Content uploaded by Janusch Blautzik
Author content
All content in this area was uploaded by Janusch Blautzik
Content may be subject to copyright.
Activated protein C protects against diabetic nephropathy
by inhibiting endothelial and podocyte apoptosis
Berend Isermann
1,10
, Ilya A Vinnikov
1,10
, Thati Madhusudhan
1,10
, Stefanie Herzog
1
, Muhammed Kashif
1
,
Janusch Blautzik
1
, Marcus A F Corat
2,9
, Martin Zeier
3
, Erwin Blessing
4
, Jun Oh
5
, Bruce Gerlitz
6
,
David T Berg
6
, Brian W Grinnell
6
, Triantafyllos Chavakis
7
, Charles T Esmon
8
, Hartmut Weiler
2
,
Angelika Bierhaus
1
& Peter P Nawroth
1
Data providing direct evidence for a causative link between endothelial dysfunction, microvascular disease and diabetic end-organ
damage are scarce. Here we show that activated protein C (APC) formation, which is regulated by endothelial thrombomodulin, is
reduced in diabetic mice and causally linked to nephropathy. Thrombomodulin-dependent APC formation mediates cytoprotection
in diabetic nephropathy by inhibiting glomerular apoptosis. APC prevents glucose-induced apoptosis in endothelial cells and
podocytes, the cellular components of the glomerular filtration barrier. APC modulates the mitochondrial apoptosis pathway via
the protease-activated receptor PAR-1 and the endothelial protein C receptor EPCR in glucose-stressed cells. These experiments
establish a new pathway, in which hyperglycemia impairs endothelial thrombomodulin-dependent APC formation. Loss of
thrombomodulin-dependent APC formation interrupts cross-talk between the vascular compartment and podocytes, causing
glomerular apoptosis and diabetic nephropathy. Conversely, maintaining high APC levels during long-term diabetes protects against
diabetic nephropathy.
Diabetic nephropathy is the leading cause of end-stage renal failure in
the western world
1
. Glomerular abnormalities are observed in early
diabetic nephropathy and include increases in glomerular filtration rate
and albuminuria. These pathological indices are in part the conse-
quence of glomerular capillary damage
2
. The cellular components of
the glomerular capillary are endothelial cells and podocytes, which,
together with the basement membrane, constitute the glomerular
filtration barrier. Podocyte integrity, an important determinant for
the permselective properties of the glomerular filtration barrier
3
,is
impaired in individuals with type 1 and type 2 diabetes
4,5
. The highly
specialized glomerular endothelium likewise contributes to the perm-
selective glomerular barrier
3
. It thus seems possible, but has not yet
been shown, that endothelial dysfunction is causally related to impair-
ment of the glomerular filtration barrier and diabetic nephropathy.
On unperturbed endothelial cells, activation of protein C that is
dependent on thrombomodulin (encoded by Thbd) inhibits coagula-
tion, inflammation and apoptosis
6
. These endothelial-protective effects
are mediated at least in part by PAR-1, the endothelial protein C
receptor (EPCR) and the sphingosine-1 receptor 1 (S1P-R1)
7–10
.
Function of the endothelial thrombomodulin–protein C system is
impaired in diabetic individuals, as shown by increased levels of
soluble thrombomodulin, thought to reflect loss of thrombomodulin
from the endothelium, and decreased levels of activated protein C
(APC)
11,12
. Thus, perturbation of the thrombomodulin–protein C
system resulting in reduced levels of APC is a potential mechanism of
glomerular capillary dysfunction in diabetes.
To study the role of APC for endothelial and glomerular capillary
dysfunction, we used mice with genetically altered in vivo APC
formation. Using mice with a reduction of thrombomodulin-
dependent protein C activation
13
and a new mouse model expressing
a hyperactivatable protein C mutation, we show that APC modulates
nephropathy in long-term experimental diabetes by modulating
apoptosis of endothelial cells and podocytes. The anti-apoptotic effect
of APC involves receptor-dependent regulation of the mitochondrial-
dependent apoptosis pathway.
RESULTS
APC modulates diabetic nephropathy
Thrombomodulin is strongly expressed in mouse glomeruli (Fig. 1a).
Persistent hyperglycemia reduces thrombomodulin expression and
Received 16 May; accepted 19 September; published online 4 November 2007; doi:10.1038/nm1667
1
Department of Medicine I and Clinical Chemistry, University of Heidelberg, INF 410, 69120 Heidelberg, Germany.
2
The Blood Research Institute, Blood Center of
Wisconsin, 8727 Watertown Plank Road Milwaukee, Wisconsin 53226, USA.
3
Department of Medicine I, Nephrology, University of Heidelberg, INF 162,
4
Department
of Medicine III, Cardiology, University of Heidelberg, INF 410, and
5
Department of Pediatric Nephrology, University of Heidelberg, INF 153, 69120 Heidelberg,
Germany.
6
BioTechnology Discovery Research, Lilly Research Laboratories, Indianapolis, Indiana 46285, USA.
7
Experimental Immunology Branch, National Cancer
Institute, Building 10, Room 4B17, National Institutes of Health, Bethesda, Maryland 20892, USA.
8
Cardiovascular Biology Research Program, Oklahoma Medical
Research Foundation, and Howard Hughes Medical Institute, 825 N.E. 13th, Room A-205, Mail Box 45, Oklahoma City, Oklahoma 73104, USA.
9
Present address:
Centro Multidisciplinar para Investigac¸a
˜
o Biolo
´
gica, Universidade Estadual de Campinas, Campinas 13083-877, SP, Brasil–Caixa Postal: 6095.
10
These authors
contributed equally to this work. Correspondence should be addressed to B.I. (berend.isermann@med.uni-heidelberg.de).
NATUR E MEDI CINE VOLUME 13
[
NUMBER 11
[
NOVEMBER 2007 1349
ARTICLES
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
APC formation in vivo in wild-type mice (Fig. 1b–e). This finding
prompted us to study the functional role of hyperglycemia-induced
loss of APC formation in experimental diabetic nephropathy.
We used genetically engineered mice to determine whether loss of
thrombomodulin-dependent PC activation and diabetic nephropathy
are causally linked. In Thbd
Pro/Pro
mice, thrombomodulin-dependent
protein C activation is impaired as a result of a targeted mutation
(Thbd
Pro
,Q404P)
13
.TheThbd
Pro
mutation impairs thrombomodulin-
dependent activation of protein C and of the thrombin-activatable
fibrinolysis inhibitor (TAFI)
14
, precluding a clear differentiation
between APC and activated TAFI-dependent effects.
To determine whether high levels of APC are sufficient to compen-
sate for the loss of thrombomodulin-dependent APC formation, we
generated a new transgenic mouse line with a gain of function in
respect to protein C activation (APC
high
mice, Supplementary Meth-
ods and Supplementary Fig. 1 online). These mice express a hyper-
activatable form of human protein C (D167F/D172K), which is
efficiently activated by mouse thrombin both in the absence and
presence of mouse thrombomodulin (181-fold and 5-fold activation
rate, respectively, as compared with mouse wild-type protein C;
Supplementary Table 1 online). Though increasing the activation
rate, the mutations reduce the anticoagulant effect to 35.3% in vitro
(Supplementary Table 2 online). Expression of the human protein C
mutant (D167F/D172K) in mice results in high levels of APC in vivo
(39.5 ng/ml) and increases blood loss following a standardized tail
injury (Supplementary Fig. 1f,g online).
Wild-type, Thbd
Pro/Pro
and APC
high
mice were made hyperglycemic
using streptozotocin (STZ). To evaluate the consequences of persistent
hyperglycemia, mice were kept diabetic for 26 weeks. Blood glucose
concentrations (Fig. 1f) and the amount of insulin required to
maintain glycemic control (data not shown) did not differ between
groups, allowing direct comparison of the different mouse models.
Persistent hyperglycemia increased albuminuria in wild-type mice
and, to an even larger extent, in Thbd
Pro/Pro
mice (Fig. 2a). Renal
hypertrophy, extracellular matrix deposition within glomeruli and
glomerular hypertrophy were associated with the increase of albumi-
nuria in Thbd
Pro/Pro
mice and were significantly exacerbated as
compared with diabetic wild-type mice (Fig. 2b–e). Thus, diabetic
nephropathy is aggravated in mice with a genetically imposed loss of
protein C activation.
In contrast, APC
high
mice were protected from diabetic nephro-
pathy. Compared with diabetic wild-type mice, albuminuria was
significantly reduced in diabetic APC
high
mice (Fig. 2a). The improved
outcome in albuminuria was paralleled by corresponding morpholo-
gical changes (Fig. 2b–e).
The renal protection in APC
high
mice cannot be attributed to
different thrombomodulin expression levels, because hyperglycemia
reduced renal thrombomodulin expression to a similar extent in all
mouse models used (Supplementary Fig. 2a online). A reduction of
in vivo APC formation was observed only in diabetic wild-type (see
above) and—though not significant—in diabetic Thbd
Pro/Pro
mice
(Supplementary Fig. 2b online), whereas plasma levels of APC
remained high in diabetic APC
high
mice (Supplementary Fig. 2c
online). Thus, diabetic APC
high
mice differ from diabetic wild-type
and Thbd
Pro/Pro
mice primarily in regard to in vivo APC production
but not renal thrombomodulin expression.
These data show that impaired APC formation is associated with
diabetic nephropathy and that increased levels of APC prevent diabetic
nephropathy. Thus, APC formation is directly involved in the patho-
genesis of microvascular complications.
APC is nephroprotective independent of blood clotting
Consistent with an impaired anticoagulant function in diabetic wild-
type mice with reduced APC levels, we observed increased blood
coagulation activation (thrombin-antithrombin (TAT), D-dimer) and
renal fibrin deposition (Supplementary Table 3 online). These indices
were further increased in diabetic Thbd
Pro/Pro
mice, whereas they
remained nearly normal in diabetic APC
high
mice (Supplementary
Table 3 online), raising the question of whether the protective effect of
APC is the consequence of anticoagulation.
120
TM expression
(protein, relative to WT, %)
*
100
80
60
40
20
0
CDM
120
TM expression
(mRNA, relative to WT, %)
*
100
80
60
40
20
0
CDM
5
4
3
2
1
APC production (ng/ml)
*
0
500
400
300
200
100
Blood glucose (mg/dl)
0
51015
Duration of diabetes (weeks)
Nondiabetic WT mice (n = 17)
Diabetic WT mice (n = 16)
20 25
CDM
Diabetic Thbd
Pro/Pro
mice (n = 12)
Diabetic APC
high
mice (n = 15)
abcde
f
Figure 1 Hyperglycemia suppresses thrombomodulin (TM) expression and protein C activation
in vivo.(a,b) Representative glomeruli obtained from a nondiabetic (a) and a diabetic (b)wild-
type (WT) mouse. TM is strongly expressed in glomerular capillaries of nondiabetic (a, arrow)
but not of diabetic (b, arrow) WT mice. Brown, TM-positive cells detected by horseradish
peroxidase-3,3¢-diaminobenzidine reaction; blue, hematoxylin counterstain. Scale bars, 15 mm.
(c,d) Expression of TM was determined in renal cortex tissue extracts using ELISA (c, n ¼ 10
each group) and RT-PCR (n ¼ 7 each group, three repeat experiments from each sample).
(e) In vivo APC formation in nondiabetic (n ¼ 10) and diabetic (n ¼ 5) WT mice. (f)Blood
glucose concentrations in nondiabetic WT, diabetic WT, diabetic TM
Pro/Pro
and diabetic
APC
high
mice. No significant differences were observed between groups of diabetic mice;
mean value ± s.e.m.; C, nondiabetic control mice; DM, diabetic mice. *P o 0.01 (t-test in c–e;
ANOVA in f).
ARTICLES
1350 VOLUME 13
[
NUMBER 11
[
NOVEMBER 2007 NATURE MED ICIN E
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
To evaluate the effect of anticoagulation on nephropathy, diabetic
wild-type mice were injected daily with low-molecular-weight heparin
(enoxaparin). Anticoagulation with enoxaparin normalized TAT,
D-dimer and renal fibrin deposition in diabetic wild-type mice
(Supplementary Table 3 online) but did not protect against diabetic
nephropathy (Supplementary Table 4 online). Therefore, inhibition
of blood coagulation is not sufficient to prevent diabetic nephropathy,
suggesting that APC prevents diabetic nephropathy independently
of anticoagulation.
To provide further evidence that the APC-dependent nephropro-
tection is independent of anticoagulation, diabetic APC
high
mice were
anticoagulated with enoxaparin as described above to prevent throm-
bin-mediated activation of the hyperactivatable protein C (D167F/
D172K). After anticoagulation of diabetic APC
high
mice, concentra-
tions of TAT and D-dimer remained low (data not shown), and
plasma concentrations of APC were reduced to 2.2 ng/ml as compared
with 54.7 ng/ml in non-anticoagulated diabetic APC
high
mice
(P o 0.01). Anticoagulation of diabetic APC
high
mice caused an
increase of albuminuria (135.5 mg/ml compared with 77.1 mg/mg in
non-anticoagulated APC
high
mice; P ¼ 0.027, Supplementary Table 4
online). Morphological indices of diabetic nephropathy in anticoagu-
lated diabetic APC
high
and diabetic wild-type mice were similar
(Supplementary Table 4 online). Thus, anticoagulation abrogates
nephroprotection in diabetic APC
high
mice, establishing that nephro-
protection in diabetic APC
high
mice cannot be attributed to an
anticoagulant effect.
APC might mediate nephroprotection through its anti-inflamma-
tory properties. Consistent with previous reports
15
, plasma levels of
interleukin-1b (IL-1b)and of IL-6 were increased in diabetic wild-type
mice (Supplementar y Table 5 online). Plasma cytokine levels were
normalized in both diabetic enoxaparin-treated and diabetic APC
high
mice (Supplementary Table 5 online). Because enoxaparin and APC
were equally efficient in controlling these markers of inflammation,
the nephroprotective effect of APC cannot be attributed to its anti-
inflammatory properties.
These results establish that APC mediates a nephroprotective effect
in diabetes independent of its anticoagulant and anti-inflammatory
properties and raise the question of alternative modes of action.
APC modulates glomerular apoptosis in diabetic nephropathy
APC inhibits apoptosis independent of its anticoagulant proper-
ties
8,16,17
. A loss of APC might therefore result in glomerular apop-
tosis, thus contributing to diabetic nephropathy.
Persistent hyperglycemia increased the frequency of apoptotic cells
in wild-type and—to an even larger degree—in diabetic Thbd
Pro/Pro
mice (Fig. 3). Conversely, the frequency of apoptotic cells within
glomeruli of diabetic APC
high
mice did not differ from that in
nondiabetic control mice and was significantly lower than in diabetic
wild-type mice (Fig. 3). Treatment with enoxaparin failed to reduce
apoptosis in glomeruli of diabetic wild-type mice (Fig. 3b)and
abolished the anti-apoptotic effect in diabetic APC
high
mice
(Fig. 3c). Endothelial as well as nonendothelial cells, including
podocytes, were positive by TUNEL staining (Fig. 3e).
Renal cortex tissue samples were analyzed to identify apoptosis
regulators modulated by APC. The pro-apoptotic proteins p53 and Bax
were expressed at significantly higher levels in diabetic wild-type mice
and to an even higher degree in diabetic Thbd
Pro/Pro
mice (Fig. 3f,g).
In diabetic APC
high
mice, expression of p53 and Bax remained at
normal levels. Enoxaparin treatment had no effect in diabetic wild-type
mice and increased expression of p53 and Bax in diabetic APC
high
mice
to the levels observed in diabetic wild-type mice. We observed
concomitant changes in the cytoplasmic concentrations of Smac and
700
600
500
400
300
200
100
Albuminuria (alb/crea, µg/mg)
**
0
CDM
WT
**
CDM
Thbd
Pro/Pro
Thbd
Pro/Pro
*
**
CDM
No diabetes (C)
WT
Diabetes (DM)
APC
high
APC
high
3
2
1
Histological score
*
0
CDM
WT
**
*
*
CDM
Thbd
Pro/Pro
CDM
APC
high
5
4
3
Glomeruli size (× 10
3
× µm
2
)
*
2
CDM
WT
*
*
*
CDM
Thbd
Pro/Pro
CDM
APC
high
16
14
12
10
8
6
4
2
Normalized kidney weight (mg/g)
0
CDM
WT
CDM
Thbd
Pro/Pro
CDM
APC
high
**
*
*
*
ac
b
d
e
Figure 2 Diabetic nephropathy is differentially modulated in mice with impaired or enhanced APC formation. (a,b) Albuminuria (a) and kidney weight,
normalized for body weight (b), in wild-type (WT), Thbd
Pro/Pro
and APC
high
mice without (open bars) or with (black bars) diabetes (n Z 10 for each group).
(c) Extracellular matrix deposition as determined by PAS stain. PAS staining was increased in diabetic wild-type mice (WT-DM, arrows) and in particular
in diabetic Thbd
Pro/Pro
mice (Thbd
Pro/Pro
-DM, arrows). Scale bars, 15 mm. (d) Bar graph summarizing the histological scores of PAS-stained sections.
(e) Glomerular size. C, nondiabetic control mice; DM, diabetic mice. *P o 0.05, **P o 0.01 (ANOVA). Panels d and e show results of 50 glomeruli from
seven different mice per group; mean value ± s.e.m.
ARTICLES
NATUR E MEDI CINE VOLUME 13
[
NUMBER 11
[
NOVEMBER 2007 1351
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
cytochrome c (data not shown). Expression levels of the anti-apoptotic
regulator Bcl-2 remained unaltered in all groups (data not shown).
These observations indicate that APC modulates the mitochondrial
apoptosis pathway in diabetic nephropathy.
PARP-1 is one of the main cleavage targets of caspase-3 during
apoptosis
18
. We observed an increase of the cleaved 89-kDa PARP-1
fragment in diabetic mice, and this increase was further aggravated in
diabetic Thbd
Pro/Pro
mice, whereas levels of PARP-1 cleavage remained
normal in diabetic APC
high
mice (Supplementary Fig. 3a online).
Again, enoxaparin treatment did not normalize PARP-1 cleavage in
diabetic wild-type mice and increased levels of cleaved PARP-1 in
diabetic APC
high
mice (Supplementary Fig. 3a online). The observed
differences of PARP-1 cleavage in mice with altered protein C activity
indicate modulation of caspase activation by APC in vivo.
We next determined the frequency of Ki-67–positive glomerular
cells. In diabetic wild-type mice, proliferation was significantly greater
than in nondiabetic wild-type mice (Fig. 3h), indicating that the
balance between proliferation and apoptosis was maintained in wild-
type mice. Similarly, the balance between proliferation and apoptosis
is maintained in diabetic APC
high
mice, given low indices for both
apoptosis and proliferation (Fig. 3h). In contrast, no significant
increase in proliferation was apparent in diabetic Thbd
Pro/Pro
mice
(Fig. 3h), reflecting a perturbed balance between proliferation and
apoptosis in favor of apoptosis in Thbd
Pro/Pro
mice. The low rate of
proliferation in diabetic Thbd
Pro/Pro
mice might reflect a role of
thrombomodulin in regulating proliferation.
Taken together, these data indicate that diabetic nephropathy is
associated with a pro-apoptotic phenotype, which is further aggra-
vated in diabetic Thbd
Pro/Pro
mice. The inhibition of glomerular
apoptosis and experimental nephropathy in diabetic APC
high
mice
establish that APC can overcome glomerular capillary dysfunction in
diabetic nephropathy.
Thrombomodulin dependent APC formation inhibits apoptosis
To demonstrate that loss of APC formation is causally linked
to aggravated nephropathy and glomerular apoptosis in diabetic
Thbd
Pro/Pro
mice, we generated and analyzed diabetic Thbd
Pro/Pro
APC
high
double-mutant mice. Albuminuria in diabetic Thbd
Pro/Pro
APC
high
mice remained normal and was similar to that observed in
nondiabetic Thbd
Pro/Pro
mice (Fig. 4a). In Thbd
Pro/Pro
APC
high
mice,
6
5
4
*
3
2
1
Frequency of apoptosis (%)
0
CDM
WT
DM +
Enox
CDM
APC
high
DM +
Enox
CDM
Thbd
Pro/Pro
WT
WT + Enox
APC
high
+ Enox
APC
high
Thbd
Pro/Pro
*
**
**
*
350
300
250
200
150
100
50
*
p53 levels, relative to
WT (%)
0
C
DM
WT
DM +
Enox
CDM
APC
high
DM +
Enox
CDM
Thbd
Pro/Pro
C
DM
WT
DM +
Enox
CDM
APC
high
DM +
Enox
CDM
Thbd
Pro/Pro
*
*
*
*
350
300
250
200
150
100
50
*
Bax levels, relative to WT (%)
0
C
DM
WT
DM +
Enox
CDM
APC
high
DM +
Enox
CDM
Thbd
Pro/Pro
C
DM
WT
DM +
Enox
CDM
APC
high
DM +
Enox
CDM
Thbd
Pro/Pro
2.0
1.5
1.0
0.5
*
Relative frequency of
Ki-67–positive cells (%)
0
C
DM
WT
C
DM
Thbd
Pro/Pro
C
DM
APC
high
*
*
*
*
No diabetes (C) Diabetes (DM)
Diabetes (DM) Diabetes (DM)
abcd
f
e
g
h
Figure 3 APC modulates glomerular apoptosis in diabetic nephropathy. (a–c) Frequency of
apoptosis in glomeruli, as determined by TUNEL assay. Apoptosis was frequently observed in
glomeruli of diabetic wild-type mice (WT-DM, arrows), diabetic Thbd
Pro/Pro
mice (Thbd
Pro/Pro
-
DM, arrows), diabetic wild-type mice treated with enoxaparin (b, arrows) and diabetic APC
high
mice treated with enoxaparin (c, arrows). (d) Bar graph summarizing frequency of apoptotic
glomerular cells. (e) Endothelial (arrows) and nonendothelial cells, including podocytes
(arrowhead), were TUNEL positive. (f,g) Expression of apoptosis regulators in diabetic
nephropathy. Representative western blots (top) and bar graphs (bottom) showing p53 (f) and
Bax (g) expression in renal cortex tissue samples. Bar graphs summarize results obtained
from six different mice in each group. (h) Proliferation of glomerular cells. Proliferating
glomerular cells were identified immunohistochemically (Ki-67). A significant increase of
proliferating cells was observed only in WT-DM mice. Panels a–c,e, TUNEL stain: brown, TUNEL-positive cells detected by HRP-DAB reaction; blue,
hematoxylin counterstain. Scale bar, 15 mm. C, nondiabetic control mice; DM, diabetic mice; Enox, enoxaparin treatment. Mean value ± s.e.m., *P o 0.05
(ANOVA); Panels d and h, results obtained by analyzing 50 glomeruli from seven different mice per group.
ARTICLES
1352 VOLUME 13
[
NUMBER 11
[
NOVEMBER 2007 NATURE MED ICIN E
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
normalization of albuminuria was associated with significantly
improved morphological indices (histological score) and reduced
glomerular size (1.33 versus 2.36 and 3,472 mm
2
versus 4,432 mm
2
,
respectively, in comparison with diabetic Thbd
Pro/Pro
mice, P o 0.01
in both cases), a reduced number of apoptotic glomerular cells (Fig. 4)
and normalized expression levels of p53 and Bax in renal cortex
extracts (data not shown).
Next, diabetic Thbd
Pro/Pro
mice were treated with the apoptosis
inhibitor minocycline
19,20
to provide independent evidence that
inhibition of apoptosis is sufficient for nephroprotection in these
mice. Minocycline normalized albuminuria (Fig. 4a), extracellular
matrix deposition (histological score) and glomerular size (1.63 and
3,065 mm
2
, respectively, P o 0.005 in comparison with diabetic
Thbd
Pro/Pro
mice in both cases). Minocycline efficiently prevented
apoptosis in glomerular cells (Fig. 4e) and normalized the expression
levels of p53 and Bax in renal cortex extracts of diabetic Thbd
Pro/Pro
mice (data not shown).
Thus, restoring APC formation or inhibiting apoptosis in diabetic
Thbd
Pro/Pro
suffice to prevent experimental diabetic nephropathy and
glomerular apoptosis. These data establish a role of anti-apoptotic
600
500
400
300
200
100
**
Albuminuria (alb/crea, µg/mg)
0
Thbd
Pro/Pro
C
Thbd
Pro/Pro
Thbd
Pro/Pro
APC
high
Thbd
Pro/Pro
C
Thbd
Pro/Pro
DM
Thbd
Pro/Pro
× APC
high
DM
Thbd
Pro/Pro
DM + Mc
DM
Thbd
Pro/Pro
+ MC
**
**
6
5
4
3
2
1
**
Frequency of apoptosis (%)
0
Thbd
Pro/Pro
C
Thbd
Pro/Pro
Thbd
Pro/Pro
APC
high
DM
Thbd
Pro/Pro
+ MC
**
**
ab
c
fd
e
Figure 4 APC-dependent inhibition of apoptosis prevents nephropathy in diabetic Thbd
Pro/Pro
mice. (a) Average albuminuria in nondiabetic Thbd
Pro/Pro
,
diabetic Thbd
Pro/Pro
, diabetic Thbd
Pro/Pro
APC
high
and minocycline-treated diabetic Thbd
Pro/Pro
mice (n ¼ 7 for minocycline-treated mice and Thbd
Pro/Pro
APC
high
mice; n Z 10 for diabetic and nondiabetic Thbd
Pro/Pro
mice). (b–f) The frequency of apoptotic nuclei did not differ between nondiabetic Thbd
Pro/Pro
(b), diabetic Thbd
Pro/Pro
APC
high
(d) or minocycline-treated diabetic Thbd
Pro/Pro
mice (e) and was significantly lower in comparison with diabetic Thbd
Pro/Pro
mice (c,f). b–e, TUNEL stain: brown, TUNEL-positive cells detected by HRP-DAB reaction; blue, hematoxylin counterstain; scale bar, 15 mm. f, results
obtained by analyzing 50 glomeruli from seven different mice per group. C, nondiabetic control mice; DM, diabetic mice; MC, minocycline-treated mice.
Mean value ± s.e.m. a,f,**P o 0.01 (ANOVA).
ab c
d
e
110
100
90
80
PC activation
(percentage of control)
Frequency of apoptotic cells
(relative to control, %)
70
60
50
40
30
0
500
400
300
200
*
100
0
–
36
Time (h)
12 24
*
*
#
#
#
#
14
12
10
8
6
4
2
Apoptosis
(frequency, %)
0
+ + + – Glucose (25 mM)
– – + – – APC (2 nM)
– – – + – Caspase-3 inhibitor
– – – – + Mannitol (25 mM)
Frequency of apoptotic cells
(relative to control, %)
400
300
200
****
100
0
–
+ + + + Glucose (25 mM)
– – 2 20 200 APC (nM)
– – – – – Mannitol (25 mM)
*
–
–
+
Frequency of apoptotic cells
(relative to control, %)
200
150
100
50
0
Frequency of apoptotic cells
(relative to control, %)
300
200
*
100
0
–
+ + + – Glucose (25 mM)
– – + – – APC (2 nM)
– – – + – Caspase-3 inhibito
r
– – – – + Mannitol (25 mM)
*
– + + + – Glucose (25 mM)
– – + – – APC (2 nM)
– – – + – Caspase-3 inhibito
r
– – – – + Mannitol (25 mM)
Figure 5 APC prevents hyperglycemia-
induced apoptosis in endothelial cells and
podocytes, but not mesangial cells, in vitro.
(a) Hyperglycemia-induced loss of
thrombomodulin-dependent protein C (PC)
activation precedes apoptosis. Time curve
of thrombomodulin-dependent protein C
activation (left scale) and apoptosis (right
scale) after exposure of HUVECs to a high
glucose concentration (25 mM). ’, protein C
activation; B, apoptosis; #, P o 0.05
compared to control for protein C activation.
*P o 0.05 compared to control for
apoptosis. (b–e) Bar graphs indicating the frequency of apoptotic HUVECs (b), HPMVECs (c), mesangial cells (d) or podocytes (e) exposed to high-glucose
conditions (25 mM glucose). Cells exposed to high-glucose conditions were cultured in the absence or presence of APC (concentrations as indicated) or
caspase-3 inhibitor (b,c,e). Panels a–e, mean value ± s.e.m. of three independent experiments; panels b–e,*P o 0.01 versus control (5 mM glucose, open
bars) (ANOVA).
ARTICLES
NATUR E MEDI CINE VOLUME 13
[
NUMBER 11
[
NOVEMBER 2007 1353
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
Bax
200
125
*
*
100
75
50
25
0
150
100
50
0
Bax
Cytochrome C
Glucose (25 mM)
APC (2 nM)
Mannitol (25 mM)
Smac
Actin
Bax
Actin
Actin
VDAC
250
200
150
100
50
0
250
150
100
50
0
VDAC
Glucose (25 mM)
Mannitol (25 mM)
Intensity (relative to control, %)
APC (2 nM)
Glucose (25 mM)
400
300
200
100
150
100
50
0
CDM
WT APC
high
Thbd
Pro/Pro
CDM CDM
0
Mannitol (25 mM)
APC (2 nM)
Glucose (25 mM)
Mannitol (25 mM)
APC (2 nM)
––
––
–––
–
++
+
+
Bcl-2
Actin
Glucose (25 mM)
APC (2 nM)
Mannitol (25 mM)
–
–
–
–
––
–
–
++
+
+
–
–
–
–
––
–
–
++
+
+
Intensity (relative to control, %)
Intensity (relative to control, %)
Frequency of apoptotic cells
(relative to control, %)TSP-1 levels, relative to WT (%)
––
––
–––
–
++
+
+
––
––
–––
–
++
+
+
Glucose (25 mM)
APC (2 nM)
Mannitol (25 mM)
Control
125
100
200
150
100
50
0
75
50
25
0
*
*
*
**
Glucose (25 mM)
SK1-inhibitor
Anti–PAR-1 Ab
Anti–EPCR Ab
APC (2 nM)
–
–
–
–
–
–
–
–
–
–––
––
–
–
––
++++
++
+
+
+
+
+
+
––
––
–––
–
++
+
+
25 mM glucose
25 mM glucose
25 mM glucose 25 mM glucose
25 mM glucose
+ 2 nM APC
+ Anti–PAR-1 Ab + Anti–EPCR Ab
+ 2 nM APC + 2 nM APC
+ 2 nM APC
+ 50 µM SK1-inhibitor
*
*
*
*
*
a
d
e
f
bc
Figure 6 APC inhibits mitochondrial-dependent apoptosis in high glucose–stressed endothelial cells through PAR-1 and EPCR. (a) APC normalizes
expression of Bax (top bar graph) and Bcl-2 (bottom bar graph) in HUVECs in vitro. A representative western blot is also shown. (b) APC prevents
translocation of Bax into mitochondria. Bar graphs summarize results in cytosolic (top) and mitochondrial (bottom) cell extracts. A representative western
blot is also shown; actin (cytosolic marker) and VDAC (mitochondrial marker) are shown as controls for subcellular fractionation. (c) Cytoplasmic
translocation of cytochrome c (top) and Smac (bottom) in glucose-stressed HUVECs is prevented by APC. Bar graphs and representative western blot showing
cytoplasmic levels of cytochrome c and Smac. (d,e) The anti-apoptotic effect of APC in glucose-stressed HUVECs is dependent on EPCR and PAR-1.
Representative images show TUNEL assay of HUVECs cultured using conditions as indicated below images (d) and a bar graph summarizes the frequency of
apoptosis (e). TUNEL stain with fluorescein-labeled nucleotides (green) and Hoechst 33258 nuclear counterstain (blue). Anti–PAR-1 Ab, Par-1 cleavage
blocking antibody; anti-EPCR Ab, EPCR blocking antibody; SK1-inhibitor, sphingosine kinase-1 inhibitor. Scale bars, 50 mm. (f) TSP-1 expression in renal
cortex extracts in mice as determined by western blot. Bar graph summarizes results obtained from six different mice in each group. Mean value ± s.e.m.
WT, wild-type mice; C, nondiabetic control mice; DM, diabetic mice. *P o 0.05 (ANOVA). Panels a–c,e, mean value ± s.e.m. of at least three independent
experiments; *P o 0.05 versus control (5 mM glucose, open bars).
ARTICLES
1354 VOLUME 13
[
NUMBER 11
[
NOVEMBER 2007 NATURE MED ICI NE
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
mechanisms via thrombomodulin-dependent protein C activation in
regulating experimental diabetic nephropathy.
APC normalizes nitrosative stress in diabetic nephropathy
Diabetic complications and apoptosis have been linked with
hyperglycemia-induced mitochondrial superoxide overproduction
21,22
.
Recently, an antioxidant effect of APC has been proposed
23
. Therefore,
we determined nitrotyrosine abundance as a marker of in vivo reactive
oxygen generation. In kidneys of diabetic wild-type mice, nitrotyrosine
levels were markedly increased (Supplementary Fig. 3b online).
Nitrotyrosine levels further increased in diabetic Thbd
Pro/Pro
mice,
whereas no increase of nitrotyrosine was observed in diabetic APC
high
mice. Treatment with enoxaparin had no effect on nitrotyrosine levels
in diabetic wild-type mice and increased nitrotyrosine levels in diabetic
APC
high
mice (Supplementary Fig. 3b online). The differential
regulation of nitrotyrosine concentrations in mice with genetically
determined APC formation is consistent with a role of APC in
regulating oxidative DNA damage and apoptosis.
APC prevents apoptosis in endothelial cells and podocytes
We next carried out in vitro experiments to gain further insight into
the mechanism through which APC prevents hyperglycemia-induced
glomerular apoptosis.
Protein C activation and apoptosis were quantified at various
time points after exposure of human umbilical vein endothelial
cells (HUVECs) to high glucose concentrations (25 mM).
Thrombomodulin-dependent activation of protein C was reduced as
early as 3 h after exposure to a high glucose concentration (Fig. 5a),
whereas an increase in apoptosis was observed only after 12 h. The
high glucose–dependent loss of thrombomodulin function followed
by apoptosis is consistent with the idea that a thrombomodulin-
dependent mechanism is critical to prevent apoptosis.
High-glucose conditions induced apoptosis in HUVECs and in
human pulmonary microvascular endothelial cells (HPMVECs),
whereas the osmotic control mannitol had no effect (Fig. 5b,c). The
caspase-3 inhibitor DEVD-CHO (100 pM) prevented glucose-induced
apoptosis, indicating that the observed cell death is dependent on
caspase-3 activation (Fig. 5b,c). Treatment of endothelial cells with
APC at 2 nM prevented the pro-apoptotic effect of high-glucose
conditions (Fig. 5b,c). Enoxaparin (at 4 mg/ml) did not inhibit
apoptosis (data not shown). Because neither thrombin (0.1–10 nM)
nor protein C zymogen inhibited glucose-induced apoptosis (data not
shown), the observed anti-apoptotic effect is dependent on the
proteolytic activity of APC.
Apoptosis occurs in endothelial and nonendothelial glomerular cells
of diabetic mice (see above, Fig. 3e). Therefore, we determined whether
APC inhibits glucose-induced apoptosis in mesangial cells and
podocytes. APC (2–200 nM) did not prevent high glucose–induced
apoptosis in mesangial cells (Fig. 5d), probably reflecting the absence
of EPCR on these cells (data not shown). Conversely, APC at 2 nM as
well as a caspase-3 inhibitor efficiently inhibited glucose-induced
apoptosis in podocytes (Fig. 5e), which express both thrombomodulin
and EPCR. These in vitro findings are consistent with normalized
nephrin levels and podocyte number in diabetic APC
high
mice (Sup-
plementary Fig. 4 online). Thus, APC targets both cell types that
constitute the glomerular filtration barrier, but not mesangial cells.
APC inhibits apoptosis via mitochondria, PAR-1 and EPCR
We next characterized the cellular mechanisms through which APC
prevents apoptosis. At concentrations as low as 2 nM, APC prevented
the glucose-dependent induction of Bax and reduction of Bcl-2
(Fig. 6a). At the same dose, APC prevented glucose-mediated trans-
location of Bax into the mitochondria (Fig. 6b) and reduced cyto-
plasmic levels of Smac and cytochrome c (Fig. 6c), key steps in
mitochondrial-dependent apoptosis. Again, neither thrombin nor
protein C inhibited translocation of Bax into the mitochondria
(data not shown). These data establish that APC mediates its cyto-
protective effect at least in part through modulation of mitochondrial-
dependent apoptosis.
To explore the role of APC-mediated cell signaling mechanisms,
we determined whether binding of APC to EPCR and activation of
PAR-1 are required for the cytoprotective effect in glucose-stressed
cells. The PAR-1 specific activation peptide TFLLRNPNDK, but not
the PAR-2 activating peptide SLIGRL or a control peptide, prevented
glucose-induced apoptosis to a similar extent as did APC (data not
shown). The anti-apoptotic effect of APC was abolished after
preincubation with the antibody RCR-252, which blocks binding of
APC to EPCR, or the PAR-1 cleavage blocking antibody ATAP-2
(Fig. 6d,e)
9
. By contrast, blocking sphingosine kinase-1 (Fig. 6d,e)
or the S1P-R1 (data not shown) did not abolish the anti-apoptotic
effect of APC.
Expression of thrombospondin-1 (TSP-1) is differentially regulated
by thrombin and APC through PAR-1 (ref. 24). As compared to
nondiabetic wild-type mice, TSP-1 expression in renal cortex was
increased in diabetic wild-type mice and was further increased in
diabetic Thbd
Pro/Pro
mice. Conversely, levels of TSP-1 in diabetic
APC
high
mice were reduced as compared with diabetic wild-type
mice (Fig. 6f). The observed expression pattern of TSP-1 is consistent
with differential in vivo effects of APC and thrombin acting
through PAR-1.
Taken together, these experiments establish that APC is causally
linked to glomerular capillary dysfunction and apoptosis of endothe-
lial cells and podocytes in experimental diabetes. The cross-talk
between the vascular compartment, endothelial cells and podocytes
via APC regulates nephropathy. The cytoprotective effect of APC is
mediated through inhibition of glucose-induced mitochondrial-
dependent apoptosis.
DISCUSSION
It is generally accepted that endothelial dysfunction is important for
diabetic microvascular disease. However, there has been little direct
evidence for a causative link between endothelial dysfunction, micro-
vascular disease and diabetic end-organ damage. Microvascular dis-
ease can lead to organ damage through impaired vascular function,
increased inflammation or apoptosis
21,25,26
. In the current study, mice
with a loss of thrombomodulin-dependent protein C activation had
aggravated glomerular apoptosis and diabetic nephropathy compared
with diabetic wild-type mice. Increased levels of APC formation
protected wild-type mice and mice with reduced thrombomodulin-
dependent protein C activation from hyperglycemia-induced
glomerular apoptosis and diabetic nephropathy. Thus, thrombo-
modulin-dependent APC formation mediates cytoprotective, anti-
apoptotic effects and prevents endothelial dysfunction, glomerular
capillary injury and diabetic nephropathy.
Microvascular complications in diabetic nephropathy have been
associated with hyperglycemia-induced apoptosis in mice
27
and
humans
28,29
. Increased glomerular apoptosis characterizes early, but
not advanced, stages of diabetic nephropathy in humans
29
.
The observed inhibition of glomerular apoptosis through APC is
therefore probably relevant during early disease stages. This supports
the concept that endothelial dysfunction is causally linked to dia-
betic nephropathy.
ARTICLES
NATUR E MEDI CINE VOLUME 13
[
NUMBER 11
[
NOVEMBER 2007 1355
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
The role of APC in modulating glomerular apoptosis is not
restricted to endothelial cells; in addition, APC modulates podocyte
apoptosis. This establishes a crucial function for endothelial
thrombomodulin-dependent APC formation in mediating a cross-
talk between the two cell types that constitute the cellular components
of the glomerular filtration barrier. Endothelial cells and podocytes
thus form a functional unit within the glomeruli.
As diabetic wild-type and diabetic APC
high
mice differed in respect
to in vivo APC formation but not thrombomodulin expression, the
observed effects on diabetic nephropathy do not reflect changes in
thrombomodulin expression. Moreover, the prevention of diabetic
nephropathy in diabetic Thbd
Pro/Pro
mice by genetic restoration of
APC formation provides experimental evidence that thrombomodulin-
dependent APC formation regulates glomerular apoptosis and
diabetic nephropathy.
Plasma levels of APC in APC
high
mice were within a therapeutically
relevant range. Plasma APC concentrations are about eightfold
higher than endogenous APC levels in septic individuals
30
and similar
to those observed in septic individuals receiving recombinant
human APC
31
. In marked contrast to diabetic APC
high
mice, in
which APC formation was maintained despite hyperglycemia,
diabetic wild-type mice had severely reduced APC formation.
The marked reduction of in vivo protein C activation in diabetic
wild-type mice exceeds what one would expect given the moderate
(B50%) reduction of thrombomodulin expression in diabetic
mice. It is likely that additional effects, such as shedding and post-
translational modification of thrombomodulin
32,33
or reduced
expression of EPCR, contribute to the loss of in vivo protein C
activation in hyperglycemia.
Previous in vitro studies have shown that APC inhibits apoptosis
through a receptor-dependent mechanism, independently of its anti-
coagulant function
34
. In the current study, several observations support
a cytoprotective role of APC in diabetic nephropathy independent of
its anticoagulant properties. First, anticoagulation with enoxaparin for
the entire study period prevented diabetes-associated blood coagula-
tion activation as efficiently as APC but did not protect against
glomerular apoptosis and diabetic nephropathy. Second, the hyper-
activatable form of human protein C with two amino acid changes
(D167F/D172K) that is expressed in the APC
high
mice efficiently
prevents diabetic nephropathy and glomerular apoptosis despite
having weaker anticoagulant efficiency (approximately 65% lower)
than mouse APC. Third, anticoagulation of diabetic APC
high
mice
with enoxaparin abrogates nephroprotection and increases glomerular
apoptosis despite maintaining anticoagulation. Fourth, in contrast to
APC, enoxaparin does not inhibit apoptosis in glucose-stressed
endothelial cells and podocytes in vitro. These data are entirely
consistent with previous reports establishing an anti-apoptotic effect
of APC, independent of anticoagulation, that involves receptor-
dependent mechanisms. In contrast to previous studies showing
cytoprotective and anti-apoptotic effects of APC in acute disease,
such as sepsis or stroke
8,35,36
, the current results identify an anti-
apoptotic effect of APC in chronic disease.
The mechanism underlying the anti-apoptotic effect of APC in
glucose-stressed endothelial cells depends on PAR-1 and EPCR,
whereas S1P-R1 is dispensable. A role of S1P-R1 in mediating the
nephroprotective effects of APC in vivo is unlikely considering that
S1P-R1 is virtually absent from the renal vasculature
37
.UsingPAR-1
knockout mice or a PAR-1 antagonist in vivo would provide little
insight into the role of APC-dependent PAR-1 activation in diabetic
nephropathy, because both thrombin- and APC-dependent effects
would be inhibited
7
. To gain insight into the role of thrombin- and
APC-dependent effects, we determined the expression of TSP-1, which
is known to be differentially regulated by thrombin and APC through
PAR-1 (ref. 24). The in vivo expression pattern of TSP-1, as well as
those of apoptosis regulators such as p53, is consistent with results
obtained by others
24
, and implies that APC-mediated nephroprotec-
tion involves PAR-1.
The prevention of diabetic nephropathy and glomerular apoptosis
in minocycline-treated Thbd
Pro/Pro
mice provides independent experi-
mental evidence for a role of APC in regulating apoptosis in diabetic
nephropathy. Potential mechanisms underlying the protective effect of
minocycline include inhibition of PARP-1, antioxidant effects and
modulation of mitochondrial function
38–40
. Thus, minocycline and
APC might both target a mitochondrial-dependent pathway in
diabetic nephropathy
41
.
Mitochondrial dysfunction and formation of oxygen radicals and
peroxynitrate are key events leading to chronic complications in
diabetes
21,42
. The differential regulation of peroxynitrite in diabetic
mice with genetically modified APC formation identifies a role of APC
in regulating mitochondrial function and reactive oxygen generation
in vivo. This finding is consistent with a recent report showing a direct
antioxidant effect of APC in vitro, albeit at much higher concentra-
tions (10–50 mg/ml)
23
. Our preliminary experiments show that APC
normalizes mitochondrial membrane potential in glucose-stressed
endothelial cells at much lower concentrations (5 nM; B125 ng/ml;
data not shown). This suggests that the normalization of mitochon-
drial function through APC is mechanistically linked with inhibition
of apoptosis.
Collectively, the results of the present study identify a previously
unknown role of thrombomodulin-dependent APC formation
in regulating glomerular apoptosis by modulating the intrinsic
mitochondrial apoptosis pathway. APC-mediated inhibition of
endothelial and podocyte apoptosis within glomeruli protects
against hyperglycemia-induced renal dysfunction, identifying a
crucial role of endothelial function in the pathogenesis of
diabetic nephropathy.
METHODS
Mice. Thbd
Pro/Pro
mice have been described elsewhere
13
. Based on previous
results, we generated and further characterized a mutated version of human
protein C (D167F/D172K)
43
, which results in increased activation in the
presence or absence of mouse thrombomodulin. We established a transgenic
mouse model expressing human protein C (D167F/D172K) in the liver,
resulting in high plasma concentrations of APC (Supplementary Methods,
Supplementary Fig. 1 and Supplementary Tables 1 and 2 online).
In the current study, we used littermates in which at least 98% of the genetic
background was C57BL/6 derived. We injected a subset of diabetic mice (see
below) either subcutaneously with enoxaparin (4 mg/g body weight) or
intraperitoneally with minocycline (5 mg/g body weight)
44
once a day from
3 weeks after the last STZ injection until 1 d before analyses (week 26). Animal
experiments were conducted following standards and procedures approved
by the local Animal Care and Use Committee (Regierungspra
¨
sidium
Karlsruhe, Germany).
Induction of diabetes using STZ. Diabetes was induced by intraperitoneal
administration of STZ at 60 mg/kg, freshly dissolved in 0.05 M sterile sodium
citrate, pH 4.5, on five successive days in 8-week-old mice. Mice were
considered diabetic if blood glucose levels were above 300 mg/dl 16–25 d after
the last STZ injection. We obtained blood and tissue samples after 26 weeks
of persistent hyperglycemia in diabetic mice. Age-matched littermates served
as controls. At the time of analysis, the weights of the mice and of the organs
were recorded.
Determination of albuminuria. Shortly before killing, individual mice
were placed in metabolic cages and 24-h urine samples were collected.
ARTICLES
1356 VOLUME 13
[
NUMBER 11
[
NOVEMBER 2007 NATURE MED ICI NE
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
We determined urine albumin using a mouse albumin ELISA according to the
manufacturer’s instructions and urine creatinine using the Jaffe
´
method.
Analyses of blood coagulation and plasma cytokines. We determined TAT
complexes, D-dimer, plasma IL-1b, plasma IL-6 and cross-linked fibrin in
tissue extracts as described elsewhere
45
.
APC capture assay. In APC
high
mice, blood samples were collected from the
vena cava into 0.38% sodium citrate and 50 mM benzamidine HCl (final
concentration). In a subset of experiments, we injected mice with a-thrombin
(2 mU/g body weight) into the tail vein 10 min before blood sampling.
Human APC was captured from these plasma samples using an antibody highly
specific for human APC (HAPC 1555), and the activity of the captured human
protein C was determined using the chromogenic substrate PCa
45
.Todeter-
mine the concentration of the mutated human protein C (D167F/D172K),
zymogen plasma samples were incubated with the direct protein C plasma
activator PROTAC for 30 min at 37 1C (1 U/ml plasma) before performing the
capture assay. In all other mice, we determined in vivo APC formation as
described elsewhere
45,46
.
Histology and immunohistochemistry. We perfused animals with ice-cold PBS
and then with 4% buffered paraformaldehyde. Tissues were further fixed in 4%
buffered paraformaldehyde for 2 d, embedded in paraffin and processed for
sectioning. Extracellular matrix deposition in glomeruli was assessed by periodic
acid–Schiff (PAS) staining. An investigator scored sections in a blinded fashion,
according to an established scoring system (range 0–4; 0, no extracellular matrix
deposition; 4, extracellular matrix deposition in all sections of the glomeruli)
47
.
TUNEL assays on tissue sections were done according to the manufacturer’s
instructions. An investigator determined in a blinded fashion the frequency of
apoptotic cells by counting TUNEL-positive cells and total cell number within a
glomerulus and calculating the percentage of TUNEL-positive cells. Immuno-
histochemical detection of thrombomodulin with the thrombomodulin-specific
monoclonal antibody 201 B was performed as described elsewhere
45
.Glomer-
ular size was determined using ImagePro image analyses software. Using a
graduated reference slide with a grid for distances, the software was adjusted for
analysis on a daily basis. Serial sections were screened, and the largest area on
these serial sections was measured. For morphological analyses at least seven
different mice per group with at least 50 glomeruli each were used.
Cell cul ture. All experiments performed with HUVECs or HPMVECs were
done at passages 4–6. Cells were grown on 0.2% gelatin-coated plates and
maintained at 37 1C in a humidified 5% CO
2
incubator using endothelial
growth medium in the presence of growth factors and supplements. Cells were
subcultured at confluence by trypsinization with 0.05% trypsin and 0.02%
EDTA and the medium was changed every other day.
Mouse mesangial cells (SV40 MES 13) were obtained from ATCC and
cultured according to the distributor’s recommendations. Briefly, cells were
maintained at 37 1C in a humidified 5% CO
2
incubator in a 3:1 mixture of
DMEM and Ham’s F12 medium with 14 mM HEPES and 5% FBS.
Conditionally immortalized wild-type podocytes were cultured as described
elsewhere
48,49
. In brief, podocytes were routinely grown on collagen type I at
33 1C in the presence of interferon g (10 U/ml) to enhance expression of a
thermosensitive T antigen. Under these conditions, cells proliferate and are
undifferentiated. To induce differentiation, podocytes were grown at 37 1Cin
the absence of interferon g for 14 d. We performed experiments 14 d after
induction of differentiation. We confirmed differentiation by determining
expression of synaptopodin and Wilms’ tumor-1 protein.
All in vitro experiments, except those where thrombin was added, were
performed in the presence of hirudin (1 mg/ml).
For further details regarding immunoblotting, subcellular fractionation and
assays for in vitro protein C activation and apoptosis, see Supplementary
Materials and Methods.
Statistical analyses. The data are summarized as the mean ± s.e.m. Statistical
analyses were performed using Student’s t-test or ANOVA (as indicated in
the figure legend or the text). StatistiXL software (http://www.statistixl.com)
was used for all statistical analyses. Statistical significance was accepted at the
P o 0.05 level.
Additional methods. Detailed methodology is described in Supplementary
Methods.
Note: Supplementary information is available on the Nature Medicine website.
ACKNOWLEDGMENTS
We thank S. Huntscha and C. Luckner for technical support. This work was
supported by grants from the Deutsche Forschungsgemeinschaft (IS 67/2-1;
IS 67/2-2) to B.I., a grant from the Novartis Stiftung to B.I. and T.C., a grant
from the Lautenschla
¨
ger Stiftung, and a European Association for the Study of
Diabetes grant to A.B.
AUTHOR CONTRIBUTIONS
B.I. conceptually designed, conducted and interpreted the experimental work and
wrote the manuscript; I.A.V. did in vivo and ex vivo work; T.M. did ex vivo and
in vitro work, S.H. did in vitro work; M.K. and J.B. did ex vivo work; M.A.F.C.
did transgenic mouse work; M.Z. contributed to data analyses; E.B. and T.C.
performed image analyses; J.O. contributed to podocyte in vitro work, B.G.,
D.T.B. and B.W.G. performed in vitro analyses of the D167F/D172K PC mutant;
C.T.E. provided antibodies and helped interpret experimental work; H.W. helped
generate APC
high
mice and made Thbd
Pro/Pro
mice available; A.B. helped design
experiments; P.P.N. designed experiments and wrote the manuscript.
COMPETING INTERESTS STATEMENT
The authors declare competing financial interests: details accompany the full-text
HTML version of the paper at www.nature.com/naturemedicine/.
Published online at http://www.nature.com/naturemedicine
Reprints and permissions information is available online at http://npg.nature.com/
reprintsandpermissions
1. Renal Data System: USRDS 1998 Annual Data Report (NIH Publ. No. 98–3176) (US
Government Printing Office, Washington, DC, 1998).
2. Wendt, T.M. et al. RAGE drives the development of glomerulosclerosis and implicates
podocyte activation in the pathogenesis of diabetic nephropathy. Am.J.Pathol.162,
1123–1137 (2003).
3. Haraldsson, B. & Sorensson, J. Why do we not all have proteinuria? An update of
our current understanding of the glomerular barrier. News Physiol. Sci. 19,7–10
(2004).
4. Dalla Vestra, M. et al. Is podocyte injury relevant in diabetic nephropathy? Studies in
patients with type 2 diabetes. Diabetes 52, 1031–1035 (2003).
5. Meyer, T.W., Bennett, P.H. & Nelson, R.G. Podocyte number predicts long-term urinary
albumin excretion in Pima Indians with Type II diabetes and microalbuminuria.
Diabetologia 42, 1341–1344 (1999).
6. Esmon, C.T. Inflammation and the activated protein C anticoagulant pathway. Semin.
Thromb. Hemost. 32 (Suppl. 1), 49–60 (2006).
7. Riewald, M., Petrovan, R.J., Donner, A., Mueller, B.M. & Ruf, W. Activation of
endothelial cell protease activated receptor 1 by the protein C pathway. Science
296, 1880–1882 (2002).
8. Cheng, T. et al. Activated protein C blocks p53-mediated apoptosis in ischemic human
brain endothelium and is neuroprotective. Nat. Med. 9, 338–342 (2003).
9. Feistritzer, C. & Riewald, M. Endothelial barrier protection by activated protein C
through PAR1-dependent sphingosine 1-phosphate receptor-1 crossactivation. Blood
105, 3178–3184 (2005).
10. Finigan, J.H. et al. Activated protein C mediates novel lung endothelial barrier
enhancement: role of sphingosine 1-phosphate receptor transactivation. J. Biol.
Chem. 280, 17286–17293 (2005).
11. Borcea, V. et al. Influence of ramipril on the course of plasma thrombomodulin in
patients with diabetes mellitus. Vasa 28, 172–180 (1999).
12. Fujiwara, Y., Tagami, S. & Kawakami, Y. Circulating thrombomodulin and hematologi-
cal alterations in type 2 diabetic patients with retinopathy. J. Atheroscler. Thromb. 5,
21–28 (1998).
13. Weiler-Guettler, H. et al. A targeted point mutation in thrombomodulin generates viable
mice with a prethrombotic state. J. Clin. Invest. 101, 1983–1991 (1998).
14. Hall, S.W., Nagashima, M., Zhao, L., Morser, J. & Leung, L.L. Thrombin interacts with
thrombomodulin, protein C, and thrombin-activatable fibrinolysis inhibitor via specific
and distinct domains. J. Biol. Chem. 274, 25510–25516 (1999).
15. Festa, A. et al. Inflammation and microalbuminuria in nondiabetic and type 2 diabetic
subjects: The Insulin Resistance Atherosclerosis Study. Kidney Int. 58, 1703–1710
(2000).
16. Isermann, B. et al. The thrombomodulin–protein C system is essential for the
maintenance of pregnancy. Nat. Med. 9, 331–337 (2003).
17. Mosnier, L.O. & Griffin, J.H. Inhibition of staurosporine-induced apoptosis of endo-
thelial cells by activated protein C requires protease-activated receptor-1 and endothe-
lial cell protein C receptor. Biochem. J. 373, 65–70 (2003).
18. Tewari, M. et al. Yama/CPP32 beta, a mammalian homolog of CED-3, is a CrmA-
inhibitable protease that cleaves the death substrate poly(ADP-ribose) polymerase. Cell
81, 801–809 (1995).
ARTICLES
NATUR E MEDI CINE VOLUME 13
[
NUMBER 11
[
NOVEMBER 2007 1357
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
19. Wang, J. et al. Minocycline up-regulates Bcl-2 and protects against cell death in
mitochondria. J. Biol. Chem. 279, 19948–19954 (2004).
20. Kelly, K.J. et al. Minocycline inhibits apoptosis and inflammation in a rat
model of ischemic renal injury. Am. J. Physiol. Renal Physiol. 287, F760–F766
(2004).
21. Brownlee, M. Biochemistry and molecular cell biology of diabetic complications.
Nature 414, 813–820 (2001).
22. Zou, M.H., Shi, C. & Cohen, R.A. High glucose via peroxynitrite causes tyrosine
nitration and inactivation of prostacyclin synthase that is associated with thromboxane/
prostaglandin H(2) receptor-mediated apoptosis and adhesion molecule expression in
cultured human aortic endothelial cells. Diabetes 51, 198–203 (2002).
23. Yamaji, K. et al. Activated protein C, a natural anticoagulant protein, has antioxidant
properties and inhibits lipid peroxidation and advanced glycation end products
formation. Thromb. Res. 115, 319–325 (2005).
24. Riewald, M. & Ruf, W. Protease-activated receptor-1 signaling by activated protein C in
cytokine-perturbed endothelial cells is distinct from thrombin signaling. J. Biol. Chem.
280, 19808–19814 (2005).
25. Goligorsky, M.S., Chen, J. & Brodsky, S. Workshop: endothelial cell dysfunction leading
to diabetic nephropathy: focus on nitric oxide. Hypertension 37, 744–748 (2001).
26. Endemann, D.H. & Schiffrin, E.L. Endothelial dysfunction. J. Am. Soc. Nephrol. 15,
1983–1992 (2004).
27. Murata, I. et al. Apoptotic cell loss following cell proliferation in renal glomeruli of
Otsuka Long-Evans Tokushima Fatty rats, a model of human type 2 diabetes. Am. J.
Nephrol. 22, 587–595 (2002).
28. Kumar, D., Robertson, S. & Burns, K.D. Evidence of apoptosis in human diabetic
kidney. Mol. Cell. Biochem. 259, 67–70 (2004).
29. Baba, K. et al. Involvement of apoptosis in patients with diabetic nephropathy: A study
on plasma soluble Fas levels and pathological findings. Nephrology (Carlton) 9,94–99
(2004).
30. Liaw, P.C. et al. Patients with severe sepsis vary markedly in their ability to generate
activated protein C. Blood 104, 3958–3964 (2004).
31. Macias, W.L. et al. New insights into the protein C pathway: potential implications for
the biological activities of drotrecogin alfa (activated). Crit. Care 9 Suppl 4, S38–S45
(2005).
32. Glaser, C.B. et al. Oxidation of a specific methionine in thrombomodulin by activated
neutrophil products blocks cofactor activity. A potential rapid mechanism for modula-
tion of coagulation. J. Clin. Invest. 90, 2565–2573 (1992).
33. Boehme, M.W. et al. Release of thrombomodulin from endothelial cells by concerted
action of TNF-alpha and neutrophils: in vivo and in vitro studies. Immunology 87 ,
134–140 (1996).
34. Mosnier, L.O., Gale, A.J., Yegneswaran, S. & Griffin, J.H. Activated protein C variants
with normal cytoprotective but reduced anticoagulant activity. Blood 104, 1740–1744
(2004).
35. Bernard, G.R. et al. Safety assessment of drotrecogin alfa (activated) in the treatment
of adult patients with severe sepsis. Crit. Care 7, 155–163 (2003).
36. Taylor, F.B. & Kinasewitz, G. Activated protein C in sepsis. J. Thromb. Haemost. 2,
708–717 (2004).
37. Chae, S.S., Proia, R.L. & Hla, T. Constitutive expression of the S1P1 receptor in adult
tissues. Prostaglandins Other Lipid Mediat. 73, 141–150 (2004).
38. Virag, L. & Szabo, C. The therapeutic potential of poly(ADP-ribose) polymerase
inhibitors. Pharmacol. Rev. 54, 375–429 (2002).
39. Cipriani, G. et al. Nuclear poly(ADP-ribose) polymerase-1 rapidly triggers mitochon-
drial dysfunction. J. Biol. Chem. 280, 17227–17234 (2005).
40. Alano, C.C., Kauppinen, T.M., Valls, A.V. & Swanson, R.A. Minocycline inhibits
poly(ADP-ribose) polymerase-1 at nanomolar concentrations. Proc. Natl. Acad. Sci.
USA 103, 9685–9690 (2006).
41. Szabo, C., Biser, A., Benko, R., Bottinger, E. & Susztak, K. Poly(ADP-ribose)
polymerase inhibitors ameliorate nephropathy of type 2 diabetic Lepr
db/db
mice.
Diabetes 55, 3004–3012 (2006).
42. Kiss, L. et al. The pathogenesis of diabetic complications: the role of DNA injury and
poly(ADP-ribose) polymerase activation in peroxynitrite-mediated cytotoxicity. Mem.
Inst. Oswaldo Cruz 100 (Suppl. 1), 29–37 (2005).
43. Richardson, M.A., Gerlitz, B. & Grinnell, B.W. Enhancing protein C interaction with
thrombin results in a clot-activated anticoagulant. Nature 360, 261–264 (1992).
44. Chen, M. et al. Minocycline inhibits caspase-1 and caspase-3 expression and delays
mortality in a transgenic mouse model of Huntington disease. Nat. Med. 6, 797–801
(2000).
45. Isermann, B. et al. Endothelium-specific loss of murine thrombomodulin disrupts the
protein C anticoagulant pathway and causes juvenile-onset thrombosis. J. Clin. Invest.
108, 537–546 (2001).
46. Kerlin, B.A. et al. Survival advantage associated with heterozygous Factor V Leiden
mutation in patients with severe sepsis and in mouse endotoxemia. Blood 102,
3085–3092 (2003).
47. Calkin, A.C. et al. PPAR-a and -g agonists attenuate diabetic kidney disease in the
apolipoprotein E knockout mouse. Nephrol. Dial. Transplant. 21, 2399–2405 (2006).
48. Mundel, P. et al. Rearrangements of the cytoskeleton and cell contacts induce process
formation during differentiation of conditionally immortalized mouse podocyte cell
lines. Exp. Cell Res. 236, 248–258 (1997).
49. Asanuma, K. et al. Synaptopodin regulates the actin-bundling activity of alpha-actinin
in an isoform-specific manner. J. Clin. Invest. 115, 1188–1198 (2005).
ARTICLES
1358 VOLUME 13
[
NUMBER 11
[
NOVEMBER 2007 NATURE MED ICI NE
© 2007 Nature Publishing Group http://www.nature.com/naturemedicine
