Proc. Natl. Acad. Sci. USAVol. 90, pp. 3033-3037, April 1993Cell Biology

Carbohydrate binding activities of Bradyrhizobium japonicum:Unipolar localization of the lectin BJ38 on the bacterial cell surface

(symbiosis/bacterial adhesion/confocal fluorescence microscopy)

JOHN T. LOH*, SIU-CHEONG Ho*, ADRIAAN W. DE FEIJTERt, JOHN L. WANG*, AND MELVIN SCHINDLER**Department of Biochemistry, Michigan State University, East Lansing, MI 48824; and tMeridian Instruments, Inc., 2310 Science Parkway, Okemos, MI48864

Communicated by N. Edward Tolbert, December 21, 1992

ABSTRACT A polyclonal antiserum generated against theBradyrhizobiumjaponicum lectin BJ38 was characterized to bespecifically directed against the protein. Treatment of B.japonicum cells with this antiserum and subsequent visualiza-tion with transmission electron microscopy and both conven-tional and confocal fluorescence microscopy revealed BJ38 atonly one pole of the bacterium. BJ38 appeared to be organizedin a tuft-like mass, separated from the bacterial outer mem-brane. BJ38 localization was coincident with the attachmentsite for (i) homotypic agglutination to other B. japoncum cells,(it) adhesion to the cultured soybean cell line SB-1, and (iu)adsorption to Sepharose beads covalently derivatized withlactose. In contrast, the plant lectin soybean agglutinin labeledthe bacteria at the pole distant from the bacterial attachmentsite. These results indicate that the topological distribution ofBJ38 is consistent with a suggested role for this bacterial lectinin the polar binding ofB.japonium to other ceUs and surfaces.

Bradyrhizobiumjaponicum normally infects roots ofsoybeanplants leading to symbiotic nitrogen fixation (1-3). We havedocumented (4) that this bacterial strain exhibits four sac-charide-specific binding activities: (i) adsorption to Sepha-rose beads derivatized with lactose (Lac-Sepharose), (ii)homotypic autoagglutination, (iii) heterotypic binding to cul-tured soybean (SB-1) cells, and (iv) heterotypic adhesion tosoybean roots. In all four of these assays, galactose inhibitedthe binding but a C-2 derivative of the monosaccharideN-acetyl-D-galactosamine failed to yield the same effect.Mutants ofB.japonicum, designated as N4 and N6, that wereisolated on the basis of a defect in one binding activity (SB-1cell binding) showed a concomitant loss in the other threebinding capacities (4). These observations suggested that allfour of the carbohydrate-specific binding processes may bemediated by the same component(s) and mechanism(s). Con-sistent with this proposal, we have purified a carbohydratebinding protein, designated BJ38 (Mr 38,000), which has anaffinity for galactose and lactose "13- and 240-fold higher,respectively, than its affinity for N-acetyl-D-galactosamine(5). This carbohydrate binding specificity correlates well withthat of the B. japonicum binding activities. It was proposed,therefore, that BJ38 may mediate the carbohydrate bindingactivities of B. japonicum.

In all four carbohydrate-specific binding activities tested,B. japonicum bound in a polar fashion (4). Thus, any hy-pothesis implicating a role of BJ38 in these binding assayswould require that the lectin be exposed at the surface of thebacterium where attachment occurs. In the present commu-nication, we report the characterization of a specific anti-BJ38 antibody that has allowed us to identify the topologicaldistribution of BJ38 on the bacterial cell surface.

MATERIALS AND METHODSCeUl Cultures. B. japonicum (RilOd) was originally ob-

tained from Barry Chelm of Michigan State University. Thisbacterial strain was maintained on agar plates containingyeast extract-sodium gluconate medium (YEG) and 50 mMlactose. YEG contained 1.28 g of K2HPO4, 0.2 g ofMg2SO4 7H20, 7.35 mg ofCaCl2 2H20, 28 mg of sequestrene,5.0 g ofgluconic acid, 1.0 g ofyeast extract, and H20 to 1 literand was supplemented with the following trace elements: 2.5mg of Na2EDTA, 4.39 mg of ZnSO4-7H20, 0.77 mg ofMnSO4'H20, 0.15 mg of CuSO4 5H20, 2.49 mg ofNa2MoO4 2H20, 0.23 mg of CoCl26H20, 0.46 mg ofNa2B407 10H20, 0.38 mg of Na3VO4, and 0.1 mg of NaSeO3(pH 6.0). The bacteria (from a 3-day-old culture) weretransferred from the agar plate to 50 ml of YEG in a 125-mlErlenmeyer flask and cultured for 1 day. The bacterialsuspension was inoculated into 2 liters ofYEG and culturedfor 2 days on a gyratory shaker (120 rpm) at 30°C. Aliquotsof this culture (300 ml) were inoculated into six Fernbachflasks, each containing 1.5 liters of YEG. The bacteria werefurther cultured for %'30 h until an A620 value of 1.7-2.0 wasobtained. The bacterial cells were used for the isolation ofBJ38, capsular polysaccharide (CPS), and lipopolysaccha-ride (LPS) and for immunolocalization studies.The SB-1 cell line, derived from soybean roots [Glycine

max (L) Merr. cv. Mandarin], was kindly provided by G.Lark (Department of Biology, University of Utah, Salt LakeCity). Cultures were grown in 50 ml of 1B5C medium at 27°Cin a gyratory shaker in the dark (6). Continuous cultures weremaintained by subdividing cells every 4 days, by transferring15 ml of culture to 50 ml of fresh 1B5C.

Isolation ofBJ38 and Generation of Antibody Reagents. Thebacterial lectin BJ38 was isolated by affinity chromatographyon a Lac-Sepharose column as described (5). Lactose-elutedfractions containing the lectin were identified by SDS/PAGEon 10%o polyacrylamide gels (7). Polypeptides were identifiedby silver staining (8) or by immunoblot analysis.

Fractions containing BJ38 isolated from 100 liters of bac-terial culture were concentrated by acetone precipitation (9)and subjected to SDS/PAGE. Gel slices containing BJ38,revealed by silver staining, were excised. This sample, con-taining ==10 ,ug of BJ38, was homogenized with Titer-Maxadjuvant (CytRx Corp., Norcross, GA) and injected into afemale New Zealand White rabbit. Booster injections of 6 .gof BJ38 in Titer-Max adjuvant were administered every 3weeks. Antiserum from the rabbit was collected 1 week aftereach booster injection. Monospecific antibodies directedagainst BJ38 were purified from the antiserum by using themethod of Smith and Fisher (10).

Abbreviations: CPS, capsular polysaccharide; LPS, lipopolysaccha-ride; BSA, bovine serum albumin; SBA, soybean agglutinin; FITC,fluorescein isothiocyanate; 3D, three dimensional; PI, propidiumiodide.

3033

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

July

29,

202

1

Proc. Natl. Acad. Sci. USA 90 (1993)

Anti-Brj antiserum was obtained from a rabbit immunizedwith heat-killed B. japonicum (6). This antiserum cross-reacted mainly with LPS and with other antigens from B.japonicum. The CPS and LPS fractions ofB.japonicum wereisolated by the methods of Mort and Bauer (11) and Carlsonet al. (12), respectively.For immunoblot analysis, proteins were transferred to the

Immobilon-P (Millipore) membrane by electrophoresis (400mA, 2 h, and 25°C). The membrane was blocked by 5%(wt/vol) bovine serum albumin (BSA) in 20mM Tris HCl/500mM NaCl, pH 7.5. Primary antibodies were used at 1:20dilution, and goat anti-rabbit IgG coupled to alkaline phos-phatase (Sigma) was used at 1:1000 dilution as secondaryantibody. The immunoreactive material was developed withnitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phos-phate as substrates.

Fluorescence Microscopy and Confocal Fluorescence Imag-ing. The bacterial cells were attached to coverslips (13), fixedin 2% (wt/vol) paraformaldehyde in phosphate-buffered sa-line (PBS) for 30 min, and washed in 1 M glycine for 20 minwith two changes. The cells were then permeabilized with50% ethanol, labeled with propidium iodide (PI; 10 ,ug/ml inPBS) for 2 h, and washed three times with PBS. The samplewas incubated for 4 h with a 1:20 dilution (in 5% BSA/YEG)of preimmune serum, anti-BJ38 antiserum, or anti-Brj anti-serum. After three 10-min washes in YEG, fluorescein iso-thiocyanate (FITC)-conjugated goat anti-rabbit antibody(1:50 dilution; Boehringer Mannheim) in 5% BSA/YEG wasadded and incubated for 1 h. The samples were then washedwith PBS. A similar analysis was also carried out on B.japonicum bacteria bound to SB-1 cells. B. japonicum cellswere incubated with SB-1 cells (2-day culture) in a Petri dishfor 3 h. The mixture was transferred to 5-ml culture tubes toallow the bound fraction to settle to the bottom of the tube.Unbound B. japonicum cells were removed and the boundcells were fixed in 2% paraformaldehyde in YEG for 30 min.Labeling with anti-BJ38 was then performed as above. B.japonicum cells bound to SB-1 cells were also labeled withFITC-conjugated soybean agglutinin (SBA, Sigma; 5 ,ug/ml)for 30 min, followed by two 10-min washes in PBS.The coverslip was placed on a microscope slide, sealed

with Paraffin, and scanned with an ACAS 570 InteractiveLaser Cytometer (Meridian Instruments, Okemos, MI),equipped with confocal optics. Two photomultipliers wereused to simultaneously detect fluorescence emission of flu-orescein (530 nm) and PI (605 nm). Images (360 x 360 pixels)were collected in 0.1-Am steps in the x and y directions.Sectioning was achieved by closing the pinhole aperture infront of the detection system to 60, 80, or 100 ,um andcollecting the fluorescence emission at intervals (specified infigure legends) in the axial direction. Three-dimensional (3D)reconstructions were generated by applying a modified ver-sion (14) of the simulated fluorescence process (SFP) algo-rithm (15) to the data. Background fluorescence was math-ematically removed, and a smoothing filter was applied to thedata prior to reconstruction. A Laplacian filter was used tosharpen the resulting reconstructions.For the analysis of the relative location of BJ38 and SBA

on B. japonicum cells that are autoagglutinated, the bacterialcells were first grown on a YEG agar plate for 5 days. Thecells were fixed onto coverslips (13), blocked in 5% BSA/YEG for 2 h, and then labeled with SBA or anti-BJ38 asdescribed above. Stained cells were visualized with a fluo-rescence microscope (Zeiss D-7082 Oberkochen; excitationfilter, 546 nm; chromatic beam splitter, 580 nm; barrier filter,590 nm).

Electron Microscopy. Both B. japonicum cells from sus-pension culture and from solid agar YEG medium (star-formning bacteria) were used. These cells were fixed for 15min on ice in 0.25% glutaraldehyde (EM grade, Electron

Microscopy Sciences, Fort Washington, PA) in YEG. Thecells were washed with PBS by centrifugation (2000 x g for15 min) and resuspension and suspended in 5% BSA/PBS.After 1 h on ice, the samples were incubated with anti-BJ38,anti-Bri, or normal rabbit sera (1:20 dilution in BSA/PBS, 4h on ice). The samples were washed three times and labeledwith goat anti-rabbit IgG coupled with colloidal gold (1 h at1:5 dilution, particle size of 20 nm; Polysciences). The cellswere washed three times, resuspended in 4 ml of PBS, andstored at 4°C overnight. The bacteria were centrifuged andresuspended in 0.2 ml of PBS before being applied to thegrids.The formvar-coated grids were first treated with 1 M HCI

(5 min) and coated with 1% polylysine (Mr = 100,000; Sigma)in water (5 min). A drop of the bacterial sample was thenapplied and allowed to settle on the grid for 15 min. Excesssample was blotted dry. The grids were washed three timeswith water to remove residual salt. The samples were eitherdirectly observed under an electron microscope or stainedwith 1% phosphotungstic acid (pH 7.2) before observation.

RESULTSAntibodies Directed Against BJ38. Purified BJ38 was sub-

jected to SDS/PAGE, and the gel slices corresponding to theMr 38,000 region were used as the immunogen for generatinga rabbit antiserum against the lectin. Rabbit anti-BJ38 yieldeda single band (Mr = 38,000) on an immunoblot of the BJ38sample (Fig. 1, lane a). This band was not observed withcontrol preimmune serum (Fig. 1, lane h). However, the sameBJ38 sample yielded multiple bands when analyzed withanti-Bri antibody (Fig. 1, lane e). This latter antibody hadbeen generated against heat-killed B.japonicum bacteria andhad been shown to react with LPS and other bacterialcomponents (6). The "ladder" pattern obtained when theBJ38 sample was analyzed with anti-Brj (Fig. 1, lane e) wassimilar to those obtained when LPS or CPS isolated from B.japonicum were analyzed with the same antibody (Fig. 1,lanes f and g). These results suggest that although the BJ38sample was purified in terms of polypeptides, there wasnonetheless a small amount of contamination by LPS andCPS.For the purpose of the intended use of the anti-BJ38

antiserum as a reagent for immunolocalization, it was impor-tant to establish that rabbit anti-BJ38 reacted only with BJ38and not with any contaminating polysaccharides. The resultsindicate that rabbit anti-BJ38 did not react with either LPS orCPS fractions derived from B.japonicum (Fig. 1, lanes b andc). As an additional precaution, we affinity-purified theanti-BJ38 bound to the Mr 38,000 band on immunoblots. Thisaffinity-purified monospecific antibody preparation yielded

a b c d e f g h i i

FIG. 1. Immunoblot analyses of BJ38, CPS, and LPS isolatedfrom B. japonicum. Lanes: a, d, e, and h, BJ38 sample (0.1 jUg), the0.1 M lactose fraction from a B. japonicum cell extract fractionatedon a Lac-Sepharose column; b, f, and i, LPS (0.4 pAg); c, g, andj, CPS(0.4 pug); a-c, anti-BJ38; d, monospecific anti-BJ38; e-g, anti-Brj; h-j,preimmune serum. The arrow on the left indicates the position ofmigration of a polypeptide ofMr 38,000, relative to known molecularweight markers.

3034 Cell Biology: Loh et al.

Dow

nloa

ded

by g

uest

on

July

29,

202

1

Cell Biology: Loh et al. ~~~~Proc. Nati. Acad. Sci. USA 90 (1993) 3035

results identical to the original anti-BJ38 antiserum in animmunoblot (e.g., Fig. 1, lane d) and in immunofluorescenceexperiments.

Confocal Immunofluorescence Microscopy of BJ38 on the B.japonicum Cell Surface. When B. japonicum cells were la-beled with anti-BJ38 antiserum, immunofluorescent stainingof BJ38 was observed uniquely at one pole of the cell (Fig.2A). We employed confocal fluorescence microscopy and 3Dreconstruction of image sections to evaluate the distributionof BJ38 on all bacteria, independent of bacterial orientationin relation to the focal plane. All reconstructed images aretwo-colored: orange identifies the DNA-binding PI label(utilized as a marker for bacterial volume) and green identi-f'ies the FITC-labeled antibody to localize BJ38. Bacteriaprobed for BJ38 show labeling in a tuft-like structure at onepole of the cell. The remainder of the cell surface was notlabeled. A similar pattern of labeling was also observed withthe monospeciflc aff'inity-purified anti-BJ38 antibody. Underour optimal conditions for labeling, -70% of the bacterialpopulation showed anti-BJ38 staining.The tuft-like structure containing anti-BJ38 staining is seen

to extend from the bacterial surface in contact with thesubstrate (coverslip). The bacteria appear to adhere at thispoint and project away from the glass surface, casting a

"shadow" (Fig. 2B). A fluorescence intensity profile acrossthe long axis of the bacteria was prepared for both stains toprovide a quantitative evaluation of the antigen localization(Fig. 2C). The red curve shows the PI fluorescence intensityand represents the bacterial length. The green curve showsthe distribution of BJ38 and confirms that BJ38 is localized ina polar fashion. There was also antibody label (green) appar-ent on the glass surface without any observable attachedbacteria (Fig. 2A). One interpretation is that this representslectin-containing tufts that have detached from the bacteriaand suggests the fragile nature of the association of thismaterial to the cell surface.When anti-Bri antibody was used to label the B.japonicum

cells, the staining was observed throughout the entire cellsurface (Fig. 2 D-fl. In viewing these micrographs, it shouldbe kept in mind that the green color is due to anti-Brj stainingand the orange color is due to the PI label. The mixing of thegreen and orange colors gives rise to a yellowish region (Fig.2E) and, in certain cases, this creates an impression thateither the green or yellow distribution may be polar. In thefluorescence intensity profile (Fig. 2F), the anti-Bri staining(green curve) reflects the localization of LPS at the outermembrane and thus defines the length of the bacterium, andthe PI intensity (red curve) is at its maximum toward the

C

F

I

FIG. 2. (A) Confocal fluorescence imaging and 3Dreconstruction of B. japonicum labeled with PI andrabbit anti-BJ38 plus FITC-conjugated goat anti-rabbitimmunoglobulin. Fluorescence: PI, orange; anti-BJ38,green. A total of 22 sections (100-Am pinhole apertureand 0.3-am intervals) were scanned to generate the 3Dreconstructions. (Bar = 5 pLm.) A single bacterial imagefrom A was enhanced by a factor of 8 through pixelinterpolation (B); the fluorescence intensity profile (C)was prepared by scanning along the length of thebacterium. Red curve, PI intensity corresponding tobacterial length; green curve, distribution of BJ38. (D)Same as A except the cells were labeled with PI(orange) and rabbit anti-Brj (green) and 25 sectionswere scanned. Enhanced view of the distribution ofanti-Bri (E) and fluorescence intensity profile (F) on asingle cell from D. Red curve, PI intensity; greencurve, localization of the LPS on the outer membrane.(G) Same as A, except the cells were labeled with PI(orange) and preimmune serum (green) and 15 sectionswere scanned. Enhanced view of the distribution ofpreimmune antibody (H) and fluorescence intensityprofile (I) on a single cell from G. Red curve, PIintensity; green curve, intensity profile due to preim-mune serum. (J) Confocal fluorescence imaging and 3Dreconstruction of B.japonicum cells bound to culturedsoybean (SB-i) cells. Fluorescence: PI, orange; anti-BJ38, green. A total of 25 sections (60-gAm pinholeaperture and 0.3-gtm intervals) were scanned for 3Dreconstruction. (K) Same as J, except the cells werelabeled with PI (orange) and FITC-conjugated SBA(green). A total of 14 sections (80-Am pinhole apertureand 0.3-kLm intervals) were scanned for 3D reconstruc-tion.

Cell Biology: Loh et al.

Dow

nloa

ded

by g

uest

on

July

29,

202

1

Proc. Natl. Acad. Sci. USA 90 (1993)

central less-curved segment of the cell. This conclusion issupported by immunofluorescence staining of B. japonicumcells with only anti-Brj (without the complication of the PIlabel). On the basis of these considerations, the pattern oflabeling with anti-Brj is consistent with a uniform distributionon the cell surface.

Control samples, in which bacteria were labeled withpreimmune serum, are shown in Fig. 2 G and H. Neitherbacteria nor the glass surface is stained. This is also reflectedby the lack of any intensity in the green curve of thefluorescence distribution profile (Fig. 2I).

Ultrastructural Confirmation of Polar Localization of BJ38.To confirm the polar localization of BJ38 on B. japonicum atthe ultrastructural level, B. japonicum cells were first incu-bated with anti-BJ38 and then incubated with colloidal gold-labeled goat anti-rabbit IgG. BJ38 was found at one pole ofthe bacterial cell (Fig. 3 A and B), as was seen with theimmunofluorescence data. Moreover, like the reconstructedconfocal images, BJ38 appeared to be in a tuft-like mass,localized at a distance away from the outer membrane of thebacteria. A similar result was obtained by negative stainingwith phosphotungstic acid after immunogold labeling. Thepositions of the gold particles did not appear to be associatedwith any filamentous or pilus structure. As expected, whenanti-Brj antibody was used as the primary antibody, immu-nogold labeling was observed throughout the cell surface(Fig. 3E). Preimmune serum showed no labeling of the cellsurface (Fig. 3F).

Identification of BJ38 at the Point of Contact. When B.japonicum was cultured on YEG agar plates, the bacteriatend to form aggregates (stars) by attaching to each other atone of their tips. Previous studies had shown that this starformation can be inhibited by galactose and lactose, thusimplicating a role for the lectin BJ38 (4). When these star-forming bacteria were labeled with anti-BJ38, immunofluo-rescence was observed at the focal point of cell-cell contact(Fig. 4A). This result suggests that BJ38 is present at thecontact point where the bacteria interact with each other.This conclusion is supported by the results of electron

A B.:1

FIG. 3. Immunogold labeling of B. japonicum cell surface byanti-BJ38, anti-Brj, and preimmune sera. (A and B) Cells fromsuspension culture. (C-F) Cells from star aggregates grown on a

YEG agar plate. Suspension cells or star aggregates were fixed withglutaraldehyde and incubated with anti-BJ38 (A-D), anti-Brj (E), or

preimmune (F) serum. The samples were then labeled with goatanti-rabbit IgG antibody coupled with colloidal gold (particle size, 20nm). (Bar = 0.5 ,tm.)

PH FL

A

.,.....

.*a :*

M. ....

j......B 2^*-

...... *4. s-W. * . .. , '\ :*,

FIG. 4. Fluorescent labeling of B. japonicum cells that wereattached to other B. japonicum cells in star formation. (A) Anti-BJ38(1:20 dilution for 4 h) followed by FITC-conjugated goat anti-rabbitimmunoglobulin (1:50 dilution for 1 h). (B) FITC-SBA (5 ,g/ml and30 min). The arrows highlight the focal point of contact in theagglutinated cells. PH, phase-contrast; FL, fluorescence. (Bar = 3AM.)

microscopic studies using immunogold labeling techniques(Fig. 3 C and D).When B. japonicum was allowed to attach to SB-1 cells,

anti-BJ38 stained the bacteria at the point ofcontact (Fig. 2J).No staining was found at the pole pointing away from theattachment site. Similarly, the point of contact was alsolabeled with anti-BJ38 when the bacteria were bound toLac-Sepharose beads (data not shown). These results ob-tained with anti-BJ38 were distinctly different from thecorresponding data using SBA.When FITC-SBA was used to label the star-forming B.

japonicum cells, fluorescence labeling was shown to be at thepole of the bacterium away from the attachment point (Fig.4B). No fluorescent staining was observed at the center ofthestars, where B. japonicum cells attached to one another.Similarly, SBA labeled B. japonicum cells bound to SB-1cells exclusively at the pole opposite from the point of contact(Fig. 2K). Again, the "shadowing" effect helps in terms oforientation; the SBA-labeled pole projects away from theattachment point on the plant cell wall. These data indicatethat BJ38 and the glycoconjugate receptors for SBA arefound on opposite poles of the bacteria.

DISCUSSIONThe experiments reported in this paper show that the lectinBJ38 is found at the cell surface. More strikingly, BJ38 islocalized at the pole of the bacterial cell that is involved in (i)homotypic adhesion to other star-forming B. japonicum, (ii)heterotypic binding to SB-1 cells, and (iii) attachment toLac-Sepharose beads. All of these binding processes occur ina polar fashion (4). Indeed, the data also indicate that BJ38can be found exclusively in the region of contact between thebacterium and the surface to which it is bound.The plant lectin SBA has been proposed to mediate the

binding of B. japonicum to the soybean (16, 17). AlthoughSBA binding to B. japonicum cells also yielded a polardistribution (17-20), it is important to emphasize that the twoends labeled by anti-BJ38 and SBA are opposite to eachother. The pole bound by SBA is probably not directlyinvolved in mediating bacterial attachment. When B. japoni-cum is bound to cultured SB-1 cells, FITC-SBA labeled thefree end of the bacteria, away from the point of contactbetween the bacterium and the plant cell. Similarly, FITC-SBA also labeled star-forming bacteria at the ends away from

3036 Cell Biology: Loh et al.

_ .#..

E

Dow

nloa

ded

by g

uest

on

July

29,

202

1

Proc. Natl. Acad. Sci. USA 90 (1993) 3037

the center of the cluster. The SBA-labeled pole, therefore,does not fit the criteria for directly mediating attachment ofthe bacteria to soybean cells or to other bacteria. Our presentobservations are consistent with, and perhaps could provideexplanations for, two key observations documented previ-ously: (i) N-acetyl-D-galactosamine, a potent hapten inhibi-tor of SBA, failed to inhibit B. japonicum binding to soybeanroots (4, 6, 21), and (ii) an excess amount of SBA, sufficientto saturate SBA receptor sites on B.japonicum, also failed toinhibit the same binding (22). It appears, therefore, that anyproposed involvement of SBA in facilitating B. japonicumnodulation activity (23) must be related to mechanisms otherthan mediating bacterial attachment.The polar localization of BJ38 at the cell surface is also

reflected by its internal subcellular structures. The bacteriumis divided into two distinct poles: the nucleoid portioncontaining chromosome and cytoplasmic components andthe reserve polymer half containing glycogen and poly(3-hydroxybutyrate) granules (24, 25). Ruthenium red stainingindicates that SBA-binding polysaccharides are found in thenucleoid portion ofthe bacteria (24). In contrast, the granularportion is the half with which the bacterium participates instar formation. In the present study, we have shown thatBJ38 and the SBA receptor reside at opposite ends of B.japonicum. The granular portion is thus the end at whichBJ38 would be located.At the tip of the granular portion of B. japonicum, Tsien

and Schmidt (25) observed tuft-like structures in B. japoni-cum and named them extracellular polar bodies. They con-tain fibrillar material of both polysaccharide and protein. Itwas suggested that these extracellular polar bodies mediatestar formation. Indeed, our own ultrastructural localizationof BJ38 in a tuft-like mass at a distance away from thebacterial outer membrane is consistent with this hypothesisand with the proposal that the lectin plays a key role in theautoagglutination event.On the other hand, neither our electron microscopic nor

confocal fluorescence studies indicate that BJ38 is associatedwith any well-defined proteinaceous filaments. Such filamen-tous pili have been found in Rhizobium lupini (26), Rhizobiummeliloti, Rhizobium leguminosarum (27), and B. japonicum(28) and have been proposed to mediate attachment. It hasbeen reported that the isolated pili of B. japonicum consistsof protein subunits with Mr 21,000 and Mr 18,000 (28). Thesubunit molecular weight of BJ38 makes it unlikely that it isa part of the isolated pilus structure. Moreover, our previousgel-filtration data, in which the purified BJ38 protein chro-matographed to a position corresponding to Mr 38,000 (5),suggest that the lectin most likely does not form polymersleading to a filamentous structure.

In toto, the results detailed in this paper provide furtherevidence to support the proposed role of BJ38 in mediatingthe carbohydrate binding properties of B. japonicum. Notonly is the BJ38 saccharide specificity similar to that of B.japonicum (4, 5) but also its cell surface distribution isconsistent with the polar binding activities of the bacteria.The importance of this latter observation is further demon-strated in studies with the N4 and N6 mutants (29). These

mutants showed no detectable amounts of BJ38 and a de-creased ability to nodulate soybean roots as compared to thewild-type B.japonicum. These results indicate that BJ38 mayplay an important role in the symbiotic infection process.

We thank Dr. Richard Hallgren for his valuable assistance in thepreparation of the 3D reconstructions and Dr. Natasha Raikhel forthe use of the Zeiss fluorescence microscope. We also thank Mrs.Linda Lang for her help in the preparation of the manuscript. Thiswork was supported by Grant GM45200 from the National Institutesof Health and by the Michigan Agricultural Experiment Station.

1. Bauer, W. D. (1981) Annu. Rev. Plant Physiol. 32, 407-449.2. Halverson, L. J. & Stacey, G. (1986) Microbiol. Rev. 50,

193-225.3. Ho, S.-C. & Kijne, J. W. (1991) in Lectin Reviews, eds.

Kilpatrick, D. C., van Driessche, E. & Bog-Hansen, T. C.(Sigma, St. Louis), Vol. 1, pp. 171-181.

4. Ho, S.-C., Wang, J. L. & Schindler, M. (1990) J. Cell Biol. 111,1631-1638.

5. Ho, S.-C., Schindler, M. & Wang, J. L. (1990) J. Cell Biol. 111,1639-1643.

6. Ho, S.-C., Ye, W., Schindler, M. & Wang, J. L. (1988) J.Bacteriol. 170, 3882-3890.

7. Laemmli, U. K. (1970) Nature (London) 227, 680-685.8. Blum, H., Beier, H. & Gross, H. (1987) Electrophoresis 8,

93-99.9. Hager, D. A. & Burgess, R. R. (1980) Anal. Biochem. 109,

76-86.10. Smith, D. E. & Fisher, P. A. (1984) J. Cell Biol. 99, 20-28.11. Mort, A. J. & Bauer, W. D. (1980) Plant Physiol. 66, 158-163.12. Carlson, R. W., Saunders, R. E., Napoli, C. A. & Albersheim,

P. (1978) Plant Physiol. 62, 912-917.13. Aplin, J. D. & Hughes, R. C. (1981) Anal. Biochem. 113,

144-148.14. Hallgren, R. C. & Buchholz, C. (1972) J. Microsc. (Oxford)

166, RP3.15. vander Voort, H. T. M., Brakenhoff, G. J. & Baarslag, M. W.

(1989) J. Microsc. (Oxford) 153, 123-127.16. Bohlool, B. B. & Schmidt, E. L. (1974) Science 185, 269-271.17. Stacey, G., Paau, A. S. & Brill, W. F. (1980) Plant Physiol. 66,

609-614.18. Bal, A. K., Shantharam, S. & Ratnam, S. (1978) J. Bacteriol.

133, 1393-1400.19. Bohlool, B. B. & Schmidt, E. L. (1976) J. Bacteriol. 125,

1188-1194.20. Calvert, H. E., Lalonde, M., Bhuvaneswari, T. V. & Bauer,

W. D. (1978) Can. J. Microbiol. 24, 785-793.21. Vesper, S. J. & Bauer, W. D. (1985) Symbiosis 1, 139-162.22. Pueppke, S. G. (1984) Plant Physiol. 75, 924-928.23. Halverson, L. J. & Stacey, G. (1986) Appl. Environ. Microbiol.

51, 753-760.24. Tsien, H. C. (1982) in Nitrogen Fixation, ed. Broughton, W. J.

(Clarendon, Oxford), Vol. 2, pp. 182-198.25. Tsien, H. C. & Schmidt, E. L. (1977) Can. J. Microbiol. 23,

1274-1284.26. Heumann, W. (1968) Mol. Biol. Biochem. 5, 47-81.27. Dazzo, F., Kijne, J. W., Haahtela, K. & Korhonen, T. K.

(1986) in Microbial Lectins and Agglutinins, ed. Mirelman, D.(Wiley, New York), pp. 237-254.

28. Vesper, S. & Bauer, W. D. (1986) Appl. Environ. Microbiol. 52,134-141.

29. Ho, S.-C. (1992) Symbiosis, in press.

Cell Biology: Loh et al.

Dow

nloa

ded

by g

uest

on

July

29,

202

1