It has been recognized for over a century that the anterior chamber of the eye is endowed with a remarkable immune privilege. One contributing component is the Ag-specific down-regulation of systemic delayed-type hypersensitivity (DTH) that is induced when Ags are introduced into the anterior chamber. This phenomenon, termed anterior chamber-associated immune deviation (ACAID), culminates in the generation of regulatory cells that inhibit the induction (afferent suppression) and expression (efferent suppression) of DTH. Since γδ T cells play a major role in other forms of immune regulation, we suspected they might contribute to the induction and expression of ACAID. Mice treated with anti-γδ Ab failed to develop ACAID following anterior chamber injection of either soluble Ag (OVA) or alloantigens (spleen cells). Additional experiments with knockout mice confirmed that mice lacking functional γδ T cells also fail to develop ACAID. Using a local adoptive transfer of DTH assay, we found that γδ T cells were required for the generation of regulatory T cells, but did not function as the efferent regulatory cells of ACAID. The importance of γδ T cells in corneal allograft survival was confirmed by blocking γδ T cells with GL3 Ab before corneal transplantation. While in vivo treatment with normal hamster serum had no effect on corneal graft survival, infusion of anti-γδ Ab resulted in a profound increase in corneal allograft rejection. Thus, γδ T cells are needed for sustaining at least one aspect of ocular immune privilege and for promoting corneal allograft survival.
The immune privilege of the anterior chamber of the eye has been recognized for over a century and is believed to be an adaptation for preventing immune-mediated injury to ocular tissues that possess little or no regenerative properties (e.g., corneal endothelium and retina). Ocular immune privilege was originally attributed to the unique absence of patent lymph vessels draining the interior of the eye, thereby resulting in a sequestration of Ag and a blockade of the afferent arm of the immune response (1). However, in recent years it has become apparent that immune privilege is a composite of multiple overlapping immunosuppressive and antiinflammatory properties of the anterior chamber. These include the extensive expression of Fas ligand (2, 3), immunosuppressive cytokines and neuropeptides in the aqueous humor (4), the production of NK-inhibitory cytokines such as TGF-β and MIF throughout the eye and in the aqueous humor (5, 6, 7, 8), limited expression of MHC class I and II molecules (9, 10, 11, 12), and the expression of nonclassical class I molecules on ocular cells possessing poor regenerative capacities (13). Dynamic regulatory mechanisms also contribute to the immune privilege of the eye. Ags introduced into the anterior chamber elicit a deviant systemic immune response, termed anterior chamber-associated immune deviation (ACAID),3 in which potentially injurious immune effector mechanisms, such as delayed-type hypersensitivity (DTH) and complement-fixing Abs, are suppressed.
The exact mechanisms involved in the induction of ACAID are poorly understood. However, Streilein and coworkers (14, 15, 16, 17, 18) have offered compelling evidence that APCs in the eye capture intraocular Ags and migrate to the spleen, where they induce at least two functionally distinct regulatory T cell populations in the ACAID spleen. One population is CD4+ and blocks the induction or afferent component of the immune response. A second population is CD8+ and inhibits the expression of DTH by previously sensitized T cells. In addition to ocular APCs, the generation of both populations of ACAID regulatory cells requires the participation of splenic B cells and NKT cells (19, 20, 21). The cellular interactions between B cells, NKT cells, and splenic regulatory T cells have not been elucidated, and it is possible that additional accessory cells are needed for the generation of ACAID.
γδ T cells represent a small population of lymphocytes that are typically CD4−/CD8−. However, CD8+ γδ T cells have been reported (22). A variety of functions have been ascribed to γδ T cells. γδ T cells have been shown to play a role in other forms of tolerance, including oral tolerance (23, 24, 25), testicular tolerance (26), and tumor-associated tolerance (27, 28, 29). Considering their role in various forms of immune tolerance, we entertained the hypothesis that γδ T cells were necessary for the induction of ACAID.
In this study, we report that γδ T cells are required for the induction of ACAID. Because ACAID can be induced by a variety of Ags (30, 31, 32, 33), we examined the role of γδ T cells in generating ACAID to a soluble Ag, as well as to a particulate Ag. We also demonstrate that γδ T cells are needed for the generation of efferent suppressors, but are not actually efferent suppressors themselves. Finally, because ACAID has been shown to be necessary for orthotopic corneal graft survival in the mouse (34, 35, 36), we also examined whether γδ T cells were needed for corneal graft survival. Our experiments utilized anti-TCR δ-chain Ab, GL3, first described by Goodman and Lefrancois (37). This Ab inhibits the function of γδ T cells by blocking the TCR δ-chain (25).
Materials and Methods
Animals and Abs
Six- to 10-wk-old BALB/c mice were obtained from the mouse colony at the University of Texas Southwestern Medical Center at Dallas. C57BL/6, NZB, BALB/c × C57BL/6 F1 (CB6F1) BALB.B, and TCR δ-chain knockout mice (TCRδ KO) (C57BL/6J-Tcrdtm1 Mom) were obtained from The Jackson Laboratory (Bar Harbor, ME). GL3 Ab was produced from hybridoma cells and purified by protein A columns. The hybridoma was graciously provided by Dr. Leo Lefrancois (University of Connecticut, Farmington, CT). Animals treated with GL3 Ab or normal hamster serum (N.S.) for ACAID experiments were dosed i.p. with 200 μg on days −3, +4, and +11. Animals treated with GL3 Ab or normal hamster serum for corneal transplantation were dosed i.p. with 200 μg two times each week starting at day −7. All animal studies were approved by the Institutional Review Board of the University of Texas Southwestern Medical Center at Dallas. UC-7 is an anti-γδ Ab (PharMingen, San Diego, CA). FITC GL3 Ab (PharMingen) was used in FACS staining. Armenian hamster serum was purchased from Cytogen Research and Development (West Roxbury, MA).
Induction of ACAID
ACAID was induced as described previously using microinjection of Ag into the anterior chamber of the eye (38). Briefly, mice were anesthetized with 0.133 mg/Kg ketamine hydrochloride (Fort Dodge Laboratories, Fort Dodge, IA) and 0.006 mg/Kg xylazine (Bayer, Shawnee Mission, KS) given i.p. A glass micropipette (approximately 80 μm diameter) was fitted onto a sterile infant feeding tube (no. 5 French; Professional Medical Products, Greenwood, SC) and mounted onto a 0.1-ml Hamilton (Hamilton, Whittier, CA) syringe. A Hamilton automatic dispensing apparatus was used to inject 5 μl of the 20 mg/ml OVA (Sigma, St. Louis, MO) in PBS (100 μg OVA) or 5 μl of the 2 × 108 cells/ml (NZB or BALB.B) nonadherent spleen cells in HBSS (1 × 106) into the anterior chamber.
Nonadherent spleen cells
Splenocytes from naive mice were harvested and the erythrocytes were lysed. Each spleen was resuspended in 5 ml complete RPMI and placed in a Primaria petri dish (Becton Dickinson Labware, Franklin Lakes, NJ). The splenocytes were incubated for 2 h at 37°C in 5% CO2. The nonadherent cells were drawn off carefully. Other nonadherent cells were collected by gentle swirling after 5 ml cRPMI was added to the side of the dish.
Mice were immunized by s.c. injection of either OVA in PBS (125–250 μg) or 1 × 106 (NZB or BALB.B) spleen cells in HBSS emulsified 1:1 in CFA (Sigma). Each animal received a total volume of 200 μl.
Both ear pinnae of experimental and control animals were measured with a Mitutoyo engineer’s micrometer immediately before challenge. For OVA experiments, OVA (400 μg) in 20 μl PBS was injected intradermally into the left ear pinnae. For alloantigen experiments, γ-irradiated (3000 rad) NZB or BALB.B spleen cells (4 × 106) were injected intradermally into the left ear pinnae. The right ear pinnae received 20 μl sterile PBS alone (negative control). Both ears were measured 24 h later, and the difference in size was used as a measure of DTH. Results were expressed as: specific ear swelling = (24-h measurement − 0-h measurement) for experimental ear − (24-h measurement − 0-h measurement) for negative control ear.
Depletion of γδ T cells
C57BL/6 splenocytes were collected and erythrocytes were lysed. Splenocytes were incubated with UC7 Ab (5 μg/107 cells) for 30 min on ice. The cells were washed three times thoroughly with HBSS. The cells were then incubated with Low-Tox rabbit complement (Accurate Chemical, Westbury, NY) (1:10) for 30 min at 37°C and 5% CO2. Depletion of γδ T cells was verified by FACS analysis using FITC-conjugated GL3 Ab. Splenic γδ T cell population fell from 3.2% to 0.9% following Ab treatment.
Assay for assessing DTH regulatory cells
Regulatory cells were generated by injecting OVA into the anterior chamber on day 0 (see above). These animals were also s.c. immunized on day 7 with OVA in CFA (see above). Spleen cells containing the regulatory cells were collected on day 14. Erythrocytes were lysed and the cells were resuspended at 5 × 107 cells/ml in 10 mg/ml OVA in PBS. Preimmune cells were generated by s.c. immunizing naive C57BL/6 mice on day 0 with OVA (250 μg) in PBS emulsified in CFA. Splenocytes were collected on day 14, erythrocytes lysed, and the splenocytes were passed over nylon wool to enrich for T cells. The preimmune T cells collected from the nylon wool column were resuspended at 5 × 107 cells/ml in 10 mg/ml OVA. The preimmune and regulatory cell populations were mixed 1:1 and injected into the left ear pinna (20 μl = 1 × 106) of a naive C57BL/6 mouse. The opposite ear was injected with 10 mg/ml OVA as a negative control. Ear swelling was measured 24 h later.
Orthotopic corneal transplantation
Penetrating orthotopic corneal transplants were performed as described previously (39, 40). NZB mouse corneas were transplanted (2.5 mm diameter) orthotopically onto the graft beds (2 mm in diameter) of naive or GL3-treated CB6F1 mice. Proparacaine HCl ophthalmic solution (USP (0.5%); Alcon Laboratories, Ft. Worth, TX) was used as a topical anesthetic. Vannas scissors were used to excise the corneas. Donor corneas were sewn onto the graft bed using 11-0 nylon sutures (Ethicon, Somerville, NJ). Sutures were removed on day 7 posttransplantation. Topical antibiotic (Bacitracin zinc and polymyxin B sulfate; Akorn, Decatur, IL) was applied immediately after surgery, as well as immediately after suture removal. Grafts were read two times each week using a slit-lamp (Carl Zeiss, Jena, Germany). Corneas were scored using a scale of 0–4 for opacity, edema, and vascularity (41). Two consecutive opacity scores of three or higher indicated graft rejection. No immunosuppressive drugs were applied topically or systemically.
The student’s t test was used to evaluate the significance of these experiments. χ2 analysis was used to test the graft survival data. p values less than 0.05 were considered significant.
γδ T cells are needed for the generation of ACAID to soluble Ag
ACAID can be generated using a variety of Ags, including the soluble Ag OVA. To induce ACAID, mice were injected intracamerally (i.c.) with OVA on day 0. On day 7, the i.c. primed animals were injected s.c. with OVA in PBS emulsified in CFA, and 1 wk later animals were challenged by intradermal injection of OVA. Positive controls were s.c. immunized with OVA in CFA 7 days before ear challenge. Negative controls were simply challenged with OVA in the ear. Ear swelling was used to assess the presence of DTH. Animals injected with OVA i.c. had decreased ear swelling, indicating the presence of ACAID. To test whether γδ T cells affected the ability of OVA to generate ACAID, we first treated BALB/c mice with 200 μg GL3 on days −3, +4, and +11 to block the TCR δ-chain of γδ T cells (25). These animals were then tested for their ability to develop ACAID to OVA. Mice treated with GL3 Ab were unable to induce ACAID, as seen by their inability to suppress DTH to OVA (Fig. 1). In contrast, mice treated with an isotype control serum developed ACAID.
γδ T cells are needed for the generation of ACAID to alloantigens
OVA is an example of a soluble Ag that can induce ACAID, but alloantigens can also be used to induce ACAID. We wanted to test whether GL3 treatment blocked ACAID to alloantigens. As before, animals were treated on days −3, +4, and +11 with 200 μg GL3 Ab. In these experiments, NZB nonadherent spleen cells were used as alloantigens for the induction of ACAID. Allogeneic NZB splenocytes were injected i.c. on day 0 into normal CB6F1 mice, as well as GL3-treated CB6F1 mice. This donor-host combination shares the same MHC haplotype, but is mismatched at multiple minor histocompatibility loci. On day 7, these animals, as well as the positive controls, were s.c. immunized with NZB spleen cells in HBSS emulsified in CFA. On day 14, the mice were challenged with intradermal injections of irradiated (3000 rad) NZB spleen cells. The absence of significant ear swelling indicated the presence of ACAID in untreated mice. However, animals treated with GL3 Ab showed significant ear swelling, and thus, did not develop ACAID (Fig. 2).
TCRδ KO animals cannot induce ACAID to soluble Ag
To ensure that the GL3 Ab was indeed blocking the γδ T cells, rather than stimulating them, we used TCRδ KO mice to test for their ability to develop ACAID to OVA. Because TCRδ KO mice are on a C57BL/6 background, the experiments were performed in C57BL/6 animals. As before, naive C57BL/6 and TCRδ KO mice were injected i.c. with OVA on day 0. On day 7, the i.c. primed animals and positive control animals were injected s.c. with OVA in PBS emulsified in CFA. One week later, animals were challenged by intradermal injection of OVA in PBS. Ear swelling was measured to detect the presence of DTH. The absence of ear swelling indicated the presence of ACAID in wild-type C57BL/6 mice primed in the anterior chamber with OVA (Fig. 3). In contrast, TCRδ KO mice were unable to suppress DTH to OVA, indicating the absence of ACAID. Therefore, γδ T cells are needed for ACAID. This supports the observation that GL3 Ab blocked γδ T cells, in vivo, and therefore inhibited the generation of ACAID.
DTH response is not exaggerated in γδ KO mice
In the above experiments, the animals with impaired γδ T cell function are compared with control mice that have intact γδ T cell function. We performed additional experiments to address the possibility that mice with impaired γδ T cell function may display an exaggerated DTH relative to the wild-type animals. In these experiments, we induced ACAID by i.c. injecting γδ KO animals on day 0. On day 7, the i.c. injected mice, as well as the γδ knockout positive controls, were s.c. immunized with Ag. Finally, on day 14, all groups, including the γδ KO negative controls, were challenged with Ag in the ear pinnae. Animals tested with soluble OVA Ag (Fig. 4,a), as well as those tested with BALB.B alloantigen (Fig. 4 b), confirmed earlier results, and demonstrated that the positive γδ KO controls did not display an exaggerated DTH response.
γδ T cells are not efferent suppressor cells
Earlier studies have shown that there are two populations of T cell suppressors in the ACAID spleen, CD4+ afferent suppressor cells and CD8+ efferent suppressor cells (14). Moreover, we have independently confirmed that the efferent suppressors are a TCR αβ+CD4−CD8+ T cell using the same LAT assay described below (data not shown). Because some γδ T cells are CD8+ (22), it is possible that the efferent suppressors are CD8+ γδ T cells. To test this hypothesis, ACAID spleens were tested for their ability to suppress the expression of a preexisting immune response in the absence of γδ T cells. The assay for efferent suppression was a local adoptive transfer assay in which ACAID efferent regulatory cells were generated using OVA as the Ag. The spleen cells, containing the putative efferent suppressors, were collected on day 14. These efferent suppressors were either used directly, or were first depleted of γδ T cells using UC7 Ab plus complement. Preimmune splenocytes were generated independently by s.c. immunizing naive C57BL/6 mice with OVA. On day 14, immune spleen cells were isolated, enriched for T cells, and mixed with Ag and the ACAID efferent regulatory cells. This inoculum was injected directly into the ear pinnae of naive C57BL/6 mice. If γδ T cells were acting as efferent suppressor cells, removing them from the suppressor cell population would inhibit the down-regulation of DTH. However, as seen in Fig. 5, eliminating γδ T cells from the suppressor cell population did not have an effect on the suppression of DTH, and therefore did not affect the expression of ACAID.
γδ T cells are needed for the generation of ACAID efferent suppressors
Although γδ T cells did not function directly as efferent suppressors in ACAID, they might play a role in the generation of ACAID efferent suppressor cells. To test this hypothesis, we attempted to generate efferent suppressor cells in γδ KO mice. As previously shown, these animals do not have γδ T cells, and do not develop ACAID. In these experiments, naive C57BL/6 mice, as well as TCRδ KO mice, were injected i.c. with OVA on day 0. These animals were s.c. immunized on day 7 with OVA in CFA, and potential ACAID suppressor cells were isolated on day 14. Preimmune splenocytes were generated independently by s.c. immunizing naive C57BL/6 mice on day 0 with OVA in CFA. On day 14, preimmune splenocytes were isolated and enriched for T cells. The ACAID regulatory cells were mixed with Ag and the preimmune T cells then injected intradermally into the ear pinnae of naive C57BL/6 mice. The results in Fig. 6 indicate that the absence of γδ T cells prevented the generation of efferent suppressors in response to i.c. OVA priming.
γδ T cells are needed for corneal graft survival
A growing body of evidence suggests that the high success rate of orthotopic corneal allografts is due to the corneal graft’s capacity to induce ACAID to the donor’s minor histocompatibility Ags (35, 36). Since γδ T cells are needed for the induction of ACAID, we suspected that they were also necessary for orthotopic corneal allograft survival. Accordingly, NZB corneas were grafted orthotopically to CB6F1 hosts that were treated with either GL3 Ab or normal hamster serum. Untreated CB6F1 mice served as the normal control and received orthotopic NZB corneal allografts. Other hosts were treated with either GL3 Ab or normal hamster serum 7 days before corneal transplantation and twice per week thereafter until day 50. The results in Fig. 7 indicate that treatment with anti-γδ Ab had a profound effect on corneal allograft survival. Corneal allografts underwent rejection in only 20% of the untreated mice and 0% of the mice treated with normal hamster serum. By contrast, 75% of the corneal grafts underwent rejection in the anti-γδ-treated mice. Thus, γδ T cells appear to play a crucial role in the survival of corneal allografts in this donor-host combination.
ACAID is an immunomodulatory phenomenon that is initiated in the eye, but culminates in systemic immune deviation. Streilein and coworkers have offered a compelling hypothesis proposing that Ag is captured by F4/80+ macrophages in the anterior chamber and transported via the venous circulation to the spleen (15, 16, 17, 18). It appears that the central processing unit for the systemic response to ocular Ags is the spleen, where two populations of suppressor T cells are generated. Studies have described their suppressor function as afferent suppression by CD4+ T cells, and efferent suppression by CD8+ T cells, but the actual events that lead to the generation of CD4+ afferent suppressor cells and CD8+ efferent suppressor cells are not understood. In addition to the suppressor cells, other cell populations of the spleen have been shown to be necessary for the generation of ACAID. Niederkorn et al. (19) first demonstrated the requirement of B cells in the generation of ACAID. D’Orazio et al. (21) confirmed these results using B cell KO mice whose ACAID phenotype could be restored with reconstitution with naive B cells. They have hypothesized that B cells assist in the generation of ACAID by presenting regurgitated Ag to efferent suppressor cells. Sonoda et al. (20) have recently described another ACAID accessory cell. Their work supports a role of NK1.1+ T cells in ACAID. Their studies show that NK1.1+ T cells are specifically needed for the generation of efferent suppressor cells in ACAID. They have hypothesized that the NK1.1 T cells may also be interacting with marginal zone B cells that express high levels of the CD1d molecule (42) that is essential for the NKT cell’s role in ACAID. Activation of these NKT cells could result in the production of TGF-β (43) that has been shown to be important in ACAID (44, 45). The role of B cells and NKT cells in ACAID is not fully understood. However, the studies performed have shown that these cell types are needed for the generation of the efferent suppressor cells that are ultimately responsible for the decreased DTH phenotype.
Our results suggest that yet another accessory cell, the γδ T cell, is required for the induction of ACAID. We demonstrate that γδ T cells are needed for the induction of ACAID to soluble Ag, as well as particulate Ag. Furthermore, the γδ T cells are not the CD8+ efferent suppressor cells themselves, but are needed for their generation.
γδ T cells play an important role in various forms of immune tolerance, and are present at the maternal-fetal interface, an immunologically privileged site (46). The induction of oral tolerance to OVA requires an intact γδ T cell population. Also, γδ T cells are needed in orally induced tolerance to experimental autoimmune uveitis (23). Additionally, Szczepanik and coworkers (47) demonstrated that in a high dose Ag-induced tolerance model, γδ T cells down-regulated IFN-γ production by immune effector T cells, and inhibited contact hypersensitivity responses in immune animals. Finally in mouse tumor models, γδ T cells act to inhibit the generation of cytotoxic NK cells and CTLs, presumably through their secretion of the immunosuppressive cytokines IL-10 and TGF-β (29).
We are attracted to the hypothesis that γδ T cells act as ancillary cytokine-producing cells to create an environment that is conducive to the induction of ACAID. It is noteworthy that γδ T cells produce biologically significant quantities of two cytokines, IL-10 and TGF-β, which are crucial for the induction and expression of ACAID (29). Moreover, γδ T cells can inhibit the production of IFN-γ, a cytokine that is closely associated with the abrogation of ACAID (45, 47, 48, 49). Potentially, the γδ T cells are contributing to the generation of ACAID efferent suppressors by generating cytokines critical for the ACAID environment of the spleen.
Our results suggest that the role of γδ T cells is to inhibit the generation of efferent suppressor cells. The efferent suppressors are the cells that define the decreased DTH phenotype of ACAID, and there are many cells that orchestrate their production. The γδ T cells may not directly influence the CD8+ efferent suppressors, but rather affect some of the other spleen cells involved in generating the ACAID phenotype. For example, γδ T cells can inhibit the proliferation of CD4+ αβ T cells (50), thus eliminating potential DTH-causing lymphocytes, and allowing for the development of efferent suppressors. This is further supported by evidence that γδ T cells can decrease the ability of αβ T cells to induce contact sensitivity (47), a DTH response.
We and others have demonstrated that ACAID has a profound effect in promoting corneal allograft survival (34, 35, 36). Therefore, if γδ T cells contribute to ACAID and thus, corneal allograft survival, depleting γδ T cells should promote corneal graft rejection. This in fact was the case, as systemic treatment with Ab to γδ T cells had a deleterious effect on corneal allograft survival. The normal rejection rate for NZB corneal allografts on CB6F1 hosts is only 20%. However, rejection rose to 86% in hosts treated with anti-γδ T cell Ab. These results further emphasize the importance of ACAID in promoting corneal allograft survival.
In summary, we have shown that γδ T cells are needed for the generation of ACAID to soluble alloantigen. TCRδ KO animals cannot generate ACAID, and γδ T cells are needed for the generation of efferent suppressors, but are not efferent suppressors themselves. Finally, because ACAID is needed for corneal graft survival, abrogating ACAID by blocking γδ T cells could be the cause of decreased corneal graft survival. These results reveal yet another role for the γδ T cells in tolerance, and another cell population needed for the generation of ACAID.
We thank Mr. Michael Hurt, Mr. Robert Ritter, and Ms. Elizabeth Mayhew for their assistance in generating the GL3 Ab. Also, we thank Ms. Sushma Hegde for her assistance with corneal transplantation. The GL3 hybridoma was generously provided by Dr. Leo Lefransois, University of Connecticut Health Science Center (Farmington, CT).
This work was supported by National Institutes of Health Grants EY05631 and EY07641, and an unrestricted grant from Research to Prevent Blindness (New York, NY).
Abbreviations used in this paper: ACAID, anterior chamber-associated immune deviation; DTH, delayed-type hypersensitivity; i.c., intracamerally; KO, knockout; MIF, macrophage migration-inhibitory factor; LAT, local adoptive transfer.