Pigment Epithelium-Derived Factor Improves Paracellular Blood-Brain Barrier Integrity in the Normal and Ischemic Mouse Brain

Abstract

Pigment epithelium-derived factor (PEDF) is a neurotrophic factor with neuroprotective, antiangiogenic, and antipermeability effects. In the brain, blood-brain barrier (BBB) function is essential for homeostasis. Its impairment plays a crucial role in the pathophysiology of many neurological diseases, including ischemic stroke. We investigated (a) whether PEDF counteracted vascular endothelial growth factor (VEGF)-induced BBB disruption in the mouse brain, (b) the time course and route of BBB permeability and the dynamics of PEDF expression after cerebral ischemia, and (c) whether intraventricular infusion of PEDF ameliorated brain ischemia by reducing BBB impairment. C57Bl6/N mice received intraparenchymal injections of CSF, VEGF, or a combination of VEGF and PEDF. PEDF increased paracellular but not transcellular BBB integrity as indicated by an increase in the tight junction protein claudin-5. In another group of mice undergoing 60-min middle cerebral artery occlusion (MCAO), transcellular BBB permeability (fibrinogen staining in the absence of a loss of claudin-5) increased as early as 6 h after reperfusion. PEDF immunofluorescence increased at 24 h, which paralleled with a decreased paracel-lular BBB permeability (claudin-5). PEDF after MCAO originated from the blood stream and endogenous pericytes. In the third experiment, the intraventricular infusion of PEDF decreased edema and cell death after MCAO, potentially mediated by the improvement of the paracellular route of BBB permeability (claudin-5) in the absence of an amelioration of Evans Blue extravasation. Together, our data suggest that PEDF improves BBB function after cerebral ischemia by affecting the paracellular but not the transcellular route. However, further quantitative data of the different routes of BBB permeability will be required to validate our findings.

Full text

Vol.:(0123456789)
1 3
Cellular and Molecular Neurobiology
https://doi.org/10.1007/s10571-019-00770-9
ORIGINAL RESEARCH
Pigment Epithelium‑Derived Factor Improves Paracellular Blood–Brain
Barrier Integrity intheNormal andIschemic Mouse Brain
ArinaRiabinska1,3 · MariettaZille2,4 · MenderesYusufTerzi1,5 · RyanCordell1· MelinaNieminen‑Kelhä1 ·
JanKlohs2,6,7 · AnaLuisaPiña1
Received: 22 August 2019 / Accepted: 2 December 2019
© Springer Science+Business Media, LLC, part of Springer Nature 2019
Abstract
Pigment epithelium-derived factor (PEDF) is a neurotrophic factor with neuroprotective, antiangiogenic, and antipermeability
effects. In the brain, blood–brain barrier (BBB) function is essential for homeostasis. Its impairment plays a crucial role in the
pathophysiology of many neurological diseases, including ischemic stroke. We investigated (a) whether PEDF counteracted
vascular endothelial growth factor (VEGF)-induced BBB disruption in the mouse brain, (b) the time course and route of
BBB permeability and the dynamics of PEDF expression after cerebral ischemia, and (c) whether intraventricular infusion
of PEDF ameliorated brain ischemia by reducing BBB impairment. C57Bl6/N mice received intraparenchymal injections
of CSF, VEGF, or a combination of VEGF and PEDF. PEDF increased paracellular but not transcellular BBB integrity as
indicated by an increase in the tight junction protein claudin-5. In another group of mice undergoing 60-min middle cerebral
artery occlusion (MCAO), transcellular BBB permeability (fibrinogen staining in the absence of a loss of claudin-5) increased
as early as 6h after reperfusion. PEDF immunofluorescence increased at 24h, which paralleled with a decreased paracel-
lular BBB permeability (claudin-5). PEDF after MCAO originated from the blood stream and endogenous pericytes. In the
third experiment, the intraventricular infusion of PEDF decreased edema and cell death after MCAO, potentially mediated
by the improvement of the paracellular route of BBB permeability (claudin-5) in the absence of an amelioration of Evans
Blue extravasation. Together, our data suggest that PEDF improves BBB function after cerebral ischemia by affecting the
paracellular but not the transcellular route. However, further quantitative data of the different routes of BBB permeability
will be required to validate our findings.
Keywords Blood–brain barrier· Cerebral ischemia· Claudin-5· Fibrinogen· MCAO· Pigment epithelium-derived factor
Introduction
The blood–brain barrier (BBB) consists of cells of the neu-
rovascular unit that serves to restrict the interaction between
the blood stream and brain parenchyma to maintain brain
homeostasis (Bernacki etal. 2008; Cardoso etal. 2010). The
Arina Riabinska and Marietta Zille contributed equally to this
study.
* Ana Luisa Piña
ana-luisa.pina@charite.de
1 Department ofNeurosurgery, Experimental Neurosurgery/
BCRT , Charite-Universitätsmedizin Berlin, Campus Mitte,
Chariteplatz 1/Virchowweg 21, Aschheim-Zondek-Haus
03-003, 10117Berlin, Germany
2 Department ofExperimental Neurology,
Charité-Universitätsmedizin Berlin, Berlin, Germany
3 Department ofInternal Medicine, Medical Clinic I,
University Hospital ofCologne, Cologne, Germany
4 Institute forExperimental andClinical Pharmacology
andToxicology, University ofLübeck, Lübeck, Germany
5 Department ofMedical Biology, Hatay Mustafa Kemal
University, Antakya, Hatay, Turkey
6 Institute forBiomedical Engineering, ETH andUniversity
ofZurich, Zurich, Switzerland
7 Neuroscience Center Zurich, University ofZurich andETH
Zurich, Zurich, Switzerland

Cellular and Molecular Neurobiology
1 3
selective permeability of the BBB ensures the proper deliv-
ery of nutrients to the brain. It also protects the brain from
toxic substances and pathogens (Abbott etal. 2010).
Dysfunction of the BBB contributes to many neurological
diseases, including cerebral ischemia, where a strong asso-
ciation between the BBB extravasation and tissue damage
has been demonstrated (Abbott etal. 2010; Chen etal. 2009;
Weiss etal. 2009). Studies investigating BBB integrity after
middle cerebral artery occlusion (MCAO, a rodent model
of ischemic stroke) reveal a biphasic opening of the BBB
in post-ischemic rat brains (Allen and Bayraktutan 2009;
Belayev etal. 1996; Huang etal. 1999; Klohs etal. 2009;
Pillai etal. 2009). Treatments to prevent or augment BBB
impairment are currently investigated.
Three characteristics of the brain endothelium are cru-
cial for barrier function: (1) tight junctions restricting the
paracellular diffusion of ions and proteins, (2) transcytosis
via endocytotic vesicles, and (3) the active transport of mol-
ecules between blood and brain (Abbott etal. 2010; Know-
land etal. 2014). The transcellular route involves caveolae-
mediated vesicular transport (Komarova and Malik 2010). It
can be assessed by observing fibrinogen or albumin leakage
as both blood-borne proteins have a high molecular weight
and therefore cross the BBB mainly via the transcellular
route (Ryu and McLarnon 2009; Schubert etal. 2001). In
contrast, claudin-5, among other tight junction proteins, can
be used to study the paracellular permeability of the BBB
(Knowland etal. 2014). However, in case tight junctions are
disrupted, albumin and fibrinogen may also enter the brain
via the paracellular route (Schubert etal. 2001). It is there-
fore necessary to use albumin and fibrinogen together with
markers of paracellular permeability to distinguish transcel-
lular (albumin/fibrinogen leakage in the absence of loss of
tight junction proteins) from paracellular BBB permeabil-
ity (loss of tight junction proteins with or without albumin/
fibrinogen leakage).
Pigment epithelium-derived factor (PEDF) is a growth
factor that reduces vessel permeability. It is a multifunc-
tional protein showing neurotrophic, neuroprotective, anti-
tumorigenic, and antiangiogenic functions (Tombran-Tink
2005). PEDF is produced in most mammalian tissues,
including the brain (Guan etal. 2003; van Wagenen 2008;
Yabe etal. 2001). Being one of the strongest antiangiogenic
factors reported, PEDF inhibits endothelial cell growth,
proliferation, migration, and tubule formation, counteract-
ing the vascular endothelial growth factor (VEGF) action
on these cells (Cai etal. 2006; Dawson etal. 1999; Duh
etal. 2002; Hutchings etal. 2002). PEDF is downregulated
in conditions associated with abnormal vessel formation,
such as proliferative diabetic retinopathy (Gao etal. 2002;
Ueda etal. 2010). When administered in an animal model
of the disease, PEDF reduced pathological endothelial cell
proliferation in a dose-dependent manner (Stellmach etal.
2001). Similar events were observed in kidneys of patients
suffering from diabetic nephropathy (Fujimura etal. 2009;
Wang etal. 2005).
Besides its antiangiogenic properties, PEDF was observed
to counteract the VEGF action on vessel permeability (Liu
etal. 2004; Yamagishi etal. 2007, 2003; Yang etal. 2010).
Due to its effect against vessel leakage, PEDF reduced the
brain edema after ischemic stroke or cold injury (Jinnouchi
etal. 2007; Sanagi etal. 2008). Pillai and co-workers showed
that intravenously injected PEDF reduced BBB integrity
between 24h and 1week after reperfusion in a rat model of
transient focal cerebral ischemia (Pillai etal. 2013).
In this study, we investigated (a) whether PEDF inhibits
VEGF-induced pathologic BBB disruption in the mouse
brain, (b) the time course and route of BBB permeability as
well as the dynamics of the PEDF expression after MCAO,
and (c) whether the intraventricular infusion of PEDF ame-
liorated brain ischemia by reducing BBB impairment.
Materials andMethods
All experimental procedures were performed by investiga-
tors blinded to the experimental conditions. We randomly
assigned the animals to the experimental groups.
Experimental Animals
All animal experiments were performed in accordance with
the national and international guidelines for the care and use
of laboratory animals (Tierschutzgesetz der Bundesrepublik
Deutschland, European directive, as well as GV-SOLAS and
FELASA guidelines and recommendations for laboratory
animal welfare). The studies were approved by an ethics
committee (Landesamt für Gesundheit und Soziales, Berlin,
Germany, permit number G 0270/10).
Eight- to eleven-weeks-old male C57BL6/N mice were
acquired from Charles River Laboratories, Germany. The
animals were housed in groups of 3–6 per cage with a
12/12-h light/dark cycle controlled environment and given
adlibitum access to food and water.
Stereotactic Intraparenchymal Injections
The solutions of VEGF (PeproTech, USA) and PEDF
(Bioproducts MD, USA) in aCSF (Harvard Apparatus,
Holliston, USA) as well as artificial cerebrospinal fluid
(CSF) vehicle were prepared for injections. For the clau-
din-5 expression study, solutions with the following
protein concentrations were prepared: 40ng/ml VEGF
(group VEGF alone), 40 ng/ml VEGF and 40 ng/ml
PEDF (group VEGF:PEDF 1:1), and 40ng/ml VEGF and
80ng/ml PEDF (group VEGF:PEDF 1:2). For Evans Blue

Cellular and Molecular Neurobiology
1 3
extravasation measurement, CSF vehicle, “VEGF alone,”
and “VEGF:PEDF 1:2” solutions were used. In each injec-
tion, 3μl of the solution was used so that at least 120ng
of each of the diluted proteins was injected.
We anesthetized the animals by an intraperitoneal injec-
tion of a mixture of ketamine (25mg/kg; Ketavet, Pfizer,
Germany) and xylazine (16mg/kg; Rompun, Bayer, Ger-
many) diluted in 0.9% sterile saline. The mice were placed
into the stereotactic frame, the head skin was incised, and
holes were drilled into the skull (Bregma: AP + 0.2mm,
Lat ± 2mm). Syringe needles (10μl syringes, Hamilton,
Nevada, USA) were carefully immersed into the brain,
down to 2mm depth. We then injected VEGF into the
left hemisphere and VEGF + PEDF solutions or CSF vehi-
cle into the right hemisphere with a speed of 0.5μl/30s.
After 1min, the needles were slowly removed and the skin
was sutured with a 4-0 polyester suture (Ethicon, USA).
The operated animals were removed from the stereotactic
frame and put onto a warm bed for 2h to recover from
the surgery.
Osmotic Pump Implantation
Osmotic pumps were implanted two days prior to MCAO.
Although not directly applicable to patients, we applied a
pre-MCAO treatment in order to investigate the effect of
PEDF infusion on stroke recovery. Due to the high compli-
cation rate in a pilot experiment, we chose not to implant
the pumps right after MCAO.
We primed micro-osmotic Alzet pumps (model 1007D)
under sterile conditions one day before implantation and
left them in an incubator at 37°C overnight. The animals
were anesthetized as described above and put into a ste-
reotactic frame. The mouse skull was exposed and the
pump cannula was implanted (0.2mm to the right from
Bregma) corresponding to the location of the right ven-
tricle. Cyanoacrylate clay (Weicon, Germany) was intro-
duced between the cannula cap and the skull. Then, the
cannula was slowly moved down until the cap touched the
skull. The animals were left for 5min to ensure drying
of the clay. Subsequently, the skin above the cannula was
sutured with a 5-0 polypropylen thread (Ethicon, USA).
The animals received lidocaine gel locally and 0.5ml
of ringer lactate solution (B Braun Melsungen, Germany)
subcutaneously to substitute the liquid loss. We placed the
animal cage onto the 37°C heating pad for 2h for recovery
after surgery.
The animals received a total amount of 36 ± 7.2μl of
PEDF (20μg/ml in CSF) or CSF at a pumping rate of
0.5 ± 0.1μl/h (experiments were terminated one day after
MCAO for immunohistochemistry).
Middle Cerebral Artery Occlusion
The animals were anesthetized by an intraperitoneal
injection of a mixture of 10mg/kg xylazine (Pfizer) and
200mg/kg ketamine hydrochloride (Bayer). Throughout
the whole procedure and during recovery, the body tem-
perature was maintained at 37°C via a heating pad. After
ligation of the left proximal common carotid artery and
external carotid artery, a 7-0-nylon monofilament (Doc-
col Co., NM, USA) with a 0.23-mm coated tip was intro-
duced into the distal internal carotid artery via an incision
in the ligated common carotid artery. The monofilament
was advanced 11mm distal to the carotid bifurcation to
occlude the middle cerebral artery. After the topical appli-
cation of lidocainhydrochlorid (Xylocain Gel 2%, Astra-
Zeneca), the neck wound was closed temporarily for a
60-min ischemic period. At reperfusion, the monofilament
was withdrawn from the carotid artery and the wound was
stitched with 4-0 non-resorbable sutures (Ethibond Excel,
Ethicon). After surgery, the animals were allowed to wake
up in a warming cage and were kept there for around 2h.
In the study using the intraventricular infusion of PEDF,
one animal was sacrificed prior to magnetic resonance
imaging (MRI) due to complications after surgery.
Quantication ofEvans Blue Extravasation
We quantified Evans Blue extravasation using the Evans
Blue fluorescence assay (Belayev etal. 1996; Uyama
etal. 1988). We administered intravenous injections of a
2%-solution of Evans blue dye in 0.9% isotonic saline (EB;
4ml/kg, Sigma-Aldrich, MO, USA) right after the end of
the surgical procedure (stereotactic injection or MCAO).
24h later (i.e., 7 h after stereotactic injections), we
perfused the animals transcardially with 0.9% normal
saline and harvested the brains. We removed the cerebelli
and olfactory bulbs, and cut the rest of the brains into
two hemispheres. Then, 50% trichloroacetic acid in dis-
tilled water (Sigma-Aldrich, Germany) was added (2ml
per gram of brain tissue). The brains were homogenized
and sonicated in trichloroacetic acid and then centrifuged
at 10,000×g for 20min. The supernatant was taken and
diluted 1:3 in 100% ethanol. We transferred the resulting
samples into 96-well black plates (Nunc, USA) and meas-
ured the fluorescence emission at 680nm (with an excita-
tion wavelength of 620nm) using a Tecan plate reader
(Infinite Series M200, Switzerland).
The amount of extravasated Evans blue dye in the
VEGF-induced hyperpermeability was normalized to
the amount of extravasated Evans blue in control condi-
tions, such as treatment with CSF vehicle or VEGF alone,
respectively.

Cellular and Molecular Neurobiology
1 3
Immunohistochemistry
For all immunohistochemistry analyses, we anesthetized the
animals as described before and perfused them transcardially
with 0.9% saline, followed by 4% PFA in saline. We removed
the brains from the skull and kept them in PFA solution for
2h. They were then transferred into 30% sucrose and sub-
sequently snap-frozen. 10-μm-thick coronal slices were cut
with a cryostat (HM560, Microm, Germany) and every sec-
ond section was preserved. The slices were arranged on glass
slides (six sections with a distance of 1mm from Bregma
2.96 to −2.54) and kept at −20°C.
For immunohistochemistry, we washed the sections with
washing solution (0.1% Triton X-100 in PBS) and blocked
for 30min with blocking solution (5% skim milk, 5% bovine
serum albumin, 0.1% Triton X-100 in PBS). Subsequently,
we incubated the sections at 4°C overnight with either
mouse anti-claudin-5 (Invitrogen, USA, 1:50), rat anti-CD31
(BD Biosciences, Germany, 1:100), rabbit anti-PEDF (Bio-
products MD, USA, 1:100), mouse anti-GFAP (Millipore,
USA, 1:100), rat anti-CD68 (AbD Serotec, Germany, 1:100),
rat anti-platelet-derived growth factor receptor (Abcam, UK,
1:100), or sheep anti-fibrinogen (US BioLogical, USA,
1:100).
After washing, the tissue was incubated for 1h at room
temperature with either anti-mouse Dye Light 549 (Jack-
son ImmunoResearch, USA, 1:100), anti-mouse Dye Light
488 (Jackson ImmunoResearch, USA, 1:100), anti-rabbit
Alexa Fluor 568, anti-rabbit Alexa Fluor 488 (both Invit-
rogen, USA, 1:200), anti-rat Dye Light 594, anti-rat Dye
Light 649, or anti-sheep fluorescein (all Jackson ImmunoRe-
search, USA, 1:200). Then, the sections were washed again
and mounted with DABCO.
To note, we did not find any unwanted background in the
corresponding negative controls when omitting the incuba-
tion with primary antibodies.
TUNEL Staining
We performed the TUNEL staining using the ApopTag kit
(Millipore, USA, Catalogue Number: S7165) according to
the manufacturer’s protocol. Briefly, we fixed the cryosec-
tions in 1% PFA at room temperature for 10min and post-
fixed them in ethanol-acetic acid mixture (2:1) in −20°C
for 5min. After washing in PBS, we shortly (10s) incubated
the specimens with the equilibration buffer supplied with the
kit. Thereafter, we incubated the sections with TdT enzyme
solution for 1h at 37°C. To stop the reaction, the stop/wash
buffer was applied. We washed the specimens and incubated
them for 30min with Rhodamine at room temperature. The
specimens were again washed in PBS and the coverslips
were mounted with a mounting medium containing DAPI.
Fluorescence Microscopy andImage Assessment
We assessed the immunohistochemical stainings with a
confocal microscope (Olympus BX-61, USA; numerical
aperture 0.3 for 10× and 0.5 for 20×). For each experiment,
we analyzed stained tissue samples from at least three ani-
mals per group. Animals with matching lesion sizes were
selected for an unbiased analysis. We quantitatively analyzed
the images in Cell^P software (Olympus, Hamburg, Ger-
many). In the VEGF-induced hypermeability, we assessed
claudin-5 staining on three slices per animal. On each slice,
we investigated four pictures per region of interest, i.e., the
ipsilateral and contralateral cortex and striatum. We counted
the TUNEL-positive cells on four to ten slices per animal
inside the lesion area as discriminated by H&E staining
using Cell^P. The number of TUNEL-positive cells per mm2
was subsequently calculated.
Magnetic Resonance Imaging
24h after MCAO, we scanned the mice in a 7 Tesla Phar-
mascan 70/16 (Bruker Biospin MRI GmbH, Ettlingen, Ger-
many) equipped with a 16-cm horizontal bore magnet and a
9-mm (inner diameter) shielded gradient (300MHz H-res-
onance frequency, maximum gradient strength 300 mT/m,
rise time 80μs). The data acquisition and image processing
were carried out with the Paravision 4.0 software (Bruker).
For the examination, the animals were anesthetized
with isoflurane (2.5% for induction, 1.5% for maintenance,
Forene, Abbot, Wiesbaden, Germany) in 70% N2O and 30%
O2 via a facemask under constant ventilation monitoring
(Small Animal Monitoring & Gating System, SA Instru-
ments, Stony Brook, New York, USA). The body tempera-
ture was kept constant at 37°C during the measurement
using a heated circulating water blanket.
For imaging the mouse brain, we used a T2-weighted 2D
turbo spin-echo sequence (TR/TE: 4200/36ms, rare factor 8,
4 averages). Twenty 0.5-mm-thick axial slices over the brain
from olfactory bulb to cerebellum were imaged with a field
of view of 2.56 × 2.56cm and a matrix size of 256 × 256,
resulting in a nominal voxel size of 100μm × 100μm. The
acquisition of the T2-weighted images lasted 6min 43s.
Infarct Volumetry andEdema Assessment
We processed the images acquired during magnetic reso-
nance scanning with ImageJ 1.42 (NIH, Bethesda, USA)
and Analyze 5.0 (BIR, Mayo Clinic, USA) software. We
outlined the ipsilateral and contralateral hemispheres as well
as the ischemic infarct represented by hyperintense areas in
T2-weighted images in each of the 20 slices. To estimate
the infarct volume, we applied an edema correction by cal-
culating the ‘indirect’ infarct volume as the volume of the

Cellular and Molecular Neurobiology
1 3
contralateral hemisphere minus the non-infarcted volume
of the ipsilateral hemisphere. To assess the contribution of
the edema, we calculated the percentage volume increase
of the ipsilateral compared to the contralateral hemisphere.
One animal was excluded from the analysis since the infarct
volume was < 10 mm3.
Statistical Analysis
We tested normality using the Kolmogorov–Smirnov test
and variance homogeneity using the Levené test. When the
data were normally distributed, variance homogeneity was
met, and two independent groups were investigated, we per-
formed the Student’s t test. In the case of two dependent
groups with the criteria of normality and homogeneity met,
we performed the dependent t test. When the data were not
normally distributed or variance homogeneity was not met
and two dependent groups were compared, the Wilcoxon
Signed-Rank test was performed. In the case of three inde-
pendent groups, we used the Kruskal–Wallis test.
Data are represented as mean ± standard deviation or
as median for non-parametric data. Differences were con-
sidered significant at p < 0.05. We performed all statistical
analyses with SPSS v.19.0.
Results
Intraparenchymal Injection ofPEDF
inConjunction withVEGF Reduces theParacellular
Hyperpermeability oftheBlood–Brain Barrier
It is known that VEGF infusion into the brain induces cer-
ebrovascular hyperpermeability (Proescholdt etal. 1999).
We here assessed whether PEDF can antagonize the VEGF-
induced vessel hyperpermeability, which was previously
described for the eye (Liu etal. 2004). We first quantified
the Evans Blue extravasation after the intraparenchymal
injections of VEGF compared to CSF to confirm the lit-
erature results. Next, we compared VEGF to a combination
of VEGF and PEDF in order to see the effects of PEDF
on VEGF-induced hyperpermeability. For both com-
parisons, the data were normally distributed (Kolmogo-
rov–Smirnov test, Z = 0.819, p = 0.514 for CSF vs. VEGF;
Z = 1.109, p = 0.171 for VEGF vs. VEGF + PEDF), but
the variances were not homogenous across groups (Lev-
ené test, F(1,4) = 15.975, p = 0.016 for CSF vs. VEGF;
F(1,10) = 6.620, p = 0.028 for VEGF vs. VEGF + PEDF).
The extent of Evans blue extravasation was not significantly
different between CSF and VEGF (median fold change of
4.039, n = 3, Wilcoxon Signed-Rank test, Z = − 1.604,
p = 0.109, r = − 0.655, Fig. 1a). However, we recognize
that the sample size was likely not large enough to show
statistical significance. Compared to VEGF alone, injection
of VEGF + PEDF did not change the extravasation of Evans
blue (median fold change of 1.026, n = 6, Wilcoxon Signed-
Rank test, Z = −0.314, p = 0.753, r = −0.091, Fig.1b).
As the tight junction proteins that mediate the paracellu-
lar BBB permeability are downregulated by VEGF (Argaw
etal. 2009), we investigated the change in the expression
of claudin-5 (Fig.1c). We injected VEGF into the left
hemisphere and VEGF + PEDF 1:1 or 1:2 into the other
hemisphere and counted the number of claudin-5-immuno-
reactive vessels. The data were normally distributed (Kol-
mogorov–Smirnov test, Z = 1.013, p = 0.256 and Z = 0.902,
p = 0.389 for VEGF + PEDF 1:1 and 1:2, respectively),
but the variances were not homogenous across groups for
VEGF + PEDF 1:1 (Levené test, F(1,20) = 7.821, p = 0.011
and F(1,16) = 0.735, p = 0.404 for VEGF + PEDF 1:1 and
1:2, respectively).
The number of claudin-5-immunoreactive vessels
was significantly higher in the hemisphere injected with
VEGF + PEDF 1:1 (median of 148 vessels/mm2) com-
pared to the hemisphere injected with VEGF alone (123
vessels/mm2) (Wilcoxon test, Z = − 2.179, p = 0.029,
r = 0.57, n = 11 per group, Fig.1d). Similarly, it was higher
upon VEGF:PEDF 1:2 injection compared to VEGF alone
(142 ± 32 vs. 115 ± 20 vessels/mm2 for VEGF + PEDF 1:2
or VEGF alone, respectively, dependent T test, t(8) = -2.780,
p = 0.024, r = 0.70, n = 9 per group, Fig.1d).
Together, our data showed that PEDF increased the num-
ber of Claudin-5-positive vessels, while it did not change
Evans Blue extravasation, indicating that PEDF has an effect
on the paracellular, but not on the transcellular route.
Disruption oftheBlood–Brain Barrier After MCAO
Since the BBB permeability to plasma constituents after cer-
ebral ischemia involves both transcellular and paracellular
routes, we aimed to determine the dynamics of these routes
within the first 24h in our model (at 6, 10, 14, and 24h
after MCAO). We first characterized the extravasation of
fibrinogen. On the contralateral side, we detected almost no
fibrinogen staining except for the circumventricular organs
where there is no BBB. In contrast, on the ipsilateral side,
double-positive staining was present, indicating leaky ves-
sels as well as broader fibrinogen-positive areas correspond-
ing to extravasation. We observed the majority of leaky ves-
sels in the striatum with a high extravasation density after
6h, which was lower at 10 and 14h, but increased again at
24h after reperfusion (Fig.2a).
We studied the paracellular BBB permeability using
claudin-5 immunohistochemistry. We found a decrease of
claudin-5 immunoreactive vessels at 24h in both core and
penumbra (Fig.2b).

Cellular and Molecular Neurobiology
1 3
Collectively, our data suggest that there is an early tran-
scellular permeability (extravasation of fibrinogen in the
presence of intact tight junctions), followed by a later par-
acellular permeability (at 24h, loss of tight junction pro-
teins and extravasation of fibrinogen). Whether the trans-
cellular route is also contributing to the BBB disruption is
unclear at the later time point, since fibrinogen may enter
through the paracellular route as well when tight junctions
are disrupted (Schubert etal. 2001). Further quantitative
analyses of the para- and transcellular permeability are
needed to characterize the extent of BBB permeability.
PEDF inthePost‑ischemic Brain Originated
fromtheBlood Stream andEndogenous Pericytes
As neurons and endothelial cells express PEDF, we hypoth-
esized that post-ischemic disruption of brain endothelium
may be associated with a drop in endogenous PEDF expres-
sion. We therefore performed PEDF immunohistochemis-
try at 6, 10, 14, and 24h after MCAO. The distribution of
PEDF in the brain tissue did not vary much between the
samples obtained at the different time points except for 24h
after MCAO. At this time point, we noticed a PEDF-positive
Fig. 1 The Co-injection of PEDF after VEGF-induced hyperperme-
ability reduces the paracellular blood–brain barrier disruption. a We
wanted to confirm that VEGF induces hyperpermeability in the brain.
Therefore, we administered intraparenchymal injection of 40 ng/
ml VEGF or CSF to the brain. VEGF did not significantly induce
the transcellular extravasation of Evans Blue (expressed as the fold
change to CSF, n = 3). However, we recognize that the sample size
was likely not large enough to show statistical significance. b Com-
pared to VEGF alone (40 ng/ml), the coadministration of PEDF
(40 ng/ml) did not change the Evans Blue extravasation (n = 6).
c Shown are representative pictures from the claudin-5 (Cln-5)
stained brain samples in mice receiving an intraparenchymal injec-
tion of CSF, VEGF, or VEGF + PEDF. Scale bar = 100 μm. d The
number of claudin-5-immunoreactive vessels significantly increased
after the VEGF + PEDF (40 ng/ml VEGF and 40 ng/ml PEDF for
VEGF:PEDF 1:1, n = 11; 40 ng/ml VEGF and 80 ng/ml PEDF for
VEGF:PEDF 1:2, n = 9) compared to the 40ng/ml VEGF-only treat-
ment, indicating an improved paracellular BBB integrity. The data are
represented as medians, except for the amount of claudin-5-immuno-
reactive vessels in VEGF + PEDF 1:2 that is presented as mean ± SD,
*p < 0.05

Cellular and Molecular Neurobiology
1 3
staining in the extracellular space. This is not typical for
PEDF immunohistochemistry in the brain under normal
physiological conditions (Fig.3a, van Wagenen 2008).
We investigated the origin of PEDF in the ischemic
brain using a triple staining of CD31 for endothelial cells,
fibrinogen for transcellular BBB permeability, and PEDF.
We found the PEDF staining around or very close to CD31-
positive vessels in a similar pattern as fibrinogen. In most
spots, the PEDF and fibrinogen staining co-localized
(Fig.3b), indicating the extravasation of PEDF from the
blood to the brain parenchyma. To seek for potential endog-
enous sources of PEDF synthesis in the brain, we performed
a series of double labeling experiments. We observed PEDF
staining in cells positive for platelet-derived growth factor
receptor (PDGFR), a marker for pericytes, in the ischemic
striatum (Fig.3c). However, we did not detect any PEDF-
positive astrocytes and microglial cells within the first 24h
post-MCAO.
Intraventricular Infusion ofPEDF Decreases Edema
andCell Death, Potentially byReducing Paracellular
butNot Transcellular BBB Impairment at24h After
MCAO
We studied the effects of PEDF application on the BBB
integrity and lesion development 24h after MCAO in mice
Fig. 2 The dynamics of the blood–brain barrier impairment after
MCAO. a Shown are representative pictures of the fibrinogen
extravasation (green) into the ischemic brain tissue at 6, 10, 14, and
24h after MCAO for the ipsilateral and at 10h after MCAO for the
contralateral striatum. The yellow staining indicates double-positive
signal for endothelial cells (CD31) and fibrinogen. The white squares
indicate the areas magnified in the lower row. Scale bars = 500µm. b
We assessed the paracellular BBB permeability using the claudin-5
immunohistochemistry co-stained with CD31. Shown are representa-
tive pictures at 10, 14, and 24h after MCAO for the ipsilateral and
at 10 h after MCAO for the contralateral striatum in the core and
penumbra. The yellow staining indicates double-positive signal. Scale
bars = 50µm

Cellular and Molecular Neurobiology
1 3
implanted with osmotic pumps which contained either CSF
or PEDF into the ventricle contralateral to the ischemic
lesion 48h prior to MCAO. For both the hemispheric vol-
ume increase due to edema and lesion volume, the data were
normally distributed (Kolmogorov–Smirnov test, Z = 0.604,
p = 0.859 for the hemispheric volume increase and Z = 0.624,
p = 0.831 for the lesion volume) and the variances were
homogenous across groups (Levené test, F(1,12) = 0.099,
p = 0.759 for the hemispheric volume increase and
F(1,12) = 0.049, p = 0.828 for the lesion volume).
Whereas the hemispheric volume increase was signifi-
cantly reduced in the PEDF-treated (1.91 ± 4.56%, n = 7)
compared to the CSF-treated animals (9.33 ± 5.40%, n = 7,
Student’s t test, t(12) = 2.779, p = 0.017, d = 1.20), lesion vol-
umes were not significantly reduced (20.62 ± 8.66 mm3 for
the PEDF- (n = 7) vs. 24.43 ± 9.74mm3 for the CSF-treated
animals (n = 7), Student’s t test, t(12) = 0.773, p = 0.454,
d = 0.15, Fig.4a).
Furthermore, we performed a microscopic analysis of
TUNEL staining in the lesion area. The data were normally
distributed (Kolmogorov–Smirnov test, Z = 0.533, p = 0.938)
and the variances were homogenous across groups (Lev-
ené test, F(1,8) = 0.783, p = 0.402). On average, signifi-
cantly fewer dying cells in animals infused with PEDF
(100.3 ± 53.1 cells) than in the control group (247.1 ± 113.7
cells) were detected (Student’s t test, t(8) = 2.614, p = 0.031,
d = 1.65, n = 5 per group, Fig.4b).
To investigate whether PEDF affects BBB integrity after
MCAO, we first assessed Evans Blue extravasation in ani-
mals receiving no pump, pumps with CSF or PEDF and
that were euthanized right after MRI acquisition. While the
variances were homogenous across groups (Levené test,
F(2,20) = 1.206, p = 0.320), the data were not normally dis-
tributed (Kolmogorov–Smirnov test, Z = 1.417, p = 0.036).
There was no statistically significant difference between
the groups with a median of 495ng Evans Blue/g tissue
for no pump, and 624 and 446ng Evans Blue/g tissue for
pumps with CSF or PEDF, respectively (Kruskal–Wallis
test, χ2(2,n = 23) = 0.858, p = 0.651, η2 = 0.04, n = 11 for no
pump group, n = 6 for CSF and PEDF groups, Fig.4c).
Fig. 3 The origin of PEDF in the post-ischemic brain. a Representa-
tive pictures of the expression of PEDF in the post-ischemic brain,
from left to right at 6, 10, 14, and 24 h after MCAO for the ipsilat-
eral and at 10 h after MCAO for the contralateral hemisphere. The
pictures were taken at the border of the cortex and striatum. Scale
bars = 1000µm. b We performed a triple staining for PEDF, fibrino-
gen, and CD31 to visualize the extravasation from blood vessels. The
double-positive signals for PEDF and fibrinogen indicated blood-
borne origin of PEDF after MCAO. c Representative pictures of the
double staining for PEDF and platelet-derived growth factor receptor
(PDGFR) taken in the ipsilateral cortex and striatum suggest that per-
icytes also express PEDF at the lesion site. Scale bars = 200µm

Cellular and Molecular Neurobiology
1 3
We then studied the paracellular BBB permeability using
claudin-5 immunohistochemistry. We found that PEDF infu-
sion putatively increased the number of claudin-5-immuno-
reactive vessels (Fig.4d) at 24h after MCAO in both core
and penumbra. However, a quantification of the paracellular
BBB impairment (claudin-5 or other markers) is needed to
make firm conclusions.
Together and similar to our data on PEDF in the VEGF
hyperpermeability model, these data suggest that PEDF
may have an effect on the paracellular but not transcellu-
lar BBB disruption after MCAO as PEDF only increased
the number of claudin-5-positive vessels while leaving
Evans Blue extravasation unchanged. Future studies
quantifying the extent of impairment of the paracellular
BBB route (claudin-5 or other markers) will be helpful
to validate the mechanism of PEDF protection through
counteracting VEGF-induced hyperpermeability in cer-
ebral ischemia.
Fig. 4 The intraventricular infusion of PEDF decreases the edema,
cell death, and paracellular, but not transcellular BBB impairment
24 h after MCAO. a Shown are the representative T2-weighted
brain images of animals receiving intraventricular infusions of CSF
or PEDF (20 µg/ml in CSF). PEDF significantly reduced the hemi-
spheric volume increase due to edema, but not the lesion volumes.
The data are presented as mean ± SD, *p < 0.05. b PEDF improved
the post-ischemic outcome through its neuroprotective properties.
The TUNEL staining was performed in the lesion area discriminated
from the healthy tissue by H&E staining. Representative pictures are
shown for the CSF and PEDF mouse brains. The quantification of
the TUNEL staining revealed a significant decrease of the amount of
dead/dying cells in the PEDF-treated compared to the control group.
The data are presented as mean ± SD, *p < 0.05. Scale bar = 200µm.
c Shown is a representative image of an extracted brain. PEDF did
not decrease the Evans Blue extravasation after MCAO. The data are
presented as medians. d We assessed the paracellular BBB perme-
ability using claudin-5 immunohistochemistry co-stained with CD31.
Shown are representative pictures of the ischemic core, penumbra,
and the contralateral striatum of the CSF- and PEDF-treated animals.
The yellow staining indicates double-positive signal. Scale bars =
50µm

Cellular and Molecular Neurobiology
1 3
Discussion
The present study provides evidence that the intraparen-
chymal injection of PEDF in VEGF-induced hyperper-
meability decreases the paracellular, but not the transcel-
lular BBB permeability. After 60-min transient MCAO,
we detected transcellular BBB permeability starting at 6h
after reperfusion, whereas the paracellular BBB integrity
was disrupted only at 24h. We also demonstrate that the
levels of PEDF are increased within the first 24h after
MCAO and that PEDF originates from the blood and
endogenous pericytes. Finally, the intraventricular infusion
of PEDF decreased edema and cell death after MCAO,
potentially mediated by the improvement of the paracel-
lular route of BBB permeability (claudin-5) in the absence
of an amelioration of Evans Blue extravasation.
The substantial contribution of the BBB integrity to
stroke outcome has been recognized in the literature
(Brouns etal. 2011; Kaur and Ling 2008). The impair-
ment of both the paracellular and transcellular BBB per-
meability plays a role after cerebral ischemia (Knowland
etal. 2014). As an antipermeability agent, PEDF may have
a potential to restore the damaged BBB. Interestingly,
PEDF inhibited the VEGF-induced vascular permeability
to Evans Blue albumin in the eye (Liu etal. 2004). Cai
and co-workers observed a 50%-reduction of the VEGF-
induced permeability in a culture of retinal microvascular
endothelial cells treated with PEDF compared to a con-
trol culture which was treated with VEGF only (Cai etal.
2011). The authors also reported the restoration of the
tight junction protein claudin-5, which is normally lost
after VEGF application (Cai etal. 2011). In mice, the
intravitreal injection of PEDF together with VEGF abol-
ished the hyperpermeability in the eye, which was caused
by the injection of VEGF alone (Liu etal. 2004).
In our study, we investigated the effect of the intra-
parenchymal injection of PEDF on BBB integrity after
VEGF-induced hyperpermeability. We showed that the
intraparenchymal PEDF injections together with VEGF
did not change the Evans Blue extravasation, but reversed
the decrease in the tight junction protein claudin-5 caused
by the VEGF injections alone (Fig.1). This suggests that
PEDF affects the paracellular (as detected by the loss
of tight junction proteins with or without Evans Blue
extravasation), but not the transcellular BBB permeabil-
ity (as detected by an unchanged Evans Blue extravasa-
tion). In line with our findings, the downregulation of the
tight junction protein claudin-5 after the intraparenchy-
mal injection of VEGF into healthy mice was previously
shown (Argaw etal. 2009). Importantly, the co-injection
of 20-fold molar excess of PEDF neutralized the retinal
vascular hyperpermeability caused by intravitreal VEGF
injection (Liu etal. 2004). Doubling the amount of PEDF
did not increase its influence onto the claudin-5 expression
in our model.
In the second part of our study, we characterized the BBB
permeability within the first day after MCAO. In our study,
we detected the extravasation of fibrinogen in the lesion
site between 6 and 24h after MCAO with a peak at 6 and
24h. In contrast, the amount of claudin-5-immunoreactive
vessels decreased only at 24h in both core and penumbra
(Fig.2). This points towards an early transcellular BBB
permeability (extravasation of fibrinogen in the presence of
intact tight junctions), followed by a later paracellular BBB
permeability (loss of tight junction proteins and extravasa-
tion of fibrinogen); however, further quantitative analysis of
paracellular and transcellular BBB permeability is needed
to confirm our findings.
The BBB has been shown to open in a biphasic manner
following MCAO. Using Evans Blue and the near-infrared
fluorescence imaging of albumin, Klohs and colleagues
demonstrated BBB impairment between 4 and 8h as well
as 12 and 16h after 60min of MCAO in the mouse (Klohs
etal. 2009). Jiao and colleagues described the BBB impair-
ment using Evans Blue between 3 and 120h after 2h MCAO
in the rat with two peaks at 3 and 120h. This paralleled with
the downregulation of claudin-5 mRNA expression as well
as immunoreactivity (Jiao etal. 2011). Zhang and colleagues
observed decreased amounts of claudin-5 protein levels at
12h after 60min MCAO in the mouse (Zhang etal. 2009).
In another study, Jin etal. showed a decrease in claudin-5
staining starting at 4h after murine 60-min MCAO, which
persisted until 24 and 48h (Jin etal. 2011),
More recently, Knowland and colleagues created a trans-
genic mouse line whose endothelial tight junctions are
labeled with eGFP and imaged the tight junction changes as
well as the fluorescent tracer leakage across the BBB after
tMCAO invivo. They demonstrated that the BBB impair-
ment measured by albumin extravasation occurred as early
as 6h following 2h of MCAO, decreased between 12 and
24h, and peaked again at 48h. In contrast, the tight junc-
tion defects appeared only at 48h. The authors therefore
suggested that a stepwise impairment of the transcellular
followed by the paracellular barrier mechanisms accounts
for the BBB deficits in stroke (Knowland etal. 2014), which
supports our findings.
As PEDF ensures vascular tightness in the healthy brain,
we hypothesized that under hypoxic conditions, the intrinsic
PEDF levels would be significantly downregulated. In con-
trast to our expectations, we observed an increase in PEDF
levels at 24h after MCAO.
We also investigated the origin of PEDF in the ischemic
brain since we have previously demonstrated that the ele-
vated levels of PEDF in the CSF of different neurologi-
cal disorders may originate from the CNS or the systemic

Cellular and Molecular Neurobiology
1 3
circulation or both and that there is a large gradient to sup-
port the influx (Lang etal. 2017). We found the PEDF stain-
ing located around or very close to vessels (Fig.3). This
observation may partially explain why the BBB integrity
cannot be maintained in the brain despite high PEDF levels:
If PEDF is improperly delivered to the neurovascular units,
it will not be able to exert its function. We also know that it
is not solely the PEDF expression that determines the BBB
integrity, but rather the ratio of VEGF to PEDF (Cai etal.
2011; Tong etal. 2013). Based on the current literature that
reports increased levels of VEGF under ischemic conditions
(Lennmyr etal. 1998; Li etal. 2013; Matsuo etal. 2013), we
assume that the rise in the natural PEDF levels may not be
enough to counteract the increase in VEGF.
In addition to the influx of PEDF from the blood, we also
provide first evidence for its internal synthesis by pericytes
(Fig.3), which needs to be further investigated. PEDF was
initially discovered to be synthesized by retinal pigment
epithelial cells (Tombran-Tink 2005). In the healthy brain,
PEDF is produced by neurons and endothelial cells (van
Wagenen 2008). In addition, we know that PEDF can also be
expressed by astrocytes in the periinfarct area after ischemia.
In another experiment of our group, we found an increase in
PEDF-expressing astrocytes starting at day 3 after MCAO
(unpublished data). Sanagi and colleagues also demonstrated
such an increase at 7days after MCAO (Sanagi etal. 2008).
Besides the ability of PEDF to inhibit angiogenesis
through the VEGF receptor 1 (Cai etal. 2006), different
PEDF receptors have been proposed: (i) the PEDF receptor
[PEDF-R or patatin-like phospholipase domain containing
2 (PNPLA2-PEDF)] and (ii) the non-integrin 37/67kDa
laminin receptor (LR-PEDF). PEDF-R has been localized
on motor neurons, while LR-PEDF was found on endothelial
cells (Manalo etal. 2011). In the adult subventricular zone,
PEDF-R was localized to cells positive for glial fibrillary
acid protein and early neuronal lineage cells immunoreactive
for doublecortin, while oligodendroglial lineage cells and
astrocytes expressed PEDF-R in the corpus callosum (Sohn
etal. 2012). LR-PEDF has also been shown to be expressed
in the subventricular zone (Castro-Garcia etal. 2015). How-
ever, the role of the PEDF receptors in stroke remains to be
elucidated in the future.
Although, to our knowledge, the expression of PEDF by
pericytes has not been described previously, though the posi-
tive effect of PEDF on the retinal pericyte survival and stress
resistance was thoroughly studied (Yamagishi etal. 2005;
Zhang etal. 2008). Pericytes are important guardians of the
BBB integrity under hypoxic conditions (Al Ahmad etal.
2009). Their loss or dysfunction contributes to disrupted-
vessel-associated disorders, such as diabetic retinopathy
(Yamagishi etal. 2005). Thus, it may be speculated that
PEDF has different roles in the acute phase after MCAO
being expressed in pericytes than in the more chronic phase
when it starts to be expressed in astrocytes. Further experi-
ments are needed to elucidate the time course and to confirm
the involvement of autocrine pericytic PEDF signaling.
Finally, we investigated the effect of the intraventricu-
lar PEDF infusion on BBB impairment, cell death, edema
formation, and lesion volume in mice after MCAO. We
delivered PEDF or CSF via osmotic pumps into the lateral
ventricle contralateral to the ischemic lesion. We found a
reduction of the edema, but not the lesion volumes in the
mice receiving intraventricular PEDF infusions compared
to the CSF-treated animals using MRI (Fig.4). The edema
reduction by PEDF administration was previously described
in rat MCAO models (Pillai etal. 2013; Sanagi etal. 2008)
and in a cold injury-induced brain edema model in mice
(Jinnouchi etal. 2007). However, in our previous study using
PEDF infusions into the ventricle of MCAO mice, we did
not observe a reduction in lesion size and edema (Zille etal.
2014).
Furthermore, we found a reduction of cell death in the
PEDF-treated group (Fig.4). This is in line with the neu-
roprotective properties of PEDF described in neuronal cell
culture (Falk etal. 2009; Taniwaki etal. 1995, 1997) as well
as in a murine model of traumatic brain injury (Zille etal.
2014) and MCAO in rats (Sanagi etal. 2008). In a previ-
ous study of our group using PEDF infusions, we did not
find a reduction of cell death at 21days after MCAO (Zille
etal. 2014). However, in the present study, cell death was
assessed at 24h following MCAO. The reason for the dif-
ference in the results may be due to the different time points
investigated.
MCAO also led to an impairment of the BBB, affect-
ing both the transcellular (Evans Blue) and the paracellular
(Claudin-5) BBB route. Similar to our findings in VEGF-
induced hyperpermeability, the infusion of PEDF potentially
attenuated the decrease in Claudin-5, but had no effect on the
Evans Blue extravasation (Fig.4). This suggests that PEDF
improves BBB function after cerebral ischemia, which
may be mediated through reducing the paracellular route
of BBB permeability as indirectly inferred by the improve-
ment of edema after PEDF in the absence of an amelioration
of Evans blue extravasation. However, a quantification of
the paracellular BBB impairment is needed to make firm
conclusions.
For the therapeutic application of PEDF, it is crucial to
consider the timing as well as the dosing of PEDF. Since
PEDF counteracts VEGF, it is likely more beneficial dur-
ing the acute stages after stroke, when VEGF induces BBB
hyperpermeability, which is detrimental. Later after stroke,
angiogenesis, which is mediated among others by VEGF, is
required for better outcomes (Ruan etal. 2015), and PEDF
may inhibit proper recovery at that time point.
In summary, our data suggest that PEDF plays a role in
the paracellular but not the transcellular BBB integrity in

Cellular and Molecular Neurobiology
1 3
the normal brain and after focal cerebral ischemia. After
MCAO, the BBB permeability appears to be first transcellu-
lar followed by a later breakdown of tight junction proteins.
Further studies need to be performed to elucidate the mecha-
nisms by which PEDF is neuroprotective and influences the
paracellular BBB integrity after stroke.
Acknowledgements We would like to thank Prof. Ulrich Dirnagl and
Prof. Peter Vajkoczy for their support and the helpful discussion of
the manuscript.
Author Contributions ALP initiated the study and provided the origi-
nal idea, designed all experiments, and performed in part the intra-
parenchymal experiments, JK developed further ideas on the study
and designed the intraparenchymal experiments. AR designed and per-
formed most experiments. AR, RC, and MNK performed the MCAO
surgery. AR and MYT performed the ELISA and staining experiments.
AR and MZ analyzed the data. AR and MZ performed the statistical
analysis and graphical artwork. AR and MZ wrote and edited the paper.
AR and MZ contributed equally to the study. All authors discussed the
results and commented on the manuscript.
Funding This work was possible thanks to the financial support
from the Departments of Neurosurgery and Experimental Neurol-
ogy, by the Charité Rahel Hirsch Habilitation Fellowship and Berlin-
Brandenburg Center for Regenerative Therapies Flexible Funds (No.
BCRTFF2008-17 and BCRTFF2009-38) all granted to Dr. Ana-Luisa
Pina. Marietta Zille was supported by a fellowship granted by the Stif-
tung der Deutschen Wirtschaft (Foundation of German Business).
Compliance with Ethical Standards
Conflict of interest The authors declare that they have no conflict of
interest.
Ethical Approval All animal experiments were performed in accord-
ance with the national and international guidelines for the care and
use of laboratory animals (Tierschutzgesetz der Bundesrepublik
Deutschland, European directive, as well as GV-SOLAS and FELASA
guidelines and recommendations for laboratory animal welfare). The
studies were approved by an ethics committee (Landesamt für Gesund-
heit und Soziales, Berlin, Germany, permit number G 0270/10).
References
Abbott NJ, Patabendige AA, Dolman DE, Yusof SR, Begley DJ (2010)
Structure and function of the blood–brain barrier. Neurobiol Dis
37:13–25. https ://doi.org/10.1016/j.nbd.2009.07.030
Al Ahmad A, Gassmann M, Ogunshola OO (2009) Maintaining blood–
brain barrier integrity: pericytes perform better than astrocytes
during prolonged oxygen deprivation. J Cell Physiol 218:612–622.
https ://doi.org/10.1002/jcp.21638
Allen CL, Bayraktutan U (2009) Oxidative stress and its role in the
pathogenesis of ischaemic stroke. Int J Stroke 4:461–470. https ://
doi.org/10.1111/j.1747-4949.2009.00387 .x
Argaw AT, Gurfein BT, Zhang Y, Zameer A, John GR (2009) VEGF-
mediated disruption of endothelial CLN-5 promotes blood–brain
barrier breakdown. Proc Natl Acad Sci USA 106:1977–1982.
https ://doi.org/10.1073/pnas.08086 98106
Belayev L, Busto R, Zhao W, Ginsberg MD (1996) Quantitative evalua-
tion of blood–brain barrier permeability following middle cerebral
artery occlusion in rats. Brain Res 739:88–96
Bernacki J, Dobrowolska A, Nierwińska K, Małecki A (2008) Physiol-
ogy and pharmacological role of the blood–brain barrier. Phar-
macol Rep 60:600–622
Brouns R, Wauters A, De Surgeloose D, Mariën P, De Deyn PP (2011)
Biochemical markers for blood–brain barrier dysfunction in acute
ischemic stroke correlate with evolution and outcome. Eur Neurol
65:23–31. https ://doi.org/10.1159/00032 1965
Cai J, Jiang WG, Grant MB, Boulton M (2006) Pigment epithelium-
derived factor inhibits angiogenesis via regulated intracellular pro-
teolysis of vascular endothelial growth factor receptor 1. J Biol
Chem 281:3604–3613. https ://doi.org/10.1074/jbc.M5074 01200
Cai J et al (2011) PEDF regulates vascular permeability by a
γ-secretase-mediated pathway. PLoS ONE 6:e21164. https ://doi.
org/10.1371/journ al.pone.00211 64
Cardoso FL, Brites D, Brito MA (2010) Looking at the blood–brain
barrier: molecular anatomy and possible investigation approaches.
Brain Res Rev 64:328–363. https ://doi.org/10.1016/j.brain resre
v.2010.05.003
Castro-Garcia P etal (2015) Defects in subventricular zone pigmented
epithelium-derived factor niche signaling in the senescence-
accelerated mouse prone-8. FASEB J 29:1480–1492. https ://doi.
org/10.1096/.13-24444 2
Chen B, Friedman B, Cheng Q, Tsai P, Schim E, Kleinfeld D, Lyden
PD (2009) Severe blood–brain barrier disruption and surrounding
tissue injury. Stroke 40:e666–674. https ://doi.or g/10.1161/STROK
EAHA.109.55134 1
Dawson DW, Volpert OV, Gillis P, Crawford SE, Xu H, Benedict W,
Bouck NP (1999) Pigment epithelium-derived factor: a potent
inhibitor of angiogenesis. Science 285:245–248
Duh EJ etal (2002) Pigment epithelium-derived factor suppresses
ischemia-induced retinal neovascularization and VEGF-induced
migration and growth. Invest Ophthalmol Vis Sci 43:821–829
Falk T, Zhang S, Sherman SJ (2009) Pigment epithelium derived factor
(PEDF) is neuroprotective in two invitro models of Parkinson’s
disease. Neurosci Lett 458:49–52. https ://doi.org/10.1016/j.neule
t.2009.04.018
Fujimura T etal (2009) Administration of pigment epithelium-derived
factor (PEDF) reduces proteinuria by suppressing decreased
nephrin and increased VEGF expression in the glomeruli of adria-
mycin-injected rats. Nephrol Dial Transplant 24:1397–1406. https
://doi.org/10.1093/ndt/gfn65 9
Gao G, Li Y, Gee S, Dudley A, Fant J, Crosson C, Ma JX (2002) Down-
regulation of vascular endothelial growth factor and up-regulation
of pigment epithelium-derived factor: a possible mechanism for
the anti-angiogenic activity of plasminogen kringle 5. J Biol Chem
277:9492–9497. https ://doi.org/10.1074/jbc.M1080 04200
Guan M etal (2003) Loss of pigment epithelium derived factor expres-
sion in glioma progression. J Clin Pathol 56:277–282
Huang ZG, Xue D, Preston E, Karbalai H, Buchan AM (1999) Bipha-
sic opening of the blood–brain barrier following transient focal
ischemia: effects of hypothermia. Can J Neurol Sci 26:298–304
Hutchings H, Maitre-Boube M, Tombran-Tink J, Plouët J (2002) Pig-
ment epithelium-derived factor exerts opposite effects on endothe-
lial cells of different phenotypes. Biochem Biophys Res Commun
294:764–769. https ://doi.org/10.1016/S0006 -291X(02)00548 -X
Jiao H, Wang Z, Liu Y, Wang P, Xue Y (2011) Specific role of tight
junction proteins claudin-5, occludin, and ZO-1 of the blood–
brain barrier in a focal cerebral ischemic insult. J Mol Neurosci
44:130–139. https ://doi.org/10.1007/s1203 1-011-9496-4
Jin R etal (2011) Phosphatidylinositol-3-kinase gamma plays a cen-
tral role in blood–brain barrier dysfunction in acute experimental
stroke. Stroke 42:2033–2044. https ://doi.org/10.1161/STROK
EAHA.110.60136 9

Cellular and Molecular Neurobiology
1 3
Jinnouchi Y etal (2007) Administration of pigment epithelium-
derived factor (PEDF) inhibits cold injury-induced brain edema
in mice. Brain Res 1167:92–100. https ://doi.org/10.1016/j.brain
res.2007.04.088
Kaur C, Ling EA (2008) Blood brain barrier in hypoxic-ischemic con-
ditions. Curr Neurovasc Res 5:71–81
Klohs J etal (2009) Near-infrared fluorescence imaging with fluores-
cently labeled albumin: a novel method for non-invasive optical
imaging of blood–brain barrier impairment after focal cerebral
ischemia in mice. J Neurosci Methods 180:126–132. https ://doi.
org/10.1016/j.jneum eth.2009.03.002
Knowland D etal (2014) Stepwise recruitment of transcellular and
paracellular pathways underlies blood–brain barrier breakdown
in stroke. Neuron 82:603–617. https ://doi.org/10.1016/j.neuro
n.2014.03.003
Komarova Y, Malik AB (2010) Regulation of endothelial permeabil-
ity via paracellular and transcellular transport pathways. Annu
Rev Physiol 72:463–493. https ://doi.org/10.1146/annur ev-physi
ol-02190 9-13583 3
Lang V etal (2017) Alzheimer’s disease: Elevated pigment epithelium-
derived factor in the cerebrospinal fluid is mostly of systemic
origin. J Neurol Sci 375:123–128. https ://doi.org/10.1016/j.
jns.2017.01.051
Lennmyr F, Ata KA, Funa K, Olsson Y, Terent A (1998) Expression
of vascular endothelial growth factor (VEGF) and its receptors
(Flt-1 and Flk-1) following permanent and transient occlusion of
the middle cerebral artery in the rat. J Neuropathol Exp Neurol
57:874–882
Li YN, Pan R, Qin XJ, Yang WL, Qi Z, Liu W, Liu KJ (2013) Ischemic
neurons activate astrocytes to disrupt endothelial barrier via
increasing VEGF expression. J Neurochem 10000:10000. https
://doi.org/10.1111/jnc.12611
Liu H, Ren JG, Cooper WL, Hawkins CE, Cowan MR, Tong PY (2004)
Identification of the antivasopermeability effect of pigment epi-
thelium-derived factor and its active site. Proc Natl Acad Sci USA
101:6605–6610. https ://doi.org/10.1073/pnas.03083 42101
Manalo KB, Choong PF, Becerra SP, Dass CR (2011) Pigment epithe-
lium-derived factor as an anticancer drug and new treatment meth-
ods following the discovery of its receptors: a patent perspective.
Expert Opin Ther Pat 21:121–130. https ://doi.org/10.1517/13543
776.2011.54534 7
Matsuo R etal (2013) Clinical significance of plasma VEGF value
in ischemic stroke—research for biomarkers in ischemic
stroke (REBIOS) study. BMC Neurol 13:32. https ://doi.
org/10.1186/1471-2377-13-32
Pillai DR etal (2009) Cerebral ischemia-reperfusion injury in rats—
a 3 T MRI study on biphasic blood–brain barrier opening and
the dynamics of edema formation. J Cereb Blood Flow Metab
29:1846–1855. https ://doi.org/10.1038/jcbfm .2009.106
Pillai DR, Shanbhag NC, Dittmar MS, Bogdahn U, Schlachetzki F
(2013) Neurovascular protection by targeting early blood–brain
barrier disruption with neurotrophic factors after ischemia-reper-
fusion in rats. J Cereb Blood Flow Metab 33:557–566. https ://doi.
org/10.1038/jcbfm .2012.201
Proescholdt MA, Heiss JD, Walbridge S, Muhlhauser J, Capogrossi
MC, Oldfield EH, Merrill MJ (1999) Vascular endothelial growth
factor (VEGF) modulates vascular permeability and inflammation
in rat brain. J Neuropathol Exp Neurol 58:613–627. https ://doi.
org/10.1097/00005 072-19990 6000-00006
Ruan L, Wang B, ZhuGe Q, Jin K (2015) Coupling of neurogenesis
and angiogenesis after ischemic stroke. Brain Res 1623:166–173.
https ://doi.org/10.1016/j.brain res.2015.02.042
Ryu JK, McLarnon JG (2009) A leaky blood–brain barrier, fibrinogen
infiltration and microglial reactivity in inflamed Alzheimer’s dis-
ease brain. J Cell Mol Med 13:2911–2925. https ://doi.org/10.111
1/j.1582-4934.2008.00434 .x
Sanagi T, Yabe T, Yamada H (2008) Gene transfer of PEDF attenuates
ischemic brain damage in the rat middle cerebral artery occlu-
sion model. J Neurochem 106:1841–1854. https ://doi.org/10.11
11/j.1471-4159.2008.05529 .x
Schubert W, Frank PG, Razani B, Park DS, Chow CW, Lisanti MP
(2001) Caveolae-deficient endothelial cells show defects in the
uptake and transport of albumin invivo. J Biol Chem 276:48619–
48622. https ://doi.org/10.1074/jbc.C1006 13200
Sohn J etal (2012) PEDF is a novel oligodendrogenic morphogen act-
ing on the adult SVZ and corpus callosum. J Neurosci 32:12152–
12164. https ://doi.org/10.1523/JNEUR OSCI.0628-12.2012
Stellmach V, Crawford SE, Zhou W, Bouck N (2001) Prevention of
ischemia-induced retinopathy by the natural ocular antiangiogenic
agent pigment epithelium-derived factor. Proc Natl Acad Sci USA
98:2593–2597. https ://doi.org/10.1073/pnas.03125 2398
Taniwaki T, Becerra SP, Chader GJ, Schwartz JP (1995) Pigment epi-
thelium-derived factor is a survival factor for cerebellar granule
cells in culture. J Neurochem 64:2509–2517
Taniwaki T, Hirashima N, Becerra SP, Chader GJ, Etcheberrigaray R,
Schwartz JP (1997) Pigment epithelium-derived factor protects
cultured cerebellar granule cells against glutamate-induced neu-
rotoxicity. J Neurochem 68:26–32
Tombran-Tink J (2005) The neuroprotective and angiogenesis inhibi-
tory serpin, PEDF: new insights into phylogeny, function, and
signaling. Front Biosci 10:2131–2149
Tong N etal (2013) Diosmin alleviates retinal edema by protecting the
blood-retinal barrier and reducing retinal vascular permeability
during ischemia/reperfusion injury. PLoS ONE 8:e61794. https
://doi.org/10.1371/journ al.pone.00617 94
Ueda S, Yamagishi SI, Okuda S (2010) Anti-vasopermeability effects
of PEDF in retinal-renal disorders. Curr Mol Med 10:279–283
Uyama O, Okamura N, Yanase M, Narita M, Kawabata K, Sugita M
(1988) Quantitative evaluation of vascular permeability in the ger-
bil brain after transient ischemia using Evans blue fluorescence.
J Cereb Blood Flow Metab 8:282–284. https ://doi.org/10.1038/
jcbfm .1988.59
van Wagenen S (2008) Protein expression and localization of Pigment
Epithelium Derived Factor in the central nervous system of the rat
during postnatal development. Universität Regensburg, Regens-
burg. https ://epub.uni-regen sburg .de/10559 /
Wang JJ, Zhang SX, Lu K, Chen Y, Mott R, Sato S, Ma JX (2005)
Decreased expression of pigment epithelium-derived factor is
involved in the pathogenesis of diabetic nephropathy. Diabetes
54:243–250
Weiss N, Miller F, Cazaubon S, Couraud PO (2009) The blood–brain
barrier in brain homeostasis and neurological diseases. Biochim
Biophys Acta 1788:842–857. https ://doi.org/10.1016/j.bbame
m.2008.10.022
Yabe T, Wilson D, Schwartz JP (2001) NFkappaB activation is
required for the neuroprotective effects of pigment epithelium-
derived factor (PEDF) on cerebellar granule neurons. J Biol Chem
276:43313–43319. https ://doi.org/10.1074/jbc.M1078 31200
Yamagishi S, Amano S, Inagaki Y, Okamoto T, Takeuchi M, Inoue H
(2003) Pigment epithelium-derived factor inhibits leptin-induced
angiogenesis by suppressing vascular endothelial growth factor
gene expression through anti-oxidative properties. Microvasc Res
65:186–190
Yamagishi S, Nakamura K, Takenaka K, Matsui T, Jinnouchi Y,
Imaizumi T (2005) Pigment epithelium-derived factor (PEDF)
promotes growth of pericytes through autocrine production of
platelet-derived growth factor-B. Microvasc Res 69:128–134.
https ://doi.org/10.1016/j.mvr.2005.04.001
Yamagishi S, Abe R, Jinnouchi Y, Matsui T, Imaizumi T, Inoue H
(2007) Pigment epithelium-derived factor inhibits vascular
endothelial growth factor-induced vascular hyperpermeability
both invitro and invivo. J Int Med Res 35:896–899

Cellular and Molecular Neurobiology
1 3
Yang J, Duh EJ, Caldwell RB, Behzadian MA (2010) Antipermeability
function of PEDF involves blockade of the MAP kinase/GSK/
beta-catenin signaling pathway and uPAR expression. Invest
Ophthalmol Vis Sci 51:3273–3280. https ://doi.org/10.1167/
iovs.08-2878
Zhang SX etal (2008) Pigment epithelium-derived factor mitigates
inflammation and oxidative stress in retinal pericytes exposed to
oxidized low-density lipoprotein. J Mol Endocrinol 41:135–143.
https ://doi.org/10.1677/JME-08-0011
Zhang C, An J, Haile WB, Echeverry R, Strickland DK, Yepes M
(2009) Microglial low-density lipoprotein receptor-related pro-
tein 1 mediates the effect of tissue-type plasminogen activator
on matrix metalloproteinase-9 activity in the ischemic brain. J
Cereb Blood Flow Metab 29:1946–1954. https ://doi.org/10.1038/
jcbfm .2009.174
Zille M etal (2014) Influence of pigment epithelium-derived factor
on outcome after striatal cerebral ischemia in the mouse. PLoS
ONE 9:e114595. https ://doi.org/10.1371/journ al.pone.01145 95
Publisher’s Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.