Leukocyte Trafficking in Experimental Autoimmune Uveitis: Breakdown of Blood–Retinal Barrier and Upregulation of Cellular Adhesion Molecules

Abstract

purpose. To clarify the order of events occurring in the breakdown of the blood–retinal barrier (BRB) in experimental autoimmune uveoretinitis (EAU) and in particular to study the relationships between increased vascular permeability, upregulation of endothelial cell adhesion molecules, and leukocyte adhesion and infiltration during EAU.

methods. B10.RIII mice were immunized with human interphotoreceptor retinoid binding protein (IRBP) peptide 161–180. Changes in the retinal microvasculature were examined on days 3, 6, 7, 8, 9, 10, 16, and 21 postimmunization (pi). Evans blue dye was administered intravenously to assess vascular permeability. Expression of intercellular adhesion molecule (ICAM)-1, vascular cell adhesion molecule (VCAM)-1, P-selectin, E-selectin, and platelet endothelial cell adhesion molecule (PECAM)-1 was evaluated by in vivo administration of antibody and subsequent immunostaining of retinal wholemounts. Lymphocytes from inguinal lymph nodes of normal and chicken ovalbumin (OVA)- or IRBP peptide–immunized mice at day 5, 6, 7, 8, and 15 pi were labeled in vitro with calcein-AM (C-AM) and infused intravenously into syngeneic recipient mice, which had been immunized with peptide at the same corresponding time point. Wholemount preparations of retinas were observed 24 hours later by confocal microscopy to determine the adhesion and infiltration of lymphocytes.

results. The first observation of an increase in vascular permeability occurred at day 7 pi and was restricted to focal areas of the retinal postcapillary venules of the inner vascular plexus. This progressively extended to the outer vascular plexus at day 9 pi. Specific adhesion of leukocytes to the endothelium of retinal venules of the inner vascular plexus was first observed at day 6 pi. Leukocyte extravasation into the retinal parenchyma from these vessels began at day 8 pi and extended to the outer vascular plexus at day 9 pi. The expression of adhesion molecules increased progressively during the development of EAU. In particular, the adhesion molecules ICAM-1, P-selectin, and E-selectin were expressed predominately in retinal venules, the sites of BRB breakdown, cell adhesion, and extravasation, from day 7 pi. The increases in expression of ICAM-1 and P-selectin were associated both spatially and temporally with breakdown of the BRB, cell adhesion, and extravasation. No increase in expression of P-selectin and ICAM-1 was observed in either the mesenteric vessels of EAU mice or the retinal vessels of OVA-immunized mice.

conclusions. The sequence of events in EAU appears to be focal adhesion of leukocytes to discrete sites on postcapillary venules, followed by upregulation of adhesion molecules, especially ICAM-1 and P-selectin, and breakdown of the BRB, leading to transendothelial migration of leukocytes and recruitment of large numbers of cells to the retinal parenchyma. These changes occur over a short period of 6 to 9 days pi and initiate the process of tissue damage during the following 2 to 3 weeks.

Experimental autoimmune uveoretinitis (EAU) is a T-cell–mediated autoimmune disease and serves as an animal model of human endogenous posterior uveitis (EPU). ¹ Breakdown of the blood–retinal barrier (BRB) and infiltration of inflammatory cells into the retina are fundamental to the development of EAU. 

The BRB is located at two sites, the retinal pigment epithelium (RPE) and the retinal vascular endothelium, which form the posterior and anterior barrier, respectively. Under normal conditions, this barrier restricts the entry of molecules and cells into the neuroretina, but during ocular inflammation, lymphocytes cross the BRB and enter the retina in large numbers. ^{2 3} At present, whether BRB breakdown is necessary before lymphocytes can infiltrate or whether lymphocyte infiltration results in BRB breakdown during EAU remains unresolved. Lightman and Greenwood ⁴ suggested that breakdown of the BRB in ocular inflammation was a direct consequence of lymphocytic infiltration. However, in another study, Luna et al. ⁵ found that breakdown of the BRB occurs before cell infiltration. 

In general, lymphocyte migration into sites of inflammation depends on the interaction between molecules expressed on the surface of the vascular endothelium and the leukocyte. The process starts with selectin-mediated rolling of leukocytes on the endothelium, followed by integrin and platelet endothelial cell adhesion molecule (PECAM)-1–mediated adhesion and transendothelial migration. ^{6 7 8 9} Matrix metalloproteinases (MMPs) are also involved in transmigration of leukocytes at the site of inflammation. ¹⁰ Within the retina, the vascular endothelial cell is in direct contact with circulating lymphocytes, and interactions between these cells can directly control leukocyte extravasation. However, little is known about the molecular process of leukocyte recruitment at the BRB during EAU. 

The purpose of this study was to determine the relationship between changes in vascular permeability, endothelial cell expression of cellular adhesion molecules, and leukocyte adhesion and infiltration and the role that these changes play in the breakdown of the anterior BRB (i.e., the retinal vasculature) in EAU. We performed the investigation on wholemount preparations of the retina ¹¹ and used confocal microscopy, which has the unique advantage of allowing access to and direct comparison of the different regions of the retinal vasculature. BRB breakdown was defined by the leakage of blood albumin from vessels, as detected by Evans blue. 

Materials and Methods

Animals

Female B10.RIII mice from the animal facility at the Medical School of Aberdeen University (8 to 12 weeks old, weight ∼20 g) were used in this study. The animals were cared for in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and under the regulations of the UK Animal License Act 1986 (UK). 

Induction of EAU

EAU was induced in B10.RIII mice, as described previously. ¹² Briefly, mice were immunized subcutaneously in the inguinal region with 50 μg IRBP peptide 161–180, (SGIPYIISYLHPGNTILHVD; purity >85%; Sigma, Cambridge, UK) emulsified with 50 μL Freund’s complete adjuvant (CFA, H37Ra; Difco Laboratories, Detroit, MI) in a total volume of 100 μL. Control mice were immunized with the same volume of phosphate-buffered saline (PBS), instead of IRBP peptide, in CFA. In addition, 50 μg chicken ovalbumin (OVA; Sigma) was used as the non–retinal antigen-specific immunized control. 

Identification of Microvascular Changes

To evaluate microvascular permeability in the retina during the development of EAU, 100 μL of 2% (wt/vol) Evans blue dye (Sigma) was injected into normal nonimmunized B10.RIII mice (n = 6) and day-5, -7, -8, -9, -12, -16, and -21 postimmunization (pi) mice (n = 5 at each time point) through the tail vein. Evans blue is an acid dye of the diazo group that binds to albumin in the blood, allowing sites of BRB breakdown to be detected readily. Animals were killed by inhalation of CO₂ 10 minutes later. The eyes were removed and were immediately immersed in 2% (wt/vol) paraformaldehyde (Agar Scientific Ltd., Cambridge, UK) for 1 hour. Retinal wholemounts were prepared according to the method of Chan-Ling. ¹³ 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 minutes and then spread on clean glass slides and mounted vitreous side up under coverslips with antifade medium (Vectashield; Vector Laboratories, Burlingame, CA). 

Evaluation of Leukocyte Adhesion and Infiltration

To evaluate leukocyte adhesion and infiltration during EAU, leukocytes from inguinal lymph nodes of normal nonimmunized B10.RIII mice and mice that had been immunized 5, 6, 7, 8, and 16 days previously with peptide were injected through the tail vein into syngeneic normal nonimmunized mice and day-5, -6, -7, -8, and -16 pi mice at the corresponding time point. Leukocytes from lymph nodes of OVA-immunized mice at day 6, 8, and 16 pi were also injected into syngeneic OVA-immunized mice at corresponding time points after immunization. Cells were labeled in vitro with calcein-AM (C-AM; Molecular Probes Europe BV, Leiden, The Netherlands) before they were injected, as described previously. ¹⁴ In brief, a single-cell suspension was prepared from the draining lymph nodes. Cells were then resuspended in 20 mL complete medium (RPMI 1640 supplemented with 10% [vol/vol] FCS, 1 mM sodium pyruvate, 4 mM l-glutamine, 100 μg/mL streptomycin, and 100 IU/mL penicillin; Gibco BRL, Paisley, UK) at 2 × 10⁶ cells/mL. Cells (2 × 10⁷) in 10 mL were incubated with 40 μg/mL C-AM at 37°C for 30 minutes. C-AM is nontoxic and has no effect on cell adhesion. ¹⁵ Cells were then washed three times with culture medium. Cells (1 × 10⁷) in 150 μL medium were immediately infused into recipient mice through the tail vein. Twenty-four hours later, the leukocyte-infused mice were injected with 2% Evans blue, as described earlier, and the retinal wholemounts were prepared. 

Immunolocalization of Adhesion Molecules in Retinal Vessels

Fluorescein isothiocyanate (FITC)–conjugated anti-mouse CD106 (vascular cell adhesion molecule [VCAM]-1, isotype control: FITC-conjugated rat IgG_2a), FITC-conjugated anti-mouse CD54 (intercellular adhesion molecule [ICAM]-1, isotype control: FITC-conjugated hamster IgG), FITC-conjugated anti-mouse CD62P (P-selectin, isotype control: FITC-conjugated rat IgG₁), purified anti-mouse CD62E (E-Selectin, isotype control: rat IgG_2a) and FITC-conjugated anti-mouse CD31 (platelet endothelial cell adhesion molecule [PECAM]-1, isotype control: FITC-conjugated rat IgG_2a) were purchased from PharMingen (BD Biosciences, Oxford, UK). The purified anti-mouse CD62E and isotype control rat IgG_2a were labeled with a fluorescent protein kit (Alexa Fluor 488; Molecular Probes Europe BV). VCAMs were stained in vivo with the above antibodies, by using a slightly modified version of the method previously described. ^{16 17} Briefly, 40 μL (20 μg) of each antibody or isotype control in 80 μL PBS was injected through the tail vein and allowed to bind for 15 minutes before injection of Evans blue. The animals were then killed and retinal wholemounts prepared as described earlier. The expression of P-selectin and ICAM-1 in mesenteric vessels was also observed after the treatment. The mesenteric tissues were fixed in 2% paraformaldehyde for 30 minutes and then were mounted on slides for confocal microscopy. 

Confocal Microscopy

All retinal wholemounts were examined for Evans blue and either C-AM or FITC by a confocal scanning laser imaging system fitted with krypton-argon lasers (MRC 1024; Bio-Rad Microsciences, Hemel Hempstead, UK). Using dual blue and green fluorescence, the Evans blue appeared red and the C-AM or FITC stain appeared green. 

Data Analysis

The fluorescence intensity of the adhesion molecules was quantified by image-analysis computer software (QWin System; Leica, Wetzlar, Germany). For each vessel analyzed, the fluorescence intensity in 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. Retinal arteries and veins were measured within 1 mm surrounding the optic disc. Vessels were considered to be independent variables. Five to six vessels in each retina were measured, and the mean ± SEM was calculated at each time point. Probabilities when control and EAU mice were compared at each time point were calculated with the Dunnett multiple comparison test. P < 0.05 was considered statistically significant. 

Results

Microscopic Changes in Retinal Vascular Permeability

Vascular Permeability in Control PBS-Immunized B10.RIII Mice.

The BRB was intact in all control PBS-injected B10.RIII mice. Thus, intravascular injection of Evans blue resulted in a sharp outline of the retinal vessels, with the dye retained within the vessel lumen and no detectable leakage into the tissue parenchyma (Fig. 1A) . The inner retinal vascular plexus, located in the ganglion cell and nerve fiber layer, consists of arteries, arterioles, capillaries, venules, and veins. Figure 1G shows the inner retinal plexus in control PBS-injected mice, which was characterized by vessels with a very distinct outline due to retention of Evans blue within the vascular lumen. Figure 1J shows the same region of the retina as shown in Figure 1G but focused at the junction of the outer plexiform layer and outer nuclear layer, where the outer retinal vascular plexus is located. The capillary-sized vessels that make up this plexus showed excellent retention of dye, with no evidence of fluorescence in the tissue parenchyma and a very distinct outline of the vessel. 

Vascular Permeability in EAU.

With the Evans blue dye technique, the first evidence of breakdown of the BRB with leakage of the dye into the retinal parenchyma occurred at day 7 pi in animals without clinical evidence of EAU at this stage. At day 7 pi, 6 of the 10 retinas contained obvious sites of focal dye leakage in venules and postcapillary venules (PCVs; Figs.1B 1H ) of the inner retinal vascular plexus, although the frequency of these sites was limited. In contrast to the inner retinal plexus, the outer retinal plexus showed no detectable changes in vascular permeability (Fig. 1K) at this time. However, by day 9 pi a marked breakdown of the BRB was evident and persisted through day 21 pi in all retinas (Figs.1C 1D E -F ) and was mainly located at the retinal veins, venules, and PCVs. Both the outer and inner retinal vascular plexi were involved at this stage with leakage of Evans blue from the vessel lumen resulting in a pronounced blurring of the vessel outline (Figs. 1I 1L) . No leakage was observed in the retinal arteries and arterioles throughout the observation period. However, these vessels lost their natural curvature and became straighter and narrower from days 10 to 21 pi (Figs. 1D E -F ), suggesting an increase in vascular tone in these vessels. 

Leukocyte Adhesion and Infiltration during EAU

Nonimmunized B10.RIII Mice.

In control nonimmunized B10.RIII mice infused with normal labeled cells, only a few leukocytes (11.67 ± 1.80/retina) were detected in wholemounts within the lumen of retinal vessels 24 hours after cell infusion, mostly in the capillaries of the outer retinal plexus (Fig. 2A , Table 1 ). Very few cells were detected in the veins or venules and none in the retinal parenchyma. Previous studies (Xu et al., manuscript submitted) also showed that infusion of cells from immunized animals (day 12 pi) into normal control mice was not associated with significant leukocyte adhesion to retinal veins or venules 1 hour or even 24 hours after cell infusion. 

IRBP Peptide-Immunized Mice.

The first evidence for leukocyte adhesion to vascular endothelium of veins and venules occurred at day 6 pi. Although the total number of detected cells in the retina was not significantly increased when compared with the normal control nonimmunized recipient mice (Table 1) , the cells’ preferential location to focal sites in retinal veins and venules of the inner vascular plexus was clear (Fig. 2B , Table 1 , P < 0.05 and P < 0.001, respectively). A significant increase in the total number of leukocytes sticking within the lumen of retinal vessels was apparent in day-7 pi mice (Table 1) , but was still restricted to the veins and venules of the inner retinal vascular plexus. No cell infiltration of the retinal parenchyma was detected. At day 8 pi, the number of intravascular sticking leukocytes was further increased (Table 1) . A small number of such cells had undergone extravasation, and the cells were apparent near the vessels of the retinal parenchyma (Table 1 , Fig. 2C ). However, this change was limited to the veins and venules of the inner retinal vascular plexus where the BRB was breached, as shown by concomitant focal leakage of Evans blue dye (Figs. 2C 2D ). The number of retinal parenchymal infiltrating cells increased dramatically from day 9 pi onward (Table 1) . Both the inner and outer retinal vascular plexi were involved at this time point (Figs. 2E 2F ). 

OVA-Immunized Mice.

Preferential location of injected leukocytes in retinal veins and venules of the inner vascular plexus was also observed in day-6 pi OVA-immunized mice (Table 1) . However, there was no further accumulation of cells in these vessels even at days 8 and 16 pi (Table 1) . No cellular infiltration was observed in OVA-immunized mice. 

Expression of Cellular Adhesion Molecules in Retinal Vessels during EAU

Selectin Expression.

P-selectin expression was negligible in the retinal vessels of normal (Fig. 3A2) and day-3 pi mice (Fig. 4) . From day 7 pi onward, the expression of P-selectin on veins and venules of the inner retinal vascular plexus increased progressively (Figs. 3F2 4 ), with only a weak increase in expression on arteries and arterioles and none on capillaries (data not shown). However, P-selectin was not upregulated in the outer vasculature during the course of the disease (data not shown). The staining pattern of P-selectin appeared as small granules located on the endothelium (Fig. 3F3) . 

E-Selectin Expression.

Similar to P-selectin, E-selectin staining was negligible in all retinal vessels of normal (Fig. 3B2) and day 3 pi B10.RIII mice (Fig. 4) . Between days 7 and 16 pi, there was a progressive increase in the expression of E-selectin on retinal veins and venules, peaking at day 9 pi (Figs. 3G2 4 ). Expression of E-selectin was also located on the endothelial surface and appeared granular (Fig. 3G3) . No expression of E-selectin was detected in the retinal arteries, arterioles, and capillaries, even at the later stage of the disease from days 9 to 16 pi (data not shown). 

ICAM-1 Expression.

In control, nonimmunized B10.RIII mice, ICAM-1 was positively identified in the veins and venules of inner retinal vessels but stained weakly in arteries and arterioles and was negative in capillaries (Fig. 3C2) . From days 7 to 16 pi, the expression of ICAM-1 increased progressively on veins and venules of the inner retinal plexus (Fig. 4) , peaking at day 9 pi (Figs. 3H2 4 ), with a lesser increase on arteries and arterioles but none on capillaries (data not shown). The outer retinal vascular plexus, however, displayed only weak upregulation of ICAM-1 expression between days 9 and 16 pi (data not shown). The expression of ICAM-1 was detected on the endothelial luminal surface (Fig. 3H3) . In ICAM-1–positive veins and venules, blood cell aggregations were observed between days 7 and 21 pi, and ICAM-1 also stained positively on the surface of aggregations (Fig. 3P) . 

VCAM-1 Expression.

In control mice, VCAM-1 was expressed more highly on retinal arteries and arterioles than on retinal veins and venules of the inner vascular plexus (Fig. 3D2) . Between days 9 and 16 pi, there was a progressive increase in the expression of VCAM-1 on retinal arteries, veins, and venules, with less increase in arterioles (Figs. 3I2 4 ). The expression of VCAM-1 was also located on the endothelial luminal surface (Fig. 3I3) . 

PECAM-1 Expression.

The expression of PECAM-1 in the retinal vasculature of normal mice was similar to that of VCAM-1 and predominated on retinal arteries and arterioles (Fig. 3E2) . Between days 9 and 16 pi, there was a slight increase in the expression of PECAM-1 on retinal arteries, veins, and venules and on capillaries of the inner vascular plexus (Figs. 3J2 4 ). No obvious expression of PECAM-1 was detected on the outer vascular plexus (Fig. 3O2) . The expression of PECAM-1 was specifically localized to the intercellular junctions of endothelial cells (Fig. 3J3) . 

None of the isotype control antibodies resulted in staining of the retinal vasculature or parenchyma. 

Expression of Adhesion Molecules in Mesenteric Vessels in EAU

To determine whether the upregulation of the studied adhesion molecules in EAU is retinal vessel specific, the expression of P-selectin and ICAM-1 in the mesenteric vessels of normal and day-9 pi B10.RIII mice was studied. P-selectin was not detected in the mesenteric vessels of normal (Fig. 3K2) or day-9 pi EAU mice (Fig. 3L2) . ICAM-1 was positively stained in the mesenteric vessels of normal mice (Fig. 3M2) . There was no significant enhancement of ICAM-1 (Fig. 3N2) expression in the mesenteric vessels of day-9 pi mice. 

Expression of Adhesion Molecules in Retinal Vessels of OVA-Immunized Mice

To investigate whether the upregulation of adhesion molecules is EAU specific, the expression of ICAM-1 and P-selectin in retinal vessels was studied in day-9 pi OVA-immunized mice. Neither ICAM-1 nor P-selectin was significantly enhanced in day-9 pi OVA-immunized mouse retinal vessels (Fig. 5) . 

Discussion

Using retinal wholemounts we have been able to covisualize changes in retinal microvascular permeability, the expression of endothelial cell adhesion molecules, and the distribution and properties of blood leukocytes. The sequence of events in EAU in the B10.RIII mouse appeared to be an initial focal adhesion of activated leukocytes to discrete sites on PCVs at 6 days pi. Simultaneously, there was a reduction in the number of leukocytes observed in small capillaries, suggesting that the normal transient arrest (trapping) in the passage of leukocytes through the small capillaries is prevented probably by capillary narrowing or occlusion. Adhesion of leukocytes to PCVs was followed 24 hours later by upregulation of adhesion molecules, especially ICAM-1 and P-selectin, and breakdown of the BRB, leading to transendothelial migration of leukocytes and recruitment of large numbers of cells to the retinal parenchyma. These changes occurred over a short period (6–9 days pi) and initiated the process of tissue damage by infiltrating lymphocytes ¹⁸ and macrophages ¹⁹ during the following 2 to 3 weeks. Breakdown of the BRB occurred 24 hours before cell infiltration in this model, in contrast with a previous study that found barrier dysfunction occurring concomitantly with lymphocyte infiltration in EAU. ⁴ The differences observed may be due to the different animal models used. Alternatively, it is more likely that this inconsistency is due to differences in tissue preparation and technical sensitivity. Compared with conventional histologic techniques, confocal microscopy of retinal wholemounts is more suited to the detection of the initial small, patchy breaches in the barrier and the initial steps toward infiltration of lymphocytes, because it allows visualization of the entire retinal vasculature. ^{11 20}  

Development of Increased Retinal Vascular Permeability in EAU

In this study of the inner BRB in EAU, we found that BRB breakdown first occurred in the inner vascular plexus and was limited to the retinal veins and venules, and in particular to the PCVs at day 7 pi, later extending to the outer vascular plexus at day 9 pi. The sites of initial BRB breakdown in the inner plexus are the same sites at which leukocyte arrest and adhesion, upregulation of adhesion molecules, and cellular extravasation were detected. Leakage from capillaries was not a significant feature. Instead, areas of capillary nonperfusion were observed, suggesting capillary closure by intravascular coagulation and/or cell plugging, although the numbers of leukocytes in capillaries was reduced (Table 1) . The BRB at the retinal arteries and arterioles was intact during the progress of EAU, but these vessels were not unaffected by disease. They lost their natural curvature and became straight, appearing less compliant during peak disease. However, they were not involved in leukocyte recruitment in EAU. Previously, a similar finding was reported in a soluble-antigen (S-Ag)-induced EAU model in rats, in which breakdown of the BRB and infiltration of inflammatory cells were observed in retinal venules. ²¹ However, the BRB of the inner and outer vascular plexi was not compared in that study. 

The process that initiates the early increased vascular permeability is unclear. It has been reported that cytokines, such as vascular endothelial growth factor (VEGF), ⁵ tumor necrosis factor (TNF)-α and interleukin (IL)-1β ^{5 22 23 24} are capable of inducing BRB dysfunction. These factors are also capable of stimulating release of nitric oxide by activated neutrophils and endothelial cells. ²⁵ Nitric oxide has been found to contribute in part to the increase in permeability of the BRB. ²⁶ The disruption of the BRB in veins, venules, and PCVs of the inner retinal vascular plexus is likely therefore to be the result of the release of such inflammatory agents from the accumulated leukocytes, as discussed in the next section. 

Leukocyte Adhesion and Infiltration

Preferential location of adhering leukocytes to the endothelium of retinal veins and venules was observed at day 6 pi in both IRBP peptide–immunized and OVA-immunized mice, indicating that this is non–antigen-specific adhesion. However, there was no detectable increase in any endothelial cell adhesion molecules examined at this time point, and what initiates this primary vessel-type specific adhesion of leukocytes is not clear. One possibility is that, 6 days after systemic immunization, the avidity of adhesion molecules in the veins and venules is increased. It is also possible that other adhesion molecules such as CD44 hyaluronan are involved. We have recently found that CD44 hyaluronan plays an important role in leukocyte homing in EAU (Xu et al., manuscript in preparation). 

At day 7 pi, further accumulation of leukocytes in retinal veins and venules was observed in IRBP peptide–immunized mice but not in OVA-immunized mice. The accumulation of leukocytes in IRBP peptide–immunized mice coincided with an increase in vascular permeability and upregulation of adhesion molecules. The mechanism by which this is achieved is not known. 

It has been shown that a small number of activated T cells enter the retina of normal rats 12 hours after intravenous infusion, ²⁷ and it is possible that these cells interact with perivascular antigen-presenting cells (APCs), with recognition of antigen resulting in the manufacture of proinflammatory cytokines and chemokines locally and amplification of the response. ²⁷ In the present study, we have also shown that intravenously injected in vivo primed cells extravasate across the BRB into the retina within 24 hours of injection. A report by Hu et al. ²⁸ in a rat model showed that systemic infusion of OVA-activated T cells could induce breakdown of the BRB with extravasation of a limited number of endogenous non–OVA-specific T cells into the retina 12 hours later. Using the same technique in the mouse, we did not detect extravasation of injected lymphocytes into the retinas of mice at day 6 pi when the BRB was still intact (see Fig. 2B ), but only from day 8 pi when there was evidence of initial breakdown of the BRB. This agrees with the data of Hu et al., suggesting that extravasation of cells into the retina after intravenous infusion may be dependent on BRB breakdown, which is likely to occur during the initial 12-hour lag phase after intravenous injection. How adhesion of leukocytes might cause breakdown of the BRB is not known, but it is possible that activated leukocytes interact either with perivascular APCs, as suggested by Prendergast et al., ²⁷ or with antigen-bearing endothelial cells directly, producing proinflammatory factors locally. In vitro studies have shown that retinal vascular endothelial cells could express major histocompatibility complex (MHC) class II and present antigen. ²⁹ However, whether this is the case in vivo needs further investigation. 

Locally produced cytokines change the microenvironment of retinal veins and venules, leading to upregulation of adhesion molecules and increase in vascular permeability. The accumulation of activated cells as a result of an increase in adhesion molecules leads to high concentrations of cell-released inflammatory mediators locally and, consequently, further localized changes in permeability and expression of adhesion molecules. 

Chemokines are thought to be crucial to leukocyte recruitment and cellular extravasation. We have found that monocyte chemoattractant protein-1 (MCP-1, CCL2), regulated on activation of normal T-cell–expressed and secreted (RANTES, CCL5), and macrophage inflammatory protein (MIP)-1α (CCL3) are associated with cells infiltrating the retina in EAU ³⁰ and that cytokines such as TNF-α and IL-1β can stimulate production of MCP-1 and IL-8 in retinal endothelial cells. ³¹ We suggest therefore that cytokines released by accumulating leukocytes in retinal veins and venules and also the interaction of the leukocyte with the endothelium itself may stimulate chemokine production by the endothelial cell. These chemokines may then enable the adherence of additional leukocytes, further activate accumulated leukocytes, and initiate cell extravasation in a process of amplification. 

Expression of Adhesion Molecules

Because adherence of leukocytes to the endothelium is a consequence of the expression of appropriate adhesion molecules, we examined the distribution of P-selectin, E-selectin, ICAM-1, VCAM-1, and PECAM-1 on the retinal microvascular endothelium during the course of EAU. Our data show that the expression of these adhesion molecules increased progressively as EAU developed, with the increase in ICAM-1 and P-selectin expression starting earlier and more closely associated with the sites of BRB breakdown, cellular adhesion, and extravasation. The upregulation of adhesion molecules was not observed in the mesenteric vessels of EAU mice and in the retinal vessels of OVA-immunized mice, indicating some degree of specificity in upregulation of adhesion molecules in retinal vessels in EAU. Localization and upregulation of P-selectin and ICAM-1 in retinal veins and venules in EAU may contribute to the specific adhesion of activated leukocytes in these vessels and subsequently the breakdown of the BRB. ICAM- and leukocyte function-associated antigen (LFA)-1 have been shown to play an important role in the pathogenesis of EAU, ³² in that ICAM-1 expression precedes histologic evidence of inflammation. This is in agreement with the present study. In addition, in vivo treatment with anti-ICAM-1 and anti-LFA-1 monoclonal antibodies completely prevents the development of the disease ³² or significantly reduces its severity. ^{32 33} Activated leukocytes circulating in the retinal vasculature may upregulate ICAM-1 and P-selectin on the endothelium through cytokines. ³⁴  

In contrast to ICAM-1 and P-selectin, increases in expression of VCAM-1 and PECAM-1 occurred later and was prominent in retinal arteries and arterioles, suggesting that they play different roles and may be less important than P-selectin and ICAM-1 for leukocyte infiltration into the retina during EAU. PECAM-1 is believed to be required for neutrophil and monocyte transmigration both in vivo and in vitro. ^{8 9} It is reported to be expressed, at higher levels, on the endothelium of all vessel types. ³⁵ However, in our study, it was not detected on the endothelium of the outer vascular plexus, even at days 9 and 16 pi when leukocyte infiltration was detected around these vessels. 

We conclude therefore that a likely course of events in EAU is as follows. As indicated earlier, the initial event in mouse EAU appears to be adhesion of activated lymphocytes which, by as yet unknown mechanisms, are followed by a series of amplification steps involving increased expression of adhesion molecules and BRB breakdown at the PCVs, with eventual more widespread increases in expression of adhesion molecules and BRB breakdown of the larger venules and veins. What initiates lymphocyte adhesion at day 6 pi is not clear, but possible mechanisms include CD40-CD40L and/or other cognate interaction between the endothelium and the lymphocyte. These require further study. 

 Supported by Grant 057311 from The Wellcome Trust.

 Submitted for publication December 3, 2001; revised May 9 and July 29, 2002; accepted August 9, 2002.

 Commercial relationships policy: N.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Evans blue-albumin complex leakage in wholemount retinas of control and peptide-immunized mice during the development of EAU. Evans blue was injected through the tail vein 10 minutes before death. (A, G, J) Normal PBS-immunized mouse: retinal vessels were sharply outlined, and there was no Evans blue-albumin complex leakage from the vessels in both inner (G) and outer (J) retinal vascular plexi. (B, H, K) Day-7 pi mouse: Evans blue-albumin complex leakage was seen focally in retinal venules and PCVs (arrows) but was limited to the inner retinal vascular plexus (H). There was no leakage in the outer retinal vascular plexus (K, same field as H). (C, D, I, L) Day-9 and -10 pi mice: the number and extent of Evans blue-albumin complex leakage increased as the disease progressed (arrows), both inner (I) and outer (L) retinal vascular plexi were involved. (E, F) Day-16 and -21 pi mice: as disease resolved, leakage of Evans blue was reduced. Scale bar, 100 μm.

Figure 2.

 

Cell infiltration in the retinal parenchyma. C-AM–labeled cells were injected into immunized mice through the tail vein 24 hours before death, and Evans blue was injected 10 to 15 minutes before death. Retinal wholemounts were observed by confocal laser microscopy. (A) Control PBS-immunized mouse showing C-AM–labeled cells sticking in the capillaries of the outer vascular plexus. No cell infiltration was detected. (B) Day-6 pi retina showing C-AM–labeled cells sticking along the venules of the inner vascular plexus with no cell infiltration. (C, D) Day-8 pi retina showing many cells sticking in a retinal vein and two cells infiltrating into retinal parenchyma (C). (D) No cell infiltration was detected in the outer vascular plexus. (E, F) Day-9 pi mouse: cells from the vessels showing BRB breakdown infiltrated both the inner (E) and outer (F) vascular plexi.

Table 1.

 

C-AM Labeled Leukocytes within the Retina of B10.RIII Mice during the Development of EAU, 24 Hours after Cell Injection

Time Point	Cell Number			Location
		Total	Infiltrating	Arteries	Veins	Venules/PCVs	Capillaries
Day 0	11.67 ± 1.80	0.00 ± 0.00	0.00 ± 0.00	1.17 ± 0.48	1.67 ± 0.21	8.83 ± 1.42
Day 6	16.33 ± 2.57	0.00 ± 0.00	0.00 ± 0.00	4.83 ± 1.38*	9.17 ± 1.25, ‡	2.17 ± 0.31, †
	(16.00 ± 5.06)	(0.00 ± 0.00)	(0.17 ± 0.41)	(4.17 ± 1.94*)	(7.17 ± 2.48*)	(4.50 ± 1.87*)
Day 7	26.50 ± 1.98*	0.00 ± 0.00	0.00 ± 0.00	6.17 ± 0.95, ‡	16.67 ± 1.71, ‡	3.67 ± 0.80, †
Day 8	51.83 ± 4.00, †	25.50 ± 3.07	0.50 ± 0.34	15.50 ± 1.48, ‡	34.00 ± 2.78, ‡	3.50 ± 0.62, †
	(15.33 ± 5.57)	(0.00 ± 0.00)	(0.33 ± 0.52)	(3.67 ± 1.21*)	(6.50 ± 2.17*)	(4.83 ± 2.56*)
Day 9	345.70 ± 14.83, ‡	301.30 ± 18.40, ‡	—	—	—	—
Day 16	401.30 ± 21.54, ‡	357.00 ± 21.64, ‡	—	—	—	—
	(13.33 ± 4.13)	(0.00 ± 0.00)	(0.67 ± 0.82)	(2.67 ± 1.37)	(3.83 ± 2.23)	(6.00 ± 1.67)

Figure 3.

 

Expression of adhesion molecules in the retinal and mesenteric vessels of normal nonimmunized and peptide-immunized mice. FITC-conjugated antibody and Evans blue dye were injected through the tail vein. Retinal wholemounts were prepared for confocal observation. Red, Evans blue showing vessels (1); green, FITC-conjugated antibody showing adhesion molecule expression (2). Expression in retinal vessels of normal nonimmunized B10.RIII of (A1, A2) P-selectin, (B1, B2) E-selectin, (C1, C2) ICAM-1, (D1, D2) VCAM-1, and (E1, E2) PECAM-1. Day 9 pi retina: (F1–F3) upregulation of P-selectin in retinal veins and venules and appearing as small granules on the endothelium; (G1–G3) upregulation of E-selectin in retinal veins and venules and appearing as granules on the endothelium; (H1–H3) upregulation of ICAM-1 on the endothelium of retinal veins and venules; (I1–I3) upregulation of VCAM-1 on the endothelium of retinal arteries; and (J1–J3) upregulation of PECAM-1 on retinal vessels and localization to the intercellular junctions of endothelial cells (J3). Expression in mesenteric vessels of normal nonimmunized mice of (K1, K2) P-selectin and (M1, M2) ICAM-1. Expression in day-9 pi mesenteric vessels of (L1, L2) P-selectin and (N1, N2) ICAM-1. PECAM-1 was expressed specifically on the inner (O1) but not the outer (O2) vascular plexus (9 days pi). (P) Day-9 pi retina showing blood aggregation in an ICAM-1–positive retinal vein. a, artery; v, vein. Scale, 50 μm.

Figure 4.

 

The fluorescence intensity of adhesion molecule staining in retinal veins (A) and venules (B) during the progression of EAU. The difference between control and peptide-immunized mice at each point was compared by the Dunnett multiple comparison test. *P < 0.05; **P < 0.01; ***P < 0.001 versus control (n = 6).

Figure 5.

 

The intensities of adhesion molecule P-selectin and ICAM-1 in retinal veins of normal nonimmunized and day 9 pi OVA- or IRBP peptide–immunized mice. Data are the mean ± SEM, n = 6. **P < 0.01 versus normal nonimmunized mice (Student’s unpaired t-test).