Identification of a Novel Macrophage Population in the Normal Mouse Corneal Stroma

Abstract

purpose. To examine the normal murine corneal stroma for the presence of bone marrow-derived leukocytes.

methods. Wholemounts of paraformaldehyde-fixed corneal stroma from normal mice at 5 to 16 weeks of age were examined in single- and double-color immunomorphologic studies performed with confocal microscopy. The phenotype, morphology, distribution, and density of immunopositive cells were determined.

results. Numerous CD45⁺ cells with pleomorphic and dendriform morphology were found within the pericentral and central region of the corneal stroma (200–300 cells/mm²). Dual-color immunostaining demonstrated that 100% of the CD45⁺ cells coexpressed CD11b and 50% coexpressed F4/80. Approximately 30% of the total cells and 50% of the F4/80⁺ cells coexpressed major histocompatibility complex (MHC) class II antigens. Very small to negligible numbers of cells expressed markers of dendritic cells (CD11c) or granulocytes (Ly6G). Markers for T-cells and NK cells were absent from the corneal stroma, indicating that all the cells identified in the stroma were of the myeloid lineage.

conclusions. The normal murine corneal stroma contains a significant number of CD45⁺ leukocytes. Most these cells express the CD11b marker, but not other dendrite, granulocyte, T-cell, or NK markers, placing them in the monocyte/macrophage lineage.

Macrophages are hematopoietic cells that take up residence in virtually every tissue of the body and form a critical component of the innate immune response against potential pathogens. By virtue of their phagocytic capabilities, inflammatory cytokine secretion, and expression of a wide variety of activating surface receptors specific for pathogens or antigenic complexes, these cells are well equipped in their role as a first line of defense. ^{1 2 3} Macrophages also express low levels of major histocompatibility class (MHC) class II and costimulatory molecules, thereby enabling them to act as antigen-presenting cells (APCs), albeit less efficiently than dendritic cells. ^{1 3 4 5 6 7} In addition, although macrophages are most commonly assumed to have proinflammatory functions, certain macrophages contribute directly or indirectly to immune suppression. ^{8 9 10 11} For example, the phenomenon of anterior chamber-associated immune deviation (ACAID) is dependent on an eye-derived, suppressor-inducing macrophage that acts through NK T cells. ^{11 12}  

Previous studies of the eye have demonstrated that resident tissue macrophages are present in the iris, ciliary body, uvea, retina, conjunctiva, and corneal limbus. ^{13 14 15 16 17 18} In contrast, the normal avascular cornea is thought to be an immune-privileged site, without functional APCs and largely devoid of any bone marrow-derived cells, including macrophages. Supporting this concept has been the high success rate with which human corneal transplants are accepted. ¹⁹ In addition, early experiments with allogeneic transplants in mice demonstrated that the cornea apparently has no accessory cells capable of stimulating acute allogeneic rejection. ²⁰  

Earlier attempts to identify potential APCs in the cornea have described low to negligible numbers of bone marrow-derived cells, and typically, these have been mainly observed in the periphery of the corneal epithelium. MHC class II⁺ cells with a dendritic morphology have been noted frequently at the edge of the epithelium of guinea pigs, ^{21 22 23} mice, ^{22 23} rats, ^{22 23} and humans, ^{22 23 24 25 26 27 28 29} but only rare MHC class II-positive cells are observed in the central epithelium. Similarly distributed cells expressing CD45, a marker of hematopoietic lineage, have also been identified in the normal human corneal epithelium. ^{25 27} These resident cells are thought to be MHC class II⁺ Langerhans cells, a type of dendritic cell with potent APC function. Rare MHC class II⁺ or CD45⁺ cells have also been observed in the normal corneal stroma of guinea pigs, ²¹ rabbits, ²⁸ and humans, ^{26 27 29 30 31} but at an even lower frequency than seen in the epithelium. The cells are typically found in the anterior third of the peripheral stroma and are absent from the central stroma. Some of these rare stromal cells have been observed to have dendritic cytoplasmic processes, but their exact lineage has not been determined. There are no published reports of similar examinations of the normal mouse corneal stroma. 

It is not inconceivable that previously undetected resident macrophages exist in the main body of the murine cornea. The widespread areas of macrophage distribution include the brain and testis, two sites that are also considered to be immune-privileged. ^{32 33} Further, passenger macrophages within corneal transplants may be functionally undetected, because certain subsets of resting macrophages express very little MHC class II and/or costimulatory molecules and are extremely inefficient at activating T lymphocytes, ^{5 34 35} or, alternatively, because macrophages from certain microenvironments can be immunosuppressive in function ^{8 9 10 11} and would not elicit an observed response. Moreover, immunohistochemical detection of significant numbers of bone marrow-derived cells in the central corneal stroma of other species may have been hampered in previous studies by the use of transverse sections. It has been noted that under these conditions, flattened macrophages embedded in connective tissue or collagen, can be difficult to detect. ^{36 37} We are unable to detect spreading dendritic cells in transverse sections of corneal epithelium (Hendricks RL, unpublished data, 1998), whereas these cells are readily visible in corneal flatmounts. ³⁸ We describe in this report the results of our studies examining flatmounts of the normal mouse corneal stroma for the presence of resident leukocytes and macrophages. 

Methods

Animals

Four- to 5-week old female BALB/c mice weighing 15 to 20 g were purchased from NCI-Frederick Cancer Research and Development (Frederick, MA). Animals were housed in a specific pathogen-free facility and maintained on a 12-hour light-12-hour dark cycle. All experiments complied with the guidelines set forth by the Institutional Animal Care and Use Committee of the University of Pittsburgh and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Experiments were performed in mice that were between 5 and 16 weeks of age, as indicated. 

Reagents and Antibodies

Chemicals and reagents were purchased from the following companies: DMEM and penicillin-streptomycin, BioWhittaker (Walkersville, MA); fetal bovine serum (FBS), Hyclone (Logan, UT); bovine serum albumin (BSA) and bovine gelatin, Sigma (St. Louis, MO); goat serum, Atlanta Biologicals (Norcross, GA); EDTA, Fisher Scientific (Fair Lawn, NJ); NP-40, Calbiochem (La Jolla, CA); paraformaldehyde, Electron Microscopy Sciences (Fort Washington, PA); ketamine hydrochloride and xylazine, Phoenix Pharmaceutical, Inc. (St. Louis, MO); halothane, Halocarbon Laboratories (River Edge, NJ). Biotinylated and FITC-conjugated anti-CD45 antibody (clone 30-F11), FITC-conjugated anti-CD11b (clone M1/70), FITC-conjugated anti-Ly6G (clone RB6-8C5), biotinylated anti-NK cells (clone DX5), PE-conjugated anti-CD3 (clone 17A2), FITC-conjugated rat IgG2b control (clone A95-1), hamster IgG control (clone A1g-3), biotinylated rat IgM (clone R4-22), Fc-Block antibody (anti-CD16/CD32), streptavidin-FITC, and streptavidin-horseradish peroxidase (HRP) were all purchased from BD PharMingen (San Diego, CA). Biotinylated anti-F4/80 (clone A-1) and biotinylated goat anti-hamster IgG antibodies were purchased from Serotec, Ltd. (Oxford, UK). Unconjugated anti-CDllc antibody (clone N418) was purchased from Endogen (Woburn, MA). Biotinylated rat IgG2b control was purchased from Caltag (Burlingame, CA). The anti-MHC class II hybridoma (clone M5/114.15.2) was obtained from the American Type Culture Collection (ATCC; Manassas, VA), and purified antibody was conjugated with dye from Alexa Fluor 488; Molecular Probes (Eugene, OR), according to the manufacturer’s protocol. Purified rat IgG (Sigma) was conjugated with Alexa Fluor 488 as a control. Streptavidin-Cy3 was purchased from Jackson Laboratories (West Grove, PA). 

Immunostaining of Corneal Tissue

Mice were euthanatized by halothane inhalation and the corneas excised. The normal corneas were examined under a dissecting microscope, and only those that showed no signs of inflammation or other abnormalities were used in the studies. Any attached lens, conjunctiva, and excess limbal tissue were removed from the corneas. The corneal stroma and epithelium were then separated after a 20-minute incubation at 37°C in PBS containing 20 mM EDTA. The following general procedure was used for all antibodies, with the exception of CD11c and DX5. After separation, the corneal stromas were fixed for 30 minutes at 4°C in 1% paraformaldehyde-PBS followed by extensive washing with PBS. After fixation, the corneal tissue was blocked for 20 minutes at 37°C with 10 μg/mL Fc-Block and 10 μg/mL rat IgG diluted in PBS-BGEN (PBS containing 3% BSA, 0.25% gelatin, 5 mM EDTA, and 0.025% Nonodet-P40, a nonionic detergent). After the blocking step, corneal tissue was incubated overnight at 4°C with 100 μL primary antibody (1–5 μg/mL) diluted in PBS-BGEN. The tissue was then washed 5 times, 5 minutes each, with PBS. Tissues stained with fluorescently labeled primary antibody were fixed again with 1% paraformaldehyde-PBS, for 30 minutes, at 4°C, rinsed with PBS, placed on slides, mounted with Immu-Mount mounting medium (Shandon, Pittsburgh, PA), and coverslipped. Alternatively, tissues that were reacted with a biotinylated primary antibody were further incubated with 100 μL of fluorescently labeled streptavidin (1–2.5 μg/mL) diluted in PBS-BGEN for 1 hour at 37°C. This was followed by five washes of 5 minutes each with PBS, fixation, and mounting onto slides. Tissue to be reacted with DX5 was kept unfixed until the antibody labeling procedure was complete. The fresh tissue was first blocked with 10 μg/mL Fc-Block, 20 μg/mL rat IgG, and 20 μg/mL rat IgM. After blocking, the tissue was incubated with biotinylated DX5 (5 μg/mL) diluted in PBS-BGEN for 3 hours, at 4°C. The tissue was then washed five times for 5 minutes each with cold PBS, followed by incubation with fluorescently labeled streptavidin (1–2.5 μg/mL) diluted in PBS-BGEN for 2 hours at 4°C. This was followed by washing with cold PBS, fixation with 1% paraformaldehyde, and mounting. For visualization of CD11c staining, tyramide amplification was performed with a kit (TSA Direct; NEN, Boston, MA) according to the manufacturer’s recommendations. In brief, tissue to be reacted with the CD11c antibody was blocked with 10 μg/mL Fc-Block, 10% human serum, and 10% goat serum diluted in TNB buffer (1M Tris-HCl, 0.15M NaCl, and 0.5% “Blocking Reagent” supplied with the TSA Direct Kit). The tissue was reacted in sequential incubation and washing steps with unlabelled CD11c antibody (1 μg/mL), followed by biotinylated goat anti-hamster IgG (5 μg/mL), streptavidin-HRP (1:100), and the fluorophore tyramide (1:50). All antibody incubations were performed in TNB buffer. Endogenous peroxidase activity in the tissue was quenched with 3% H₂O₂ before staining. No staining was observed with isotype-matched negative control antibodies, which were included in each experiment. All slides were examined by fluorescence microscopy on an 1 × 70 microscope (Olympus, Tokyo, Japan) equipped with a confocal imaging system (Radiance Plus; Bio-Rad, Hercules, CA). Digital images were captured using the scanning confocal laser and the accompanying software (Lasersharp 2000; Bio-Rad). 

Enumeration of Cell Number in the Corneal Stroma

Series of multiple Z-sections were generated, which collected images from all depths of the stroma that contained antibody-reactive cells. Merging the stacked Z-sections using the software (Lasersharp 2000; Bio-Rad) then created a single image. The manual counting tool of an image-analysis program (Image-Pro Plus; Media Cybernetics, Silver Spring, MD) was then used to determine the number of positive cells within the merged image. For some analyses, the presence of high background, low-intensity stain, or high density of cells made it difficult to enumerate cells accurately within a merged image. In these cases, the images were evaluated directly within the confocal imaging program (Lasersharp 2000; Bio-Rad). A transparency sheet was placed on the computer screen, and individual Z-sections within a series were examined, one at time and in order. Cells staining for one or both markers examined were scored individually on the transparency with different colored labeling pens for each marker. The entire Z-series was then enumerated as one score. Examining separate Z-sections in this manner made it easier to visualize positive cells and at the same time to ensure that they were not scored more than once. 

Virus Infection

The RE strain of herpes simplex virus (HSV)-1 was used for corneal infection. Viral particles were purified, as previously described, ³⁹ and the viral titer determined by a standard viral plaque assay. BALB/c mice, 7 weeks old, were anesthetized by intramuscular injection of 1.8 mg ketamine hydrochloride and 0.18 mg xylazine. Corneas were scarified 10 to 15 times in each direction of a crisscross pattern with a sterile 30-gauge needle. A 3-μL volume of HSV-1 suspension (containing 1 × 10⁵ plaque forming units) in RPMI-1640 was applied topically to the scarified cornea. The inflamed corneas were then harvested at 3 or 14 days after infection, as indicated in the Results section. 

Results

Bone Marrow-Derived Cells in the Normal Corneal Stroma

To determine whether the corneal stromas of normal mouse eyes contain bone marrow-derived cells, we performed fluorescent immunohistology with antibodies to various leukocyte markers. The results of confocal microscopic analyses of corneal stroma flatmounts reacted with antibodies to CD45, CD11b, F4/80, MHC class II, and CD11c are shown in Figure 1 . Numerous cells that expressed the pan-leukocyte marker CD45 ⁴⁰ were observed in the pericentral corneal stroma (Fig. 1A) . This finding was surprising, given previous reports that few, if any bone marrow-derived cells are detectable by immunohistology on transverse sections of the normal cornea. Expression of CD45 was not limited to the peripheral stroma, where cellular reactivity was densest, but extended through the pericentral region (see Fig. 2A ) and into the central innermost third of the corneal stroma (Fig. 2B) . Few cells were present at the very epicenter of the stroma. CD45⁺ cells were observed throughout the entire depth of the stroma. The highest density of cells occurred approximately in the anterior most third of the stroma and the posterior third closer to the endothelium, with the fewest cells seen in the stroma midsection. The lateral distribution of CD45⁺ cells in the pericentral region of the corneal stromas was fairly uniform, but in some corneas, sporadic, densely populated areas were interspersed with areas that had fewer positive-reacting cells. The morphology of the leukocytes in the stroma was highly variable. Cells with round, irregularly oblong, and dendriform shapes were visualized within a single cornea. In the anterior stroma, the cells frequently appeared flattened, diffusely stained, and highly dendriform in shape. These may represent cells at the epithelium-stroma interface. The cells in the central area were frequently more difficult to detect, because of a very spread-out morphology and much weaker and patchy staining. However, by using the confocal microscope to section optically through the tissue, it was possible to identify the cells in this region positively. The density of CD45⁺ cells in the corneal stroma was determined by enumerating the number of cells present in confocal images having a known area (Table 1) . Stromas from mice that were 5 to 6 weeks in age averaged 213 cells/mm² (range, 150–392) in the pericentral area and 199 cells/ mm² (range, 92–423) in the center. Stromas from mice 16 weeks in age had 323 cells/mm² (range, 287–371) in the pericentral area and 308 cells/ mm² (range 257–337) in the center. The difference between the pericentral counts of the two age groups was statistically significant (P < 0.05, unpaired two-tailed t-test). There was a trend toward fewer cells in the central area, but this variation was not statistically significant. 

Coexpression of Leukocyte Markers on Bone Marrow-Derived Cells in the Corneal Stroma

To identify the leukocyte subpopulations present, normal corneal stromas were stained with antibodies to various leukocyte differentiation markers (Fig. 1B) . The antibody CD11b, which detects complement receptor 3 (CR3) on monocytes, macrophages, dendritic cells, granulocytes, NK cells, and a subpopulation of T lymphocytes, ⁴¹ identified large numbers of cells in the stroma. Simultaneous two-color staining with CD45 and CD11b determined that 100% of the CD45⁺ cells expressed CD11b (Table 2) . The stroma also contained cells expressing the marker F4/80 (Fig. 1C) , a glycoprotein that is specifically restricted to subsets of macrophages, monocytes and dendritic cells. ^{42 43} There was a wide variation noted in the surface intensity of F4/80 on cells. Some cells were quite brightly stained, whereas others expressed very low levels of this antigen. The distribution of F4/80-reactive cells in the corneal stroma was less uniform than the distribution of the total CD45/CD11b⁺ population. The F4/80⁺ cells tended to occur in clusters of relatively high density, whereas other portions of the same stroma contained fewer positive cells. There was no apparent radial gradient of F4/80 expression between the periphery and the center. Overall, approximately 52% ± 12% of leukocytes in the corneal stroma were F4/80⁺, as determine by combined staining with CD45 and F4/80 (Table 2) . That all the cells in the stroma reacted with CD11b and many reacted with F4/80 suggests that most of the cells probably belonged to either the macrophage/monocyte or dendritic cell lineage. These two populations were distinguished by staining corneas with antibody to the integrin CD11c, which in the mouse is expressed on all dendritic cells. ^{43 44 45 46} However, only rare CD11c⁺ cells (<20 cells per cornea) were observed in the pericentral or central stroma (Fig. 1D) . Some CD11c⁺ cells were observed at the limbus and peripheral edge of the cornea, consistent with previous reports describing dendritic cells in this area. As a positive control, corneal stromas from mice with inflamed HSV-1-infected corneas (day 14 after infection) were reacted with the CD11c antibody. In contrast to normal corneas, large numbers of CD11c⁺ dendritic cells were readily observed in the pericentral region of infected corneas (Fig. 1E) . The slight blur to the images shown is due to the tyramide amplification process used to aid in visualizing CD11c antibody staining. Thus, it appears that dendritic cells represent only a very minor portion of the leukocytes that are resident in the normal mouse corneal stroma. Antibody to the pan-granulocyte marker Ly-6G ⁴⁷ occasionally detected a scattered grouping of small numbers of cells in the pericentral stroma, but the overall density per corneal stroma was lower than 25 cells per mm² (data not shown). T lymphocytes and NK cells were completely absent from the normal corneal stroma, in that no staining was observed with antibodies to the pan-T-cell marker CD3 ⁴⁸ or the pan-NK cell marker DX5 ⁴⁹ (data not shown). As a positive control for these antibodies, we examined HSV-1-infected corneas. Numerous CD3⁺ cells were detected in corneas 14 days after HSV-1 infection. We were also able to image DX5⁺ cells specifically in unfixed corneas at 3 days after HSV-1 infection (data not shown, see the Methods section for techniques used). Taken together, the data indicate that most of the cells detected in the normal corneal stroma were myeloid in lineage. 

The leukocytes’ potential for antigen presentation in the stroma was assessed by staining with antibody to MHC class II (Fig. 1F) . MHC class II was coexpressed on 30% ± 12% of CD45⁺ cells (Table 1) . Two populations of MHC class II-expressing cells were noted: One stained very weakly (MHC II^dim), whereas another exhibited intense staining (MHC II^bright). Approximately 70% of MHC class II⁺ cells were MHC II^dim, and the remaining 30% were MHC II^bright. Expression was limited to CD45⁺ leukocytes and was not observed on stromal keratocytes. The MHC class II stain revealed cells with a remarkably heterogeneous morphology, and it appeared that this marker efficiently stained projecting dendritic processes. Many cells had a flattened, slightly dendritic shape, others were more rounded or oblong, and some had an extremely elongated shape. To determine whether MHC class II was restricted to a subset of CD45⁺ cells, coexpression of MHC class II with F4/80 was also examined. It was calculated that 50% ± 15% of F4/80⁺ cells were MHC class II⁺, accounting for approximately 80% of the total MHC class II⁺ cells. Cells that were MHC class II⁺ (both MHC II^dim and MHC II^bright) but F4/80⁻ were also observed but were fewer in number than F4/80⁺/MHC class II⁺ cells. 

Discussion

In this study, the normal murine corneal stroma contained a significant population of resident leukocytes, as defined by the leukocyte common antigen, CD45. Virtually all these cells coexpressed the integrin marker CD11b, which detects monocytes, macrophages, dendritic cells, polymorphonuclear neutrophils, NK cells, and a subpopulation of T cells. We were unable to detect cells bearing markers of NK cells or T cells, and only rare cells expressed granulocyte markers. Very few cells were found to express the pan-dendritic cell marker CD11c. Based on this phenotype, we conclude that most of the endogenous leukocytes in the normal corneal stroma are monocytes or macrophages. Our finding that approximately 50% of the CD45⁺ cells in the stroma also expressed the F4/80 antigen, a lineage-restricted marker found only on subsets of monocytes, macrophages, and dendritic cells, ^{42 43} provided further evidence of a monocyte/macrophage lineage. MHC class II was expressed by 30% of the CD45⁺ cells in the stroma and was found on 50% of the F4/80⁺ cells. Approximately one third of the MHC class II⁺ cells were MHC II^bright, whereas the remaining MHC II⁺ cells expressed very low levels of this antigen and were considered MHC II^dim. 

Our demonstration that the normal mouse corneal stroma contains a significant number of macrophages appears to contradict results demonstrating very few, if any, leukocytes in the stroma in other species (guinea pig, rabbit, rat, and human). ^{21 26 27 28 29 30 31} It is possible that macrophages are a unique feature of the mouse corneal stroma. However, macrophages have been found in virtually every other tissue of the body, including the brain and testis, which are also considered to be immune-privileged sites. ^{32 33} It seems likely that the difference in results relates to technical aspects of the studies. Our use of whole corneal stromal flatmounts for immunohistology, as opposed to the commonly used transverse cross sections, appears to be critical for visualization of stromal macrophages. Many of the endogenous leukocytes appeared spread out and flattened in the stroma, possibly because of the lamellar structure of the stromal matrix. ⁵⁰ This could have impeded previous attempts to identify cells in transverse sections. Further, typical cross sections are only 4 to 6 μm thick, whereas most of the cells that we observed were larger than 10 μm in lateral diameter and were highly irregular in shape, thus making it likely that much of the defining exterior membrane of an individual cell could be lost during sectioning. Other investigators have also reported that macrophages spread in the plane of a surface or flattened between collagen fibers are difficult to image in cross sections. ^{36 37} We also determined that light fixation of the tissue with 1% paraformaldehyde elicited optimal signal with the antibodies used, in particular anti-CD11c and anti-CD45, which possess epitopes that are sensitive to 4% paraformaldehyde fixation. The use of confocal microscopy was another important aspect in our identification of the leukocytes that populate the normal corneal stroma. Confocal imaging of consecutive Z-sections through the stroma (en face to the plane of the flattened cornea) enabled us to distinguish weakly stained cells from background. 

Another extremely critical observation was that the endogenous leukocytes in the corneal stroma appear to express high levels of Fc-receptors, and staining of the stroma with most rat antibodies (control or specific) in the absence of Fc-Block gave rise to high numbers of cells staining nonspecifically. Pretreatment of the tissue with Fc-Block eliminated nonspecific staining, enhancing resolution of positive cells. This Fc-mediated artifact, however, supports our conclusion that most of the leukocytes in the stroma are macrophages or monocytes, because these are known to express Fc receptors. ⁵¹  

Our results are reminiscent of those in previous reports demonstrating the presence of significant numbers of macrophages and dendritic cells in the uveal tract of the rat and murine eye. Two main groups of leukocytes have been identified in these tissues: cells expressing high levels of MHC class II which have no macrophage-restricted markers and are presumably dendritic cells, and resident macrophages, which in the rat were defined by the macrophage markers ED1 or ED2, ^{13 14 15 52 53} and in the mouse were defined by markers for F4/80 or sialoadhesin. ^{14 18} Approximately half of the myeloid cells we identified in the stroma expressed F4/80, whereas all the cells reacted with the marker CD11b. In blood, lymphoid tissues, and many nonlymphoid sites (including the kidney, liver and gastrointestinal tract), the F4/80 antigen is expressed by many resident monocytes and macrophages. ^{37 38 44 54 55} However, not all macrophages are F4/80⁺, and F4/80⁻/CD11b⁺ macrophages have been described in distinct areas, such as the marginal zone of the spleen, the optic nerve and the connective tissue of the lung. ^{56 58 59} The heterogeneity in the F4/80 phenotype of myeloid cells of the corneal stroma may suggest potential functional differences in these subpopulations, although a specific function for the F4/80 antigen itself has not been identified. Expression of F4/80 antigen is downregulated after activation of macrophages with inflammatory stimuli, ^{44 60 61 62} suggesting that the F4/80⁻ cells of the corneal stroma could be activated macrophages. This seems doubtful, however, because the mice were not exposed to any known inflammatory stimuli. Furthermore, stimuli that downregulate F4/80 expression are associated with upregulation of MHC class II expression. ^{60 61 62} The majority (∼90%) of the F4/80⁻ cells in the stroma did not express detectable MHC class II and are thus unlikely to be activated cells. 

Prior studies of uveal tract tissues reported very little coexpression of MHC class II on cells that expressed macrophage markers. We determined that 30% of the myeloid cells in the corneal stroma were MHC class II⁺, although approximately 70% of these cells were MHC class II^dim. It is likely that the highly sensitive confocal microscopy technique used on clear corneal tissue allowed identification of MHC class II^dim cells, which may have counterparts in other ocular tissues. In the iris, the number of macrophages exceeds the number of dendritic cells, and the density of each cell type appeared to exceed that which we found in the corneal stroma. Approximately 600 to 700 macrophages/mm² and 400 to 500 MHC II⁺ cells (presumed dendritic)/mm² were observed in the iris and ciliary body stroma, ^{14 52 63 64 65} compared with the approximately 200 to 300 CD45⁺/CD11b⁺ cells/mm² that we enumerated in the corneal stroma. However, the morphology of the cells that were identified as macrophages in the uveal tract was strikingly similar to our observations. The macrophages were described as having pleomorphic or dendriform morphology and often had very elongated veillike processes. ¹³ Smaller rounded cells were also observed. The resident macrophages of the iris and ciliary body have been observed to have a predominantly perivascular location, suggesting that they are distributed in the tissue primarily after emerging from the supplying vasculature. ¹⁴ In this respect, the presence of macrophages in the central corneal stroma is somewhat enigmatic. Because the normal cornea is avascular, our observations may suggest selective chemoattraction of macrophages into the normal cornea. 

A recent abstract reported that significant numbers of CD11c⁺ dendritic cells are present in the corneal stroma of normal BALB/c mice. ⁶⁶ However, we observed very few CD11c⁺ cells in the corneal stroma of these mice. In preliminary experiments, we observed very weak staining of CD11c on subpopulations of cells in HSV-inflamed stromas, although this staining could be readily detected by confocal microscopic analysis. To rule out the possibility that we failed to detect weakly stained CD11c⁺ cells in normal corneal stromas, we used tyramide amplification to augment the CD11c signal. With tyramide amplification, strong staining for CD11c was observed in HSV-1-infected corneal stromas. However, even under these conditions, we observed few to no CD11c⁺ cells in the normal stroma. Although CD11c is thought to be expressed on all mouse dendritic cells, ^{43 44 45 46} it has been suggested that some peripheral dendritic cell populations in nonlymphoid organs such as the heart may express no CD11c. ⁶⁷ However, a study that used more sensitive detection techniques reported that heart dendrites expressed low levels of CD11c. ⁶⁸ Given the highly sensitive detection technique used in our studies, we conclude that few true dendritic cells are present in the normal mouse corneal stroma. 

We have not yet investigated the function of the endogenous macrophages in the cornea. Resident tissue macrophages exist in many sites, but their exact roles are not well defined. This is in part because cells of the macrophage lineage perform an exceptionally wide range of critical activities. Macrophages are highly phagocytic, facilitating their role in clearing tissue of damaged, infected, or senescent cells. ^{1 2 3} Macrophages also perform essential tasks in wound healing by directing tissue remodeling through the secretion of appropriate enzymes and growth factors, such as collagenase, elastase, and fibroblast growth factor. ^{69 70 71} This could be a particularly critical function for macrophages in the cornea, whose anatomic structure requires rapid healing after physical damage. 

Macrophages perform numerous important immunologic functions. Resident stromal macrophages may provide a critical first line of defense against pathogens that breach the epithelial barrier of the cornea by producing antimicrobial substances, such as nitric oxide and TNFα and other inflammatory cytokines and chemokines that attract and activate additional macrophages and leukocytes. ^{1 2 3} Recent evidence suggests that endogenous macrophages may play a role in the early control of HSV-1 infections in the cornea. Subconjunctival injection of clodronate liposomes that selectively deplete macrophages resulted in an increase in the severity and duration of HSV-1 epithelial keratitis, as well as a delay in clearance of live virus from the infected eyes. ⁷² Resident macrophages may also have an APC function. Although most of the endogenous macrophages in the normal stroma do not express MHC class II, they may upregulate expression of these molecules after encounter with pathogens, permitting direct antigen presentation to infiltrating T cells. Also, the macrophages in the corneal stroma may possess APC function for activated/memory T cells, despite low MHC class II expression. For instance, antigen pulsed macrophages from the iris and choroid (most without detectable MHC class II) can activate antigen-primed, but not naive T cells. ^{15 73} Thus, the macrophages in the stroma, perhaps even those with low MHC class II expression, may have similar potential as APCs for secondary immune responses. Alternatively, resident macrophages may indirectly influence APC function in the cornea in two ways: by producing cytokines such as IL-1, IL-6, and TNF-α that induce neovascularization ^{74 75 76} and dendritic cell migration ^{77 78} and by transferring membrane antigens or antigenic peptides from phagocytosed pathogens to infiltrating APC through the process of cross-presentation. ^{79 80} A recent study demonstrated that subconjunctival injection of macrophage-depleting clodronate liposomes at the time of HSV-1 corneal infection reduces the severity of stromal keratitis and reduces the delayed-type hypersensitivity (DTH) response to HSV antigens. ^{72 81} Because HSV-induced stromal keratitis and DTH in the mouse involves activation and cytokine production by CD4⁺ T cells, ^{82 83 84 85 86} these findings are consistent with a direct or indirect APC function for resident macrophages. 

Instead of activating T cells, macrophages in the stroma may contribute to the immune-privileged status of the cornea by downregulating the T-cell response to antigens acquired within the tissue. Macrophages that are exposed to transforming growth factor (TGF)-β can induce tolerance to processed antigens when adaptively transferred to naïve recipients. ^{87 88 89} Indeed, iris macrophages that are exposed to TGF-β in the aqueous humor are thought to contribute to the immune privilege of the anterior chamber through induction of ACAID. ^{11 87} TGF-β is also present in the cornea ^{89 90 91} and may confer an ACAID-inducing phenotype on the corneal macrophages. However, the development of ACAID can be prevented by the inflammatory cytokines IL-1, IL-6, and TNF-α. ^{92 93} Corneal cells have been shown to produce IL-1, IL-6, and TNF-α in response to certain stimuli, such as bacteria or lipopolysaccharide. ^{94 95 96} Thus, resident corneal macrophages may enhance or inhibit T cell responses, depending on the particular conditions in the cornea at the time antigens are acquired. Characterization of the immunostimulatory or immunosuppressive properties of resident corneal macrophages could have important implications for understanding immunity and immunopathology in the cornea, as well as immunologically mediated corneal graft rejection. 

In summary, the observations reported herein demonstrate that the corneal stroma contains numerous endogenous leukocytes, the extent of which has been previously unappreciated. Based on single- and dual-color phenotype analysis, we conclude that most of these cells are monocytes or macrophages. The potential function of these cells in maintaining corneal homeostasis and protection against pathogens remains to be determined. 

 Supported by Grant EY10359 and Core Grant EY08098 from the National Eye Institute, National Institutes of Health, Bethesda, Maryland; Research to Prevent Blindness; and the Eye and Ear Foundation of Pittsburgh, Pennsylvania.

 Submitted for publication November 16, 2001; revised February 19, 2002; accepted March 11, 2002.

 Commercial relationships policy: N.

 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.

 

Immunofluorescent staining of corneal stroma flatmounts: the normal stroma contains numerous cells expressing leukocyte markers. (A) CD45⁺ cells in the pericentral stroma (21 stacked Z-sections); (B) CD11b staining in the pericentral stroma (15 stacked Z-sections); (C) cluster of F4/80⁺ cells in the pericentral stroma, with the edge of a radial cut seen at the bottom and left corner (13 stacked Z-sections); (D) rare CD11c⁺ cell in the central stroma (6 stacked Z-sections); (E) CD11c⁺ cells in pericentral stroma of HSV-1 infected cornea (8 stacked Z-sections); and (F) MHC class II⁺ cells in the pericentral stroma (14 stacked Z-sections). All images were taken at medium-power magnification with a 40× objective. Contrast has been enhanced in some images to illustrate cells with weaker staining. Corneas were obtained from mice that were 5 to 6 weeks in age.

Figure 2.

 

Low-power view of the corneal stroma demonstrating the location of CD45⁺ cells. (A) View of the pericentral region (PC) in one of the four corneal leafs created after the placement of four radial cuts in the corneal stroma. Arrowheads: indicate the edge of one such cut. Dashed line: an approximate demarcation between the peripheral and pericentral stroma. The periphery and edge of the cornea lies above the line in this image. The area referred to as pericentral lies below the line. (B) View of the central region (C) in a stroma. This was considered to be the radial epicenter of the corneal stroma, having an approximate diameter of 1200 to 1400 μm, and lying inside the radial cuts created in the stroma. Dashed line: circumference of part of this region. Arrowhead: terminus of one radial cut extending inward from the edge. Both images were taken with a 10× objective.

Table 1.

 

Enumeration of CD45-Expressing Cells in the Corneal Stroma

Age*	Pericentral (n), †	Central (n), †
5–6 Weeks	213 ± 86 (16)	199 ± 120 (16)
16 Weeks	323 ± 36 (4)	308 ± 36 (4)

 Data are expressed as average density per square millimeter ± SD.

*  Age of the animals at the time of corneal harvest.

†   n, Number of stromas evaluated, with a total of four counted fields per pericentral region, and one count from the central area per corneal stroma.

Table 2.

 

Coexpression of Leukocyte Antigens on Cells in the Corneal Stroma

	CD45+CD11b*	CD45+MHC Class II, †	CD45+F4/80, ‡	F4/80+MHC Class II, §
% Cells expressing first marker that coexpress the second marker	100% (6), ∥	30 ± 12% (7)	52 ± 20% (7)	50 ± 15% (6)

 The antibody combinations were used in dual staining experiments.

*  Stroma costained with anti-CD45-biotin + Sa-Cy3 and anti-CD11b-FITC.

†  Stroma costained with anti-CD45-biotin + Sa-Cy3 and anti-MHC class II-488.

‡  Stroma costained with anti-CD45-FITC and anti-F4/80-biotin + SA-Cy3.

§  Stroma costained with anti-F4/80-biotin + SA-Cy3 and anti-MHC class II-488.

∥  Number of stromas examined. Four to five field images were enumerated per stroma, with the exception of the F4/80 versus MHC II group. In these, the number of images evaluated per corneal stroma ranged from one to six, due to variations in frequency of F4/80⁺ cell clusters in each stroma.