Expression of Proteoglycan Decorin in Neural Retina

Abstract

purpose. To identify the expression of chondroitin/dermatan sulfate proteoglycan decorin in retina and to elucidate its changes during development and ischemia–reperfusion.

methods. Expression of decorin in rat retina was investigated by reverse transcription–polymerase chain reaction (RT-PCR) and immunohistochemistry. Distributional changes during development and transient ischemia in model eyes also were investigated by immunohistochemical experiments.

results. Gene expression of decorin core protein was identified in rat retina by RT-PCR. Decorin immunoreactivities were shown throughout the retina, especially in the ganglion cell layer. In developing rat retinas, at embryonic stages (embryonic day 16), decorin was distributed uniformly throughout the retina. As retina matured, the intensity of decorin immunostaining in retinal inner layers and retinal pigment epithelium increased. Furthermore, in experimental transient retinal ischemia, after transient downregulation of the decorin core protein gene between 24 and 48 hours after the ischemia, recovered (or increased) expression was shown by semiquantitative RT-PCR experiments. Immunohistochemical studies revealed strong decorin immunoreactivities in the damaged inner layers 1 week later.

conclusions. The expression of decorin was identified in adult and developing rat retina. The distributional changes of decorin during the retinal development suggest that this proteoglycan may play a role in the differentiation of retinal ganglion cells. Moreover, in rat ischemia–reperfusion models, the alterations in gene expression and immunohistochemical localization showed the contribution of this proteoglycan to the damage and repair processes in diseased retina.

Proteoglycans are composed of a core protein molecule to which glycosaminoglycans (GAGs) are covalently linked as side chains. ^{1 2} Until recently, proteoglycans were classified into groups based on their GAG side chains. Varner et al. ³ and Tawara et al. ^{4 5} have clearly demonstrated, by histochemical and biochemical studies, the presence of chondroitin sulfate and heparan sulfate in mammalian retinas, especially in interphotoreceptor matrix. ^{3 4 5} Recent molecular biologic studies have characterized that the molecular structure of a number of proteoglycan core proteins ¹ and additionally have shown that a diverse set of proteoglycans are present and are precisely regulated in the developing brain. ^{6 7} Proteoglycans specifically bind many cell surface molecules and extracellular matrix molecules that are involved in various developmental events in the brain. ^{8 9 10} These phenomena include proliferation and migration of neuroblasts, neurite outgrowth, and formation of specific synapses. Additionally, in neural retina, it has been hypothesized that alterations in the expression of proteoglycans are involved in a number of pathologic conditions of the ocular fundus. For example, aberrant expression of proteoglycans has been reported in retinal tissues in cases of retinal degeneration, ^{11 12} retinitis pigmentosa, ¹³ age-related macular degeneration, ¹⁴ and experimental myopia. ¹⁵ Although many studies on the relationships between proteoglycans and numerous retinal diseases have been reported, molecular biologic studies for expressional regulation of proteoglycan core proteins in normal and pathologic retinas ^{16 17 18} are more limited. Moreover, there are only a few reports on the identification of proteoglycan core proteins expressed in retina. ^{19 20 21} Decorin, also known as PG II, is a chondroitin/dermatan sulfate proteoglycan and has been isolated from mammalian connective tissues. ²² The mature proteoglycan consists of a 36-kDa core protein, ²³ a single GAG chain located near the amino terminus, ²⁴ and two to three asparagine-linked oligosaccharides. ^{25 26} It is one of the major small proteoglycans related to cell proliferation and extracellular matrix assembly. ^{27 28} In addition, decorin has been reported to play a role in the regeneration of nervous tissues in central nervous system injury. ²⁹ Herein, we report gene expression and localization of decorin core protein in retina and show alterations in the distribution of decorin in the developing retina and a transient retinal ischemia model. 

Materials and Methods

Isolation of Retinal RNA and Synthesis of cDNA

Male Wistar rats (6 weeks after birth) were anesthetized by diethyl ether and intraperitoneal injection of pentobarbital (50 mg/kg). After enucleation of the eyeballs, neural retina was removed by scissors and forceps guided by an operating microscope. Enucleated animals were killed by overdose injection of pentobarbital. All animals were given water and food ad libitum. All studies were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Total RNAs were isolated from the retina by the acid guanidium thiocyanate-phenol chloroform extraction methods. ³⁰ The extracted RNAs were used to synthesize template cDNAs for subsequent reverse transcription-initiated polymerase chain reaction (RT-PCR) experiments with the use of reverse transcriptase (SuperScript RNase H-Reverse Transcriptase; Gibco, Life Technologies, Gaithersburg, MD). Each cDNA sample was synthesized from 5 μg of the extracted total RNA. 

RT-PCR Experiments

PCR was performed by the method of Saiki et al., ³¹ modified slightly, as described previously. ³² cDNA preparation (1 μl) was added to the following mixture (Gene Taq, Nippon Gene, Toyama, Japan): 1 μl (5 U) Taq DNA polymerase, 4 μl 2.5 mM dNTP mixture, 5 μl Taq universal buffer, 2 μl 10 picomole/μl sense primer, 2 μl 10 picomole/μl antisense primer, and 35.5 μl distilled water. For PCR experiments, the following conditions were used: denaturation at 95^oC for 30 seconds, annealing at 65^oC for 30 seconds, and polymerization at 72^oC for 1 minute for 35 cycles. After the reaction, 9 μl of the PCR product was mixed with 1 μl loading buffer. The mixture was separated by 2% agarose gel electrophoresis. Primers used in this study were designed from reported sequences for decorin genes. ^{33 34} Each sequence of the sense primer and the antisense primer was 5′-TTGCAGGGAATGAAGGGTCT-3′ and 5′-TGTGGGTGAATTTGCCAATA-3′, respectively. Possible contamination of the genomic sequence was examined by control experiments without the use of reverse transcriptase. Rat brain RNAs were used as positive controls, because the expression of mRNA for decorin had been reported. ³⁵ In Southern blot analysis, the electrophoresed PCR samples were transferred to a nylon filter (Amersham, Buckinghamshire, UK) by the capillary transfer method with 20× SSC. Synthesized internal oligonucleotide probe (the sequence; 5′-CCTCAAGGTCTGCCCACTTC-3′) was labeled and used for enhanced chemiluminescence 3′-oligolabeling and detection (Amersham) in an attempt to confirm the identity of the PCR products. 

In semiquantitative RT-PCR experiments, cDNA preparations, which were created from 5 μg of the total retinal RNAs, were normalized toβ -actin gene expression using PCR experiments with 2 μCi radiolabeled dCTP, as described previously. ³⁶ The specific primers for β-actin were 5′-AGCTGAGAGGGAAATCGTGC-3′ (sense) and 5′-ACCAGACAGCACTGTGTTGG-3′ (antisense). ³⁷ The following conditions for PCR experiments were used: denaturation at 95^oC for 30 seconds, annealing at 65^oC for 30 seconds, and polymerization at 72^oC for 1 minute for 25 cycles (β-actin primers) or 30 cycles (decorin primers). After the reaction, 2% agarose gel electrophoresis and subsequent Southern blot analysis were performed, as described earlier. Loading buffer (1 μl) was added to 9μ l PCR sample, and then applied to each lane of the gel. To investigate relative levels of decorin gene expression, PCR products of a series of diluted standard samples (PCR products from cDNA of adult normal rat retina) were applied on the same gel. Densities of hybridizing bands of Southern blot analysis were measured (Image 1.59, National Institutes of Health, Bethesda, MD). A standard curve was generated from the hybridizing bands of PCR products amplified from serial dilutions of template cDNAs, and the semiquantitative concentrations of the PCR products from cDNA samples were determined from the standard curve. 

Preparation of Tissue for Immunohistochemistry

Postnatal Wistar rats were anesthetized by diethyl ether and intraperitoneal injection of pentobarbital. Embryonic rats were extracted from uteruses after the pregnant rats were killed, followed by decapitation and enucleation. All postnatal animals were killed by overdose injection of pentobarbital. After the anesthesia of pentobarbital (50 mg/kg), adult rats (male Wistar rat 6 weeks after birth) were perfusion fixed by 4% paraformaldehyde in phosphate-buffered saline (PBS) before enucleation. The enucleated eyes were fixed with gentle shaking for 2 hours at 4°C with 4% paraformaldehyde in PBS, and washed for 5 minutes in PBS, then gently shaken overnight at 4°C in 30% sucrose-PBS before freezing on powdered dry ice. Sections (16 μm) were cut using a cryostat, collected onto silanized slides (Dako Japan, Kyoto, Japan). 

Antibodies

The polyclonal rabbit anti-decorin core protein antiserum used in this study has been described elsewhere in detail. ^{25 38} The specificity of the antibody for decorin has already been confirmed in a previous report. ^{25 39} The polyclonal rabbit anti-glial fibrillary acidic protein (GFAP) antiserum and the monoclonal mouse anti-neurofilament IgG₁ were purchased from Dako. It is known that horizontal cells and the optic axons in ganglion cells are stained by the anti-neurofilament antibody that reacts with phosphorylated neurofilament proteins. ^{40 41} The polyclonal rabbit anti-S-100β, the marker protein of retinal glial cells, antiserum was a generous gift from Dr. Masayo Takahashi. All antibodies were diluted with 0.5% bovine serum albumin (BSA) in PBS. The dilution of all antibodies was 500:1. 

Immunohistochemistry

Retinal sections were circled with a PAP pen (Dako Pen, Dako) to form a hydrophobic barrier to solutions and were fixed for 30 minutes in 3% paraformaldehyde and 1% sucrose in PBS, then rinsed for 3 minutes twice in PBS. Slides were covered with 3% hydrogen peroxide in PBS, followed by washing for 3 minutes twice in PBS. Sections were covered with 50 mM glycine in PBS, then rinsed for 3 minutes in PBS before each slide was covered for 1 hour with blocking solution (2% BSA, 2% horse normal antiserum, and 2% goat normal antiserum in PBS). After slides were washed for 3 minutes in PBS, sections were incubated overnight at 4°C with primary antibody. Sections were rinsed three times for 3 minutes in PBS after removal of the primary antibody. Sections were then incubated for 1 hour at room temperature with the secondary antibody; biotinylated anti-rabbit IgG (Vector Laboratories, Burlingame, CA) for polyclonal primary antibody, or rat-absorbed biotinylated anti-mouse IgG (Vector) for a monoclonal primary antibody, and washed (using the same procedure as for the primary antibody). Slides were covered for 45 minutes at room temperature with avidin DH. and biotinylated horseradish peroxidase H reagents (Vectastain Elite ABC Kit, Vector). After washing the slides three times for 3 minutes in PBS, diaminobenzidine tetrahydrochloride (DAB; Dako) was used for staining the sections. The retinal sections for each comparison were immunolabeled during the same experiment; DAB substrate incubation time in all developmental retinal sections was 3 minutes. 

For immunohistochemistry using fluorescein-conjugated secondary antibodies, slides were incubated in cold methanol (−20°C) for 15 minutes after fixation for 30 minutes in 3% paraformaldehyde and 1% sucrose in PBS, followed by washing for 3 minutes twice in PBS. Then, sections were covered for 15 minutes with 50 mM glycine in PBS followed by the blocking procedure and the incubation of primary antibody, as described earlier. In the incubation of secondary antibody, rat-absorbed fluorescein-conjugated horse anti-mouse IgG and fluorescein-conjugated goat anti-rabbit IgG (both from Vector) were used for the primary monoclonal antibody and the primary polyclonal antibody, respectively. After incubation with the secondary antibody, sections were washed six times for 3 minutes each in PBS, then mounted (Vectashield, Vector), and the slides examined under a confocal microscope (model LSM410; Carl Zeiss, Oberkochen, Germany). 

Western Blot Analysis

Western blot analysis for decorin and sample preparation were performed by a slight modification of the method described previously. ⁴² Briefly, the retinal tissue (42 mg wet weight) from adult rats was extracted with 400 μl extraction buffer (20 mM Tris-HCl buffer [pH 7.2], containing 7 M urea, 1% Triton X-100, 0.15 M sodium chloride and protease inhibitors) by end-over-end rotation for 18 hours at 4°C. The extract (320 μl) was applied to a 250-μl column of diethylaminoethyl (DEAE)-trisacryl M equilibrated with the extraction buffer. The column was washed with the urea-free extraction buffer. The breakthrough was collected. Bound material was desorbed by applying 750 μl urea-free extraction buffer containing 1.0 M sodium chloride. The breakthrough and the bound material (375μ l each) were precipitated with 375 μl methanol and 95 μl chloroform. The upper phase was removed, and the chloroform phase was washed with 50% methanol and dried. The pellets were dissolved in 20μ l chondroitinase buffer and treated with 50 mU chondroitinase ABC at 37°C for 2 hours. The samples were subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) in a 12.5% acrylamide gel. The separated proteins were electrotransferred onto a nitrocellulose membrane. After blocking the membrane with 3% casein and 0.002% Tween 20 in Tris-buffered saline (TBS; pH 7.4) at 4°C for 16 hours, the membrane was incubated in a 1:500 dilution of anti-decorin antibody for 2 hours at room temperature, then washed with TBS for 5 minutes five times. The blot membrane was treated with a 1:1000 dilution of peroxidase-conjugated goat anti-rabbit IgG antibody (Bio-Rad, Munich, Germany). After washing the membrane with TBS for 5 minutes five times, the membrane was stained with DAB solution. 

Rat Retinal Ischemia Model

As described previously, ³⁶ retinal ischemia and reperfusion were induced using the method of Stefánsson et al. ⁴³ Briefly, male Wistar rats (6 weeks after birth) were anesthetized by diethyl ether and intramuscular injection of xylazine (total 2 mg) and ketamine (total 5 mg). After conjunctival peritomy, the optic nerve was exposed by blunt dissection. A 6-0 nylon suture was passed behind the surgically exposed optic nerve sheath, then tightened under direct observation by an operating microscope. The suture was removed after 2 hours, and reperfusion of the retinal vessels was confirmed subsequently. During the 2-hour constriction of the optic nerve sheath, additional xylazine (1 mg) and ketamine (2.5 mg) were injected to continue the anesthesia. To perform the molecular biologic analyses, after 2 hours of retinal ischemia, , rats were killed by overdose injections of pentobarbital, and the eyes were enucleated at 0, 3, 6, 24, 48, and 96 hours and 1 week. 

Results

RT-PCR Experiments

RT-PCR experiments using decorin-specific primers showed that cDNA fragments of the expected length (440 bp) were amplified in experiments using rat retina and brain cDNAs, whereas control PCR experiments without the use of reverse transcriptase did not reveal any bands (Fig. 1 A). Southern blot analysis using an internal oligonucleotide probe was performed in an attempt to confirm the origin of the products amplified by RT-PCR experiments (Fig. 1B) . It showed that the amplified products had the expected lengths hybridized with internal probes, which indicated that they were derived from the expected rat decorin core protein sequences. Additionally, the subcloned PCR product for the decorin probe was sequenced by the dideoxynucleotide chain–termination procedure, ⁴⁴ and the product was confirmed to be a part of the decorin exon gene sequence (data not shown). Taken together, the results clearly indicated the existence of gene expression for decorin core protein in neural retina and brain. 

Immunohistochemical Studies

Immunohistochemical studies demonstrated that decorin core protein immunoreactivities were present throughout rat retinal tissue from retinal pigment epithelium (RPE) to inner layers. Among the retinal layers, strong decorin immunoreactivities were observed in retinal inner layers such as the nerve fiber layer (NFL) and ganglion cell layer (GCL). Also, moderate immunoreactivities were shown in the inner plexiform layer (IPL) and inner nuclear layer (INL). In contrast, decorin immunoreactivities in the outer nuclear layer (ONL) were relatively faint (Fig. 2) . In summary, immunohistochemical studies showed that decorin expression was much more enhanced in inner retinal layers than in outer layers. The same procedure was performed using normal rabbit IgG instead of anti-decorin antibody to confirm that the decorin immunoreactivities were not false positive. No immunoreactivities were observed in the retinal section. 

Western Blot Analysis for Decorin in Adult Rat Retina

To characterize some molecular properties in decorin retinal tissue, Western blot analyses were performed. All immunoreactive material was able to bind to an anionic exchange resin, thus providing evidence of the absence of GAG-free core protein in the tissue (Fig. 3) . When the proteoglycan sample was treated with chondroitinase ABC, a single protein band of approximately 48 kDa could be observed that agreed with the expected molecular weight of decorin core protein containing three asparagine-bound oligosaccharides, as described previously. ²⁵ Additionally, another band of 30 kDa was detected. 

Distributional Changes in Decorin Immunoreactivities during Rat Retinal Development

In an effort to elucidate the relationship between decorin expression and retinal differentiation, distributional changes in decorin immunoreactivities were investigated. On embryonic day (E)16, homogeneous retinal cells of similar appearance were observed throughout the retina. In early postnatal stages (postnatal days[ P]0–7), inner retinal cells appeared to become differentiated from the homogeneous retinal cells and composed conspicuous layers, which were possibly consistent with GCL and IPL. Formation of outer retinal layers was shown between P14 and P21 (Fig. 4 A). Neurofilament immunoreactivity was not found in early developmental stages from E16 to P7, but it appeared in the NFL on P14. Afterward, in addition to more intensive immunoreactivities in the NFL, the OPL also became slightly positive for neurofilament immunostaining on P21 (Fig. 4B) . 

Moderate decorin immunoreactivities were distributed from RPE to NFL on E16 (Fig. 4C) . The decorin immunoreactivities were homogeneous in retinas at the early developmental stage, and there were no significant spatial changes in decorin expression in neural retina. At the time of birth and in early postnatal stages, on P0 and P7, as ganglion cells became differentiated, intensities of immunoreactivities in inner layers were increased, and localization of positive immunostaining in these layers became more conspicuous. On P14, neurofilament immunoreactivities were found in the NFL, which implies that the axons of ganglion cells had matured (Fig. 4B) . At the same time, decorin immunoreactivities were found intensively in the NFL and GCL, but not in the ONL (Fig. 4C) . On P21, when the morphologic appearance of retinal layers had already matured, immunohistochemical studies showed a similar distribution of decorin immunoreactivities to that seen in adult rat eyes. Thus, at this late developmental stage, strong decorin immunoreactivities were observed intensively in the innermost layers, such as the NFL and GCL; moderate immunoreactivities in the IPL, INL, the rod–cone layer, and RPE; but only faint immunoreactivities in the ONL. 

Semiquantitative Examination of Decorin Gene Expression in Retinal Ischemia

The data obtained from a series of immunohistochemical studies on decorin in normal rat retinas showed its distribution in the innermost retinal layers in both mature and developing retinas, as we have described. Thus, in an effort to investigate the possible roles of decorin in the survival and/or regenerating processes in these layers of diseased retinas, we created a transient retinal ischemia model in rat eyes with the use of ligation of the optic nerve, to serve as an experimental animal model of severe damage in the inner retinal layers. PCR experiments for decorin core protein were conducted using rat ischemic retina cDNAs as templates. The PCR experiments showed that, even after normalization to β-actin gene expression, the gene expression for decorin decreased after transient ischemia and recovered after 4 days to 1 week (Fig. 5) . With the use of Southern blot analysis and subsequent densitometric analysis, the relative expression of decorin was determined in ischemic retina. Semiquantitative RT-PCR and Southern blot analysis using normalization to β-actin showed that the mean levels (± SE) of gene expression for decorin were 1.06 ± 0.61-fold and 1.11 ± 0.08-fold compared with the control (normal) retinas, respectively, immediately and at 3 hours after the cessation of transient ischemia (2 hours). Thus, the time course of decorin gene expression after transient retinal ischemia did not show any significant changes immediately after reperfusion. After that, it showed that decorin gene expression decreased from 6 to 48 hours after cessation of ischemia, reached a minimum at 24 hours (0.20 ± 0.13-fold), and then recovered to near-normal or increased levels over 1 week (1.30 ± 0.38-fold; Fig. 6 ). 

Immunohistochemical Studies on Decorin in Rat Transient Retinal Ischemia

To ascertain which layers have relationships with the changes in gene expression for decorin core protein in transient retinal ischemia, immunohistochemical studies on decorin core protein were performed. As described in a previous report, ³⁶ inner retinal layers were seriously damaged in the experimental eyes with transient retinal ischemia. The time course of morphologic changes in the ischemic retinas showed that ganglion cell disappearance occurred early (12–24 hours), and was followed by the loss of retinal cells in the INL (48–168 hours), whereas the outer layers in the retina remained almost intact for 1 week. ³⁶ In retinas early in the reperfusion time (within 24 hours), decorin immunoreactivities were found in the same pattern as in control (normal) retina. However, in retinas during the late stage (168 hours after reperfusion) when serious cell loss of inner retinal cells was observed, inner retinal tissue showed strong immunoreactivities for decorin (Fig. 7 A). In an effort to identify retinal cell type related to the increased decorin expression in the retinas in the late reperfusion stage, immunohistochemical studies using anti-retinal cell-marker antibodies were performed. Among the cell type markers, strong immunoreactivities for S-100β, a specific marker for glial cells, were found in partial cells (astrocytes and Müller cells) present in the GCL and INL in the retina 24 hours after reperfusion, whereas they also were stained in control (normal) retina but were fainter than in the retina 24 hours after reperfusion (Fig. 7B) . Moreover, strong immunoreactivities for S-100β were also detected in inner retinal layers with serious cell loss in the late stage (168 hours after reperfusion). In contrast, immunoreactivities for another cell marker, GFAP, which is expressed from activated glial cells, were hardly shown in normal retina, although glial cells across the retinal layers (Müller cells) and retinal glial cells in inner layers (astrocytes) were stained considerably at 24 hours’ reperfusion (Fig. 7C) . At 168 hours after reperfusion, strong GFAP immunoreactivities were detected throughout damaged GCL, IPL, and INL. Additionally, neurofilament immunoreactivities by the anti-neurofilament antibody that reacts with phosphorylated neurofilament proteins were detected in NFL and OPL in normal retina and 24 hours after reperfusion, which implies that horizontal cells and the optic axons of ganglion cells almost remained. On the contrary, at 168 hours after reperfusion, immunoreactivities for neurofilament were found in OPL but not in NFL (Fig. 7D) . Taken together, in inner retinal layers with increased decorin immunoreactivities at 168 hours after reperfusion, glial cell marker immunoreactivities were increased despite the cell loss in the inner layers. 

Discussion

Proteoglycans play pivotal roles in various cellular processes such as proliferation, migration, cell–substratum adhesion, and survival. ¹ Their influence on neurite extension ^{7 45 46} makes them potential candidates for interactions with growing axons during development and after injury of the mammalian central nervous system. Classic histochemical and biochemical studies have shown the presence of several types of GAGs in mammalian retina. For example, Varner et al. ³ and Tawara et al. ^{4 5} reported the presence of heparan sulfate and chondroitin sulfate proteoglycans in mammalian interphotoreceptor matrices. ^{3 4 5} Proteoglycan has been shown to play an important role in neurite outgrowth of retinal ganglion cells ^{47 48 49} and to be a neurotrophic factor for retinal cells. ⁵⁰ Moreover, altered expression of proteoglycans has been reported in retinas with degenerative diseases, ^{11 12 13 14} angiogenic and proliferative diseases, ¹⁸ and enlargement of the eye. ¹⁵ Changes in the expression of retinal proteoglycans have been hypothesized to be involved in a number of pathologic events that occur in the ocular fundus. Thus, in the present study, we investigated expression of proteoglycans in pathologic and developmental situations in an attempt to elucidate their roles. 

Our molecular biologic studies showed the presence of mRNA expression for decorin core protein in rat retinas and brains. In addition, immunohistochemical studies have shown the existence of this core protein in rat retinas, which is consistent with the results from RT-PCR and subsequent Southern blot analysis (or sequencing of the amplified products). The specificity of polyclonal anti-decorin antibody in this study has already been shown in other tissues except retina, as described previously. ^{25 39} Our western blot analysis showed the presence of a core protein of approximately 48 kDa, as expected for decorin. Moreover, a 30-kDa core protein fragment was present in the same sample. Because the retention of this fragment on diethylaminoethyl-trisacryl M proves the presence of a GAG chain, it is likely that it was generated by limited proteolysis in the C-terminal part of the decorin core protein. The presence of similar fragments in tissue extracts has been described before. ^{42 51}  

Our results indicated strong decorin immunoreactivities in the inner retina, especially in the GCL and NFL, in comparison with that in the outer retina. Previous investigation has shown that chondroitin sulfate proteoglycans can be neurotrophic factors for retinal ganglion cells, ⁵⁰ suggesting that decorin, a chondroitin/dermatan sulfate proteoglycan, may be one of the neurotrophic factors surrounding retinal ganglion cells and their neurites. The expression of decorin in the GCL and NFL becomes more conspicuous after birth as the inner retinal cells make progress in their differentiation. The temporal and spatial colocalization of decorin with retinal ganglion cells and their own neurites suggests that the expression of this proteoglycan may also be related to the differentiation of ganglion cells in addition to survival of retinal cells. Additional studies of expression of neurofilament protein supported this hypothesis, because the intensity of decorin immunoreactivities around retinal ganglion cells increased at P14 when the expression of neurofilament appeared in the NFL. So far, the expression of proteoglycan core protein in mammalian retina has not been elucidated enough, although some proteoglycans, such as versican and agrin, have been investigated in chick embryonic retinas. ^{17 20} To our knowledge, this study is the first report on mRNA and protein expression of decorin core protein in mammalian retina. 

Our results from semiquantitative RT-PCR studies show that mRNA expression of decorin changed after transient retinal ischemia. Alteration in mRNA expression included no apparent change in the very early stages (within 3 hours after reperfusion), a decrease in the early stages (6–48 hours after reperfusion), and recovery to near-normal or increased levels during the late stage (4 days–1 week after reperfusion). That there was no significant change in mRNA expression for decorin in the very early stages (within 3 hours after reperfusion) suggests that alteration in expression of this proteoglycan may not relate to ischemia itself but to reperfusion after its cessation. After that, decreased mRNA expression for decorin may reflect damage and subsequent cell death (disappearance) in most retinal cells in inner layers where decorin is produced and/or deposited, because retinal cell loss in the inner layers was observed in the early stage (after 24 hours). One of the most important points to be discussed may be the question of how and why the decorin mRNA expression can recover during the late stage (4 days–1 week). The partial answer to this question may be found in the changed cell components in the diseased retinas by reperfusion after transient retinal ischemia. Our immunohistochemical study using retinal cell markers showed that the population of glial cells among the surviving cells increased in inner retinal layers during the late stage. This hypothesis seems to be contrary to the important association of retinal ganglion cells with decorin expression during retinal development, as described earlier. Retinal ganglion cell loss was observed between 6 and 48 hours after reperfusion, which agrees with decreased mRNA expression for decorin. Thus, a responsive increase in mRNA expression for decorin appears to be derived from the increased population of another decorin-expressing retinal cell type (glial cells) in a pathologic situation. 

Furthermore, it may be possible that proliferation and activation of retinal glial cells exert another influence on the expression of decorin, because numerous cytokines can be secreted by activated retinal glial cells in addition to inflammatory cells. ⁵³ In transient retinal ischemia models, however, increased expression of numerous cytokines such as interleukin-1 and transforming growth factor-β has been shown. ^{36 53} These cytokines have been known to upregulate numerous extracellular matrices including decorin. ^{54 55} Taken together, the possible second answer to the question of decorin upregulation may be changed cellular behavior in response to secreted cytokines in retinas with reperfusion status. Some investigators have suggested that increased expression of decorin may relate to regenerative processes in the damaged nervous tissues. ^{29 56} Hagedorn et al. ⁵⁷ have reported that decorin immunoreactivities are detected in epiretinal membrane in proliferative vitreoretinopathy and proliferative diabetic retinopathy. Thus, increased decorin expression in inner layers in the late stage of reperfusion after transient retinal ischemia may be part of the repair and/or regeneration mechanism in the retina. 

In conclusion, in our studies decorin core protein was expressed in retina, and its expression was altered temporally and spatially in retinal development and ischemia–reperfusion situations. 

 Reprint requests: Hidenobu Tanihara, the Section of Ophthalmology, Tenri Hospital, 200 Mishima-cho, Tenri, Nara, 632-8552 Japan.

 Supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Science, Sports and Culture, Japan.

 Submitted for publication June 30, 1998; revised January 28, 1999; accepted March 23, 1999.

 Proprietary interest category: N.

 

Figure 1.

 

RT-PCR experiments and Southern blot analysis of decorin. (A) RT-PCR experiments. (B) Southern blot analysis. Lane 1, adult rat retina cDNA; lane 2, adult rat retina cDNA without the use of reverse transcriptase (negative control); lane 3, rat brain cDNA (positive control); lane M, a marker. HincII-digested φX174 DNA was used as a marker. The arrow shows positive bands of the expected lengths (440 bp).

Figure 2.

 

Immunohistochemistry of rat adult retina using anti-rat decorin antibody. Strong decorin immunoreactivities were observed in retinal inner layers such as the NFL and GCL. Moderate immunoreactivities were detected in the IPL and INL. In contrast, decorin immunoreactivities in the ONL were relatively faint. RC, rod–cone layer.

Figure 3.

 

Western blotting for decorin in adult rat retina. A single band of approximately 48 kDa was detected on digestion with chondroitinase ABC in the material bound to diethylaminoethyl-trisacryl M (open arrowhead). Another band (30 kDa) was also shown by the same procedure (closed arrowhead). Note that the appearance of both bands required chondroitinase ABC treatment and that immunoreactive material was absent in the breakthrough (unbound) fraction from the ion exchange column. ABC+ or −, with or without chondroitinase ABC treatment; Bound, the materials bound to the column of diethylaminoethyl-trisacryl M; Unbound, the breakthrough fraction.

Figure 4.

 

Expression of decorin in developing rat retinas. (A) On E16, homogenous retinal neuroblast cells were observed throughout the retina, although the NFL and GCL were slightly separated from the other layers. From P0 to P7, GCL and IPL were formed. Formation of outer retinal layers was shown between P14 and P21. Hematoxylin–eosin staining. (B) Immunohistochemistry against neurofilament; neurofilament immunoreactivity was negative in early developmental stages from E16 to P7, but positive in the NFL on P14. In addition to more intensive immunoreactivities in the NFL, the OPL also became slightly positive for neurofilament immunostaining on P21. (C) Immunohistochemistry against decorin: Decorin immunoreactivities were homogenous in retinas on E16. On P0 and P7, intensities of immunoreactivities in inner layers were increased. On P14, decorin immunoreactivities were found intensively in the NFL and GCL, but not in the ONL. On P21, the distribution of decorin immunoreactivities was similar to that in adult rat retinas. RC, rod–cone layer.

Figure 5.

 

Representative PCR experiments and Southern blot analyses of decorin gene expression after transient retinal ischemia. The gene expression for decorin decreased after transient ischemia and recovered after 4 days to 1 week. (A) Southern blot analysis of decorin. (B) PCR experiments on decorin. (C) PCR experiments on β-actin. Lane N, normal retina; lane 1, retina immediately after cessation of 2 hours of ischemia; lane 2, after 3 hours; lane 3, after 6 hours; lane 4, after 1 day; lane 5, after 2 days; lane 6, after 4 days; lane 7, after 1 week of reperfusion.

Figure 6.

 

The time-course of decorin gene expression. The relative expression (mean ± SEM) of decorin in ischemic retina was evaluated by Southern blot analysis and subsequent densitometric analysis. Decorin gene expression decreased from 6 to 48 hours after cessation of ischemia, reached a minimum at 24 hours; 0.20 ± 0.13 (the mean levels ± SE) folds relative to the mean of controls, and then recovered to near-normal or increased levels over 1 week (1.30 ± 0.38 folds). Error bars, SE. *P < 0.05 by unpaired Student’s t-test, 24 hours and 48 hours after reperfusion versus control and 168 hours after reperfusion.

Figure 7.

 

Immunohistochemistry in rat retinal ischemia. (A) Immunohistochemical analysis of decorin in rat retinal ischemia. In the retinas at 168 hours after reperfusion, the thickness of the inner layers had decreased. Decorin immunoreactivities showed no significant changes even after reperfusion of 24 hours in comparison with normal retina. The inner layers at 168 hours after reperfusion showed strong immunoreactivities of decorin (*). (B) S-100β immunostaining in rat retinal ischemia. Retinal glial cells in GCL (closed arrowheads) and INL (open arrow heads) were especially stained on normal retina. Increased immunoreactivities in GCL and INL were shown at 24 hours after reperfusion. Damaged inner layers at 168 hours after reperfusion, where cell loss was observed, showed strong immunostaining of S-100β (*). (C) GFAP immunostaining in rat retinal ischemia. Immunoreactivities were hardly found in normal retina, whereas retinal glial cells in inner layers (closed arrowheads) and across the retinal layers (open arrowheads) were dramatically stained at 24 hours’ reperfusion. At 168 hours after reperfusion, strong GFAP immunoreactivities were detected throughout damaged inner layers (*). (D) Neurofilament (NF) immunostaining in rat retinal ischemia. Because phosphorylated neurofilament is expressed in horizontal cells and the optic axons of ganglion cells in normal rat retina, the NFL and OPL were positive in normal rat retina. Note that immunoreactivities in the NFL at 168 hours after reperfusion disappeared.