Investigative Ophthalmology & Visual Science, Vol. 30, No. 12, December 1989
Copyright © Association for Research in Vision and Ophthalmology
Time-Lapse Videomicroscopic Study of In Vitro
Wound Closure in Rabbit Corneal Cells
Kwan Y. Chan,* Dorothy L. Patron,f and Yvonne T. Cosgrovef
This study used video time-lapse recording to characterize the dynamic features of corneal epithelial
cells and keratocytes during in vitro wound closure. Confluent cultures of these two cell types from
rabbits were established in Rose chambers. A wound 4 or 10 mm in diameter was produced in the
center of each culture by mechanical removal of cells. Wound closure was recorded by videomicroscopy for 2-3 days and reviewed at a playback speed of 400 times normal. The epithelial cells at the
wound margin initiated migration by extending lamellipodia with undulatory motions. Successive tiers
of cells moved as a continuous sheet in a unified and coordinated manner while maintaining intercellular linkage. The migration was unidirectional, toward the wound center. The mean migration rate of
the leading cells was 104 jim/hr. The trailing cells migrated at successively slower rates, inversely
proportional to their distance from the wound margin. Mitosis was rare during migration but did occur
simultaneously. The mitotic rate was 3.7 mitoses/100 cells. The relative mitotic frequency was 0.23
mitosis/hr. By contrast, in keratocyte cultures, the cells around the wound margin migrated individually and asynchronously without intercellular connection. Initially the cells moved generally toward
the wound space, but later, different cells migrated in different directions. The mean migration rate
was 15 fim/hr. Mitosis occurred frequently. The mitotic rate was 25.3 mitoses/100 cells, and the
relative mitotic frequency was 1.33 mitoses/hr. The cell cycle duration was 9.9 hr. Thus corneal
epithelial cells and keratocytes showed fundamentally different characteristics and mechanisms of
wound closure in vitro. Invest Ophthalmol Vis Sci 30:2488-2498,1989
A primary step in corneal wound healing is the
migration of surrounding epithelial cells into the
wound.1 Various studies have investigated the kinetics of migration, the cellular and structural features of migration, and the effect of various substances on migration. However, the dynamic characteristics of epithelial migration during wound closure
have not been visualized directly. There have been
time-lapse cinematographic studies of epithelial
movement (including that of corneal epithelium) in
tissue or organ culture,2"4 none of which pertains to
wound closure in particular. In the present study,
time-lapse videomicroscopy was used to characterize
the dynamic features of corneal epithelial cells and
keratocytes during their migration in cell culture to
close an artificially produced wound. Part of the
video recording has been presented previously.5
Materials and Methods
Cell Culture Techniques
Corneal cells were isolated from adult albino rabbits (2.0-2.5 kg). All experiments with these animals
were performed according to the ARVO Resolution
on the Use of Animals in Research. The epithelial
cells were isolated by the dispase method as described
previously.6 Keratocytes were isolated by collagenase
digestion of the stroma.7 The purity of the cells isolated by these techniques has also been verified.7
Corneal epithelial cells were plated at densities of
2-3 million cells per culture. Keratocytes were plated
at approximately 4 X 105 cells per culture. Because of
differences in the plating efficiency and mitotic rate,7
the epithelial cells and keratocytes were confluent
5-6 and 3 days respectively, after plating.
Cells were plated into Rose chambers8 (Fig. 1). For
the initial culture of cells, the top and bottom coverslips of the Rose chamber were plastic and gelatincoated glass, respectively. The culture chamber consisted of a growth surface 25 mm in diameter and
approximately 2 ml in volume.
The culture medium consisted of Ml99 medium
supplemented with hormones and trace chemicals, as
described previously,6 and containing 2% fetal calf
From the Departments of ""Ophthalmology and fObstetrics and
Gynecology, University of Washington School of Medicine, Seattle, Washington.
Supported by NIH Grant EY-04538 and Edna McConnell Clark
Foundation Grant 13187. Some facilities were supported in part by
HDO-2274 and EY-01730, and in part by a departmental award
from Research to Prevent Blindness, Inc., New York, New York.
Reprint requests: Kwan Y. Chan, PhD, Department of Ophthalmology, RJ-10, University of Washington, Seattle, WA 98195.
2488
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933145/ on 06/16/2017
VITRO CORNEAL WOUND CLOSURE / Chan er al
No. 12
v
cated by pH changes in the dye. A Dage MTI videocamera (Dage-MTI, Michigan City, IN) and a Gyyr
time-lapse videocassette recorder (Odetics, Anaheim,
CA) were used to record images at one picture per 7
sec on VHS videotapes. Microscope objectives of 16X
and 25X were used. The recording field either followed the same migration front (a wound margin)
and so required manual repositioning of the field
with time, or fixed at the same area on the growth
surface and thereby showed a continuous passing of
cells.
STAINLESS
''STEEL
SI L I CONE RUBBER
S
GLASS
COVERSLIP
( GELATIN-COATED
2489
)
Fig. 1. Rose chamber assembly for cell culture. The culture area
is indicated by the broken circle. Two hypodermic needles (25gauge) are inserted through the silicone rubber gasket to provide
access of liquid to the culture chamber. One half of the upper part
of the assembly is omitted to provide a cutaway view. Overall
dimensions are 75 X 50 X 10 mm.
serum. Culture medium was replaced every 1-2 days.
The cells grown in Rose chambers (sealed) were incubated at 37°C in humidified air.
Wounding of Cells
To produce a wound area in the confluent monolayer of epithelial cells or keratocytes, the Rose
chamber was partially disassembled at room temperature inside a flow hood by removing the screws, the
top stainless steel plate, and the top (plastic) coverslip. In the exposed culture chamber (containing culture medium), a plastic cylinder (10 mm in diameter)
was positioned at the center of the culture surface and
the cells inside the boundary of the cylinder were
scrubbed away with a cotton applicator.9 To produce
a 4-mm-diameter wound, the cells were scrubbed
away with a cotton applicator by freehand. Complete
removal of cells in the wound area was ascertained by
microscopic examination. The culture chambers
were rinsed with culture medium to remove cell
debris. A new, glass coverslip was applied to the top
of the silastic gasket, and the Rose chamber was reassembled. The interior of the chamber wasfilledwith
fresh culture medium, and incubation at 37°C was
resumed.
Time-Lapse Videomicroscopy
Video recording of cell migration in the wounded
cultures was started either immediately or a few hours
after wounding. The Rose chamber was positioned
on the stage of an inverted microscope and maintained at 37°C during the recording period (2-3
days). Culture medium was replaced daily as indi-
Data Analysis
The recorded image of cell migration was examined at a playback speed of 400 times normal. Calculation of time was based on the digital timer generated on-screen during each episode of recording. Distance was calibrated from the image of the divisions
in a hemocytometer.
The rate of cell migration was calculated based on
the distance traveled per unit time (/um/hr). The same
marker point in the nucleus or cell membrane of a
migrating cell was tracked across the screen of the
monitor, from one side of the screen to the opposite
side, using the freeze-frame technique. The distance
traveled on the screen was measured and converted
to the actual distance in micrometers. Travel time
was noted from the on-screen timer. In each image
field, four representative cells were measured
(10-15% variability), each with duplicate measurements (1-2% variability). The mean rate was calculated for each field. When the same migration front
was tracked with time, the image field was repositioned at appropriate time intervals. When the field
was fixed (see previous section), sequential groups ol
migrating cells were measured as they appeared in the
field, starting with the migration front. The separation between adjacent groups was less than the width
of the field. The distance of each group from the
migration front was estimated by summation of the
distance between adjacent groups.
Cell mitosis was evaluated by three measurements.
Mitotic cells were identified by specific features in
each phase of mitosis (see Results). The relative frequency of mitosis (mitoses/hr) was determined by the
number of mitoses occurring in a given field within a
given period (5-18 hr). Mitotic rate (mitoses/10C
cells) was determined by the number of mitoses occurring in a cell population which was enumerated ir
a given field. The duration of a cell cycle (in hours"
was determined from the time taken by the daughtei
cells of one mitosis to undergo another mitosis, with
cytokinesis used as a marker of mitosis.
Statistical analysis was based on the student t-test.
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933145/ on 06/16/2017
2490
INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / December 1989
n
CEI
Vol. 30
1
Fig. 2. Phase micrograph of corneal epithelial cells during in
vitro wound closure. The leading cells surrounding the wound
margin (W) typically extend lamellipodia (arrows), and the nucleus
(star) is displaced to the posterior. Sometimes intercellular linkage
appears as a broad band of membranous structure (arrowhead)
between cell borders. Calibration bar = 50 fitn.
Results
Epithelial Cells
The use of Rose chambers for videomicroscopy
necessitated the culturing of rabbit corneal epithelial
cells on gelatin-coated glass coverslips. The growth
and morphology of these cells were similar to the
previously studied cells grown in plastic tissue culture
dishes.79 During confluency, the monolayered cells
did not show any gross migratory movements, except
for an infrequent, transient shifting motion of local
cell groups suggesting minor adjustment of cell positions on the substratum.
When a cell-free area (a wound) was artificially
produced in the center of the confluent cells, cell migration was initiated around the wound margin. This
occurred from 1 half hour to several hours after
wounding. The first tier of cells surrounding the
wound margin (the leading cells) extended broad,
fanlike lamellipodia toward the open wound area
(Fig. 2). The nucleus of these cells was located offcenter and posterior to the lamellipodia (Fig. 2), Cell
diameter was 50-80 jtim, depending on the extent of
lamellipodia. In the compressed video playback
time-frame (equivalent to fast-motion), the lamellipodia undulated in quick successions like waves. The
trailing cells also exhibited undulating and streaming
motions, although the forward position of the lamellipodia and dislocation of the nucleus were more difficult to distinguish. Cell diameter was 30-40 fj,m.
The migration of the tiers of cells was unidirectional,
toward the wound center. The cells moved as a continuous sheet in a unified and coordinated manner
while maintaining intercellular linkage and relative
position to one another (Fig. 3). This linkage some-
Fig. 3. Videomicrographs of migrating epithelial cells. The migration front is moving from right to left, pushing two air bubbles
(asterisks) along the way. (The Rose chamber was normally placed
upside down on the microscope stage.) Lamellipodia of the leading
cells appear irregular in outline (arrows) during undulation. Broad
bands of intercellular linkage are visible (arrowheads). Relative
time (min:sec): (A) 0:00; (B) 53:01; (C) 121:18. Calibration bar
= 50 /xm.
times appeared as a broad, membranous structure at
the cell boundaries (Figs. 2, 3). During cell migration,
intracellular organelles (refractive vesicles of various
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933145/ on 06/16/2017
No. 12
IN VITRO CORNEAL WOUND CLOSURE / Chan er al
2491
Fig. 4, Videomicrographs showing the reunion of a separated leading epithelial cell. The migration front is moving from right to left. (A) A
leading cell (large arrow) is separated from the others (relative time, min:sec = 0). (B) Numerous smaller lamellipodia (arrowheads) are
extended as the cell turns around to move back (18:34). (C) Afilopodium(small arrow) is extended to anchor the cell onto the closest neighbor
(39:44). (D) The cell pulls itself back to join the other leading cells. Thefilopodium(small arrow) is still visible (62:16). (E) The cell attempts to
establish normal cellular contacts (83:19). (F) The cellfinallyreestablishes intercellular linkage and extends typical lamellipodia (large arrow)
(111:17). Asterisks indicate air bubbles. Calibration bar = 50 nm.
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933145/ on 06/16/2017
2492
INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / December 1989
sizes) also moved about inside the cytoplasm. Infrequently, a leading cell moved ahead of the migration
front and became separated from other leading cells
(Fig. 4A). When this occurred, the cell moved back,
using lamellipodia (Fig. 4B). A filopodium was then
extended to contact the closest cell in the migration
front (Fig. 4C), which the separated cell used as an
anchor to pull itself back into the group (Fig. 4D).
The effort to reestablish contact lasted approximately
1 hr. Another 40 min were required for the cell to
reestablish normal cellular linkage with the surrounding cells (Fig. 4E, F). In some cases, the separated cell rejoined the group using lamellipodia action alone, without extending a filopodium. On another occasion, a separated leading cell attempted to
rejoin the group but failed to establish sufficient contact with the migrating cells. The cell became spherical and was finally ejected into the culture medium.
The migration rate was analyzed by following the
same migration front with time or byfixingthe frame
at one area to monitor successive tiers of cells trailing
behind the migration front. The migration rate of the
leading cells was 104 ± 7 /im/hr (mean ± SEM; N
= 11 fields, from four cultures). The trailing cells
moved successively more slowly the farther they were
from the wound margin (Fig. 5). The half-maximal
rate occurred at 264-438 /um behind the wound margin, as calculated by regression analysis of the two
cultures shown in Figure 5. It is unlikely that this
differential migration rate was dependent on time (eg,
aging of cells during culture or fatigue of migrating
cells) rather than distance from the wound margin.
As shown in Figure 5, multiple measurements that
spanned several hours on the leading cells yielded
similar migration rates, whereas measurements of a
similar time span on the trailing cells resulted in differential rates. In addition, as shown in Figure 5B, the
leading cells migrated on the second day at a rate
similar to that of the first day. During completion of
wound closure, the leading cells from all sides converged at a similar rate (Fig. 5B). On visual inspection, the leading cells merged together centripetally to
cover the remaining open space. Even after complete
closure, the trailing cells still moved centripetally for
approximately 1 hr with gradual reduction in speed,
until all movement stopped. This residual motion
caused some compression of cells in the center, which
appeared to have a higher cell density, as shown previously (Fig. 7F in ref. 9). Afterwards, the cells did not
move any further, with the exception of some transient shifting motion as was noticed in normal confluent cells.
Time-lapse videomicroscopy of wound closure also
offered an opportunity to visualize mitosis of corneal
epithelial cells. Figure 6 shows an epithelial cell un-
Vol. 30
Day 1
Day 0
i i i f
it
i
i
100
80
<o 60
o
40
5
20
Wound
margin
0
2
4
6
8
Distance from wound margin
10
( x 1 0 ~ 2 urn )
B
Day 0
Day 1
V
120
_ 10 °
\_ 80
<o
J.
60
c
o
I 4°
i
20
VAWound
margin
4
6
0
2
Distance from wound
2
margin ( x 1 0 " i i m )
X Y Z
Center
Closure
Fig. 5. Migration rates of various tiers of epithelial cells. The
migration rates of leading cells at the same wound margin and
trailing cells at various distances from the wound margin are plotted as histograms. The times at which the rates were measured
during wound closure are indicated (scale not in proportion). Error
bars indicate SEM (N = 4 cells). Broken line indicates the slope
obtained from regression analysis. (A) Wound size was 10 mm-diameter. The half-maximal rate occurred 438 Mm from the wound
margin (arrows). (B) Wound size was 4 mm-diameter. Half-maximal rate occurred 264 /*m from the wound margin (arrows). The
rates of leading cells on different sides of the wound center during
completion of closure are also shown. X = upper side; Y = left side;
Z = lower side.
dergoing mitosis during cell migration. This cell was
located a few cells behind the migration front. During
prophase, the nucleus rotated a few degrees, after
which the cytoplasm retracted centripetally approximately one third of the way toward the nucleus. This
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933145/ on 06/16/2017
No. 12
VITRO CORNEAL WOUND CLOSURE / Chan er al
2493
Fig. 6A); the chromosomes migrated toward opposite
poles (anaphase, Fig. 6B); and the cytoplasm divided
(cytokinesis) to form two daughter cells (telophase,
Fig. 6C). During the whole process, the dividing cell
apparently maintained the same rate of migration as
the adjacent migrating cells. Figure 7 shows a rare
occasion when two adjacent cells divided simultaneously. These cells maintained their relative positions just behind the leading cells during the whole
process. Mitosis usually lasted 15-30 min. The relative frequency of mitosis in cultured epithelial cells
was 0.23 ± 0.09 mitoses/hr (mean ± SEM; N = five
fields, from four cultures). The mitotic rate was 3.7
± 1.2 mitoses/100 cells (mean ± SEM; N = five
fields, from four cultures).
H-f5 .•11*1:31
HDD
A
Fig. 6. Videomicrographs of epithelial mitosis during wound
closure. The migration front is moving toward the lower right
corner. The leading cells show lamellipodia (arrow) and a posteriorly located nucleus (arrowhead). (A) Metaphase in a dividing cell
(asterisk). Note the alignment of chromosomes (relative time,
min:sec, 0:0). (B) In anaphase, two sets of chromosomes separate
(asterisk) (9:22). (C) In telophase, two daughter cells are formed
(stars) (11:56). Calibration bar = 50 jim,
partial rounding produced a refractive outline (Fig.
6A). Then, in sequence, the chromosomes arranged
themselves linearly along the equator (metaphase,
Keratocytes
When wound closure was studied on keratocytes,
different features were observed. Approximately 1
half hr after wounding, only the first few tiers of cells
at the wound margin migrated into the wound space
(Fig. 8A, B). The cells moved individually and
asynchronously. The movement involved sliding of
the whole cell, which maintained its spindle shape.
There was no pronounced undulation in the filopodium or in the flattened cytoplasmic process. The
soma diameter was under 20 /im. None of the cells
was interconnected. Although the general direction of
migration was toward the wound center, individual
cells could move sideways or even backwards (Fig. 8).
As the cells spread out in the wound space, they did
not cover the substratum completely, but rather left
pockets of blank space behind. The cells behind the
migration front remained confluent and stationary,
until 5-6 hr later, when they began to move into the
wound area. The migration rate of the leading cells
(generalized over an imaginary front) was 15 ± 1.5
/i.m/hr (mean ± SEM; N = four fields, from two cultures).
Mitosis was commonly observed in the keratocytes. The cell, normally spindle-shaped (Fig. 8D),
rounded up completely into a refractive sphere (Fig.
8E), which subsequently divided into two (Fig. 8F).
Sometimes a filopodium remained unretracted during the process (Fig. 8E, F). The daughter cells flattened into the normal spindle shape after division
and moved away from one another (Fig. 8G, H).
Under favorable optical conditions, detailed observation of each phase of mitosis was possible on some
keratocytes. The events were similar to those described above for epithelial cells, except that each
phase, including cytokinesis, occurred while the cell
was rounded. The duration of keratocyte mitosis was
15-30 min. The relative frequency of mitosis was
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933145/ on 06/16/2017
2494
INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / December 1989
r
Vol. 00
two cultures). In some episodes of videomicroscopy,
the daughter cells from an earlier mitosis remained in
the same field and underwent another mitosis at a
later time. Thus, it was possible to determine the duration of a cell cycle in keratocytes, which was 9.9
± 0.6 hr (mean ± SEM; N = six cells). Some daughter
cells divided at exactly the same time, while some
divided within 1 half hr of each other.
Table 1 summarizes the different features of in
vitro wound closure in corneal epithelial cells and
keratocytes.
l-flH5U:n:2T-
nnn
Discussion
—
-Dl-BB 11:22:12
11:31:51
Fig. 7. Videomicrographs of two simultaneous mitoses in migrating epithelial cells. The migration front is moving toward the
lower left corner. Ruffles {small arrows) appear in the undulating
lamellipodia of leading cells. (A) Two cells appear refractive as they
progress into metaphase (arrowhead and large arrow) {relative
time, min:sec, = 0:0). (B) One of the cells is in anaphase (large
arrow) (8:15). (C) Two pairs of daughter cells are formed (asterisks
and stars) (23:32). Calibration bar = 50 fim.
1.33 ± 0.17 mitoses/hr (mean ± SEM; N = six fields,
from two cultures). The mitotic rate was 25.3 ± 5.2
mitoses/100 cells (mean ± SEM; N = six fields, from
This study has characterized certain dynamic features of corneal cell migration during in vitro wound
closure. Numerous studies have provided useful information on the cellular and structural features associated with corneal cell migration during either in
vivo or in vitro wound healing.1(M5 However, most of
these studies employed conventional techniques,
such as gross measurement of change in wound size,
and various conventional microscopic methods, such
as light,15 transmission," or scanning10-1213 microscopy. We used videomicroscopy to record cell migration in real time and obtained information that confirms some aspects of the hypothesis raised in previous studies. In addition, our data provide new
information not available previously.
This study shows that the hallmarks of epithelial
cell migration during wound closure include lamellipodia, sliding in sheets, and interconnection between
cells. Previous studies have described surface ruffling,
lamellae, and filopodia on the marginal cells of a
wound, and have suggested that these are responsible
for cell movement.10"13 Our data confirm that lamellipodia provide cell motility. The ruffles are caused by
the undulatory motion of the lamellipodia. The filopodia described previously are small in size10 and are
different in appearance from the larger filopodium
we observed when a separated cell rejoined the migration front. This study also provides direct evidence
to support the previous suggestion that epithelial cells
migrate as a continuous sheet while maintaining interconnections between cells. Previous ultrastructural
studies have shown a reduction of microvilli, a widening of interdesmosome space, and a lack of hemidesmosomes on marginal cells.121415 In addition, our
observation shows that each tier of cells migrates
within a narrow range of rates (10-15% variability).
We also observed flexibility between the cell borders
while the cells undulate individually and yet slide
together. Thus it appears that during mass migration,
the intercellular linkage between epithelial cells is
transformed into a flexible condition. Whether this
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933145/ on 06/16/2017
No 12
IN VITRO CORNEAL WOUND CLOSURE / Chan er al
2495
Fig. 8. Videomicrographs of keratocyte migration and mitosis during wound closure. (A) The freshly made wound is located in the upper
half of the frame (relative time, hr:min:sec, 0:0:0). (B) Thefirstfew tiers of cells move into the wound space (2:10:13). (C) Genera! movement
of the cells begins to appear more variable in direction and speed (4:35:35). For example, the cell marked with the arrowhead is moving
sideways toward the right in subsequent frames (C-H). (D) Another cell (arrow) moves towards the left in subsequent frames (D-H). The cell
marked with an asterisk undergoes mitosis in subsequent frames (5:49:33). (E) A keratocyte rounds up and appears refractive during mitosis
(asterisk). Note that theftlopodiumis not completely retracted (relative time of mitosis, min:sec, 0:0). (F) Cytokinesis in anaphase (asterisk)
(time of mitosis 9:02). (G) Two daughter cells separate (stars) (mitosis time 14:36). (H) The daughter cellsflattenand lose refractivity (stars).
Calibration bar = 50 ftm.
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933145/ on 06/16/2017
2496
INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / December 1989
Table 1. Summarized features of in vitro wound
closure in rabbit corneal cells
Features
Epithelial cells
Keratocytes
Directionality
Cellular relation
Timing
Motility structure
Migration rate (jtm/hr)
(leading cells)
Mitotic rate (mitoses/100
cells)
Relative mitotic frequency
(mitoses/hr)
Cell cycle duration (hr)
Unidirectional
Interconnected
Synchronous
Lamellipodia
Multidirectional
Separate
Asynchronous
Nonspecific
104 ± 7*
15 ± 1.5
3.7 ± 1.2f
25.3 ± 5.2
0.23 ± 0.09*
—
1.33 ±0.17
9.9 ± 0.6
The values are mean ± SEM. See text for N.
•P<0.001.
f P < 0.005 when compared to keratocytes.
condition compromises the normal barrier function
of epithelial layers is not clear.
Previous studies have discussed whether the sliding
cells push or pull during wound closure.1013 Two observations in our study suggest that this may be a
complex issue. First, starting with the leading cells at
the wound margin, the migration rate was inversely
proportional to the distance of cells from the wound
margin. If an open space provides a signal for the
marginal cells to initiate migration, this inverse relationship ensures that the trailing cells do not outrun
the leading cells. The differential rates may be a consequence of an intercellular signal mechanism and
the location of open space in one direction. Thus, the
leading cells may constitute a pulling force. Second,
after completion of closure at the center of the
wound, when the leading cells essentially stopped, the
trailing cells continued to migrate with reducing
speed for 1 hr before stopping completely. Thus, the
trailing cells may exert a pushing force. Alternatively,
the slowdown time may be related to the time required for execution of a signal mechanism for cessation of cell motility. It is noteworthy that a previous
study, in which trailing cells were monitored with ink
marks, described a crowded appearance of cells in the
center of a completed closure.13 The dynamic significance of these images is resolved by the video technique used in this study.
The migration rate (104 ixm/hr) of the leading epithelial cells measured in this in vitro model is the
highest ever recorded. Buck has summarized the rates
obtained in various studies through 1979 (in vivo
wounds produced by various methods); the rates
range within 60-70 /xm/hr.13 Recently, a more elaborate study yielded a rate of 64 ^m/hr.16 There has
been speculation that neutrophils, including poly-
Vol. 00
morphonuclear leucocytes, retard the initial rate of
epithelial migration.1214 It is possible that the simplified environment of our in vitro model avoided this
type of interfering factor and therefore permitted the
maximal migration rate. In a model of cultured epithelial cells similar to the one in this study, with the
exception that wounding was produced by freezing,
the migration rate was approximately 70 /xm/hr (recalculated from mm2/nr).17 However, in organ culture studies in which the sliding of epithelium towards the cut edge of excised cornea blocks was measured, rates of 20-26 ^m/hr were obtained."18'19 In
the present study, we sometimes encountered lower
rates of 50-60 /*m/hr in the leading cells. In each of
these instances, however, the cells either quickly
changed from confluency to a differentiated state, or
appeared unhealthy. Thus it is likely that subtle
changes in the culture condition or differentiation of
cells in vitro9 affects the migration rate.
Also relevant to the above discussion is the possibility that a latent period occurs before the initiation
of epithelial migration. Crosson et al16 have made a
definitive measurement of the lag period (5.5 hr) that
previous in vivo studies have repeatedly encountered." 13 In our in vitro study, initiation of epithelial
migration was also variable among different cultures
and among different locations of wound margins in
the same culture. It is possible that under in vivo or in
vitro conditions, biologic variability is an intrinsic
problem and may not be related directly to the mechanism of cell migration. However, since the minimum time required in our studies for the initiation of
migration was approximately 1 half hour for both
epithelial cells and keratocytes, a minimum period
for the induction of cell migration may be required.
Under in vivo conditions, the presence of damaged
tissues and cell debris, the variable nature of the exposed basal lamina or stroma surface, and the influence of inflammatory reaction may all contribute to
lengthen the latent period.
Perhaps the most surprising observation in this
study is the ability of migrating epithelial cells to undergo mitosis without missing a step in the sheetform
migration. Previous studies in this laboratory have
indicated that cultured epithelial cells are derived
from basal and wing cells, with very little contribution from superficial cells.7 Although the mitotic rate
of the cultured cells was low (similar to in vivo epithelium), our observation indicates that an epithelial cell
can execute both cell migration and mitosis almost
simultaneously, even at the migration front. This
concept has not been recognized previously.1 One
unique feature of epithelial mitosis was the partial
and limited retraction of perikaryon throughout the
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933145/ on 06/16/2017
No. 12
VITRO CORNEAL WOUND CLOSURE / Chan er ol
process. In contrast, keratocytes rounded up completely during mitosis. Thus, the epithelial cell is able
to maintain an interconnection with adjacent cells
and to slide forward while dividing.
A companson of the different features of epithelial
cells and keratocytes during in vitro wound closure
indicates that the process in each cell type is based on
a fundamentally different biologic program. The epithelial cells normally are interconnected by desmosomes in confluent cultures and show little motility.
Upon induction by unknown mechanisms to migrate
toward a wound space, the cells were mobilized in
coordination, with their interconnections intact but
flexible. In essence, these cultured cells behaved as an
organized tissue layer rather than as a collection of
individual cells. In contrast, keratocytes are a collection of separate cells in confluent cultures. Even
though they responded to a migration induction, the
movement was not coordinated and lacked a common directionality. Previous in vivo studies have
suggested that keratocytes align themselves in the
wound,1 and that in unwounded conditions, keratocyte processes interconnect to form a network structure.20 In this regard, it is possible that the in vitro
keratocyte model is far different from keratocyte
wound healing in vivo. Another difference between
epithelial and keratocyte migration is the difference
in migration rate. Epithelial cells moved as much as
seven times faster than keratocytes. On the other
hand, the mitotic rate and relative frequency of mitosis in migrating keratocytes were approximately six
times higher than those of migrating epithelial cells.
Thus, epithelial wound closure is based predominantly on cell migration, whereas keratocyte wound
closure relies heavily on cell replication. This conclusion is consistent with the concept derived from in
vivo wound healing studies.1 The present study did
not evaluate the aspect of increased epithelial mitosis
after the wound closure is completed.
Numerous studies have focused on the effect of
extracellular matrix proteins on epithelial migration1819 and on the role of actin in epithelial migration.21"24 In light of the above discussion on the susceptibility of the latent period and migration rate to
variability according to experimental conditions,
some precautions in the interpretation of data are
worthy of mention. For example, the organ culture
model used in some studies attained a migration rate
of only 20 ixm/hr, which was enhanced by fibronectin
or epidermal growth factor (EGF) to a maximum of
30 jum/hr.18'19 These values are only one third to one
half of those reported in animal studies. It appears
that this model does not provide optimal conditions
to attain the normal rate of epithelial migration.
2497
Thus, whether fibronectin or EGF remains effective
at the maximal migration rate of epithelial cells is yet
to be concluded. Another precaution involves the localization of actin filaments in migrating epithelial
cells. Several studies have identified actin filaments in
the basal cell membrane of marginal cells surrounding a wound.21"24 It is crucial in studies of this type to
ascertain that the cells being examined are actively
migrating. The hallmark features of a migrating epithelial cell identified in our study would be useful for
this purpose. In one study, which evaluated the difference between rat and rabbit epithelial migration in
vitro, the rabbit cells, shown in a micrograph, did not
exhibit lamellipodia and ruffles typical of migrating
cells.24 It is possible that the lack of stress fibers and
the perikaryal location of actin observed in these cells
were not related to migration. In addition, the migration rate measured in the model was 17-22 /im/hr.
The model itself was based on the movement of epithelial cells during subconfluent culture and did not
involve a deliberate wounding of confluent cells.
Thus, it is unclear whether the findings in that study
are applicable to normal wound healing. From the
large body of information obtained from previous
studies, some common criteria for various model
systems (whether in vivo or in vitro) may be established for studying epithelial wound healing in cornea. Such criteria should include the presence of lamellipodia and ruffles, movement in sheetform, and
a migration rate of at least 60-70 ^m/hr in the leading cells. The lag period should not be significantly
longer than 6 hr. Adherence to these criteria would
ensure that the model system operates in optimal
conditions in a way similar to the natural process of
wound healing itself.
The dynamic features of corneal cell migration
shown in this study emphasize the difficulty and incompleteness of studying wound healing when static
techniques are used. A leading epithelial cell (50-80
fxm in diameter) migrating at 104 ^m/hr will move a
distance equivalent to its diameter in 30-45 min.
Each cycle of undulation in the ruffles of lamellipodia
possibly happens in minutes, if not seconds. It is
likely that the structural and molecular components
that form the basis of epithelial migration are changing rapidly with time and space. It will be necessary to
employ dynamic techniques, such as time-lapse recording, to investigate the dynamic components of
wound closure in the same time frame as the phenomenon itself. This study has further recognized the
various finer components of epithelial migration,
such as initiation of migration (signal transduction),
modification of intercellular linkage, intercellular
communication (synchronization among cells and
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933145/ on 06/16/2017
2498
INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / December 1989
reunion of out-of-step cells), lamellipodia motility,
and termination of migration (slowdown and cessation). To understand fully the migratory function of
corneal epithelium, it may be necessary in future investigations to differentiate between the above components when determining the effect of various substances (eg, matrix proteins, growth factors, drugs) on
wound closure.
Key words: cell migration, corneal epithelium, keratocytes,
time-lapse technique, wound healing
Acknowledgments
10.
11.
12.
13.
14.
The authors are grateful to Peter Cummings for preparing the videomicrographs.
15.
References
16.
1. Binder PS, Wickham MG, Zavala EY, and Akers PH: Corneal
anatomy and wound healing. In Symposium on Medical and
Surgical Diseases of the Cornea, Barraquer JI, Binder PS, Buxton JN, Fine M, Jones DB, Laibson PR, Nesburn AB, Paton D,
and Troutman RC, editors. St Louis, C. V. Mosby, 1980, pp.
1-35.
2. Vaughan RB and Trinkaus JP: Movements of epithelial cell
sheet in vitro. J Cell Sci 1:407, 1966.
3. Dipasquale A: Locomotory activity of epithelial cells in culture. Exp Cell Res 94:191, 1975.
4. Gibbins JR: Epithelial migration in organ culture: A morphological and time lapse cinematographic analysis of migrating
stratified squamous epithelium. Pathology 10:207, 1978.
5. Chan KY and Patton D: Time-lapse video microscopic study
of in vitro wound closure in rabbit corneal epithelial cells and
keratocytes. ARVO Abstracts. Invest Ophthalmol Vis Sci
27(Suppl):52, 1986.
6. Chan KY and Haschke RH: Specificity of a neuronotrophic
factor from rabbit corneal epithelial cultures. Exp Eye Res
41:687, 1985.
7. Chan KY and Haschke RH: Isolation and culture of corneal
cells and their interactions with dissociated trigeminal
neurons. Exp Eye Res 35:137, 1982.
8. Rose G: A separable and multipurpose tissue culture chamber.
Tex Rep Biol Med 12:1074, 1954.
9. Chan KY: Release of plasminogen activator by cultured cor-
17.
18.
19.
20.
21.
22.
23.
24.
Vol. 30
neal epithelial cells during differentiation and wound closure.
Exp Eye Res 42:417, 1986.
Pfister RR: The healing of corneal epithelial abrasions in the
rabbit: A scanning electron microscope study. Invest Ophthalmol 14:648, 1975.
Kuwabara T, Perkins DG, and Cogan DG: Sliding of the epithelium in experimental corneal wounds. Invest Ophthalmol
15:4, 1976.
Haik BG and Zimny ML: Scanning electron microscopy of
corneal wound healing in the rabbit. Invest Ophthalmol
16:787, 1977.
Buck RC: Cell migration in repair of mouse corneal epithelium. Invest Ophthalmol Vis Sci 18:767, 1979.
Srinivasan BD, Worgul BV, Iwamoto T, and Eakins KE: The
reepithelialization of rabbit cornea following partial and complete epithelial denudation. Exp Eye Res 25:343, 1977.
Yamanaka H and Eguchi G: Regeneration of the cornea in
adult newts: Overall process and behavior of epithelial cells.
Differentiation 19:84, 1981.
Crosson CE, Klyce SD, and Beuerman RW: Epithelial wound
closure in the rabbit cornea: A biphasic process. Invest Ophthalmol Vis Sci 27:464, 1986.
Jumblatt MM and Neufeld AH: A tissue culture assay of corneal epithelial wound closure. Invest Ophthalmol Vis Sci 27:8,
1986.
Nishida T, Nakagawa S, Awata T, Ohashi Y, Watanabe K, and
Manabe R: Fibronectin promotes epithelial migration of cultured rabbit cornea in situ. J Cell Biol 97:1653, 1983.
Watanabe K, Nakagawa S, and Nishida T: Stimulatory effects
offibronectinand EGF on migration of corneal epithelial cells.
Invest Ophthalmol Vis Sci 28:205, 1987.
Nishida T, Yasumoto K, Otori T, and Desaki J: The network
structure of corneal fibroblasts in the rat as revealed by scanning electron microscopy. Invest Ophthalmol Vis Sci 29:1887,
1988.
Gipson IK and Anderson RA: Actin filaments in normal and
migrating corneal epithelial cells. Invest Ophthalmol 16:161,
1977.
Gipson IK, Westcott MJ, and Brooksby NG: Effects of cytochalasin B and D and colchicine on migration of the corneal
epithelium. Invest Ophthalmol Vis Sci 22:633, 1982.
Nakagawa S, Nishida T, and Manabe R: Actin organization in
migrating corneal epithelium of rabbits in situ. Exp Eye Res
41:335, 1985.
Soong HK and Cintron C: Different corneal epithelial healing
mechanisms in rat and rabbit: Role of actin and calmodulin.
Invest Ophthalmol Vis Sci 26:838, 1985.
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933145/ on 06/16/2017