This article describes a step-by-step protocol to set up an ex vivo porcine model of bacterial keratitis. Pseudomonas aeruginosa is used as a prototypic organism. This innovative model mimics in vivo infection as bacterial proliferation is dependent on the ability of the bacterium to damage corneal tissue.
When developing novel antimicrobials, the success of animal trials is dependent on accurate extrapolation of antimicrobial efficacy from in vitro tests to animal infections in vivo. The existing in vitro tests typically overestimate antimicrobial efficacy as the presence of host tissue as a diffusion barrier is not accounted for. To overcome this bottleneck, we have developed an ex vivo porcine corneal model of bacterial keratitis using Pseudomonas aeruginosa as a prototypic organism. This article describes the preparation of the porcine cornea and protocol for establishment of the infection. Bespoke glass molds enable straightforward setup of the cornea for infection studies. The model mimics in vivo infection as bacterial proliferation is dependent on the ability of the bacterium to damage corneal tissue. Establishment of infection is verified as an increase in the number of colony forming units assessed via viable plate counts. The results demonstrate that infection can be established in a highly reproducible fashion in the ex vivo corneas using the method described here. The model can be extended in the future to mimic keratitis caused by microorganisms other than P. aeruginosa. The ultimate aim of the model is to investigate the effect of antimicrobial chemotherapy on the progress of bacterial infection in a scenario more representative of in vivo infections. In so doing, the model described here will reduce the use of animals for testing, improve success rates in clinical trials and ultimately enable rapid translation of novel antimicrobials to the clinic.
Corneal infections are important causes of blindness and occur in epidemic proportions in low- and mid-income countries. The etiology of the disease varies from region to region but bacteria account for a large majority of these cases. Pseudomonas aeruginosa is an important pathogen that causes a rapidly progressive disease. In many cases, patients are left with stromal scarring, irregular astigmatism, require transplant or in the worst case scenario, lose an eye1,2.
Bacterial keratitis caused by P. aeruginosa is a difficult eye infection to treat particularly due to the increasing emergence of antimicrobial resistant strains of P. aeruginosa. Within the last decade, it has become apparent that testing and developing new treatments for corneal infections, in general, and those caused by Pseudomonas sp., in particular, are essential to combat the current trend in antibiotic resistance3.
For testing the efficacy of new treatments for corneal infections, conventional in vitro microbiological methods are a poor surrogate due to the difference in bacterial physiology during laboratory culture and during infections in vivo as well as due to the lack of the host interface4,5. In vivo animal models, however, are expensive, time-consuming, can only deliver a small number of replicates and raise concerns about animal welfare.
In this article, we demonstrate a simple and reproducible organotypic ex vivo porcine model of keratitis that can be used to test various treatments for acute and chronic infections. We have used P. aeruginosa for this experiment but the model also works well with other bacteria, and organisms such as fungi and yeast which cause keratitis.
Albino laboratory rabbits were sacrificed in the laboratory for other planned experimental work under home office approved protocols. The eyes were not required for experimental use in those studies so they were used for this protocol.
1. Sterilization
2. Sample collection
3. Preparation of the corneoscleral button
4. Maintenance of the corneoscleral buttons
5. Preparation of an inoculum
6. Infecting the corneoscleral button
7. Homogenization of the cornea to harvest the bacteria
The design of the glass molds are an innovative and original idea, the use of which allowed us to set up the model in a consistent fashion with minimal/no issues with contamination. The molds were prepared by a glass blower at the University of Sheffield based on a design (Figure 1A). The experimental setup maintains the convex shape of the cornea and holds bacteria on the top of the epithelium where infection takes place (Figure 1B).
Porcine corneas usually swell after few days in medium. This is normal and we found that there was no significant difference between corneas with and without addition of dextran, which is usually added to prevent swelling of the cornea (Figure 1H). The corneas are typically wounded to help the bacteria penetrate the epithelium. Although there was no significant difference in the progress of infection between wounded (cut) and unwounded (uncut) corneas, we noticed more variations between replicates in uncut corneas (Figure 1C). Washing the corneas twice with PBS removes excess bacteria that did not attach to the epithelium. There was a significant difference in CFU between washed and unwashed porcine corneas infected with P. aeruginosa PAO1 for 24 hours (Figure 1D). There was no significant difference in CFU counts between porcine and rabbit corneas infected with PA14 and PAO1 (Figure 1E,1F). The results for both models were reproducible. After 24 hours, the cornea infected with either Pseudomonas strain always develop opacity and the cut area becomes more visible and open in comparison to the uninfected cornea (Figure 1G).
Figure 1: Ex vivo cornea infected with Pseudomonas aeruginosa. (A) Schematic picture of a glass mold used for maintaining the shape of the cornea and facilitating the introduction of bacteria and treatments. The thickness of the glass molds is 1.5 mm and is the same as the thickness of test tubes made from borosilicate glass. (B) Schematic picture of the experimental set up. (C) Testing the effect of wounding on the final CFU count after homogenization. Uncut (n = 16) and cut (n = 28) corneas were infected with P. aeruginosa PAO1 and P. aeruginosa PA14 for 24 hours. The corneas were washed with 1 mL of PBS before homogenization. Error bars indicate standard deviation. (D) Testing the effect of washing corneas with 2 x 1 mL of PBS (n = 6) and not washing (n = 6) on the final CFU count after infection with P. aeruginosa PAO1 for 24 hours. Error bars indicate standard deviation. (E) Final CFU count in porcine corneas infected with P. aeruginosa PAO1 and P. aeruginosa PA14 for 24 hours (n = 10). Corneas were washed and cut. Error bars indicate standard deviation. (F) Final CFU count in rabbit corneas infected with P. aeruginosa PAO1 and P. aeruginosa PA14 for 24 hours (n = 6). Corneas were washed and cut. Error bars indicate standard deviation. (G) Pictures of ex vivo porcine corneas infected with P. aeruginosa PAO1 for 24 hours. The control was wounded but no bacteria were added. The infected corneas were wounded and 107 CFU were added to the cut side. No CFU were recovered from the control cornea. (H) Final CFU recovered after 24 hours of infection with P. aeruginosa PAO1 from corneas treated with dextran (n = 2) and those without dextran (n = 9). Corneas were washed and cut. Error bars indicate standard deviation. Please click here to view a larger version of this figure.
The main driver behind the development of this keratitis model using ex vivo porcine cornea is to provide researchers developing novel antimicrobials with a representative in vitro model to more accurately determine antimicrobial efficacy at the preclinical stages. This will provide researchers involved in developing new antimicrobials greater control over drug design and formulation at the pre-clinical stages, increase success at clinical trials, reduce use of animals by enabling targeted studies and result in faster translation of new antimicrobials to clinic.
A number of studies have investigated the effect of infections on ex vivo corneas from various animals such as: rabbit6, dog7, goat8 and pigs9,10,11. Most of these studies focus on ways of establishing6 and visualizing an infection9 but so far there have only been a few publications focusing on drug testing and accurate quantification of bacteria6,7,8,12.
The primary advantage of our model is the availability of the porcine corneas as part of the food chain. The use of ex vivo porcine corneas therefore aligns with the principle of 3Rs, which is to replace, refine and reduce the use of animals in research, whilst providing a representative model of the host interface. We have observed no issues with contamination of the corneal explants if the protocol is strictly followed. The glass molds are very easy, quick and straightforward to use without any requirement for specialized equipment. The narrow ring at the top makes the addition of a small quantity of a tested drug (100 µL) or bacteria convenient. The ring of the glass mold allows PBS with bacteria or a drug solution to be retained in the central part of the cornea and prevents the bacteria from getting underneath the cornea. The ring is easy to clean and sterilize, and allows the observation of the changes that occur on the top of the cornea during infection. Strains of fluorescently-tagged bacteria can be used to visualize infection or quantify the spread of infection in the tissue using fluorescent confocal microscopy. The whole corneas can be further processed for histology or electron microscopy imaging.
The critical steps are marked in the protocol. Extra attention must be paid to these steps when carrying out the protocol to ensure successful infection. The most critical steps within the protocol are ensuring that the corneas are treated with sufficient antibiotics to prevent infection during preparation and then that the antibiotics are sufficiently eliminated before the introduction of the infective organism, in this case P. aeruginosa. When setting up the experiments using this protocol, in some instances, turbidity developed during incubation in the antibiotic-free medium. This turbidity was indicative of growth of microorganisms in the antibiotic-free medium. This might be due to incomplete treatment of the cornea using the antibiotics or due to contamination during handling. These corneas were not taken forward for further experiments and were discarded. Development of turbidity when incubating corneas in antibiotic-free medium was avoided by employing frequent sterilization runs in the incubator, using disposable pipette tips with a filter and taking adequate care when sterilizing the tools used for excising the cornea from the porcine eyes. Another critical step is when the corneas are placed in the glass mold prior to infection. The glass mold enables one to maintain the convex shape of the cornea. The convexity of the cornea is a challenge for retention of either the infective dose or the therapeutic agent on the surface of the cornea. Therefore, it is essential to ensure the presence of adequate seal between the cornea and the glass mold. When there is adequate seal between the cornea and the glass mold, the ring structure above the mold creates a reservoir to retain either the infective dose or the therapeutic agent. An adequate seal is ensured by completely filling the wide section of the glass mold with DMEM agar up to the brim.
As is the case with any model, there are limitations associated with the ex vivo porcine cornea model described. The model described herein does not mimic the composition, flow and replenishment of the tear film across the cornea. The mechanical action provided by blinking is also not incorporated into the model. There is agreement in the literature that tear film composition and dynamics, and blinking are important defense mechanisms that remove foreign particles and microorganisms from the eye13. Indeed, the model also lacks an immune response that is triggered during infection in vivo. It is likely that the progression of infection in vivo in the presence of these defense mechanisms is different to that observed in the ex vivo model described here. Despite these limitations, the ex vivo porcine corneal model is relevant for testing the effectiveness of existing and emerging antimicrobials for two main reasons: 1) the physiology of the bacteria in the ex vivo model mimics the in vivo conditions as bacterial proliferation is dependent on their ability to damage the corneal tissue, and 2) the model incorporates the three dimensional tissue as a diffusion barrier for therapeutics much like in the in vivo situation. Therefore, the ex vivo model is advantageous over conventional techniques for antimicrobial susceptibility testing.
The ex vivo porcine cornea model described here can be also used for studying different strains of bacteria, fungi and yeast that cause keratitis. This ex vivo cornea model is reproducible and allows one to generate replicates within a short time unlike in vivo models. Instead of PBS, artificial tears or host immune defense cells can theoretically be added to mimic the live scenario. Corneas are obtained from the same breed of pigs and about 21-23 weeks old when slaughtered. Therefore, there is less variability between replicates compared to those obtained from human cadavers. The concept of using a porcine ex vivo cornea model for biomedical applications has gained more popularity within the last few years because of its biological similarity to the human eye which makes this model easier to compare14. There is increased interest in using porcine corneas for transplantation15,16 or as a model for dry eye17 or wound healing18.
The authors have nothing to disclose.
The authors would like to thank Elliot Abattoir in Chesterfield for providing porcine eyes. The glass rings were made based on our design by the glass blower Dan Jackson from the Department of Chemistry at the University of Sheffield. The authors would like to thank the Medical Research Council (MR/S004688/1) for funding. The authors would like to also thank Mrs Shanali Dikwella for technical help with cornea preparation. The authors would like to thank Mr Jonathan Emery for help with formatting pictures.
50 mL Falcon tube | SLS | 352070 | |
Amphotericin B | Sigma | A2942 | |
Cellstar 12 well plate | Greiner Bio-One | 665180 | |
Dextran | Sigma | 31425-100mg-F | |
Distel | Fisher Scientific | 12899357 | |
DMEM + glutamax | SLS | D0819 | |
Dual Oven Incubator | SLS | OVe1020 | Sterilising oven |
Epidermal growth factor | SLS | E5036-200UG | |
F12 HAM | Sigma | N4888 | |
Foetal calf serum | Labtech International | CA-115/500 | |
Forceps | Fisher Scientific | 15307805 | |
Handheld homogeniser 220 | Fisher Scientific | 15575809 | Homogeniser |
Heracell VIOS 160i | Thermo Scientific | 15373212 | Tissue culture incubator |
Heraeus Megafuge 16R | VWR | 521-2242 | Centrifuge |
Insulin, recombinant Human | SLS | 91077C-1G | |
LB agar | Sigma | L2897 | |
Multitron | Infors | Not appplicable | Bacterial incubator |
PBS | SLS | P4417 | |
Penicillin-Streptomycin | SLS | P0781 | |
Petri dish | Fisher Scientific | 12664785 | |
Petri dish 35x10mm CytoOne | Starlab | CC7672-3340 | |
Povidone iodine | Weldricks pharmacy | 2122828 | |
Safe 2020 | Fisher Scientific | 1284804 | Class II microbiology safety cabinet |
Scalpel blade number 15 | Fisher Scientific | O305 | |
Scalpel Swann Morton | Fisher Scientific | 11849002 |