrather limited attention. The diagnostic andeventual prophylactic possibilities of an anti-gen shared by many common pathogens areobvious. This was promptly recognized by

Neter and his associates, who have since de-voted much of their research to various aspectsof the "common antigen" (135-149; 208-229).General interest in it may still be hampered bylack ofknowledge of its chemical nature and byapparent confusion about its identity. Recentstudies have uncovered new facts that mayhelp to clarify some points of confusion and atthe same time define subjects requiring furtherwork. We hope that we can here describe whatis hitherto known of the "common antigen" in away that will stimulate further research.

DISCOVERY AND DEFINITIONSThe enterobacterial common antigen was

found by Kunin in the course of studies of uri-nary tract infections, of Escherichia coli typesresponsible for them, and of the presence ofanti-E. coli antibodies in the patients (99, 101,

591

on May 21, 2020 by guest

http://mm

br.asm.org/

Dow

nloaded from

http://mmbr.asm.org/

592 MAKELA AND MAYER

102). He used indirect hemagglutination (asapplied for use with enterobacterial[lipo]polysaccharides by Neter et al. [139, 144])for detection of E. coli anti-O antibodies andwas led to investigate the extent of serologicalcross-reactions between the various E. coli 0serotypes.When thus screening extracts of 145 E. coli

strains representing almost all the known 0serotypes for their ability to sensitize erythro-cytes (RBC) to agglutination not only by homol-ogous but also by heterologous antisera, it wasfound (102) that, unexpectedly, almost all theseextracts were capable of such sensitizationwhen certain heterologous antisera were usedfor the hemagglutination. The best serum forthis hemagglutination was anti-E. coli 014,whereas anti-E. coli 056 and anti-E. coli 0144showed less reaction. The cross-reactive prop-erty of the anti-014 serum could be removed byabsorption with extracts of any E. coli strain,whereas the absorbed serum retained its capac-ity to agglutinate RBC coated with the homolo-gous 014 extract. Thus the anti-014 serumseemed to have antibodies to an antigen com-mon to all E. coli strains and a different sort ofantibody specific for the 014-type lipopolysac-charide (LPS) not shared by other strains. Thatmost other sera did not contain the first kind ofantibodies, namely those reactive with the com-mon antigen, whereas all anti-014 sera pre-pared by different immunization schemes did,was quite surprising. It could be accounted forby assuming that the common antigen in mostbacteria is present in a haptenic form, whereasin the 014-type strains (and probably in someothers as well) it is in an immunogenic form.This apparent existence of the common anti-

gen in two different forms has caused muchconfusion in further research. The chemicalbasis of these two forms has only recentlystarted to be understood (see below, "Immuno-genicity"). In both forms, the immunodetermi-nant part is apparently identical, but may beassociated with different carrier structures.The "immunogenic" form of the common

antigen is present in only few bacterial strains,of which E. coli 014 is an example. It nowseems -that in this form the immunodeter-minant structure of the common antigen isprobably covalently linked to LPS. The "hap-tenic" form of the common antigen is probablypresent in all common antigen-positive bac-teria, of which most are "non-immunogenic,"i.e., do not elicit anti-common antigen anti-bodies when injected as whole heat-killedbacterial suspensions. This form is character-ized by its ability to sensitize RBC for hemag-glutination. Our current working hypothesis isthat in this haptenic form the immunodeter-

BACTERIOL. REV.

minant structure is linked to a "carrier"molecule, which is not LPS. The carrier may bemostly responsible for the ability to coat RBC.(The haptenic form of the common antigen mayunder certain conditions also elicit an anti-common antigen response, and these conditionswill be discussed later on in the section on"Immunogenicity".)This common antigen was found, besides in

E. coli, in several other members of the familyEnterobacteriaceae (102). It was not present inother gram-negative bacteria such as Pseudom-onas andBrucella or in any gram-positive orga-nisms, and later studies have confirmed it as aproperty typical of the Enterobacteriaceae (seebelow, "Distribution of ECA"). We should thusdefine the enterobacterial common antigen inaccordance with the original description by Ku-nin as a cross-reactive antigen detectable inseveral species of enteric bacteria by indirecthemagglutination when using anti-E. coli 014sera. Other sera and other detecting methodscan be used, but in each case the identity oftheantigen detected must be confirmed by compar-ison with the original definition. We would alsolike to suggest the term ECA for the enterobac-terial common antigen to indicate its restric-tion to enteric bacteria. So far the abbreviationCA has been generally used, but it is moreambiguous and can be confused with Ca, forcancer.

SEROLOGICAL METHODSHemagglutination (HA), Hemolysis, and

Hemagglutination Inhibition (HAI)Fresh human (blood group 0) or sheep RBC

have been used mostly, but probably other RBCand formalin-preserved cells work as well (22,139, 216). These are coated by the sensitizingECA preparations and thus become reactivewith anti-ECA sera. This reaction can be visu-alized as agglutination of the RBC (HA). It canalso be seen as lysis of the coated RBC in thepresence of complement, but, peculiarly, thereare restrictions as to the kind of cells used-human RBC are not lysed in this test (144, 216).Both reactions can in turn be inhibited by freeECA.The usual sensitizing antigen is a superna-

tant of cultures grown in broth or syntheticmedia (61, 102), or suspended in saline or bufferfrom agar surfaces. The latter are more reac-tive if heated at 1000C (usually for 1 h) -pre-sumably to release more antigen to the super-natant- before removing the bacteria by cen-trifugation (216). LPS prepared by the hotphenol-water extraction method or by trichloro-acetic acid extraction as the Boivin antigen(109) is also reactive (102), whereas LPS ex-


http://mm

br.asm.org/

Dow

nloaded from


ENTEROBACTERIAL COMMON ANTIGEN 593

tracted with the phenol-chloroform-petroleumether method is not (126). Brodhage (20-22)used urea extracts ofbacteria in a similar assayand detected a common antigen-probablyECA, although the exact correlation has notbeen studied (see also "Other cross-reactivitiesand related antigens").The sensitization step in the standard assay

takes place at 37"C (for 30 to 60 min), usuallywith a 2.5% suspension of washed RBC. Afterexcess antigen is washed away, the sensitizedcells are suspended to a concentration of ap-proximately 2.5% in phosphate-buffered salineor other buffer. The agglutination proceeds at37TC for 30 min and is preferably read by thenaked eye after a light centrifugation (e.g., 3min at 1,300 x g) (216). Hemolysis is a slightlymore sensitive assay than HA (144, 216) whensensitized sheep RBC, guinea pig complement,and rabbit anti-ECA sera are used. It is de-pendent on the use of correct complementsource and sera. A Jerne plaque modification ofthe hemolysis assay has also been developed(39). Hemolysis of sensitized cells also occurs invivo in immunized animals (136, 193), a phe-nomenon that could be important in naturalinfection.The classical antiserum for ECA detection is

rabbit anti-E. coli 014 serum obtained by var-ious immunization schedules using heated(100TC for 1 to 2 h) suspensions of whole bacte-ria as immunogen. Similar antisera are ob-tained using certain rough (R) Shigella or E.coli mutants or a semipurified ECA-"ethanol-soluble fraction"-isolated from any ECA-con-taining bacteria as immunogens (192) (see "in-munogenicity"). All these sera seem to measureECA equally well. When used with a heterolo-gous source ofECA, they also seem to be equalin specificity.When measuring anti-ECA antibodies, most

studies have used ECA from Salmonella or E.coli strains (014, 0111, 07, etc.) for sensitizingthe RBC with essentially the same results, andit was expected that any ECA-containingstrain would do as well. However, it now seemsprobable that extracts of the Proteus rettgeristrain 6572 used by McCabe et al. (110, 112, 113,115) have predominantly another antigen (see"Chemical identity of ECA") also capable ofsensitizing RBC for agglutination by anti-E.coli 014 serum (W. McCabe, personal com-munication). This other antigen may explainsome differences in the findings of McCabe etal. as compared with most other studies.Therefore, it should be determined as soon aspossible to what extent this other antigen maybe responsible for some reactions ascribed toECA. HAI, the ability of ECA to inhibit HA or

hemolysis, is a good measure of its antigenic(immunodeterminant) part and should be thebest way to look for the presence or absence ofECA in unknown materials. The inhibitingcapacity is not affected by the ability orinability of ECA to sensitize RBC, to elicit animmune response, or to be precipitated. Thisinhibitory capacity has been successfully usedin following ECA during purification steps (85,100, 192). Until chemical identification of ECAhas progressed further, HAI remains the prin-cipal method of quantitating ECA.

Other Serological MethodsImmunoprecipitation would be especially

useful for monitoring an antigen during chemi-cal purification and as a final test ofpurity andidentity. It was therefore disappointing that inthe initial studies ofECA the results were neg-ative: no precipitation was seen in test tube oragar-gel diffusion (100-102). More recently ithas become evident that the failure to detectprecipitation is dependent on the antiserumused- some sera contain precipitating antibod-ies, others do not. The first negative results inprecipitation had a more trivial cause: the anti-sera used were standard sera for routine agglu-tination tests and contained 50% glycerol. Thismade them very hygroscopic, and consequentlydiffusion in agar took place only toward theantiserum well, not away from it (85).Mayer and Schmidt found two precipitating

sera among several anti-Ri-type sera tested(126). Both double-diffusion precipitation andimmunoelectrophoresis have been used (85,126, 211). The precipitation method is now avaluable help in following purification of ECA,as first shown by Johns et al. (85). It is ob-viously necessary to select suitable antisera forthis purpose, and negative findings must beinterpreted with caution; e.g., a change in thesize of the ECA can alter its precipitabilitywhile the immunodeterminant remains intact.Specificity of the reaction must be ascertainedby HAI.An immunofluorescence (IF) test for detect-

ing ECA was developed because of clinical in-terest, in hopes of detecting the antigen in tis-sue specimens. Using fluorescein-labeled goatanti-rabbit gamma globulin and anti-E. coli014 rabbit serum, Aoki et al. (8) obtainedstrong fluorescence in smears of all ECA-posi-tive bacteria and weak or no fluorescence withECA-negative bacteria. The identity of thereactants in IF is based on these cross-reactionsonly and cannot therefore be considered as cer-tain as recognition of ECA on the basis of HA.

Bacterial agglutination is not a usablemethod, since anti-ECA antibodies do not ag-

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from



glutinate intact bacteria (102). The reason forthis is not known. Intact bacteria seem able tobind anti-ECA (as suggested by IF and by theirability to remove anti-ECA from immune sera[219] or to become opsonized [38]). Anti-ECAantibodies are mostly immunoglobulin M (224)and would therefore be expected to be capableof causing agglutination.

Agglutination of coated latex particles doesnot take place either. Latex particles adsorbboth ECA and LPS from crude bacterial ex-tracts (100, 219), and the presence ofboth can bedemonstrated by the ability of the coated parti-cles to remove the corresponding antibodiesfrom immune sera (37, 146) and to be opsonizedfor phagocytosis (38). However, the latex parti-cles become agglutinable by anti-LPS sera only(217).Anti-ECA antibodies obtained by immuniz-

ing rabbits with either E. coli 014 or variousimmunogenic preparations ofSalmonella or E.coli 0111 were bactericidal (in a heterologouscomplement-mediated system) to E. coli 014but not to E. coli 01, 055, or 0111 nor toShigella dysenteriae or Salmonella species (38,101). Accepting the notion that in E. coli 014the ECA immunodeterminant is covalentlylinked to LPS, these findings can be interpretedto mean that the killing ofE. coli 014 is causedby a reaction ofanti-ECA antibodies with LPS,whereas such a reaction between the same anti-bodies and ECA (non-LPS linked) does not leadto killing. Unfortunately, we do not know therequirements for antibody + complement-me-diated killing, although the topological locationofthe antigen in the bacterial cell is believed tobe important (160). The reaction of ECA withits antibody can activate complement as evi-denced by the hemolysis reaction, but whetherthis reaction takes place on the intact bacteriahas not been tested.

Several anti-ECA sera could be shown to op-sonize ECA-positiveE. coli and Salmonella butnot ECA-negative Pseudomonas for phagocyto-sis by rabbit peritoneal exudate cells; the ho-mologous system E. coli 014-anti-E. coli 014(containing both anti-0 and anti-ECA antibod-ies) was only slightly more effective than aheterologous (ECA-dependent) system basedsolely on anti-ECA antibodies (38). The opso-nizing capacity of the antisera could be re-moved by absorption with ECA-coated RBC(38). In a similar experiment, ECA-coated latexparticles were opsonized for phagocytosis byanti-ECA serum but not by normal serum. FreeLPS interfered with the reactivity of ECA, aphenomenon that will be discussed in detail inthe section on immunogenicity.

BACTERIOL. REV.

CHEMISTRYIn spite of several recent studies on the im-

munochemical characterization of ECA, itschemical identity is still far from clear. Eachgroup of workers has developed its own extrac-tion and purification procedures, and they oftenuse different strains as starting material. Thepurified products have not yet been sufficientlycompared. Table 1 summarizes the isolationprocedures used and gives a summary on theproperties and characteristics of the final prod-ucts.

It seems reasonable to discuss the work ofKunin (100) and of Hammarstrom et al. (70)first, since both studied, at least predomi-nantly, the immunogenic strainE. coli 014:K7.We will then discuss the isolation proceduresdescribed by Johns et al. (85), Suzuki et al.(192), and McLaughlin and Domingue (116) andMarx and Petcovici (122), who used nonimmu-nogenic strains for ECA isolation. Finally, wewill describe and discuss in more detail therecent work of D. Mannel and H. Mayer (un-published data) on genetically defined ECA-positive and -negative mutants of Salmonella.

Studies on Immunogenic StrainsAll LPS preparations, no matter how ex-

tracted, obtained from the ECA-immunogenicstrains elicit the synthesis of anti-ECA in rab-bits, whereas LPS of nonimmunogenic strainsdoes not. This strongly suggests that in theimmunogenic strains the ECA immunodeter-minant is part ofthe LPS molecule. In addition,these strains, like all other ECA-positivestrains, probably contain non-LPS-linked ECA.The first report on the isolation and immuno-

chemical characterization of ECA goes back toKunin (100), who used for ECA extraction thehot phenol-water method (207), which was orig-inally designed for the isolation of 0 antigen.The aqueous-phase material obtained by thisprocedure was first dialyzed, then concen-trated, and finally precipitated with 10 volumesofethanol. The redissolved precipitate was sub-jected to diethylaminoethyl (DEAE)-cellulosechromatography, using a gradient of NaCl forelution.The individual fractions were tested for the

presence of ECA by HAI and for 0 antigen byboth HAl and precipitability in the agar dou-ble-difflsion test. The main peaks ofECA or 0antigen activity eluted at slightly differentNaCl molarities (0.16 versus 0.21 to 0.28, re-spectively), indicating that both activities areat least partly separable. Similar results wereobtained with extracts of E. coli 01, a nonde-


http://mm

br.asm.org/

Dow

nloaded from



fined E. coli R mutant, Salmonella typhosa,Shigella flexneri, and Proteus OX19 (100). Sev-eral other methods, such as ethanol fractiona-tion, preparative ultracentrifugation at 136,000x g, or chromatography on columns of Sepha-dex G-100 or G-200, resulted in only incompleteseparation of ECA from LPS (100), suggestingthat these two antigens are complexed togetheror are of similar molecular weight.

Detailed chemical analyses were carried outonly with the E. coli 014 material, becausemuch more ECA-specific material could be ob-tained from this strain than from any of theothers investigated (100). As shown in Table 1,the ECA material, purified by DEAE-cellulosechromatography, contained a small percentageof neutral sugars besides a fairly large amountofamino sugar (GlcN). Neither uronic acids nor6-deoxyhexoses were detectable. The ratherhigh nitrogen content was not accounted for byprotein, which showed large fluctuation amongindividual preparations (1.7 to 8.4%). From thenature of the principal amino acids identified,i.e., glutamic acid, alanine, glycine, asparticacid, and (probably) diaminopimelic acid, itwas concluded that ECA was most likely of cellwall origin (100). The ECA activity of thepurified material was not affected by periodateoxidation or by trypsin digestion. From theelution of ECA into the VO fraction of aSephadex G-100 column, the molecular weightof ECA was assumed to exceed 40,000.This purified ECA material failed to coat

RBC in spite of its good HA-inhibiting capacity,nor did it engender ECA-specific antibodies inrabbits by intravenous immunization. In boththese respects it differed from the phenol-waterextract used as starting material. In the light ofpresent knowledge, it seems probable that thepurified product represented the haptenic, non-LPS-linked form of ECA, which is the minorform of ECA in this strain (E. coli 014). Thelack of activity in HA is more difficult to ex-plain; it may indicate a loss of the carrier partof ECA.The immunochemical studies on ECA of E.

coli 014 reported by Hammarstrom et al. (70)seem to contradict the above findings in essen-tial points. They also used the hot phenol-watermethod (207) for ECA extraction from carefullywashed E. coli 014 bacteria, and again bothECA and 0 antigen were demonstrated in theaqueous phase. After removal of coextractedRNA by treatment with pancreatic ribonucle-ase, the residual material was found to be ofhigh molecular weight (in the VO fraction inSepharose 4B column chromatography). Analy-sis of the material before further separation

showed the presence of sugars of the LPS core(see Fig. 7): glucose, galactose, heptose, keto-deoxyoctonic acid (KDO), and some glucosa-mine beside a large amount ofO-acetyl. Most ofthe latter could be removed by treatment withalkali (0.25 N NaOH, for 3 h at 50°C) withoutaffecting the serological ECA specificity. Al-kali-treated material was subjected to mild acidhydrolysis (1% acetic acid for 1.5 h at 100°C),which cleaved the ketosidic linkage betweenKDO and lipid A. After removal of the lipidprecipitate, the acid-soluble fraction waspassed over Sephadex G-50 with pyridine-aceticacid ofpH 5.4 as eluting buffer. A major peak inthe molecular weight region of about 2,000 to3,000 showed inhibition of an ECA HA systemin contrast to a number of minor peaks elutedeither before or after this major peak. Sugaranalysis of the inhibiting fraction showed aratio of galactose-glucose-heptose of 2:1:5:1. Theglucosamine content was low; the nitrogen con-tent, however, was higher than expected. Theauthors concluded that in E. coli 014 the ECAspecificity is related to the 014 core region.The data obtained by Hammarstrom et al.

(70) are in agreement with immunogenic ECAbeing LPS linked. They show in addition thatthis linkage is not cleaved by hydrolysis with1% acetic acid used for splitting the KDO-lipidA linkage (109). Kunin (100) apparently wasworking with a mixture of the free and theLPS-linked ECA of E. coli 014, judging fromthe reported chromatographic separation ofECA and 014 specificities. A comparison of theelution patterns given by Kunin (100) shows awide overlap of both specificities in E. coli 014and a clear separation of ECA and 0 specifici-ties in the nonimmunogenic strains.

Studies on Nonimmunogenic StrainsSalmonella typhosa strain 0:901 was se-

lected by Johns et al. (85) for ECA extractionbecause of its lack of possibly interfering Vi, H,and K antigens. Acetone-killed bacteria wereextracted at room temperature with distilledwater. The aqueous extract obtained was thenmixed with picric acid to 90% saturation, re-sulting in removal of several protein antigensas followed by agar-gel precipitation with ahomologous antiserum. Two volumes of acetonewas added to the picric acid supernatant toprecipitate residual antigenic material and toremove most of the picric acid. The precipitatewas dissolved in bicarbonate buffer, dialyzed,concentrated, passed over a Sephadex G-200column, and thereby separated into four frac-tions, which were tested for 0 antigen and ECAspecificity by immunoelectrophoresis and im-

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


z +

_.j~oII + +

5 coU Z } + + + +,0. M,4

+

zc

* *

Q

co

C; oz z

[ 432Tit ztoCD

, z

eq

0~~~~~~~0z~~~~~~'co

cli Ci~~~

z I C030

p4U ~~~~U

-~~~~~~~~~~~~~5

44~~~~~~~~~.E~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-- 0664)tw ¢I

oo'&~~~~~~~~~~~~~ G)r~~~~iL0. iaS4RiiE3

;~~~~~~~~t7aanX2 t ~Ot&0~~~~ ~ ~ ~ ~ ~ ~~~~.

______p° p4| g °O

(i2e6 oC01c -

44-1 co0 1 ,

I . il-40 4) -4 ia : A ,M- 0

A

+ +

0_ 0

.Oo :0Oc.

m Y OL m 8

I

.

5 ,--

4.00

0.4

t

.0

m

0)

.03a

A

Cs)

*04

0

llco

'-

Si

44m!Eqs

m I gfI 8 .,


http://mm

br.asm.org/

Dow

nloaded from


+ o- t sc);$

0 *V

* ++'ri

co

*co

0 n

c) o

. .0

U ^b

0

8.X.5

*4)0

co0

04


http://mm

br.asm.org/

Dow

nloaded from



munoprecipitation. The third fraction con-

tained no 0 antigen but two other antigens; one

reacted only with the homologous anti-S. typhiantiserum, and the other reacted with anti-E.coli 014 serum. The latter was taken to repre-

sent ECA, and was further purified by prepara-tive gel electrophoresis. Sodium chloride ex-

traction of the appropriate gel fractions yieldeda material that strongly precipitated with an E.coli 014 antiserum. It was homogeneous on thebasis of a single precipitation arc and had a

high negative charge (fast anodic migration) atpH 8.6. From the rather high mobility in agar-

gel diffusion, its molecular weight was assumedto be far below the 40,000 reported by Kunin(100). Identical materials were demonstrable insimilar extracts of Salmonella minnesota S,Ra, and Rb chemotypes, but not in the mutantsof Rc to Re chemotypes (see Fig. 4; all the S.minnesota mutants of chemotypes Rc to Re in-vestigated also contain a rfe mutation [see be-low "Genetic determination"]).

Chemical analyses of the electrophoreticallypure material revealed a high percentage ofhexose, mostly galactose, but also some glucose(M. A. Johns, Ph.D. thesis, Boston Univ., Bos-ton, Mass., 1973) and a lower percentage of bothhexosamines (GlcN) and protein. Phosphorus,heptose, and KDO were completely absent. Theisolated material failed to sensitize humanRBC; it was, however, active in inhibiting an

ECA hemagglutinating system, although a

rather high dose (200 ug/ml) of the inhibitorwas required. No anti-ECA antibodies were en-

gendered by intravenous immunization of rab-bits with the purified product (85; Johns, Ph.Dthesis).Making use of the important discovery of

Neter and his group (192) that LPS and ECAare separable by ethanol fractionation, Mc-Laughlin and Domingue (116) obtained a partlypurified ECA preparation from E. coli 06 and075 and Klebsiella pneumoniae. Cells were

harvested with phosphate buffer of pH 7.3, andthe resulting suspension was heated for 1 h at100°C. The supernatant fluid containing bothECA and 0-antigenic material was mixed with95% ethanol to give a final ethanol concentra-tion of 85%. Ethanol-soluble and -insoluble ma-terials were separated by centrifugation. Theformer contained ECA, whereas most of the 0antigen was in the ethanol-insoluble fraction.After removal of the ethanol at room tempera-ture, the resulting residue was dissolved in wa-ter. A partial chemical characterization of theethanol-soluble, non-desalted material showedmainly protein and nucleic acids; some carbo-hydrate material (about 1% by the phenol-sul-furic acid method) and some chloroform-metha-

nol-soluble lipid were also present (see Table 1).The material was active in HAI as well as incoating RBC for HA by heterologous anti-ECAsera. It also engendered ECA antibodies byintravenous immunization of rabbits. The heatstability of ECA was considered to contradictits being a protein, contrary to the analyticaldata presented.In a recent study by Marx and Petcovici (122)

heated ethanol was used as extractant of ECAfrom Salmonella typhimurium TV149 (a roughmutant of type Ra; see Fig. 4). A suspension of200 g of wet bacteria in 2,000 ml of 96% ethanolwas heated at 60'C for 20 min. After centrifuga-tion of the cooled suspension, the supernatantwas collected, evaporated to near dryness, andthen taken up in 40 ml of 85% aqueous ethanoland precipitated with 3 volumes of acetone. Thesediment collected by centrifugation was redis-solved in 4 ml of water and applied to a Sepha-dex G-75 column. ECA was found mainly in theV0 fraction and was recovered in a yield of 5 mgfrom the 200 g of bacterial wet weight. PurifiedECA obtained by this technique was of highmolecular weight (excluded from Sepharose2B), retained its RBC-coating ability, and wasimmunogenic in rabbits using intravenous im-munization. Chemical analyses (see Table 1)revealed mainly protein, glucosamine, and glu-cose, as well as constituents characteristic of acephaline-type phosphoglyceride, i.e., glycerol,ethanolamine, phosphorus, and fatty acids(mostly palmitic acid, but also some stearic,myristic, and unidentified fatty acids). Thehighly acidic character of the isolated productwas documented by its ready absorption to var-ious anion exchangers and by its precipitationwith lysozyme (the latter as a basic protein isknown to precipitate polyanions, e.g., ribonu-cleic acid [1821). The material, in accordancewith an earlier report, lost its RBC-sensitizingability during incubation with phospholipase A(121). In contrast to the earlier report, phospho-lipase A was found to also destroy the capacityof ECA to inhibit HA, suggesting that the im-munodeterminants were also destroyed. TheECA materials tested in these two studies wereadmittedly different, yet it is difficult to recon-cile the conflicting findings. The action of phos-pholipase A was accompanied by a release ofabout half of the fatty acids present, presum-ably from the phosphoglyceride.A new extraction and purification method

(Fig. 1) for ECA was developed by Mannel andMayer (D. Mannel, thesis, Universitat Frei-burg i.Br., in preparation). It is based on ear-lier observations showing that phenol-water-extracted LPS usually contains the ECA deter-minant, whereas a phenol-chloroform-petro-

BACTERIOL. RimV.


http://mm

br.asm.org/

Dow

nloaded from



phenol-killed (100%)bacteria

phenol-waterextraction

phenol phase waer phaseE(1 (10%)

PCP extraction

PCP soluble PCP insolubleWL/GI WL/Gu | (2%)

phenol soluble (0.4%) water precipitate (5.0%)ultra ntrifugation ultrac ntrifugationultr= i.fugation

P1-S r1-L (0.3%) G1-S (4.9%) G1-L Gu

I LPS (1.8%)DEAE-celluloseelution at 0.9 M NaCI

[P-L DEAE ECA(0.2%)

FIG. 1. Extraction procedure (D. Mdnnel and H. Mayer, to be published) for ECA and ECA-free LPS. ECAis accumulated in fractions P1-S and P1 -L; P1-L further purified gives P1-L DEAE. ECA-free LPS is presentin G1-S. The relative yields as percentages ofdry weight ofbacteria (the smooth strain Salmonella MontevideoSH94 in this case) are given in parentheses.

leum ether (PCP)-extracted LPS (51) does not(126). Subsequent extraction with phenol-waterof previously PCP-extracted R mutants givesan aqueous-phase material that is considerablyenriched for ECA (126). The newly devisedmethod uses the same two extractants but in areverse order, and this makes it possible toextract ECA from both S and R forms (PCPdoes not extract LPS from S-form bacteria [51]).C. Gilanos (personal communication) obtainedhighly purified LPS of Salmonella S forms bytreating the aqueous phase of phenol-water ex-tracts with the PCP mixture. ECA and LPS areboth soluble in this mixture. However, on addi-tion of some drops of water to the phenol phase(obtained after removal of the volatile chloro-form and petroleum ether), only LPS precipi-tates, whereas ECA remains in solution. Ex-tensive dialysis and a final lyophilization stepresults in the so-called P1 fraction (see Fig. 1),which can be separated by ultracentrifugationto give a sediment (P1-S) and a supernatantfraction (P1-L).

Although both fractions showed high ECA

activity (HAI), only the Pl-L fraction was usedfor further purification. P1-L material was dis-solved by ultrasonic treatment in a smallamount of 0.5 M ammonium acetate-methanolbuffer (31) and was then applied to a DEAE-cellulose column previously equilibrated withthe same buffer. Stepwise elution with ammo-nium acetate-methanol of increasing molarity(0.5, 1.0, and 1.6) eluted ECA in the middlefraction. Rechromatography of this fractionwith 0.9 M buffer yielded ECA as a symmetri-cal peak as assayed by differential refractome-try. Investigation of individual fractions byHAI and gel precipitation showed that the peakeluted with 0.9 M buffer contained the entireECA specificity. Corresponding fractions werecombined, dialyzed, and electrodialyzed (50)and were then used for chemical analysis.The P1-L DEAE-purified ECA formed a pre-

cipitate in immunoelectrophoresis in agar gelsat pH 8.6, usually in the form of a bird-wingpattern similar to that observed with acidiccapsular polysaccharides of E. coli (152). Mildalkali treatment resulted in a single precipita-

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from



tion arc, thus indicating that ester linkagesmight have been cross-linking individualchains of ECA. The alkali-treated, more homo-geneous P1-L DEAE fraction was used for de-termining the sedimentation coefficient. Itshowed a single peak in analytical ultracentrif-ugation; the sedimentation coefficient of 0.5Ssuggests a rather small molecular weight, farbelow 10,000.

Chemical characterization of this purifiedECA material is being continued (D. Manneland H. Mayer, manuscript in preparation).Thus far, it has been shown that as the maincomponent this material contains a heteropoly-mer composed of two amino sugars, i)-glucos-amine and r)-mannosaminuronic acid (Man-NUA), which are partly esterified by palmiticacid. After hydrolysis with hydrochloric acid,the ManNUA is present almost completely inthe lactone form (124), which is easily demon-strated by high-voltage electrophoresis (Fig. 2and 6). Its unequivocal identification wasachieved by reduction of the esterified materialwith NaBH4, resulting in the formation ofmannosamine. Its D-configuration was provenby phosphorylation of the newly formed man-nosamine with- hexokinase-adenosine 5'-tri-phosphate (124).The importance of ManNUA for the antigen-

icity of ECA was indicated by complete disap-pearance of the precipitation line (Mannel andMayer, unpublished data) upon esterification ofits carboxylic groups with diazomethane. Sub-sequent saponification of the esterified ECArestored its precipitability. Also, reduction ofthe purified ECA material changed its precipi-tation. An independent line of proof ofManNUA being part of ECA is provided bymutants which, because of genetic blocks, lackECA: they all also lack ManNUA (see "Geneticdetermination of ECA"). Third, ManNUA-con-taining polymers are found in ECA-positive en-terobacterial strains of different genera. Figure2 shows an electrophoretic separation of hydrol-ysates of semipurified ECA fractions from dif-ferent strains.The palmitic acid is apparently not part of

the immunodeterminant of ECA, but is impor-tant for its RBC-coating ability: the removal ofthe palmitic acid by alkali treatment does notaffect the ability of the material to inhibit HA,but completely abolishes the sensitizing ability.This observation is analogous to the demonstra-tion by Hammerling and Westphal (72) thatlipid-free polysaccharides-with the exceptionof several acidic capsular polysaccharides ofhigh molecular weight (67, 84)- usually are notattached to the surface of RBC, whereas theirartificially prepared O-stearoyl derivatives are.

FIG. 2. Electrophoretic separation of hydrolysates(N HCl, 4 h, 100'C) ofsemipurified ECA-containingfractions (pH 2.8, 50 V/cm, 1 h, AgNO3INaOH). (A)ManNU (lactone of mannosaminuronic acid); B,glucosamine; C, oligosaccharide of A and B. (a)Citrobacter 1658; (b) Salmonella arizonae F628(09a-, 09c-); (c) S. minnesota Rl; (d) E. coli F870(K-12); (e) standards ofglucosamine and glucose.

Other components, reported previously asconstituents of ECA, like protein, ribonucleicacid, or neutral sugars (see Table 1), are pres-ent in this material at most in minimalamounts. It is not certain, however, that allconstituents of this ECA material have beendiscovered.

Chemical Identity of ECAAs a whole, the data summarized in Table 1

vary widely, making it difficult to draw conclu-sions as to the chemical identity of ECA.Most workers have found ECA to be highly

negatively charged. Its antigenicity is not de-

BACTERIOL. REV.

ALLLA.r!!.v

V


http://mm

br.asm.org/

Dow

nloaded from



rfe, rff 7( met E

rfa ilvpyrExyl

\ ~~his

rf bFIG. 3. Simplified chromosome map of Salmo-

nella showing the positions of rf. genes (190; P. H.Makeli, G. Schmidt, H. Mayer, H. Nikaido, H. Y.Whang, and E. Neter, J. Bacteriol., in press). rfa,rfb, rfc, and rfe genes participate in the synthesis ofLPS, rif, and rfe, and probably some rtb genesparticipate in the synthesis ofECA.

stroyed by heat, even 12000 (216), or by trypsinor Pronase treatment (219), suggesting that itis not a protein; however, protein constituentsof the outer membrane of enteric bacteria areextremely resistant to such treatments (e.g.,references 16, 77, 105, 150, 165). Since ECAactivity is also resistant to periodate oxidation(100, 121) and is present in mutants that areunable to synthesize uridine 5'-diphosphate(UDP)-glucose, UDP-galactose, or guanosine5'-diphosphate (GDP)-mannose (see "Geneticdetermination of ECA"), ECA is probably not acarbohydrate composed of the usual commonhexoses. Acid hydrolysis (0.1 N HCl) destroysECA only slowly compared with LPS (100). APseudomonas "factor," probably an enzyme,which unfortunately has not been further char-acterized, destroyed ECA as measured by HAIbut did not affect the LPS-linked ECA (219,221).The data of Mannel and Mayer described

above suggest that ECA is a heteropolymercomposed predominantly ofN-acetylated aminosugars, the ManNUA providing the negativecharge and being part of the immunodetermi-nant structure. The strongest evidence for thisview is the lack of ManNUA in mutants lack-ing ECA and its presence in various strainshaving ECA. A partial esterification by pal-mitic acid adds hydrophobic groups to the oth-erwise hydrophilic polymer. Such amphiphilicpolymers are expected to be surface active (75,231), and ECA can easily attach to all kinds ofsurfaces (RBC, lymphocytes, and latex parti-

cles as demonstrated by the acquisition of sero-logical ECA specificity). Alkali treatment de-stroys the sensitizing capacity of ECA in crudeextracts (58, 126) as well as in the P1-L DEAE-purified preparation, from which it removespalmitic acid. In contrast, the ability of LPS tosensitize RBC for agglutination by anti-O anti-bodies is enhanced by and often requires thisalkali treatment (145). Some agents that in-hibit or destroy the capacity of LPS to sensitizeRBC for agglutination with anti-O antibodieshave much less effect on the sensitizing capac-ity of ECA, e.g., normal rabbit serum and poly-myxin B (60, 143, 227). Treatment with phos-pholipase A (but not C or D) destroyed thesensitizing capacity ofECA (121). ECA purifiedby certain methods (DEAE chromatography ofphenol-water extracts [100], Sephadex G-200chromatography of picric acid-soluble water ex-tracts [85]), although active in HAI or precipi-tation, does not sensitize and may have lost thehydrophobic part of the molecule.The proposed structure could also well ac-

count for the apparent presence of ECA as ag-gregates of different sizes in crude bacterialextracts as well as in purified preparations.These aggregates could be micelles of ECA andalso co-micelles with other amphiphilic com-pounds, e.g., LPS. The possible importance ofaggregate size will be discussed below under"Immunogenicity." However, how can thisstructure be reconciled with the results ofchemical analysis of the other ECA prepara-tions?One obvious possibility is that what is called

ECA is in actuality not one antigen but ratherseveral (at least two) independent immunologi-cally different antigens that have the commonproperty of being shared by several species ofenteric bacteria. Another possibility may be therelative impurity of most ECA preparations.Some discrepancies can be accounted for byassociation of the ECA immunodeterminantgroups with different carrier structures in thevarious preparations.A heteropolymer built mostly of amino

sugars could easily be overlooked, especiallywhen present in rather small amounts: theusual colorimetric assays for carbohydrates (or-cinol or phenol-sulfuric acid, etc.) do not meas-ure amino sugars, nor do aminouronic acidsreact with carbazole-sulfuric acid (154), whichis normally used for quantitation of unsubsti-tuted uronic acids (35). Glucosamine has beenfound as a constituent of ECA by nearly allinvestigators (Table 1), although to various ex-tents. High values were reported by Kunin(100) and by Marx and Petcovici (122). Thelatter authors have kindly provided us with a

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from



sample of their purified ECA preparations,which we could show to contain appreciableamounts of ManNUA in addition to glucosa-mine (122).We also studied a sample of the material

isolated by Johns et al. (85) in which, however,we found no or only trace amounts of eitheramino sugar. Reciprocally, McCabe and Johns(personal communication) compared the P1-LDEAE material of MAnnel and Mayer preparedfrom Salmonella Montevideo SH 94 with theirECA in immune precipitation with E. coli 014antiserum. This comparison revealed that thetwo ECA preparations are serologically differ-ent. Whereas the Johns et al. preparation gaveone sharp precipitation band with the anti-E.coli 014 serum in double diffusion in agar, theP1-L DEAE material gave two bands, only oneof which showed a reaction of identity with theJohns et al. material. The other one was com-pletely separate, closer to the antigen well. Inimmunoelectrophoresis the same picture wasretained, the Johns et al. material giving onesharp band with fast movement to the anodeand the P1-L DEAE material giving two bands,both less sharp, one in the same position as theband of the Johns et al. preparation and theother closer to the antigen well. A further dif-ference between these two preparations is lessactivity in HA and HAI of the Johns et al.material. Obviously, the two preparations aredifferent both serologically and chemically.The genetic evidence speaks for the ManNUApolymer being ECA. For future work it will benecessary to establish the identity of bothantigens.

GENETIC DETERMINATION OF ECAMutations blocking the biosynthesis of a

component of ECA would be expected to blockits synthesis and result in strains lacking ECA.Conversely, determining the ECA content ofknown mutants might help to deduce the com-position of ECA. Unfortunately, no way ofselecting ECA-negative mutants is known, andall such mutants we now know were eitherselected indirectly or found accidentally.Mutants defective in the synthesis of UDP-

glucose, UDP-galactose, and GDP-mannose(pmi), which interfere with the synthesis ofLPS, were readily available in Salmonella andE. coli (134, 158, 190). They were studied for thepresence of ECA (121; our unpublished data).They all were positive, suggesting that the cor-responding hexoses are not a part of ECA. IfManNUA is a constituent ofECA, the presenceof ECA in pmi mutants indicates thatManNUA is not synthesized via the GDP-man-nose pathway. This agrees with the demon-

BACTERIOL. REV.

strated mode of biosynthesis of ManNUA viaglucosamine nucleotides (see below, "Othercross-reactivities and related antigens").When the extensively studied series of rough

(R) mutants ofS. Minnesota (108, 109, 173, 175)were tested for the presence of ECA, a sugges-tive pattern emerged (85, 120): the R mutantswith the most defective LPS were devoid ofECA, whereas the smooth parent strain andthe R mutants with more complete LPS corestructures were ECA positive. This indicated arelationship between the synthesis of ECA andthat ofLPS and suggested even structural simi-larities (85). However, at that time it wa's notpossible to propose a coherent hypothesis be-cause there were no reliable data on the chemi-cal composition of ECA. The finding that an-other type of R mutant, those with a completeLPS core but lacking 0-specific side chains be-cause of deletion of a long chromosomal seg-ment in the his-rfb region, were also ECA neg-ative (P. H. Makela, G. Schmidt, H. Mayer, H.Nikaido, H. Y. Whang, and E. Neter, J.Bacteriol., in press) caused further difficultiesin interpretation.A genetic study of these and other mutants

has now established a connection between thebiosynthesis of ECA and LPS. First, rfe genes,close to ilv (Fig. 3) on the genetic map, arerequired for the synthesis of ECA as well as forthe synthesis of the LPS side chains of somechemotypes. Second, rff genes, located veryclose to rfe, are required for ECA synthesis butdiffer from rfe by not participating in LPS syn-thesis. Third, some so-far undefined part of thertb gene cluster also participates in thesynthesis or regulation of synthesis of ECA.Furthermore, the linkage of ECA to LPS inimmunogenic strains requires the function ofrfa genes to produce the correct complete LPScore and to transfer the ECA immunodetermi-nant to it.

The rfe GenesA genetic analysis of the S. Minnesota R

mutant series (mR3, mR5, mR8, etc.) men-tioned above revealed that all of the more defec-tive mutants (LPS chemotypes Rc, Rd, or Rd2,and Re; Fig. 4) that were also ECA negative infact had two separate rf. mutations, both ofwhich affected LPS synthesis (87). One muta-tion in each strain was of the known rfa type,i.e., affecting a specific transferase of a coreconstituent (sugar or phosphate) and therebyresponsible for the LPS chemotype of each mu-tant (190). The other mutation, however, was ofthe less well-understood rfe type, which pre-vents the synthesis of the 0 side-chain polysac-charide in certain Salmonella groups (118): rfe


http://mm

br.asm.org/

Dow

nloaded from



H KDO0 poly- Glc Gal GIc Hep Hep- lipid Asaccharide / /EAP

GlcNAc Gal Hep P

P

Re

Rd2

Rd,

Rc

Rb

Ra

S

rtb orrfe orrfaL

FIG. 4. Schematic structure of the LPS in smooth (S) and rough (R) strains of Salmonella, the Rchemotypes Ra to Re, and the rf. mutations leading to them (84, 109, 190). EA, ethanolamine; Gal, D-galactose; Glc, D-glucose; GlcNAc, N-acetylglucosamine; Hep, heptose, KDO, ketodeoxyoctonic acid; P,phosphate.

mutants have a complete LPS core of chemo-type Ra. Operationally, rfe mutants can be dis-tinguished from rfa- mutants by the location ofthe affected gene (Fig. 3): the rfe genes are closeto ilv at min 122, and the rfa genes are close topyrE at min 116.

It turned out that the lack of ECA in the S.Minnesota R strains was associated with the rfetype of mutation (120): all the single rfa mutantderivatives of chemotypes Re, Rd2, Rd,, and Rcwere ECA positive, contrary to the rfe- deriva-tives, which all were ECA negative. A clearimplication of this finding was that the LPScore could not be part of ECA, contrary to pre-vious hypotheses (70, 85).

It was less clear what the requirement of anintact rfe function for ECA synthesis meant,because the function of rfe genes is still un-known. However, more proof of the role of rfegenes in ECA synthesis has been accumulat-ing. rfe mutants had been originally found as Rmutants in S. Montevideo (0-6,7 of group Cl) aswell as in S. Minnesota (0-21 of group L) (118).All these rfe mutants were devoid of ECA orcontained it in minimal amounts only (119,120). These trace amounts could be explainedby assuming that the rfe mutation in questionhad not completely inactivated the rfe gene

function and that the remaining function wasdetectable as "leakiness" of the mutation. Suchleaky mutations are very common among othergenes of LPS synthesis (103, 158, 190).

In Salmonella of 0 group B (0-4,12), rfe mu-tations were not known, and it was not likelythat a comparable rfe function would be re-quired for the synthesis of the 0-4,12-type poly-saccharide. Hybrid strains derived from S. ty-phimurium 0 group B by introducing intothem by conjugation a mutant rfe (C-rfe-) genefrom S. Montevideo were smooth with LPS oftype 0-4,12 as determined by the group B rfband rfc genes (119). These hybrids were never-theless ECA negative, indicating that there arein S. typhimurium genes allelic to rfe of groupC and that these are also necessary for its ECAsynthesis. This hypothesis was tested by fur-ther genetic manipulations. The C-rfegenes ofthe ECA-negative hybrid were replaced in fur-ther crosses by the C-rfe+ genes from S. monte-video or by the corresponding (B-rfe+) regionfrom S. typhimurium (119); in both cases therecombinants had become ECA positive.A further study of such hybrids, obtained by

conjugation or transduction and containingvarious combinations of rfe and rib plus rfcfrom groups B and Cl, either functional or mu-

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from



tated, has made it clear that the rfe genes ofeither group B or group C are required for ECAsynthesis (118, 119; P. H. Mdkeld, unpublishedobservations). They are not necessary for thesynthesis of the 0-4,12 polysaccharide but par-ticipate in the synthesis of the 0-6,7 LPS ofgroup Cl. In all these cases the rfe mutantswere obtained by selecting for R mutants instrains having the rtb+ genes of group C. Thusit could be concluded that all rfe mutations thatimpair the synthesis of 0-6,7-type polysaccha-ride also impair the production of ECA. At thattime it was not clear, however, whether thereverse was true, i.e., whether all mutations ingroup C that prevent the production of ECAwould also affect LPS; no method was availablefor a direct selection of ECA-negative mutants.To find out whether rfe genes are also present

in E. coli, rfegenes were transferred from S.Montevideo or S. typhimurium into E. colistrains (G. Schmidt, H. Mayer and, P. H. Mak-ela, J. Bacteriol., in press). These recombi-nants were all ECA negative and rough, withthe complete R core. This suggested that thefunction of rfe genes is required for the produc-tion of ECA and for the synthesis of the E. coli09, 08, and 0100 polysaccharide in much thesame way as it is needed in group C Salmo-

nella. When R mutants with the complete LPScore were selected in a strain of E. coli08:K27-, some of them were also ECA negativeand ilv linked (P1 transduction), and thereforeprobably of rfe type.Some F' episomes ofE. coli K-12 carry the ilv

genes and probably also the nearby rfe genes.When E. coli K-12 strains with the F' factorsF14 or F133 were used as genetic donors toseveral ilv- ECA-negative rfe (or rif; see be-low), mutants ofS. typhimurium, ilv+ recombi-nants were ECA positive, indicating that F14and F133 carry rfe (rff) genes that are function-ally homologous to Salmonella rfe (G. Schmidtet al., J. Bacteriol., in press).

Role of the rib RegionA series ofR mutants with various lengths of

their chromosome in the rib-his region deletedhas been isolated and characterized in S. typhi-murium (151). By analyzing several of the en-zyme products of the rib genes, it was possibleto deduce how far the deletion in each mutantstrain extended into the rfb cluster. When thisseries was tested for the presence ofECA (Mak-ela et al., in press), a clear pattern emerged: thestrains with the longer deletions were ECAnegative, and those with shorter deletions were

rib genes

MAN RHA ABE MAN

2 2 Y-1 3 1 *2 3 o 3 2

+

+

+

+

FIG. 5. Presence (+) or absence (-) of ECA in deletion mutants of the his-rib region of Salmonellatyphimurium (151; Makeld et al., J. Bacteriol., in press). MAN 2, etc., refer to enzymes determined by thertb genes (MAN 2 = phosphomannomutase; MAN 3 = GDP-mannose pyrophosphorylase; ABE 3 = 'CDP-abequose synthetase"; ABE 2 = CDP-glucose oxidoreductase; ABE 1 = CDP-glucose pyrophosphorylase;RHA-3 'thymidine diphosphate [TDP]-rhamnose synthetase"; RHA 1 = TDP-glucose pyrophosphorylase;RHA 2 = TDP-glucose oxidoreductase; Y = unidentified, probably regulatory, function concerning anotherrib gene).

his operon,-A

BACTERIOL. Rzv.


http://mm

br.asm.org/

Dow

nloaded from



ECA positive (Fig. 5). This finding suggestedthe possibility of defining at least one of theenzymes required for ECA biosynthesis. When,however, some of these long-deletion strainswere used in genetic crosses, in which thedeleted rib region together with his wasreplaced by its wild-type allele (as evidenced bythe production of a smooth 0-4,12-type LPS),these recombinants were still ECA negative.

It was therefore suspected that their ECA-negative phenotype was due to a second muta-tion outside the rfb cluster. A likely possibilitywas an rfe type of mutation. This was tested byfirst isolating ilv- mutants of these deletionstrains and then crossing them with an ECA-positive S. typhimurium donor. Many ilv+ andtherefore probably also rfe+ recombinants weretested and found to be still ECA negative. How-ever, recombinants in which both the rfe andrib chromosomal regions came from an ECA-positive S. typhimurium donor were ECA posi-tive (P. H. Makeld, G. Schmidt, H. Mayer, H.Nikaido, H. Y. Whang, and E. Neter, J. Bac-teriol., in press). The inevitable conclusion wasthat the long-deletion strains contained muta-tions in (at least) two separate genes, one in thehis-rib region and the other close to ilv, bothrequired for the synthesis of ECA.That the rib region of S. typhimurium was

playing a role in the production of ECA hadbeen suggested by the absence of ECA in hy-brids between S. typhimurium and S. monte-video in which the group B rib region of theformer was replaced by the group C rib regionof the latter (119). This also suggested a differ-ence between Salmonella groups B and C inthis respect: only the group B rib region fur-nished the function required for ECA synthesis.When deletions of the his-rib region were stud-ied in group C (S. Montevideo), they all wereECA positive, supporting the notion that C-ribdoes not participate in ECA synthesis (Mikelaet al., in press).The mutant and hybrid strains described

above, which were missing a function of thegroup B rib region required for the productionof ECA, were, on closer analysis, not com-pletely devoid of ECA. Instead, trace amountsof ECA could be found in each case in strainsthat did not have additional mutations affect-ing ECA (119). This trace reaction is probablytrue ECA, since it is detected by HAI as well asby the ability of bacterial extracts (especiallyafter concentration) to sensitize RBC for HA.Also, some ManNUA-containing material canbe isolated from these strains (results to bepublished). It cannot be caused by conventionalleakiness of a mutation, since it is found instrains with large chromosomal deletions. It

could, however, result from a situation inwhich another enzyme, specified by a separategene, was providing the same function, al-though in a less efficient form; examples of thisare known for some rib--determined enzymes(134, 151). Another possibility is that the role ofB-rib genes in ECA synthesis is regulatory.Another clue to the function of the rib genes

may be provided by the phenotype of the ribdeletion mutants. Many of these mutants areslightly more sensitive to a detergent, deoxy-cholate, than are normal strains. When theirderivatives, which contained only the rfb dele-tion mutation, were prepared, they turned outto have more dramatic phenotypic defects. Suchstrains grew poorly on ordinary media andwere sensitive not only to deoxycholate but alsoto sodium dodecyl sulfate at concentrations towhich wild-type Salmonella are completely re-sistant. These changes in detergent sensitivitysuggest a defect in the organization of the cellwall, probably the outer membrane, of whichLPS is a part. Generalized defects of the outermembrane have been described in deep LPSmutants (those with a Re type core; Fig. 4): theprotein content of their outer membrane is re-duced and they are highly sensitive to deoxy-cholate and many other toxic agents (3, 170,171, 185, 230).

Mutants of a New, "rif," TypeA mutant strain ofS. Minnesota, smooth but

ECA negative, was accidentally found, andshowed immediately that not all mutations pre-venting the synthesis of ECA also affect LPS(52). The mutation in this strain was closelylinked to ilv, like rfe. Whether this locus shouldbe called rfe is disputable: it shares two proper-ties (map position, ECA-negative mutant phe-notype) with rfe but differs in a third, participa-tion in LPS synthesis. We propose to call it rifuntil it is further characterized. It seems likelythat the function and mode of action of thegenes of the rfe cluster will soon be clarified,and this problem can then be solved.More mutations of the above-mentioned "rif'

type have been found recently (Mdkela et al.,in press). The ilv-linked mutation found inmany his-rib-deletion strains seems to be ofthistype. When the his-rib region of such strains isreplaced by the wild-type allele of either groupB or even group C1, the recombinants aresmooth (0-4,12 or 0-6,7) but ECA negative.The ilv-linked mutation therefore behaves likerif: it prevents the biosynthesis of ECA withoutinterfering with the synthesis of the 0 polysac-charide.The rif mutation in the his-rib-deletion

strains alleviates the phenotypic effect of the

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from



deletion on detergent sensitivity of the bacte-ria. The rff rfb--deletion mutants are less sen-

sitive to deoxycholate than are the rtb-deletionmutants, and single rff mutant bacteria de-rived from them are as resistant as wild-typestrains. New rff mutations could be easily se-

lected from his-rtb-deletion strains by selectingfor resistance to sodium dodecyl sulfate. Howthe rff mutations counteract the pleiotropic ef-fects of the rtb deletion has not been clarifiedyet, and a closer study of the outer membraneof these various mutant types is needed.

Genetics of ECA in Immunogenic StrainsFor immunogenicity, the ECA immunodeter-

minant must be linked to LPS. As described inthesection on immunogenicity, this requires a

certain type of LPS, essentially the completeLPS core of unusual type present only in a fewenterobacterial strains. A rfa mutant of such a

strain (E. coli 014) was found to be ECA posi-tive but nonimmunogenic, presumably becauseECA could not be linked to an incomplete core(123).The gene rfaL codes for a translocase, which

is required for the attachment of 0 side chainsto LPS core (190). rfaL mutants of immuno-genic strains of E. coli and Shigella also haveECA but are not immunogenic, suggesting thatthe rfaL function is required for the transloca-tion of ECA to LPS (174).Genes required for the synthesis of the ECA

immunodeterminant are also expected to be re-

quired for immunogenicity. This has beentested for rfe: rfe mutants of an immunogenicE. coli strain had no detectable ECA and didnot elicit the production of anti-ECA (G.Schmidt et al., in press).

ConclusionsWhat we now know of the role of each of the

genes of ECA and/or LPS synthesis can be de-scribed as summarized in Table 2. The tablecontains two main columns, one for the Salmo-nella 0 group B, as exemplified by S. typhimu-rium, and the other for several other Salmo-nella and E. coli serogroups. These differ in therole of rfe genes, required for the synthesis of 0polysaccharide in most 0 groups but not in S.typhimurium. They also differ in the role of rfbgenes, which apparently participate in ECAsynthesis only in S. typhimurium.

In all strains so far examined, rfe and rffgenes are required for the synthesis of ECA.The immediate products of any of these genes

have unfortunately not yet been isolated andidentified, and consequently only hypothesescan be presented of their mode of function. Therfe genes should concern a step common to thesynthesis of ECA as well as to T1 and 0 sidechains of so many different 0 groups that com-

mon components to all are not known. Thehypothesis best fitting the existing data is thatrfe genes are involved in the synthesis or modi-fication of a carrier molecule used in the assem-

bly of a number of polymers, such as ECA, T1,and several types of 0 side chains. This carrierwould have to be different from the "ACL" car-

rier lipid in the assembly of 0 side chains ofgroup B Salmonellae, since these do not requirethe rfe function. Some evidence for a differentmode of synthesis for the 0 side chain of S.montevideo, 0 group C, has indeed been pre-

sented (54). The rff genes not participating inLPS synthesis could be involved in the synthe-sis and assembly of such ECA-specific com-

pounds as ManNUA (at least two enzymes re-

TABLE 2. ECA and LPS characteristics of mutants in different rf. genes

Salmonella Montevideo of groupCta. ^ '~~~~~

Salmnonella typhimurium ofSalmonella minnesota of group L; g B

Reference Mutation Escherichia coli 08, 09, 0100

ECA 0 hap- LPSc EC hp LPStenb tnECA0None + (+) HS + (+) S

119, 120 rfb + - R, complete core + - RaAd rtb long deletion +e _ R, complete core Trace - Ra119, 120, Bd rfe - - R, complete core - (+) SA rff - (+) S - (+) S119, 120, 123, B rfaL to rfaE + + R, complete or + + Rato Re

incomplete core

a Most but not all points tested for each species/serotype.b 0-specific polysaccharide detected as hapten if not transferred to the LPS core; small amounts present,

even in smooth forms as intermediates of biosynthesis.e See Fig. 4 for different R structures in rf. mutants.d A, P. H. Makela, G. Schmidt, H. Mayer, H. Nikaido, H. Y. Whang, and E. Neter, J. Bacteriol., in press.

B, G. Schmidt, H. Mayer, and P. H. Makela, J. Bacteriol., in press.e Several his-rfr-deletion strains tested, but length of deletion was not known.

BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from


ENTEROBACTERIAL COMMON ANTIGEN

quired for its synthesis; see "Other cross reac-tivities and related antigens").ECA, then, behaves in many ways like both

the 0 and T1 polysaccharides (190). All theseantigens (ECA, 0, and T1) are linked to LPS,provided the core is complete and of the correcttype and the product of the rfaL gene (appar-ently part of translocase) is present. If they arenot linked to LPS, they stay as 'hapten," linkedto the carrier molecule on which they weresynthesized. They all (in the case of 0 polysac-charide, this applies to certain 0 groups only)require the function of the rfe gene, which maybe providing the correct form of the carrier. Inaddition, there are genes involved in the syn-thesis of structures specific for each one ofthem: rff (and possible other types, not yetfound) for ECA, rifb for 0, and rft for the Tpolysaccharide.The ECA-negative mutants have provided

the best opportunity so far to test the chemicalidentity ofECA. The presence or absence of theManNUA-containing material has been testedin all the three mutant classes (rfe, rff, and rfbdeletion) found ECA negative by serologicalHA and HAI tests (D. Mannel, H. Mayer, andP. H. MAkela, manuscript in preparation). Ineach case a complete correlation was found be-tween lack of ManNUA and the ECA-negativephenotype. Figure 6 shows the lack of Man-NUA (lactone of ManNUA) in all three ECA-negative mutants of S. minnesota studied (R5and R8 = rfe- mutants; SH 3786 = a rff-mutant) and its presence in a S. minnesotawild-type strain (S1114) as well as in ilv+rfe+rffrecombinants of the three ECA-negative mu-tants. Their P1-L fractions (preparation as inFig. 1) were also compared serologically (Table3). Similar results were obtained with ECA-negative mutants of S. typhimurium and S.montevideo, and they are in our opinion a verystrong proof of the identity of ECA with theManNU/GlcN polymer.

IMMUNOGENICITYImmunogenic Strains

Anti-ECA was originally (102) found in thesera of rabbits that had been immunized with afew E. coli serotypes (014, 056, 0124, 0144),whereas numerous other E. coli serotypes wereapparently unable to immunize the rabbits toECA. When whole killed bacteria are used asimmunogens, very few strains (although ECApositive by various serological tests) are able toelicit the production of anti-ECA antibodies(Table 4). Although it was later found possibleto prepare immunogenic ECA also from nonim-munogenic strains as a fraction soluble in 85%ethanol (192), the immunogenic strains re-

FIG. 6. Electrophoretic separation of hydrolysates(4 N HCl for 2 h at1OOC ofPl -L fractions (250 pg)from Salmonella minnesota strains (120; D. Mannel,H. Mayer, and P. H. Makela, manuscript in prepara-tion) compared with glucosamine and glucose stan-dards (a). Electrophoretic conditions and staining asin Fig. 2. (b) ilv+ rfe+ recombinant of R5, ECApositive; (c) R mutant R5 (rfa- rfe-), ECA negative;(d) ilv+ rfe+ recombinant ofR8, ECA positive; (e) Rmutant R8 (rfa- rfei), ECA negative; (f) ilv+ rff+recombinant ofSH3786, ECA positive; (g) SH3786,a rff- mutant, ECA negative; (h) parent strainS1114, smooth, ECA positive.

mained different: part of their immunogenicECA was ethanol insoluble and present in thesame fraction with LPS.Although the overall chemical composition of

the LPS (= 0 antigen) of the two hitherto in-vestigated immunogenic serotypes (014 and056) shows that they both belong to the mostsimple chemotype I (84, 109, 153), as also do com-plete R-type mutants, this did not readily sug-gest why they were immunogenic. A clue to thechemistry of immunogenicity was found whenMayer et al. discovered (126-128) that certainrough mutants of E. coli and Shigella wereimmunogenic in the same sense as was E. coli014. Heat-killed bacteria of these R types wereimmunogenic, and most of their immunogenicECA was ethanol insoluble (214). Immunogen-icity was limited to certain core types. Thecomplete Salmonella-type core is described asRa, whereas amongE. coli and Shigella strainsseveral different core types are found, desig-nated from R1 to R4, plus the core type found inthe geneticist's favorite strain, E. coli K-12 (84;

607VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


TABLE 3. Comparison ofPl-L fractions from ECA-positive and ECA-negative smooth strains ofSalmonellaminnesotaa

Minimal in- .Strain Derivation Pl-L (mg) Titer in HA' hibiting in tion in I ManNUA,

Hgg/l)i pH 8.6'S1114 Wild type 35.5 (2.0)c 1:5120 7.8 + +SH3786 rif mutation 10.7 (0.9) 1:1280d 250 - TracedSH5657 ilv+ rff 28.5 (2.0) 1:5120 3.9 + +

recombinant ofSH3786

a Based on reference 12D and D. Mannel, H. Mayer, and P. H. Makela, manuscript in preparation.Abbreviations: HA, hemagglutination; HAI, hemagglutination inhibition; IE, immunoelectrophoresis;HVE, high-voltage electrophoresis on paper.

b ECA antiserum was rabbit antiserum against heat-killed whole cells ofShigella boydii type 3- (F3140)(128).

c Numbers in parentheses indicate percentage of aqueous phase.d Probably caused by leakiness of the rff mutation, which allows the synthesis of a small amount ofECA

detected in the Pl-L fraction in which it is enriched. No ECA activity was detected in this strain byexamining the supernatant of a heat-killed suspension for HA.

TABLE 4. ECA-immunogenic strainsa

Organism Core type LPS ECA immuno- ReferencegenicityEscherichia coli 014:K7 R4 + + 102, 123, 172, 214E. coli 056 + 102, 112E. coli 0124 + 102, 112E. coli 0144 + 102E. coli K-12 R-K-12 + 174, A, BbShigella sonnei phase II R1 + + 44, 126

R mutant of Shigella boydii type 3 R1 + + 214R mutant ofE. coli 08:K27+ R1 + + 214R mutant ofE. coli 08:K27- R1 ++ 126, 214R mutant ofE. coli 09:K29- R1 + + 126R mutant ofE. coli 09:K31- R1 ++ 126

a I. e., strains that elicit an anti-ECA response when injected, as heat-killed suspensions, intravenouslyin rabbits.

b A, P. Prehm, S. Stirm, B. Jann, K. Jann, and H. G. Boman, Evr. J. Biochem. 66:369-377, 1976. B, H.Mayer, A. M. C. Rapin, G. Schmidt, and H. G. Boman, Eur. J. Biochem. 66:357-368, 1976.

Fig. 7). Of these, only R1, R4, and at leastpartly K-12 are associated with ECA immuno-genicity (126-128, 172, 214). The latter mutantsare chemically very similar and even serologi-cally cross-reacting to some degree (127). TheRa, R2, and R3 core types have a terminal N-acetylglucosamine, whereas the R1 and R4cores end in an unsubstituted glucose. In the K-12-type core, the terminal glucose is partiallysubstituted by glucosamine. The immunogenicand nonimmunogenic core types seem to differ.also in the anomeric linkages of the terminal orsubterminal glucose: it is probably a in thenonimmunogenic strains and 8 in the immuno-genic ones (125). Perhaps this is the reason whya Salmonella mutant of core type Rb (Fig. 4)was not ECA immunogenic (H. Mayer, unpub-lished observations).The classically immunogenic E. coli 014 was

recently found to fit into this scheme (172) inso-

far as it was shown to be in fact a rough strainwhose cultural roughness was masked by thecapsule (K antigen). Its acapsular mutants arerough, and LPS extracted from either the cap-sulated strain or acapsular mutants is R, corre-sponding to the new R type R4. The serologicalspecificity called 014 would then represent thespecificity of the R4 core. In fact, F. Kauffmann(Copenhagen), who established the E. coli typ-ing scheme, hesitated to identify E. coli 014 asa smooth strain since it had features of an Rmutant (215). The other immunogenic 0 typeshave not been studied in equal detail; knowinghow easily R mutants arise and accumulate inpreserved cultures of the enteric bacteria, itwould not be surprising if they, too, were R orpartly R. The anti-ECA titers obtained whenusing them as immunogens are rather low (102,85, 126).We may conclude that the immunodetermi-

608 MAKELA AND MAYER BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from


ENTEROBACTERIAL COMMON ANTIGEN

NONIMMUNOGENIC STRAINS

GlcNAc Gal

a 1.2 al 1.61.2 1_3___ _ 1.3G c 1 -2 )Gal 1.3 ) Glc l o (Hep,KDO,EA,P) Salmonella Ra

GlcNAc Gal

a 1.2 a 1.6(14,12 1.2 1.3 N, 1.3Glc --.- GIc -3 Glc -3 (Hep,KDO,EA,P) Escherichia coli R2

a an

Glc Glc

1 4GlcNAc - Gal - GIc - (Hep,KDO,EA,P) E. coli R3

ECA-IMMUNOGENIC STRAINS

GIc Gal

G - GcGl--HeK2PEci1. 3 1 3 1.

al 1. lc 1-3 >- Glc 1. * (Hep,KDO,EA,P) E. coli R1

GlcNAc Gal-

1.6 1.6

GIc 1.2- Glc 1-.3> Glc 1.3 > (Hep,KDO,EAP) E. coli K-12

RhaFIG. 7. Schematic structure of the LPS core types in ECA-nonimmunogenic strains Salmonella Ra (109),

E. coli R2 (71), and E. coli R3 (86), and in ECA-immunogenic strains E. coli Ri (84) and E. coli K-12 (P.Prehm, S. Stirm, B. Jann, K. Jann, and H. G. Boman, Eur. J. Biochem., 66:369-377; H. Mayer, A. M. C.Rapin, G. Schmidt, and H. G. Boman, Eur. J. Biochem. 66:357-368.

nant of ECA can be linked to the LPS core,providing the core is ofa suitable configuration,and in this linkage it is immunogenic for theECA determinant. We would like to believethat this is a covalent linkage because of theassociation of ECA with LPS in the followingprocedures (126, 128): (i) in phenol-water andPCP extraction (the PCP extract of nonimmu-nogenic strains does not contain ECA); (ii) inultracentrifugation; (iii) in the purificationscheme combining these principles and pre-sented in Fig. 1, in which LPS and ECA from

nonimmunogenic strains can be separated; (iv)in the fraction insoluble in 85% ethanol (192);and (v) in retaining its RBC-coating abilityafter alkali treatment (126). These tests cer-tainly establish a close association betweenLPS and ECA, but a direct proof of covalentlinkage is still missing.Some further information has been provided

by studies of mutants of the immunogenicstrains (see "Genetics of ECA in immunogenicstrains"). The core must be free, not substitutedby 0 side chains: heat-killed suspensions of

609VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from



smooth forms of E. coli 08:(K27) or Shigellaboydii type 3+, core type R1, were not immuno-genic (128, 215). Schmidt et al. (172) prepared ge-netic hybrid strains in which the R4-type 014core was combined with 0 side chains of speci-ficity 08. Such hybrids were stated to be "atbest slightly immunogenic." The core must alsobe complete; a rfa mutant of E. coli 014 inwhich the core was incomplete, lacking 1 moleach of galactose and glucose, contained ECAin ethanol-soluble but not in ethanol-insoluble(LPS-linked) form, and its heat-killed cells didnot elicit the synthesis of anti-ECA in rabbits,contrary to the behavior of the parent strain(123).These very exact requirements for the LPS

structure suggest that linkage of ECA to LPS isenzyme mediated and covalent. The fact thatthe function of a rfaL gene, probably a translo-case, is also required for ECA immunogenicityprovides a further argument for the view thatECA is covalently linked to the LPS core, prob-ably to the terminal glucose in the same way asare the 0 side chains of smooth forms. For afinal proof of this, the linkage region should,however, be isolated and its structure deter-mined.

Interference by LPSAlthough covalent or other specific linkage of

LPS to ECA in immunogenic strains results inimmunogenicity of ECA, it has, largely throughthe work of E. Neter and his associates,become clear that a different sort of inter-action is responsible for the lack of immunizingability in nonimmunogenic strains. This inter-action is probably a nonspecific aggregation ofECA with LPS, and this ECA can be renderedimmunogenic when fractionated away fromLPS.Suzuki et al. (192) did just this: they sepa-

rated extracts of nonimmunogenic strains ofE.coli or Salmonella into ethanol-soluble and-insoluble fractions, of which the ethanol-solu-ble fraction was immunogenic for ECA,whereas the whole extract or the ethanol-insol-uble fraction (which contained LPS) was not.Mixing the ethanol-soluble with the ethanol-insoluble fraction resulted in a marked de-crease in its immunogenic capacity.

It must be stressed at this point that whenspeaking of immunogenicity and nonimmuno-genicity, we in fact only state something ofquantitative differences, and it might be moreappropriate to talk of strong and weak immu-nogenicity. For example, in many cases a "non-immunogenic" preparation of ECA could bemade to immunize when administered

properly. Subcutaneous immunization with acrude bacterial extract containing ECA + LPS,especially with complete or incomplete Freundadjuvant, resulted in an anti-ECA antibody re-sponse (65). The difference between the immu-nogenic and nonimmunogenic forms is bestseen when weak immunization methods (intra-venous injection as opposed to subcutaneousinjection with adjuvant) are used. And even the"immunogenic" or "strongly immunogenic"ethanol-soluble ECA is a rather weak antigen,usually requiring several injections a few daysapart to produce good anti-ECA titers.LPS is generally recognized as an adjuvant of

immune responses (137). It increases antibodyproduction to weak antigens, prevents the de-velopment or maintenance of tolerance (28),and converts T-dependent antigens to T-inde-pendent antigens (29, 30). These functions ofLPS are thought to correlate with its mitogenic-ity to B cells (5, 27, 183). Suppression of immu-nity has, however, been reported in certainspecific instances (2, 137, 177).The inhibitory effect of LPS has been mostly

studied in rabbits, partly because mice respondvery poorly to ECA immunization (see "Preva-lence of antibodies to ECA"). The necessity ofusing rabbits has obvious disadvantages, sincelarge-scale immunizations are not possible. Anexact quantitation of the parameters ofthe LPSeffect is missing, partly because of this andpartly because of difficulties of quantitatingECA. The immune response has been measuredas titers in indirect hemagglutination, which isnot a very exact measure either. And althoughLPS can now be obtained in fairly pure form,special care must be taken to obtain it free ofECA (the best method probably being the ECApreparation method of Fig. 1, which gives ECA-free LPS as a water-precipitated fraction, G1-S), and this has usually not been done.

In summary, ECA coaggregated with LPS bymixing in the test tube, coextracted from bacte-ria or present together on heat-killed bacteria,gives rise to a minimal anti-ECA response(HA), if any at all, when injected intrave-nously. Nevertheless, it has had an apparentpriming effect on the animal, as seen when asubsequent intravenous injection of ECA (freeof LPS) results in a secondary type of response.After this single booster injection, the antibodylevel rises more rapidly and reaches a muchhigher peak (e.g., HA titer of 1:6,000 comparedwith 1:34) than without the priming (39, 147).Both inhibition of the response by LPS and thebooster effect can also be seen by measuring thenumbers of lymph node or spleen cells synthe-sizing anti-ECA and giving hemolytic plaques

BACTERIOL. Rzv.


http://mm

br.asm.org/

Dow

nloaded from



with RBC sensitized with ECA (39). The inhibi-tion cannot be due to direct toxicity ofLPS sinceit is not seen ifLPS is injected at a separate sitefrom ECA (194).The quantitative and kinetic aspects of the

system were studied by Whang and Neter (222).The extract of E. coli 0111 could be used forpriming over a 1,000-fold range of dilutions.Similarly, the booster antigen (= the ethanol-soluble fraction) could be used undiluted ordown to a dilution of 1:1,000. LPS was effectivefor priming at 100 pug or higher doses. Theinterval between the priming and booster injec-tions had to be 2 or more days to produce thesecondary response. The strength and kineticsof the secondary response remained unalteredwhen the interval was increased up to 8 weeks(the maximum tested). The specificity of thepriming was tested by using such ECA-nega-tive organisms as Pseudomonas or Staphylo-coccus for the primary immunization; a subse-quent injection of ethanol-soluble ECA causedonly a minimal response (= weak primary re-sponse to a single suboptimal dose of antigen).Many LPS preparations have been tested for

their immunosuppressive effect towards ECA.The LPS present in the crude extracts or intheir ethanol-insoluble sediment, as well asseveral purified LPS preparations, act as im-munosuppressants (194, 195, 222). The lipid Apart of LPS, which is responsible for the manybiological effects of endotoxin (= LPS), was alsofound to be responsible for this specific immu-nosuppression (212, 222). If the lipid A in someexperiments (194) was less active than LPS, thelipid A was probably poorly soluble.The ethanol-insoluble LPS from the immuno-

genic strain E. coli 014-probably LPS linkedwith ECA-was immunogenic for ECA, i.e.,did not show the immunosuppressive effect(192, 195). When LPS from other sources wasmixed (coaggregated) with it, its capacity toelicit the production of anti-ECA was reduced(195).

Importance of the Form of AggregationBesides LPS, several other materials have

been found to have an immunosuppressive ef-fect toward ECA immunization. Cardiolipin in54- to 540-,ug/ml amounts, mixed with ethanol-soluble ECA, prevents the primary responsebut not immunological priming to ECA (226). Italso has the same effect on the LPS-linked ECAof E. coli 014. Whole serum from various ani-mal species (227); membranes of a Mycoplasma(13) and chlorphenesin (228) (3-p-chlorphenoxy-1,2-propanediol, a compound shown to causeimmunosuppression when injected together

with sheep RBC [12]); gangliosides; methyl pal-mitate; and detergents such as Triton X-100and Tween 20 (1) have a similar effect, in manycases dependent on the dose of the immunogenand/or inhibitor. Several other lipids (glyceryltrioleate, sphingomyelin, and cerebrosides)have been reported to be without effect on theimmunogenicity of ECA (1). Cholesterol wasshown to prevent the immunosuppressive effectof LPS (crude bacterial extracts) (225) and ofTriton X-100 (1).Most of the above-mentioned immunosup-

pressants are known to be membrane activecompounds, i.e., lipids or ampholytes. As suchthey can be expected to complex readily withcellular membranes on the one hand and withlipidic or amphiphilic compounds (such as ECAmay well be) on the other (75). The first mecha-nism may affect early events in the immuneresponse (stimulation of the precursors of anti-body-forming cells probably occurs as a mem-brane reaction), whereas the second may beexpected to modify the antigenic properties ofECA. This modification could involve maskingof antigenic groups by steric hindrance or alter-ation of the solubility properties and therebythe aggregate size or the conformation of ECA.Some experimental data bear on these possi-

bilities. LPS (195), serum (227) and cardiolipin(226) were shown to interfere only minimally ornot at all with the ability ofECA to coat RBC orto react with anti-ECA in HAI. LPS or lipid Ainhibited antibody-enhanced phagocytosis of la-tex particles sensitized with mixtures of theseand ECA, although they did not reduce theamount ofECA available on the latex particlesto adsorb anti-ECA antibodies (37). These datasuggest that the antigenic groups of ECA wereexposed to soluble antibodies and that the in-hibitory action could take place at a stage ofactive cell surface events such as occur in phag-ocytosis (or stimulation of cells).The importance of the particle size of the

ECA-containing immunogen was best demon-strated by Whang et al. (208). The ethanol-soluble ECA fraction was further divided bycentrifugation (23,500 x g for 20 min) into apellet (larger aggregates) and the supernatant.Of these, only the pellet was a good immuno-gen, although both the resuspended pellet andthe supernatant contained equal amounts ofECA measured by HAI activity. Reducing theparticle size of the ethanol-soluble fraction bypassing it through membrane filters reducedand finally abolished its immunogenicity (208).A change in particle size or in conformation

could be responsible for the effects of heating,freezing, and alkali treatment. Heating (at

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from



1000C for 1 h) of the immunogenic ethanol-soluble ECA reduces its immunogenicity,which can be restored by repeated freezing andthawing (213). This freeze-thaw procedure alsorestores the lost ECA immunogenicity of mem-brane filtrates (208). Alkali treatment (0.1 NNaOH for 1 h at 3700) also abolishes ECAimmunogenicity, which cannot be restored byfreezing and thawing. This treatment probablyremoves part of the palmitic acid and therebyreduces the hydroghobicity and aggregationtendency of ECA. LPS, treated similarly byheat or alkali, also loses its immunogenicity,and poor immunogenicity of phenol-water-ex-tracted LPS preparations can be improved byrepeated freezing and thawing (148, 213). Vary-ing mitogenic efficiencies of different LPS prep-arations can be similarly explained (184). Thesensitivity ofECA and LPS to these procedureswas similar but not exactly the same, except forthe LPS-linked ECA of immunogenic strains.All these nonimmunogenic preparations never-

theless primed the animals to give a secondaryresponse to either ECA or 0 antigens (LPS)after a booster injection of immunogenic mate-rial (148, 208, 213).

Intimate contact between ECA and the inhib-itory compound is necessary for the inhibition.Both inhibitor and ECA must be injected simul-taneously at the same site (1, 194, 227). A more

clear-cut inhibition is obtained if they are

mixed thoroughly, in ethanol (in which they atleast partly solubilize) for example, and incu-bated (allowing time for rearrangement) beforeinjection. It is this requirement of intimate con-

tact that led Neter et al. to propose the term"antigen-associated" immunosuppression (227)for this effect.

If it is more the form of antigen presentationthan the presence or absence of a specific inhib-itor that determines the immunogenicity ofECA, some apparently contradictory findingsfall in place. Among these are the following. (i)

The suppression of ECA immunogenicity isseen most clearly with heat-killed cells,whereas living bacteria or bacteria killed withformalin, merthiolate, or phenol do elicit theproduction of anti-ECA antibodies (214). (ii)Several fractions of disrupted (French press)bacteria were isolated from nonimmunogenicstrains without prior heating (36). Most ofthesefractions, as well as the crude disrupted mate-rial, elicited a good anti-ECA response wheninjected intravenously into rabbits. Heating re-duced their ECA immunogenicity only to a mi-nor extent. Significantly, they all containedLPS (as judged by their ability to evoke an anti-O antibody response), which apparently did notsuffice to suppress the immunogenicity of ECA

in this situation. (iii) The physicochemicalstate of ECA and LPS as well as their relativeamounts can easily be expected to be abnormalin spheroplasts, which lack the rigid cell wall;this could account for the results (whole sphero-plasts immunogenic, their supernatants or ly-sates not) obtained with these cells (220). (iv)RBC or fibroblasts coated with ECA from crudeextracts of nonimmunogenic strains are immu-nogenic for ECA in rabbits (58, 149).The immunosuppression by LPS and other

similar immunosuppressants has also beenshown to affect the "common antigen" of gram-positive bacteria (226, 227). As discussed below("Other cross-reacting or related antigens"),this antigen is ill defined but may be a teichoicor lipoteichoic acid. Teichoic acids have beenshown to have aggregation tendencies with sev-eral compounds, e.g., polysaccharides and albu-min (41). In any case, the immunosuppressionwith priming is very similar to that seen withECA coaggregated with LPS. The 0 antigen(the antigenic property of the polysaccharidepart of LPS) is not affected by addition of LPSfrom the same or other source or by addition ofserum or detergents (1, 39, 147, 195, 213). Wheninjected intravenously as a heat- or alkali-treated preparation, it results, however, in im-munosuppression with priming. The above rea-soning ofreduced aggregate size as the basis forthis immunosuppression could fit the case ofLPS too, although the requirements for a suita-ble form of presentation may be different forLPS than for ECA. We are not aware of otherantigens that have been tested for this effect.The observations described above speak for a

decisive role of the proper form of ECA forimmunogenicity. The aggregation tendency isconsistent with the probable chemical structureof ECA as an amphiphilic compound (75, 231).However, it is not possible to define the exactrequirements for good immunogenicity. In sev-eral cases, it is clear that the small size of theaggregates is responsible for poor immunogen-icity. Whether LPS and other similarly actingcompounds always reduce particle size orwhether they also or under other circumstancesmodify the conformation in other ways, whichare incompatible with immunogenicity, cannotbe decided at this time. An experimental possi-bility may be offered by the recent work ofGalanos et al. (49, 50), showing that differentsalt forms (e.g., triethylamine, sodium, cal-cium) of LPS have very different solubilityproperties. The size of the LPS aggregates-vesicles or micelles-is altered, and it is verypossible, although not studied in detail yet,that the affinity of the LPS to animal cell mem-branes or to ECA could be altered as well.

BACTERIOL. REzV.


http://mm

br.asm.org/

Dow

nloaded from



Immunological Tolerance?

Another obvious possibility to account for thedescribed immunosuppression is the well-docu-mented phenomenon of immunological toler-ance or paralysis (42, 80). Reduction of the mo-lecular size reduces immunogenicity of mostantigens, both proteins and polysaccharides.The smaller molecules (below a molecularweight of 10,000) of fairly monotonous polysac-charides such as glucans or levan are nonim-munogenic, but produce a state of tolerance orunresponsiveness to further injections of theantigen even in its immunogenic form. Thetolerance may fade with time, but requiresmuch longer periods, even weeks in mice, thanthe intervals used before the booster injectionsof ECA. Also, the animal species plays an im-portant role in determining the outcome- animmune response or not- of immunization(e.g., reference 66).Tolerance production is favored by weak im-

munization, such as intravenous injection ofantigen (where the ECA immunosuppression isbest seen). The LPS preparations, with whichthe ECA immunosuppression was demon-strated, were obtained by phenol-water extrac-tion and contained ECA. It is possible thatcoaggregates of ECA and LPS are especiallyfavorable for tolerance induction, even withvery small amounts of ECA. One experimentalobservation (215) speaks strongly for a tolero-genic mechanism: whereas LPS isolated fromECA-positive strains of Salmonella producedthe above-described immunosuppression toECA, LPS from ECA-negative mutant strainsdid not. This finding is difficult to explain onthe hypothesis of aggregate size but easy on thehypothesis of tolerance: the LPS from ECA-negative strains is not expected to have anyECA immunodeterminants for tolerance induc-tion. However, the mode of action of the rfe-rffmutations responsible for the ECA negativityof the strains is not known. Therefore it ispossible that these mutations also affect LPS,for instance in a way that would alter its confor-mation or affinity to ECA or the size of theaggregates produced.Experiments using whole bacteria (215) also

contradict the simple hypothesis that the ag-gregate size is the sole basis ofimmunosuppres-sion. When immunogenic E. coli or Shigellaand nonimmunogenic Salmonella were injectedtogether at the same site, anti-ECA productionwas suppressed. This did not happen if the twokinds of bacteria were injected in separate earveins. Under these conditions, most of the im-munizing ECA would be expected to be boundto the bacterial cells and not affected by ECA or

LPS on other bacteria. On the other hand, iftherole of the Salmonella bacteria was tolerogenic,it is difficult to see why they would not have thesame effect when injected at separate sites.What speaks most strongly against the toler-

ogenic effect of LPS-ECA mixtures is the factthat no tolerance to a subsequent injection ofimmunogenic ECA is produced, but instead theanimals are primed to give a heightened "sec-ondary" response. In conventional tolerance,some kind of priming is sometimes seen witheither protein or polysaccharide antigens onrecovery from the tolerant state, and called"overshoot" (42, 181). The time required for thetransition from unresponsiveness to an in-creased response is, however, longer than themere 2 days required before a booster dose ofECA.

In several cases, a T-cell-dependent antigencaused an apparent priming effect in animalsthat were lacking functional T cells and weretherefore unable to respond with antibody pro-duction (32, 164, 176, 180, 196). These resultssuggested that the antigen had neverthelesssucceeded in stimulating B cells to proliferationand production of memory cells. A comparablesituation was described by Chiller and Weigle(28) in mice tolerant to a T-cell-dependent anti-gen (human gamma globulin). At a certainstage of recovery, the tolerance could be brokenby injecting the antigen in a T-cell-independentform (together with LPS), and this resulted in asecondary response type of increased antibodysynthesis. They suggested that because of dif-ferent sensitivities of T and B cells to a toleriz-ing antigen, it may be possible that at a certaindose T cells become tolerant whereas B cells areimmunized (primed), although no antibody pro-duction takes place since it would require T-cellparticipation. If these B cells then meet animmunogenic, T-cell-independent stimulus,they respond with rapid antibody production.These examples have attractive similarities

to the ECA case. However, so far there are nodata to demonstrate the presence of memorycells to ECA, and the fact that anti-ECA (mea-sured, it is true, mainly by HA) even after thiskind of priming is almost exclusively 19S (224)speaks against memory. Nor do we knowwhether ECA or coaggregates of ECA and LPSwould stimulate B cells in rabbits, although thefact that LPS alone can be a T-cell-independentantigen in mice suggests that this could be thecase. Furthermore, little is known of rules ofimmunity with such amphiphilic materials asECA is likely to be. True tolerance to ECAcould never be demonstrated, for example byvarying the dose of the antigen (208). Unfortu-nately, the poor immunogenicity of ECA in

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from



mice precludes experiments with T-cell-defi-cient mice (166) and the C3H/HeJ strain, whichis unresponsive to many effects of LPS (191),experiments which would help to clarify theseproblems and to demonstrate how far the ob-served "antigen-associated" immunosuppres-sion represents a new immunological principle.

In view of the many seemingly irreconcilableobservations, we believe +hat both theories mayturn out to be partly correct. The form of aggre-gation (or conformation) may be decisive insome situations, and some sort oftolerance mayoperate in others.

ECA IN THE BACTERIAL CELLVery little is known of the localization of

ECA in bacteria. At least parts of it must resideon the surface of the cell; some of it is veryeasily eluted into the surrounding medium(102), and anti-ECA antibodies attach to intactbacteria, as shown by fluorescence elicited bysubsequent treatment with fluorescent anti-gamma globulin (8). The "ECA-dependent" flu-orescence is less intense than that caused byLPS-anti-LPS, which may suggest that ECA isless exposed than is LPS or (which is more

likely in view of the large amounts of LPS inthe outer membrane) that there is less ECApresent. However, the specificity of the IF reac-tion for ECA is not finally established.Whang and Neter (220) have looked for ECA

in glycine-induced spheroplasts of Salmonellatyphi and have detected some activity by HA,by HAI, and by immunization. Unexpectedly,they could demonstrate ECA only in sphero-plast suspensions, not in the supernatant or inlysed suspensions, perhaps suggesting thatECA was present in reduced amounts. Thiscould be taken to support the localization ofECA in the outer membrane, which is partlydisintegrated in spheroplasts, but the evidenceis weak and circumstantial.The fact that anti-ECA cannot kill bacteria

in the presence of complement may suggestthat ECA is not embedded in the outer mem-

brane in the way LPS is. On the other hand,antibodies towards capsular polysaccharidescan be bactericidal (66), suggesting that themolecule mediating a bactericidal reaction doesnot need to be an integral component of thebacterial membrane. (It must be admitted,though, that the location of the capsular poly-saccharides is not well defined either. However,the absence of any lipid-soluble part in thepolysaccharides makes it unlikely that theywould be embedded in the predominantly lip-idic membrane.)Domingue and Johnson (36) and Johnson, Do-

mingue, and Klein (submitted for publication)have made the most direct attempts at localiz-ing ECA. They used Salmonella strains, E. coli014, and Klebsiella and isolated different frac-tions, including the outer membrane and thecytoplasmic membrane. The ECA activity ofE.coli 014 was more clearly present in the cellenvelope fraction, but in most other cases allfractions, also the cytoplasm, contained ECA.It seems likely that the location of the LPS-linked ECA in E. coli 014 followed the locationof LPS, whereas the haptenic ECA of the otherstrains was solubilized in the process and con-taminated all fractions.

CLINICAL IMPLICATIONSMany ofthe bacteria of the family Enterobac-

teriaceae are important pathogens of man, ani-mals, and plants. In some disease groups (uri-nary tract infections) they are the main sourceof infection, and in others they are very com-mon causative organisms (all infections of theperitoneal cavity and its organs). A componentcommon to all Enterobacteriaceae, therefore,would have great promise as a means of devel-oping generally applicable diagnostic methodsfor these diseases or for measuring immunity tothem. Questions of the possible role of ECA inpathogenesis are obvious. Is it a virulence fac-tor (by direct toxic action, by protecting bacte-ria from host defenses, by affecting tissue affin-ity of the bacteria)? Do antibodies to ECA pro-tect the host animal (man)? Are there antigeni-cally similar components in the host tissuesand, if so, how does this affect the acute infec-tion or its sequelae?

Immediate Biological Effects of ECAEndotoxin (LPS) has well-documented toxic

effects, which can be demonstrated in the wholeanimal and partly also at the cellular level andwhich profoundly affect the course of infectionscaused by gram-negative bacteria (89).Kunin (100) tested his DEAE-purified ECA

from E. coli 014 for pyrogenic and lethal effectsin rabbits; at a dose of 500 ug it produced feverbut no other harmful effects, whereas LPS waslethal at the tested dose of 25 ug. The sameobservation was made by Johns et al. (85) withtheir picric acid-Sephadex G-200-purified mate-rial from S. typhi. It produced fever at a dose of250 but not 100 jig; LPS was stated to be about1,000 times more active as pyrogen. This puri-fied material was also 1,000-fold less activethan LPS in the limulus coagulation assay,which is another, apparently rather unspecificbut extremely sensitive, test for endotoxin ac-tivity. It may be remembered that in both these

BACTERMOL. REV.


http://mm

br.asm.org/

Dow

nloaded from



cases the ECA preparations were not in the"native" state a shown by their lack of RBC-sensitizing capacity.

In a mouse toxicity test (material injectedintraperitoneally into mice kept at 370C), theethanol-soluble ECA (192) was nonlethal at adose of 1,000 jug, whereas LPS still showed 20%lethality at a dose of 1 pAg (96). The same prepa-ration was 100- to 1000-fold less efficient thanLPS in several other endotoxin assays: promo-tion of a Shwartzman-like reaction by epineph-rine in rabbits, immediate production of non-specific resistance to Salmonella infection inmice, and direct cytotoxicity on monolayers ofguinea pig peritoneal macrophages (96).

In all these tests, the doses of ECA requiredto produce any of the pyrogenic or other toxiceffects were so high that the effects could beaccounted for by contaminating LPS present inthe ECA preparations. A different toxic mecha-nism has been suggested for ECA: in vivohemolysis of RBC coated with ECA releasedduring infection (136, 193), but we have no databearing on this possibility in human infections.

Is ECA a Virulence Factor?A role ofECA in the pathogenic properties of

enteric bacteria was recently demonstrated.ECA-positive and ECA-negative (rfe-) deriva-tives of the mouse pathogen S. typhimuriumwere injected intraperitoneally in mice (202).Whereas the LDO (dose lethal to 50% of the testanimals) of ECA-positive strains was 104, theLD50 of their ECA-negative sisters (all strainssmooth, 04,12, constructed by conjugation)was 10-fold higher, indicating that ECA isneeded for the expression of the full pathogeniccapacity of the bacteria.The mode of action of the ECA as a virulence

factor has also been studied, so far with nega-tive results (202; unpublished observations).There was no difference in the growth ratebetween ECA-positive and -negative sisterstrains. After intravenous injection, both var-iants were removed slowly from the blood-atest presumed to measure phagocytosis by thereticuloendothelial system. In this test nonvi-rulent rough (R) bacteria disappear more rap-idly than virulent smooth (S) bacteria (98).However, it is likely that the test is not sensi-tive enough to detect small differences in bacte-rial susceptibility to phagocytosis; e.g., no dif-ference was seen between S bacteria, whichhad only small differences in their LPS butdiffered approximately 10-fold in their viru-lence (200, 201).

Conversely, low content of ECA (measuredby HAI) has been reported as a characteristic of

E. coli strains causing diarrhea in younginfants in Mexico, as compared with strainsisolated before or after the diarrheal episode(24a, 98a). This was equally true of entero-pathogenic and nonenteropathogenic serotypes;a correlation with enterotoxin production hasnot been reported.More studies of the role ofECA in pathogen-

icity of enteric bacteria are obviously needed.As a surface component it can be expected toreact with the defense mechanisms of the hostand thereby influence the outcome of infection.The possible antigenic interrelationships ofECA and some tissue components (see below)suggest a role in specific tissue affinity. Thepossibility of preparing ECA-negative mutantsof strains to be studied should now make suchstudies more meaningful.

Prevalence of Antibodies to ECAA low level of anti-ECA antibodies seems to

be present in normal human sera. Kunin (99-101) measured hemagglutinin titers of a largenumber of human sera using RBC sensitizedwith several E. coli extracts; this would meas-ure both anti-0 and anti-ECA antibodies.When the sera were absorbed with an ECApreparation (E. coli 014 extract), the titers tomany other strains were reduced by a factor of 2to 4, suggesting that as a rule the sera con-tained both anti-ECA and many different anti-0 activities. The presence of anti-ECA deducedby this means in various age groups would beparallel to the presence of anti-0. There was alow level of anti-ECA in cord blood and verylittle at 2 to 6 months of age, with increasingtiters above that age. The levels in cord bloodwere much lower than in the maternal serum.In colostrum, both anti-O and anti-ECA levelswere high (100). Similarly, Whang and Neter(218) found anti-ECA in the cord blood of 8 outof 18 infants tested.

Pools of sera from healthy blood donors wereshown to regularly contain anti-ECA (again bycomparing heterologous hemagglutinin titersbefore and after adsorption by ECA-containingextracts of several enteric bacteria or a non-ECA-containing extract of Staphylococcus au-reus) (218). For example, an initial titer of 1:40was reduced by ECA-absorption to 1:10. Severalgamma globulin preparations from Japan,South Africa, India, and the U.S.A. could like-wise be shown to have anti-ECA as well as anti-0 antibodies to several indicator bacteria (218).

Pools of sera from horses, cattle, dogs, andpigs all contained appreciable amounts of anti-ECA antibodies (101). Normal rabbit serum, onthe other hand, was practically free of anti-

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from



ECA activity (tested at a dilution of 1:10 withseveral E. coli preparations) (101). Since itseems reasonable to believe that the anti-ECAactivity in normal sera is the result of continu-ous immunization caused by products of intes-tinal bacteria, the absence of anti-ECA in rab-bit serum may reflect the predominantly gram-positive flora of the rabbit gut.

Active immunization with a suitable immu-nization schedule (see above, "Immunogenic-ity") leads to production of relatively high lev-els of anti-ECA antibody, although animal spe-

cies and strains differ in responsiveness.Most of the studies with ECA have used rab-

bit sera. Fractionation of rabbit anti-ECA sera

by filtration through Sephadex G-200 has beenperformed (224). Several sera obtained with dif-ferent immunization schedules showed anti-ECA (HA) in the 19S fraction only. A "late"serum, taken at day 14 of a secondary responsewhen antibody levels were maximal, had 90%of its anti-ECA activity at 19S, the rest distrib-uted among a large number of fractions. Allantibody in all these sera was susceptible to 2-mercaptoethanol. No hemagglutinating (orblocking) antibodies were detected in the 7Sfractions. The anti-O antibodies (measured bybacterial agglutination) in the same sera were

found in both the 19S and 7S fractions, and allwere susceptible to mercaptoethanol. Whenpregnant rabbits had been immunized, consid-erable amounts of anti-ECA were found in thepooled sera of the fetuses at days 27 to 28 ofpregnancy. Again, this anti-ECA was mainly19S. Further proof of transplacental transfer ofanti-ECA antibodies was obtained by passiveimmunization of the mother, which resulted inmeasurable anti-ECA titers in the newborn butno anti-ECA titer in the amniotic fluid.Mice respond relatively poorly to ECA, be it

the LPS-linked, immunogenic form in E. coli014 or the ethanol-soluble, non-LPS-linkedECA (56, 58, 62, 63; E. A. Gorzynski and E.Neter, Bacteriol. Proc., p. 86, 1968). The bestresponder among mouse strains was C57Bl/6HA; the CBA/St strain was fairly good,whereas DBA/2Jax and outbred Swiss albinomice were very poor responders. In mice, thenon-LPS-linked ECA was usually a better im-munogen, eliciting higher titers and a responsein more mice than did an E. coli 014 extract.Mice also differed from rabbits in not showing a

secondary response (booster injection ofethanol-soluble ECA after a rest period of 52days).

Contrary to this experience, McCabe andGreely (112) obtained good anti-ECA titers inthe mouse strain CH1 after immunization withheat-killed E. coli 014. The antibody was pres-

ent in both 7S and 19S fractions. The differencemay be related to their use of an extract ofProteus rettgeri strain 6572 to sensitize the RBCfor HA (see "Serological methods").Guinea pigs have been immunized by both E.

coli 014 extracts and ethanol-soluble ECA inFreund adjuvant and tested for cutaneous de-layed hypersensitivity and antibody production(132). No or only minimal anti-ECA activitywas detected in the sera 9 days after a singledose of immunogen. A booster on day 20 re-sulted in low anti-ECA titers in all animals. Adelayed-type hypersensitivity reaction was elic-ited on day 9 as well as on day 30 by intrader-mal injection of various ECA preparations. APseudomonas extract not containing ECA alsocaused erythema and induration in the animalsimmunized with ethanol-soluble ECA, and itwas not completely clarified whether the de-layed hypersensitivity reactions were caused byECA or by small amounts of LPS contaminat-ing the ECA preparations.Some immunizations have been performed in

humans with ethanol-soluble ECA from E. coli0111 (64). The anti-ECA titers (HA) in preim-munization sera ranged from below 1:10 to 1:40;those after immunization ranged from 1:160 to1:1,280 (nine out often immunized subjects; onedid not respond). The post-immunization serahad also more opsonic activity than did the pre-immunization sera (203).

ECA and Anti-ECA Antibodies in Relation toDisease

The possible role of anti-ECA antibodies invarious infections usually caused by entericbacteria has been tested. The general impres-sion has been rather negative: the levels ofanti-ECA are low, and only a fraction of thepatients respond with a rise (Table 5). The titervalue may, however, not be very significant;only HA has been used, and therefore (almost)only IgM antibody has been measured. In thecase of anti-0, only IgG antibodies have beenshown to be protective in human disease (114).When anti-O titers (to the 0 antigen of thecausative organism) have been measured in thesame samples as ECA, they have been consist-ently higher (4, 33, 114, 141, 146, 204, 218).

Diseases in which anti-ECA responses havebeen found only occasionally, and at a low titer,are enteritis (Salmonella or E. coli), bactere-mia, and acute urinary tract infections (Table5). The only conditions in which high anti-ECAlevels and/or a rise in anti-ECA have beenconsistent features are shigellosis (33, 218), per-itonitis (141, 146), and chronic urinary tractinfection (4, 167, 218). In these, over 50% of thepatients had high anti-ECA titers. In relation

BACTERIOL. REV.-


http://mm

br.asm.org/

Dow

nloaded from



TABLE 5. Anti-ECA in human infections caused by gram-negative bacteria

Infection No. of patients Anti-ECA Anti-ECA in controls Reference

Urinary tract Chronic, 6 1:80-1:320 in 5/6 1:40 218Acute, 13Acute, 10 below 1 yr oldand 10 above 1 yr old;chronic, 2

Acute or chronic, 50

Children, 111?

152

1:160, no rise0-1:81:8-5121:12818% -1:160 or more16% -fourfold rise1-1:160 or moreNo difference to con-troP

Elevated in 63%

0-1:16

1:40

Not elevated

204

4

33

141113

167

Fourfold rise to 1:16024% - 1:160 or more16% -fourfold rise7-below 1:400-above 1:80

56%-1:160 or more38% -fourfold rise

Bacteremia 3108 with fatal underly-ing disease, 65 ofthese followed

98 without fatal under-lying disease, 43 ofthese followed

Peritonitis 27

1 (rough E. coli, core

type R1)

No changeInitial titers low29% -fourfold rise

Initial titers not differ- 1:160-1:2,56(0ent from controls

36% -fourfold rise

8%-below 1:4055% -1:160 or more82% -fourfold riseRise from 0 to 1:640

aCells sensitized for HA by an extract ofProteus rettgerii 6572; for discussion see "Serological methods."

to shigellosis, one is reminded of the frequencyof R mutants with the R1 type of core in Shi-gella strains, especially Shigella sonnei (44,127)-such R strains belong to the strains thatare "immunogenic" for ECA. In the other twoconditions a long-continuing or especiallystrong antigenic stimulus is probable. Over60% of peritonitis patients showed a fourfold or

greater rise in antibody (141), indicating an

active process of immunization. Over 80% ofpatients with chronic or repeated urinary tractinfections (4, 167, 218) had high and persistentanti-ECA titers. In view of the clinical impor-tance and frequency of this condition, this diag-nostic possibility should be studied further.ECA was detected by IF in kidney tissue in

confirmed pyelonephritis (six cases out of six)and in six cases out ofseven ofpyelonephritis inwhich bacteria had not been cultivated, but notin patients with other types of chronic renaldisease (7). However, other workers could findECA in kidney tissue in only one out of ninecases of chronic pyelonephritis and three out of

five cases of acute pyelonephritis (178). Inexperimental pyelonephritis of the rat, ECAcould be demonstrated in acute infection butnot later on (197a). The reasons for the differentfindings are not apparent; the techniques andantisera were very similar, and proper IFcontrols with bacteria (8, 178) or tissue fromexperimental animal infections (9, 178) were

performed (however, see discussion of the IFmethod in "Other serological methods").McCabe et al. (113, 115) have studied a large

series of patients with bacteremia comparedwith healthy blood donors in an attempt to findcorrelations between anti-O, anti-R, and anti-lipid A and anti-ECA antibodies with the occur-

rence or prognosis of the infection. The bactere-mia patients had lower anti-ECA titers thanthe controls, but this difference was almost en-

tirely accounted for by the approximately halfof the patients who had an ultimately fatalunderlying disease. A rise (fourfold or more) inanti-ECA was seen in one-third of the patientsirrespective of the basic disease or the species of

150

Salmonellosisor other en-teritis

Shigellosis

10

50

1:401:40

21833

141

1:40 33

33

113, 115

141

146

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from



TABLz 6. Presence (+) or absence (-) ofECA in extracts of various bacteria as tested by hemagglutinationand/or inhibition of hemagglutinationa

Name ECA strains Reference

Gram-negative facultatively anaerobic rodsFamily EnterobacteriaceaeEscherichia

"alcalescens-dispar "ShigellaEdwardsiellaSalmonella

S. arizona (=Arizona)KlebsiellaEnterobacterCitrobacter (including Levinea)SerratiaProteusProvidenciaYersiniaPectobacteriumErwinia (herbicola, carotovora, amylovora)Erwinia chrysanthemi

Family VibrionaceaePlesiomonas? (Vibrio? Aeromonas?)shigelloidesAeromonasVibrio

Genera of uncertain affiliationFlavobacteriumHaemophilusPasteurella

Gram-negative aerobic rods and cocciFamily PseudomonadaceaePseudomonas

Genera of uncertain affiliationAlcaligenes faecalisBrucellaBordetellaFrancisella tularensis

Other gram-negative bacteriaAcinetobacter calcoaceticusRhodopseudomonas capsulata

Gram-positive bacteriaStaphylococcusStreptococcusSarcinaBacillus

4

4

+b 166 8, 102, 106, 229+ 4 106+ 44 102, 106, 229+ 33 106, 229+b 73 102, 106, 229+ 12 229+ 26 8, 102, 106, 229+ 56 8, 34, 102, 106, 229

33 106, 22926 34, 106, 22950 8, 102, 106, 229

+ 19 106, 229+ 70 106, 117+ 15 34, 229

24 102, 106- 6 106

+ 15 106, 210- 46 106, 210- 17 106

- 19 34- 1 106- 6 106, 117

- 14 8, 102, 106, 117

_ 4_ 4_ 2_ 1

_ 1- 1

_ 2_ 2_ 1_ 2

106102, 106, 117102, 106117

117126

102, 216102, 216102102, 216

a Classification according to Bergey's Manual (23).b Occasional ECA-negative strains have been reported: one strain ofSerratia marcescens (102), sixProteus

strains (P. vulgaris, morgani, rettgeri, mirabilis) (8, 106), and mutants ofE. coli and Salmonella (119, 120;G. Schmidt, H. Mayer, and P. H. Makela; J. Bacteriol., in press; P. H. Mikela, G. Schmidt, H. Mayer, H.Nikaido, H. Y. Whang, and E. Neter, J. Bacteriol., in press) among many positive strains of the samespecies.

enteric bacteria (E. coli, Klebsiella group, Pro-teus, or Salmonella) causing the infection. Thelevel of anti-ECA was not correlated with prog-nosis when patients were examined in separategroups depending on whether or not they had afatal basic disease. By the same methods, poorprognosis (death or development of shock) couldbe correlated with low anti-O titer (immuno-globulin G only), low anti-R (of the Re type ofbasal core, Fig. 4; see also "Other cross-reactivi-

ties and related antigens") and less clearly lowanti-lipid A. Despite the careful analysis madeand the clearly important division of the pa-tients in groups of basically different capacitiesto resist infection (those with versus thosewithout an underlying fatal disease), these con-clusions of the correlation of the different anti-body levels with prognosis should probably notyet be accepted as the final truth. First, thecorrelations are not of very high level of signifi-

BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from



cance, and larger groups are still needed- non-homogeneity of the patient group has, as dem-onstrated, a large influence on the prognosis.Second, the use of a Proteus rettgeri extract tosensitize the RBC for HA (113) has not beencompletely correlated with ECA standards (seeabove, "Serological methods"). And third, sincea correlation of anti-O with prognosis in humanbacteremia could be shown only when immuno-globulin G antibodies were examined (114), al-though anti-O has been repeatedly shown toprotect in experimental infection in mice (95,107, 163), it may be questionable to study onlyHA titers (predominantly, at least, immuno-globulin M) of the other antibodies.

Anti-ECA in Experimental InfectionThe possible protection afforded by ECA im-

munization has been studied in either general-ized bacteremic infection of mice, caused byintravenous or intraperitoneal injection of S.typhimurium, E. coli, orKlebsiella, or pyelone-phritis of rabbits, caused by retrograde or hema-togenous infection by E. coli, Klebsiella, orProteus mirabilis. The results in mice havebeen equivocal and those in rabbits have beenmore promising, but it is not at all clearwhether this difference is related to the differ-ent infection types or to the animal species,mice being much poorer responders to immu-nization with ECA than rabbits (see above,"Prevalence of antibodies to ECA"). In somecases, an ECA-negative Pseudomonas strainwas used as control challenge (40), increasingthe significance of results obtained. It can beargued, however, that Pseudomonas differsfrom enteric bacteria in so many ways that itcannot be regarded as a valid control. ECA-negative mutants are now available, and theyshould provide better controls.

Gorzynski et al. (56, 58, 63; Gorzynski andNeter, Bacteriol. Proc., p. 86, 1968) used miceimmunized with E. coli 014 or the ethanol-soluble ECA of E. coli 0111 and challengedwith S. typhimurium. Both C57Bl/Ha6 mice,shown to be relatively good at producing anti-ECA, and poorly responding Swiss albino mice(62) gave almost identical protection resultsafter intraperitoneal immunization (63; Gor-zynski and Neter, Bacteriol. Proc., p. 86, 1968);anti-ECA was not recorded. RBC (of horse ori-gin; mouse RBC did not work) sensitized withECA could also be used for immunization (58).Passive immunization with rabbit anti-ECAgave similar results (56). The protection wasnever complete. A significant difference couldbe demonstrated between the ECA-immunizedand control animal (receiving normal rabbitserum [56] or extracts of Staphylococcus au-

reus or Pseudomonas) only by following thekinetics of the infection. Survival rates in thetwo groups were compared each day, a 50%comparison value indicating equality betweenthe groups. In this way the ECA-immunizedgroup got approximately 60 to 70% scores,which in most cases were significant (P <0.001).McCabe and Greely (112) immunized mice

(CFI) intraperitoneally with E. coli 014 andobtained good anti-ECA titers (measured byHA with Proteus rettgeri strain 6572 as sensi-tizing agent). Upon challenge with Klebsiellapneumoniae or E. coli (unrelated strain), theLD50 values were exactly the same in the im-munized and the control groups, whereas im-munization with the challenge organism wasclearly protective, increasing the LD50 30- toover 1,000-fold. Passive immunization with tworabbit anti-ECA sera gave no protection either,but a third serum did. This protective serumwas shown to contain anti-Re antibodies (Fig. 4and "Other cross-reactivities and related anti-gens") in addition to anti-ECA; absorption ofthe serum with ECA did not abolish its protec-tive capacity, whereas absorption with Re did.This finding strongly suggests that anti-Re an-tibodies may be important as cross-reactive pro-tecting factors. This same conclusion has beenreached from other studies with Re immuniza-tion in mice (114) and from examination of thecorrelation of different antibodies with prog-nosis of human bacteremia caused by entericbacteria (114, 115).These findings make it necessary to examine

the possible effect of Re and lipid A antibodiesin each case where protection by anti-ECA isclaimed. For instance, in the above-mentionedstudies by Gorzynski et al. (56, 58, 63; Gorzyn-ski and Neter, Bacteriol. Proc., p. 86, 1968), Reantibodies could apparently be involved. Al-though ethanol-soluble and therefore relativelyLPS-free ECA was often used as immunizingagent, it contains enough LPS to elicit the pro-duction of anti-O antibodies, although not veryefficiently (192). On the other hand, since miceare poor anti-ECA producers, the heavy im-munization necessary could have magnified theanti-LPS responses (no values are given).Domingue et al. (40, 48, 116) used the rabbit

pyelonephritis model (infection either via thebladder supported with a foreign body or intra-venously after local kidney damage caused byoxamide feeding). They immunized the rabbitswith the ethanol-soluble ECA, in one case (116)partly purified, and obtained good anti-ECAtiters. They found definite protection from pye-lonephritis irrespective of route of infection andirrespective ofwhether the diagnosis was estab-

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from



lished on the basis ofbacteremia or histopathol-ogy. Both active immunization of the rabbitsand rabbit anti-ECA antiserum (but not antise-rum absorbed with ethanol-soluble ECA coatedon sheep RBC) were effective when the chal-lenge organism was ECA-positive Proteus mi-rabilis but not when it was ECA-negative Pseu-domonas (40). A capsulated Klebsiella strain,which was not sensitive to opsonization by anti-ECA, was not sensitive to the anti-ECA-me-diated protection either. It may be worth men-

tioning that anti-ECA antibodies did not in-crease in any of the animals during the course

of their kidney infection, although anti-O anti-bodies did.

Antigenic Similarities between ECA andAnimal Tissues

Antigenic cross-reactions between microor-ganisms and host tissues are believed to beinvolved in the pathogenesis of certain autoim-mune diseases, of which the best-known exam-

ple may be rheumatic fever with cross-reactionsbetween streptococcal antigens and heart tissue(91). Sera from patients with ulcerative colitis,a chronic disease in which autoimmune phe-nomena are suspected of being important, haveantibodies reacting with antigen(s) of colon tis-sue. This was demonstrated also with colontissue of germfree animals, since otherwisemany bacterial antigens contaminating ex-

tracts of normal colon would have interferedwith interpretations (43, 104, 155, 156). Thesesera, as well as sera of rats made autoimmuneby injections of rabbit colon, also react withECA (E. coli 014 extract) in HA, HAI, and thetest for a leucocyte migration inhibition factor(24, 43). An equally high frequency of antibod-ies apparently against ECA, although theiridentity has not been finally established, was

present in liver cirrhosis (43) but was not foundin other intestinal or specific colon diseases (43,104). The cross-reacting tissue antigen was

fairly specific to colon, and resembled ECA inbeing present in phenol-water extracts of thetissue, in its heat resistance, and in its abilityto adsorb to RBC (104).These findings obviously suggest a possible

role of ECA of intestinal bacteria in triggeringan autoimmune reaction, which could lead tosymptoms such as those of ulcerative colitis.The reactivity of the cross-reacting antigen inthe migration inhibition factor test points at itspossible participation in reactions of cellularimmunity as well.

Gorzynski and Krasny (55, 57, 59) looked forpossible tissue components cross-reacting withECA in mice. They found cross-reacting mate-rial, when measured by HAI, in extracts of

livers and to a less extent of spleens and kid-neys. When a more sensitive assay, the abilityof the extracts to prime rabbits for a low dose ofimmunogenic ethanol-soluble ECA, was used,many tissues of most mouse strains seemed tocontain cross-reactive material. It was sug-gested that the presence of the cross-reactivematerial in mouse tissues could account for thegenerally poor response of mice to immuniza-tion with ECA, and also for differences betweenmouse strains (62). However, other animalshave not yet been studied in the same way,although the authors apparently plan to do so.In their tests colon was never a very goodsource of the cross-reacting material, contraryto the studies in rats and humans by Perlmannet al. (155).Holmgren et al. (79) have found cross-reac-

tivities between human kidney tissue and E.coli and suggest that these may have a role inthe pathogenesis of renal lesions in pyelone-phritis. Antisera to several (but not all) E. coliserotypes precipitated extracts of kidney tissuebut not of liver or spleen. Anti-human kidneyserum gave precipitation reactions with ex-tracts of the same E. coli types. One of theprecipitation lines showed a reaction of identitybetween the three E. coli strains, one of whichwas 014. With the data given, the possibilitythat this could be ECA is not excluded, al-though it is made somewhat unlikely by the ab-sence of precipitation with several other E. colistrains.

OTHER CROSS-REACTIVITIES ANDRELATED ANTIGENS

Various "Common" AntigensOther common or "cross-reactive" antigens

between different bacteria have been described,some of which may be identical with ECAwhereas others can be clearly differentiatedfrom it.Brodhage (20-22) described a serological

cross-reactivity probably caused by ECA, whichhe called a common, or "C," antigen. It wasdetected in enteric bacteria by indirect HA us-ing urea extracts of the bacteria. The antise-rum showing these cross-reactions was an anti-Shigella sonnei serum probably containinganti-ECA; S. sonnei cultures usually have aproportion of R mutants (44), and these havethe R1 type of LPS core, connected with immu-nogenic form of ECA.The outer membrane of all gram-negative

bacteria has a similar overall composition ofLPS, (lipo)proteins, and lipids (16, 105, 150).The LPS is responsible for the O-antigenic dif-ferences that form the basis of the hundreds of

BACTERIUL. REV.


http://mm

br.asm.org/

Dow

nloaded from



serotypes inE. coli, Salmonella ("species"), etc.(109). These 0-antigenic specificities reside inthe distally located polysaccharide chains ofLPS. The other end of the LPS molecule consti-tutes of a lipid, designated lipid A, which isprobably embedded in the outer membrane.The lipid A has, with a few exceptions, a verysimilar structure in all gram-negative bacteria(161) and could thus represent a common anti-gen. This is, however, not easily demonstra-ble-whole bacteria do not elicit anti-lipid Aantibodies, and such antibodies react with iso-lated lipid A and only to a small extent withwhole bacteria or with the complete LPS (52).The polysaccharide chains are attached to lipidA via the R-core oligosaccharide region. Thiscore is identical in all Salmonella and rathersimilar (but not always serologically cross-re-acting) in all members of Enterobacteriaceae.Best conserved in phylogeny is the "deep core,"the part closest to lipid A. This deep-core regionis represented by the Re LPS (or rather Reglycolipid) of the most deficient still viable Rmutants (rfaE or "heptoseless") (108, 109, 173,175). The isolated Re structure as well as therfaE mutant bacteria are immunogenic. Theresulting anti-Re antibodies react with Re-typeLPS and the rfaE mutant bacteria. They do notagglutinate normal smooth bacteria, but stainthem for indirect IF (233). This method gavepositive reactions with many Enterobacteria-ceae strains but also with several other gram-negative organisms. The specificity of the reac-tion was not controlled so as to exclude partici-pation of other cross-reacting antigens. Anti-lipid A and anti-Re antibodies seem to occur innormal human sera, and their levels areincreased in gram-negative infections (49,113-115, 141). Their possible role in either en-hancing (206) or, hopefully, protecting against(111, 114, 133) infection is being studied.The outer-membrane proteins have only re-

cently become amenable to detailed studies,which are showing even more similarities be-tween all these bacteria. A lipoprotein (15)linked to peptidoglycan on the one hand andexposed enough on the cell surface to be accessi-ble to antibodies on the other hand is a compo-nent apparently common to many gram-nega-tive bacteria. Rabbit antisera prepared againstwhole bacteria ofE. coli, Salmonella, and Shi-gella all have antibodies to this lipoprotein (17).It remains to be seen how far its fine structureis conserved among gram-negative organismsto support immunological cross-reactivity.

Other proteins of the outer membrane mayalso be fairly similar between different species,although differences in the molecular weightsof the major proteins are known (77). Cross-

reactivity due to such a major outer-membraneprotein has been described between separategroups (grouping being based on the capsularpolysaccharides) ofNeisseria meningitidis (47).

Cross-reactivity extending from a wide rangeof gram-negative organisms to the gram-posi-tive Bacillus subtilis and Staphylococcus au-reus detectable with an anti-Hydrogenomonasfacilis phospholipoprotein serum has been de-scribed (162). However, according to presentknowledge, the exterior surfaces of gram-nega-tive and gram-positive bacteria are so differentthat such a cross-reactivity is rather unex-pected and should be carefully ascertained; sofar no details of absorption studies have beengiven.Seltmann found an acidic, thermolabile anti-

gen common to all gram-negative bacteria andpossibly glycoprotein in nature (179). Holmgrenet al. and Kaijser (78, 79, 90) used immunodiffu-sion, immunoprecipitation, and immunoelec-trophoresis techniques in screening gram-nega-tive bacteria for common antigens. They de-tected (i) several components common betweenmany E. coli strains and (ii) an electrophoreti-cally fast-moving acidic antigen (probably thesame described by Seltmann [179]) apparentlycommon to E. coli and some Proteus, Pseudom-onas, and Neisseria meningitidis strains butnot to Staphylococci. This specific distributionseemed to preclude its identity with ECA. Alsothe fact that it was demonstrable with antiseraother than anti-E. coli 014 speaks against itbeing ECA.

Several common antigens detectable by pre-cipitation have been described in water extracts(85) or high-salt extracts (25, 26, 159) ofSalmo-nella strains. They are apparently often causedby lipoprotein antigens and separate fromECA.A phenomenon apparently similar to that

observed with ECA was described in gram-posi-tive organisms by Rantz et al. (157) and con-firmed by Neter and others (6, 138, 140, 142).Simple extracts ofmany gram-positive bacteria(including Streptococcus, Staphylococcus, Lis-teria, and Bacillus species), but not of gram-negative ones, sensitized RBC for HA in certainhuman sera or in sera of rabbits that had beenimmunized with gram-positive but not withgram-negative organisms. The non-species-spe-cific substance was characterized only very in-completely but appeared to behave like a poly-saccharide. Whether the non-species-specificsubstance was a teichoic acid has so far notbeen clarified, but teichoic acids have sinceemerged as substances shared in identical orclosely similar forms by many gram-positivebacteria and responsible for many of their

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from



cross-reactions (for review, see reference 97).The staphylococcal common antigen is suscepti-ble to the antigen-associated immunosuppres-sion by LPS and many other substances as de-scribed for ECA (see "Immunogenicity").Quite apart from these cross-reactive anti-

gens shared by many phylogenetically relatedbacteria are another sort of cross-reactivitieswhich we would like to call accidental. Theseare exemplified by the cross-reaction betweenthe capsular polysaccharides of group B menin-gococci and E. coli capsular type K1 based on asimilar polysaccharide, poly-N-acetylneura-minic acid (11, 14, 68, 92). It can be assumedthat identical structural features among poly-saccharides arise by chance with a reasonablefrequency; the possible numbers of monosac-charides to be used as their building blocks arelimited. Such cross-reactions are known notonly between different bacteria but also be-tween bacteria and animal or human cells: theForssman antigen present in humans, guineapigs, horses, sheep, chickens, etc., and alsofound in bacteria (19), and the immunodetermi-nants of several human blood group antigensshared also by plants and bacteria (83, 187-189). Such cross-reactions have been ascribedroles in the pathogenesis of infectious diseases,and especially of their reactive sequelae (79,91). Many of these cross-reactions have beensummarized in a review (84).

ManNUA-Containing Bacterial AntigensAlthough ManNUA is a very rare sugar, it is

not restricted to ECA. ManNUA was first dis-covered by Perkins (154) as a constituent of apolysaccharide antigen of Micrococcus lyso-deikticus, in which it replaces the usual tei-choic acids as teichuronic acid (73). ManNUAhas also been found in other bacterial antigens.Together with glucose it forms the above-men-tioned teichuronic acid of M. lysodeikticus (73,74, 199) and also the capsular K7 antigen of E.coli 07 and 014 (81, 94, 124). In combinationwith fucosamine, it is found in a surface anti-gen of the H variant of Staphylococcus aureus(232). ManNUA also occurs in the K15 antigenof Vibrio parahaemolyticus (168, 198).The apparent presence of two independent

ManNUA-containing antigens in E. coli014:K7, i.e., in the strain in which Kunin (100)first found the ECA, is surely noteworthy. Pre-liminary investigations carried out with acap-sular mutants ofE. coli 014 and 07 (i.e., withE. coli 014:K7- and 07:K7-) revealed that bothmutants still contain the ECA determinant infair amounts (G. Schmidt, D. Mannel, and H.Mayer, unpublished data); however, K- muta-tions are often leaky, and therefore this evi-

dence is not very strong. ManNUA-containingECA preparations are obtained from many Sal-monella, which have no K antigens, and inthem the ECA could not have been mixed upwith a capsular polysaccharide. Absorption ofan ECA-antiserum with heat-killed M. lyso-deikticus showed no lowering of the ECA titer,suggesting that no cross-reaction exists be-tween ECA and the teichuronic acid ofM. lyso-deikticus, although ManNUA is also part oftheserological determinant of the latter (199; ourunpublished results).The biosynthesis of ManNUA starting from

UDP-N-acetylglucosamine (UDPGlcNAc) wasrecently elaborated in E. coli 014 by Kawa-mura et al. (94), who were able to separate theenzymes responsible for epimerization and de-hydrogenation. UDPGlcNAc is apparently (94)first epimerized to UDPManNAc and then con-verted to UDPManNAcUA, the activated formof ManNUA. These findings are expected tofacilitate future studies of the biosynthesis ofECA as well as interpretation of genetic data(see P. Kiss, thesis, University of Freiburg,1975).

DISTRIBUTION OF ECA: TAXONOMICIMPLICATIONS

Kunin (101) defined the distribution of ECAin a way that turned out to be precisely correct:he found it in a "large variety of groups belong-ing to Enterobacteriaceae" but not in othergram-negative organisms (Brucella, Pseudom-onas, Bordetella pertussis) nor in gram-positivebacteria (Staphylococcus, Streptococcus, Sar-cina, Bacillus). Table 6 shows the bacteria thathave been tested for the presence of ECA inbuffer extracts. In most cases the ability of theextract to sensitize RBC for HA and in somecases also the inhibition of HA was tested.Many strains from all known genera of En-

terobacteriaceae (23, 45) have been shown tocontain ECA. Gram-positive strains and manygram-negative strains not classified as Entero-bacteriaceae have been recognized as ECA neg-ative. Representatives of some families thatmay be closest to Enterobacteriaceae (such asAzotobacteraceae, Rhizobiaceae, Halobacteri-aceae, Nitrobacteraceae, and Bacteroidaceae)are unfortunately missing from the list of exam-ined bacteria, which is medically oriented.

In some instances the presence or absence ofECA may be an additional taxonomic criterionas discussed by Le Minor et al. (106) and Bader(10). Yersinia have recently been recognized onthe basis of their biochemical properties as sep-arate from Pasteurella (186, 197) and tenta-tively transferred to Enterobacteriaceae (130,131); this transfer will very probably be ac-

BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from



cepted. The presence ofECA in Yersinia and itsabsence in Pasteurella supports this classifica-tion.The positions ofErwinia and Pectobacterium

are not completely clear; they are included inthe same genus ofEnterobacteriaceae in the 8thedition ofBergey's Manual (23). It is quite pos-sible that the group is very heterogeneous andcontains members close to Enterobacteriaceae(also ECA positive) and others that should notbe so classified (six strains of "Erwinia chry-santhemi" examined were all ECA negative)(18, 53, 169).Although Aeromonas resembles Enterobac-

teriaceae in some of its fermentative properties,it is classified in Vibrionaceae (23) and is alsoECA negative. Several Vibrio strains have alsobeen found to be ECA negative. In contrast,bacteria variously classified as Plesiomonas orAeromonas or Vibrio shigelloides (46, 69, 76)are ECA positive. The specific name shigel-loides refers to antigenic cross-reactivity withShigella sonnei, which seems to involve the 0antigen (46). This applies to both the 0-specificpolysaccharide, which is serologically similarto that of S. sonnei (phase I, the smooth form)and to the R core, which is of the R1 type (T.Kontrohr, P6cs, personal communication). TheECA was apparently not LPS linked, and onlythe ethanol-soluble fraction was immunogenic(210), as expected of ECA-positive smoothstrains. We propose that the taxonomic positionof Plesiomonas be reexamined for inclusion inEnterobacteriaceae as originally suggested(46).ECA-positive bacteria can understandably

lose through mutation their capacity to produceECA. Such ECA-negative mutants have beendescribed in Salmonella (119, 120) and E. coli(see "Genetic determination of ECA"). A fewECA-negative strains have been encounteredin surveys among mainly ECA-positive strainsof a species (e.g., one Serratia) (101) and someProteus strains (106) (Table 6), and it is likelythat they also represent such mutants.

SUMMARYIt has become clear that all strains (with the

exception of defective mutants) of bacteria be-longing to the family of Enterobacteriaceaeshare at least one antigenic component-theenterobacterial common antigen, ECA-notpresent in other gram-negative or gram-posi-tive organisms. This antigen is demonstrableby indirect hemagglutination and by othermeans. A pure anti-ECA serum is not yet avail-able, and tests on ECA are always based ondemonstrating cross-reactivity between differ-ent genera. Whenever doubt of the identity of

ECA arises, rabbit antis. coli 014 serum shouldbe the standard anti-ECA and HA and HAIshould be the standard methods, (using RBCcoated with extracts of several different entero-bacterial strains) as defined by Kunin (101).There may be more than one antigen shared bythe enteric bacteria, and this should beclarified as soon as possible. We tend to believethat the ECA defined as above is one antigenicentity, but the evidence is mostly indirect, anda final chemical identification is urgentlyneeded.

It seems that the ECA in most bacteria ispresent in a "haptenic" form. This form ofECAhas a propensity to aggregate with various hy-drophobic structures, for instance with LPS incell extracts or with the RBC membranes tosensitize them for hemagglutination. In somebacteria, however, ECA is linked, probably co-valently, to LPS.

Recently a preparation of ECA free of LPSand active in various serological tests for ECAhas been obtained. Even if it is not completelypure (two lines are detected by immunopre-cipitation), at least 50% of its weight is ac-counted for by a heteropolymer of GlcNAc andManNAcUA. The latter sugar appears essentialfor its serological specificity. This material ispresent in various ECA-positive species ofenteric bacteria but absent from their ECA-negative mutants. The behavior of these mu-tants and ECA-positive recombinants obtainedfrom them constitutes the strongest evidencefor this material representing ECA. Some fattyacid, mainly palmitic, is linked to the GlcN-ManNUA polymer. ECA therefore seems to bean amphiphiic molecule in which the sugarunits (at least the ManNUA) are responsiblefor the immunological specificity and the hy-drophobic part is responsible for the aggregationtendency.The LPS-linked form of ECA has the same

immunological specificity and therefore shouldcontain ManNUA. This is probably linked to thenonreducing terminus of the LPS core, since therequirements for this linkage are very strict. Anunusual type of core structure (without theterminal glucosamine present in most othertypes of cores) is required; furthermore, this coremust be complete but unsubstituted by 0 sidechains. The linkage region has not yet beenisolated to prove that a covalent linkage exists.However, suggestive evidence is provided by thefact that this linkage requires the function oftherfaL gene, which determines a translocaseneeded to link the 0 side chain to core.

Recently, several kinds ofmutants blocked inthe biosynthesis ofECA have been found. Theyindicate connections between the biosynthesis

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from



of ECA and LPS. The rfe genes (close to ilv)participating in the synthesis of the 0 sidechain of several (but not all) Salmonella and E.coli serogroups are required for the synthesis ofECA in all groups. It seems possible that thecommon step is the synthesis or modification ofa carrier molecule for the synthesis of boththese polysaccharides. Other, rff, genes, alsoclose to ilv, are required for the synthesis ofECA only; they may be responsible for stepsspecific to ECA, for instance in the synthesisand transfer ofManNUA. Further, the rtb genecluster of Salmonella typhimurium is also in-volved in determining the presence of ECA.Some ECA-negative mutants are very sensitiveto detergents, suggesting an altered (outer)membrane structure. Further work with thesemutants can be expected to give information ofthe biosynthesis of ECA as well as of its func-tion in the bacteria and in relation to diseasecaused by these bacteria.Only a few kinds of whole heat-killed bacte-

ria-those in which ECA is linked to LPS, pro-visionally called "immunogenic strains" - elicitthe production of anti-ECA antibodies. TheLPS-linked ECA is a fairly good immunogen inrabbits and poor in mice. The non-LPS-linkedECA can also immunize rabbits when pre-sented in a suitable form; a large aggregate sizeis a definite requirement. LPS and some lipidsand detergents prevent this immunization; theinjection of such mixtures does, however, notlead to true tolerance, as the animals becomeprimed to a subsequent injection of an immuno-genic form of ECA. The nature ofthis phenome-non urgently needs clarification.ECA is probably present on the outer surface

of the bacteria and can therefore be expected tobe involved in reactions of the bacteria withtheir animal hosts. Anti-ECA antibodies areelevated in severe or chronic infections causedby enteric bacteria; their diagnostic potentiali-ties would still require more study. ECA-nega-tive mutants have a slightly reduced virulence.A protective function of anti-ECA antibodieshas been suggested in experimental infection,but a correlation with presence or prognosis ofhuman disease has so far not been shown. Fur-ther studies are, however, indicated, especiallyif better immunizing methods can be devel-oped.

ACKNOWLEDGMENTSWe wish to dedicate this study to E. Neter, Buf-

falo, N.Y., whose continued interest and research inthe common antigen has been decisive for the devel-opment of our knowledge in this field.We thank Marianne Hovi and Marita Antila for

their most valuable help in the preparation of themanuscript, W. McCabe and E. Johnsson for mak-

ing their newest results known to us before publica-tion, 0. Makela for discussions on the immunologi-cal aspects, and E. Neter, G. Schmidt, 0. Luderitz,and the editors of this journal for valuable sugges-tions on improving this paper.

LITERATURE CITED1. Agarwal, M. K., and E. Neter. 1971. Effect of

selected lipids and surfactants on immuno-genicity of several bacterial antigen. J. Im-munol. 107:1448-1456.

2. Ahlstedt, S., and J. Holmgren. 1975. Alterationof the antibody response to Escherichia coli 0antigen in mice by prior exposure to varioussomatic antigens. Immunology 29:487-496.

3. Ames, G. F.-L., E. N. Spudich, and H. Ni-kaido. 1974. Protein composition of the outermembrane of Salmonella typhimurium: ef-fect of lipopolysaccharide mutations. J. Bac-teriol. 117:406-416.

4. Andersen, H. J. 1966. Studies of urinary tractinfection in infancy and childhood. VII. Therelation ofE. coli antibodies in pyelonephri-tis as measured by homologous and common(Kunin) antigens. J. Pediatr. 68:542-550.

5. Andersson, J., G. Moller, and 0. Sj6berg. 1972.Selective induction of DNA synthesis in Tand B lymphocytes. Cell. Immunol. 4:381-393.

6. Anzai, H., E. Neter, and E. A. Gorzynski.1960. Demonstration of Rantz antigen inmembers of the genus Bacillus. J. Bacteriol.80:142-143.

7. Aoki, S., S. Imamura, M. Aoki, and W. R.McCabe. 1969. "Abacterial" and bacterialpyelonephritis. Immunofluorescent localiza-tion of bacterial antigen. N. Engl. J. Med.281:1375-1382.

8. Aoki, S., M. Merkel, and W. R. McCabe. 1966.Immunofluorescent demonstration of thecommon enterobacterial antigen. Proc. Soc.Exp. Biol. Med. 121:230-234.

9. Aoki, S., M. Merkel, M. Aoki, and W. R. Mc-Cabe. 1967. Immunofluorescent localizationof bacterial antigen in pyelonephritis. I. Theuse of antisera against the common entero-bacterial antigen in experimental renal le-sions. J. Lab. Clin. Med. 70:204-212.

10. Bader, R.-E. 1972. Ein neuer gemeinsamerFaktorder Typen I bis V von Pasteurellapseudotuberculosis. Zentralbl. Bakteriol.Parasitenkd. Infektionskr. I Abt. Orig. ReiheA 221:327-333.

11. Barry, G. T., W. F. Goebel. 1957. Colominicacid, a substance of bacterial origin relatedto sialic acid. Nature (London) 179:206.

12. Berger, F. M., G. M. Fukuri, E. G. Golden-baum, M. DeAngelo, and G. C. Chandlee.1969. The effect of chlorphenesin on the im-mune response. J. Immunol. 102:1024-1031.

13. Bergquist, L. M., B. H. S. Lau, and C. E.Winter. 1974. Mycoplasma-associated immu-nosuppression: effect on hemagglutinin re-sponse to common antigen in rabbits. Infect.Immun. 9:410-415.

BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from



14. Bhattacharjee, A. K., H. J. Jennings, C. P.Kenny, A. Martin, and I. C. P. Smith. 1975.Structural determination of the sialic acidpolysaccharide antigens of the Neisseriameningitidis serogroups B and C with '3Cnuclear magnetic resonance. J. Biol. Chem.250:1926-1932.

15. Braun, V. 1975. Covalent lipoprotein from theouter membrane of Escherichia coli. Bio-chem. Biophys. Acta 415:335-377.

16. Braun, V., and K. Hantke. 1974. Biochemistryof bacterial cell envelope. Annu. Rev. Bio-chem. 43:89-121.

17. Braun, V., V. Bosch, E. R. Klumpp, I. Neff,H. Mayer, and S. Schlecht. 1976. Antigenicdeterminants of murein-lipoprotein and itsexposure at the surface of EnterobacteriaceaeEur. J. Biochem. 62:555-566.

18. Brenner, D. J., G. R. Fanning, and A. G.Steigerwalt. 1974. Deoxyribonucleic acid re-latedness among erwiniae and other Entero-bacteriaceae: the gall, wilt, and dry-necrosisorganisms (genus Erwinia Winslow et al.,sensu stricto). Int. J. Syst. Bacteriol. 24:197-204.

19. Briles, E. B., and A. Tomasz. 1973. Pneumo-coccal Forssmann antigen. J. Biol. Chem.248:6394-6397.

20. Brodhage, H. 1961. Harnstoff-Extrakte ausgramnegativen Bakterien in der indirektenHamagglutinationsreaktion. I. Harnstoff-Extrakte aus S. typhi, paratyphi B, choleraesuis und Shig. sonnei. Z. Hyg. 148:94-104.

21. Brodhage, H. 1962. Harnstoff-Extrakte ausgramnegativen Bakterien in der indirektenHamagglutinationsreaktion. II. Die serolo-gische Beziehung zwischen dem "C-Antigen"und dem Rauh-Antigen. Z. Hyg. 148:208-213.

22. Brodhage, H. 1962. Urea-treated Gram-nega-tive bacilli in the indirect hemagglutinationreaction. Nature (London) 193:501-502.

23. Buchanan, R. E., and N. E. Gibbons (ed.).1974. Bergey's manual of determinative bac-teriology, 8th ed. The Williams & WilkinsCo., Baltimore.

24. Bull, D. M., and T. F. Ignaczak. 1973. Entero-bacterial common antigen-induced lympho-cyte reactivity in inflammatory bowel dis-ease. Gastroenterology 64:43-50.

24a. Carrillo, J., B. Hashimoto, and J. Kumate.1966. Content of heterogenetic antigen inEscherichia coli and its relationship todiarrhea in newborn infants. J. Infect. Dis.116:285-296.

25. Chermann, J. C., M. Digeon, and M. Raynaud.1967. Non-identit4 des antigenes h6thro-logues rl, r2, r3 et de l'antigene commun deKunin. Ann. Inst. Pasteur Paris 112:375-398.

26. Chermann, J. C., M. Raynaud, and M. Digeon.1967. Etude des divers antigenes elaborespar Salmonella typhi rough (souche R2).Ann. Inst. Pasteur Paris 112:375-398.

27. Chiller, J. M., J. B. Skidmore, D. C. Morrison,and W. 0. Weigle. 1973. Relationship of the

structure of bacterial lipopolysaccharides toits function in mitogenicity and adjuvantic-ity. Proc. Natl. Acad. Sci. U.S.A. 70:2129-2133.

28. Chiller, J. M., and W. 0. Weigle. 1973. Termi-nation of tolerance to human gamma globu-lin in mice by antigen and bacterial lipopoly-saccharide (endotoxin). J. Exp. Med.137:740-750.

29. Coutinho, A., E. Gronowich, W. W. Bullock,and G. Moller. 1974. Mechanism of thymus-independent immunocyte triggering. Mito-genic activation of B cells results in specificimmune responses. J. Exp. Med. 139:74-92.

30. Coutinho, A., and G. Moller. 1975. Thymus-independent B-cell induction and paralysis.Adv. Immunol. 21:113-236.

31. Dankert, M., A. Wright, W. S. Kelly, and P.W. Robbins. 1966. Isolation, purification andproperties of the lipid-linked intermediatesof 0 antigen biosynthesis. Arch. Biochem.Biophys. 116:425-435.

32. Diamantstein, T., and E. Blitstein-Willinger.1974. T cell-independent development of Bmemory cells. Eur. J. Biochem. 4:830-832.

33. Diaz, F., and E. Neter. 1968. Antibody re-sponse to the common enterobacterial anti-gen of children with shigellosis, salmonel-losis or urinary tract infection. Amer. J.Med. Sci. 256:18-24.

34. Diaz, F., and E. Neter. 1969. Production ofcommon enterobacterial antigen by pigment-producing, gram-negative bacilli. J. Bacte-riol. 97:453454.

35. Dische, Z. 1947. A new specific color reaction ofhexuronic acids. J. Biol. Chem. 167:189-198.

36. Domingue, G., and E. Johnsson. 1974. Isola-tion of subcellular fractions containing im-munogenic enterobacterial common antigen.Z. Immunitaets forsch. Bd. 148:23-38.

37. Domingue, G. J., and E. Neter. 1966. Inhibi-tion by lipopolysaccharide of immune phago-cytosis of latex particles modified with com-mon antigen of enteric bacteria. Proc. Soc.Exp. Biol. Med. 121:133-137.

38. Domingue, G. J., and E. Neter. 1966. Opsoniz-ing and bactericidal activity of antibodiesagainst common antigen of Enterobacteria-ceae. J. Bacteriol. 91:129-133.

39. Domingue, G. J., and E. Neter. 1967. Theplaque test for the demonstration of antibod-ies against the enterobacterial common anti-gen produced by lymph node cells. Immunol-ogy 13:539-545.

40. Domingue, G. J., A. Salhi, C. Rountree, andW. Little. 1970. Prevention of experimentalhematogenous and retrograde pyelonephritisby antibodies against enterobacterial com-mon antigen. Infect. Immun. 2:175-182.

41. Doyle, R. J., A. N. Chatterjee, U. N. Streips,and F. E. Young. 1975. Soluble macromolec-ular complex involving bacterial teichoicacids. J. Bacteriol. 124:341-347.

42. Dresser, D. W., and A. Mitchison. 1968. Themechanism of immunochemical paralysis.Adv. Immunol. 8:129-181.

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from



43. Eckhardt, R., M. Heinisch, and K. H. Meyerzum Buschenfelde. 1975. Zellulare Immun-reaktionen gegenuber Common Antigen undhumanen Dunn-bzw. Dickdarmantigenenbei Patienten mit Enteritis regionalis, Coli-tis ulcerosa und Leberzirrhose. Z. Gastroen-terol. 13:413417.

44. Edwards, P. R., and W. H. Ewing. 1969. Iden-tification ofEnterobacteriaceae, 2nd ed. Bur-gess Publishing Co., Minneapolis.

45. Ewing, W. H., R. Hugh, and J. G. Johnson.1967. Revised definitions for the family En-terobacteriaceae, its tribes and genera. Mon-ogr. Communicable Disease Center, At-lanta.

46. Ferguson, W. W., and N. D. Henderson. 1947.Description of strain C27: a motile organismwith the major antigen of Shigella sonneiphase I. J. Bacteriol. 54:179-181.

47. Frasch, C. E., and E. C. Gotschlich. 1974. Anouter membrane protein ofNeisseria menin-gitidis group B responsible for serotype spec-

ificity. J. Exp. Med. 140:87-104.48. Frentz, G., and G. Domingue. 1973. Effects of

immunization with ethanol-soluble entero-bacterial common antigen in vivo bacterialclearance and hematogenous pyelonephritis.Proc. Soc. Exp. Biol. Med. 142:246-252.

49. Galanos, C. 1975. Physical state and biologicalactivity of lipopolysaccharides. Toxicity andimmunogenicity of the lipid A component. Z.Immunitaetsforsch. 149:214-229.

50. Galanos, C., and 0. Luderitz. 1975. Electrodi-alysis of lipopolysaccharides and their con-

version to uniform salt forms. Eur. J. Bio-chem. 544:603-610.

51. Galanos, C., 0. Luderitz, and 0. Westphal.1969. A new method for the extraction of Rlipopolysaccharides. Eur. J. Biochem. 9:245-249.

52. Galanos, C., 0. Luderitz, and 0. Westphal.1971. Preparation and properties of antiseraagainst the lipid-A component of bacteriallipopolysaccharides. Eur. J. Biochem. 24:116-122.

53. Gardner, J. M., and J. Kado. 1972. Compara-tive base sequence homologies ofthe deoxyri-bonucleic acids ofErwinia species and otherEnterobacteriaceae. Int. J. Syst. Bacteriol.22:201-209.

54. Gmeiner, J. 1975. Glyceraldehyde phosphate atthe reducing terminus ofSalmonella 0 hap-tens. Eur. J. Biochem. 51:449-457.

55. Gorzynski, E. A. 1976. Cross-reactivity be-tween mouse tissue and enterobacterial com-mon antigen (CA). Mil. Med, in press.

56. Gorzynski, E. A., J. L. Ambrus, and E. Neter.1971. Effect of common enterobacterial anti-serum on experimental Salmonella typhimu-rium infection of mice. Proc. Soc. Exp. Biol.Med. 137:1209-1212.

57. Gorzynski, E. A., and S. A. Krasny. 1975.Cross-reactivity between organ extracts ofgnotobiotic mice and enterobacterial com-mon antigen. J. Reticuloendothel. Soc. 17:346-352.

58. Gorzynski, E. A., and S. A. Krasny. 1975. Ef-fect of erythrocytes treated with enterobac-terial common antigen on experimental Sal-monella typhimurium infection of mice.Med. Microbiol. Immunol. 161:163-170.

59. Gorzynski, E. A., and S. A. Krasny. 1975. Im-munological mimicry between mouse tissueand enterobacterial common antigen. Im-mun. Commun. 4:39-49.

60. Gorzynski, E. A., and E. Neter. 1966. Influenceof polymyxin B and normal rabbit serum onerythrocyte modifying capacity of commonenterobacterial antigen. J. Bacteriol. 91:2110-2111.

61. Gorzynski, E. A., and E. Neter. 1970. Produc-tion of common antigen by enteric bacteriagrown in a synthetic culture medium. Infect.Immun. 2:767-771.

62. Gorzynski, E. A., E. Neter, and J. L. Ambrus.1970. Differences in antibody responses ofmouse strains to enterobacterial common an-tigen. Proc. Soc. Exp. Biol. Med. 134:776-779.

63. Gorzynski, E. A., R. L. Priore, and E. Neter.1972. Effect of immunization with commonenterobacterial antigen on experimentalSalmonella typhimurium infection of mice.Immun. Commun. 1:123-130.

64. Gorzynski, E. A., C. J. van Oss, J. L. Ambrus,and E. Neter. 1972. The hemagglutinin re-sponse of human subjects to common entero-bacterial antigen. Infect. Immun. 5:625-626.

65. Gorzynski, E. A., H. Y. Whang, T. Suzuki,and E. Neter. 1963. Differences in antibodyresponse of rabbit to enterobacterial antigenafter intravenous and subcutaneous injec-tion with adjuvant. Proc. Soc. Exp. Biol.Med. 114:700-704.

66. Gotschlich, E. C., I. Goldschneider, and M. S.Artenstein. 1969. Human immunity to themeningococcus. IV. Immunogenicity ofgroup A and group C meningococcal polysac-charides in human volunteers. J. Exp. Med.129:1367-1384.'

67. Gotschlich, E. C., T. Y. Liu, and M. S. Arten-stein. 1969. Human immunity to the menin-gococcus. III. Preparation and immunochem-ical properties of the group A, group B andgroup C meningococcal polysaccharides. J.Exp. Med. 129:1349-1365.

68. Grados, O., and W. H. Ewing. 1970. Antigenicrelationship between Escherichia coli andNeisseria meningitidis. J. Infect. Dis.122:100-103.

69. Habs, H., and R. H. W. Schubert. 1962. Oberdie biochemischen Merkmale und die taxo-nomische Stellung von Pseudomonas shigel-loides (Bader). Zentralbl. Bakteriol. Parasi-tenkd Infectionskr. Hyg. Abt. Orig. 186:316-327.

70. Hammarstrom, S., H. E. Carlsson, P. Perl-mann, and S. Svensson. 1971. Immunochem-istry of the common antigen ofEnterobacteri-aceae (Kunin). J. Exp. Med. 134:565-576.

71. Hammerling, G., 0. Lfideritz, 0. Westphal,and P. H. Makela. 1971. Structural investi-

BACTBICIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from



gation on the core polysaccharide of Esche-richia coli 0100. Eur. J. Biochem. 22:331-344.

72. Hammerling, U., and 0. Westphal. 1967. Syn-thesis and use of O-stearoyl polysaccharidesin passive hemagglutination and hemolysis.Eur. J. Biochem. 1:46-50.

73. Hase, S., and Y. Matsushima. 1970. Structuralstudies on a glucose-containing polysaccha-ride obtained from Micrococcus lysodeikticuscell walls. I. J. Biochem. 68:723-730.

74. Hase, S., and Y. Matsushima. 1972. Structuralstudies on a glucose-containing polysaccha-ride obtained from Micrococcus lysodeikti-cus. III. Determination of the structure. J.Biochem. 72:1117-1128.

75. Helenius, A., and K. Simons. 1975. Solubiliza-tion of membranes by detergents. Biochem.Biophys. Acta 415:29-79.

76. Hendrie, M. S., J. M. Shewan, and M. Vbron.1971. Aeromonas shigelloides (Bader) Ewinget al.: a proposal that it be transferred to thegenus Vibrio. Int. J. Syst. Bacteriol. 21:25-27.

77. Hindennach, I., and U. Henning. 1975. Themajor proteins of the Escherichia coli outercell envelope membrane' Preparative isola-tion of all major membrane proteins. Eur. J.Biochem. 59:207-213.

78. Holmgren, J., G. Eggersten, L. A. Hanson,and K. Lincoln. 1969. Immunodiffusion stud-ies on Escherichia coli. I. Identification of 0,K and H antigens in an 06 strain. Acta Pa-thol. Microbiol. Scand. 76:304-318.

79. Holmgren, J., L. A. Hanson, S. E. Holm, andB. Kaijser. 1971. An antigenic relationshipbetween kidney and certain Escherichia colistrains. Int. Arch. Allergy 41:463-474.

80. Howard, J. G., and N. A. Mitchison. 1975. Im-munological tolerance. Progr. Allergy 18:43-96.

81. Ichihara, N., N. Ishimoto, and E. Ito. 1974.Enzymatic incorporation ofN-acetyl-D-man-nosaminuronic acid and D-glucose into a pol-ysaccharide of Escherichia coli 014:K7:H-.FEBS Lett. 40:309-311.

82. Iino, T. 1963. Genetics of flagellar antigen inSalmonella, p. 731-740. Genetics today,Proc. XI Int. Congr. Genet., The Hague,The Netherlands.

83. Iseki, S. 1952. Blood group substances in bacte-ria. Gunma J. Med. Sci. 1:1-21.

84. Jann, K., and 0. Westphal. 1975. Microbialpolysaccharides, p. 1-125. In M. Sela (ed.),The antigens, vol. III. Academic Press Inc.,New York.

85. Johns, M. A., R. E. Whiteside, E. E. Baker,and W. R. McCabe. 1973. Common entero-bacterial antigen. I. Isolation and purifica-tion from Salmonella typhosa 0:901. J. Im-munol. 110:781-790.

86. Johnston, J. H., R. J. Johnston, and D. A. R.Simmons. 1967. The immunochemistry ofShigella flexneri O-antigens. The biochemi-cal basis of smooth to rough mutation. Bio-chem. J. 105:79-87.

87. Jousimies, H., and P. H. Makela. 1974. Genetic

analysis ofSalmonella minnesota R mutantswith defects in the biosynthesis of the lipo-polysaccharide core. J. Bacteriol. 119: 753-759.

88. Joys, T. M. 1968. The structure of flagella andthe genetic control of flagellation in Eubac-teriales. Antonie van Leeuwenhoek J. Mi-crobiol. Serol. 34:205-225.

89. Kadis, S., G. Weinbaum, and S. J. Ajl (ed.).1971. Microbial toxins, vol. V. AcademicPress Inc., New York.

90. Kaiser, B. 1975. Immunological studies of anantigen common to many Gram-negativebacteria with special reference to E. coli.Characterization and biological significance.Int. Archs Allergy Appl. Immunol. 48:72-81.

91. Kaplan, M. H. 1965. Autoantibodies to heartand rheumatic fever: the induction of auto-immunity to heart by streptococcal antigencross-reactive with heart. Ann. N.Y. Acad.Sci. 124:904.

92. Kasper, D. L., J. L. Winkelhake, W. D. Zollin-ger, B. L. Brandt, and M. S. Artenstein.1973. Immunochemical similarity betweenpolysaccharide antigens of Escherichia coli07:K1(L): NM and group B Neisseria menin-gitidis. J. Immunol. 110:262-268.

93. Kauffmann, F. 1966. The bacteriology ofEnter-obacteriaceae. Munksgaard, Copenhagen.

94. Kawamura, T., N. Ichihara, N. Ishimoto, andE. Ito. 1975. Biosynthesis of uridine diphos-phate N-acetyl-D-mannosaminuronic acidfrom uridine diphosphate N-acetyl-D-gluco-samine in Escherichia coli: separation of en-zyme responsible for epimerization anddehydrogenation. Biochem. Biophys. Res.Commun. 66:1506-1512.

95. Kenny, K., and M. Herzberg. 1968. Antibodyresponse and protection induced by immuni-zation with smooth and rough strains in ex-perimental salmonellosis. J. Bacteriol. 95:406-417.

96. Kessel, R. W. I., E. Neter, and W. Braun. 1966.Biological activities of the common antigenof Enterobacteriaceae. J. Bacteriol. 91:465-466.

97. Knox, K. W., and A. J. Wicken. 1973. Immuno-logical properties of teichoic acids. Bacteriol.Rev. 37:215-257.

98. Krishnapillai, V., and K. Karthigasu. 1969.Salmonella abony-Salmonella typhimuriumrecombinant nonvirulent for the mouse. J.Bacteriol. 97:1343-1351.

98a. Kumater, J., J. Cravioto, B. Hashimoto, L.Vega, and J. Carrillo. 1971. Content of com-mon antigen ofEscherichia coli and diarrheaof newborns and infants in a Mexicanpreindustrial community. Ann. N.Y. Acad.Sci. 176:350-359.

99. Kunin, C. M. 1962. Distribution of antibodiesagainst various non-enteropathic E. coligroups. II. Relation to age, sex, and breed.Arch. Intern. Med. 110:676-686.

100. Kunin, C. M. 1963. Separation, characteriza-tion, and biological significance of a commonantigen in Enterobacteriaceae. J. Exp. Med.

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from



118:565-586.101. Kunin, C. M., and M. V. Beard. 1963. Serologi-

cal studies of 0 antigens of Escherichia coliby means of the hemagglutination test. J.Bacteriol. 85:541-548.

102. Kunin, C. M., M. V. Beard, and N. E. Halma-gyi. 1962. Evidence for a common hapten as-sociated with endotoxin fractions of E. coliand other enterobacteriaceae. Proc. Soc. Exp.Biol. Med. 111:160-166.

103. Kuo, T. T., and B. A. D. Stocker. 1972. Map-ping of rfa genes in Salmonella typhimuriumby ES18 and P22 transduction and by conju-gation. J. Bacteriol. 112:48-63.

104. Lagercrantz, R., S. Hammarstrom, P. Perl-mann, and B. E. Gustafsson. 1968. Immuno-logical studies in ulcerative colitis IV. Originof autoantibodies. Exp. J. Med. 128:1339-1352.

105. Leive, L. 1973. Bacterial membranes andwalls. Marcel Dekker, Inc., New York.

106. Le Minor, L., A.-M. Chalon, and M. Veron.1972. Recherches sur la presence del'antigene commun des "Enterobacteriaceae"(antigene Kunin) chez les "Yersinia", "Levi-nea", "Aeromonas" et "Vibrio." Ann. Inst.Pasteur Paris 123:761-774.

107. Lindberg, A. A., L. T. Rosenberg, A. Ljung-gren, P. J. Garegg, S. Svensson, and N.-H.Wallin. 1974. Effect of synthetic disaccha-ride-protein conjugate as an immunogen inSalmonella infection in mice. Infect. Immun.10:541-545.

108. Luderitz, O., C. Galanos, H. J. Risse, E.Ruschmann, S. Schlecht, G. Schmidt, H.Schulte-Holthausen, R. Wheat, 0. West-phal, and J. Schlosshardt. 1966. Structuralrelationship of Salmonella 0 and R antigen.Ann. N.Y. Acad. Sci. 133:349-374.

109. Luderitz, O., 0. Westphal, A. M. Staub, andH. Nikaido. 1971. Isolation and chemical andimmunological characterization of bacteriallipopolysaccharides, p. 145-233. In G. Wein-baum, S. Kadis, and S. J. Ajl (ed.), Microbialtoxins, vol. IV. Academic Press Inc., NewYork.

110. McCabe, W. R. 1972. Immunization with Rmutants of S. minnesota. I. Protectionagainst challenge with heterologous gram-negative bacilli. J. Immunol. 108:601-610.

111. McCabe, W. R. 1975. Humoral immunity togram-negative bacillary infections in experi-mental animals and man. Infection 3:115-118.

112. McCabe, W. R., and A. Greely. 1973. Commonenterobacterial antigen. II. Effect of immuni-zation on challenge with heterologous bacilli.Infect. Immun. 7:386-392.

113. McCabe, W. R., M. Johns, and T. DiGenio.1973. Common enterobacterial antigen. III.Initial titers and antibody response in bacte-remia caused by gram-negative bacilli. In-fect. Immun. 7:393-397.

114. McCabe, W. R., M. A. Johns, S. H. Zinner, T.DiGenio and A. Greely. 1973. Type specificand cross reactive antigens of Gram-negative

bacilli in protection against infection, p. 112-120. In B. Urbaschek, R. Urbaschek, and E.Neter (ed.), Gram-negative bacterial infec-tions and mode of endotoxin action. Springer,New York.

115. McCabe, W. R., B. E. Kreger, and M. Johns.1972. Type-specific and cross-reactive anti-bodies in gram-negative bacteremia. N.Engl. J. Med. 287:261-267.

116. McLaughlin, J. C., and G. J. Domingue. 1974.The immunologic role of the ethanol-solubleenterobacterial common antigen versus ex-perimental renal infection. Immun. Com-mun. 3:51-75.

117. Maeland, J. A., and A. Digranes. 1975. Com-mon enterobacterial antigen in Yersinia en-terocolitica. Acta Pathol. Microbiol. Scand.Sect. B 83:382-386.

118. Makela, P. H., M. Jahkola, and 0. Luderitz.1970. A new gene cluster rfe concerned withthe biosynthesis of Salmonella lipopolysac-charide. J. Gen. Microbiol. 60:91-106.

119. Makela, P. H., and H. Mayer. 1974. Participa-tion of lipopolysaccharide genes in the deter-mination ofthe enterobacterial common anti-gen: analysis in Salmonella groups B and C.J. Bacteriol. 119:765-770.

120. Makela, P. H., H. Mayer, H. Y. Whang, and E.Neter. 1974. Participation of lipopolysaccha-ride genes in the determination of the com-mon enterobacterial antigen: analysis of Rmutants of Salmonella minnesota. J. Bacte-riol. 119:760-764.

121. Marx, A., and H. Mayer. 1974. The action ofphospholipase A on the common enterobacte-riaceae antigen of the Kunin-type. Zentralbl.Bakteriol. Parasitenkd. Infektionskr. Hyg. IAbt. Orig. 226:224-231.

122. Marx, A., and M. Petcovici. 1975. Immuno-chemical studies on purified common entero-bacterial antigen. Zentralbl. Bakteriol. Par-asitenkd. Infektionskr. Hyg. I Abt. Orig.233:486-494.

123. Marx, A., and M. Petcovici. 1976. Role of therfa locus in immunogenicity of common en-terobacterial antigen (Kunin). Infect. Im-mun. 13:360-364.

124. Mayer, H. 1969. D-mannosaminuronsaure-Baustein des K7-Antigens von Escherichiacoli. Eur. J. Biochem. 8:139-145.

125. Mayer, H. 1972. Reaktivitat und Differenzi-erung der kompletten enterobakteriellen R-Basaltypen mit Concanavalin A. Zentralbl.Bakteriol. Parasitenkd. Infektionskr. Hyg. IAbt. Orig. Reihe A 220:477-483.

126. Mayer, H., and G. Schmidt. 1971. Hamagglu-tinine gegen ein gemeinsames Enterobacteri-aceen-Antigen in E. coli Ri-Antiseren. Zen-tralbl. Bakteriol. Parasitenkd. Infektionskr.Hyg. I Abt. Orig. Reihe A 216:299-313.

127. Mayer, H., and G. Schmidt. 1973. The occur-rence of three different lipopolysaccharidecores in Shigella and their relationship toknown enterobacterial core types. Zentralbl.Bakteriol. Parasitenkd. Infektionskr. Hyg. I.Abt. Orig. Reihe A 224:345-354.

BACTRIUOL. REV.


http://mm

br.asm.org/

Dow

nloaded from



128. Mayer, H., G. Schmidt, H. Y. Whang, and E.Neter. 1972. Biochemical basis of the immu-nogenicity of the common enterobacterial an-tigen. Infect. Immun. 6:540-544.

129. Milner, K. C., J. A. Rudbach, and E. Ribi.1971. General characteristics, 1-65. In G.Weinbaum, S. Kadis, and S. J. Ajl (ed.),Microbial toxins, vol. IV. Academic PressInc., New York.

130. Mollaret, H. H., and W. Knapp. 1971. Report(1966-1970) of the subcommittee on Pasteu-rella, Yersinia, and Francisella to the inter-national committee on nomenclature of bac-teria. Int. J. Syst. Bacteriol. 21:157.

131. Mollaret, H. H., and W. Knapp. 1972. Interna-tional committee of systematic bacteriologysubcommittee on the taxonomy ofPasteurella,Yersinia, and Francisella. Int. J. Syst. Bacte-riol. 22:401.

132. Morgenstern, M. A., and E. Gorzynski. 1973.Immune response of guinea pigs to commonenterobacterial antigen. Immun. Commun.2:495-506.

133. Mullan, N. A., P. M. Newsome, P. G. Cun-nington, P. H. Palmer, and M. E. Wilson.1974. Protection against gram-negative infec-tions with antiserum to lipid A from Salmo-nella Minnesota R595. Infect. Immun. 10:1195-1201.

134. Nakae, T., and H. Nikaido. 1971. Multiple mo-lecular forms of uridine diphosphate glucosepyrophosphorylase from Salmonella typhi-murium. II. Genetic determination of multi-ple forms. J. Biol. Chem. 246:4397-4403.

135. Neter, E. 1956. Bacterial hemagglutinationand hemolysis. Bacteriol. Rev. 20:166-188.

136. Neter, E. 1965. Indirect bacterial hemaggluti-nation, and its application to the study ofbacterial antigens and serologic diagnosis.Pathol. Microbiol. 28:859-877.

137. Neter, E. 1969. Endotoxins and the immuneresponse. Curr. Top. Microbiol. Immunol.47:82-124.

138. Neter, E., H. Anzai, and E. A. Gorzynski.1960. Identification of an antigen common toListeria monocytogenes and other bacteria.Proc. Soc. Exp. Biol. Med. 105:131-143.

139. Neter, E., L. F. Bertram, D. A. Zak, M. R.Murdock, and C. E. Arbesman. 1952. Studieson hemagglutination and hemolysis by Esch-erichia coli antisera. J. Exp. Med. 96:1-15.

140. Neter, E., and E. A. Goryznski. 1959. Studieson indirect staphylococcal hemagglutination.Z. Immunitaetsforsch. 118:269-288.

141. Neter, E., E. A. Kennedy, and Th. C. Jawett,Jr. 1973. Antibody response to common enter-obacterial antigen of children with pyogenicperitonitis. Infection 1:12-16.

142. Neter, E., E. Rajnovich, and E. A. Gorzynski.1960. Study of staphylococcal antibodies ofthe Rantz type: placental transfer and titersin sera of children of various ages. Pediatrics25:21-26.

143. Neter, E., 0. Westphal, and 0. Luderitz. 1955.Effects of lecithin, cholesterol, and serum onerythrocyte modification and antibody neu-

tralization by enterobacterial lipopolysaccha-rides. Proc. Soc. Exp. Biol. Med. 88:339-341.

144. Neter, E., 0. Westphal, 0. Luderitz, and E. A.Gorzynski. 1956. The bacterial hemaggluti-nation test for the demonstration of antibod-ies to Enterobacteriaceae. Ann. N.Y. Acad.Sci. 66:141-156.

145. Neter, E., 0. Westphal, 0. Luderitz, E. A.Gorzynski, and E. Eichenberger. 1956. Stud-ies of enterobacterial lipopolysaccharides. Ef-fects of heat and chemicals on erythrocyte-modifying, antigenic, toxic and pyrogenicproperties. J. Immunol. 76:377-385.

146. Neter, E., and H. Y. Whang. 1972. The com-mon antigen ofgram negative bacteria, p. 14.In A. Nowotny (ed.), Cellular antigens.Springer-Verlag, Heidelberg/New York.

147. Neter, E., H. Y. Whang, 0. Lfideritz, and 0.Westphal. 1966. Immunological primingwithout production of circulating bacterialantibodies conditioned by endotoxin and itslipoid A component. Nature (London)212:420-421.

148. Neter, E., H. Y. Whang, and H. Mayer. 1973.Immunogenicity and antigenicity of endo-toxic lipopolysaccharides: reversible effects oftemperature on immunogenicity. J. Infect.Dis. 28:56-60.

149. Neter, E., H. Y. Whang, T. Suzuki, and E. A.Gorzynski. 1964. Differences in antibody re-sponse of rabbit to intravenously injected sol-uble and cell-attached enterobacterial anti-gen. Immunology 7:657-664.

150. Nikaido, H. 1973. Biosynthesis and assembly oflipopolysaccharide and the outer membranelayer of gram-negative cell wall, p. 131-288.In L. Leive (ed.), Bacterial membranes andwalls. Marcel Dekker, Inc., New York.

151. Nikaido, H., M. Levinthal, K. Nikaido, and K.Nakane. 1967. Extended deletions in the his-tidine-rough-B region ofthe Salmonella chro-mosome. Proc. Natl. Acad. Sci. U.S.A.57:1825-1832.

152. Orskov, F., I. Orskov, B. Jann, K. Jann. 1971.Immunoelectrophoretic patterns of extractsfrom all Escherichia coli 0 and K antigen teststrains. Correlation with pathogenicity. ActaPathol. Microbiol. Scand. 79:142-152.

153. 0rskov, F., I. 0rskov, B. Jann, K. Jann, E.Mfller-Seitz, and 0. Westphal. 1967. Immu-nochemistry of Escherichia coli 0 antigens.Acta Pathol. Microbiol. Scand. 71:339-358.

154. Perkins, H. R. 1963. A polymer containing glu-cose and aminohexuronic acid isolated fromcell walls of Micrococcus lysodeikticus. Bio-chem. J. 86:475-483.

155. Perlmann, P., S. Hammerstrom, R. Lager-crantz, and D. Campbell. 1967. Autoantibod-ies to colon in rats and human ulcerativecolitis: cross reactivity with Escherichia coli014 antigen. Proc. Soc. Exp. Biol. Med. 125:975-980.

156. Perlmann, P., S. Hammarstrom, R. Lager-crantz, and B. E. Gustafmson. 1965. Antigenfrom colon of germfree rats and antibodies inhuman ulcerative colitis. Ann. N.Y. Acad.

VoL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from



Sci. 124:377-394.157. Rantz, L. A., E. Randall, and A. Zuckerman.

1956. Hemolysis and hemagglutination bynormal and immune serums of erythrocytestreated with a nonspecific bacterial sub-stance. J. Infect. Dis. 98:211-222.

158. Rapin, A. M., and H. M. Kalckar. 1971.The relation of bacteriophage attachment tolipopolysaccharide structure, p. 267-307. InG. Weinbaum, S. Kadis, and S. J. AjI (ed.),Microbial toxins, vol. IV. Academic PressInc., New York.

159. Raynaud, M., B. Kouznetzova, M. J. Navarro,J. C. Chermann, M. Digeon, and A. Petiprez.1973. A common antigenic constituent in var-ious purified Salmonella endotoxins. J. In-fect. Dis. 128:35-42.

160. Reynolds, B. L., and H. Pruul. 1971. Protec-tive role of smooth lipopolysaccharide in theserum bactericidal reaction. Infect. Immun.4:764-771.

161. Rietschel, E. Th., and 0. Luderitz. 1975.Chemical structure of lipopolysaccharidesand endotoxin immunity. Z. Immunitaets-Forsch. 149:201-213.

162. Rittenhouse, H. G., J. B. Rodda, and B. A.McFadden. 1973. Immunologically cross-re-acting proteins in cell walls ofmany bacteria.J. Bacteriol. 113:1400-1403.

163. Roantree, R. J. 1971. The relationship of lipo-polysaccharide structure to bacterial viru-lence, p. 1-37. In S. Kadis, G. Weinbaum,and S. J. Ajl (ed.), Microbial toxins, vol. V.Academic Press Inc., New York.

164. Roelants, G. E., and B. A. Askonas. 1972. Im-munological B memory in thymus deprivedmice. Nature (London) New Biol. 239:63-64.

165. Rosenbusch, J. P. 1974. Characterization ofthemajor envelope protein from Escherichia coli.Regular arrangement on the peptidoglycanand unusual dodecyl sulfate binding. J. Biol.Chem. 249:8019-8029.

166. Rygaard, J. 1969. Thymus and self. Immuno-biology of the mouse mutant nude. F.A.D.L.,Copenhagen.

167. Saito, I. 1967. Serological study of chronic pye-lonephritis. Especially on the diagnosticvalue of the estimation of enterobacterialcommon antigen response. Fukushima J.Med. Sci. 14:45-53.

168. Sakakibara, K., M. Torii, and K. Kuroda. 1972.Immunochemical studies on K antigens ofVibrio parahaemolyticus A55 (05:K15). Isola-tion and identification of hexosaminuronicacid. Jpn. J. Bacteriol. 27:223-232.

169. Schaad, N. 1974. Comparative immunology ofribosomes and disc gel electrophoresis of ribo-somal proteins form erwiniae, pectobacteria,and other members of the family Enterobacte-riaceae. Int. J. Syst. Bacteriol. 24:42-53.

170. Schlecht, S., and G. Schmidt. 1970. Moglich-keiten zur Differenzierung von Salmonella-R-Formen mittels Antibiotica und anti-bakterieller Farbstoffe. Zentralbl. Bakteriol.Parasitenkd. Infektionskr. Hyg. I Abt. Orig.212:505-511.

171. Schlecht, S., and G. Schmidt. 1970. Unter-suchungen zur Typisierung von Salmonella-R-Formen. 5. Untersuchungen an Salmo-nella ruiru-Mutanten. Zentralbl. Bakteriol.Parasitenkd. Infektionskr. Hyg. I Abt. Orig.215:196-202.

172. Schmidt, G., B. Jann, and K. Jann. 1974. Ge-netic and immunochemical studies on Esche-richia coli 014:K7:H-. Eur. J. Biochem. 42:303-309.

173. Schmidt, G., and 0. Luderitz. 1969. Untersu-chungen zur Typisierung von Salmonella-R-Formen. 2. Typisierung von S. minnesotaMutanten durch Phagen. Zentralbl. Bakte-riol. Parasitenkd. Infektionskr. Hyg. Abt. IOrig. 210:381-387.

174. Schmidt, G., D. Mannel, H. Mayer, H. Y.Whang, and E. Neter. 1976. The role of alipopolysaccharide gene for immunogenicityof the enterobacterial common antigen. J.Bacteriol. 126:579-586.

175. Schmidt, G., S. Schlecht, 0. Luderitz, and 0.Westphal. 1969. Untersuchungen zur Typisi-erung von Salmonella-R-Formen. I. Mikro-biologische und serologische Untersuchun-gen an Salmonella minnesota Mutanten.Zentralbl. Bakteriol. Parasitenkd. Infek-tionskr. Hyg. Abt. I Orig. 209:483-496.

176. Schrader, J. W. 1975. The role ofT cells in IgGproduction; thymus-dependent antigens in-duce B cell memory in the absence ofT cells.J. Immunol. 114:1665-1669.

177. Schwab, J. H. 1975. Suppression ofthe immuneresponse by microorganisms. Bacteriol. Rev.39:121-143.

178. Schwarz, M. M., and R. S. Cotran. 1973. Com-mon enterobacterial antigen in humanchronic pyelonephritis and interstitial ne-phritis. N. Engl. J. Med. 289:830-835.

179. Seltmann, G. 1971. Untersuchungen zur Anti-genstrukturen von Shigella. IV. Isolierungund Reinigung eines thermolabilen Antigensvon Sh. sonnei. Zentralbl. Bacteriol. Parasi-tenkd. Infektionskr. Hyg., I Abt. Orig. ReiheA 219:324-335.

180. Sinclair, N. R. St. C., and H. Elliot. 1968.Neonatal thymectomy and hemolysin re-sponses in inbred mice. J. Immunol. 101:251-255.

181. Siskind, G. W., P. Y. Paterson, and L.Thomas. 1963. Induction of unresponsivenessand immunity in newborn and adult micewith pneumococcal polysaccharide. J. Immu-nol. 90:929-933.

182. Skarnes, R. C., and D. W. Watson. 1955. Theinhibition of lysozyme by acidic polymersfrom pathogenic bacteria. J. Bacteriol.70:110-112.

183. Skidmore, B. J., J. M. Chiller, D. C. Morrison,and W. 0. Weigle. 1975. Immunologic proper-ties of bacterial lipopolysaccharide (LPS):correlation between the mitogenic, adjuvant,and immunogenic activities. J. Immunol.114:770-775.

184. Skidmore, B. J., D. C. Morrison, J. M. Chiller,and W. 0. Weigle. 1975. Immunogenic prop-

BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from



erties of bacterial lipopolysaccharide (LPS).II. The unresponsiveness of C3H/HeJ mousespleen cells to LPS-induced mitogenesis isdependent on the method used to extractLPS. J. Exp. Med. 142:1488-1508.

185. Smit, J., Y. Kamio, and H. Nikaido. 1975.Outer membranes of Salmonella typhimu-rium. Chemical analysis and freeze-fracturestudies with lipopolysaccharide mutants. J.Bacteriol. 124:942-958.

186. Smith, J. E., and E. Thal. 1965. A taxonomicstudy of the genus Pasteurella using a nu-meral technique. Acta Pathol. Microbiol.Scand. 64:213-223.

187. Springer, G. F. 1956. Inhibition of blood-groupagglutinins by substances occurring inplants. J. Immunol. 76:399407.

188. Springer, G. F. 1971. The importance of blood-group and Forssman antigenic determinantsin interactions between man and microbes, p.39-90. In S. Kadis, G. Weinbaum, and S. J.Ajl (ed.), Microbial toxins, vol. V. AcademicPress Inc., New York.

189. Springer, G. F., P. Williamson, and W. C.Brandes. 1961. Blood group activity of gram-negative bacteria. J. Exp. Med. 113:1077-1093.

190. Stocker, B. A. D., and P. H. Makela. 1971.Genetic aspects of biosynthesis and structureof Salmonella lipopolysaccharide, p. 369-438.In G. Weinbaum, S. Kadis, and S. J. Ajl(ed.), Microbial toxins, vol. V. AcademicPress Inc., New York.

191. Sultzer, B. M. 1969. Genetic control of leuco-cyte response to endotoxin. Nature (London)219:1253-1254.

192. Suzuki, T., E. A. Gorzynski, and E. Neter.1964. Separation by ethanol of common andsomatic antigens of Enterobacteriaceae. J.Bacteriol. 88:1240-1243.

193. Suzuki, T., E. A. Gorzynski, H. Y. Whang,and E. Neter. 1964. Hemolysis in immunerabbits of autologous erythrocytes modifiedwith common enterobacterial antigen. Ex-perientia 20:75-76.

194. Suzuki, T., H. Y. Whang, E. A. Gorzynski,and E. Neter. 1964. Inhibition by lipopolysac-charide (endotoxin) of antibody response ofrabbit to common antigen of enterobacteria-ceae. Proc Soc. Exp. Biol. Med. 117:785-789.

195. Suzuki, T., H. Y. Whang, and E. Neter. 1966.Studies of common antigen of enterobacteria-ceae with particular reference to Escherichiacoli 014. Ann. Immunol. Hung. 9:283-292.

196. Takeya, K., and K. Nomoto. 1967. Characteris-tics of antibody response in young or thymec-tomized mice. J. Immunol. 99:831-836.

197. Talbot, J. M., and P. H. A. Sneath. 1960. Ataxonomic study of Pasteurella septica es-pecially of strains from human sources. J.Gen. Microbiol. 22:303-311.

197a. Thomsen, 0. F., and T. Hjort. 1973. Immuno-fluorescent demonstration of bacterial anti-gen in experimental pyelonephritis withantiserum against common enterobacterialantigen. Acta Pathol. Microbiol. Scand.

Sect. A 81: 474-482.198. Torii, M., and K. Kuroda. 1971. Vibrio para-

haemolyticus AS5 (05:K15). Hexosaminuronicacids as constituents sugars. Jpn. J. Bacte-riol. 26:379-380.

199. Toni, M., S. Sakakibara, S. Iida, K. Kobyashi,S. Hase, and Y. Matsushima. 1973. Immuno-chemical studies on a polysaccharide fromMicrococcus lysodeikticus containing glucoseand N-acetyl-mannosaminuronic acid. BikenJ. 16:11-16.

200. Valtonen, V. V., J. Aird, M. Valtonen, 0.Makela, and P. H. Mikela. 1971. Mouse viru-lence of Salmonella. Antigen-dependent dif-ferences are demonstrable also after immu-nosuppression. Acta Pathol. Microbiol.Scand. Sect. B 79:715-718.

201. Valtonen, M. V., M. Plosila, V. V. Valtonen,and P. H. Mikela. 1975. Effect of the qualityof the lipopolysaccharide on mouse virulenceof Salmonella enteritidis. Infect. Immun.12:828-832.

202. Valtonen, M. V., U. M. Larinkari, M. Plosila,V. V. Valtonen, and P. H. Mikela. 1976.Effect of enterobacterial common antigen onmouse virulence ofSalmonella typhimurium.Infect. Immun. 13:1601-1605.

203. van Ons, C. J., J. L. Ambrus, E. A. Gorzynski,and E. Neter. 1972. The opsonin response ofhuman subjects to common enterobacterialantigen. Immun. Commun. 1:69-75.

204. Vosti, K. L., A. S. Monto, J. J. Older, and L. A.Pantz. 1964. The serologic specificity of crudeand purified antigen extracts of Escherichiacoli in hemagglutination reactions with rab-bit and human antisera. J. Immunol. 93:199-204.

205. Weil, E., and A. Felix. 1918. Untersuchungenuiber die gewohnlichen Proteus-stimme undihre Beziehungen zu den X-Stammen. Wien.Klin. Wochenschr. 23:637-639.

206. Westenfelder, M., C. Galanos, and P. 0.Madsen. 1975. Experimental lipid A-inducednephritis in the dog. Invest. Urol. 12:337-345.

207. Westphal, O., 0. Luideritz, and F. Bister. 1952.Ober die Extraktion von Bakterien mitPhenol/Wasser. Z. Naturforsch. 7b:148-155.

208. Whang, H. Y., D. Chun, Y. Yagi, and E.Neter. 1970. Immunogenicity of soluble andaggregated common enterobacterial antigen.Int. Arch. Allergy 38:57-67.

209. Whang, H. Y., E. Cohen, and E. Neter. 1965.Immunogenic studies of antigenically modi-fied lymphocytes from patients with leuke-mia. Vox Sang. 10:161-168.

210. Whang, H. Y., M. E. Heller, and E. Neter.1972. Production by Aeromonas of commonenterobacterial antigen and its possible taxo-nomic significance. J. Bacteriol. 110:161-164.

211. Whang, H. Y., U. Loza, E. Neter, and F. Mil-grom. 1973. Gel precipitation of common en-terobacterial antigen by its antibody. Int.Arch. Allergy Appl. Immunol. 45:905-914.

212. Whang, H. Y., 0. Luderitz, 0. Westphal, andE. Neter. 1965. Inhibition by lipid A of forma-tion of antibodies against common antigen of

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from



enterobacteriaceae. Proc. Soc. Exp. Biol.Med. 120:371-374.

213. Whang, H. Y., H. Mayer, and E. Neter. 1971.Differential effects on immunogenicity andantigenicity of heat, freezing and alkalitreatment of bacterial antigens. J. Immunol.106:1552-1558.

214. Whang, H. Y., H. Mayer, G. Schmidt, and E.Neter. 1972. Immunogenicity of the commonenterobacterial antigen produced by smoothand rough strains. Infect. Immun. 6:533-539.

215. Whang, H. Y., H. Mayer, G. Schmidt, and E.Neter. 1976. Strain-related differences in im-munosuppressive effects of Enterobacteria-ceae and their lipopolysaccharides on produc-tion in rabbits of antibody to enterobacterialcommon antigen. Infect. Immun. 13:1074-1079.

216. Whang, H. Y., and E. Neter. 1962. Immuno-chemical studies of a heterogenetic entero-bacterial antigen (Kunin). J. Bacteriol. 84:1245-1250.

217. Whang, H. Y., and E. Neter. 1963. Compara-tive study of latex agglutination and hemag-glutination tests for demonstration of entero-bacterial antibodies. Am. J. Clin. Pathol.40:136-142.

218. Whang, H. Y., and E. Neter. 1963. Study ofheterogenetic (Kunin) antibodies in serum ofhealthy subjects and children with entericand urinary tract infections. J. Pediatr.63:412-419.

219. Whang, H. Y., and E. Neter. 1964. Selectivedestruction by Pseudomonas aeruginosa ofcommon antigen of enterobacteriaceae. J.Bacteriol. 88:1244-1248.

220. Whang, H. Y., and E. Neter. 1965. Demonstra-tion of common antigen in Salmonella typhispheroplasts. Can. J. Microbiol. 11:598-601.

221. Whang, H. Y., and E. Neter. 1965. Destructionof common antigen ofEnterobacteriaceae by apsychrophilic Pseudomonas. J. Bacteriol.89:1436-1437.

222. Whang, H. Y., and E. Neter. 1967. Furtherstudies of effect of endotoxin on antibody re-sponse of rabbit to common antigen ofEntero-bacteriaceae. J. Immunol. 98:948-957.

223. Whang, H. Y., and E. Neter. 1967. Immuno-suppression by endotoxin and its lipoid Acomponent. Proc. Soc. Exp. Biol. Med.124:919-924.

224. Whang, H. Y., Y. Yagi, and E. Neter. 1967.Characterization of rabbit antibodies againstcommon bacterial antigens and their pres-ence in the fetus. Int. Arch. Allergy 32:353-365.

225. Whang, H. Y., and E. Neter. 1968. Effect ofcholesterol on immunogenicity of commonenterobacterial antigen. Proc. Soc. Exp. Biol.Med. 128:956-959.

226. Whang, H. Y., and E. Neter. 1968. Inhibitionby cardiolipin of the antibody response tobacterial antigens. J. Immunol. 100:501-506.

227. Whang, H. Y., and E. Neter. 1969. Antigen-associated immunosuppressant: effect of se-rum on immune response. Science 163:290-291.

228. Whang, H. Y., and E. Neter. 1970. Chlorphe-nesin: an antigen-associated immunosuppre-sant. Infect. Immun. 2:60-64.

229. Whang, H. Y., and E. Neter. 1973. Productionof common enterobacterial antigen by mem-bers ofthe family Enterobacteriaceae. Exper-ientia 29:96-97.

230. Wilkinson, R. G., P. Gemski, and B. A. D.Stocker. 1972. Non-smooth mutants of Sal-monella typhimurium: differentiation byphage sensitivity and genetic mapping. J.Gen. Microbiol. 70:527-554.

231. Winsor, P. A. 1968. Binary and multicompo-nent solutions of amphiphilic compounds.Solubilization and the formation, structure,and theoretical significance of liquid crystal-line solutions. Chem. Rev. 68:1-40.

232. Wu, T. C. M., and J. T. Park. 1971. Chemicalcharacterization of a new surface antigenicpolysaccharide from a mutant ofStaphylococ-cus aureus. J. Bacteriol. 108:874-884.

233. Young, L. S., K. R. Hoffman, and P. Stevens.1975. "Core" glycolipid ofEnterobacteriaceae:immunofluorescent detection of antigen andantibody. Proc. Soc. Exp. Biol. Med. 149:389-396.

BACTICRIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from


Documents

Enterobacterial Common Antigen · minant structure of the common antigen is probably covalently linked to LPS. The "hap-tenic" formofthe commonantigen is probably present in all common