Lack of IFN-γ Synthesis in Aqueous Humor during Corneal Graft Rejection Correlates with Suppressed Nitric Oxide Production by Macrophages

Abstract

purpose. To investigate cytokine production by leukocytes in aqueous humor (AH) during corneal graft rejection and nitric oxide (NO) production by macrophages as a potential mediator of graft damage.

methods. Rats received corneal allotransplants and were killed during acute rejection. Leukocytes in AH that expressed tumor necrosis factor (TNF)-α, interferon (IFN)-γ, and interleukin (IL)-10 were quantified by flow cytometry. Isograft and further allograft recipients were killed, and sectioned corneas with conjunctivae were examined by histology for production of inducible nitric oxide synthase (iNOS), NO, and nitrotyrosine (NT).

results. Between 80% and 90% of T cells, NK cells, and macrophages in AH expressed TNF-α, and at least 20% expressed IL-10. However, IFN-γ was undetectable unless cells were first stimulated in vitro with PMA and ionomycin, which yielded IFN-γ in 25% of cells. iNOS⁺ macrophages were identified in donor cornea and AH, correlating precisely with rejection. Cells producing low levels of NO (NO^dim cells) were found in donor stroma, but NT⁺ cells were rare. Both NT⁺ and NO⁺ cells were rare in the anterior chamber (AC) or attached to corneal endothelium. NT⁺ macrophages that were also NO^bright were associated with sutures in allograft and isograft recipients and within conjunctivae, either scattered or in leukocyte aggregates.

conclusions. IFN-γ synthesis is lacking in the AC during rejection, correlating with lack of NO but not of iNOS expression. NO does not appear to mediate endothelial cell death. NT and high levels of NO production are associated with nonspecific inflammatory cells.

A corneal graft is unusual among transplants in that influx of inflammatory cells can occur via two routes: blood vessels in conjunctiva and/or a neovascularized cornea convey leukocytes to the stroma and epithelium, whereas the endothelium is most likely to be targeted by cells released into the aqueous humor (AH) from the ciliary body and/or iris. Corneal graft rejection is considered to be mediated by CD4⁺ T cells, but precise effector mechanisms have yet to be established, and it is possible that these differ within each cell layer. ^{1 2} AH, in particular, contains a range of agents of proven immunoregulatory potential, ^{3 4 5 6 7 8} some or all of which may influence the composition and activity of cells entering the anterior chamber (AC). However, whether or in what fashion they can alter the manifestations of an established immune response such as graft rejection has been little studied. 

Analysis of the leukocyte content of AH and thereby of corneal endothelial rejection has been constrained by the small yield of cells. ^{9 10 11} Flow cytometry has so far been used to quantify lymphocytes and natural killer (NK) cells in the AH of rats. ^{10 12} The most comprehensive study to date noted a high proportion of CD161⁺ (NK) cells (25%), but did not address cytokine production. ¹²  

Macrophages are potential effector cells in a CD4-mediated response, but evidence of a cytotoxic role of these cells in rejection is conflicting. A major cytotoxic product of macrophages is nitric oxide (NO) and expression of the enzyme inducible nitric oxide synthase (iNOS) has been demonstrated by immunohistochemistry in rejected grafts in mice ¹³ and rats. ¹⁴ Use of nitrotyrosine (NT) as a surrogate marker for the cytotoxic NO product peroxynitrite revealed NT⁺ cells in the stroma and endothelium, the presence of which correlated with rejection, ¹⁴ and corneal graft survival was considered prolonged (at P = 0.67) after treatment of recipients with the iNOS inhibitor aminoguanidine. ¹³ Other studies have cast doubt on macrophages as effector cells: depletion studies in both rats ¹⁵ and mice ¹ suggest that macrophages are involved in the initial activation of T cells rather than as effector cells, and cytotoxicity against donor splenocytes has been attributed to the NK cell fraction of AH rather than to T cells or macrophages. ¹²  

The purpose of this study was to investigate cytokine and cell surface phenotypes of cells infiltrating the AH at maximum rejection and whether macrophages mediate the cytotoxicity of graft cells via NO synthesis. The cytotoxic action of NO must be distinguished from potentially protective effects (e.g., inhibition of lymphocyte proliferation ¹⁶ or of inflammation provoked by apoptotic cells). ¹⁷ Cytotoxicity is achieved predominantly via peroxynitrite-mediated lipid peroxidation of cell membranes, which cannot be visualized. However, peroxynitrite synthesis can be inferred by its effect in nitrating protein tyrosine residues, detectable by immunohistochemistry with anti-NT antibodies. To facilitate analysis, a rat strain combination with rapid-onset rejection (LEW to PVG) over a short time frame (days 10–15 after transplantation) was used. ²  

Methods

Rats

Inbred female LEW (RT1^l) and PVG (RT1^c) specific pathogen-free rats were purchased from Harlan (Bicester, UK). Procedures were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Corneal Transplantation and Clinical Evaluation

Corneal transplantation (LEW or PVG recipients of PVG allografts) and clinical evaluation were performed as previously described. ¹⁸ Donors and recipients were aged 8 to 16 weeks. Briefly, a 3.5-mm graft was transplanted to a 3.00-mm graft bed and secured with 12 interrupted 11-0 nylon sutures. The sutures were cut as short as possible, but were not removed. The corneas were evaluated and rejection defined as previously described. 

Analysis of Leukocyte Infiltrates in Aqueous Humor by Flow Cytometry

Rats were terminally anesthetized and perfused through the left ventricle with at least 60 mL PBS, to remove blood from corneal vessels. Transplanted eyes were excised and washed in Hanks’ balanced salt solution (HBSS; Invitrogen-Gibco, Paisley, UK) containing 1 mM HEPES buffer and 0.1% NaN₃, corneas were removed, and cells adhering to endothelium gently scraped off into HBSS. The AC was washed out with further HBSS. After one further wash, the Fc receptors were blocked with a mixture of heat-inactivated 10% mouse and 10% rat serum for 30 minutes at 4°C, followed by one or more PE-, FITC-, or PerCP-conjugated antibodies specific for CD4, CD25, CD68 (ED1), CD163 (ED2), CD11b, CD161 (Serotec, Oxford, UK), αβTCR, CD8, MHC class II (BD Biosciences, Oxford, UK) or isotype-matched negative controls. Before CD68 staining, the cells were permeabilized according to the manufacturer’s instructions. The cell counts were acquired on a flow cytometer (FACSCalibur; BD Biosciences, San Jose, CA). The leukocyte subpopulation was defined according to size (forward scatter) and granularity (side scatter), the gate being set initially on live (propidium iodide negative) CD45⁺ cells. For cytokine expression, the cells were incubated after harvesting for 2 to 2.5 hours at 37°C in complete RPMI (Invitrogen-Gibco) containing 3% rat serum and 1 μL/mL brefeldin A (GolgiPlug; BD Biosciences). Detection of IFN-γ expression was achieved by prestimulating the cells for 4 hours at 37°C with phorbol myristate acetate (PMA,10 ng/mL) and ionomycin (400ng/mL; Sigma-Aldrich, Poole, UK), with 1 μL/mL brefeldin A being added for the final 2 hours. The cells were then incubated with cell surface antibodies, fixed, and permeabilized according to the manufacturer’s instructions and incubated for 30 minutes at 4°C with PE-labeled hamster anti-rat/mouse TNF-α, mouse anti-rat IL-10, mouse anti-rat IFN-γ, or the appropriate control antibodies (BD Biosciences), followed by flow cytometric acquisition. Positive control (RiCK-2) cells for cytokine labeling, unlabeled antibody, and cytokines for blocking were obtained from BD Biosciences. Flow cytometric analysis was performed with commercial software (FlowJo; Tree Star Inc., Ashland, OR). Gates were set with reference to fluorescence intensity of cells labeled with isotype-matched negative control antibodies. 

Detection of iNOS, NO, and Nitrotyrosine by Fluorescence Histology

The rats were killed and the eyes excised with attached conjunctivae and snap frozen in liquid nitrogen. Sagittal sections (6–8 μm) were cut across the graft, air dried on to glass slides, and double labeled in the following combinations: iNOS and cell subtype; NO and cell subtype; NT and cell subtype; NO and iNOS; and NO and NT. NO was visualized by incubation with 5 μM 4-amino 5-methylamino-2′,7′-difluorofluorescein diacetate (DAF-FM DA; Invitrogen-Molecular Probes, Paisley, UK) in DMEM. NT and iNOS were visualized with rabbit antibodies to NT (1:500) and mouse iNOS (1:4000), respectively (Upstate Biotechnology, Watford, UK) followed by biotinylated goat anti-rabbit Ig and streptavidin-conjugated rhodamine red-X (Stratech Scientific, Newmarket, UK). Cell subtypes were visualized using mouse anti-rat antibodies to myeloid cells (CD11a, CD11b, CD68, CD163, or MHC class II; Serotec, Oxford, UK) followed, as appropriate, by FITC-conjugated goat anti mouse Ig (Stratech) or by biotinylated goat anti-mouse Ig (Vector Laboratories, Peterborough, UK) and streptavidin-conjugated rhodamine red-X. Fixation before labeling consisted of acetone or a mixture of 95% ethanol/5% acetic acid (NT only). In the case of dual labeling involving DAF-FM DA, the sections were incubated with DAF-FM DA before fixation, which was followed by incubation with the second label. The sections were mounted in antifade medium (Vectashield; Vector Laboratories) containing 4′,6-diamidino-2-phenylindole (DAPI) Negative controls for NO labeling consisted of sections incubated with DMEM and/or preincubation with 1 mM NO scavenger, 2-phenyl-4,4,5,5,-tetramethylimidazoline-1-oxyl 3-oxide (PTIO) followed by 1 mM PTIO with the DAF-FM DA. Controls for iNOS consisted of sections in which nonimmune rabbit Ig (Upstate Biotechnology) was substituted for the primary antibody (negative) and sections of immature thymus (positive). Controls for NT consisted of sections treated with primary antibody preincubated with 10 mM NT (negative) and artificial nitration of tissue (positive), according to the manufacturer’s instructions. Allospecificity controls consisted of isografts (n = 4; day 15 only) and nontransplanted corneas (n = 3). 

Statistical Analysis

Differences in iNOS and NO production over time were determined by analysis of variance (ANOVA; SPSS software; SPSS, Chicago, IL). 

Results

Phenotype of Cells in Aqueous Humor

Histologic studies have indicated that the number of leukocytes in the stroma are increasing by day 10 after transplantation, but the endothelium is not yet infiltrated. However, by day 15 an infiltrate covers the entire endothelium and is resolving by day 18. ² Initial analysis of AH cells in the present study indicated a much lower infiltrate on day 10 (threefold fewer cells) than on day 15, with almost twofold fewer cells on day 18. Therefore, analysis was performed, unless otherwise stated, on day 15. Cells from four to six eyes were pooled to obtain a sufficient number for flow cytometry. The percentage of each of the major leukocyte subsets (except B cells, which we have previously shown to be sparse, both in the cornea and AH ² ) was assessed by single staining and repeated at least once (Table 1) . Double and triple staining was then performed as appropriate (at least twice for each antibody or combination of antibodies) to further characterize cell surface phenotype and cytokine production. 

In the AC on day 15, T cells represented approximately half the total number of cells (Table 1) . They were more numerous relative to other cells in the stroma, ² and included a population of CD161⁺ NK T cells (Fig. 1A) . The remainder comprised TCR⁻ NK cells, granulocytes, and myeloid/antigen presenting cells of multiple overlapping phenotype, expressing CD11b, CD68, CD163, and MHC class II (Table 1) . Granulocytes and CD 163⁺ macrophages were relatively less numerous than we ² and others ¹⁰ have found in rat stroma, whereas NK cells were more numerous, consisting of both TCR⁺CD161^low and TCR⁻CD161^high cells (Fig. 1A) . Approximately half the CD161⁺ cells were CD8α⁺, comprising more than 30% of the CD8α⁺ population (Fig. 1B) . CD25⁺ cells were also heterogeneous (CD4⁺ and CD4⁻, with a proportion of the CD4⁺CD25⁺ expressing high levels of CD25; Fig. 1C ). Approximately one third of CD163⁺ macrophages were CD4⁺, and most expressed MHC class II. High levels of class II expression correlate with higher levels of CD163 (Fig. 1D) . 

AH cells were then examined for expression of the cytokines TNF-α, IFN-γ, and IL-10. It was found that 90% to 95% of TCR⁺ and CD161⁺ cells and approximately 85% of CD11b⁺ cells produced TNF-α, whereas up to 40% of total cells produced IL-10 (Fig. 2) . TNF-α was not detected unless the cells were exposed to Golgi blocking agent (data not shown), excluding the possibility that the labeling was due to TNF’s binding to its receptor expressed on other cells; only 50% of the cells expressed TNF-α on day 12. Labeling for TNF-α and IL-10 could be specifically blocked (Fig. 2) . By contrast, very few cells were labeled with the IFN-γ antibody (0%–1%; Fig. 3A ), equivalent to the irrelevant antibody negative controls (data not shown). However, incubation with PMA and ionomycin induced IFN-γ production in approximately 20% of total cells, comprising approximately 25% of each subtype tested (CD4⁺, CD8⁺, and NK cells; Figs. 3B 3C 3D ). IFN-γ production was also negligible on day 12 (data not shown), when previous studies have shown that leukocytes are beginning to adhere to the corneal endothelium but morphologic damage is still minimal. ²  

Evidence of Cytotoxic NO Production by Fluorescence Histology

Despite indirect evidence of IFN-γ mRNA in AH cells during acute rejection, IFN-γ protein was not produced. As IFN-γ is a major inducer of iNOS transcription and NO synthesis, we wanted to determine whether macrophages in AH express iNOS and generate NO and NT. Given previous documentation of IFN-γ production in the stroma during rejection, ^{2 10} we compared stroma with AC for synthesis of these molecules. As there was consistent acute allograft rejection in this strain combination, the recipients were killed at three time points: days 10 (early rejection), 15 (maximum rejection), and 18 (late rejection). The location and phenotype of iNOS-expressing cells in histologic sections were determined by immunolabeling and correlated with NO production by labeling with DAF-FM DA. Peroxynitrite production was assessed by labeling for NT. 

iNOS immunoreactivity was detected in allografts in a patchy distribution and at variable intensity on days 10, 15, and 18 after transplantation (Figs. 4 5A) , but not in isografts or nonsurgical corneas. iNOS⁺ cells were restricted almost entirely to donor stroma and endothelium and adjacent cells in the AC (i.e., were very rarely found in donor epithelium), recipient cornea or conjunctiva. There was a change in the number of labeled cells over time (P < 0.001), reaching a maximum on day 15 (Fig. 4) , and no difference in iNOS expression between stroma and endothelium (P = 0.18). Thus, there appeared to be no inhibition of iNOS synthesis in the AC compartment. Labeled cells were of the myeloid phenotype (i.e., expressed CD11b, CD11a, and CD68 [Figs. 5A 5B 5C 5D ], but less frequently CD163, especially in posterior stroma and endothelium [Fig. 5E ], or MHC class II). 

NO labeling of corneas was variable but of generally low intensity. It resembled that of iNOS in being found in the allograft stroma in a patchy distribution on days 10, 15, and 18 (Fig. 6)and not in isografts or nonsurgical corneas. However, in certain other respects, NO production did not correlate with iNOS expression. The pattern of labeling was significantly different from that of iNOS over time (Fig. 4vs. Fig. 6 , P = 0.006), the number of NO⁺ cells in the corneal stroma being substantially higher than the number of iNOS⁺ cells on day 10 and NO⁺ cells being rarely seen adhering to the endothelium. The difference in NO production between the stroma and endothelium was significant (P < 0.001). Second, NO labeling in the stroma, unlike iNOS, was the strongest around sutures (Fig. 5F)and often included epithelium (Figs. 5G 5H)and recipient cornea adjacent to the graft–host junction (not shown). Finally, dual labeling revealed that cells exhibiting strong iNOS expression frequently did not show a strong NO signal and vice versa (Fig. 5G) . The lack of NO⁺ cells adhering to endothelium was confirmed by colabeling of two endothelial sheets, removed, and processed as described previously, ² for iNOS and NO (Fig. 5I) . Thus, it appeared that the lack of IFN-γ in AH cells correlated with a lack of NO in AC and endothelium, but not a lack of iNOS. 

Despite the production of NO in an allospecific fashion in the corneal stroma, a deleterious effect of NO in the form of NT was not evident, NT⁺ cells being rare at all time points (usually none, but no more than four per corneal section; Fig. 5H ). Colabeling identified these rare NT⁺ cells as expressing brightly (NO^bright). To achieve images of equivalent brightness NT⁻NO⁺ cells required a 50% to 90% longer exposure time and their NO signal was virtually abrogated by fixation (NO^dim). Furthermore, these NT⁺NO^bright cells were conspicuously round in morphology, whereas the NT⁻NO^dim cells were of variable morphology (Fig. 5H) . NT⁺ cells were also rare on endothelium (1 or 2 cells per sheet, n = 2; data not shown). 

Although there were few NT⁺ cells in donor tissue, they were more numerous adjacent to sutures and in conjunctiva, where they were consistently of greater NO signal (NO^bright). The number of NT⁺NO^bright cells (unlike NT⁻NO^dim cells) was very similar in allograft and isograft recipients, but there were very few such cells in the conjunctivae of nonsurgical eyes. Their distribution in the conjunctivae was relatively sparse at the limbus, increased in density toward the fornix, and was most concentrated in occasional leukocyte aggregates in the palpebral conjunctiva (Figs. 7A 7B 7C) , which we have previously shown to develop after transplantation in this model. ¹⁹ Punctate NT labeling around sutures, in the fornix, and in the palpebral conjunctival aggregates was frequently surrounded by more extensive, diffuse NO labeling (Figs. 7A 7B 7C) . Although many of these NO⁺ cells were CD11b⁺, the surface phenotype differed from that of iNOS⁺ cells in that they were frequently MHC class II⁺ or CD163⁺, but rarely CD11c⁺ (data not shown). Labeling with the negative control antibody for iNOS was restricted to glandular material in the palpebral conjunctiva. NO labeling in cornea and conjunctiva was completely or almost completely abolished by incubating sections with DAF-FM DA and the NO scavenger PTIO (Fig. 7Dversus 7E). NT labeling was completely or almost completely abolished by preincubating the anti-NT antibody for 1 hours with 10 mM NT before application to the sections (Fig. 7Dversus 7F). 

Discussion

Both flow cytometric and histologic data showed that myeloid cells infiltrating the AH at maximum corneal graft rejection were of varied phenotype. The infiltrate was donor specific, as isograft recipients had minimal, if any, infiltrate in the AC at this stage after transplantation. The higher proportion of CD68 (ED1)⁺ monocytes compared with CD163 (ED2)⁺ tissue macrophages (21% vs. 11% by flow cytometry) and variable intensity of CD163 expression in AH cells are consistent with the de novo expression and upregulation of CD163 on CD68⁺ monocytes after entry into the AC. This upregulation was accompanied by an increase in macrophage MHC class II expression and abundant macrophages adhering to endothelium. However, there was no evidence of IFN-γ or NO synthesis in AH, despite iNOS protein expression, or of NO-mediated damage to corneal cells. 

The proportions of T-cell subsets and NK cells found in AH were very similar to those reported by others at comparable stages of rejection, ^{10 12} particularly the heterogeneity and relatively high proportion of NK cells (25% of total leukocytes) reported by Claerhout et al. ¹² The ratio of αβTCR⁺CD161^low to αβTCR⁻CD161^high cells in the present study was virtually identical with their ratio of CD3⁺CD161⁺ cells to CD3⁻CD161⁺ cells, as was the proportion of CD8⁺ NK cells. 

To elucidate macrophage function, we examined cytokine production in AH cells by using available anti-rat antibodies to TNF-α, IFN-γ, and IL-10. An advantageous feature of the AH compared with solid tissues is that mechanical manipulation or enzymatic digestion, which might affect metabolism of cells, are not required. Thus, the cell content of proteins such as cytokines is more likely to reflect the status quo in vivo. In our study, at maximum rejection, most AH cells synthesized TNF-α, and a substantial number of cells produced IL-10, but they did not synthesize IFN-γ. The percentages of cells producing TNF-α (85%–95%, depending on cell type) and IL-10 (20%–40%) indicate that most or all IL-10-producing cells must coexpress TNF-α. TNF-α protein has been documented within the cornea of both rat and mouse during rejection. ^{11 20} The rat cornea has not been tested for IL-10, but, in the mouse, ELISA failed to show IL-10 protein in the cornea ²¹ and immunohistology revealed relatively few IL-10⁺ cells ¹¹ (although mRNA was recorded). This suggests a possible difference in IL-10 production between the stroma and AH. 

Our failure to find IFN-γ-producing cells in the AH was unexpected, as by immunohistochemistry, these cells have been detected in the stroma of the rat ^{2 10} and the mouse ¹¹ as well as in the AH of the mouse. ¹¹ It is also abundant in other pathologic models. ^{22 23} The ability of AH cells in our study to produce IFN-γ when stimulated for 4 hours in vitro indicated a block in cytokine production in vivo at the translational level and therefore does not contradict evidence in the mouse of IFN-γ mRNA in AH cells. ¹¹ This lack of IFN-γ and high production of IL-10 in AH in vivo supports the concept of altered immune responses in the AC compared with stroma and the possibility of different effector mechanisms operating in these two compartments. 

IFN-γ, ^{24 25} with TNF-α, ²⁶ is a major inducer of iNOS transcription and NO synthesis by macrophages, and IFN-γ⁺ cells are concentrated in the upper stroma during corneal graft rejection. ^{2 10} Having failed to find IFN-γ⁺ cells in AH, we therefore expected to find iNOS in stroma, but not in AH cells. However, there was no such correlation, as iNOS was expressed by cells in both the stromal and AC compartments. By contrast, IFN-γ⁺ cells correlated with NO⁺ cells in cornea, in that NO⁺ cells were most frequently found and were brightest in the upper stroma and were not seen in the AC compartment. The lack of NO in the AC and the lack of peroxynitrite production in the donor cornea did not support the involvement of NO in endothelial or stromal cell death. The NT signal, indicative of peroxynitrite synthesis, was almost entirely restricted to the vicinity of morphologically rounded cells in sutures and conjunctivae. The similar number of such cells in allograft and isograft recipients and their absence in normal corneas and conjunctivae suggests that peroxynitrite was produced by cells responding to a nonspecific inflammatory stimulus, most likely the presence of sutures, rather than to alloantigens. The consistent association of NT with rounded, higher signaling NO^bright cells, whereas NO^dim cells in stroma were negative for NT and morphologically normal, suggests that NO^bright cells were undergoing apoptosis, presumably by an NO-dependent mechanism. Protective roles have been attributed to low levels of NO synthesis, such as the control of leukocyte recruitment ¹⁷ and proliferation ^{16 27} and may explain the lack of peroxynitrite production by NO^dim cells in the stroma. Alternatively, peroxynitrite synthesis may have been prevented by efficient scavenging of the oxygen radicals required for its generation. This is mediated by the enzyme superoxide dismutase, abundant in the cornea and protective against light-induced free radical damage. ²⁸  

Although there was a close correlation between greater NO signaling cells and NT production, a close temporal or spatial correlation between either NO^bright or NO^dim cells and iNOS clearly did not exist. In some circumstances, particularly in the AC, there was an excess of iNOS over NO, in others vice versa (shown graphically in Fig. 4vs. 6 and in the images in Figs. 5G 5I ). An excess of iNOS over NO may in part be accounted for by intermittent pulsatile release ²⁹ or rapid neutralization of NO. However, posttranslational control of iNOS activity would be a more likely explanation of the lack of NO in the AC environment and would be consistent with the regulation of iNOS activity by factors within the AH. These could be mediated at several levels. First, upregulation of the enzyme arginase (e.g., by abundant IL-10) would deplete the arginine substrate of iNOS. Second, dimerization of iNOS is essential for activity; therefore, depletion or sequestration of any of a range of metabolites essential for this process could prevent NO production (reviewed by Alderton et al. ³⁰ ). Third, proteins capable of directly inhibiting dimerization have been identified in other systems. ³¹  

Diffusion of NO from its site of synthesis may account for the presence of NO in the absence of iNOS. NO can react reversibly with SH groups of proteins and low-molecular-weight thiols (e.g., glutathione) to produce nitrosothiols, which may act as sinks of NO for more distant delivery. ³² Alternatively, some of the NO generated in this model may have been derived from other NOS enzymes (e.g., endothelial (e)NOS or mitochondrial (m)NOS). Production of eNOS has been documented in macrophages ³³ and corneal endothelial cells. ³⁴ Activation of macrophages by proteins released by damaged cells and/or bacteria may account for the strong NO signal around the sutures. 

Our findings highlight the complexities of NO production in pathologic situations such as graft rejection. They show that the presence of iNOS protein per se is not proof of NO synthesis, much less of whether NO exerts a cytotoxic function. Although we recognize that detection of both NO and NT is limited by the sensitivity of detection methods used, these data suggest that NO is not a mediator of endothelial cell death in corneal graft rejection. In vitro studies have shown that the neuropeptides α-MSH ⁴ and vasointestinal peptide (VIP) ⁵ found in AH, can downregulate IFN-γ production by T cells. Furthermore, it has been shown that the neuropeptide calcitonin gene-related peptide (CGRP) can inhibit NO production by macrophages without inhibiting iNOS synthesis. ⁶ Thus, it is plausible that neuropeptide(s) play a pivotal role in the downregulation of both IFN-γ and NO production in AH in vivo. TNF-α induces NO generation by bone-marrow–derived macrophages at much lower levels without synergy from IFN-γ. ²⁶ We cannot, of course, exclude the possibility that some IFN-γ reaches AH by diffusion from stroma, although if TNF-α predominates, IFN-γ responses will be blunted. ²⁶  

Our findings do not exclude a cytotoxic role for NO in stromal rejection by a mechanism other than peroxynitrite production or killing by macrophages of stromal or endothelial cells via a mechanism independent of NO. However, they are more consistent with those in previous reports in rats ¹⁵ and mice ¹ suggesting a pathologic role of macrophages in the affector rather than effector stage of the immune response. 

We ² and others ¹ have suggested that different layers of the cornea undergo rejection by different mechanisms, and our current findings support this conclusion. As well as differences between the stromal and AC compartments with respect to IFN-γ and NO production, there was a substantially greater proportion of NK cells (and fewer neutrophils) than in the stroma, ² and a role for NK cells in endothelial rejection ¹² requires further investigation. Certainly, the complexities revealed by accumulating studies of the effector response reinforce the notion that CD4⁺ T cells remain the most appropriate targets for therapeutic intervention. 

 Supported by the Royal College of Surgeons and National Eye Research Centre, UK.

 Submitted for publication March 1, 2008; revised April 22 and May 27, 2008; accepted August 25, 2008.

 Disclosure: S.M. Nicholls, None; A.D. Dick, 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.

 

Cells Infiltrating Aqueous Humor at Maximum Rejection

Cell Surface Antigen (Clone)	Mean % Cells (Range)*
αβTCR (R73)	52 (42–62)
CD4 (W3/25)	29 (27–30)
CD8α (OX8)	45 (40–50)
CD25 (OX39)	13 (8–18)
CD161 (10/78; NK)	27 (25–28)
MHC II (OX6)	30
CD163 (ED2; mannose receptor on tissue macrophages)	11 (10–14)
CD 68 (ED1; macrosialin on monocytes/macrophages)	21
CD11b (OX42)	13 (9–17)
Granulocytes (HIS 48)	14 (13–15)

*  Cells were pooled from 4 to 6 eyes on each occasion, yielding a mean of 7 × 10⁴–2.2 × 10⁵ cells per eye.

Figure 1.

 

Analysis of leukocytes in AH at maximum rejection. (A) NK cells were CD161^highαβTCR⁻ or CD161^lowαβTCR⁺ in equal ratios. (B) NK cells were CD161⁺CD8⁺ and CD161⁺CD8⁻ in approximately equal ratios. (C) CD25 was expressed at higher levels on CD4⁺ than on CD4⁻ cells. (D) High levels of CD163 expression correlated with high levels of MHC class II expression. Nonspecific fluorescence obtained with isotype control antibodies varied from 0.5% to 2% of gated events.

Figure 2.

 

Cells expressed TNF-α and IL-10 in AH at maximum rejection. Top: TNF-α was expressed by most of the infiltrating myeloid, NK, and T cells; bottom: 20% to 40% of cells expressed IL-10. Labeling could be specifically blocked (right). TNF-α was blocked by preincubation of cells with excess unlabeled anti-TNF antibody and IL-10 by preincubation of labeled anti-IL-10 antibody with recombinant IL-10. TNF-α labeling of positive control–activated rat lymphoid (RiCK) cells was reduced from 85% to 0.7% of cells by preincubation of labeled anti-TNF antibody with recombinant TNF-α. Approximately 20% of RiCK cells expressed IL-10 (data not shown). Nonspecific fluorescence obtained with isotype control antibodies was not more than 2% of gated events.

Figure 3.

 

(A) Cells in AH do not synthesize IFN-γ at maximum rejection. After stimulation with PMA and ionomycin (PMA/I), approximately 25% of CD4⁺ cells (B), CD8⁺ cells (C), and NK cells (D), synthesized IFN-γ. Between 40% and 45% of activated rat lymphoid (RiCK) cells produced IFN-γ (data not shown).

Figure 4.

 

Mean iNOS labeling scores in donor corneal stroma and in endothelial cells adhering to corneal endothelium. The cells were counted to a maximum of 30 (score 1); higher scores denote the maximum percentage of tissue area of donor labeled. This method was used to avoid difficulties in counting individual cells when accumulated in large numbers: 1, 0 cells to ≤30 cells; 2, >30 cells to ≤25%; 3, >25% to ≤50%; 4, >50% to ≤100%. At least 15 sections were examined across each cornea, the scores of section(s) with maximum area labeled were determined and used to calculate the mean labeling score of replicate corneas (day 10, n = 7; day 15, n = 7; day 18, n = 4).

Figure 5.

 

iNOS, NO, and NT production in donor corneas. iNOS and NT appear red, cell surface phenotypic markers and NO appear green, colabeled cells appear orange/yellow, cell nuclei (DAPI) appear blue. (A) iNOS⁺ cells are CD11b⁺; graft margin is at right. (B) Clump of iNOS⁺CD11b⁺ cells adhering to corneal endothelium. (C) CD11c⁺ cells express iNOS. (D) Subepithelial CD68⁺ cells express iNOS. (E) CD163⁺ cells express little or no iNOS and vice versa. (F) Strong NO production around a suture (arrow), diminishing into the cornea at right; note the lack of NO at the endothelium and in the AC. (G) Central donor showing lack of correlation between strong iNOS expression and NO; note NO in the basal epithelium at right. (H) Extensive NO production in the anterior stroma and epithelium, but only one cell producing NT (arrow). (I) Corneal endothelial sheet showing extensive iNOS labeling in cells with dendritic morphology, but only one cell producing NO (arrow). (A–G, I) Peak rejection on day 15, (H) day 10. ep epithelium, en endothelium. Bar: (A, C, F, G) 100 μm; (B, D, E, H, I) 20 μm.

Figure 6.

 

Mean NO labeling scores in donor corneal stroma and cells adhering to corneal endothelium. Note that NO production peaked earlier than iNOS expression (Fig. 4)and before maximum rejection. Scoring was performed as in Figure 4 . Day 10, n = 4; day 15, n = 4; day 18, n = 2.

Figure 7.

 

Production of iNOS and NT in the conjunctiva. NT appears red, NO appears green, colabeled cells appear orange/yellow, cell nuclei (DAPI) appear blue. (A–C) NO⁺ cells in a leukocyte aggregate are also NT⁺; note additional diffuse NO labeling and subepithelial location of aggregate in merged image (C). NO and NT labeling of conjunctival cells was specific: (D) scattered colabeled cells; (E) NO labeling was abolished in a nearby section if the NO scavenger PTIO was applied to tissue with DAF-FM-DA. (F) NT labeling was abolished in a nearby section by preincubation of the anti-NT antibody with NT. ep epithelium. Bar, 100 μm.