MicroMeeting: Who's talking to whom? Epithelial–bacterial pathogen interactions


  • Phillip D. Aldridge,

    Corresponding author
    1. Institute for Cell and Molecular Biosciences, University of Newcastle upon Tyne, The Medical School, Newcastle upon Tyne, NE2 4HH, UK.
      E-mail p.d.aldridge@ncl.ac.uk; Tel. (+44) 191 222 7704; Fax (+44) 191 222 7424.
    Search for more papers by this author
  • Michael A. Gray,

    1. Institute for Cell and Molecular Biosciences, University of Newcastle upon Tyne, The Medical School, Newcastle upon Tyne, NE2 4HH, UK.
    Search for more papers by this author
  • Barry H. Hirst,

    1. Institute for Cell and Molecular Biosciences, University of Newcastle upon Tyne, The Medical School, Newcastle upon Tyne, NE2 4HH, UK.
    Search for more papers by this author
  • C. M. Anjam Khan

    1. Institute for Cell and Molecular Biosciences, University of Newcastle upon Tyne, The Medical School, Newcastle upon Tyne, NE2 4HH, UK.
    Search for more papers by this author

E-mail p.d.aldridge@ncl.ac.uk; Tel. (+44) 191 222 7704; Fax (+44) 191 222 7424.


Our perception that host–bacterial interactions lead to disease comes from rare, unsuccessful interactions resulting in the development of detectable symptoms. In contrast, the majority of host–bacterial interactions go unnoticed as the host and bacteria perceive each other to be no threat. In July 2004, a focused international symposium on epithelial–bacterial pathogen interactions was held in Newcastle upon Tyne (UK). The symposium concentrated on recent advances in our understanding of bacterial interactions at respiratory and gastrointestinal mucosal epithelial layers. For the host these epithelial tissues represent a first line of defence against invading bacterial pathogens. Through the discovery that the innate immune system plays a pivotal role during host–bacterial interactions, it has become clear that epithelia are being utilized by the host to monitor or communicate with both pathogenic and commensal bacteria. Interest in understanding the bacterial perspective of these interactions has lead researchers to realize that the bacteria utilize the same factors associated with disease to establish successful long-term interactions. Here we discuss several common themes and concepts that emerged from recent studies that have allowed physiologists and microbiologists to interact at a common interface similar to their counterparts – epithelia and bacterial pathogens. These studies highlight the need for further multidisciplinary studies into how the host differentiates between pathogenic and commensal bacteria.


Host–bacterial interactions are often wrongly assumed to be detrimental to both host and bacteria as a result of ‘dead-end’ interactions leading to the development of visible disease symptoms. Research into why these interactions have been labelled ‘wrong’, has led researchers to realize that the majority of host–bacterial interactions go unnoticed and that the development of disease symptoms is actually a rare event (Merrell and Falkow, 2004). Irrespective of what happens, the immense selective pressure at the host–bacterial interface means that both the host and bacteria have evolved numerous ways in which to communicate their wishes/intentions with each other. This communication is essential for the survival of the host and bacterium, even if it is just to tell one side ‘sorry wrong number’. The evident complexity of host–bacterial interactions has generally resulted in only one side of the interaction being studied or discussed separately, focusing on how the bacteria react to host defence systems or how the host comes to terms with the presence of colonizing bacterial pathogens.

On 22–23 July 2004 an international symposium on epithelial–bacterial pathogen interactions took place in Newcastle upon Tyne (UK) supported by The Physiological Society (http://www.ncl.ac.uk/camb/research/conference/EBPI.htm). This symposium focused on recent advances in the understanding of the extent to which bacteria and epithelial cells communicate with each other during their interactions at a molecular, cellular and whole tissue level and the outcome of these communications. The aim of this symposium was to provide an opportunity for both cell physiologists and microbiologists to present their views of how epithelia and bacteria react once contact has been made. The abstracts of the communications presented at the meeting are published in the Journal of Physiology (Vol. 559P, 2004: http://www.physoc.org/publications/proceedings/archive/).

In animal hosts, the epithelial layer of mucosal surfaces in the lung and gastrointestinal tract represents a primary contact environment for bacteria. Rather than just representing a physical barrier to prevent bacteria gaining unwanted access to essential organs, the epithelial layer also provides a surface where the host can interact with bacteria. Both the host and bacteria also use it as a neutral location where decisions can be made without significant damage to either side. Recent advances have shown that epithelia are of great importance to the innate immune system and these cells are constantly screening the bacterial population present for unwanted pathogens, as well as screening the commensal bacterial population. Interestingly, gastrointestinal epithelium can tolerate a significant degree of bacterial colonization, whereas the lung epithelia do not even tolerate their presence. How commensal and pathogenic bacteria are differentiated by gastrointestinal epithelia is a major question in this field.

In response to contact with host cells, bacterial pathogens have evolved numerous methods to modulate the physiological function of epithelia to their advantage to prevent clearance or detection by the host defence mechanisms. They do this by producing a wide range of virulence factors including exotoxins, extracellular enzymes and the direct translocation of proteins into the host cells (Merrell and Falkow, 2004). There is no universal pathway that all bacteria utilize to achieve this goal; another reflection of the selective pressure exerted upon them to survive in host environments. As well as the production of virulence factors, pathogenic bacteria evade host defence mechanisms by either invading host epithelial cells or inhibiting active uptake by macrophages (Sansonetti, 2002). In this meeting report it is our intention to review some of the major aspects presented at the symposium. Rather than a comprehensive review of the whole meeting, we have chosen to focus on aspects where common themes and concepts emerged. We cannot of course go into significant detail on everything and the reader is pointed towards more specific reviews on some of the topics covered (Basset et al., 2003; Merrell and Falkow, 2004; Ratjen and Döring, 2003; Sansonetti, 2002).

The lung – a story of opportunists

The lungs are subject to all sorts of foreign bodies including dust, airborne microorganisms and other pollutants. In normal healthy individuals, a very effective and complex innate defence system keeps the lungs clean and sterile. The presence of foreign bodies that are not readily degradable and acquired or inborn defects of this defence system can lead to lung disease. A good example of the latter is cystic fibrosis (CF), which affects 1 in 2500 newborn individuals. CF is caused by mutations in a plasma membrane chloride channel known as CFTR (Ratjen and Döring, 2003). In airways, lack of CFTR function leads to a viscous mucus layer on the epithelium that impairs mucociliary activity. This results in CF patients being susceptible to chronic bacterial infection by opportunistic pathogens such as Pseudomonas aeruginosa, Staphylococcus aureus or Burkholderia spp. (Döring and Worlitzsch, 2000). Using lung biopsies and primary respiratory epithelial cells from CF patients, Gerd Döring (University of Tübungen, Germany) presented evidence that S. aureus and P. aeruginosa are located on the respiratory epithelium of CF patients in the mucus layer rather than at the epithelial membrane (Ulrich et al., 1998) (Fig. 1). These findings question previous hypotheses that explain the presence of chronic lung infections in CF patients with a structural modified respiratory epithelial membrane to which the opportunistic pathogens adhere directly. Their data also suggest that the lack of mucus clearing is the reason why bacteria can infect CF lungs.

Figure 1.

Pseudomonas aeruginosa (A) and Staphylococcus aureus (B) on ciliated respiratory epithelium of CF patients. False coloured scanning electron microscopy pictures showing the mucus layer and either P. aeruginosa cells (rods) in A or S. aureus cocci coloured yellow in B (from M. Ulrich, G. Döring and J. Berger, unpubl.).

A major milestone in the understanding of chronic bacterial lung infections in CF patients was the finding of Döring and coworkers that microaerobic/anaerobic environmental conditions in the mucus layer on the respiratory epithelium of CF patients induce a phenotypic switch from non-mucoid to mucoid, alginate producing, P. aeurginosa, leading to macrocolony growth resembling a biofilm (Worlitzsch et al., 2002). Such macrocolonies are notoriously resistant to elimination by phagocytic cells or antibiotic therapy. Döring also presented data showing that single P. aeruginosa cells, presumably derived from endobronchial macrocolonies, reach the alveoli where higher amounts of rigid collagen is found rather than elastin as observed in healthy tissue. These findings may not be that surprising as most previous studies on CF infections have concentrated on the proximal airway (Ratjen and Döring, 2003). This work has implications for our understanding of how P. aeruginosa establishes a chronic infection, as CF patients would have great difficulty in clearing bacteria residing in terminal alveoli.

Although P. aeruginosa may not directly adhere to epithelial membranes according to the findings of Döring and coworkers, the pathogen is still able to interact with or affect the physiological function of the respiratory epithelial layer by producing extracellular virulence factors such as pyocyanin. Although the major role of pyocyanin is thought to be as an antibiotic against competing bacteria for sites of colonization, it is also known to exhibit toxic effects upon epithelial cells (O’Malley et al., 2003). Brad Britigan (University of Cincinnati College of Medicine, USA) presented recently published data that defines the action of pyocyanin on epithelial cells (O’Malley et al., 2003; Reszka et al., 2004). The cytotoxicity of pyocyanin to bacteria or epithelial cells is known to be via its ability to redox cycle. It does this by accepting electrons, usually from NADH or NADPH, and passing them onto to O2, resulting in the production of superoxide and hydrogen peroxide. The result of this redox cycle is that cells exposed to pyocyanin are under significant oxidative stress. Cell functions may be directly impacted by the oxidants or indirectly affected via an influence of redox sensitive signalling cascades. Microscopic evaluation of pyocyanin treated respiratory epithelia shows that the main site of action is in, or around, the mitochondria (O’Malley et al., 2003) (Fig. 2). The same study also showed that epithelial cells are able to metabolize pyocyanin, leading to the conclusion that one defence mechanism against the action of pyocyanin is simply to catabolize it. More recent unpublished data also suggest that pyocyanin can still penetrate through intact epithelial cell monolayers avoiding degradation and impact the cells located below. A possible metabolic pathway for pyocyanin has recently been published by Britigan and coworkers. They have shown that peroxidases are able to irreversibly inactivate pyocyanin in vitro (Reszka et al., 2004). This has led them to propose that naturally occurring peroxidases, such as lactoperoxidase, in the lung could also do the same (Reszka et al., 2004).

Figure 2.

Effect of pyocyanin on mitochondrial ultrastructure. A549 monolayers were incubated with buffer (A), 50 µM pyocyanin (B) or 100 µM pyocyanin (C) for 2 h, following which the cells were fixed and their mitochondria were examined by transmission electron microscopy (×100 000). At the highest concentration of pyocyanin, a decrease in mitochondrial matrix structure was observed. Reproduced from O’Malley et al. (2003) with kind permission of the American Physiological Society.

Quorum sensing globally regulates the majority of virulence factors produced by P. aeruginosa (Pesci and Iglewski, 1997). As well as regulating virulence factor production, quorum sensing plays a significant role in the life cycle of P. aeruginosa, whether it is infecting a CF lung or is simply in the environment. Well-established in vitro studies have shown that quorum sensing plays an important role in the ability of bacteria to form a biofilm efficiently (Pesci and Iglewski, 1997). P. aeruginosa is not only an opportunistic pathogen of CF patients but also causes infections of the cornea and burn wounds (Rumbaugh et al., 1999). Comparable in vivo studies using both lung and burn wound models of P. aeruginosa infections have shown that biofilm formation and quorum sensing are essential for the establishment of infection (Rumbaugh et al., 1999). All of these studies have concentrated on how P. aeruginosa reacts to quorum sensing signals during pathogenesis. Two major quorum sensing molecules produced by P. aeruginosa are responsible for the regulation of virulence factors: 3OC12-HSL (also known as OdDHL) and C4-HSL (also known as BHL) (Chun et al., 2004). Pete Greenberg (University of Iowa, USA) presented data showing that respiratory epithelia also take note of the global communication occurring amongst the bacterial population present in the lung. Using cultures of primary respiratory cells, Greenberg and coworkers observed that C4-HSL diffuses across the epithelial cell layer. The second molecule, 3OC12-HSL, mysteriously disappeared from their reactions. Further research has shown that 3OC12-HSL is, in fact, metabolized by the epithelial cells. Greenberg and coworkers have recently published data showing that the ability to metabolize 3OC12-HSL varies between cell type, suggesting that only host cells that come in to regular contact with 3OC12-HSL have evolved the ability to metabolize it (Chun et al., 2004). This finding of Greenberg and coworkers has significant implications for models of P. aeruginosa pathogenesis. In vitro studies of biofilm formation have shown that P. aeruginosa strains deficient in producing 3OC12-HSL form only thin biofilms that are susceptible to detergent, whereas C4-HSL deficient strains could still form detergent resistant biofilms (Davies et al., 1998). It has been proposed that detergent susceptibility may reflect biofilm susceptibility to antimicrobials or the innate immune defence system (Rumbaugh et al., 2000). Therefore, one reason why P. aeruginosa colonization is observed (by Döring) preferentially in the mucus layer, rather than the epithelial layer, could be that the host cells metabolize 3OC12-HSL. This leads to a feasible and testable working hypothesis in that the host cells inhibit biofilm formation at the epithelial interface to such an extent that the hosts defence systems have a chance to clear any attempted colonization.

Arguably the most feared bacterial pathogen for CF patients is the opportunistic pathogen Burkholderia cepacia (De Soyza et al. 2001). It now appears that B. cepacia comprises a mixture of at least nine different but closely related genomovars of variable virulence that are now collectively referred to as Burkholderia cepacia complex (BCC). B. cenocepacia (formerly genomovar III) is the most clinically prevalent and also the most virulent. The pro-inflammatory potential of members of the BCC is mainly a property of the lipopolysaccharide (LPS), the structure and composition of which are essential for this proinflammatory activity (De Soyza et al., 2004). Tony De Soyza in the group of Anjam Khan (University of Newcastle, UK) and Paul Corris (Freeman Hospital, Newcastle, UK) presented a poster describing the persistence of pretransplant BCC strains in CF patients that have undergone lung transplants. Significantly, distinct changes in the LPS pattern were observed within one week of transplantation in two thirds of the patients, suggesting that BCC distinguishes between CF and healthy lung tissue and can adapt its LPS composition accordingly. This observation is very similar to that made in recent studies of P. aeruginosa infections of CF patients, where changes in LPS composition were also identified (Guina et al., 2003). This would suggest that the structure of LPS is highly dynamic and responds to a changing environment. This has a significant impact on the role of LPS during bacterial pathogen – host epithelial cell interactions.

The related bacterium Burkholderia pseudomallei is the aetiological agent of melioidosis, a severe disease commonly associated with fatal septicaemia and acute pneumonia (White, 2003). Melioidosis can also be manifested as a chronic infection and symptoms often do not develop for a significant period of time, up to 26 years after exposure in some reported cases. In endemic areas the incidence of melioidosis is extremely high in CF patients (O’Carroll et al., 2003). The predisposed susceptibility of CF and diabetic patients to B. pseudomallei, the ability it has to spread via aerosols and cause chronic infections have led to a dramatic increase in research on the pathogenesis of this bacterium in recent years (Stevens and Galyov, 2004).

Type III secretion systems (TTSS) facilitate the translocation of effector proteins from gram-negative bacterial pathogens directly into host cells. These effector proteins are responsible for the modulation of the host cells cytoskeleton and function to the pathogens benefit. Many TTSS effector proteins are required for the active invasion of host cells by the invading pathogens (Thomas and Finlay, 2003). One interesting feature of B. pseudomallei is that it contains two chromosomes of 4.07 and 3.17 Mb (White, 2003). Analysis of the B. pseudomallei genome sequence has revealed three virulence-related TTSS plus the flagellar associated type III secretion apparatus. Up until now the highest number of virulence-associated TTSS found in one bacterium has been two in Salmonella enterica (Nguyen et al. 2000). One of the three TTSS in B. pseudomallei (bsa) is very similar to the Inv/Mxi-Spa loci of Salmonella and Shigella respectively (Stevens et al., 2002). The other two show stronger similarity to the hrp TTSS loci in plant pathogens (Stevens et al., 2002). Ed Galyov (Institute of Animal Health, Compton, UK) presented an overview of recent work from his laboratory concerning the molecular function of the bsa locus from B. pseudomallei and reported current findings on the identification and characterization of the proteins required for the actin-based motility of B. pseudomallei. Like Listeria monocytogenes and Shigella spp., B. pseudomallei can move between host cells via actin driven motility (Stevens et al., 2002). Interestingly, B. pseudomallei uses a different pathway for actin polymerization than that used by either L. monocytogenes and Shigella (Breitbach et al. 2003).

The gastrointestinal tract – how to tell the difference

Unlike the lungs, the human gastrointestinal tract is colonized by a wide range of commensal bacteria. The majority of these bacteria are found in the colon, with relatively few present in the small intestine (Ayabe et al., 2004). To defend against unwanted pathogens, the epithelial layer of the small intestine must monitor the total bacterial flora to weed out the bad guys. In doing so, the epithelium acts as an early warning system against infection and, on recognizing an unwanted pathogen, has the ability to induce both the innate and adaptive immune response. Recent advances in understanding how epithelial cells recognize pathogens, irrespective of their location, has led to the discovery that host epithelia continually monitor for pathogen-associated molecular patterns (PAMPs) rather than waiting for a pathogen to arrive (Ayabe et al., 2004). The most common patterns scanned by the host include LPS, flagellins and bacterial DNA (Akira and Hemmi, 2003). However, the host is faced with a problem in the gastrointestinal tract in that it is continually exposed to these PAMP's from commensal bacteria. This has lead researchers to question how we, as hosts, differentiate between ‘friend’ and ‘foe’ in the bacterial flora of the gastrointestinal tract. One way would be to limit signalling through pathways found to react dramatically to bacteria in other parts of the body. For example, LPS activates the innate immune response via the toll-like receptor TLR-4 (Akira and Hemmi, 2003). However, TLR-4 expression is extremely low in intestinal epithelial cells, suggesting that continual stimulation by LPS has caused these cells to downregulate TLR-4 expression (Abreu et al., 2001). Similarly, activation of innate immune responses in intestinal epithelial cells via TLR-5-mediated flagellin recognition is averted by restricting TLR-5 localization to the basolateral side of the cells (Gewirtz et al., 2001). This suggests that, in this case, epithelial cells use flagellin as a control PAMP to monitor for any bacteria that succeed in overcoming the epithelial barrier, for example S. enterica serovar Typhimurium (Akira and Hemmi, 2003).

Now the intestinal epithelial cells face a further problem: if pathogenic bacteria do not activate the most common monitoring pathways, how may they be detected? Martin Kagnoff (University of California, San Diego, USA) presented recently published data from his laboratory showing that invasive bacteria that are not recognized by toll-like receptors in intestinal epithelial cells can nonetheless be recognized via the cytosolic recognition molecule Nod1. Recognition via Nod1 leads to activation of NF-κΒ that is required for the essential signals to induce a mucosal inflammatory response (Kim et al., 2004). They did this by expressing a dominant negative form of Nod1 in intestinal epithelial cells that are not responsive to bacterial LPS via TLR-4. They then monitored the activation of NF-κΒ on challenge with invasive, but non-flagellated, Escherichia coli 029:NM (non-motile) cells that do not activate TLR-5. NF-κΒ activation was inhibited in the presence of the dominant negative allele of Nod1. However, the same cells were still responsive to TLR-5 activation of NF-κΒ when challenged with bacterial flagellin or flagellated bacteria, suggesting that Nod1 activation of NF-κΒ is important as a backup to detect bacteria that are able to evade TLR monitoring pathways (Chamaillard et al., 2003; Kim et al., 2004).

On recognizing PAMPs, TLRs and Nod1 feed into signal transduction cascades that activate both adaptive and innate immune responses as well as inducing an inflammatory response. One branch of the innate immune system is the production of small peptides known as defensins (Ouellette, 2004). There are three families of defensins, α-, β- and θ-, of which only the α-defensins are produced in the small intestine. α-Defensins are produced by Paneth cells in the small intestine of mice and humans (Ayabe et al., 2004). The Paneth cells also produce lysozyme in response to PAMP recognition. The chicken gut produces three lysozymes, including a novel chicken lysozyme g2 (Nile et al., 2004). Claire Townes from the group of Judith Hall (University of Newcastle, UK) presented data from a study of the expression of lysozyme production in the chicken gastrointestinal tract that suggested chickens do not possess Paneth cells (Fig. 3). Their data showed that lysozyme was expressed and secreted from intestinal epithelial cells per se and not from specialized cells, as found in mice, rats, rhesus macaques and humans.

Figure 3.

Lysozyme expression in chicken intestine reveals absence of Paneth cells.
A. Immunohistochemical staining localizes lysozyme to Paneth cells in the crypts at the base of mouse intestinal villi.
B. In the chicken small intestine, lysozyme is localized to the enterocyte brush-border luminal surfaces of villi, with little staining at the base, consistent with the absence of Paneth cells. Adapted from Nile et al. (2004).

Andre Ouellette (University of California, Irvine, USA) presented recently published data on the processing of α-defensins in Paneth cells of mice. Interestingly, even though the same specialized cells produce α-defensins in mice and humans, post-translational processing and activation of the α-defensin precursors is mediated by two unrelated proteases in the two mammals: for mice the matrix metalloproteinase MMP-7 activates Paneth cell α-defensins, whereas trypsin processes α-defensin in humans (Ayabe et al., 2004). The α-defensins are characterized by the presence of a unique tridisulphide array (Ouellette, 2004). Ouellette and coworkers have shown that these disulphide bridges are not required for the bactericidal activity of the α-defensins as is widely thought. Instead their data show that the disulphide bonds are crucial in protecting the α-defensins from indiscriminate degradation by the activating protease MMP-7 (Maemoto et al., 2004). Even though trypsin is the protease responsible for α-defensin activation in humans, MMP-7 still plays a role in the battle against bacterial pathogens in humans, but in the stomach rather than the small intestine. Andrea Varro (University of Liverpool, UK) presented data showing that gastric epithelial cells increase their production of MMP-7 during infection with Helicobacter pylori. In this situation MMP-7 plays a role in epithelial cell migration and spreading. In addition, MMP-7 released from the epithelium stimulates a specific mesenchymal cell type, the myofibroblast, to migrate and proliferate. They propose that MMP-7 plays a role not only in the epithelium but also acts as a signalling molecule between epithelial and mesenchymal cells in response to infection in the stomach (Wroblewski et al., 2003).

The gastrointestinal tract – a bacterial view

During this report we have, until now, concentrated on the epithelial–bacterial interaction in the lung and the mechanisms by which gastrointestinal epithelial cells differentiate between commensal and pathogenic bacteria. Several presentations on the bacterial perspective of interactions with host epithelia in the gut also identified new and interesting aspects of bacterial pathogenesis. Significant advances in our understanding of pathogenesis were presented with respect to infections for the well-studied pathogens Salmonella and E. coli.

Salmonella is a facultative intracellular pathogen capable of invading intestinal epithelial cells and macrophages. Understanding the complexities of bacterial pathogen gene expression in vivo is essential to understanding the biology of these organisms and the mechanism by which they cause disease in the host. The identification of genes expressed in vivo and the high throughput identification of virulence genes have been revolutionized by the development of in vivo expression technology (IVET) and signature-tagged mutagenesis (STM) (Hensel et al., 1995; Mahan et al., 1993). The next leap forward in our understanding of pathogen gene expression will come with the application of DNA microarray technology to obtain transcriptional profiles of host cells and invading pathogens. Isabelle Hautefort from the laboratory of Jay Hinton (BBSRC IFR, Norwich, UK) provided some exciting unpublished data on the Salmonella transcriptional landscape within epithelial HeLa cells, complementing their previous studies in macrophages (Eriksson et al., 2003). To investigate host cell-type specificity of Salmonella gene expression, Hautefort and coworkers used an integrative approach that combined DNA microarrays with a GFP chromosomal transcriptional reporter system (Hautefort et al., 2003). New evidence was presented to show that, not too unexpectedly, the SPI1 and SPI2 pathogenicity islands are differentially expressed in macrophages and epithelial cells. Within epithelial cells Salmonella replicate within the Salmonella-containing vacuole (SCV), a unique phagosome. Hautefort presented surprising preliminary observations suggesting that Salmonella can escape the vacuolar compartment and multiply within the cytoplasm. These results are certainly of great importance for the Salmonella pathogenicity field. Why would Salmonella choose to escape the vacuole? Studies to investigate this vacuolar escape by Salmonella using different epithelial cell-types and experimental conditions are already underway.

Infections with pathogenic strains of E. coli can often be fatal. Two of the most prevalent pathogenic E. coli groups include the enterohaemorrhagic E. coli (EHEC) and enteropathogenic E. coli (EPEC). EHEC is a major problem in the developed world, where it causes ‘Hamburger disease’ and other food contaminations (Deng et al., 2004). EPEC, on the other hand, is a greater problem in the developing world. Both EPEC and EHEC possess the locus of enterocyte effacement (LEE) pathogenicity island. LEE encodes a TTSS responsible for the formation of attaching and effacing lesions (A/E lesions) on intestinal epithelium that allow E. coli to sit on the cell surface on a pedestal, formed by effector protein-induced rearrangements of the host cell cytoskeleton. Gadi Frankel (Imperial College London, UK) presented data from studies of the interaction of EHEC and the closely related bacterium Citrobacter rodentium with intestinal epithelium. Using a lux-based assay, they were able to show, to their surprise, that LEE expression is reduced upon attachment to epithelial cells. The increasing interest in EHEC compared to EPEC for infection biology has lead to the discovery of up to 20 effector proteins some of which are not linked to the LEE locus. This has led Frankel and colleagues to realize that, although the major function of the LEE locus is the formation of A/E lesions, a much greater level of communication and modulation of host cell functions occurs during these interactions. Exactly how these effector proteins contribute to pathogenesis is now of great interest.

Studies on EPEC interaction with immortalized undifferentiated HeLa cells have identified the multifunctional nature of LEE-encoded effector molecules as well as the existence of synergistic and/or antagonistic effector relationships (Jepson et al., 2003; Kenny and Jepson, 2000). In his presentation Brendan Kenny (University of Bristol) described studies with polarized layers of Caco-2 cells, which mimic many aspects of the gut epithelium, revealing that two effector molecules (Map and EspF) and the Intimin outer membrane protein function together to enable EPEC to compromise barrier function – a process linked with diarrhoeal disease. Amazingly, in EPEC the outer membrane protein, Intimin, the bacterial inserted plasma membrane receptor, Tir, and an additional unidentified receptor(s) control the action of the effectors. This new work not only reveals new functions for some effector molecules but also demonstrates redundant, synergistic and antagonistic effector relationships within this gut model system (Dean and Kenny, 2004) making it likely that such relationships also occur during infections in vivo.

Concluding remarks

In this micromeeting report, we have emphasized several common themes that emerged as major talking points during the symposium. The symposium was able to bring to the forefront some excellent examples of successful interactions of scientists from two separate disciplines (physiology and microbiology) in a similar manner to their counterparts, the epithelial cells and pathogenic bacteria. The role of the innate immune system in defending the host from invading pathogens is becoming clearer, while simultaneously identifying another level of complexity as research delves deeper into epithelial–bacterial interactions. Clearly, not all cells in the body use the same signalling pathways and current evidence suggests this might result from selective pressure that depends on the frequency with which certain cells come into contact with bacterial PAMPs. While some aspects of the innate immune system are common amongst all animals, there are examples, such as Paneth cells, that are not. Further research into such differences in the innate immune systems of certain animals might increase our understanding of why certain species are reservoirs for human bacterial pathogens while others are as susceptible to infection as humans.

The use of alternative model systems and approaches to answer a question is an important factor to better understand the intricacies of epithelial–bacterial interactions. These studies might not lead directly to the development of a new therapy or drug but they are still key in furthering our understanding of this important interaction. Similarly, well-established model systems of epithelial–bacterial interactions can still reveal unexpected surprises, such as the ability of epithelial cells to fight back by simply metabolizing bacterial signalling molecules and virulence factors, thereby rendering them inactive. These studies have also highlighted that much remains to be learned about this aspect of the pathogenic bacterial life cycle, as well as looking further into uncharacterized stages for specific pathogens. Further research on epithelial–bacterial pathogen interactions, incorporating collaborative efforts from multiple biological research disciplines, is still necessary. With time, such collaborative efforts will uncover even more fascinating aspects of these interactions, enabling us to identify alternative strategies to combat the manifestation of disease symptoms.


The authors would like to thank all those approached during the preparation of this report for their helpful discussions and comments. The Physiological Society sponsored this meeting, with additional support from Novartis and Trends in Microbiology. We would like to thank Dr Gwen Averley and Emma Chaffin for their excellent support during the organization of the meeting. The Wellcome Trust, BBSRC, Nuffield Foundation and The Newcastle upon Tyne Hospitals NHS Trust support current research in the laboratories of the authors’.