Abstract
purpose. Retinal ischemia-reperfusion injury induces apoptosis of retinal neurons. The
purpose of this study was to examine the association of c-Jun,
caspase-1, -2, and -3 immunoreactivities and neuronal apoptosis in the
retinal ganglion cell layer (GCL) and to study the effects of
intravitreal brain-derived neurotrophic factor (BDNF) on the expression
of these gene products in a rat model of retinal ischemia-reperfusion
injury.
methods. After 60 minutes of ischemia, eyes were enucleated after 3, 6, 12, 24,
and 168 hours of reperfusion. The numbers of c-Jun-, caspase-1-,
caspase-2-, caspase-3, and TdT-dUTP terminal nick-end labeling
(TUNEL)–positive cells in the GCL were counted. Recombinant human BDNF
(5 μg) or vehicle was injected intravitreally immediately after
reperfusion. At 6, 24, and 168 hours, the numbers of immunoreactive
cells in BDNF- and vehicle-treated groups were compared.
results. Expression of c-Jun and caspase-2 was found in dying cells in
flat-mounted retinas. The numbers of caspase-1– and
caspase-3–positive cells were fewer than c-Jun– or
caspase-2–positive cells. Cell death in the retinal GCL was suppressed
by an intravitreal injection of BDNF. The numbers of TUNEL- and
caspase-2–positive cells were lower in the BDNF-treated group at 6
hours after reperfusion (P < 0.01). The number of
c-Jun–positive cells in the treated retinas was not altered by the
treatment.
conclusions. Expression of c-Jun and caspase-2 is associated with neuronal cell
apoptosis in the GCL. Suppression of caspase-2 expression may explain
the neuroprotective effects of BDNF.
Brain-derived neurotrophic factor (BDNF) is a neurotrophin
that suppresses neuronal death from various injuries. Although little
is known about how BDNF protects neurons from such injuries, it is
likely that it modifies some mechanism(s) in the pathway of cell death.
Retinal ganglion cell (RGC) death occurs in several ocular and systemic
diseases including glaucoma, ischemic optic neuropathy, and
Alzheimer’s disease. The RGC death in these diseases has been shown to
be by apoptosis.
1 In neuronal apoptosis, the caspase
family proteases play an important role in the execution phase.
Although BDNF is known to suppress RGC death, it is not clear how BDNF
prevents apoptosis and whether BDNF suppresses the activation of
caspase family proteases.
To investigate the mechanism of RGC death, experiments have been
carried out on the rat axotomy and on the retinal ischemia-reperfusion
models. In these models, the expression of immediate early gene
products such as c-Jun is found in dying cells.
2 We have
shown previously that expression of cell cycle–related gene products
such as c-Jun and cyclin D1 takes place in the dying neurons,
especially those in the inner nuclear layer.
3
The retinal ischemia-reperfusion model is known to induce apoptosis of
cells in all layers of the retina, and we used this model to study the
mechanism of cell death. We chose to study the expression of cell
cycle–related gene products that have been shown to be associated with
cell death and to study only cells (RGCs and amacrine) in the retinal
ganglion cell layer (GCL). We shall show that cell death after
ischemia-reperfusion injury in the retinal GCL is associated with the
expression of c-Jun and caspase family proteases and describe the
neuroprotective effects of BDNF.
Adult female Sprague–Dawley rats (Japan SLC, Hamamatsu, Japan)
weighing 160 to 180 g (
n = 350) were used. The
experiments were conducted in compliance with the ARVO Statement for
the Use of Animals in Ophthalmic and Vision Research. Rats were
anesthetized by intraperitoneal pentobarbital (50 mg/kg), and retinal
ischemia was induced by cannulating the anterior chamber with a
27-gauge needle connected to a bag containing normal
saline.
4 The intraocular pressure was increased to 110 mm
Hg for 60 minutes by elevating the bag. The temperature of the rats was
kept at 37°C by a heating pad. Sham operations were performed without
increasing the intraocular pressure.
The following antibodies were used: rabbit anti-mouse c-Jun/AP-1
(N) antibody raised against a peptide that corresponds to amino acids
91–105 mapping within the amino terminal domain; goat anti–caspase-1
(M-19) antibody against amino acids 276–294 mapping to the carboxyl
terminus of the 20-kDa subunit of caspase-1 of mouse origin; goat
anti–caspase-2(N-19) antibody against amino acids 3–21 mapping to
the amino terminus of the precursor of caspase-2 of mouse origin; and
goat anti–caspase-3 (L-18) antibody against amino acids 157–174
mapping to the carboxyl terminus of the 20-kDa subunit of the protease
precursor of human origin. All antibodies were purchased from Santa
Cruz Biotechnology (Santa Cruz, CA).
At 3, 6, 12, 24, and 168 hours after reperfusion, rats were deeply
anesthetized and perfused through the heart with 4% paraformaldehyde
in 0.1 M phosphate-buffered saline (PBS). The eyes were enucleated, and
the retinas were separated from the eyecups in plastic petri dishes
containing PBS. The retinas were then fixed for 2 hours in 4%
paraformaldehyde. After rinsing in PBS, retinas were incubated for 30
minutes at room temperature with 0.2% Triton-X in PBS in plastic petri
dishes.
For the detection of caspase-1 and caspase-3 immunoreactivities, the
retinas were incubated with NeuroPore (Trevigen, Gaithersburg, MD) for
30 minutes (caspase-1) or 3 hours (caspase-3) at room temperature. The
retinas were rinsed with PBS and then incubated with rabbit anti-mouse
c-Jun antibody (1 μg/ml), goat anti–caspase-1 antibody (20 μg/ml),
goat anti–caspase-2 antibody (20 μg/ml), or goat anti–caspase-3
antibody (20 μg/ml) for 48 hours at 4°C. The working concentrations
of antibodies were determined after preliminary experiments with
various concentrations. The retinas were then rinsed with PBS and
incubated with fluorescein isothiocyanate (FITC)–conjugated
anti-rabbit or anti-goat IgG antibody for 75 minutes.
To count the numbers of cells in the retinal GCL, the retinas were
stained with propidium iodide (PI, 20 μg/ml) for 10 minutes at room
temperature, and the retinas were flatmounted vitreous side up on glass
slides. The retinas in which the primary antibodies were omitted were
used as negative controls. Positive controls were generated by using
retinal sections or flatmounted retinas in which these proteins were
known to be expressed.
3 5
After fixation, the retinas were subjected to three cycles of
freezing and thawing and then incubated overnight with NeuroPore in
plastic petri dishes. The DNA nick-end labeling was performed by a
Mebstain apoptosis kit (MBL, Nagoya, Japan). The retinas were incubated
with TdT and biotinylated dUTP in TdT buffer for 2 hours at 37°C,
then rinsed with termination buffer containing 30 mM sodium chloride
and 3 mM sodium citrate for 30 minutes at room temperature followed by
rinsing with PBS. The retinas were incubated with avidin–FITC for 2
hours at 37°C and rinsed with PBS. After PI staining, the retinas
were flatmounted on glass slides. Positive controls were generated by
incubating the specimens with DNase1 in water (1 μg/ml, for 1 hour at
37°C) before incubation with TdT and biotinylated dUTP. The control
eyes without ischemia-reperfusion injury were used as negative
controls.
The flatmounted retinas were photographed with a scanning laser
confocal microscope (model LSM 410; Zeiss, Oberkochen, Germany) using a
green filter to detect FITC and a red filter for PI; the focus was in
the retinal GCL plane. The numbers of FITC-labeled c-Jun– and
caspase-positive cells, TdT-dUTP terminal nick-end labeling
(TUNEL)–positive cells and PI-stained cells were counted in six areas
(0.2 × 0.2 mm), 1 and 2 mm away from optic disc and every 30°
of the circle in each quadrant. Thus, data from 24 areas from one eye
were obtained. Regions with thick nerve fiber or blood vessels were
avoided, and a more central or peripheral area was chosen. Of the
PI-stained cells, white blood cells, which have fragmented nuclei, and
red blood cells, which have oval-shaped and spindle-shaped endothelial
cells, were excluded from cell counts. The cell count was done by two
examiners in a masked fashion. Data are expressed as the number of
cells per square millimeter at each time point, and results are
expressed as mean ± SEM.
To label the RGCs, rats were anesthetized and placed in
a stereotaxic frame (Narishige, Tokyo, Japan). A part of the skull and
cerebral cortex was removed to expose the superior colliculi
bilaterally. Multiple injections of neurotracer, FluoSpheres, or 1%
DiI (Molecular Probe, Eugene, OR) were made in different regions of
each superior colliculus with a glass micropipette attached to a 1-μl
Hamilton syringe, at a depth of 0.5 or 1.0 mm. Seven days later,
ischemia-reperfusion injury was induced, and the eyes were treated as
described for TUNEL and immunostaining.
Recombinant human brain-derived neurotrophic factor (BDNF;
N-terminal methionine-free, 2.0 mg/ml in 10 mM sodium phosphate and 150
mM NaCl, pH 7.0) was obtained from Regeneron Pharmaceuticals
(Tarrytown, NY).
6 Immediately after reperfusion, 2.5 μl
(5.0 μg) of BDNF was injected into the vitreous cavity with a 5-μl
Hamilton microsyringe attached to a 30-gauge needle. In the control
group, 2.5 μl of vehicle was injected. Care was taken not to injure
the lens and retinal vessels, and eyes that exhibited any complications
were excluded. At 6, 24, and 168 hours after reperfusion, rats were
deeply anesthetized, and the eyes taken for immunohistochemical and
TUNEL studies as described.
Statistical analysis was done using a two-way ANOVA followed by
Fisher’s post hoc test. P < 0.05 was considered
statistically significant.
The number of cells in the retinal GCL decreased from 3250.2 ± 131.3 cells/mm
2 in sham-operated eyes
(mean ± SEM,
n = 5) to 2098.2 ± 45.3 at 6
hours, to 1858.5 ± 63.2 at 24 hours, and 1439.2 ± 50.8 at
168 hours after reperfusion
(Fig. 1) .
TUNEL-positive cells were present at 3 hours after reperfusion
(442.5 ± 47.2 cells/mm
2,
n = 5), and the number peaked at 6 hours (545.3 ± 67.0;
Fig. 1 ).
Cells that were positive for c-Jun or caspase-1, -2, and -3
immunoreactivities were also present at 3 hours after reperfusion
(Fig. 1) . Among these four antibodies, anti–c-Jun and anti–caspase-2 showed
consistent and strong immunostaining. Much weaker staining was obtained
with anti–caspase-1 and anti–caspase-3 antibodies. The number of
immunoreactive cells was highest at 6 hours for c-Jun at 202.2 ±
24.0 cells/mm
2 (
n = 5), for
caspase-1 at 51.2 ± 12.5 (
n = 5), for caspase-2
at 230.2 ± 15.7 (
n = 5), and for caspase-3 at
60.3 ± 9.8 (
n = 5;
Fig. 1 ). Expression of c-Jun
and the caspases was also confirmed in retrogradely labeled RGCs
(Fig. 2) . Pertinent to this study, the expression of c-Jun and the caspases was
found in cells with shrunken cell bodies and condensed chromatin (i.e.,
dying cells;
Fig. 2 ).
In the vehicle-treated group, the numbers of cells in the retinal
GCL were 2126.8 ± 38.2 cells/mm
2 at 6
hours, 1970.8 ± 56.7 at 24 hours, and 1432.5 ± 78.2 at 168
hours after reperfusion
(Fig. 4) . After intravitreal BDNF, there were
2370.3 ± 122.5 cells/mm
2 (
n = 5) at 6 hours (
P < 0.05, compared with vehicle
treatment), 2310.0 ± 51.3 at 24 hours (
P <
0.01), and 2261.8 ± 16.1 at 168 hours (
P < 0.01;
Figs. 3 and 4 ). Thus, there were 10.3% more neurons in the retinal GCL at 6 hours,
14.7% more at 24 hours, and 36.7% more at 168 hours in the
BDNF-treated animals.
Without BDNF treatment, the number of TUNEL-positive cells was
545.2 ± 29.7 cells/mm
2 (
n =
5) at 6 hours, and the number was 357.4 ± 67.7 (
n = 5) with BDNF. This difference was statistically significant
(
P < 0.01,
Figs. 3 and 4 ). At 24 hours after
reperfusion, however, the number of TUNEL-positive cells was not
markedly different from the controls
(Fig. 4) .
The number of c-Jun–positive cells in the control group was 226.5 ± 24.3 cells/mm
2 (
n = 5) at 6
hours after reperfusion and decreased to 68.0 ± 14.3 at 24 hours
(
n = 5;
Figs. 3 and 4 ). The number of c-Jun–positive
cells in the BDNF-treated group was 224.8 ± 24.9
cells/mm
2 (
n = 5) at 6 hours and
82.0 ± 18.1 at 24 hours
(Figs. 3 and 4) . These values were not
significantly different from those of the controls.
The number of caspase-2–positive cells in the vehicle control also
showed a peak at 6 hours after reperfusion (244.6 ± 15.7
cells/mm
2 ;
n = 5) and decreased
to 57.8 ± 12.6 at 24 hours
(Fig. 4) . In the BDNF-treated groups,
there were fewer caspase-2–positive cells than in the vehicle group:
124.4 ± 35.4 cells/mm
2 (
n =
5) at 6 hours and 35.4 ± 11.2 at 24 hours
(Figs. 3 and 4) . The
difference between the control and BDNF-treated groups at 6 hours was
statistically significant (
P < 0.01). In the
quantitative study of caspase-1–positive and caspase-3–positive
cells, there were no statistically significant differences between
these groups and the BDNF-treated group (graphic data not shown).
These results showed that c-Jun and the caspase family proteases,
especially caspase-2, were expressed in cells in the retinal GCL that
had morphologic and histochemical signs of dying cells
(Figs. 2 and 3) .
Although the observations made with the retrograde labeled retinas
suggest that the dying cells were mainly RGCs, we cannot rule out
amacrine cells as being part of this group. In any case, these
observations suggest an association between the cell death and c-Jun
and/or caspase-2 expression. This relationship agrees with the role
played by c-Jun in cell death in the central and peripheral nervous
systems and in the RGC death in the axotomy model.
2
Among the caspase family of proteases, expression of caspase-2 was
detected in the highest number of cells, with much fewer cells
expressing caspase-1 and -3. There is good evidence that caspase-2
induces apoptosis,
7 8 and more recently, an independent
processing of caspase-2 and upregulation of caspase-3 activity have
been shown in trophic factor–deprived PC12 cell death. This type of
cell death required caspase-2 activation, and the upregulation of
caspase-3–like activity was neither necessary nor sufficient to induce
cell death.
9 Our data are in keeping with these findings.
Expression of c-Jun and caspase-2 was already present at 3 hours after
reperfusion
(Fig. 1) , and their peak expression was seen at 6 hours.
Thus, most of the c-Jun– and caspase-2–related cell death occurred
within 24 hours after reperfusion, which also corresponds to the time
course of TUNEL-positive cells
(Fig. 1) . However, there were a number
of TUNEL-positive cells that did not express c-Jun or caspase
immunoreactivities. One possible explanation for this discrepancy is
that the window for the expression of c-Jun and caspases is shorter
than the window to detect dying cells by TUNEL staining. Another
possible explanation is that cell death may occur independent of c-Jun
expression, caspase expression, or both.
The protective effects of BDNF were evident as early as 6 hours after
reperfusion
(Fig. 4) . Intravitreal injection of BDNF decreased the
number of TUNEL-positive cells in the GCL, but the number of
c-Jun–positive cells remained unchanged. This suggests that BDNF does
not influence the expression of c-Jun, which agrees with De Felipe and
Hunt,
10 who suggested that BDNF does not regulate c-Jun
expression in damaged neurons. However, BDNF did suppress the
expression of caspase-2, both directly and indirectly. Similar
regulation of caspase-2 expression by nerve growth factor occurred in
sympathetic neurons and PC12 cells.
8
The relationship between c-Jun– and caspase-2–dependent pathways is
not known: They may be distinct pathways or they may be linked. In this
experiment, the number of caspase-2–positive cells was significantly
decreased by BDNF but not the c-Jun–positive cells. These findings
suggest two hypotheses: One is that c-Jun and caspase-2 are linked and
BDNF works between the c-Jun and caspase-2 steps and the other is that
c-Jun and caspase-2 have distinct pathways and BDNF only blocks
expression of caspase-2. From our experiments, it is not possible to
select which of these hypotheses is correct because we do not know
whether a single cell expresses c-Jun and caspase-2 sequentially.
Expression of c-Jun and caspase-2 was detected in cells with a shrunken
cell body and condensed chromatin (i.e., morphological signs of cell
death;
Figs. 2 and 3 ). Thus, even if BDNF works between c-Jun and
caspase-2, it is unlikely that the process of apoptosis can be
reversed. We hypothesize that c-Jun–dependent and caspase-2–dependent
cell death occurs via distinct pathways and that the cascade associated
with caspase-2 is blocked by BDNF but c-Jun–dependent pathways are
not.
In this study, retrograde labeling of the RGCs was not done routinely;
instead, we used confocal microscopy and counted the number of cells
stained with PI in the retinal GCL. Because amacrine cells are also
found in the retinal GCL, their presence will alter the overall numbers
of cells. Thus, the cell counts presented represent both RGCs and
amacrine cells. Even when the RGCs were labeled retrogradely, it was
not easy to differentiate cells that lost the dye due to the
destruction of cell body from displaced amacrine cells
(Fig. 2) .
Because our analysis consisted of the cell numbers in the retinal GCL,
and because our conclusions depended on changes in the cell numbers,
the presence of amacrine cells should not alter our conclusions.
In summary, we have shown that the expression of c-Jun and caspase-2 is
associated with cell death in the retinal GCL in a rat
ischemia-reperfusion injury model. BDNF had a neuroprotective effect on
cell death, possibly by suppressing caspase-2 expression, but had no
influence on c-Jun expression. Although the pathway linked to caspase-2
is suppressed by BDNF, there may exist other pathways through which
BDNF works.