ILAR Journal Online, Volume 40(2) 1999: Animal Models of Human Vision

Online Issues

<< All Back-issues

<< This Issue's Table of Contents

ILAR Journal V40(2) 1999
Animal Models of Human Vision

Use of an Animal Model in Studies of Bacterial Corneal Infection
Brigitte A. Cowell, Christine Wu, and Suzanne M. J. Fleiszig

Brigitte A. Cowell, Ph.D., Postdoctoral Fellow, Christine Wu, B.S., Research Assistant, and Suzanne M. J. Fleiszig, O.D., Ph.D., Assistant Professor, are in the School of Optometry, University of California-Berkeley, Berkeley, California.

Introduction

Bacterial corneal infection (bacterial keratitis) is the most common cause of suppurative corneal ulceration (Wilhelmus 1996). Despite continual improvement of available medical therapies, permanent corneal opacity frequently results. Occasionally the infection will be so destructive that a transplant is required. Approximately 0.5 to 1.0% of US cases of corneal infection require surgery. More seriously, bacterial keratitis is a major cause of vision loss in developing countries (trachoma and xerophthalmia being the most common causes).

Immune Mechanisms of the Eye

The corneal surface is continually exposed to microorganisms from the air and skin yet is infrequently colonized by bacteria. The eye is normally an immunologically privileged organ, and unique features such as eyelid function and bulk flow of tears are thought to be significant factors in protection against corneal infection (Foster ! 987). A breakdown or subversion of normal host defense mechanisms is required for infection of the cornea by bacteria to occur. These defenses commence with eyelid function and lashes, which provide some protection. This protection is augmented by the rinsing action of tear flow and antimicrobial constituents of tears (Smolin 1987). Lysozyme is an antimicrobial tear protein that cleaves bacterial cell wall peptidoglycan of susceptible bacteria, which are mainly Gram positive. Lactoferrin has a direct bactericidal effect as well as iron-binding properties that inhibit bacterial growth. Immunoglobulins (especially secretory immunoglobulin A) and complement components add to the array of host defenses encountered in the tear film (Franklin 1989). Furthermore, corneal epithelial cells are capable of phagocytosis (Fleiszig and others 1995). Potentially, this would be another defense mechanism to use against bacterial contamination in which frequent sloughing of the outermost layers of epithelial cells would remove ingested bacteria. Ocular mucin plays a role in defense against infection because of its ability to bind bacteria and by constant exchange during blinking (Fleiszig and others 1994). Although the cornea is normally avascular, immune cells such as polymorphonuclear leukocytes are rapidly recruited from iris and limbal vessels in response to the presence of infecting bacteria.

Bacterial Infection of the Cornea

Clearly, impairment of normal corneal defenses increases the risk of bacterial infection. The anterior route of access accounts for most cases of suppurative keratitis. One of the most common predisposing factors for corneal infection is contact lens wear, which may stagnate the tear film and predispose to infection by decreasing clearance of bacteria from the corneal surface (Fleiszig and others 1992). Lens wear may also directly compromise the integrity of corneal epithelium. The risk of infection is significantly enhanced by overnight wear of lenses compared with a daily wear schedule (Schein and others 1989). Indeed, approximately 40% of corneal infections recorded over a 3-yr period in New York were contact lens related (Levine and Serdarevic 1998). A national study in the Netherlands estimated the infection risk for contact lens wearers to be 1.1 per 10,000 for rigid gas-permeable lens wear, 3.3 per 10,000 for daily soft lens wear, and 22.9 per 10,000 for extended wear (Cheng and others 1998). In 1992, the number of Americans wearing contact lenses was reported to be approximately 24 million (Buehler and others 1992). Although the risk of infection for any particular lens wearer is low, these estimates nevertheless translate into a significant number of infections that are potentially sight threatening.

Other factors that may increase the risk of corneal infection include dry eye syndrome and blepharitis, which are reported to predispose to infection, and other ocular diseases, which alter the normal tear film (Lemp 1990). Trauma and surgery disrupt the protective epithelial barrier and allow microorganisms direct access to the stroma (Kreger 1983). Typically the corneal stroma is avascular; therefore, significant bacterial proliferation may occur before immune cell arrival. Occasionally bacterial infection of the cornea may occur via limbal blood vessels during sepsis from infectious scleritis, or through the endothelium during intraocular infection (Wilhelmus 1996).

Bacterial Virulence Mechanisms

Pathogens have evolved strategies to avoid killing by the host defense system. Although the presence of antibacterial proteins such as lactoferrin and lysozyme in the tear film prevents colonization by some bacteria, ocular pathogens such as Pseudomonas aeruginosa have been demonstrated to be resistant to their effect (Cowell and others 1997). The production of an exopolysaccharide capsule or slime layer has been reported to allow pathogens such as Streptococcus pneumoniae and Staphylococcus epidermidis to resist phagocytosis (Bowman and McCulley 1994). By masking complement binding, the O-side chains of bacterial lipopolysaccharide can allow P. aeruginosa to avoid phagocytosis (Cryz and others 1984).

P. aeruginosa is one of the most commonly isolated pathogens from contact lens-associated ulcerative incidents (Levine and Serdarevic 1998). Patients infected with P. aeruginosa develop severe ulcers and often require more extensive treatments than patients with infections caused by other pathogens (Cheng and others 1998). Corneal destruction during P. aeruginosa infection is rapid, and perforation and/or loss of vision is possible within 24 hr (Hazlett 1995).

P. aeruginosa is an opportunistic Gram-negative pathogen that is ubiquitous in the environment and commonly found contaminating lens cases and solutions (Holland and others 1993). Once adhered to the lens or trapped under the lens proximal to the cornea, bacteria are more difficult to remove by the normal mechanical action of the eyelid and may additionally resist the host defenses of the tear film. P. aeruginosa has recently been demonstrated to cause damage to the intact cornea in vitro after only hours of contact (Fleiszig and others 1998). This damage indicates that stagnation of tear fluid under the lens is potentially involved in the development of some corneal infections.

Both bacterial and host factors are considered to be involved in corneal damage resulting from P. aeruginosa infection (Holland and others 1993). P. aeruginosa produces many toxins and proteases that cause tissue damage, while host proteases released by neutrophils (especially elastase) are also destructive. Exotoxin A, exoenzyme S, and phospholipase C are some of the identified virulence factors of P. aeruginosa reported to be important in disease progression (Holland and others 1993). ExoU has recently been demonstrated to be essential for acute cytotoxicity of some strains of P. aeruginosa (Finck-Barbançon and others 1997). Protease production (such as alkaline protease and elastase) is thought to be responsible for corneal damage during infection (Hobden and others 1993), and pill and lipopolysaccharide have also been reported to be involved (Gupta and others 1994).

The host immune system (which includes antibodies, complement, and phagocytic cells) has a significant effect on disease progression (Berk and others 1991). The inflammatory cell response, in particular, is critical to development of disease (Hazlett and others 1991). The primary immune cells recruited in the initial stages of corneal infection are neutrophils. Lysosomal products released during activation and phagocytosis include cationic proteins, acid proteases, and neutral proteases (Bowman and McCulley 1994). Cationic proteins increase vascular permeability and are chemotactic for mononuclear phagocytes. Acid proteases degrade basement membrane, and neutral proteases are active against fibrin, elastin, and collagen. Neutrophils, when exposed to bacteria, release abundant reactive oxygen species that damage ocular tissue in addition to the targeted bacteria (Henson and Johnston 1987). Upregulation of expression of numerous cytokines and chemokines has been detected in response to bacterial corneal infection, possibly contributing to corneal damage by augmenting or prolonging the inflammatory response (Kernacki and others 1998).

We need to better understand the pathogenesis of corneal infections if we are to develop new strategies for prevention or therapy. P. aeruginosa infections do not always respond well to antibiotics to which they are susceptible in vitro. Furthermore, bacteria such as P. aeruginosa are rapidly becoming resistant to current antibiotic agents (Ciofu and others 1994).

Clinical Features of Bacterial Keratitis

Contact lens wearers who develop bacterial keratitis generally do not report recent trauma (Wilson 1996). Symptoms usually commence with a foreign body sensation or irritation that increases in intensity to persistent pain. This development is accompanied by redness, photophobia, and blurred vision. Focal stromal suppuration is characteristic of early bacterial keratitis, which consists of a dense yellow-white or gray-white stromal infiltrate with an overlying epithelial defect and adherent mucopurulent exudate (Wilhelmus 1996). Multiple infiltrates are sometimes seen in contact lens wearers. Stromal ulceration and liquefactive necrosis may develop if the disease progresses without treatment. Stromal edema and striae usually surround the area of ulceration. A ring-shaped infiltrate may develop, indicating an inflammatory reaction to endotoxin produced by Gram-negative bacteria (Wilhelmus 1996). Active migration of inflammatory cells in the adjacent stroma can blur the edges of the focal infiltrate. Intraocular inflammation can range from a few aqueous cells and flare to hypopyon formation (accumulation and settling of white blood cells in the anterior chamber) correlating with vascular congestion of the iris.

Corneal infection induced by P. aeruginosa typically presents as a rapidly progressing suppurative stromal infiltrate with marked mucopurulent exudate (Wilhelmus 1996). Yellowish coagulative necrosis surrounded by epithelial edema resulting from inflammation is characteristic of infection caused by this pathogen. Rapidly escalating ulcer size and degree of suppuration can result in substantial stromal loss. A hypopyon is usually present, and descemetocele formation (protrusion of Descemet' s membrane due to thinning of overlying tissue) or perforation may occur.

In Vitro Versus in Vivo Infection Models

Like most tissues of the human body, the cornea is complex in nature. The corneal epithelium is a highly organized and regular, stratified squamous epithelial cell layer (Figure 1; also Gipson and Sugrue 1994). Basal cells of the epithelium adhere to a basal lamina called Bowman's layer. The underlying stromal layer is composed of collagen fibrils tightly packed in parallel arrays and fibroblasts in a proteoglycan-containing matrix (Olsen and McCarthy 1994). The single layer of endothelial cells that complete the corneal complex secretes a thick basement membrane (Descemet' s membrane) that separates the endothelium from the stroma. A mucin-containing glycocalyx is secreted by corneal and conjunctival cells that stabilize the overlying tear film. An intact tear film is necessary for corneal cell health, for maintenance of clear vision, and for protection of the corneal surface from foreign materials, including bacteria (Dartt 1994). Blinking augments the spreading of tears and promotes corneal epithelial cell sloughing. Although the cornea is normally avascular, immune cells are readily recruited during infection and play a critical role both in the development of inflammation and in the resolution of bacterial infection. The complexities of the cornea, the immune system, and their interaction with invading bacteria make the use of animal models of corneal infection critical to developing an understanding of pathogenesis of corneal infection and subsequently devising effective therapies for this potentially blinding disease.

The use of animals in models of bacterial infection must be fully justifiable, and the ethics of animal experimentation have been vigorously debated for a number of years (Goodwin and Gordis 1991; Naquet 1993; Parker 1994). Whenever possible, scientists are obliged to reduce the numbers of animals used, refine models to reduce stress to the animals, and finally seek alternative methods (Russell and Burch 1959). Relative replacement, a strategy widely practiced in biomedical research (Balls 1994), involves the humane killing of animals to provide cells, tissues, and organs for in vitro studies. Although animals are still required for the initial source of tissue, development of an immortalized cell line significantly reduces the requirement for animals in many areas of research. Use of organs in vitro eliminates infecting a living animal but does not necessarily reduce the number of animals required to complete a study.

One approach to studying pathogenesis of infection of the cornea in vitro that has been used extensively by researchers is the separation of cell types by using individual cell lines such as cultured corneal epithelial cells. This approach allows investigation of the specific interactions between bacteria and the particular cell type. However, cultured cells are homogeneous, may be unpolarized, and lack surrounding structures that are seen in vivo with stratified tissue such as the cornea. These differences can result in altered bacterial interactions in the artificial system. For example, bacteria associate in higher numbers with cultured epithelial cells than with cells on whole cornea (Fleiszig and others 1995). In the eye, the epithelial layer is composed of polarized epithelial cells that vary in shape and function as they approach the apical surface. Polarity is one cell characteristic that strongly influences susceptibility of cells to bacterial invasion (Fleiszig and others 1997b). Cell to cell communication can also influence the behavior of epithelial cells. Other cell types present in corneal tissue apart from epithelial cells (such as keratocytes and endothelial cells) may be involved in the infectious process directly or via effects on epithelial cell function.

Bacterial interactions with epithelial cells also may be studied using human corneal cells collected from living subjects by ocular irrigation (Fleiszig and others 1992; Fullard and Wilson 1986). The advantages of this model are that it allows investigators to study human cells and to examine differences between people who have differing risks of infection. Although this method may provide information about initial events occurring after contact of the bacteria with corneal epithelial cells, cells collected in this manner are similar to cultured cells in that they lack polarity and surrounding structures of the tissue from which they originated as soon as they are removed from the eye. Furthermore, only a small number of cells can be collected from each subject.

Organ culture of excised tissue allows investigation of bacterial infection of whole tissue in which cells have grown in vivo and in which important characteristics such as stratification and polarity are retained. Use of enucleated eyes enables investigators to exclude the activity of the host immune system in the design of experimental protocols, which is useful when the direct effects of bacteria on host tissues are studied.

Significant insights into the pathogenesis of P. aeruginosa infection of the cornea have been gained using an immortalized rabbit corneal epithelial cell line and organ culture of rat eyes (Fleiszig and others 1995, 1996, 1998). Initial attachment of P. aeruginosa has also been studied using human corneal epithelial cells harvested by corneal irrigation (Fleiszig and others 1992). With these methods, bacterial virulence factors shown to be important for infection of cells in vitro have been demonstrated to correlate with virulence in vivo in the mouse model described below (Fleiszig and others 1994, 1996).

Thus, the infectious process can be dissected in vitro in various ways. Nevertheless, investigators must ultimately also study living systems to fully understand the complexity of the infectious process. The study of pathogenesis of infectious disease in the cornea has progressed largely because of the development of animal models for in vivo studies.

Animal Models of Bacterial Corneal Infection

Rabbits and mice have been the most commonly utilized laboratory animals for the study of corneal infections (Gerke and Magliocco 1971; Hazlett and others 1976; Hobden and others 1988), although models using rats and Chinese hamsters have also been described (Twining and others 1986; van Klink and others 1996). Many of the clinical features of the human corneal disease can be satisfactorily reproduced in the rodent models (Berk 1993). The rabbit is an attractive model because contact lenses may be fitted easily, allowing design of the animal model to be as close to the human situation as possible. The disadvantages of using rabbits are the substantial cost of the number of animals required for statistical power in analysis of experimental results and the time required to handle the corresponding number of animals of this size. The advantages of using mice for in vivo experiments are that they are more convenient to handle, require less housing space, and are significantly less expensive both to purchase and to maintain. Furthermore, their immune system is well characterized, and gene-knockout mice that lack various important immune system components are readily available (Berk 1993; Berk and others 1987). Corneal scarification before infection, intrastromal injection of bacteria, or overnight contact lens wear is considered to be a requirement for the establishment of bacterial infection in an animal model.

The mouse model has been the most extensively used for studies of P. aeruginosa corneal infection (Beisel and others 1983; Gerke and Magliocco 1971; Preston and others 1995). In this model, mice are anesthetized, and one cornea of each mouse is scratched with a sterile needle to produce a corneal defect. A suspension of P. aeruginosa is inoculated onto the damaged corneal surface, and infection is allowed to progress for various time periods before examination and sacrifice. Corneal disease caused by P. aeruginosa is characterized in this model by corneal opacity with extensive infiltration of neutrophils (Berk and others 1979). Severe infection normally develops within 12 hr of scarification and inoculation; the infection remains localized and does not spread to the contralateral uninfected eye or to the bloodstream (Hazlett and others 1976). Depending on the strain and the inoculum used, the infection either will resolve or will progress to corneal perforation, phthisis bulbi, or severe ocular destruction (Berk 1993).

In the infection model described by Beisel and coworkers (1983), mice anesthetized with ether (no longer approved for use as an anesthetic agent) remained unconscious for a short time deemed sufficient for scarification and inoculation with a bacterial suspension (less than 1 min). Only a limited number of bacterial strains have been reported to infect corneas using this method (Hazlett and others 1991). A modified protocol has been described by Preston and others (1995), wherein mice are injected intraperitoneally with an anesthetic cocktail, resulting in a more prolonged state of anesthesia. This procedure increases contact of the bacterial suspension with the wounded cornea before the animal would typically wake and resume blinking (approximately 20 min). Another advantage of this modified model is that the investigator has more time to carefully infect the eye. With this model, a wider range of bacterial strains infected the cornea, and at a reduced inoculum. This model is described in detail below.

Need for a New Scoring System

Since the early 1980s, the murine corneal infection model has been scored using a simple scoring system (see Beisel and others 1983). In this system, infection is graded and scored from 0 to 4, depending mainly on the size of the resultant ulceration. A grade of 0 represents an eye identical to the uninfected contralateral control eye; grade 1, faint opacity partially covering the pupil; grade 2, dense opacity covering the pupil; grade 3, dense opacity covering the entire anterior segment; and grade 4, perforation of the cornea and/ or phthisis bulbi (shrinkage of the eyeball).

As described above, the appearance of corneal ulceration in humans is complex, and assessment by the clinician generally takes into account various features apart from the size of the ulcer (see Harrison 1975). Human ocular disease is also commonly scored using a grading scale of 0 to 4, but many aspects are graded, including area, depth, density, edema, folding, position, and lid reaction. Some of these aspects are difficult to evaluate or are not useful in the mouse scratch model. For example, scarification of the central cornea determines the central position of the resultant ulcer. Furthermore, the small size of the mouse eye leads to practical difficulties. Nevertheless, a grading system that describes mainly the area of ulceration does not account for other features that could be relevant in the interpretation of data, such as the distinction between moderate infiltration and dense ulceration if both cover the same area of cornea.

Area, density, and surface regularity are three attributes of ulceration of corneal epithelium that might be affected by more than one factor of bacteria and/or host immune system. Area is likely to be due not only to the spread of the organism from the initial site of infection, but also to the amount and location of chemotactic factors produced. The density will be determined by the amount of neutrophil infiltration and the amount of corneal damage by proteases and other degradative products of both bacteria and neutrophils. Lack of surface regularity during infection is likely to reflect both the loss of structural integrity resulting from stromal damage or edema and roughness of the surface that could be due to epithelial cell disruption or loss. Thus, it is likely that different virulence factors of P. aeruginosa could have different effects on these clinical features.

In this article, we describe a new scoring system wherein four attributes of corneal infection are graded and scored 0 to 4: area of major opacity, density of this opacity, density of infiltration of the surrounding cornea, and surface regularity. To demonstrate the usefulness of this system, we illustrate its use in a study investigating the role of a bacterial transcriptional regulator in virulence of P. aeruginosa.

Design of a New Scoring System

The modified grading system shown in Table 1 addresses various features of the gross appearance of corneal infection. This system is likely to be more sensitive in evaluating virulence of a bacterial strain, the influence of the immune system, and the effects of potential therapeutic agents. Since it incorporates an assessment of surface regularity, it may be particularly useful for examining epithelial health. For this reason, this new system is likely to be useful for studies of virulence factors that affect interactions with epithelial cells, such as the ExsA pathway in P. aeruginosa. P. aeruginosa
has an impressive range of potential virulence factors that would be expected to contribute to its pathogenesis.

Background

Corneal isolates of P. aeruginosa may be grouped into two phenotypes, invasive and cytotoxic, according to their interaction with corneal epithelial cells in vitro. It has been shown in vitro using cultured corneal epithelial cells that cytotoxic strains of P. aeruginosa remain outside the host cells and induce acute cytotoxicity from this extracellular location within 3 hr (Evans and others 1998; Fleiszig and others 1996). This acute cytotoxicity is dependent on the expression of ExoU. In contrast, invasive P. aeruginosa enter epithelial cells and kill cells from an intracellular location, and this killing occurs only after prolonged incubation (at least 20 hr; Andika and others 1998). This "chronic" form of cytotoxicity is independent of ExoU, since invasive strains lack the gene that encodes this protein.

Both invasive and cytotoxic P. aeruginosa cause disease in the scarified cornea model, but the pathologies have been shown to be distinct. In one study, an invasive strain of P. aeruginosa(6294) produced massive stromal infiltration (Cole and others 1998). The overlying epithelial ulceration that developed after 1 day was larger in area than the initial corneal defect, and there were moderate corneal edema and severe conjunctival hyperemia. A cytotoxic strain, 6206, caused a dense ring infiltrative response to infection. In this case, there were extreme corneal edema, anterior chamber response, and conjunctival hyperemia. The epithelial defect was confined to the original area and had not enlarged, as was observed during infection with the invasive strain.

The phenotype of cytotoxic strains of P. aeruginosa is regulated at the genetic level by ExsA, a transcriptional activator of a number of virulence-associated proteins, including ExoS and ExoU (Fleiszig and others 1997a). Interestingly, an exsA mutant of the invasive strain PAO1 shows significantly less chronic cell death in vitro than wild type PAO 1, indicating that the ExsA-regulated pathway may also be critical to virulence of invasive phenotypes of P. aeruginosa (Andika and others 1998). For this reason, we studied the effect of the exsA mutation on disease caused by PAO1.

Procedure

Bacteria

P. aeruginosa PAO 1 (invasive phenotype) and PAO 1 exsA::W were stored at -70°C in glycerol stocks before inoculation onto tryptic soy agar plates and incubation overnight at 37°C. Bacteria were resuspended in buffered minimal essential medium (BMEM¹)to OD₆₅₀ = 0.1, corresponding to a concentration of 10⁸ cfu/mL. Suspensions were diluted to final concentrations of 2 x 10⁷ cfu/mL in BMEM. Viable counts of serial dilutions of bacterial suspensions were made to confirm this concentration.

Infection of Mice

C57/BL6 female mice (5 to 6 wk old) were obtained from Jackson Laboratories (Bar Harbor, Maine). Animals were anesthetized by intraperitoneal injection with an anesthetic cocktail of 21 mg/mL of ketamine, 2.4 mg/mL of xylazine, and 0.3 mg/mL of acepromezine. Eyes were checked for corneal clarity under a stereomicroscope before the left eye was scarified by three parallel incisions to the central cornea with a sterile 25 5/8 G needle. Care was taken not to penetrate the corneal stroma. Bacterial suspension (5mL, 10⁵ cfu) or BMEM (negative control) was applied to the scarified cornea. Animals were observed daily for 7 days. At 1, 2, and 7 days, infection was graded by a masked investigator using a stereomicroscope, and photographs were taken. The initial grading scheme (0 to 4) is described above and in Beisel and others (1983). At 7 days, animals were sacrificed by cervical dislocation after anesthesia. Animals were treated in accordance with the Resolution on the Use of Animals by the Association for Research in Vision and Ophthalmology, and the protocol used for infection was approved by the University of California-Berkeley Animal Care and Use Committee.

Data were expressed as median values with upper and lower quartiles. Statistical significance was assessed non-parametrically using the Mann-Whitney test.

Results and Discussion

Results using the original scoring system that assessed mainly the area of the opacity were identical for the PAO 1 parent strain and exsA- mutant strain at day 1 (Table 2). Nevertheless, it was obvious from photographs of representative eyes from infected mice that the infection was less severe in the group infected with exsA- mutant (Figure 2). By day 7, the assigned scores reflected this difference (Table 2, p < 0.05).

Using the modified scoring system described in Table 1, the photographs were used to rescore the same infected eyes. The results (Table 3) show differences in infection intensity between the exsA- mutant and the wild type PAO 1 at all three time points, including the 24- and 48-hr time points, which were not detected by the conventional scoring system.

Since the virulence of P. aeruginosa is multifactorial, eliminating one factor (such as the ExsA-regulated pathway) only reduced the severity of infection but did not render the pathogen avirulent. By grading three different attributes of the appearance of corneal ulceration, the ability to detect differences between the mutant and the parent strain was more sensitive. Scores for density of major opacity and surface regularity were lower at all time points for the exsA-mutant compared with the parent PAO1 (Table 3, p < 0.05). The reduced corruption of surface regularity may reflect reduced killing of epithelial cells, as has been shown in vitro (Andika and others 1998). This new system allowed us to confirm in vivo that the ExsA pathway is important in chronic cytotoxicity caused by invasive P. aeruginosa. Although significant infiltration by neutrophils would have contributed to the density grade initially, this score had decreased by 7 days, and the infection was close to resolution. In contrast, the parent strain showed significant opacity of cornea, swelling, and loss of corneal integrity at 7 days. This evidence is likely to reflect growth of P. aeruginosa, cytotoxicity toward corneal cells, and continued chemotaxis and activation of immune cells.

It is clear that experimentation using currently available in vitro models can indicate bacterial and host factors that are important to development of corneal disease. One limitation of these systems is that they do not replicate the complex organization of corneal tissue or the balanced host defense system that will be involved in interactions with the pathogen that leads to bacterial keratitis. In vivo models allow experimenters to further elaborate how virulence factors interact during the complex sequence of events that occur during corneal infection. Thus, carefully designed studies that rationally utilize both in vitro and in vivo systems are necessary for the investigation of bacteria-host interactions crucial to disease formation and for testing of future therapies. Development of better animal models, and improving the accuracy of the measurements derived from these studies, will improve our ability to make in vivo studies as relevant as possible to disease in humans.

¹Abbreviation used in this paper: BMEM, buffered minimal essential medium.

Acknowledgments

B.A.C. was supported by a Bausch & Lomb postdoctoral fellowship and S.M.J.F., by National Institutes of Health grant RO1 EY11221. The authors thank Douglas Hamaker for expert graphical assistance.

References

Andika RC, Wu C, Fleiszig SMJ. 1998. Chronic cytotoxicity induced by invasive Pseudomonas aeruginosa is invasion dependent and involves Exs A. (Abstract). Invest Ophthalmol Vis Sci 39:S143.

Balls M. 1994. Replacement of animal procedures: Alternatives in research, education and testing. Lab Anim 28:193-211.

Beisel KW, Hazlett LD, Berk RS. 1983. Dominant susceptibility effect on the murine corneal response to Pseudomonas aeruginosa. Proc Soc Exp Biol Med 172:488-491.

Berk RS. 1993. Genetic regulation of the murine corneal and non-corneal response to Pseudomonas aeruginosa. In: Campa M, Berdinelli M, Friedman H, editors. Pseudomonas aeruginosa as an Opportunistic Pathogen. New York: Plenum Press. p 186-206.

Berk RS, Leon MA, Hazlett LD. 1979. Genetic control of the murine corneal response to Pseudomonas aeruginosa. Infect Immun 26:1221-1223.

Berk RS, Hazlett LD, Beisel KW. 1987. Genetic studies on resistant and susceptibility genes controlling the mouse cornea response to infection with Pseudomonas aeruginosa. Antibiot Chemother 39:83-91.

Berk RS, Preston MJ, Montgomery IN, Hazlett LD, Tse TY. 1991. Antibody hyporesponsiveness in resistant BALB/CJ intracorneally infected with Pseudomonas aeruginosa. Reg Immunol 3:186-191.

Bowman RW, McCulley JP. 1994. Ocular bacteriology. In: Albert DM, Jakobiec FA, editors. Principles and Practice of Ophthalmology. Philadelphia: WB Saunders. p 816-834.

Buehler PO, Schein OD, Stamler JF, Verdier DD, Katz J. 1992. The increased risk of ulcerative keratitis among disposable soft contact lens users. Arch Ophthalmol 110:1555-1558.

Cheng KH, Leung SL, Hoekman JW, Beekhuis WH, Geerards AJM, Kijlstra A. 1998. Incidence of contact lens-associated ulcerative keratitis and its related morbidity. (Abstract). Invest Ophthalmol Vis Sci 39:S337.

Ciofu O, Giwercman B, Pedersen SS, Hoiby N. 1994. Development of antibiotic resistance in Pseudomonas aeruginosa during two decades of antipseudomonal treatment at the Danish CF center. Apmis 102:674-680.

Cole N, Willcox MDP, Fleiszig SMJ, Stapleton F, Bao B, Tout S, Husband A. 1998. Different strains of Pseudomonas aeruginosa isolated from ocular infections or inflammation display distinct corneal pathologies in an animal model. Curr Eye Res 17:730-735.

Cowell BA, Willcox MDP, Schneider RP. 1997. Growth of Gram-negative bacteria in a simulated ocular environment. Aust NZ J Ophthalmol 25:S23-S26.

Cryz SJ Jr, Pitt TL, Fürer E, Germanier R. 1984. Role of lipopolysaccharide in virulence of Pseudomonas aeruginosa. Infect Immun 44:508-513.

Dartt DA. 1994. Tear physiology and biochemistry. In: Albert DM, Jakobiec FA, editors. Principles and Practice of Ophthalmology. Philadelphia: WB Saunders. p 1043-1049.

Evans DJ, Frank DW, Finck-Barbançon V, Wu C, Fleiszig SMJ. 1998. Pseudomonas aeruginosa invasion and cytotoxicity are independent events, both of which involve protein tyrosine kinase activity. Infect Immun 66:1453-1459.

Finck-Barbançon V, Goranson J, Zhu L, Sawa T, Wiener-Kronish JP, Fleiszig SMJ, Wu C, Mende-Mueller L, Frank DW. 1997. ExoU expression by Pseudomonas aeruginosa correlates with acute cytotoxicity and epithelial injury. Mol Microbiol 25:547-557.

Fleiszig SMJ, Fletcher EL, Lowe R. 1992. Microbiology of contact lens care. In: Bennett ES, Weissman BA, editors. Clinical Contact Lens Practice. Philadelphia: JB Lippincott. p 1-29.

Fleiszig SMJ, Zaidi TS, Fletcher EL, Preston MJ, Pier GB. 1994. Pseudomonas aeruginosa invades corneal epithelial cells during experimental infection. Infect Immun 62:3485-3493.

Fleiszig SMJ, Zaidi TS, Pier GB. 1995. Pseudomonas aeruginosa invasion of and multiplication within corneal epithelial cells in vitro. Infect Immun 63:4072-4077.

Fleiszig SMJ, Zaidi TS, Preston MJ, Grout M, Evans DJ, Pier GB. 1996. Relationship between cytotoxicity and corneal epithelial cell invasion by clinical isolates of Pseudomonas aeruginosa. Infect Immun 64:2288-2294.

Fleiszig SM J, Wiener-Kronish JP, Miyazaki H, Vallas V, Mostov KE, Kanada D, Sawa T, Yen TS, Frank DW. 1997a. Pseudomonas aeruginosa-mediated cytotoxicity and invasion correlate with distinct genotypes at the loci encoding exoenzyme S. Infect Immun 65:579-586.

Fleiszig SMJ, Evans DJ, Do N, Vailas V, Shin S, Mostov KE. 1997b. Epithelial cell polarity affects susceptibility to Pseudomonas aeruginosa invasion and cytotoxicity. Infect Immun 65:2861-2867.

Fleiszig SMJ, Lee EJ, Wu C, Andika RC, Vailas V, Portoles M, Frank DW. 1998. Cytotoxic strains of Pseudomonas aeruginosa can damage the intact corneal surface in vitro. CLAO J 24:41-47.

Foster CS. 1987. Eye and immunity. Antibiot Chemother 39:77-82. Franklin RM. 1989. The ocular secretory immune system: A review. Curr Eye Res 8:599-606.

Fullard RJ, Wilson GS. 1986. Investigation of sloughed corneal epithelial cells collected by non-invasive irrigation of the corneal surface. Curr Eye Res 5:847-856.

Gerke JR, Magliocco MV. 1971. Experimental Pseudomonas aeruginosa infections of the mouse cornea. Infect Immun 3:209-216.

Gipson IK, Sugrue SP. 1994. Cell biology of the corneal epithelium. In: Albert DM, Jakobiec FA, editors. Principles and Practice of Ophthalmology. Philadelphia: WB Saunders. p 3-16.

Goodwin FK, Gordis E. 1991. Correspondence. New Biologist 3:907-908.

Gupta SK, Berk RS, Masinick S, Hazlett LD. 1994. Pili and lipopolysaccharide of Pseudomonas aeruginosa bind to the glycolipid asialo GMI. Infect Immun 62:4572-4579.

Harrison SM. 1975. Grading corneal ulcers. Ann Ophthalmol 7:537-542.

Hazlett LD. 1995. Analysis of ocular microbial adhesion. Methods Enzymol 253:53-66.

Hazlett LD, Rosen D, Berk RS. 1976. Experimental eye infections caused by Pseudomonas aeruginosa. Ophthalmic Res 8:311-318.

Hazlett LD, Moon MM, Singh A, Berk RS, Rudner XL. 1991. Analysis of adhesion, piliation, protease production and ocular infectivity of several Pseudomonas aeruginosa strains. Curr Eye Res 10:351-362.

Henson PM, Johnston RB Jr. 1987. Tissue injury in inflammation. J Clin Invest 79:669-674.

Hobden JA, Rootman DS, O'Callaghan RJ, Hill JM. 1988. Ionophoretic application of tobramycin to uninfected and Pseudomonas aeruginosa-infected rabbit corneas. Antimicrob Agents Chemother 32:978-981.

Hobden JA, Engel LS, Callegan MC, Hill JM, Gebhardt BM, O'Callaghan RJ. 1993. Pseudomonas aeruginosa keratitis in leukopenic rabbits. Curr Eye Res 12:461-467.

Holland SP, Pulido JS, Shires TK, Costerton JW. 1993. Pseudomonas aeruginosa ocular infections. In: Fick RB Jr, editor. Pseudomonas aeruginosa the Opportunist: Pathogenesis and Disease. Boca Raton: CRC Press. p 159-176.

Kernacki KA, Goebel DJ, Poosch MS, Hazlett LD. 1998. Early cytokine and chemokine gene expression during Pseudomonas aeruginosa corneal infection in mice. Infect Immun 66:376-379.

Kreger AS. 1983. Pathogenesis of Pseudomonas aeruginosa ocular diseases. Rev Infect Dis 5:S931-S935.

Lemp MA. 1990. Is the dry eye contact lens wearer at risk'? Yes. Cornea 9:S48-S50.

Levine M, Serdarevic O. 1998. Gram positive organisms and ciprofloxacin in contact lens induced microbial keratitis. (Abstract). Invest Ophthalmol Vis Sci 39:S337.

Naquet R. 1993. Ethical and moral considerations in the design of experiments. Neuroscience 57:183-189.

Olsen BR, McCarthy MT. 1994. Molecular structure of the sclera, cornea and vitreous body. In: Albert DM, Jakobiec FA, editors. Principles and Practice of Ophthalmology. Philadelphia: WB Saunders. p 38-63.

Parker J. 1994. Help from philosophy: Responding to the animal rights challenge. FASEB J 8:375-377.

Preston M J, Fleiszig SMJ, Zaidi TS, Goldberg JB, Shortridge VD, Vasil ML, Pier GB. 1995. Rapid and sensitive method for evaluating Pseudomonas aeruginosa virulence factors during corneal infections in mice. Infect Immun 63:3497-3501.

Russell WMS, Burch RL, editors. 1959. The Principles of Humane Experimental Technique. London: Methuen.

Schein OD, Glynn RJ, Poggio EC, Seddon JM, Kenyon KR, and the Microbial Keratitis Study Group. 1989. The relative risk of ulcerative keratitis among users of daily-wear and extended-wear soft contact lenses. A case-control study. N Engl J Med 321:773-778.

Smolin G. 1987. The role of tears in the prevention of infections, lnt Ophthalmol Clin 27:25-26.

Twining SS, Zhou X, Schulte DP, Wilson PM, Fish B, Moulder J. 1986. Effect of vitamin A deficiency on the early response to experimental Pseudomonas keratitis. Invest Ophthalmol Vis Sci 37:511-522.

van Klink F, Taylor WM, Alizadeh H, Jager M J, van Rooijen N, Niederkorn JY. 1996. The role of macrophages in Acanthamoeba keratitis. Invest Ophthalmol Vis Sci 37:1271-1281.

Wilhelmus KR. 1996. Bacterial keratitis. In: Pepose JS, Holland GN, Wilhelmus KR, editors. Ocular Infection and Immunity. St Louis: Mosby. p 970-1032.

Wilson LA. 1996. Biomaterials and ocular infection. In: Pepose JS, Holland GN, Wilhelmus KR, editors. Ocular Infection and Immunity. St Louis: Mosby. p 215-231.

Table 1 Modified scoring system for bacterial corneal infection

	Grade
	0	1	2	3	4
Area	None	1-25%	26-50%	51-75%	76-100%
Density of opacity	Clear	Slight cloudiness, details of pupil and iris discernible	Cloudy, but outline of iris and pupil remains visible	Cloudy, opacity not uniform	Uniform opacity
Surface regularity	Smooth	Slight surface irregularity	Rough surface, some swelling	Significant swelling, crater or descemetocele formation	Perforation or seroius descemetocele

Table 2 Involvement of ExsA in pathology caused by an invasive strain of P. aeruginosa (PAO1)

No. of days after infection	Median score (lower:upper quartile)
	PAO1 (wild type)	PAO1 exsA::W
1	3 (3:3)	3 (3:3)
2	3 (3:3)	2 (0.5:2.75)
7	3 (3:3)	0^a (0:1.5)

^aSignificantly different from wild-type strain (p<0.05).

Table 3 Detection time of exsA-mutation effects using the modified scoring system

			Median score (lower:upper quartile)
		PAO1 (wild type)	PAO1 exsA::W
1	Area of major opacity	3 (3:3)	1 (1:2.5)
	Density of opacity	3 (3:3)	2^a (1.25:2)
	Density of surrounding area	2 (1.25:2)	1 (0.25:1)
	Surface regularity	2 (2:2)	1^a (1:1)
2	Area of major opacity	2 (2:2)	2 (0.5:2.75)
	Density of opacity	4 (4:4)	1^a (0.25:1.75)
	Density of surrounding area	2 (2:2)	0^a (0:0.75)
	Surface regularity	2 (2:2)	0^a (0:0.75)
7	Area of major opacity	3 (3:3)	0^a (0:1.5)
	Density of opacity	4 (4:4)	0^a (0:0.75)
	Density of surrounding area	1 (1:1)	0^a (0:0)
	Surface regularity	3 (3:3)	0^a (0:0)

^aSignificantly different from wild-type strain (p<0.05).

Figure 1 Structure of the cornea.

Figure 2 Representative mouse eyes infected with (a) PAO1 (an invasive strain of P. aeruginosa) and (b) PAO1exsA::W (mutant, lacking the transcriptional regulator ExsA) 24 hr after infection.