Vulnerability of Corneal Epithelial Cells to P aeruginosa In Vitro

Considering how resistant the healthy cornea is to P aeruginosa , it is striking how vulnerable the epithelial cells that line the corneal surface become when grown in vitro. More than 50% of clinical and laboratory isolates of P aeruginosa have the capacity to invade and then replicate within cultured corneal epithelial cells. Once inside the cell, they induce the formation of, and then traffic to, plasma membrane blebs, which can detach and carry the bacteria swimming within them to distant locations. This sequence of events requires ExoS, a toxin that P aeruginosa can inject across host cell membranes (a type III secretion system). Cytotoxic strains of P aeruginosa , which constitute about half of isolates that cause contact lens–related infection, lack ExoS and instead encode ExoU. Although ExoU is also a type III–secreted toxin, it causes a much more rapid form of cell death than ExoS, and it exerts its pathogenic effects while the bacteria are outside of the target cell.

In Vivo Factors and Corneal Resistance to P aeruginosa

P aeruginosa is ubiquitous in nature. As such, we are often exposed to it as we go about our daily activities. The same is true for most pathogens that cause corneal infections. Thus, it is fortuitous that the healthy cornea, in contrast to cultured corneal epithelial cells, is exquisitely resistant to microbial attack. Indeed, the inoculation of extremely large inocula (a thick bacterial suspension) of either invasive or cytotoxic P aeruginosa onto intact mouse or rat corneas in vivo results in rapid bacterial clearance from the ocular surface without pathology. Thus, defense mechanisms exist in the healthy eye that protect against corneal infection, which are absent from laboratory culture conditions. These defenses are likely to differ from the type of host immune responses that are activated when an infection occurs, since they are constantly present under conditions of health. Studying health, and factors involved in maintaining it, requires the use of completely different models and methods from those used to study disease, the latter being used for most research to date in this field. Importantly, studying parameters that maintain health is a significant challenge because of the lack of observable changes when disease is absent. To address this problem, our laboratory has developed multiple models to mimic the intrinsic resistance of the in vivo cornea in a research setting, and we have also begun to use these models to dissect the mechanisms involved in defense of the healthy cornea.

In Vivo Factors and Corneal Resistance to P aeruginosa

P aeruginosa is ubiquitous in nature. As such, we are often exposed to it as we go about our daily activities. The same is true for most pathogens that cause corneal infections. Thus, it is fortuitous that the healthy cornea, in contrast to cultured corneal epithelial cells, is exquisitely resistant to microbial attack. Indeed, the inoculation of extremely large inocula (a thick bacterial suspension) of either invasive or cytotoxic P aeruginosa onto intact mouse or rat corneas in vivo results in rapid bacterial clearance from the ocular surface without pathology. Thus, defense mechanisms exist in the healthy eye that protect against corneal infection, which are absent from laboratory culture conditions. These defenses are likely to differ from the type of host immune responses that are activated when an infection occurs, since they are constantly present under conditions of health. Studying health, and factors involved in maintaining it, requires the use of completely different models and methods from those used to study disease, the latter being used for most research to date in this field. Importantly, studying parameters that maintain health is a significant challenge because of the lack of observable changes when disease is absent. To address this problem, our laboratory has developed multiple models to mimic the intrinsic resistance of the in vivo cornea in a research setting, and we have also begun to use these models to dissect the mechanisms involved in defense of the healthy cornea.

Tear Fluid

One approach that we have used to determine which in vivo factors confer resistance to microbial attack is to consider what is missing in cell culture that makes cells vulnerable in vitro, but not in vivo. A very obvious factor missing in cell culture is the tear film. We have confirmed that tear fluid can protect corneal epithelial cells in culture against both invasive and cytotoxic P aeruginosa . Importantly, we have found that human tears can protect the injured and healing mouse cornea from infection by P aeruginosa in vivo.

How does tear fluid protect? It is well recognized that tear fluid and blinking can physically cleanse the ocular surface and wash away potential pathogens, and that tear fluid also contains molecules with direct antimicrobial activity against many microbes, for example, lysozyme and lactoferrin ( Figure , Table ). However, our data have shown that the capacity of tear fluid to protect cells against P aeruginosa is independent of direct antimicrobial activity. In fact, we found that many P aeruginosa strains, including clinical isolates from microbial keratitis, grow readily in undiluted human tear fluid, yet tear fluid can still protect corneal epithelial cells against them.

Schematic overview of intrinsic corneal defenses against P aeruginosa and other microbial pathogens during health. Several of these defenses are dependent on MyD88, an adaptor protein associated with TLR and IL-1R signaling. MyD88-dependent defenses include anti-adhesive factors of the epithelial surface, which can be removed by tissue paper blotting, and they also include epithelial barrier function against P aeruginosa traversal after bacterial adhesion is enabled (also see Table ).

If tear fluid does not inhibit bacteria growth, how does it protect cells against P aeruginosa ? Our data show that at least part of the answer to that question is that tear fluid acts directly upon corneal epithelial cells to make them more resistant to P aeruginosa virulence strategies. We showed this experimentally by pretreating human corneal epithelial cells with human tear fluid and then washing the tear fluid away before adding a bacterial inoculum. The results showed that human tear pretreatment rendered human corneal epithelial cells more resistant to invasion by invasive P aeruginosa strains and cell death caused by cytotoxic strains. Tear fluid–induced resistance was associated with an upregulation of stress response transcription factors nuclear factor κB (NF-κB) and activating protein 1 (AP-1) and the upregulated and downregulated expression of many epithelial genes. The latter includes genes encoding cytokines, transcription factors, and junctional proteins. Importantly, that work also showed that tear fluid upregulated the antimicrobial ribonuclease 7 (RNase7) and the immunomodulator ST2 (a member of the interleukin 1 receptor [IL-1R] family), and that both factors contributed to tear fluid–induced corneal epithelial cell defense against P aeruginosa . We have also shown that tear fluid increased transepithelial resistance (barrier function) of corneal epithelial cells in vitro. That phenomenon is likely to help the multilayered corneal epithelium protect itself against microbial traversal, a key event for the pathogenesis of infection. Whether or not mucosal fluids elsewhere in the body also regulate the immunity of the epithelia that they bathe is yet to be determined. However, it is also possible that this protective function of tear fluid serves to replace the now well-established roles played by commensal microbes at other sites in modulating innate defense and homeostasis.

Although tear fluid does not consistently inhibit P aeruginosa viability, it remains possible that tears suppress bacterial virulence strategies, which would augment its effects on epithelial cell immunity. Tear fluid contains mucins, secretory immunoglobulin A (sIgA), and surfactant proteins, for example, surfactant protein D (SP-D), each of which can bind microbes and potentially alter their interactions with corneal epithelial cells. Other tear components may also help defend the corneal surface against infection, including tear lipocalin, an endonuclease, and other, as-yet-unidentified, factors.

The Corneal Epithelium

For more than 3 decades, researchers who study the pathogenesis of corneal infection have used either corneal scarification or stromal injection as methods to enable susceptibility in animal infection models. The principle upon which this practice is based is that the corneal epithelium is a formidable barrier to infecting microbes, so it needs to be bypassed for infection to be initiated. However, models that bypass this layer do not enable study of the mechanisms for its resistance. To address this problem, we have experimented with more subtle manipulations of the corneal epithelium. Our goal has been to make the epithelium more susceptible to bacterial binding, both with and without susceptibility to bacterial penetration (traversal), so that we can study these events while also deciphering the defenses that protect against them. To enable us to track bacteria as they penetrate, we have developed a suite of imaging technologies that allow accurate localization of live bacteria within living mouse eyeballs over time relative to the epithelial surface, individual epithelial cells, and the underlying basal lamina.

It is commonly thought that tight junctions, which reside within the superficial cell layer, are responsible for barrier function of the corneal epithelium against penetrating microbes. However, using the methods described above we have found that this is only part of the story. Using tissue paper blotting of the corneal surface, we have shown that subtle injury to the superficial epithelium, resulting in loss of barrier function to fluorescein, allows P aeruginosa to adhere to the cornea but not penetrate beyond the epithelial surface. Thus, the tight junctions that exclude fluorescein are not needed for the corneal epithelium to stop adherent bacteria from penetrating. The fact that we are able to promote bacterial adhesion without bacterial penetration shows that defenses against these first 2 steps in corneal infection are separable, and that they are likely to involve different players.

Of course, it remains possible that some type of cell-to-cell junction(s) beyond the superficial surface are involved in stopping bacteria from penetrating the epithelium, and that the reason fluorescein, but not bacteria, go through is that they are less “tight” than the superficial tight junctions. In fact, treating the cornea with a calcium chelator, ethylene glycol tetra-acetic acid (EGTA), after tissue paper blotting does allow bacteria to penetrate the epithelium. This result could implicate the involvement of some type of cell-to-cell junction(s), since their integrity is generally calcium-dependent. However, other cellular functions that could protect against bacterial traversal are also calcium-dependent, for example, the roles of SP-D in innate defense. Indeed, one of our recent studies showed that P aeruginosa could partially traverse the tissue paper–blotted corneal epithelium of SP-D knockout mice in vivo.

Other data support the possibility that either junctional structures or antimicrobial peptides are involved in epithelial defense against P aeruginosa traversal. The corneas of mice deficient in myeloid differentiation primary response protein 88 (MyD88), a key adaptor protein of innate immunity, are susceptible to P aeruginosa penetration without the need for tissue paper blotting or EGTA treatment. MyD88 is an essential component of Toll-like receptor (TLR) signaling and IL-1R signaling, which enables corneal cells to respond to microbial antigens through the activation of cytokines and chemokines, secretion of antimicrobial peptides, and the recruitment of phagocytic cells.

MyD88 regulation of defenses against bacterial adhesion to, and bacterial penetration of, the corneal epithelium would be consistent with junctional structure involvement in defense, since TLR signaling (dependent on MyD88 for most TLRs), along with other pattern recognition receptors, help regulate the function of tight junctions in other cell types. However, MyD88 involvement in defense against bacterial adhesion and traversal may also be attributable to its importance in regulating the expression of antimicrobial peptides, including human β-defensin-2 (hBD-2) and the cathelicidin LL-37, both of which are expressed by corneal epithelial cells after stimulation with TLR or IL-1R agonists. Indeed, we have already shown that hBD-2 is important in protecting the corneal epithelium against P aeruginosa colonization. Our ongoing studies are investigating the relative roles of individual TLRs, and the IL-1R, in defense against P aeruginosa corneal adhesion and epithelial traversal, and the relative role of epithelial cells vs other resident corneal cell types that also express MyD88-dependent receptors (eg, macrophages and dendritic cells).

Our most recent studies have revealed that corneal epithelial cells express other novel antimicrobial compounds. Specifically, we have found that peptide fragments of the intermediate filament protein cytokeratin 6A, keratin-derived antimicrobial peptides (KDAMPs), isolated from lysates of human corneal epithelial cells, were rapidly bactericidal against multiple clinical isolates of P aeruginosa and against other bacterial pathogens (eg, Streptococcus pyogenes and Staphylococcus aureus ). Importantly, knockdown of cytokeratin 6A from which KDAMPs are derived reduces the antimicrobial activity of human corneal epithelial cell lysates, and in vivo renders the mouse corneal epithelium significantly more susceptible to bacterial adhesion. Cytokeratin 6A knockdown did not enable fluorescein staining, suggesting that tight junctions remained intact. Whether KDAMP expression or function is MyD88 dependent is to be determined.

The fact that MyD88 regulates the anti-adhesive nature of the corneal epithelium is interesting. Mucins (soluble and membrane-bound) are thought to be important in preventing adhesion of bacteria, such as P aeruginosa and Staphylococcus aureus, to corneal epithelial cells. The fact that tissue paper blotting enables bacterial adhesion is consistent with that assumption, since it is likely to remove, or at least disrupt, mucins at the corneal surface. Loss of corneal defense against P aeruginosa adhesion in the MyD88 knockout mouse corneas suggests either that mucin expression is MyD88 dependent or that the role of mucins is indirect, perhaps via their capacity to sequester MyD88-dependent antimicrobial factors, as shown for other tissues.

Corneal epithelial cells can internalize bacteria and can subsequently traffic them to perinuclear vacuoles within the cell, where they fail to thrive. Our more recent unpublished data indicate that vacuolar acidification reduces the viability of intracellular P aeruginosa . Whether this is involved in defense against microbial penetration through the healthy corneal epithelium is yet to be determined. Supporting that possibility, however, is our observation that when P aeruginosa is inoculated onto a healthy rat cornea, most internalized bacteria are found in cells that are readily shed from the eye with rinsing. That result supports the notion that internalization/cell shedding is a mechanism for clearing bacteria that manage to adhere to the surface.

If P aeruginosa does manage to traverse the multilayered corneal epithelium and all of its defenses, the epithelial basement membrane (the basal lamina), composed mostly of extracellular matrix proteins (eg, laminin and collagen type IV), prevents them from actually entering the corneal stroma. The basal lamina does this in 2 ways: one physical and the other biochemical. The basal lamina acts as a physical filter because it is a mesh containing pores smaller than the size of most bacteria. This filtering role played by the basal lamina explains why making the corneal epithelium susceptible to bacterial adhesion/traversal (using either EGTA or MyD88 knockout mice) does not necessarily result in microbial keratitis (disease/pathology). The pathology that occurs during microbial keratitis requires bacterial entry into the stroma, which then leads to the activation of inflammatory and immune responses and their subsequent damaging sequelae (eg, [refs ]). In the laboratory, the filtering role of the basal lamina can easily be observed using in vitro or in vivo models. Even within corneas made susceptible to disease by scratch injury, penetrating bacteria distant from the scratch-injured area can be seen aligned on the anterior surface of still-intact basal lamina. After scarification, the cornea regains its resistance to infection within 12 hours, which corresponds to the time that bacteria are no longer able to take that final step into the stroma. Interestingly, reacquisition of resistance to corneal infection occurs before barrier function to fluorescein staining is completely reestablished. These results provide further evidence that fluorescein staining is a poor predictor of susceptibility to infection, and that other defense mechanisms (eg, the basal lamina) can still protect the cornea against bacterial penetration when superficial tight junctions are compromised.

Our in vitro modeling experiments confirmed that basal lamina extracellular matrix proteins can form a barrier to bacterial passage. However, that study also showed another role for the basal lamina in defense, which was to improve the barrier function of the epithelial cells growing on top of them. The mechanism(s) by which these proteins impact the barrier function of cells on the opposite side of the multilayer is yet to be determined, but could involve effects on junctional integrity or antimicrobial peptide/mucin expression. Whatever the case, the intact basal lamina is another factor that is lacking in standard cell culture assays that could relate to the increased susceptibility of corneal epithelial cells to bacteria when grown in vitro.

In summary, corneal epithelial–associated barriers to P aeruginosa consist of defenses against adhesion and defenses against microbial penetration (traversal). The players involved likely include junctional complexes, secreted and internal antimicrobial peptides, mucins, and the basal lamina foundation that provides a physical barrier while also supporting epithelial homeostasis. During and after P aeruginosa challenge, corneal epithelial defenses are enhanced and regulated by epithelial-derived cytokines and chemokines that can facilitate communication with cells of the immune system to boost corneal defenses.