Microglia/macrophages responses to kainate-induced injury in the rat retina

Microglia/macrophages responses to kainate-induced

injury in the rat retina

Min-Lin Chang

, Ching-Hsiang Wu

, Hsiung-Fen Chien

, Ya-Fen Jiang-Shieh

,

Jeng-Yung Shieh

, Chen-Yuan Wen

*

a

Department of Anatomy and Cell Biology, College of Medicine, National Taiwan University, 1, Section 1, Jen Ai Road, Taipei 100, Taiwan

b

Department of Biology and Anatomy, National Defense Medical Center, Taipei 114, Taiwan

c

Department of Surgery, College of Medicine, National Taiwan University, Taipei 100, Taiwan

d

Department of Anatomy, College of Medicine, National Cheng Kung University, Tainan 701, Taiwan Received 30 June 2005; accepted 29 November 2005

Available online 3 February 2006

Abstract

The present study was aimed to elucidate how retinal microglia/macrophages would respond to neuronal death after intravitreal kainate

injection. An increased expression of the complement receptor type 3 (CR3) and an induction of the major histocompatibility complex (MHC) class

II and ED-1 antigens were mainly observed in the inner retina after kainate injection. Prominent cell death revealed by Fluoro Jade B (FJB) staining

and ultrastructural examination appeared at the inner border of the inner nuclear layer (INL) at 1 day post-injection. Interestingly, some

immunoreactive cells appeared at the outer segment of photoreceptor layer (OSPRL) at different time intervals. Our quantitative analysis further

showed that CR3 immunoreactivity was drastically increased peaking at 7 days but subsided thereafter. MHC class II and ED-1 immunoreactivities

showed a moderate but steady increase peaking at 3 days and declined thereafter. Double labeling study further revealed that retinal microglia/

macrophages expressed concurrently CR3 and ED-1 antigens (OX-42

/ED-1

) or MHC class II molecules (OX-42

/OX-6

) and remained

branched in shape at early stage of kainate challenge. By electron microscopy, microglia/macrophages with CR3 immunoreactivity displayed

abundant cytoplasm containing a few vesicles and phagosomes. Other cells ultrastructurally similar to Mu¨ller cells or astrocytes could also engulf

exogenous substances. In conclusion, retinal microglia/macrophages responded vigorously to kainate-induced neuronal cell death that may also

trigger the recruitment of macrophages from neighboring tissues and induce the phagocytotic activity of cells other than retinal microglia/

macrophages.

# 2005 Elsevier Ireland Ltd and the Japan Neuroscience Society. All rights reserved.

1. Introduction

The microglial cells were one of the constituent glial cell types

in the retina and thought to be related to the mononuclear

phagocytic system (

Ogden, 1994

). The retinal microglial cells,

often branched, were characterized by an oval cell body bearing

an elongated nucleus (

Ogden, 1994

). Previous studies have

reported that the retinal microglia in rabbits and rats were mainly

located in the nerve fiber layer (NFL), the inner plexiform layer

(IPL) and, occasionally, in the outer plexiform layer (OPL)

(

Wang et al., 1996

). The cells were believed to serve as the

intrinsic immunocompetent cells in the retinal tissues (

Matsu-bara et al., 1999

) and may participate in antigen presentation,

especially when they were exposed to IFN-g (

Zhang et al., 1997;

Matsubara et al., 1999

). They were also activated in different

retinal injuries such as excitotoxic neurodegeneration (

Zeevalk

et al., 1989

) due to excessive accumulation of glutamate

(

Vorwerk et al., 2000

). High levels of extracellular glutamate

were probably associated with the degeneration of photoreceptor

cells, ischemic retina and glaucoma (

Ulshafer et al., 1990; Dreyer

et al., 1996; Tamai et al., 1997

).

A conformationally restricted analog of glutamate viz.

kainate has also been reported to have potent excitotoxicity on

neurons in the hippocampus and the retina (

Kleinschmidt et al.,

1986

). In the retina, kainate destroyed the ganglion cells

leading to severe morphological changes in the IPL and the

www.elsevier.com/locate/neures

* Corresponding author. Tel.: +886 2 3562211; fax: +886 2 3915292. E-mail address: wency@ntu.edu.tw (C.-Y. Wen).

0168-0102/$ – see front matter # 2005 Elsevier Ireland Ltd and the Japan Neuroscience Society. All rights reserved. doi:10.1016/j.neures.2005.11.010

inner nuclear layer (INL) (

Honjo et al., 2000

). Along with the

neuronal cell death induced by kainate, the retinal glial cells

exhibited signs of reactive changes (

Dreyfus et al., 1998

). In

this situation, the Mu¨ller cells were induced to show

phosphoinositide hydrolysis (

Lopez-Colome et al., 1993

) and

upregulated expression of glial fibrillary acidic protein (

Honjo

et al., 2000

). Furthermore, retinal microglial cells as revealed

by RCA-1 histochemistry were activated and followed a

distribution pattern similar to the terminal transferase-mediated

dUTP nick-end-labeling staining (

Shin et al., 2000

). Hence, it

was concluded that microglia were the only phagocyte involved

in kainate-induced retinal apoptosis (

Shin et al., 2000

). In the

light of this finding and in order to gain further insights on the

effect of kainate on retinal neurons, a better understanding of

microglia/macrophages reaction is vital, especially in regard to

the time and spatial patterns of the glial type. Indeed, the

changes of immunomolecules associated with the reactive

retinal microglia/macrophages have remained to be explored.

Using a panel of antibodies and ultrastructural examination,

this study sought to investigate thoroughly and quantitatively

the time and spatial patterns in the changes of

immunoexpres-sion of the complement receptor type 3 (CR3), major

histocompatibility complex (MHC) class II, and lysosome

antigens in the retinal microglia/macrophages, marked by

OX-42, OX-6 and ED-1, respectively, following the treatment of

kainate. Results of the present study may provide insight into

the relationship between the structural changes of retinas and

glial activation as well as some clues for rescuing neuronal

death in ischemic retinas or glaucoma suffering from a high

level of neurocytotoxic glutamate.

2. Materials and methods

2.1. Tissue processing

Adult male Wistar rats weighting 250–350 g (approximately 8–10 weeks old; n = 60) were used in this study. All experiments were conducted in accordance with the standards of the Association for Research in Vision and Ophthalmology. Kainic acid (KA; Sigma, K0250) was dissolved in sterile normal saline solution administered at the concentration of 5 nM. Each rat was given with an intravitreal injection of 1.4 ml KA into the right eye. The same volume of sterile 0.9% normal saline was injected into the left eye as the controls. The experimental animals were sacrificed at 1, 3, 7, 14 and 28 days after injections. Eight rats were used for each time point (including untreated group). Following deep anesthesia with an intraperitoneal injection of 7% chloral hydrate, all rats were perfused with Ringer’s solution, followed by 4% paraformaldehyde in 0.1 M phosphate buffer (PB; pH 7.4). After fixation, the eyeballs were removed and then the lens were detached before transferred into 30% sucrose solution and kept in the same solution overnight at 4 8C. Each eyeball was cut into sagittal sections in a cryostat at the thickness of 12 mm and the frozen sections obtained from different experimental groups were mounted on micro slides (DAKO; 5116).

2.2. Immunohistochemistry

Mounted sections were washed in Tris buffer solution (TBS) and treated in a solution containing 10% methanol and 1% hydrogen peroxide in TBS for 1 h to remove possible endogenous peroxidase and to improve the permeability of the cell membrane. After TBS washing, sections were blocked in a combina-tion of 0.1% Triton X-100 and 10% normal horse serum (NHS) for 1 h. They

were then incubated overnight with the following primary antibodies: mono-clonal antibody OX-42 (1:100; Serotec, MCA275R) for the detection of CR3, monoclonal antibody OX-6 (1:400; Serotec, MCA46R) for MHC class II antigen, and monoclonal antibody ED-1 (1:100; Serotec, MCA341R) for lysosomal protein of macrophages. After incubation, sections were washed in TBS and then treated with the secondary antibody, biotinylated horse anti-mouse IgG (1:200; Vector, BA2001), for 1 h. The reaction was amplified with streptavidin–biotin-peroxidase complex (1:300; DAKO, P0387) and visua-lized with 3,3-diaminobenzidine tetrahydrochloride (DAB; Sigma, D-5637). All sections were counterstained by cresyl fast violet and examined under the light microscope (Zeiss, Axiophot). For double labeling, mounted sections were blocked in TBS containing 10% normal goat serum (NGS) for 1 h, and then incubated with the first primary antibody: mouse anti-rat OX-6 (1:100; Serotec, MCA46R) or mouse anti-rat ED-1 (1:100; Serotec, MCA341R) IgG overnight, followed by a dilution (1:200) of fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse IgG (Jackson ImmunoResearch, 115-095-100) or tetramethylrhodamine isothiocyanate (TRITC)-conjugated goat anti-mouse IgG (Jackson ImmunoResearch, 115-025-166) in TBS for 2 h. After washing, the section was treated with one of the second primary antibodies: mouse anti-rat CD11b IgG-conjugated R-phycoerythrin (RPE) (1:20; Serotec, MCA275PE) or biotinylated mouse anti-rat CD11b IgG (1:50; Serotec, MCA275B) overnight. The later antibody was then detected with dichloro-triazinylaminofluorescein (DTAF)-conjugated streptavidin (Jackson Immu-noResearch, 016-010-084) in TBS for 2 h. These sections were then counterstained by TOTO-3 iodide (1:5000, Molecular Probes, T3604), and then examined under a confocal laser scanning microscope (Zeiss, LSM510, Carl Zeiss, Germany) or fluorescence light microscope (Zeiss, Axiophot, Carl Zeiss, Germany).

2.3. Fluoro-Jade B (FJB) staining

FJB staining is well-established method to mark degenerating neurons and fibers (Schmued et al., 1997). Mounted sections were first immersed in a solution containing 1% sodium hydroxide in 80% alcohol for 5 min, followed by 70% alcohol and distilled water each for 2 min. The sections were then transferred to a solution of 0.06% potassium permanganate for 20 min, and rinsed several times in distilled water. The staining solution was prepared from a 0.01% stock solution of FJB (Chemicon, AG310) that was made by adding 10 mg of the dye powder to 100 ml of distilled water. To make up 100 ml of working solution, 4 ml of the stock solution was added to 96 ml of 0.1% acetic acid vehicle, resulting in a final dye concentration of 0.0004%. After 20 min incubation in the working solution, the sections were washed several times in distilled water each for 5 min, and examined in a fluorescence light microscope using green light (500–570 nm) excitation.

2.4. Immunoelectron microscopy

Eight rats were used for immunoelectron microscopy. The experimental animals were subjected to a similar process as mentioned above and sacrificed by perfusing with 4% paraformaldehyde at 1 and 3 days after KA injection. Following perfusion, the eye-cups were carefully dissected and immersed in 10% sucrose solution and kept in the same solution overnight at 4 8C. By a vibratome, the retina was cut sagittally into serial 100 mm thick sections. These floating sections were rinsed several times in 0.1 M PB, and then treated in a solution containing 1% hydrogen peroxide in TBS for 1 h. Sections were blocked for 1 h in a combination of 0.01% Triton X-100 and 10% NHS. They were then incubated overnight with the monoclonal antibody OX-42 (1:100; Serotec, MCA275R). After incubation, sections were washed in TBS, and then treated with the secondary antibody, biotinylated horse anti-mouse IgG (1:200; Vector, BA2001) for 1 h. The reaction was amplified with streptavidin–biotin-peroxidase complex (1:300; DAKO, P0387), and visua-lized with DAB (Sigma, D-5637). The brown-colored sections were post-fixed in 1% OsO4for 30 min. After osmication, floating sections were briefly rinsed

in PB twice, and then dehydrated in an ascending series of ethanol. After dehydration, all specimens were embedded in pure Epon–Araldite mixture. Ultra-thin sections of the retina were double-stained with uranyl acetate and lead citrate. The stained sections were examined in a JOEL-2000 electron microscope.

Other four adult rats received intravitreal injection of 5 nM in a volume of 1.4 ml KA mixed with 1 ml of 1% horseradish peroxidase (HRP, type VI–A; Sigma, T6782). The experimental animals were sacrificed at 3 days after injection then perfused with 2% paraformaldehyde mixed with 2.5% glutar-aldehyde in 0.1 M PB. Using a vibratome, the retina was cut sagittally into serial 100 mm thick sections for HRP reaction and electron microscopy preparation.

2.5. Quantitative study and statistical analysis

In the present study, the intensity of different labeled profiles was not estimated, instead of their total area and density being analyzed. Immunola-beled sections from the above-mentioned groups were used for enumeration of labeled profiles (three animals for each group). Three sections through the level of the optic disc from one retina of each rat were collected. A total of three images from the area about 1 mm width around the optic disc in three sections per retina for the respective immunostaining (OX-42, OX-6 and ED-1) were scrutinized using a light microscope at the magnification of 400. The values of the total area of labeled profiles were obtained from the image analyzer (Image-Pro Plus) that defined the sum of gray-level pixels selected at a fixed threshold as the total area of labeled profiles. The constant threshold would be selected to cover all profiles of positive immunolabeling and applied to all grayscale digital images from sections with the same immunostaining. Because of a significant loss of retinal tissues starting at 3 days post-injection, the alterations of each immunomolecule in the atrophy retinas may not be reflected enough by the changes of the total area of labeled profiles. Therefore, the density of labeled profiles defined by the total area of labeled profiles against the area of the retinal profiles analyzed would be considered and may provide more information in understanding the revolution of immunomolecules in KA-challenged retinas. The area of the retinal profiles excluding the ONL and photoreceptor layer in each image captured was then measured and concomitantly the total area of labeled profiles was obtained from the same area. The areas of the retinal tissues at different time intervals were compared in a unit length (mm) of the retinal section. All data were expressed as mean S.E.M. in this study. Comparison of the total area of labeled profiles and the density of labeled profiles at various time points among groups was first analyzed using one-way ANOVA. At a fixed time point, differences between control and KA groups were analyzed using Student’s t-test. Statistical difference was considered significant if p < 0.05.

3. Results

When compared with that of untreated/normal rats (data not

shown), the retinas in rats receiving saline injection and killed

at 1, 3, 7, 14 and 28 days appeared structurally normal

(

Figs. 1A, 2A and 3A

). Although the retinal morphology at 1

day post-injection with KA was comparable to that in the

saline-treated rats, many neurons at the inner border of the INL

showed a pyknotic nucleus (

Figs. 1B, 2B and 3B

). A similar

topography of degenerating profiles was revealed by FJB

staining that also showed some degenerated cells at the GCL

(

Fig. 1

C). At 3 days post-injection with KA, the NFL, GCL and

IPL began to show signs of shrinkage with occasional pyknotic

neurons. At 7, 14 and 28 days, the attenuated NFL, GCL and

IPL collectively referred to as ANGI were hardly discernible.

3.1. Immunohistochemistry

3.1.1. OX-42 immunostaining

At different time intervals after saline injection (

Fig. 1

A),

OX-42 positive microglial cells showed a similar distribution pattern

in the NFL, GCL and IPL to the untreated/normal retinas (data

not shown). At 1 day following KA injection, OX-42 labeled

microglia/macrophages were noticeably hypertrophied but

remained ramified. Furthermore, the immunoreactivity and

profiles of cells especially those in the NFL, GCL and IPL were

markedly enhanced when compared with those in the control

retinas (

Fig. 1

B). At 3 days, many heavily stained cells were

localized in the ANGI, INL and OPL, notably in areas adjacent to

the inner border of the INL (

Fig. 1

D). At 7 days, microglia/

macrophages exhibited the strongest OX-42 immunoreactivity

and were distributed in the innermost retina (

Fig. 1

E).

Interestingly, a variable number of OX-42 positive cells occurred

between the outer segment of photoreceptor layer (OSPRL) and

the pigmented epithelium (

Fig. 1

H), and remained populated in

the ANGI (

Fig. 1

F,G) and the OSPRL (

Fig. 1

I and J) at 14 and 28

days post-injection. By quantitative analysis, labeled profiles of

saline-treated retinas at all time intervals examined were

comparable to those of the untreated/normal ones (data not

shown). However, when compared with that in the controls, the

total area of OX-42 immunoreactive profiles across all layers of

the retina showed a drastic increase after KA injection. The

upsurge reached its peak at 7 days, and after this, the value

declined drastically till 28 days (

Fig. 1

K). In view of the

significant loss of retinal tissue from 3 days after KA injection

(

Fig. 1

L), the density (total area of labeled profiles/tissue area) of

immunolabeled profiles with different antibodies was further

estimated. The results confirmed that the increase in different

immunolabeled profiles peaked at 7 days. In parallel to the total

area of OX-42 immunoreactive profiles, the density of labeled

profiles also showed a marked increase during the first week after

KA injection; thereafter it declined drastically (

Fig. 1

M).

3.1.2. OX-6 immunostaining

OX-6 immunoreactive cells were undetected in the

untreated/normal (data not shown) and saline-treated retinas

at all time intervals (

Fig. 2

A). At 1 day after KA injection, OX-6

immunoreactive cells were ramified and detected in the NFL,

GCL and IPL (

Fig. 2

B). At 3 days, many heavily labeled cells

were observed with the majority of them in the ANGI, and most

of them were rounded or emitted stout processes (

Fig. 2

C). At 7

days, OX-6 positive cells in the ANGI displayed an intense

immunoreactivity and were concentrated at the inner margin of

the ANGI and sporadically at the OSPRL (

Fig. 2

D and G). This

was also true for OX-6 labeled cells at 14 (

Fig. 2

E and H) and

28 days (

Fig. 2

F and I), but the cell populations were markedly

reduced when compared with earlier time intervals. Our

quantitative study showed that the total area of OX-6 labeled

profiles peaked at 3 days after KA injection and declined from

14 days onwards (

Fig. 2

J). On the other hand, the density of

OX-6 immunoreactive profiles showed a gradual increase

during the first 3 days after KA treatment. It peaked at 7 days

and decreased gradually from 14 days onwards (

Fig. 2

K).

3.1.3. ED-1 immunostaining

Except for some ED-1 labeled cells in the choroid, labeled

cells were not observed in the untreated/normal (data not

shown) and saline-treated retinas at all time intervals examined

(

Fig. 3

A). At 1 day after KA injection, punctuate ED-1

immunoreactive products were confined to labeled cells in the

NFL and GCL (

Fig. 3

B). At 3 days, ED-1 positive cells

appeared round or pleomorphic with some cells showing thin

and stout processes, and were distributed in the ANGI, INL and

OPL, more frequently at the junction between the ANGI and

INL (

Fig. 3

C). ED-1 immunoreactive cells were sparsely

distributed at 7 days and they tended to populate at the inner

half of the ANGI (

Fig. 3

D). This distribution pattern of ED-1

positive cells was also observed at 14 (

Fig. 3

E) and 28 (

Fig. 3

F)

days post-injection. As with OX-42 and OX-6 immunoreactive

Fig. 1. OX-42 immunoreactivity in the retinas following saline (A) and kainate (B–I) injection. In rats receiving saline injection, OX-42 immunoreactivity is distributed in the NFL, GCL and IPL at various time points (A, arrow). One day following kainate injection, OX-42 immunoreactivity is enhanced (B, arrow). Meanwhile, many neurons showing pyknotic nuclei appear at the inner border of the INL (B, arrowhead) and are labeled by Fluoro-Jade B (FJB) staining (C, arrow). Note other FJB-positive neurons at the GCL (C, arrowhead). At 3 days, cells intensely stained for OX-42 are observed in the region adjacent to the inner border of the INL (D, arrow) and, at 7 days in the inner ANGI (E, arrow). OX-42 immunoreactive cells persist in the ANGI at 14 (F, arrow) and 28 (G, arrow) days. Note that a few OX-42 positive cells occur occasionally in the OSPRL at 7 (H), 14 (I) and 28 (J) days post-injection (arrows). Quantitative analysis shows that the total area of OX-42 immunoreactive profiles is significantly elevated at 1 day after kainate injection, peaked at 7 days but declined drastically thereafter (K). Note the considerable loss of retinal tissue beginning at 3 days till 7 days after kainate injection; subsequently the value being to plateau till 28 days (L). The density of OX-42 labeled profiles reaches its peak rapidly at 7 days after kainate injection and then declines drastically in the same manner (M). #, significant difference ( p < 0.05) when compared with the saline-treated groups; P, significant difference ( p < 0.05) when compared with the preceding time interval; S, significant difference ( p < 0.05) when compared with the subsequent time interval; B, significant difference ( p < 0.05) when compared with both the preceding and subsequent time intervals. NFL, nerve fiber layer; GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; ANGI, attenuated NFL, GCL and IPL; OSPRL, outer segment of photoreceptor layer; scale bars = 50 mm.

cells, a variable number of ED-1 positive cells were observed at

the OSPRL at different time intervals (

Fig. 3

G–I). Similar to

those of OX-6 immunoreactive profiles, quantitative analysis

indicated a peak total area of ED-1 labeled profiles at 3 days

after KA injection and a steady decrease thereafter (

Fig. 3

J). It

is also true for the density of ED-1 immunoreactive profiles that

showed a gradual increase at beginning of injury after KA

treatment following by a climax at 7 days and a gradual

reduction from 14 days onwards (

Fig. 3

K).

3.1.4. Immunophenotypes of microglia/macrophages in

KA-challenged retinas

In order to get insight into the nature of microglia/

macrophages in KA-challenged retinas, double labeling of

the abovementioned immunomolecules was applied. We found

that cells concurrently expressing CR3 and unknown

cyto-plasmic/lysosomal antigens (OX-42

/ED-1

,

Fig. 4

A–C)

existed mainly in the inner retinas and remained ramified at

the first 3 days post-injection. At the same period of KA

Fig. 2. OX-6 immunoreactivity in the retinas following saline (A) and kainate (B–I) injection. OX-6 immunoreactivity is absent in the saline-treated retina (A). At 1 day after kainate injection, OX-6 immunoreactivity (B, arrow) is markedly induced in ramified cells at the NFL, GCL and IPL with the occurrence of pyknotic nuclei (B, arrowhead) in the inner border of the INL. At 3 days, most of the OX-6 immunoreactive cells are round with some bearing stout processes (C, arrow). The distribution of OX-6 positive cells including those in OSPRL parallels that of OX-42 cells at 7 days (D and G, arrows). OX-6 immunoreactive cells persist at 14 (E and H, arrows) and 28 (F and I, arrows) days post-injection. By quantitative analysis, the total area of OX-6 labeled profiles (J) peaks at 3 days and declines steadily till 28 days. Note that the density of OX-6 labeled profiles (K) rises gradually in the initial stage, and decreases gradually after peaking at 7 days. Abbreviations and symbols same as inFig. 1; scale bar = 50 mm.

challenge, other ramified cells expressing CR3 also contained

MHC class II molecules (OX-42

/OX-6

,

Fig. 5

A–C) and

occurred frequently at the NFL and GCL.

3.2. Electron microscopic study

By immunoelectron microscopy, OX-42 positive cells in the

untreated/normal retinas showed immunoreaction product at

the plasma membrane and displayed characteristic features of

microglia (

Fig. 6

A). At 1 day after KA injection, OX-42 labeled

microglia showed more abundant cytoplasm containing

numerous OX-42 positive vesicles and a few lysosomes and

phagosomes (

Fig. 6

B). Meanwhile, degenerating neurons found

at the inner border of the INL showed a pyknotic nucleus

containing irregular clump of electron-dense chromatin and

disorganized cytoplasm with some vacuoles and swollen

mitochondria (

Fig. 7

A). Cell processes either positive or

negative for OX-42 frequently surrounded the degenerating

cells (

Fig. 7

A and B). Similar degenerating neurons (

Fig. 7

B)

were observed at 3 days when OX-42 labeled microglia/

Fig. 3. ED-1 immunoreactivity in the retinas following saline (A) and kainate (B–I) injection. ED-1 immunoreactivity is undetected in saline-injected rats (A). At 1 day after kainate injection, punctate ED-1 immunoreaction products (B, arrow) are induced mainly in the NFL with concurrent occurrence of a few pyknotic nuclei at the inner border of the INL (B, arrowhead). At 3 days, round ED-1 positive cells with some of them showing thin and stout processes (C, arrow) appear in vicinity of the inner INL. As with OX-42 and OX-6 immunoreactive cells, ED-1 immunoreactive cells appear to populate at the inner half of the ANGI and occasionally in the OSPRL at 7 (D and G, arrows) 14 (E and H, arrows) and 28 (F and I, arrows) days. Similar to the expression pattern of OX-6 labeled profiles, the total area of ED-1 labeled profiles (J) also peaks at 3 days and declines steadily till 28 days. Quantitative data further show a gradual increase of the density of ED-1 labeled profiles (K) in the initial stage of kainate stimulation and a steady decrease after its peak expression at 7 days. Abbreviations and symbols same as inFig. 1; scale bar = 50 mm.

macrophages were characterized by their copious cytoplasm

endowed with immunoreactive vesicles and phagosomes

(

Fig. 8

A). Some labeled cells were rounded or amoeboidic

with an indented nucleus. Its moderate amount of cytoplasm

contained OX-42 labeled vesicles, tubule-like structures and

vacuoles (

Fig. 8

B). In addition, many microglia/macrophages

were laden with lipid droplets and phagosomes containing

reaction products of exogenous HRP that was co-administrated

with KA into the vitreum (

Fig. 9

A and B). Occasionally HRP

products were also found in lysosome-like structures of other

cells resembling Mu¨ller cells or astrocytes. The latter cells also

appeared to enclose degenerating debris (

Fig. 9

A–C).

4. Discussion

In optic nerve axotomy model,

Garcia-Valenzuela et al.

(2005)

detected an alteration of OX-42 labeled microglial cells

in the affected retinas at day 5 after lesioning. The number of

reactive microglia in the GCL appeared to reach its peak at day

12 post-lesion. Using fluorescent microspheres to label

circulating monocytes, the authors concluded a limited

contribution of blood borne macrophages in the axotomized

retinas (

Garcia-Valenzuela and Sharma, 1999;

Garcia-Valen-zuela et al., 2005

). In the same injury model,

Zhang and Tso

(2003)

also demonstrated that OX-42 positive cells in the inner

retina could phagocytose dying ganglion cells and migrate to

the subretinal space. It was suggested that OX-42 and ED-1

positive cells in the normal and lesioned retinas were resident

microglia. In retinal ischemia/reperfusion injury,

Zhang et al.

(2005)

reported the occurrence of numerous intensely labeled

OX-42, ED-1 and OX-6 cells in the inner retina early at 1 day

after ischemia and the distribution of labeled cells in the

subretinal space during 3–14 days of recovery. Using RCA-1

histochemistry to label phagocytes,

Shin et al. (2000)

stated that

microglia were the only phagocytes involved in clearance of

apoptotic debris from the inner retinas at 1 day after kainate

injection. We confirm here an early response of OX-42, ED-1

and OX-6 labeled cells in the inner retina at 1 day after kainate

injection and, furthermore, a significant distribution of them in

the OSPRL (subretinal space) during the late stage of

excitotoxic injury. It would appear therefore from studies by

others as well as by us that the rapid reactive response of

microglia to retinal injury is a general phenomenon. The

present ultrastructural results however have added the fact that

the retinal Mu¨ller cells other than microglia/macrophages also

partake in the removal of cellular debris induced by kainate.

Fig. 4. Fluorescent images of retinal microglia/macrophages double-labeled with OX-42 (A) and ED-1 (B) at 3 days after kainate injection. Microglia/macrophages in kainate-challenged retinas are labeled with anti-OX-42 conjugated RPE (A, red) followed by anti-ED-1 antibody visualized with goat anti-mouse IgG-conjugated FITC (B, green). The merged image (C) shows that microglia/macrophages in lesioned retinas concurrently express OX-42 and ED-1 immunoreactivities (C, yellow) and are ramified in form (C, arrows). Abbreviations same as inFig. 1; scale bar = 50 mm.

Fig. 5. Confocal images of retinal microglia/macrophages double-labeled with OX-42 (A) and OX-6 (B) at 3 days after kainate injection. Retinal microglia/ macrophages are subjected to mark with biotinylated anti-OX-42 antibody detected with streptavidin-conjugated DTAF (A, green) and subsequently with anti-OX-6 antibody imagined with goat anti-mouse IgG-conjugated TRITC (B, red). In the merged image (C) with TOTO-3 nuclear staining (blue), it demonstrates the co-expression (yellow) of OX-42 and OX-6 immunoreactivities on microglia/macrophages that still bear cell processes (C, arrows). Abbreviations same as inFig. 1; scale bar = 50 mm.

Our quantitative study has further demonstrated a distinct

expression pattern between OX-42 and ED-1/OX-6

immunor-eactivities (not numbers of OX-42, ED1 and OX-6 labeled

cells) in lesioned retinas suggesting that this may be a

differential or specific response of subsets of retinal microglia/

macrophages to kainate stimulation.

In untreated or control retinas, microglia bearing CR3 were

widely distributed across the different layers of retina, which

were devoid of MHC class II and ED-1 unknown cytoplasmic/

lysosomal antigens. Following kainate injection, CR3

immu-noreactivity of cells especially those in the NFL, GCL and IPL

was markedly enhanced. Furthermore, the cells were clearly

hypertrophied when compared with those in the control retinas.

Moreover, the present study has shown that kainate induced the

expression of MHC class II and ED-1 antigens in retinal

ramified cells coincident to those labeled by CR3 antibody in

terms of their spatial distribution and colocalization of

immunomolecules. The characteristic immunophenotypes of

retinal microglia have been reported in different injury or

disease models of the retina (

Leon et al., 2000; Zeng et al.,

2000; Fauser et al., 2001

). For example, in the diabetic retina

(

Zeng et al., 2000

) or the retina of experimental autoimmune

uveitis (

Fauser et al., 2001

), microglia were activated to express

MHC class II and ED-1 antigens that were increased across the

Fig. 6. Electron micrographs showing OX-42 immunoreactive cells in the normal rat retina (A) and at 1 day post-injection with kainate (B). The cell in the normal retina shows OX-42 immunoreactivity at its plasma membrane (A, arrows) and emits processes (A and P). Its nucleus (N) is small and irregular and bears heterochromatin masses (A, arrowheads) beneath the nuclear envelope. OX-42 positive cell (B) at 1 day after kainate injection displays a rod-shaped soma that contains more abundant cytoplasm and emits numerous thinner processes (B and T). The OX-42 immunoreactivity is increased at its plasma membrane (B, arrows) and vesicles (B, arrowheads); scale bars = 1 mm.

Fig. 7. Electron micrographs of degenerating neurons in the retinas at 1 (A) and 3 days (B) after kainate injection. At 1 day, the degenerating neuron in the inner border of the INL (A) undergoes necrosis with a pyknotic nucleus (A and N) and disrupted cytoplasm with vacuoles (A and V). Arrows in (A) indicate OX-42 negative cytoplasmic processes that appear to surround the degenerating neuron. At 3 days, a degenerating neuron (B) showing disintegrated nucleus (B and N) and cytoplasm (B and Cp) is in close apposition to an OX-42 labeled cytoplasmic process (B, arrows); scale bars = 1 mm.

different retinal layers and extended into the ONL at later

period of the disease (

Zeng et al., 2000

). An increased level of

cells positive for ED-1 in the retina by lens injury or by

intraocular injection of a macrophage activator (Zymosan) had

been further demonstrated (

Leon et al., 2000

). The induction of

MHC class II and ED-1 antigens may therefore reflect a

common signal of microglial activation in the affected retina

and function in antigen presentation, sampling/cleaning

endogenous and exogenous intraocular molecules and/or

probably promoting tissue regeneration (

Leon et al., 2000

).

By quantitative analysis, we have shown that the total area of

labeled profiles of either MHC class II or ED-1 antigens in

retinal microglia/macrophages peaked at 3 days after kainate

injection. The density of the labeled profiles, however, reached

its peak at 7 days. This may be attributed to the severe loss of

retinal tissues at 7 days. Both the density and total area of CR3

immunoreactive profiles were coincident peaking at 7 days and

they decreased drastically thereafter. In contrast, the retinal

microglia/macrophages expressing either MHC class II or

ED-1 antigens showed a slow decline after peaking at 3 and 7 days

in the total area and density of labeled profiles, respectively, at

the later time intervals. The significance of differential

Fig. 8. Electron micrographs of OX-42 positive cells in the retina at 3 days after kainate injection. OX-42 immunoreactive cells at 3 days are amoeboidic (A) or rounded (B) in outline. The immunoreactive cell in (A) shows broad processes (P) and abundant cytoplasm containing vesicles (A, arrows) and phagosomes (A, double arrows) with OX-42 immunoreaction products. The round immu-noreactive cell (B) has an indented nucleus (N) and moderate amount of cytoplasm. Its vesicles (B, arrows), tubule-like structures (B, arrowheads) and vacuoles (B, double arrows) are positive for OX-42; scale bars = 1 mm.

Fig. 9. Electron micrographs of the retinal structures at 3 days after co-injection with kainate and HRP. Note full-blown microglia/macrophages (A and B, M) possessing numerous lipid droplets (A and B, L) and phagosomes incorporating HRP reaction products (A and B, arrows). Near to the above cells are cell processes that show a slightly darker cytoplasm resembling Mu¨ller cells or astrocytes. These processes also include HRP reaction products in lysosome-like structures (A–C, arrowheads) and appear to surround some degenerating tissues (B and C, double arrows); scale bars = 1 mm.

expressions of CR3 and MHC class II/ED-1 antigens in

microglia/macrophages following injection is unclear. This

may hypothetically be attributed to the constitutive and

inducible nature of CR3 and MHC class II/ED-1 antigens,

respectively, in retinal microglia that demanded these

immunomolecules differentially in response to tissue injury.

A noteworthy feature in this study was that cells expressing the

respective or two immunomolecules remained ramified at 1–3

days after kainate injection. At 3 days, however, more round

and amoeboid microglia/macrophages were observed. It is

suggested that the round and amoeboidic

microglia/macro-phages are derived from activation and transformation from the

ramified microglial cells. The possibility of invasion of

circulating monocytes giving rise to these cells at later time

points cannot be excluded.

It is well documented that microglial cells are pathologic

sensor as they respond vigorously and swiftly to subtle changes

in their ambient environment (

Wu et al., 1997; Ling et al., 2001;

Chang et al., 2003

). A novel finding in the present study was the

increase in immunomolecules of microglia/macrophages in the

inner border of the INL in kainate-treated rats especially in

those killed at 3 days. The robust microglia/macrophages

reaction in the inner border of the INL is consistent with

previous results by

Morgan and Ingham (1981)

, who showed

that low doses (6 nM) of kainate significantly produced lesion

confined to the IPL and INL where amacrine cells were

markedly reduced in number. It is therefore speculated that

microglia/macrophages reaction in lesioned retina may be

elicited by kainate-induced amacrine cell deaths that are

concentrated at the inner border of the INL. Previous studies

have also shown that optic axotomy significantly caused

ganglion cell death in the GCL and induced microglial

population changes in the restricted layer such as the NFL, GCL

or IPL (

Zhang and Tso, 2003

). This is in agreement with our

present results showing an increased immunomolecules of

microglia/macrophages adjacent to the vitreal surface of

kainate-treated retina at later time periods. It is noteworthy

that the main retinal vasculatures are distributed in the same

region. Perivascular accumulation of microglia/macrophages

has been evidenced during the later stages of various brain

injuries or diseases (

Wu et al., 1997; Grossmann et al., 2002

). In

this connection, we proposed that retinal

microglia/macro-phages accumulation in the inner ANGI may be recruited

toward the blood vessels at later time points of kainate-induced

injuries. An interesting feature in the kainate-treated retina was

the occurrence of some OX-42, OX-6 and ED-1 labeled

microglia/macrophages in the region between the OSPRL and

the retinal pigment epithelium at various time intervals a

feature that was reported by

Shin et al. (2000)

who also

examined microglial reaction in the retina after application with

kainate. It has been reported that kainite-induced degenerative

changes in other brain tissues were accompanied by monocyte

invasion from blood circulation (

Sola et al., 1997

). In diabetic

retina, neuronal degeneration was accompanied by microglial

migration either from resident microglia in the NFL and GCL

or infiltration of monocytes from the retinal blood vessels (

Zeng

et al., 2000

). In the present study, labeled cells in the OSPRL

may be derived from the circulating monocytes in the

vessel-rich choroid. Indeed, in the study on photocoagulated rabbit

retina it was reported that monocytes invaded the retina from

the choroid through the Bruch’s membrane (

Ishikawa et al.,

1983

).

Previous studies have shown that phagocytes can be induced

under a variety of microenvironmental disturbances in the

retina. Microglia, which eliminate the degenerating debris from

the central nervous system (

Shin et al., 2000

) thus functioning

as phagocytes, have also been regarded as important immune

effector cells in the retina (

Zeng et al., 2000

). Recently,

Shin

et al. (2000)

have suggested that microglial cells are likely to be

the one and only type of phagocytes in the kainate-induced

retinal apoptosis. Besides phagocytes, Mu¨ller cells, the

principal neuroglial cells in the retina (

Newman and

Reich-enbach, 1996

), are also involved in phagocytosis during the

development (

Egensperger et al., 1996

) and in various

pathologic conditions of the retina (

Roque and Caldwell,

1990; Kim et al., 1998

). Our immunoelectron microscopic

study has shown that microglia/macrophages endowed with

CR3 and which are known to be involved in phagocytosis

contained a significant amount of phagosomes at 1 and 3 days

after kainate-induced degeneration. The present ultrastructural

examination further showed that cell elements other than

microglia/macrophages particularly those resembling Mu¨ller

cells or astrocytes frequently wrapped degenerating neurons or

structures and contained exogenous HRP reaction products in

their lysosome-like structures, indicating their phagocytic

capability.

In conclusion, microglia/macrophages reaction, probably

indicative of cell activation, is linked to neuronal cell loss/

degeneration, both occurring between 1 and 28 days of

kainate-induced excitotoxicity. On the other hand, the present results

suggest that microglia/macrophages are not the only type of

phagocytes in kainate-lesioned retina in rats. Their

phagocy-tosis capacity may not be sufficient enough to eliminate

kainate-induced degenerating debris that was removed by other

retinal cells of non-phagocyte nature.

Acknowledgements

This study was supported in part by research grants (NSC

92-2320-B002-113

and

NSC

93-2320-B002-033)

from

National Science Council, Taiwan.

References

Chang, C.Y., Chien, H.F., Jiangshieh, Y.F., Wu, C.H., 2003. Microglia in the olfactory bulb of rats during postnatal development and olfactory nerve injury with zinc sulfate: A lectin labeling and ultrastrucutural study. Neurosci. Res. 45, 325–333.

Dreyer, E.B., Zurakowski, D., Schumer, R.A., Podos, S.M., Lipton, S.A., 1996. Elevated glutamate levels in the vitreous body of humans and monkeys with glaucoma. Arch. Ophthalmol. 114, 299–305.

Dreyfus, H., Sahel, J., Heidinger, V., Mohand-Said, S., Guerold, B., Meuillet, E., Fontaine, V., Hicks, D., 1998. Gangliosides and neurotrophic growth factors in the retina: Molecular interactions and applications as neuropro-tective agents. Ann. N. Y. Acad. Sci. 845, 240–252.

Egensperger, R., Maslim, J., Bisti, S., Hollander, H., Stone, J., 1996. Fate of DNA from retinal cells dying during development: uptake by microglia and macroglia (Mu¨ller cells). Brain Res. Dev. Brain Res. 97, 1–8.

Fauser, S., Nguyen, T.D., Bekure, K., Schluesener, H.J., Meyermann, R., 2001. Differential activation of microglial cells in local and remote areas of IRBP1169–1191-induced rat uveitis. Acta Neuropathol. 101, 565–571. Garcia-Valenzuela, E., Sharma, S.C., 1999. Laminar restriction of retinal

macrophagic response to optic nerve axotomy in the rats. J. Neurobiol. 40, 55–66.

Garcia-Valenzuela, E., Sharma, S.C., Pina, A.L., 2005. Multilayered retinal microglial response to optic nerve transaction in rats. Mol. Vis. 11, 225–231. Grossmann, R., Stence, N., Carr, J., Fuller, L., Waite, M., Dailey, M.E., 2002. Juxtavascular microglia migrate along brain microvessels following activa-tion during early postnatal development. Glia 37, 229–240.

Honjo, M., Tanihara, H., Kido, N., Inatani, M., Okazaki, K., Honda, Y., 2000. Expression of ciliary neurotrophic factor activated by retinal Mu¨ller cells in eyes with NMDA- and kainic acid-induced neuronal death. Invest. Ophthal-mol. Vis. Sci. 41, 552–560.

Ishikawa, Y., Momoeda, S., Yoshitomi, F., 1983. Origin of macrophage in photocoagulated rabbit retina. Jpn. J. Ophthalmol. 27, 138–148. Kim, I.B., Kim, K.Y., Joo, C.K., Lee, M.Y., Oh, S.J., Chung, J.W., Chun, M.H.,

1998. Reaction of Mu¨ller cells after increased intraocular pressure in the rat retina. Exp. Brain Res. 121, 419–424.

Kleinschmidt, J., Zucker, C.L., Yazulla, S., 1986. Neurotoxic action of kainic acid in the isolated toad and goldfish retina: I. Description of effects. J. Comp. Neurol. 254, 184–195.

Leon, S., Yin, Y., Nguyen, J., Irwin, N., Benowitz, L.I., 2000. Lens injury stimulates axon regeneration in the mature rat optic nerve. J. Neurosci. 20, 4615–4626.

Ling, E.A., Ng, Y.K., Wu, C.H., Kaur, C., 2001. Microglia: its development and role as a neuropathology sensor. Prog. Brain Res. 132, 61–79.

Lopez-Colome, A.M., Ortega, A., Romo-de-Vivar, M., 1993. Excitatory amino acid-induced phosphoinositide hydrolysis in Mu¨ller glia. Glia 9, 127–135. Matsubara, T., Pararajasegaram, G., Wu, G.S., Rao, N.A., 1999. Retinal microglia differentially express phenotypic markers of antigen-presenting cells in vitro. Invest. Ophthalmol. Vis. Sci. 40, 3186–3193.

Morgan, I.G., Ingham, C.A., 1981. Kainic acid affects both plexiform layers of chicken retina. Neurosci. Lett. 21, 275–280.

Newman, E., Reichenbach, A., 1996. The Mu¨ller cell: a functional element of the retina. Trends Neurosci. 19, 307–312.

Ogden, T.E., 1994. Glia. In: Stephen, J.R. (Ed.), Retina. Mosby, St. Louis, pp. 54–57.

Roque, R.S., Caldwell, R.B., 1990. Mu¨ller cell changes precede vascularization of the pigment epithelium in the dystrophic rat retina. Glia 3, 464–475. Schmued, L.C., Albertson, C., Slikker Jr., W., 1997. Fluoro-Jade: a novel

fluorochrome for the sensitive and reliable histochemical localization of neuronal degeneration. Brain Res. 751, 37–46.

Shin, D.H., Lee, H.Y., Lee, H.W., Lee, K.H., Lim, H.S., Jeon, G.S., Cho, S.S., Hwang, D.H., 2000. Activation of microglia in kainic acid induced rat retinal apoptosis. Neurosci. Lett. 292, 159–162.

Sola, C., Tusell, J.M., Serratosa, J., 1997. Calmodulin is expressed by reactive microglia in the hippocampus of kainic acid-treated mice. Neuroscience 81, 699–705.

Tamai, K., Toumoto, E., Majima, A., 1997. Local hypothermia protects the retina from ischemic injury in vitrectomy. Br. J. Ophthalmol. 81, 789–794.

Ulshafer, R.J., Sherry, D.M., Dawson, R., Wallace Jr., D.R., 1990. Excitatory amino acid involvement in retinal degeneration. Brain Res. 531, 350–354. Vorwerk, C.K., Naskar, R., Schuettauf, F., Quinto, K., Zurakowski, D., Goche-nauer, G., Robinson, M.B., Mackler, S.A., Dreyer, E.B., 2000. Depression of retinal glutamate transporter function leads to elevated intravitreal gluta-mate levels and ganglion cell death. Invest. Ophthalmol. Vis. Sci. 41, 3615– 3621.

Wang, C.C., Wu, C.H., Shieh, J.Y., Wen, C.Y., Ling, E.A., 1996. Immunohis-tochemical study of amoeboid microglial cells in fetal rat brain. J. Anat. 189, 567–574.

Wu, C.H., Wang, H.J., Wen, C.Y., Lien, K.C., Ling, E.A., 1997. Response of amoeboid and ramified microglial cells to lipopolysaccharide injections in postnatal rats—a lectin and ultrastructural study. Neurosci. Res. 27, 133– 141.

Zeevalk, G.D., Hyndman, A.G., Nicklas, W.J., 1989. Excitatory amino acid-induced toxicity in chick retina: amino acid release, histology, and effects of chloride channel blockers. J. Neurochem. 53, 1610–1619.

Zeng, X.X., Ng, Y.K., Ling, E.A., 2000. Neuronal and microglial response in the retina of streptozotocin-induced diabetic rats. Vis. Neurosci. 17, 463– 471.

Zhang, C., Tso, M.O., 2003. Characterization of activated retinal microglia following optic axotomy. J. Neurosci. Res. 73, 840–845.

Zhang, C., Lam, T.T., Tso, M.O., 2005. Heterogeneous populations of micro-glia/macrophages in the retina and their activation after retinal ischemia and reperfusion injury. Exp. Eye Res. (Epub ahead of print).

Zhang, J., Wu, G.S., Ishimoto, S., Pararajasegaram, G., Rao, N.A., 1997. Expression of major histocompatibility complex molecules in rodent retina: Immunohistochemical study. Invest. Ophthalmol. Vis. Sci. 38, 1848–1857.