To access the private area of this site, please log in.

Username: 

Password:  

Imaging Techniques for the Study of Escherichia coli and Salmonella Infections

ELISABETH TORSTENSSON, PETER KJÄLL, AND AGNETA RICHTER-DAHLFORS*

[SECTION EDITOR: BRETT FINLAY]

Posted May 27, 2005

Microbiology and Tumour Biology Center (MTC), Karolinska Institutet, 171 77 Stockholm, Sweden

*Corresponding author. Phone: +46-8-52487425, Fax: +46-8-342651, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

 

INTRODUCTION

Infectious diseases are among the leading causes of mortality worldwide, and numerous bacterial species are included in the vast array of causative agents. Although antibiotics have saved innumerable lives and forever changed the practice of medicine, an urgent need exists for novel treatments of bacterial infections, especially in light of the increasing occurrence of bacterial strains resistant to commonly used antibiotics. It is therefore essential that we improve our understanding of the complex cross-talk that occurs between bacteria and infected host cells and tissues to identify novel targets for drug design.

This chapter describes microscopy-based techniques that can be used to study interactions between bacteria and infected host cells, bacterial gene expression in the infected animal, and bacteria-induced cell signaling in eukaryotic cells.

As infectious model systems, urinary tract infections caused by uropathogenic Escherichia coli (UPEC) and a mouse model of typhoid fever caused by Salmonella enterica serovar Typhimurium are used. Since this chapter does not intend to review the full clinical pictures of the diseases, readers are referred to recent reviews by Kaper et al. (13) describing urinary tract infections and by Parry et al. (18) describing the murine model of typhoid fever.

 

INTERACTIONS BETWEEN BACTERIA AND INFECTED HOST CELLS

To study the interaction mechanism between bacteria and eukaryotic cells, one commonly uses cell lines, primary cells, and animal models. In this section, we provide examples of how these various model systems can be used for this purpose.

 

Urinary Tract Infections Caused by E. coli

Cell Lines.

Cell lines are among the most commonly used model systems. A big selection of cell lines originating from various organs exists, and they are accordingly used to mimic, for example, epithelial cells of the various mucosal surfaces (respiratory, gastrointestinal, and urogenital tract), cells of the immune system and the brain (American Type Culture Collection). One major advantage is that cell lines are convenient to work with since they provide a reliable source of material that can be kept in the laboratory for years. In addition, this model system is very conformed, thus resulting in repeatable results. One must remember, however, that cell lines originate from transformed cells of cancer tumors and may therefore not properly reflect the situation in healthy tissues.

Figure 1 shows an example of UPEC’s interaction with human renal epithelial cells (A498; American Type Culture Collection) using indirect immunofluorescence microscopy. An antibody specific for the O-antigen of the lipopolysaccharide (LPS) (the major constituent of gram-negative bacterial outer membrane), in combination with a secondary fluorophore-tagged (Cy3) antibody reveals bacterial localization (yellow) on the renal cells. The cytoskeleton of the renal cells is outlined using fluorescein isothiocyanate (FITC)-phalloidin (Sigma-Aldrich) and is shown in green. By using antibodies that recognize various bacterial epitopes in combination with isogenic mutants, this experiment can be varied immensely to identify those bacterial structures that are essential for the interaction between bacteria and host cells.

Primary Cells.

Primary cells can be prepared from selected host tissues, and they are most often used immediately after preparation. Since the isolation procedure often is laborious, these experiments are sometimes not as reproducible as when working with cell lines. Once the experiments are set up in a robust fashion, however, they are often very informative, and the results obtained from primary cells are considered as representing the normal situation better than results obtained from experiments using cell lines. This is because primary cells often retain essential physiological features, for example, cell polarization. While cell lines most commonly have lost this feature, primary epithelial cells isolated from renal proximal tubules grow in clusters with intact cell polarity. Figure 2a shows a cluster of primary renal proximal tubule cells outlined in green by FITC-phalloidin. It is apparent that the centrally located cell is in close contact with the neighboring cells. This feature is in sharp contrast to the cells depicted in Fig. 1, which mainly grow as single cells. When the ZO-1 antibody is used to stain the primary cells, a marked green line is seen around them (Fig. 2b). This line represents the tight junctions between the cells, and it becomes visible because the ZO-1 antibody recognizes the protein zonula occludens, one component of the multiprotein complex that constitutes the tight junctions. By analyzing this type of staining with confocal microscopy it is possible to differentiate between the apical and basolateral side of the cells. This may be a fruitful approach when analyzing cellular localization of proteins, since anti-ZO-1 antibodies can be used in combination with an antibody specific for the epitope of choice.

Primary cells also preserve their original features regarding the highly variable glycosylation pattern, and these cells may therefore be better suited for bacterial-binding studies than cell lines. Primary renal proximal tubule cells have, for example, been used to identify galactosyl-ceramide as a receptor for bacterial type 1c fimbriae (2). Two other fimbriae are associated with the pathogenesis of urinary tract infection. Type 1 fimbriae are a major virulence factor in cystitis and binds to mannosylated receptors present in the apical membrane of bladder epithelial cells. The P fimbriae are predominantly expressed by E. coli strains causing pyelonephritis, and it binds to the Galα(1-4)Gal disaccharide on renal cells (3, 15). Because of the small size of the fimbriae (nanometers), these structures are best visualized by using electron microscopy (EM), as well as modified EM techniques such as freeze-fracture EM (16). Figure 3 shows an EM image of bacteria that express fimbriae.

Animal Models.

The first point of contact between a pathogen and host cells is often at mucosal epithelial surfaces. From there, bacteria may disseminate to reach deeper tissues where infections are established. To gain understanding of the mechanisms that are responsible for such progression of disease, animal models are commonly used where the pathogenesis mimics the pathogenesis in humans as closely as possible. The animals most commonly used for infection models are rodents (mice and rats) and rabbits. Depending on the normal route of infection that causes a specific disease, animals are infected via the intranasal, oral, and intragastric routes, or in the urinary bladder, and via intraperitoneal or intravenous injections. After infection, animals are observed for a desired time (from minutes to days and weeks) before they are prepared for analysis.

Confocal laser-scanning microscopy and computerized image analysis techniques using immunostained tissue sections is a powerful tool for such in vivo analysis. Unlike conventional microscopy, this method allows infection with small infectious doses to examine early time points in the infection and to focus on a single bacterium in a large tissue area. The relative ease of detection of bacteria is facilitated by the use of thick sections (30 μm), which is 7 to 30 times the thickness used in conventional immunohistochemistry (1 to 4 μm) and 300 to 600 times that used in electron microscopy (50 to 100 nm). This approach is easily applied to the study of various species of bacteria infecting different organs in the animal model of choice.

Rats and mice are most commonly used for experimental urinary tract infections (14, 22). Figure 4 is a schematic depicting how animals are infected experimentally by the infusion of bacteria into the bladder via a catheter. Here, bacteria may colonize the bladder, thus causing cystitis. Alternatively, bacteria may ascend to the kidneys, where pyelonephritis can be established. When studying pyelonephritis, animals are commonly killed within a week after infection, when the kidneys are removed, fixed, and frozen before further analysis. Thick tissue sections are permeabilized and stained with fluorescent antibodies specific for the epitopes of choice. Epitopes on the bacterium as well as markers present in the tissue may be used. Since the confocal microscope allows the user to study individual focal planes separated with a defined size, structures deep in the tissues can be identified.

Figure 5 shows an image collected on the confocal microscope when FITC-phalloidin is used to visualize the renal cortex. Characteristic structures of the cortex are clearly visible, such as glomerulus, Bowman’s capsule, and the proximal tubules (indicated in Fig. 5). Using this technique, proximal tubules were identified as the renal site that UPEC interacts with at early stages of pyelonephritis (23). As the infection progresses, bacteria can break the epithelial barrier and enter the renal parenchyma. This process is visualized in Movie 1. Again, the renal parenchyma is outlined in green (FITC-phalloidin), and red bacteria can be seen within the lumen of the proximal tubules, as well as forming microcolonies on the tubular epithelial lining. A couple of bacteria are also caught in the process of crossing the epithelial barrier. During the process of breaking the epithelial barrier to reach the interstitium, massive numbers of immune cells are attracted to the site of infection, and this infiltration ultimately disrupts the tissue morphology. This process is obvious when comparing the normal morphology of the uninfected renal cortex (Fig. 6a) to that of a severely infected tissue in Fig. 6b. In the latter image, massive numbers of bacteria (red) are dispersed throughout the parenchyma, and tissue destruction has occurred as a consequence of immune cell infiltration.

 

Serovar Typhimurium in a Mouse Model of Typhoid Fever

Confocal microscopy as a tool to study pathogenesis in vivo was originally developed to analyze typhoid fever in a mouse model (19). Unlike previously used methods, a very small inoculum of only 100 CFU was injected intravenously and the progression of disease could be observed where serovar Typhimurium translocated to the liver, one of the organs where Salmonella proliferates during typhoid fever. Thick sections of mouse liver were prepared, and different fluorescent-labeled antibodies were used for simultaneous visualization of bacterial infiltration and changes in tissue morphology. Figure 7a shows the intact tissue morphology in normal, uninfected liver tissue, where the resident Kupffer cells (stained with anti-CD18, pink) are dispersed throughout the liver parenchyma (FITC-phalloidin, purple). Three days after bacterial inoculation, infiltration of immune cells has occurred (Fig. 7b, anti-CD18, pink), and they are preferentially clustered in foci where they colocalize with bacteria (anti-Salmonella, green-yellow). Five days after infection (Fig. 7c), large foci of infection have been established where the massive numbers of bacteria (green-yellow) and the extensive infiltration of immune cells (anti-CD18, pink) cause deranged tissue morphology with severe pathological consequences.

By using antibodies that differentiate between different types of immune cells, it is possible to gain further understanding of the progression of infection. If an anti-CD18 antibody (recognizing all immune cells) is used in combination with an antibody that specifically recognizes neutrophils (RB6-8C5), the temporal pattern of immune cell recruitment is observed. Figure 8a shows the Kupffer cells (stained with anti-CD18, red) in uninfected tissue. At 3 days after infection, a purple stain appears (Fig. 8b), indicating that the major cell type present in the foci at this time point is neutrophils. Five days after the initiation of infection, however, neutrophils have been replaced by macrophages. This is shown in Fig. 8c, where the purple has disappeared (no neutrophils present), and the red staining reveals that, besides the resident Kupffer cells, a massive infiltrate of macrophages is dispersed throughout the tissue. Taken together, the results obtained from the use of confocal microscopy of immunostained tissues reflect very well the previously reported pathology linked to typhoid fever.

 

Intracellular or Extracellular Bacteria?

Bacterial pathogens can either reside extracellularly, or they can invade cells and multiply intracellularly. Cell lines have been used to address this question in vitro; however, to analyze the situation in vivo requires more sophisticated techniques. Again, confocal microscopy has been proven as a useful tool. Figure 9a shows a liver section stained with anti-Salmonella antibodies (red). Numerous bacteria are present in the liver at day 5 after infection, and the increased number of macrophages is shown using anti-CD18 antibody (green). The picture is a projection of several images collected at defined focal planes throughout a 30-μm-thick tissue section using confocal microscopy, thus the x-y projection represents a volume of data. When image-analysis programs are used to project the boxed macrophage to visualize the side view (y-z) and the bottom view (x-z) (Fig. 9b), it is obvious that all orange bacteria colocalize to the green macrophage in all three dimensions, thus proving that the bacteria are intracellularly located.

An alternative method is used to provide the image that is presented in Fig. 9c, where volume rendering has been performed on the same stack of data as represented by the box in Fig. 9a. Here, the green represents the membrane surface structure of the macrophage, and bacteria (red/orange) are not visible until the volume is cut open, as represented by a 1.8-μm tilted section of the macrophage (Fig. 9d). Again, this demonstrates that serovar Typhimurium is intracellularly located during typhoid fever.

In addition to the liver, Salmonella also reside within the spleen during typhoid fever. A similar approach to the one described previously has successfully been used to study the location of serovar Typhimurium in infected mouse spleens, and it was concluded that the vast majority of bacteria were present within specific subsets of splenic macrophages (20).

Collectively, it can be considered proven that confocal microscopy is a valuable tool that aids in the study of pathogenesis in vivo. Not only bacterial localization in the tissue but also the accompanying changes in tissue morphology that occur during progression of disease can be addressed. Thus, the technique can be applied to solve a vast array of scientific problems. The major limitation is probably the availability of antibodies recognizing the objects of interest, but many of these are available commercially as well as from within the research community.

 

BACTERIAL GENE EXPRESSION IN THE INFECTED ANIMAL

During their lifetime, bacteria are exposed to a multitude of microenvironments outside and inside a host. Within the host, bacteria can be located in various organs where they are exposed to different cell types, ranging from epithelial cells at the mucosal linings to phagocytic white blood cells. In each site, bacteria are exposed to specific sets of innate immune defense mechanisms, and to survive these threats, bacteria must rapidly adapt their gene expression profile to maximize their chance of survival in any situation.

 

Methods for Measurement of Gene Expression In Vivo

Previously, it has been difficult to perform experiments to investigate which bacterial genes are expressed when bacteria multiply in a host. However, novel techniques have evolved that aid in this problem. Laser capture microdissection (LCM) is a technique that allows isolation of selected cells and/or bacteria from tissue slices, and the expression level of the gene of interest can be analyzed by performing quantitative real-time RT-PCR on extracted mRNA (12). An alternative is to perform microarray analysis on mRNA from bacteria grown in cells or tissues. Microarrays provide information about the total transcriptional profile in bacteria at a given time point of infection, and a typical expression DNA-microarray is shown in Fig. 10. A technique was recently developed to successfully identify the full transcriptome of serovar Typhimurium growing intracellularly in J774-A.1 macrophage-like cells (10). Finally, in situ hybridization can be used, where selected mRNA can be detected in cells and tissues based on a probe-target interaction. Combined with immunohistochemistry, this semiquantitative assessment method can be a powerful tool in measuring gene expression (6).

 

GFP as a Marker To Detect Bacterial Gene Expression in Urinary Tract Infections

Recent developments concerning fluorescent proteins have resulted in possibilities to use these as reporters to monitor gene expression in individual bacterial cells (21). Usually gfp (green fluorescent protein) constructs are cloned into multicopy plasmids that are transformed into bacteria and used in infection studies. This approach poses several problems, such as varied plasmid copy numbers in individual bacteria and the requirement of antibiotic treatment of the animal to ensure that bacteria maintain the plasmids. Finally, multicopy plasmids produce high concentrations of GFP, which are potentially toxic for the bacteria.

To overcome these obstacles, one can use a modified gfp gene that produces a brighter and, at the same time less toxic, protein. This enables detection of gene expression even if the gfp-promoter fusion is present only as a single copy on the bacterial chromosome (11). Thus, by fusing the promoter of the gene of interest (Px) to the gene encoding GFP, and using a one-step allelic replacement procedure based on phage λ Red recombinase (7), the Px-gfp construct can be cloned as a single copy into the bacterial chromosome (Fig. 11). One example of successful use of this methodology in vitro is shown in Fig. 12. Fluorescent microscopy shows that UPEC harboring Px-gfp as a single copy on the chromosome are readily detected (green, Px-gfp, upper panel), while isogenic bacteria which contain the gfp gene without promoter are not (gfp, lower panel). This result is not simply due to the presence and absence of bacteria, since an antibody recognizing LPS on UPEC (red, orange) shows that bacteria are present in both samples.

The same bacterial constructs can be used to identify bacterial gene expression in vivo using the rat model of pyelonephritis. Using confocal microscopy of thick tissue slices from UPEC-infected kidneys, bacteria can be identified that express GFP under the control of a bacterial promoter (Fig. 13a). However, the GFP-expressing bacteria only constitute a fraction of the total number of bacteria present in the tissue, as shown when an antibody recognizing LPS is used on the same tissue section (Fig. 13b, red). This problem can be caused by the low signal-to-noise ratio of the fluorescence intensity in the infected tissue or by the fact that not all bacteria have the Px promoter turned on, but there may be other explanations as well.

 

BACTERIUM-INDUCED CELL SIGNALING IN EUKARYOTIC CELLS

Just as bacteria sense the environment and regulate their gene expression accordingly, eukaryotic cells react when exposed to bacteria. Studies in the recent years have revealed that eukaryotic cells express Toll-like receptors (TLRs) that recognize microbial structural motifs such as LPS on gram-negative bacteria. Signaling via TLRs is rapidly transmitted to alert the immune system (17). However, TLRs are organ- and cell-specifically expressed, but cells/tissues that do not express these receptors are still capable of eliciting eukaryotic signaling via alternative pathways (1). Such interactions may lead to a rapid onset of innate immune responses and to other host defense mechanisms. Also, cells may respond by undergoing apoptosis or die from osmotic lysis. The cellular output depends on which signaling pathways are activated by bacterial factors.

 

Signaling Events during Pyelonephritis

Alpha-hemolysin (HlyA) is a toxin encoded by a majority of pyelonephritogenic E. coli isolates. It belongs to the family of RTX toxins (repeats in toxin), whose members are known to contribute to the pathogenesis of different diseases by causing dysfunction of a variety of general cellular reactions (24). The major role of HlyA as a virulence factor has for a long time been ascribed to its pore-forming ability, causing osmotic lysis of cells. This is visualized in Fig. 14a, where HlyA-producing UPEC cause cells to round up, detach, and eventually die. In contrast, none of these morphological changes occur in cells exposed to an isogenic bacterial mutant unable to express HlyA (Fig. 14b).

When a fluorescent imaging system was used to study whether exposure to HlyA caused any alterations of the Ca²⁺ homeostasis in primary renal epithelial cells, it was found that HlyA has a dual, concentration-dependent effect on target cells. In addition to the cytolytic effect seen when cells are exposed to high concentrations of the toxin, it was found that sublytic doses of HlyA elicited a slow oscillation of the intracellular Ca²⁺ concentration in renal cells (23). This effect was revealed because a system was used that allowed recording of the intracellular Ca²⁺ concentration at a single-cell level. Cells grown on glass coverslips were loaded with the Ca²⁺-sensitive fluorescent dye FURA-2AM and then mounted in a microscope equipped with a cooled charge-coupled device (CCD) camera. The basal level of Ca²⁺ (0.1 μM) in a cluster of approximately 15 cells is shown as dark blue in Fig. 15. Once a low concentration of HlyA is added to the cells, the fluorescence of FURA-2AM changes, thus reflecting the alteration of the intracellular Ca²⁺ concentration. The color changes in Fig. 15 visualize how the intracellular Ca²⁺ concentration fluctuates in individual cells over time during a 60-min recording. An alternative way of presenting similar data is shown in Movie 2, where the color of the FURA-2AM-loaded cell changes immediately when HlyA is added to the cell. A single-cell tracing is shown in Fig. 16, which represents the Ca²⁺ concentration in the cell in Movie 2.

Oscillation of the intracellular Ca²⁺ concentration is commonly used by cells to avoid the lethal effects of sustained increases of Ca²⁺. Also, the temporal feature of the oscillation provides specificity to the signal (8, 9). It is well known that oscillations with a periodicity in the microsecond range direct exocytosis of, for example, neurotransmitters at nerve synapses, while slower periodicities in the minute range control gene transcription (4, 5). The HlyA-induced Ca²⁺ oscillations were found to occur with a periodicity of 12 min. Via frequency-modulated activation of the transcription factor NF-κB, specific up-regulation of the proinflammatory cytokine interleukin-6 (IL-6) and chemokine IL-8 was detected (23). This is the first pathophysiologically relevant protein that has been shown to direct eukaryotic gene expression via induction of Ca²⁺ oscillations. At present, some components of the signal-transduction pathway have been identified, but work is in progress to identify the cellular components involved.

Measurement of Ca²⁺ fluxes provides a very sensitive method to measure cytolytic events as well. When cells are exposed to an increased concentration of HlyA, a rapid influx of Ca²⁺ occurs, leading to a sustained increase in Ca²⁺ concentration and eventually cell lysis. This event is shown in Movie 3, and a corresponding single-cell tracing is shown in Fig. 17.

 

Serovar Typhimurium Causes Apoptosis of Macrophages in the Infected Liver

Induction of apoptosis is a relatively common host response to bacterial infections, and it is possible to probe this physiological event fluorescently by using commercially available kits. Use of confocal microscopy of 30-μm liver sections stained with such a kit shows that apoptotic cells occur only occasionally in the uninfected liver (Fig. 18a, green). However, mice with typhoid fever show an increased number of bacteria (pink) in the liver at day 3 (Fig. 18b) and day 5 (Fig. 18c) after infection, and this is accompanied by infiltration of immune cells (blue). The green staining in the two latter images indicates that progression of typhoid fever is accompanied with an increased number of apoptotic macrophages in the liver (19).

 

CONCLUSION

The rapid development of fluorescent reporter proteins and advances in microscopy-based techniques have provided new and promising approaches not only to locate bacteria in tissues, but also to analyze expression of virulence factors in individual bacteria and host cells during the progression of disease. These techniques enable, for the first time, studies of the complex microenvironments within infected organs and will reveal the alterations of bacterial physiology that occur during bacterial growth within a host. Knowledge in the fields of cell biology and microbial pathogenesis will indeed rapidly advance as a result of technical improvements, which bring promise of identification of, for example, novel drug targets that can be used to develop a new generation of antimicrobial agents.

 

Acknowledgments

The work performed in the Richter-Dahlfors laboratory is supported by grants from the Royal Swedish Academy of Sciences (KVA) and the Swedish Foundation for Strategic Research (SSF).

Figure 3 was kindly provided by A. von Euler, and Fig. 10 was provided by S. Eriksson and M. Rhen.