Abstract
purpose. To develop a topical inoculation model of Staphylococcus
aureus keratitis in which scarification, contact lenses, and
spermidine are used to inhibit the host defenses and to investigate the
role of α-toxin in this infection.
methods. An α-toxin–positive parent strain (8325-4), its isogenicα
-toxin–negative mutant (DU1090), and a genetically rescued form of
the mutant (DU1090/pDU1212) were bound to rabbit-specific contact
lenses, treated with spermidine (50 mM), and applied to scarified
rabbit corneas. Eyes were treated topically with spermidine before and
after lens application. Eyes were graded for disease by slit lamp
examination (SLE) every 6 hours until 24 hours PI (PI), and erosion
diameters were measured. Histopathologic changes and colony forming
units (CFUs) of bacteria were determined.
results. Spermidine treatment and inoculation of eyes with Staphylococcus on contact lenses resulted in significant
increases in both CFUs per cornea (P = 0.0041) and
SLE score (P ≤ 0.0001), compared with eyes
inoculated without spermidine treatment. The CFUs in eyes infected with
8325-4, DU1090, or DU1090/pDU1212 demonstrated a similar
(P ≥ 0.1959) multilog increase in CFUs over the
inoculum at 24 hours PI. The α-toxin–producing strains, 8325-4 and
DU1090/pDU1212, caused significantly more disease than theα
-toxin–deficient mutant DU1090 at 24 hours PI
(P ≤ 0.0001). Histopathology revealed bacteria in
scarified regions of the corneas and, for 8325-4 and DU1090/pDU1212,
extensive epithelial sloughing and severe inflammation.
conclusions. A new topical model of infection has been developed, and α-toxin is
an important virulence factor in this model.
Infection of the cornea by
Staphylococcus aureus often involves the spread of bacteria from the skin or lid margin to
the tear film and then to the cornea.
1 2 3 The host defense
system of the tear film, which protects the eye from such infections,
has long been recognized to include lysozyme, lactoferrin, complement,
and IgA antibody.
4 5 6 7 8 More recently, Qu and
Lehrer
9 have demonstrated that the human eye is also
protected by the action of phospholipase A
2.
Recent in vitro studies from our laboratory have demonstrated that
phospholipase A
2 in the rabbit eye is a potent
defense against
S. aureus.
10 Host defenses were
so effective that topical inoculation of a scarified cornea with 10
million colony forming units (CFUs) of
S. aureus adhering to
a contact lens failed to cause a productive infection and resulted in
efficient bacterial killing within the tear film.
The killing of
Staphylococcus on contact lenses after
topical corneal application is in contrast to the extensive bacterial
replication and pathologic effect of keratitis that results from the
intrastromal injection of bacteria into the cornea. The injection of
only 100 CFUs into the corneal stroma results in severe keratitis, an
indication that the stroma does not have the host defense systems found
in the tear film. Keratitis resulting from intrastromal injection is
characterized by bacterial replication and severe ocular changes,
including corneal edema, corneal epithelial cell destruction, and
iritis, as well as migration of polymorphonuclear neutrophils (PMNs)
from the eyelid to the tear film.
11 12 13 Genetic,
immunologic, and histopathologic studies have shown that the major
cause of these pathologic events is the action of α-toxin, a lytic
toxin produced by
S. aureus in the late log phase of
growth.
11 14 15
Bacterial infection of the cornea has been hypothesized to occur in a
two-step process.
13 Bacteria first interact with the
surface of corneal epithelial cells and then penetrate into the stroma
where toxins mediate severe inflammation and tissue damage. The
intrastromal model of
Staphylococcus keratitis has been
useful for studies of chemotherapy and pathologic reactions that occur
during the intrastromal stage of infection. However, the intrastromal
model cannot address the events occurring earlier in the infectious
process at the epithelial cell surface. To study these reactions, a
model of keratitis initiated by topical inoculation is needed.
The results of recent studies on the interaction of
Staphylococcus and tears have been applied in the present
study to develop a topical inoculation model of keratitis. Topical
inoculation of the rabbit cornea with
Staphylococcus and the
incubation of bacteria in tears in vitro have been found to quickly
kill large numbers of
Staphylococcus.
10 In
vitro studies have determined that spermidine inhibits the
phospholipase activity of rabbit tears and prevents bacterial killing
by tears in vitro.
10 Accordingly, in the present study
spermidine treatment was used to augment bacterial survival in the tear
film in an effort to produce a topical inoculation model of
Staphylococcus keratitis. In addition, the importance ofα
-toxin in the infectious process that occurs after topical
inoculation of the rabbit cornea was tested.
S. aureus strain 8325-4 is an α-toxin–producing
strain analyzed previously in the rabbit intrastromal injection
keratitis model.
11 16 S. aureus strain DU1090 is
an α-toxin mutant in which the α-toxin gene has been inactivated by
allele replacement.
16 17 S. aureus strain
DU1090/pDU1212 is the α-toxin–negative mutant that contains the
plasmid pDU1212 that codes for α-toxin and restores α-toxin
production to wild-type levels.
17 18 DU1090 and
DU1090/pDU1212 have previously been analyzed in the rabbit intrastromal
injection keratitis model.
11 16
New Zealand White rabbits (2.0–3.0 kg) were maintained in
strict accordance with the institutional guidelines and the ARVO
Statement for the Use of Animals in Ophthalmic and Vision Research. All
rabbits were anesthetized by subcutaneous injection of a 1:5 mixture of
xylazine (100 mg/ml; Rompum; Miles Laboratories, Shawnee, KS) and
ketamine hydrochloride (100 mg/ml; Ketaset; Bristol Laboratories,
Syracuse, NY). Proparacaine hydrochloride (0.5% Alcaine; Alcon
Laboratories, Fort Worth, TX) was topically applied to each eye before
scarification.
Bacteria were grown overnight in tryptic soy broth (TSB; Difco
Laboratories, Detroit, MI) and subcultured 1:100 in TSB to
approximately 108 CFU/ml (optical density at 650
nm = 0.285; midlog phase). Contact lenses (61.4% polymacon,
38.6% water) designed for the rabbit eye with a base curve of 7.5 and
diameter of 14.4 mm (Gelflex Laboratories, Perth, Australia) were
incubated for 1 hour at room temperature in a 1.5-ml suspension of
midlog-phase bacteria diluted to 105 CFU/ml.
After 30 minutes, 1 ml 50 mM spermidine (Sigma-Aldrich, St. Louis, MO)
was added to each lens already soaking in bacterial culture. Contact
lenses (n ≥ 4) used to determine the number of
bacteria bound were selected at random, rinsed in sterile PBS, and
assayed by homogenizing each in sterile PBS (3.0 ml), diluting the
homogenate in PBS, and inoculating onto tryptic soy agar (TSA; Difco).
Control lenses for 8325-4, DU1090, and DU1090/pDU1212 each contained a
similar number of adherent bacteria per contact lens (3.95 ± 0.10
log CFU, 3.94 ± 0.07 log CFU, and 4.15 ± 0.05 log CFU,
respectively; P ≥ 0.0801).
Sixty and 30 minutes before and at the time of application of
contact lenses, 40 μl spermidine (50 mM), dissolved in distilled
water and filter sterilized, was applied to each rabbit eye.
Scarifications, sufficient to penetrate the corneal stroma, were
performed on rabbit corneas using a 22-gauge needle making three
parallel scratches (approximately 8 mm long) in the center of each
cornea. Contact lenses with adherent bacteria were then placed on each
rabbit eye (n ≥ 5 per strain), underneath the
nictitating membrane and remained for the duration of the experiment.
After lens insertion, rabbit eyes were again treated with 40 μl
spermidine (50 mM) at 30, 60, and 90 minutes postinfection (PI).
Rabbit eyes, with lenses in place, were examined for changes by
slit lamp examination (SLE) every 6 hours until 24 hours PI. SLE of
rabbit eyes was performed by biomicroscope (Topcon; Koaku Kikai KK,
Tokyo, Japan) by two masked observers. Each of seven ocular parameters
(injection, chemosis, corneal infiltrate, corneal edema, fibrin in the
anterior chamber, hypopyon formation, and iritis) was graded on a scale
of 0 (none) to 4 (severe), as previously described.
19 The
parameter grades were totaled to produce a single SLE score, ranging
from 0 (normal eye) to a theoretical maximum of 28. Corneal erosions
were detected with fluorescein (Fluor-I-Strip AT; Wyeth-Ayerst
Laboratories Inc., Philadelphia, PA) and erosion diameters measured.
Average diameters were expressed in millimeters.
Rabbits were killed at 24 hours PI by an injection of
pentobarbital sodium (100 mg/ml; The Butler Company, Columbus, OH).
Corneas (
n ≥ 5 per strain) were aseptically removed
and cut in half. One half was used for histopathology, and the other
half was cultured to enumerate viable bacteria. The number of viable
S. aureus per cornea was determined by culturing corneal
homogenates in triplicate, as previously described.
11 16 Contact lenses were also collected, homogenized, and plated on TSA at
24 hours PI. CFUs were expressed as base 10 logarithms. Bacteria
recovered from infected eyes were tested for hemolysin production by
inoculating individual colonies onto rabbit blood agar plates (PML
Microbiologicals, Wilsonville, OR).
One half of each cornea harvested at 24 hours PI was subjected
to histopathologic analyses, as previously described.
15 Briefly, corneas were fixed immediately in 10% formalin (EK
Industries, Joliet, IL). A tissue processor (Hypercenter XP Processor;
Shandon, Pittsburgh, PA) was used to prepare the corneal tissue as
follows: immersion overnight in 10% zinc formalin, dehydration in
alcohol (70%, 80%, 95%, and three changes of absolute alcohol), and
immersion in xylene three times to clear the tissue. Corneal tissues
were embedded in paraffin. The resultant paraffin blocks were then cut
into 4-μm-thick sections with a rotary microtome and stained with
hematoxylin and eosin for pathologic examination.
Data were analyzed by computer (Statistical Analysis System;
SAS, Cary, NC).
20 For CFU determination, analysis of
variance and Student’s
t-tests between least-squares means
from each group showing statistical variances were performed. For SLE
scores, nonparametric one-way analysis of variance (Kruskal-Wallis
test) and the Wilcoxon test were used for comparison among groups.
P ≤ 0.05 was considered significant.
Previous studies from this laboratory have determined that rabbit
tears kill
S. aureus in vitro, and spermidine protects the
bacteria from this killing action.
10 Therefore,
experiments were conducted to determine the ability of spermidine (50
mM) to weaken host defenses and protect
S. aureus 8325-4 on
contact lenses applied to scarified rabbit corneas. Application of
lenses without the use of spermidine resulted in no significant change
in the number of bacteria on the lens from the time of inoculation
(10
4 CFU per lens) through 24 hours
(
P = 0.9355). However, similar inoculations augmented
by the use of spermidine resulted in more than a 2-log increase in CFUs
per lens by 24 hours PI (
P = 0.0027;
Table 1 ). More important, spermidine treatment resulted in an increase
in the number of bacteria in the cornea, an increase that was
significantly greater than the count in the inoculum (
P = 0.0002). Corneas of inoculated eyes not treated with spermidine did
not show a significant increase in CFUs over the inoculum
(
P = 0.4129). The bacteria on spermidine-treated lenses
caused an infection that increased the SLE score significantly over
that produced by bacteria on lenses not treated with spermidine
(
P ≤ 0.0001;
Table 1 ).
Infection of the rabbit cornea by
S. aureus 8325-4 adherent
to contact lenses treated with spermidine was confirmed by
histopathologic analysis
(Figs. 1A 1B) . The corneas at 24 hours after inoculation showed destruction of
the corneal epithelium and bacterial penetration into the stroma. An
association of bacteria and neutrophils was seen within the corneal
fissures created by scarification
(Fig. 1C) . Thus, rabbit corneas
topically inoculated with
S. aureus adherent to contact
lenses and treated with spermidine resulted in keratitis, as measured
by bacterial replication, inflammation, and pathologic changes.
To determine the role of α-toxin in the topical model of infection,
the replication and virulence of the parent 8325-4 strain was compared
with that of the α-toxin–deficient mutant, DU1090, and its rescued
strain, DU1090/pDU1212. The CFUs in the cornea of the parent,α
-toxin–deficient, and rescued strains were not significantly
different at 24 hours PI (6.59 ± 0.13, 6.65 ± 0.16,
6.12 ± 0.25 CFU/ml, respectively;
P ≥ 0.1959).
Each colony recovered from eyes infected with the parent or rescue
strain retained the hemolytic activity of an α-toxin–producing
strain, whereas those colonies recovered from eyes inoculated with the
mutant strain did not have the hemolytic activity characteristic ofα
-toxin. The SLE scores in eyes infected with the rescued strain were
comparable to those of eyes infected with the parent strain
(
P ≥ 0.1123) and scores with both strains were higher
than those of the α-toxin–deficient mutant at all times tested from
6 to 24 hours PI (
P ≤ 0.0081;
Fig. 2 ). Eyes inoculated with the parent or the rescued strains showed severe
ocular inflammation, including conjunctival injection, chemosis,
corneal infiltration, corneal edema, hypopyon formation, fibrin
accumulation in the anterior chamber, and iritis. The changes in the
eyes infected with the α-toxin–deficient mutant were trace to mild
and were limited to conjunctival injection, chemosis, corneal
infiltrate, and iritis. Corneas infected with the parent or rescued
strain had severe epithelial erosions, whereas each cornea infected
with the α-toxin–negative mutant had an intact corneal epithelium
(parent, 9.56 ± 0.87 mm; rescue, 8.40 ± 0.87 mm; mutant,
0.0 ± 0.0 mm;
P ≤ 0.0001). Inflammatory
changes in eyes infected with each strain, as described earlier, were
visible grossly
(Fig. 3) .
Histopathologic analysis of eyes infected with either of theα
-toxin–producing strains demonstrated the loss of the corneal
epithelium and the accumulation of PMN in the tear film within the
fissures created by scarification
(Figs. 4A 4C) . In contrast, eyes infected with the α-toxin–negative mutant
retained the corneal epithelium and demonstrated few PMNs in the tear
film associated with the cornea
(Fig. 4B) .
This investigation has shown that Staphylococcus adherent to rabbit-specific contact lenses, in conjunction with
spermidine treatment, can cause keratitis when applied to scarified
rabbit corneas. Such topical inoculations result in bacterial
replication on lenses and in the cornea, inflammation, and pathologic
changes typically associated with keratitis. The inflammatory and
pathologic changes produced include destruction of the corneal
epithelium, influx of PMNs into the tear film and corneal stroma,
chemosis and injection within the conjunctiva, edema of the cornea,
hypopyon formation, fibrin accumulation in the anterior chamber, and
iritis.
Previous studies have shown that
Staphylococcus on
contact lenses designed for human use are killed in the rabbit eye and
that phospholipase A
2 has a critical role in this
potent defense reaction in the host.
10 Thus, the uses of
the lenses specifically designed for the rabbit eye and spermidine
treatment were important factors in the development of keratitis after
this inoculation procedure. Rabbit-specific contact lenses, unlike
human-specific lenses, conform to the curvature of the rabbit cornea
assuring a closer interaction with the cornea.
Staphylococcus replication after topical inoculation is
apparently fostered in part by the barrier effect of the contact
lenses, which limits tear film flow into fissures created during
scarification. In addition, bacterial replication is aided by the
inhibition of the bactericidal action of phospholipase
A
2 as a result of multiple applications of
spermidine. The lenses may also protect bacteria from some of the PMNs
that reach the tear film from the overlying eyelid or
limbus.
12
The severe inflammatory and pathologic changes associated with this new
topical model of
Staphylococcus keratitis are dependent on
the production of α-toxin during infection. Corneal epithelial
destruction and severe ocular inflammatory responses were present in
eyes infected with the toxin-producing strains, yet absent in eyes
inoculated with the α-toxin–deficient mutant. We have previously
demonstrated that α-toxin causes extensive pathologic damage in an
intrastromal injection model of
Staphylococcus keratitis.
11 15 16 α-Toxin, however, is not essential
for the growth of bacteria within the cornea in the intrastromal
model,
11 and α-toxin was not essential for growth of
bacteria in the cornea after topical inoculation.
The topical model of keratitis is expected to be useful in the
analysis of the early stages of infection. One key question that could
be addressed in this model is the mechanism by which
Staphylococcus enters the corneal stroma. Johnson et
al.
21 found that
S. aureus adhered to rabbit
corneal epithelial cells in vitro; however, this finding was not
demonstrated in vivo. One explanation could be that bacteria replicate
in corneal fissures and some are able to enter the stroma. The fibrous
collagen bundles of the stroma are cut during scarification, and the
spaces created between bundles could provide a route for the bacteria
to diffuse from the fissures to the stroma. The penetration of theα
-toxin–deficient mutant into the stroma suggests that the toxic and
inflammatory effects of α-toxin are not needed for bacterial invasion
of the stroma.
The topical model could also be useful for determining the
ability of antibody to protect against corneal infection. Antibody toα
-toxin has been found to protect the cornea from tissue damage and
inflammation after intrastromal injection.
14 However,
antibody to α-toxin does not interfere with
Staphylococcus replication in the cornea. It is possible that antibody, particularly
IgA class antibody in the tear film, specific for a
Staphylococcus antigen other than α-toxin can prevent
Staphylococcus corneal infection. The topical model could
provide evidence of such protection and a determination of the
immunogen as well as the immunization procedure that best induces a
protective immune state.
This study has shown that Staphylococcus can infect the
scarified rabbit cornea, provided the phospholipase
A2 defense of the tear film is compromised. The
model should allow analysis of the early bacterial infectious process
and of the ability of innate as well as acquired host defenses
(antibodies) to protect against keratitis.
The authors thank Kiana Nelson for technical assistance, Brett
Thibodeaux for help with the figure preparations, and Mary Marquart and
Megan Austin for critical review of the manuscript.