Vol. 155, No. 3JOURNAL OF BACTERIOLOGY, Sept. 1983, p. 1382-13920021-9193/83/091382-11$02.00/0Copyright © 1983, American Society for Microbiology

Monoclonal Antibodies Reveal LamB Antigenic Determinantson Both Faces of the Escherichia coli Outer Membrane

SERGIO SCHENKMAN,1 EVELYNE COUTURE,2 AND MAXIME SCHWARTZ'*Unite de Genetique Mol6culaire1 and Unite de Microscopie Electronique,2 Department de Biologie

Moleculaire, Institut Pasteur, 75724 Paris Cedex 15, France

Received 11 April 1983/Accepted 7 June 1983

LamB protein is involved in the transport of maltose across the outer membraneand constitutes the receptor for phage X. In this study we characterized sixpreviously described anti-LamB monoclonal antibodies (mAbs). Four of these,the E-mAbs, recognized determinants that were exposed at the cell surface,whereas the other two, the I-mAbs, recognized determinants which were notexposed. Competition experiments demonstrated that the domains recognized bythese two classes of mAbs were completely distinct. In addition, the E-mAbsprevented LamB from neutralizing phage A in vitro and protected LamB againstproteolytic degradation, whereas the I-mAbs had no such effects. The E-mAbshave been shown previously to constitute two classes; some E-mAbs inhibitmaltose transport in vivo, and others do not. Immunoelectron microscopydemonstrated that the I-mAbs also define at least two types of determinants. Oneof these, which is accessible in membrane fragments from a mutant (lpp) devoid oflipoprotein but not in membrane fragments from an lpp+ strain, probablycorresponds to a region of LamB that is involved in the interactions withpeptidoglycan. The other determinant, which is fully accessible in LamB-peptidoglycan complexes and in LamB-containing phospholipid vesicles but onlyslightly accessible in membrane fragments from an lpp mutant, is probably locatedvery close to the inner surface of the outer membrane. LamB also contains at leastone additional determinant, which (i) is exposed at the inner surface of themembrane, (ii) is accessible to antibodies in membrane fragments from an lpp+strain, and (iii) may be involved in the interaction of LamB with the periplasmicmaltose-binding protein.

LamB protein is one of the major cellularproteins (approximately 105 copies per cell)when Escherichia coli is grown in the presenceof maltose (2, 24). The primary function of thisprotein is to permit the diffusion of maltose andmaltodextrins across the outer membrane (2, 17,30). In detergent solutions (14, 20) and probablyin membranes (19, 23), LamB is a trimer com-posed of three identical subunits containing 421amino acids (5). The amino acid sequence ofLamB was deduced from the nucleotide se-quence of its structural gene (5), whereas circu-lar dichroism studies (14) indicated a predom-inance of p-sheets in its secondary structure.Like other pore-forming proteins (22, 26), LamBis believed to span the entire outer membrane.At the outer face of the membrane LamB acts asa receptor for phages, such as X (24) and K10(25), is recognized by specific antibodies (8), andbinds maltose and maltodextrins (6). Evidencethat LamB is exposed on the inner face of theouter membrane is more circumstantial. Genetic(11, 31) and biochemical (1, 18) experiments

have indicated that this protein interacts withthe periplasmic maltose-binding protein, andthis feature has been included in models describ-ing the mechanism of maltose transport (11, 28).The interaction with peptidoglycan also suggeststhat LamB may be exposed on the inner face ofthe outer membrane. LamB remains associatedwith the peptidoglycan sacculus when cells aredissolved in sodium dodecyl sulfate at 60°C inthe presence of Mg2+ ions (9); it can be elutedfrom the sacculus by removing the Mg2" ions orby increasing the salt concentration and can bereassociated with the sacculus under adequateconditions (32). However, reassociation re-quires the presence of lipoprotein, which iscovalently bound to peptidoglycan and which isbelieved to be at least partially embedded in theouter membrane (3, 22, 32). Therefore, the asso-ciation of LamB with the sacculus could resultfrom LamB-lipoprotein interactions that occurwithin the outer membrane and does not neces-sarily imply that LamB protrudes at the innerface of this membrane.

1382

on October 19, 2017 by guest

http://jb.asm.org/

Dow

nloaded from

LamB ANTIGENIC DETERMINANTS 1383

To examine how LamB is inserted in themembrane, we used monoclonal antibodies(mAbs) directed against the protein, as recentlydescribed by Gabay and Schwartz (8). Some ofthese mAbs were shown previously to recognizedeterminants that were exposed at the bacterialsurface; they inhibited the adsorption of X andK10, and some also inhibited maltose transport.The other mAbs did not bind to intact cells; onebound to LamB-peptidoglycan complexes, andbecause of this it was proposed that the antigen-ic site recognized by this mAb is located withinthe membrane. In other cases the mAbs whichdid not bind to the surfaces of intact cells onlyrecognized solubilized LamB that was obtainedafter elution from the peptidoglycan by extrac-tion with detergent and EDTA; it was assumedthat these mAbs recognized sites involve in theLamB-peptidoglycan interaction (8).

In this paper we further characterize the sitesrecognized by the different classes of mAbs andprovide more direct evidence that LamB isaccessible at the periplasmic face of the outermembrane.

MATERIALS AND METHODSBacterial strains and medium. Strain popll30 (ompR

his) was used for LamB purification (9). The followingother strains were also used in this work: pop3(araD139 AlacU169 rpsL relA thi; strain MC4100 ofCasadaban [4]); pop3205, a derivative of pop3 carryingthe nonsense mutation IamB102 (12); JE5506 (pps hisproA argE thi lac xyl mtl tsx); and JE5505, an Ippmutant of strain JE5506 (29). These bacteria weregrown in M63 medium (24) supplemented with 0.4%maltose and the necessary amino acids and vitamins.

Purification of LamB. LamB-peptidoglycan com-plexes were obtained as described by Gabay andYasunaka (9). These complexes were dissociated in 10mM Tris-hydrochloride (pH 7.4) containing 5 mMEDTA and 2% octylpolyoxyethylene (octyl-POE) agift from J. P. Rosenbusch (26). After 1 h of centrifu-gation at 100,000 x g, the supernatant was adjusted to20 mM MgCl2, phenylmethylsulfonyl fluoride-treatedDNase I (0.01 mg/ml; Sigma Chemical Co.) was added,and the mixture was incubated for 1 h at 37°C. Duringthis step, DNA hydrolysis led to a reduction in theviscosity of the preparation, and the addition of Mg2+ions induced precipitation of the LamB protein. Afteradditional centrifugation at 100,000 x g for 1 h, theresulting pellet was suspended in 10 mM Tris-hydro-chloride (pH 7.4)-S5 mM EDTA-2% octyl-POE andincubated for 1 h at 37°C. The insoluble material wasremoved by centrifugation at 100,000 x g for 1 h, andthe supernatant was loaded onto a column containingmAb3O2 coupled (13) to Sepharose 4B. LamB waseluted by washing the column with 0.1 M glycinehydrochloride buffer (pH 2.8) containing 1% octyl-POE. The protein solution was brought to pH 7 with 1M Tris-hydrochloride buffer (pH 8.3) and dialyzedagainst 10 mM Tris-hydrochloride buffer (pH 7.4)containing 1% octyl-POE. In a standard preparation, 2liters of culture and a 2-ml Sepharose 4B columncontaining 7 mg of immobilized mAb yielded approxi-

mately 1 mg of LamB. The concentration of purifiedLamB was determined spectrophotometrically by us-ing 24.5 as the extinction coefficient of a 1% solution at280 nm (20).

Radioactively labeled LamB was prepared as de-scribed above, after the bacteria were grown in a low-sulfur medium containing 0.1 mCi of 35S (25 to 40 Ci/ml; Amersham France) per ml (8). More than 98% ofthe radioactivity present in the final preparation couldbe precipitated by adding any of the mAbs directedagainst LamB and Staphylococcus aureus cells. Thespecific activity of the purified protein was generallybetween 5 and 20 nCi/,umol of LamB monomer.

Preparation of antibodies and Fab fragments. Aconventional anti-LamB serum (polyclonal antibodies)had been obtained previously by injecting Triton X-100-solubilized LamB into a rabbit (9). This serum waspreadsorbed (27) in two different ways. Preadsorptionwith strain pop3205 cells (lamB102) yielded a prepara-tion of polyclonal antibodies (total serum [tS]) which,when tested by immunofluorescence (8), reacted withthe surfaces of lamB+ cells, but not with the surfacesof cells of strain pop3205, a mutant devoid of LamBprotein. Preadsorption with strain pop3 cells (lamB+)yielded a preparation of antibodies (adsorbed serum[aS]) which failed to react either with lamB+ cells orwith lamB mutants. When assayed for its ability toimmunoprecipitate purified LamB, the aS had a titerwhich was approximately one-tenth the titer of the tS.For electron microscopy, aS was further adsorbedwith vesicles from strain pop3205 because this serumcontained contaminating antibodies which reactedwith components of the E. coli membranes which wereinaccessible in whole cells and unrelated to LamB.After this additional adsorption step the serum reactedwith membrane fragments from lamB+ cells, but notwith fragments from strain pop3205.The monoclonal anti-LamB antibodies (8) were puri-

fied from ascitic fluids by protein A-Sepharose 4Bchromatography (13), and the total immunoglobulinconcentration was determined (16) by using bovineserum albumin as a standard. To prepare Fab frag-ments, the purified immunoglobulin G preparationswere digested with papain (Worthington Diagnostics)(13). The reaction was stopped by dialyzing the mix-ture against 5 mM sodium phosphate buffer (pH 8.0).The insoluble Fc fragments were eliminated by centrif-ugation, and the Fab fragments were purified byWhatman DE-52 chromatography (13) followed by asingle passage through a protein A-Sepharose 4Bcolumn.As a control we used an mAb (IDA 7) directed

against an antigen which is totally unrelated to LamB(a gift from G. Buttin).

Immunoprecipitation and competition experiments.For all immunoprecipitation experiments 35S-labeledLamB was diluted in 0.6 ml of 50 mM Tris-hydrochlo-ride (pH 8.0)-0.15 M NaCl-0.5 mM EDTA-2% TritonX-100 (8) (Triton buffer) containing the desired amountof immunoglobulin. After 4 h at 37°C, the immunocom-plexes were precipitated by adding 0.05 ml of a 10%suspension of formolized S. aureus cells. The precipi-tate was collected by centrifugation, washed withTriton buffer, and added to 7 ml of Biofluor (NewEngland Nuclear Corp.). The radioactivity of thispreparation was determined with an Intertechniquescintillation counter.

VOL. 155, 1983

on October 19, 2017 by guest

http://jb.asm.org/

Dow

nloaded from

1384 SCHENKMAN, COUTURE, AND SCHWARTZ

For competition experiments "S-labeled LamB wasincubated first for 4 h at 37°C in Triton buffer with theFab fragment and then for an additional 4 h with theintact immunoglobulin. The Fab fragment was used inlarge excess (approximately 100 to 1,000 times theamount of antibody necessary to precipitate 50% ofthe 35S-labeled LamB). The intact immunoglobulinwas used in a limiting amount at a concentration thatwas independently determined to precipitate about30% of the 35S-labeled LamB in the absence of com-peting Fab fragment.For proteolysis experiments the preparations were

incubated in 50 mM Tris-hydrochloride (pH 8.0) con-taining 1% octyl-POE at 370C. 3S-labeled LamB waspreincubated for 4 h with the antibodies, and subtilisin(protease type VII; Sigma) was then added at a molarratio of 0.5 with respect to the concentration of LamBmonomer. The reaction was stopped by adding phen-ylmethane sulfonyl fluoride to a final concentration of1 mg/ml and incubating the mixture for 30 min at roomtemperature. The samples were then dissolved insodium dodecyl sulfate sample buffer (0.0625 M Tris-chloride [pH 6.8], 2% sodium dodecyl sulfate buffer),boiled for 3 min, and electrophoresed in 19.3% poly-acrylamide gels containing 0.1% sodium dodecyl sul-fate and 6 M urea (15). The radioactive bands wererevealed by fluorography after the gel was treated withEn3hance (New England Nuclear Corp.).Membrane preparation and electron microscopy. E.

0~32

0.l6

co

E

0.2 0.4LamBbw1*/LamBj.

FIG. 1. Relative affinities of the mAbs for LamB(Scatchard plots). Several concentrations of 3"S-la-beled LamB were incubated for 4 h at 37°C with 0.14nM mAb 302 (O), 0.5 nM mAb 347 (-), 2.2 nM mAb436 (0), or 3 nM mAb 141 (0). Bound LamB wasdetermined after immunoprecipitation, and free LamBwas calculated by determining the difference from thetotal radioactivity present before precipitation. ALamB molecular weight of 141,000 was used for Kddeterminations.

coli membrane fragments were obtained from maltose-grown cells. The procedure used was essentially thatof Osborn et al. (21) and involved treatment of cellswith lysozyme and EDTA to produce spheroplasts,sonication to fragment spheroplasts, and ultracentrifu-gation. No attempt was made to separate outer andinner membranes. Each preparation contained a mix-ture of large and small fragments. Since only theformer could be labeled with anti-LamB antibodies(see below), they were believed to correspond to outermembrane fragments.LamB-containing reconstituted vesicles were pre-

pared by the dialysis method (26). Briefly, 1 ml ofLamB dissolved in 1 ml of a solution containing 3%octyl-POE, 0.1 M NaCl, 1 mM MgCl2, 0.02% NaN3,and 10 mM HEPES (N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid) (pH 7.4) was transferred to atube that was internally coated with a dried film ofsoybean phosphatidylcholine (type III; Sigma). Theresulting suspension was vigorously mixed with aVortex mixer for S min and dialyzed first for 16 h andthen twice for 4 h against 300 ml of the same bufferwithout detergent.Both types of vesicles were diluted 1:10 in phos-

phate-buffered saline (PBS) and adsorbed onto glow-discharged Formvar-coated carbon grids for 1 min atroom temperature. The grids were rinsed extensivelywith PBS containing 0.5% bovine serum albumin (Mbuffer) and then incubated for 1 h in a solution ofantibody diluted in M buffer. The grids were thenwashed once more with M buffer and floated for 1 h atroom temperature over a solution of a protein A-goldcomplex (10). The grids were washed again with Mbuffer and then with PBS, floated for 30 s over double-distilled water, and stained with a 2% sodium phos-photungstate solution at pH 7.0. Micrographs wereobtained with a Siemens model 101 electron micro-scope.

RESULTSAffinity of the mAbs for LamB. Gabay and

Schwartz distinguished four classes of mAbsdirected against LamB. The mAbs which recog-nize determinants exposed at the cell surface(mAbs 302, 177, 347, and 72) are referred tocollectively as E-mAbs. The other mAbs, whichare believed to recognize internal determinants,are referred to as I-mAbs. An estimate of theaffinity of the mAbs for LamB protein wasobtained by using radioactive LamB protein andprecipitating the immune complexes with S.aureus cells. With the four mAbs tested weobtained linear Scatchard plots from which ap-parent Kd values were calculated by using141,000 as the molecular weight of the LamBtrimer. The apparent Kd values were 0.27, 0.24,1.1, and 1.4 nM for mAbs 302, 347, 436, and 141,respectively (Fig. 1). With such high affinitiesthe interactions between LamB and the mAbscould be considered essentially irreversible.This greatly facilitated the design of the competi-tion experiments described below.

Competition between Fab fragments and anti-bodies. The possible overlapping of the binding

J. BACTERIOL.

on October 19, 2017 by guest

http://jb.asm.org/

Dow

nloaded from

LamB ANTIGENIC DETERMINANTS 1385

sites for the different mAbs were determined incompetition studies. We used the fact that all ofthe available anti-LamB mAbs are recognizedby protein A (8). Fab fragments, which are notrecognized by protein A, were used as competi-tors in immunoprecipitation experiments withintact antibodies and S. aureus cells. The resultsobtained with the Fab fragments from one E-mAb (E-mAb 302) and one I-mAb (1-mAb 436)are shown in Fig. 2. An excess of E-mAb 302Fab inhibited the immunoprecipitation observedin the presence of the E-mAbs, but had no effecton the immunoprecipitation induced by the I-mAbs. Conversely, the I-mAb 436 Fab fragmentinhibited the immunoprecipitation observed inthe presence of mAbs 436 and 141, but had littleor no effect in the presence of the E-mAbs. Thisresult indicates that the domain recognized by I-mAbs 141 and 436 is completely distinct fromthe domain recognized by the E-mAbs.

Additional information was obtained by per-forming competition experiments involving twopreparations of polyclonal anti-LamB antise-rum. One preparation (tS) had been preadsorbedwith cells devoid of LamB protein and thereforeprobably contained antibodies that recognizedall domains of LamB. The other preparation (aS)had been preadsorbed with lamB+ cells and wasthereby depleted of the antibodies which recog-

-0

EcU-J

w

a.

r-0.

0z

A

0

B

0I1rKI hr _1 mts aS 302 347 177 72 141 436

FIG. 2. Fab competition experiments. 35S-labeledLamB (0.1 ,ug/ml) was preincubated for 4 h at 37°Cwith or without 50 ,ug of mAb 302 Fab (A) or mAb 436Fab (B) per ml. Antibodies (as shown on the abscissa)were then added, and, after an additional 4 h ofincubation at 37°C, immunoprecipitation was obtainedby adding S. aureus cells. The percentages correspondto the relative amounts of immunoprecipitation in thepresence and absence of Fab. tS and aS represent thetotal anti-LamB serum and the serum preadsorbedwith lamB+ cells, respectively. The values correspondto the means of three determinations.

nized determinants located at the cell surface.The Fab fragment from mAb 302 failed to com-pete with aS. This was predictable since mAb302 recognized an external determinant, where-as aS was specific for internal determinants.However the mAb 302 Fab fragment was alsovery inefficient in competing with tS. This indi-cates that LamB has E determinants which donot overlap with the determinant recognized bymAb 302 and are therefore different from thedeterminants recognized by the four available E-mAbs. Similarly, the incomplete competitionbetween mAb 436 and aS indicates that LamBhas I determinants which do not overlap with thedeterminant recognized by mAb 436.

Effect of the mAbs on the interaction betweenLamB and phage X. The E-mAbs were shownpreviously to prevent the adsorption of phage Ain vivo. Therefore, it was not surprising to findthat mAbs 302 and 347 prevented LamB frominactivating XVho in vitro (Fig. 3). In contrast,the two 1-mAbs failed to prevent LamB frominactivating AVho in vitro; this result was lesspredictable. Indeed, although these two mAbshad no effect on X adsorption in vivo, this was anobvious consequence of their inability to reactwith the cell surface. In vitro, where they couldinteract with LamB, they could have altered theconformation of this protein in such a way that itwould have lost its ability to inactivate thephage.

Proteolytic digestion of LamB: effect of themAbs. Like the other pore-forming proteins ofE. coli, LamB is extremely resistant to proteoly-sis (22). However, we found that by using highconcentrations of subtilisin (ratio of enzyme toLamB, 1:2) and long incubation times (48 h at37°C) it was possible to obtain essentially com-plete inactivation of LamB, as assayed by XVhoneutralization, and conversion of LamB tosmaller peptides. The major hydrolysis productwas a 22,000-dalton peptide, but 14,000- and34,000-dalton peptides were also detected.When incubated under the same conditions butin the absence of subtilisin, LamB remainedfully active.To study the effect of the mAbs on this

proteolytic reaction, we used a somewhatshorter incubation time (24 h). Under theseconditions approximately 80% of LamB wasconverted to smaller peptides (Fig. 4). Prein-cubation of LamB with I-mAb 436 or 141 had noeffect on this reaction. On the other hand, the E-mAbs protected LamB against proteolytic deg-radation. Results identical to those shown inFig. 4 were obtained when we used 100-fold-higher concentrations of mAbs or Fab frag-ments. When the gel shown in Fig. 4 was stainedwith Coomassie blue, we found that practicallyall of the immunoglobulin G had been converted

VOL. 155, 1983

on October 19, 2017 by guest

http://jb.asm.org/

Dow

nloaded from

1386 SCHENKMAN, COUTURE, AND SCHWARTZ

50-5

0-12 -10 -8

Log [ IgG ] (M)

FIG. 3. Effect of mAbs on LamB activity in vitro.LamB (1 Fg/ml) was incubated for 4 h at 37°C withdifferent amounts of mAb 302 (0), mAb 347 (0), mAb436 (D), or mAb 141 (O) in 50 mM Tris-hydrochloride(pH 8.0)-1% octyl-POE. Samples (15 ,ul) were dilutedinto 1 ml of a Tris-hydrochloride (pH 7.4)-10 mMMgCI2 solution containing 7 x 103 plaque-formingunits of XVho phage. LamB activity was calculatedfrom the kinetics of XVho inactivation (24), and thevalues correspond to the remaining amounts of activi-ty relative to the activity observed in the absence ofantibodies or in the presence of IDA 7 control mAb.IgG, Immunoglobulin G.

to 22,000- to 26,000-dalton peptides similar tothe Fab and Fc fragments (data not shown).

All of the experiments described above rein-force the notion that the E-mAbs recognize adomain of the LamB protein which is differentfrom the domain recognized by the 1-mAbs. Thedomain recognized by the E-mAbs is clearlylocated at the cell surface. The experimentsdescribed below provided information regardingthe locations of the sites recognized by the I-mAbs.

Locations of the antigenic determinants whenLamB is inserted in a membrane. The E-mABdeterminants were known to be accessible whenLamB was inserted in the bacterial outer mem-brane, but very little was known about theaccessibility of the sites recognized by the I-mAbs or even about the existence of these sitesin the absence of detergent. A first set of experi-ments was performed after LamB was insertedin artificial phosphatidylcholine vesicles. At-tempts to evaluate the accessibility of the anti-genic sites by immunoprecipitation were unsuc-cessful due to the retention of the vesicles by theimmunoadsorbent. Instead, we developed a rapid immunoelectron microscopy technique inwhich the vesicles were preadsorbed to the gridsbefore reaction with the antibodies. As Fig. 5shows, two kinds of vesicles were identified.One population was composed of small clear

vesicles which were not labeled, and the otherpopulation was composed of large isolated vesi-cles, which were labeled by all of the anti-LamBmAbs, including 1-mAbs 436 and 141, but not bythe IDA 7 control immunoglobulin.The same type of experiment was then per-

formed with natural membrane fragments thatwere obtained after sonication of bacterialspheroplasts. These fragments were labeled byE-mAb 302 (Fig. 6a), showing that the externalface of the outer membrane was exposed inthese fragments. However, two lines of evi-dence indicated that these membranes were alsoaccessible from the periplasmic face. First,OmpA, an outer membrane protein which has aprotease-sensitive site at the inner face of themembrane (22), could be cleaved by trypsin(data not shown). Second, the aS serum whichdid not recognize external determinants ofLamB did interact with these fragments (Fig.6b).Although the results described above indicat-

ed that the periplasmic face of the membranewas accessible in the outer membrane fragmentsand even that some LamB determinants wereexposed on this face, no labeling was obtainedwith mAbs 436 and 141 (Fig. 6c and d). Thisresult was somewhat surprising since the deter-minants recognized by these mAbs had been

1 2 3 4 5

68

443

30

_I83

FIG. 4. mAb protection against proteolysis. 35S_labeled LamB (10 pLg/ml) was preincubated with 80 .Lgof the control IDA 7 mAb (lane 2), mAb 302 (lane 3),mAb 347 (lane 4), mAb 436 (lane 5), or mAb 141 (lane6) per ml. Subtilisin (17.5 p.g/ml) was then added, andthe samples were incubated for 24 h at 37°C. Lane 1was a control where LamB was incubated under thesame conditions but in the absence of antibodies andsubtilisin. Arrowheads indicate the positions of molec-ular weight marker proteins (in kilodaltons).

J. BACTERIOL.

on October 19, 2017 by guest

http://jb.asm.org/

Dow

nloaded from

VOL. 155, 1983 LamB ANTIGENIC DETERMINANTS 1387

'.~~~~~~~~~~~~~~~~~~~~~~kis,s <.0.~~~~~~,~~~~~-IJ'..'.' ,' t_ .

.to~~~~~~~~~~~::'. "

0*~~~~~~~~~~~~~~~~~~~~~~~~~0

2.~ ~~~~~.c d.

FIG. 5. Reaction of the mAbs with phosphatidylcholine vesicles containing inserted LamB. Reconstitutedvesicles were adsorbed onto grids and, after washing, incubated with 8 jig of mAb 302 per ml (a), 5 jig of controlmAb per ml (b), 5 R±g of mAb 141 per ml (c), or 10 pLg of mAb 436 per ml (d). The grids were then incubated withgold-labeled protein A as described in the text.

shown to be accessible in artificial vesicles (Fig. presence in the latter of lipoprotein, which an-5). chors the outer membrane to the peptidoglycanOne major difference between the artificial sacculus (22). To test whether the lack of reac-

membrane and the E. coli outer membrane is the tivity of mAbs 436 and 141 with the outer

on October 19, 2017 by guest

http://jb.asm.org/

Dow

nloaded from

1388 SCHENKMAN, COUTURE, AND SCHWARTZ

a . I. a

0. . .

.

0

*I.

0

a * * .^~~~~~

0,1tt c

..I.

. ~ S

sw+,{ *

* S

bI:

dFIG. 6. Immunoelectron microscopy of wild-type (Ipp+) membrane fragments. Vesicles prepared as de-

scribed in the text from strain pop3 were diluted 10-fold in PBS, adsorbed onto grids, and incubated with 8pug ofmAb 302 per ml (a), a 1/10 dilution of aS serum (b), 5 ,g of mAb 141 per ml (c), or 10 pug of mAb 436 per ml (d).The grids were washed, treated with gold-labeled protein A, washed, and negatively stained. Results identical tothose shown were obtained by using vesicles from strain JE5506, which is the lpp+ parent of strain JE5505.

J. BACTERIOL.

I'll

:110

0

on October 19, 2017 by guest

http://jb.asm.org/

Dow

nloaded from

LamB ANTIGENIC DETERMINANTS 1389

membrane fragments might be related to thepresence of lipoprotein, we prepared vesiclesfrom a mutant devoid of this protein (strainJE5505), as well as from its lpp+ parent (strainJE5506). The results obtained with strain JE5506were the same as those obtained with strainpop3 (data not shown), but they differed fromthose obtained with strain JE5505. I-mAb 436yielded strong labeling of the vesicles from thislpp mutant (Fig. 7c). Since mAb 436 failed togive a positive immunofluorescence responsewith intact bacteria, this labeling of JE5505membranes most probably corresponded to aninteraction with the inner face of the outermembrane. The other I-mAb, mAb 141, reactedvery poorly with the membrane fragments fromthe Ipp mutant (Fig. 7d).

DISCUSSION

The properties of the six mAbs which westudied are summarized in Table 1. These prop-erties allow us to define the following two dis-tinct, nonoverlapping domains in the LamB pro-tein: the E domain, which is exposed at the cellsurface, and the I domain, which is not.The E domain was previously shown to con-

tain at least two antigenic sites, one defined bymAbs 347 and 72, which inhibit maltose trans-port in intact bacteria, and the other defined bymAbs 302 and 177, which do not (8). In thispaper we show that these two determinants mustoverlap or at least be very close to one another.Indeed, the Fab fragment from mAb 302 inhibit-ed the binding of the three other E-mAbs. Inaddition, the four E-mAbs had identical effectson the X phage-neutralizing activity of LamB invitro and on LamB proteolysis. The simplestinterpretation of these results is that the sitesrecognized by the four E-mAbs share a regionwhich (i) is exposed at the bacterial surface invivo, (ii) is involved in the interaction of LamBwith phage X, and (iii) contains the primarytarget for proteolytic attack of LamB in vitro.

The I domain contains at least three antigenicsites. Two of these sites are defined by mAbs141 and 436. The third, which may consist ofseveral antigenic sites, is defined by antibodieswhich are present in aS, a polyclonal anti-LamBantiserum which has been preadsorbed withintact lamB+ bacteria. The sites recognized bymAbs 141 and 436 overlap or are close to oneanother since Fab fragments from mAb 436inhibit the binding of mAb 141. However, thesetwo sites are clearly distinct from those recog-nized by the E-mAbs since (i) the Fab fragmentsfrom mAb 436 failed to inhibit the binding of E-mAbs, (ii) the Fab fragments from E-mAb 302failed to inhibit the binding of mAbs 141 and 436,and (iii) mAbs 141 and 436 have no effect on thesensitivity of LamB to proteolysis or on itsability to neutralize phage X.

The sites recognized by the two I-mAbs areboth exposed when LamB is inserted into artifi-cial vesicles, but not when LamB is present inouter membrane fragments obtained from wild-type strains. Despite these common properties,the two sites are nevertheless clearly distinct.First, the site recognized by mAb 141 is exposedin LamB-peptidoglycan complexes, whereasthat recognized by mAb 436 is not (8). Second,the site recognized by mAb 436 is fully exposedin outer membrane fragments isolated from an

lpp mutant, whereas the site recognized by mAb141 is not. These results indicate that the siterecognized by mAb 436 is located on the peri-plasmic face of the membrane and is presumablyinvolved in the interaction of LamB with pepti-doglycan. More specifically, this site may corre-

spond to a region of LamB that interacts withlipoprotein, which is itself covalently bound topeptidoglycan. Indeed, the inaccessibility of thissite in outer membrane fragments from Ipp+bacteria is probably due to the presence of thelipoprotein itself and not to the presence of largepeptidoglycan fragments which might have re-sisted the lysozyme treatment used to prepare

TABLE 1. Main characteristics of the anti-LamB antibodies

Binding to LamB- Effect on A Protection of Binding to Binding to Binding toAntibody Domaina peptidoglycan neutralizing LamB artificial wild-type E. lpp mutant

b ~~~~~against veies clvsces eicscomplexesb activity proteolysis vesicles coli vesicles vesicles

mAb302 E + + + + + +mAb 177 E + + + NTC NT NTmAb 347 E + + + NT NT NTmAb 72 E + + + NT NT NTmAb 141 I + - - + - -/+mAb 436 I - - - + - +

a Domains E and I are defined based on immunofluorescence labeling of intact bacteria (8) and from Fabcompetition experiments.

b From reference 8.c NT, Not tested.

VOL. 155, 1983

on October 19, 2017 by guest

http://jb.asm.org/

Dow

nloaded from

1390 SCHENKMAN, COUTURE, AND SCHWARTZ J. BACTERIOL.

&P

I

2zL cSd,.''-.' '_

FIG. 7. Immunoelectron microscopy of membrane fragment from lpp strain JE5505. Membrane fragmentsprepared as described in the text were diluted 10-fold in PBS, adsorbed onto grids and incubated with 8 p.g ofmAb 302 per ml (a), 16 p.g of control mAb per ml (b), 5 pLg of mAb 141 per ml (c), or 10 p.g of mAb 436 per ml (d).The grids were washed, treated with gold-labeled protein A, washed, and negatively stained.

on October 19, 2017 by guest

http://jb.asm.org/

Dow

nloaded from

LamB ANTIGENIC DETERMINANTS 1391

FIG. 8. Model showing antigenic sites in LamBprotein. Lpo, Lipoprotein.

the vesicles. If such large fragments had beenpresent (21), they would have been expected toprevent trypsin from attacking the carboxyl endof the OmpA protein (22) and to prevent aS frominteracting with the membrane fragments. Thiswas not the case.The site recognized by mAb 141 is clearly not

involved in the interaction of LamB with pepti-doglycan since it is accessible in LamB-peptido-glycan complexes. Gabay and Schwartz (8) havesuggested that this site could be buried withinthe membrane. However, since (i) this site isfully accessible when LamB is inserted in phos-pholipid vesicles and (ii) it overlaps the site ofmAb 436, which is fully accessible in the samemembrane fragments it seems that the mAb 141site is not deeply embedded in the lipid bilayer;rather, it may be mostly located on the periplas-mic face of the outer membrane.The existence of at least one additional anti-

genic site in the I domain is suggested by the factthat unlike mAbs 141 and 436, the LamB-specif-ic serum which has been preadsorbed withlamB+ cells, can react with outer membranevesicles from a wild-type strain. The fact that aSreacts with outer membrane vesicles of lpp+bacteria indicates that a region of LamB must beaccessible at the periplasmic face of the outermembrane, even in the presence of lipoprotein.Part or all of this region could be the regionwhich interacts with the periplasmic maltose-binding protein (1).Figure 8 shows our present view of the loca-

tion of the antigenic sites studied in this workwith respect to the outer membrane and tofunctional sites on the LamB protein. This anal-ysis is now being pursued in two directions. Oneinvolves the mapping of the antigenic sites onthe LamB polypeptide. In this respect the siterecognized by E-mAbs 177 and 302 has been

shown to be located in the vicinity of the COOH-terminal end of the polypeptide (7). The otherdirection is the characterization of additionalantigenic determinants. The results of competi-tion experiments performed with Fab fragmentsand complete sera, both preadsorbed with wholecells and not preadsorbed (Fig. 2), indicate thatthere are determinants which do not overlapthose described here.

ACKNOWLEDGMENTS

We gratefully acknowledge J. Gabay, who made this workpossible by isolating the hybridoma cell lines. We thank J.Gabay, A. P. Pugsley, A. Ryter, and H. Shuman for helpfulsuggestions.

This work was supported by grants LA 269 and 270 from theCentre National de la Recherche Scientifique and by grant 82V 1279 from the Ministbre de l'Industrie et de la Recherche.S.S. was a postdoctoral fellow from Conselho Nacional deDesenvolvimento Cientffico e Tenol6gico (CNPq, Brazil).

LITERATURE CITED1. Bavoil, P., and H. Nikaido. 1981. Physical interaction

between the phage lambda receptor and the carrier immo-bilized maltose-binding protein of Escherichia coli J. Biol.Chem. 256:11385-11388.

2. Braun, V., and H. J. Krieger-Brauer. 1977. Interrelation-ship of the phage X receptor protein and maltose transportin mutants of Escherichia coli K12. Biochim. Biophys.Acta 469:89-98.

3. Braun, V., and K. Rehn. 1969. Chemical characterization,spatial distribution and function of a lipoprotein (murein-lipoprotein) of the E. coli cell wall. The specific effect oftrypsin on the membrane structure. Eur. J. Biochem.10:426-436.

4. Casadaban, M. J. 1976. Transposition and fusion of thelac genes to selected promoters in Escherichia coli usingbacteriophage lambda and Mu. J. Mol. Biol. 104:541-555.

5. Clement, J. M., and M. Hofnung. 1981. The sequence ofthe X receptor, an outer membrane protein of E. coli K12.Cell 27:507-514.

6. Ferenci, T., M. Schwentorat, S. Ullrich, and J. Vilmart.1980. Lambda receptor in the outer membrane of Esche-richia coli as binding protein for maltodextrins and starchpolysaccharides. J. Bacteriol. 142:521-526.

7. Gabay, J., S. Benson, and M. Schwartz. 1983. Geneticmapping of antigenic determinants on a membrane pro-tein. J. Biol. Chem. 258:2410-2414.

8. Gabay, J., and M. Schwartz. 1982. Monoclonal antibodyas a probe for structure and function of an Escherichia coliouter membrane protein. J. Biol. Chem. 257:6627-6630.

9. Gabay, J., and K. Yasunaka. 1980. Interaction of theLamB protein with the peptidoglycan layer in Escherichiacoli K12. Eur. J. Biochem. 104:13-18.

10. Geoghegan, W. D., and G. A. Ackermann. 1977. Adsorp-tion of horseradish peroxidase, ovomucoid and anti-immunoglobulin to colloidal gold for the indirect detectionof concanavalin A, wheat germ agglutinin and goat anti-human immunoglobulin G on cell surfaces at the electronmicroscopic level. A new method, theory and application.J. Histochem. Cytochem. 25:1187-1200.

11. Heuzenroeder, M. W., and P. Reeves. 1980. Periplasmicmaltose-binding protein confers specificity on the outermembrane maltose pore of Escherichia coli. J. Bacteriol.141:431-435.

12. Hofnung, M., A. Jezierska, and C. Braun-Breton. 1976.lamB mutations in E. Coli K12: growth of X host rangemutants and effect of nonsense suppressors. Mol. Gen.Genet. 145:207-213.

13. Hudson, L., and F. C. Hay. 1980. Practical immunology.2nd ed. Blackwell Scientific Publications, Oxford.

VOL. 155, 1983

on October 19, 2017 by guest

http://jb.asm.org/

Dow

nloaded from

1392 SCHENKMAN, COUTURE, AND SCHWARTZ

14. Ishii, J. N., Y. Okajina, and T. Nakae. 1981. Characteriza-tion of LamB protein from the outer membrane of Esche-richia coli that forms diffusion pores selective for maltose-maltodextrins. FEBS Lett. 134:217-220.

15. Ito, K., T. Date, and W. Wickner. 1980. Synthesis,assembly into the cytoplasmic membrane, and proteolyti-cal processing of the precursor or coliphage M13 coatprotein. J. Biol. Chem. 255:2123-2130.

16. Lowry, 0. H., N. J. Rosebrough, A. L. Farr, and R. J.Randall. 1951. Protein measurement with the Folin phenolreagent. J. Biol. Chem. 193:265-275.

17. Luckey, M., and H. Nikaido. 1980. Specificity of diffusionchannels produced by X phage receptor of Escherichiacoli. Proc. Nati. Acad. Sci. U.S.A. 77:167-171.

18. Luckey, M., and H. Nikaido. 1983. Bacteriophage lambdareceptor protein in Escherichia coli K12: lowered affinityof some mutant proteins for maltose-binding protein invitro. J. Bacteriol. 153:1056-1059.

19. Marchal, C., and M. Hofnung. 1983. Negative dominancein gene lamB: random assembly of secreted subunitsissued from different polysomes. EMBO J. 2:81-86.

20. Neuhaus, J. M. 1982. The receptor protein of phage X:purification, characterization and preliminary electricalstudies in planar lipid bilayers. Ann. Microbiol. (Paris)133A:27-32.

21. Osborn, M. J., J. E. Ganden, E. Parisi, and J. Carson.1982. Mechanism of assembly of the outer membrane ofSalmonella typhimurium. Isolation and characterizationof cytoplasmic and outer membrane. J. Biol. Chem.247:3962-3972.

22. Osborn, M. J., and H. C. P. Wu. 1980. Proteins of theouter membrane of Gram-negative bacteria. Annu. Rev.Microbiol. 34:369-422.

23. Palva, E. T. 1979. Protein interactions in the outer mem-brane of Escherichia coli. Eur. J. Biochem. 93:495-503.

24. Randall-Hazelbauer, L., and M. Schwartz. 1973. Isolationof the bacteriophage lambda receptor from Escherichiacoli. J. Bacteriol. 116:1436-1446.

25. Roa, M. 1979. Interaction of bacteriophage K1O with itsreceptor, the lamB protein of Escherichia coli. J. Bacteri-ol. 140:680-686.

26. Rosenbusch, J. P., B. M. Garavito, D. L. Dorset, and A.Engel. 1982. Structure and function of a pore-formingtransmembrane protein: high resolution studies of a bacte-rial porin, p. 171-174. In H. Peeters (ed.), Protides of thebiological fluids, 29th Colloquium-1981. PergamonPress, Oxford.

27. Schwartz, M., and L. Le Minor. 1975. Occurrence of thebacteriophage lambda receptor in some Enterobac-teriaceae. J.Virol. 15:679-685.

28. Shuman, H. A. 1982. The maltose-maltodextrin transportsystem of Escherichia coli. Ann. Microbiol. (Paris)133A:153-159.

29. Suzuki, H., Y. Nishimura, S. Yasuda, A. Nishimura, M.Yamada, and Y. Hirota. 1978. Murein-lipoprotein of Esch-erichia coli: a protein involved in the stabilization ofbacterial cell envelope. Mol. Gen. Genet. 167:1-9.

30. Szmelcman, S., and M. Hofnung. 1975. Maltose transportin Escherichia coli K12: involvement of the bacteriophagelambda receptor. J. Bacteriol. 124:112-118.

31. Wandersman, D., M. Schwartz, and T. Ferenci. 1979.Escherichia coli mutants impaired in maltodextrin trans-port. J. Bacteriol. 140:1-13.

32. Yamada, H., T. Nogami, and S. Mizushima. 1981. Ar-rangement of bacteriophage lambda receptor protein(LamB) in the cell surface of Escherichia coli: a reconsti-tution study. J. Bacteriol. 147:660-669.

J. BACTERIOL.

on October 19, 2017 by guest

http://jb.asm.org/

Dow

nloaded from