Expression of Collagen I, Smooth Muscle a-Actin, and Vimentin ...

Expression of Collagen I, Smooth Muscle a-Actin, and
Vimentin During the Healing of Alkali-Burned and
Lacerated Corneas
Masamichi Ishizaki, GuangZhu, Takeshi Haseba, ShaunS. Shafer, and Winston W.-Y. Kao
Purposes. Alkali-burned corneas can seldom heal properly to restore corneal transparency. To
provide a better understanding of this devastating corneal injury, we compared the expression
of collagen I, smooth muscle a-actin (a-SMA), and vimentin in lacerated and alkali-burned
rabbit corneas.
Methods. A radiolabeled cDNA probe of a 1(1) chain was used in slot-blot hybridization to
determine the levels of a 1(1) mRNA in alkali-burned corneas. In situ hybridization was used to
identify the cell types that express the a 1(1) chain. Antibodies against collagen I, a-SMA, and
vimentin were used in immunohistochemical studies to determine the tissue distribution of
collagen I and to identify cells expressing a-SMA and vimentin.
Results. The levels of al(I) mRNA in alkali-burned corneas increased steadily after the alkali
burn and reached a plateau within 2 weeks. One day after alkali burn, specific in situ hybridization
signals were detected in stromal cells immediately surrounding the edge of the corneal
injury. As the healing proceeded, the fibroblastic cells migrated into the injured stroma, and
they showed positive reactions by in situ hybridization and by immunostaining with anti-collagen
I probes. In alkali-burned corneas, retrocorneal membranes were formed 1 week after
injury. This fibrillar membrane was stained by anti-collagen I antibody, and the fibroblastic
cells in the membrane were hybridized by the 3 H-labeled al(I) cDNA probe. No retrocorneal
membrane was formed in the lacerated corneas, even after the injured corneas were allowed to
heal for 3 weeks. The epithelial cells in the epithelial plug of lacerated corneas were positive by
in situ hybridization, whereas the epithelial cells in the regenerated epithelium of alkali-burned
cornea was not. Antibodies against a-SMA reacted with the migrating fibroblastic cells but did
not react with epithelial cells or endothelial cells in the injured corneas. Anti-vimentin antibody
reacted with fibroblastic cells, endothelial cells, and keratocytes in normal and injured
corneas, and with the basal epithelial cells of injured corneas.
Conclusions. During wound healing, the keratocytes that migrate to injured stroma transform
into myofibroblasts. These myofibroblasts express high levels of «1(I) mRNA, a-SMA, and
vimentin. The healing of alkali-burned corneas differ from that of lacerated corneas in that the
retrocorneal membranes are formed in the former but not in the latter. In addition, the
epithelial cells of alkali-burned corneas lack al(I) mRNA, whereas it is found in the epithelium
of lacerated corneas. These differences may result from the persistence of inflammatory cells
in the alkali-burned corneas. Invest Ophthalmol Vis Sci. 1993; 34:3320-3328.
Alkali burn causes one of the most severe corneal
injuries. The most serious alkali burns often involve
From the Department of Ophthalmology, University of Cincinnati, Cincinnati, Ohio.
Supported in part by National Eye Institute grant EY05629 and by unrestricted
departmental grants from Research to Prevent Blindness, New York, New York,
and Lions Eye Research Foundation, Columbus, Ohio.
Submitted for publication September 21, 1992; accepted March 22, 1993.
Proprietary interest category: N.
Reprint requests: Winston W.-Y. Kao, Ph.D., Department of Ophthalmology,
University of Cincinnati, Eden and Bethesda Ave, M.L. 527, Cincinnati, OH
45267-0527.
limbal epithelium and other surrounding ocular tissues
(conjunctiva, sclera, and so on) and lead to the
loss of vision from corneal ulceration and perforation
within a short time (1 to 3 weeks). 1 " 5 In contrast, less
severe alkali burns can heal, but the healing is characterized
by the persistence of inflammatory cells in the
injured corneas, recurrent epithelial defects, and neovascularization.
These characteristics interfere with
the proper healing process and result in the formation
of scar tissue, recurrent corneal erosions, or both. 1 ' 2 ' 5 " 8
3320
Investigative Ophthalmology & Visual Science, November 1993, Vol. 34, No. 12
Copyright © Association for Research in Vision and Ophthalmology

Expression of Collagen I in Injured Corneas 3321
The cause or causes of this devastating process is not
well understood, and the treatment of alkali-burned
corneas is usually not satisfactory. Seldom do alkaliburned
corneas heal properly to restore vision.
Collagen constitutes about 80% of the organic
constituents of the corneal stroma. 9 " 10 Collagen I is
the major collagenous component found in stroma, 10
with the well-organized collagen I fibrils contributing
to corneal transparency. 11 However, after an alkali
burn, polymorphonuclear leukocytes infiltrate the injured
corneas, and the proteolytic enzymes, oxidative
derivatives, or both, released by the inflammatory cells
can cause severe loss of the extracellular matrix. 34
Meanwhile, the stromal cells (keratocytes) that survive
the alkali burn may proliferate and synthesize components
of extracellular matrix for the repair of the injured
corneas. Stromal ulceration takes place when
the rate of degradation of extracellular matrix components
(e.g., collagen, proteoglycans) exceeds the rate
of synthesis. Many studies have focused on the degradation
of the extracellular matrix after corneal alkali
burns. 3412 " 14 However, there is little information available
regarding the synthesis of the extracellular matrix
during the healing of the alkali-burned corneas. In
contrast, many investigators have examined the metabolism
of fibrillar collagens during the healing of lacerated
corneas. 15 " 17 For example, increases in the synthesis
of collagen I, III, and V were reported by Cintron,
et al. 1617 We previously reported that stromal cells
play a major role in the healing of lacerated corneas.
The stromal cells not only synthesize collagen I to repair
the injured tissues but actively participate in the
process of remodeling the extracellular matrix by phagocytizing
collagen fibrils during'the healing of lacerated
corneas. 15
It has been suggested that myofibroblasts participate
in the healing of mechanical wounds. Myofibroblasts
that synthesize and secrete collagen I during
wound healing are characterized by the expression of
smooth muscle (a-SMA). 18 " 19 It has been suggested
that myofibroblasts contribute to wound contraction
of mechanical injuries. 1920 Thus, it is of interest to
examine whether myofibroblasts may play a similar
role in the healing of alkali burns.
In the present study, we compared the expression
of collagen I by alkali-burned corneas and lacerated
corneas after allowing them to heal for various periods
of time. Slot-blot hybridization and in situ hybridization
with radiolabeled cDNA of al (I) mRNA was used
to analyze the expression of collagen I. Antibodies
against collagen I, smooth muscle a-actin (a-SMA),
and vimentin were used in immunohistochemical studies
of the injured corneas. Our results indicate that
fibroblastic cells are the major cell type that synthesizes
collagen I in the stroma of alkali-burned and lacerated
corneas. The fibroblastic cells have the characteristics
of myofibroblasts because they express a-
SMA in addition to vimentin. Alkali-burned corneas,
formed a retrocorneal membrane consisting of collagen
I, whereas the lacerated corneas did not form such
a membrane.
MATERIALS AND METHODS
32 P-Labeled dNTP, 7-ATP, and [ 3 H]dCTP were purchased
from DuPont-New England Nuclear (Boston,
MA). Restriction endonucleases were obtained from
New England Biolabs (Beverly, MA). BAS membranes
were purchased from Schleicher and Schuell, Inc.
(Keene, NH). Polyclonal anti-collagen I antibody was
purchased from Southern Biotechnology Associates
(Birmingham, AL). Monoclonal antibodies against a-
SMA, vimentin, and desmin were obtained from Dakopatts,
(Copenhagen, Denmark). Biotinylated second
antibodies and ABC (avidin biotin peroxidase complex)
reagents were purchased from Vector (Burlingame,
CA). All other chemicals and reagents were obtained
from either Sigma (St. Louis, MO) or Fisher
Scientific (Pittsburgh, PA), unless otherwise specified.
Animal Experiments
Adult albino rabbits of either sex weighing 3 to 4 kg
were purchased from Clerco Research Farm (Cincinnati,
OH). Animal experiments were performed in
compliance with the ARVO Resolution on the Use of
Animals in Research. Rabbits were anesthetized with a
combined intramuscular administration of ketamine
(30 mg/kg) and rompun (5 mg/kg). A drop of proparacaine-HCl
(0.5%) was applied directly to the eye.
Only one eye of each experimental rabbit was injured.
The contralateral eyes were discarded because occasionally
pathologic changes were detected in the uninjured
eyes (our unpublished observation). For control
experiments, corneas from untreated animals were
used. To create the alkali burn, a filter paper 8 mm in
diameter soaked in 1 M NaOH was applied to the
center of the cornea for 1 minute and followed by a
rinse with 20 ml of phosphate-buffered saline (PBS)
containing 0.15 M NaCl and 0.01 M sodium phosphate,
pH 7.5. 21 To create a laceration wound, an 8
mm penetrating incision was made in the center of the
cornea with a Micro-Sharp blade (Becton-Dickinson,
Franklin Lakes, NJ). 15 Buprenorphine (0.3 mg/kg) was
administered subcutaneously immediately after injury
and 3 times daily for 3 days.
The injured corneas were allowed to heal for
various periods of time from 1 to 21 days. The rabbits
were killed with an overdose of sodium pentobarbital
intravenously (65 mg/kg). Corneal buttons (10 mm in
diameter) were excised so that each excised cornea
contained a portion of uninjured tissue. 21 The excised
corneas were immediately frozen in liquid nitrogen

3322 Investigative Ophthalmology & Visual Science, November 1993, Vol. 34, No. 12
and stored at — 70°C until use or were subjected to in
situ hybridization as described below.
Extraction of Total RNA
All solutions were autoclaved in the presence of 0.1%
diethylpyrocarbonate (DEPC), and all glassware and
appliances were either baked overnight or soaked in 1
M NaOH for 2 hours. Total RNA was isolated from
alkali-burned rabbit corneas and normal rabbit corneas
with RNAzol (Molecular Research Center, Inc.,
Cincinnati, OH), as previously described. 21 The RNAs
were resuspended in a solution containing 0.5% SDS
and 0.1 U/ml RNasin (Promega, Madison, WI) and
stored at —70°C until use.
Slot-Blot Hybridization
About 20 fig of total RNA from each sample were
denatured in 50% formamide at 65°C for 5 minutes
and were twofold serially diluted. The RNA was then
transblotted to BAS membranes with a slot-blotter
(Schleicher & Schuell) 21 and hybridized with the 32 P-
labeled cDNA of a 1(1) mRNA, as described previously.
22
In Situ Hybridization
The normal, alkali-burned, and lacerated rabbit corneas
were fixedin 4% methanol-free EM-grade formaldehyde
(Polysciences, Inc., Warrington, PA) in 0.05 M
sodium PBS, pH 7.4, at 4°C for 1 hour, and then
immersed in 30% sucrose in DEPC-treated water for 2
hours at 4°C. The specimens were embedded in OCT
compound (Miles, Elkhart, IN), quick frozen in ethanol/dry
ice, and stored at -70°C. Frozen sections of
5 fim thickness were cut by a cryostat and mounted on
Superfrost/Plus microscope slides (Fisher Scientific,
Pittsburgh, PA). The sections were then subjected to
hybridization with 3 H-labeled cDNA probes of rabbit
al(I) mRNA, as previously described. 21
A 3 H-labeled DNA probe was prepared by [ 3 H]-
dCTP incorporation using random primers as previously
described. 21 The specificity of the probe was 2 X
10 7 cpm/fxg. The hybridization reaction was carried
out at 45°C overnight. The sections were washed
under stringent conditions to reduce the background,
as previously described. 21 The slides were then dehydrated
in graded ethanol, air dried, and dipped in 50%
Kodak NT-2 Nuclear emulsion (Eastman-Kodak, Manchester,
NY); after exposure for 2 to 4 weeks, the
slides were developed for 3 minutes in a Kodak D-19
developer at 15°C, counterstained with hematoxylin
and eosin, and mounted.
Immunohistochemistry
For detection of collagen I, frozen sections of 5 /um
thickness were prepared from the OCT-embedded,
unfixed rabbit corneas by a cryostat and mounted on
Superfrost/Plus microscope slides (Fisher Scientific).
Cryosections were subjected to immunostaining using
the ABC method (avidin-biotin-peroxidase complex),
described by Chida et al. 23 For detection of a-SMA,
vimentin, and desmin, sections (3 /xm) of paraffin-embedded
corneas were mounted on precleaned glass
slides without adhesive. The sections were deparaffinized
and rehydrated with PBS before incubation with
antibodies.
Briefly, the tissue sections were incubated with
0.3% H 2 0 2 in methanol at room temperature for 30
minutes to eliminate endogenous peroxidase activity
in tissue. The sections were rinsed with PBS and incubated
at room temperature for 120 minutes with
preimmune serum of the source of biotinylated second
antibody (10X diluted) to block nonspecific absorption
of second antibodies to the tissue sections.
The sections were incubated with goat polyclonal anticollagen
I antibodies (1 jug/ml) or appropriately diluted
monoclonal antibodies (0.1 to 1 jig/ml) at room
temperature for 60 minutes or 4°C overnight. The
sections were washed with PBS and further incubated
with biotinylated second antibody for 60 minutes at
room temperature, followed by incubation with streptavidin-biotin-peroxidase
complex at room temperature
for 60 minutes. The sections were soaked in a
solution containing 0.2 mg/ml of 3-3'-diaminobenzidine
hydrochloride, 0.005% of H 2 0 2 , and 50 mM Tris-
HC1 buffer, pH 7.6, for 3 to 5 minutes and counterstained
with Mayer's hematoxylin. Photomicrograms
were prepared with a Nikon Diaphot microscope (Nikon,
Garden City, NY).
RESULTS
Slot-Blot Hybridization With 32 P-Labeled
cDNAofal(I)mRNA
Total RNA was extracted from alkali-burned rabbit
corneas that had healed for 1, 2, 3, 5, 7, 14, and 21
days with RNAzol, and it was subjected to slot-blot
hybridization with 32 P-labeled cDNA of a 1(1) mRNA
as previously described. 21 Figure 1 demonstrates that
levels of a 1(1) mRNA steadily increase and reach a
plateau as the injured corneas are allowed to heal for
more than 14 days. These results indicate that the alkali-burned
corneas actively synthesize collagen I in
attempts to repair the destroyed extracellular matrix
induced by the alkali treatment.
In Situ Hybridization With 3 H-Labeled cDNA
Probe of al (I) mRNA
To identify the cells synthesizing collagen I in alkaliburned
corneas, the injured corneas that had healed
for 1 day and for 1,2, and 3 weeks were subjected to in

Expression of Collagen I in Injured Corneas 3323
E
4.0
3.0
2.0
1.0
0
-
• • • • i • ' • • i • • • • i • • • • i • • • • i • • • •
:
I
/
N
' 1 $ l/f J
r
0 5 10 15
days after Injury
FIGURE l. Slot-blot hybridization of «1 (I) mRNA from alkaliburned
corneas by 32 P-labeled cDNA insert of ARC35. Total
RNA was isolated from alkali-burned corneas healed for 1 to
21 days as indicated. Twenty jug of total RNA from each
sample were twofold serially diluted and blotted to BASmembranes
(10 Mg in the first well). The niters were then
hybridized with 32 P-labelecl cDNA insert of ARC35, 22 and
autoradiograms were prepared. The relative intensities of
the samples were determined with a Helena Quick Scan densitometer
(Helena Laboratories, Beaumont, TX). The figures
are an average of four corneas; the bars indicate the
standard deviations.
situ hybridization with 3 H-labeled cDNA probe of
cd(I) mRNA as previously described. 18 Figure 2a
shows that 1 day after alkali burn, the keratocytes adjacent
to the injury express a low level of al(I) mRNA-
As the injured corneas heal, the keratocytes migrate
into the injured stroma and express high levels of al (I)
mRNA. The observation is consistent with the results
shown in Figure 1 in which the levels of a 1(1) mRNA
increase as alkali-burned corneas heal for up to 21
days. Results of the histologic examination and the
bright field of in situ hybridization indicate that retrocorneal
fibrillar membranes often form in the alkaliburned
corneas healed for more than 1 week (Fig. 2f).
The presence of the retrocorneal membrane in alkaliinjured
corneas can also be seen in Figures 4 and 7.
The formation of a retrocorneal membrane starts at
the edge of injury and extends toward the center as the
alkali-burned corneas healed progressively. The fibroblastic
cells in the retrocorneal membranes in alkaliburned
corneas that healed for 1 to 3 weeks also express
high levels of al(l) mRNA (Figs. 2b, 2c, 2d, 3f).
Figures 3b and 3c show that no specific hybridization
signals can be readily identified in the regenerating
epithelium of alkali-burned corneas that have healed
for 1 day and 3 weeks. The results indicate that the
regenerating epithelium either does not express or expresses
a very low level of al(i) mRNA.
20
•
In an attempt to examine whether the lacerated
cornea may heal differently, a series of similar experiments
of in situ hybridization was performed with lacerated
corneas healed for 1, 2, and 3 weeks. Fibroblastic
cells migrate into the wound of the lacerated cornea
and express high levels of al(I) mRNA. Figure 2e
shows a typical dark-field microgram from a lacerated
cornea healed for 2 weeks. Similar results were obtained
with lacerated corneas that had healed for 1
and 3 weeks (data not shown). It is of interest to note
that lacerated corneas do not form the retrocorneal
membrane (Fig. 2e). The endothelial cells in the uninjured
portion of the cornea either do not express or
express a very low level of al(l) mRNA (Fig. 3d),
whereas the epithelial cells in the regenerated epithelial
plug have a low level but specific expression of
al(l) mRNA (Figs. 2e, 3a).
Immunostaining With Antibody Against
Collagen I
To verify whether collagen I exists in the retrocorneal
membrane, goat antibody against collagen I was used
in immunohistochemical studies with alkali-burned
and lacerated corneas. Figure 4a shows the reaction of
anti-collagen I antibody with the denatured collagenous
components in the alkali-injured stroma healed
for I day. Figure 4b demonstrates the presence of collagen
I in the granulomatous tissue between the regenerated
epithelium and stroma of alkali-burned corneas
healed for 3 weeks. Figure 4c shows that the antibody
reacts with the retrocorneal membrane but does
not react with the Descemet's membrane. Similar immunohistochemical
studies were performed with lacerated
corneas healed for 1, 2, and 3 weeks. The results
indicate that the lacerated corneas did not form retrocorneal
membrane, and no fibrous structure can be
detected underneath the endothelium of the uninjured
portion of the lacerated corneas (data not
shown).
Immunostaining of Alkali-Burned Rabbit
Corneas With Antibodies Against Smooth
Muscle a-Actin, Vimentin, and Desmin
To characterize the fibroblastic cells in the alkaliburned
corneas, antibodies against a-SMA, vimentin,
and desmin were used in immunohistochemical studies
with alkali-injured cornea healed for 3 weeks. Figure
5a shows that anti-a-SMA does not react with either
keratocytes or epithelial cells in the normal corneas.
In alkali-burned cornea healed for 2 and 3
weeks, the antibody reacts with the fibroblastic cells in
stroma and retrocorneal membrane (Figs. 5b, 7a, 7c).
The anti-a-SMA antibody does not react with the epithelial
cells in alkali-burned and lacerated corneas
(Figs. 5b, 7a, 7e). The endothelial cells in both normal

3324 Investigative Ophthalmology 8c Visual Science, November 1993, Vol. 34, No. 12
1
f
RCM
0.1 mm
FIGURE 2 In situ hybridization of injured corneas with 3 H-labeled cDNA of al(I) mRNA. The
alkali-injured corneas were allowed to heal for 1 day and for 1, 2, and 3 weeks. The corneas
were then subjected to in situ hybridization with 3 H-labeled ARC35 cDNA probes encoding
the 3'-end of the al(I) mRNA. Dark-field micrograms: Panel a, alkali-burned corneas healed
for 24 hours; panel b, alkali-burned corneas 1 week after injury; panel c, alkali-burned
corneas 2 weeks after injury; panel d, alkali-burned corneas 3 weeks after injury; panel e,
lacerated cornea healed for 2 weeks; panel f, bright-field microgram of panel b. Large arrows
indicate the sites of injury. Small arrows and stars indicate cells hybridized by the 3 H-labeled
probe. D, Descemet's membrane; RCM, retrocorneal membrane.
(data not shown) and lacerated corneas (Fig. 7g) are
not stained by the anti-a-SMA antibody.
In another experiment, antibody against vimentin
was used to stain the normal, alkali-injured, and lacerated
corneas. The anti-vimentin antibody reacts with
keratocytes in normal corneas and fibroblastic cells in
alkali-injured stroma (Figs. 6, 7b), fibroblastic cells in
retrocorneal membranes (Fig. 7d), endothelial cells of
normal corneas (data not shown), and in lacerated corneas
healed for 2 weeks (Fig. 7h). The epithelial cells in

Expression of Collagen I in Injured Corneas 3325
FIGURE 3. In situ hybridization of injured corneas with 3 H-labeled cDNA of al(I) mRNA,
Panels a and d, lacerated cornea healed for 2 weeks; panels b and e, alkali-burned cornea
healed for 1 day; panels c and f, alkali-burned cornea healed for 3 weeks, epi, epithelium; D,
Descemet's membrane. Magnification, bar = 0.2 mm.
01 mm
0.1mm
FIGURE 4. Immunostaining of alkali-burned corneas with antibodies
against collagen I. Injured corneas healed for 1 day
and 3 weeks were subjected to immunostaining with specific
anti-collagen I antibodies. Panel a, cornea 24 hours after
injury. Degenerated stroma still reacts to the anti-collagen I
antibodies (X270). Panels b and c, 3 weeks after injury;
panel b, the newly formed granulation tissue between epithelium
and stroma (*) (X270); panel c, retrocorneal fibriilar
membrane. Magnification (X350), bar = 0.1 mm.
FIGURE 5. Immunostaining of normal and alkali-burned corneas
with antibody against smooth muscle a-actin. 0.1 fig/ml
of anti-a-actin monoclonal antibody was used in the immunostaining.
Panel a, normal cornea; panel b, alkali-burned corneas
healed for 3 weeks. Magnification, bar = 0.1 mm.
FIGURE 6. Immunostaining of normal and alkali-burned corneas
with antibody against vimentin. 0.88 fig/ml of anti-vimentin
monoclonal antibody was used. Panel a, normal cornea;
panel b, alkali-burned cornea healed for 3 weeks. Magnification,
bar = 0.1 mm.
0.1 mm

3326 Investigative Ophthalmology & Visual Science, November 1993, Vol. 34, No. 12
Viiiuillin
e
9
0.1 mm
FIGURE 7. Immunostaining of alkali-burned and lacerated corneas with antibodies against
smooth muscle a-actin and vimentin. Alkali-burned and lacerated corneas healed for 2 weeks
were subjected to immunostaining as described in Figures 5 and 6. Panels a, b, c, and d,
alkali-burned cornea; panels e, f, g, and h, lacerated cornea; panels a, c, e, and g, anti-smooth
muscle a-actin; panel b, d, f, and h, anti-vimentin. RCM, retrocorneal membrane; arrows
indicate Descemet's membrane; *, granulation tissues in lacerated cornea. Magnification, bar
= 0.1 mm.
normal corneas are not stained by anti-vimentin antibody
(Fig. 6a), but the basal epithelial cells in alkaliburned
and lacerated corneas are stained by the antibody
{Figs. 7b, 7f). The anti-desmin antibody does not
react with any cell type in normal and injured corneas,
but it does react with ocular muscle cells (data not
shown).
DISCUSSION
In the present study, we measured the amounts of
al(I) mRNA in alkali-injured rabbit corneas. Our results
indicate that the amount of al(I) mRNA steadily
increases as the alkali-burned corneas heal, and it
reaches a plateau after 14 days of injury (Fig. 1). We
previously reported that the increase of the amount of
al(I) mRNA in lacerated rabbit corneas followed a
biphasic kinetics. 15 The difference in the levels of a 1(1)
mRNA can be explained by the fact that alkali burn
causes severe cell death, whereas laceration does not.
The increase of al(l) mRNA reflects the proliferation
of fibroblastic cells and increased cellular activities
when the alkali-burned corneas are allowed to heal
(Fig. 2).
The cell types that express a 1(1) mRNA are not
exactly identical in alkali-burned and lacerated corneas,
as judged by in situ hybridization (Figs. 2, 3). The
alkali-burned corneas often form a retrocorneal membrane
within a week after injury, and the fibroblastic
cells in this membrane express high levels of a 1(1)

Expression of Collagen I in Injured Corneas 3327
mRNA. However, the lacerated corneas do not form
such retrocorneal membrane. Another interesting observation
is that although the epithelial cells of lacerated
corneas have a relatively low but specific expression
of a 1(1) mRNA, the epithelial cells of alkaliburned
corneas do not have detectable activity in
expressing a 1(1) mRNA. The reasons for the differences
in the expression of a 1(1) mRNA by the epithelial
cells in alkali-burned and lacerated corneas remain
unknown. It is possible, however, that the persistence
of PMN or other inflammatory cells 1 " 5 in alkali-burned
corneas may be responsible for the variation in the
expression of a 1(1) mRNA by the epithelial cells. It is
known that chemotactants derived from the proteolytic
reaction and oxidative burst of PMN, as well as
cytokines released by other inflammatory cells, can
modulate the functions of cells residing in the injured
tissues. 24 Kay et al have demonstrated that PMN secrete
factors that stimulate production of basic fibroblast
growth factor by cultured corneal endothelial
cells. 25 ' 26 Basic fibroblast growth factor alters the synthesis
of collagen I by the endothelial cells. The
scheme is consistent with our observations of the persistence
of inflammatory cells in alkali-burned corneas
and the synthesis of collagen I by the fibroblastic cells
in the retrocorneal membrane. Many investigators
have also demonstrated that cytokines (interleukin 2
and 7-interferon) modulate the expression of collagen
genes in vivo and in vitro. 27 " 31
Our immunohistochemical studies indicate that
the fibroblastic cells in the injured stromas and in the
retrocorneal membrane have characteristics of myofibroblasts,
in that these fibroblastic cells are stained by
the antibodies against vimentin and a-SMA (Figs. 5, 6,
7) but not by anti-desmin antibody. Vimentin has been
shown to be expressed by most mesenchymal cells. 18
However, the expression of a-SMA is restricted to the
smooth muscle cells and some fibroblastic cells—the
so-designated myofibroblasts. 19 " 20 It should be mentioned
that stromal cells (keratocytes) in the normal
cornea do not react with the anti-a-actin antibody (Fig.
5), indicating that there are phenotypic changes of
stromal fibroblastic cells in the injured stroma and in
the newly formed retrocorneal membrane. Recently, it
has been suggested that myofibroblasts play a role in
the contraction of incision cornea wounds. 20 It is likely
that these fibroblastic cells contribute to wound contraction
in our experimental model of alkali burns.
However, further studies are needed to elucidate this
hypothesis.
The fibroblastic cells found in the injured stroma
are most likely derived from the keratocytes in the injured
tissue. However, the source of the fibroblastic
cells in the retrocorneal membrane is not known. It is
possible that the rabbit corneal endothelial cells may
proliferate and transform to the fibroblastic cell type.
This hypothesis is especially intriguing when one considers
that corneal endothelial cells are derived from
mesenchymal neurocrest during embryonic development,
just as are the stromal keratocytes. 11 The results
of immunohistochemical studies of anti-a-actin and
anti-vimentin are consistent with the notion that the
fibroblastic cells in retrocorneal membrane are of endothelial
cell origin. 18 It has been demonstrated recently
that the endothelial cells can enter the cell cycle,
possibly because of the presence of cytokines, growth
factors, or both. 32 - 33 It is plausible to speculate that
cytokines secreted by PMN or other inflammatory
cells persisting in the alkali-burned corneas may induce
the proliferation and transformation of endothelial
cells to myofibroblasts, which form the retrocorneal
membranes. Although the alkali burns (8 mm in
diameter) in our studies are relatively mild and do not
usually involve other ocular tissues (e.g., trabecular
meshwork or iris ciliary body), it is possible that the
fibroblastic cells in the retrocorneal membrane may
derive from these surrounding ocular tissues.
The anti-vimentin antibody does not react with the
epithelial cells of normal corneas. Sundar-Raj et al recently
demonstrated that vimentin is transiently expressed
by epithelial cells in nonpenetrating lacerated
corneas that healed for fewer than 5 days. 34 However,
in the present study we found that the basal epithelial
cells still express vimentin 2 weeks after injury (Fig. 7b,
7f). The reason for the differences between our observations
and that of Sundar-Raj et al is not known. However,
experimental designs (alkali burn, penetrating
incision versus nonpenetrating incision) may account
for our differing results.
Key Words
alkali-burned corneas, lacerated corneas, wound-healing,
collagen I, myofibroblast
References
1. Hughes FWF, Jr. Alkali burns of the eye: II: Clinical
and pathological course. Arch Ophthalmol. 1946; 36:
124-189.
2. Matsuda H, Smelszer GK. Epithelium and stroma in
alkali-burned corneas. Arch Ophthalmol. 1973; 89:296-
401.
3. Kao WWY, Ebert J, Kao CWC, Covington H, Cintron
C. Development of monoclonal antibodies recognizing
collagenase from rabbit PMN: The presence of the
enzyme in ulcerating corneas. Curr Eye Res. 1986; 5:
801-805.
4. Burns FR, Gray RD, Paterson CA. Inhibition of alkaliinduced
corneal ulceration and perforation of a thiol
peptide. Invest Ophthalmol Vis Sci. 1990; 31:107-114.
5. Omerod LD, Abelson MB, Kenyon KB. Standard models
of corneal injury using alkali-immersed filter discs.
Invest Ophthalmol Vis Sci. 1989;30:2148-2153.
6. Omerod LD, Garsd A, Reddy CV, et al. Dynamics of

3328 Investigative Ophthalmology 8c Visual Science, November 1993, Vol. 34, No. 12
corneal epithelial healing after an alkali burn. Invest
OpMhalmol Vis Sci. 1989; 30:1784-1793.
7. Jughans BM, Collin HB. The limbal vascular response
to corneal injury: Autoradiography study. Cornea.
1989;67:685-693.
8. Chung JH, Fagerholm P. Corneal alkali wound healing
in the monkey. Ada Ophthalmol. 1989; 67:685-
693.
9. Kao WWY, Vergnes JP, Ebert J, Sundar-Raj CV,
Brown SI. Increased collagenase and gelatinase activities
in keratoconus. Biochem Biophys Res Comm. 1982;
107:929-936.
10. Kao WWY, Mai SH, Lee Chou KL. Biosynthesis of
procollagens and collagens by tissue explants and matrix
free cells from embryonic chick cornea. Invest Ophthalmol
Vis Sci. 1982;23:787-795.
11. Hay ED. Development of the vertebrate cornea. In:
Bourne GH, Danielli JF, eds. International Review of
Cytology. New York: Academic Press; 1980:263-322.
12. Matsubara M, Zieski JD, Fini ME. Mechanism of basement
membrane dissolution preceeding corneal ulceration.
Invest Ophthalmol Vis Sci. 1991;32:3221-3237.
13. Fini ME, Cui TY, Mouldovan A, et al. An inhibitor of
the matrix metalloproteinase synthesized by rabbit
corneal epithelium. Invest Ophthalmol Vis Sci.
1991;32:
2997-3001.
14. Hayashi K, Berman M, Smith D, et al. Pathogenesis of
corneal epithelial defects: Role of plasminogen activator.
CurrEyeRes. 1991; 10:381-398.
15. Sakai J, Hung J, Zhu G, et al. Collagen metabolism
during healing of lacerated rabbit corneas. Exp Eye
Res. 1991;52:237-244.
16. Cintron C, Hong B-S, Kublin CL. Quantitative analysis
of collagen from normal developing corneas and
corneal scars. Curr Eye Res. 1981:1-8.
17. Cintron C, Hong B-S, Covington HI, Macarak EJ. Heterogeneity
of collagens in rabbit cornea, type III collagen.
Invest Ophthalmol Vis Sci. 1988; 29:767-775.
18. Sappino AP, Schurch W, Gabbiani G. Biology of disease:
Differentiation repertoire of fibroblastic cells:
Expression of cytoskeletal proteins as marker of phenotypic
modulations. Lab Invest. 1990;63:144-161.
19. Majno G, Gabbiani G, Hirschel BJ, Rayan GB, Statkov
PR. Contraction of granulation tissue in vitro-similarity
to smooth muscle. Science. 1971; 173:548-550.
20. Garana RMR, Petroll WM, Chen W-T, et al. Radial
keratotomy: II: Role of the myofibroblast in corneal
wound contraction. Invest Ophthalmol Vis Sci. 1992;
33:3271-3282.
21. Zhu G, Ishizaki M, HasebaT, et al. Expression of K12
keratin in alkali-burned rabbit corneas. Curr Eye Res.
1992;11:
22. Haseba T, Nakazawa M, Kao CWC, Murthy R, Kao
WWY. Isolation of wound specific cDNA clones from
a cDNA library prepared with mRNAs of alkaliburned
rabbit corneas. Cornea. 1991; 10:322-329.
23. Chida Y, Ishizaki M, Kao WWY. Expression and methylation
of the /?-subunit gene of prolyl 4-hydroxylase:
In erythrocytes, tendon and cornea of chick embryos.
Connect Tiss Res. 1992;28:191-204.
24. Gallin JI. Inflammation. In: Paul WE, ed. Fundamental
Immunology. New York: Raven Press; 1989; 721-733.
25. Kay EDP, Rivela L, He YG. Corneal endothelium modulation
factors released by polymorphonuclear leukocytes.
Invest Ophthalmol Vis Sci. 1990;31:313-322.
26. Kay EDP, Gu X, Ninomiya Y, Smith RE. Corneal endothelial
modulation: A factor released by leukocytes induces
basic fibroblast growth factor that modulates
cell shape and collagen. Invest Ophthalmol Vis Sci.
1993;34:663-672.
27. Lankat-Buttgereit B, Kulozik M, Hunzelmann N,
Krieg T. Cytokines alter mRNA steady state levels of
basement membrane proteins in human skin fibrob\asts.
J Dermatol Sci. 1991; 2:300-307.
28. Kingnorth AN, Slavin J. Peptide growth factors and
wound healing. BrJSurg 1991;78:1286-1290.
29. Dinarello CA. Inflammatory cytokines, interleukin-1
and tumor necrosis factor as effector on molecules in
autoimmune diseases. Curr Opin Immunol. 1991;3:
941-948.
30. Freundlich J, Bomalaski JS, Neilson E, Jimenez SA.
Regulation of fibroblast proliferation and collagen
synthesis by cytokines. Immunol Today. 1986; 7:303-
307.
31. Thomas KA. Fibroblast growth factors. FASEB J.
1987; 1:434-440.
32. Laing RA, Neubauer L, Oak SS, Kayne HL, Leibowitz
HM. Evidence for mitosis in the adult corneal endothelium.
Ophthalmology. 1984;91:1129-1134.
33. Couch JM, Cullen P, Casey TA, Fabre JW.Mitotic activity
of corneal endothelial cells in organ culture with
recombinant human epidermal growth factor. Ophthalmology.
1987; 94:1-6.
34. Sundar Raj N, Rizzo JD, Anderson SC, Gesiotto JP.
Expression of vimentin by rabbit corneal epithelial
cells during wound repair. Cell Tissue Res. 1992;267:
347-356.