The ephrin ligands (Efn) and their
Eph receptors (Eph) are membrane-
bound and interact upon cell-to-cell
contact, allowing bidirectional signal-
ling (Himanen et al. 2007; Arvanitis &
Davy 2008; Lackmann & Boyd 2008).
Up to now, a total of nine ephrins
and 16 Ephs have been divided into
groups A and B. In general, ephrinA
proteins interact with EphA receptors
and ephrinB proteins interact with
EphB receptors but exceptions to this
rule exist. It is thought that cluster
formation of ephrins or Eph receptors
at the cell surface are a prerequisite
for successful signal transduction.
The ephrin⁄Eph ligand receptor sys-
tem was first investigated in the con-
text of axon guidance (Butler & Tear
2007; Egea & Klein 2007; Reber et al.
2007). But during recent years, their
importance for angiogenesis became
evident (Adams & Klein 2000; Cheng
et al. 2002; Augustin & Reiss 2003;
Brantley-Sieders & Chen 2004; Hero-
ult et al. 2006; Zhang & Hughes 2006;
expressed on arteries, while EphB4 is
expressed on veins (Gerety et al.
1999). Mice deficient for either gene
show a lethal phenotype at early
embryonic stages with severe defects
in vascular development (Wang et al.
1998). Similar defects are observed
Therapeutic interference with
EphrinB2 signalling inhibits
Christoph Ehlken,1Gottfried Martin,1Clemens Lange,1
Eleni G. Gogaki,1Ulrike Fiedler,2Florence Schaffner,2
Lutz L. Hansen,1Hellmut G. Augustin2,* and Hansju ¨ rgen T. Agostini1
1University Eye Hospital, Freiburg, Germany
2Department of Vascular Biology, Tumor Biology Center Freiburg, Germany
Purpose: To investigate whether EphrinB2 (EfnB2) or EphB4 influence retinal
angiogenesis under physiological or pathological conditions.
Methods: Using the mouse model of oxygen-induced proliferative retinopathy
(OIR), the expression of EfnB2, EphB4, vascular endothelial growth factor
(VEGF), VEGFR1 and VEGFR2 was quantified by quantitative polymerase
chain reaction (qPCR) and localized in EfnB2- and EphB4-lacZ mice. Angio-
proliferative retinopathy was manipulated by intravitreal injection of dimeric
EfnB2 and monomeric or dimeric EphB4.
Results: Dimeric EphB4 (EphB4-Fc) and EfnB2 (EfnB2-Fc) enhanced
hypoxia-induced angioproliferative retinopathy but not physiological angiogen-
esis. Monomeric EphB4 (sEphB4) reduced angiogenesis. The messenger RNA
(mRNA) level of EfnB2 increased significantly in the hyperoxic phase (P7–
P12), while EphB4, VEGF, VEGFR1 and VEGFR2 showed a significant – up
to fivefold – increased expression at P14, the start of morphologically visible
vasoproliferation caused by relative hypoxia.
Conclusion: The ephrin⁄Eph system is involved in angioproliferative retinopa-
thy. Stimulation of EphB4 and EfnB2 signalling using EfnB2-Fc and EphB4-
Fc, respectively, enhanced hypoxia-induced angiogenesis. In contrast, sEphB4
inhibited hypoxia-induced angiogenesis. Therefore, angiogenesis is enhanced by
signalling through both EphB4 (forward) and EfnB2 (reverse). The distinction
in the expression kinetics of EphB4 and EfnB2 indicates that they govern two
different signalling pathways and are regulated in diverse ways. sEphB4 might
be a useful drug for antiangiogenic therapy.
Key words: angiogenesis – molecular biology – retinal diseases – retinopathy of prematurity
Acta Ophthalmol. 2011: 89: 82–90
ª 2009 The Authors
Journal compilation ª 2009 Acta Ophthalmol
*Current address: Joint Research Division Vascular Biology of the Medical Faculty Mannheim,
University of Heidelberg and the German Cancer Research Centre, Heidelberg, Germany
Acta Ophthalmologica 2011
in EphB2⁄EphB3 double knockout
mutants, while the single EphB2 or
EphB3 mutants have no vascular phe-
notype (Adams et al. 1999).
In this study, we used the soluble,
monomeric EphB4 (sEphB4), which
contains only the extracellular part of
the receptor. It binds to the corre-
sponding membrane-bound ligand, but
it does not elicit signalling (Martiny-
Baron et al. 2004; Kertesz et al. 2006).
Thus, it interferes with membrane-
bound receptor signalling and might
therefore have an inhibitory effect. For
functional dimerization, the extracellu-
lar domain of the receptor or ligand,
respectively, was fused to the Fc part
of immunoglobulin G (IgG) (EphB4-
Fc or dimeric EphB4, EfnB2-Fc or
dimeric EphrinB2). The dimeric recep-
tor bodies stimulate signalling of their
binding partners (Fuller et al. 2003;
Hamada et al. 2003). EfnB2-Fc inhib-
its but EphB4-Fc induces sprouting
angiogenesis in an in vitro angiogenesis
assay (Fuller et al. 2003; Martiny-
Baron et al. 2004).
In ophthalmic disease, immunohis-
tochemical staining for EfnB2, EphB2
and EphB3 expression was found on
endothelial cells of fibroproliferative
membranes from human eyes suffering
from proliferative diabetic retinopathy
or retinopathy of prematurity (Umeda
et al. 2004). Human retinal endothelial
cells (HREC) express EfnB2 but only
low levels of EphB4 (Steinle et al.
2003). In these cells, stimulation of
EfnB2 by an EphB4-Fc fusion protein
resulted in phosphorylation of PI3K,
signalling via the mitogen-activated
protein kinase (MAPK) pathway is
involved. Similarly, sEphA2 inhibits
retinal neovascularization in the oxy-
gen-induced proliferative retinopathy
(OIR) mouse model (Chen et al.
2006). In age-related macular degener-
ation an over-expression of EphA7 in
likely associated with late-stage dis-
ease, was shown (Martin et al. 2004).
More than ten years ago, Smith
et al. (1994) introduced the mouse
model of oxygen-induced retinopathy.
It is a well-established tool to inves-
aches (Unsoeld et al. 2004; Agostini
et al. 2005; Maier et al. 2005; Lange
et al. 2007). In addition, physiological
retinal vascularization can be followed
easily in mice, because they are born
with only a basic vascular network
around the papilla and develop their
definitive vascular network during the
first three postnatal weeks. In this
EphB4 expression during the normal
development of the murine retinal vas-
culature and under pathological con-
ditions, the localization of EfnB2 and
EphB4 in the murine eye, and the
EfnB2 and EphB4 on retinal vascular-
ization in vivo.
Materials and Methods
C57BL⁄6J mice (Charles River Labo-
ratories, Hamburg, Germany) were
used. All animal procedures adhered to
the animal care guidelines of the Insti-
tute for Laboratory Animal Research
(Guide for the Care and Use of Labo-
ratory Animals) in accordance with the
Association for Research in Vision and
Ophthalmology (ARVO) Statement for
the Use of Animals in Ophthalmic and
Vision Research and were approved by
the animal welfare committee of the
University of Freiburg.
OIR mouse model
The OIR mouse model and the scor-
ing system to evaluate retinopathy in
mice were described by Unsoeld et al.
(2004). Briefly, 7-day-old mice (P7)
were incubated in 75% oxygen for
5 days and then returned to normal
air. At P12, 2 ll of 0.3 lm EfnB2-Fc
(40 ng), EphB4-Fc (66 ng, both from
R&D Systems, Wiesbaden, Germany),
1.6 lm sEphB4 (Martiny-Baron et al.
2004) (200 ng) or 0.3 lm Fc (16 ng)
were injected into the right eye, while
the left eye received a 2 ll injection of
phosphate-buffered saline (PBS) only.
After five additional days, at P17,
mice were perfused with fluorescein
retinae were flat-mounted. Features of
the proliferative retinopathy – includ-
ing vascular tuft or cluster formation,
central avascular zone, blood vessel
tortuosity and peripheral vasculariza-
tion – were evaluated in individual
pairs based on a modified scoring
system introduced by Higgins and col-
leagues (Higgins et al. 1999; Yossuck
et al. 2000) in a masked manner, as
detailed in Fig. 1. Differences in reti-
nopathy scores between treated and
control eyes were evaluated statisti-
cally using the Wilcoxon signed ranks
test using R [http:⁄ ⁄www.r-project.org⁄
with the command wilcoxsign_test (V1
V2,data = dR,
‘exact’)]. Differences with p < 0.05
were judged to be significant.
Intravitreal injections were performed
(Zeiss, Jena, Germany) with glass pip-
ettes with a diameter of approximately
150 lm at the tip (Unsoeld et al.
2.5% of isoflurane in oxygen (Vapor
19.3;Dra ¨ ger,Lu ¨ beck,
Each eye was punctured at the upper
nasal limbus and a volume of 2 ll of
test or control solution (as described
earlier) was injected. The pipette was
kept in place for about 10 seconds to
allow diffusion of the solution. Previ-
ously, it was shown that a labelled
substance was distributed well in the
eye after intravitreal injection (Agos-
tini et al. 2005). As seen in humans
reflux at the puncture site could be
observed. The mouse eye is 3 mm in
diameter, from which a volume of
14 ll is calculated. Although the vol-
ume of the murine vitreous is difficult
to measure, sections are estimated to
be 2 ll. This is about the same vol-
ume as for the retina and is in accor-
dance with measurements in the rat
(Dureau et al. 2001). Consequently, if
2 ll of a substance is injected into the
vitreous, it will be diluted to half of
the original concentration by diffusion
within the vitreous and to one third
when spreading evenly into vitreous
Quantitative polymerase chain reaction
Quantitative polymerase chain reac-
tion (qPCR) was performed using
standard methods, as described by
Martin et al. (2004). Whole retinae
from mice treated with oxygen or from
control mice were prepared for RNA
isolation using the RNeasy Kit (Qia-
gen, Hilden, Germany). cDNA was
generated using Superscript II reverse
transcriptase (Invitrogen, Karlsruhe,
Acta Ophthalmologica 2011
(RT)-PCR was performed using the
SYBR green assay (Eurogentec, Ko ¨ ln,
Germany). Primers were designed to
span an intron of the target gene for
distinction from DNA. For primer
sequences see Table 1. The PCR reac-
tion mix contained 1.5 mm MgCl2, and
the annealing temperature was 56?C in
an ABI 7700 sequence detection sys-
tem (Applied Biosystems, Weiterstadt,
checked on agarose gels and by disso-
ciation curve analysis for single bands.
sequencing. The ABI 7700 software
Efficiency was calculated by the Lin-
RegPCR program (Ramakers et al.
calculation were determined by the
REST384 program (Pfaffl et al. 2002)
(both at http:⁄ ⁄www.gene-quantifica-
tion.info⁄). Ratios for each time-point
and treatment were calculated as mean
from at least three independent mea-
surements using retinae from 13–22
mice (3–5 l). mRNA levels of b-actin
(Actb) were used as reference.
We used EfnB2-lacZ and EphB4-lacZ
mice (Wang et al. 1998; Gerety et al.
1999) (Charles River Laboratories).
LacZ staining was based on the proce-
dure of Lobe et al. (1999). Eyes were
fixed for 4 h in 0.2% glutaraldehyde,
50 mm ethylene glycol tetraacetic acid
(EGTA), pH 7.3, 100 mm MgCl2 in
PBS, pH 7.3 on ice. Samples were
washed three times for 15 min in wash
buffer [2 mm MgCl2, 0.01% sodium
(NP-40) in PBS, pH 7.3] and stained
Table 1. Primers.
Gene Forward primerreverse primerFragment length (bp)
Actb, b-actin; EfnB2, EphrinB2; VEGF, vascular endothelial growth factor; ICAM1, intercellu-
lar adhesion molecule 1.
Fig. 1. Retinal vascularization. (A, B) Scoring system used for the evaluation of retinopathy. (1) Central avascular zone (A), one point for every
field that is more than 50% avascular, maximum 13 points. (2) One point for every field containing at least one large cluster of blood vessel tufts
(A), maximum 13 points. (3) Blood vessel tufts (B), one point for every field containing at least one blood vessel tuft, maximum 29 points. (4)
Tortuosity of the vessels, maximum six points. All points were totalled to make the final score (maximum 61 points). Images are of oxygen-
induced proliferative retinopathy (OIR) controls without injection. (C) Typical retinal flat mounts from the experiments shown in Fig. 4 prepared
after perfusion with fluorescein isothiocyanate (FITC)-dextran on P17. A large central avascular zone (asterisk) is formed in the OIR retina
(retinopathy score 15), which is filled in partially by neovascularization in the OIR, sEphB4 retina (score 8). Neovascular tufts and clusters
(arrow) are formed, especially at the border of the avascular zone, as seen in OIR and enhanced by dimeric EphrinB2 (EfnB2-Fc) (score 23).
Acta Ophthalmologica 2011
in 0.5 mg⁄ml X-gal, 5 mm potassium
ferricyanide in wash buffer at 30?C
overnight with protection from light.
Eyes were then embedded in paraffin,
sectioned and mounted on slides for
Vascular endothelial growth factor
Mice were kept under physiological
conditions. On P13, they received an
intravitreal injection of 2 ll of PBS
containing 100 ng of vascular endothe-
lial growth factor (VEGF)-A164(R&D
Systems, Wiesbaden, Germany). The
control eye received PBS only. RNA
was isolated from the retina 6 h later,
as described earlier. qPCR was per-
formed for intercellular adhesion mole-
cule 1 (ICAM)1, VEGF, EfnB2 and
EphB4. For primer
Table 1. The expression in the VEGF-
treated eye was compared to the con-
trol eye receiving a buffer injection.
Kinetics of EfnB2, EphB4, VEGF,
VEGFR1 and VEGFR2 expression in the
As shown in Fig. 2, mRNA of EphB4
was expressed constantly during phys-
iological retinal development. Oxygen
exposure altered neither the expression
level nor the kinetics of EphB4. Only
at the time of the first visible angio-
proliferativechangesin the OIR
model at P14, 2 days after return to
normal oxygen, was there a statisti-
expression of EphB4. The expression
of EfnB2 was different. It showed a
twofold significant increase towards
the end of the high-oxygen phase
and returned to normal levels after-
For comparison, the mRNA levels
of VEGF, VEGFR1 and VEGFR2
were determined. During physiological
retinal development, VEGF expression
exposure in the OIR model, VEGF
expression did not increase until P10.
The increase first observed at P12
peaked at P14 (fivefold increase) and
returned slightly afterwards. Under
was expressed almost constantly and
from P13 with a significant peak at
P15. In OIR mice, VEGFR2 increased
threefold during hyperoxia, decreased
after return to room air and showed a
significant peak at P14. VEGFR1
increased constantly up to 10-fold from
retinal development. In OIR mice, it
also increased as under physiological
conditions but there was a sharp, signif-
icant peak at P14. Interestingly, the
increase of VEGF expression started
earlier thanthatofVEGFR1, VEGFR2
and EphB4 in the OIR model (compare
datafor P13inFig. 2).
Retinal localization of EfnB2 and EphB4
The expression pattern of EfnB2 and
EphB4 in the eye was determined in
heterozygous mice that had the EfnB2
or EphB4 gene, respectively, replaced
by lacZ. EphB4 showed expression in
assumed to be veins (Fig. 3). In addi-
tion, EphB4 was found on preretinal
neovascularizations after OIR treat-
ment. EfnB2 was expressed in the
inner layer of the choroid as well as in
small retinal vessels. In addition, there
was a rather homogenous expression
in the retinal ganglion layer and inner
The peak of EphB4 expression fol-
lowed a rise of VEGF when mice
returned from oxygen exposure to nor-
mal room air (Fig. 2). Hypothesizing
Fig. 2. Kinetics of gene expression in the retina. Expression of EphrinB2 (EfnB2), EphB4, vas-
cular endothelial growth factor (VEGF), VEGFR1 and VEGFR2 during physiological angio-
genesis [no oxygen-induced proliferative retinopathy (OIR), squares] and after oxygen exposure
between P7 and P12 (OIR, triangles) measured by quantitative polymerase chain reaction
(qPCR). A peak is found at P14 for EphB4, VEGF, VEGFR1 and VEGFR2 in the OIR sche-
dule. EfnB2 showed an increased expression during hyperoxia. No proteins were injected. The
grey area indicates the period of elevated oxygen concentration. Values are related to b-actin
and to the expression level of each gene at P5. Error bars indicate standard error. Significant
differences between OIR and no OIR pairs are marked with a black asterisk.
Acta Ophthalmologica 2011
recombinant VEGF164 was used for
intravitreal injection under physiologi-
cal conditions. At P13, VEGF164was
injected into one eye and buffer in the
control eye. ICAM1 mRNA in the
retina increased fivefold compared to
the individual control eyes, indicating
that the injected VEGF was active.
(Table 2), indicating that there is no
direct regulation by VEGF.
Intravitreal injection of dimeric or
monomeric EfnB2 and EphB4 under
physiological conditions or in the OIR
Under physiological conditions with-
out oxygen exposure, the retinopathy
scores in mice were at the expected
low levels. A typical retina is shown in
Fig. 1C. As summarized in Fig. 4 and
Table 3, intravitreal application of
not change this low level [median
retinopathy score (mRS): EfnB2, 0.5;
EphB4, 3] compared to the partner eye
treated with an equal volume of buffer
solution (mRS: EfnB2, 1; EphB4, 2).
EphB4-Fc did not change physiologi-
cal retinal angiogenesis significantly.
retinopathy scores. The median score
in the EfnB2 control group was 9, in
the EphB4 group 13 as shown in
Fig. 4 and Table 3. This higher level
of angioproliferative retinopathy was
Fig. 3. EphrinB2 (EfnB2) and EphB4 expression in the eye. (A) Adult mice. Expression patterns were investigated in adult mice having EfnB2 or
EphB4 genes, respectively, replaced by lacZ. EphB4 was expressed (blue staining) in some of the retinal vessels and capillaries of both the inner
(rgc) and outer (op) retinal vessel layer as well as in the choroid. The two images show different parts of the same eye. EfnB2 was found in the
choroid as well as in retinal vessels. In addition, EfnB2 was expressed evenly in the retinal ganglion cell layer and in the inner plexiform layer.
Retinal ganglion cell layer (rgc), inner plexiform layer (ip), inner nuclear layer (in), outer plexiform layer (op), outer nuclear layer (on), photore-
ceptors (pr), retinal pigment epithelium (rpe), choroid (ch). All images were taken with differential interference contrast (DIC) optics. Scale bars
correspond to 50 lm. (B) P19 oxygen-induced proliferative retinopathy (OIR) mice. Expression pattern of EphB4 in P19 OIR mice. Serial sections
show intense EphB4 expression in a preretinal neovascularization (see arrow).
Acta Ophthalmologica 2011
increased further by the application of
EfnB2-Fc or EphB4-Fc (mRS: EfnB2,
12; EphB4, 22). Injection of a control-
Fc alone did not result in a significant
change of retinopathy. Corresponding
to the increase by EphB4-Fc, intravi-
treal application of sEphB4 decreased
the retinopathy score (mRS: treated, 9;
control, 12). The statistical significance
in our experiments was not changed if
only tufts and clusters were scored as
indicators for retinal angioprolifera-
tion. This shows that sEphB4 is an
inhibitor of hypoxia-induced angiogen-
esis in the retina, indicating that
EphB4-mediated EfnB2 signalling is
essential for angiogenesis in the retina.
Claxton and Fruttiger showed that
EfnB2 is repressed in retinal vessels
when mice are kept under hypoxic con-
ditions (10% oxygen) (Claxton & Frut-
physiological postnatal growth of the
retinal vessels under normoxic condi-
tions is not associated with a significant
change in the expression of EfnB2 or
EphB4 in mice, as shown in Fig. 2.
However, there was a twofold increase
of EfnB2 expression during the incuba-
tion at 75% oxygen in the OIR model
(Fig. 2, P10). During this phase of oxy-
gen exposure the central avascular zone
reaches its maximal size. Oxygen expo-
sure also leads to vessel regression.
Thus, EfnB2 expression is affected by
the oxygen concentration, leading to
upregulation in hyperoxia and repres-
sion in hypoxia. EfnB2 is expressed in
retinal arteries (Claxton & Fruttiger
2005), although additional expression
in the retinal ganglion cell layer and
inner plexiform layer was observed in
the lacZ-stain (Fig. 3). Therefore, we
cannot define whether the increased
EfnB2 expression during hyperoxia
originates from the vessels or from the
neuronal cells of the inner retina.
In contrast, when looking at the
ischaemia-induced proliferative phase
in the OIR model 2–5 days after
return to room air (P14–P17), we
found a peak expression of EphB4
whereas EfnB2 remained at a physio-
logical level. This indicates that the
respond differently to oxygen. In a
similar investigation, the authors did
Table 2. Expression of intercellular adhesion
molecule 1 (ICAM1), vascular endothelial
growth factor (VEGF), EphrinB2 (EfnB2)
and EphB4 after VEGF injection.
5.0 ± 1.6
1.0 ± 0.14
0.99 ± 0.22
0.79 ± 0.15
On P13, mice (not treated with oxygen) were
injected with 2 ll (100 ng) of VEGF into one
eye and phosphate-buffered saline in the con-
trol eye. After 6 h, RNA was isolated and
measured by quantitative polymerase chain
reaction. Values indicate the ratio of treated
eye⁄control eye and are related to b-actin
(Actb) as a standard.
EfnB2-Fc, No OIR
EphB4-Fc, No OIR
0 1020 30 40
Fig. 4. EphrinB2 (EfnB2) and EphB4 effects on angiogenesis in the normal retina and in oxy-
gen-induced retinopathy. Each dot represents both eyes of one mouse: the x value of a dot indi-
cates the retinopathy score of the treated eye, its y value the score of the control eye that
received phosphate-buffered saline (PBS). Dots above the diagonal mean reduced retinopathy,
and the median is marked by an asterisk. While retinopathy is low in the physiological develop-
ment, it is high after oxygen treatment. This high level is increased further by EfnB2-Fc or
EphB4-Fc while it is reduced by sEphB4. Fc injection results in a non-significant reduction of
the retinopathy score.
Acta Ophthalmologica 2011
not find an elevated EfnB2 expression
during hyperoxia, and the EphB4 peak
at P14 appeared only as a modulation
(Zamora et al. 2005). This difference
can be explained by technical differ-
ences using semi-quantitative RT-PCR
(Zamora et al. 2005), which might not
be sensitive enough to detect small dif-
ferences in expression levels, compared
to the quantitative PCR used in this
To compare the peak expression of
EphB4 and EfnB2 with known mark-
ers of hypoxia-driven angiogenesis,
the mRNA levels of VEGF, VEGFR1
and VEGFR2 were quantified during
physiological retinal development and
in the OIR model. A peak expression
similar to that of EphB4 was observed
for all three factors (Fig. 2). At the
end of the high-oxygen phase, VEGF
expression increased and reached its
highest level at P14. This is in accor-
dance with other reports (Ozaki et al.
1999; Robinson et al. 2001; Yao et al.
2005). During hyperoxia, a constant
level of VEGF was found while the
control mice without oxygen exposure
showed a twofold increase in VEGF
mRNA (Fig. 2, P10). These findings
are in accordance with those of Pierce
and colleagues (Pierce et al. 1996) and
confirm the well-described role of this
VEGF-mediated retinal angiogenesis
during physiological development.
Because the expression kinetics of
EphB4 all show a peak at P14 in the
OIR model, we wondered whether the
increase of EphB4 might be linked to
the increased expression of VEGF
that preceded the upregulation of the
other factors in the OIR model. How-
ever, a single intravitreal injection of
VEGF at P13 under physiological
conditions with ICAM1 expression as
a positive control did not show any
effect on either EphB4 or EfnB2
expression (Table 2).
Under normoxic conditions, intravi-
treal injection of dimeric EfnB2 or
EphB4 did not change the physiologi-
cal development of the retinal vascula-
alterations in the morphology of the
retinal vessels, and the retinopathy
scores remained at low levels (Fig. 4).
The outcome was different in OIR
mice when EfnB2-Fc or EphB4-Fc was
injected intravitreally at P12. After
induction of the proliferative retinopa-
thy in the OIR model, both proteins
enhanced retinopathy with increased
vasoproliferation. Control Fc alone
reduced the retinopathy slightly but
not significantly, as was also shown by
Zamora and colleagues (Zamora et al.
2005). However, using the monomeric
isoforms for intravitreal injection at
P12, sEphB4 showed a marked reduc-
tion of proliferative retinopathy. This
effect parallels the observation in the
spheroidal angiogenesis assay: EphB4-
Fc enhanced VEGF-induced sprouting,
while sEphB4 abolished the enhance-
ment of EphB4-Fc (Martiny-Baron
et al. 2004). Similarly, sEphB4 reduces
migration of choroidal endothelial cells
and their tube formation (He et al.
2005). In accordance with our findings,
sEphB4 was shown to reduce laser-
induced choroidal neovascularization
in the rat (He et al. 2005). EphB4 and
EfnB2 are expressed in the laser lesion.
Furthermore, EfnB2-Fc strongly pro-
moted angiogenesis in corneal neovas-
cularization and Matrigel plug assays
in mice (Maekawa et al. 2003).
In contrast to these results and the
in vitro data obtained by the spheroidal
angiogenesis assay, Zamora and col-
retinopathy with the dimeric forms
(EfnB2-Fc or EphB4-Fc) in the OIR
model (Zamora et al. 2005). After hav-
ing injected EfnB2-Fc or EphB4-Fc at
P12 and P14, they evaluated the num-
ber of preretinal nuclei in the neovascu-
lar tufts. In the OIR protocol used
injection is performed on P12 only.
Puncturing an eye can reduce neovas-
cularization [the authors’ unpublished
results and Stitt et al. (2004)]. Because
Zamora and colleagues injected twice,
this could mask the proangiogenic
effect of EfnB2-Fc or EphB4-Fc. In a
comparable experiment, the reduction
of retinopathy that was detected by
PTK787 on P12 (Maier et al. 2005) dis-
repeated for both the treated and the
control eye on P14 (unpublished data),
demonstrating that double injection is
not comparable to a single injection. In
the study of Zamora et al., up to four
times more EfnB2-Fc or EphB4-Fc was
used per injection. However, all dose–
response curves in cell adhesion or tube
formation experiments did not result in
a reverse effect at low concentrations
compared to higher ones (Fuller et al.
2003; Hamada et al. 2003; He et al.
2005). Thus, different concentrations
of EfnB2-Fc or EphB4-Fc are unlikely
to explain the opposite results. Most
importantly, however, our results and
those of Zamora et al. show that the
inducible effect on retinal angiogenesis
EphB4 system does not depend on
Table 3. EphrinB2 (EfnB2) and EphB4 signalling enhances angiogenesis in the oxygen-induced
proliferative retinopathy (OIR) model.
Without O2(physiological)With O2(OIR)
EfnB2-FcEphB4-Fc EfnB2-Fc EphB4-FcsEphB4 Fc
Control (PBS injection)
1825 27 23 2625
n, number of animals; score treated, median score of treated eyes; PBS, phosphate-buffered sal-
ine; score control, median score of control eyes treated with PBS; IQR, interquartile range;
p-value, p-value of the Wilcoxon signed rank test concerning the difference between the treated
and control eye of each animal. p < 0.05 is significant at the 95% level.
The low retinopathy scores under physiological conditions (without O2) are not increased by
dimeric ephrins but by oxygen exposure in the OIR model (compare control scores). EfnB2-Fc
or EphB4-Fc further increased the retinopathy while it was reduced by sEphB4.
The Wilcoxon signed rank test analyses the differences between two related measurements, in
this case the retinopathy scores of the treated and control eyes of each animal. The p-value for
tufts was calculated to get results comparable to those where the extraretinal nuclei were
counted from histological slices.
Acta Ophthalmologica 2011
primary target. Both have an effect in
the same but not the opposite direction
of angiogenesis, indicating that not
either EfnB2 signalling or EphB4 sig-
nalling but both are involved in vessel
Blocking angiogenesis by sEphB4 is
not complete. Possible reasons are
that the dose of sEphB4 used –
although the highest possible – was
insufficient, that sEphB4 is not stable
or that it is removed too quickly from
the tissue. Intravitreal injection of the
tyrosin kinase inhibitor PTK787, for
example, blocks retinopathy only par-
tially (Maier et al. 2005). Complete
inhibition of retinal neovascularization
in OIR mice was observed only when
PTK787 was given systemically over
5 days (Ozaki et al. 2000).
Our current understanding of the
effects of dimeric
EfnB2 and EphB4 on retinal vascular
cells is summarized in Fig. 5. While
EphB4-Fc enhanced angiogenesis by
stimulating reverse signalling through
EfnB2, sEphB4 inhibited angiogenesis
by competing with EphB4 (Fig. 5).
Similarly, VEGFR2-Fc inhibits angio-
genesis by competing with VEGFR1
and VEGFR2 as receptors for VEGF
(Agostini et al. 2005). Surprisingly,
EfnB2-Fc did not inhibit angiogenesis
as would be expected if signalling
through EfnB2 was the only responsi-
ble pathway: EfnB2-Fc is a competitive
inhibitor of EfnB2 signalling. EfnB2-
indicating that signalling
EphB4 is also promoting angiogenesis.
These observations add to the picture
of why and how the ephrin⁄Eph system
and especially the monomeric, soluble
form EphB4 offers therapeutic options
in ischaemia-induced retinopathy. Tar-
geting an element other than VEGF is
a reasonable approach for combination
therapy with anti-VEGF agents as
currently under investigation for retinal
vein occlusions, diabetic retinopathy or
used clinically in age-related macul-
The authors thank A. Mattes and B.
Flu ¨ gel for technical assistance. This
work was supported by grants from
the Deutsche Forschungsgemeinschaft
[DFG, HA 2537⁄3-1 to LLH and
Au83⁄9-2 (SPP1190 ‘The tumour-ves-
sel interface’) to HGA] and the Nußs-
tiftung Freiburg. All authors declare
that they have no commercial or finan-
cial interest in the subject of the study.
Adams RH & Klein R (2000): Eph receptors
and ephrin ligands. Essential mediators of
vascular development. Trends Cardiovasc
Med 10: 183–188.
Adams RH, Wilkinson GA, Weiss C, Diella F,
Gale NW, Deutsch U, Risau W & Klein R
(1999): Roles of ephrinB ligands and EphB
receptors in cardiovascular development:
demarcation of arterial⁄venous domains,
angiogenesis. Genes Dev 13: 295–306.
Agostini HT, Boden K, Unsold A, Martin G,
Hansen L, Fiedler U, Esser N & Marme D
(2005): A single local injection of recombi-
nant VEGF receptor 2 but not of Tie2
inhibits retinal neovascularization in the
mouse. Curr Eye Res 30: 249–257.
Arvanitis D & Davy A (2008): Eph⁄ephrin
signaling: networks. Genes Dev 22: 416–
Augustin HG & Reiss Y (2003): EphB recep-
tors and ephrinB ligands: regulators of vas-
Tissue Res 314: 25–31.
Brantley-Sieders DM & Chen J (2004): Eph
receptor tyrosine kinases in angiogenesis:
from development to disease. Angiogenesis
Butler SJ & Tear G (2007): Getting axons
onto the right path: the role of transcrip-
tion factors in axon guidance. Develop-
ment 134: 439–448.
Chen J, Hicks D, Brantley-Sieders D, Cheng
N, McCollum GW, Qi-Werdich X & Penn
J (2006): Inhibition of retinal neovascular-
ization by soluble EphA2 receptor. Exp
Eye Res 82: 664–673.
Cheng N, Brantley DM & Chen J (2002): The
ephrins and Eph receptors in angiogenesis.
Cytokine Growth Factor Rev 13: 75–85.
Claxton S & Fruttiger M (2005): Oxygen
modifies artery differentiation and network
morphogenesis in the retinal vasculature.
Dev Dyn 233: 822–828.
Cell B Cell A
Fig. 5. Theoretical effect of dimeric or monomeric ephrins on signalling. Forward signalling as well as reverse signalling is induced if two adjacent
cells (A and B) carrying EphrinB2 (EfnB2) and EphB4, respectively, come into contact (bold horizontal line). Even without this contact, EphB4-
Fc induces signalling through EfnB2 (reverse signalling) while sEphB4 inhibits EfnB2 by competition for EphB4. In addition, both EphB4-Fc and
sEphB4 inhibit EphB4 signalling (forward signalling) competitively by preventing binding of EfnB2 to EphB4. The respective interactions are true
for EfnB2-Fc and sEfnB2 in this theoretical scheme.
Acta Ophthalmologica 2011
Dureau P, Bonnel S, Menasche M, Dufier JL
& Abitbol M (2001): Quantitative analysis
of intravitreal injections in the rat. Curr
Eye Res 22: 74–77.
Egea J & Klein R (2007): Bidirectional Eph-
ephrin signaling during axon guidance.
Trends Cell Biol 17: 230–238.
Fuller T, Korff T, Kilian A, Dandekar G &
Augustin HG (2003): Forward EphB4 sig-
naling in endothelial cells controls cellular
repulsion and segregation from ephrinB2
positive cells. J Cell Sci 116: 2461–2470.
Gerety SS, Wang HU, Chen ZF & Anderson
DJ (1999): Symmetrical mutant phenotypes
of the receptor EphB4 and its specific trans-
membrane ligand ephrin-B2 in cardiovascu-
lar development. Mol Cell 4: 403–414.
Hamada K, Oike Y, Ito Y, Maekawa H,
Miyata K, Shimomura T & Suda T (2003):
Distinct roles of ephrin-B2 forward and
EphB4 reverse signaling in endothelial cells.
Arterioscler Thromb Vasc Biol 23: 190–197.
He S, Ding Y, Zhou J et al. (2005): Soluble
EphB4 regulates choroidal endothelial cell
function and inhibits laser-induced choroi-
dal neovascularization. Invest Ophthalmol
Vis Sci 46: 4772–4779.
Heroult M, Schaffner F & Augustin HG
(2006): Eph receptor and ephrin ligand-
mediated interactions during angiogenesis
and tumor progression. Exp Cell Res 312:
Higgins RD, Yu K, Sanders RJ, Nandgaon-
kar BN, Rotschild T & Rifkin DB (1999):
Diltiazem reduces retinal neovasculariza-
tion in a mouse model of oxygen induced
retinopathy. Curr Eye Res 18: 20–27.
Himanen JP, Saha N & Nikolov DB (2007):
Cell-cell signaling via Eph receptors and
ephrins. Curr Opin Cell Biol 19: 534–542.
Leshanski L, Kumar SR, Zozulya S & Gill
PS (2006): The soluble extracellular domain
of EphB4 (sEphB4) antagonizes EphB4-
EphrinB2 interaction, modulates angiogen-
esis, and inhibits tumor growth. Blood 107:
Kuijper S, Turner CJ & Adams RH (2007):
Regulation of angiogenesis by Eph-ephrin
interactions. Trends Cardiovasc Med 17:
Lackmann M & Boyd AW (2008): Eph, a pro-
tein family coming of age: more confusion,
insight, or complexity? Sci Signal 1: re2.
Lange C, Ehlken C, Martin G, Konzok K,
Moscoso DP, Hansen LL & Agostini HT
(2007): Intravitreal injection of the heparin
reduces retinal neovascularization in mice.
Exp Eye Res 85: 323–327.
Lobe CG, Koop KE, Kreppner W, Lomeli
H, Gertsenstein M & Nagy A (1999):
Z⁄AP, a double reporter for cre-mediated
recombination. Dev Biol 208: 281–292.
Maekawa H, Oike Y, Kanda S, Ito Y,
Yamada Y, Kurihara H, Nagai R & Suda
T (2003): Ephrin-B2 induces migration of
endothelial cells through the phosphatidyl-
inositol-3 kinase pathway and promotes
angiogenesis in adult vasculature. Arterios-
cler Thromb Vasc Biol 23: 2008–2014.
Maier P, Unsoeld AS, Junker B, Martin G,
Drevs J, Hansen LL & Agostini HT
(2005): Intravitreal injection of specific
receptor tyrosine kinase inhibitor PTK787⁄
ZK222 584 improves ischemia-induced reti-
nopathy in mice. Graefes Arch Clin Exp
Ophthalmol 243: 593–600.
Agostini HT (2004): Differential expression
of angioregulatory factors in normal and
CNV-derived human retinal pigment epi-
thelium. Graefes Arch Clin Exp Ophthal-
mol 242: 321–326.
Martiny-Baron G, Korff T, Schaffner F,
Esser N, Eggstein S, Marme D & Augustin
HG (2004): Inhibition of tumor growth
and angiogenesis by soluble EphB4. Neo-
plasia 6: 248–257.
Ozaki H, Yu AY, Della N et al. (1999):
Hypoxia inducible factor-1alpha is incre-
ased in ischemic retina: temporal and
spatial correlation with VEGF expression.
Invest Ophthalmol Vis Sci 40: 182–189.
Ozaki H, Seo MS, Ozaki K et al. (2000):
growth factor receptor signaling is suffi-
cient to completely prevent retinal neovas-
cularization. Am J Pathol 156: 697–707.
Pfaffl MW, Horgan GW & Dempfle L (2002):
Relative expression software tool (REST)
for group-wise comparison and statistical
analysis of relative expression results in real-
time PCR. Nucleic Acids Res 30: e36.
Pierce EA, Foley ED & Smith LE (1996):
Regulation of vascular endothelial growth
factor by oxygen in a model of retinopathy
114:1219–1228. [Published erratum appears
in Arch Ophthalmol 1997; 115: 427.].
Ramakers C, Ruijter JM, Deprez RH &
analysis of quantitative real-time polymer-
ase chain reaction (PCR) data. Neurosci
Lett 339: 62–66.
Reber M, Hindges R & Lemke G (2007):
Eph receptors and ephrin ligands in axon
guidance. Adv Exp Med Biol 621: 32–49.
Robinson GS, Ju M, Shih SC, Xu X, McMa-
hon G, Caldwell RB & Smith LE (2001):
Nonvascular role for VEGF: VEGFR-1, 2
activity is critical for neural retinal devel-
opment. FASEB J 15: 1215–1217.
Kostyk SK, D’Amato R, Sullivan R &
D’Amore PA (1994): Oxygen-induced reti-
nopathy in the mouse. Invest Ophthalmol
Vis Sci 35: 101–111.
G, Hansen LL&
Steinle JJ, Meininger CJ, Chowdhury U, Wu
G & Granger HJ (2003): Role of ephrin B2
in human retinal endothelial cell prolifera-
tion and migration. Cell Signal 15: 1011–
Stitt AW, Graham D & Gardiner TA (2004):
Ocular wounding prevents pre-retinal neo-
expression in the inner retina. Mol Vis 10:
Umeda N, Ozaki H, Hayashi H & Oshima K
(2004): Expression of ephrinB2 and its
receptors on fibroproliferative membranes
in ocular angiogenic diseases. Am J Oph-
thalmol 138: 270–279.
Unsoeld AS, Junker B, Mazitschek R, Martin
G, Hansen LL, Giannis A & Agostini HT
(2004): Local injection of receptor tyrosine
kinase inhibitor MAE 87 reduces retinal
neovascularization in mice. Mol Vis 10:
Wang HU, Chen ZF & Anderson DJ (1998):
Molecular distinction and angiogenic inter-
action between embryonic arteries and
veins revealed by ephrin-B2 and its recep-
tor Eph-B4. Cell 93:661–664.
Yao YG, Yang HS, Cao Z, Danielsson J &
Duh EJ (2005): Upregulation of placental
tional mechanism. FEBS Lett 579: 1227–
Yossuck P, Yan Y, Tadesse M & Higgins
RD (2000): Dexamethasone and critical
effect of timing on retinopathy. Invest
Ophthalmol Vis Sci 41: 3095–3099.
Zamora DO, Davies MH, Planck SR, Rosen-
baum JT & Powers MR (2005): Soluble
forms of EphrinB2 and EphB4 reduce reti-
nal neovascularization in a model of prolif-
erative retinopathy. Invest Ophthalmol Vis
Sci 46: 2175–2182.
Zhang J & Hughes S (2006): Role of the eph-
rin and Eph receptor tyrosine kinase fami-
lies in angiogenesis and development of the
cardiovascular system. J Pathol 208: 453–
Received on November 5th, 2008
Accepted on March 8th, 2009.
Hansju ¨ rgen T. Agostini
Universita ¨ ts-Augenklinik Freiburg
Tel: + 49 761 270 4001
Fax: + 49 761 270 4174
Acta Ophthalmologica 2011