CD45-Positive Cells of the Retina and Their Responsiveness to In Vivo and In Vitro Treatment with IFN-γ or Anti-CD40

Abstract

purpose. The ability to mount antigen-specific immune reactions in the retina demonstrates local recognition of retinal antigens. However, properties of antigen-presenting cells (APCs) of the retina are uncertain. The current study was undertaken to look for evidence of CD45⁺ cells with APC potential in the retina and to examine their in situ and in vitro responses to IFN-γ and anti-CD40, two stimuli known to upregulate activities associated with antigen presentation.

methods. Mice were pretreated with systemic or intracameral (IC) inoculations of IFN-γ or anti-CD40. Retinas were harvested, enzymatically dissociated, positively selected with anti-CD45, and analyzed by flow cytometry with antibodies known to identify APCs.

results. The most common CD45⁺ retinal cells were CD11b⁺, F4/80⁺, CD8α⁺, CD80⁺, and major histocompatibility complex (MHC) class II^lo, a phenotype characteristic of central nervous system (CNS) microglia (MG). There was also a small population of DEC-205⁺ cells and a smaller number of CD11c⁺ cells, both markers of dendritic cells (DCs). IC inoculation of IFN-γ led to an increase in the number of CD45⁺ cells and a modest upregulation of MHC class II on CD11b⁺ cells. IC inoculation of anti-CD40 also increased the total number of CD45⁺ cells and the number of CD11b⁺ cells, but increases in CD80 and MHC class II expression on CD11b⁺ cells were insignificant. After anti-CD40 treatment, CD45^hi11c⁺ cells increased in number and altered their expression of CD11b.

conclusions. Retinal MG were readily identified as the most numerous population. A small population of cells with perivascular cell (PVC)–like properties was found. They were CD45^hi11c⁺, some had elevated MHC class II, and they were affected by anti-CD40 treatment in vivo. No conventional DCs were found, although there was a distinct DEC-205⁺ population. Overall, the effects of IFN-γ and anti-CD40 treatment were attenuated in the retina in vivo and also on CD45⁺ cells in culture, compared with the control.

Immature dendritic cells (DCs) residing in nonlymphoid tissues are known to sample the local environment for antigens (Ags). On initiation of a maturation program, they migrate to peripheral lymphoid tissues, where these molecules are processed and expressed as peptides in major histocompatibility complex (MHC) molecules, together with accessory molecules that enable priming of Ag-specific naïve T cells. ¹ The activated T cells emigrate to tissue sites where Ag is found, leading to a local response. In a number of studies of nervous system tissues, resident dendritic cells (DCs) are thought to be absent from the immunologically quiescent central nervous system (CNS) parenchyma, but are recruited during inflammation. ² There are recent reports of DCs in the choroid plexus and meninges. ³ Infiltrating DCs in the CNS appear to play a role in initiating local inflammation in experimental autoimmune encephalomyelitis (EAE) ⁴ and as effectors, ² but microglia (MG) and infiltrating macrophages (MΦs) appear to be most crucial in the pathogenesis of EAE. ^{5 6}  

We wanted to study CD45⁺ cells with potential antigen-presenting cell (APC) activity in the retina. The retina is the target of the induced autoimmune disease, experimental autoimmune uveoretinitis (EAU), directed at any of several photoreceptor cell proteins, much as CNS myelin Ags are targets for EAE. The retina is also an immune-privileged site and lies behind the blood–retina barrier. The barriers formed by the retinal vascular endothelium and the retinal pigmented epithelium (RPE) provide a degree of sequestration, limiting the circulation of resting T cells, even those with specificity for a retinal Ag, through the parenchyma of the quiescent retina. ⁷ Movement of proteins and macromolecules in and out of the retina across the barriers is restricted. Retinal immune privilege also appears to include Ag-dependent immunoregulatory processes, ⁸ which could either depend on specific delivery and presentation of these Ags to the immune system or result from leakage of Ag from the retina. In the EAU model, retinal CD45⁺ cells, including parenchymal MG and perivascular cells (PVCs), ⁹ or infiltrating MΦs ¹⁰ play roles in presenting local Ags to activated T cells that have extravasated and/or serve as mediators of tissue destruction. 

Whether the immunologically quiescent retinal microenvironment is sampled by cells with DC-like properties is unclear. The presence of draining lymphatics is not established, although some evidence of drainage has been reported. ¹¹ Retinal MG have not been reported to exit and circulate to lymph nodes (LNs) or spleen. Despite the obstacles presented by the blood–retina barrier, the occurrence of immunoregulatory activity resulting from retinal expression of an Ag ⁸ and the immunopathology resulting from autoimmune responses to retinal Ags (EAU), require that retinal Ag be recognized by T lymphocytes. In the case of EAU, recognition must occur in the retina, on resident or recruited APCs, to produce local tissue destruction. At present, the most likely candidate for a retinal DC equivalent may be the PVCs. ¹² Immunoregulatory or peripheral tolerance processes could result from the recognition of retinal Ag delivered to lymphoid tissues outside the retina, as well as from local recognition of retinal Ag on APCs programmed to inhibit T-cell responses. ¹³  

The results we report herein confirm that most CD45⁺ cells we identified in the retina were CD45^intCD11b⁺ MG. We also found that they expressed several other markers associated with CNS MG. Cells that shared some phenotypic properties with DCs were found in much smaller numbers. Eyes were inoculated with IFN-γ or anti-CD40 to determine whether activation of the CD45⁺ cells was inhibited. Doses of IFN-γ that induced inflammation when administered intradermally and upregulated expression of markers on spleen cells in vivo and in vitro had only a moderate effect on the retina. Modest increases in cell number and/or expression levels of CD80 and class II MHC were found in the retina in vivo. Much larger doses of IFN-γ administered systemically had no detectable effect in the retina. Cells that expressed the DC marker CD11c appeared in greater numbers after treatment with anti-CD40, but changes in expression levels of CD8α, CD80, and class II MHC were small. The environment of the retina appears to attenuate the activity of these agents, which are known to have activating effects on MG, MΦs, and DCs in other tissues and in vitro. 

Materials and Methods

Mice

All experiments were performed with B10.A mice. Use of mice was in accordance with the ARVO Statement for Use of Animals in Ophthalmic and Vision Research and received local institutional review board approval. 

Collection of Tissues for Preparation of CD45⁺ Cells

B10.A mice were killed by CO₂ inhalation and transcardially perfused with cold PBS-heparin (2 U/mL) to displace blood from the retinal vasculature. Perfusion was continued until the effluent was clear—approximately 20 mL. Eyes, brain, spleen, and LNs were removed and placed in Ca²⁺-Mg²⁺–free phosphate-buffered saline (CMF-PBS) on ice until all tissues were collected. The initial incision in the eye was made with a microvitreoretinal (MVR) knife posterior to the junction of the retina and ciliary body. Scissors were used to extend the cut around the eye so that the entire anterior segment could be removed. The lens and vitreous were then removed. No attempt was made to remove RPE from the retina, because cells of interest have been reported in the subretinal space ¹⁴ and could be lost if the RPE were removed. If the DCs observed in the choroid near the basal RPE surface ^{15 16 17} were adherent to the RPE, they could be carried along. This did not appear to be a problem in practice, because CD11c⁺ cells were few in all samples. Retina was detached from the optic nerve with an MVR knife. Much care was taken to avoid contamination of the retina with uveal tissue, which is known to be well-populated with CD45⁺ cells, especially ED2⁺ MΦs and OX6⁺ DCs. ¹⁶ The retinas were pooled on ice. Fifty microliters of PBS containing collagenase (0.5 mg/mL) and DNase (40 μg/mL) was added per retina. When all retinas were collected in this tube, they were warmed to 37°C and mixed gently and frequently to dissociate the tissue. Six to eight retinas were collected and pooled for most studies. Samples of brain and spleen were dissociated with the same enzymatic procedure. Preliminary studies showed that after approximately 30 minutes, release of cells was complete. Cell suspensions were diluted with 10 volumes of CMF-PBS containing 5 mM EDTA and 1% BSA, and incubation was continued for 5 minutes with gentle mixing. The cells were pelleted and washed twice with CMF-PBS containing 5 mM EDTA and 1% BSA. The cells were resuspended in buffer and applied to a density gradient. Cells at the 1.0875 g/cm³ interface were collected for positive selection of CD45⁺ cells, or assayed without use of column enrichment. 

Positive Selection for CD45⁺ Cells

Enrichment of CD45⁺ cells was achieved with a magnetic separation system (MACS; Miltenyi Biotec, Auburn, CA). The cell suspensions were incubated with anti-CD45 conjugated with paramagnetic beads before application to MS⁺ or LS⁺ columns per the manufacturer’s recommendations. 

Flow Cytometry

Single-cell suspensions were made in 100 μL of staining buffer (PBS with 5% fetal calf serum [FCS], 0.02% Na azide) containing the labeled antibodies listed in Table 1 . After 20 minutes on ice, cells that were incubated with biotin-labeled Abs were washed twice and resuspended in streptavidin conjugates. After 15 minutes, all cells were washed twice and resuspended for flow cytometry (FACS Calibur; BD Biosciences, San Jose, CA), conducted with the accompanying software (CellQuest; BD Biosciences). Experiments were based on acquiring 5 × 10⁴ to 1 × 10⁶ total events. Staining intensity was expressed as geometric mean fluorescence intensity (GMFI). 

In Vivo Treatment of Mice to Upregulate Markers

Some mice were given a single intracameral (IC) inoculation of 100 to 200 U IFN-γ in 1 μL 24 hours before harvest of retina. Control mice received an injection of saline in one eye. Some mice received intraperitoneal (IP) injections of IFN-γ, 20,000 U/d for 2 days, before harvesting tissues for flow cytometry. An antibody with agonist activity for CD40 ^{18 19} (mAb FGK45, gift from Stephen Schoenberger, La Jolla Institute for Allergy and Immunology) was also administered to some mice. These mice were given an IC injection of 1 μL (1.5 μg) of FGK45 24 hours before harvest. An equal amount of saline or control Ig was administered to control eyes. 

Results

Recovery of CD45⁺ Cells from Retina

To establish that the frequency of CD45⁺ cells was increased by magnetic separation, aliquots were made to compare unseparated cells with the column-adherent and nonadherent populations (Table 2) . The fraction of CD45⁺ cells increased from approximately 1% in the unseparated, ungated population to approximately 20% in the CD45enriched, size-gated population. The fraction of CD45⁺ cells in the cells that did not adhere to the column was dramatically reduced. Similar levels of enrichment and depletion were found in most subsequent experiments. A large quantity of rod outer segment fragments and other cellular debris dominated scatterplots of the cells and were only partially removed by a single column pass. However, the enrichment of CD45⁺ cells obtained by use of the column and size-gating permitted accurate analysis of the cell surface markers. 

Properties of Retinal MG

Approximately 80% of the total CD45⁺ cells from quiescent retina were CD11b⁺, and most of these were tightly concentrated in the CD11b^hiCD45^int region identified as R4 in Figure 1A . The CD11b⁺ cells of the retina and brain, as well as those from iris/ciliary body, constituted similar discrete populations of brightly stained cells compared with the broad peak of CD11b⁺ cells found in spleen or LNs (data not shown). We divided the CD45⁺ cells into three populations based on the level of CD45 expression (Fig. 1) . The cells in the R4 region correspond to the MG population reported by others. ²⁰ Cells bearing F4/80 also made up a significant portion of the CD45⁺ cells. Most of these (approximately 85%) were also CD11b⁺ (Fig. 1C) . F4/80 and CD11b cells were not distinguished by CD80, MHC class II, or CD8α expression. Although F4/80 is often associated with DCs, it appears that F4/80 in the retina largely identified the CD45^intCD11b⁺ MG population. 

A moderate level of CD8α expression was concentrated in the R4 CD11b⁺45^int population that defines retinal MG and differs from the low CD8α expression in the R5 CD45^hi population (Figs. 2A 2B) . Constitutive expression of CD8α is a property that retinal MG share with CNS MG, but which distinguishes both of these populations from CD11b⁺ cells in spleen, which are CD8α^lo (R7 region in Figs. 2C 2D 2E ). A low level of Fas expression on a significant portion of the CD11b⁺ cells of retina and brain was another property that distinguished them from splenic CD11b⁺ cells, of which few cells were Fas⁺ (R9 region in Fig. 3A ). Nearly one half of retinal CD11b⁺ cells were Fas⁺. CD80, or B7-1, an important molecule for Ag presentation leading to CD4 T-cell activation, was present at low to moderate levels on approximately 80% of CD45^intCD11b⁺ MG. It was also expressed on approximately 60% of CD45^hi cells (Fig. 4B) , distinguishing it from the distribution of CD8α. The fraction of CD45⁺ cells that were CD80⁺ in the retina was much higher than in the spleen (Fig. 4D) . Class II (I-A^k) was also expressed at higher average levels on CD11b⁺ cells compared with CD11b⁻ cells, but at much lower levels than on CD45⁺11b⁺ LN cells (Fig. 5) . A few retinal CD11b⁺ cells were also CD11c⁺, a marker that is often expressed on DCs (see later discussion and Fig. 11A ). 

DEC-205⁺ Cells in the Retina

A substantial portion of the remaining CD45⁺ cells in the retina were found to express DEC-205. These cells constituted 10% to 30% of the total CD45⁺ population (Fig. 6A) . They were not CD11c⁺ (not shown). The DEC-205⁺ cells expressed a very low level of CD11b (Fig. 6A) , and their expression of class II (I-A^k) was low (Fig. 6B) . Approximately 40% of the DEC-205⁺ cells (Fig. 6C ; comparison of top right to borrom right quadrants) were positive for CD80, and one third were positive for CD8α (Fig. 6D ; comparison of top right to bottom right quadrants). A correlation was noted between CD80 and CD8α staining on DEC-205⁺ cells (Fig. 6E , arrow). If the DEC-205⁺ cells were CD8α⁺, they were likely also to be CD80⁺ (Fig. 6E) . Approximately one third of the DEC-205⁺ cells were double positive for CD8α and CD80. It is interesting to note that expression of these molecules has been found to change with maturation of DCs from other tissues. ^{21 22}  

Effect of IFN-γ on Phenotypes of the CD45⁺ Retinal Cells In Vivo

The T lymphocytes most likely to enter the parenchyma of the retina are those that have been activated elsewhere. Resting T cells are not known to gain access to the retina. Activated T cells are likely to secrete IFN-γ and to express CD40L, and could interact most productively with APCs that are activated by these factors. To determine the effect of these activators of DCs, MΦs, and MG in the retina, mice were inoculated with either IFN-γ or an anti-CD40 Ab (FGK45) which has agonist activity for CD40 (as described later). Analysis of CD45⁺ retinal cells from mice given IP injections of 20,000 U of IFN-γ for 2 days before harvest on the third day indicated that this treatment had no discernible effect on expression using the panel of Abs described in Table 1 . This treatment upregulated MHC class II expression on B220⁺ cells from the spleen (data not shown). Apparently, bioactive levels of IFN-γ did not penetrate the blood–retina barrier, or if it did, its activity was attenuated by factors in the retinal microenvironment. An alternative route of inoculation was taken in an effort to deliver IFN-γ to the retina with minimal trauma. IFN-γ (100–200 U in 1 μL saline) was inoculated into the anterior chamber of the eye, and the retina was harvested the next day as described. Retina harvested from saline-inoculated eyes was used as the control. Histologic analysis of IFN-γ–inoculated eyes showed no evidence of inflammation (data not shown), consistent with a previous report that showed no evidence of infiltration despite the induction of sensitivity to a delayed-type hypersensitivity (DTH)–like response after similar IFN-γ inoculation. ²³ Conversely, intradermal inoculation of 200 U IFN-γ in 10 μL into the mouse ear pinna causes a small but detectable inflammation that can be measured by ear swelling after 24 hours (net swelling of 0.0351 ± 0.0130 mm; P = 0.05 compared with saline injection). Evidence of mononuclear cell infiltration was also found by histology of IFN-γ–injected ears after 24 hours (data not shown). 

The recovery of CD45^int and CD45^hi cells from eyes treated with IFN-γ was increased by approximately twofold, with the largest increase in the CD45^hi population, which is consistent with IFN-γ activation of the MG population (Fig. 7) . Most of this increase was in CD11b⁺ cells, with an associated increased fraction of I-A^k-positive cells (from 1.2% to 4.4%; Figs. 8A 8B ). A small increase in the fraction of CD80⁺CD11b⁺ cells was found after treatment (from 5.8% to 10.5%), although the average level of CD80 expression was not significantly changed (Figs. 8C 8D) . The CD8α expression level and the proportion of CD8α⁺ cells, which might change if the MG had been induced to enter a differentiation program, as has been observed in DCs, ²¹ were not altered (not shown). There was little or no change in the number of CD11c⁺ cells (not shown). 

Anti-CD40 Activation In Vivo

Ligation of CD40 on MG, DCs, and B lymphocytes is known to produce changes associated with increased Ag-presenting activity, including increased expression of MHC class II and CD80/86. ^{24 25} Retinal MG are known to be CD40⁺. ²⁶ Although inoculation of 1.5 μg of FGK45 antibody into the ear did not cause significant ear swelling (net swelling of 0.0121 ± 0.0129 mm; P > 0.05 compared with saline injection), several other assays showed that the antibody was active. The FGK45 Ab was inoculated into the anterior chamber in a effort to gain access to the retina, because systemic inoculation of 100 to 200 μg had no effect on retinal cells (data not shown). Injection of 1.5 μg in 1 μL led to a small but significant increase in the number of CD45⁺ cells, compared with the isotype and saline control injections (Fig. 9) . Although the number of CD11b⁺ cells increased with anti-CD40 treatment, their expression of CD80, I-A^k and CD8α was not significantly changed (data not shown). 

CD11c⁺ Cells

Although the effect on CD45^intCD11b⁺ MG was mostly due to an increase in number, the CD45^hi population (gated on the CD45^hi peak in Fig. 9 ) of anti-CD40–treated eyes contained an increased number of CD11c⁺ cells compared with the CD45^lo/int population (Fig. 10) . The small number of CD11c⁺ cells found in the CD45^hi fraction of untreated eyes increased significantly after anti-CD40 treatment, but not by IC inoculation of the mAb isotype control or inoculation of saline (Fig. 10) . In the experiment shown in Figures 9 and 10 , the frequency of CD11c⁺ cells (number of cells denoted by the bracket in Fig. 10 ) increased from less than 1% in the control eyes to approximately 5% (110 of 2072 events) of the total CD45⁺ cells in anti-CD40–treated eyes. 

Before treatment, the small number of existing CD11c⁺ cells was approximately 80% CD11b^hi (Figs. 11A 11B) . After treatment, almost half were CD11b^lo/int. Expression of CD80 on CD11c⁺ cells was similar to that on CD11b⁺ cells and was not significantly changed by anti-CD40 treatment (Figs. 11C 11D) . Although there were relatively few MHC class II^hi cells in the retina, they appeared in the CD11c⁺ population, regardless of whether the eyes were treated with anti-CD40 (Figs. 11E 11F) . The number of class II⁺ cells was increased by anti-CD40 treatment, but the average level of class II expression was unchanged. CD8α was found on approximately 20% of the cells, before and after treatment, and the expression level did not change (data not shown). The number of cells expressing another dendritic cell marker, DEC-205, did not show a significant increase in numbers after anti-CD40 treatment (data not shown). 

Activation of Retinal Cells In Vitro

Column-enriched CD45⁺ cells were collected from retinas as described and cultured for 24 hours with and without IFN-γ (1000 U/mL) or anti-CD40 (3 μg/mL). The cells were resuspended, collected, and the wells washed again with CMF-PBS to remove adherent cells. Incubation with anti-CD40 had no significant effect on any of the markers, although we have observed in other studies that inclusion of this dose of antibody significantly increases the in vitro response of naïve T cells to specific Ag using splenic APC (data not shown). IFN-γ treatment increased the intensity of CD45 and CD11b staining compared with cells incubated in medium alone, showing that the IFN-γ was present at a biologically significant level (Table 3) . A small, but significant increase in I-A^k labeling was also observed, but none of the other labels was affected. Even when CD45⁺ cells are removed from the retina, washed, and cultured, their expression of molecules associated with Ag presentation was not much affected by IFN-γ or anti-CD40. 

Discussion

The search for retinal APCs has generally concentrated on CD45⁺ cells, even though some other non–bone-marrow–derived cells, including RPE ²⁷ and Müller cells, ²⁸ express MHC class II under some conditions. Previous studies of CD45⁺ cells from retina have provided evidence for at least two distinct populations, PVCs and resident parenchymal MG. One distinction between the two is the level of MHC class II expression. At least a portion of the PVCs constitutively express class II at a higher level, whereas parenchymal MG demonstrate low levels of expression. ^{20 29 30 31} Other criteria are the levels of CD45 and CD11b expression: PVCs are CD45^hi and MG are CD45^loCD11b^hi. ^{20 32} This MG population was readily apparent in our flow cytometry studies as the CD45^intCD11b⁺ population. Our use of a PE-labeled anti-CD45 mAb produced very bright staining so that the CD45 low staining intensity observed by others appeared more intense in our study, but was clearly lower than, and distinguishable from, staining in CD45^hi cells. MG were mostly DEC-205⁻ and CD11c⁻. The F4/80⁺ population was indistinguishable from the CD11b⁺ MG cells by all measures we have made to date. 

The identity of the cells that provide the initial Ag presentation in retina is of particular interest. The PVC are candidates for this function. Using radiation bone marrow chimeras, it has been found that donor bone marrow confers susceptibility to adoptive transfer of EAU and EAE by T cells restricted to donor MHC class II. ^{33 34} Because PVCs exhibit turnover, whereas MG turnover very slowly, if at all, these results argue that a required step in Ag recognition in the retina depends on PVCs. This required step is possibly the first step in local Ag recognition. If PVCs are characterized by elevated class II expression, the DEC-205⁺ cells do not satisfy that criterion. The DEC-205⁺ population was significantly CD80⁺, but it expressed little class II and was little affected by either the IFN-γ or anti-CD40 intraocular inoculations within the time course of the study. These cells may represent an immature DC-like cell in an Ag-gathering mode. 

The identity of the PVC population is suggested from our studies. Based on our flow cytometry results, we propose that CD45^hi11c^hi cells (Figs. 10 and 11 , untreated eyes) contain PVCs. These cells are also CD11b^hi, and a portion of them express significant levels of MHC class II (Fig. 11E) . The CD11c⁺ population increased after anti-CD40 inoculation and expressed CD80, a wide range of MHC class II, and was divided between CD11b^lo and CD11b^hi. T cells activated elsewhere would have the ability to emigrate from retinal blood vessels, and due to their expression of CD154, would be able to engage CD40 on these cells. Together with CD80 and MHC class II expressed by these cells, local Ag recognition and activation could occur. Whether these cells were recruited from the circulation or arose from an existing retinal population is uncertain. The CD11c⁺ and DEC-205⁺ cells could represent different stages of differentiation of the same population. Because they are so few in number, functional studies based on purification of CD11c⁺ cells have been difficult. 

The retina, like the CNS, appears not to be endowed with typical immature, interstitial DCs in the Ag-gathering state that are commonly found in other tissues. Furthermore, there is no known circulation of DCs or MG from the retina to LNs that would deliver retinal Ag to naïve T cells. In experiments in progress, we have examined the activation status of naïve, β-gal–specific CD4 T cells in the cervical and submandibular LNs of hi-arr-β-gal mice, which express β-galactosidase in the retina. No evidence for a response has been found (unpublished observations). Still, recognition of retinal Ag does occur, but it is not certain if that recognition is limited to APCs in the retina. Clearly, the phenotype expected of immature DCs in other tissues is not well replicated in the retina. One explanation is that parenchymal MG are induced to differentiate into DCs or DC-like cells under certain conditions. 

Studies in brain lend support to this idea. MG isolated from neonatal mouse brain and further cocultured with astrocytes and/or granulocyte-macrophage–colony-stimulating factor (GM-CSF) show development of many DC-like properties, including APC activity. ³⁵ , ³⁶ CD40 ligation of these DC-like cells induces terminal differentiation into mature DCs, leading to speculation that DCs in brain derive from resident MG. ³⁷ MG from brain express CD40. ³⁸ That expression is downregulated by TGF-β ³⁹ and upregulated by IFN-γ. ⁴⁰ CD40 expression in CNS is needed for development of severe disease in EAE. ⁴¹ CD40 ligation on CNS MG induces IL-12p70, which may skew the outcome of the interaction with T cells toward a Th1 response. ⁴² Depending on cytokine pretreatment, brain MG may acquire the ability to prime or anergize naïve cells. ⁴³ IFN-γ and anti-CD40 treatment of MG are both needed to upregulate NOS. ⁴⁴  

Several factors limit the applicability of the studies just cited to the question of which cells initiate local responses in vivo. The MG isolation techniques are hard pressed to separate MG from PVCs, and it is not entirely satisfying to attribute the findings to MG, rather than contaminating populations. These studies were also performed with cultured MG. Although this allows more control of the conditions under which the cells are assessed, it also removes them from the local tissue environment, which appears to be an important factor in setting their responsiveness and phenotype. This includes the local production of TGF-β, known to downregulate expression of B7 and CD40. ³⁹ Furthermore, if the DEC-205⁺ or CD11c⁺ cells we describe serve as retinal (or brain) DCs, then the in vitro results obtained with MG may have unknown significance in vivo. Finally, even though these examples of CNS MG provide evidence for activities associated with mature DCs, our long-term question concerns the mechanism of transport of Ag from the retina to a regional LN or other site where a response to Ag can be developed, whether regulatory or effector. This requires the Ag-gathering function of immature DCs and their subsequent trafficking to secondary lymphoid tissue where this Ag can be presented. This aspect has not been investigated in studies to date. 

APCs and their expression of surface markers are known to be regulated by a variety of factors that produce substantial changes in the nature of the response of the lymphocytes that bind Ag on these APCs. ^{45 46 47} In contrast, IC inoculation of IFN-γ and anti-CD40 in the current study had relatively modest consequences for CD45⁺ cells in retina. Histology showed no evidence of infiltrates in eyes injected with these doses of IFN-γ or FGK45. Because the entire retina was harvested, migration within the retina was not a factor. The results are consistent with observations that tissue-resident MΦs have little response to stimulants, and that infiltrating MΦs mediate the disease. ^{16 47 48} We have considered the possibility that the increase in number of CD45⁺ cells after the inoculation of IFN-γ or anti-CD40 was due to infiltration of the retina rather than changes induced in existing retinal cells. To minimize contamination by cells in the vasculature, the mice were perfused with saline before the retinas were harvested. Furthermore, there was no evidence for an increase in the number of T or B cells by flow cytometry after the intraocular inoculations. To the extent that activation of MG ^{49 50 51} and PVCs ¹² results in loss of dendriform morphology and assumption of ameboid form, it is possible that extraction of the ameboid form during tissue dissociation is more efficient, leading to enhanced recovery from IFN-γ or anti-CD40–treated eyes. 

Some consequences of intraocular IFN-γ inoculation have been examined by others. These studies show that intracameral, intravitreal, or intraperitoneal IFN-γ inoculation in mice has little ability to upregulate MHC class II expression to levels detectable by immunohistochemistry and results in little infiltration. A single intraocular dose of 76,000 U followed by staining for MHC class II 5 days later produced only rare positive retinal cells and no infiltration. ⁵² Intraperitoneal inoculation of 50,000 U IFN-γ daily for 7 days in BALB/c mice had no effects in the retina, although the RPE was positive for MHC class II and substantial induction of expression was found in other ocular tissues. ⁵³ These studies did not further characterize the affected cells. Constitutive, transgenic expression of a small amount of IFN-γ in murine retinal photoreceptor cells had much more effect, producing significant class II⁺, Mac-1⁺, and NK⁺ infiltration in the inner retina and ganglion cell layers and subsequent retinal degeneration. ⁵⁴ In another study, IC inoculation of 100 U IFN-γ in mice promoted subsequent Ag-specific DTH response in the anterior chamber, but IFN-γ alone did not induce histologically detectable changes. ²³ This result further shows that the dose and route of IFN-γ we have used is able to produce a biologically significant effect in the eye. 

Our results are consistent in part with those of Robertson et al., ⁴⁷ who found that retinal MG were relatively insensitive to the developing inflammatory environment, probably due to their preconditioning in the TGF-β–rich retinal environment. They found that infiltrating MΦs function as though activated by TNF-α or IFN-γ and produce NO. Because our treatments were performed in vivo, their model would predict that TGF-β in the retinal environment would limit the response to IFN-γ, as we found. 

 Supported by National Eye Institute EY11542; Research to Prevent Blindness, Inc.; the Anna Heilmaier Foundation, St. Paul, Minnesota; and the Minnesota Lions and Lionesses Clubs.

 Submitted for publication October 2, 2002; revised December 20, 2002, and March 4, 2003; accepted March 11, 2003.

 Disclosure: D.S. Gregerson, None; J. Yang, 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.

 

Table 1.

 

Antibodies

Specificity	Clone/Ab	Conjugate	Source
B220	RA3-6B2	PE	BD-PharMingen*
CD3ε	145-2C11	Per-CP	BD-PharMingen
CD8α	53-6.7	Per-CP	BD-PharMingen
CD11b	M1/70	APC	BD-PharMingen
CD11c	HL3	FITC	BD-PharMingen
CD45	30-F11	Paramagnetic beads	Miltenyi, †
CD45	30-F11	PE	BD-PharMingen
CD80	16-10A1	APC	BD-PharMingen
CD95	Jo2	Per-CP	BD-PharMingen
DEC-205	NLDC-145	PE	Cedarlane, ‡
F4/80	Cl:A3-1	PE	Caltag, §
I-Ak	11.5.2.1	FITC, Per-CP	BD-PharMingen
CD40	FGK45	None	S. Schoenberger, †

Table 2.

 

Enrichment of CD45⁺ Cells by Retention on Magnetic Columns

Retinal Cells	Ungated Events			Size-Gated Events
		% CD45+	t-Test*	% CD45+	t-Test*
Unseparated	1.3 ± 0.4, †	—	2.3 ± 0.9	—
Column adherent	8.1 ± 2.9	0.028	20.1 ± 7.6	0.025
Column nonadherent	0.4 ± 0.3	0.051	0.2 ± 0.2	0.028

Figure 1.

 

Most CD45⁺ cells in retina were CD11b⁺. CD45 column-enriched cells were stained with the indicated mAbs. After flow cytometry, the events were size gated to minimize debris and numbers of non–bone-marrow–derived cells. (A) CD45⁺ cells divided into CD11b-positive (R1) or low/negative (R2). (B) CD45⁺ cells are divided into expression levels; R3 low (lo), R4 intermediate (int), or R5 high (hi). (C) The F4/80⁺ population substantially overlapped the CD11b⁺ population, as represented by the large double-positive population in the top right quadrant. n = 5 experiments.

Figure 2.

 

CD8α expression was concentrated in the CD11b⁺45^int population. CD45 column-enriched cells from retina, brain, and spleen were stained with the indicated mAbs. After flow cytometry, the events were size gated. (A) CD45^int11b⁺ cells from the R4 region of Figure 1 . (B) CD45^hi (contains CD11b^hi and CD11b^lo cells) cells from R5 region of Figure 1 . CD8α expression on the total CD11b⁺ cells from retina (C), brain (D), and spleen (E). (A, B) n = 5 experiments; (C–E) n = 2 experiments.

Figure 3.

 

Fas expression was found at higher levels on the total CD11b⁺ cells of retina (A, D) and brain (B, E), than on CD11b⁺ spleen (C, F) cells. CD45 column-enriched cells from retina, brain, and spleen were stained with the indicated mAbs. After flow cytometry, the events were size gated. (A–C) GMFI of Fas expression on all CD11b⁺ cells. (D–F). Percentage of Fas⁺ cells in retina and brain compared with spleen. The CD11b⁺ cells from the R9 regions of (A–C), respectively, were analyzed for Fas expression. n = 2 experiments.

Figure 4.

 

CD80 expression on MG versus CD45^hi cells. (A) CD45^int11b⁺ MG; (B) CD45^hi retinal cells; (C) control CD45⁻ retinal cells; and (D) CD45⁺ spleen cells. CD45 column-enriched cells from retina, brain, and spleen were stained with the indicated mAbs. After flow cytometry, the events were size gated. The CD45^int and CD45^hi cells were gated as shown in Figure 1B (regions R4 and R5, respectively), and analyzed for CD80. The CD45⁻ cells in (C) correspond to the R6 region of Figure 1A . n = 3 experiments.

Figure 5.

 

MHC Class II (I-A^k) expression on retinal CD11b⁺ cells compared with retinal CD11b⁻ cells or to CD11b⁺ LN cells. CD45 column-enriched cells were stained with the indicated mAbs. After flow cytometry, the events were size gated and analyzed for I-A^k expression. (A) CD11b⁺ cells corresponding to the cells in the R1 region of Figure 1A ; (B) CD11b⁻ cells corresponding to regions R2 and R6 of Figure 1A ; and (C) CD11b⁺ LN cells. (A, B) n = 6 experiments; (C) n = 3 experiments.

Figure 6.

 

DEC-205⁺ cells in the retina. CD45 column-enriched cells were stained with the indicated mAbs. After flow cytometry, the events were size-gated. (A) DEC-205 and CD11b expression; (B) staining for class II (I-A^k) and DEC-205. (C–E) Staining of DEC-205⁺ cells for CD80 (C), CD8α (D), or both (E). (A) n = 4 experiments; (B–E) n = 2 experiments.

Figure 7.

 

Intraocular inoculation of IFN-γ increased the number of CD45⁺ cells recovered from retina by approximately twofold. CD45 column-enriched cells were stained with anti-CD45. After flow cytometry, the events were size gated and analyzed for CD45⁺ cells. (A) Saline inoculated eyes; (B) eyes inoculated with 100 U IFN-γ. These are representative profiles from three experiments containing a total of seven determinations. Comparison of the percent recovery of CD45⁺ cells from the control versus IFN-γ–inoculated eyes gave P = 0.02, by t-test.

Figure 8.

 

Changes in the phenotype of CD45⁺ cells from retina after intraocular inoculation of IFN-γ. CD45 column-enriched cells were stained with mAbs specific for CD11b, CD80, CD8α, and I-A^k. After flow cytometry, the events were size gated and analyzed for expression of these markers. (A, B) The proportion of total CD11b⁺ cells (percent in both top quadrants: 7.2% ± 1.2% vs. 14.2% ± 5.7%; P = 0.015), and the percentage of CD11b⁺ cells positive for I-A^k expression (top right quadrant: 7.1% ± 7.0% vs. 19.6% ± 7.9%; P = 0.02) increased in the treated eyes. The results are representative of three experiments with a total of six determinations. (C, D) The level of CD80 expression (top right quadrant) did not increase significantly (two experiments with three total determinations).

Figure 9.

 

Inoculation of FGK45 (anti-CD40) caused a small increase in the recovery of CD45⁺ cells. Retinal cells were recovered from a density gradient (1.0875 g/cm³) and stained with anti-CD45. After flow cytometry (1 × 10⁶ events collected), the events were size gated and analyzed for CD45⁺ cells. Experimental conditions include: IC inoculation of saline (A); IC inoculation of mAb isotype (B); and IC inoculation of anti-CD40 (C). (D) Results of the isotype control mAb staining for the CD45 used in flow cytometry. Representative profiles from three experiments containing a total of seven determinations are shown. Comparison of the percentage recovery of CD45⁺ cells from the control (isotype and saline) versus anti-CD40 inoculated eyes. P = 0.01; by t-test.

Figure 10.

 

Increase in the recovery of CD11c⁺ cells after FGK45 inoculation. Retinal cells were recovered from a density gradient as shown in Figure 9 and stained with the indicated mAbs (anti-CD11c or isotype control). After flow cytometry, the events were size gated and gated on the CD45^hi versus CD45^int populations from Figure 9 for analysis of CD11c expression. (A, B) IC saline-inoculated eyes; (C, D) IC isotype control mAb-inoculated eyes; (E, F) IC anti-CD40–inoculated eyes; (G, H) isotype control mAbs for the anti-CD45 and anti-CD11c used in the flow cytometry. The CD11c⁺ cells were found in the CD45^hi region (F), but not in the CD45^int region (E), of the anti-CD40 samples. Because no difference was found between saline and isotype control inoculated eyes, these experiments were combined. The average increase in CD11c⁺ cells was approximately sixfold (four experiments; six total determinations; P < 0.01).

Figure 11.

 

Changes in the CD11c⁺ cells after treatment of the eyes with anti-CD40. Retinal cells recovered from a density gradient as in Figure 9 were stained with the indicated mAbs. After flow cytometry, the events were size-gated, gated on CD11c⁺ cells, and analyzed for the other markers. (A, B) CD11b; (C, D) CD80; and (E, F) I-A^k. The approximately twofold increase in CD45^hi11c⁺11b^lo cells after anti-CD40 treatment was significant (three experiments, four total determinations; P = 0.024).

Table 3.

 

Modulation of Surface Markers by Treatment In Vitro with IFN-γ or Anti-CD40

Specificity	Exp.	Samples	Medium (GMFI ± SD)	IFN-γ			Anti-CD40
								GMFI ± SD	t-Test*	GMFI ± SD	t-Test
CD45	3, †	10, ‡	609 ± 122	1088 ± 168	<0.001	574 ± 191	NS
CD11b	3	9	732 ± 501	1745 ± 863	0.017	859 ± 577	NS
CD11c	3	6	71 ± 47	54 ± 20	NS	49 ± 11	NS
I-Ak	2	5	35 ± 4	56 ± 10	0.002	56 ± 32	NS
CD80	2	6	60 ± 32	70 ± 24	NS	92 ± 56	NS
CD8α	3	5	55 ± 26	58 ± 25	NS	59 ± 23	NS

The authors thank Kristin Mitchell for assistance with the histology, Thien Sam for technical assistance, and Gary Birnbaum for a thoughtful critique of the manuscript.