Vascular Damage in a Mouse Model of Diabetic
Retinopathy: Relation to Neuronal and Glial Changes
Rachel A. Feit-Leichman,1 Reiko Kinouchi,1,2 Masumi Takeda,1,2 Zhigang Fan,1
Susanne Mohr,3 Timothy S. Kern,3,4 and Dong Feng Chen1,4
PURPOSE. Lack of information about the development of dia-
betic retinopathy in mice has greatly hindered the use of
genetic mouse models for the study of disease mechanisms and
the development of therapeutic strategies. The objective of
this study was to characterize the occurrence and pathologic
progression of diabetic retinopathy in C57Bl/6J mice.
METHODS. Diabetes was induced with five consecutive injec-
tions of streptozotocin (STZ). The retinas were collected at
different time points (2 weeks to 22 months) after the induc-
tion of diabetes and examined by using molecular, histologic,
and immunohistochemical techniques and morphometric anal-
ysis.
RESULTS. There was transient induction of cell apoptosis and
caspase-3 activation in retinal neurons of C57Bl/6 mice within
days of diabetes induction. Glial fibrillary acidic protein
(GFAP), a marker of glial activation, likewise was transiently
upregulated, seemingly in astrocytes but not in Müller cells.
These abnormalities quickly returned to normal; ultimately, no
detectable loss of retinal ganglion cells (RGCs) was noted by
any of three independent methods (number of cells in ganglion
cell layer of retinal cross-sections, retrograde labeling of retinal
ganglion cells with fluorescent dye, or TUNEL staining) after up
to a 1-year duration of diabetes. Despite this apparent lack of
evidence for progressive damage in neurons and glial cells,
diabetic mice developed vascular disease characteristic of the
early stage of diabetic retinopathy beginning at 6 months after
the onset of disease. The vascular damage—formation of acel-
lular capillaries and pericyte ghosts— continued to increase
through the 18 months examined.
From the 1Schepens Eye Research Institute and Department of
Ophthalmology, Harvard Medical School, Boston, Massachusetts; the
2
Department of Ophthalmology, Asahikawa Medical College, Asa-
hikawa, Japan; and the 3Department of Medicine and Ophthalmology,
Center for Diabetes Research, Case Western Reserve University, Cleve-
land, Ohio.
4
Contributed equally to the work and therefore should be consid-
ered equivalent authors.
Supported by National Eye Institute Grants EY012983 (DFC),
EY00300 (TSK), and EY014380 (SM); the Massachusetts Lion’s Eye
Research Fund, the Kristin C. Dietrich Diabetes Research Award; the
Animal Models of Diabetes Complications Consortium (AMDCC;
funded by the National Institute of Health); Juvenile Diabetes Research
Foundation Grant CDA-2-2000-390 (SM); and the Juvenile Diabetes
Research Foundation Center for Diabetic Retinopathy at the Schepens
Eye Research Institute.
Submitted for publication November 22, 2004; revised June 22,
2005; accepted September 16, 2005.
Disclosure: R.A. Feit-Leichman, None; R. Kinouchi, None; M.
Takeda, None; Z. Fan, None; S. Mohr, None; T.S. Kern, None; D.F.
Chen, None
The publication costs of this article were defrayed in part by page
charge payment. This article must therefore be marked “advertise-
ment” in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Corresponding author: Dong Feng Chen, Schepens Eye Research
Institute, Harvard Medical School, 20 Staniford Street, Boston, MA
02114; dfchen@vision.eri.harvard.edu.
Investigative Ophthalmology & Visual Science, November 2005, Vol. 46, No. 11
Copyright © Association for Research in Vision and Ophthalmology
CONCLUSIONS. Diabetic C57Bl/6J mice develop capillary lesion
that are characteristic of the early stages of diabetic retinopathy
in patients. The data suggest that diabetes-induced degenera-
tion of retinal capillaries can develop independent of neuronal
loss or chronic GFAP upregulation in glial cells. (Invest Oph-
thalmol Vis Sci. 2005;46:4281– 4287) DOI:10.1167/iovs.04-1361
D iabetic retinopathy is a leading cause of adult blindness
and is the most common complication of diabetes. It
affects more than 90% of people with diabetes, ultimately
leading to retinal edema, neovascularization, and vision loss in
some patients.1,2 Vascular changes characteristic of diabetic
retinopathy in humans, including breakdown of the blood–
retinal barrier, thickening of the capillary basement membrane
(BM), reduction in the number of pericytes, and an increase in
the number of acellular capillaries, have been widely docu-
mented in diabetic rats, dogs, and cats.3–5 However, mice as a
model of diabetic retinopathy have been less studied.
Recently, it has been reported that diabetes also induces
damage in nonvascular retinal neurons and Müller glial cells.6,7
Impaired retinal electrophysiology and neurodegeneration
have been detected in at least some diabetic patients.8 –10 In
diabetic rats, changes in retinal physiology and biochemistry
have been reported at as early as 1 to 2 months of hypergly-
cemia.8,11,12 In addition, retinal glial cells, primarily Müller glia
change from quiescent to an injury-associated phenotype and
express high levels of GFAP—a hallmark of glial cell activa-
tion—in the human retina during early diabetes.13 In rat Müller
glia, alterations in GFAP expression patterns7,13–15 and trans-
location of GAPDH to the nucleus (a proapoptotic change)16
have been observed after 2 to 5 months of experimental dia-
betes. In contrast, capillary obliteration first becomes signifi-
cantly greater than normal after years of diabetes in humans or
after approximately 6 months of diabetes in rats. Thus, diabe-
tes-induced nonvascular abnormalities precede the develop-
ment of the vascular cell changes in rats and may contribute to
the pathogenesis of the vascular disease.6 To date, the relation
of the nonvascular abnormalities to the development of the
vascular lesion characteristic of diabetic retinopathy is only
beginning to be explored.
Because of its great potential for genetic manipulation, the
mouse offers a unique opportunity to study the molecular
pathways involved in disease development; however, descrip-
tions of the retinal disease that develops in the diabetic mouse
are incomplete and often contradictory. In some studies, STZ-
induced diabetic mice showed no distinct vascular retinopathy
other than thickening of the retinal capillary BM for up to 20
months,17 whereas other reports have demonstrated that the
early stages of vascular diabetic retinopathy develops18 and are
inhibited in diabetic or galactosemic mice deficient in intercel-
lular adhesion molecule (ICAM) or CD18.18,19 Other studies
that focused on diabetes-induced changes in endothelial cells
and pericytes yielded mixed results.20 –22 Reports on effects of
diabetes on retinal neurons in mice also have been contradic-
tory. Increased frequency of apoptotic retinal neurons has
been reported in some18,23,24 but not all25 studies of diabetic
mice. To clarify these contradictions and provide systematic
4281
4282 Feit-Leichman et al.
characterization of retinal disease after induction of diabetes,
we investigated the occurrence and time course of diabetes-
induced changes in neurons, glia, and vasculature in the retina
of C57Bl/6J mice, a commonly used mouse strain.
MATERIALS AND METHODS
Experimental Animals
All experiments were performed in accordance with the protocol
approved by the Animal Care and Use Committee of The Schepens Eye
Research Institute (SERI) and Case Western Reserve University and the
tenets of the ARVO Statement for the Use of Animals in Ophthalmic
and Vision Research. C57Bl/6J mice (Jackson Laboratory, Bar Harbor,
ME) aged 7 to 10 weeks were rendered diabetic with five consecutive
daily intraperitoneal injections of STZ (55 mg/kg) freshly dissolved in
citrate buffer (pH 4.5). Development of diabetes (defined by blood
glucose greater than 250 mg/dL) was verified 1 week after the first STZ
injection (Glucometer Elite XL; Bayer Corp., Elkhart, IN). Mice were
housed in the SERI animal facility with 12-hour light/dark cycle and
allowed free access to food and water. Body weight was recorded
fortnightly. Glycemic control was estimated on multiple occasions
from the measurement of glycohemoglobin (GHb), using either a GHb
assay (Glyc-Affin; PerkinElmer, Norton, OH) or a glycohemoglobin
assay (Helena Glyco Tek Laboratory, Beaumont, TX).
Immunohistochemical Staining
The eyes or fresh retinal wholemounts were fixed in 4% paraformal-
dehyde for 1 hour. For staining of retinal sections, the eye was then
immersed in 30% sucrose for 4 hours, embedded in optimal cutting
temperature (OCT) compound (Tissue-Tek; Sakura Finetek, Torrance,
CA), and sectioned at 30 ␮m. Retinal wholemounts or sections were
then incubated in 1% Triton X-100 in 0.1% citrate for 1 hour and
stained with GFAP antibody conjugated to Cy-3 (1:5000; Sigma-Aldrich,
St. Louis, MO) at 4°C overnight. Immunofluorescence labeling was
then observed under a microscope equipped with fluorescence illumi-
nation (TE300; Nikon, Tokyo, Japan).
Western Blot Analysis
Freshly isolated mouse retinas were frozen in liquid nitrogen. They
were then homogenized with 1⫻ lysis buffer (Reporter Lysis 5⫻
Buffer; Promega, Madison, WI) and kept at 4°C for 1 hour. The protein
concentration was determined by a BCA kit (Micro BCA Assay Reagent
Kit; Pierce Chemical, Rockford, IL). Ten micrograms of protein from
each sample were electrophoresed in 10% SDS-PAGE with protein
markers (BMA ProSieve Color Protein Markers; BioWhittaker Molecular
Applications, Rockland, ME). After electrophoresis, proteins were
transferred to nitrocellulose membrane. The blot was then incubated
in blocking solution (1% bovine serum albumin, 2.5% milk protein,
0.2% Tween-20 in PBS) for 1 hour, followed by reactions with primary
antibodies against GFAP (1:5000; Sigma-Aldrich) or Ras-GAP as a con-
trol (1:2000; Sigma-Aldrich) in blocking buffer at 4°C overnight. The
membrane was incubated with goat anti-rabbit conjugated with horse-
radish peroxidase (1:5000; Chemicon International, Temecula, CA)
and goat anti-mouse conjugated with horseradish peroxidase (1:5000;
Chemicon International) in blocking buffer for 1 hour at room tem-
perature. Antibody detection was performed with enhanced chemilu-
minescence (Chemiluminescence System and SuperSignal West Pico
Chemiluminescent Substrate; Pierce Chemical). Densitometric analysis
was performed with Image J software (available by ftp at zippy.nimh.
nih.gov/ or at http://rsb.info.nih.gov/nih-image; developed by Wayne
Rasband, National Institutes of Health, Bethesda, MD).
Caspase Activity Assay
Caspase-3 activity was measured as described previously.26 Briefly,
equal amounts of retinal protein extracts (usually 15 ␮g) were incu-
bated in a lysate buffer containing the fluorogenic caspase-3 substrate
IOVS, November 2005, Vol. 46, No. 11
(DEVD-AFC; 2.5 ␮M) at a total volume of 100 ␮L at 32°C for 1 hour.
Cleavage of the substrate emitted a fluorescence signal that was quan-
tified by a fluorescence plate reader (excitation: 400 nm, emission: 505
nm; Spectra FluorPlus; Tecan, Durham, NC).
Preparation of Trypsin-Digested
Retinal Vasculature
The enucleated eye was placed in 10% buffered formalin for 4 to 5
days. The retina was dissected and incubated in a 3% crude trypsin
solution (DIFCO, Detroit, MI) containing 0.2 M sodium fluoride at 37°C
for 2 hours, as described elsewhere.27 The neuroretinal tissue was
gently brushed away, and the resultant isolated vascular tree was air
dried onto a glass microscope slide.
Tdt-dUTP Terminal Nick-End Labeling
To evaluate apoptosis in retinal vascular cells, the slide-mounted retinal
vascular digests were incubated with 1% Triton X-100 in 0.1% sodium
citrate for 1 hour, to enhance penetration of reagents into cells,
followed by staining with TUNEL. TUNEL-positive capillary nuclei
were then counted, and the data were expressed per retina. To eval-
uate cell apoptosis in the neuroretina, the mouse eye was fixed in 4%
paraformaldehyde for 15 minutes. The retina was then isolated and
placed in 2% Triton X-100 in PBS for 1 hour at room temperature.
Retinas treated with DNase were used as a positive control for TUNEL.
The retinas were flatmounted, and all samples for TUNEL were incu-
bated with terminal deoxynucleotidyl transferase conjugated to fluo-
rescein (In Situ Cell Death Detection Kit; Fluorescein Roche Diagnos-
tics, Indianapolis, IN), which labels free 3⬘OH DNA, at 37°C for 1 hour.
TUNEL-positive cells were counted within the entire retina of each
mouse.
Quantitation of Acellular Capillaries and
Pericyte Ghosts
After TUNEL-positive cells were counted, coverslips were soaked off,
and the retinal vasculature then was stained with hematoxylin-periodic
acid-Schiff. Acellular capillaries were counted in four to seven field
areas in the mid retina. Acellular capillaries were identified as capillary-
sized vessel tubes (not ⬍20% the diameter of surrounding healthy
capillaries) having no nuclei anywhere along their length. Pericyte
ghosts were estimated from the prevalence of spaces in the capillary
BMs from which pericytes had disappeared. Although counting the
ghosts, at least 1000 capillary cells (endothelial cells and pericytes)
were counted in five mid-retinal field areas. Ghosts on any acellular
vessel were excluded. All measurements were made in a masked
manner.
Electron Microscopy
After enucleation, the eye was incubated in half-strength Karnovsky
overnight. The anterior segment and vitreous were removed, and the
eyecup was washed in 0.1 M cacodylate buffer for 5 to 10 minutes.
Tissue was postfixed in 2% osmium tetroxide (Sigma-Aldrich), dehy-
drated, and embedded in Spurr’s resin. Capillaries were selected at
random from the inner nuclear layer and photographed under a trans-
mission electron microscope (model 410; Philips, Eindhoven, The
Netherlands). Approximately 20 electron micrographs were taken of
each retinal sample. The area and perimeter of the outer and inner BM
were measured with Image J software (NIH). BM thickness was calcu-
lated by dividing its areas by its length; results are reported in square
micrometers per 1000 micrometers.
Retrograde Labeling of Retinal Ganglion Cells
Mice were anesthetized by intraperitoneal administration of a mixture
of xylazine (2.5 mg/mL) and ketamine (12.5mg/mL). The skin over the
cranium was incised to expose the scalp. A hole (2 ⫻ 1 mm) was cut
with a scalpel, 4 mm posterior to the bregma and 1 mm lateral to the
midline on both sides of the midline raphe. The superior colliculi were
IOVS, November 2005, Vol. 46, No. 11 Diabetic Retinopathy in C57Bl/6J Mice 4283

exposed by gentle aspiration of the overlying occipital cortex. A piece

of Gelfoam (Pharmacia & Upjohn; Kalamazoo, MI) soaked in a 5%
solution of fluorescent gold tracer (Fluorogold; Fluorochrome, Denver,
CO) was directly applied to the superior colliculus. The skull was
replaced, and the overlying skin was sutured and covered with topical
antibiotic ointment. Seven days after application of the tracer (time to
allow retrograde uptake of dye and labeling of the RGC somata) mice
were euthanatized in a CO2 gas chamber and the eyes enucleated.
Retinas were isolated and placed in 4% paraformaldehyde for 10 min-
utes (Sigma-Aldrich). They were then flattened as wholemounts on
microscope slides (VWR, West Chester, PA). Ten fields at 20⫻ magni-
fication (area of 0.09 mm2) were photographed randomly from each
retina with a fluorescence microscope. The total number of RGC
bodies was counted in each field using Image J software (NIH).

Cell Counts in the Ganglion Cell Layer

Formalin-fixed whole eyes were embedded in paraffin and sectioned
sagittally (4 ␮m), so that each section passed through or next to the
optic nerve and the center of the cornea were collected. Sections were
stained with hematoxylin and eosin (H&E), and the number of cells in
the ganglion cell layer (GCL) was counted for a 250-␮m linear distance
on each side of the optic nerve (adjacent to the optic nerve). The
counts from the two sides were averaged and reported per unit length
of retina.

Statistical Analysis
Results are presented as mean ⫾ SD. The Mann-Whitney test or two-
paired Student’s t-test was used to assess significance between two
groups. One-way ANOVA followed by the post hoc Tukey (Fisher’s
protected least significant difference) test was used to assess statistical
significance between multiple groups. P ⱕ 0.05 was regarded as
statistically significant.
RESULTS
Induction of Diabetes in C57Bl/6J Mice
To induce diabetes in C57Bl/6J mice, we administered intra-
peritoneal STZ at 55 mg/kg for five consecutive days. For most
animals, the levels of blood glucose levels reached maximum
elevation at 4 weeks after STZ injection and remained elevated
throughout the time course of study. GHb levels in diabetic
mice were significantly higher (⬎3-fold) than control animals
(3.7 ⫾ 0.6%; P ⬍ 0.01), and all diabetic mice evaluated retained
elevated GHb until death. As male mice consistently developed
and maintained higher blood sugar levels than female mice
(Chen DF, unpublished observation, 2000), male mice were
selected for the studies. Some animals maintained body weight
and survived without insulin supplements, whereas others that
tended to lose weight were given small amounts (0.2 U, 0 –3
times per week) of insulin to prevent weight loss. A small
number of animals did not respond to STZ and were used as
the control for effects of the drug.
Neural Apoptosis in the Early Diabetic Retina
TUNEL was performed in wholemount retinas at 2 weeks and
1, 2, and 6 months after induction of diabetes (Fig. 1A). A
significant increase in TUNEL-positive cells, primarily localized
in the GCL was observed in retinas of mice at 2 weeks of
diabetes (P ⬍ 0.05). However, at 1 month, the number of
TUNEL-positive cells in diabetic mice declined toward normal,
although remaining slightly increased compared with the con-
trol group (P ⫽ 0.17). At 2 and 6 months of diabetes, no
significant differences in neuronal apoptosis were observed
between groups. To preclude that the transient induction of
neuronal apoptosis is caused by a possible toxic effect of STZ,
we performed TUNEL in retinas of mice in which diabetes did
FIGURE 1. Lack of evidence for significant neuronal loss in the mouse
retina in the early stage of diabetes. (A) Counts of TUNEL-positive cells
in the STZ-treated and control mouse retinas at various time points
after diabetes (n ⫽ 5/group). (B) Photomicrograph of representative
retinal sections stained with H&E. (C) Quantification of neuronal loss
in the GCL, assessed by number of cells counted in the GCL at 6 and
12 months after induction of diabetes (C). (D) Photomicrograph of
retrogradely labeled RGCs (arrowhead) in a retinal wholemount. (E)
Counts of retrogradely labeled RGCs in the diabetic and control mouse
retinas (n ⫽ 4/group). No observable difference in RGC loss was noted
between the control and diabetic groups.
not develop after STZ injection. We found no significant dif-
ference of the number of TUNEL-positive cells in STZ-treated
mice in which diabetes did not develop in comparison with the
age-matched control, even at 2 weeks after STZ injection
(4.0 ⫾ 1.3; P ⬎ 0.1 compared with age-matched control
group). Thus, the results indicate that the mouse neuroretina
undergoes a transient and small amount of cell apoptosis im-
mediately after induction of diabetes.
As the TUNEL-positive cells were localized primarily in the
GCL, we further assessed diabetes-induced neuronal loss by
counting cells in the GCL at 6 and 12 months of diabetes, using
retinal sections taken from diabetic and age-matched control
mice (Figs. 1B, 1C). After these long durations of diabetes, we
found no significant difference in the number of cells in the
GCL between the diabetic and nondiabetic mice (Fig. 1C). To
corroborate this result, we also performed retrograde labeling
of RGCs in diabetic (n ⫽ 4) and control (n ⫽ 4) mice at 3
months of diabetes (Figs. 1D, 1E). Labeled RGCs from both
eyes were counted. Again, we observed that the retinas of
diabetic mice contained a number of RGCs (4980 ⫾ 1663
RGC/mm2) similar to that of the control group (4942 ⫾ 1445
RGC/mm2; P ⬎ 0.5; Fig. 1E).
Because of the unexpected transient increase in the number
of TUNEL-positive neuronal cells in the mouse retina, we ex-
amined the activity of caspase-3, a downstream inducer of
apoptosis. Coinciding with the transient induction of neural
apoptosis, we noted a significant increase in caspase-3 activity
4284 Feit-Leichman et al.
in diabetic mouse retinas compared with age-matched control
group over the first month of diabetes, which likewise dimin-
ished during the ensuing 2 to 4 months after induction of
diabetes (Fig. 2). Caspase-3 activity increased again after 5
months, which appeared to correlate with the induction of
retinal vascular cell apoptosis at the later stage, as indicated
later in this study. Thus, these results suggest that retinal
neurons undergo a small and transient apoptosis soon after the
development of diabetes in mice, but this abnormality is not
maintained at longer durations of diabetes, nor does it result in
a significant loss of RGCs in diabetic C57Bl/6J mice.
Glial Cell Changes in the Diabetic Retina
Activation of Müller glial cells, characterized by upregulation of
GFAP, is another component of early diabetic retinopathy in
diabetic rats.13 To evaluate retinal glial cell changes, we exam-
ined expression of GFAP in mice from 2 weeks to 12 months
after diabetes induction. With Western blot analysis, the largest
increase of GFAP expression was observed at 1 month after the
induction of diabetes (Fig. 3), and this abnormality remained
significant at 2 months but quickly returned toward normal
with longer durations of diabetes. To determine further which
cell type was responsible for the increase in GFAP level after
the onset of diabetes, we performed immunofluorescence
staining in retinal sections, at 1, 2, and 6 month of diabetes.
Increased GFAP expression at 1 and 2 months of diabetes was
consistent with the labeling pattern of astrocytes in the GCL
(Fig. 4). In agreement with the quantification of GFAP levels
with Western blot, after 6 months of diabetes, the intensity of
GFAP labeling in the diabetic retinas returned to normal. At any
time point examined, no upregulation of GFAP expression was
observed in Müller glial cell bodies or other retinal layers that
were consistent with Müller glial cell labeling.
Development of Vascular Features Characteristic
of Diabetic Retinopathy in the Diabetic
Mouse Retina
Next, we studied the vascular disease in the mouse retina at
various time points, from 6 to 19 months, after STZ-injection.
Features of early vascular damage associated with diabetic
retinopathy include the formation of acellular capillaries and
pericyte ghosts and induction of vascular cell apoptosis. At 6
months after diabetes induction, a statistically significant in-
crease in the number of acellular capillaries was noted in the
FIGURE 2. Activity of caspase-3 measured from retinal homogenates
prepared at various time points after diabetes induction (n ⫽ 5/group).
Caspase-3 activity was significantly greater than normal at 1 month
after induction of diabetes, then diminished to normal, and began to
increase again after approximately 6 months. *P ⬍ 0.05.
IOVS, November 2005, Vol. 46, No. 11
FIGURE 3. Diabetes induced transient upregulation of GFAP expres-
sion in the mouse retina. (A) Western blot analysis of GFAP levels in the
retinas of diabetic (D) and age-matched control (C) mice at various
time points after STZ injection. Anti-RAS-GAP was used as a control. (B)
Densitometric measurement of band densities in Western blot analysis.
For each time point, the band-density level of control mouse retinas
was set to 100%, and the band-density level of diabetic mouse retinas
is presented as a percentage of the age-matched control retinas (n ⱖ
3/group). Note that the level of GFAP expression increased in the
diabetic mouse retinas beginning at 1 month after induction of diabetes
and continued through 2 months, but returned to normal at 6 months.
diabetic mice compared with the age-matched control animals
(P ⬍ 0.01; Fig. 5). With increasing duration of diabetes, the
frequency of acellular capillaries increased in diabetic animals
but not in nondiabetic control subjects. A significant increase
of capillary cell apoptosis and prevalence of pericyte ghosts
were also observed in diabetic mice after 6 to 9 months of
diabetes, and the frequency of the lesion likewise increased
with increasing duration of diabetes (Fig. 6). These results
indicate that the mouse exhibits vascular abnormalities char-
acteristic of early diabetic retinopathy in rats and humans.
BM Thickening
Thickening of the capillary BM is a phenomenon that is often
observed in diabetic retinopathy. To evaluate whether retinal
capillaries from diabetic mice undergo BM thickening, we
examined by transmission electron microscopy the mouse ret-
ina after 6, 12, and 15 months of induced diabetes. To mini-
mize variability, the examination was conducted only in capil-
laries located in the outer nuclear layer of the retina. We found
that capillary BM thickness tended to increase at longer dura-
tions of diabetes compared with the age-matched control, but
the results did not achieve statistical significance (Fig. 7).
DISCUSSION
Several important findings emerged from the present study.
First, diabetic C57Bl/6J mice exhibit development of vascular
disease characteristic of early diabetic retinopathy in rats and
humans,4 notably induction of acellular capillaries and vascular
cell apoptosis and formation of pericyte ghosts, as early as 6
months after induction of diabetes. The severity of these le-
sions is progressive, becoming more severe with longer dura-
tion of diabetes. Second, vascular lesions develop despite only
mild and transient abnormalities in the neural retina, unlike the
degeneration of RGCs and Müller glial changes that have been
reported in humans and rats.6,9,13–15
The mouse has not been studied extensively as a model of
diabetic retinopathy. The small size of the mouse eye presents
unique challenges in determining capillary lesions in the
IOVS, November 2005, Vol. 46, No. 11
FIGURE 4. Absence of apparent Mül-
ler glia activation in the diabetic mouse
retina. Representative photomicro-
graphs of retinal sections that were
prepared from diabetic (E–H) and age-
matched control (A–D) mice at 1, 2,
and 6 months after STZ-injection (n ⫽
3/group). GFAP upregulation was ob-
served in the GCL, with a pattern re-
sembling that of astrocytes (arrow-
heads), in the diabetic retinas at 1 and
2 months of diabetes. However, GFAP
labeling was not detected in Müller
glial cell bodies that normally appear in
the inner nuclear layer or their pro-
cesses in other retinal layers at any age
examined.
mouse retina. For example, pericyte ghosts are much more
difficult to detect in mice than they are in rats or larger species.
Although the mouse brings special problems, the ability to use
genetically modified mice to study the pathogenesis of diabetic
retinopathy gives a major advantage over the use of any other
animal species, offsetting the potential hardships. Herein, we
report a systematic study of the occurrence and progression of
retinal disease of diabetic retinopathy in the mouse. Our results
and those of others28 indicate that the C57Bl/6J mouse has
vascular lesions consistent with the early stages of diabetic
retinopathy, similar to findings in diabetic rats and human.
Likewise, the Akita mouse shows development of capillary
lesions in the early stages of diabetic retinopathy.18 Of note,
we found only slight thickening of retinal capillary BMs in our
mice, even as capillary degeneration was developing. This
suggests that processes leading to capillary BM thickening are
in some way different from those leading to hyperglycemia-
mediated death of retinal capillary cells.
In the present study, we quantified neuronal cell death in
the mouse retina at various times after the induction of diabe-
FIGURE 5. STZ-treated mice exhibited acellular capillaries, beginning
at 6 months after induction of diabetes. (A, B) Low- and high-magnifi-
cation photomicrographs of trypsin-digested retinas stained with peri-
odic acid-Schiff hematoxylin. Arrowhead: an acellular capillary. (C)
Quantitative analysis of acellular capillaries counted per mm2 trypsin-
digested retinas indicating a significant increase in the number of
acellular capillaries compared to control mice beginning at 6 months
after STZ-injection (n ⬎ 6/group). *P ⬍ 0.001; **P ⬍ 0.0001 with
ANOVA test.
Diabetic Retinopathy in C57Bl/6J Mice 4285
tes, and the results were corroborated using three methods: (1)
TUNEL, (2) counting the number of cells in the GCL in H&E-
stained retinal sections, and (3) counting surviving RGCs in
retinal wholemounts using a retrograde tracer (Fluorogold;
Fluorochrome). In each case, we did not detect significant loss
of retinal neurons in diabetic mice. Our inability to detect RGC
loss in the diabetic mouse seems not to be due to a method-
ological problem, because quantifying ganglion cells by the
same method (in cross sections) revealed a significant decrease
in the number of RGCs in diabetic rats after 8 months of
diabetes (Kern TS, unpublished results, 2004). Altogether, our
studies indicated a transient activation of proapoptotic pro-
cesses in the mouse retina after induction of diabetes, but it did
not lead to significant loss of neurons in long-term diabetes.
These findings disagree with a much shorter study by Mar-
tin et al.23 who found a progressive neuronal loss in the closely
related mouse strain (C57Bl/6) used in our study, which be-
came significantly greater than normal at 10 weeks after induc-
tion of diabetes. Differences between these studies that could
account for the different conclusions seemed to be few. First,
we provided low doses of insulin to the mice to prevent
diabetes-induced weight loss and failure to grow, whereas
Martin et al.23 administered no insulin to any of the experimen-
tal mice. Second, in our procedure, diabetes was induced in
FIGURE 6. Induction of capillary apoptosis and formation of pericyte
ghosts in the diabetic mouse retina. (C) The merged image of light
microscopy (A) and TUNEL (B) of trypsin-digested retina. Arrows:
TUNEL⫹ cells. (D, E) Counts of apoptotic capillary cells (D) and
pericyte ghosts (E) in trypsin-digested retinas prepared from STZ-
treated and age-matched control mice (n ⬎ 5/group). *P ⬍ 0.05, **P ⬍
0.01 (Mann-Whitney test).
4286 Feit-Leichman et al.
FIGURE 7. Absence of capillary BM thickening in the retinas of dia-
betic mice. Quantification of BM thickness at various time points after
STZ-injection, using transmission electron microscopy (n ⱖ 6/group).
No significant difference was detected between the control and STZ-
treated groups at 6 to 15 months after induction of diabetes. P ⬎ 0.1
(Student’s t-test).
mice at the age of 7 to 10 weeks, by using five consecutive
injections of low-dose STZ (55 mg/mL), whereas Martin et al.
induced diabetes in 3-week-old mice with three injections of a
higher dose (75 mg/mL). Finally, the minor differences in the
genetic background of the mice used (C57Bl/6J versus
C57Bl/6) might also result in the different observations. In a
study of retinal sections of Akita mice, significantly fewer cells
were also reported in the GCL than in nondiabetic control
animals.18
Neuronal lesions commonly evoke a response of glial cells,
which become activated and undergo reactive gliosis, a pro-
cess characterized by proliferation and changes in intermediate
filament (e.g., GFAP) production.29 In rat retina, GFAP upregu-
lation in Müller glia (but not astrocytes) develops soon after
diabetes, and continues for many months.14 In the present
study, we observed a transient response (caspase-3) in RGCs of
the diabetic mouse retina at 1 to 2 months of diabetes, which
correlated with the transient presence of TUNEL-positive neu-
ronal cells. Likewise, GFAP upregulation in diabetic mice was
transient and limited. Moreover, the upregulation of GFAP in
the mouse retina seemed to occur in astrocytes, but not Müller
glia, unlike what happens in diabetic rats.15 These data suggest
that although diabetes transiently triggers death signals in a
small population of neurons of the mouse retina, the damage is
probably not strong enough to evoke the response of Müller
glial cells in the mice,6 or GFAP upregulation is not a sufficient
marker of glial activation in this model. Müller glial cells do
develop proapoptotic changes in retinas of diabetic rats and
mice, as evidenced by translocation of GAPDH to the nucleus16
(and unpublished data, 2004), a change that has been closely
identified with apoptosis in other cell types.30 Whether these
abnormalities detected in neurons and glia were sufficient to
contribute to the vascular lesions that developed at longer
durations of diabetes cannot be positively determined at
present.
Recent reports comparing biochemical sequelae of hyper-
glycemia in retinas and other tissues between mice and rats
indicated several similarities31 and differences.25,32,33 At the
similar glucose levels, mice had much lower polyol pathway
activity in the retina than did rats.25,31,34,35 Activation of the
polyol pathway has been shown to account for the induction
of early neuronal apoptosis, GFAP upregulation in Müller cells,
and capillary BM thickening in the retina of diabetic
rats.25,36 –38 Thus, lower activity of the polyol pathway in the
diabetic mouse retina might be a factor in the modest evidence
of diabetes-induced damage of retinal neurons, Müller glial
cells, and BM thickening.22
IOVS, November 2005, Vol. 46, No. 11
It is intriguing that diabetes-induced microvascular lesions
develop in the mouse retina, despite the absence of significant
neuronal loss (as assessed by three independent methods) and
persistent glial cell activation (as assessed by GFAP induction).
These observations provide no support to a postulate that
diabetes-induced early changes in neurons and Müller glia
contribute critically to the later development of vascular le-
sions in diabetic retinopathy.6,9
Acknowledgments
The authors thank Mara Lorenzi for critical comments, Debra Shaum-
berg for statistical analysis, Patricia Pearson for electron microscopy,
and Casey Miller and Todd Hoehn for supervision and maintenance of
diabetic animals.
References
1. Cai J, Boulton M. The pathogenesis of diabetic retinopathy: old
concepts and new questions. Eye. 2002;16:242–260.
2. Archer DB. Bowman Lecture 1998. Diabetic retinopathy: some
cellular, molecular and therapeutic considerations. Eye. 1999;13:
497–523.
3. Engerman RL, Kern TS. Retinopathy in animal models of diabetes.
Diabetes Metab Rev. 1995;11:109 –120.
4. Yu DY, Cringle SJ, Su EN, et al. Pathogenesis and intervention
strategies in diabetic retinopathy. Clin Exp Ophthalmol. 2001;29:
164 –166.
5. Engerman RL. Pathogenesis of diabetic retinopathy. Diabetes.
1989;38:1203–1206.
6. Barber AJ. A new view of diabetic retinopathy: a neurodegenera-
tive disease of the eye. Prog Neuropsychopharmacol Biol Psychi-
atry. 2003;27:283–290.
7. Rungger-Brandle E, Dosso AA, Leuenberger PM. Glial reactivity, an
early feature of diabetic retinopathy. Invest Ophthalmol Vis Sci.
2000;41:1971–1980.
8. Barber AJ, Lieth E, Khin SA, et al. Neural apoptosis in the retina
during experimental and human diabetes: early onset and effect of
insulin. J Clin Invest. 1998;102:783–791.
9. Lieth E, Gardner TW, Barber AJ, Antonetti DA. Retinal
neurodegeneration: early pathology in diabetes. Clin Exp Ophthal-
mol. 2000;28:3– 8.
10. Parisi V, Uccioli L. Visual electrophysiological responses in persons
with type 1 diabetes. Diabetes Metab Res Rev. 2001;17:12–18.
11. Aizu Y, Oyanagi K, Hu J, Nakagawa H. Degeneration of retinal
neuronal processes and pigment epithelium in the early stage of
the streptozotocin-diabetic rats. Neuropathology. 2002;22:161–
170.
12. Agardh E, Bruun A, Agardh CD. Retinal glial cell immunoreactivity
and neuronal cell changes in rats with STZ-induced diabetes. Curr
Eye Res. 2001;23:276 –284.
13. Mizutani M, Gerhardinger C, Lorenzi M. Muller cell changes in
human diabetic retinopathy. Diabetes. 1998;47:445– 449.
14. Lieth E, Barber AJ, Xu B, et al. Glial reactivity and impaired
glutamate metabolism in short-term experimental diabetic retinop-
athy. Penn State Retina Research Group. Diabetes. 1998;47:815–
820.
15. Barber AJ, Antonetti DA, Gardner TW. Altered expression of retinal
occludin and glial fibrillary acidic protein in experimental diabe-
tes. The Penn State Retina Research Group. Invest Ophthalmol Vis
Sci. 2000;41:3561–3568.
16. Kusner LL, Sarthy VP, Mohr S. Nuclear translocation of glyceralde-
hyde-3-phosphate dehydrogenase: a role in high glucose-induced
apoptosis in retinal Müller cells. Invest Ophthalmol Vis Sci. 2004;
45:1553–1561.
17. Cuthbertson RA, Hopper JL, Mandel TE. Difference in effect of
cultured fetal pancreas transplants on retinal and renal capillary
basement membrane thickness in diabetic mice. Transplantation.
1989;48:218 –223.
IOVS, November 2005, Vol. 46, No. 11
18. Tamura H, Miyamoto K, Kiryu J, et al. Intravitreal injection of
corticosteroid attenuates leukostasis and vascular leakage in ex-
perimental diabetic retina. Invest Ophthalmol Vis Sci. 2005;46:
1440 –1444.
19. Joussen AM, Poulaki V, Le ML, et al. A central role for inflammation
in the pathogenesis of diabetic retinopathy. FASEB J. 2004;18:
1450 –1452.
20. Agren A, Rehn G, Naeser P. Morphology and enzyme activities of
the retinal capillaries in streptozotocin-diabetic mice. Acta Oph-
thalmol (Copenh). 1979;57:1065–1069.
21. Midena E, Segato T, Radin S, et al. Studies on the retina of the
diabetic db/db mouse. I. Endothelial cell-pericyte ratio. Ophthal-
mic Res. 1989;21:106 –111.
22. Kern TS, Engerman RL. A mouse model of diabetic retinopathy.
Arch Ophthalmol. 1996;114:986 –990.
23. Martin PM, Roon P, Van Ells TK, et al. Death of retinal neurons in
streptozotocin-induced diabetic mice. Invest Ophthalmol Vis Sci.
2004;45:3330 –3336.
24. Ning X, Baoyu Q, Yuzhen L, et al. Neuro-optic cell apoptosis and
microangiopathy in KKAY mouse retina. Int J Mol Med. 2004;13:
87–92.
25. Asnaghi V, Gerhardinger C, Hoehn T, et al. A role for the polyol
pathway in the early neuroretinal apoptosis and glial changes
induced by diabetes in the rat. Diabetes. 2003;52:506 –511.
26. Mohr S, Xi X, Tang J, Kern TS. Caspase activation in retinas of
diabetic and galactosemic mice and diabetic patients. Diabetes.
2002;51:1172–1179.
27. Kern TS, Engerman RL. Galactose-induced retinal microangiopathy
in rats. Invest Ophthalmol Vis Sci. 1995;36:490 – 496.
28. Joussen AM, Poulaki V, Le ML, et al. A central role for inflammation
in the pathogenesis of diabetic retinopathy. FASEB J. 2004;18:
1450 –1452.
Diabetic Retinopathy in C57Bl/6J Mice 4287
29. Bringmann A, Reichenbach A. Role of Müller cells in retinal de-
generations. Front Biosci. 2001;6:E72–E92.
30. Chuang DM, Hough C, Senatorov VV. Glyceraldehyde-3-phosphate
dehydrogenase, apoptosis, and neurodegenerative diseases. Annu
Rev Pharmacol Toxicol. 2005;45:269 –290.
31. Kowluru RA. Retinal metabolic abnormalities in diabetic mouse:
comparison with diabetic rat. Curr Eye Res. 2002;24:123–128.
32. Kowluru RA, Kowluru A. Erythrocyte sodium-potassium ATPase
activity and thiol metabolism in genetically hyperglycemic mice.
Metabolism. 1992;41:160 –164.
33. Llewelyn JG, Thomas PK, Mirrlees DJ. Aldose reductase activity
and myo-inositol levels in sciatic nerve and dorsal root ganglia of
the diabetic mutant mouse [C57/BL/Ks (db/db)]. Metabolism.
1991;40:1084 –1087.
34. Agardh CD, Agardh E, Hultberg B, Ahren B. Long-standing hyper-
glycemia in C57BL/6J mice does not affect retinal glutathione
levels or endothelial/pericyte ratio in retinal capillaries. J Diabetes
Compl. 2000;14:146 –153.
35. Lee AY, Chung SK, Chung SS. Demonstration that polyol accumu-
lation is responsible for diabetic cataract by the use of transgenic
mice expressing the aldose reductase gene in the lens. Proc Natl
Acad Sci USA. 1995;92:2780 –2784.
36. Robison WG Jr, Kador PF, Akagi Y, et al. Prevention of basement
membrane thickening in retinal capillaries by a novel inhibitor of
aldose reductase, tolrestat. Diabetes. 1986;35:295–299.
37. Akita M, Mizuno K, Matsubara A, et al. Effects of an aldose reduc-
tase inhibitor, SNK-860, on the histopathological changes of retinal
tissues in a streptozotocin-induced diabetic rat model. Acta Med
Okayama. 1993;47:299 –304.
38. Kojima K, Matsubara H, Harada T, et al. Effects of aldose reductase
inhibitor on retinal microangiopathy in streptozotocin-diabetic
rats. Jpn J Ophthalmol. 1985;29:99 –109.