Although there is evidence that altering the Th1/Th2 balance toward Th2 cells may be important in the resolution of Th1-type autoimmune disease, adoptive transfer of Th2 cells is not effective in protecting against Th1-type disease and may cause disease. Therefore, we examined the recruitment of Th1- and Th2-like cells into the retina in the murine autoimmune disease experimental autoimmune uveoretinitis. CD4 T cells were polarized in vitro to IFN-γ-producing Th1-like cells and non-IFN-γ-producing Th2-like cells, labeled, and adoptively transferred. Trafficking to the retina in vivo was evaluated by scanning laser ophthalmoscopy and infiltration by confocal microscopy. There were more rolling and adherent Th1-like cells and they rolled more slowly than did Th2-like cells. Th1-like cells were preferentially recruited into the retinal parenchyma at both initiation and resolution. Surface P-selectin glycoprotein ligand 1 (PSGL-1) and LFA-1 were up-regulated on both populations but were expressed at higher levels on Th1-like cells. Up-regulation of CD44 expression was higher on Th2-like cells. P-selectin, E-selectin, and ICAM-1 are up-regulated on postcapillary venules in the retina. Pretreatment of Th1-like cells with anti-PSGL-1 inhibited rolling and infiltration of Th1-like cells but not Th2-like cells, providing direct in vivo evidence for the inability of Th2 to respond to P/E-selectin despite increased expression of PSGL-1. Anti-LFA-1 pretreatment inhibited infiltration of both Th1- and Th2-like cells, but more so Th-1. We suggest that random trafficking of activated T cells (both Th1 and Th2) across the blood-retina barrier is mediated by CD44:CD44R and LFA-1:ICAM-1, whereas preferential recruitment of Th1 cells is mediated by PSGL-1:P/E-selectin.
CD4 Th cells have been shown to be polarized into functionally distinct subsets that are characterized by the patterns of cytokines they produce, with Th1 cells producing IFN-γ and Th2 cells producing IL-4, IL-5, and IL-10 (1, 2). The action of these subsets leads to different types of immune response, with a Th1-type immune response being phagocyte mediated, as exemplified by the delayed-type hypersensitivity reaction. Th2 cells, in contrast, are associated with allergic reactions via eosinophils and IgE production. Th2 cells are also able to negatively regulate Th1 cell-mediated responses, thus acting in an anti-inflammatory capacity.
It has been suggested that altering the Th1/Th2 balance in vivo toward Th2 function could protect against Th1-type autoimmune disease (3, 4, 5). However, adoptive transfer of Th2 cells failed to protect against Th1-induced autoimmune disease (6, 7, 8). In addition, any protective effect that Th2 cells may have appears to function at the disease induction phase but not the effector phase of Th1-type autoimmune disease (3, 9, 10).
The maintenance of functionally polarized immune responses requires different subsets of lymphocytes to localize to distinct sites of inflammation. It has been shown that adoptively transferred Th2 cells can migrate into target tissue, but only when recipients have been sublethally irradiated (8), are immune compromised (scid mice) (11), or lack αβ T cells (12). Thus, in a Th1-mediated inflammatory reaction in unmodified recipients, the passage of Th2 cells from the circulation into the tissue may be restricted and this restriction may be regulated by αβ T cells. This restriction is likely to be brought about at least in part by adhesion molecule expression. The localization of lymphocytes to tissue during immune responses involves complex interactions between adhesion molecules and chemoattractants on the lymphocytes and the endothelium in a flow situation (13, 14). The regulation of these molecules is likely to determine the subset of T lymphocyte able to cross the endothelium and enter the tissue. Studies in vitro on transendothelial migration in a static situation have indicated that the increased ability of Th1 cells over Th2 to cross the endothelium is due in part to the adhesion molecules LFA-1/ICAM and CD44 (15). In addition, it has been shown that there is a preferential accumulation of P-selectin-positive, α4β7-positive Th1-like IFN-γ-producing cells in the tissues in a murine model of autoimmune gastritis (16).
Experimental autoimmune uveoretinitis (EAU)4 is a Th1-type, organ-specific autoimmune disease induced by immunization with retinal Ags such as retinal soluble Ag (S-Ag) and interphotoreceptor retinoid binding protein (IRBP) in susceptible strains of rats or mice or by adoptive transfer of retinal Ag-specific T cells (17, 18, 19, 20, 21, 22). It serves as an animal model of human endogenous posterior uveitis (17, 19). EAU is also a self-limited autoimmune disease. Resolution of the inflammation during the late stage of disease has been suggested to be due to the Th2 cell response (23, 24), with Th2-type cytokines such as IL-4, TGF-β (25, 26), and IL-10 (5, 27) shown to suppress EAU despite the fact that adoptive transfer of Th2 cells failed to protect against Th1 cell-induced EAU (8).
Previous studies that have investigated the role of Th1 vs Th2 cells in pathogenesis of various inflammatory diseases have suggested that Th1 cells accumulate in the tissues in response to specific adhesion molecule:ligand interactions, particularly involving P- and E-selectin (28, 29). However, evidence for the role of these molecules in these conditions to date has been indirect. In the present study, using in vivo visualization by scanning laser ophthalmoscopy (SLO) of cells trafficking to the retina (30) combined with confocal microscopy of retinal whole mounts, we show directly for the first time preferential recruitment of IFN-γ-producing Th1-like cells over non-IFN-γ-producing Th2-like cells to the retina in the autoimmune model EAU, and also that this preferential recruitment was mediated by P-selectin glycoprotein ligand 1 (PSGL-1):P/E-selectin and LFA-1:ICAM-1 interactions. In particular, Th1-like cells show reduced rolling velocity and increased rolling behavior and adherence to the endothelium compared with Th2-like cells, indicating a definitive role for mediators of rolling/adhesion in initiating recruitment of Th1 cells to sites of inflammation.
Materials and Methods
Inbred female B10.RIII mice 8–12 wk old and 18–24 g were obtained from the animal facility at the Medical School of Aberdeen University. All animals were managed in accordance with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research and under the regulations of the United Kingdom Animal License Act 1986 (U.K.).
Induction of EAU
EAU was induced in B10.RIII mice as described previously (22, 31). Briefly, mice were immunized s.c. with 50 μg of human IRBP peptide 161-180 (SGIPYIISYLHPGNTILHVD; purity >85%; Sigma-Aldrich, Cambridge, U.K.) emulsified with 50 μl of CFA, H37Ra (Difco Laboratories, Detroit, MI) in a total volume of 100 μl. Naive mice were untreated. Immunization with the same volume of PBS, instead of IRBP peptide, in CFA, H37Ra had no detectable effect clinically or histologically in the retina, and trafficking, in these mice or using cells from these mice, did not differ from that in which animals were untreated (data not shown).
Animals were observed using an ophthalmic operating microscope and slit-lamp for clinical evaluation of the ocular fundus and the anterior segment of the eye, respectively, daily from day 7 to day 21 postimmunization (pi). Using this method, we have shown previously that clinical signs of disease appear from day 9 or 10 pi, and that active disease lasts for 2–3 wk (32, 33). Animals at day 9 pi (Ag-primed, disease initiation stage) and day 18 pi (disease recovery stage) were chosen for study of in vivo cell trafficking. Histological examination has shown that inflammatory cell infiltration and perivasculitis occur at day 9 pi, with infiltrating cells distributed in the ganglion layer and vitreous. Most of the photoreceptor layer remains normal at this time, but there is some focal photoreceptor cell damage (34). No rolling of Ag-primed leukocytes is seen before day 9 pi EAU (34).
Detection of Th1/Th2 cytokine mRNA expression in the retina
Retinas were dissected from the eyes of both naive animals and those immunized with IRBP peptide after perfusion with 30 ml of PBS containing 10 U/ml heparin under terminal anesthesia. Retinas from the same animal were pooled and RNA was isolated using RNA-Bee (Biogenesis, Poole, Dorset, U.K.) according to the manufacturer’s protocol.
Poly(A)+ RNA from 5 μg of total RNA was reverse transcribed with 200 U of Moloney murine leukemia virus reverse transcriptase (Promega, Southampton, U.K.). One microliter of this cDNA was used in the PCR. Each PCR was conducted in a total volume of 25 μl containing 12.5 μl of master mix (Promega) and 2.5 μl of primer mix (10 μM). GAPDH (35), IFN-γ, and IL-4 (36) primers were obtained from ThermoHybaid (Ulm, Germany). Primers for IFN-γ and IL-4 were intron-spanning to allow discrimination of any genomic DNA.
Thirty-three cycles of amplification were performed, with each cycle consisting of a denaturation step at 94°C for 50 s, annealing at 55°C for 1 min, and polymerization at 72°C for 1 min 30 s. In the first cycle denaturation was conducted for 2 min, and in the final cycle polymerization was for 5 min. After amplification, samples were run on a 1.8% agarose gel (molecular biology grade; Promega) in TBE (0.045 M Tris-borate, 0.001 M EDTA) containing 0.4 μg/ml ethidium bromide. Relative abundance of product was assessed by calculating the ratios of the cytokine band to the GAPDH band for each sample using GeneGenius software (Syngene, Cambridge, U.K.).
Cell preparation and culture
Naive CD4+ T lymphocytes were isolated from lymph node cells using MACS positive selection protocol (Miltenyi Biotec, Bergisch Gladbach, Germany). A total of 92–95% of these cells were CD4+ when tested with anti-CD4 mAb (BD Biosciences, Oxford, U.K.). A modified version of the method described by Siveke and Hamann (37) was used for the establishment of polarized T cells. CD4+ cells were treated for 2 days in six-well plates precoated with 2 μg/ml anti-mouse CD3e (clone 145-2C11; BD Biosciences) in RPMI 1640 plus 10% FCS supplemented with anti-mouse CD28 (2 μg/ml, 37.51 clone; BD Bioscience), IL-12 (5 ng/ml; BD Biosciences), IFN-γ (20 ng/ml; BD Biosciences), and anti-IL-4 (1 μg/ml, 11B11 clone; BD Biosciences) for generation of IFN-γ-producing, Th1-like cells or with anti-mouse CD28 (2 μg/ml, 37.51 clone; BD Biosciences), IL-2 (5 ng/ml; BD Biosciences), IL-4 (10 ng/ml), and anti-IFN-γ (2 μg/ml, clone XMG1.2; BD Biosciences) for non-IFN-γ-producing, Th2-like cells. The cells were then transferred onto uncoated plates without a change of medium and were cultured for an additional 4–6 days to allow the cells to return to a resting state (37). The resulting effector cells express levels of L-selectin and CD45RB comparable with those of naive CD4 cells and not memory cells, with low CD25 expression (data not shown) (37) indicating a resting state and increased levels of CD44 showing that they are activated cells. The return to a resting state is important because resting cells have been shown to actively traffic to a much greater extent than fully activated cells (29, 37). These cells are thought to resemble mature effector T cells in vivo, which re-enter the circulation after a sessile, proliferative phase (29).
Intracellular staining of T cell polarized subsets was performed after restimulation on a plate precoated with 2 μg/ml anti-CD3 and anti-CD28 overnight. Stained cells were analyzed using a FACSCalibur flow cytometer (BD Biosciences). In the generated Th1-like cell population, 30–60% of the cells produced high amounts of IFN-γ and no IL-4 (Fig. 1). In the Th2-like cell culture, a significant fraction of cells produced high levels of IL-4 but not IFN-γ (Fig. 1). Although polarized populations of T cells prepared using this method are generally referred to as Th1 and Th2 cells, it is not possible to prove that they contain equal numbers of Th1 and Th2 cells. It has been shown that it is not possible to stimulate all of the cells in these types of primary culture (37), and the percentage of cells seen to produce cytokine in this analysis will be a result of various factors, including the method of restimulation, the transient nature of cytokine production, and the presence of some uncommitted cells (37, 38). Studies with T cell clones have confirmed that IL-4-producing cells are detected at lower frequency in Th2 clones than are IFN-γ producing cells in a Th1 clone due to the transient nature of IL-4 synthesis (38). As such, we have referred to these populations as IFN-γ producing and non-IFN-γ producing.
Analysis of cell surface molecules
Cell surface expression of adhesion molecules on in vitro polarized T cells was detected by flow cytometric analysis. T cells were stained with specific Abs to mouse CD62L (L-selectin, rat IgG2a), CD45RB (rat IgG2a), CD25 (rat IgG1), CD44 (rat IgG2b), CD11a (LFA-1, rat IgG2a), CD162 (PSGL-1, rat IgG1), and their isotype controls (all were purchased from BD Biosciences) in 1% (w/v) BSA/PBS at 4°C for 20 min. Flow cytometry was performed on a FACSCalibur and was analyzed using CellQuest software (BD Biosciences).
Adhesion molecule expression (P, E-selectin, ICAM-1, VCAM-1, and platelet endothelial cell adhesion molecule-1 (PECAM-1)) on retinal vascular endothelial cells was detected using an in vivo intravascular staining protocol as described previously (33). Briefly, 80 μl (20 μg) of FITC-conjugated specific Ab to mouse VCAM-1, ICAM-1, P-selectin, E-selectin, and PECAM-1 (all purchased from BD Biosciences) and their relevant isotype controls were injected via the tail vein and allowed to bind for 15 min and were followed by an injection of 100 μl of 2% (w/v) Evans blue (Sigma-Aldrich). Evans blue is an acid dye of the diazo group that binds to albumin in the blood. The animals were then killed, and retinal whole mounts were prepared as described elsewhere (33, 39). In brief, the anterior segment of the globe was removed and the retina was peeled from the choroid. Retinas were washed twice in PBS for 15 min and then were spread on clean glass slides and mounted vitreous side up under coverslips with Vectashield (Vector Laboratories, Burlingame, CA). Samples were observed using a confocal scanning laser imaging system (Zeiss LSM510; Carl Zeiss, Gotingen, Germany). Evans blue appeared red and FITC green.
Expression of adhesion molecule CD44 and its ligand hyaluronic acid (HA) on retinal vascular endothelial cells was performed in vitro in whole mounts of retina. The whole retina was dissected from the mouse eye according to the method of Chan-Ling (39). Unfixed retinal tissues were treated with 6% BSA for 1 h and then were incubated with biotinylated HA binding protein (bHABP; 2 μg/ml; Seikagaku, Tokyo, Japan) in 2% BSA and 0.5% Triton X-100 at 4°C overnight. After washing extensively with TBS, sections were incubated with PE-conjugated streptavidin (Caltag Laboratories, Burlingame, CA) and FITC-conjugated anti-mouse CD44 (IM7; BD Bioscience) for 1 h at room temperature. Specificity of bHABP for binding to HA was determined by evaluating the degree of bHABP binding to tissues predigested with hyaluronidase before applying the bHABP probe. Samples were incubated with Streptomyces hyaluronidase (Seikagaku; 100 turbidity reducing units/ml sodium acetate buffer (pH 6)) at 37°C for 3 h, and then they were washed thoroughly with TBS. Retinal whole mounts were analyzed by confocal laser microscopy (Zeiss LSM510).
In vivo leukocyte trafficking assay
A total of 2 × 107 in vitro generated polarized cells in 10 ml were incubated with 40 μg/ml calcein-acetoxymethyl ester (C-AM) (Molecular Probes, Leiden, The Netherlands) at 37°C for 30 min. C-AM (green) is nontoxic and has no effect on cell adhesion (40). For blocking experiments, cells were incubated with 5 μg/ml anti-mouse CD44 (IM7; BD Biosciences) or anti-mouse PSGL-1 (2PH1; BD Biosciences) or with 10 μg/ml anti-LFA-1 (M17/4; BD Biosciences) or rat IgG control for 30 min at 37°C before C-AM labeling. Previous studies have shown that IM7 can block the binding of HA to cell surface CD44 (41) and can down-regulate leukocyte cell surface CD44 expression (42, 43), whereas 2PH1 can block the binding of mouse leukocytes to P-selectin (42, 44). M17/4 has been shown to be able to block LFA-1/ICAM-1 interaction (45).
Scanning laser ophthalmoscopy.
The technique of SLO in mice has been described in detail elsewhere (30, 46). In brief, mice were anesthetized with an i.m. injection of 0.4 ml/kg Hypnorm (0.315 mg of fentanyl citrate and 10 mg of fluanisone/ml; Janssen-Cilag, High Wycombe, U.K.) and i.p. with 1 ml/kg Diazepam (Phoenix Pharmaceuticals, Gloucester, U.K.). Fifty microliters of 0.05% (v/v) sodium fluorescein (Sigma-Aldrich) was injected via the tail vein to outline the vessels, followed by 1 × 107 fluorescently labeled polarized cells, either untreated or Ab treated, in 150 μl of complete medium. SLO images were recorded simultaneously on videotape (S-VHS) and digitally at 25 frames per second as described previously (46). For each eye, three regions of interest containing one to three veins/venules were recorded for at least 30 min.
Video analysis was conducted offline as described elsewhere (30). Rolling leukocytes and those not interacting with the endothelium were counted in each venule. Rolling cells were defined as those cells with a velocity below the critical velocity as calculated by the following equation (47, 48).
and Ε is the ratio of the leukocyte diameter to vessel diameter. The rolling efficiency was calculated as the percentage of labeled rolling cells among the total number of labeled leukocytes that entered a venule. The sticking efficiency was determined as the percentage of labeled leukocytes becoming firmly adherent for at least 20 s, compared with the total number of labeled leukocytes that rolled in a venule during the same time interval. Rolling velocities of 40 randomly chosen, rolling, labeled cells in retinal venules were measured in digital images (46).
A total of 1 × 107 (150 μl) C-AM-labeled, Ab-treated or untreated, polarized cells were injected via the tail vein into normal or primed day 9 and day 18 pi EAU mice. At 1, 4, and 16 h after cell infusion, 100 μl of 2% Evans blue (Sigma-Aldrich) was injected via the tail vein and allowed to bind for 5–10 min. The eyes were then harvested and fixed in 2% (w/v) paraformaldehyde (Agar Scientific, Cambridge, U.K.) for 1 h. Retinal whole mounts were prepared as described elsewhere (33, 39).
Analysis of CD44 expression and HA binding
In vitro-generated, polarized cells were incubated with 5 μg/ml high-m.w. HA (Seikagaku) at 37°C for 30 min. After washing with PBS, cells were incubated with 2% BSA/PBS to block nonspecific binding followed by bHABP at room temperature for 30 min. Samples were then double stained with PE-conjugated streptavidin (Caltag Laboratories) and FITC-conjugated anti-mouse CD44 (IM7; BD Biosciences) and were analyzed by flow cytometry as described above.
Endothelial cell adhesion molecule fluorescence intensity was quantified using image analysis computer software (QWin System; Leica, Cambridge, U.K.). For each vessel analyzed, the fluorescence intensity for a region of the parenchyma (without any vessels or artifacts) was measured and subtracted from the fluorescence intensities measured inside the lumen of the vessels. Vessels were considered as independent variables. The difference between control and IRBP peptide-immunized mice at each time point was compared using Dunnett’s multiple comparison test. The mean values for infiltrating cells were compared using Tukey’s multiple comparison test. Differences in the numbers of rolling and sticking cells between Ab-treated and control groups were compared using the χ2 test. Differences with p < 0.05 were considered statistically significant.
Trafficking of polarized cells in mouse retina
EAU is described as a Th1-type autoimmune disease (20). To investigate whether there is a selective recruitment of T lymphocyte subsets during EAU development, in vitro-generated, non-Ag-specific, IFN-γ-producing and non-IFN-γ-producing polarized cell populations were adoptively transferred into primed day 9 pi (disease initiation stage) and day 18 pi (disease recovery stage) EAU.
When labeled, polarized cells were adoptively transferred into a primed mouse at disease initiation stage. As many as 52.6 ± 5.7% cells from the IFN-γ-producing population rolled on retinal venules, whereas only 14.2 ± 0.8% of the non-IFN-γ-producing population rolled. Non-IFN-γ-producing cell rolling, however, was still statistically higher than that of naive CD4 cells (p < 0.05; Fig. 2,A). At the disease recovery stage in which the inflammation had begun to diminish, again more labeled cells from the IFN-γ-producing population rolled than from the non-IFN-γ-producing population (Fig. 2 A). There was no significant difference between naive CD4 cells and the non-IFN-γ-producing population in rolling cells at this time point.
When the rolling velocities were compared in primed mice (day 9 pi), cells from the IFN-γ-producing population rolled at a much lower speed compared with naive cells and were also slower than the non-IFN-γ-producing population (Fig. 2,B). The rolling velocities of cells from the non-IFN-γ-producing population, in contrast, were significantly lower than that of naive CD4 cells (Fig. 2 B). However, these effects were only observed when the polarized cells were transferred into IRBP peptide-immunized mice. When either population of labeled polarized cells was adoptively transferred into normal nonimmunized mice, no rolling cells were observed (data not shown). Similarly naive splenocytes have higher rolling velocities than do IRBP peptide in vivo-primed splenocytes (34).
Sticking is the second step of leukocyte migration (13, 14), resulting from further activation (i.e., via chemokine-chemokine receptor interaction) during the rolling process (14, 49). When cells roll more slowly, they have more time to be further activated by chemokines secreted or presented by the endothelium and to become firmly attached (50). We have previously shown that naive cells do not adhere to normal retinal vessels and that only very few activated T cells (5–10 cells/retina/30 min) adhere to normal retinal vessels (51). In the present study, our data for cell sticking efficiency (sticking cells related to total rolling cells) show that significantly more labeled cells from the IFN-γ-producing population adhered to activated endothelium than did either the naive or non-IFN-γ-producing populations (Fig. 2,C). There was no significant difference between naive and cells of the non-IFN-γ-producing population in sticking efficiency (Fig. 2 C). All rolling and sticking cells were observed exclusively in retinal venules and postcapillary venules, as previously reported (30).
When infiltrating cells were counted in retinal whole mounts in primed mice (day 9 pi) as early as 1 h after cell infusion, many labeled cells from the IFN-γ-producing population were detected in the retinal parenchyma (Fig. 3, A and D2), and this number of infiltrating cells increased dramatically from 4 to 16 h after cell infusion (Fig. 3, A and D3). Labeled cells of the non-IFN-γ-producing population were also detected infiltrating the retina in these primed mice; however, the number was significantly lower (p < 0.05) than that for the IFN-γ-producing cell population at each time point (Fig. 3, A and D4). A small number of naive CD4 cells was detected in primed (day 9 pi) retinal parenchyma 4 h after cell infusion, and this was increased by 16 h after cell infusion (Fig. 3,A) but remained less than for cells of the non-IFN-γ-producing population. When labeled polarized cells were adoptively transferred into primed mice at the disease recovery stage (day 18 pi), although less infiltrating cells were detected compared with those in day 9 pi recipient mice at each time point, again more cells from the IFN-γ-producing population had infiltrated than from the naive and non-IFN-γ-producing cell populations (Fig. 3,B). In naive recipient mice, no infiltration of labeled polarized cells was observed at 1 and 4 h after cell infusion (Fig. 3, C and D1). Sixteen hours after cell injection, a few cells were detected in the retinal parenchyma and there was no significant difference between the two populations of polarized cells in the number of retinal infiltrating cells (Fig. 3 C). No cell infiltration was observed when naive CD4 cells were adoptively transferred into naive mice.
Cytokine mRNA in retina
To confirm Th1 and Th2 cell entry into the retina during the normal course of EAU (i.e., without tail vein injection of polarized cells), the presence or absence of IFN-γ and IL-4 mRNA in perfused retinas was determined by routine RT-PCR. Neither IFN-γ nor IL-4 mRNA could be detected in control animals injected with PBS, which were not diseased (Fig. 4). However, both IFN-γ and IL-4 were present in retinas from mice with EAU at day 10, but at day 18 pi this was reduced such that IL-4 was undetectable (Fig. 4).
Adhesion molecule expression
Because adhesion molecules are of critical importance in the regulation of leukocyte transmigration across the endothelial cells of the blood vessels, we have examined the adhesion molecule expression on retinal vascular endothelial cells and on the two populations of in vitro polarized cells.
Retinal vascular endothelial cells.
Previously we have shown that CD44/HA is involved in leukocyte trafficking in EAU (52). In the current experiment, we were able to show that both CD44 and HA were positively stained in retinal vessels (Fig. 5), with the expression of CD44 under the staining layer of HA (Fig. 5,B). This supports the notion that HA anchors to vascular endothelial cells through the binding of CD44 (53). There was a slight but statistically significant up-regulation of CD44 in retinal vessels in primed mice at the disease initiation stage (day 9 pi; Fig. 5,A), whereas there was no significant increment in HA expression compared with that in the retinal vessels of naive mice (Fig. 5,A). We have previously shown that, although VCAM-1, PECAM-1, ICAM-1, P-selectin, and E-selectin are significantly up-regulated on the retinal vasculature in these primed mice (day 9 pi), only ICAM-1, P-selectin, and E-selectin are expressed and up-regulated preferentially in retinal venules, the main site of leukocyte extravasation (33). Therefore, ICAM-1, P-selectin, and E-selectin expression were compared in mice at the disease initiation and recovery stages (day 9 and day 18 pi) and were shown to be significantly lower at the disease recovery stage (Fig. 5 A).
CD44, PSGL-1, and LFA-1 were all significantly up-regulated on both populations of polarized cells compared with naive CD4 cells. Up-regulation of PSGL-1 and LFA-1 was significantly greater in the IFN-γ-producing population compared with the non-IFN-γ-producing cell population (Fig. 6).
P-selectin/PSGL-1 and LFA-1/ICAM-1 in the trafficking of polarized cells at the blood-retina barrier (BRB)
Cell rolling is believed to be mediated by the selectin family (14, 54, 55). As we have shown that P- and E-selectin and ICAM-1 are up-regulated on retinal venules in EAU and that PSGL-1 (P/E-selectin ligand) and LFA-1 are up-regulated on the polarized cells, we have examined the involvement of these ligand/receptor pairs in Th cell infiltration of the retina in more detail. Pretreatment of cells with anti-PSGL-1 Ab in vitro significantly suppressed rolling of the IFN-γ-producing cells but not the non-IFN-γ-producing cell population in retinal venules of primed mice (day 9 pi; Fig. 7,A). Cell infiltration in these primed mice, 16 h after adoptive transfer, was also significantly reduced by anti-mouse PSGL-1 Ab pretreatment of the IFN-γ-producing cell population, but not the non-IFN-γ-producing population (Fig. 7,C). In contrast, blocking LFA-1 with anti-LFA-1 Ab did not reduce rolling efficiency (Fig. 7,B), and although it resulted in a dramatic reduction (92%) in the number of IFN-γ-producing cells infiltrating the retina, it also produced a statistically significant reduction (64%) in the number of non-IFN-γ-producing cells infiltrating (Fig. 7 D).
CD44/HA in the trafficking of polarized cells at the BRB
CD44/HA has been reported as a ligand-receptor pair important for adhesion and in mediating the rolling of activated leukocytes (56, 57, 58). Previously we have shown that CD44/HA is involved in the leukocyte trafficking in EAU (52). Whether it is involved in the rolling of Th cells is not known. The results of the in vitro HA binding assay show that both the IFN-γ-producing and nonproducing cell populations are able to bind HA (Fig. 8). The binding ability of the IFN-γ-producing cell population is significantly higher than that of the non-IFN-γ-producing population (p < 0.05; Fig. 8). Pretreatment of labeled polarized cells with anti-mouse CD44 Ab IM7 significantly suppressed the rolling of both cell populations (Fig. 9,A). Subsequent transmigration into the retina of both cell populations was also suppressed by anti-CD44 Ab treatment (Fig. 9 B).
In the present study, we have demonstrated using an animal model of Th1-type, self-limited, organ-specific autoimmune disease, EAU, that cells from the IFN-γ-producing, Th1-like population are preferentially recruited into the inflamed retina at both the disease initiation stage and disease resolving stage. There was no selective recruitment of cells from the non-IFN-γ-producing, Th2-like population into the retina even during the clinical disease recovery stage, suggesting that this selectivity applies throughout the disease process.
Polarizing CD4 cells in vitro to IFN-γ-producing and non-IFN-γ-producing Th1- and Th2-like populations, using specific cytokines and then allowing them to return to a resting state, provided physiologically relevant populations of cells capable of efficient trafficking (29, 37), which was confirmed by flow cytometric analysis of IFN-γ and IL-4 production. Our data showing the expression of IFN-γ and IL-4 mRNA in perfused retinas from day 10 pi, but not in naive mice, confirm the results obtained with these in vitro polarized T cells: that both Th1 and Th2 cells are able to traffic to the retina at the disease initiation stage, but that both are reduced at the disease recovery stage (day 18 pi) and there is no selective recruitment of Th2 cells at this stage.
Evidence has suggested that disease recovery in EAU may be due to a shift in the immune response toward Th2 (23, 24, 27, 59). In general, Th2 cells can antagonize the Th1 immune response either by blocking the generation of Th1 cells in the lymph node or by blocking the effector functions of Th1 cells (60, 61, 62). In Th1-type, organ-specific autoimmune disease, Th1 effector cells are selectively recruited into the inflamed organs. To block the effector functions of these Th1 cells, Th2 cells need to be able to infiltrate into the inflammatory tissue rich in Th1 effector cells. Our study shows that, in EAU, Th2 cells are not selectively recruited into the inflamed retina and only infiltrate in low numbers even during the disease recovery stage. The lesser ability of Th2 cells to be recruited into an inflamed site may contribute to their weaker pathogenicity in addition to differences in the mediators they produce (6, 8, 12, 63).
Central to disease recovery may be the significantly reduced trafficking of the IFN-γ-producing cell population to, and accumulation in, the retina during the late stage of EAU, compared with that during the disease initiation stage. Studies on genetic susceptibility to EAU have also suggested that resistance is associated more with an inhibited Th1 response than an elevated Th2 response (20). During the spontaneous remission of experimental allergic encephalomyelitis, although Th1 cytokines decline after clinical recovery, this is not associated with Th2 cytokine release by infiltrating cells (64).
Both the selectivity of IFN-γ-producing, Th1-like cell recruitment to the retina over Th2-like cell recruitment and the reduction of Th1-like cell recruitment as disease resolves are likely to be regulated at least in part by adhesion molecule expression. Our experiments have been conducted with polarized T cell populations that have not been stimulated with Ag and are from naive mice that have not been stimulated with Ag. In addition, the early selective infiltration of cells of the IFN-γ-producing population (1 h; see Fig. 3 A) occurred at a time when the likelihood of encountering Ag and developing antigenic specificity in the draining lymph node would be low because studies have shown that several hours are required for Ag-specific T cell activation (65). This shows that the selective accumulation of IFN-γ-producing cells in the retina is not Ag-related but is due to changes in the microenvironment at the BRB and within the retina during EAU. Therefore, we examined adhesion molecule expression in this system to determine those adhesion molecules most influential in this regulation.
Previous studies have shown that homing of Th1 cells into inflamed sites was attributable to the interaction of P/E-selectin and their ligand PSGL-1 (29, 66). Although both Th1 and Th2 cells express high levels of PSGL-1 (P/E-selectin ligand), only Th1 cells were able to bind P/E-selectin (66, 67) due to their differential expression of α3 fucosyltransferases (67, 68), a family of enzymes that modify carbohydrate moieties decorating PSGL-1 and other surface receptors (69). Our study showed that blocking of cell surface PSGL-1 with specific Ab significantly suppressed the rolling and final transmigration of the IFN-γ-producing, Th1-like cell population but not the Th2-like cell population. Thus, we provide the first direct in vivo evidence supporting a critical role for PSGL-1:P/E-selectin in Th1-like cell recruitment in an unmanipulated model of leukocyte-endothelial cell rolling behavior and transendothelial migration into inflamed tissues (29, 66). The up-regulation of P/E-selectin expression on retinal endothelial cells at disease initiation in EAU and its reduction as disease resolves therefore appear to be key factors in regulating the entry of specific T cell subsets into the retinal tissue.
However, what controls the adhesion molecule expression and particularly P/E-selectin in retinal vessels during EAU is not known. Previous in vitro studies have shown that the adhesion molecules ICAM-1, PECAM-1, and P/E-selectin can be up-regulated by Th1-type cytokines IL-1β, TNF-α, and IFN-γ and down-regulated by Th2 cytokines IL-4, IL-10, and IL-6 (70). In the present study, it is therefore possible that during the disease initiation stage of EAU, P/E-selectin and ICAM-1 are up-regulated by Th1-type cytokines (71) and during the late stage of EAU these adhesion molecules are down-regulated by increased Th2-type cytokines (23, 24, 71). We have shown that up-regulation of adhesion molecules on retinal vessels can be induced simply by the presence of intravascular, circulating, activated T cells and that as few as 1 × 105 activated T cells are sufficient to achieve this effect (51). Even the limited adherence of Th2 cells to vessels seen during EAU in this study may be sufficient to alter adhesion molecule expression. Thus, Th2 cells, via their cytokine production, may be able to suppress the Th1-type immune response in EAU, but this may be at the endothelial cell surface by altering adhesion molecule expression rather than within the retina.
The demonstration that adoptive transfer of Th2 cells only causes EAE when αβ T cells are not present in the recipient (12) may be a reflection of the ability of certain αβ T cells to maintain the selectivity of the endothelium in favor of Th1 cells via cytokine regulation of adhesion molecule expression.
In addition to PSGL-1, certain integrins likely to be important for passage across the endothelium have also been reported to be differentially expressed on Th1 and Th2 cells (72, 73). We have shown αLβ2 (LFA-1), a receptor for ICAM-1, at higher levels on IFN-γ-producing, Th1-like cells. Ab to LFA-1 did not reduce the rolling efficiency of either Th1- or Th2-type cells, although it did inhibit extravasation. Although both cell types were inhibited, this inhibition was greater for the Th1-like cells. This suggests that LFA-1 is not important for rolling of cells but is critical for transendothelial migration. This is consistent with a recent report in humans showing that LFA-1 is a ligand for junctional adhesion molecule-1 (JAM-1) and is involved in leukocyte transendothelial migration (74). That Th2-type cells were inhibited by anti-LFA-1 to a lesser extent may indicate a difference in the degree to which Th1 and Th2 cells use JAM-1. Our data show that CD44-HA interaction contributes to the rolling and infiltration of both Th1 and Th2 cells, suggesting that random trafficking of activated Th1- and Th2-like cells is mediated by CD44:CD44R interactions.
The differential expression of adhesion molecules on the polarized T cells and corresponding ligands on the endothelium will act in concert with the local expression of chemokines and chemokine receptors (75, 76, 77) to allow selective trafficking of Th1 and Th2 cells. This interplay at the BRB is the subject of an ongoing study. However, studies in vitro on transendothelial migration in a static situation have indicated that the increased ability of Th1 cells over Th2 to cross the endothelium is due more to changes to adhesion molecules than to chemoattractants and is reflected in differential regulation of signaling mechanisms independently controlling transendothelial migration of Th1 cells vs Th2 cells (15).
In conclusion, we show that in a model of organ-specific autoimmune disease, EAU, selective recruitment of IFN-γ-producing, Th1-like cells takes place, with resolution of the inflammation due primarily to the reduced recruitment of these cells into the inflamed retina. There is no preferential recruitment of Th2 cells even during the disease recovery stage of EAU. The selective recruitment of Th1-like cells is brought about by P/E-selectin up-regulation on the retinal endothelium and the differential expression of PSGL-1 on the polarized cells. PSGL-1:P/E-selectin regulation of T cell recruitment appears to be critical because Th2 cells are unable to respond to P/E-selectin despite up-regulating PSGL-1. In contrast, random recruitment of all activated T cells (both Th1 and Th2) appears to be mediated by CD44:CD44R and LFA-1:ICAM-1 interactions. The permissive vascular environment for Th1 cell recruitment is likely to be regulated by the Th subsets themselves via cytokine production.
We thank Carol Wallace and Graeme Lamont for technical assistance.
This work was supported by the Wellcome Trust (Grant 057311).
Abbreviations used in this paper: EAU, experimental autoimmune uveoretinitis; IRBP, interphotoreceptor retinoid binding protein; SLO, scanning laser ophthalmoscopy; PSGL-1, P-selectin glycoprotein ligand 1; pi, postimmunization; PECAM-1, platelet endothelial cell adhesion molecule-1; C-AM, calcein-acetoxymethyl ester; HA, hyaluronic acid; bHABP, biotinylated HA binding protein; BRB, blood-retina barrier.