T Cell Traffic and the Inflammatory Response inExperimental Autoimmune Uveoretinitis

Robert A. Prendergast,1 Charles E. Iliff,1 Nezih M. Coskuncan,1 Rachel R. Caspi,2

Gil Sartani,2 Teresa K. Tarrant,2 Gerard A. Lutty,1 and D. Scott McLeod1

PURPOSE. TO quantify S-antigen-specific (S-Ag) T cells in the retina after adoptive transfer, and toevaluate their role in the initiation and progress of destructive ocular inflammation in experimentalautoimmune uveoretinitis (EAU).

METHODS. Lewis rats were administered 10 X 106 S-Ag-specific T cells from the SP35 cell line or10 X 106 concanavalin A-stimulated syngeneic spleen cell lymphoblasts labeled with lipophilicPKH26 fluorescent dye immediately before intravenous inoculation. Labeled cells in each retinawere counted at various times from 4 to 120 hours after cell transfer by fluorescence microscopicanalysis of each dissociated retina. Recipient eyes were examined within the same period by lightand confocal microscope.

RESULTS. SP35 T cells showed a biphasic distribution in the retina. The first peak of 160 cells/retinawas noted at 24 hours. A steady decline of labeled cells at 48 and 72 hours was followed by a rapidincrease at 96 and 120 hours. Concanavalin A-stimulated, control-labeled cell populations showedan identical peak at 24 hours but a persistent decline thereafter; only two or three T cells werepresent in each retina at 120 hours. Concurrent inoculation of SP35 cells and nonspecific T cellblasts did not produce more SP35 cells than control cells in the retina at any time. Microscopicanalysis showed mononuclear cell infiltration of the iris, ciliary body, and aqueous humor at 48hours, which intensified rapidly and persisted through 120 hours. Retinal inflammation did notbegin until 80 hours. Mononuclear cell adherence to vascular endothelium and perivascularmacrophage infiltration of the innermost layers progressed to edema, and profound destructiveinflammation and loss of retinal stratification were observed at 120 hours.

CONCLUSIONS. There is no evidence of a blood- ocular or blood-retinal barrier to activated T cellblasts. Autologous S-Ag does not provoke a more rapid entry of specific T cells at that site. The dataconfirm that anterior segment inflammation precedes retinal inflammation, even though S-Ag-specific T cells were present in the retina within a few hours after cell transfer. Because S-Ag isclearly present in the retina, delay in antigen presentation at that site may account for the temporaldifference between retinal and anterior segment inflammation. (Invest Ophthalmol Vis Sci. 1998;39:754-762)

Experimental autoimmune uveoretinitis (EAU) is a well-characterized, lymphocyte-mediated autoimmune dis-ease model that can be induced by active immunization

of susceptible species with whole S-antigen (S-Ag) protein(arrestin) or by immunization with certain immunopathogenicpeptides derived from this molecule.1"3 It can be induced innaive syngeneic recipients by transfer of activated CD4+ major

From 'The Johns Hopkins University, Baltimore; and the 2NationalEye Institute, National Institutes of Health, Bethesda, Maryland.

Supported in part by Research to Prevent Blindness and Fight ForSight Research Division of Prevent Blindness America (CEI), New York,New York; National Institutes of Health National Research ServiceAward Institutional Training grant EY07047 (NMC); Research to Pre-vent Blindness Foreign Fellowship (NMC); and National Institutes ofHealth Core Center grant EY01765 (Wilmer Institute). GAL is an Amer-ican Heart Association Established Investigator.

Submitted for publication July 14, 1997; revised November 26,1997; accepted December 9, 1997.

Proprietary interest category: N.Reprint requests: Robert A. Prendergast, The Wilmer Eye Institute,

The Johns Hopkins University School of Medicine, 600 North WolfeStreet, Baltimore, MD 21287-9142.

histocompatibility complex-restricted T cells specific for reti-nal S-Ag.4"6 The presence of S-Ag-specific T cells within theglobe is requisite for the initiation and rapid development ofretinal and uveal inflammation.7'8 Because the retina is a majortarget organ, this exquisitely defined and readily isolated tissuepresents a unique structure for the study of the traffic andspecificity of T cells within the target tissue at sequential timesafter immunization.

Antigen-specific migration or retention of immune effec-tor T cells at sites of cell-mediated immunity, including delayedhypersensitivity, allograft rejection reactions, and autoimmuneT cell-driven lesions, has been studied for many years.9"1' Ingeneral, tritiated thymidine-labeled lymphocytes, transferredfrom immunized donors to syngeneic recipients for quantita-tive analysis of the number of transferred cells at antigen-specific or appropriate nonspecific challenge sites, have beenused in these studies. In the case of skin allograft rejection inthe rabbit, cells from a specific draining lymph node reactingto one allograft were labeled in vivo, and the presence of thesecells calibrated at test transplant rejection sites of identical orallogeneic skin grafts in the same animal.10 In none of theseinstances has a clear preponderance of labeled sensitized lym-

754Investigative Ophthalmology & Visual Science, April 1998, Vol. 39, No. 5Copyright © Association for Research in Vision and Ophthalmology

IOVS, April 1998, Vol. 39, No. 5 T Cell Traffic in EAU 755

phocytes been demonstrated at the specific immune reactivesite in excess of the number of these cells at appropriatecontrol sites.

Several objections could be raised to these experiments,however. The specific donor lymphocytes from animals ini-tially responding to antigen had, in general, been accumulatedfrom disrupted lymphoid tissue in which antigen-respondingcells were actively engaged in clonal expansion and wereharvested at various stages of maturity. The specific cellswithin the donor population were present in low concentra-tions in a highly random, immunologically naive group. Donorcells were usually examined autoradiographically at reactivesites in minute samples taken from inflammatory lesions, or thenumber of donor cells at the reactive inflammatory site wasinferred by analysis of total radioactivity of that area when thedonor population was labeled ex vivo with a radioactive tagbefore transfer.12 In the latter instance, it is unclear preciselyhow radiation comes to this site, whether from the originallabeled lymphocyte population, through generation of thesecells in the host animal, or by leakage of label and possibleuptake by nonspecific inflammatory cells.

The precise mechanism of T cell-mediated inflammatorydisease is not fully understood. Furthermore, in EAU, the com-plete temporal sequence of ocular inflammation, includingchemotaxis and activation of mononuclear cells, is also un-clear. In the experiments described in the present report, twoexperimental techniques were used to examine T cell traffic inthe EAU model of ocular autoimmune disease. First, the precisenumber of T cells reactive to retinal antigen were calibrated byusing a uniformly labeled S-Ag-specific T cell line, initiallycharacterized by Savion et al.,6 for adoptive transfer of EAU.The second technique used was cytologic examination of theentire end organ—that is, the retina, for the presence of effec-tor cells. Thus, it was possible to quantify all labeled T cellsinvolved at the end organ-reactive site and so to obviate thesampling error inherent in previous studies. These data dem-onstrate that activated T cells of S-Ag or random specificityinvade retinal tissue early after transfer, and that the totalnumber of these T cells can be directly measured during earlyphases before and after development of destructive disease.The sequence of pathologic changes induced by these T cellswill also be shown throughout the entire sequence from thetime of T cell transfer through complete destruction of theretina.

MATERIALS AND METHODS

Animals

Young male Lewis rats, 175 to 225 g, and adult female BALB/cmice were obtained from Charles River (Wilmington, MA) andwere maintained in the animal facilities at the Wilmer Institute.The maintenance, care, and all experimental use of theseanimals were in strict compliance with guidelines issued by theNational Institutes of Health and the ARVO Statement for theUse of Animals in Ophthalmic and Vision Research.

Cells

SP35, the T cell line derived from reactive lymph nodes ofLewis rats immunized with bovine S-Ag plus complete Freund'sadjuvant, was used throughout this study. SP35 is a classII-restricted T cell line specific to the P35 peptide composed

of amino acids 337 to 356 of human S-Ag. This line wasmaintained by restimulation with S-Ag peptide P35 antigen,then by interleukin-2 (IL-2)- driven expansion, according to themethod of Savion et al.6 After culture, all cells were frozenwhile in the active blast state and were maintained in liquidnitrogen for 3 to 6 months before their use in these experi-ments. Aliquots from tissue culture harvests of these effector Tcells were tested for their pathogenicity by adoptive cell trans-fer to naive Lewis rats, in which 0.2 X 106 cells injectedintraperitoneally consistently induced EAU.

Fluorescent Labeling of T Cells and Induction ofExperimental Autoimmune Uveoretinitis

Vials of frozen SP35 and y-irradiated splenic feeder cells wererapidly thawed under 56°C tap water and were allowed torecover at 5 million SP35 cells/ml for 4 hours in RPMI mediumcontaining 10% activated spleen cell supernatant and antibiot-ics. Cells were harvested, washed in Hanks' balanced saltsolution (HBSS), and placed over Ficoll (Sigma, St. Louis, MO)to separate viable SP35 lymphoblasts from irradiated rat feedercells used during proliferative expansion. After centrifugationat 400g for 22 minutes, SP35 blast cells were harvested,washed three times in HBSS, and labeled, using the red fluo-rescein dye cell linker kit (PKH26; Sigma), precisely as directedby the manufacturer. PKH26 is a patented fluorescent celllinker technology that incorporates fluoresceinated lipophilicmolecules into the cell membrane by selective partitioning.The precise nature and chemical composition of all of thecomponents of the cell linker kit are unavailable from Sigma orfrom the original manufacturer.13>l4 Briefly, a suspension of 20to 25 X 106 HBSS-washed SP35 cells was exposed to 2 X 10"6

M PKH26 dye in PKH26 buffer for 4 minutes in polypropylenetubes. The reaction was stopped by addition of an equal vol-ume of fetal bovine serum for 1 minute, and the cell solutionwas diluted 1:1 with complete medium and washed threetimes before injection. The cells were brilliantly red with asmooth, uniformly fluorescent membrane label, visible underepifluorescent microscope equipped with a rhodamine detec-tion filter system. Labeled SP35 cells were concentrated to10 X 106 cells in 0.3 ml HBSS and injected into the penile veinof normal Lewis rats anesthetized by xylazine hydrochloride(Rompun; Bayer Animal Health, Leverkusen, Germany) andketamine hydrochloride (Ketaset; Bristol-Myers Squibb, NewYork, NY). Injection was simple to perform; the nits awakenedapproximately 15 to 30 minutes later and displayed no evi-dence of injury. A control population of normal Lewis ratspleen cells, stimulated by concanavalin A (Con A; Sigma) at35 /Jig/ml for 60 hours, was used after PKH26 labeling at thesame concentration and was injected intravenously in an iden-tical fashion.

Enumeration of S-Antigen T Cell Population inRetinal Tissue

Labeled SP35 or control cell recipient rats were killed using anoverdose of sodium pentobarbital delivered intraperitoneallyand the vascular system was flushed by left ventricular injec-tion of 200 ml HBSS. The eyes were carefully dissected, theanterior segment removed, and the retinas separated from thechoroid within 30 minutes after death. All ocular tissues werefixed for 2 days in 4% paraformaldehyde in phosphate-bufferedsaline containing 6% sucrose (pH 7.2). After washing, fixed

756 Prendergast et al. IOVS, April 1998, Vol. 39, No. 5

retinas were digested with elastase, according to the methoddescribed by Laver et al.15 After removal of intact retinal ves-sels, the total number of disaggregated retinal cells isolated bythis technique from normal Lewis retinas was 12 X 106 ± 8%cells/retina, measured by Coulter counter. Cells from one ret-ina were concentrated to a volume of 150 A HBSS. Labeled cellswere counted, using a standard hemocytometer and a Zeiss(Carl Zeiss, Thornwood, NY) epifluorescent microscope with adry X16 planapochromatic objective. Labeled cells in the pe-ripheral blood were counted in a similar fashion after Ficollseparation of mononuclear cells from 2 ml heparinized periph-eral blood.

HistopathologyThe sequence of pathologic changes was studied at varioustimes after injection of labeled SP35 or labeled Lewis ratCon A-stimulated splenic lymphocytes. At the time of death,the eyes were removed, stabbed with a number 11 blade atfour sites in the pars plana, and fixed in 2% paraformaldehydein 0.1 M cacodylate buffer (pH 7.2) for 20 hours. The retinaswere removed, washed in cacodylate buffer, hemisectioned,and embedded in glycol methacrylate. Two-micrometer sec-tions were stained with toluidine blue and basic fuchsin (Poly-sciences, Warrenton, PA). Anterior segments and posteriorsclera with overlying choroid and retinal pigment cells wereembedded in paraffin, sectioned at 5 jLim, and stained byhematoxylin-eosin. Confocal microscopic study of phosphate-buffered saline-washed, paraformaldehyde-fixed retinal wholemounts was performed, using a Zeiss laser scanning confocalmicroscope, courtesy of Marshall Montrose, Department ofMedicine, Johns Hopkins University.

Lectin-Induced Proliferation and CytokineProduction

PKH26-labeled and unlabeled rat and mouse spleen cells werecultured at 2 X 106 cell/ml in 5-ml quantities in 6-well plates(Costar, Cambridge, MA). Con A was added to these cultures atconcentrations of 3-5 /xg/ml for rat and 2 ju-g/ml for mousespleen cultures. Control and Con A culture supernatants weretested for IL-2, IL-4, and interferon-y, by using enzyme-linkedimmunosorbent assay kits (R&D Systems, Minneapolis, MN).Tritiated thymidine incorporation in control and lectin-stimu-lated spleen cells was determined for PKH26-labeled and un-labeled cells. A mean of 5 wells was used for each measure-ment in a standard microculture technique.'6

RESULTS

T Cell TrafficThese studies were carried out in Lewis rats, using S-Ag-specific T cell line SP35, generated from S-Ag-immune Lewisrats and stimulated in vitro with the human S-Ag peptide P35and lectin-stimulated lymphoblasts. The presence of fluores-ceinated SP35 cells in the retina was examined in rats injectedintravenously with 10 X 106 PKH26-labeled cells. The animalswere killed from 12 to 120 hours thereafter. Removal of theintact retinal vasculature after elastase treatment permittedepifluorescent microscopic analysis of labeled cells in the en-tire retinal cell population. As shown in Figure 1, S-Ag-specificcells demonstrated a biphasic population density within theretina as time elapsed. An early first peak occurred at 24 hours.

250

CECDQ- 150

O100

50

1 12 24 48 72 96 120

Hours After Cell Transfer

FIGURE 1. Measurements of extravascular PKH26-labeled SP35T cells in the retina at various times after intravenous transferof 10 X 106 cells. Each bar represents the mean number oflabeled cells in four to six individual enzyme-dissociated retinasexamined.

There was a steady decrease during the next 2 days, then arapid increase in labeled cells at 96 and 120 hours after SP35 Tcell transfer. Retinal inflammation was not detected during theinitial peak or during its decline, as will be shown. Severalfeatures of this portion of the experiment should be noted. Thenumber of labeled cells in each retina from a single animal wasvery similar. In addition, comparison of the labeled cell numberin retinas examined at the same time after adoptive transferalso showed close adherence to the mean.

The first peak in this biphasic time course experiment at24 hours showed approximately 160 labeled SP35 cells withineach retina. This peak was followed by a steady decline, to theextent that at 72 hours, fewer than 40 cells were presentwithin the retinal tissue. The rapid increase at 96 and 120hours shown in Figure 1 marks only the specifically labeled Tcells. The pathologic condition of the retina at those timesdemonstrates that this represents only a small portion of infil-trating mononuclear cells (see later discussion).

Normal Lewis rat spleen cells activated by Con A stimula-tion were labeled with PKH26 and were used as a controlpopulation. Injection of 10 X 106 labeled T cell blasts induceda peak of activity very similar to that seen with S-Ag-specific Tcells during the first phase of the response 24 hours afterinjection (Fig. 2). Subsequently, however, the cells decreasedto much lower numbers at 96 and 120 hours, the time of thesecond peak after SP35 labeled cell transfer (Fig. 1). No inflam-mation was present at any time after injection of normalCon A-activated T cell blasts.

To study the specificity of cells in the reactive retinaltarget, an experiment was performed in which inoculationof SP35 S-Ag-specific T cells and Con A Lewis rat T cellblasts was used. Only one of these two cell populations waslabeled with PKH26 dye. Thus, 10 X 106 labeled S-Ag-specific cells were injected with 10 X 106 unlabeledCon A-activated spleen cells. The converse experiment wasperformed in which the label was placed on the Con A-stim-ulated control T cell blasts. Experimental autoimmune uveo-retinitis was induced in all animals. No recognizable prepon-derance of specific or nonspecific lectin-activated T cells at24 hours or later times could be demonstrated in this ex-periment in which 4 to 8 eyes were examined for each timepoint for each labeled population (Fig. 3). The differences in

IOVS, April 1998, Vol. 39, No. 5 T Cell Traffic in EAU 757

150 -

75

12 24 48 72 96

Hours After Cell Transfer120

FIGURE 2. Measurements of extravascular PKH26-labeled con-canavalin A-stimulated syngeneic spleen cells in the retina afterintravenous transfer of 10 X 106 cells. Each bar represents themean number of four to six enzyme-dissociated retinas exam-ined.

the number of cells from Con A-activated or S-Ag-activatedsources at each time point were less than 20%, and thedifferences between the left and right eyes, with rare ex-ceptions, were less than 5% (data not shown). The meannumber of SP35 cells present at each time point was veryslightly higher, compared with the number of nonspecificblasts, until 96 hours, when the reverse was seen (Fig. 3).This difference is not statistically significant, probably be-cause, cytologically, all SP35 cells were blasts in an activatedstate, whereas Con A-stimulated normal Lewis spleen cellswere only 90% to 95% blast cells at the time of transfer. At96 hours, a larger population of nonactivated but labeledCon A cells come into the retina than do labeled S-Ag-specific cells, accounting for the increased number of non-specific cells.

The state of activity of injected cells is critical. Whenhighly activated SP35 T cells were taken from their frozen stateand placed in culture without antigen or mitogenic stimulus for48 hours, they decreased in size but remained viable. AfterPKH26 labeling, 10 X 106 of these cultured SP35 cells wereinjected into normal Lewis rats and the retinas examined at 24hours. Labeled cells of this rested population (Fig. 3; hatchedbar) were only rarely detected in retinas at the time of the firstpeak of the SP35 blast cell incursion.

The ability of PKH26-labeled lymphocytes to produce cy-tokines and to undergo mitosis was tested with mouse and ratspleen cells stimulated by Con A. PKH26 labeling did notinhibit thymidine incorporation after lectin stimulation (Table1). In these experiments, normal BALB/c mouse spleen cellswere used, labeled with PKH26 by the standard techniqueadvised by the instruction sheet (Sigma), or treated with halfthis concentration of dye, or treated as though they were to belabeled using the buffer system provided for label dilution butwithout the label itself. In all cases, the stimulation index wasnot substantially decreased. The generation of IL-2 and IL-4 wastested with supernatants of labeled or unlabeled murine Con A-treated lymphocytes and Lewis rat labeled or unlabeled lym-phocytes for interferon-y. In the production of cytokines, therewas no substantial evidence of the label's interfering with theirmanufacture (Fig. 4).

300 r

is 200DC

0_

<» 150

II I ConA Activated Spleen Cells

| S Antigen Specific T Cells

* ^ a s Antigen Specific T Cells post 48 h culture

i4 8 12 24 48 72 96

Hours After Cell Transfer

FIGURE 3. Measurements of labeled cells in the retina of Lewisrats administered PKH26-labeled SP35 cells plus unlabeled con-canavalin A-stimulated syngeneic spleen lymphoblasts, or ad-ministered PKH26-labeled syngeneic lymphoblasts plus unla-beled SP35 cells. Injections contained 10 X 106 cells of eachpopulation. The mean number of labeled cells from six to eightretinas are depicted at each time point. Note the paucity oflabeled quiescent (rested) SP35 cells in retinas examined 24hours after inoculation.

HistopathologyThe development of pathology after adoptive transfer of SP35cells required examination of the retina and other ocular tis-sues, including the anterior segment and choroid, beginningearly and continuing sequentially after transfer of SP35 T cells.No retinal inflammation was detected during the first 72 hoursafter SP35 T cell transfer. A section from the retina of an animalkilled at 48 hours shows a completely normal retina cut incross-section (Fig. 5A). This section is representative of theentire retina from all animals killed at this time. At 80 hours,mononuclear cells adherent to the endothelial surface areshown in a retinal venule. Peri vascular infiltration in the nervefiber layer extends to the subjacent inner plexiform layer, theinner nuclear layer, and the outer plexiform layer directlybeneath this vessel (Fig. 5B). After this time, retinal pathologychanged dramatically. At 96 hours after cell transfer, there was

TABLE 1. Mitogenic Response to Lectin Stimulus byControl and Labeled Lewis Rat Spleen Cells

Label

—PKH26*PKH26*PKH26fPKH26f(0)*(0)*

Lectin

—

ConA—ConA—ConA—ConA

Counts per Minute(mean x 10~2)

151820

101700

132400

111800

StimulationIndex

121

170

185

164

* PKH26 at 2 X 1O~6 M concentration.t PKH26 at 1 X 10"6 M concentration.i Spleen cells treated but not exposed to PKH26 dye.

758 Prendergast et al.

IL - 2 fL - 4

1000 -

IFN-Y

O 500 -

Leclin Stimulus 0 +PKH-26 Label 0 0

FIGURE 4. Cytokine production of PKH26-labeled or control mu-rine spleen cells (interleukin-2 and interleukin-4) and Lewis rat spleencells (interferon-y) cultured with or without concanavalin A.

IOVS, April 1998, Vol. 39, No. 5

disruption of many layers and edema was present within thenerve fiber layer and surrounding capillaries penetrating theinner plexiform layer to the inner nuclear layer (Fig. 5C). Itshould be noted that there was no evidence of inflammatorycell infiltration at this time within the photoreceptor layer. Thepigment epithelium and underlying choroid were uninvolvedup to this point in our adoptive EAU model. The retina shownin Figure 5D is representative of retinas from animals killed 120hours after SP35 cell transfer and shows massive destruction inthe retina and infiltration of mononuclear cells within thephotoreceptor layer. Small vessels within the outer plexiformlayer were surrounded by areas of destruction that extended tothe photoreceptor layer, subjacent to nuclear layer destruc-tion. Most infiltrating cells were mononuclear in these laterperiods. An even band of lightly stained material was presentbetween the inner limiting membrane and the margin of thevitreous (Figs. 5B, 5C, 5D). Edema from the retina appeared toseparate the vitreous from the retinal surface. At 120 hours, as

FIGURE 5. Retinal inflammation after transfer of 10 x 106 SP35 cells. (A) No evidence of disease is seen 48 hours after T celltransfer. Original magnification, X280. (B) Eighty hours after T cell transfer, adherent mononuclear cells are present within aninner retinal venule. Perivascular infiltratation of mononuclear and rare polymorphonuclear cells extends though the outerplexiform layer margin. Original magnification, X35O. (C) Edema and mononuclear cell infiltrations are present in the ganglioncell layer and focally in the inner and outer nuclear layers in which compression of cells is apparent 96 hours after T cell transfer.Original magnification, X280. (D) One hundred twenty hours after T cell transfer, retinal inflammation is present in all retinallayers. Dilated capillaries in the outer plexiform layer are surrounded by acellular necrotic zones. Mononuclear cell infiltrate ispredominant in the rod outer segments, contiguous to the pericapillary destruction of the outer nuclear layer. Originalmagnification, X200.

IOVS, April 1998, Vol. 39, No. 5 T Cell Traffic in EAU 759

FIGURE 6. Confocal microscopic computer-generated images of a paraformaldehyde-flxed,unstained retinal whole mount obtained 96 hours after transfer of 10 X 106 PKH26-labeledSP35 cells. (A) A dividing capillary contains an elongated red blood cell visualized by phasemicroscope. (B) The presence of a PKH26 (red>labeled SP35 cell is appreciated in the samearea when laser illumination is used. This cell appears to be breaking out of the capillarylumen. The adjacent red blood cell measures approximately 9 /win in its longest dimension.

inflammation became more severe, this loculated space wasoccupied by mononuclear inflammatory cells. Examinationwith scanning confocal laser microscope of a retinal wholemount taken at 96 hours (Fig. 6) showed the transit of aPKH26-labeled SP35 T cell from the capillary lumen to thesurrounding nerve fiber layer.

Inflammatory disease in the iris and ciliary bodies pro-gressed in a different time course than did retinal disease. InFigure 7, anterior segments are shown at 48, 90, and 120 hoursafter adoptive transfer. It is clear that inflammation of theanterior segment precedes retinal inflammation. The vast ma-jority of cells comprising this infiltrate are mononuclear cells,with principally macrophage characteristics. At no time wasthere evidence of an initial inflammatory response in the cho-roid in any of the examples from 24 to 96 hours after SP35 celltransfer. However, at 120 hours, there was massive infiltrationof the choroid (data not shown).

In summary, findings in histologic examination indicatethat the anterior segment, including the iris and ciliary body,was the first ocular structure to become involved with signif-icant inflammation, 48 hours after cell transfer. The onset ofretinal inflammation at 80 hours was signaled, by adhesion ofinflammatory cells to vascular endothelium and infiltration inthe innermost perivascular regions of the nerve fiber layer.Inflammation rapidly progressed through the inner and outernuclear layers, and only in the last stages affected rod outersegments. The choroid was the last structure to be involved,beginning only after involvement of the outer retina. At notime was there evidence of disease in Lewis rats that received10 X 106 Con A-induced syngeneic blast cells.

DISCUSSION

According to the data provided by our experiments on adop-tive transfer of EAU, syngeneic S-Ag-specific T cells enter the

retina within hours after intravenous inoculation. Transfer ofhighly fluorescent labeled S-Ag-speciric T cell blasts and studyof the entire extravascular end organ has made it possible tomeasure the total number of T cells at the target site in earlyphases of EAU. These cells exhibit a biphasic accumulation inthe retina, the first peak occurring 24 hours after SP35 T celltransfer. The number of these cells waned during the next 2days, but they reaccumulated rapidly after 80 hours, concur-rent with the onset of destructive retinal inflammation. Ofparticular interest was the finding that control populations ofCon A-activated syngeneic spleen cell blasts showed preciselythe same initial entry time and concentration in the retina asdid the first phase of SP35 EAU-specific cells. Control cellspeaked in the retina at 24 hours after transfer and decreased atthe same rate as SP35 cells, but this decrease persisted for theentire length of time of the experiment. Only rare cells werepresent at 120 hours. There was no evidence of inflammatorychanges as a result of control cell occupation of the retina.

In cotransfer experiments in which S-Ag-specific and ConA syngeneic T cell blast populations were injected at the sametime, no substantial difference in cell kinetics could be de-tected between either population in the retina. These dataestablish that there is no propensity for specific T cell entryinto the retina that depends on antigen recognition for pene-tration of vascular endothelium in the initial phase. The secondincrease in S-Ag-specific and control cells is probably causedby cytokine activation of vascular endothelium; the quantifi-able T cell blasts are only part of a major inflammatory change,primarily the influx of macrophages, which occurs at thistime.17"19 These data confirm and extend the observations onT cell-mediated cellular immune responses shown in otherexperimental model systems in which delayed hypersensitivi-ty,9 the induction of autoimmune inflammatory thyroiditis,'l

and allograft rejection10 have been examined.

760 Prendergast et al. IOVS, April 1998, Vol. 39, No. 5

FIGURE 7. Anterior segment inflammation after transfer of10 X 106 SP35 cells. (A) Mononuclear ceUs infiltrate tlie iris 48hours after cell transfer. The anterior chamber contains anbrinous exudate with mononuclear inflammatory cells. Orig-inal magnification, X200. (B) Inflammation of the iris, ciliarybody, and anterior chamber is more marked at 90 hours.Protein exudate and disruption of the corneal endotheliumoverlies the inflamed iris. Original magnification, X250. (C) At120 hours, chronic inflammation has obliterated many featuresof normal iris morphology. Original magnification, X25O.

It should be noted that the number of labeled cells in theperipheral blood remained relatively stable through the 120-hour course, comprising 0.5% to 1% of the total circulatingmononuclear cells, as previously demonstrated.20 Of particularinterest is that injected T cells must be in an activated state topenetrate the vascular endothelium. When SP35 cells are al-lowed to rest in culture for 48 hours without antigenic ormitogenic stimulus, the number of such cells present withinthe retina at the time of the first peak decreases to less than 4%of that seen with an activated T cell population. It is unlikelythat labeled cells would shed dye and still be present withinthe retina during the time course of our experiment, in that ithas been shown that PKH26-labeled cells retain their dye for100 days in vivo.'3'14

After the initial peak at of labeled SP35 cells within theretina at 24 hours, there is a decline in population at 48 and 72hours. This may be because of reentry of these cells into thecirculation through the basilar portion of endothelium. It isalso conceivable that these specific T cells may undergo rapidapoptosis within the retina itself. Griffith et al.21 and Stuart etal.22 have noted that lymphocyte membrane Fas-Fas ligandinteraction, which leads to effector T cell death, is involved insurvival of corneal allografts21'22 and may well be responsiblefor the anterior chamber-associated immune deviation phe-nomenon.23 Nevertheless, it is clear from the results of ourexperiment and from results of those conducted in other lab-oratories that antigen-specific T cells are capable of inducingEAU6 and experimental autoimmune encephalomyelitis.2'* Inthe experiments reported here, we used 10 X 106 labeledlymphocytes, more than 25 times the blinding dose, to induceEAU. We observed fewer than 40 of these cells in the retina justbefore the onset of destructive inflammatory retinitis. How canso few T cells be responsible for inducing ocular disease whenapoptosis through Fas-Fas ligand interaction has been demon-strated? In A recent report by Zhang et al.,25 it was noted thatThl effectors of transgenic origin undergo rapid Fas-Fas ligand-mediated apoptosis.25 These investigators state that the trans-genic Th cells used in their experiments "do not have thedisadvantage of being selected in vitro by multiple restimula-tion, which is likely to select for cells that are relatively resis-tant to cell death." We were aided in our experiment by this"disadvantage." The SP35 cell line was derived from Lewis ratsimmunized with bovine S-Ag, and their lymphocytes wereintermittently stimulated with P35 peptide or with cytokines,including IL-2, provided by activated spleen cell supernatantsin vitro.6 Thus, it is probable that this system of specific T cellexpansion favors growth of an effector T cell line resistant toFas-Fas ligand-induced apoptosis.

Anterior segment inflammation occurs more rapidly af-ter S-Ag-specific T cell transfer than does retinal inflamma-tion. It is precisely this delay in retinal inflammation thatpermits the following hypothesis: In the anterior segment,class II antigen-presenting cells (APCs) are present, as hasbeen demonstrated by several observers, particularly Mc-Menamin et al.26"28 Inflammation within the anterior seg-ment occurs soon after the arrival of inflammatory CD4+

S-Ag-specific cells to that site. Preliminary studies indicatedthat labeled SP35 cells enter the iris at precisely the sametime that they enter the retina (data not shown), and inflam-mation of the anterior segment begins before 48 hours.Using in situ hybridization, Singh et al.29 observed mRNA forS-Ag in the iris and ciliary epithelium and the retina. Immu-nohistochemical analysis of these tissues also confirms thepresence of the arrestin family of molecules.30 Thus, al-though transferred SP35 T cell blasts were present at bothsites simultaneously, we ascribe the delay in retinal inflam-mation to the absence or inadequacy of constitutive APCs inthe normal Lewis rat retina. Results in recent experiments inour laboratory, including those of immunohistologic analysisof retinal whole mounts, and results reported by others havenot unequivocally established the presence of APC-bearingclass II major histocompatibility complex antigens in normalLewis rat retinas.31"33 Therefore, we hypothesize that thedelay in inflammation of the retina compared with time ofonset in the anterior segment in this model was caused by adelay in antigen presentation in the retina of the immuno-

IOVS, April 1998, Vol. 39, No. 5 T Cell Traffic in EAU 76l

genie S-Ag peptide to specific T cells already present at thatsite. Only after class II molecules are displayed by retinalAPCs can cytokines be generated for activation of retinalendothelium for the accumulation of destructive inflamma-tory cells.

The entry of effector T lymphocytes into central nervoustissue was studied by several investigators in the model systemof experimental autoimmune encephalomyelitis.24'34'35 Theseinvestigators noted that transfer of myelin basic protein-immu-nized Lewis rat effector T cells would induce disease in naivechimeric rat recipients.34 They also reported that control lec-tin-activated T cells could enter the spinal cord in essentiallythe same number and at the same time as specific T cells. In themethods used by Hickey et al.,24 multiple sections of spinalcord were analyzed to enumerate donor cells in an adoptivetransfer study of experimental autoimmune encephalomyelitis.In their experiment, an initial peak of T cells was recorded atapproximately 12 hours, which diminished during the suc-ceeding 3 days and was followed by rapid mononuclear cellinflammation, producing histologically definitive experimentalautoimmune encephalomyelitis. These results are consistentwith the data reported in the present study. Hickey et al. alsodemonstrated that nonactivated effector cells with the samemyelin basic protein specificity failed to enter the centralnervous system or to cause disease.34

In summary, the results of our experiment have demon-strated that there is no evidence for a blood-ocular or blood-retinal barrier to activated T cells. No evidence has beenaccumulated to show a greater propensity of T cells to leavethe bloodstream initially and enter the retina on the basis ofantigen specificity. The second phase of destructive retinalinflammation is dependent on in situ antigen recognition, be-cause it occurs only after transfer of specific, but not nonspe-cific, T cells. After adoptive T cell transfer, immunologicallyinduced inflammation begins in the anterior segment wellbefore it begins in the retina. We hypothesize that the temporaldifference between the anterior segment and retinal inflamma-tion is caused by the absence of mature APCs in the normalLewis rat retina. The acquisition of major histocompatibilitycomplex class II-specific peptide-bearing molecules on retinalAPCs is necessary for T cell interaction and cytokine produc-tion which, in turn, activates retinal vascular endothelium. Thiscritical step is required for the influx of nonspecific inflamma-tory cells and retinal destruction. Experiments are currently inprogress to identify retinal APCs in this model of adoptive EAU.

References1. Wacker WB. Experimental allergic uveitis. Investigations of retinal

autoimmunity and the immunopathogenic responses evoked. In-vest Ophthalmol Vis Sd. 1991;32:3119-3128.

2. Caspi RR. Basic mechanisms in immune-mediated uveitic disease.In: Lightman SL, ed. Immunology of Eye Diseases. Lancaster, UK:Kluwer; 1989:61-86.

3. Gery I, Streilein JW. Autoimmunity in the eye and its regulation.Curr Opin Immunol. 1994;6:938-945.

4. Caspi RR, Roberge FG, McAllister CG, et al. T cell lines mediatingexperimental autoimmune uveoretinitis (EAU) in the rat. / Immu-nol. 1986; 136:928-933.

5. Gregerson DS, Obritsch WF, Fling SP, Cameron JD. S-antigen-specific rat T cell lines recognize peptide fragments of S-antigenand mediate experimental autoimmune uveoretinitis and pineali-tis. J Immunol. 1986;136:2875-2882.

6. Savion S, Oddo S, Grover S, Caspi RR. Uveitogenic T lymphocytesin the rat. Pathogenicity vs lymphokine production, adhesion mol-

ecules and surface antigen expression. J'Neuroimmunol, 1994;55:35-44.

7. Greenwood J, Howes R, Lightman S. The blood-retinal barrier inexperimental autoimmune uveoretinitis: Leukocyte interactionsand functional damage. Lab Invest. 1994;70:39-52.

8. Wang YF, Calder VL, Greenwood J, Lightman SL. Lymphocyteadhesion to cultured endothelial cells of the blood retinal barrier.J Neuroimmunol. 1993;48:l6l-l68.

9. McCluskey RT, Benacerraf B, McCluskey JW. Studies on the spec-ificity of the cellular infiltrate in delayed hypersensitivity reactions.J Immunol. 1963;90:466-477.

10. Prendergast RA. Cellular specificity in the homograft reaction. /ExpMed. 1964;119:377-388.

11. Werdelin O, McCluskey RT. The nature and the specificity ofmononuclear cells in experimental autoimmune inflammationsand the mechanisms leading to their accumulation. / Exp Meet.1971;133:1242-1263.

12. Lightman SL, Caspi RR, Nussenblatt RB, Palestine AG. Antigen-directed retention of an autoimmune T cell line. Cell Immunol.1987;110:28-34.

13. Horan PK, Slezak SE. Stable cell membrane labelling. Nature.1989;340:l67-l68.

14. Slezak SE, Horan PK. Fluorescent in vivo tracking of hematopoieticcells. Blood. 1990;74:2171.

15. Laver NM, Robison WG, Pfeffer BA. Novel methods for isolatingintact retinal vascular beds from diabetic humans and ani-mal models. Invest Ophthalmol Vis Sd. 1993;34:2097-2104.

16. James SP. Measurement of proliferative responses in cultured lym-phocytes. In: Coligan JE, Kruisbeck AM, Margulies DH, ShevachEM, Strober W, eds. Current Protocols in Immunology. NewYork: Wiley; 1994;7.10.1-7.10.10.

17. Whitcup SM, DeBorge LR, Caspi RR, Harning R, Nussenblatt RB,Chan CC. Monoclonal antibodies against lCAM-1 and LFA-1(CDlla/CD18) inhibit experimental autoimmune uveitis. Clin Im-munol Immunopathol. 1993;67:l43-150.

18. Picker LJ, Kishimoto TK, Smith CW, Warnock RA, Butcher EC.ELAM-1 is an adhesion molecule for skin-homing T cells. Nature.1991 ;349:796-798.

19- Springer TA. Traffic signals for lymphocyte recirculation and leu-kocyte emigration: the multistep paradigm. Cell. 1994;76:301-3l4.

20. Prendergast, RA, Coskuncan NM, McLeod DS, Lutty GA, Caspi RR.T cell traffic and the pathogenesis of experimental autoimmuneuveoretinitis. In: Nussenblatt RB, Whitcup SM, Caspi, RR, Gery I,eds. Advances in Ocular Immunology. New York: Elsevier; 1994:59-62.

21. Griffith TS, Brunner T, Fletcher SM, Green DR, Ferguson TA. Fasligand-induced apoptosis as a mechanism of immune privilege.Science. 1995;270:1189-1192.

22. Stuart PM, Griffith TS, Usui N, Pepose JS, Yu X, Ferguson TA. CD95ligand (FasL)-induced apoptosis is necessary for corneal allograftsurvival. / Clin Invest. 1997;99:396 - 402.

23- Griffith TS, Ferguson TA. The role of FasL-induced apoptosis inimmune privilege. Immunol Today. 1997; 18:240-244.

24. Hickey WF, Kimura H. Perivascular microglia are bone marrowderived and present antigen in vivo. Science. 1988;239:290-292.

25. Zhang X, Brunner T, Carter L, et al. Unequal death in T helper cell(Th) 1 and Th2 effectors: Thl, but not Th2, effectors undergorapid Fas/FasL-mediated apoptosis. / Exp Med. 1997; 185:1837-1849.

26. McMenamin PG, Crewe J, Morrison S, Holt PG. Immunomorpho-logical studies of macrophages and MHC class II-positive dendriticcells in the iris and ciliary body of the rat, mouse and human eye.Invest Ophthalmol Vis Set. 1994;35:3234-3250.

27. Steptoe RJ, Holt PG, McMenamin PG. Demonstration of the imrnu-nostimulatory capacity of dendritic cells isolated from the rat iris.Immunology. 1995;85:630-637.

28. Butler TL, McMenamin PG. Resident and infiltrating immune cellsin the uveal tract in early and late stages of experimental autoim-muune uveoretinitis. Invest Ophthalmol Vis Sci. 1996;37:2195-2210.

762 Prendergast et al. IOVS, April 1998, Vol. 39, No. 5

29- Singh AK, Kumar G, Shinohara T, Shichi H. Porcine S-antigen.cDNA sequence and expression in retina, ciliary epithelium andiris. Exp Eye Res. 1996;62:299-308.

30. Nicolas C, Ghedira I, Faure JP, Mirshahi M. /3-arrestin relatedproteins in ocular tissues [ARVO Abstract]. Invest Ophthalmol VisSci. 1997;38:S320. Abstract nr 1494.

31. Prendergast RA, Caspi RR, Tarrant TK, Iliff CE. Class II positiveantigen-presenting cells in retina and iris in T cell-inducedEAU [ARVO Abstract]. Invest Ophthalmol Vis Sci. 1997;38:SI 159. Abstract nr 5407.

32. Forrester JV, McMenamin PG, Holthouse I, Lumsden L, LiversidgeJ. Localization and characterization of major histocompatibility

complex class II-positive cells in the posterior segment of the eye:Implications for induction of autoimmune uveoretinitis. InvestOphthalmol Vis Sci. 1994;35:64-77.

33- Yang P, de Vos AF, Kijlstra A. Macrophages in the retina of normalLewis rats and their dynamics after injection of lipopolysaccharide.Invest Ophthalmol Vis Sci. 1996;37:77-85.

34. Hickey WF, Hsu BL, Kimura H. T-lymphocyte entry into the centralnervous system. / Neurosci Res. 1991 ;28:254 -260.

35. Bauer J, Huitinga I, Zhao W, Lassmann H, Hickey WF, Dijkstra CD.The role of macrophages, perivascular cells, and microglia in thepathogenesis of experimental autoimmune encephalomyelitis.Glia. 1995;15:437-446.