Abstract
purpose. To determine whether it is possible to induce proliferation in the
endothelium of older donor corneas by treatment of the intact monolayer
with EDTA.
methods. Corneas from donors 52 to 75 years of age were obtained from an eye
bank and were usually cut in quarters to increase sample size. The
effect of EDTA dose (0.02–2.0 mg/ml) and incubation time (6, 30, and
60 minutes) on endothelial cell–cell contacts was evaluated by
staining for ZO-1, a cell junction marker. Cell death was tested by a
commercial live–dead assay. Corneal pieces were incubated for 0, 24,
48, or 60 hours in culture medium (M-199, 10% fetal bovine serum, 10
ng/ml epidermal growth factor, 20 ng/ml fibroblast growth factor)
before EDTA treatment. After treatment, pieces were incubated in the
same medium for 24, 48, 72, or 96 hours to permit cell cycle entry.
Tissue was fixed, stained for Ki67 (a marker for late G1-phase through
the M-phase), and mounted in medium containing propidium iodide to
visualize all nuclei. Confocal images were evaluated by computer (Image
software; NIH, Bethesda, MD) to count Ki67-positive and propidium
iodide–stained cells.
results. EDTA released corneal endothelial cell–cell contacts in a dose- and
time-dependent manner. At doses and incubation times tested, EDTA did
not induce significant cell death. Preincubation in culture medium for
24 hours was needed for endothelial cells to efficiently initiate
proliferation in response to EDTA. The endothelium of corneas incubated
in mitogen-containing medium for up to 108 hours without EDTA treatment
did not stain for Ki67. EDTA at 2.0 mg/ml for 60 minutes appeared
optimal and stimulated 16% to 18% of the cells to proliferate.
Ki67-positive mitotic figures were visible 48 hours after exposure to
EDTA. Formation of daughter cells was visible after double-staining for
Ki67 and ZO-1.
conclusions. EDTA released cells from contact inhibition and promoted proliferation
in corneal endothelium from older donors. The authors hypothesize that
corneal endothelium from older individuals divide in situ when exposed
to positive growth factors under conditions in which cells have been
transiently released from contact inhibition.
Human corneal endothelial cell density decreases with
age,
1 2 indicating that these cells do not replicate
sufficiently to replace dead or injured cells. Transfection studies
using viral oncoproteins have demonstrated that human corneal
endothelium has an intrinsic high proliferative
capacity.
3 4 The fact that endothelial cells possess the
capability to divide but do not normally replicate in vivo suggests
that they are actively maintained in a nonreplicative state. Cell cycle
studies indicate that human corneal endothelial cells in vivo are
arrested in the G1-phase of the cell cycle.
5 6 Among the
factors that may be responsible for maintaining these cells in G1-phase
arrest are the nonavailability of positive growth factors
7 and the presence in aqueous humor of transforming growth factor
(TGF)-β2, which suppresses S-phase entry in cultured corneal
endothelial cells.
8 9
In addition, studies of the developing cornea in neonatal rats suggest
that a contact-inhibition–like mechanism may actively suppress
replication in the mature endothelial monolayer.
10 Other evidence for cell-contact–mediated regulation of proliferation
is that corneal endothelial cells divide in response to
wounding.
11 12 In tissue culture and in organ cultured
corneas, only endothelial cells adjacent to the wound edge or cells
that have migrated into the wound bed proliferate, demonstrating the
importance of releasing cell–cell contacts to promote proliferation.
Formation of cell–cell contacts is mediated by a number of proteins
that are associated with different types of junctional complexes,
including cadherins (adhering junctions),
13 ZO-1 (tight
junctions),
14 and connexin-43 (gap
junctions).
15 These proteins all require calcium for
maintenance of their adhesion function. In low-calcium environments,
junctional complexes mediated by these proteins disassemble, and
cell–cell contact is broken. Exposure of the corneal endothelium to
calcium-free medium causes disruption of apical junctional complexes,
increased transendothelial perfusion, and corneal
edema.
16 17 18 These changes can be reversed by replacing
calcium in the medium
17 18 or by exposing cells to
ionophores that release intracellular calcium stores.
18
Previous studies from this laboratory
19 have demonstrated
that corneal endothelium from older donors (>50 years of age) is
capable of proliferation in response to a mechanical wound. Cells from
older donors responded in fewer numbers and entered the cell cycle more
slowly than those from younger donors (<30 years of age) when corneas
were cultured in medium containing 10% serum and 20 ng/ml fibroblast
growth factor (FGF). Addition of 10 ng/ml epidermal growth factor (EGF)
to this medium induced significantly more older cells to proliferate
and increased the rate of cell cycle entry. As in previous studies,
only cells within the wound bed or at the wound edge proliferated.
Together, these data indicate that endothelial cells from older donors
can proliferate if cell–cell contacts are broken by wounding and if
given sufficient mitogenic stimulation. The present study was designed
to determine whether it is possible to induce proliferation in the
endothelium without mechanical wounding of the monolayer. Because EDTA,
a known chelator of calcium and magnesium, has been used in many
laboratories to release cultured cells from contact inhibition, it was
tested for its ability to release endothelial cells from cell–cell
contact and promote proliferation in an organ culture model.
Donor human corneas were obtained from National Disease Research
Interchange (Philadelphia, PA) and from the Central Florida Lions Eye
Bank. All corneas were maintained at 4°C in preservation medium for 1
week or less before study. Endothelial cell counts were more than 2000
cells/mm2. Criteria for exclusion of corneas from
these studies included history of endothelial dystrophy, presence of
central guttata, low endothelial cell density, and ocular inflammation
or disease.
Twenty-one human corneas were obtained from donors 52 to 75 years
of age (mean age, 65.1 years). Whole corneas were usually cut in
quarters to increase sample size. In some cases, whole corneas were
used as control samples. Corneal pieces were placed endothelial-side-up
in individual wells of a 24-well tissue culture plate (Falcon, Lincoln
Park, NJ). Pieces were incubated for 24 hours in medium-199 containing
10% fetal bovine serum (FBS), 10 ng/ml epidermal growth factor (EGF;
Upstate Biotechnologies, Lake Placid, NY), 20 ng/ml fibroblast growth
factor (FGF; Biomedical Technologies, Stoughton, MA), and 50 mg/ml
gentamicin to stabilize the endothelium before study. EDTA (di-sodium
EDTA.2H20) was prepared in Hanks’ balanced salt
solution (HBSS; without calcium chloride, magnesium chloride, or
magnesium sulfate; Life Technologies, Grand Island, NY), adjusted to
pH7.4, and added to the culture medium at a final concentration of
0.02, 0.2, or 2.0 mg/ml. Corneas were treated with EDTA for 10, 30, or
60 minutes and then returned to culture medium for up to 96 hours. EDTA
treatment controls included exposing corneal pieces to all
manipulations and incubation conditions, including 1 hour in HBSS, but
without EDTA. All corneas were maintained at 37°C in a 5% carbon
dioxide, humidified atmosphere until removal for analysis of cell cycle
progression.
Immunostaining for ZO-1 detected corneal endothelial cell
boundaries, whereas Ki67 staining detected actively cycling cells.
Immunolocalization was performed using the same antibodies and
protocols as described previously for ZO-1
10 and
Ki67.
19 Corneal pieces stained for Ki67 were mounted in
medium containing propidium iodide (PI; Vector Laboratories,
Burlingame, CA) to visualize all nuclei. In some cases, samples were
double-stained for both Ki67 and ZO-1. Slides were viewed using a
confocal microscope (model TCS 4D, equipped with a DMRBE laser; Leitz,
Wetzlar, Germany; and SCANware ver. 4.2 software; Leica Lasertechnik,
Heidelberg, Germany). Images were collected from the central region of
the corneal specimens, away from the cut edges, by using a ×16, ×40,
or ×100 oil-immersion lens. Laser power and gain controls were
adjusted to achieve an optimal range of output signal intensity for
each channel. Confocal images were collected, and micrographs were
printed by computer (Photoshop ver. 4.0; Adobe, San Jose, CA). For some
micrographs, the printing contrast was adjusted to provide a clearer
image.
Fluorescence confocal immunocytochemistry for Ki67 was used to
evaluate corneal endothelial cells for their ability to enter and
complete the cell cycle. All nuclei were stained with PI. Positive Ki67
staining patterns detected actively cycling cells and also acted as
markers for specific phases of the cell cycle. Completion of the cell
cycle was determined by observation of mitotic figures stained with
Ki67. Three representative confocal micrographs were taken per corneal
quarter with a ×40 objective lens. A software program (Image ver.
1.62; NIH) was used to count total PI-stained nuclei, total
Ki67-positive cells, and cells in the G1, S/G2, and M phases of the
cell cycle. Cells were counted in five 100 μm
2 areas of each micrograph (15 areas counted per corneal piece). Counts
were averaged, and the percentage of actively cycling cells and of
cells in each phase of the cell cycle was calculated. Each study was
conducted using corneas from two to three donors. Statistical
comparisons were made by computer, using a paired Student’s
t-test (StatView ver, 4.11; Abacus Concepts, Berkeley, CA).
Results reported in
Table 1 are expressed as percentages of Ki67-positive cells ± SD.
Results reported in
Figures 3B and 3C are expressed as percentages of
Ki67-positive cells ± SEM.
Studies were first conducted to determine the effect of EDTA on
the integrity of endothelial cell–cell contacts. Corneal pieces were
incubated for 10, 30, or 60 minutes in 0.02, 0.2, or 2.0 mg/ml EDTA,
then washed, and stained for ZO-1 to visualize cell boundaries.
Representative micrographs in
Figure 1 show the dose- and time-dependent effect of EDTA on cell–cell
contacts. In the absence of EDTA, cells maintained normal lateral
associations
(Fig. 1A) . With increasing EDTA concentration or
incubation time, there was an increase in the lateral separation of
cells, visualized as a double line of positive ZO-1 staining
demarcating the two cell membranes
(Figs. 1B 1C 1D 1E) . Incubation in
the presence of 2.0 mg/ml for 60 minutes caused cells to become rounded
with only thin cytoplasmic strands connecting adjacent cells. This
treatment, however, did not cause cells to lose contact with
Descemet’s membrane. EDTA, at the concentrations and incubation times
tested, did not significantly induce cell death as indicated by a
commercially available live–dead staining assay (data not shown).
Previous studies
19 revealed that corneal endothelial
cells incubated in medium containing 10% serum, 20 ng/ml FGF, and 10
ng/ml EGF stained positively for Ki67 in response to wounding, but no
positive staining was observed in unwounded areas of the endothelium.
Similar absence of Ki67 staining is shown in
Figure 2A which shows corneal quarters that were incubated in this medium for
108 hours. Thus, exposure of the intact endothelium to this
mitogen-containing medium was not sufficient to induce cell cycle
entry. Similarly, incubation in up to 2.0 mg/ml EDTA alone did not
promote proliferation, even if cells were postincubated for 48 hours in
mitogen-containing medium
(Figs. 2B 2C) . In contrast, preincubation in
mitogen-containing medium and subsequent exposure to EDTA followed by a
48-hour postincubation promoted cell cycle progression in the
endothelium, revealed by the presence of Ki67-stained nuclei and
mitotic figures
(Fig. 2D) . Negative control corneas were manipulated in
a manner similar to EDTA-treated corneas. These corneas were incubated
for 1 hour in HBSS alone to control for the EDTA incubation and then
maintained for 48 hours in the culture medium described. Fluorescence
microscopy showed only a very rare Ki67-positive cell (data not shown).
Ki67 is expressed in cells from the mid-to-late G1 phase through
mitosis, making it an excellent marker of actively cycling
cells.
20 21 Immunostaining with antibody to Ki67 also
permits semiquantitative analysis of the relative number of cells in
late G1 phase through mitosis.
19 22 The graph in
Figure 3A shows typical endothelial cell cycle kinetics in corneal specimens
from a single donor preincubated for 24 hours in mitogen-containing
medium, treated with 0.2 mg/ml EDTA for 60 minutes, and postincubated
in medium for 0, 24, 48, or 72 hours. Cell proliferation began by 24
hours after EDTA treatment, and the cell cycle was completed in
approximately 48 hours, as indicated by the presence of mitotic figures
at that time point. By 72 hours after EDTA treatment, there was a mixed
population of Ki67-stained cells with evidence of new cell cycle entry.
With this method of analysis, it was not possible to determine whether
this population of Ki67-positive cells represented a second round of
proliferation or whether a subpopulation of cells responded more slowly
to the EDTA+mitogen treatment. Double-staining of corneal pieces with
Ki67 and ZO-1 revealed that in cells treated with EDTA and mitogens the
cell cycle is completed and daughter cells formed (data not shown).
Studies were conducted to determine incubation conditions that
induce cell cycle entry in the largest number of endothelial cells, as
determined by counting Ki67-positive cells. The requirement for
preincubation of corneas in mitogen-containing medium was first
determined. For these studies, preincubation times of 0, 24, 48, and 60
hours were used. EDTA concentration and treatment times were held
constant, and corneal samples were postincubated in mitogen-containing
medium for 48 hours.
Figure 3B provides a representative example of the
results, which indicate that, of the preincubation times tested, 24
hours was sufficient to yield the maximal percentage of Ki67-positive
cells. This preincubation time was then used for all subsequent
studies. In studies to determine the optimal postincubation time,
corneal pieces were preincubated in medium for 24 hours, followed by
EDTA treatment. Postincubation times tested included 0, 24, 48, 72, and
96 hours.
Figure 3C provides a representative example, indicating that
a maximum number of Ki67-positive cells was obtained 48 hours after
incubation. Depending on the specific donor cornea, some Ki67-positive
cells could be detected 72 hours after EDTA treatment; however, no
samples showed actively cycling cells by 96 hours after incubation.
As indicated in
Table 1 , EDTA at either 0.2 or 2.0 mg/ml was capable of
promoting cell cycle progression. The relative percentage of
Ki67-positive cells observed under the four EDTA incubation conditions
was significantly higher than negative controls in which corneas were
incubated in HBSS alone. Corneas treated with 2.0 mg/ml EDTA for 60
minutes yielded a significantly higher number of actively cycling cells
(
P = 0.007) than those treated with 0.2 mg/ml EDTA for
30 minutes. Repeated EDTA treatment cycles promoted proliferation over
a longer period, but the treatment became gradually less effective
(data not shown).
The present study demonstrates that EDTA, a known chelator
of calcium and magnesium, is able to disrupt lateral cell contacts
within the corneal endothelial monolayer without apparent interference
with cell–substrate associations or negative effects on viability.
ZO-1-staining of endothelium treated for 60 minutes with 2.0 mg/ml EDTA
(the highest concentration and longest treatment time tested) revealed
thin cytoplasmic strands between rounded cells in a pattern very
similar to that observed by Stern et al.,
18 who perfused
the endothelium with calcium-free medium. Exposure of the intact
endothelium to EDTA or to mitogens alone was not sufficient to induce
proliferation. In contrast, a combination of the two, with appropriate
timing, clearly stimulated a proliferative response.
Treatment of the monolayer with EDTA presumably interfered sufficiently
with maintenance of junctional complexes to release cells from contact
inhibition, making them sensitive to mitogenic stimulation. The reason
preincubation of the endothelium in mitogen-containing medium was
required is not clear. Pretreatment with mitogens may initiate cellular
responses that prepare for cell cycle entry once cell–cell contacts
have been released. Preincubation times shorter than the 24 hours used
in this study may be even more effective, because receptor
downregulation can occur with prolonged growth factor incubation.
Future studies will test the effectiveness of preincubation times
shorter than 24 hours and also identify any cell cycle–related changes
that occur during this preincubation period.
The possibility that preincubation with mitogens prepares cells to
enter the cycle is supported by the fact that cells appeared to
initiate proliferation as an almost synchronous population. This
response differed from cell cycle entry in wounded endothelium exposed
to the same culture medium.
19 Under those conditions,
cells continued to enter the cycle until the wound bed was completely
repopulated.
An important finding from these studies was that, after treatment
of the intact endothelial monolayer with EDTA and mitogens,
approximately 17% of the cell population entered the cell cycle. It is
expected that further refinement of the EDTA and/or mitogen treatment
protocol will increase the relative percentage of proliferating cells.
Because the relative percentage of Ki67-positive cells tended to differ
with the specific treatment, it is possible that we may not have
obtained optimal efficiency in stimulating proliferation. The ability
to induce proliferation in endothelial cells does not appear to be
restricted by the presence of a senescent population; we have not
observed positive staining for β-galactosidase, a marker of cell
senescence,
23 in the endothelium of older individuals
(personal observation). Future studies will determine whether
EDTA+mitogen treatment consistently increases cell density in donor
corneas with low endothelial cell counts.
The induction of proliferation in these studies occurred without
mechanical wounding and without apparent cell damage. This underscores
the integral role of cell–cell contact in maintaining the corneal
endothelium in a nonreplicative state and indicates that temporary
interference with this important antiproliferative mechanism could
induce transient proliferation in this physiologically important
monolayer. These findings suggest that it may be possible to induce
proliferation and thereby increase the density of endothelial cells.
This treatment could be applied directly to the endothelium to increase
endothelial cell density in corneas to be used for transplantation. In
addition, it may be possible to stimulate endothelial cell division in
situ by using EDTA (or similar reagents that temporarily interfere with
cell–cell contact) plus mitogen to increase cell density in
individuals at risk for vision loss due to low endothelial cell counts.
The authors thank Deshea Harris for excellent technical guidance
and assistance in various phases of this work.