HARUO TAKAMIYA,'* STEPHEN BATSFORD,' HANS GELDERBLOM,2 AND ARNOLD VOGT'. Abteilung fur ... The 0-antigens of gram-negative bacteria are.
JOURNAL OF BACTERIOLOGY, Oct. 1979, p. 261-266 0021-9193/79/ 10-0261/06$02.00/0
Vol. 140, No. 1
Immuno-Electron Microscopic Localization of Lipopolysaccharide Antigens on Ultrathin Sections of Salmonella typhimurium HARUO TAKAMIYA,'* STEPHEN BATSFORD,' HANS GELDERBLOM,2 AND ARNOLD VOGT' Abteilung fur Immunologie im Zentrum fur Hygiene der Universitat Freiburg im Breisgan,1 and Robert Koch-Institut des Bundesgesundheitsamtes, Berlin,2 Germany Received for publication 27 July 1979
Lipopolysaccharide antigens were demonstrated on ultrathin sections of styrene-embedded Salmonella typhimurium by direct postembedding staining with ferritin-labeled antibodies. The antigenicity, partially masked in the embedding process, could be satisfactorily recovered by treatment of ultrathin sections with nonspecific protease. As judged from the reaction site of the ferritin-labeled antibodies, the lipopolysaccharides were localized in two zones. The broader zone of densely distributed ferritin molecules was superimposed over the whole outer cell wall, and a smaller zone revealing antigenicity was found over the cell membrane, which strongly supports the concept that the latter is the site of synthesis of lipopolysaccharides. The well-defined labeled areas between these two antigenic zones may be the routes whereby the synthesized polysaccharide molecules reach the cell wall. The 0-antigens of gram-negative bacteria are located within the outer layer of the cell wall. From biochemical, immunological, and electron microscopic studies, including work on cell wall fractions and suitable mutants, it could be deduced that the lipopolysaccharide antigens are synthesized at the cytoplasmic membrane (1, 6, 9). By using intact cells, direct visualization of the antigens by immuno-electron microscopy was till now not possible because antigens are usually not accessible for antibody on ultrathin sections after plastic embedding (11). We have recently developed a postembedding staining method by using styrene as embedding material and protease treatment for subsequent reexposure of the antigens which enables us to localize antigens on the electron microscopic level. The results obtained with postembedding staining for the 0-4,5 antigens on sections of Salmonella typhimurium confirm and extend the results reported with the techniques mentioned above. MATERIALS AND METHODS S. typhimurium isolated from human stool specimens were cultured in glucose broth for 16 h. The bacteria were pelleted by centrifugation at 2,500 x g for 15 min and resuspended in 0.05 M cacodylate buffer, pH 7.2. Glutaraldehyde diluted in 0.05 M cacodylate buffer (pH 7.2) was added to the chilled bacteria suspension to give a final concentration of 2.0% glutaraldehyde and kept for 1.5 h at 0°C (ice bath). The fixed bacteria were washed once, resuspended in cacodylate buffer, and divided for pre- and postembed261
ding staining. For the former, the bacteria were immersed in ferritin-labeled anti-0-4,5 conjugate, washed, and block embedded in Epon after agar enclosure (4). For postembedding immune labeling, the bacteria suspension was brought to 45°C in a water bath, and an equal volume of 2% agar at 45°C was added, thoroughly mixed for 20 s, and transferred to an ice bath for solidification. Small pieces were cut, dehydrated with acetone, and embedded in styrene after the method of Kushida (3). Sections were cut on an LKB Ultrotome and treated with protease type V (Sigma Chemical Co., catalogue no. P 5005) at 0.01 mg/ml in phosphate-buffered saline, pH 7.4, for 1 h in a moist chamber at 37°C. After washing three times with phosphate-buffered saline, the treated sections were incubated with 10% fetal calf serum in 0.1 M phosphate buffer, pH 7.6, for 15 min and then transferred to the same buffer solution which contained 2.0 mg of apoferritin as well as 1.0 mg of normal rabbit immunoglobulin G per ml for 30 min. The sections were stained for 60 min with ferritin-labeled rabbit anti-0-4,5 antiserum (immunoglobulin G fraction) applying the isolated, small-size 1:1 conjugate, which consisted of a single ferritin molecule linked to one antibody globulin molecule. Isolation of the small-size conjugate was performed by Pevikon electrophoresis followed by gel filtration on Sepharose 6B as described elsewhere (S. R. Batsford, H. Takamiya, and A. Vogt, J. Immunol. Methods, in press). Ferritin labeling was performed as described previously (7). The conjugates were adjusted to a ferritin concentration of 1.0 mg/ml. Controls were done with conjugates consisting of the immunoglobulin G fraction of normal rabbit serum conjugated with ferritin. The sections were mounted finally on Formvar-coated grids and viewed with a Siemens Elmiskop IA at 80 kV after poststaining with
262
TAKAMIYA ET AL.
osmium tetroxide vapor for 2 min, methanolic uranylacetate for 5 min, and lead citrate for 30 s consecutively.
RESULTS For each technique several sections with 100 to 200 bacteria were examined. Preembedding staining of Salmonella for lipopolysaccharide antigens resulted in the ferritin labeling of the external surface of the bacteria. The ferritin conjugate was located close to the outer cell wall, mostly randomly distributed. In some places, the ferritin molecules were lying in a monolayer like a string of pearls parallel to the visible electron-dense outermost surface; at other places, the ferritin molecules were scattered in small piles two or three layers thick (Fig. 1). No ferritin molecules were seen within the confines of the cell wall. With the postembedding staining technique, ferritin molecules were found to be distributed over the whole width of the cell wall (Fig. 2). At some places small external protrusions were seen (Fig. 2 to 5). A second labeled zone was observed over extensive parts of the cytoplasmic membrane (Fig. 2 to 5). The labeling of the latter was less constant and usually weaker than the staining
J. BACTERAIOL.
of the cell wall. Both labeled zones, which due to the retraction of the cytoplasmic membrane were lying wide apart, were connected by zones of ferritin labeling. The frequency of these zones and their widths varied (Fig. 2 to 5). Such zones were usually found only at places where the cytoplasmic membrane was labeled too. The cytoplasm of the cells was mostly unstained. Only a few ferritin molecules were found inside the cytoplasmic membrane. As judged from the background of the control sections (Fig. 6), these ferritin molecules were probably nonspecifically bound. A relatively high background was often seen in the areas surrounding the bacterial cells; with increasing distance from the bacteria, the background staining decreased. Since in the control sections treated with ferritin conjugated with normal rabbit immunoglobulin G virtually no background at all was observed, the ferritin conjugate molecules found in the vicinity of the bacteria probably have reacted with antigenic material released from the cell wall during the fixation and embedding processes. The imperfect cytoplasmic fine structure is due to poor fixation which is needed for antigenic preservation. Bacteria with cytoplasmic retraction were
FIG. 1. Preembedding staining of S. typhimurium with ferritin-labeled antibodv to 0-4,5 antigen. The staining is confined to the external cell surface. The reacted ferritin molecules are randomlv distributed. At some places, up to three layers of ferritin can be seen forming antigenic protrusions (arrowheads). 105,000x. In this and all other figures, bars represent 100 nm.
VOL. 140, 1979
POSTEMBEDDING STAINING OF LII'O01OLYSACCHARIDE
selected for clearer visualization of the ferritin labeling. DISCUSSION Preembedding staining of bacteria with ferritin-labeled antibodies is restricted to antigens located on the outer surface, because the labeled antibody is too large for penetration on the cell wall. The results presented here demonstrate that postembedding staining of Salmonella is possible when styrene is employed. The thus-embedded polysaccharide antigen contrasted with proteins, i.e., was accessible to specific antibodies (manuscript submitted). Subsequent treatment with protease led to further exposure of the antigenicity. The mechanism of action is not clear (10). With this technique we achieved a specific localization of intracellular lipopolysaccharide antigens at the electron microscopic level. The findings thus obtained confirm and substantially extend the results reported previously with preembedding staining. After postembedding staining of S. typhimurium with ferritin-conjugated anti-0-4,5 antibody two distinct labeled zones were always found. The outer zone of densely distributed ferritin molecules corresponded exactly to the cell wall. In addition, some threadlike protrusions of antigenic material from the outer surface were seen. Since we used exclusively small-size ferritin immunoconjugates for the staining, it is unlikely that the extended protrusions are artefacts. That the protrusions in the postembedding technique are more pronounced may be due to the different handling. It could be that in the preembedding process ferritin conjugate reacted with antigenic filaments extending from the outer surface that are torn off during washing due to the mass of the ferritin. Protrusions have already been described by Shands (8). This author believed such structures to be fibrils of somatic antigens ex-
263
tending out from the confines of the cell wall. Apart from these protrusions, the polysaccharide antigens seem to be distributed rather homogeneously throughout the width of the cell wall, because in contrast to the results obtained from preembedding staining, antigen was not restricted to the outer layer of the cell wall. Another new feature revealed by postembedding staining was a second zone of more or less densely labeled ferritin molecules superimposed on the cytoplasmic membrane. This observation is a further strong indication that polysaccharide antigen is synthesized at the cell membrane, a finding which had already been deduced from biochemical and immuno-electron microscopic studies on subcellular fractions (1, 6). Figures 2 through 5 show that the intensity of staining for polysaccharide antigen over the cytoplasmic membrane varies, possibly indicating different stages of activity in the synthesis of the polysaccharide antigens. Close examination of the distribution of the labeled antibody between the cell wall and the cell membrane in bacteria where the latter had been retracted from the cell wall indicated that the lipopolysaccharide molecules within this region are not uniformly distributed in the space but tend to occur in clumps forming bridges. A possible interpretation is that newly synthesised polysaccharide antigens reach the cell wall by travelling along preformed pathways and that a random movement of antigenic material across the gap is therefore unlikely. The idea of distinct pathways for antigen transport was first put forward by Muhlradt et al. (5), who found, working with plasmolyzed cells, that newly synthesised lipopolysaccharide is found initially at the sites where the cytoplasmic and outer membrane adhere to one another. Additional biochemical evidence for this concept was presented by Kulpa and Leive (2), who worked with Escherichia coli.
Continued on following page
FIG. 2-5. Postembedding staining of S. typhimurium with ferritin-labeled antibody to 0-4,5 antigen. Cells were selected where the cytoplasmic membrane had been retracted fromn the cell wall. Staining of the entire cell wall is seen; protrasions are also present (arrowheads). Itn addition the cytoplasmic mem branes are stained to vlarying extents (Fig. 2, 3, and 5). At setveral positions, antigenic material appears to be bridging the gap between the cytoplasmic membrane and the cell wall (an example i's arrowed). Figur-e 4 is an enlargement of a portion of Fig. 3. Note that the bridging antigens occur at the sites uhere the cytoplasmic membrane is also labeled. The backgr-ound staining in the vicinity of the cell surface may be caused by antigenic material released during the fixation and embedding process. Fig. 2, 141,000x; Fig. 3, 150,000x; Fig. 4, 405,000x; Fig. 5, 156,OOOx. 26.4
VOL. 140, 1979
POSTEMBEDDING STAINING OF LIPOPOLYSACCHARIDE
265
266
'ITAKAMIYA El AL.
*J. B A( I E IOL.
FIG. 6. Postembedding stalining of S. typhimurium with ferritin conjugated with normal rabbit linmunoglobulin G as a control. Only isolated non.specificallv reacted ferritin molecules can be seeni (arrous). 120,000x. ACKNOWLEDGMENT This work
was
supportetl
by
the Deutsche Forschungsge-
meinschaft. LITERATURE CITED
Hama, S., T. Katsumoto, and A. Takagi. 1973. Investigations on the subcellular localization of 0-antigens in Salmonella enteritidis by nmeans of ferritin-conjugated antibody technique. Yonago Acta Med. 17:20(7-215. 2. Kulpa, L. F., and L. Leive. 1976. Mode of insertion of lipopolysaccharide into the outer nmemibrane of Escherichia coli. J. Bacteriol. 126:467-477. 3. Kushida, H. 1961. A styrene-methacrylate resin enmbedding method for ultrathin staining. J. Electronmicrosc. 20:16-19. 4. Luft, J. H. 1961. Improvements in epoxy resin embedding miethods. J. Biophys. Biochem. Cytol. 9:409-414. 5. Muhlradt, P. F., F. Menzel, J. R. Golecki, and V. Speth. 1973. Otuter menmbrane of Salmonella. Sites of export of newly synthesised lipopolvsaccharide on the bacterial surface. Eur. J. Biochem. 35:471-481.
6. Osborn, M. J., J. E. Gander, and E. Parisi. 1972. Mechanisnm of assembly of the outer membrane of Sal-
monella typhimurium. ,J. Biol. Chem. 247:3973-3986. 7. Otto, H., H. Takamiya, and A. Vogt. 19783. A two-stage method for cross-linking antibody globulin to ferritin by glutaraldehyde. Comparison between the one-stage and the two-stage method. ,J. InMLmunol. Methods 3:137-146. 8. Shands, J. W. 1965. Localization of somatic antigen on gram-negative bacteria by electron microscopy. J. Bacteriol. 90:266-270 9. Shands, J. W. 1966. Localization of somatic antigen on gram-negative bacteria using ferritin antibody conjugate. Ann. N. Y. Acad. Sci. 133:292-298. 10. Takamiya, H., W. Bodemer, and A. Vogt. 197-8. Masking of protein antigen by nmodification of amino groups with carbobenzoxychloride (benzyl chloroformate) and demasking by treatment with nonspecific protease. 1J. Histochem. Cytochem. 26:914-920. 11. Vogt, A., H. Takamiya, and W. A. Kim. 1976. Some problems involved in post-embediding staining, p. 1()9115. Itn G. Feldman, P. D)ruet, ,J. Bignon, and S. Avrameas (ed.), Immunoenzvnmatic techniques. North-Hollan(d l'ublishing Co., Amrsterdamii.
