Granulocyte Colony-Stimulating Factor Protects Retinal Photoreceptor Cells against Light-Induced Damage

Abstract

purpose. Granulocyte colony stimulating factor (G-CSF) has been shown to have neuroprotective and anti-inflammatory effects in cerebral damage models. In addition, bone-marrow–derived hematopoietic cells, which can be mobilized with G-CSF, have a neuroprotective effect in hereditary retinal cell death. The present study was conducted to investigate whether G-CSF protects photoreceptors from light-induced cell death.

methods. G-CSF or vehicle was systemically injected before the light exposure and for four consecutive days after the exposure. Morphologic and electrophysiologic examinations were performed 1 week after the exposure to light. Gamma ray irradiation (6.5 Gy) was used to examine the involvement of bone marrow-derived cells increased by G-CSF injection. The expression of G-CSF receptor in the retina was analyzed by immunohistochemistry and quantitative RT-PCR.

results. The outer nuclear layer thickness was partially preserved in G-CSF–treated mice (measured at 300 μm superior from the optic disc, G-CSF: 14.9 ± 6.3 μm versus control: 6.7 ± 2.5 μm), and an electroretinogram confirmed the preservation of wave amplitudes (maximum scotopic a-wave G-CSF: 97.7 ± 48.0 μV versus control: 14.4 ± 21.9 μV, maximum scotopic b-wave G-CSF: 298.1 ± 145.3 μV versus control: 33.2 ± 50.1 μV). The effect was not lost, even with leukocyte depletion by irradiation. G-CSF receptor was expressed in retinal cells and upregulated by the light exposure (1.8-fold upregulation 2 hours after light exposure).

conclusions. G-CSF protects photoreceptor cells against light-induced damage, possibly via G-CSF receptor expressed on retinal cells. These findings may lead to a novel treatment strategy for neural degenerating diseases of the retina.

Retinal diseases characterized by photoreceptor cell death, such as retinitis pigmentosa and age-related macular degeneration, are the leading causes of blindness in developed countries. The light-induced retinal damage model is used to study the mechanisms and treatment strategies for these diseases, since the model shares the pathologic feature of photoreceptor cell death. The main pathway of light-induced photoreceptor cell death is apoptosis. ¹ The causative mechanisms of the apoptosis include activation of nitric oxide synthase, calcium overload, disturbance of mitochondrial function, and generation of reactive oxygen species. A multitude of treatment strategies (e.g., inhibition of nitric oxide synthase, use of calcium channel blocker, modulating mitochondria, or administration of antioxidants) have been performed in animal models and found to mediate beneficial effects. ²  

Granulocyte colony-stimulating factor (G-CSF) is a potent and specific growth factor for neutrophilic granulocytes. It is commercially available and used in the treatment of neutropenia. It increases granulocytes by promoting the differentiation of granulocyte precursors ³ and reducing apoptosis. ⁴ In addition to these hematopoietic effects, antiapoptotic and neuroprotective effects of G-CSF were identified in models for cerebral ischemia ^{5 6 7 8 9} and Parkinson’s disease. ¹⁰ In fact, a preliminary clinical trial investigating the effect of G-CSF for stroke patients has showed favorable results. ¹¹ The antiapoptotic effect seems to be directly mediated via the G-CSF receptor (G-CSFR) expressed on neural cells. ¹²  

G-CSF also has an anti-inflammatory effect, which is beneficial for central nervous system injury. The administration of G-CSF to an animal model after intracerebral hemorrhage reduced brain edema, inflammation, and blood–brain barrier permeability. ¹³ The effect was also seen in a stroke model ^{8 14} and in experimental encephalomyelitis. ¹⁵ The administration of G-CSF alters the expression level of several cytokines including IL-4, TGF-β1, interferon-γ, TNF-α, IL-1β, and inducible nitric oxide synthase (iNOS). ¹⁶ The regulation of these cytokines and subsequent reduction of T-cell migration to the injury site is considered to be a main factor in the anti-inflammatory effect. 

In addition, G-CSF mobilizes hematopoietic stem cells from bone marrow into peripheral blood. ¹⁷ The bone marrow–derived stem cells participate in angiogenesis in the retina. ¹⁸ Intravitreous injection of these cells provides a neuroprotective effect in the model of retinitis pigmentosa through angiogenesis. ¹⁹ A recent study even suggested the potential of these cells to differentiate into neural cells, ²⁰ and some reports have supported this notion in the recovering process of injured retina. ^{21 22}  

We hypothesized that the neuroprotective and anti-inflammatory effects seen in the cerebral model could be beneficial for the retina as well as in the central nervous system. In addition, G-CSF-mobilized hematopoietic stem cells can possibly contribute to the protection or regeneration of injured retina. We examined whether G-CSF exhibits neuroprotective/antiapoptotic effects in a light-induced retinal damage model to investigate the potential of G-CSF treatment for retinal disease with photoreceptor loss. 

Materials and Methods

Animals

Animals were treated according to the regulations in the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the Guideline for Animal Experiments of Kyoto University. All animal experiments were conducted with the approval of the Animal Research Committee, Graduate School of Medicine, Kyoto University. The BALB/c mice used in all experiments were purchased from Nihon SLC (Shizuoka, Japan). 

Light Damage

BALB/c albino mice were housed in 12-hour light–dark cycle conditions with a daytime light intensity of 12 lux. After the overnight dark adaptation, 6-week-old male mice were SC injected with vehicle or recombinant human G-CSF (Kirin, Tokyo, Japan) at a concentration of 100 or 10 μg/kg and were exposed to 5000 lux cool white fluorescent light (PL36W; Aqua System, Tokyo, Japan) for 2 hours. The light exposure was started at 10 AM. The temperature in the cage was checked and did not exceed 27°C. The mice were again housed under a 12-hour light–dark cycle. The daily injection of G-CSF or vehicle continued for the following 4 days. 

Irradiation

To examine the effect of activated bone marrow-derived cells by G-CSF treatment, the mice were irradiated using γ-ray (Nordion, Ottawa, Ontario, Canada) 5 days before light exposure. The total irradiated dose was 6.5 Gy, and the dose rate was 1.11 Gy/min. Hematologic parameters including erythrocyte, leukocyte, and platelet counts were determined with a particle counter (PCE 170; Erma Inc., Tokyo, Japan) in a preliminary experiment to confirm the effect of G-CSF and irradiation. 

Electroretinogram

Seven days after exposure to light, ERGs of the mice were recorded. After overnight dark adaptation, the mice were anesthetized with intraperitoneal injection of ketamine (50 mg/kg) and xylazine (20 mg/kg). The pupils were fully dilated with 1% tropicamide, 2.5% phenylephrine HCl, and 1% cyclopentolate HCl (Santen Pharmaceutical, Osaka, Japan). During the ERG recording, the mice were kept on a heating pad to maintain a constant body temperature. ERGs were recorded (PowerLab System 2/25; AD Instruments, Nagoya, Japan), by using a gold wire loop placed on the right cornea. Reference electrodes were placed in the cheek, and the ground leads were placed on the hip. 

Full-field scotopic ERGs were measured in response to 3-ms flashes at an intensity ranging from 0.329 to 30 cd-s/m² at 1-minute intervals without averaging multiple responses. Subsequently, photopic ERGs were recorded in response to 0.329 to 30 cd-s/m² flash and 30 cd/m² background light after 7 minutes of light adaptation, filtered between 0.3 and 500 Hz, and stored. Thirty-two ERG measurements were obtained and averaged. 

For the statistical analysis, we used scotopic ERGs recorded at intensities of 0.0127, 0.3279, 3.288, and 30 cd-s/m², and photopic ERGs recorded at 3.288 and 30 cd-s/m². The amplitudes of a-waves were measured from the baseline to the trough in the cornea-negative direction; b-waves were measured from the cornea-negative peak to the major cornea-positive trough. Photopic peak responses were measured from the baseline to the maximum cornea-positive peak. These analyses were performed with commercial software (Scope 3 ver. 3.8.2 software; AD Instruments). 

Histologic Analysis

After recording the ERGs, the mice were transcardially perfused with 4% paraformaldehyde (Wako, Osaka, Japan) and 0.25% glutaraldehyde (Wako). A marking suture was placed on the upper conjunctiva, and the eyes were enucleated, fixed with the same solution, and embedded in paraffin. Then, 2-μm-thick vertical sections were made through the optic disc. These specimens were stained with hematoxylin and eosin and then photographed with a microscope (Axio Imager; Carl Zeiss Meditec, Jena, Germany). The outer nuclear layer (ONL) thickness was measured at the superior and inferior retina 150, 300, and 600 μm from the optic disc. The analysis was performed with the microscope system software (Axiovision ver. 4.3 software; Carl Zeiss Meditec). Measurements from both eyes were averaged as one sample for the statistical analysis. 

For immunostaining, the eyes were enucleated from mice without light exposure. The eyes were fixed, sequentially immersed with 10%, 20%, and 30% sucrose, and then embedded in OCT compound (Miles, Elkhart, IN). Cryostat sections (5 μm thick) were mounted on silanized slides (Dako, Glostrup, Denmark). They were incubated at room temperature for 1 hour with 20% blocking solution (Block Ace; Dainihon-Seiyaku, Osaka, Japan) in 0.1 M PBS containing 0.03% Triton-X (Sigma-Aldrich, St. Louis, MO) to block nonspecific antibody binding. Specimens were incubated for 24 hours at 4°C with primary antibody diluted in 5% blocking solution in 0.1 M PBS. The following primary antibodies were used: rabbit anti–G-CSF (1:500; Santa Cruz Biotechnology, Santa Cruz, CA), rabbit anti–G-CSFR (1:500; Santa Cruz Biotechnology), mouse anti-rod opsin (1:5000; Sigma-Aldrich), mouse anti-calbindin (1:1000; Sigma-Aldrich), mouse anti-glutamine synthetase (1:1000; Chemicon, Temecula, CA), mouse anti-neurofilament M (1:1000; Chemicon). The secondary antibodies were anti-mouse IgG conjugated with Alexa 488 and anti-rabbit IgG conjugated with Alexa 546 (1:500, Invitrogen, Carlsbad, CA). Cell nuclei were counterstained with 4′,6-diamidino-2-phenylindole (1 μg/mL; Invitrogen). For negative controls, the isotype control of rabbit IgG (1:100; Abcam, Tokyo, Japan) was used as the primary antibody. The specimens were imaged with a laser-scanning confocal microscope (LSM 5 Pascal; Carl Zeiss Meditec). 

Reverse Transcription–Polymerase Chain Reaction

Retinas were collected from mice without light exposure or 2, 6, or 24 hours after light exposure. Total RNA was isolated (RNeasy; Qiagen, Hilden, Germany), treated with RNase-free DNase I, and reverse transcribed with a first-strand cDNA synthesis kit (GE Healthcare, Buckinghamshire, UK), according to the manufacturers’ protocols. cDNA from bone marrow was obtained and used as the positive control. The expression of G-CSFR was confirmed with conventional PCR using the following primers: G-CSFR (csf3r, accession number: NM_007782.2) forward TGTGCCCCAACCTCCAAACCA and reverse GCTAGGGGCCAGAGACAGAGACAC. Amplification was performed for 30 cycles. One cycle was 30 seconds at 95°C, 30 seconds at 60°C, and 40 seconds at 72°C. The PCR products were analyzed with 2% agarose gel electrophoresis. Quantitative real-time PCR was performed with a sequence detection system (Prism 7000; Applied Biosystems, Inc. [ABI] Foster City, CA). Reactions were performed in duplicate with SYBR Green PCR master mix (ABI). PCR probes for G-CSFR and intrinsic control β-actin were purchased from ABI. Cycling conditions were as follows: 10 minutes at 95°C, 15 seconds at 95°C, and 1 minute at 60°C for 40 cycles. Amplification plots and cycle threshold values from the exponential phase of the PCR were analyzed (Prism SDS 1.7; ABI). Relative regulation levels were determined after normalization to β-actin. 

Statistical Analysis

Statistical analysis was performed with commercial software (SPSS, Chicago, IL). Data from each group were compared by one-way ANOVA followed by the Bonferroni test or unpaired t-test, as appropriate. 

Results

The administration of G-CSF efficiently increased the number of leukocytes: baseline, 3.0 ± 0.7 × 10³/μL (n = 9); 10 μg/kg G-CSF for 5 days, 3.8 ± 1.2 × 10³/μL (n = 8, P = 0.15 compared to base line); 100 μg/kg G-CSF for 5 days, 6.0 ± 3.0 × 10³/μL (n = 9, P < 0.01). The number of erythrocytes and platelets was not significantly changed. An almost total depletion of leukocytes was confirmed 5 days after irradiation: 0.1 ± 0.1 × 10³/μL (n = 5, P = 0.01). The number of leukocytes did not recover, even after 5 days of G-CSF injection after irradiation: 12 days after irradiation, 0.1 ± 0.1 × 10³/μL (n = 5, P = 0.01). These results confirm that the G-CSF injection and irradiation worked well at the settings used. 

Light exposure–induced photoreceptor cell death occurred as previously reported. ONL and photoreceptor outer segments (OS) were markedly thinned and disorganized. The damage was more severe in the superior or central retina than in the inferior or peripheral retina. Although injection of vehicle did not show a significant effect on cell death and retinal thinning, ONL and OS were partially preserved in the mice treated with G-CSF (Fig. 1A) . Thickness measurement of the ONL confirmed the statistically significant difference (Fig. 1B) . These results indicate that G-CSF decreased photoreceptor cell loss, at least at the morphologic level. 

We used an electrophysiologic method, ERG, to determine whether there is a functional preservation. In normal mice, photopic and scotopic ERG responses increased with increasing light intensities (Fig. 2A , left). In contrast to the markedly decreased response in vehicle-treated mice (Fig. 2A , right), G-CSF–treated mice showed significantly preserved responses (Fig. 2A , middle). The amplitudes of scotopic a-waves, b-waves, and photopic responses were determined and compared, demonstrating a statistically significant difference between vehicle-treated and G-CSF–treated mice (Fig. 2B) . The result confirms the preservation of photoreceptor function as well as the photoreceptor structure. Although there was no statistically significant difference, injection of 100 μg/kg G-CSF seemed to be more effective than 10 μg/kg G-CSF. Therefore, we used a dose of 100 μg/kg in the remaining experiments. 

Next, we examined whether bone marrow-derived cells increased by G-CSF played a role in photoreceptor cell preservation. As already mentioned, peripheral leukocytes were almost totally depleted with 6.5 Gy of irradiation and did not recover by the time of enucleation. We used these irradiated mice to determine the involvement of bone marrow–derived cells. After the irradiation, G-CSF–treated mice still showed a photoreceptor protective effect compared with vehicle-treated mice (Fig. 3A) . Functional protection, as determined by ERGs, was also observed in the G-CSF–treated condition (Fig. 3B) . These results suggest that G-CSF neuroprotective effects may be mediated directly via retinal cells rather than indirectly via leukocytes. 

Since our data suggested that G-CSF works directly on retinal cells, we examined whether G-CSFR was expressed in retinal cells. Immunohistochemical analysis revealed that G-CSFR was universally expressed in retinal cells. Double staining showed that ganglion cells, bipolar cells, amacrine cells, Müller cells, and photoreceptor cells express G-CSFR, even in the normal state (Fig. 4) . 

We used RT-PCR to further examine the gene expression of G-CSFR in the retina. The expression of G-CSFR was evident and was transiently upregulated by light exposure compared with nontreated retina (1.8-fold upregulation 2 hours after exposure to light, P < 0.001; Fig. 5 ). 

Discussion

The administration of G-CSF alleviated the light-induced damage of retinal photoreceptor cells. This effect was confirmed by histologic and electrophysiologic studies. These results provide the first evidence that G-CSF has a neuroprotective effect in a retinal cell damage model as was shown in the central nervous system damage models. Our findings may be a step toward a novel treatment strategy for retinal diseases caused by photoreceptor cell loss. 

The effect of G-CSF seemed to be mediated via G-CSFR on retinal cells, since our result showed that the neuroprotective/antiapoptotic effect was not lost with leukocyte depletion. We confirmed that G-CSFR is expressed in normal adult retina and is upregulated in response to light exposure. Two hours after light exposure, the expression of G-CSFR was transiently increased 1.8-fold. This rapid response to the injury suggests a natural defense mechanism mediated by the G-CSF/G-CSFR pathway. In fact, several neurotrophic factors are upregulated in response to retinal injuries. Prolonged exposure to light increases the levels of basic fibroblast growth factor and ciliary neurotrophic factor in the retina. ²³ The upregulation of these neurotrophic factors is considered to be a cause of the preconditioning paradigm, which is the resistance to light damage exhibited by animals exposed to bright light for a short time before prolonged exposure. ²⁴ Although the intrinsic expression of G-CSF in the retina was not confirmed in the present study, the upregulation of G-CSFR may be another example of the endogenous protective system. 

The exact mechanism by which each cell responds to G-CSF/G-CSFR signal has not yet been elucidated. The phosphatidylinositol 3-kinase (PI3K)/Akt pathway, ^{12 25} Janus kinase 2 (JAK2)/ signal transducer and activator of transcription 3 (STAT3) pathway, ²⁶ and extracellular signal–regulated kinase (ERK) pathways ¹² have all been implicated in the inhibition of G-CSF–mediated apoptosis in cerebral models. Future studies are needed to elucidate the molecular and cellular mechanisms underlying the protective effect seen in the present study. 

The anti-inflammatory effect of G-CSF could also contribute to the preservation of retinal cells. Park et al. ¹³ showed that G-CSF suppresses inflammation and blood–brain barrier permeability in brain injury. This effect would decrease the secondary cell damage after light exposure. Our data also suggest the putative positive effect of leukocyte depletion: Irradiated mice showed a tendency toward better preservation of retinal cells, although it was not statistically investigated. In addition, G-CSF inhibits iNOS gene expression and decreases iNOS levels, ^{9 27} which should be beneficial in the light-induced damage model. ^{28 29} We speculate that these anti-inflammatory effects could be another explanation for the protection of retinal cells, although this idea remains to be examined by future studies. 

Hematopoietic stem cells are another possible factor in the G-CSF–mediated neuroprotective effect. The administration of G-CSF mobilizes hematopoietic stem cells from the bone marrow into peripheral blood. ³⁰ We considered it plausible that the stem cells mobilized in response to G-CSF contribute to photoreceptor survival, as was demonstrated for the neuroprotective effect in stroke ³¹ and spinal cord injury models. ³² In fact, bone marrow-derived hematopoietic stem cells have been shown to mediate a neuroprotective effect in a retinitis pigmentosa model. ¹⁹  

However, our findings suggest that the effect mediated by the stem cells was minimal in our light-induced damage model. When bone marrow-derived cells were depleted with sublethal irradiation, G-CSF still showed a protective effect. Moreover, the depletion of leukocytes itself seemed to lessen the light-induced damage. This result seems inconsistent with that of the retinitis pigmentosa model. ¹⁹ We assume that the difference lies in the apoptotic process in different models. In acute injury, such as bright-light–induced damage, the inflammatory harm from the leukocytes would overwhelm the putative protective effect from a small number of stem cells. However, despite this result, the G-CSF strategy may be effective in cases of genetic retinal degeneration. 

The results of the present study are applicable to retinal diseases. Although the effect of G-CSF was investigated in a light-induced retinal damage model instead of a retinal degeneration model or a macular degeneration model, the conditions share the main final pathway of photoreceptor apoptosis. In addition, G-CSF seems to mediate a neuroprotective effect on chronic neurodegenerative diseases, as shown in a Parkinson’s disease model, ¹⁰ as well as in cases of stroke/cerebral ischemia, as shown in an acute damage model. This evidence implies that the neuroprotective effect in the acute retinal damage model can be applied to chronic retinal degenerative diseases, including retinitis pigmentosa. 

Investigation of the neuroprotective effect of G-CSF in other ocular diseases such as glaucoma can be rationalized, since G-CSFR is expressed in ganglion cells (Fig. 4) . Glaucoma, which is characterized by ganglion cell death, is a major cause of blindness, and the prevention of disease progression is of great clinical importance. 

A combination of other hematopoietic factors would be a rational strategy to increase the effect of G-CSF. The nonhematopoietic roles of hematopoietic factors are currently being examined by many researchers. In addition to G-CSF, erythropoietin is also recognized as a neuroprotectant; in fact, erythropoietin has a neuroprotective effect in light-induced retinal damage, ^{33 34} as well as CNS disease models. Granulocyte–monocyte colony stimulating factor (GM-CSF) has also been shown to have a neuroprotective effect in ischemic cerebral injury. ³⁵ Since these cytokines seem to activate different antiapoptotic cascades, ^{5 12 35 36 37} the combination of these cytokines may have an additive effect. Further studies are needed to examine the safety and additive effect of combination therapy. 

Finally, we showed the beneficial effect of G-CSF on light-induced retinal damage. G-CSF has already been proven safe for clinical use and has been extensively used for more than 10 years. Systemic administration of G-CSF is expected to sufficiently activate this neuroprotective pathway since G-CSF passes across the intact blood–brain barrier. ^{12 38} Further in vivo studies are needed to confirm the efficacy of the therapy in isolation or in combination with other hematopoietic cytokines. 

 Presented at the annual meeting of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Florida, May 2008.

 Supported by Grant 17689045 from the Ministry of Education, Science, and Culture of Japan.

 Submitted for publication January 9, 2008; revised May 9, June 26, and July 21, 2008; accepted October 3, 2008.

 Disclosure: A. Oishi, None; A. Otani, None; M. Sasahara, None; H. Kojima, None; H. Nakamura, None; Y. Yodoi, None; N. Yoshimura, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

(A) Photographs of mouse retina. Left: untreated control mouse; middle: G-CSF–treated mouse; right: vehicle-treated mouse. (B) ONL retinal thickness. G-CSF–treated mice had thicker ONL than did vehicle-treated mice.

Figure 2.

 

(A) Representative results of scotopic and photopic ERG measured 1 week after light exposure. Left: normal untreated mouse; middle: G-CSF–treated mouse; right: vehicle-treated mouse. (B) Quantitative results of each component of scotopic and photopic ERGs. ERG amplitudes were retained in G-CSF–treated mice to a greater extent than in vehicle-treated mice.

Figure 3.

 

ONL thickness (A) and ERG amplitudes (B) of mice 1 week after light exposure. Leukocytes were depleted by irradiation 5 days before the light exposure. The G-CSF–treated mice exhibited better preservation of ONL both morphologically and electrophysiologically.

Figure 4.

 

Double staining confirmed that the G-CSF receptor was expressed on ganglion cells (top left), bipolar cells (middle left), horizontal and amacrine cells (bottom left), Müller cells (top right), and photoreceptor cells (bottom right).

Figure 5.

 

G-CSFR was expressed in normal retina. (A) The expression level of G-CSFR was transiently increased in light-exposed retina compared with normal control (B).