Graefes Arch Clin Exp Ophthalmol

DOI 10.1007/s00417-016-3546-0

BASIC SCIENCE

Ex vivo rabbit and human corneas as models for bacterial

and fungal keratitis
Abigail Pinnock 1 & Nagaveni Shivshetty 2 & Sanhita Roy 2 & Stephen Rimmer 3 &
Ian Douglas 1 & Sheila MacNeil 1,4 & Prashant Garg 2

Received: 8 August 2016 / Revised: 23 October 2016 / Accepted: 31 October 2016

# The Author(s) 2016. This article is published with open access at Springerlink.com

Abstract majority of organisms from infected corneas. Histology of

Purpose In the study of microbial keratitis, in vivo animal the scalpel-wounded and injection models indicated extensive
models often require a large number of animals, and in vitro infiltration of P. aeruginosa throughout the entire cornea, with
monolayer cell culture does not maintain the three- less infiltration observed for S. aureus, C. albicans and
dimensional structure of the tissues or cell-to-cell communi- F. solani. The models also supported dual infections.
cation of in vivo models. Here, we propose reproducible Conclusions Both scalpel wounding and injection methods
ex vivo models of single- and dual-infection keratitis as an are suitable for inducing infection of ex vivo rabbit and human
alternative to in vivo and in vitro models. cornea models. These simple and reproducible models will be
Methods Excised rabbit and human corneoscleral rims main- useful as an alternative to in vitro and in vivo models for
tained in organ culture were infected using 108 cells of investigating the detection and treatment of microbial kerati-
Staphylococcus aureus, Pseudomonas aeruginosa, Candida tis, particularly when this might be due to two infective
albicans or Fusarium solani. The infection was introduced organisms.
by wounding with a scalpel and exposing corneas to the mi-
crobial suspension or by intrastromal injection. Post-inocula- Keywords Ex vivo cornea . Microbial keratitis .
tion, corneas were maintained for 24 and 48 h at 37 °C. After Colony-forming units . Corneal model
incubation, corneas were either homogenised to determine
colony-forming units (CFU)/cornea or processed for histolog-
ical examination using routine staining methods. Single- and Introduction
mixed-species infections were compared.
Results We observed a significant increase in CFU after 48 h Microbial keratitis is a major problem worldwide and is an
compared to 24 h with S. aureus and P. aeruginosa. However, important cause of vision loss and blindness. In vivo animal
no such increase was observed in corneas infected with models, in vitro cell culture and ex vivo models have been
C. albicans or F. solani. The injection method yielded an used for investigating different aspects of this disease, includ-
approximately two- to 100-fold increase (p < 0.05) in the ing pathogenicity and treatment strategies [1–7].
In vivo studies require the use of a large number of animals
to answer a research question. The welfare of these animals
* Sheila MacNeil has become an important ethical issue [8], leading to the pro-
s.macneil@sheffield.ac.uk motion of the philosophy of ‘replacement, reduction and re-
finement’ in the use of animals in research [9].
1
One way to overcome these issues is to use in vitro mono-
University of Sheffield, Sheffield S10 2TA, UK
layer cultures of cells, including immortalised [5] or primary
2
LV Prasad Eye Institute, Banjara Hills, Hyderabad 500034, India [4] corneal epithelial cells. However, these are not representa-
3
University of Bradford, Bradford BD7 1DP, UK tive of the in vivo situation. They lack a three-dimensional
4
The Kroto Research Institute, North Campus, University of (3D) structure and cross-talk between different epithelial cells,
Sheffield, Broad Lane, Sheffield S3 7HQ, UK limbal cells and keratocytes. Consequently, advances have
 Graefes Arch Clin Exp Ophthalmol

been made in 3D multi-layered tissue-engineered corneal con- (Mumbai, India). All other reagents were obtained from
structs, and a few of these (EpiOcular™ from Mattek Sigma-Aldrich (Dorset, UK) unless otherwise stated.
Corporation, Ashland, MA, USA, and HCE/corneal epitheli- Calcofluor-white was obtained from Sigma-Aldrich (Dorset,
um (SkinEthic) from Episkin, Lyon, France) are commercially UK) and from HiMedia (Mumbai, India).
available [10]. These have been used to study corneal patho-
genesis [6, 11]. Whilst they possess the 3D architecture of Isolation of rabbit corneas
their in vivo counterparts, these models often use
immortalised cell lines and lack intrinsic innate immune mol- Corneas with sclera rims were dissected using a standard pro-
ecules which occur in vivo. cedure including decontamination with povidone iodine, and
Recently, there has been some interest in the use of ex vivo were immediately placed into phosphate-buffered saline
corneal models to study keratitis [12, 13]. Although these (PBS) [15].
models lack immune elements, the 3D architecture remains,
as do the intracellular innate immune molecules and cellular– Ex vivo corneal organ culture
stromal components. These models have been used for study-
ing wound healing [14], microbial adherence [12] and molec- Organ cultures were as previously described [15, 18]. Rabbit
ular microbial pathogenicity [13]. In our laboratory, we have and human corneoscleral buttons were placed epithelial side
used ex-vivo corneal models to study corneal epithelial regen- down in 35-mm petri dishes, and 500 μl Dulbecco’s modified
eration [15, 16] and to develop models of inflammation. We eagle’s medium (DMEM)–agarose (0.5 % w/v) solution was
have also shown that by gently rocking media over the cor- pipetted into the endothelial side of the cornea. The solution
neas, they can be maintained in culture for at least 4 weeks was allowed to solidify, and the buttons were then inverted so
[17]. that the epithelium was facing up (Fig. 1a and b). Culture
To the best of our knowledge, a comparison of bacterial medium (DMEM: Ham’s F12 [1:1] supplemented with 10 %
and fungal infections and of mixed infections has not been fetal calf serum [FCS], 100 U ml−1 penicillin and 100 U ml−1
undertaken in ex vivo corneal models. We report a comparison streptomycin, 2.5 μg ml−1 amphotericin B, 5 μg ml−1 insulin
of single- and mixed-species infections in both rabbit and and 10 ng ml−1 epidermal growth factor [EGF]) was added to
human corneas to better understand the use of these models submerge the ex vivo corneas. Prior to infection, corneas were
in microbial keratitis. washed three times with PBS and incubated in antibiotic- and
antifungal-free medium for at least 24 h to remove residual
antimicrobials.
Materials and methods All experimental work was performed on rabbit corneas in
the UK and on human corneas in India.
Materials
Culture of bacteria and fungi
We used corneas from two types of rabbits—wild brown rab-
bits (Blackface Meat Company, Dumfries, Scotland) and New For rabbit corneas, laboratory strains of S. aureus (S-235),
Zealand rabbits (University of Sheffield, from rabbits P. aeruginosa (SOM-1), C. albicans (SC5314) and F. solani
sacrificed at the end of a licenced study). There was no differ- strain (NCPF 2699), purchased from the National Collection
ence in the performance of corneas from these two types of of Pathogenic Fungi (UK), were used. For human corneas,
rabbits. Cadaveric human corneas unsuitable for transplant ATCC cultures of S. aureus (25923), P. aeruginosa (27853)
were acquired from the Ramayamma International Eye and C. albicans (90028) were used. All bacterial and fungal
Bank, L V Prasad Eye Institute, Hyderabad, India. All corneas strains were cultured on brain-heart infusion (BHI) agar at
were obtained following procedures approved by the institu- 37 °C overnight and then maintained at 4 °C. For use in ex-
tional review board for the protection of human subjects. periments, one colony was sub-cultured from agar into BHI
Dispase II was obtained from Roche Diagnostics (Burgess broth and incubated overnight at 37 °C. Stationary-phase mi-
Hill, UK), and Videne® antiseptic solution was purchased crobes were used in rabbit cornea experiments. For human
from Ecolab (St. Paul, MN, USA). Mouse 3T3 fibroblasts corneal experiments, on the day of corneal inoculation, a fresh
(used in India) were from the American Type Culture broth was inoculated, and exponential-phase bacteria/fungi
Collection (ATCC; Manassas, VA, USA), and those used in were used based on predetermined growth curves.
the UK were an established J2 3T3 cell line originally from
Professor Howard Green, USA. Epidermal growth factor was Infection of ex vivo corneas
obtained from Invitrogen (Paisley, UK). For the culture of
microorganisms, brain-heart infusion (BHI) agar and broth Corneas were wounded with a scalpel (3 slashes vertically and
were purchased from Oxoid (Hampshire, UK) or HiMedia 3 slashes horizontally), and a metal ring was placed on the
Graefes Arch Clin Exp Ophthalmol

Fig. 1 Schematic representation a b Cornea

of a cross section (a) and a top- Cornea
down image (b) of a corneoscleral
button in organ culture. To infect Culture
Culture medium
corneas, a metal ring was placed
medium
on the corneoscleral button after Sclera
wounding to form a seal, and
bacteria/fungi were added to the Sclera Petri-
surface of the cornea (c) Petri-dish
dish

Agarose-culture medium
c

corneoscleral button, creating a watertight seal. Into the centre Results

of the ring, 108 S. aureus, P. aeruginosa, C. albicans or
F. solani were added (Fig. 1c), or the corneas were injected Macroscopic view of rabbit and human corneas
intrastromally (using a 26-gauge needle; Becton Dickinson,
Oxford, UK) with the same number of organisms. Corneas infected with bacteria and fungi showed a visible
The infected corneas were incubated for 24 or 48 h at increase in haze compared with uninfected corneas (Fig. 2).
37 °C, and were then homogenised and the resulting suspen- The scratches were visible in all corneas, but were more evi-
sion serially diluted and spotted onto agar plates for colony dent in infected corneas than uninfected corneas (Fig. 2).
enumeration. A set of infected corneas was also processed for FITC-labelled bacteria within infected rabbit and human
histology and sections stained using Gram (bacteria) and pe- corneas are shown in Fig. 3. It was observed that S. aureus
riodic acid–Schiff (PAS) stains (fungi). Corneas not exposed cells covered the surface of the cornea at 24 h and 48 h, and at
to microbes were used as controls. Histological sections were certain locations, clumps of bacteria ranging from 5 to 25 μm
imaged using a BX51 upright microscope and cell3D imaging in diameter were detected. On the other hand P. aeruginosa-
software (Olympus, Essex, UK) in the UK or the ProgRes infected corneas showed fewer clumps than observed with
CapturePro 2.5 software (Jenoptik) in India. S. aureus, after 24 and 48 h. A microscopic view of the surface
of calcofluor white-stained C. albicans-infected corneas
showed a more uniform spread of yeast and a few hyphal
Imaging of microorganisms on the corneal surface forms on the surface of the cornea after 24 h, which increased
in both the distribution of individual yeast cells and the spread
To visualise bacteria, 108 S. aureus or P. aeruginosa were of hyphae by 48 h (Fig. 3). The distribution of F. solani at 24 h
labelled using 1 mg ml−1 fluorescein isothiocyanate (FITC) was more punctuated with hyphae at distinct places in both
for 1 h at 4 °C, followed by four washes with PBS. For fungi, rabbit and human corneas (Fig. 3). After 48 h, the surface of
whole corneas were covered with 1:1 calcofluor white in 10 % the cornea was covered with a mat of fungi, where the hyphae
(v/v) potassium hydroxide for 10 min and washed three times could be observed extending into the scratch and in all direc-
with PBS. Bacteria- and fungi-infected corneas were imaged tions, away from the fungal bulk (Fig. 3). The coverage of
using fluorescence microscopy as described above. bacteria and fungi over the corneal surface was similar be-
tween rabbit and human corneas.
Statistical analysis
Single-species infection of rabbit and human corneas
Box-and-whisker plots of colony-forming units (CFU) per
cornea were plotted using GraphPad Prism 6 software. All After 24 h, CFU recovered per cornea for S. aureus,
comparisons were analysed using Student’s unpaired two- P. aeruginosa, C. albicans and F. solani were as follows
tailed t test, using Microsoft® Excel (Microsoft® Office, (Fig. 4): 5.1 ± 1.0 × 105, 1.9 ± 0.3 × 107, 3.0 ± 0.6 × 105 and
2010). A p value ≤ 0.05 was considered significant. 2.5 ± 0.9 × 105 CFU/rabbit cornea, respectively, and 3.8 ±
 Graefes Arch Clin Exp Ophthalmol

Fig. 2 Fluorescein-stained rabbit S.aureus P.aeruginosa C.albicans F.solani Control

and human corneas showing
turbidity of infected versus non-
infected corneas. Corneas were
scalpel-wounded and exposed to RABBIT
S. aureus, P. aeruginosa,
C. albicans or F. solani for 24 h.
Corneas were briefly washed and
stained with 0.5 mg ml−1 of
fluorescein isothiocyanate for
30 min, washed again and
photographed. Arrows indicate
scalpel wounds HUMAN

0.8 × 106, 4.4 ± 0.6 × 108, 1.9 ± 0.3 × 105 and 1.8 ± 0.1 × 103 and 2.1 ± 0.1 × 103(p = 0.081) CFU/human cornea, respec-
CFU/human cornea, respectively. A significantly higher num- tively. In addition, there was approximately tenfold greater
ber of S. aureus and P. aeruginosa were recovered after 48 h recovery of bacteria from human corneas than from rabbit
of incubation in both rabbit (1.7 ± 0.3 × 106 (p = 0.00005), 4.4 corneas after both 24 and 48 h.
± 0.7 × 107 (p = 0.0009) and human corneas (1.5 ± 0.4 × 107 The injection method involved the introduction of bacteria
(p = 0.0004), 6.5 ± 3.0 × 108 (p = 0.0057), respectively, com- and fungi into the stroma. Compared to the scalpel method,
pared to yields at 24 h. There was no significant difference in after 24 h, injection of a single-species organism resulted in
the recovery of C. albicans or F. solani after 48 h, with 5.1 ± higher CFU/cornea (p < 0.05) for all organisms, with the ex-
1.5 × 105 (p = 0.159) and 1.6 ± 0.7 × 106 (p = 0.090) CFU/ ception of C. albicans in human corneas, where no significant
rabbit cornea, respectively, and 5.3 ± 1.6 × 105(p = 0.108) difference was observed (p = 0.057).

Fig. 3 FITC-labelled S. aureus or a S.aureus P.aeruginosa C.albicans F.solani

P. aeruginosa were incubated
with scalpel wounded rabbit (a)
and human (b) corneas for 24 and
48 h, washed and imaged using a
24h
fluorescent microscope.
Unlabelled C. albicans and
F. solani were incubated with
rabbit and human corneas for 24
and 48 h, washed, the model
stained with Calcofluor White
Scratch
and imaged using a fluorescent
48h
microscope. The distribution of
bacteria and fungi over the
surface of the cornea and located
within the scratch wound can be
observed
b
Scratch
24h

50µm 50µm 20µm

50µm

48h

50µm 50µm 20µm

50µm
Graefes Arch Clin Exp Ophthalmol

(Fig. 5a), which increased after 48 h in both number and depth

of penetration.

Two-species infection of corneas

In the two-species infection model, we were able to recover

both bacterial species from infected rabbit and human corneas
(Table 1). Histological examination confirmed the quantitative
colony count data (Fig. 6). As with the single-species model,
infiltration of P. aeruginosa cells was found throughout the
stroma, including Descemet’s membrane, whereas S. aureus
showed little spread beyond the injection site.
A mixed infection involving C. albicans and P. aeruginosa
is the most commonly observed clinically [19]. In ex vivo
models of this mixed infection, both organisms were recov-
ered after 24 h following scalpel wounding, with
P. aeruginosa showing dominance within the tissue by both
colony counting (4.20 ± 1.6 × 106 and 2.12 ± 0.9 × 108 CFU/
rabbit cornea and 5.15 ± 6.6 × 105 and 1.00 ± 1.2 × 108 CFU/
human cornea for C. albicans and P. aeruginosa, respectively)
and histology (Table 1 and Fig. 6). The level of recovery of
both organisms was approximately the same regardless of the
method of introduction of organisms.

Fig. 4 Single-species infection of ex vivo rabbit (a) and human (b)

corneas. Ex vivo corneas were scratched six times with a scalpel and Discussion
exposed to single-species inoculum of S. aureus, P. aeruginosa,
C. albicans or F. solani for 24 h at 37 °C. The models were then
Ex vivo models have previously been used to study corneal–
homogenised, and the resulting suspension serially diluted and plated
onto agar plates. The number of colony-forming units per cornea were microbial interactions [12, 13]. However, there have been no
plotted for each cornea. The box plot shows the minimum and maximum direct comparisons of single and mixed bacterial and fungal
values depicted by the bars; the upper quartile, median and lower quartile infections or a comparison of the infection of rabbit and hu-
are depicted by the top, middle and bottom horizontal lines, respectively
man corneas. Here, we describe the numbers of viable organ-
isms recovered from rabbit and human corneas after 24 and
48 h, showing histological images from scalpel wounding and
The histology of single-species-infected corneas after 24 h intrastromal injection as ways of introducing organisms to the
is shown in Fig. 5a. Here, vast infiltration of P. aeruginosa can cornea.
be seen covering the epithelium and entire stroma and infil- A variety of methods have been described in the literature
trating the Descemet’s membrane. This was independent of for introducing bacteria or fungi to experimental (in vivo,
the method of inoculation (data not shown). The histology in vitro or ex vivo) corneas. These include the use of
of the S. aureus-infected corneas was characterised by the bacterial/fungal-inoculated contact lenses [20, 21], blotting
concentration of the majority of organisms within the paper and ethylene glycol tetraacetic acid (EGTA) [7], and
scratches (Fig. 5a). The number of bacteria within tissue sec- mechanical removal of the epithelial surface [22, 23].
tions correlated with the CFU/cornea data (Fig. 4). However, the most commonly described methods are corneal
The distribution of C. albicans within the corneal tissue scratch [24, 25] and intrastromal injection [26, 27]. Therefore,
was similar between the human and rabbit corneas. Yeast cells we chose to scratch the corneas with a scalpel six times, so that
and hyphal elements were observed close to the scratch site, the scratch revealed the upper stromal compartment, and also
with no infiltration beyond 150 μm into the stroma (Fig. 5a). to introduce organisms intrastromally using an injection meth-
The number of hyphae and the infiltration of C. albicans cells od. Although the injection method gave a greater yield of
into the stroma after 48 h did not differ from that observed organisms than the scalpel method, we observed that the scal-
after 24 h. pel method mimicked clinical infection in which infection is
The tissue penetration by F. solani after scalpel wounding initiated from an abrasion on the corneal surface [28]. Because
was less than that for C. albicans at 24 h, with infiltration not it was thought that the infiltrative capacity of P. aeruginosa
more than 10 μm into the stroma from the site of inoculation within the tissue might inhibit or prevent the growth and
 Graefes Arch Clin Exp Ophthalmol

a S.aureus P.aeruginosa C.albicans F.solani

RABBIT

HUMAN

b S.aureus P.aeruginosa C.albicans F.solani

RABBIT

HUMAN

Fig. 5 Histology of single-species infection of ex vivo rabbit and human and stained using Gram stain (S. aureus and P. aeruginosa) or PAS stain
corneas. Ex vivo rabbit and human corneas were scratched six times with (C. albicans and F. solani). Gram-positive (purple) cocci (S. aureus),
a scalpel and exposed to a single-species inoculum of S. aureus, Gram-negative (pink) rods (P. aeruginosa), and purple round yeast and
P. aeruginosa, C. albicans or F. solani for 24 h (a) or 48 h (b). Corneas hyphae (C. albicans and F. solani) can be observed at the epithelial
were fixed in 10 % buffered formalin, embedded in paraffin, sectioned surface and within the scratch, and are present in the stroma

Table 1 Numbers of CFU/cornea recovered from multi-bacterial/fungal infections after 24 h

A
RABBIT HUMAN
S. aureus - P. aeruginosa S. aureus P. aeruginosa S. aureus P. aeruginosa
6 7 5
1.88 ± 0.6 × 10 3.09 ± 0.9 × 10 5.25 ± 1.6 × 10 2.43 ± 3.2 × 108
C. albicans - P. aeruginosa C. albicans P. aeruginosa C. albicans P. aeruginosa
4.20 ± 1.6 × 106 2.12 ± 0.9 × 108 5.15 ± 6.6 × 105 1.00 ± 1.2 × 108
B
RABBIT HUMAN
S. aureus - P. aeruginosa S. aureus P. aeruginosa S. aureus P. aeruginosa
9.26 ± 4.2 × 106 5.83 ± 2.4 × 108 2.50 ± 5.7 × 104 3.72 ± 1.08 × 108
C. albicans - P. aeruginosa C. albicans P. aeruginosa C. albicans P. aeruginosa
4.25 ± 1.1 × 106 6.84 ± 0.6 × 108 4.05 ± 9.6 × 105 7.65 ± 1.0 × 108

Two multi-pathogen infections were investigated. These were a mixed S. aureus/P. aeruginosa and a mixed C. albicans/P. aeruginosa infection. A: EX
vivo rabbit and human corneas were wounded with a scalpel and exposed to a mixture of 108 of both organisms for 24 h B: Ex vivo rabbit and human
corneas were intrastromally injected with 108 cells of the first organism at 3–5 distinct locations, and then 108 cells of the second organism were
similarly injected at different sites. The corneas were incubated for 24 h, washed, homogenised, serially diluted and plated onto agar plates, and the CFU/
cornea calculated. Data is expressed in CFU/cornea ± SEM of at least three independent experiments performed in triplicate
Graefes Arch Clin Exp Ophthalmol

Scalpel Injection

P.aeruginosa S.aureus
RABBIT
S.aureus
P.aeruginosa

S.aureus
S.aureus
HUMAN

P.aeruginosa P.aeruginosa

Fig. 6 Histology of rabbit and human ex vivo models showing a mixed P. aeruginosa were added to the surface of the cornea for 24 h (scalpel).
S. aureus and P. aeruginosa infection. At different sites within the same Sections were Gram-stained and imaged to visualise S. aureus and
cornea, ex vivo corneas were intrastromally injected with 108 S. aureus P. aeruginosa within their injection sites at distinct locations within the
and 108 P. aeruginosa and incubated for 24 h (injection). Alternatively, stroma. P. aeruginosa shows widespread infiltration into the tissue,
corneas were wounded with a scalpel, and 108 S. aureus and 108 whereas S. aureus shows less infiltration

propagation of S. aureus and/or C. albicans when introduced enhanced tissue disruption compared to a cna− isogenic
together through a scalpel wound, we compared this method mutant. The cna protein is considered to be a virulence
with intrastromal injection, injecting organisms at separate factor mediating bacterial adherence to the epithelial sur-
sites and thus preventing their interaction. However, the scal- face and the stroma, and neutrophil recruitment to the
pel method did not prevent the recovery of S. aureus or infection site. Of the two strains of S. aureus that we used
C. albicans, suggesting that either method is suitable for es- in this study, the ATCC 25923 strain is known to express
tablishing a mixed infection model. Therefore, we show that this gene [31], which could be the reason for the higher
an infection can be induced in the ex vivo corneas for both level of binding within the scratch than at the surface (it is
single and multiple species using either scalpel wounding or unknown whether this gene is expressed in the local clin-
intrastromal injection. ical strain, S235). According to reports in the literature,
The following aspects of this model need further in vivo corneas infected with S. aureus by intrastromal
discussion: injection returned bacterial counts of approximately
104–107 CFU/cornea [32, 33], depending on the number
1. A large bacterial/fungal inoculum was used, because this of bacteria in the starting inoculum and the length of time
corneal model does not have a blood supply or immune the bacteria were incubated with the eye. These values are
system. Consequently, the damage that the inflammation in line with the recovery we obtained from ex vivo
causes to the local tissue, and which provides additional models, suggesting that our model is representative of
nutrients via the vasculature for the bacteria/fungi, was an in vivo infection in terms of the number of bacteria
not present. recovered.
2. S. aureus was not typically found at the epithelial surface, 3. We did not observe any ulceration or corneal edema. This
but rather within the scratches, commonly in clusters, and is because ex vivo corneas lack inflammatory cells that are
not migrating into surrounding and deeper cornea. This primarily responsible for epithelial ulceration [32], stro-
has been described previously as well [29, 30]. The ob- mal polymorphonuclear neutrophil (PMN) infiltration
served attachment of S. aureus at the stromal surface is [34–36], and ulcer formation [37] seen in clinical
also supported by the observations of Rhem et al. [29], S. aureus infection.
who demonstrated that collagen-binding clinical S. aureus 4. Compared to the S. aureus model, P. aeruginosa-infected
isolates expressing the cna collagen-binding gene showed corneas yielded a greater number of CFU/cornea and were
 Graefes Arch Clin Exp Ophthalmol

seen infiltrating the entire cornea, despite having the same from each cornea, i.e. there was an approximately tenfold
inoculum. This high level of infiltration has been shown increase in the recovery of bacteria from human versus
to be the result of proteolytic bacterial enzymes, including rabbit corneas. This may have been due to the use of
type III secretion system-associated cytotoxins, exoen- different bacterial strains (ATTC strains [human] and lo-
zyme U and exoenzyme S [38, 39], alkaline protease cal clinical strains [rabbit]), intrinsic differences between
and elastase [40], which have been shown to contribute the corneas of the two species, including anatomical and
to corneal erosion [41]. In addition, host proteolytic en- molecular differences such as differences in the
zymes also contribute to corneal ulceration [26]. We ob- Bowman’s membrane [44] arrangement of collagen fibres
served a softening of P. aeruginosa-infected corneas and [53], size, thickness [54], secretion of antimicrobial pep-
increased opacity compared with control corneas, but no tides [55], or surface mucin modifications [56].
ulceration. As mentioned previously, this was due to the Furthermore, the difference in bacterial recovery could
lack of an immune cell component in these ex vivo be due to the use of stationary-phase organisms in rabbit
cultures. experiments and log-phase bacteria in human corneal ex-
5. Previous studies have reported an increase in the recovery periments. However, in comparing these two types of in-
of P. aeruginosa from corneas compared with the initial ocula, we have established that a similar number of bac-
inoculum [42, 43]. However, we did not find this in- teria in either phase still results in a clinically relevant
creased recovery of P. aeruginosa. Although we have no level of infection for both single- and multi-species infec-
definitive explanation for this observation, one possible tion in both corneas, showing comparable histology.
explanation is that only a portion of these organisms ac- 9. The length of time these models were cultured in vitro
tually adhere to the corneal surface and are able to invade/ was short. The acute nature of such an infection limits
colonise. The data presented suggest that 106–107 CFU/ its use to short-term experiments involving, for example,
cornea is the maximum number that can be recovered investigation of treatment strategies [57], innate immune
from an ex vivo cornea, and this maximal amount occurs responses [58], detection of organisms [59] and host–mi-
after 24 h. crobe interactions [18]. These models were not intended
6. In contrast to the bacterial infections, single-species infec- to replicate the clinical outcome of infection that can be
tions with C. albicans and F. solani did not show a sig- observed in vivo, which may develop over several weeks
nificant increase in the recovery of organisms after 48 h when not treated effectively. In these ex vivo models,
versus 24 h, and the numbers of organisms recovered there is an obvious lack of a host immune component,
were lower, despite the same inoculum. This has been and the presence and infiltration of inflammatory cells is
described briefly in the literature, where as many as thought to contribute to the severity of disease [26]. As
109–106 CFU/cornea for C. albicans and 103 CFU/ such, these ex vivo models do not form corneal ulcers as
cornea for F. solani were inoculated into in vivo or typically observed clinically [18, 41, 48]. They also lack
ex vivo murine, rabbit or rat models, with recovery of as tear films, which play a defensive role.
little as 105–103 C. albicans CFU/cornea [12, 44–46], and
103 F. solani CFU/cornea [47], respectively. The reason In summary, we achieved our aim of establishing a repro-
for this is not fully understood. ducible infection of both human and animal corneas. It is
7. From the 24-h histology images presented here, little in- certain that no model system (including animals) is a perfect
filtration of fungi into the corneal tissue is seen, with or- surrogate for the natural human infection. Nonetheless, useful
ganisms remaining predominantly at the surface. data can still be obtained. We show that we can establish a
However, particularly for F. solani, there was vast infiltra- reproducible in vitro bacterial and fungal infection, with the
tion of fungal cells throughout the stroma after 48 h. The final number of recoverable bacteria/fungi comparable to that
histology of infected in vivo cultures mimics histological from natural in vivo experiments. These models are now being
images of clinical infection, with a dense white fungal used in the evaluation of microbial detection systems.
plaque, corneal opacity, corneal infiltration, oedema, ulcer
formation, satellite lesions, corneal neovascularisation Compliance with ethical standards
and hypopyon [47–51]. Yeast forms of C. albicans and
conidia of F. solani are shown to adhere to the stroma, and Human and animal rights and informed consent This article does
after a period of time, hyphae form that penetrate the not contain any studies with human participants or animals performed by
stroma [12, 45, 52] to a depth of approximately 150 μm any of the authors.
[44]. This was observed in our ex vivo rabbit and human
Funding The Wellcome Trust provided financial support in the form of
cornea models.
a grant through the Affordable Healthcare in India Initiative (no.
8. Differences were observed between rabbit and human 0998800/B/12/Z). The sponsor had no role in the design or conduct of
corneas primarily in the number of organisms recovered this research.
Graefes Arch Clin Exp Ophthalmol

Conflict of interest All authors certify that they have no affiliations 13. Hua X, Yuan X, Di Pietro A, Wilhelmus KR (2010) The molecular
with or involvement in any organization or entity with any financial pathogenicity of Fusarium keratitis: a fungal transcriptional regula-
interest (such as honoraria; educational grants; participation in speakers’ tor promotes hyphal penetration of the cornea. Cornea 29:1440–
bureaus; membership, employment, consultancies, stock ownership, or 1444. doi:10.1097/ICO.0b013e3181d8383a
other equity interest; and expert testimony or patent-licensing arrange- 14. Castro-Combs J, Noguera G, Cano M, Yew M, Gehlbach PL,
ments) or non-financial interest (such as personal or professional relation- Palmer J, Behrens A (2008) Corneal wound healing is modulated
ships, affiliations, knowledge or beliefs) in the subject matter or materials by topical application of amniotic fluid in an ex vivo organ culture
discussed in this manuscript. model. Exp Eye Res 87:56–63. doi:10.1016/j.exer.2008.04.010
15. Deshpande P, Notara M, Bullett N, Daniels JT, Haddow DB,
MacNeil S (2009) Development of a surface-modified contact lens
for the transfer of cultured limbal epithelial cells to the cornea for
Open Access This article is distributed under the terms of the Creative ocular surface diseases. Tissue Eng A 15:2889–2902. doi:10.1089
Commons Attribution 4.0 International License (http:// /ten.TEA.2008.0528
creativecommons.org/licenses/by/4.0/), which permits unrestricted use, 16. Ortega Í, Deshpande P, Gill AA, MacNeil S, Claeyssens F (2013)
distribution, and reproduction in any medium, provided you give appro- Development of a microfabricated artificial limbus with
priate credit to the original author(s) and the source, provide a link to the micropockets for cell delivery to the cornea. Biofabrication 5:
Creative Commons license, and indicate if changes were made. 025008. doi:10.1088/1758-5082/5/2/025008
17. Deshpande P, Ortega Í, Sefat F, Sangwan VS, Green N, Claeyssens
F, MacNeil S (2015) Rocking media over ex vivo corneas improves
this model and allows the study of the effect of proinflammatory
References cytokines on wound healing. Invest Ophthalmol Vis Sci 56:1553–
1561. doi:10.1167/iovs.14-15308
1. Chandra J, Pearlman E, Ghannoum M (2014) Animal models to 18. Alekseev O, Tran AH, Azizkhan-Clifford J (2012) Ex vivo
investigate fungal biofilm formation. Methods Mol Biol 1147:141– organotypic corneal model of acute epithelial herpes simplex virus
157. doi:10.1007/978-1-4939-0467-9_10 type I infection. J Vis Exp e3631. doi: 10.3791/3631
2. Evans DJ, Fleiszig SM (2013) Why does the healthy cornea resist 19. Ray M, Nigel LC, Tan AM (2014) Triple infection keratitis. Eye
Pseudomonas aeruginosa infection? Am J Ophthalmol 155:961– Contact Lens 40:123–126. doi:10.1097/icl.0000000000000022
970. doi:10.1016/j.ajo.2013.03.001 20. Lawin-Brussel CA, Refojo MF, Leong FL, Hanninen L, Kenyon
KR (1993) Effect of Pseudomonas aeruginosa concentration in
3. Yang K, Wu M, Li M, Li D, Peng A, Nie X, Sun M, Wang J, Wu Y,
experimental contact lens-related microbial keratitis. Cornea 12:
Deng Q (2014) miR-155 suppresses bacterial clearance in
10–18
Pseudomonas aeruginosa keratitis by targeting Rheb. J Infect Dis
21. Cole N, Hume EB, Vijay AK, Sankaridurg P, Kumar N, Willcox
89–98. doi: 10.1093/infdis/jiu002
MD (2010) In vivo performance of melimine as an antimicrobial
4. Zaidi T, Zaidi T, Yoong P, Pier GB (2013) Staphylococcus aureus
coating for contact lenses in models of CLARE and CLPU. Invest
corneal infections: effect of the Panton-valentine leukocidin (PVL)
Ophthalmol Vis Sci 51:390–395. doi:10.1167/iovs.09-4068
and antibody to PVL on virulence and pathology. Invest
22. Stangogiannis-Druya E, Stangogiannis-Druya C, Naranjo-Tackman
Ophthalmol Vis Sci 54:4430–4438. doi:10.1167/iovs.13-11701
R, Vanzzini V, Villar-Kuri J (2009) Bacterial corneal ulcer treated
5. Kolar SSN, Luca V, Baidouri H, Mannino G, McDermott AM,
with intrastromal antibiotic. Experimental model in vivo. Arch Soc
Mangoni ML (2014) Esculentin-1a (1-21) NH2: a frog skin-
Esp Oftalmol 84:123–132
derived peptide for microbial keratitis. Cell Mol Life Sci 1–11.
23. Sun Y, Hise AG, Kalsow CM, Pearlman E (2006) Staphylococcus
doi: 10.1007/s00018-014-1694-0
aureus-induced corneal inflammation is dependent on Toll-like re-
6. Alarcon I, Evans DJ, Fleiszig SM (2009) The role of twitching ceptor 2 and myeloid differentiation factor 88. Infect Immun 74:
motility in Pseudomonas aeruginosa exit from and translocation 5325–5332. doi:10.1128/iai.00645-06
of corneal epithelial cells. Invest Ophthalmol Vis Sci 50:2237– 24. Blaylock WK, Yue BY, Robin JB (1990) The use of concanavalin A
2244. doi:10.1167/iovs.08-2785 to competitively inhibit Pseudomonas aeruginosa adherence to rab-
7. Alarcon I, Tam C, Mun JJ, LeDue J, Evans DJ, Fleiszig SM (2011) bit corneal epithelium. CLAO J 16:223–227
Factors impacting corneal epithelial barrier function against 25. Kwong MS, Evans DJ, Ni M, Cowell BA, Fleiszig SM (2007)
Pseudomonas aeruginosa traversal. Invest Ophthalmol Vis Sci Human tear fluid protects against Pseudomonas aeruginosa kerati-
52:1368–1377. doi:10.1167/iovs.10-6125 tis in a murine experimental model. Infect Immun 75:2325–2332.
8. Badyal DK, Desai C (2014) Animal use in pharmacology education doi:10.1128/iai.01404-06
and research: the changing scenario. Indian J Pharm 46:257–265. 26. Kessler E, Mondino BJ, Brown SI (1977) The corneal response to
doi:10.4103/0253-7613.132153 Pseudomonas aeruginosa: histopathological and enzymatic charac-
9. Editorial (2010) Reduce, refine, replace. Nat Immunol 11:971. doi: terization. Invest Ophthalmol Vis Sci 16:116–125
10.1038/ni1110-971 27. Barequet IS, Bourla N, Pessach YN, Safrin M, Yankovich D,
10. Shah A, Brugnano J, Sun S, Vase A, Orwin E (2008) The develop- Ohman DE, Rosner M, Kessler E (2012) Staphylolysin is an effec-
ment of a tissue-engineered cornea: biomaterials and culture tive therapeutic agent for Staphylococcus aureus experimental ker-
methods. Pediatr Res 63:535–544. doi:10.1203/PDR.0b013 atitis. Graefes Arch Clin Exp Ophthalmol 250:223–229.
e31816bdf54 doi:10.1007/s00417-011-1822-6
11. Augustin DK, Heimer SR, Tam C, Li WY, Le Due JM, Evans DJ, 28. Deorukhkar S, Katiyar R, Saini S (2012) Epidemiological features
Fleiszig SM (2011) Role of defensins in corneal epithelial barrier and laboratory results of bacterial and fungal keratitis: a five-year
function against Pseudomonas aeruginosa traversal. Infect Immun study at a rural tertiary-care hospital in western Maharashtra, India.
79:595–605. doi:10.1128/iai.00854-10 Singapore Med J 53:264–267
12. Zhou Q, Chen H, Qu M, Wang Q, Yang L, Xie L (2011) 29. Rhem MN, Lech EM, Patti JM, McDevitt D, Hook M, Jones DB,
Development of a novel ex vivo model of corneal fungal adher- Wilhelmus KR (2000) The collagen-binding adhesin is a virulence
ence. Graefes Arch Clin Exp Ophthalmol 249:693–700. factor in Staphylococcus aureus keratitis. Infect Immun 68:3776–
doi:10.1007/s00417-010-1601-9 3779
 Graefes Arch Clin Exp Ophthalmol

30. Hume EB, Dajcs JJ, Moreau JM, Sloop GD, Willcox MD, multidrug-resistant Pseudomonas aeruginosa keratitis. J Infect
O’Callaghan RJ (2001) Staphylococcus corneal virulence in a Chemother 20:343–349. doi:10.1016/j.jiac.2013.10.015
new topical model of infection. Invest Ophthalmol Vis Sci 42: 44. Goldblum D, Frueh BE, Sarra GM, Katsoulis K, Zimmerli S (2005)
2904–2908 Topical caspofungin for treatment of keratitis caused by Candida
31. Almeida LM, de Almeida MZP, Mendonça CL, Mamizuka EM albicans in a rabbit model. Antimicrob Agents Chemother 49:
(2013) Comparative analysis of agr groups and virulence genes 1359–1363. doi:10.1128/aac.49.4.1359-1363.2005
among subclinical and clinical mastitis Staphylococcus aureus iso- 45. Jackson BE, Mitchell BM, Wilhelmus KR (2007) Corneal virulence
lates from sheep flocks of the Northeast of Brazil. Braz J Microbiol of Candida albicans strains deficient in Tup1-regulated genes.
44:493–498. doi:10.1590/S1517-83822013000200026 Invest Ophthalmol Vis Sci 48:2535–2539. doi:10.1167/iovs.06-
32. Oguz H, Ozbilge H, Oguz E, Gurkan T (2005) Effectiveness of 0909
topical taurolidine versus ciprofloxacin, ofloxacin, and fortified 46. Yuan X, Hua X, Wilhelmus KR (2010) The corneal expression of
cefazolin in a rabbit Staphylococcus aureus keratitis model. Curr antimicrobial peptides during experimental fungal keratitis. Curr
Eye Res 30:155–161. doi:10.1080/02713680490908733 Eye Res 35:872–879. doi:10.3109/02713683.2010.495812
33. Barequet IS, Ben Simon GJ, Safrin M, Ohman DE, Kessler E 47. Yavas GF, Ozturk F, Kusbeci T, Cetinkaya Z, Ermis SS, Kiraz N,
(2004) Pseudomonas aeruginosa LasA protease in treatment of Inan UU (2008) Antifungal efficacy of voriconazole, itraconazole
experimental staphylococcal keratitis. Antimicrob Agents and amphotericin b in experimental Fusarium solani keratitis.
Chemother 48:1681–1687 Graefes Arch Clin Exp Ophthalmol 246:275–279. doi:10.1007
34. Hume EB, Cole N, Khan S, Garthwaite LL, Aliwarga Y, Schubert /s00417-007-0687-1
TL, Willcox MD (2005) A Staphylococcus aureus mouse keratitis 48. Das S, Sharma S, Mahapatra S, Sahu SK (2015) Fusarium keratitis
topical infection model: cytokine balance in different strains of at a tertiary eye care centre in India. Int Ophthalmol 35:387–393.
mice. Immunol Cell Biol 83:294–300. doi:10.1111/j.1440- doi:10.1007/s10792-014-9961-5
1711.2005.01326.x 49. Ledbetter EC, Norman ML, Starr JK (2015) In vivo confocal mi-
35. Hsiao CH, Ong SJ, Chuang CC, Ma DH, Huang YC (2015) A croscopy for the detection of canine fungal keratitis and monitoring
comparison of clinical features between community-associated of therapeutic response. Vet Ophthalmol. doi:10.1111/vop.12287
and healthcare-associated methicillin-resistant Staphylococcus
50. Thomas PA, Kaliamurthy J (2013) Mycotic keratitis: epidemiology,
aureus keratitis. J Ophthalmol 2015:923941. doi:10.1155/2015
diagnosis and management. Clin Microbiol Infect 19:210–220.
/923941
doi:10.1111/1469-0691.12126
36. Shetty R, Kaweri L, Nuijts RM, Nagaraja H, Arora V, Kumar RS
51. Zhu JL, Gao XR, Cui HP, Lang LL, Li Q, Liao X (2011)
(2014) Profile of microbial keratitis after corneal collagen cross-
Experimental model of Fusarium solani keratitis in rats. Int J
linking. Biomed Res Int 2014:340509. doi:10.1155/2014/340509
Ophthalmol 4:371–376. doi:10.3980/j.issn.2222-3959.2011.04.09
37. Sueke H, Shankar J, Neal T, Winstanley C, Tuft S, Coates R,
52. Pan X, Wang Y, Zhou Q, Chen P, Xu Y, Chen H, Xie L (2011)
Horsburgh MJ, Kaye S (2013) lukSF-PV in Staphylococcus aureus
Activation of focal adhesion kinase enhances the adhesion of
keratitis isolates and association with clinical outcome. Invest
Fusarium solani to human corneal epithelial cells via the tyrosine-
Ophthalmol Vis Sci 54:3410–3416. doi:10.1167/iovs.12-11276
specific protein kinase signaling pathway. Mol Vis 17:638–646
38. Zhu H, Conibear TC, Bandara R, Aliwarga Y, Stapleton F, Willcox
MD (2006) Type III secretion system-associated toxins, proteases, 53. Ojeda JL, Ventosa JA, Piedra S (2001) The three-dimensional mi-
serotypes, and antibiotic resistance of Pseudomonas aeruginosa croanatomy of the rabbit and human cornea. A chemical and me-
isolates associated with keratitis. Curr Eye Res 31:297–306. chanical microdissection-SEM approach. J Anat 199:567–576
doi:10.1080/02713680500536746 54. Marquart ME (2011) Animal models of bacterial keratitis. J Biomed
39. Fleiszig SM, Wiener-Kronish JP, Miyazaki H, Vallas V, Mostov Biotechnol 2011:680642. doi:10.1155/2011/680642
KE, Kanada D, Sawa T, Yen TS, Frank DW (1997) Pseudomonas 55. Durr UH, Sudheendra US, Ramamoorthy A (2006) LL-37, the only
aeruginosa-mediated cytotoxicity and invasion correlate with dis- human member of the cathelicidin family of antimicrobial peptides.
tinct genotypes at the loci encoding exoenzyme S. Infect Immun 65: Biochim Biophys Acta 1758:1408–1425. doi:10.1016/j.
579–586 bbamem.2006.03.030
40. Lomholt JA, Poulsen K, Kilian M (2001) Epidemic population 56. Royle L, Matthews E, Corfield A, Berry M, Rudd PM, Dwek RA,
structure of Pseudomonas aeruginosa: evidence for a clone that is Carrington SD (2008) Glycan structures of ocular surface mucins in
pathogenic to the eye and that has a distinct combination of viru- man, rabbit and dog display species differences. Glycoconj J 25:
lence factors. Infect Immun 69:6284–6295. doi:10.1128 763–773. doi:10.1007/s10719-008-9136-6
/iai.69.10.6284-6295.2001 57. Sriram S, Gibson DJ, Robinson P, Pi L, Tuli S, Lewin AS, Schultz
41. Thibodeaux BA, Caballero AR, Marquart ME, Tommassen J, G (2014) Assessment of anti-scarring therapies in ex vivo organ
O’Callaghan RJ (2007) Corneal virulence of Pseudomonas cultured rabbit corneas. Exp Eye Res 125:173–182. doi:10.1016/j.
aeruginosa elastase B and alkaline protease produced by exer.2014.06.014
Pseudomonas putida. Curr Eye Res 32:373–386. doi:10.1080 58. Marino A, Pergolizzi S, Lauriano ER, Santoro G, Spataro F, Cimino
/02713680701244181 F, Speciale A, Nostro A, Bisignano G (2015) TLR2 activation in
42. Zaidi TS, Zaidi T, Pier GB, Priebe GP (2012) Topical neutralization corneal stromal cells by Staphylococcus aureus-induced keratitis.
of interleukin-17 during experimental Pseudomonas aeruginosa Acta Pathol Microbiol Immunol Scand 123:163–168. doi:10.1111
corneal infection promotes bacterial clearance and reduces pathol- /apm.12333
ogy. Infect Immun 80:3706–3712. doi:10.1128/iai.00249-12 59. Posch LC, Zhu M, Robertson DM (2014) Multipurpose care
43. Tajima K, Miyake T, Koike N, Hattori T, Kumakura S, Yamaguchi solution-induced corneal surface disruption and Pseudomonas
T, Matsumoto T, Fujita K, Kuroda M, Ito N, Goto H (2014) In vivo aeruginosa internalization in the rabbit corneal epithelium. Invest
challenging of polymyxins and levofloxacin eye drop against Ophthalmol Vis Sci 55:4229–4237. doi:10.1167/iovs.14-14513