Inflammatory Mediators in Uveitis: Differential Induction of Cytokines and Chemokines in Th1- Versus Th2-Mediated Ocular Inflammation

The immune system is the body’s principal defender against disease, but it can also perpetrate severe disease. Immune defenses can destroy vital tissues by two mechanisms: 1) by directly attacking the body’s own tissues, and 2) by creating an inflammatory response to a foreign Ag that is more deleterious than the Ag itself. The former mechanism is implicated in autoimmune diseases, while the latter is exemplified by allergic reactions.

In the eye, both of these destructive mechanisms contribute to vision loss. Misdirected autoimmune responses against ocular Ags, with ensuing inflammation, can occur spontaneously or after ocular injury. Also, defensive responses to foreign Ags in the eye can cause inflammation. Whatever the inciting event, even transient inflammation threatens vision, as it can destroy the transparency of the lens and cornea and the integrity of retinal components. Currently, ocular inflammation is responsible for ∼10–15% of vision loss in the United States (1, 2).

To control deleterious immune responses, it is essential to identify the molecules that promote these responses. Proinflammatory cytokines are key players in deleterious autoimmune responses (3). Other important candidates are the chemokines, a family of 8- to 14-kDa proteins that mediate leukocyte emigration from blood vessels into tissues. Over 40 human and mouse chemokines have been identified. In the last several years, chemokines have been shown to play critical roles in the development of inflammation in diverse disease models (4, 5, 6). Chemokines are produced by ocular cells cultured in vitro, and several chemokines have been detected in ocular fluids and tissues (7, 8, 9, 10, 11, 12, 13, 14). However, a complete analysis of chemokine involvement in the development of ocular inflammation has not been conducted.

Quantitative identification of multiple mRNA transcripts in a small tissue sample has recently become possible with the advent of kinetic (real time) RT-PCR (15). In this study, we utilized this technique to characterize the expression of 74 mRNA transcripts of cytokines, chemokines, and chemokine receptors during T cell-induced ocular inflammation. We examined a murine model in which transgenic (Tg)² mice expressing hen egg lysozyme (HEL) in the lens of the eye are injected with T cells specific for HEL (16). In this model, both Th1- and Th2-polarized cells induce eye inflammation when introduced into irradiated mice. The inflammatory infiltrates induced by Th1 and Th2 cells differ, containing primarily mononuclear (MNL) or polymorphonuclear (PMN) cells, respectively. By analyzing these mice, we were able to compare the complement of cytokine, chemokine, and chemokine receptor mRNA transcripts expressed in the eye during Th1- and Th2-polarized responses attracting distinct leukocyte subsets. We compared these patterns to mediator expression patterns in experimental autoimmune uveitis (EAU), a well-characterized animal model of human autoimmune eye disease. To our knowledge, the findings detailed in this work represent the most comprehensive analysis to date of 1) inflammatory mediator transcript expression during the development of ocular inflammation, and 2) chemokine transcript expression in Th1- vs Th2-induced inflammation. The results of this large-scale study offer a new perspective on the role of individual mediators in the development of inflammation.

HEL-Tg mice, expressing membrane-bound HEL, on the FVB/N background, were generated as detailed elsewhere (17). HEL-specific TCR Tg mice, on the B10.BR background, designated 3A9 (18), were a generous gift from M. Davis (Stanford University, Stanford, CA). Tg mice from each of the two lines were mated to produce (FVB/N × B10.BR)F₁ hybrids, expressing either one of the two transgenes. Only such F₁ hybrid mice were used in all experiments of the present study. The mice were housed in a pathogen-free facility, and all manipulations were conducted in compliance with the National Institutes of Health Resolution on the Use of Animals in Research.

Th1 and Th2 cells expressing a HEL-specific TCR were prepared as follows: spleen and lymph node cells of 3A9 mice were pooled and the T cell fraction was partially purified on enrichment columns (R&D Systems, Minneapolis, MN). CD4 cells were then purified by magnetic sorting with SuperMACS, using beads directly coupled to anti-mouse CD4 (Miltenyi Biotec, Sunnyvale, CA). Purified (>97%) CD4 cells were then cultured at 2.5 × 10⁵/ml in RPMI 1640 medium supplemented with 50 μM 2-ME, antibiotics, and 10% FBS (complete medium) with 2.5 × 10⁶/ml syngeneic wild-type splenocytes, irradiated with 30 Gy, as APC, in the presence of either 2 μg/ml HEL (Sigma-Aldrich, St. Louis, MO), 10 ng/ml IL-12 (Sigma-Aldrich), and 10 μg/ml anti-IL-4 Ab (BD PharMingen, San Diego, CA) for Th1, or 0.2 μg/ml HEL, 10 ng/ml IL-4 (BD PharMingen), 10 μg/ml anti-IFN-γ Ab (BD PharMingen), and 10 μg/ml anti-IL-12 Ab (BD PharMingen) for Th2. After 3 days, cultured cells were expanded with 40 IU/ml IL-2 (Chiron, Emeryville, CA) for 3–5 days and then restimulated at 2.5 × 10⁵/ml with 2.5 × 10⁶ irradiated syngeneic APC in the presence of either 2 μg/ml HEL, 40 IU/ml IL-2, and 10 ng/ml IL-12 for Th1, or 0.2 μg/ml HEL, 40 IU/ml IL-2, and 10 ng/ml IL-4 for Th2. Three days later, cells were harvested, washed, resuspended in RPMI 1640, and injected i.v. into recipient mice, as indicated.

Upon injection, the T cell preparations were >90% alive and blastic, and >97% CD4 positive. Recipient HEL-Tg mice were irradiated 4–5 h before cell injection with 4.5 Gy. Recipient mice were sacrificed immediately before cell injection, or 2, 4, or 7 days thereafter. At the time of sacrifice, eyes were removed. One eye was frozen for later RNA preparation or microdissection, and the other eye was used for histological analysis. Eye sections were stained by conventional H&E or by an eosinophil stain that highlights cyanide-resistant eosinophil peroxidase (19, 20). Briefly, eye frozen sections were fixed in 1% Formalin in acetone for 30 s, then stained for 10 min in PBS containing 0.4 mg/ml NaCN, 3 μl/ml 3% hydrogen peroxide, and 0.75 mg/ml diaminobenzidine (Sigma-Aldrich). Slides were then counterstained in Gills hematoxylin for 1 min. For comparison, serial sections from the same eyes were stained with H&E.

The severity of ocular inflammation in recipient mice was scored by two investigators, who separately evaluated the level of inflammation in the anterior segment, vitreous, and retina, on a scale of 0–3. The final score consisted of the sum of the three subscores, with a maximum value of 9. Severity of changes was assessed according to tissue structural changes, intensity of cellular infiltration, and levels of proteinaceous exudates.

EAU was induced as follows. B10.A mice (The Jackson Laboratory, Bar Harbor, ME) were immunized with 50 μg bovine interphotoreceptor retinoid-binding protein (IRBP, a generous gift from B. Wiggert, National Eye Institute, National Institutes of Health), emulsified in CFA containing 2.5 mg/ml Mycobacterium tuberculosis H37RA. The emulsion was injected into the base of the tail and two thighs in a total volume of 0.2 ml, and the mice were concurrently injected i.p. with 0.5 μg pertussis toxin (Sigma). Mice were sacrificed on day 21 postimmunization, and their eyes were processed for RNA analysis and histological examination, as described above.

Real time RT-PCR assays were performed to specifically quantify levels of RNA transcripts. RNA was prepared from frozen eyes as follows. First, each eye was solubilized using TRIzol reagent (Life Technologies, Gaithersburg, MD) in a tissue homogenizer (Kontes Glass Company, Vineland, NJ). RNA was extracted with chloroform/isoamyl alcohol, precipitated with ethanol, and resuspended in diethyl pyrocarbonate water using standard procedures. Isolated RNA was incubated with 10 U DNase I (Boehringer Mannheim, Indianapolis, IN) in the presence of RNasin (Promega, Madison, WI) for 30 min at 37°C. The samples were then heat inactivated at 95°C for 10 min, chilled, and reverse transcribed with Superscript II reverse transcriptase (Life Technologies) with random hexamers, according to the manufacturer’s protocol. Primers were either obtained from Perkin-Elmer (Foster City, CA), or generated with Primer Express software (Perkin-Elmer) and were synthesized in the DNAX core facility, as reported previously (21, 22). Whenever possible, primer pairs were designed to span intron/exon borders. Samples were then subjected to 40 cycles of amplification at 95°C for 15 s, followed by 60°C for 1 min using an ABI Geneamp 5700 sequence detection system and SYBR green buffer, according to the manufacturer (Perkin-Elmer). PCR amplification of the housekeeping gene ubiquitin was performed for each sample to control for sample loading and to allow normalization between samples, according to the manufacturer’s instructions (Perkin-Elmer). Both water and genomic DNA controls were included to ensure specificity. Each data point was examined for integrity by analysis of the amplification plot and dissociation curves.

In adoptive transfer experiments, replicate samples were treated as follows. For experimental samples, mRNA from eyes of three replicate mice was extracted and analyzed separately. For control mice, mRNA from three replicate eyes was extracted and reverse transcribed, then equivalent amounts of individual cDNA reactions were combined to create pooled samples for real time RT-PCR.

For quantitative RT-PCR experiments, mRNA levels for each primer pair were normalized to the housekeeping gene ubiquitin, according to the manufacturer’s instructions (Perkin-Elmer). Ubiquitin-normalized values from experiments performed with identical reaction conditions can be compared with each other, although variations in primer pair efficiency make these comparisons approximate. Quantitative RT-PCR for all adoptive transfer experiment samples and controls were performed at the same time to enable comparisons of ubiquitin-normalized values. Similarly, all PCR reactions for EAU experiments and relevant controls were performed at the same time to facilitate comparison. Note, however, that the ubiquitin-normalized values between these two data sets should not be directly compared.

Calculations of the fold increase in mRNA level from baseline (shown in Table I) were performed as follows. For adoptive transfer experiments, the maximal mean mRNA level detected at any time point (usually at day 4 for Th1-induced disease and day 7 for Th2-induced disease) was used as the peak mRNA level. The mRNA level present in pooled samples at time zero (after irradiation, but before T cell injection) was used as the baseline level. Presented in Table I is the following ratio: peak mRNA level/baseline mRNA level. For EAU experiments, peak level was the mean mRNA level at day 21, and the baseline was the mean mRNA level in unmanipulated B10A mouse eyes. All values were rounded to two significant digits. Ratios of the fold increase in Th1-induced disease to the fold increase in Th2-induced disease (which also represents the ratio of the peak value in Th1-induced disease to the peak value in Th2 disease) are also shown in Table I. In compiling the data, molecules that were undetectable at time zero were assigned a baseline value of 1 for the purpose of calculating a ratio. Most values represent the average of three mRNA levels determined from three replicate mouse eyes; some represent the average of two levels if one reaction failed.

Retinal pigment epithelium (RPE) cells or infiltrating leukocytes were isolated from mouse eye frozen sections using a PixCell II laser microdissection microscope (Arcturus, Mountain View, CA). Infiltrating leukocytes or RPE cells were identified by light microscopy and removed to a membrane using laser settings of power 35 mW, duration 3 ms, spot size 7.5 μm, or power 40 mW, duration 3.5 ms, spot size 7.5 μm, respectively. The captured RPE samples also contained small numbers of adjacent resident cells, mostly choroidal capillary endothelial cells and fibroblasts.

RNA was extracted from RPE cells or infiltrating leukocytes captured by laser microdissection as follows. A total of 200 μl TRIzol (Life Technologies) was used to lyse the cells. RNA was isolated by chloroform extraction and isopropanol precipitation, according to the manufacturer’s instructions. After digestion with DNase, total RNA was used for cDNA synthesis. A Superscript II RNase H⁻ reverse transcriptase system (Life Technologies) and random primers (Promega) were employed. PCR was performed with 2 μl cDNA, 3 pmol of each ³²P-labeled primer, 4 nmol of each dNTP, 1× GeneAmp buffer, 1 U AmpliTag Gold Polymerase (Perkin-Elmer), and a final concentration of 1.5 mM MgCl₂. The following primer sequences were used for PCR amplification: monokine induced by IFN-γ (MIG)/CXCL9, 5′-GATCAAACCTGCCTACATCC-3′ and 5′-GGCTCTGTAGAACACAGAGT-3′; macrophage-inflammatory protein (MIP)-1γ/CCL9, 5′-GCCCACTAAGAAGATGAAGCCT-3′ and 5′-CCTTCTCTAAAGCAAATGTAA-3′; eotaxin/CC chemokine ligand (CCL)11, 5′-TAGGT AAGCAGTAACTTCCATCTGTCTC-3′ and 5′-TGACTAAATCAAGCAGTTCTTAGGCTCTG-3′; RANTES/CCL5, 5′-CCTCACCATCATCCTCACTGC-3′ and 5′-TCTTCTCTGGGTTGGCACACA-3′; C-10/CCL6, 5′-ATAACGCGTATGCAGGCCTCATACAAGAAATGG-3′ and 5′-TAC TGCAGTCAAGCAATGACCTTGTTC-3′; 18S, 5′- AGGAATTGACGGA AGGGCAC-3′ and 5′-GGACATCTAAGGGCATCACA-3′.

Primer sequences for RANTES/CCL5 (23), C-10/CCL6 (24) eotaxin/CCL11, MIP-1γ/CCL9, and MIG/CXCL9 (25) were described in the corresponding cited publications. MIG/CXCL9, MIP-1γ/CCL9, and C10/CCL6 primers were obtained from Life Technologies, RANTES/CCL5 primers were a generous gift from N. Tuaillon (National Eye Institute, National Institutes of Health), and eotaxin/CCL11 primers were a generous gift from J. Farber (National Institute of Allergy and Infectious Diseases, National Institutes of Health). The 18S primers were purchased from Ambion (Austin, TX).

Primers were labeled with ³²P before PCR. Reactions were conducted in PCR Express Thermal Cycler (Hybaid, Middlesex, U.K.) for 40 cycles with an annealing temperature of 60°C, a denaturing temperature of 94°C, and an extension temperature of 72°C. PCR products were separated on polyacrylamide gel, and radioactive bands were detected on a PhosphorImager Storm 860 (Molecular Dynamics, Sunnyvale, CA) using ImageQuant software.

Ocular inflammation was induced by adoptively transferring HEL-specific polarized Th1 or Th2 cells into Tg recipient mice expressing HEL in their lens. Fig. 1 summarizes the mean histological severity of ocular inflammation in groups of recipient mice examined on day 2, 4, or 7 after T cell injection. Eyes of mice injected with Th1 cells showed no changes on day 2, mild to moderate inflammation on day 4, and moderate to severe inflammation on day 7 postinjection. Mice injected with Th2 cells had no ocular changes on day 2, none to mild inflammation on day 4, and mild to moderate ocular inflammation on day 7. It is of note that Th1 cells are more potent than Th2 cells in this system, inducing more severe disease even when injected at a much smaller number than Th2 cells (1 × 10⁶ vs 20 × 10⁶ per recipient, respectively). Fig. 2,A demonstrates typical histological changes in eyes of recipient mice. Th1 cell-induced disease is characterized by MNL infiltration, proteinaceous exudate, and retinal folding, whereas Th2 recipient eyes show infiltration with mostly PMNs, many of which are eosinophils (Fig. 2 B). These ocular changes in Th1 and Th2 recipient mice are consistent with those observed and described in more detail in our previous study (16)

To identify inflammatory mediators involved in eye inflammation, we measured mRNA levels of 34 cytokines, 26 chemokines, and 14 chemokine receptors by real time RT-PCR in eyes of recipient mice on days 0, 2, 4, and 7 post cell injection. Fig. 3 summarizes the kinetics of changes in mRNA levels of molecules that increased significantly in inflamed eyes, or exhibited high levels of baseline expression. Table I shows the complete list of tested molecules and records their baseline levels at day 0, as well as their highest measured mRNA levels, presented as fold increase over the baseline. In Fig. 3, molecules are divided into those that were preferentially expressed in Th1 (a) or Th2 (b) recipients, or those that were similarly expressed in eyes of both recipient groups (c).

As shown in Fig. 3,a, the inflammatory molecules preferentially expressed in Th1 recipient eyes included mRNA transcripts of eight cytokines (IL-1α, IL-1β, IL-1R antagonist, IL-6, IL-18, TNF-α, IFN-γ, and osteopontin), eleven chemokines (TCA-3/CCL1, MCP-1/CCL2, MIP-1α/CCL3, MIP-1β/CCL4, RANTES/CCL5, MCP-3/CCL7, MIP-3β/CCL19, MIP-2/CXCL1–3, MIG/CXCL9, IFN-γ-inducible protein-10 (IP-10)/CXCL10, and lymphotactin/XCL1), and five chemokine receptors (CCR1, CCR2, CCR5, CXCR3, and CCR7). In the eyes of Th1 recipients, significant increases were also observed in the mRNA levels of IL-12p40 and IL-12p35 (Table I), but the peak normalized units of these cytokines (240 and 416, respectively, on day 4) were quite low and, therefore, were not included in Fig. 3. Considerably fewer molecules were preferentially expressed in eyes of Th2 cell recipients (Fig. 3,b). These included transcripts of four cytokines (IL-4, IL-5, IL-9, and IL-13), two chemokines (eotaxin/CCL11 and TARC/CCL17), and one chemokine receptor (CCR-3). Molecules that were similarly expressed in recipients of Th1 or Th2 cells (Fig. 3 c) included transcripts of four cytokines (TGF-β2, TGF-β3, platelet-derived growth factor (PDGF)-β, and PDGF-R), six chemokines (C-10/CCL6, MIP-1γ/CCL9, MDC/CCL22, CTACK/CCL27, BRAK/CXCL14, and fractalkine/CX3CL1), and two chemokine receptors (CXCR6 and CX3CR1).

Examination of the data shown in Fig. 3 and Table I reveals several points of interest. 1) In general, cytokine mRNA expression patterns in this model fit known Th1 and Th2 associations. 2) Chemokine mRNA expression patterns were consistent with the type of inflammatory infiltrate seen in each disease (i.e., eyes with Th1 disease expressed transcripts of T cell and monocyte chemoattractants, whereas eyes with Th2 disease up-regulated the mRNA of eosinophil attractant eotaxin/CCL11). 3) Chemokine mRNA expression patterns seen in this study contribute new information regarding associations with Th1- or Th2-induced inflammation (see Discussion). 4) Minimal or no expression of cytokine or chemokine transcripts was detected in recipient eyes on day 2 post cell transfer, whereas steep increases in mRNA for these mediators were observed by day 4. 5) Peak mRNA expression was observed on day 4 for most mediators in Th1 recipient eyes, but on day 7 for mediators preferentially expressed in Th2 recipient eyes. 6) Chemokine receptor transcripts seen in Th1- or Th2-induced disease states matched the chemokine ligand mRNAs that were present. Furthermore, all chemokine receptor transcripts expressed in eyes of either Th1 or Th2 recipients reached their highest expression level on day 7, i.e., after peak mRNA expression of their chemokine ligands (in most cases). 7) mRNA of several inflammatory molecules were expressed at significant levels in uninflamed control eyes. These included six cytokines (IL-18, TGF-β2, TGF-β3, PDGF-β, PDGF-R, and osteopontin), five chemokines (C-10/CCL6, MIP-1γ/CCL9, Eotaxin/CCL11, BRAK/CXCL14, and fractalkine/CX3CL1), and two chemokine receptors (CXCR6 and CX3CR1). Fig. 3 and Table I show the measured levels of these constitutively expressed mRNA molecules in mice on day 0. Similar constitutive expression of these thirteen molecules was observed in eyes of two other control groups, namely, intact (nonirradiated) HEL-Tg mice (data not shown) and naive B10.A mice (see below). Interestingly, the expression levels of nine of these thirteen molecules did not change during the inflammatory process in the eyes of Th1 or Th2 recipients (Fig. 3).

Multiple studies have indicated that ocular cells have the potential to produce chemokines (7, 9, 10, 11). To define the origin of abundantly expressed mRNA transcripts in our model, we isolated both infiltrating leukocytes and RPE from frozen sections of inflamed eyes using laser capture microdissection. Fig. 4,A shows a representative microdissection of an eye section with Th1-induced inflammation. Microdissected samples from this and other sections of inflamed eyes of Th1 or Th2 recipients were used to detect specific chemokine transcripts, employing conventional RT-PCR (Fig. 4 B). As expected, MIG/CXCL9 and RANTES/CCL5 transcripts were detected in Th1 disease samples only, eotaxin/CCL11 transcripts were detected in Th2 samples only, and C-10/CCL6 transcripts were detected in all samples. Notably, RANTES/CCL5 transcripts were detected in RPE extracts only, whereas MIG/CXCL9 transcripts were detected in both RPE and leukocyte extracts. Eotaxin/CCL11 mRNA was also detected in both RPE and leukocyte extracts, with a larger signal from the RPE extracts. C-10/CCL6 transcript was detected in both leukocyte and RPE extracts at approximately equal levels.

Interestingly, MIP-1γ/CCL9 transcripts were detected only in eye samples with Th2-induced disease, and only in the infiltrating leukocytes. These data indicate that the predominantly eosinophilic infiltrate seen in Th2-induced inflammation was a source of MIP-1γ/CCL9. Furthermore, MIP-1γ/CCL9 mRNA was present in whole eye extracts from Th1-induced inflammation (Table I), but this mRNA was not detected in leukocytes or RPE from these eyes. This remained true in four replicate experiments (two shown). This observation suggests that another cell type within the eye was producing MIP-1γ/CCL9 mRNA in Th1-induced inflammation (and possibly in Th2 disease as well).

EAU is a well-characterized model of human autoimmune eye disease, in which susceptible animals develop ocular inflammation when immunized with eye-specific proteins (26). To determine which inflammatory mediators play a role in the development of EAU, we studied the mRNA complement in eyes of B10.A mice in which EAU was induced by immunization with IRBP. These eyes showed the characteristic inflammation and tissue damage, with an inflammatory infiltrate comprised mainly of MNLs and a small number of PMNs (27).

The measured mRNA levels of cytokines, chemokines, and chemokine receptors in eyes with EAU are compared in Fig. 5 with those in control naive B10.A mice. Individual molecules of each of the three groups of inflammatory mediators are shown in this figure in order of their detected mRNA level. In addition, the calculated fold increase from baseline for all tested molecules in EAU eyes is recorded in Table I, allowing comparison among the molecule expression pattern in EAU with that in the Th1- and Th2-adoptive transfer models. The mediator transcript expression pattern in EAU eyes resembles that seen in Th1 recipient eyes, but enhanced expression of several molecules that characterize Th2-induced inflammation was also seen in EAU (Table I, Fig. 5).

The cytokines that characterized both Th1-induced ocular inflammation and EAU included in particular IL-1α, IL-1β, IL-1R antagonist, IL-6, TNF-α, and osteopontin. The expression of IFN-γ mRNA in the EAU eyes was also elevated, although to a lesser degree than in Th1 recipient eyes. On the other hand, eyes with EAU resembled Th2 recipient eyes in showing increased expression of IL-5 transcript (Table I).

A close similarity was clearly seen between the chemokine transcript expression patterns in EAU eyes and in Th1 recipient eyes. The most highly expressed chemokine transcripts in EAU eyes were the same as those seen in Th1 recipient eyes, namely, MIG/CXCL9, RANTES/CCL5, and IP-10/CXCL10, which characterized Th1-induced disease, and C-10/CCL6, BRAK/CXCL14, MIP-1γ/CCL9, and fractalkine/CX3CL1, which were expressed in both Th1- and Th2-induced ocular inflammation. Other chemokine transcripts found in Th1-induced inflammation were detected in EAU eyes at moderate levels. These include molecules such as MIP-1α/CCL3, MIP-1β/CCL4, or MIP-2/CXCL1–3. In contrast, no increase in expression of Th2-specific chemokine transcripts was observed in eyes with EAU.

The pattern of chemokine receptor transcript expression in eyes with EAU also showed similarities to that seen in Th1-induced inflammation, with increased expression of CCR1, CCR5, CCR7, CXCR3, and CXCR2 transcripts. Interestingly, CCR6 mRNA was expressed in EAU at remarkably higher levels than in Th1 or Th2 recipient eyes. CCR3 mRNA was only mildly elevated in EAU, in contrast to its elevation in Th2 recipient eyes. Transcripts of two receptors, CX3CR1 and CXCR6, were highly expressed in eyes with EAU as well as in eyes of naive control mice.

Ocular inflammation, leading to scarring and vision loss, occurs in a variety of human disease states. In many of these, inflammation is thought to be initiated by eye-reactive T cells (26, 28). In this study, we have dissected the pathogenesis of T cell-induced ocular inflammation by examining the complement of transcripts of major cytokines, chemokines, and chemokine receptors expressed in each of three animal models. The Th1- and Th2-adoptive transfer models allowed us to identify inflammatory mediators induced by these distinct T cell subsets. Comparison with EAU, a more physiologic model of autoimmune eye disease, allowed us to better understand the pathogenesis of the human diseases mimicked by EAU.

In the Th1- and Th2-adoptive transfer models, analysis of inflammatory mediator transcripts revealed three patterns of expression, i.e., preferential elevation in recipients of Th1 or of Th2 cells, or similar expression in eyes of both recipient groups. Notably, the number of chemokines and chemokine receptor transcripts that were preferentially expressed in Th1-induced ocular inflammation was much higher than that in Th2-induced disease (Fig. 3). This difference could be attributed in part to a greater capacity of Th1 cells to produce or induce chemokines (29), a capacity that may also explain in part the observation that these lymphocytes are profoundly more efficient than Th2 cells in inducing ocular inflammation in this experimental system (Fig. 1) (16).

The preferential expression of cytokine transcripts that we see in Th1- or Th2-mediated ocular inflammation is generally in line with well-established Th1/Th2 associations in other systems (30, 31, 32). Less is known, however, about the Th1/Th2 specificity of chemokines, and our results thus contribute new information to this issue (6).

Some of the Th1/Th2 associations we uncovered were expected based on previous studies. Seven chemokines that were found in the present study to be strongly associated with Th1-induced inflammation include MIG/CXCL9, IP-10/CXCL10, monocyte chemoattractant protein-1 (MCP-1)/CCL2, MCP-3/CCL7, RANTES/CCL5, MIP-1α/CCL3, and MIP-1β/CCL4. The affiliation of these seven chemokines with Th1 cells has been indicated in other studies by the findings that these chemokines 1) are induced by Th1 cytokines (33, 34, 35, 36), 2) are ligands for chemokine receptors specific to Th1 cells (37, 38, 39), and 3) are associated with Th1-induced inflammatory processes such as experimental autoimmune encephalomyelitis (EAE) (24, 40, 41, 42) or inflammatory bowel disease (43). Similarly, the two chemokines whose transcripts were preferentially expressed in Th2 recipient eyes are known to be Th2 associated. Both eotaxin/CCL11 and TARC/CCL17 are associated with other Th2 disease states (44, 45, 46, 47, 48) and are ligands for receptors found on Th2 cells (37, 38, 39). Furthermore, eotaxin/CCL11 has been shown to be induced by Th2 cytokines (45, 49, 50, 51).

Other chemokines characterized in our study have exhibited more equivocal Th1/Th2 associations. These include four chemokines that were preferentially associated with Th1-induced eye disease. MIP-2/CXCL1–3, a neutrophil chemoattractant, is expressed in other Th1-associated disease states (25, 40, 52, 53), but can also be up-regulated by Th2-specific cytokines IL-4 or IL-5 and inhibited by the Th1 cytokine IFN-γ (49, 54). TCA-3/CCL1 has been found in both Th1- and Th2-associated disease models (24, 25, 53), and the TCA-3/CCL1 receptor, CCR4, is selectively expressed on Th2 cells. In our model, mRNA of this chemokine was highly expressed in Th1 recipients on day 4, but its expression rose considerably in Th2 recipients on day 7. MIP-3β/CCL19 is known for its role in lymphocyte homing to lymph nodes (55), but little is known about its role in Th-induced inflammation. Lymphotactin/XCL1 was clearly Th1 associated in the present study (Fig. 3) and other systems (25, 29, 53), but it has been reported to down-regulate Th1 responses in some situations (56).

The three chemokine transcripts up-regulated in both Th1 and Th2 cell recipients are also not clearly Th1 or Th2 associated. Two of these, C-10/CCL6 and MIP-1γ/CCL9, have been proposed to play a role in Th2-polarized responses, but C-10/CCL6 may also be involved in EAE development (57, 58, 59). MDC/CCL22, a chemokine whose transcript was up-regulated in both Th1 and Th2 recipient eyes, although with different kinetics, has been affiliated mainly with Th2-mediated responses (46).

Conflicting evidence regarding the Th1/Th2 associations of many chemokines suggests that the elaboration of these molecules may be tissue and context dependent. Indeed, previous work has shown that different cell types can produce different chemokines in response to the same extracellular signal (60). In the eye, RPE cells have been shown to elaborate chemokines in vitro (7, 9, 10, 11). Laser capture microdissection enabled us to demonstrate, for the first time, chemokine transcript expression by these cells in vivo, during an ongoing inflammatory response. Furthermore, we found that RPE and infiltrating leukocytes differed partially in their chemokine expression patterns, even though the context (the inflamed eye) was the same (Fig. 4). This finding supports a role for RPE in shaping ocular inflammatory responses in an eye-specific manner. It is also of interest that different chemokine transcripts were expressed by RPE cells in eyes developing inflammation induced by Th1 or Th2 cells (Fig. 4).

It is particularly noteworthy that in addition to the mediator transcripts up-regulated during the inflammatory response, we found multiple mediator transcripts that were expressed constitutively in intact mouse eyes and whose expression levels did not change much during the inflammatory process. These included trancripts of three chemokines (fractalkine/CXC3CL1, BRAK/CXCL14, and CTACK/CCL27), two chemokine receptors (CXCR-6 and CXC3R1), and five cytokines (TGF-β2, TGF-β3, PDGF-β, osteopontin, and PDGF-R). Constitutive expression of TGF-β2, TGF-β3, and PDGF-β in the eye is well known (61, 62), but the expression of chemokines and chemokine receptors in the normal eye has not previously been described. While the role of these chemokines is not known, it is possible that they participate in the physiologic process of immune surveillance, i.e., lymphocyte migration through eye tissues. Previous analyses of adoptively transferred T cells inducing EAE (63, 64) or EAU (65) demonstrated that small numbers of activated lymphocytes migrate into the target organ shortly after injection. This primary wave of infiltrating cells is postulated to release mediators that initiate the massive cell invasion that begins 2–3 days later (64). It is proposed, therefore, that the constitutively expressed chemokines we identified help mediate the poorly defined primary process in the eye, and that similar batteries of chemokines carry out this function in other organs. The need for constitutively expressed chemokines for the invasion of activated lymphocytes was indicated by the recent finding that this process is inhibited by treatment with pertussis toxin, an agent that inhibits chemotaxis (66). In addition to the aforementioned functions, some of the constitutively expressed mediators may play a role in ocular immune privilege. TGF-β2 is well known for its immunosuppressive activity and presumed role in protecting the eye from damaging immune responses (62). Interestingly, fractalkine/CXC3CL1 is constitutively expressed in the brain and has recently been implicated in limiting CNS inflammation (67, 68).

Also of interest are our findings with osteopontin. This molecule, also known as early T cell activation gene-1, exerts both cytokine and chemoattractant activities, and was recently shown to play a role in immunopathogenic processes of cell-mediated diseases (69, 70). Exceedingly high levels of osteopontin mRNA were found in intact eyes, and these levels increased considerably in EAU (Fig. 5) and in Th1-mediated inflammation, but not in Th2-induced disease (Table I and Fig. 3).

A major goal of the present study was to better understand the pathogenesis of ocular inflammation by comparing the inflammatory molecule expression pattern in recipients of Th1 or Th2 cells with that in mice developing EAU. EAU serves as an animal model for ocular inflammatory conditions with presumed autoimmune etiology, such as sympathetic ophthalmia, birdshot retinochoroidopathy, Behcet’s disease, or the Vogt-Koyanagi-Harada syndrome (2, 26, 28). IRBP-induced EAU in the mouse has been thought to be a T cell-mediated disease, with prominent involvement of Th1 cells, based on an association with Th1-specific cytokines (71, 72), as well as on the direct relationship between the susceptibility to the disease and the lymphokine profile of the animal (73). Data collected in this study confirmed these observations and profoundly extended the scope of the investigation by including characterization of an extensive battery of Th1- and Th2-specific cytokines, as well as chemokines and chemokine receptors. The pattern of increased expression of transcripts of the three families of inflammatory molecules in the EAU eyes closely resembled that of Th1 recipient eyes (Table I). The elevated expression of IL-5 mRNA and marginal up-regulation of eotaxin/CCL11 mRNA may indicate a minor involvement of Th2 cells in the pathogenic process.

The data reported in this work underscore the multiplicity of mediators that participate in the ocular inflammatory process. Analysis of these data shows that significant levels of up-regulated transcripts of cytokines and chemokines are first detected on day 4. It is probable, however, that the inflammatory process is triggered earlier by low levels of cytokines released by the small number of HEL-specific T cells that invade the eye immediately following cell injection. In this model, the primary cytokines initiate a cycle in which cytokines, chemokines, and adhesion molecules are up-regulated, leading to additional cell recruitment, leading to additional inflammatory mediator production, and so on. An important implication of this model is that inhibition of sight-damaging ocular inflammation should be targeted at upstream cytokines. This approach has been successful in inflammatory diseases such as Crohn’s disease and rheumatoid arthritis, which can be ameliorated by blocking TNF-α, an upstream cytokine (3, 74, 75, 76). It is of interest, however, that the anti-TNF-α agents effective in these two diseases were considerably less effective in treatment of ocular inflammation (76). Furthermore, an anti-TNF-α Ab was recently found to exacerbate EAU in mice when administered during the efferent phase of the disease (77). Our data point to other upstream cytokines that are potential targets for therapeutic inhibition. For example, both IL-1 and IL-6 were greatly up-regulated in eyes with Th1-induced inflammation and in EAU. Agents that block these two cytokines are already under development for use in humans (3). This study provides an impetus for characterizing the efficacy of these and other agents in combating ocular inflammatory disease.

We are grateful to Dr. Joost J. Oppenheim for critical reading of the manuscript and for his thoughtful comments and suggestions. We also thank Rashid Mahdi, Barbara P. Vistica, and Lang Hung for superb technical assistance; Robert S. Lee for tail DNA analysis; Rick Dreyfuss for digitized microphotography; and the staff of National Eye Institute Histolab for tissue section preparations.