Download - Death of Hypothalamic Astrocytes in Poorly Controlled Diabetic Rats is Associated with Nuclear Translocation of Apoptosis Inducing Factor

ORIGINAL ARTICLE

Death of Hypothalamic Astrocytes in Poorly Controlled Diabetic Rats isAssociated with Nuclear Translocation of Apoptosis Inducing FactorC. Garcıa-Caceres,*�� A. Lechuga-Sancho,* J. Argente,*�� L. M. Frago*��1 and J. A. Chowen*�1

*Hospital Infantil Universitario Nino Jesus, Servicio de Endocrinologıa, Madrid, Spain.

�Universidad Autonoma de Madrid, Departamento de Pediatrıa, Madrid, Spain.

�CIBER Fisiopatologıa Obesidad y Nutricion (CIBEROBN), Instituto de Salud Carlos III, Madrid, Spain.

Many of the secondary complications arising from poorly controlled

diabetes mellitus (DM) involve increased cell death (1, 2). Indeed,

the hormonal imbalances that occur in poorly controlled diabetic

subjects (3) may be due, at least in part, to increased cell death in

the anterior pituitary (4). In addition, the increase in cell death at

the level of the hypothalamus (5, 6) may also be involved. Neuronal

death is reported to occur in the hypothalamus after 3 months (6),

whereas astrocytes are affected as early as 4 weeks after the onset

of diabetes (5). Not only is there increased death of these glial cells,

but also their proliferation is reduced, resulting in a significant

reduction in the number of astrocytes in the arcuate nucleus. As

hypothalamic glial cells modulate neuroendocrine functions (7),

these changes, as well as the reduction in the projections per astro-

cyte (5), could participate in the hormonal changes observed in

poorly controlled diabetes.

Astrocytes are involved in the maintenance of the neuronal

homeostatic environment (8), provide metabolic and trophic support

to neurones (9), regulate neurogenesis and participate in neuropro-

tection (10). Synaptic number and activity are also modulated by

glial cells (7, 11–13) and, in the hypothalamus, this is associated

with physiological changes in endocrine function (7, 12, 13). Thus,

a reduction in hypothalamic astrocytes and their projections could

modulate endocrine function, as well as render neurones more sus-

ceptible to adverse situations. Furthermore, as astrocytes modulate

Journal ofNeuroendocrinology

Correspondence to:

Julie A. Chowen, Servicio de

Endocrinologıa, Hospital Infantil

Universitario Nino Jesus, Avenida

Menendez Pelayo 65, 28009 Madrid,

Spain (e-mail: jachowen@

telefonica.net).1These authors contributed equally to

this study.

Astrocytes in the hypothalamus of poorly controlled diabetic rats are reduced in number, due to

increased apoptosis and decreased proliferation, and undergo morphological changes, including

a decrease in projections. These changes are associated with modifications in synaptic proteins

and most likely affect neuroendocrine signalling and function. The present study aimed to deter-

mine the intracellular mechanisms underlying this increase in hypothalamic cell death. Adult

male Wistar rats were injected with streptozotocin (70 mg ⁄ kg, i.p) and controls received vehicle.

Rats were killed at 1, 4, 6 and 8 weeks after diabetes onset (glycaemia > 300 mg ⁄ dl). Cell death,

as detected by enzyme-linked immunosorbent assay, increased at 4 weeks of diabetes. Immuno-

histochemistry and terminal dUTP nick-end labelling (TUNEL) assays indicated that these cells

corresponded to glial fibrillary acidic protein (GFAP) positive cells. No significant change in frag-

mentation of caspases 2, 3, 6, 7, 8, 9, or 12 was observed with western blot analysis. However,

enzymatic assays indicated that caspase 3 activity increased significantly after 1 week of diabe-

tes and decreased below control levels thereafter. In the hypothalamus, cell bodies lining the

third ventricle, fibres radiating from the third ventricle and GFAP positive cells expressed frag-

mented caspase 3, with this labelling increasing at 1 week of diabetes. However, because no

nuclear labelling was observed and this increase in activity did not correlate temporally with the

increased cell death, this caspase may not be involved in astrocyte death. By contrast, nuclear

translocation of apoptosis inducing factor (AIF) increased significantly in astrocytes in parallel

with the increase in death and AIF was found in TUNEL positive cells. Thus, nuclear translocation

of AIF could underlie the increased death, whereas fragmentation of caspase 3 could be associ-

ated with the morphological changes found in hypothalamic astrocytes of diabetic rats.

Key words: astrocytes, caspase, apoptosis.

doi: 10.1111/j.1365-2826.2008.01795.x

Journal of Neuroendocrinology 20, 1348–1360

ª 2008 The Authors. Journal Compilation ª 2008 Blackwell Publishing Ltd

Journal of NeuroendocrinologyFrom Molecular to Translational Neurobiology

glucose metabolism and glucose sensing in the central nervous sys-

tem (9), their affectation in diabetes could contribute to further

deregulation of glucose homeostasis. Hence, understanding the

mechanisms underlying glial apoptosis in metabolic conditions such

as poorly controlled diabetes mellitus could afford valuable infor-

mation regarding the role of these cells in hypothalamic function.

In diabetic animals, cell death by classical apoptotic pathways

involving caspases occurs in numerous tissues (1, 2, 5, 14). In the

intrinsic cell death pathway, initiator caspases such as caspase 9

are activated and these in turn activate effector caspases, including

caspases 3, 6 and 7. Activation of these effector caspases then trig-

gers processes that ultimately result in cell death. This pathway is

triggered by external signals or internal changes, with the mito-

chondria being a very important component of this cascade (15).

The balance between pro- and anti-apoptotic members of the Bcl-2

protein family has a crucial role in determining the integrity of the

mitochondria and its release of apoptogenic factors, and hence, cell

death (15).

The extrinsic cell death pathway is activated by ligand binding to

extra-cellular death receptors that belong to the tumour necrosis

factor (TNF) receptor superfamily (15, 16). This pathway, which

entails cleavage of procaspase-8 that can subsequently either acti-

vate executioner caspases or act directly to induce apoptosis (17),

is activated in the anterior pituitary of diabetic rats in a cell specific

manner (18, 19).

These pathways, however, are not independent because there is

extensive cross-talk between them, with many of the same intracel-

lular proteins being involved in both processes. Furthermore, apop-

totic cell death can also occur by caspase independent mechanisms

with specific mitochondrial factors also playing a fundamental role

(20, 21). For example, apoptosis-inducing factor (AIF) is normally

located in the mitochondria but, when confronted with a fatal

insult, can be translocated to the nucleus where it induces chroma-

tin condensation and DNA fragmentation (21). The present study

aimed to determine the intracellular mechanisms involved in the

increased death of hypothalamic astrocytes in response to poorly

controlled diabetes.

Materials and methods

Materials

All chemicals and reagents were purchased from Sigma or Merck (Barcelona,

Spain) unless otherwise indicated.

Animals

Young adult male Wistar rats (weighing approximately 250 g) were housed

under a 12 : 12 h light ⁄ dark cycle and given free access to rat chow and tap

water. Rats were injected with streptozotocin (70 mg ⁄ kg, i.p.) or vehicle (0.1 M

citrate buffer, pH 4.5). Blood glucose levels were measured (Glucocard Memory

2; Menarini Diagnostic, Florence, Italy) at baseline and then daily to determine

the onset of diabetes (glycaemia > 300 mg ⁄ dl) and then when killed. Rats

were killed by asphyxiation with CO2 and decapitation 1, 4, 6 or 8 weeks later.

All rats were killed between 10.00 h and 12.00 h. The brains were removed

and rapidly frozen on dry ice and stored at )70 �C until processed.

The following groups of rats were established: diabetic for 1 week

(DB1W; n = 6), diabetic for 4 weeks (DB4W; n = 6), diabetic for 6 weeks

(DB6W; n = 6) and diabetic for 8 weeks (DB8W; n = 6). Control rats were

killed at each time point but, as no differences in any of the studied vari-

ables were found (data not shown), control data were analysed as one

group for statistical analysis (n = 8). Three independent studies were per-

formed and hypothalamic cell death was verified in each study.

For all studies, rats were treated according to the European Community

laws for animal care and the study was approved by the appropriate institu-

tional ethical committee.

Protein extraction

For western blotting and enzyme-linked immunosorbent assay (ELISA), hypo-

thalami were isolated on ice using the following boundaries: an anterior cut

was made at the level of the optic chiasm, a posterior coronal section ante-

rior to the mammilary bodies, two sagittal cuts parallel to the lateral ventri-

cles, and a dorsal horizontal cut at the level of the anterior commissure.

Each hypothalamus was then divided into two symmetric halves. One hemi-

hypothalamus was employed for cell death detection ELISA assays and the

other half for western blotting. Tissue for ELISA was homogenised in lysis

buffer provided by the manufacturer of the commercial kit (Roche Diagnos-

tics, Mannheim, Germany). Tissue for western blotting was processed as pre-

viously described (5). Total protein concentration was determined by the

method of Bradford (Protein Assay; Bio-Rad Laboratories, Hercules, CA,

USA).

Insulin ELISA

This assay was performed according to the manufacturer’s instructions

(Linco Research, Inc., St Charles, MO, USA). Briefly, microtitre plates

coated with pre-titred monocloncal mouse anti-rat insulin antibody were

washed three times with wash buffer [50 mM Tris buffered saline (TBS)

containing Tween-20]. Ten microlitres of prediluted standards, quality con-

trol samples and serum samples were added to wells in duplicate. Detec-

tion antibody (biotinylated anti-insulin, 80 ll) was added and the plate

sealed and incubated at room temperature for 2 h while shaking. The

wells were then washed, 100 ll of enzyme solution (streptavidin-horse-

radish peroxidase conjugate) added and incubated (30 min). After wash-

ing, 100 ll of substrate solution (3,3¢,5,5¢-tetramethylbenzidine) was

added and the colour allowed to develop (approximately 15 min). Stop

solution (0.3 M HCl) was added and the plates were read at 450

and 590 nm on an automatic plate reader (Infinite M200; Tecan, Grodig,

Austria). The intra- and inter-assay coefficients of variation were 1.9 and

7.6, respectively. The limit of sensitivity of this assay is 0.2 ng ⁄ ml. For

those diabetic samples below the limit of detection, 0.2 ng ⁄ ml was used

for statistical analysis.

Cell death detection ELISA

This photometric enzyme immunoassay for the quantitative in vitro determi-

nation of cytoplasmic histone-associated DNA fragments (mono- and oligo-

nucleosomes) after induced cell death was carried out according to the

manufacturer’s instructions (Cell Death Detection ELISA; Roche Diagnostics)

and as previously described (5). Each sample was measured in duplicate in

each assay. Background measurements were made and this value was sub-

tracted from the mean value of each sample. This assay has a detection

limit of approximately 50 dead cells ⁄ well and results were normalised to

protein levels in each sample and are reported as relative levels of cell death

compared to controls. The inter- and intra-assay coefficients of variation

were 8.5% and 4.3%, respectively.

AIF induced death of astrocytes 1349

ª 2008 The Authors. Journal Compilation ª 2008 Blackwell Publishing Ltd, Journal of Neuroendocrinology, 20, 1348–1360

Caspase activity

Caspase 9, 8 and 3 ⁄ 7 activities were measured by using Caspase-Glo Assays

(Promega, Madison, WI, USA). These assays provide a luminogenic substrate

that, in the presence of the specific caspase in question, is cleaved to

release aminoluciferin, a substrate for luciferase. The luminescent signal pro-

duced is proportional to the amount of caspase activity present. Briefly,

10 lg of sample protein were added in a total volume of 25 ll of distilled

water in duplicate to a white-walled 96-well assay plate. Caspase-Glo

reagent (25 ll) was then added to each sample. The plate was then covered

and mixed gently at 300–500 r.p.m. for 30 s and incubated at room temper-

ature for 20–180 min. Luminescence was measured by using an Infinite

M200 automatic plate reader (Tecan).

Western blotting

Depending on the specific protein to be detected, either 20, 40 or 60 lg of

protein were resolved on a 12% SDS-polyacrylamide gel under denaturing

conditions. The proteins were then electro-transferred to poly(vinylidene di-

fluoride) membranes (Bio-Rad). Membranes were blocked in TBS (20 mM)

containing 5% nonfat dried milk or 5% bovine serum albumin (BSA) and

0.1% Tween 20 for 2 h. Primary antibodies, used at a concentration of

1 : 1000 unless otherwise stated, were: Anti-caspase 8, anti-caspase 7, anti-

caspase 2, anti-p53 and anti-Bax (1 : 5000), all purchased from Neomarkers

(Fremont, CA, USA). Anti-caspase 9, anti-caspase 3, anti-phosphorylated (p)-

Akt and anti-p-glycogen synthase kinase (GSK)3b (1 : 500) were from Cell

Signaling Technology (Beverly, MA, USA). Anti-caspase 12 was from Sigma

(St Louis, MO, USA), anti-caspase 6 from Medical & Biological Laboratories

(Woburn, MA, USA), anti-c-Jun and anti-pMAPK from Upstate Cell Signaling

Solutions (Lake Placid, NY, USA), anti-Bcl2alpha from Thermo Scientific (Fre-

mont, CA, USA), anti-PARP, anti-calpain-I and anti-AIF from Chemicon Inter-

national (Temecula, CA, USA). Anti-pJNK and anti-p-p38 (1 : 750) were from

Promega, anti-JNK1 and anti-p38 (1 : 500) from Santa Cruz Biotechnology

(Santa Cruz, CA, USA), anti-X-linked apoptosis inhibiting protein (XIAP) was

from BD Transduction Laboratories (Franklin Lakes, NJ, USA), anti-heat shock

protein (HSP)70 and HSP25 from Stressgen (Ann Arbor, MI, USA) and anti-

Bcl-2phosphoTyr129 was purchased from Labvision (Fremont, CA, USA). All

primary antibodies were incubated overnight at 4 �C under agitation. The

membranes were washed and incubated with the corresponding secondary

antibody conjugated with peroxidase (Pierce, Rockford, IL, USA). Bound per-

oxidase activity was visualised by chemiluminiscence (Perkin Elmer Life Sci-

ence, Boston, MA, USA) and quantified by densitometry using the BIO-1D

system (Vilber Lournat, Marne la Vallee, France). All results were first norma-

lised to actin levels in each lane (anti-actin 1 : 1000, Santa Cruz Biotechnol-

ogy) and then to control values on each blot. All experiments were

performed a minimum of two times.

Tissue sectioning

The brains were allowed to equilibrate in the cryostat chamber ()17 �C),

trimmed and embedded in OCT (Tissue-Tek, Elkhart, IN, USA). Coronal sec-

tions were cut at 20 lm throughout the entire arcuate nucleus and thaw-

mounted onto positively-charged slides. The slides were then stored at

)70 �C until immunohistochemistry was performed.

Double immunohistochemistry

After fixation in 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4),

sections were washed in phosphate buffer and then equilibrated in TBS

(0.1 M, pH 7.4) or phosphate-buffered saline (PBS; 0.1 M, pH 7.4) with 0.1%

Triton X-100 and 0.1% BSA for 20 min. The same buffer was used in all

subsequent washes. Sections were blocked for 90 min at room temperature

in TBS or PBS with 1% Triton X-100 and 3% BSA and then incubated for

48 h at 4 �C in a humid chamber with the primary antibodies. Antibodies

included anti-cleaved-caspase 3 antiserum (1 : 250; Cell Signaling Technol-

ogy), anti-pJNK (1 : 500), anti-Bax (1 : 500), anti-AIF (1 : 1000), anti-Bcl-

2alpha (1 : 500), anti-pBcl-2 (1 : 500) or anti-calpainI (1 : 250) combined

with either rabbit anti-GFAP (Neomarkers) or mouse anti-GFAP (Sigma) at a

dilution of 1 : 1000 or mouse anti-NeuN (Chemicon) at a dilution of

1 : 250. Sections were washed and incubated for 2 h under dark conditions

with the appropriate secondary antibodies that included Alexa Fluor 633

goat anti-mouse IgG (1 : 2000), Alexa Fluor 488 goat anti-mouse IgG

(1 : 1000), Alexa Fluor 633 goat anti-rabbit IgG (1 : 1000) and Alexa Fluor

488 goat anti-rabbit IgG (1 : 1000) from Molecular Probes (Leiden, Nether-

lands), and biotin conjugated goat anti-mouse IgG (1 : 1000) and biotin

conjugated goat anti-rabbit IgG (1 : 1000) from Pierce. The slides were

washed and incubated in the dark during 90 min with Alexa Fluor 633 con-

jugated streptavidin or Alexa Fluor 488 conjugated streptavidin (Molecular

Probes) at a dilution of 1 : 2000.

Preliminary assays were performed to determine the concentration of

antibodies to be used and specificity. For every experiment, sections for all

groups were incubated in parallel. In each assay, control slides consisted of

the omission of primary antibodies and verification that immunofluores-

cence was absent. Immunofluorescence was visualised by using a confocal

microscope (Leica model DMIRB; Leica, Wetzlar, Germany).

Terminal dUTP nick-end labelling (TUNEL) plusimmunohistochemistry

Cell death detection assays by TUNEL were performed following the manu-

facturer’s instructions (Promega). Briefly, sections were fixed in 4% parafor-

maldehyde in 0.1 M phosphate buffer (pH 7.4) for 20 min, washed three

times in buffer and incubated twice with PBS containing 0.3% BSA for

5 min. Sections were incubated in permeabilisation solution (PBS containing

1% Triton X-100) for 60 min and the labelling was performed with the ter-

minal deoxynucleotidyl transferase enzyme in a buffer containing fluores-

cein-12-dUTP for 60 min at 37 �C. Sections were washed, blocked with PBS

containing 3% BSA and 1% Triton X-100 for 60 min and incubated over-

night in a humid chamber at 4 �C with primary antibodies for mouse anti-

GFAP (1 : 500) or NeuN (1 : 250) in PBS containing 3% BSA and 1% Triton

X-100. After washing, the slides were incubated with Alexa Fluor goat anti-

mouse 633 in blocking buffer at a dilution of 1 : 1000. Finally, the slides

were again washed three times before mounting in glycerol. The results

were visualised by confocal microscopy.

Negative controls for TUNEL staining and immunohistochemistry were

performed by omitting terminal deoxynucleotidyl-transferase from the label-

ling mixture and the primary antibody, respectively, and resulted in no spe-

cific labelling.

Statistical analysis

All experiments were performed a minimum of two times. When a sample was

analysed more than once in separate assays (western blots) or repeated mea-

sures in the same assay (ELISA), the mean value was used for statistical analy-

sis; hence, ‘n’ represents the number of animals used in each group and no

pooling of samples was performed (n = 6 for diabetic groups and n = 8 for

controls unless otherwise stated). Testing for normality was performed by the

Lilliefor’s test. Bartlett’s test was used to determine that the groups have equal

variances. A one-way ANOVA followed by a Scheffe F-test was performed to

determine difference between experimental groups. P < 0.05 was considered

statistically significant. All results are reported as the mean � SEM. All wes-

tern blot results are reported as the percent of control value.

1350 C. Garcıa-Caceres et al.


Results

Blood glucose levels

Glucose levels remained significantly elevated in diabetic rats com-

pared to controls throughout the study (ANOVA: P < 0.0001) and no

significant differences were found between the diabetic groups. At

the time when the rats were killed, mean morning blood glucose

levels were: Control = 81.3 � 2.1 mg ⁄ dl; DB1W = 495.9 � 29.4

mg ⁄ dl; DB4W = 515.3 � 39.0 mg ⁄ dl; DB6W = 521.1 � 32.8 mg ⁄ dl;

DB8W = 549 � 23.6 mg ⁄ dl (n = 18 for each diabetic group, n = 24

for controls).

Insulin levels

In diabetic rats, insulin levels remained significantly reduced com-

pared to controls throughout the study (ANOVA: P < 0.01) and no sig-

nificant differences were found between the diabetic groups. At the

time when the rats were killed, mean morning blood insulin levels

were: Control = 1.89 � 0.23 ng ⁄ ml; DB1W = 0.43 � 0.04 ng ⁄ ml;

DB4W = 0.41 � 0.01 ng ⁄ ml; DB6W = 0.40 � 0.02 ng ⁄ ml; DB8W =

0.45 � 0.05 ng ⁄ ml (n = 18 for each diabetic group, n = 24 for

controls).

Cell death

There was a significant increase in cell death in the hypothalamus

of diabetic rats starting at 4 weeks of diabetes evolution [Con-

trol = 100 � 11 arbitrary units (AU); DB1W = 99 � 9 AU;

DB4W = 152 � 19 AU; DB6W = 160 � 8 AU; DB8W = 165 � 12

AU; ANOVA: P < 0.01 (n = 12 for each diabetic group, n = 16 for

controls)] as detected by ELISA, thereby confirming our previously

reported results (5).

At all stages of diabetes, the majority of TUNEL positive hypotha-

lamic cells were found in the arcuate nucleus. In control animals,

very few or no TUNEL positive cells were found. As demonstrated in

Fig. 1, TUNEL labelling found in the hypothalamus of DB4W rats

was associated with GFAP immunostained cells (Fig. 1A–C, solid

arrow). In the hypothalamus of DB8W rats, GFAP immunostaining

was also found in the majority of TUNEL labelled cells (Fig. 1D–F,

solid arrows). In addition, some TUNEL positive cells were not GFAP

positive (open arrow); however, these cells were few in number and

were not demonstrated to be NeuN positive (data not shown), sug-

gesting that they were not neurones.

Caspase activation

Although caspase 9 activation tended to increase at 1 week of dia-

betes and decrease at 4 weeks, these changes were not statistically

significant either by the Caspase-Glo assay (Fig. 2A) or western blot

analysis (data not shown).

Caspase 3 ⁄ 7 activation was significantly increased at 1 week of

diabetes and decreased thereafter as detected by the Caspase-Glo

assay (ANOVA: P < 0.05; Fig. 2B). In an attempt to differentiate

between activation of caspase 3 and 7, western blot analysis was

performed. There was no significant change in the levels of frag-

mented caspase 7 (Control = 100 � 10; DB1W = 113 � 21;

DB4W = 96 � 17; DB6W = 98 � 7; DB8W = 87 � 26). Western

blot analysis indicated that the fragmented form of caspase 3

increased after 1 week of diabetes, although this change was not

significant (Control = 100 � 8; DB1W = 141 � 21; DB4W =

106 � 9; DB6W = 117 � 17; DB8W = 116 � 17). These results

suggest that the more sensitive Caspase-Glo assay was most likely

detecting activation of caspase 3.

Immunohistochemistry for cleaved caspase 3 also suggested an

increase in the activation of this caspase in the hypothalamus after

1 week of diabetes. In control rats, immunoreactivity for cleaved cas-

pase 3 was low in the hypothalamus, with infrequent colocalisation

of GFAP and cleaved caspase 3 being found (Fig. 3A–C). Immuno-

reactivity for both GFAP and cleaved caspase 3 increased in the

TUNEL

TUNEL

GFAP

GFAP

Merge

Merge

DB4W

DB8W

(A) (B) (C)

(D) (E) (F)

DB4W

DB8W

DB4W

DB8W

Fig. 1. Terminal dUTP nick-end labelling (TUNEL) plus immunohistochemistry for glial fibrillary acidic protein (GFAP) in the hypothalamus of a rat diabetic for

4 weeks (DB4W; A–C) or for 8 weeks (DB8W; D–F). Cells that are positive for both TUNEL and GFAP are indicated by solid arrows. TUNEL positive cells that were

not GFAP positive were infrequently observed in DB8W rats (open arrow). Scale bar = 50 lm.



hypothalamus of DB1W rats, with colocalisation of immunoreactivity

for these two proteins also increasing (Fig. 3D–F). In diabetic rats,

cleaved caspase 3 labelling remained cytoplasmic. Projections origi-

nating at the base of the third ventricle were also immunopositive for

cleaved caspase 3. These projections were sparse in control rats and

did not colocalise with GFAP immunostaining (Fig. 3G–I, open arrow).

Cleaved caspase 3 immunolabelling of these fibres increased in DB1W

rats, but they remained GFAP negative (Fig. 3J–L, open arrows). Cells

lining the third ventricle were more highly immunoreactive for

cleaved caspase 3 in diabetic rats (Fig. 3P–R, asterisks) compared to

control rats (Fig. 3M–O, asterisks), with these cells being GFAP negative

in both experimental groups. Immunoreactive cleaved caspase 3 cells

were also negative for NeuN (data not shown).

Activation of caspase 8 was not detected in any of the experi-

mental groups by the Caspase Glo assay, nor was caspase 8 frag-

mentation detected by western blot analysis (data not shown).

There was no significant change in the activation levels of cas-

pase 6 (Control = 100 � 12; DB1W = 84 � 21; DB4W = 93 � 15;

DB6W = 88 � 11; DB8W = 91 � 29), caspase 12 (Control =

100 � 12; DB1W = 96 � 6; DB4W = 66 � 7; DB6W = 116 � 9;

DB8W = 67 � 15) or caspase 2 (Control = 100 � 8; DB1W =

138 � 23; DB4W = 110 � 12; DB6W = 95 � 20; DB8W = 130

� 29) as determined by western blot analysis.

Bcl family

Relative levels of the pro-apoptotic factor Bax increased signifi-

cantly at 1 week of diabetes compared to control rats and then

decreased (ANOVA: P < 0.0001; Fig. 4A). In the control group, Bax

immunostaining was not found to colocalise with GFAP. In DB1W

rats, Bax immunostaining increased, with the majority of Bax posi-

tive cells being GFAP negative, although an occasional cell immuno-

positive for both GFAP and Bax was found (data not shown). After

4 weeks of diabetes, when increased cell death could be detected,

relative Bax levels in the hypothalamus were not significantly

different from control rats (Fig. 4A) and differences in the cellular

localisation as detected by immunohistochemistry were not seen

(data not shown).

Bcl-2a protein levels increased at 4 weeks of diabetes, becoming

significant at 6 weeks and remaining so at 8 weeks (ANOVA:

P < 0.05; Fig. 4B). This anti-apoptotic protein was found in both

neurones and glia in all experimental groups (data not shown).

The anti-apoptotic properties of Bcl-2 can be blocked by phos-

phorylation (22); however, pBcl-2 levels did not change in response

to diabetes (Fig. 4C). Immunohistochemistry indicated that the

phosphorylated form of this protein was found predominantly in

GFAP positive cells in the hypothalamus of all experimental groups

(data not shown).

HSP and inhibitors of apoptosis

HSP can prevent both neuronal and glial cell death (23) and there

was an increase in HSP70 as early as 1 week after diabetes onset,

which gradually declined to control levels by 8 weeks of diabetes

(ANOVA P < 0.01; Fig. 4D). In both control and diabetic rats, HSP70

was localised mainly with NeuN positive cells, whereas colocalisa-

tion with GFAP positive cells was infrequent. Although HSP25 levels

tended to increase at 4 weeks of diabetes, this change was not sig-

nificant (Control = 100 � 12; DB1W = 97 � 20; DB4W = 172 �18; DB6W = 114 � 30; DB8W = 99 � 12).

There was no difference between any of the experimental groups

in the levels of XIAP (Control = 100 � 3; DB1W = 103 � 19;

DB4W = 116 � 12; DB6W = 103 � 2; DB8W = 125 � 13).

Phosphokinases

No significant change in the levels of pAkt (Control = 100 � 8;

DB1W = 119 � 8; DB4W = 94 � 5; DB6W = 112 � 22; DB8W =

119 � 15), pERK1 ⁄ 2 (Control = 100 � 11; DB1W = 89 � 18;

DB4W = 98 �11; DB6W = 117 � 8; DB8W = 109 � 7) or p-p38

levels (Control = 100 � 3; DB1W = 96 � 8; DB4W = 81 � 4;

DB6W = 88 � 7; DB8W = 106 � 14) were found in the hypothala-

mus at any stage of diabetes.

Cas

pase

3/7

act

ivat

ion

(% c

ontr

ol)

Ct DB1W DB4W DB6W DB8W0

100

200

300

400

(A)

(B)*

* *

Cas

pase

9 a

ctiv

atio

n (%

con

trol

)

0

50

100

150

200

Ct DB1W DB4W DB6W DB8W

NS

Fig. 2. (A) Caspase 9 activity in the hypothalamus of control and diabetic

rats as determined by a caspase activity assay. (B) Caspase 3 ⁄ 7 activity in

the hypothalamus of control and diabetic rats. There was a significant

increase at 1 week of diabetes, whereas, at 4 and 8 weeks of diabetes,

caspase 3 ⁄ 7 activity was significantly below control levels (ANOVA: P < 0.05 =

significant compared to control groups). NS, not significant; Ct, control;

DB1W, diabetic for 1 week; DB4W, diabetic for 4 weeks; DB6W, diabetic for

6 weeks; DB8W, diabetic for 8 weeks.



Cl- Casp 3 GFAP Merge




(A)

(D)

(G)

(J)

3V 3V 3V

3V 3V 3V

Control Control Control



DB1W DB1W DB1W

DB1W DB1W DB1W

DB1W DB1W DB1W

3V 3V 3V

3V 3V 3V

*

**

*

*

*

**

*

*

*

**

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*


Cl. Casp 3 GFAP Merge

(M)

(P)

(B)

(E)

(H)

(K)

(N)

(Q)

(C)

(F)

(I)

(L)

(O)

(R)

Fig. 3. Immunolabelling for cleaved caspase 3 (Cl-Casp3; green) was found to colocalise with glial fibrillary acidic protein (GFAP; red) in the arcuate nucleus

of both control groups (A–C) and rats that were diabetic for 1 week (DB1W; D–F) as indicated by the solid arrows. Cl-Casp3 labelled fibres (open arrows) were

seen radiating from the third ventricle (3V) in both control (G–I) and DB1W (J–L) rats. The labelling of these fibres clearly increased in DB1W rats. Cells lining

the 3V (asterisks) were also found to express Cl-Casp3 in both control (M–O) and DB1W (P–R) rats, with the intensity of this labelling increasing in diabetic rats.



There was an increase in the phosphorylation of JNK at 4 weeks

of diabetes, becoming significant at 6 weeks and remaining so at

8 weeks (ANOVA: P < 0.05; Fig. 4E). Although this kinase has been

implicated in the death of astrocytes in response to other insults,

its phosphorylated form was not detected in GFAP positive cells in

the hypothalamus (data not shown). No change in the levels of

Ct DB1W DB4W DB6W DB8W Ct DB1W DB4W DB6W DB8W



50

100

150

200

250

0

300(A) (B)

(C) (D)

(E) (F)

*

Bax

b-actin300

Bcl-2

a (%

con

trol

)

0

100

200

*

*

Bcl-2

b-actin

pBcl-2 Hsp70

Hsp

70 (%

con

trol

)

0

20

40

60

80

100

120

140

160

180

*

*

b-actinb-actin

0

Phos

pho-

Bcl-2

(% c

ontr

ol)

Bax

(% c

ontr

ol)

100

200

NS

pJNK

pJN

K (%

con

trol

)

b-actin

100

200

300

400

0

500

**

0

1

100

200

300

400

500

600 *

*

*

pGSK

3b (%

con

trol

)

b-actinpGSK3 b

50

150

250

Fig. 4. (A) Relative Bax protein levels in the hypothalamus of control and diabetic rats as determined by western blot. There was a significant increase at

1 week of diabetes compared to control levels (ANOVA: P < 0.0001). (B) Bcl-2a levels increased at 4 weeks of diabetes, becoming significantly elevated compared

to controls at 6 and 8 weeks of diabetes (ANOVA: P < 0.05). (C) Phosphorylated Bcl-2 levels did not differ significantly between the experimental groups. (D) In

the hypothalamus, relative levels of heat shock protein (HSP) 70 increased significantly at 1 week of diabetes, remaining so at 4 weeks and decreasing to

control levels at 8 weeks of diabetes (ANOVA: P < 0.01). (E) Relative levels of pJNK as detected by western blot analysis in the hypothalamus of control and

diabetic rats. There was a significant increase at 6 and 8 weeks of diabetes (ANOVA: P < 0.05). (F) Levels of phosphorylated glycogen synthase kinase (pGSK)3bincreased significantly at 4 weeks and remained increased 6 and 8 weeks after the onset of diabetes (ANOVA: P < 0.02). Representative blots of one of the

experiments for each protein are shown. NS, not significant; Ct, control; DB1W, diabetic for 1 week; DB4W, diabetic for 4 weeks; DB6W, diabetic for 6 weeks;

DB8W, diabetic for 8 weeks.



p-cJun, a downstream target of JNK, were found (Control =

100 � 11; DB1W = 77 � 12; DB4W = 84 � 3; DB6W = 82 � 10;

DB8W = 122 � 13).

GSK3b, p53 and PARP

There was an approximately fivefold increase in the levels of

pGSK3b at 4 weeks that remained significantly elevated at 6 and

8 weeks of diabetes (ANOVA: P < 0.02; Fig. 4F). The overall levels of

p53 were not significantly different between any of the experimen-

tal groups (Control = 100 � 8; DB1W = 83 � 25; DB4W =

121 � 27; DB6W = 83 � 5; DB8W = 94 � 3). In addition, no

phosphorylation of p53(SER15) could be detected in any of the

experimental groups.

Levels of the intact form of PARP increased at 4 weeks of diabe-

tes, becoming significantly elevated after 6 weeks of diabetes and

remained elevated at 8 weeks of diabetes (ANOVA: P < 0.05; Fig. 5A).

Levels of fragmented PARP decreased significantly at 1 week of dia-

betes and then increased significantly at 4 weeks of diabetes (ANOVA:

P < 0.05; Fig. 5B).

Factors involved in non-caspase dependent cell death

Mean calpain-I levels were significantly reduced in the hypothala-

mus at 1 week of diabetes and then returned to control levels for

the remainder of the study (Control = 100 � 17; DB1W = 49 � 9;

DB4W = 105 � 17; DB6W = 99 � 19; DB8W = 98 � 24; ANOVA:

P < 0.05, n = 4 in each experimental group). No change in intracel-

lular location of this protein was discerned by immunohistochemis-

try at any time throughout the study (data not shown).

No significant change in the levels of apoptosis inducing fac-

tor (AIF) was detected in response to diabetes (Fig. 5C). However,

AIF

(% c

ontr

ol)

PARP/116b-actin

0

50

100

150

200PARP/85

b-actin

*

*

PARP

116

(% c

ontr

ol)

0

50

25

125

150

100

225

(A) (B)

(C)

200

*

PARP

85 (%

con

trol

)

Ct DB1W DB4W DB6W

0

50

100

150

AIF

b-actin

NS

Ct DB1W DB4W DB6W DB8W

DB8W Ct DB1W DB4W DB6W DB8W

Fig. 5. (A) The relative levels of the proform of pol (ADP-ribose) polymerase (PARP) were significantly increased 6 weeks after the onset of diabetes (ANOVA:

P < 0.05). (B) The 85KD fragmented form of PARP were significantly decreased at 1 week of diabetes and then increased significantly at 4 weeks of diabetes

(ANOVA: P < 0.05). (C) Relative levels of apopotosis inducing factor (AIF) in the hypothalamus did not differ between the experimental groups. Representative

blots of one of the experiments for each protein are shown. NS, not significant; Ct, control; DB1W, diabetic for 1 week; DB4W, diabetic for 4 weeks; DB6W,

diabetic for 6 weeks; DB8W, diabetic for 8 weeks.



the intracellular localisation of this protein differed between

experimental groups. In the hypothalamus of control rats, AIF

was not found to be translocated to the nucleus of GFAP posi-

tive cells (Fig. 6C,G). In rats that were diabetic for only 1 week,

AIF was not found in the nucleus of GFAP positive cells

(Fig. 6H). After 4, 6 and 8 weeks of diabetes, AIF was found in

the nucleus of numerous GFAP positive cells, as well as non-

nuclear parts in others (Fig. 6 I, J, F, K, respectively). Cells with AIF

translocated to the nucleus were mainly localised to the arcuate

nucleus.

AIF immunolabelling was also found associated with TUNEL posi-

tive cells; although not all TUNEL positive cells were AIF positive

(Fig. 6L–M).

Discussion

In diabetes mellitus, apoptotic cell death occurs in numerous tissues

(1, 2, 4–6, 14, 18, 19) with increased glucose levels (24, 25),

decreased insulin or insulin-like growth factor signalling (26) or an

increase in cytokines such as TNFa (27) being the triggering factor.

Astrocytes in the arcuate nucleus are activated by changes in glu-

cose concentrations (28) and are involved in glucose sensing at the

brain level (9). Furthermore, glial cells are reported to control the

supply of glucose and its metabolites to neurones (9), indicating

that they are the first line of defence against changes in glucose

concentrations. Indeed, when glucose availability is reduced, glyco-

gen stored in astrocytes serves as a fuel source for neurones (29).

Control

AIF

DB

DB8W

AIF

GFAP

GFAP AIF GFAP

AIF GFAP

AIF GFAP DB8WControl DB6WDB4W

(A) (B) (C)

(D) (E) (F)

(G) (K)

DB1W AIF GFAP AIF GFAP AIF GFAP AIF GFAP

TUNELDB6W AIFDB6W AIF TUNELDB6W

(H) (I) (J)

(L) (M) (N)

Fig. 6. Immunoreactivity for AIF (green) was not found in the nucleus of glial fibrillary acidic protein (GFAP; red) positive cells in the arcuate nucleus of con-

trol rats (A–C, G). In the arcuate nucleus of rats after 8 weeks of diabetes, AIF immunolabelling was observed to be non-nuclear in some GFAP positive cells

and nuclear in others (D–F, K). In DB1W rats, AIF was found to be mainly non-nuclear in GFAP positive cells (H). In DB4W (I) and DB6W (J) rats, AIF was non-

nuclear in some GFAP positive cells and translocated to the nucleus of others. Solid arrows indicate GFAP positive cells with non-nuclear localisation of AIF.

Open arrows indicate GFAP positive cells with AIF localised to the nucleus. AIF (red) colocalization in terminal dUTP nick-end labelled (TUNEL, green) positive

cells. DB1W, diabetic for 1 week; DB4W, diabetic for 4 weeks; DB6W, diabetic for 6 weeks; DB8W, diabetic for 8 weeks. Scale bar = 30 lm.



Thus, it is possible that prolonged exposure to elevated glucose lev-

els underlies the increased death of hypothalamic astrocytes.

Indeed, high glucose levels have been shown to directly induce

death of different glial cell types (30, 31). However, other circulat-

ing factors, such as increased TNFa (18), decreased insulin (19) or

insulin growth factor (IGF)-I (32), or decreased central IGF-I as well

as its receptor (L. M. Frago, J. A. Chowen; unpublished observation),

could also be involved.

Activation of caspases 9 and 3 (14, 25, 33), as well as caspase 6

(14, 34) and 8 (18, 33, 34) have been implicated in diabetes induced

cell death. However, these classical apoptotic pathways cannot

explain the death of astrocytes observed in the present study

because no activation of caspases 8, 2, 12 or 6 was detected.

Changes in caspase 9 activity temporally paralleled caspase 3 ⁄ 7activity, but did not reach statistical significance. Although the

enzymatic method employed to detect caspase activity did not dis-

tinguish between caspases 3 and 7, western blot analysis suggested

that caspase 3 was activated, as no fragmentation of caspase 7

was detected, whereas changes in fragmentation of caspase 3

expressed the same temporal pattern as caspase activity. This was

further supported by our immunohistochemical data.

Caspase 3 activity increased 1 week after diabetes onset, when

no increase in cell death was observed, and then decreased below

control levels at 4 weeks of diabetes, when cell death increased. The

lack of temporal coordination and also the absence of nuclear local-

isation of cleaved caspase 3 suggest that this caspase is most likely

not inducing the death of astrocytes. In control rats, cleaved cas-

pase 3 was found in the cytoplasm of some GFAP positive cells in

the arcuate nucleus, in GFAP negative cellular projections originat-

ing at the base of the third ventricle, as well as in GFAP negative

cell bodies lining the third ventricle. After 1 week of diabetes,

immunoreactivity of cleaved caspase 3 increased in all of these cel-

lular locations. Caspase 3 is implicated in functions other than

induction of cell death, including cleavage of substrates involved in

cell proliferation and migration (35, 36), or cellular plasticity

through cleavage of cytoskeletal components, including actin,

vimentin and GFAP (37–39). Indeed, cleaved caspase 3 is expressed

in astrocytes in the normal adult brain where it is not associated

with cell death (40, 41). In the normal adult cerebellum, caspase 3 is

suggested to participate in the differentiation or proliferation of

Bergman glia (42), although, in diabetic rats, the proliferation of

hypothalamic astrocytes is reduced (5). Caspase 3 mediated cleavage

of GFAP has also been associated with turnover of GFAP in reactive

astrocytes (38, 39). In our experimental paradigm, not only do GFAP

protein levels decline in the hypothalamus, but also the mean num-

ber of GFAP positive projections per astrocyte in the arcuate nucleus

also decreases significantly (5). Whether caspase 3 activation is

involved in this cytoskeletal reorganisation in hypothalamic astro-

cytes remains to be demonstrated. However, because cytoplasmic

caspase 3 can also act an as initiator caspase, activating down-

stream apoptosis inducing factors that are then translocated to the

nucleus, participation of this increase in caspase 3 activation at

1 week of diabetes in cell death processes cannot be ruled out.

Proteins of the Bcl-2 family were also modified in response to

diabetes. The anti-apoptotic protein Bcl-2a increased at 4 weeks

and remained significantly increased at 6 and 8 weeks of diabetes,

precisely when cell death increased, and was found in both neuro-

nes and astrocytes. However, phosphorylation of Bcl-2, which can

inhibit its anti-apoptotic properties (22), was found mainly in astro-

cytes. Hence, although overall levels of phosphorylated Bcl-2 did

not change in response to diabetes, its inactivation in some astro-

cytes could render them more susceptible to further assaults and

possible death. Indeed, streptozotocin induced diabetes results in

astrocytes being more susceptible to death in response to ischaemia

(43).

Overall, levels of the pro-apoptotic protein Bax increased signifi-

cantly at 1 week of diabetes and were found mainly in neurones.

As neuronal death was not observed at this time, translocation of

this protein to the mitochondrial membrane and its apoptotic

actions must have been impeded. One possibility is inhibition of the

intrinsic cell death pathway by HSP (44). Indeed, changes in the

levels of HSP70 paralleled those of Bax, increasing significantly at

1 week of diabetes and declining gradually thereafter. Furthermore,

HSP70 was found mainly in neurones, with very few GFAP positive

cells expressing this protein in control or diabetic rats. By contrast,

XIAP, which can inhibit both the intrinsic and extrinsic cell death

pathways, was not modified.

The ubiquitous serine ⁄ threonine kinase GSK3b is constitutively

active in resting cells, with its actions being tightly controlled

through inactivation by phosphorylation. When active, GSK3b can

phosphorylate members of the Bcl-2 family and other mitochon-

drial proteins to promote cell death (45). Phosphorylation of GSK3bincreased significantly in the hypothalamus at 4 weeks of diabetes,

in accordance with changes in other mitochondrial proteins moving

towards an anti-apoptotic balance. Taken together, the data

obtained in the present study indicate that the intrinsic cell death

pathway is not involved in the increased cell death in the hypothal-

amus and that, indeed, anti-apoptotic processes are activated that

could explain the lack of neuronal death at this time.

Activation of distinct MAP kinases involved in cell survival or

induction of apoptosis were analysed to further understand the

intracellular changes occurring in this experimental paradigm. No

change in the activation of Akt or ERKs was observed. These pro-

teins are generally involved in cell survival and proliferative pro-

cesses in both neurones and glia (46), although ERKs have also

been associated with induction of glial death (47, 48). Similarly, no

modification was found in the activation of p38, a kinase involved

in both neuronal (49) and glial (50) cell death. JNK has been associ-

ated with both induction and protection (46, 48, 50) of astrocytic

cell death, as well as glial induced neurone cell death (51). A signif-

icant increase in phosphorylation of JNK occurred at 4 weeks of

diabetes, coincident with the increase in cell death, but was found

almost exclusively in non-GFAP positive cells. Indeed, phosphoryla-

tion of JNK occurs in neurones in response to a decrease in insulin

(52) and participates in neuronal protection in diabetes and in

response to glucose ⁄ oxidative stress (53). In addition, JNK has

recently been shown to modulate mitochondrial bioenergetics

through phosphorylation of pyruvate dehydrogenase (54). Thus, the

observed activation of JNK could be involved in changes in neuro-

nal metabolism and protection from the diabetic state.



Nuclear translocation of AIF is involved in caspase-independent

cell death in response to a variety of signals, including oxidative

stress, glutamate toxicity, and ischaemia (21). Loss of mitochon-

drial membrane integrity results in the release of this factor,

which can then be translocated to the nucleus where it induces

DNA fragmentation and chromatin condensation (21). Some stud-

ies suggest that this mechanism may be especially involved in cell

death in the adult brain (55). Nuclear translocation of AIF has

also been shown to induce caspase-independent cell death in reti-

nal neurones in response to high glucose levels (56), suggesting

that this mechanism may be involved in diabetes induced cell

death. Although we found no overall increase in AIF levels,

increased nuclear localisation of this factor was observed at the

same experimental time points where increased astrocytic death

was found. Furthermore, AIF was found to colocalise in most,

although not all, TUNEL positive cells, indicating that it is most

likely one of the mechanisms involved in diabetes induced cell

death in the hypothalamus.

Elucidation of the molecular events involved in the release of

AIF from the mitochondria remains an active area of investigation,

with many aspects of this process yet to be determined. Activa-

tion of p53 and its downstream proteins such as Bax (21) have

been implicated in AIF translocation from the mitochondria; how-

ever, we found no activation of p53 and Bax was not found to

be increased when AIF translocation was seen. Calpain I has

recently been shown to be involved in AIF translocation in ischae-

mia induced neuronal cell death (57), but in rat hypothalamus

calpain I was actually decreased after 1 week of diabetes, return-

ing to control levels thereafter. Furthermore, no changes in its

subcellular distribution were detected. Under basal conditions,

PARP1 is involved in the detection of DNA damage and repair

(58). However, in conditions of severe cellular damage, increased

PARP activation can result in cellular energy depletion and thus

lead to increased cell death, and one mechanism by which PARP

induces cell death is through liberation of AIF (21). In the hypo-

thalamus of diabetic rats, an increase in overall PARP1 levels, as

well as increased fragmentation of PARP, was coincident with the

increase in cell death. Furthermore, when PARP levels were

increased, translocation of AIF to the nucleus was found in hypo-

thalamic astrocytes. Together, these observations suggest that

PARP induced translocation of AIF may be involved in the

increased death of astrocytes in the hypothalamus of poorly con-

trolled diabetic rats. Indeed, primary astrocyte cultures have been

used to demonstrate that AIF-mediated cell death in astrocytes

involves PARP1 activation and requires NAD+ depletion, although

how PARP1 activation is associated with AIF release remains to

be determined (59).

The results reported in the present study demonstrate that

numerous intracellular signalling mechanisms are modulated in a

time specific manner in the hypothalamus in response to poorly

controlled diabetes and that many of these changes are cell type

specific (Table 1). The original aim of the study was to determine

the intracellular mechanism underlying the increased death of as-

trocytes. We report that this process most likely involves nuclear

translocation of AIF, although other mechanisms may also be

involved. In addition, we demonstrate that intracellular mechanisms

involved with cell protection are also activated and that many of

these processes are neurone specific, which is congruent with the

delay in neuronal death.

Acknowledgements

This work was funded by grants from Fondo de Investigacion Sanitaria

(PI040817, PI051268 and PI070182), CIBER Fisiopatologıa de Obesidad y

Nutricion (CIBEROBN) Instituto de Salud Carlos III and Fundacion de Endo-

crinologıa y Nutricion. C.G.-C. is supported by a predoctoral fellowship from

the Ministerio de Educacion y Ciencia, A.M.L.-S. was supported by a post-

doctoral fellowship from Fondo de Investigacion Sanitaria. J.A.C. is supported

by the biomedical investigation programme of the Consejerıa de Sanidad y

Consumo de la Comunidad de Madrid. The authors would like to thank San-

dra Canelles and Francisca Dıaz for the excellent technical support.

Received 30 May 2008,

revised 11 August 2008,

accepted 26 August 2008

Table 1. Summary of the Major Changes in Pro- and Anti-Apoptotic Proteins Observed in the Hypothalamus of Diabetic Rats Compared to Control Rats.

DB1 DB4 DB6 DB8 Cellular localisation

Cell death M › › › Astrocytes, mainly arcuate nucleus

Caspase 3 ⁄ 7 activation (pro-apoptotic) › fl M fl Astrocytes (cytoplasmic), projections from and cells lining third ventricle

Bcl-2a (anti-apoptotic) M › › › Neurones and astrocytes

pBcl-2a (pro-apoptotic) M M M M Astrocytes

Bax (pro-apoptotic) › M M M Mainly neurones, few astrocytes

Hsp70 (anti-apoptotic) › › M M Mainly neurones, few astrocytes

pJNK (can be either) M M › › Mainly neurones, few astrocytes

pGSKb (anti-apoptotic) M › › › ND

Calpain 1 (pro-apoptotic) fl M M M Cytoplamic localisation

PARP116 (pro-apoptotic) M M › M ND

PARP85 (pro-apoptotic) fl › M M ND

AIF nuclear translocation (pro-apoptotic) M › › › Astrocytes

DB1W, diabetic for 1 week; DB4W, diabetic for 4 weeks; DB6W, diabetic for 6 weeks; DB8W, diabetic for 8 weeks; ND, not done.



References

1 Nishikawa T, Araki E. Impact of mitochondrial ROS production in the

pathogenesis of diabetes mellitus and its complications. Antioxid Redox

Signal 2007; 9: 343–353.

2 Allen DA, Yaqoob MM, Harwood SM. Mechanisms of high glucose-

induced apoptosis and its relationship to diabetic complications. J Nutr

Biochem 2005; 16: 705–713.

3 Steger RW, Rabe MM. The effect of diabetes mellitus on endocrine and

reproductive function. Proc Soc Exp Biol Med 1997; 214: 1–11.

4 Arroba AI, Frago LM, Paneda C, Argente J, Chowen JA. The number of

lactotrophs is reduced in the anterior pituitary of streptozotocin-induced

diabetic rats. Diabetologıa 2003; 465: 634–638.

5 Lechuga-Sancho AM, Arroba AI, Frago LM, Garcıa-Caceres C, Delgado

Rubin de Celix A, Argente J, Chowen JA. Reduction in the number of as-

trocytes and their projections is associated with increased synaptic pro-

tein density in the hypothalamus of poorly controlled diabetic rats.

Endocrinology 2006; 147: 5314–5324.

6 Klein JP, Hains BC, Craner MJ, Black JA, Waxman SG. Apoptosis of vaso-

pressinergic hypothalamic neurons in chronic diabetes mellitus. Neurobi-

ol Dis 2004; 15: 221–228.

7 Garcia-Segura LM, McCarthy MM. Minireview: role of glia in neuroendo-

crine function. Endocrinology 2004; 145: 1082–1086.

8 Walz W. Role of astrocytes in the clearance of excess extracellular

potassium. Neurochem Int 2000; 36: 291–300.

9 Pellerin L, Bouzier-Sore AK, Aubert A, Serres S, Merle M, Costalat R,

Magistretti PJ. Activity-dependent regulation of energy metabolism by

astrocytes: an update. Glia 2007; 55: 1251–1262.

10 Azcoitia I, Garcia-Ovejero D, Chowen JA, Garcia-Segura LM. Astroglia

play a key role in the neuroprotective actions of estrogen. Prog Brain

Res 2001; 132: 469–478.

11 Ullian EM, Sapperstein SK, Christopherson KS, Barres BA. Control of syn-

apse number by glia. Science 2001; 291: 657–660.

12 Theodosis DT, Trailin A, Poulain DA. Remodeling of astrocytes, a prere-

quisite for synapse turnover in the adult brain? Insights from the oxyto-

cin system of the hypothalamus. Am J Physiol Regul Integr Comp

Physiol 2006; 290: R1175–R1182.

13 Pinto S, Roseberry AG, Liu H, Diano S, Shanabrough M, Cai X, Friedman

JM, Horvath TL. Rapid rewiring of arcuate nucleus feeding circuits by

leptin. Science 2004; 304: 110–115.

14 Lechuga-Sancho AM, Arroba AI, Frago LM, Paneda C, Garcıa-Caceres C,

Delgado Rubın de Celix A, Argente J, Chowen JA. Activation of the

intrinsic cell death pathway, increased apoptosis and modulation of as-

trocytes in the cerebellum of diabetic rats. Neurobiol Dis 2006; 23: 290–

299.

15 Green DR, Droemer G. The pathophysiology of mitochondrial cell death.

Science 2004; 305: 626–629.

16 Wajant H. Death receptors. Essays Biochem 2003; 39: 53–71.

17 Krammer PE. CD95’s deadly mission in the immune system. Nature

2000; 407: 789–795.

18 Arroba AI, Frago LM, Argente J, Chowen JA. Activation of caspase 8 in

the pituitaries of streptozotocin-induced diabetic rats: implication in

increased apoptosis of lactotrophs. Endocrinology 2005; 146: 4417–

4424.

19 Arroba AI, Lechuga-Sancho AM, Frago LM, Argente J, Chowen JA. Cell-

specific expression of X-linked inhibitor of apoptosis in the anterior pitu-

itary of streptozotocin-induced diabetic rats. J Endocrinol 2007; 192:

215–227.

20 Hetz CA, Torres V, Quest AF. Beyond apoptosis: nonapoptotic cell death

in physiology and disease. Biochem Cell Biol 2005; 83: 579–588.

21 Krantic S, Mechawar N, Reix S, Quirion R. Apoptosis-inducing factor: a

matter of neuron life and death. Prog Neurobiol 2007; 81: 179–196.

22 Srivastava RK, Srivastava AR, Korsmeyer SJ, Nesterova M, Cho-Chung YS,

Longo DL. Involvement of microtubules in the regulation of Bcl2 phos-

phorylation and apoptosis through cyclic AMP-dependent protein kinase.

Mol Cell Biol 1998; 18: 3509–3517.

23 Yenari MA, Liu J, Zheng Z, Vexler ZS, Lee JE, Giffard RG. Antiapoptotic

and anti-inflammatory mechanisms of heat-shock protein protection.

Ann NY Acad Sci 2005; 1053: 74–83.

24 Romero G, Liu WH, Asnaghi V, Kern TS, Lorenzi M. Activation of

nuclear factor-kappaB induced by diabetes and high glucose regulates

a proapoptotic program in retinal pericytes. Diabetes 2002; 51: 2241–

2248.

25 Anitha M, Gondha C, Sutliff R, Parsadanian A, Mwangi S, Sitaraman SV,

Srinivasan S. GDNF rescues hyperglycemia-induced diabetic enteric neu-

ropathy through activation of the PI3K ⁄ Akt pathway. J Clin Invest 2006;

116: 344–356.

26 Ishii DN. Implication of insulin-like growth factors in the pathogenesis

of diabetic neuropathy. Brain Res Brain Res Rev 1995; 20: 47–67.

27 Chen G, Goeddel DV. TNF-R1 signaling: a beautiful pathway. Science

2002; 296: 1634–1635.

28 Guillod-Maximin E, Lorsignol A, Alquier T, Penicaud L. Acute intracarotid

glucose injection towards the brain induces specific c-fos activation in

hypothalamic nuclei: involvement of astrocytes in cerebral glucose-sens-

ing in rats. J Neuroendocrinol 2004; 16: 464–471.

29 Brown AM, Ransom BR. Astrocyte glycogen and brain energy metabo-

lism. Glia 2007; 55: 1263–1271.

30 Xi X, Gao L, Atala DA, Smith DG, Codispoti MC, Gong B, Kern TS, Zhang

JZ. Chronically elevated glucose-induced apoptosis is mediated by inacti-

vation of Akt in cultured Muller cells. Biochem Biophys Res Commun

2005; 326: 548–553.

31 Delaney CL, Russell JW, Cheng HL, Feldman EL. Insulin-like growth

factor-I and over-expression of Bcl-xL prevent glucose-mediated

apoptosis in Schwann cells. J Neuropathol Exp Neurol 2001; 60:

147–160.

32 Busiguina S, Chowen JA, Argente J, Torres-Aleman I. Specific alterations

of the insulin-like growth factor I system in the cerebellum of diabetic

rats. Endocrinology 1996; 137: 4980–4987.

33 Al-Mashat HA, Kandru S, Liu R, Behl Y, Desta T, Graves DT. Diabetes

enhances mRNA levels of proapoptotic genes and caspase activity,

which contribute to impaired healing. Diabetes 2006; 55: 487–495.

34 Bojunga J, Nowak D, Mitrou PS, Hoelzer D, Zeuzem S, Chow KU. Antioxi-

dative treatment prevents activation of death-receptor- and mitochon-

drion-dependent apoptosis in the hearts of diabetic rats. Diabetologia

2004; 47: 2072–2080.

35 Yan XX, Najbauer J, Woo CC, Dashtipour K, Ribak CE, Leon M. Expression

of active caspase-3 in mitotic and postmitotic cells of the rat forebrain.

J Comp Neurol 2001; 433: 4–22.

36 Kuranaga E, Miura M. Nonapoptotic functions of caspases: caspases as

regulatory molecules for immunity and cell-fate determination. Trends

Cell Biol 2007; 17: 135–144.

37 Morishima N. Changes in nuclear morphology during apoptosis correlate

with vimentin cleavage by different caspases located either upstream or

downstream of Bcl-2 action. Genes Cells 1999; 4: 401–414.

38 Mouser PE, Head E, Ha KH, Rohn TT. Caspase-mediated cleavage of glial

fibrillary acidic protein within degenerating astrocytes of the Alzheimer’s

disease brain. Am J Pathol 2006; 168: 936–946.

39 Acarin L, Villapol S, Faiz M, Rohn TT, Castellano B, Gonzalez B. Caspase-

3 activation in astrocytes following postnatal excitotoxic damage corre-

lates with cytoskeletal remodeling but not with cell death or prolifera-

tion. Glia 2007; 55: 954–965.

40 Noyan-Ashraf MH, Brandizzi F, Juurlink BH. Constitutive nuclear localiza-

tion of activated caspase 3 in subpopulations of the astroglial family of

cells. Glia 2005; 49: 588–593.



41 Delgado-Rubın del Celix A, Chowen JA, Argente J, Frago LM. Growth

hormone releasing peptide-6 acts as a survival factor in glutamate-

induced excitotoxicity. J Neurochem 2006; 99: 839–849.

42 Oomman S, Strahlendorf H, Finckbone V, Strahlendorf J. Non-lethal

active caspase-3 expression in Bergmann glia of postnatal rat cerebel-

lum. Brain Res Dev Brain Res 2005; 160: 130–145.

43 Muranyi M, Ding C, He Q, Lin Y, Li PA. Streptozotocin-induced diabetes

causes astrocyte death after ischemia and reperfusion injury. Diabetes

2006; 55: 349–355.

44 Stankiewicz AR, Lachapelle G, Foo CP, Radicioni SM, Mosser DD. Hsp70

inhibits heat-induced apoptosis upstream of mitochondria by preventing

Bax translocation. J Biol Chem 2005; 280: 38729–38739.

45 Beurel E, Jope RS. The paradoxical pro- and anti-apoptotic actions of

GSK3 in the intrinsic and extrinsic apoptosis signaling pathways. Prog

Neurobiol 2006; 79: 173–189.

46 Ciccarelli R, D¢Alimonte I, Ballerini P, D¢Auro M, Nargi E, Buccella S,

Di Iori P, Bruno V, Nicoletti F, Caciagli F. Molecular signalling mediat-

ing the protective effect of A1 adenosine and mGlu3 metabotropic

glutamate receptor activation against apoptosis by oxygen ⁄ glucose

deprivation in cultured astrocytes. Mol Pharmacol 2007; 71: 1369–

1380.

47 Shinozaki Y, Koizumi S, Ohno Y, Nagao T, Inoue K. Extracellular ATP

counteracts the ERK1 ⁄ 2-mediated death-promoting signaling cascades

in astrocytes. Glia 2006; 54: 606–618.

48 Kawasaki T, Kitao T, Nakagawa K, Fujisaki H, Takegawa Y, Koda K, Ago Y,

Baba A, Matsuda T. Nitric oxide-induced apoptosis in cultured rat astro-

cytes: protection by edaravone, a radical scavenger. Glia 2007; 55:

1325–1333.

49 Redman PT, He K, Hartnett KA, Jefferson BS, Hu L, Rosenberg PA, Levitan

ES, Aizenman E. Apoptotic surge of potassium currents is mediated by

p38 phosphorylation of Kv2.1. Proc Natl Acad Sci USA 2007; 104:

3568–3573.

50 Wang Y, Luo W, Stricker R, Reiser G. Protease-activated receptor-1 pro-

tects rat astrocytes from apoptotic cell death via JNK-mediated release

of the chemokine GRO ⁄ CINC-1. J Neurochem 2006; 98: 1046–1060.

51 Xie Z, Smith CJ, Van Eldik LJ. Activated glia induce neuron death via

MAP kinase signaling pathways involving JNK and p38. Glia 2004; 45:

170–179.

52 Schechter R, Beju D, Miller KE. The effect of insulin deficiency on tau

and neurofilament in the insulin knockout mouse. Biochem Biophys Res

Commun 2005; 334: 979–986.

53 Price SA, Onusom L, Purves-Tyson TD, Fernyhough P, Tomlinson DR. Acti-

vation of JNK in sensory neurons protects against sensory neuron cell

death in diabetes and on exposure to glucose ⁄ oxidative stress in vitro.

Ann NY Acad Sci 2003; 1010: 95–99.

54 Zhou Q, Lam PY, Han D, Cadenas E. c-Jun N-terminal kinase regulates

mitochondrial bioenergetics by modulating pyruvate dehydrogenase

activity in primary cortical neurons. J Neurochem 2008; 104: 325–335.

55 Cao G, Clark RS, Pei W, Tin W, Zhang F, Sun FY, Grahm SH, Chen J.

Translocation of apoptosis-inducing factor in vulnerable neurons after

transient cerebral ishcemia and neuronal cultures after oxygen-glucose

deprivation. J Cereb Blood Flow Metab 2003; 23: 1137–1150.

56 Santiago AR, Cristovao AJ, Santos PF, Carvalho CM, Ambrosio AF. High

glucose induces caspase-independent cell death in retinal neural cells.

Neurobiol Dis 2007; 25: 464–472.

57 Cao G, Xing J, Xiao X, Liou AKF, Gao Y, Yin X-M, Clark RSB, Graham SH,

Chen J. Critical role of calpain I in mitochondrial release of apoptosis-

inducing factor in ischemic neuronal injury. J Neurosci 2007; 27: 9278–

9293.

58 Kauppinen TM. Multiple roles for poly(ADP-ribose)polymerase-1 in neu-

rological disease. Neurochem Int 2007; 50: 954–958.

59 Alano CC, Ying W, Swanson RA. Poly(ADP-ribose) polymerase-1-mediated

cell death in astrocytes requires NAD+ depletion and mitochondrial per-

meability transition. J Biol Chem 2004; 279: 18895–18902.

