Báo cáo khoa học: AKAP12 in astrocytes induces barrier functions in human endothelial cells through protein kinase Cf pptx

AKAP12 in astrocytes induces barrier functions in human
endothelial cells through protein kinase Cf
Yoon Kyung Choi and Kyu-Won Kim
NeuroVascular Coordination Research Center, College of Pharmacy and Research, Institute of Pharmaceutical Sciences,
Seoul National University, Korea
The interaction of astrocytes and blood vessels plays
an important role in retinal vascular development and
angiogenesis [1]. Astrocytes act as a sensor and guide
for the developing retinal vasculature by detecting
hypoxia, and responding by increasing the expression
of hypoxia-inducible angiogenic factors, such as
hypoxia-inducible factor-1a (HIF-1a) and vascular
endothelial growth factor (VEGF) [2,3]. Astrocyte-
derived VEGF is the most potent angiogenic factor,
and promotes endothelial cell proliferation, migration
and permeability during hypoxia [3–6]. Thrombospon-
din-1 (TSP-1) is an antiangiogenic factor that inhibits
angiogenesis in vivo [7]. TSP-1 mRNA and protein
levels are significantly reduced by oxygen ⁄ glucose
deprivation in cerebral endothelial cells [8]. TSP-1
belongs to a family of secreted glycoproteins and is a
constitutive component of the basement membrane,
which plays an integral role in the differentiation and
migration of endothelial cells during angiogenesis [9,10].
Src-suppressed C kinase substrate is the rodent
ortholog of human gravin. Src-suppressed C kinase sub-
strate and gravin have been redesignated as A-kinase
anchor protein 12 (AKAP12) [11]. AKAP12 acts as a
multivalent scaffold protein, and has been shown to
associate with protein kinase C (PKC), protein kina-
se A, calmodulin, cyclins, F-actin, and b-adrenergic

receptors [11–13]. Thus, AKAP12 functions as a
dynamic platform for signal transduction. AKAP12 is
Keywords
AKAP12 (A-kinase anchor protein 12);
blood–neural barrier; protein kinase Cf;
thrombospondin-1; vascular endothelial
growth factor
Correspondence
K W. Kim, NeuroVascular Coordination
Research Center, College of Pharmacy,
Seoul National University, Seoul 151-742,
Korea
Fax: +82 2 872 1795
Tel: +82 2 880 6988
E-mail:
(Received 12 November 2007, revised 2
March 2008, accepted 7 March 2008)
doi:10.1111/j.1742-4658.2008.06387.x
Interactions between astrocytes and blood vessels are essential for the for-
mation and maintenance of the blood–neural barrier (BNB). Astrocyte-
derived A-kinase anchor protein 12 (AKAP12) influences BNB formation,
but the mechanism of regulation of BNB functions by AKAP12 is not fully
understood. We have defined a new pathway of barriergenesis in human
retina microvascular endothelial cells (HRMECs) involving astrocytic
AKAP12. Treatment of HRMECs with conditioned media from AKAP12-
overexpressing astrocytes reduced phosphorylation of protein kinase Cf
(PKCf), decreased the levels of vascular endothelial growth factor (VEGF)
mRNA and protein, and increased thrombospondin-1 (TSP-1) levels, which
led to antiangiogenesis and barriergenesis. Transfection of a small interfer-
ence RNA targeting PKCf decreased VEGF levels and increased TSP-1

levels in HRMECs. Rho is a putative downstream signal of PKCf, and
inhibition of Rho kinase with a specific inhibitor, Y27632, decreased
VEGF levels and increased TSP-1 levels. We therefore suggest that
AKAP12 in astrocytes differentially regulates the expression of VEGF and
TSP-1 via the inhibition of PKCf phosphorylation and Rho kinase activity
in HRMECs.
Abbreviations
AKAP12, A-kinase anchor protein 12; Ang1, angiopoietin-1; CM, conditioned medium; COMP, cartilage oligomeric matrix protein; DAPI, 4¢,
6-diamidino-2-phenylindole; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; H, hypoxia; HIF-1a, hypoxia-inducible factor-1a; HRMEC,
human retina microvascular endothelial cell; N, normoxia; NC, negative control; PKC, protein kinase C; RITC, rhodamine B isothiocyanate;
si, small interfering; TSP-1, thrombospodin-1; VEGF, vascular endothelial growth factor.
2338 FEBS Journal 275 (2008) 2338–2353 ª 2008 The Authors Journal compilation ª 2008 FEBS
involved in the regulation of the actin cytoskeleton,
cell morphology, cell adhesion, and cell spreading
[11,12,14]. Cytoskeletal remodeling is regulated by
intracellular signaling pathways that involve PKC, cal-
cium-regulated signaling, and the Rho family GTPases
[15–18]. In addition, Rho has been shown to induce
HIF-1a transactivation and VEGF expression [19].
Previously, we showed that AKAP12 in astrocytes is
important for regulating the formation of the mouse
blood–brain and human blood–retinal barriers by
downregulating HIF-1a-mediated VEGF expression
[20,21].
The effects of PKC activation on endothelial cell
permeability are of great interest. PKC belongs to a
family of serine ⁄ threonine kinases that are involved in
signal transduction. Isoforms of PKC are classified
according to their structure, activity, and substrate
requirements, and comprise three main classes: classic

PKCs (a, bI, bII, c), novel PKCs (d, e, g, h), and atyp-
ical PKCs (f, k) [22]. The activity of PKCs is regulated
by localization to the plasma membrane and phos-
phorylation status [22]. Several studies have shown
that PKCs increase vascular permeability [23–27]. One
of the atypical PKCs, PKCf, upregulates HIF-1a
activity and VEGF expression in renal cell carcinoma
cells [28], and increases thrombin-induced vascular
permeability in human umbilical vein endothelial
cells [25].
In the current study, we investigated the mechanism
of reduced vascular permeability and angiogenesis
caused by astrocytic AKAP12. We show that astro-
cytic AKAP12 inhibits phosphorylation of PKCf in
neighboring human retina microvascular endothelial
cells (HRMECs), which leads to a decrease in VEGF
levels and an increase in TSP-1 levels in HRMECs,
resulting in reduced vascular permeability, decreased
endothelial cell migration, and upregulation of tight
junction proteins. In addition, VEGF
165
treatment of
HRMECs induced VEGF mRNA levels via a positive
feedback mechanism of regulation and decreased TSP-
1 levels, suggesting that the differential regulation of
VEGF and TSP-1 in HRMECs by astrocytic AKAP12
could stem from the astrocyte-secreted factor VEGF.
Results
Effects of AKAP12-overexpressing astrocytes on
endothelial cell migration, vascular permeability,

and the levels of tight junction proteins
In the unvascularized retina during eye development,
astrocytes detect hypoxia, and respond by inducing
the expression of VEGF, which stimulates new vessel
formation [2,3]. To better understand the effect of
astrocytic AKAP12 on the retinal vasculature, we trea-
ted HRMECs with conditioned medium (CM) from
mock-transfected astrocytes that were exposed to
normoxia (N-mock-CM) or hypoxia (H-mock-CM), or
Akap12-transfected astrocytes that were exposed to
hypoxia (H-AKAP12-CM). After the treatment with
CM, we examined endothelial cell migration, vascular
permeability, and the levels of tight junction proteins
and an adhesion molecule. H-mock-CM markedly
increased HRMEC migration as compared to N-mock-
CM (Fig. 1A,B). When cells were incubated with
H-AKAP12-CM, reduced migration in comparison
with that of cells incubated with H-mock-CM was
observed, and this reduced migration was similar to
that in N-mock-CM-treated cells (Fig. 1A,B). We also
examined the passage of rhodamine B isothiocyanate
(RITC)–dextran through monolayers of HRMECs as a
measure of permeability, and found that vascular per-
meability was increased by H-mock-CM, and signifi-
cantly reduced by H-AKAP12-CM (Fig. 1C). We next
investigated whether AKAP12 regulated the expression
of tight junction proteins and an adhesion molecule,
vascular endothelial (VE)-cadherin. H-mock-CM
strongly decreased the expression of the junction pro-
teins ZO-1 and ZO-2, as well as VE-cadherin, whereas

H-AKAP12-CM significantly increased the expression
of ZO-1, ZO-2 and VE-cadherin in HRMECs
(Fig. 1D). The increased levels of the junction proteins
in the H-AKAP12- CM-treated cells were almost same
as the levels in the N-mock-CM-treated cells (Fig. 1D).
In our previous study, transfection of Akap12 into
human astrocytes increased angiopoietin-1 (Ang1) lev-
els in CM under hypoxic conditions, and this played a
role in barrier properties in HRMECs [20]. Therefore,
we examined whether the junction proteins claudin-1
and VE-cadherin were also regulated by Ang1 in CM
from hypoxic astrocytes. H-AKAP12-CM strongly
increased the expression of claudin-1 and VE-cadherin
as compared to H-mock-CM in HRMECs (Fig. 1E).
These effects were blocked when the H-AKAP12-CM
was pretreated with an antibody to Ang1 (Fig. 1E).
These results suggest that astrocytic AKAP12 plays a
role in endothelial barrier function through astrocyte-
derived secretion factor(s), including Ang1.
AKAP12 in astrocytes differentially regulates the
expression of VEGF and TSP-1 in HRMECs
VEGF is an important signal for retinal vessel migra-
tion and permeability [29]. TSP-1 is involved in endo-
thelial cell differentiation and migration [30,31].
Observing that CM from AKAP12-overexpressing
Y. K. Choi and K W. Kim Regulation of the blood–neural barrier by AKAP12
FEBS Journal 275 (2008) 2338–2353 ª 2008 The Authors Journal compilation ª 2008 FEBS 2339
N-mock-CM
A
B

D
E
C
H-mock-CM H-AKAP12-CM
0
1
2
3
4
5
RITC passed (µg·mL
–1
)
*
#
ZO-1
ZO-2
Actin
VE-cadherin
Actin
Cytosol
Membrane
CM :
VE-cadherin
Actin
Claudin-1
0
50
100
150

VE-cadherin
Claudin-1
H-M
H-A
H-A/Ang1
: CMH-A/IgG
#
*
*
**
*
*
0
50
100
150
ZO-1
ZO-2
VE-cadherin
#
*
*
*
**
**
H-M
H-A
H-A/Ang1
CM :
H-A/I

g
G
Relative expression
of proteins (%)
Relative expression
of proteins (%)
N-mock H-mock H-AKAP12 N-mock H-mock H-AKAP12
CM :
N-mock H-mock H-AKAP12
CM :
N-mock H-mock H-AKAP12CM :
0
50
100
Migrated area (%)
125
25
75
#
*
Fig. 1. AKAP12 in astrocytes regulates endothelial cell migration, vascular permeability, and the expression of tight junction proteins in
HRMECs. (A) CM from mock-transfected astrocytes exposed to normoxia (N-mock-CM) or hypoxia (H-mock-CM) and CM from Akap12-trans-
fected astrocytes exposed to hypoxia (H-AKAP12-CM) were collected and concentrated (4·). HRMECs were treated with N-mock-CM,
H-mock-CM or H-Akap12-CM for 24 h, and migration was observed. (B) HRMECs were marked with an injury line and the distance of the injury
line was measured. After 24 h, the quantification of the area of migration was performed (n = 4). The area was set to 100% in the H-mock-CM
treatment condition.
#
P < 0.001 as compared to the N-mock-CM condition; *P < 0.05 as compared with the H-mock-CM condition. (C) RITC–
dextran permeability as a marker for vascular permeability was analyzed in HRMECs. *P < 0.005 as compared to the N-mock-CM condition;
#

P < 0.005 as compared to the H-mock-CM condition (n = 4). (D) HRMECs were treated for 24 h with N-mock-CM, H-mock-CM or H-AKAP12-
CM. The expression of ZO-1 and ZO-2 in cytosolic fractions and the expression of VE-cadherin in membrane fractions were analyzed by wes-
tern blot. The quantification of these protein levels from three independent experiments is shown on the right. The expression was set to
100% in the N-mock-CM treatment condition. *P < 0.05 as compared to control;
#
P < 0.01 as compared to control; **P < 0.005 as compared
to control. In the case of the H-mock-CM condition, the control is the N-mock-CM condition; in the case of the H-AKAP12-CM condition, the
control is the H-mock-CM condition. (E) The expression of claudin-1 and VE-cadherin was detected in the membrane fraction in the H-M
(H-mock-CM), H-A (H-AKAP12-CM), H-A ⁄ Ang1 (H-AKAP12-CM neutralized by Ang1 antibody) and H-A ⁄ IgG (H-AKAP12-CM neutralized by IgG
antibody) treatment conditions by western blot assay. RMECs were treated for 24 h with these CMs. The quantification of these protein levels
from three independent experiments is shown on the right. The expression was set to 100% in the H-AKAP12-CM treatment condition.
*P < 0.05 as compared to control;
#
P < 0.01 as compared to control; **P < 0.005 as compared to control. In the case of H-A-CM, the control
is H-M-CM; in the case of H-A ⁄ Ang1-CM, the control is H-A-CM; in the case of H-A ⁄ IgG-CM, the control is H-A ⁄ Ang1-CM.
Regulation of the blood–neural barrier by AKAP12 Y. K. Choi and K W. Kim
2340 FEBS Journal 275 (2008) 2338–2353 ª 2008 The Authors Journal compilation ª 2008 FEBS
hypoxic astrocytes reduced endothelial cell migration
and vascular permeability (Fig. 1A–C), we investigated
the effects of AKAP12 on VEGF and TSP-1 expres-
sion in endothelial cells. HRMECs were treated with
CM from mock-transfected or Akap12-transfected hyp-
oxic astrocytes, and fluorescence immunohistochemis-
try was carried out to assess the levels of VEGF and
TSP-1 in cells. The level of VEGF was increased and
N-mock-CM H-mock-CM H-AKAP12-CM
0
50
100
150

*
#
GAPDH
VEGF
165
RT-PCR
RT-PCR
TSP-1
VEGF
TSP-1
VEGF
TSP-1
**
N-mock
H-mock H-AKAP12
CM :
VEGF
121
Relative expression
of proteins (%)
Relative expression
of mRNA (%)
Relative expression
of mRNA (%)
N
VEGF
165
VEGF
121
GAPDH

H/VEGF H/IgG
H
CM :
TSP-1
0
50
100
150
*
#
VEGF
TSP
-1
N-mockCM :
H-mock H-AKAP12
N-mockCM :
H-mock H-AKAP12
*
0
50
100
150
CM : N H/VEGF H/IgG
H
VEGF
TSP-1
*
*
*
*

*
*
**
A
B
C
Fig. 2. AKAP12 in astrocytes regulates the expression of VEGF and TSP-1 in HRMECs. (A) The expression of VEGF and TSP-1 (green) was
analyzed by fluorescence immunocytochemistry. Nuclei (blue) were stained with DAPI. Scale bar, 50 lm. Data from four independent experi-
ments were quantified, and are presented on the right. Quantification of immunohistochemical staining area was analyzed using
IMAGE-PRO
PLUS
(Media Cybernetics). Each stained area is presented relative to the area with the highest staining intensity. The level of VEGF in
H-mock-CM-treated cells was set at 100%. The level of TSP-1 in H-AKAP12-CM-treated cells was set at 100%. *P < 0.05 as compared to
N-mock-CM-treated cells;
#
P < 0.005 as compared to H-mock-CM-treated cells; **P < 0.05 as compared to H-mock-CM-treated cells. (B)
The mRNA levels of VEGF, TSP-1 and GAPDH were analyzed by RT-PCR. The quantification of VEGF and TSP-1 mRNA levels from three
independent experiments is shown on the right. The level of VEGF in H-mock-CM-treated cells was set at 100%. The level of TSP-1 in
N-mock-CM-treated cells was set at 100%. *P < 0.05 as compared to N-mock-CM-treated cells;
#
P < 0.001 as compared to H-mock-CM-
treated cells; **P < 0.05 as compared to H-mock-CM-treated cells. (C) The mRNA levels of VEGF, TSP-1 and GAPDH were analyzed in the
N (CM from astrocytes exposed to normoxia), H (CM from astrocytes exposed to hypoxia), H ⁄ VEGF (H-CM neutralized by an antibody to
VEGF) and H ⁄ IgG (H-CM neutralized by an antibody to IgG) treatment conditions by RT-PCR. The quantification of VEGF and TSP-1 mRNA
levels from three independent experiments is shown on the right. The level of VEGF in H ⁄ IgG-treated cells was set at 100%. The level of
TSP-1 in N-treated cells was set at 100%. *P < 0.05 as compared to control. In the case of H, the control is N; in the case of H ⁄ VEGF, the
control is H; in the case of H ⁄ IgG, the control is H ⁄ VEGF. GAPDH served as an internal control.
Y. K. Choi and K W. Kim Regulation of the blood–neural barrier by AKAP12
FEBS Journal 275 (2008) 2338–2353 ª 2008 The Authors Journal compilation ª 2008 FEBS 2341
the level of TSP-1 was decreased in the H-mock-CM-

treated cells as compared to the levels in the N-mock-
CM-treated cells (Fig. 2A). H-AKAP12-CM resulted
in decreased VEGF levels and increased TSP-1 levels,
and these levels were similar to the levels in N-mock-
CM-treated cells (Fig. 2A). These results indicate that
there is an inverse relationship between the levels of
VEGF and TSP-1 in endothelial cells, and that the
relationship is regulated by AKAP12-overexpressing
astrocytes. We further examined VEGF and TSP-1
mRNA levels in HRMECs, using RT-PCR. H-mock-
CM induced a marked increase in VEGF mRNA levels
as compared to N-mock-CM, whereas H-AKAP12-
CM decreased VEGF mRNA levels as compared to
H-mock-CM (Fig. 2B). TSP-1 mRNA levels exhibited
an inverse pattern of expression as compared to VEGF
(Fig. 2B). These results were consistent with the immu-
nocytochemical analysis (Fig. 2A), and indicated that
AKAP12 might regulate the secretion of astrocyte-
specific signal(s) that could reduce VEGF expression
and increase TSP-1 expression in endothelial cells.
Because astrocytes secrete a high level of VEGF during
hypoxia [20], we hypothesized that astrocyte-derived
VEGF functions as a stimulator of its own expression
in HRMECs. CM from hypoxic astrocytes (H-CM)
induced a marked increase in VEGF mRNA level as
compared to N-CM, whereas VEGF-neutralizing
H-CM (H ⁄ VEGF) decreased VEGF mRNA levels as
compared to H-CM (Fig. 2C). TSP-1 mRNA levels
exhibited an inverse pattern of expression as compared
to VEGF (Fig. 2C). These results suggest that the dif-

ferential regulation of VEGF and TSP-1 in HRMECs
by astrocytic AKAP12 could stem from the astrocyte-
secreted factor VEGF.
VEGF regulates the expression of VEGF and
TSP-1 in HRMECs
To confirm the effects of VEGF on the differential reg-
ulation of VEGF and TSP-1 in HRMECs, we treated
HRMECs with recombinant VEGF
165
. As shown in
Fig. 3A, VEGF mRNA levels in HRMECs were
increased following treatment with recombinant
VEGF
165
in a concentration-dependent manner. We
investigated whether treatment of HRMECs with
VEGF
165
influenced cell survival, because treatment
with VEGF
165
increased endothelial VEGF expression
(Fig. 3A). As shown in Fig. 3B, treatment with
VEGF
165
induced a 10% increase in cell viability
(P < 0.005). However, hypoxia did not increase cell
viability as compared to normoxia (Fig. 3B), although
hypoxia induces endothelial VEGF expression. Next,
we pretreated HRMECs with an antibody to Flk-1

(VEGF receptor 2), and followed this with VEGF
165
treatment. Treatment of HRMECs with VEGF
165
strongly increased the VEGF mRNA level, and this
effect was abolished by antibody blockade of
VEGF
165
–Flk-1 interactions (Fig. 3C). These results
suggest that treatment of HRMECs with VEGF
165
increased the expression of VEGF mRNA in part via
Flk-1 in HRMECs. We observed that treatment of
HRMECs with VEGF
165
induced intracellular VEGF
Fig. 3. VEGF regulates the expression of VEGF and TSP-1 in HRMECs. (A) HRMECs were treated with recombinant VEGF
165
(0, 10,
50 ngÆmL
)1
) for 24 h, and VEGF mRNA levels were analyzed by RT-PCR. Data from four independent experiments were quantified, and are
presented on the right. The level of VEGF in cells treated with 50 ngÆmL
)1
VEGF was set as 100%. *P < 0.05 as compared to cells treated
with 0 ngÆmL
)1
VEGF;
##
P < 0.0001 as compared to cells treated with 0 ngÆmL

)1
VEGF. (B) HRMECs were treated with recombinant
VEGF
165
(50 ngÆmL
)1
) or subjected to hypoxia for 24 h, and viable cells were evaluated by a cell viability assay (n = 5). The levels of viable
cells among cells treated with 50 ngÆmL
)1
VEGF or normoxic cells were set as 100%. **P < 0.005 as compared to cells treated with
0ngÆmL
)1
VEGF. (C) The VEGF mRNA level was detected by RT-PCR in the conditions with or without VEGF
165
treatment. Cells were pre-
treated with an antibody to Flk-1 or an antibody to IgG for 3 h, and this was followed by VEGF
165
treatment for 24 h. The quantification of
VEGF mRNA level from three independent experiments is shown on the right. The expression of VEGF mRNA in cells treated with VEGF
165
(50 ngÆmL
)1
) was set at 100%. **P < 0.005 as compared to control; *P < 0.05 as compared to control. In the case of the VEGF
165
treat-
ment condition, the control is no treatment condition; in the case of the Flk-1 neutralizing condition, the control is the VEGF
165
treatment
condition; in the case of the IgG neutralizing condition, the control is the Flk-1 neutralizing condition. (D) HRMECs were treated with a com-
bination of recombinant VEGF

165
(50 ngÆmL
)1
) and PKC inhibitors such as GF109203X (GF; 0.1 lM or 6 lM) and chelerythrine chloride (C;
1 l
M) for 24 h. The VEGF mRNA level was detected by RT-PCR. (E) HRMECs were treated with recombinant VEGF
165
(50 ngÆmL
)1
) for 4 or
24 h, and the level of TSP-1 secretion was analyzed by western blot. The data from five independent experiments were quantified, and are
presented on the right. The level of TSP-1 in the 0 ngÆmL
)1
VEGF treatment condition (control) was set at 100%. **P < 0.005 as compared
to control. (F) HRMECs were pretreated with an antibody to Flk-1 or an antibody to IgG for 3 h, and this was followed by VEGF
165
treatment
for 24 h. The TSP-1 secretion level was analyzed by western blot (n = 3). b-Actin, GAPDH and Ponceau Red staining served as the internal
controls. (G,H) The levels of PKCf and phosphorylated PKCf in HRMECs treated without (control) or with recombinant VEGF
165
(+VEGF) for
15 min were analyzed by fluorescence immunocytochemistry (G) and western blot (H). Scale bar, 50 lm. Nuclei (blue) were stained with
DAPI. Data from four independent experiments were quantified, and are presented on the right. Expression levels in control cells were set
at 100%. **P < 0.005 as compared to control.
Regulation of the blood–neural barrier by AKAP12 Y. K. Choi and K W. Kim
2342 FEBS Journal 275 (2008) 2338–2353 ª 2008 The Authors Journal compilation ª 2008 FEBS
VEGF
+
p-PKCζ
Actin

PKCζ
0
50
100
150
200
Relative expression
of proteins (%)
PKCζ
p-PKCζ
VEGF
+
**
0
120
Relative expression
of TSP-1 protein (%)
4 h 24 h
VEGF
165
(50 ng·mL
–1
)
40
80
TSP-1
4 h 24 h
VEGF
Ponceau
CM

**
**
10
A B
C D
E
G
H
F
50
VEGF
165
(ng·mL
–1
)
β-actin
VEGF
165
VEGF
121
0
120
Relative expression
of VEGF mRNA (%)
40
80
10 50
VEGF
##
*

p-PKCζ
Nuclei Merge
PKCζ
Nuclei Merge
+VEGF
Control
GAPDH
VEGF
165
VEGF
121
+
IgG
++
Flk-1
VEGF
165
(50 ng·mL
–1
)
+
IgG
++
Flk-1
VEGF
165
(50 ng·mL
–1
)
TSP-1

Ponceau
CM
0
40
80
Viable cells (relative
ratio to control %)
+
/
+
/
VEGF hypoxia
120
+
IgG
++
Flk-1
0
40
Relative expression
of VEGF mRNA (%)
80
120
VEGF
*
**
**
**
+
+++

β-actin
VEGF
0.1
6
1C (μ
M
)
VEGF
165
GF (μ
M
)
Y. K. Choi and K W. Kim Regulation of the blood–neural barrier by AKAP12
FEBS Journal 275 (2008) 2338–2353 ª 2008 The Authors Journal compilation ª 2008 FEBS 2343
mRNA levels (Fig. 3A), and this increase was blocked
by PKC inhibitors such as bisindolylmaleimide I
(GF109203X) and chelerythrine chloride (Fig. 3D).
Chelerythrine chloride, an inhibitor of all PKC iso-
forms, decreased VEGF
165
-stimulated VEGF mRNA
levels in HRMECs (Fig. 3D). GF109203X has an
indolylmaleimide structure that inhibited all PKC
isoforms in cells treated with 6 lm, and inhibited classic
and novel PKCs in cells treated with 0.1 lm [25].
VEGF mRNA levels were more reduced in cells trea-
ted with 6 lm GF109203X than in cells treated with
with 0.1 lm GF109203X (Fig. 3D). These results sug-
gest that atypical PKC isoforms may play a role in
VEGF

165
-stimulated VEGF induction. We also exam-
ined whether VEGF treatment influenced the expres-
sion of TSP-1, as it has been shown that the
expression of VEGF and TSP-1 are inversely regulated
[32,33]. We observed a significant reduction in the
levels of secreted TSP-1 at 4 h after VEGF treatment,
followed by a partial recovery at 24 h after VEGF
treatment (Fig. 3E). In addition, treatment of
HRMECs with VEGF
165
decreased TSP-1 secretion
levels, and this effect was blocked when cells were
pretreated with an antibody to Flk-1 (Fig. 3F). These
results suggest that astrocyte-derived VEGF differen-
tially regulates VEGF and TSP-1 levels via Flk-1 in
HRMECs.
VEGF induces phosphorylation of PKCf in
HRMECs
We examined whether VEGF is involved in the activa-
tion of phosphorylated PKCf in HRMECs, because
vascular permeability has been linked to PKCf activa-
tion [25], and the increase in VEGF mRNA levels after
VEGF
165
treatment could be mediated by atypical
PKCs (Fig. 3D). We treated HRMECs with recombi-
nant VEGF
165
and examined the effect on phos-

phorylated PKCf levels using fluorescence
immunocytochemistry. The level of PKCf was
unchanged by VEGF
165
, whereas that of phosphory-
lated PKCf was remarkably increased by VEGF
165
treatment (Fig. 3G). Western blot analysis confirmed
these results (Fig. 3H), and also showed that Flk-1
coimmunoprecipitates with PKCf (data not shown).
These results suggest that VEGF secreted from astro-
cytes binds to its cognate receptor on HRMECs and
induces the phosphorylation of PKCf.
AKAP12-overexpressing astrocytes inhibit
phosphorylation of PKCf in HRMECs
We next investigated whether PKCf activation is
affected by astrocytic AKAP12. When cells were incu-
bated with N-mock-CM, H-mock-CM or H-AKAP12-
CM, the levels and distribution of PKCf were
unchanged (Fig. 4A). Thereafter, we examined the
levels of phosphorylated PKCf in HRMECs using an
antibody that specifically recognizes phosphorylated
Thr410 within the activation loop of PKCf. When cells
were incubated with H-mock-CM, phosphorylated
PKCf levels were significantly increased as compared
to N-mock-CM-treated cells (Fig. 4A). H-AKAP12-
CM decreased phosphorylated PKCf levels, in a simi-
lar pattern to that seen with N-mock-CM (Fig. 4A).
These results were confirmed by western blot analysis
(Fig. 4B). When H-CM was pretreated with an anti-

body to VEGF, this effect was abolished (Fig. 4C).
Fig. 4. AKAP12 from astrocytes regulates the level of phosphorylated PKCf in HRMECs (A) Levels of PKCf and phosphorylated PKCf in
HRMECs treated with N-mock-CM, H-mock-CM, and H-AKAP12-CM for 15 min were analyzed by fluorescence immunocytochemistry
(green). Nuclei (blue) were stained with DAPI. Scale bar, 50 lm. (B) PKCf (7 lg) and phosphorylated PKCf (25 lg) levels under the indicated
conditions were analyzed by western blot. Data from four independent experiments were quantified, and are presented on the right. The
expression in N-mock-CM-treated cells was set at 100%. *P < 0.05 as compared to N-mock-CM-treated cells;
#
P < 0.05 as compared to
H-mock-CM-treated cells. (C) PKCf and phosphorylated PKCf levels were detected in the N (CM from normoxic astrocytes), H (CM from
hypoxic astrocytes), H ⁄ VEGF (H-CM neutralized by an antibody to VEGF) and H ⁄ IgG (H-CM neutralized by an antibody to IgG) treatment con-
ditions by western blot. HRMECs were treated for 24 h with these CMs. The quantification of protein levels from three independent experi-
ments is shown on the right. The levels of PKCf and phosphorylated PKCf in N-CM-treated cells were set at 100%. *P < 0.05 as compared
to control;
#
P < 0.001 as compared to control. In the case of H, the control is N; in the case of H ⁄ VEGF, the control is H; in the case of
H ⁄ IgG, the control is H ⁄ VEGF. (D–E) Human astrocytes were transfected with NC RNA or siRNA targeting Akap12 (siAkap12) and incubated
for 36 h, and CM then was collected and concentrated (4·). HRMECs were treated with CM for 15 min, and the levels of phosphorylated
PKCf and PKCf were analyzed by immunocytochemistry (D) and western blot (E). Scale bar, 50 lm. Nuclei (blue) were stained with DAPI.
Data from four independent experiments were quantified, and are presented on the right. The expression in NC-CM-treated cells was set at
100%. *P < 0.05 as compared to control. Actin was used as an internal control. (F) HRMECs were pretreated with or without COMP-Ang1
(100 ngÆmL
)1
) for 15 min, and this was followed by treatment with NC-CM or siAkap12-CM for 15 min. PKCf and phosphorylated PKCf
immunoblots were analyzed by western blot. The quantification of the immunoblots from three independent experiments is shown on the
right. The expression in NC-CM-treated cells was set at 100%.
#
P < 0.01 as compared to cells treated with NC-CM only; *P < 0.05 as com-
pared to siAkap12-CM-treated cells.
Regulation of the blood–neural barrier by AKAP12 Y. K. Choi and K W. Kim
2344 FEBS Journal 275 (2008) 2338–2353 ª 2008 The Authors Journal compilation ª 2008 FEBS

p-PKCζ
Actin
PKCζ
0
50
100
150
200
Relative expression
of proteins (%)
PKCζ
p-PKCζ
#
*
N-mock H-mock H-AKAP12 : CM
N-mock H-mock H-AKAP12CM :
p-PKCζ
Actin
PKCζ
: CMN H H/VEGF H/IgG
PKCζ
N-mock-CM H-mock-CM H-AKAP12-CM
p-PKCζ
PKCζ
p-PKCζ
siAkap12-CMNC-CM
p-PKCζ
Actin
siAkap12-CM
+

PKCζ
0
50
100
150
200
Relative expression
of proteins (%)
PKCζ
p-PKCζ
siAkap12-CM
+
*
PKCζ
Actin
siAkap12-CM
+
Ang1
+
+
+
p-PKCζ
PKCζ
p-PKCζ
*
#
Relative expression
of proteins (%)
0
50

100
150
200
+
+
+
+
siAkap12-CM
Ang1
PKCζ
p-PKCζ
*
#
Relative expression
of proteins (%)
0
50
100
150
200
*
CM : N H H/VEGF H/IgG
AD
B
C
E
F
Y. K. Choi and K W. Kim Regulation of the blood–neural barrier by AKAP12
FEBS Journal 275 (2008) 2338–2353 ª 2008 The Authors Journal compilation ª 2008 FEBS 2345
We next treated HRMECs with CM from small inter-

fering (si)Akap12-transfected cells (siAkap12-CM) and
examined phosphorylated PKCf levels by fluorescence
immunocytochemistry. The levels of PKCf were not
changed in siAkap12-CM-treated cells, whereas phos-
phorylated PKCf levels were strongly increased as
compared to negative control-CM (NC-CM)-treated
cells (Fig. 4D). The result was confirmed by western
blot analysis (Fig. 4E). A chimeric form of Ang1 is
soluble and more potent than native Ang1 in Tie2 (an
Angl receptor on endothelial cells) phosphorylation,
and its N-terminal portion has the short coiled-coil
domain of cartilage oligomeric matrix protein (COMP)
(COMP-Ang1) [34]. When HRMECs were pretreated
with COMP-Ang1 and then treated with siAkap12-
CM, phosphorylation of PKCf was significantly
decreased as compared to what was seen with siA-
kap12-CM treatment alone (Fig. 4F), suggesting that
siAkap12-CM-induced PKCf activation is partly
blocked by Ang1.
PKCf regulates the expression of VEGF and
TSP-1 in HRMECs
Astrocytic AKAP12 downregulates phosphorylation of
PKCf (Fig. 4), decreases the level of VEGF (Fig. 2A,B)
and increases the level of TSP-1 in HRMECs
(Fig. 2A,B). We next examined the relationship
between PKCf activation and the expression of VEGF
and TSP-1 in HRMECs. We transfected HRMECs
with an siRNA that targets PKCf (siPKCf), and found
that VEGF and TSP-1 secretion and mRNA levels
were markedly downregulated and upregulated, respec-

tively (Fig. 5A,B). Therefore, we propose that AKAP12
in astrocytes differentially regulates the secretion levels
of VEGF and TSP-1 in HRMECs via PKCf.
CM
PKCζ
p-PKCζ
Actin
TSP-1
Ponceau
siPKCζ
+
VEGF
++
Hypoxia
0
50
100
150
VEGF
TSP-1
noisserpxeevitaleR
)%(snietorpfo
+
++
siPKCζ
Hypoxia
*
**
RT-PCR
siPKCζ

+
++
Hypoxia
VEGF
β-actin
TSP-1
0
50
100
150
VEGF
TSP-1
n
ois
s
e
r
p
x
e
evitale
R
)
%
(ANR
m
fo
+
++
siPKCζ

Hypoxia
*
**
A
B
Fig. 5. PKCf regulates the expression of VEGF and TSP-1 in HRMECs. (A,B) HRMECs were transfected with an siRNA targeting PKCf (siP-
KCf), incubated for 24 h, and then exposed to hypoxia for 24 h. CM was collected and concentrated (4·). (A) The levels of PKCf and phos-
phorylated PKCf in cell lysates, and the levels of VEGF and TSP-1 in CM, were analyzed by western blot. The data from four independent
experiments were quantified, and are presented on the right. The level of VEGF in cells transfected with an NC RNA was set at 100%. The
level of TSP-1 in siPKCf-transfected cells was set at 100%. *P < 0.05 as compared to control; **P < 0.005 as compared to control. (B)
VEGF and TSP-1 mRNA levels were analyzed by RT-PCR. The data from three independent experiments were quantified, and are presented
on the right. The level of VEGF in cells transfected with a negative control RNA was set at 100%. The level of TSP-1 in siPKCf-transfected
cells was set as 100%. *P < 0.05 as compared to control; **P < 0.005 as compared to control. b-Actin and Ponceau Red staining served as
the internal controls.
Regulation of the blood–neural barrier by AKAP12 Y. K. Choi and K W. Kim
2346 FEBS Journal 275 (2008) 2338–2353 ª 2008 The Authors Journal compilation ª 2008 FEBS
Rho kinase is involved in the regulation of VEGF
and TSP-1 levels in HRMECs
Recent studies have shown that Rho family GTPases
are activated by atypical PKCf during cell motility
[35,36]. As our data indicated that PKCf regulates
both VEGF and TSP-1 in HRMECs (Fig. 5), we
examined the effect of a specific Rho kinase inhibitor,
Y27632, on VEGF and TSP-1 levels. VEGF and TSP-
1 levels are differentially regulated by hypoxia [8,37].
Cells were grown under normoxic or hypoxic condi-
tions, and treated or not treated with Y27632. Under
both normoxic and hypoxic conditions, Y27632 treat-
ment decreased VEGF and increased TSP-1 mRNA
and secretion levels as compared to untreated cells

(Fig. 6A,B). These results show that differential regula-
tion of VEGF and TSP-1 levels in response to oxygen
levels may result from Rho kinase activity. Because
treatment of HRMECs with siAkap12-CM activated
PKCf (Fig. 4D,E), we examined whether a similar
pattern of regulation of VEGF and TSP-1 occurred.
siAkap12-CM upregulated VEGF expression and
downregulated TSP-1 levels in HMRECs (Fig. 6C,D).
When cells were treated with a combination of Y27632
and siAkap12-CM, VEGF levels were significantly
decreased and TSP-1 levels were increased (Fig. 6C,D).
These results suggest that AKAP12 in astrocytes inver-
sely regulates VEGF and TSP-1, and that this regula-
tion is mediated by Rho kinase activity.
Discussion
Astrocytes are very complex cells, capable of respond-
ing to hypoxia in the developing retina [2], and secret-
ing the angiogenic factor VEGF, which subsequently
induces angiogenesis [3–5]. The initial event in angio-
genesis is the loosening of cell–cell contacts, and this is
followed by the migration of endothelial cells to form
a capillary tube network [38]. Oxygen delivered by the
developing brain and retinal vasculature stops angio-
genesis, and the blood–neural barrier is formed
[20,21,37,39]. Elevated oxygen levels (reoxygenation)
upregulate AKAP12 expression in astrocytes, resulting
in decreased VEGF secretion and increased Ang1
secretion [20,21]. In the current study, we demon-
strated that AKAP12-overexpressing hypoxic astro-
cytes induce endothelial tightening and

antiangiogenesis (Fig. 1). We showed that AKAP12 in
astrocytes decreases VEGF levels and increases TSP-1
levels in HRMECs (Fig. 2A,B). VEGF is a key factor
in the induction of vascular permeability during angio-
genesis and in barrier disruption [4–6,40–43]. TSP-1 is
an angiostatic factor that is upregulated during reoxy-
genation [8,37], and that plays a key role in vessel dif-
ferentiation and migration [7,30,31]. These data show
that AKAP12 plays a critical role in mediating impor-
tant interactions between retinal endothelial cells and
astrocytes during retinal angiogenesis and barriergene-
sis. Our previous studies revealed that AKAP12 in
astrocytes reduces vascular permeability [20,21].
However, the mechanism by which AKAP12 regulates
vascular permeability is unclear. In this study, we dem-
onstrated that AKAP12 from astrocytes inversely regu-
lates VEGF and TSP-1 levels via PKCf activation in
HRMECs.
We showed that AKAP12 in astrocytes decreases
VEGF levels and increases TSP-1 levels in HRMECs
(Fig. 2A,B), and that these effects are blocked in the
VEGF-neutralizing condition (Fig. 2C). We found that
VEGF
165
treatment of HRMECs induces VEGF
expression and reduces TSP-1 expression (Fig. 3A,E),
suggesting that the differential regulation of VEGF
and TSP-1 in HRMECs by astrocytic AKAP12 could
stem from the astrocyte-secreted factor VEGF. These
results show that development of new blood vessels

during retinal development may be triggered not only
by secreted VEGF from hypoxic astrocytes, but also
by autocrine-stimulated VEGF from retinal endothelial
cells. The increase in VEGF levels and the decrease in
TSP-1 levels after VEGF
165
treatment could be medi-
ated by Flk-1 (Fig. 3C,F). VEGF
165
treatment of
HRMECs increased cell viability by 10% (Fig. 3B).
However, we did not observe a significant increase in
cell viability in the hypoxic condition as compared to
the normoxic condition (Fig. 3B), although hypoxia
induces endothelial VEGF expression (Fig. 6A).
Considering that both VEGF
165
treatment and hypoxia
increase endothelial VEGF expression, further studies
are necessary to clarify the differential effects of VEGF
and hypoxia on endothelial cell survival. We observed
that treatment of HRMECs with recombinant
VEGF
165
induced intracellular VEGF mRNA levels
via a positive feedback mechanism of regulation
(Fig. 3A). These effects were reduced when HRMECs
were treated with a combination of VEGF
165
and

PKC inhibitors such as GF109203X and chelerythrine
chloride (Fig. 3D). According to our data, VEGF
mRNA levels were more reduced in cells treated with
6 lm GF109203X than in cells treated with 0.1 lm
GF109203X, suggesting that atypical PKCs may be
involved in this VEGF induction. We also found that
recombinant VEGF
165
treatment increased the levels
of phosphorylated PKCf, one of the atypical PKCs, in
HRMECs (Fig. 3G,H), which suggests that the secre-
tion of VEGF from astrocytes is involved in the acti-
vation and phosphorylation of PKCf in HRMECs
Y. K. Choi and K W. Kim Regulation of the blood–neural barrier by AKAP12
FEBS Journal 275 (2008) 2338–2353 ª 2008 The Authors Journal compilation ª 2008 FEBS 2347
(Fig. 3G,H). Interestingly, our data showed that PKCf
activation by VEGF in HRMECs is induced by the
phosphorylation of PKCf (Fig. 3F), whereas PKCf
translocation to the membrane was observed either fol-
lowing treatment with thrombin in endothelial cells
[25], or following treatment with VEGF in retinal
pigment epithelial cells [44]. To determine whether the
expression of VEGF and TSP-1 was regulated by
PKCf, we transfected HRMECs with siPKCf, and
observed that the expression of VEGF was decreased
and that the expression of TSP-1 was increased in
siPKCf-transfected cells (Fig. 5). PKCf had a signifi-
cant effect on the levels of secreted VEGF and TSP-1
(Fig. 5) as compared with intracellular VEGF and
VEGF

TSP-1
NC-CM siAkap12-CM siAkap12-CM+YNC-CM+Y
Y27632
+ +
++
siAkap12-CM
VEGF
TSP-1
Ponceau
CM
noisserpxeevitaleR
)%(snietorpfo
0
50
100
150
TSP-1
VEGF
Y27632
+ +
++
siAkap12-CM
##
*
**
##
**
**
CM
TSP-1

Ponceau
Y27632
+
+
++
Hypoxia
*
+
+
++
Y27632
Hypoxia
no
is
s
er
pxe
evit
aleR
)%(nietorp1-PSTfo
0
40
80
120
*
VEGF
#
TSP-1
Y27632
+

+
++
Hypoxia
VEGF
β-actin
RT-PCR
0
120
noisserpxeevitaleR
)%(nietorp1-PSTfo
40
80
+ +
++
Y27632
siAkap12-CM
*
#
#
A
B
C
D
Regulation of the blood–neural barrier by AKAP12 Y. K. Choi and K W. Kim
2348 FEBS Journal 275 (2008) 2338–2353 ª 2008 The Authors Journal compilation ª 2008 FEBS
TSP-1 levels (data not shown), because the signal
transduction requires the binding of secreted VEGF
and TSP-1 to endothelial cell receptors [45,46]. On the
basis of these data, we suggest that astrocyte-derived
VEGF activates PKCf, resulting in differential

regulation of VEGF and TSP-1 in HRMECs. A recent
study showed that PKCa induced angiogenesis
via induction of VEGF in human umbilical vein endo-
thelial cells (HUVECs) [47]. Therefore, we propose
that PKCf, as well as PKCa, can induce VEGF in
endothelial cells.
As astrocytic AKAP12 inhibited phosphorylation of
PKCf in HRMECs (Fig. 4) and increased Ang1 secre-
tion levels [20], we examined whether PKCf activation
could be reduced by Ang1. COMP-Ang1 induced par-
tial inhibition of phosphorylation of PKCf induced by
siAkap12-CM (Fig. 4F). In addition, when HRMECs
were incubated with H-AKAP12-CM, claudin-1 and
VE-cadherin levels were upregulated as compared to
those seen with H-mock-CM (Fig. 1E), and this effect
was blocked by pretreatment of H-AKAP12-CM with
an antibody to Ang1 (Fig. 1E). These results indicate
that Ang1 derived from astrocytes may induce junction
proteins by inhibition of phosphorylation of PKCf in
HRMECs.
Several studies have shown that atypical PKCf regu-
lates Rho activity [35,36]. In the current study, we
investigated the effect of a specific Rho kinase inhibi-
tor, Y27632, on VEGF and TSP-1 levels in HRMECs.
We demonstrated that hypoxia increases the secretion
of VEGF and decreases the secretion of TSP-1, and
that this effect is blocked by Y27632 (Fig. 6A,B). The
same effect was also observed in HRMECs when these
cells were treated with CM from siAkap12-transfected
astrocytes (Fig. 6C,D). siAkap12-CM activated PKCf

in HRMECs (Fig. 4D,E), leading to increased VEGF
levels and decreased TSP-1 levels, and this effect was
blocked by Y27632 (Fig. 6C,D). Our results suggest
that Rho kinase activity plays a key role in the differ-
ential regulation of VEGF and TSP-1 by AKAP12 or
oxygen tension.
In summary, we showed that VEGF and TSP-1 lev-
els in HRMECs are differentially regulated by
AKAP12 in astrocytes (Fig. 2A,B), leading to reduced
vascular permeability, decreased endothelial cell
migration, and upregulation of tight junction proteins.
We also demonstrated that the regulation of VEGF
and TSP-1 by astrocytic AKAP12 is mediated by
phosphorylation of PKCf and Rho kinase activity.
The most prominent features of the blood–neural bar-
rier are the presence of complex tight junctions, and
the interaction of adhesion molecules of central ner-
vous system endothelial cells, which together form an
endothelial barrier [48–51]. Our results may elucidate
a pathway to restoring barrier function in central ner-
vous system diseases that are associated with
increased VEGF expression and decreased TSP-1
expression.
Experimental procedures
Immunofluorescence staining
HRMECs were incubated overnight at 4 °C with the indi-
cated primary antibodies, and this was followed by incuba-
tion with Alexa Fluor antibodies as secondary antibodies.
Nuclei were stained with 4¢,6-diamidino-2-phenylindole
(DAPI) (Sigma, St Louis, MO, USA). Images were obtained

with an Axiovert M200 (Carl Zeiss, Oberkochen, Germany)
microscope, and analyzed using image-pro plus (Media
Cybernetics, Bethesda, MD, USA). We counted the area
stained with green fluorescence according to the
Fig. 6. The involvement of Rho kinase in the expression of VEGF and TSP-1 in HRMECs. (A) HRMECs were treated with 10 lM Y27632 for
24 h under both normoxic and hypoxic conditions. VEGF and TSP-1 mRNA levels were detected by RT-PCR. (B) VEGF and TSP-1 secretion
levels were detected by western blot. The data from five independent experiments were quantified, and are shown on the right. The level
of expression of TSP-1 in Y27632-treated cells under normoxia was set at 100%. *P < 0.05 as compared to control;
#
P < 0.005 as compared
to control. The control for Y27632 treatment under normoxia was no Y27632 under normoxia; the control for no Y27632 treatment under
hypoxia was no Y27632 under normoxia; the control for Y27632 treatment under hypoxia was no Y27632 under hypoxia. (C) HRMECs were
treated with NC-CM, NC-CM plus 10 l
M Y27632 (NC-CM + Y), siAkap12-CM or siAkap12-CM plus Y27632 (siAkap12-CM + Y) for 24 h. The
levels of VEGF and TSP-1 (green) were detected by immunocytochemistry. Scale bar, 50 lm. Nuclei (blue) were stained with DAPI. Quantifi-
cation of the immunohistochemical staining area was performed using
IMAGE-PRO PLUS. The expression levels from four independent experi-
ments were quantified, and are shown on the right. The level of VEGF in siAkap12-CM-treated cells was set at 100%. The level of TSP-1 in
NC-CM + Y27632-treated cells was set at 100%. *P < 0.05 as compared to control; **P < 0.005 as compared to control;
##
P < 0.001 as
compared to control. The control for NC-CM + Y was NC-CM; the control for siAkap12-CM was NC-CM; the control for siAkap12-CM + Y
was siAkap12-CM. (D) Secretion levels of VEGF and TSP-1 were detected by western blot under the same conditions as for (C). The data
from four independent experiments were quantified, and are shown on the right. The level of TSP-1 in NC-CM + Y27632-treated cells was
set at 100%. *P < 0.05 as compared to control;
#
P < 0.01 as compared to control. The control for NC-CM + Y was NC-CM; the control for
siAkap12-CM was NC-CM; the control for siAkap12-CM + Y was siAkap12-CM. b-actin and Ponceau Red staining served as the internal
controls.
Y. K. Choi and K W. Kim Regulation of the blood–neural barrier by AKAP12

FEBS Journal 275 (2008) 2338–2353 ª 2008 The Authors Journal compilation ª 2008 FEBS 2349
manufacturer’s instructions. Each stained area was pre-
sented relative to the area with the highest staining intensity.
Cell culture
Primary human brain astrocyte cells, dissociated from nor-
mal human brain cortex tissue, were purchased from the
Applied Cell Biology Research Institute (Kirkland, WA,
USA). Primary human brain astrocyte cells were cultured
in DMEM supplemented with 10% fetal bovine serum (In-
vitrogen, San Diego, CA, USA) and antibiotics. HRMECs
were purchased from the Applied Cell Biology Research
Institute and grown in M199 medium supplemented with
20% fetal bovine serum, 3 ngÆmL
)1
basic fibroblast growth
factor (Invitrogen) and 10 UÆmL
)1
heparin (Sigma). For
hypoxia experiments, astrocytes were incubated in a hyp-
oxic chamber (Forma Scientific, San Bruno, CA, USA),
which maintained the cells under low oxygen tension (5%
CO
2
with 1% O
2
, balanced with N
2
).
Migration assay
HRMECs were seeded on gelatin-coated 12-well culture

dishes. At 90% confluence, the endothelial monolayers were
marked with an injury line and wounded with the end of a
200-lL tip. Plates were rinsed with serum-free medium to
remove cellular debris. CM from human astrocytes tran-
siently transfected with mock or Akap12 vectors under
normoxic or hypoxic conditions was then added. HRMECs
were allowed to migrate for 24 h, rinsed with NaCl ⁄ P
i
,
fixed with absolute methanol for 5 min, and stained with
Giemsa (Sigma).
Western blot analysis
Cellular protein and CM protein were analyzed by western
blot assay. Western blot analysis was performed as
described previously [52]. We used antibodies specific for
Ang1, VEGF, VE-cadherin, ZO-2, Flk-1, PKCf and phos-
phorylated PKCf (Snata Cruz Biotechnology, Santa Cruz,
CA, USA), ZO-1 (Zymed, San Francisco, CA, USA), TSP-
1 (Neomarkers), and b-actin (Sigma). Recombinant
VEGF
165
was purchased from R&D Systems (Minneapolis,
MN, USA), and COMP-Ang1 was a generous gift from
G. Y. Koh (Korea Advanced Institute of Science and Tech-
nology). Y27632, a specific Rho kinase inhibitor, was
purchased from Calbiochem (La Jolla, CA, USA). Ponceau
S solution was purchased from Sigma.
Cell fractionation
Cell fractionation was performed as described previously
[25] with minor modifications. Cells were washed with

NaCl ⁄ P
i
and harvested by scraping into 50 lL of homoge-
nization buffer (20 mm Tris ⁄ HCl, pH 7.4, 0.5 mm EDTA,
0.5 m m EGTA, 10 mm b-mercaptoethanol, 5% glycerol,
2mm NaF, 1 mm Na
3
VO
4
, and proteinase inhibitor mix-
ture). Cells were mechanically homogenized and centrifuged
for 15 min at 1200 g at 4 °C. The supernatant was further
centrifuged for 80 min at 28 700 g at 4 °C. The resulting
supernatant containing the cytosolic components was
removed, and the pellet containing the membrane compo-
nents were resuspended in 50 lL of homogenized buffer
supplemented with 0.5% Triton X-100 and 100 mm NaCl.
Proteins in 20 lL of each fraction were separated by
SDS ⁄ PAGE.
Transient transfection and CM preparation
The full-length rat Akap12 cDNA (from I. H. Gelman,
Roswell Park Cancer Institute, Buffalo, NY, USA) was
subcloned into pcDNA3. Transient transfections were per-
formed using Lipofectamine and Plus reagent (Invitrogen).
For the preparation of CMs for treating HRMECs, med-
ium from transfected human astrocytes was changed to
M199 medium containing 1% fetal bovine serum for 24 h,
collected and filtered through a 0.22 lm pore membrane
(Millipore, Beverly, MA, USA), and then concentrated four
times using centrifugal filters (Millipore). For preparation

of CM for western blot analysis, M199 medium containing
1% fetal bovine serum from transfected cells was collected
and concentrated through Ultra-4 centrifugal filters
(Millipore).
Cell viability assay [2,3-bis(2-methoxy-4-nitro-
5-sulfophenyl)-2H-tetrazolium-5-carboxanilide
inner salt assay]
Cell viability was determined using CellTiter 96 Aqueous
One Solution (Promega). Cells were seeded into gelatin-
coated 96-well plates, and incubated with or without
reagents for 24 h. Each culture condition was analyzed in
triplicate. The absorbance values at 492 nm were corrected
by subtracting the average absorbance from the control
wells containing ‘no cells’.
Permeability assay
Permeability across the endothelial cell monolayer was mea-
sured by using type I collagen-coated transwell units
(6.5 mm diameter, 3.0 lm pore spolycarbonate filter; Corn-
ing, Corning, NY, USA). After HRMECs become conflu-
ent, CM was treated for 24 h. Permeability was measured
by adding 0.1 mg of RITC-labeled dextran (relative molec-
ular mass  10 000) ⁄ mL to the upper chamber. After incu-
bation for 15 min, 100 lL of sample from the lower
compartment was diluted with 100 l L of NaCl ⁄ P
i
and mea-
sured for fluorescence at 635 nm when excited at 540 nm
Regulation of the blood–neural barrier by AKAP12 Y. K. Choi and K W. Kim
2350 FEBS Journal 275 (2008) 2338–2353 ª 2008 The Authors Journal compilation ª 2008 FEBS
with a spectrophotometer (Tecan Spectra Fluor; Tecan

Durham, NC, USA).
RNA interference
Astrocytes and HRMECs were grown to 80% confluence,
and siRNAs (siAkap12 50 nm; siPKCf 150 nm) were trans-
fected into the cells using Lipofectamine and Plus reagent
(Invitrogen). All transfections were performed according to
the manufacturer’s instructions. siRNAs and the control
nonsilencing RNAs were designed by Dharmacon (Lafay-
ette, CO, USA). The human Akap12 target sequence used
was: 5¢-AGACGGATGTAGTGTTGAA-3¢. The siRNA
targeting PKCf was designed by Dharmacon (Catalog
number, M-003526-04).
RT-PCR
Total RNA was isolated from the indicated cells using Tri-
zol reagent (Invitrogen). RT-PCR analysis was performed
as described previously [53]. The following sets of primers
were used: VEGF, 5¢-GAGAATTCGGCCTCCGAAA
CCATGAACTTTCTGT-3¢ (forward) and 5¢-GAGCATG
CCCTCCTGCCCGGCTCACCGC-3¢ (reverse); TSP-1,
5¢-CGTCCTGTTCCTGATGCATG-3¢ (forward) and
5¢-GGCCCTGTCTTCCTGCACAA-3¢ (reverse); glyceralde-
hyde-3-phosphate dehydrogenase (GAPDH), 5¢-CAGGG
CTGCTTTTAACTCTG-3¢ (forward) and 5¢-TAGAGG
CAGGGATGATGTTC-3¢ (r everse); and b-actin, 5¢-GACTA
CCTCATGAAGATC-3¢ (forward) and 5¢-GATCC
ACATCTGCTGGAA-3¢ (reverse). The PCR products were
separated on 1.2% agarose gels and visualized by ethidium
bromide staining under a transilluminator (LAS 3000; Fuji-
film, Tokyo, Japan).
Data analysis and statistics

Quantification of band intensity was analyzed using imagej
( and normalized to the density of
the b-actin or Ponceau staining band. All data are pre-
sented as mean ± SD changed into relative percentage.
Statistical comparisons between groups were done using
Student’s t-test. P < 0.05 was considered to be statistically
significant.
Acknowledgements
This work was supported by the Creative Research
Initiatives (NeuroVascular Coordination Research
Center) of the Ministry of Science and Technology.
We thank Dr. G. Y. Koh (Korea Advanced Institute
of Science and Technology, Daejeon, Republic of
Korea) for providing COMP-Ang1. Y. K. Choi is
grateful to Kyu Han Kim (NeuroVascular Coordina-
tion Research Center, College of Pharmacy, Seoul
National University) for a useful discussion. We
declare that we have no competing financial interests.
References
1 Abbott NJ, Ronnback L & Hansson E (2006) Astro-
cyte–endothelial interactions at the blood–brain barrier.
Nat Rev Neurosci 7, 41–53.
2 Zhang Y, Porat RM, Alon T, Keshet E & Stone J
(1999) Tissue oxygen levels control astrocyte movement
and differentiation in developing retina. Brain Res Dev
Brain Res 118, 135–145.
3 Stone J, Itin A, Alon T, Pe’er J, Gnessin H, Chan-Ling
T & Keshet E (1995) Development of retinal vascula-
ture is mediated by hypoxia-induced vascular endothe-
lial growth factor (VEGF) expression by neuroglia.

J Neurosci 15, 4738–4747.
4 Provis JM, Leech J, Diaz CM, Penfold PL, Stone J &
Keshet E (1997) Development of the human retinal vas-
culature: cellular relations and VEGF expression. Exp
Eye Res 65, 555–568.
5 Dvorak HF, Nagy JA, Feng D, Brown LF & Dvorak
AM (1999) Vascular permeability factor ⁄ vascular endo-
thelial growth factor and the significance of microvascu-
lar hyperpermeability in angiogenesis. Curr Top
Microbiol Immunol 237, 97–132.
6 Alon T, Hemo I, Itin A, Pe’er J, Stone J & Keshet E
(1995) Vascular endothelial growth factor acts as a sur-
vival factor for newly formed retinal vessels and has
implications for retinopathy of prematurity. Nat Med 1,
1024–1028.
7 Vailhe B & Feige JJ (2003) Thrombospondins as anti-
angiogenic therapeutic agents. Curr Pharm Des 9, 583–
588.
8 Hu CJ, Chen SD, Yang DI, Lin TN, Chen CM, Huang
TH & Hsu CY (2006) Promoter region methylation and
reduced expression of thrombospondin-1 after oxygen–
glucose deprivation in murine cerebral endothelial cells.
J Cereb Blood Flow Metab 26, 1519–1526.
9 Kalluri R (2003) Basement membranes: structure,
assembly and role in tumour angiogenesis. Nat Rev
Cancer 3, 422–433.
10 Rastinejad F, Polverini PJ & Bouck NP (1989) Regula-
tion of the activity of a new inhibitor of angiogenesis
by a cancer suppressor gene. Cell 56, 345–355.
11 Gelman IH & Gao L (2006) SSeCKS ⁄ Gravin ⁄ AKAP12

metastasis suppressor inhibits podosome formation via
RhoA- and Cdc42-dependent pathways. Mol Cancer
Res 4, 151–158.
12 Gelman IH, Lee K, Tombler E, Gordon R & Lin X
(1998) Control of cytoskeletal architecture by the src-
suppressed C kinase substrate, SSeCKS. Cell Motil
Cytoskeleton 41, 1–17.
Y. K. Choi and K W. Kim Regulation of the blood–neural barrier by AKAP12
FEBS Journal 275 (2008) 2338–2353 ª 2008 The Authors Journal compilation ª 2008 FEBS 2351
13 Lin X, Tombler E, Nelson PJ, Ross M & Gelman IH
(1996) A novel src- and ras-suppressed protein kinase C
substrate associated with cytoskeletal architecture.
J Biol Chem 271, 28430–28438.
14 Cheng C, Liu H, Ge H, Qian J, Qin J, Sun L & Shen A
(2007) Essential role of Src suppressed C kinase sub-
strates in endothelial cell adhesion and spreading.
Biochem Biophys Res Commun 358, 342–348.
15 Kaibuchi K, Kuroda S & Amano M (1999) Regulation
of the cytoskeleton and cell adhesion by the Rho family
GTPases in mammalian cells. Annu Rev Biochem 68,
459–486.
16 Amano M, Chihara K, Kimura K, Fukata Y, Nakam-
ura N, Matsuura Y & Kaibuchi K (1997) Formation of
actin stress fibers and focal adhesions enhanced by
Rho-kinase. Science 275, 1308–1311.
17 Park JH, Okayama N, Gute D, Krsmanovic A, Battar-
bee H & Alexander JS (1999) Hypoxia ⁄ aglycemia
increases endothelial permeability: role of second mes-
sengers and cytoskeleton. Am J Physiol 277, C1066–
C1074.

18 Chapline C, Cottom J, Tobin H, Hulmes J, Crabb J &
Jaken S (1998) A major, transformation-sensitive PKC-
binding protein is also a PKC substrate involved in
cytoskeletal remodeling. J Biol Chem 273, 19482–19489.
19 Hayashi M, Sakata M, Takeda T, Tahara M, Yamam-
oto T, Minekawa R, Isobe A, Tasaka K & Murata Y
(2005) Hypoxia up-regulates hypoxia-inducible factor-
1alpha expression through RhoA activation in tropho-
blast cells. J Clin Endocrinol Metab 90, 1712–1719.
20 Choi YK, Kim JH, Kim WJ, Lee HY, Park JA, Lee
SW, Yoon DK, Kim HH, Chung H, Yu YS et al.
(2007) AKAP12 regulates human blood–retinal barrier
formation by downregulation of hypoxia-inducible fac-
tor-1alpha. J Neurosci 27, 4472–4481.
21 Lee SW, Kim WJ, Choi YK, Song HS, Son MJ, Gel-
man IH, Kim YJ & Kim KW (2003) SSeCKS regulates
angiogenesis and tight junction formation in blood–
brain barrier. Nat Med 9, 900–906.
22 Parekh DB, Ziegler W & Parker PJ (2000) Multiple
pathways control protein kinase C phosphorylation.
EMBO J 19, 496–503.
23 Wu HM, Yuan Y, Zawieja DC, Tinsley J & Granger
HJ (1999) Role of phospholipase C, protein kinase C,
and calcium in VEGF-induced venular hyperpermeabili-
ty. Am J Physiol 276, H535–H542.
24 Haller H, Ziegler W, Lindschau C & Luft FC (1996)
Endothelial cell tyrosine kinase receptor and G protein-
coupled receptor activation involves distinct protein
kinase C isoforms. Arterioscler Thromb Vasc Biol 16,
678–686.

25 Li X, Hahn CN, Parsons M, Drew J, Vadas MA &
Gamble JR (2004) Role of protein kinase Czeta in
thrombin-induced endothelial permeability changes:
inhibition by angiopoietin-1. Blood 104, 1716–1724.
26 Haller H, Hempel A, Homuth V, Mandelkow A, Bus-
jahn A, Maasch C, Drab M, Lindschau C, Jupner A,
Vetter K et al. (1998) Endothelial-cell permeability and
protein kinase C in pre-eclampsia. Lancet 351, 945–949.
27 Pal S, Datta K, Khosravi-Far R & Mukhopadhyay D
(2001) Role of protein kinase Czeta in Ras-mediated
transcriptional activation of vascular permeability fac-
tor ⁄ vascular endothelial growth factor expression.
J Biol Chem 276, 2395–2403.
28 Datta K, Li J, Bhattacharya R, Gasparian L, Wang E
& Mukhopadhyay D (2004) Protein kinase C zeta trans-
activates hypoxia-inducible factor alpha by promoting
its association with p300 in renal cancer. Cancer Res 64,
456–462.
29 Ferrara N, Gerber HP & LeCouter J (2003) The biol-
ogy of VEGF and its receptors. Nat Med 9, 669–676.
30 Li Z, Wang C, Jiao X, Lu Y, Fu M, Quong AA, Dye
C, Yang J, Dai M, Ju X et al. (2006) Cyclin D1 regu-
lates cellular migration through the inhibition of throm-
bospondin 1 and ROCK signaling. Mol Cell Biol 26,
4240–4256.
31 Weinstat-Saslow DL, Zabrenetzky VS, VanHoutte K,
Frazier WA, Roberts DD & Steeg PS (1994) Transfec-
tion of thrombospondin 1 complementary DNA into a
human breast carcinoma cell line reduces primary
tumor growth, metastatic potential, and angiogenesis.

Cancer Res 54, 6504–6511.
32 Zhang YW, Su Y, Volpert OV & Vande Woude GF
(2003) Hepatocyte growth factor ⁄ scatter factor mediates
angiogenesis through positive VEGF and negative
thrombospondin 1 regulation. Proc Natl Acad Sci USA
100, 12718–12723.
33 Greenaway J, Gentry PA, Feige JJ, LaMarre J & Petrik
JJ (2005) Thrombospondin and vascular endothelial
growth factor are cyclically expressed in an inverse pat-
tern during bovine ovarian follicle development. Biol
Reprod 72, 1071–1078.
34 Cho CH, Kammerer RA, Lee HJ, Steinmetz MO, Ryu
YS, Lee SH, Yasunaga K, Kim KT, Kim I, Choi HH
et al. (2004) COMP-Ang1: a designed angiopoietin-1
variant with nonleaky angiogenic activity. Proc Natl
Acad Sci USA 101, 5547–5552.
35 Kuribayashi K, Nakamura K, Tanaka M, Sato T, Kato
J, Sasaki K, Takimoto R, Kogawa K, Terui T, Takay-
ama T et al. (2007) Essential role of protein kinase C
zeta in transducing a motility signal induced by super-
oxide and a chemotactic peptide, fMLP. J Cell Biol
176, 1049–1060.
36 Van Kolen K & Slegers H (2006) Atypical PKCzeta is
involved in RhoA-dependent mitogenic signaling by the
P2Y(12) receptor in C6 cells. FEBS J 273, 1843–1854.
37 Song HS, Son MJ, Lee YM, Kim WJ, Lee SW, Kim
CW & Kim KW (2002) Oxygen tension regulates the
maturation of the blood–brain barrier. Biochem Biophys
Res Commun 290, 325–331.
Regulation of the blood–neural barrier by AKAP12 Y. K. Choi and K W. Kim

2352 FEBS Journal 275 (2008) 2338–2353 ª 2008 The Authors Journal compilation ª 2008 FEBS
38 Jackson C (2002) Matrix metalloproteinases and angio-
genesis. Curr Opin Nephrol Hypertens 11, 295–299.
39 West H, Richardson WD & Fruttiger M (2005) Stabil-
ization of the retinal vascular network by reciprocal
feedback between blood vessels and astrocytes. Develop-
ment 132, 1855–1862.
40 Provis JM (2001) Development of the primate retinal
vasculature. Prog Retin Eye Res 20, 799–821.
41 Qaum T, Xu Q, Joussen AM, Clemens MW, Qin W,
Miyamoto K, Hassessian H, Wiegand SJ, Rudge J,
Yancopoulos GD et al. (2001) VEGF-initiated blood–
retinal barrier breakdown in early diabetes. Invest Oph-
thalmol Vis Sci 42, 2408–2413.
42 Witmer AN, Vrensen GF, Van Noorden CJ & Schlinge-
mann RO (2003) Vascular endothelial growth factors
and angiogenesis in eye disease. Prog Retin Eye Res 22,
1–29.
43 Carmeliet P, Ferreira V, Breier G, Pollefeyt S, Kieckens
L, Gertsenstein M, Fahrig M, Vandenhoeck A, Harpal
K, Eberhardt C et al. (1996) Abnormal blood vessel
development and lethality in embryos lacking a single
VEGF allele. Nature 380, 435–439.
44 Young TA, Wang H, Munk S, Hammoudi DS, Young
DS, Mandelcorn MS & Whiteside CI (2005) Vascular
endothelial growth factor expression and secretion by
retinal pigment epithelial cells in high glucose and
hypoxia is protein kinase C-dependent. Exp Eye Res 80,
651–662.
45 Millauer B, Wizigmann-Voos S, Schnurch H, Martinez

R, Moller NP, Risau W & Ullrich A (1993) High affin-
ity VEGF binding and developmental expression sug-
gest Flk-1 as a major regulator of vasculogenesis and
angiogenesis. Cell 72, 835–846.
46 Murphy-Ullrich JE & Mosher DF (1987) Interactions
of thrombospondin with endothelial cells: receptor-med-
iated binding and degradation. J Cell Biol 105, 1603–
1611.
47 Xu H, Czerwinski P, Hortmann M, Sohn HY, Forster-
mann U & Li H (2008) Protein kinase C {alpha} pro-
motes angiogenic activity of human endothelial cells via
induction of vascular endothelial growth factor. Cardio-
vasc Res, doi:10.1093/cvr/cvm/085
48 Morcos Y, Hosie MJ, Bauer HC & Chan-Ling T (2001)
Immunolocalization of occludin and claudin-1 to tight
junctions in intact CNS vessels of mammalian retina.
J Neurocytol 30, 107–123.
49 Staddon JM & Rubin LL (1996) Cell adhesion, cell
junctions and the blood–brain barrier. Curr Opin Neu-
robiol 6, 622–627.
50 Gumbiner B & Simons K (1986) A functional assay for
proteins involved in establishing an epithelial occluding
barrier: identification of a uvomorulin-like polypeptide.
J Cell Biol 102, 457–468.
51 Gavard J & Gutkind JS (2006) VEGF controls endothe-
lial-cell permeability by promoting the beta-arrestin-
dependent endocytosis of VE-cadherin. Nat Cell Biol 8,
1223–1234.
52 Moon EJ, Jeong CH, Jeong JW, Kim KR, Yu DY,
Murakami S, Kim CW & Kim KW (2004) Hepatitis B

virus X protein induces angiogenesis by stabilizing
hypoxia-inducible factor-1alpha. FASEB J 18, 382–384.
53 Lee MS, Moon EJ, Lee SW, Kim MS, Kim KW &
Kim YJ (2001) Angiogenic activity of pyruvic acid in in
vivo and in vitro angiogenesis models. Cancer Res 61,
3290–3293.
Y. K. Choi and K W. Kim Regulation of the blood–neural barrier by AKAP12
FEBS Journal 275 (2008) 2338–2353 ª 2008 The Authors Journal compilation ª 2008 FEBS 2353