JOURNAL OF BACTERIOLOGY, JUlY, 1965Copyright @ 1965 Amiierican Society for Microbiology

Vol. 90, No. 1Printed in U.S.A

Localization of Somatic Antigen on Gram-NegativeBacteria by Electron Microscopy

J. W. SHANDSDepartment of Microbiology, College of Medicine, University of Florida, Gainesville, Florida

Received for publication 16 March 1965

ABSTRACT

SHANDS, J. W. (University of Florida, Gainesville). Localization of somatic antigenon gram-negative bacteria by electron microscopy. J. Bacteriol. 90:266-270. 1965.-Antisera specific for the somatic antigens of Salmonella typhimuriutn and Escherichiacoli 0113 were prepared, and globulins isolated from these antisera were labeled withferritin. Micrographs of labeled, sectioned bacteria show that somatic antigen is lo-cated in considerable quantities on the surface of the bacteria, and, furthermore, that itcan extend up to 150 m,u beyond the confines of the cell wall. The arrangement of theferritin on the bacteria suggests that the antigenic site3 are located on fibrillar struc-tures.

Using a partially rough strain of Escherichiacoli B, Weidel and Primosigh (1958) and Weidel,Frank, and Martin (1960) reported biochemicalevidence that the cell wall of a gram-negativebacterium is a three-lavered structure. Theirconcept of the gram-negative cell wall includes asuperficial lipoprotein layer covering a lipopoly-saccharide layer and an innermost rigid or "R"layer. Electron micrographs of sectioned E. coliare at least consistent with these findings, re-vealing three layers in the cell wall (Kellenbergerand Ryter, 1958).

It is firmly established that the lipopolysac-charides, or somatic antigens, of gram-negativebacteria are part of the cell wall (Ribi, Milner,and Perrine, 1959), but it is less certain that theyare confined to the cell wall or much less to one ofthe three layers. Indeed, some findings indicatethat this may be an erroneous concept. Agglutina-tion of bacteria by antibod- to the somatic anti-gen would be impossible if the somatic antigenwere covered by a lipoprotein layer. Furthermore,Wardlaw (1964) compared rough and smoothstrains of Escherichia coli in their susceptibilityto the lethal activity of complement. His findingsled him to propose that eiidotoxin (in the smoothstrain) is located on the cell-wall surface as wellas the middle laver.The present approach to the problem was at

the morphological level, with the use of ferritin-conjugated antibodies specific for the somaticantigens of Salmonella typhinuitrium and E. coli0113. Intact cells of both organisms were labeledand studied by electron microscopy.

MATERI.ALS AND METHODS

Bacteria. S. typhimnurium 7, a smooth, virulentorganism, was obtained from M. Herzberg (De-partment of Bacteriology, University of Florida).A smooth strain of E. coli 0113 was obtained fromE. Rlibi (Rocky Mountain Laboratories, Hamilton,Mont.). The bacteria were grown for 5 hr in glu-cose-salts medium prior to labeling.

Antisera. Antisera to the somatic antigens wereprepared in New Zealand rabbits. The antigenswere whole organisms of S. typhimnurium or E.coli which had been boiled for 2.5 hr in saline todenature flagellar and K antigens. The immuniza-tions consisted of weekly intravenous injections of5 X 101 to 1 X 109 organisms, and the animals werebled 6 days after the last injection. The globulinswere precipitated from the antisera by 37% am-monium sulfate, and were coupled to ferritin bythe method of Sri Rlam et al. (1963).

Labeling. The culture of bacteria to be labeledwas washed once and reconstituted in Hanks' bal-anced salt solution. Four to six drops of ferritin-conjugated antibody were added to 2 to 3 ml ofculture, and the mixture was allowed to stand untilagglutinated. The agglutinated bacteria were thenwashed three times in Hanks' balanced salt solu-tion, fixed by the method of Kellenberger andRyter (1958), and embedded in Maraglas (Mara-blette Corp., Long Island City, N.Y.) by themethod of Spurlock, Kattine, and Freeman (1963).Thin sections were cut on the Porter-Blum MT-2microtome (Ivan Sorvall, Inc., Norwalk, Conn.),stained with lead citrate (Reynolds, 1963), andexamined with a Siemens Elmiskop I electronmicroscope.

266

Urwp '*s' ^ 4 _ b. e <> *, _ss .. v

,. * z ;9IA. b z Fas..... ^ ;1S Xe: .. ;

w tr !.FX .g ' ;s JiP ' '^lil} 4. A Si *Se * @ #

e ar'} ffi.. . ^f . .. ,5>4 z<fffis > '' 't-

R >* v sw w.. s . n5 . e . ....... , . *s X #a_ tyi§ig>lg NSErsS + .'

-kSwh l, . , w^ 9z b

3i1:JI: s.'.>>l b* EX: . . 6' .+ w S.At ........ zw.ff * ^ :}Jk. j {f;_teS.,,. j; i #,^ t{ ' . *$ if 2 r ffi eF -t4eStrs%;SF xs i r s%tcF sL i . o ; 4sI.s,8;3E, e F wS*z rS;t s.; .S u | t B - ..X .... a. t* w , *3i4,t s N 'P't..N,'., 4F .. ,,, ..... ,.s.w,.o, ;*.

-#s j fD sl..+ ,.: .# : iSrFIG. 1. Escherichia coli agglutinated by ferritin-conjugated antibody to somatic antigen. The bacteria are

widely separated and appear to be connected by bridges of ferritin. X60,000.FIG. 2. Salmonella typhimurium agglutinated by ferritin-conjugated antibody to somatic antigen.

X60,000.267

0.1 9

1-'''''~~ ~ ~---"-----'-----'"

4

3:~ ~ ~ , ,'im.:

4 " i1* a,,. , !!!!!! #r . 4

3S.* SJ5'" VS;A46 % 4~~~~~~~~~~~~~~~~~~~!;Af, FA 4s1~#..J,

aa*-0,,..:',.4,","

Nt WM |..

R.~ ~ ~ ~ ~ ~~~.

skt'.*

i,A".:4

.5t~~~ ~ ~ ~ ~ ~ ~ ~~ e

T.~ ~ ~ ~ ~ .

W1 a

::w: ts ':*~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~s

B ~~~~~~~~~~'.

5*, ? t i i t . X r . ?? | .g?,,;...

FIG. 3. Surface of Escherichia coli labeled with ferritin-conjugated antibody showing a fibrillar orienta-tion of the ferritin. X200,000.FIG. 4. Surface of labeled Salmonella typhimurium. X400,000.

268

LOCALIZATION OF SOMATIC ANTIGENS

o0.25gI -,,,-- I-v- -1._

:..A I

a, 'I-,

*~ f B". *,?

FIG. 5. Labeled Escherichia coli showing "capsule" of ferritin surrounding the organism. X80,000.

RESULTSThe specificity of the ferritin-conjugated anti-

bodies was assayed in several ways. To rule outnonspecific labeling of the bacteria, crossed ex-periments were performed in which S. typhi-murium was incubated with ferritin-labeled anti-E. coli globulin and vice versa. No labeling of thebacteria was found in either experiment. Theantibody activity of the S. typhimuirium anti-serum could be completely removed by absorptionwith a potent endotoxin prepared from the sameorganism by aqueous ether extraction (Ribi et al.,1959). Similarly, all the activity of the anti-E.coli conjugate could be absorbed by the sameorganism after being killed in alcohol, digestedwith trypsin, and boiled for 2.5 hr. Consequently,it was concluded that virtually all the antibodyactivity was directed against the somatic antigensand none against either the flagellar or K anti-gens.Study of labeled, agglutinated organisms re-

vealed that the agglutination of these gram-nega-tive organisms almost never involved juxtap)osi-tion of the cell wall of one organism againstanother. The organisms were almost always

separated by a distance of at least severalhundred Angstroms and appeared to be heldtogether by bridges of ferritin-conjugated anti-body (Fig. 1 and 2). Pictures at higher magnifica-tion strongly suggested that the basic structureslabeled by the ferritin antibody conjugate werefibrils which extended out from the cell wall, somecurling back on themselves or cross-linked toother fibrils by antibody (Fig. 3 and 4). E. coliwas so densely labeled that the somatic antigenapleared to form a structure analogous to acalpsule about the organism (Fig. 5).

DIscussioN

Although the major portion of the somaticantigen or endotoxin is found in the cell wall ofgram-negative bacteria (Ribi et al., 1959), thepresent findings show that it is certainly notconfined to this structure, but rather that itextends for distances up to 150 my4 beyond theconfines of the cell wall. It is, therefore, clear thatthe somatic antigen should not be designatedsimply as a layer in the gram-negative cell wall.The somatic antigen also appears to be struc-

turally a fibril. This finding is consistent with

269\ ()L. 90, 1965

4

i!ql,

4;,

J. BACTERIOL.

those of Milner et al. (1963), who demonstrateda fibrillar structure in the lipopolysaceharidederived from Bordetella pertussis. The agglutina-tion of gram-negative bacteria by antisomaticantibody appears to be effected by cross-linkingof these fibrils from individual cells and not bycross-linking of cell walls.The existence of long fibrils of somatic antigen

or endotoxin streaming from the surface of gram-negative organisms implies that quantities of thismaterial may be easily removed without killingand disrupting the integrity of the bacterium.This may be one explanation for the presence ofendotoxin in culture filtrates. These fibrils mayalso be the source of endotoxin obtained byaqueous ether extraction (Ribi et al., 1959), a pro-cedure which does not disrupt the cell, and whichrequires viable, initact bacteria to produce a goodyield.The portion of endotoxin extending beyond

the confines of the cell probably plays an impor-tant role in host-parasite relationships of gram-negative infection. In the pathogenesis of infec-tion, the initial host-parasite interaction wouldtake place at the bacterial surface. It is quitepossible that at the time of the initial encounterrelease of endotoxin from the bacterial surface,without damage to the bacterium, may serve todecrease local host defenses and thereby facilitatethe establishment of infection. Once infection isestablished, continued endotoxin productioncould lead to endotoxemia and its consequentmanifestations, even in the absence of death anddisintegration of large numbers of organisms.The superficial localization of somatic antigen

may also be one explanation for the relativeresistance of smooth gram-negative organisms tothe bactericidal activity of normal serum (AMi-chael and Landy, 1961). Where the concentrationof antibody is limited, most of it may be takenup by antigenic sites distant from the cell wall.The subsequent fixation of complement at thesesites would presumably result in less bactericidalactivity.

ACKNOWLEDGMENTSThe technical assistance of Lorenza Brown is

gratefully acknowledged.This investigation was supported by Public

Health Service research grant Al 03798-05 from theNational Institute of Allergy and Infectious Dis-eases.

LITERATURE CITED

KELLENBERGERt, E., AND A. RYTER. 1958. Cell walland cytoplasmic membrane of Escherichia coli.J. Biophys. Biochem. Cytol. 4:323-326.

MICHAEL, J. G., AND M. LANDY. 1961. Endotoxicproperties of gram-negative bacteria and theirsusceptibility to the lethal effect of normalserum. J. Infect. Diseases 108:90-94.

MILNER, K. C., R. L. ANACKER, K. FUKUSHI, W. T.HASKINS, M. LANDY, B. MALMGREN, AND E.RIBI. 1963. Symposium on relationship of struc-ture of microorganisms to their immunologicalproperties. III. Structure and biological prop-erties of surface antigens from gram-negativebacteria. Bacteriol. Rtev. 27:352-368.

REYNOLDS, E. S. 1963. The use of lead citrate athigh pH as an electron-opaque stain in electronmicroscopy. J. Cell Biol. 17:208-212.

RIBI, E., K. C. MILNER, AND T. 1). PERRINE. 1959.Endotoxic and antigenic fractions from the cellwall of Salmonella enteritidis. Methods for sepa-ration and some biologic activities. J. Immunol.82:75-84.

SPURLOCK, B. O., V. C. KATTINE, AND J. A.FREEMAN. 1963. Technical modifications inmaraglas embedding. J. Cell Biol. 17:203-207.

SRI RAM, J., S. S. TAWDE, G. B. PIERCE, JR., ANDA. R. MIDGLEY, JR. 1963. Preparationi of anti-body-ferritin conjugates for immuno-electronmicroscopy. J. Cell Biol. 17:673-675.

WARIDLAW, A. C. 1964. Endotoxin and complementsubstrate, p. 81-88. In M. Landy and W. Braun[ed.], Bacterial endotoxins. Rutgers UniversityPress, New Brunswick, N.J.,

WEIDEL, W., AND J. PJUIMOSIGH. 1958. Biochemicalparallels between lysis by virulent phage andlysis by penicillin. J. (Gen. Microbiol. 18:513-517.

WEIDEL, W., H. FRANK, AND H. H. MARTIN. 1960.The rigid layer of the cell wall of E.scherichiacoli strain B. J. Gen. Microbiol. 22:158-166.

270 SHANDS