British Journal of Ophthalmology 1995; 79: 834-840

ORIGINAL ARTICLES - Laboratory science

Flow cytometric identification of a minoritypopulation ofMHC class II positive cells inthe normal rat retina distinct fromCD45lowCD1 lb/c+CD4low parenchymal microglia

A D Dick, A L Ford, J V Forrester, J D Sedgwick

AbstractAims-This study aimed to isolate andclassify by flow cytometry, the cell surfacephenotype of microglia in the normal ratretina with a view to identifying putativeantigen presenting cells (APC) within theretina, which has to date not been possibleby immunohistochemistry.Methods-Normal rat retinal microgliawere isolated and classified using a modi-fication of an isolation technique employ-ing graduated Percoll density gradient cellseparation and flow cytometric pheno-typic criteria used for CNS microglia.Results-Retinal microglia can be definedby flow cytometry on the basis of theirCD45oWCDllb/c+CD4lOw cell surfaceexpression. Constitutive MHC class IIexpression in the normal rat retina wasconfined almost exclusively to a veryminor population of cells expressingneither low (microglia) nor high levels ofCD45. Three colour flow cytometricanalysis confirmed that these MHC classII positive cells were ED2+.Conclusions-Using this sensitive isolationtechnique we have identified the cell sur-face characteristics of ramified, residentmicroglia, and found that they do notconstitutively express MHC class II. Thereis, however, constitutive MHC class IIexpression on a phenotypically distinctpopulation of cells (CD4510wIi&hED2+). Wepropose these cells are the counterpart ofthe perivascular macrophages found inthe CNS which present antigen toextravasating T cells, although their exactretinal location can only be confirmed byimmunohistochemical analysis. The roleofparenchymal microglia as APC remainsundefined. Future isolation of microgliaand putative perivascular cells using thistechnique will help identify the role thesecells play in the initiation and perpetuationofimmune responses within the retina.(Br_J Ophthalmol 1995; 79: 834-840)

Experimental autoimmune uveoretinitis(EAU) is an established model for auto-

immune endogenous posterior uveitis inhumans.' 2 This model of intraocular inflam-mation is mediated by MHC class II restrictedCD4+ T cells which infiltrate the retina earlyin the course of the disease and, histologically,can be seen entering the retina through boththe inner retinal vessels and choroid, ultimatelycausing destruction of the photoreceptor outersegments where the major uveitogenic antigensare found.1-3 Local antigen presentation byMHC class II positive antigen presenting cells(APC) is required to activate systemicallyprimed CD4+ T cells and initiate the autoim-mune inflammatory response. Recently therehave been reports of inducible MHC class IIantigen expression on non-haemopoietic struc-tural elements within the eye including retinalpigment epithelium, vascular endothelium,and Muller cells.4-6 This is analogous to thesituation in the CNS where both astrocytes7and endothelial cells8 9 can be induced toexpress MHC class II molecules and processand present protein antigen to T lymphocytes,at least in vitro (although do not necessarilystimulate the T cells to divide8 10 11). However,there is no evidence that any of these cell typesfunction as APC in vivo.12 Morphologicalidentification of haemopoietically derivedcandidate APC within the eye have includedthe rich network of classic dendritic cellsexpressing high levels of MHC class II in thechoroidl3 and iris/ciliary body.'4 The presenceof perivascular cuffing in uveitis also argues forthe existence of retinal vessel associated APCto enable antigen recognition. However, nosuch APC has been definitively identifiedwithin the rat retina. Thus, to date, the majorcandidate APC within the rat retina areparenchymal ramified microglia, similar toCNS microglia. These express the CDllb/cmolecule, are thought to be bone marrowderived,'5 but are not MHC class II positive.'4In contrast, immunohistochemical analysis ofadult human retina has revealed constitutiveMHC class II expression on two classes ofmacrophage lineage cells, expressing low andhigh levels of the CD45 molecule and repre-senting resident microglia and perivascularmacrophage populations respectively,'6 17analogous to well defined CNS populations in

Centenary Institute ofCancer Medicine andCell Biology, Sydney,AustraliaA D DickAL FordJ D Sedgwick

Department ofOphthalmology,Medical School,AberdeenA D DickJ V Forrester

Correspondence to:Dr A D Dick, Department ofOphthalmology, Universityof Aberdeen, MedicalSchool, Foresterhill,Aberdeen AB9 2ZD.

Accepted for publication24 April 1995

834

on June 7, 2020 by guest. Protected by copyright.

http://bjo.bmj.com

/B

r J Ophthalm

ol: first published as 10.1136/bjo.79.9.834 on 1 Septem

ber 1995. Dow

nloaded from

Flow cytometric identification ofa minority population ofMHC class II positive cells in the normal rat retina

Retina CNS

LEW

1"

. ':'K... 2 (8-9%).. 3 (1 2%)

101 102 103 104

2

.,

255

PVG

03 310' 10

1 2(62/)~ ~ 1 2 (55.4%)

10 1010° 101 102 103 104 10 10 10 10 10

CD45

2

..

I

255:

0

2

.:::,

255

Forward scatter

Figure 1 CD45 expression and scatter profile of retinal microglia. Labelling of both retinal and CNS preparations withCD45 mAb subdivides the cells into three main populations (1-3). Number 2 denotes CD45'o population (microglia)whose scatter profile seen in the bottom three dot plots is similarfor retina (of both LEWand PVG) and CNS preparations.

rat.18 19 In the past, not only has immuno-histochemistry of the rat retina failed toeither distinguish populations of CD45+ cellsor demonstrate MHC class II expression butthere has been no in vitro or in vivo assess-

ment of the putative APC capacity of humanretinal microglia/macrophages. Here we

analyse the cell surface phenotype of retinalmicroglia in the rat by flow cytometry in an

attempt to classify putative APC. A minoritynon-microglial population was identifiedwhich has a similar phenotype to CNS peri-vascular macrophages and is the predominantMHC class II antigen expressing cell withinthe retina.

Materials and methods

ANIMALSInbred adult rats of the Lewis (LEW) andPVG strains were obtained from the AnimalResources Centre (Perth, Australia) or bredin local facilities (Blackburn Animal House,University of Sydney). Six-week-old (150-200 g) animals were used in the experimentswhich were conducted according to the codeof practice for experiments on animals andguidelines set by the National Health andMedical Research Council (Australia), whichadhered to the ARVO statement for the use ofanimals in ophthalmic and vision research.Each experiment was conducted on two or

three rats (four to six retinas) in order toachieve enough cell numbers for flow cyto-metric analysis and was repeated at least three

times except where indicated (n refers tonumber of experiments).

PREPARATION OF RETINAL AND CNS MICROGLIAAnimals were killed by carbon dioxide asphyx-iation then perfused slowly through the leftventricle by intracardiac catheterisation with250 ml of cold phosphate buffered saline (PBS)with heparin (250 U) to remove circulat-ing mononuclear cells. The retinas fromenucleated eyes were microscopically dissectedin cold PBS 0f02% (w/v)/bovine serum

albumin (BSA) taking care to remove the retinaposterior to the ciliary body/pars plana toexclude ciliary body/choroidal contamination,and then disrupted mechanically through a

stainless steel sieve. The dissociated materialwas washed by centrifugation at 400x g for 10minutes in 50 ml of PBS/BSA. Subsequent iso-lation steps were modified from the previouslydescribed technique for CNS microglia.18 19 Inbrief, a 2 ml cell suspension of dissociatedretinas in PBS/BSA (4-8x 105 cells/retina) was

added to a graduated Percoll (Pharmacia)gradient prepared in a 10 ml conical bottomedtube. This consisted of 2 ml steps of Percoll at1-21 g/ml, 1-088 g/ml, 1b072 g/ml, and1-03 g/ml. This gradient was centrifuged at1250X g for 20 minutes at 20°C with slowacceleration time and no brake. Cells were

collected from both the 1-072 and 1 088 g/mllayers although the majority of cells were

present at 1 072 g/ml. Viability was determinedby trypan blue exclusion after a further wash inPBS/BSA. At this stage 1-2X105 cells/retina

*- ALT0

01)+o

0)

z

04

103

102

101

10 -Xt*- :*-..I..... . . . .... .

100

255

a)

01)C0)

UI)

v0

835

on June 7, 2020 by guest. Protected by copyright.

http://bjo.bmj.com

/B

r J Ophthalm

ol: first published as 10.1136/bjo.79.9.834 on 1 Septem

ber 1995. Dow

nloaded from

Dick, Ford, Forrester, Sedgwick

Retina CNS

LEW104

3103

2

101 102 103 104

PVG

10

10° 10 10 10 10

CD45Figure 2 CDllb/c+ expression on CD4510w retinal microglia. Two colourflow cytometric analysis of both retina andCNS showing that the entire CD45'O population (8% in retina and 39% in CNS ofgated populations, arrow) express theCDllb/c molecule (complement receptor 3/P150-95) known to be present on microglia. This further differentiates theCD45Shgh cell population (population 3, Fig 1) which is spl* into CD 1b/c+ and CDl lb/c- ceUls.

were isolated after density centrifugation foranalysis. CNS microglia were isolated as pre-viously described,18 19 but only from Lewisstrain rats. The significant difference betweenthe CNS and retinal procedures is that theformer requires collagenase/DNAse digestionto enhance cell yield and a low density gradientto remove myelin. The non-myelinated retinado not require these steps.

MONOCLONAL ANTIBODIES (mAb) AND FLOWCYTOMETRYMouse mAb specific for rat cell surface markerswere originally generated from hybridomasprovided by the MRC, Cellular ImmunologyUnit, Oxford which include MRC OX1 (anti-CD45), MRC OX42 (anti-CD1lb/c), W3/25(anti-CD4), MRC OX6 (anti-MHC class II,

I-A), MRC OX21 (anti-human C3bi and notrat cells), and R73 (anti-a,B TCR) from DrThomas Hiunig, Wiirzburg, Germany (seeSedgwick et a118 and Ford et al19 for crossreference details). ED2 mAb20 was kindlyprovided by Dr Christine Dijkstra (Amster-dam, the Netherlands). Phycoerythin (PE)conjugated anti-CD45 was purchased directlyfrom Serotec, UK. mAb MRC OX21 was usedas an isotype control for flow cytometry. mAbwere prepared as tissue culture supematant(TCS) and as ascites (ED2). For direct stain-ing, mAb were purified and conjugated tobiotin or fluorescein isothiocyanate (FITC).TCS mAb was detected with rat absorbedreagent, FITC conjugated sheep F(ab')2 anti-mouse Ig (Sigma, USA) and biotinylated mAbwere detected either with streptavidin-PE(SA-PE, Caltag, USA) for two colour flowcytometry or streptavidin-Texas Red (SA-TxR,Molecular Probes Inc, USA) for three colourflow cytometry.

Staining was performed as previouslydetailed18 19 using FACS buffer (PBS/BSA/10mM NaN3) for washes. All reagents, buffers,and incubations were kept and performed at4°C. In brief, for two colour labelling a cellsuspension of between 104 and 105 cells wasincubated sequentially with TCS mAb or a1:50 dilution of ED2 ascites, FITC con-jugated sheep F(ab')2 anti-mouse Ig in the

presence of 10% normal rat serum (NRS),blocked with 10% NRS/normal mouse serum(NMS), biotinylated second mAb in thepresence of both NMS and NRS and finallySA-PE. For three colour labelling, stepswere the same except the second mAb wasdirectly conjugated to PE (OX1-PE) andbiotinylated mAb were detected with SA-TxR.Cells were fixed with a final 1% (v/v) formalinsolution and stored at 4°C until analysiswas performed. Negative controls (identicalspecies and isotype) and single fluorochromecontrols were performed to allow accuratebreakthrough compensation. Two colourphenotypic analysis of both retina and CNSpreparations were performed using FACScan(Becton Dickinson, USA). Three colourphenotypic analysis was performed using aFACStarPIus sorter (Becton Dickinson, USA)and LYSIS II acquisition and analysis soft-ware. A total of 10 000 events were collectedand both microglial gates and instrumentvariables were set according to their forwardand side scatter characteristics. Analysis offluorescence was performed after further back-gating to exclude dead cells and aggregates.Thus, in most cases, profiles represent lessthan 50% of collected events.

Results

SCATTER PROFILES AND CD45 EXPRESSIONCells were isolated from the retina of two ratstrains, the albino LEW and the PVG.Staining of retinal cells as well as those iso-lated from the normal LEW rat CNS with anantibody specific for the leucocyte commonantigen (CD45) found only on cells ofhaemopoietic origin, revealed three majorpopulations (top, Fig 1). The characteristiccell surface phenotype of microglia is basedon their CD45 expression.'8 19 There are twopopulations of CD45 expressing retinal cells;CD451ow microglia (population 2), account-ing for 6-2-9% (n=4) of gated cells andCD45high cells (population 3) accounting for0-6-19% of gated cells (n=4). In CNS boththese populations are present at a higher per-centage of the total and this can be attributed

AL

U

100 i100

836

on June 7, 2020 by guest. Protected by copyright.

http://bjo.bmj.com

/B

r J Ophthalm

ol: first published as 10.1136/bjo.79.9.834 on 1 Septem

ber 1995. Dow

nloaded from

Flow cytometric identification of a minority population ofMHC class II positive cells in the normal rat retina

io

I-)

;. . 3 (11%)

;5 4 (8%)

2101

101

10° 101 102 103 104

1(2%)

CD45Figure 3 CNS derived CD45high cells can be proportionally reduced by removal ofmeninges and choroid plexus. (A) Two colour labelling with CD45 and CDl l b/c ofwhole CNS preparation reveals a characteristic four population profile (see Fig 2).(B) Analysis of tissue after removal of meninges, tissue dissection away from theventricles, and then needle biopsies taken from the remaining tissue shows an increase inthe microglial yield (population 2) with a corresponding reduction in CD45high cells(populations 3 and 4).

to the significantly lower proportion of non-

haemopoietic (CD45-, population 1) cellsrecovered in the CNS compared with theretina. In retinas, population 1 is most likelyto be neural retinal elements and bipolar andganglion cells because of their scatter profileand lack of CD45 expression. Comparablecells are not recovered with similar efficiencyfrom the CNS and we believe this is due tothe fact that many would be lost during theinitial low density purification step to remove

myelin (see methods). Cell size may also bea relevant factor. Forward and side scatter(Fig 1, bottom) are a measure of cell size andheterogeneity respectively, and illustrate therelatively small size of population 2(microglia). They also show population 2 inall preparations has very similar characteris-tics. Additionally, when analysing cells by thistechnique, electronic forward and side scattergates are placed around the major populationof cells to exclude dead (small) or aggregated(large) cells and population 1 cells in theretina obviously fall into a gate size that is notsignificantly different from microglia, whereaslarger motor neurons in the CNS, forexample, may be excluded. It is not possible,nor is it the intention of this study to compareprecisely the populations isolated from retinasand CNS, but the latter was included here inview of the substantial characterisation ofmicroglia and other associated macrophagesalready documented for this tissue.18 19 Inparticular, the whole CNS preparation con-tains relatively higher numbers of CD45highcells (Fig 1). In both CNS and retinas, thispopulation can be further subdivided intoCD llb/c positive (other macrophages) andnegative populations (Fig 2). If the CNS isprepared by dissecting frontal lobe away fromthe ventricles, followed by removal ofmeninges, including dura mater, arachnoid,and pia mater, and needle biopsy of theremaining tissue the CD45high population issubstantially reduced (Fig 3). Whether theremaining CD45high cells are exclusivelyperivascular/parenchymal and a parallel

population to the CD45high cells of the retinais difficult to determine and for this reason wehave not further characterised their pheno-type. Figure 2 also shows that all the CD451owmicroglial population are CD1 1b/c+ (arrow),a monocyte/granulocyte marker which isknown to be expressed by both CNS and reti-nal microglia.14 18 19 There was no apparentspecies difference between LEW and PVGwith respect to proportion of cell populationsand cell surface phenotype. The slightly dif-ferent levels of CD45 expression in LEW andPVG retinal microglia represents the use ofaltered fluorescence gain settings only.

PHENOTYPIC CHARACTERISATION OF RETINALMICROGLIATwo colour flow cytometric analysis was usedto assess CD4, MHC class II, and ED2 cellsurface expression on isolated retinal CD4510w(microglial) and CD45higf (mixed) popula-tions. CD4/CD45 staining (Fig 4A) subdividesthe populations into four parts (1-4).Population 4 (CD45-) are CD4- as expected.Populations 1-3 which are all CD45+ are acomplex mixture of cells. Microglia (CD4510w,population 3) are CD4lOw with average CD4levels around two times above background.Population 2 is clearly CD4+ although levelsvary, particularly in the CNS, where there is amixture of CD4+ T lymphocytes (CD4high,around 20 times above background, arrow)with the remainder being macrophages whichexpress slightly less CD4 (5-6 times abovebackground, arrowhead). Most ofpopulation 2in the retina are also macrophage lineage cellson this basis and ot, T cell receptor positivecells are difficult to detect in retinal prepara-tions because of their very low numbers (datanot shown). Population 1 is CD4- whichwould include some CD8+ T lymphocytesparticularly in the CNS. Flow cytometricanalysis reveals a very small population ofMHC class III cells in the retina (036-042%of gated cells, n=3: Rl, Fig 4B) which is sub-stantially less than the equivalent population ofcells seen in the whole CNS (6-3% of gatedcells). In both retina and CNS, MHC class IIexpressing cells are positioned midwaybetween CD45high and CD451ow populations.In a similar location, a small population ofED2+ cells (0 35-0 41% of gated cells, n=3)have been identified in the retina (R1; Fig 4C)and in the CNS (around 1O% of total).

RETINAL MHC CLASS II EXPRESSION ISCONFINED TO ED2+ CELLSConstitutive MHC class II expression in theCNS cannot be accounted for by ED2+ cellsalone (Figs 4B and C) and in view of the wholeCNS preparations used here (see above), thisis not surprising. Moreover, published dataexamining MHC class II expression in the ratCNS at the immunohistological level, indicatethat not all MHC class III cells are ED21.22However, the similarity in percentage positiveMHC class II and ED2 expressing cells in theretina implied these were the same population.

837

on June 7, 2020 by guest. Protected by copyright.

http://bjo.bmj.com

/B

r J Ophthalm

ol: first published as 10.1136/bjo.79.9.834 on 1 Septem

ber 1995. Dow

nloaded from

Dick, Ford, Forrester, Sedgwick

A Retina

104

3~~~~03 1 .'t2.It

CNS

.

102

101

0AK ,10° 10123.I 4n'

102 103 10410°

CD4

Rl 0.37%

'.I..

2 103 10410 10 10

MHC clas

c

103 Rl 0.39%

1o2

101X ' -*.

10010° 10 102 103 104

ED2

Figure 4 CD4, MHC class II, and ED2 expressionX(CD4Skiw microglia) are CD4l*w. The true negativecoby the negative ED2 staining in (C). Both anti-CD4IgGI isotype. Thus the shift ofpopulation 3 ta the righcontains some CD4+ T cells in the CNS (arrow) with(arrowhead). The CD45- population 4 are CD4- (Bcels (Rl) which are CD45+ MHC class I+ and CD,

To confirm this,was required andin Figure 5. By ga

CD45+ED2- polthese for MHC cthat the majorityMHC classII+ wcells are. The lattw

Substantial Mfound in the u

tamination of retiican be excluded a

fully dissected posalso a frill of retiinerve head prever

the retina was re

easily identified,I

. 3

E E:.

PVG, as separate from the choroid because ofthe presence of retinal pigment epithelium. (3)Choroidal MHC class III cells are, in themajority ED2 negative,13 a phenotype we didnot identify in the pure retinal preparations.

DiscussionThe eye has been considered an immune

4 privileged site because of the lack of dedicatedlymphatics and the absence of professional

10 10 10 10 104 APC. With the recent evidence that the uveahas cells which at least morphologically andphenotypically resemble cells known to act ascompetent APC,13 1421 the question arises as towhether the retina itself contains any immuno-competent cells which may play a role in the

D4 *. inflamed retina. We have defined by flow cyto-metric analysis, rat retinal microglia which have

o3 . . 6.3% markers in common with haemopoietic derived..,.. . ; ;/7.cells, in particular, CD45 and CD1 lb/c. This

2 ..tt).;, . ' resident microglial population has a similarO t4... -....jJ phenotype (CD45IowCDl lb/c±CD4l0w) toCNS microglia.18 19 Retinal microglia can be

D : _ p distinguished from another minor populationof cells within the rat which have not been

0o previously described and these express the10 10 102 103 104 ED2 marker20 and intermediate levels of

CD45 (CD45Iow/wh). Most significantly, three;S 1I colour flow cytometric analysis provided

unequivocal evidence that the majority ofretinal ED2+ cells areMHC class II+, clearlydistinguishing them from ED2+ cells in the

34 adjacent choroid where most are MHC classII negative.13 Retinal derived ED2+ cells

33. s -' R1 1% account for almost all the MHC class II stain-ingin the normal rat retina.Two aspects are of note. Firstly, the ED2+

cell population is present at a rate of approxi-mately 1:250 cells, and this assumes an equal

.s:s*>z<] .; efficiency of isolation of all cell types. In fact, itis likely that highly processed cells includingo0;;;; Miller cells and many neuronal elements may

10° 101 102 103 10 be isolated with lower efficiency, so the 1:250Po frequency is probably an overestimate. Thus,

the paucity of ED2+ cells may account for theon CD45+ cells. (A) Population 3 documented absence ofMHC class II stainingmtrolfor this population is exhibited in retina by immunohistochemistry, a tech-and anti-ED2 mAb are of the mouse nique13 14 which is unable to sample entireIt in (A) is real. Population 2 tissue as we have done here. Secondly, pre-the remainder being macrophages) and (C). Minor populations of vious reports on isolation, phenotype, and cul-45+ED2+ respectively. ture of microglia from the CNS and retina

which have proposed the microglia as a candi-three colour flow cytometry date APC within these tissues have usedthe results of this are shown markers to define the microglial populationiting aroundCD45+ED2+ or (for example, CD1 b) which are also sharedpulations and then assessing by some peripheral blood leucocytes as well as,lass II expression it is clear the minority perivascular cell population. It is(>85%) of ED2+ cells are clearly difficult to distinguish between resident

rhile few, if any of the ED2- microglial cells (parenchymal), perivascularer includes all microglia. cells, other associated macrophages, and those.HC class II expression is cells infiltrating from the blood by means otheryvea of the eye.'3 14 Con- than flow cytometry.nal preparations by this tissue The animal model of posterior uveitis,s follows: (1) retina was care- EAU, demonstrates cells which enter thesterior to the ciliary body and retina from the choroid, where a large inflam-na was left around the optic matory response is seen and also the inneriting choroidal seeding when retinal vessels where large perivascular cuffsmoved. (2)The retina was develop. The severity of choroiditis and innerarticularly in thepigmented retinal vasculitisvaries with type and dose of

B

ALftIta)

838

I.

14

11

la

la

10

10

4=10 =

3I.- 110 .. , ...

1(

1(

on June 7, 2020 by guest. Protected by copyright.

http://bjo.bmj.com

/B

r J Ophthalm

ol: first published as 10.1136/bjo.79.9.834 on 1 Septem

ber 1995. Dow

nloaded from

Flow cytometric identification of a minority population ofMHC class II positive cells in the normal rat retina

104

mt

aI

100-

0I100 10 102 103 104

10°EDo1o2 10ED2

CD45+ECD2-(Rl)A CD45+ED2+(R2)

101

MHC class 1.Figure 5 Most constitutive MHC class II expression in the normal retina is on ED2cells. Retinal cells were stainedfor three colourflow cytometry for the markers CD45,and MHC class II. The dot plot (left) is a representative example of expression ofEDversus CD45 by two colourflow cytometry. The MHC class II expression of twopopulations, Rl (CD45+ED2-) and R2 (CD45+ED2+) was assessed byforming anregions as shown.

antigen given1 2 and between activelypassively induced disease.3 As we

mentioned, it has been suggested thatchoroidal MHC class II+ cells act as local.for the circulating autoreactive T cells wreach the choroid, but as yet no cell hasidentified which may act as an APC for Twhich have entered the retina throughinner retinal vessels. We have shown tismall population of MHC class II+ cellpresent in perfused retinal preparationboth LEW and PVG rats and that these cethe retina are confined to the ED2+ phtype. We propose these are the counte:of the perivascular cell (or perivas(macrophage) found in the CNS,12 22 alth4their location can only be confirmecimmunohistochemical analysis. Studierirradiation chimeric rats have shown thatperivascular cells can present autoantigencells in vivo,23 and it is likely the ED2+detected here may similarly act as APC wthe retina. Nevertheless it remains for x.isolate ED2+ cells as well as retinal micrto establish their APC role. Such studie,obviously difficult given the small nunof cells involved but are currently Eattempted.Our findings are at least in part, consi

with previous immunohistochemical studiadult human retina16 17 which have derstrated MHC class II+ perivascular (

Unlike human retina, however, we couldidentify in rat, any pronounced MHC claexpression on resident microglia. Remicroglia have been implicated in induphagocytic, cellular, and vascular changldystrophic retina and also have phagocapacity within the subretinal space d&development.24 25 Their role as professiAPC and in immunoregulation has yet tdefined but flow cytometric analysisidentify and has the potential to then is(cell populations for further study. Alth(upregulation of MHC class II expressiolmicroglia occurs in the inflamed rat r((J V Forrester, unpublished data) as it doothe CNS,18 our recent studies with freshlylated microglia from rat CNS indicate t

cells are relatively poor APC for secondary Tcell responses compared with other non-microglial CNS macrophages.19 The lattermay include at least some of the perivascularcell populations. Finally, the range of CD4expressing cells in rats and humans appearscomparable, unlike mice, in that both expressthe CD4 antigen on cells of the macrophagemonocyte lineage.26 27 Thus the finding ofCD4 expression in both the perivascular andmicroglial populations of the retina, though at

102 a lower level than CD4 expression of T cells(Fig 4), suggests that comparable cells inhumans may be targets ofhuman immunodefi-

+ ciency virus (HIV) infection in the retina,ED2, similar to the CD4 dependent HIV infection2 on CNS microglia.28zalysis

The authors would like to thank Joseph Webster and PatrickTaylor for their technical assistance with the FACStarplus,Karen Knight and her team for breeding and care of experi-mental animals, and especially Anna Goodsall for producingand titrating monoclonal and polyclonal antibodies used in this

and study. We would also like to thank Dr Paul McMenamin forhave helpful discussions. The studies were supported by grants to

JDS from the National Health and Medical Research Council,the the Multiple Sclerosis Society of Australia, Commonwealth

APC AIDS Research Grants and to ADD from the LeverhulmeTrust, UK. A D Dick is a Wernher-Pigott MRC travellingrhich fellow, J D Sedgwick is supported by a Wellcome Trust senior

been research fellowship in medical sciences in Australia, and J Vcells Forrester is supported by Guide Dogs for the Blind, UK.

the 1 Gery I, Mochizuki M, Nussenblatt RB. Retinal antigens andiat a the immunopathologic process they provoke. Prog Ret Ress are 2 1986; 5:,5-109.2 Forrester JV, Liversidge J, Dua HS, Towler H,is of McMenamin PG. Comparison of clinical and experimen-lls in tal uveitis. Curr Eye Res 1990; 9: 75-84.

3 Caspi RR, Roberge FG, Mcallister CG, El-Said M,leno- Kuwabara T, Gery I, et al. T-cell lines mediating experi-

rpart mental autoimmune uvoretinitis (EAU) in the rat.art g Immunol 1986; 136: 928-33.

cular 4 Liversidge J, Sewell HF, Forrester JV. Interactions betweenough lymphocytes and cells of the blood retinal barrier:mechanisms of T-lymphocyte adhesion to human retinalI by capillary endothelial cells and retinal pigment epithelials in cells in vitro. Immunology 1990; 71: 390-6.5 Chan CC, Hooks JJ, Nussenblatt RB, Detrick B. Expressionsuch of Ia antigen on retinal pigment epithelium in experimen-to T tal autoimmune uveoretinitis. Curr Eye Res 1986; 5:325-32.cells 6 Roberge FG, Caspi RR, Nussenblatt RB. Retinal Mullerithin cells produce IL-1 activity and have a dual effect onautoimmune T-helper lymphocytes. IJmmunol 1988; 140:Is to 2193-9.

oglia 7 Fontana A, Fierz W, Wekerle H. Astrocytes present myelinbasic protein to encephalitogenic T-cell lines. Natures are 1984; 307: 273-6.ibers 8 Pryce G, Male D, Sedgwick J. Antigen presentation in

brain: brain endothelial cells are poor stimulators ofT cell)eing proliferation. Immunology 1989; 66: 207-12.

9 Fabry Z, Sandor M, Gajewski TF, Herlein JA, WaldscmidtMM, Lynch RG, et al. Differential activation of ThI and

stent Th2 CD4+ cells by murine brain microvessel endothelialies in cells and smooth muscle/pericytes. J7 Immunol 1993; 151:38-43.non- Io Risau W, Engelhardt B, Wekerle H. Immune function ofthecells. blood-brain barrier: incomplete presentation of protein(auto-)antigens by rat brain microvascular endothelium inI not vitro. J CellBiol 1990; 110: 1757-64.II 1 Fabry Z, Waldschmidt MM, Moore SA, Hart MN. Antigen

presentation by brain microvessel smooth muscle andtinal endothelium.J Neuroimmunol 1990; 28: 63-70.icing 12 Sedgwick JD, Hickey WF. Antigen presentation in the

central nervous system. In: Keane RW, Hickey WF, eds.es in Immunology of the nervous system. Oxford: Oxfordtcytic University Press, 1995.13 Forrester JV, McMenamin PG, Holthouse I, Lumsden L,iring Liversidge J. Localisation and characterisation of majorional histocompatibility complex class II-positive cells in theposterior segment of the eye: implications for induction ofDo be autoimmune uveoretinitis. Invest Ophthalmol Vis Sci 1994;can 14 35: 64-77.

14 McMenamin PG, Holthouse I, Holt PG. Class II majorolate histocompatability complex (Ia) antigen bearing dendriticOugh cells in the iris and ciliary body of the rat eye: distribution,phenotype and relation to retinal microglia. Immunologyn on 1992; 77: 385-93.etina 15 Perry VH, Gordon S. Macrophages and microglia in the

nervous system. Trends Neurosci 1988; 11: 273-7.es in 1 6 Penfold PL, Provis JM, Liew SCK. Human retinalriso- X XXXXXmicroglia express phenotypic characteristics in common

with dendritic antigen-presenting cells. J Neuroimmunolhese 1993; 45: 183-92.

839

on June 7, 2020 by guest. Protected by copyright.

http://bjo.bmj.com

/B

r J Ophthalm

ol: first published as 10.1136/bjo.79.9.834 on 1 Septem

ber 1995. Dow

nloaded from

Dick, Ford, Forrester, Sedgwick

17 Penfold PL, Madigan MC, Provis JM. Antibodies to humanleukocyte antigens indicate subpopulations of microglia inhuman retina. Vis Neurosci 1991; 7: 383-8.

18 Sedgwick JD, Schwender S, Imrich H, Dorries R, ButcherGW, ter Meulen V. Isolation and direct characterisationof resident microglial cells from the normal and inflamedcentral nervous system. Proc Nad Acad Sci USA 1991; 88:7438-42.

19 Ford AL, Goodsall AL, Hickey WV, Sedgwick JD. Normaladult ramified microglia separated from other CNSmacrophages by flow cytometric sorting: phenotypic dif-ferences defined and direct ex vivo antigen presentation tomyelin basic protein-reactive CD4+ T cells compared.Immunol (in press).

20 Dijkstra CD, Dopp EA, Joling P, Kraal G. The hetero-geneity of mononuclear phagocytes in lymphoid organs:distinct macrophage subpopulations in the rat recognisedby monoclonal antibodies ED1, ED2 and ED3.Immunology 1985; 54: 589-96.

21 McMenamin PG, Crewe J, Morrison S, Holt PG.Immunomorphologic studies of macrophages and MHCclass II+ dendritic cells in the iris and ciliary body of therat, mouse and human eye. Invest Ophthalmol Vis Sci1994; 35: 3234-50.

22 Streit WF, Graeber MB. Heterogeneity of microglial andperivascular cell populations: insights gained from thefacial nucleus paradigm. Glia 1993; 7: 68-74.

23 Hickey WF, Kimura H. Perivascular microglial cells of theCNS are bone marrow-derived and present antigen invivo. Science 1988; 239: 290-2.

24 Thanos S. Sick photoreceptors attract activated microgliafrom the ganglion cell layer: a model to study the inflam-matory cascades in rats with inherited retinal dystrophy.Brain Res 1992; 588: 21-8.

25 Loffler K, McMenamin PG. Evaluation of subretinalmacrophage-like cells in the human fetal eye. InvestOphthalmol Vis SCi 1990; 31: 1628-35.

26 Jeffries WA, Green JR, Williams AF. Authentic T helperCD4 (W3/25) antigen on rat peritoneal macrophages.Exp Med 1985; 162: 117-27.

27 Wood GS, Warner NL, Warnke RA. Anti-leu-3IT4antibodies react with cells of monocyte-macrophageand Langerhans lineage. Jf Immunol 1983; 131:212-7.

28 Jordan CA, Watkins BA, Kufta C, Dubois-Dalcq M.Infection of brain microglial cells by human immunodefi-ciency virus type I is CD4-dependent. Virol 1991; 65:736-42.

840

on June 7, 2020 by guest. Protected by copyright.

http://bjo.bmj.com

/B

r J Ophthalm

ol: first published as 10.1136/bjo.79.9.834 on 1 Septem

ber 1995. Dow

nloaded from