Abstract
Freeze-fracture replica labeling has become a versatile tool to visualize both membrane components and other cell structures using SDS-replica cleaning before specific immunogold labeling of proteins or lipids. We report here for the first time the adoption and optimization of the method to studies of bacterial envelopes, as applied to structural analysis of the distribution of the unique BabA-adhesin of the gastric pathogen Helicobacter pylori. BabA is important for bacterial adherence to the human epithelial cell lining of the stomach. The adhesin was found to be distributed all over the bacterial cell surfaces. Our results suggest that the SDS-replica labeling allows assessment of protein localization to distinct cell compartments and analysis of co-localization with neighboring membrane structures.
Keywords
Helicobacter pylori is a spiral-shaped, Gram-negative bacterium about 2-5 μm long, 0.5-1 μm in diameter, and with 1-6 unipolar flagella. In recent years these bacteria have been implicated, with convincing evidence, as the major cause of gastritis, duodenal ulcer, and gastric ulcer, and also appear to be associated with gastric cancer (Sipponen et al. 1992; Dixon 1997; Ota and Genta 1997).
Microbial adherence is a prerequisite for bacterial colonization and is thereby a virulence-promoting factor, particularly in changing environments such as the gastrointestinal tract.
The bacterial surface molecules responsible for the interactions with specific eukaryotic host cell molecules are adherence proteins usually called adhesins. The adhesins are either anchored in the bacterial outer membrane or are loosely surface-associated. Therefore, these proteins are of particular interest for understanding the biology and pathophysiology of host cell-microbial interactions. H. pylori is suggested to express at least five different adhesins for attachment to the gastric mucosa (Labigne and de Reuse 1996).
On eukaryotic cells, an ensemble of receptor molecules, such as glycoconjugates or proteins, mediate interaction with different adhesins (Doyle and Ofek 1994). In the intestine, a wide variety of potential epithelial receptors for bacterial adhesins have been described (Cover and Blaser 1996).
The fucosylated histo-blood group antigen Lewisb (Leb) has been reported to be a receptor for adherence of H. pylori (Borén et al. 1993). The corresponding adhesin, BabA, was recently identified, which binds to the blood group antigen Leb with high affinity (Ilver et al. 1998).
To gain deeper understanding of microbial pathogenesis, including the involvement of H. pylori in the development of acid peptic disease, it is of great relevance to be able to visualize and measure expression of adhesive traits. Therefore, it is also of interest to analyze the location, number, and detailed distribution of the BabA adhesin molecules on H. pylori. Thus far, these aspects have never been demonstrated on an en face view of the H. pylori bacterial envelope. Moreover, from a methodological point of view, there have been few publications concerning freeze-fracture of H. pylori bacteria (Noach et al. 1994; Lupetti et al. 1996; Cover et al. 1997; Lanzavecchia et al. 1998): None of these concerns the visualization of specific adhesins or outer membrane receptors.
In this study, we have undertaken a new approach to visualize bacterial adhesins. This was done utilizing the recently developed SDS-digested freeze-fracture labeling technique SDS-FRL (Fujimoto 1995; for a detailed review see Fujimoto 1997). The SDS-FRL technique has previously been applied only to studies of eukaryotic cells (e.g., Furuse et al. 1996; Dunia et al. 1998; Takizawa et al. 1998). Here we demonstrate for the first time the applicability of the method to prokaryotic cells, and more specifically to the surface analysis of the bacterium H. pylori.
Using the SDS-FRL technique together with specific polyclonal antibodies generated against the BabA adhesin, we have been able to examine the localization and distribution of the bacterial adhesin protein. Finally, we have investigated different detergents with the aim of assessing whether the technique can be further optimized for prokaryotic cells.
Materials and Methods
Bacterial Strains and Growth Conditions
The bacterial strains of H. pylori used in this study were CCUG 17874 and CCUG 17875 from the Culture Collection at the University of Göteborg. A BabA knockout mutant of strain CCUG 17875 was used (Jafar-Sondén et al. unpublished observations), a specific mutant strain in which both the BabA genes are inactivated, here denoted 17875babA1A2.
The H. pylori strains were grown on GC II agar base plates (Becton-Dickinson; Mountain View, CA), supplemented with 2.5% (w/v) Lab M agar no. 2 (Topley House; Bury, UK), 7% (v/v) horse blood citrate, 8% (v/v) heat-inactivated horse serum, 0.03% (v/v) Iso-Vitalex, 12 μg/ml vancomycin, 2 μg/ml amphotericin B, 20 μg/ml nalidixic acid, and a final concentration of HCl to 2.5 mM.
For growth of the 17875babA1A2 strain, the plates were also supplemented with 15 μg/ml chloramphenicol to allow selective growth of this specific mutant.
The bacteria were incubated under microaerophilic conditions (3% O2, 5% CO2, and 92% N2) at 37C for 2 days before use.
Freeze-fracture and Labeling of H. pylori Replicas
Bacteria were inoculated from the agar plate to the double replica device discs (Pfeiffer Vaccum Scandinavia, Kungsbacka, Sweden; BU 012 133-T). The pellets were quick-frozen in liquid propane and rapidly transferred to the freeze-fracture plant (Balzers BAF 400; Liechtenstein) and fractured at −115C. The fractured specimens were shadow-casted with Pt and C as for conventional freeze-fracture replication.
The discs with the replicas were then gently placed in PBS, pH 7.3, to separate the replica from the disc. In most cases, the replicas had to be gently touched with a spatula to detach them from the supporting disc. The replicas were transferred to 2.5% SDS (Pharmacia Biotech; Uppsala, Sweden) in 10 mM Tris-HCl, 30 mM sucrose, pH 8.3. The SDS digestion was carried out overnight at room temperature (RT).
After the treatment with SDS, the replicas were washed three times in PBS, pH 7.3. The samples were placed in wells of a porcelain dish and blocked with 1% bovine serum albumin (BSA Fraction V; Roche Diagnostics, Bromna, Sweden) in PBS, pH 7.3, for 30 min at RT. Replicas were placed in drops of affinity-purified polyclonal antibodies from rabbit directed against the BabA adhesin, diluted 1:100 in PBS with 1% BSA, and incubated for 1 hr at RT. The labeling procedure was accomplished by putting small drops (25 μl) on Parafilm in a moisture chamber created inside a Petri dish. The replicas were washed three times in PBS with 1% BSA and 0.05% Tween-20, pH 7.3.
To detect the bound primary antibody, the replicas were placed in drops of biotinylated swine anti-goat, -mouse, and -rabbit antibody (E 0453; Dakopatts, Stockholm, Sweden) diluted 1:100 in PBS with 1% BSA, and incubated for 1 hr at RT in a moisture chamber. The replicas were washed three times in PBS with 1% BSA and 0.05% Tween-20, pH 7.3, and then placed in drops of 10-nm streptavidin-conjugated colloidal gold RPN 442 (Amersham; Poole, UK) diluted 1:50 in PBS, and incubated for 1 hr at RT in a moisture chamber. Thereafter the replicas were washed three times in PBS, pH 7.3, and fixed in 0.5% glutaraldehyde in 0.1 M phosphate buffer, pH 7.3, for 10 min at RT. Finally, the replicas were washed in distilled water, mounted on gold-coated grids, and dried gently with filter paper. Control immunolabeling incubations were made, comprising omission of the primary antibody used in the labeling procedure. The internal controls in the study were the BabA-negative strains CCUG 17874 and 17875babA1A2.
Replicas were examined with a JEM-2000 EX electron microscope (Jeol; Tokyo, Japan). Negatives of electron micrographs were taken and then processed to positive prints. All micrographs were photographed at the same magnification (X 20,000). The photos were digitized by scanning (Snapscan 600; Agfa-Gevaert AB, Kista, Sweden) using the software FotoLook 32 v2.09.04 and iPhoto Express v1.0 (Ulead Systems; Torrance, CA). The pictures were sent over the Ethernet, arranged by using the Adobe Photoshop software, and printed directly from the computer on a digital color printer Mavigraph UP-D8800 (Sony; Tokyo, Japan).
We also tested the possibility of using other detergents, i.e., 10 mM Nonidet P-40 (not shown), 20 mM N-octyl-glucoside, 5 mM Triton X-114 (not shown), 20 mM Zwittergent 3–12, 20 mM Zwittergent 3–10, 10 mM Tween-80, 5 mM Triton X-100, and 10 mM CHAPS for digesting replicas for subsequent immunogold labeling. The replicas were digested in PBS solution, pH 7.3, for 1 hr at RT before immunolabeling. Detergents were purchased from Roche Diagnostics Scandinavia, except for the following which were acquired from Sigma (St Louis, MO): Triton X-114, Zwittergent 3-10, and Tween-80. All chemicals were of the highest grade available.
In parallel experiments with the different detergents, we carried out 2.5% SDS digestion of replicas as reference control and as immunolabeling controls the primary antibodies were omitted in the labeling procedure.
Measurement of Binding Activity of the BabA Adhesin
The binding assay used was carried out as previously described by Falk et al. (1994), with some modifications. The Leb glycoconjugate (IsoSep AB; Tullinge, Sweden) was labeled with 125I by the chloramine T method. Bacteria were cultured as previously described.
Bacteria were harvested from the agar plate in 1 ml PBS, supplemented with 0.05% Tween-20 and 1% BSA, pH 7.3. After centrifugation (2000 × g, 5 min) and resuspension two times in PBS with 0.05% Tween-20 and 1% BSA, the optical density was calibrated to 0.10 at a wavelength of 600 nm. One ml of the solution was added to microcentrifuge tubes along with 300 ng, ≍13,000 cpm, of 125I-labeled Leb conjugate.
The mixture was incubated at RT on a rocking shaker for 30 min. Then bacteria were pelleted by centrifugation (6000 × g, 10 min) and the amount of bound conjugates in the bacteria pellet was measured by gamma scintillation (1282 Compugamma CS; LKB Wallac, Oy, Finland). Each time, triplicate samples were used.
Results
Expression of BabA Adhesin
To ascertain the expression of the Leb antigen-binding adhesin BabA, strains used during the study were analyzed binding the soluble Leb conjugate. The experiment showed reproducibility and confirmed expression of BabA activity for the strain that was stipulated to demonstrate binding activity.
In comparison with results previously obtained on the blood group antigen binding activity for the CCUG 17875 strain, we obtained a similar level of binding (Ilver et al. 1998) (Table 1). However, a minor variation in expression of the adhesins is to be expected in response to the environmental and growth conditions, which can differ from one laboratory to another.
Surface Distribution of the BabA Adhesin
Morphological study of the BabA adhesin distribution and localization was performed using the SDS-FRL technique. As shown in Figures 1A and 1B, the SDS-fracture labeling shows that BabA is clearly localized and distributed on the outer membrane (OM), or more specifically, on the extracellular face (EF) of the outer membrane of the Gram-negative H. pylori bacteria.
Our results also showed that the adhesin can be visualized through the outer membrane by the colloid gold particles, i.e., at the extracellular face (EF) of the cytoplasmic or inner membrane (see Figure 2C).
The data presented here (Figures 1A and 1B) strongly imply that the BabA adhesins are regularly distributed on the parent 17875 strain, which expresses the molecule at its membrane. To some extent the particles are closely packed, but the majority of gold particles show a dispersed distribution. For strains lacking the BabA adhesin, immunogold labeling is completely absent (as shown in Figures 1C and 1D).
Measurement of the activity of the BabA adhesin in different strains of Helicobacter pylori a
aBinding of bacteria to soluble 125I-labeled Leb glycoconjugate was measured by gamma-counting of pelleted bacteria. Binding is expressed as percentage of the total amount of 125I conjugate added to the bacteria. Data are presented as mean percent of maximal radioactivity ± SEM and range of three separate experiments.
The 17875babA1A2 strain also served as an internal control, owing to its non-expression of the BabA adhesin. The strain showed very interesting morphological features. In the EF of the cytoplasmic membrane, several blebs were observed (arrowheads in Figure 1D).
Efficacy of Different Detergents
In a further effort to evaluate the usefulness of the SDS-FRL method on H. pylori, we tested a variety of detergents, some previously reported to be useful on eukaryotic cells, e.g., octyl-glucoside (Takizawa et al. 1998). In the test series with different detergents, there were cases with unfractured cell components, which were incompletely digested and remained attached to immunolabeled replicas, marked in the micrographs with an asterisk (∗). These remains of biological material on the replica gave the picture an unwanted blurring, complicating the observation of fine structures.
In comparing the eight tested detergents the zwitterionic detergents, i.e., Zwittergent 3-12 (Figure 2F) and Zwittergent 3-10 (Figure 2E), were found to give the most satisfying results compared with the original 2.5% SDS digestion. In contrast, the octyl-glucoside, previously reported to be efficient for eukaryotic cells (Takizawa et al. 1998; Takizawa 1999), showed nonspecific labeling and insufficient digestion of the replicas (Figure 2D).
Discussion
SDS-digested freeze-fracture labeling (SDS-FRL) combines freeze-fracture electron microscopy with colloidal gold immunocytochemistry, which has proved to be a powerful technique for visual identification of membrane topology and the spatial distribution of various components of eukaryotic biomembranes (e.g., Fujimoto et al. 1997; Dunia et al. 1998; Takizawa 1999).
The SDS-FRL methodology was first established (Fujimoto 1995) on the hypothesis that the split membrane halves were physically fixed in the Pt/C cast. There the SDS treatment dissolves unfractured cell components, leaving the stabilized portion of the membrane essentially unaffected. This opens up the possibility of labeling both the inner protoplasmic face (P-face, PF) and the outer exoplasmic face (E-face, EF).
(
Tests with different detergents in combination with freeze-fracture replica labeling of the outer membrane of the CCUG 17875 strain of H. pylori. The BabA adhesin is gold-labeled (black arrows). Binding to nonspecific components is visualized (white arrowheads). Remains of undissolved cell components marked (∗). (
Gram-negative bacteria exhibit very different external structures in comparison with eukaryotic cells, owing to their specialized outer membrane structure. The cell wall of a Gram-negative bacterium contains two layers external to the cytoplasmic membrane. Immediately external to the cytoplasmic membrane is a thin peptidoglycan layer. Exterior to the peptidoglycan layer is the outer membrane, and the space between the external surface of the cytoplasmic membrane and the outer membrane inner surface is called the periplasmic space. The outer membrane has a bilayer structure but differs from other membranes in the structure of the outer leaflet, which is primarily composed of an amphipathic molecule called lipopolysaccharide or LPS (Murray et al. 1998). Although this is the general structure for Gram-negative bacteria, there are observations implying that the two membranes in H. pylori might be in closer contact than in Gram-negative bacteria in general (Doig and Trust 1994).
The greatest benefit of the present method is that integral membrane proteins (IMPs), which previously could be seen only as morphological indistinguishable IMPs, can now be selectively displayed on the micrographs, as a result of the use of the SDS-FRL technique combined with specific antibodies (Fujimoto 1997).
In the present study, we have shown that the SDS-FRL method is fully transformable and can also be used on prokaryotic cells, particularly in Gram-negative bacteria such as H. pylori.
We demonstrate in this study an observation of new morphological features in the mutant strain of H. pylori, 17875babA1A2, i.e., blebs with associated transmembrane spicules (see inset in Figure 1D). On the basis of this observation, we suggest three hypotheses.
According to the first hypothesis, the blebs are found on a mutant strain, in which compensatory overexpression of some protein or an alternative structure may be likely. This should be manifested in these “exporting blebs” observed. This is in accordance with earlier reports of adhesion sites on E. coli serving as excretion apparatuses for membrane proteins and precursor molecules, from the cytoplasm to the outer membrane (Bayer 1968,1991; Bayer et al. 1990).
Alternatively, the blebs may be DNA-carrying vesicles, based on a corresponding feature of another Gram-negative bacterium, i.e., Neisseria gonorrhoeae, which previously has been reported to form exporting blebs (Dorward and Garon 1989; Dorward et al. 1989). Similar DNA-containing vesicles are detectable in many Gram-negative strains (Dorward and Garon 1990).
Finally, the protrusions in the EF of the cytoplasmic membrane could make up so-called transformasomes. The transformasome mediates uptake of double-stranded donor DNA, which binds sequence-specifically. This has been observed in the Gram-negative bacterium Haemophilus influenzae (Concino and Goodgal 1981; Kahn et al. 1982). The expression of this DNA uptake apparatus may be a result of the loss of gene expression in the 17875babA1A2 strain.
Definite data and observations in favor of one of these hypotheses are not yet available, and because they are beyond the scope of this report we must leave these remaining questions to further studies. Concerning the evaluation of detergents for replica cleaning, a concluding comment should be that SDS appears to be a working detergent for bacterial envelope labeling but that the zwitterionic detergents appear to yield equivalent results and efficacy.
Some immunogold labeling of nonspecific components can be seen in the micrographs presented here (white arrowheads). It is possible that the detergent manages to solublize the adhesin, which then tends to stick to the replica surface, where it becomes labeled with the antibody-gold complex. Another explanation for the phenomenon is that there are parts of bacteria from beneath that protrude small parts of their envelopes, not visible in the replica plane but still distinct enough to support gold labeling and visualization in the TEM.
To conclude, despite the major differences between eukaryotic and prokaryotic cells, our present results strongly indicate that the SDS-FRL is most useful for ultrastructural investigation of bacterial envelopes. With available specific adherence protein antibodies, we have succeeded in extracting new information about the exact localization of these membrane-associated adhesins on H. pylori, and in addition have observed a system of compensatory blebs in the adhesindeficient mutant. We believe that these findings are of importance for further understanding of the basis of host-parasite interactions.
Footnotes
Acknowledgements
Supported by the program Infection and Vaccinology, Foundation for Strategic Research, Stockholm, Sweden, by the Swedish Medical Research Council (project nos. 6251 and 11218), by the Swedish Research Council for Engineering Science, and by the King Gustaf V 80-Year Foundation.
We wish to thank Andreas Lundgren for critical linguistic reading of the manuscript.