Retinal Microglia Differentially Express Phenotypic Markers of Antigen-Presenting Cells In Vitro

Abstract

purpose. Retinal microglial cells of newborn Lewis rats were isolated and cultured, and the effect of macrophage colony-stimulating factor (M-CSF), granulocyte–macrophage colony-stimulating factor (GM-CSF), and interferon-γ (IFN-γ) on microglial expression of the accessory molecules required for antigen presentation were studied.

methods. Retinal microglia were isolated from newborn Lewis rats and cultured in media supplemented with either M-CSF or GM-CSF. Immunohistochemical tests using anti-macrophage complement receptor 3 (OX42) or anti-monocyte–macrophage (ED1) and DiI-ac-low-density lipoprotein (LDL) uptake were used to identify microglia. The effect on accessory molecule expression of microglial cells cultured under varying conditions (M-CSF, GM-CSF, and M-CSF plus IFN-γ) was analyzed by fluorescence-activated cell sorter, using one of the following antibodies: anti-OX3, anti-OX6, anti-rat intercellular adhesion molecule (ICAM)-1, anti-rat B7-1, or anti-rat B7-2.

results. The cultured retinal microglia were positive for macrophage-related antigens (ED1 and OX42) and also showed uptake of LDL. Furthermore, ICAM-1 and B7-2 were expressed constitutively on these cells, and MHC class II and B7-1 were also expressed after IFN-γ stimulation.

conclusions. In vitro, the retinal microglia express the molecules required for effective antigen presentation to CD4-positive T cells. These findings suggest that microglia may play a role in local antigen presentation, especially when they are exposed to IFN-γ.

Retinal microglia are believed to be the intrinsic immunocompetent cells of retinal tissues. Whereas the development of the retina from the neural tube is similar to the development of the gray matter in the brain, the ontogeny of microglia is still a matter of controversy. Resident macrophages and microglia in many tissues are known to differ from circulating macrophages, both in their morphology and in some of the functional roles they play in local injury or infection. ^{1 2} In addition to a number of surface antigens that have been shown to be common for both microglia and macrophages in the brain, ^{3 4 5 6} a close similarity of their functional properties has also been noted. ^{7 8 9} Therefore, it has been hypothesized that during embryonic development of the retina and brain, microglial precursors derived from bone marrow enter these organs and differentiate into microglia through a series of morphologic transitions. ^{10 11 12}  

In 1983, Ling et al. ¹³ first succeeded in isolating ameboid microglial cells from rat brain and culturing these cells in vitro. These cultured microglia were found to possess functional characteristics similar to those of macrophages, including the expression of surface antigens and low-density lipoprotein (LDL) receptors as well as other immunoregulatory functions. ^{7 8} Similarity to macrophages was also found in the responsiveness of microglial cells to hemopoietic colony-stimulating factors (CSFs), ^{9 14} and an antigen-presenting function has also been reported after stimulation of brain microglia. ^{14 15 16} Therefore, it is evident that in the brain, microglia play a role in both homeostasis and immune responses. 

Because the population of retinal microglia is considerably smaller than the population of brain microglia and, additionally, because it is difficult to induce the proliferation of mature ramified microglial cells in vitro, there have been few reports of cultured retinal microglial cells. In 1993, when Roque and Caldwell ¹⁷ first isolated and cultured retinal microglial cells from adult Royal College of Surgeons (RCS) rats, they found these cells to express several surface markers common to brain microglia. These microglial cultures were also found to be highly phagocytic and to proliferate swiftly in response to macrophage colony-stimulating factor (M-CSF). However, attempts to isolate these cells from normal rats were unsuccessful until recently, when de Kozak et al. ¹⁸ isolated microglial cells from the retinas of dystrophic and nondystrophic control rats. 

A subpopulation of retinal microglia from rodents and humans has recently been reported to express major histocompatibility complex (MHC) class II molecules. ^{19 20 21 22} In this regard we have also demonstrated that a small number of retinal microglial cells showed donor class II molecules in chimeric rats injected with interferon (IFN)-γ. ²³ Others have reported similar observations in brain microglia. ^{10 24} The results of these studies suggest that retinal microglial cells are bone marrow–derived and may act as professional antigen-presenting cells (APCs). 

In the present study, we succeeded in isolating retinal microglial cells from normal newborn Lewis rats and in culturing these cells for at least several weeks using media supplemented with hemopoietic CSFs. We also confirmed their identity by immunohistochemical and functional studies (uptake of LDL). The effect of hemopoietic CSF and IFN-γ on expression of accessory molecules such as MHC class II, intercellular adhesion molecule (ICAM)-1 and the B7 family of molecules was also examined by flow cytometry. 

Materials and Methods

Cell Cultures

All animals used for the cell cultures were maintained and treated in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Newborn Lewis rats, 5 to 7 days of age, were obtained from Charles River (Wilmington, MA). The eyes were enucleated and hemisected, and the lens and vitreous were removed. The retina was carefully removed, taking care to avoid pigment epithelium contamination, and placed in Dulbecco’s modified Eagle’s medium (DMEM; Gibco, Grand Island, NY), containing 10% fetal bovine serum (FBS; Atlanta Biologicals, Norcross, GA) and 1% penicillin-streptomycin (Gibco). Retinas were digested with 0.05% trypsin in 0.53 mM EDTA (Gibco) for 1 hour at 37°C. After DMEM with 10% FBS (DMEM-FBS) was added to terminate trypsinization, the retinal pieces were dissociated by trituration and centrifuged. The retinal cells were cultured in DMEM-FBS supplemented with either 5 ng/ml human M-CSF (Sigma, St. Louis, MO) or 1 ng/ml murine granulocyte–macrophage (GM)-CSF (Sigma). Retinas were allowed to grow in 75-cm² flasks for at least 3 days and then were replenished with additional medium. For purification, after washing, the cells were incubated with Ca²⁺-Mg²⁺–free Hanks’ balanced salt solution (HBSS; Sigma) containing 0.2% EDTA and 5% FBS for 1 hour at 4°C and detached by vigorous pipetting. The resultant cell suspension was placed in plastic flasks and allowed to adhere for 30 minutes at 37°C. Afterward, loosely adhering and suspended cells were removed by gently shaking the flasks at room temperature. Cultures were maintained in a humidified atmosphere of 95% air-5% CO₂ at 37°C. The average density of the microglial cells was 3 × 10⁶ cells/flask and 7.5 × 10⁶ cells/flask for cells grown in M-CSF and GM-CSF, respectively. 

For fluorescence-activated cell sorter (FACS) analysis, microglial cells were cultured in M-CSF for 7 days, after which the cells were incubated with 100u/ml IFN-γ for 24 hours. 

Characterization of Isolated Microglia

Immunohistochemical Study.

Indirect immunohistochemical analysis was used to confirm the identity of microglia. Cells were plated onto chamber slides (Laboratory-Tek; Nunc, Naperville, IL) and allowed to grow for 2 days. After three washes with HBSS, the cells were fixed in acetone for 10 minutes at 4°C. A standard biotin-avidin immunocytochemical technique was then performed using the appropriate kits (Vector, Burlingame, CA). Murine monoclonal antibodies to rat macrophage complement receptor 3 (OX42, 1:100; Serotec, Oxford, England, UK) or rat monocytes and macrophages (ED1, 1:100; Serotec) were used as primary antibodies. Cultures were also tested with anti-cow glial fibrillary acidic protein (GFAP, 1:100; Dako, Carpinteria, CA) for staining of astrocytes and anti-bovine cellular retinaldehyde-binding protein (1:1000, a gift from John Saari, University of Washington, Seattle) for staining of Müller cells and retinal pigment epithelial cells. Negative controls were incubated with phosphate-buffered saline (PBS) in place of primary antibodies. 

DiI-ac-LDL Study.

The acetylated LDL, labeled with the fluorescent probe 1, 1′-dioctadecyl-3, 3, 3′,3′-tetramethyl-indocarbocyanate (DiI-ac-LDL; Biomedical Technologies, Stoughton, MA), was used to identify the functional characteristics of mononuclear phagocytic cells. Cells grown on chamber slides were incubated for 6 hours at 37°C with DiI-ac-LDL at a concentration of 10 μg/ml in culture medium. After three washes with PBS, cells were fixed with 3% formaldehyde in PBS (pH 7.2) at room temperature for 30 minutes and examined under confocal laser scanning microscopy (Carl Zeiss, Oberkochen, Germany). 

FACS Analysis of the Accessory Molecule Phenotype

The effect on accessory molecule expression of microglial cells cultured under varying conditions was analyzed by FACS (FACStar System; Becton Dickinson, San Jose, CA). After removal of dead cells using a Ficoll (Pharmacia, Uppsala, Sweden)-gradient, microglia grown for 24 hours in the presence of M-CSF, GM-CSF, or M-CSF plus IFN-γ were washed with PBS-1% bovine serum albumin (BSA) and pelleted by centrifuging at 1200 rpm for 10 minutes. Cells were labeled by indirect immunostaining with one of the following antibodies: anti-Lewis rat MHC class II (OX3), anti-rat MHC class II (OX6), anti-rat ICAM-1, anti-rat B7-1, or anti-rat B7-2 (Pharmingen, San Diego, CA). All other monoclonal antibodies were purchased from Serotec. The cell pellets were suspended and incubated with 20 μl of the appropriate antibodies for 30 minutes at 4°C. After the pellets were washed, they were incubated with fluorescein isothiocyanate–conjugated monoclonal anti-mouse IgG at 4°C for 30 minutes. Cells were washed and then resuspended in PBS-1% BSA. As a negative control, cells were incubated with irrelevant mouse IgG₁ in place of primary antibodies. For each sample, 1 × 10⁶ cells were analyzed. 

Results

Rat Microglia in Culture

After 3 days there were large cellular aggregates, cell clumps, and debris in the culture. Within these, many cells adhered to the flask. Some of these adherent cells appeared as single long processes or were ramified (Fig. 1A ). After 7 days, the cultures grew to confluence, and numerous microglial cells could be seen that were either ameboid or ramified. The most common shape appeared to resemble tear drops with a single long process (Fig. 1B) . 

After purification, the cells loosely adhered to the flasks and immediately assumed an ameboid shape. Within several hours, these cells showed a long single process. Later, some of these occasionally emitted several short processes, resembling a ramified cell (Fig. 2A ). These cells were successfully cultured for 1 to 3 weeks in hemopoietic M-CSF– or GM-CSF–supplemented medium. The proliferative effects of GM-CSF differed sufficiently from that of M-CSF after 7 days in culture, in that the number of microglia driven by addition of GM-CSF was almost twice that grown in the presence of M-CSF (data not shown). A morphologic comparison before and after stimulation with IFN-γ was also undertaken. After IFN-γ stimulation, cells changed morphology and became smaller and rounder (Fig. 2B) . 

Characterization of Rat Microglia

In the immunohistologic characterization of the cultured cells, more than 95% of the cells showed a positive reaction for both ED1 and OX42 (Fig. 3) . None of the cells showed a positive reaction for GFAP and cellular retinaldehyde-binding protein. These results indicate that the isolated cells were pure microglia and were not contaminated with astrocytes, retinal pigment epithelial cells, or Müller cells. 

To evaluate the functional characteristics of the cultured cells, uptake of DiI-ac-LDL was examined. The cells, regardless of their morphology, either ameboid or ramified, showed a strong uptake of DiI-ac-LDL, which could be observed as a punctate dye accumulation in the cytoplasm of the cell body (Fig. 4) . 

Effect of Hemopoietic CSF and IFN-γ on the Expression of Accessory Molecules by Microglia

As shown in Figure 5 , there was a marked upregulation of both Lewis rat–specific (OX3) and rat MHC class II (OX6) expression with IFN-γ in M-CSF, whereas no expression of class II molecules was observed with M-CSF alone. GM-CSF caused a smaller degree of upregulation. This trend was the same for either OX3 or OX6. The expression of ICAM-1 revealed the same high levels in all cultures. There were no differences, either with or without IFN-γ in M-CSF. 

The FACS analysis revealed expression of B7-2 at detectable levels in cells cultured under all three conditions. No shift in mean fluorescence intensity was observed for the B7-1 molecule in cells cultured with either M-CSF or GM-CSF. However, a detectable shift was observed after IFN-γ stimulation (Fig. 6) . 

Discussion

In the present study, we succeeded in isolating retinal microglia from newborn Lewis rats, and we established culture systems using media supplemented with hemopoietic CSFs. The cultured retinal microglia, regardless of their morphology, have macrophage-associated properties. They are positive for macrophage-related antigens (ED1 and OX42), uptake of LDL, and costimulatory molecules (ICAM-1, B7-2). Furthermore, cultured microglia were found to express MHC class II and B7-1 after IFN-γ stimulation. 

In earlier experiments, we attempted to isolate microglia from adult rat eyes. However, the isolated cells did not proliferate and were unresponsive to hemopoietic CSFs, resulting in very few cells. Roque and Caldwell, ¹⁷ who first succeeded in culturing retinal microglial cells from dystrophic retina, have also reported the difficulty of culturing these cells from normal retina. Based on these earlier studies, it was evident that the appropriate time to isolate and culture microglial cells from normal rat eyes was within the second postnatal week. The transformation of ameboid microglia into ramified microglia, which occurs between the second and third postnatal week, is considered to be a regressive phenomenon, manifested by the diminution of their content of hydrolytic enzymes and the downregulation of surface antigens. ^{12 25}  

With regard to the ontogeny of microglia, the more widely accepted view is that during the embryonic development of the retina and brain, microglial precursors derived from bone marrow enter these organs and differentiate into microglia through a series of morphologic transitions. ^{10 12 24} Recently, the expression of macrophage-related antigens during development of ameboid microglia in fetal rat brain demonstrated that ED1, OX6, and OX42 were detected from the 14th embryonic day to birth and that these cells began to emit short processes. ²⁶ In the early postnatal rat brain, OX42 was found, but OX6 was rarely detectable. Late in the process, those cells that expressed these antigens gradually downregulated them. In addition, a morphologic change into an oval or elongated form occurred by postnatal day 21. ^{27 28 29} We isolated microglial cells between the fifth and the seventh postnatal days, and our cultured cells showed a positive reaction for both ED1 and OX42 by immunocytochemical techniques. Further, from a morphologic standpoint, the cells showed an ameboid form with some of them demonstrating long processes. These findings, which were similar to previous in vivo studies of brain microglia, suggested that our cultured cells could also have been derived from bone marrow. 

It was noted after a 24-hour incubation with IFN-γ that retinal microglial cells cultured in M-CSF became smaller and rounder (Fig. 2B) . The transition of brain microglia from an ameboid to a ramified morphology has been reported to be due to the formation of microtubules. ²⁵ It has also been noted that the ameboid microglia releases more tumor necrosis factor (TNF) than the ramified form, after stimulation with lipopolysaccharide (LPS). de Kozak et al. ¹⁸ have demonstrated the synthesis of both TNF and nitric oxide by retinal microglial cells after stimulation with IFN-γ and LPS. It is possible therefore that the change in morphology of the retinal microglial cells subsequent to incubation with IFN-γ is a result of the production of TNF and/or nitric oxide. Because the manner in which microtubule stabilization is controlled remains unclear, this point needs further study. 

We have demonstrated that a small number of retinal microglial cells express donor class II molecules in chimeric rats and that most retinal microglial cells express ED1 and OX42. ²³ These reports also support the concept that retinal microglia are derived from bone marrow. 

DiI-ac-LDL has been considered more convenient and more reliable than histochemical staining for nonspecific esterase for identifying ameboid microglia in dissociated brain cell cultures. ⁷ The retinal microglial cells described in this study showed strong uptake of DiI-ac-LDL, which was observed as punctate dye accumulation in the cytoplasm of the cells. This indicates that retinal microglia have scavenger receptors for acetylated LDL and that they possess a phagocytic function. 

In the present study an attempt was made, using FACS analysis, to determine the effect of hemopoietic CSF and IFN-γ on the expression of accessory molecules, such as MHC class II, ICAM-1, and the B7 family of molecules. Our results show that there was a remarkable upregulation of both Lewis rat (OX3) and rat (OX6) MHC class II expression on microglia cultured in M-CSF plus IFN-γ when compared with cells cultured with M-CSF alone. These data support our previous in vivo results with Lewis rats showing that IFN-γ injection causes an increase in OX6-positive microglial cells when compared with uninjected rats. ²³ There was a slight upregulation of OX3 and OX6 in GM-CSF–driven cultures although FACS analysis demonstrated no apparent effect of M-CSF alone on MHC class II expression on the retinal microglia. Preliminary immunostaining of these cells cultured on chamber slides showed that 30% of the cells were OX6 positive, and 80% were OX3 positive. ³⁰ This difference is similar to the results of previous reports showing that HLA-DR (class II) expression on brain microglia increases after culture. ²⁶ It is assumed that adhesion of microglia to the culture flask acts as a pseudoactivation process, prompting increased expression of the surface antigen. 

ICAM-1 antigens have been known to act as costimulatory molecules ³¹ and to be expressed on isolated mature microglia and cultured immature microglia. ³¹ In our FACS studies, the expression of ICAM-1 had the same high levels in all cultures and showed no difference with or without IFN-γ. However, in situ immunostaining of cells from naive chimeric rats and those receiving IFN-γ injection did not demonstrate expression of ICAM-1 (unpublished data, April 1997). Similar results have also been reported in the brain. ³³ Other studies by Fischer et al. ¹⁶ also compared the expression of ICAM-1 on cultured brain microglia using similar methods. They demonstrated low levels of ICAM-1 in both M-CSF- and GM-CSF-driven microglial populations that were then enhanced after IFN-γ stimulation. We suspect that one reason that there was no difference in ICAM-1 expression in our system may be that our basal culture condition activated the expression of ICAM-1 before IFN-γ stimulation. 

Expression of the B7 family of molecules is generally restricted to fully immunocompetent APCs of the lymphoid system, such as activated monocytes and macrophages. ^{34 35 36 37} In brain microglia, B7-1 mRNA expression is markedly increased after exposure to IFN-γ or GM-CSF, whereas B7-2 mRNA is expressed in untreated microglia constitutively. ³⁸ Our results demonstrate that B7-2 was expressed in cultures under all three conditions, whereas B7-1 could not be found in the presence of either M- or GM-CSF basal culture conditions. However, an expression of B7-1 after IFN-γ stimulation was detectable. Because the dosage of GM-CSF used in our studies was 10 times lower than that in the report on brain microglia, it cannot be concluded categorically that GM-CSF does not induce B7-1 expression of retinal microglia. 

In rodent eyes, constitutive expression of MHC class II has been demonstrated on the cells of the conjunctiva, cornea, trabecular meshwork, iris, ciliary body, and choroid. ^{39 40 41 42 43} A subpopulation of retinal microglia from rodents and humans has recently been reported to express MHC class II molecules, ^{19 20 21 22 23} suggesting the possibility that microglia may act as APCs. In the present study, cultured microglia were shown to express ICAM-1 and B7-2 molecules constitutively. In addition, these cells had the ability to express MHC class II and B7-1 molecules when activated. All these molecules are required for an APC to effectively present antigen to CD4-positive helper T cells. ⁴⁴ Therefore, these studies further suggest that retinal microglia may play a role in local antigen presentation, especially when levels of IFN-γ are increased. However, further studies are required to determine the role of these cells in retinal antigen-specific T-cell responses. 

 Supported in part by Grants EY 10727 and EY 03040 from the National Institutes of Health.

 Submitted for publication August 18, 1998; revised February 2, 1999; accepted February 16, 1999.

 Commercial relationships policy: N.

 

Figure 1.

 

Phase-contrast photomicrographs of mixed retinal cultures from newborn Lewis rats (M-CSF–supplemented medium). (A) Day 3 retinal cultures: adherent single cells could be observed within aggregates, cell clumps, and debris. Some of these cells appeared as a single, long process or ramified. (B) Day 7 confluent retinal cultures: numerous microglial cells could be seen that were either ameboid or ramified. The most common shape resembled tear drops with a single, long process. Magnification, ×100.

Figure 2.

 

Phase-contrast photomicrographs of a morphologic comparison on purified microglial cell cultures before and after stimulation with IFN-γ (M-CSF–supplemented medium). (A) Purified microglial cell cultures, day 1 after purification: both ameboid and ramified forms could be observed, and some of the microglial cells showed a long, single process or several short processes, resembling a ramified cell. (B) Purified microglial cell cultures, after stimulation with IFN-γ: the cells changed their morphology and were smaller and rounder. Magnification, ×200.

Figure 3.

 

Immunocytochemical characterization of retinal microglial cells in purified cultures (M-CSF–supplemented medium). Cells grown on chamber slides were incubated with OX42 (1:100) monoclonal antibodies (A) or ED1 (1:100) monoclonal antibodies (C). Negative controls (B) were incubated with PBS in place of primary antibody. A standard biotin-avidin immunocytochemical technique was then performed. More than 95% of the cells showed a positive reaction for both ED1 and OX42. Magnification, ×200.

Figure 4.

 

Functional characteristics of retinal microglial cell in purified culture (M-CSF–supplemented medium). Confocal images of DiI-ac-LDL uptake. Cells were incubated with 10 mg/ml DiI-ac-LDL for 6 hours at 37°C. All the cells showed strong uptake of DiI-ac-LDL, which was seen as the intracellular accumulation of the dye. Magnification,× 500.

Figure 5.

 

Comparison of expression of OX3, OX6, ICAM-1, and B7-2 on cultured microglia grown in the presence of M-CSF, GM-CSF, or M-CSF plus IFN-γ for 24 hours. Cells were labeled with one of the following antibodies: OX3, OX6, ICAM-1, or anti-rat B7-2. Fluorescein isothiocyanate–conjugated monoclonal antibody was used to detect the bound antibodies. Dashed lines indicate relative fluorescence for immunostaining with the antibodies to the antigen indicated on the left. Solid lines indicate immunostaining with isotype control antibodies for negative controls. A marked upregulation of both OX3 and OX6 expression was demonstrated on microglia with IFN-γ in M-CSF. GM-CSF caused a smaller degree of upregulation, whereas no expression of class II molecules was observed with M-CSF alone. This trend was the same for OX3 and OX6. ICAM-1 was expressed at the same high levels in all cultures. There were no differences in M-CSF, with or without IFN-γ. B7-2 was also expressed at detectable levels in cells cultured under all three conditions.

Figure 6.

 

Comparison of expression of B7-1 molecule on cultured microglia grown in the presence of M-CSF (A), GM-CSF (B), or M-CSF plus IFN-γ (C) for 24 hours. Cells were labeled with biotin-conjugated anti-rat B7-1, followed by r-phycoerythrin-streptavidin. Dotted lines indicate relative fluorescence for immunostaining with the antibodies to B7-1. Solid lines indicate immunostaining with biotin-conjugated isotype control antibodies. No shift in mean fluorescence intensity was observed for the B7-1 molecule in microglial cells cultured with M-CSF or GM-CSF. However, a detectable shift was observed after IFN-γ stimulation. FL2-height indicates fluorescence was measured by signal height.