Roles of Vascular Endothelial Growth Factor and Astrocyte
Degeneration in the Genesis of Retinopathy of Prematurity
Jonathan Stone,* Tailoi Chan-Ling,* Jacob Pe'er,^ Ahuva Itin,% Hadassah Gnessin,\
and Eli Keshet%
Purpose. To assess the role of vascular endothelial growth factor (VEGF) in the feline model
of retinopathy of prematurity (ROP).
Methods. Retinopathy of prematurity was induced in neonatal cats by raising them in an
oxygen-enriched (70% to 80%) atmosphere for 4 days to suppress vessel formation and then
returning them to room air for 3 to 27 days. In situ hybridization was used to detect the
expression of VEGF and its high-affinity receptor, flk-1, in the retina of neonatal cats, and
glial fibrillary acidic protein immunocytochemistry was used to assess astrocyte status.
Results. The expression of VEGF in the innermost layers of retina fell in hyperoxia and
increased on return to room air. Vascular endothelial growth factor expression was transient;
it was maximal where vessels were about to form, and it rapidly downregulated after vessels
had formed. During the proliferative vasculopathy of ROP, VEGF expression was stronger
than in the normally developing retina, and the astrocytes that normally express VEGF degenerated. After the degeneration of astrocytes, VEGF was expressed by neurones of the ganglion
cell layer, flk-1 was expressed by intraretinal and preretinal vessels. Supplemental oxygen
therapy reduced or eliminated the overexpression of VEGF expression, astrocyte degeneration, and formation of preretinal vessels.
Conclusions. Regulation of VEGF by tissue oxygen mediates the inhibition of vessel growth
during hyperoxia and the subsequent proliferative vasculopathy. Degeneration of retinal astrocytes creates conditions for the growth of preretinal vessels. Invest Ophthalmol Vis Sci.
1996;37:290-299.
A he pathogenesis of retinopathy of prematurity
(ROP) has been of intense interest since the disease
was recognized in the 1940s. Work by ophthalmologists, beginning with Campbell,1 pointed to the importance of oxygen in the etiology of the disease, and the
experimental work reviewed by Patz2 and Ashton and
colleagues3"6 showed the interplay of hyperoxia and
hypoxia in the pathogenic cascade. Ashton's work3"6
drew attention to the role of retinal oxygen levels in
From the Departments of%Molecular Biology and tOphthalmology, Hebrew
University and Hadassah University Hospital, Jerusalem, and the Department of
* Anatomy and Histology, University of Sydney, Australia.
Supported by grants from the National Health and Medical Research Council of
Australia, the Australian Retinitis Pigmentosa Association, the Sir Zelman Cowen
Universities Fund (Sydney), and the Mireille and James Levy Foundation
(Jerusalem), 'the Ramariotti Foundations (Sydney), and the Government Employees
Medical Research Fund.
Submitted for publication April 5, 1995; revised July 6, 1995; accepted October 4,
1995.
Proprietary interest category: N.
Reprint requests: Jonathan Stone, Department of Anatomy and Histology, University
of Sydney F13, Sydney, Netu South Wales 2006, Australia.
290
Downloaded From: http://iovs.arvojournals.org/ on 06/16/2017
the normal development of the retina, and he suggested that the mechanisms that cause vasculopathy,
such as ROP, may be exaggerations of normal mechanisms. Chan-Ling et al,7 Chan-Ling and Stone,8 and
Chan-Ling et al910 showed that retinal astrocytes degenerate in feline ROP and that their degeneration
is associated with, and may cause, a loss of barrier
properties in the vessels that develop during ROP.
In an earlier study,11 we presented evidence that
the endothelial-specific mitogen and chemoattractant,
vascular endothelial growth factor (VEGF), plays a key
role in the normal development of the retinal vasculature, linking retinal hypoxia to the formation of blood
vessels. In this study, we examine the role of VEGF
and astrocytes in the feline model of ROP, extending
previous analyses of this disease (reviewed in ChanLing and Stone12). Vascular endothelial growth factor
expression was examined in retinas undergoing the
proliferative vasculopathy of ROP and in a number of
control retinas. Astrocyte status was assessed from glial
Investigative Ophthalmology & Visual Science, February 1996, Vol. 37, No. 2
Copyright © Association for Research in Vision and Ophthalmology
Vascular Endothelial Growth Factor in Retinopathy of Prematurity
fibrillary acidic protein (GFAP)-labeling of sections
adjacent to those hybridized. Results show that during
the period of hyperoxia used to induce ROP, the normal expression of VEGF is inhibited strongly and that,
during the proliferative vasculopathy that follows return to room air, VEGF is expressed at levels well
above normal in the area of retina at and beyond the
edge of the reforming vasculature. In addition, during
the proliferative vasculopathy of feline ROP, it was
noted that degenerative changes occurred in astrocytes, that the expression of VEGF was no longer specific to astrocytes, and that preretinal vessels were common.
We hypothesize that changes in VEGF expression
mediate the suppression of vessel growth during the
period of hyperoxia as well as the proliferation of vessels that follows return to room air. We also propose
that a breakdown of astrocytes after return to room
air, first reported by Chan-Ling and Stone,8 contributes to the formation of retinal vessels. These hypotheses were tested by examination of the expression of
VEGF receptors during proliferative vasculopathy and
examination of the expression of VEGF during supplemental oxygen therapy (SOT) of ROP, as developed
by Chan-Ling et al.10
METHODS
All procedures were in accord with the guidelines established by the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.
Oxygen Exposure
Kittens were placed in incubators in which oxygen
tension was controlled by mixing air and oxygen and
were monitored by a polargraphic oxygen sensor. Approximately half the time, the mother was moved from
the incubator to a free-run room for exercise. Retinas
were studied from two animals not exposed to hyperoxia; they were 4 and 10 days of age. Retinas were
studied from two animals placed in hyperoxia (70%
to 80% oxygen) 1 to 24 hours after birth, kept in
hyperoxia until 4 days of age, and returned to room
air for 3, 11, and 27 days. Retinas were studied from
two animals exposed to hyperoxia and not returned
to room air. One was placed in hyperoxia 24 hours
after birth and remained in hyperoxia for 3 days from
birth in hyperoxia, thus representing the state of the
retina after the oxygen exposure used to induce ROP.
The other was raised in room air for 3 days and exposed to hyperoxia for 1 day; this shorter period of
exposure left much of the vessel structure intact and
allowed identification of the critical region at the edge
of the developing vasculature. Retinas were examined
from an animal raised in hyperoxia for 4 days and
then in 45% oxygen for 3 days.
Downloaded From: http://iovs.arvojournals.org/ on 06/16/2017
291
Preparation of Retinas
After appropriate periods in oxygen-enriched air,
room air, or both, each kitten was administered an
overdose (60 mg/kg) of sodium pentobarbitone. Eyes
were enucleated and fixed by immersion in 4% paraformaldehyde in phosphate-buffered saline. Procedures used were not fixation sensitive. Eyes were embedded in paraffin and sectioned horizontally at 10
fim.
In Situ Hybridization
Paraffin-embedded eyes were sectioned (5 to 10 mm
thick), processed, and hybridized in situ as previously
described.13 Autoradiographic exposure was for 5 to
9 days. As a VEGF-specific probe, we used a 1.8-kb
long cDNA fragment containing approximately the 3'
two thirds of the coding region, as well as the entire
3'-untranslated region of mouse VEGF165. As a VEGF
receptor-specific probe of the KDR/Flk-1 type,14 we
used KDR cDNA.15 cDNAs were subcloned onto the
polylinker of a PBS vector (Stratagene, Lajolla, CA)
and were linearized by digestion with the appropriate
restriction endonuclease to allow synthesis of a 35Slabeled complementary RNA in either the antisense
or sense orientation (using T3 or T7 RNA polymerase,
respectively). RNA probes were fragmented by mild
alkaline treatment before use for in situ hybridization.
Extensive testing of the sense probes produced consistently negative results.
Glial Fibrillary Acidic Protein
Immunochemistry
Paraffin-embedded sections were incubated in a polyclonal anti-GFAP antibody (Dako, Carpinteria, CA)
and then in a peroxidase-conjugate anti-rabbit IgG
antibody. Bound antibodies were made visible with
the diaminobenzidine reaction, and the retina was
counterstained with cresyl violet.
Quantification of Vascular Endothelial Growth
Factor Signal
Lengths of VEGF-hybridized sections were viewed under dark-field illumination using a 20X objective, and
the dark-field images were digitized. Illumination conditions were standardized, using set positions for field
and aperture diaphragms, for illumination brightness
and condenser position. The brightness of the darkfield image was measured in 50-//m steps along the
inner surface of the retina. The area measured at each
step was a rectangle 150 jum by 50 //m, oriented with
its long axis along the length of the section and centered on the VEGF signal at the inner surface. The
signal shown is the difference in signal intensity between each measurement at the inner surface of the
292
Investigative Ophthalmology & Visual Science, February 1996, Vol. 37, No. 2
FIGURE l. Vascular endothelial growth factor (VEGF) expression in normoxia, hyperoxia,
and retinopathy of prematurity. All sections were hybridized with the probe for VEGF mRNA
and are shown with the inner surface uppermost. Arrows in A, E, F, and G are positioned
so that their stubs are located at the peripheral limits of blood vessels, and they point to
the periphery of the retina. (A) Region from the edge of vessels in a normally developing
3-day-old retina. The signal is found internal to the inner plexiform layer (i). <B) The same
region under bright-field illumination. The retina has separated during processing from
the retinal pigment epithelium. (C) Region from the edge of surviving vasculature from
the retina of a kitten raised for 3 days in room air, then for 24 hours in 70% to 80% oxygen.
The signal at the inner surface of the retina has been suppressed; signal is apparent at the
outer surface, over the pigment epithelium. (D) Region from the retina of a kitten kept in
room air for 24 hours, then in 70% to 80% oxygen for 3 days. The signal at the inner
surface of the retina has been suppressed; signal is apparent at the outer surface, over the
pigment epithelium. (E) Region from the edge of the developing vasculature in the retina
of a kitten raised for 4 days in high (70% to 80%) oxygen, then in room air for 3 days.
Signal is strong at the inner surface and apparent over the pigment epithelium. (F) Region
from the edge of the developing vasculature in the retina of a kitten raised for 4 days in
high (70% to 80%) oxygen, then in room air for 11 days. Signal is very strong at the inner
surface. Signal at the pigment epithelium is not distinct from light reflected by the pigment
granules in the pigment epidielium. (G) Region from the edge of the developing vasculature
in the retina of a kitten raised for 4 days in high (70% to 80%) oxygen, then in room air
for 27 days. Signal remains strong at the inner surface. A weak signal is apparent at the
outer surface.
retina and the background signal at a point anterior to
the retina, separate from any retinal or ocular tissue.
RESULTS
Vascular Endothelial Growth Factor Expression
in Normal Development
Figures 1A and IB show the expression of VEGF
during normal development of the cat retina. Figure
Downloaded From: http://iovs.arvojournals.org/ on 06/16/2017
1A shows in dark-field illumination the expression
of VEGF at the inner surface of the retina of a 4day-old kitten. The signal is found close to the inner
surface of the retina, internal to the inner plexiform
layer (i). The arrow points toward the periphery of
the retina; the stub of the arrow is located at the
edge of the vessels forming in this retina, identified
by examination at high magnification. Figure IB
shows the region of retina in Figure 1A in bright-
Vascular Endothelial Growth Factor in Retinopathy of Prematurity
293
P4/llROP(hypoxic)
P4 nonnoxic
P4/3 SOT
field illumination. These observations confirm our
previous findings."
Vascular Endothelial Growth Factor Expression
During Retinopathy of Prematurity
During the period in hyperoxia used to induce feline
ROP, the growth of new vessels stops; within 4 days,
existing vessels are obliterated.6'7 Figures 1C and ID
show that hyperoxia downregulates the expression of
VEGF at the inner surface of the retina. Figure 1C
shows the VEGF signal in the retina of a 4-day-old
kitten exposed to oxygen for 24 hours; the VEGF signal present at 4 days in the normally developing retina
(Fig. 1A) is suppressed strongly. Figure ID shows that
after the longer exposure used to induce ROP (in this
case, 3 days), the VEGF signal is suppressed strongly
(Fig. ID).
When a kitten exposed to hyperoxia was returned
to room air, revascularization of the retina began. As
in normal development, the vessels spread from the
optic disc, but the new vasculature was proliferative,
fast spreading, and lacking in barrier properties.7 In
the retina that survived 3 days in room air after 4 days
in hyperoxia, vessels had regrown in the retina and
extended 4 mm nasal to the optic disc; 5.5 mm of
retina remained unvascularized. Figure IE shows
VEGF expression at the edge of the spreading vasculature; the stub of the arrow in Figure IE is located at
the most peripheral vessel (observed by inspection at
high power), and the arrow points to the peripheral
margin of the retina. Vascular endothelial growth factor expression was maximal near the edge of the vessels and extended peripheral to the edge, toward the
margin of the retina. Just central to the left of the
maximum, expression fell off sharply to a low level
that was maintained toward the optic disc.
In the retina that survived 11 days in room air
after 4 days in hyperoxia, the vessels extended 7.5 mm
nasal to the optic disc, still 3 mm from the edge of
the retina. At the edge of the developing vasculature
(arrow in Fig. IF), the VEGF signal was strong and
extended several millimeters into the peripheral retina (direction of the arrow). Central to the edge of
the vessels, the VEGF signal remained high for a short
distance, then was reduced sharply. The strength of
VEGF expression at this stage of proliferative vasculopathy was markedly higher than in normal development or after 3 days of recovery in air.
In the retina that survived 27 days in room air after
4 days in hyperoxia, vessels had extended 9.5 mm nasal
to the optic disc; only 0.5 mm remained without vasculature. A similar margin at the peripheral edge of the
normal retina remained avascular in adulthood.16 At this
nasal edge, therefore, vascularization seemed complete
and, correspondingly, the VEGF signal was low (not
shown). On the temporal side of the retina, the vessels
Downloaded From: http://iovs.arvojournals.org/ on 06/16/2017
peripheral
-1200
-1000
-800
-600
-400
-200
0
200
400
600
800
1000
Distance from the edge of vessels (microns)
FIGURE 2.
Measures of vascular endothelial growth factor expression in the inner (axon and ganglion cell) layers of
retina during normal development (4-day-old normoxic kitten), during the proliferative vasculopathy of retinopathy
of prematurity (P4/11 ROP) and in supplemental oxygen
therapy (P4/3 SOT). The signal, expressed in arbitrary
units, is the difference in brightness between the signal and
background measured on a scale of 0 to 256. The sections
were hybridized by standard procedure, section thickness
was 10 /im, illumination conditions of the microscope (lamp
brightness, field and aperture condensers, condenser positions) were held constant. The vertical dashed line represents the position of the most peripheral vessel in the section; more central retina is to the left, more peripheral to
the right.
extended 8.5 mm from the optic disc, leaving 2 mm of
retina unvascularized. At this margin, the VEGF signal
was strong peripheral to the edge of the vessels (stub of
the arrow in Fig. 1G) and toward the margin of the
retina (direction of the arrow). Central to the edge of
the vessels, VEGF expression was low. Overall, the VEGF
signal at this margin of the vessels was less than after 11
days in air, but it remained higher than normal (compare Figs. 1A, IF, 1G).
Effects on VEGF expression are shown more quantitatively in Figure 2. In normal development, the
VEGF signal peaks peripheral to the edge of the developing vessels, as described previously," and decreases
toward the optic disc and toward the periphery. In the
material from the 4- to 11-day-old kittens, the VEGF
signal shows similar trends: It reaches a maximum
peripheral to the edge of the vessels and decreases
more centrally and peripherally. However, the overall
level of expression is considerably higher than normal.
Status of Astrocytes in Proliferative
Vasculopathy
Figures 3A to 3C show the status of astrocytes in the
retinas studied. Confirming Ling and Stone,17 in the
m
Investigative Ophthahnology 8c Visual Science, February 1996, Vol. 37, No. 2
D:P10VEGF
!•: riOVIX.ll'
«
Kcl "...
ipl
•*M&k
B: P4/11 GFAP
£P4/11 VEGF
#
CP4/27GFAP
•".
H:P4/27VEGF
1.1*4
r .
3dQLC;FAP
r'
gel
ipl
ipl
normally developing 10-day-old retina (Fig. 3A),
GFAP+ processes are prominent at the inner surface
of the retina, wrapping axons and forming part of the
inner glia limitans. They lie principally in the axon
layer, internal to neurones of the ganglion cell
layer.1819 Only occasional GFAP+ processes extend
Downloaded From: http://iovs.arvojournals.org/ on 06/16/2017
ipl
into the ganglion cell layer. As reported previously,8
however, the astrocytes undergo degeneration during
the proliferative phase of retinopathy of prematurity.
Three days after return to room air, GFAP+ structures
at the edge of the spreading vasculature appeared abnormally patchy, and GFAP+ elements had spread into
Vascular Endothelial Growth Factor in Retinopathy of Prematurity
295
FIGURE 3. Localization of vascular endothelial growth factor (VEGF) expression and glial
fibrillary acidic protein (GFAP) label in normoxia, retinopathy of prematurity, and hyperoxia. All sections are positioned with the inner limiting membrane uppermost. (A) In a
normal P10 retina, GFAP label concentrated in astrocyte-like cells lying superficial to the
neurons (n) of the ganglion cell layer (gel). (B) In the retina of an animal that survived in
room air for 11 days after 4 days in high (70% to 80%) oxygen, the GFAP-labeled structures
appeared granular and were spread among neuronal somas throughout the gel. (C) In the
retina of an animal that survived 27 days in room air after 4 days in high (70% to 80%)
oxygen, the GFAP-labeled structures appeared more normal. They concentrated internal to
the gel, although some labeling persisted in the gel. (D,E) Bright- and dark-field views of
the labeling of VEGF mRNA in a section adjacent to that in A. The VEGF signal lies superficial
to the neurones (n) of the gel. The arrows in D and E point to the same position in the
section. (F,G) Bright- and dark-field views of the labeling of VEGF mRNA in a section
adjacent to that in B. The VEGF signal is distributed throughout the ganglion cell layer and
is intense over many neurone-like somas. (H) Bright-field view of the labeling of the VEGF
mRNA probe hybridized to a section adjacent to that in C. The VEGF signal is strong over
the neurones of the gel and weak over the more internal layer of retina, where GFAP+
labeling is strong. (I) GFAP labeling in the retina of a 4-day-old animal exposed to high
oxygen for 3 days. The labeling is normal in intensity and in its location internal to the gel.
ipl = inner plexiform layer.
the ganglion cell layer, blurring the normal separation
of GFAP+ elements and neurones (data not shown).
Eleven days after return to room air, GFAP+ elements
at the edge of the spreading vasculature appeared increasingly pale and disorganized and had moved away
from the inner glia limitans, leaving a GFAP-free margin near the inner limiting membrane (ILM) (Fig.
3B). By 27 days after exposure to high oxygen, astrocytes had reinvaded the retina and spread past the
region of vessel growth.7 Seen in section (Fig. 3C),
the distribution of GFAP+ material was more normal
than at 4 to 11 days, concentrating internal to the
ganglion cell layer. These observations confirm our
earlier conclusion7 that functioning astrocytes are absent from the site of vessel formation during much of
the period of proliferative vasculopathy of feline ROP.
Vascular Endothelial Growth Factor Expression
by Neurones in Proliferative Vasculopathy
Figures 3D to 31 show evidence that after the degeneration of astrocytes, VEGF is expressed by neurones. In
normal development, GFAP+ astrocytes are located
superficial to the ganglion cell layer (Fig. 3A). Seen
in an adjacent section, the VEGF signal was also superficial to neurones (n) of the ganglion cell layer (Figs.
3D, 3E), and thus coextensive with the GFAP+ material. At days 4 to 11, the GFAP+ structures are degenerative and mixed with the neurones of the ganglion
cell layer (Fig. 3B), and the VEGF signal is coextensive
with the ganglion cell layer (Figs. 3F, 3G). Although
GFAP labeling is present in the ganglion cell layer,
the structures labeled are granular and disorganized,
confirming our previous evidence7 that this is debris
of degenerating astrocytes. The situation is clearer 27
Downloaded From: http://iovs.arvojournals.org/ on 06/16/2017
days after return to room air, when GFAP labeling
largely has been resegregated to the axon layer (Fig.
3C), internal to the ganglion cell layer, and the VEGF
signal remains strong over neurones in the ganglion
cell layer (Fig. 3H) and appears not to have developed
in the astrocytes lying more superficially. Confirming
Chan-Ling and Stone,20 Figure 31 presents a control
observation that the astrocytes were not affected by
hyperoxia itself. In this case, the astrocytes appeared
normal in the intensity of their staining and in their
superficial location after 3 days in 70% to 80% oxygen.
Where endothelial cells could be discerned in regions of high VEGF expression, they did not appear
to be associated with VEGF signals. We conclude that
after the degeneration of astrocytes during proliferative vasculopathy of feline ROP, VEGF is strongly expressed by non-glial cells, principally by neurones of
the ganglion cell layer.
Preretinal Vessels
The formation of preretinal vessels is a relatively severe abnormality of ROP, reported previously in the
feline model by Chan-Ling et al.7 In the current
study, preretinal vessels were prominent in the animal
returned to room air for 11 days. This is the animal
in which astrocyte breakdown was clearest (Fig. 3B)
and in which VEGF expression was highest (Figs. IF,
2, 3F, 3G). A preretinal vessel is shown in Figure 4A,
which depicts a region of retina near the optic disc.
Astrocytes reentered this part of the retina, as described by Chan-Ling and Stone.8 They are normal
in their location, internal to the inner plexiform layer
(i), and clearly mark the position of the inner limiting
membrane (ILM) (arrow in Figs. 4A, 4C). Preretinal
Investigative Ophthalmology & Visual Science, February 1996, Vol. 37, No. 2
\
.
I ~t 1
;'
I
1
L
C.FAP
r
P
Downloaded From: http://iovs.arvojournals.org/ on 06/16/2017
«+ *••
B:P4/11GFAP
Vascular Endothelial Growth Factor in Retinopathy of Prematurity
297
FIGURE 4. (A) Section of the retina of the animal exposed to hyperoxia (day 4) and then
kept in room air for 11 days. Glial fibrillary acidic protein (GFAP) labeling shows the axon
layer, which is thick because this region is near the optic disc. The GFAP label marks the
inner limiting membrane (ILM), at the level indicated by the arrow. Internal to the ILM is
a profuse formation of vessels (pr). They are not GFAP+. (B) At higher power, the profuseness of the preretinal vessels (pr) is apparent. The GFAP label marks the ILM. (C) The
same area of an adjacent section hybridized with the vascular endothelial growth factor
(VEGF) mRNA probe. (D) The same area of an adjacent section hybridized with the flk-1
mRNA probe. The arrows in C and D indicate the position of the ILM. (E to G) Retinal
vessels from inner layers of the retina of a kitten raised for 4 days in high (70% to 80%)
oxygen and then in room air for 11 days, in sections hybridized with the probe for the flk1 mRNA. In E and F, the signal concentrates over the endothelial cells of a blood vessel.
In G, the signal concentrates over a retinal vessel {lower arrow) and over preretinal vessels
{upper two arrows). (H to J) Adjacent sections from the retina of a kitten raised for 4 days
in high (70% to 80%) oxygen and then in 45% oxygen for 3 days. (H) The distribution of
GFAP+ cells is normal, lying around vessels and along axons superficial to neurons of the
ganglion cell layer (gel). (IJ) Bright- and dark-field views show that, as in the normally
developing retina, the VEGF signal concentrates over structures that lie superficial to the
neurones (n) of the ganglion cell layer.
vessels (pr) lie internal to the ILM and appear to be
free of GFAP labeling. Figure 4B shows the profuseness of preretinal vessels, again confirming ChanLing and Stone.8 Figure 4C shows that the VEGF signal in the preretinal vessels (pr) is low, close to background retinal levels, and is much lower than the intense VEGF signal expressed more peripherally in this
retina (Fig. IF).
Testing the Roles of Vascular Endothelial
Growth Factor and Astrocytes
These observations led us to formulate and test a
model of the role of VEGF and astrocytes in the pathogenesis of ROP. When the neonate is placed in an
oxygen-enriched atmosphere, the inner layers of the
retina become hyperoxic, and that hyperoxia downregulates VEGF at the inner surface, where it is expressed by astrocytes.'' The consequent reduction of
VEGF secretion inhibits vessel growth. When the neonate is returned to room air, the severe hypoxia of
the inner layers of the retina, caused by the lack of
vessels,12 precipitates the degeneration of astrocytes
and induces high levels of VEGF secretion by cells
remaining in the inner layers, principally neurones.
The VEGF causes proliferative vessel growth in the
retina and from the retina into the vitreous humor.
Oxygen brought by the vessels limits the hypoxia and
thereby the vasculopathy. This model was tested in
two ways.
flk-1 Expression
If VEGF is acting as a growth factor to produce the
vessel growth in ROP, its receptor should be expressed
by the growing vessels. Sections adjacent to those hy-
Downloaded From: http://iovs.arvojournals.org/ on 06/16/2017
bridized for VEGF were hybridized for its high-affinity
flk-1 receptor,flk-1was expressed by vessels developing
in normoxia," and during ROP in preretinal vessels
(Figs. 4D, 4G) and in vessels within the retina (Figs.
4E to 4G).
Supplemental Oxygen Therapy Allows Astrocyte
Survival and Prevents Overexpression of
Vascular Endothelial Growth Factor
If high VEGF expression causes the proliferative vasculopathy of ROP, then SOT, which largely eliminates
proliferative vasculopathy,10 should normalize VEGF
expression. This was demonstrated in a kitten raised
for 4 days in 70% to 80% oxygen and then for 3 days
in 45% oxygen, the level shown by Chan-Ling et al10
to normalize the formation of vessels for several days
after exposure to hyperoxia. At the edge of the reforming vessels, astrocytes are normal in morphology
and in their location in the axon layer, internal to the
ganglion cell layer (Fig. 4H), confirming Chan-Ling
et al.10 Further, the expression of VEGF is normal in
its location in the axon layer, superficial to the neurones of the ganglion cell layer (Figs. 41, 4J). Finally,
the expression of VEGF during SOT is relatively normal. The pattern of VEGF expression across the edge
of the developing vessels is compared with the normal
pattern in Figure 2. Peak VEGF expression during
SOT was found lateral to the edge of the developing
vessels, as in the normally developing retina, and was
not elevated; it is suggested in the Discussion that SOT
prevents proliferative vasculopathy by preventing overexpression of VEGF, which occurs if the animal is returned to room air. However, differences are apparent
in VEGF expression between SOT and normal devel-
298
Investigative Ophthalmology & Visual Science, February 1996, Vol. 37, No. 2
opment. The peak signal appears reduced in SOT,
VEGF expression central to the edge of vessels is relatively low, and expression peripheral to the vessels is
relatively high.
DISCUSSION
Evidence is presented of the roles of VEGF and of
astrocytes in the pathogenesis of the feline form of
ROP. The involvement of VEGF in feline ROP can
be stated simply. Inhibition of vessel growth during
hyperoxia used to induce ROP results, the current
study suggests, from a downregulation of VEGF expression induced by hyperoxia, and the proliferative
vasculopathy that follows the return of the kitten to
room air results from an upregulation of VEGF secretion induced by hypoxia of the inner retina caused12
by the lack of vessels. Once an area of retina is revascularized, the oxygen brought by the vessels downregulates VEGF expression toward the low levels found in
the fully vascularized normal retina.
The role of astrocytes in ROP is more complex to
state. Astrocytes are the normal source of the VEGF
signal that drives the formation of the inner layer of
vasculature." They are not affected morphologically
by hyperoxia, though their expression of VEGF is downregulated. Astrocytes are involved more critically in
the pathogenesis of the proliferative vasculopathy that
follows the return of the animal to room air. Retinal
astrocytes degenerate under the intense hypoxia that
follows the return to room air.6 Their absence has two
consequences. First, as noted by Chan-Ling et al,7 the
proliferative vessels lack barrier properties normally
induced by astrocytes21'22 until a second wave of astrocytes reaches them from the optic disc. Second, the
high levels of VEGF expression induced by the intense
hypoxic stimulus are associated with cells, particularly
neurones, that do not secrete the factor in normal
development.''
Current data confirm, but also differ from, the
observations reported recendy by Pierce and colleagues,23 who reported an upregulation of VEGF expression by the mouse retina during proliferative vasculopathy induced by hyperoxia during development.
Miller et al24 reported an upregulation of VEGF expression by the adult monkey retina when hypoxia
was induced by experimental venous occlusion. Our
results show a comparable upregulation of VEGF in
the developing cat retina during hyperoxia-induced
retinopathy. At a more detailed level, however, the
current results show features in the cat not reported
for the mouse or monkey. In particular, we note a
change in the cell class expressing VEGF at the inner
surface of the retina, from astrocytes in normal development" to neurones in the proliferative phase of
feline ROP. Upregulation of VEGF expression in neu-
Downloaded From: http://iovs.arvojournals.org/ on 06/16/2017
rones may occur independently of the degeneration
of astrocytes but represents a significant change in the
cellular source of VEGF.
Preretinal Vessels: Evidence of the Paracrine
Action of Vascular Endothelial Growth Factor
According to our results, preretinal vessels do not express VEGF strongly, but they do express the highaffinity receptor for VEGF termed flk-1.Nl5 This observation makes it possible to comment on two questions
related to preretinal vessels: Why are they profuse?
What are the preconditions for their formation?
The profuseness of preretinal vessels is a paradox.
Why do vessels entering the acellular vitreous humor
become more profuse than vessels in the retina, which
the major sink for oxygen? The answer may lie in the
paracrine action of VEGF, i.e., in its secretion by one
class of cell, normally astrocytes, to act on nearby endothelial cells." Preretinal vessels are profuse, we suggest, because of a lengthening of the feedback path
from functioning vessels to the cells that secrete VEGF.
In normally developing retina, vessels form directly
adherent to the glial cells that secrete VEGF." Oxygen
delivered by the vessels must diffuse only across the
minimal distance from the vessel to the glial cell to
which it is adherent. Preretinal vessels presumably are
induced by VEGF that has accumulated in the vitreous
humor. Such accumulation has been demonstrated
in one of the neovascularizing diseases of the retina,
diabetic retinopathy.23'25 Once formed, preretinal vessels are some distance from the cells that secrete
VEGF, and the oxygen they deliver must diffuse over
a relatively long distance to reach those cells. This
distance may weaken the feedback link so that preretinal vessels become profuse before the oxygen they
diffuse reaches the retina in sufficient concentration
to limit VEGF expression.
In what conditions do vessels grow from the retina
into the vitreous humor? Clinical experience suggests
that those conditions include hypoxia of the inner
retina, in retinopathy of prematurity,"3 or in diabetic
retinopathy.25 We make the more specific suggestion
that preretinal vessels form when intense hypoxia
causes the degeneration of astrocytes and the strong
expression of VEGF by other cells, particularly neurones (above). Because astrocytes are part of the inner
glia limitans of the retina,18 the glia limitans presumably is damaged. With VEGF expression high and the
glia limitans breached, VEGF may diffuse into the vitreous humor, inducing vessel growth away from the
retina. Ashton's4 insight that proliferative vasculopathy is caused by exaggeration of the mechanisms that
cause normal vascularization of the retina is thus
largely correct; preretinal vessels may form, as do normal vessels, by the hypoxia-induced secretion of VEGF
and may be proliferative because of the high level of
Vascular Endothelial Growth Factor in Retinopathy of Prematurity
VEGF expression. There appears, however, to be a
qualitative element in the causation of preretinal vessels, i.e., the breakdown of astrocyte function and the
substitution of neurones for astrocytes as the class of
cell expressing VEGF.
These hypotheses emphasize the importance of
the integrity of astrocytes for the normal development
of retinal vessels. That importance has been demonstrated empirically in the creation of regimes of supplemental oxygen therapy for the management of retinopathy of prematurity. 10 The current hypotheses provide a cellular and molecular basis for planning and
testing improved treatment of retinas threatened by
n e ovascularization.
Key Words
astrocytes, retinal development, retinal vasculature, retinopathy of prematurity, vascular endothelial growth factor
(VEGF)
Acknowledgments
The authors thank Dr. Ping Hu for technical assistance.
References
1. Campbell K. Intensive oxygen therapy as a possible
cause of retrolental fibroplasia: A clinical approach.
MedJAustr. 1951; 48-50.
2. Patz A. Oxygen studies in retrolental fibroplasia: V:
The effect of rapid vs gradual withdrawal from oxygen
on the mouse eye. Arch Ophthalmol. 1957;57:727-729.
3. Ashton NA. Experimental retrolental fibroplasia. Ann
ReuMed. 1957; 8:441-454.
4. Ashton N. Oxygen and the growth and development
of retinal vessels. Am J Ophthalmol. 1966; 62:412-435.
5. Ashton N. Pathological basis of retrolental fibroplasia.
Br J Ophthalmol. 1954;38:385-396.
6. Ashton N, Ward B, Serpell G. Role of oxygen in the
genesis of retrolental fibroplasia. Br J Ophthalmol.
1953;37:513-520.
7. Chan-Ling T, Tout S, Hollander H, Stone J. Vascular
changes and their mechanisms in the feline model of
retinopathy of prematurity. Invest Ophthalmol Vis Set.
1992;33:2128-2147.
8. Chan-Ling T, Stone J. Degeneration of astrocytes in
feline retinopathy of prematurity causes failure of the
blood-retinal barrier. Invest Ophthalmol Vis Sci.
1992; 33:2148-2159.
9. Chan-Ling T, Gock B, Stone J. The effect of oxygen
on vasoformative cell division: Evidence that 'physiological hypoxia' is the stimulus for normal retinal vasculogenesis. Invest Ophthalmol Vis Sci. 1995; 36:1201214.
10. Chan-Ling T, Gock B, Stone J. Supplemental oxygen
therapy: Basis for noninvasive treatment of retinopathy of prematurity. Invest Ophthalmol Vis Sci.
1995;36:1215-1230.
Downloaded From: http://iovs.arvojournals.org/ on 06/16/2017
299
11. Stone J, Itin A, Alon T, Gnessin H, Chan-Ling T,
Pe'er Y, Keshet E. Development of retinal vasculature
is mediated by hypoxia-induced vascular endothelial
growth factor (VEGF) expression by neuroglia./iWwrosd. 1995; 15:4738-4747.
12. Chan-Ling T, Stone J. Retinopathy of prematurity:
Its origins in the architecture of the retina. Prog Retinal
Res. 1993; 12:155-178.
13. Shweiki D, Itin A, Softer D, Keshet E. Vascular endothelial growth factor induced by hypoxia may mediate
hypoxia-initiated angiogenesis. Nature. 1992; 359:843845.
14. Millauer B, Wizigmann-Voos S, Schniirch H, et al.
High affinity VEGF binding and developmental expression suggest Flk-1 as a major regulator of vasculogenesis and angiogenesis. Cell. 1993;72:835-846.
15. Terman BI, Carrion ME, Kovacs E, Rasmussen BA,
Eddy RL, Shows TB. Identification of a new endothelial cell growth factor receptor tyrosine kinase. Oncogene. 1991;6:1677-1683.
16. Chan-Ling T, Halasz P, Stone J. Development of retinal vasculature in the cat: Processes and mechanisms.
CurrEyeRes. 1990;9:459-478.
17. Chan-Ling T, Stone J. The development of astrocytes
in the cat retina: Evidence of migration from the optic
nerve. Dev Brain Res. 1988; 44:73-85.
18. Hollander H, Makarov F, Dreher Z, van Driel D,
Chan-Ling T, Stone J. Functions of the macroglia of
the retina: The sharing and division of labour between
astrocytes and Miiller cells. / Comp Neurol. 1991;
313:587-603.
19. Stone J, Dreher Z. Relationship between astrocytes,
ganglion cells and vasculature of the retina. / Comp
Neurol. 1987; 255:35-49.
20. Chan-Ling T, Stone J. Factors determining the migration of astrocytes into the developing retina: Migration does not depend on intact axons or patent vessels,
/ Comp Neurol. 1991;303:375-386.
21. Janzer RC, Raff MC. Astrocytes induce blood-brain
barrier properties in endothelial cells. Nature.
1987;325:253-257.
22. Tout S, Chan-Ling T, Hollander H, Stone J. The role
of Miiller cells in the formation of the blood retinal
barrier. Neurosdence. 1993;55:291-301.
23. Pierce EA, Avery RT, Foley ED, Aiello LP, Smith L.
Vascular endothelial growth factor/vascular permeability factor expression in a mouse model of retinal
neovascularisation. Proc Natl Acad Sci USA. 1995;
92:905-909.
24. Miller JW, Adamis AP, Shima DT, et al. Vascular endothelial growth factor/vascular permeability factor is
temporally and spatially correlated with ocular angiogenesis in a primate model. Am J Pathol. 1994;
145:574-584.
25. Adamis AP, Miller JW, Bernal MT, et al. Increased
vascular endothelial growth factor levels in the vitreous of eyes with proliferative diabetic retinopathy. Am
J Ophthalmol. 1994; 118:445-450.