Using Tolerance Induced Via the Anterior Chamber of the Eye to Inhibit Th2-Dependent Pulmonary Pathology1

Anterior chamber-associated immune deviation (ACAID), a manifestation of ocular immune privilege, prevents Th1-dependent delayed hypersensitivity from developing in response to eye-derived Ags, thereby preserving vision. Since Th2-type cells have recently been shown to mediate destructive inflammation of the cornea, we wondered whether pre-emptive induction of ACAID could inhibit Th2 responses. Using a murine model of OVA -specific, Th2-dependent pulmonary inflammation, we pretreated susceptible mice by injecting OVA alone into the anterior chamber, or by injecting OVA-pulsed, TGF-β2-treated peritoneal exudate cells i.v. These mice were then immunized with OVA plus alum strategy that generates Th2-mediated OVA-specific pulmonary pathology. When pretreated mice were challenged intratracheally with OVA, their bronchoalveolar lavage fluids contained far fewer eosinophils and significantly less IL-4, IL-5, and IL-13 compared with that of positive, nonpretreated controls. Similarly, lung-draining lymph node cells of pretreated mice secreted significantly less IL-4, IL-5, and IL-13 when challenged in vitro with OVA. Moreover, sera from pretreated mice contained much lower titers of OVA-specific IgE Abs. We conclude that Ags injected into the anterior chamber of the eye impair both Th1 and Th2 responses. These results reduce the likelihood that ACAID regulates Th1 responses via a Th2-like mechanism. Thus, immune privilege of the eye regulates inflammation secondary to both Th1- and Th2-type immune responses.

A nterior chamber-associated immune deviation (ACAID) 4 is induced when antigenic material is placed in the anterior chamber of the eye. ACAID is defined as an Agspecific, stereotypic, deviant systemic immune response in which the effectors of Th1-type immunogenic inflammation (T cells that mediate delayed hypersensitivity and B cells that secrete complement-fixing Abs (1)) are selectively deficient (2)(3)(4)(5). At the same time, animals with ACAID generate large amounts of IgG1 Abs (3). Since Th1 cells and complement-fixing Abs are also deficient in mice with predominant Th2 responses to an Ag (6), it has been proposed that ACAID is merely a Th2-type response elicited by Ags injected through this unusual route (7,8). To that end, Kosiewicz et al. (9) reported that mice that first received OVA in the anterior chamber followed by s.c. sensitization with OVA in CFA acquired OVA-specific T cells that secrete IL-4 and IL-10, but not IFN-␥. Moreover, D'Orazio and Niederkorn (8) showed that ACAID could not be induced in mice deficient in IL-10. Because Th2 cells can secrete IL-10 and because IL-10 is regarded as an immune inhibitory cytokine, the failure of ACAID to develop in IL-10-deficient mice is consistent with the view that ACAID is dependent upon Th2 responses. Moreover, since Th2-type immune responses have been often characterized as anti-inflammatory (10) and since it has been proposed that immune privilege in the eye exists to suppress inflammation, it is reasonable to propose that ACAID could be mediated by Th2 cells because ACAID is linked to the phenomenon of immunologic privilege in the eye and in the brain (11).
Although several experimental diseases of the eye are known to be mediated by Th1 effector cells (experimental autoimmune uveitis (12), herpes simplex keratitis (13)), evidence reported during the past decade has also implicated Th2 cells in the pathogenesis of ocular inflammatory disease. Foster and colleagues (14) reported that inflammation of the cornea in the setting of experimental herpes virus infection of the ocular surface can be mediated by HSV-specific Th2 cells. More recently, Pearlman et al. (15) have characterized a Th2-mediated inflammation of the cornea secondary to ocular infection with Onchocercal volvulus. Thus, since Th2 cells are capable of promoting inflammation of the ocular surface that leads to blindness, it is of interest to know whether functionally similar cells have the ability to mediate ACAID, a process that is believed to inhibit the blinding consequences of inflammation. To this point, it has recently been demonstrated that ACAID can readily be induced in mice in which the IL-4 gene had been disrupted, a genetic lesion that prevents these mice from acquiring Th2-type responses to exogenous Ags (8,9).
In the present study, we examined the potential role of Th2 cells in ACAID by determining whether induction of ACAID with OVA promoted or interfered with the development of a well-characterized model of OVA-induced experimental allergic lung disease in mice, a Th2-mediated disease (16 -18). Our results indicate that mice pretreated with an anterior chamber injection of OVA or with an i.v. injection of OVA-pulsed ACAID-inducing APCs exposed in vitro to TGF-␤2 (19,20) acquired OVA-specific ACAID and failed to develop the typical signs of OVA-dependent experimental allergic lung disease. These findings exclude the possibility that ACAID is mediated by Th2 regulatory cells.

Materials and Methods
Mice BALB/c mice were bred in our animal facility or were purchased from Taconic Farms (Germantown, NY) and were used at 8 -12 wk of age. All animals were treated according to the Association for Research in Vision and Ophthalmology Statement on the Use of Animals in Ophthalmic and Vision Research. All protocols were preapproved by the Animal Care and Use Committee of the Schepens Eye Research Institute in accordance with National Institutes of Health guidelines.

Administration of Ag by various routes and immunization regimens
For induction of experimental allergic lung disease, the procedure of sensitization and challenge was modified from the method of Kung et al. (16). Briefly, mice were immunized i.p. with 10 g of OVA (Sigma-Aldrich, St. Louis, MO) mixed with 2.25 mg of aluminum hydroxide (Imject Alum; Pierce, South Iselin, NJ) (alum) in 100 l of PBS. The animals received a booster i.p. injection of this alum-OVA mixture 7 days later. Five days after the second i.p. injection, mice were intubated and challenged intratracheally (i.t.) with 50 g of OVA dissolved in 50 l of PBS.
For intraocular injection, each mouse received a 3-l inoculation of OVA (50 g) dissolved in HBSS into the anterior chamber (AC) of the right eye 7 days before sensitization. In some experiments, 50 g of keyhole limpet hemocyanin (KLH; Calbiochem, Darmstadt, Germany) dissolved in PBS was inoculated into the AC.
For conventional sensitization, mice received a s.c. injection of 100 g of OVA emulsified in CFA (Life Technologies, Grand Island, NY).

Bronchoalveolar lavage (BAL)
Mice were euthanized by an i.p. injection of sodium pentobarbital 1 or 3 days after i.t. challenge with OVA. The trachea was dissected free from the underlying soft tissues and an 0.58-mm diameter tube was inserted through a small incision in the trachea. BAL was performed 10 times through the tracheal cannula with 1-ml aliquots of pyrogen-free PBS warmed to 37°C. BAL fluid harvested in the first 1 ml was centrifuged and the supernatant was collected and kept at Ϫ70°C until use for cytokine assay. A hemocytometer was used to count the total number of BAL cells under phasecontrast microscopy. For differential cell count, cytospin preparations were made and stained with Diff-Quik (Green Cross, Osaka, Japan). A total of 1000 cells was counted to calculate the differential populations of alveolar macrophages, neutrophils, eosinophils, lymphocytes, and monocytes that were identified by standard morphology.

Cytokine production of lung-draining lymph node (LN) cells
Lung-draining LN cells were harvested 1 or 3 days after i.t. challenge with OVA. The LN cells (2 ϫ 10 5 /well) were cultured for 120 h in the absence or presence of OVA (1 mg/ml) in 96-well plates, as first described by Janssen et al. (21). The LN cells were also cultured in a plate coated with anti-CD3 Ab (2C11, 10 g/ml in PBS). Cells were cultured in medium consisting of RPMI 1640 (BioWhittaker), 10% heat-inactivated FCS (Hy-Clone, Logan, UT), 2 mM L-glutamine, 10 mM HEPES buffer, 100 U/ml penicillin G sodium, 100 g/ml streptomycin sulfate, and 1 ϫ 10 Ϫ5 M 2-ME (Sigma-Aldrich). Supernatants were harvested and stored at Ϫ70°C until use for cytokine assays.

Cytokine assays
Content of cytokines in BAL and culture supernatants was analyzed by quantitative capture ELISA, according to the manufacturer's instructions for IL-4, IL-5, IFN-␥ (BD PharMingen, San Diego, CA), and IL-13 (R&D Systems). The detection limits of the ELISAs were 16 pg/ml for IL-4 and IL-5, 62.5 pg/ml for IL-13, and 290 pg/ml for IFN-␥. In some experiments, samples were diluted with PBS containing 2% BSA to an appropriate concentration.

Delayed hypersensitivity assay
Mice were challenged by intradermal injection of Ag (200 g of OVA/10 l of HBSS) into the ear pinnae 7 days after final exposure to OVA. Ear swelling was measured 24 and 48 h later with an engineer's micrometer (Mitutoyo; MTI, Paramus, NJ). Mice sensitized with OVA emulsified with CFA 7 days before challenge were used as positive control.

Statistical analyses
In experiments with groups of mice, each group contained at least five animals, and all experiments (both in vitro and in vivo) were repeated at least twice with similar results. The results displayed in the figures are representative of three or more experiments; SEM were calculated from a single experiment. Data were subjected to analysis by Student's t test as appropriate. A p Ͻ 0.05 was considered to be significantly different.

Effects of OVA injected intracamerally or of OVA-pulsed PEC pretreated with TGF-␤2 on OVA-specific humoral immune responses of mice with Th2 immunity
The first experiments were designed to determine whether mice initially exposed to OVA via an intracameral injection were able to develop the typical Th2-dependent spectrum of Abs when immunized subsequently with OVA and alum. Panels of adult BALB/c mice received OVA (50 g/3 l) into the AC of one eye. Seven days later these mice received an i.p. injection of OVA (10 g) in aluminum hydroxide (alum), followed 7 days later by a second i.p. injection of OVA plus alum. Three days later tail vein blood was collected, and sera were separated and assayed quantitatively by ELISA for levels of OVA-specific IgG1, IgG2a, IgG2b, IgG3, and IgE Abs. Positive control mice received two i.p. immunizations with OVA plus alum, but no intracameral injection of OVA. An additional panel of BALB/c mice received a s.c. immunization with OVA (50 g) in CFA; their sera were collected 10 days later. The results of this experiment are presented in Fig. 1A. Mice that received only i.p. injections of OVA plus alum (positive controls) generated high titers of OVA-specific IgE and IgG1, but low levels of IgG2a, 2b, and 3 Abs. By contrast, mice that similarly received i.p. injections of OVA plus alum following an AC injection of OVA (ACAID group) had barely detectable levels of OVA-specific IgE, but otherwise their levels of serum OVA-specific Abs were similar to the positive controls. As anticipated, mice immunized with OVA plus CFA produced high levels of IgG2a, IgG2b, and IgG3 Abs, but low levels of IgE and IgG1 (data not shown).
Similar experiments were performed in which a panel of BALB/c mice was pretreated with an i.v. injection of BALB/c PEC that were incubated overnight with TGF-␤2 and OVA. As before, these mice were immunized subsequently with two i.p. injections of OVA plus alum. Positive controls received only two i.p. injections of OVA plus alum, while an OVA-CFA control panel received a single s.c. injection of OVA plus CFA. The sera obtained from these mice were assayed for OVA-specific Abs. As revealed in Fig. 1B, mice that first received OVA-pulsed PEC treated with TGF-␤2 displayed low levels of OVA-specific IgE, but levels of IgG1 and IgG3 that were higher than the positive controls. Together these findings indicate that mice first exposed to OVA via the AC or via OVA-pulsed, TGF-␤2-treated PEC displayed an impaired capacity to produce IgE Abs when subjected subse-quently to an immunizing regimen that, in normal mice, generated high titer IgE responses. Thus, the humoral immune response of ACAID is distinctly different from a typical Th2-type humoral response, and ACAID suppresses IgE Abs to the Ag first encountered through the AC.

Effects of OVA injected intracamerally or of OVA-pulsed PEC pretreated with TGF-␤2 on airway inflammation of mice with a Th2 OVA pulmonary response
As before, positive control panels of BALB/c mice received an immunizing regimen of OVA plus alum designed to elicit an experimental model of allergic asthma. Five days after the second i.p. injection of OVA plus alum, the mice were challenged i.t. with OVA (50 g). Three days later, the mice were euthanized, their bronchoalveolar spaces lavaged, and the fluid analyzed for its content of cells and cytokines. Experimental panels were pretreated as before with either an AC injection of OVA or an i.v. injection of OVA-pulsed, TGF-␤2-treated PEC. Negative control mice received only an i.t. challenge with OVA. The results of these experiments are summarized in Fig. 2. As anticipated, the BAL of positive control mice contained a significant increase in total cells, compared with negative controls. By contrast, the BAL obtained from mice pretreated with OVA in the AC contained no more cells than the BAL of negative controls. The increased cellular content of positive control BAL was ascribable to eosinophils, lymphocyte, neutrophils, and monocytes. Similar results were observed with BAL from mice pretreated with an i.v. injection of OVApulsed, TGF-␤2-treated PEC (data not shown). The accumulation of eosinophils in BAL is a characteristic feature of airway inflammation in experimental allergic lung disease of this type. Therefore, the virtual absence of eosinophils in the BAL of mice pretreated with OVA in the AC or by OVA-pulsed, TGF-␤2-treated PEC indicates that these pretreatments mitigated this aspect of Th2-mediated pathology.
To determine whether the changes in cell content of BAL of mice pretreated with OVA in the AC before i.p. immunization with OVA plus alum were Ag specific, panels of BALB/c mice received an AC injection of OVA or KLH. Seven days later, both panels Mice sensitized with OVA/alum twice without any pretreatment were used as positive control and naive mice were used as negative control. Serum was harvested 10 days after first sensitization for estimations of OVAspecific Abs by ELISA. Mean OD readings for each group of sera are presented Ϯ SEM. OVA-specific IgE values of mice with ACAID and of mice that received OVA-pulsed, TGF-␤2-treated PEC were less than those of positive control ‫,ء(‬ p Ͻ 0.0000001; ‫,ءء‬ p Ͻ 0.0001, respectively). OVAspecific IgG1 and IgG3 values of mice that received tolerogenic PEC were greater than those of positive controls and the latter was less than that of mice sensitized with OVA and CFA ‫,ءءء(‬ p Ͻ 0.05). received i.p. immunizing regimens of OVA plus alum. Three days after i.t. challenge with OVA, BAL was collected and the cellular content was assessed. As the results presented in Fig. 3 reveal, only the BAL from mice pretreated with OVA in the AC lacked high numbers of eosinophils. By contrast, the number of total cells and eosinophils were comparable to positive controls in the BAL of mice pretreated in the AC with KLH, but immunized and challenged i.t. with OVA. Thus, the changes observed in pulmonary inflammation of ocularly pretreated mice were Ag dependent and specific.
To determine the content of relevant cytokines in airway inflammation of the Th2 type, BAL was harvested 1 day after i.t. challenge of experimental and positive control mice described above. These BAL fluids were analyzed by ELISA for quantitative levels of IL-4, IL-5, and IL-13. As the results in Fig. 4 reveal, the BAL fluids from positive control mice contained large amounts of all three cytokines, whereas the BAL fluids obtained from mice pretreated with OVA in the AC contained much lower levels of IL-4, IL-5, and IL-13. Similar results were obtained with BAL from mice pretreated with OVA-pulsed, TGF-␤2-treated PEC (data not shown). Together, the fluids obtained from lavaging the bronchoalveolar spaces of mice pretreated with OVA in the AC or with OVA-pulsed, TGF-␤2-treated PEC were deficient in the cells (eosinophils) and cytokines (IL-4, IL-5, and IL-13) that are believed to play a prominent role in the pathogenesis of Th2-mediated allergic asthma.

Effects of OVA injected intracamerally or of OVA-pulsed PEC pretreated with TGF-␤2 on phenotype of OVA-specific lymphoid cells in lung-draining LN
In experimentally induced allergic lung disease in mice, the LN that drain the airways of the lung (hilar and mediastinal) are documented to contain T cells of the Th2 phenotype that are responsible for causing the disease (21,22). Therefore, we postulated that intracameral OVA pretreatment of mice destined to develop experimental allergic lung disease would impact the phenotype of OVA-specific cells in the draining LN. As before, panels of BALB/c mice were immunized i.p. with two injections of OVA plus alum; some panels were pretreated with AC injection of OVA, while other panels were pretreated with i.v. injections of OVA-pulsed, TGF-␤2treated APCs. All mice were then challenged i.t. with OVA. After 1 or 3 days, their lung-draining LN were harvested and cultured in vitro with anti-CD3 Abs or with OVA, respectively, for 120 h. The supernatants were then harvested and assayed by ELISA for IL-4, IL-5, IL-13, and IFN-␥. The results are displayed in Figs. 5 and 6. When   exposed in vitro to anti-CD3, LN cells from positive controls secreted all four cytokines at easily measurable levels. However, when lymphoid cells were obtained from mice pretreated with OVA in the AC (Fig. 5A) or with i.v. injected OVA-pulsed, TGF-␤2-treated PEC (Fig.  5B), anti-CD3-responding cells produced significantly less IL-5 and IL-13, but not less IFN-␥, compared with positive controls. Similarly, positive control draining LN cells exposed in vitro to OVA (Fig. 6) produced large amounts of IL-5 and IL-13 compared with mice pretreated with OVA in the AC. Increased expression of Th2 cytokine mRNA levels (IL-4, IL-5, and IL-13) in lung draining LN correlated positively with levels of their protein production (data not shown). Thus, ACAID inhibits the generation of lymphoid cells in draining LN that produce IL-5 and IL-13, two cytokines strongly implicated in the pathogenesis of experimental allergic lung disease.
An alternative explanation for the inhibition of IL-5 and IL-13 production in draining LN is that IFN-␥-producing Th1-type cells are present. To examine this possibility, BALB/c mice received first an AC injection of OVA, followed 1 wk later by immunization with OVA plus alum. The delayed hypersensitivity of these mice was then assessed and compared with that of mice that received OVA plus alum or OVA plus CFA immunization alone. The mice pretreated with OVA in the AC displayed no evidence of OVAspecific delayed hypersensitivity (data not shown). We conclude that Th1 cells are not present in mice pretreated with OVA in the AC and that ACAID is responsible for the failed IL-5 and IL-13 production in the lung draining LN of mice pretreated with OVA in the AC before immunization and intratracheal challenge with OVA.

Discussion
The systemic unresponsiveness that is elicited by introducing Ag first through the AC of the eye was first reported by Kaplan et al. in 1975 (23). Since then, a wide array of chemically and biologically diverse antigenic materials placed in the AC have been reported to elicit ACAID (5,24,25). Because the first reports of ACAID arose from tissue transplantation experiments and because delayed hypersensitivity and Th1-type immune responses have come to be considered dominant effectors of graft rejection, vir-tually all studies of ACAID have emphasized suppression of Th1type immunity as a cardinal feature of the phenomenon. Since Th1 responses were first described as the reciprocal of Th2 responses and since Th2 cells were shown to suppress Th1-type responses (26), various investigators consider ACAID to be merely a Th2 response to Ags introduced into the AC of the eye. Fragmentary support for this possibility was generated by the experiments of Kosiewicz et al. (9) and of Niederkorn and D'Orazio (8), although other results cast doubt on the notion that ACAID is similar to Th2 responses. First, Kosiewicz et al. (9) also reported that unless mice that receive an AC injection of OVA are subsequently immunized systemically with OVA plus CFA, no evidence of OVA-responding Th2 cells can be found in spleen and LN. Instead, the spleens of these mice contain cells that secrete only TGF-␤ when stimulated with OVA in vitro. Second, mice deficient in IL-4 (because the relevant gene has been disrupted) readily acquire ACAID when Ag is injected into the AC. Because these genetically manipulated mice are incapable of developing Th2 responses, it seems unlikely that the ACAID they acquire can be Th2 mediated.
The results of the studies reported here diminish the possibility that the regulation of ACAID and Th2 are identical, at least for the heterologous protein Ag, OVA. Mice pretreated with either an AC injection of OVA or an i.v. injection of OVA-pulsed, TGF-␤2treated APCs failed to generate OVA-specific IgE responses when immunized i.p. with OVA plus alum. Moreover, when these mice were challenged i.t. with OVA, their BAL fluids contained few if any traces of a Th2-dependent inflammatory response: sparse eosinophils and lymphocytes were present, and only trivial amounts of IL-4, IL-5, and IL-13 were detected. By contrast, the positive control mice immunized with OVA plus alum generated robust OVA-specific IgE responses, and their BAL following intratracheal challenge with OVA was rich in cells, especially eosinophils, and in IL-4, IL-5, and IL-13.
The failure of mice pretreated with OVA in the AC or with the surrogate ACAID-inducing APCs generated in vitro to generate intrapulmonary Th2 responses to OVA challenge was also evident when lymphoid cells were evaluated in the lung-draining LN. Whether stimulated in vitro with anti-CD3 Abs or with OVA, draining LN cells of OVA-pretreated mice failed to secrete significant amounts of IL-5 or IL-13. RT-PCR analysis of LN cells and BAL cells showed that the genes for IL-5, IL-13, and IFN-␥ were either silenced or greatly repressed by OVA-AC treatment. These findings lead us to conclude that pretreatment of mice with AC injection of OVA or with an i.v. injection of OVA-pulsed, TGF-␤2-treated APCs suppressed OVA-specific Th1 and Th2 responses comparably.
Our finding that induction of ACAID inhibits Th2-dependent pathology extends our knowledge of the range of immune effector systems altered in ocular immune privilege. Suppression of Th2dependent pathology joins suppression of delayed hypersensitivity and suppression of complement-dependent inflammation. Yet even with the addition of suppressed Th2 responses to the immune privilege repertoire, there are immune effector systems that remain intact in ACAID. Mice with ACAID produce large amounts of IgG1 Abs. In that regard, it was of interest to learn in our present studies that, even though IgG1 is usually included among the Abs promoted by Th2 cells, IgG1 production persisted at high levels in mice that were pretreated with OVA in the AC followed by an i.p. immunization with OVA plus alum. Only IgE production, another Ab thought to be facilitated by Th2 cells, was diminished in these mice. Previous reports have demonstrated that mice with ACAID induced by minor histocompatibility Ag-bearing tumor cells acquire large numbers of activated CD8 ϩ cytotoxic T cell precursors in their secondary lymphoid organs (27). Thus, on the cell-mediated arm of immune responses, mice with ACAID can still mount cytotoxic T cell responses and, on the humoral immune side, mice with ACAID can still mount robust IgG1 responses to eye-derived Ags (3). In these animals, promotion of CD8 ϩ T cell and IgG1 responses at the expense of Th1, Th2, and complement-fixing Ab responses serves to emphasize that the systemic immune response to eye-derived Ags is "deviant," ergo the term ACAID.
Mice with Th2-biased immune responses produce large amounts of noncomplement-fixing IgG1 Abs as well as IgE. Our evidence indicates that induction of ACAID in mice that are subsequently exposed to a Th2-biasing immunization regimen prevents their production of IgE, while the production of IgG1 is preserved. At the very least, this outcome suggests that the regulations by T cells of IgG1 and IgE are distinct and that the regulatory T cells of ACAID suppress both complement-fixing Abs (IgG2a, IgG2b, IgG3) and IgE, whereas Th2 cells suppress complement-fixing IgG Abs, but not IgE.
It is worth pointing out that induction of ACAID offers for consideration a novel strategy with which to suppress an important Th2-dependent immunopathologic condition, allergic asthma. Previously, ACAID has been used experimentally to suppress or prevent rejection of orthotopic corneal allografts in mice (28,29) as well as experimental autoimmune uveitis (30,31), both of which are Th1-dependent pathologic conditions. Alternatively, attempts to use ACAID experimentally to prevent rejection of solid tissue allografts such as skin and heart have met with no success (J. W. Streilein, unpublished observations). We believe that the explanation for this conundrum resides in the fact that ACAID is immune deviation, i.e., a selective deficiency of one or more immune effector modalities, but not of all effector modalities. Whereas experimental allergic lung disease appears to be mediated solely by Th2 responses, and experimental autoimmune uveitis and acute rejection of orthotopic corneal allografts appear to be mediated solely by Th1 responses, rejection of skin or heart allografts can also be achieved by CD8 ϩ T cells and perhaps by Ab-dependent cell-mediated cytotoxicity that uses noncomplement-fixing Abs. One prediction from these considerations is that induction of ACAID may be a useful therapeutic strategy, but only if the pathologic condition in question is mediated purely by the effector modalities that ACAID suppresses.