Immunohistochemical and topographic studies of dendritic cells and ...

Immunohistochemical and Topographic Studies ofDendritic Cells and Macrophages in Human Fetal Cornea

Claudia M. Diaz-Araya* Michele C. Madigan,-\ Jan M. Provis,*~fand Philip L. Penfold\

Purpose. To investigate the distribution and phenotype of major histocompatibility complex(MHC) class II-positive dendritic cells and macrophages in normal human fetal cornea inthe age range 10 to 25 weeks gestation.

Methods. Peroxidase and gold immunohistochemistry were used to visualize MHC class II andmacrophage antigen (S22) immunoreactive cells. Cell distributions were analyzed quantita-tively, and topographic maps were produced.

Results. Immunoreactive cells, concentrated centrally, were present at 10 weeks gestation inthe corneal epithelium and stroma. Average densities increased steadily up to 25 weeks gesta-tion. Two morphologic forms of MHC class II and S22 immunoreactive cells were observed—large, dendritiform cells and small, rounded cells with short processes. Electron microscopyrevealed that most MHC class II-positive cells were morphologically consistent with previousultrastructural descriptions of corneal Langerhans cells. Immunoreactive cells were morenumerous in immunogold-labeled specimens than in peroxidase-labeled specimens of similarages. However, quantitative analysis of both techniques revealed that S22-positive cells com-prised 30% to 50% of MHC class II-positive cells.

Conclusions. This study provides a detailed description of heterogeneous populations of MHCclass II and S22 immunoreactive cells in the human fetal cornea. In contrast to the adultcornea, which is typically devoid of MHC class II-positive cells, immunoreactive cells in thefetal cornea are concentrated centrally and increase in density up to at least 25 weeks gestation.These results indicate that reduction in Langerhans cell numbers to adult levels must occurafter 25 weeks gestation. The presence of dendritic cells and macrophages in the fetal corneahas important implications for the understanding of corneal immunology. Invest OphthalmolVis Sci. 1995;36:644-656.

I t has been established that Langerhans cells (LC) are apopulation of dendritic cells (DC) that mediate antigenpresentation in vitro1'2 and in vivo in the skin and cor-nea.3"5 Constitutive expression of major histcompatibilitycomplex (MHC) class II antigens is a characteristicfeature of DC, including LC in the skin,6'7 theconjunctivum, and the adult corneal epitheliumof several animal species9'8'9 and humans.8'10"14 Althoughthe presence of LC throughout the epithelium of die

From the * Departments of Anatomy ami Histology and ^Clinical Ophthalmology,University of Sydney, New South Wales, Australia.Supported by a Medical Foundation Postdoctoral Fellowship, University of Sydney,New South Wales, Australia. (MCM).Submitted for public12, 1994.Proprietary interest t

njuly 5, 1994; revised October 11, 1994; accepted October

Reprint requests: Claudia M. Diaz-Araya, Department of Anatomy and Histology,University of Sydney, New South Wales, Australia.

central cornea in normal infants has previously beenreported using nucleotidase reactivity,15 little is knownabout these cells in the developing human cornea, andthe identity of MHC class II-positive cells in the cornealstroma remains controversial.3'811'131416"20

Subsets of macrophages and DC have been identifiedpreviously using monoclonal antibodies in normal hu-man tissues, including brain and skin.21 In addition, popu-lations of resident DC and macrophages have been re-ported in the rat ciliary body and iris,22 the rat choroid,23

and in the human retina.24'25 The presence of heteroge-neous populations of DC and macrophages in the humanfetal cornea, however, has not been previously docu-mented. In the present study, we have used antibodies toMHC class II and macrophage (S22) antigens to studythe morphology and distribution of DC and macrophagesin wholemounted fetal cornea.

644Investigative Ophthalmology & Visual Science, March 1995, Vol. 36, No. 3Copyright © Association for Research in Vision and Ophthalmology

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933409/ on 02/13/2018

Dendritic Cells and Macrophages in Fetal Cornea 645

MATERIALS AND METHODS

Specimens and Tissue Preparation

Human tissues used in this study were obtained (afterobtaining informed consent) according to the tenetsof the Declaration of Helsinki and approval by theHuman Ethics Committee at the University of Sydney,Australia. Normal human fetal eyes (n = 74) in the agerange 10 to 25 weeks gestation were obtained within 15minutes of pregnancy termination. Fetuses at 21 weeksgestation or more died of natural causes, usually afterpremature delivery. Approximate gestational ageswere determined from medical records and ultra-sound examinations. Specimens were fixed in either2% paraformaldehyde in 0.1 M phosphate buffer (pH7.4) or in a periodate-lysine-paraformaldehyde solu-tion containing 3% paraformaldehyde, 1.8% lysine,and 0.3% periodate (pH 7.4) for 24 hours at 4°C andsubsequently washed in 0.1 M Tris-buffered saline(TBS). Eyes were dissected into two portions by anincision at the ora serrata. Corneas with the conjuncti-vum attached, the latter used as a positive control,were dissected away from the ciliary body and lens.

Immunohistochemistry

Dendritic cells and macrophages in fetal corneas andconjunctivum were identified using immunohisto-chemical procedures. Specimens were soaked in 0.2%Triton X-100 (BDH, Sydney, Australia) in TBS (pH7.6) for peroxidase labeling or in 0.4% saponin inTBS for immunogold labeling for 1 to 6 days at 4°C(depending on corneal size and thickness) before in-cubation in primary antibody for 3 to 6 days at 4°C.Monoclonal mouse anti-human class II HLA-DR,CR3/43, 1:50 dilution (Dako, Sydney, Australia) andmouse anti-human macrophage (S22 antigen), 1:5 di-lution (Amersham, Sydney, Australia) antibodies wereused in this study. Primary and secondary antibodieswere diluted in TBS-0.4% saponin-2% normal sheepserum (Dako) for peroxidase labeling or in TBS-0.4%saponin—2% normal goat serum (Dako) for immu-nogold labeling. Specimens were then rinsed in TBS.

Avidin-Biotin-Peroxidase Labeling

Specimens were incubated in a biotinylated secondaryantibody, anti-mouse, 1:50 dilution (Amersham) for afurther 24 hours at 4°C. After another rinse in TBS,bound antibody was detected with an avidin-biotin-peroxidase labeling technique (Vectastain, VectorLaboratories, Burlingame, CA) using a nickel-en-hanced 3,3'diaminobenzidine tetrahydrochloride so-lution.26 Radial cuts were made in whole corneas toassist flattening; specimens were wholemounted, epi-thelium upward, onto gelatinized slides, placed in ahumidified chamber overnight at 4°C to avoid sudden

dehydration, air dried, dehydrated, and mounted inDePeX (BDH).

Immunogold Labeling

Specimens were labeled using 1-nm gold particles en-hanced with silver for light microscopy. Specimenswere incubated overnight at 4°C in 1 nm gold conju-gated to goat F(ab')2 anti-mouse IgG + IgM (BioCellResearch Laboratory, Cardiff, UK), diluted 1:50. Afterfurther washing in 0.4% saponin-TBS, TBS, and 0.1M sodium cacodylate buffer (pH 7.4), tissue was fixedin 2.5% glutaraldehyde in 0.1 M sodium cacodylatebuffer for 1 hour at room temperature or overnightat 4°C. After rinsing in distilled water, bound goldparticles were visualized using silver enhancing solu-tion (BioCell Research); the reaction was stopped bywashing in distilled water, and specimens were flat-tened onto glass slides and mounted in glycerol.

ControlsDelay between death and fixation was less than 6 hoursfor all specimens used in this study. However, to exam-ine for the effect of postmortem delay on MHC classII expression, fixation of some human fetal corneaswas intentionally delayed at intervals of 45 minutesand at 6,12, and 18 hours (at 4°C), and the intensitiesof staining were compared qualitatively between sam-ples. In all cases, the intensity of MHC class II stainingwas similar.

Nonspecific antibody binding was excluded bysubstituting the primary antibody with a nonimmunemonoclonal mouse IgG isotype, 1:50 dilution (Dako).In addition, the primary and secondary antibodieswere excluded in some experiments to control fornonspecific binding of secondary antibodies and thepossible expression of endogenous peroxidase activity.

Electron MicroscopyPieces of periodate-lysine-paraformaldehyde-fixed cor-nea (10 to 20 weeks gestation; n = 6) were cut 1 to 2mm wide in a cross-section from the limbus to thecentral cornea. These pieces were immunogold la-beled for MHC class II and macrophage antigens asdescribed above. After fixation in 2.5% glutaraldehydein 0.1 M sodium cacodylate buffer (pH 7.4), speci-mens were postfixed in 2% osmium tetroxide, dehy-drated through a series of alcohols and acetone, em-bedded in epon-araldite resin, and cured at 60°C.Semithin sections (0.5 fxm) were stained with tolu-idine blue and examined with a 100X objective. Ultra-thin sections were examined at 100 kV in a Hitachi(Tokyo, Japan) H500 transmission electron micro-scope.

AnalysesThe morphology, size, and distribution of immunore-active cells in cornea and conjunctiva were assessed by


646 Investigative Ophthalmology & Visual Science, March 1995, Vol. 36, No. 3

TABLE l. Avidin-Biotin Peroxidase Labeled Cells

Age(weeks gestation)

MHC Class II101314151617-18192025

S22

1213-1415161718

n

36446

10692

221111

Average Area (mm2)(mean ± SD)

4.4 ± 26.5 ± 49.5 ± 3

11.3 ± 215.7 ± 217.2 ± 217.8 ± 319.5 ± 431.6 ± 1

2.5 ± 0.74.4

12.212.414.217.0

Density/mm2

(range)

1-52-689-134

15-21314-18610-3024-3109-372

63-240

2-2512-5236-1038-40

40-8732-142

Average Density/mm2 (mean ± SD)

2.5 ± 145 ± 21

112 ± 53126 :114 :155 :196 :202 :

t 57t 46t 68t 52t 75

188 ± 80

15 ± 234 ± 21

61215886

TotalNumber ofPositive Cells(range)

10-1284-390

558-1092934-2223867-2400994-4703

2018-51421653-59864204-7627

33-4185-214

745262816

1456

Average TotalNumber ofPositive Cells(mean ± SD)

11 j261 i934 j

1399 j1741 j2676 j3546 i3787 j

t 1t 133t 252t 609t 564t 1252t 1265t 1386

5916 ± 2420

37 ± 6150 ± 91

745262816

1456

light microscopy in wholemounts and sections. Somadiameters and dendritic field diameters of labeledcells were measured in samples of 100 cells from super-ficial and deep planes of focus of wholemounted cor-nea (n = 20) and conjunctiva (n = 1).

Density distributions of MHC class II and S22-posi-tive cells (number per mm2) were mapped by lightmicroscopy using a squared ocular graticule at a finalmagnification of X125. Counts of positive cells weremade in consecutive squares, completely sampling thewholemounted corneas. Positively labeled cells eitherwithin the graticule square or touching its upper andleft borders were included in the counts, and the datawere plotted onto an outline of the wholemountdrawn to scale from which the corneal area was also

estimated. The full thickness of each cornea was sur-veyed in two planes of focus (superficial and deep)for peroxidase-labeled specimens and in approxi-mately four planes of focus for immunogold labeledspecimens; the final densities were obtained by addingthe counts from each plane of focus.

These data were analyzed in two ways. For popula-tion analyses, the total number of immunoreactivecells for each specimen was determined, and the meandensity per mm2 was calculated. Data for all specimensin each age group were pooled, and the average den-sity per mm2 and the average total population werecalculated (Tables 1 and 2). For topographic analyses,distribution maps of MHC class II and S22-positivecells in each corneal wholemount were prepared using

TABLE 2. Immunogold Labeled CellsAge(weeks gestation)

MHC Class II12

1518

S22

12

151718

Average Area(mm2)

2.63.3

10.813.9

4.04.53.1

11.216.617.0

Density/mm2

(range)

10-3515-190

148-231115-190

12-1720-3625-4860-11440-14732-191

AverageDensity/mm2

4016718192

1628528897

156

Total Numberof Positive Cells

104550

19561283

63124164983

16132650



N

v:

,E

fFIGURE I. Semithin sections (0.5 fim) of human fetal cornea reacted for MHC class II antigens(16 weeks gestation, A, C, E) and macrophage (S22) antigens (20 weeks gestation, B, D,F). Immunoreactive cells were distributed in a nonlaminar fashion throughout fetal corneas.Immunoreactive cells with distinct silver-particle labeling of the cell surface are indicated(arrows) in anterior stroma (A, B), mid-stroma (C, D), and posterior stroma (E, F). Ep =epithelium; En = endothelium; K = keratocytes. (Immunogold histochemistry; toluidineblue.) Scale bar = 30 fim.

the Magellan program for Macintosh (Apple; Cuper-tino, CA).27

RESULTS

MorphologyIn all fetal corneal wholemounts flattened onto glassslides, the anterior cornea lies uppermost and the pos-

terior cornea underneath. We will, therefore, subse-quently refer to the anterior cornea as the superficialcornea and to the posterior cornea as the deep cornea.

MHC class II and S22-positive cells were presentin all specimens studied, displaying uniformly intensereactivity. Immunoreactive cells were distributedthroughout the fetal corneal epithelium and stroma,and in the immunogold preparations they were dis-



tributed in a nonlaminar fashion (Fig. 1). In cornealwholemounts, most MHC class II or S22-positive cellshad a dendritic morphology with multiple varicosecell processes extending from a small oval cell body(Figs. 2A to 2E). Positive cells with a more rounded,condensed morphology were also present, particularlyin specimens prepared using immunogold histochem-istry (Fig. 2F). At 25 weeks gestation, MHC class II-positive cells, evident at the junction of the peripheralcornea and conjunctiva, exhibited long streamer-likeprocesses oriented perpendicularly to the limbus (notillustrated).

Generally, the average diameters of the cell fieldsand somata of MHC class II and S22-positive cells weresimilar at all gestational ages. Positive cells had fielddiameters ranging from 13 to 81 fj,m and average fielddiameters of 36 ± 9 //m at 12 weeks gestation, 38 ±10 //m at 15 weeks gestation, 36 ± 9 fim at 18 weeksgestation, and 36 ± 9 /t/m at 20 weeks gestation. Somadiameters of positive cells ranged from 4 to 15 fim,and average soma diameters were 8.5 ± 2 fxm at 12weeks gestation, 9 ± 2 fim at 15 weeks gestation, 8 ±2 fim at 18 weeks gestation, and 7.5 ± 2 ^m at 20weeks gestation. Average field and soma diameters ofMHC class II or S22-positive cells at four representativegestational ages are presented in Figure 3. The dataindicate that, with the exception of MHC class II cellsat 12 weeks gestation, the field diameters of both S22and MHC class Il-positive cells were larger in the su-perficial cornea than the deep cornea (P < 0.0001;unpaired (-test). Furthermore, S22-positive cells con-sistently had larger fields (P< 0.0001; unpaired (-test)and larger somata (P < 0.0001; unpaired (-test) thanMHC class Il-positive cells. There was no statisticallysignificant difference in cell soma diameter in the su-perficial versus the deep cornea.

Conjunctival immunoreactive cells had a ramifiedmorphology similar to the immunoreactive cells of thecornea (Fig. 2G). In contrast to immunoreactive cellsof the cornea, however, many conjunctival immunore-

active cells were elongated parallel to the corneal cir-cumference, displaying larger average field diametersthan immunoreactive cells of the cornea. Class Il-posi-tive cells of the conjunctiva had an average soma diam-eter of 6.1 ± 1.4 /im (range, 4.4 to 10.6 ^m) and anaverage field diameter of 44.2 ± 9 /xm (range, 20 to71 fim). No staining was observed in corneas in whicha nonspecific IgGl isotype control was substituted forthe primary antibody (Fig. 2H). Similarly, no stainingwas apparent in corneas in which primary and second-ary antibodies were excluded.

At the electron microscopic level, silver-enhancedimmunogold labeling was detected on the surface ofcells distributed throughout the full thickness of thecornea (Fig. 4). Immunoreactive cells had mono-lobed or bi-lobed nuclei with condensed heterochro-matin in the periphery. The Golgi apparatus andrough endoplasmic reticulum were well developed.Similar morphologic forms were apparent throughoutfetal corneas. There was no evidence of MHC class IIor S22 reactivity on keratocytes or endothelial cells,although some epithelial cells appeared to displayMHC class II or S22 reactivity.

Corneal Area

Corneal area increased steadily with increasing gesta-tional age. Mean corneal area for MHC class II andS22-labeled specimens at each gestational age arelisted in Tables 1 and 2. These data were pooled toobtain average corneal areas for all specimens ana-lyzed at each gestational age (Fig. 5), and revealed asixfold increase in corneal area over the period ofdevelopment studied. Analysis of corneal areas in fel-low eyes at 13, 15, and 17 weeks gestation, in whichone eye was prepared using avidin-biotin-peroxidaseimmunohistochemistry and the other using immuno-gold histochemistry, indicated shrinkage of the perox-idase-labeled corneas of 17.7%, 15.8%, and 14.4%,respectively.

FIGURE 2. Photomicrographs of wholemounted human fetal cornea and conjunctiva. MHCclass Il-positive cells (arrows) were evident in planes of focus through (A) the basal epithelialcells (b) and (B) in the posterior corneal stroma (peroxidase histochemistry, 19 weeks gesta-tion). Macrophage antigen S22-positive cells (arrows) were also evident in planes of focusthrough the (C) basal epithelial cells (b) and (D) in posterior corneal stroma (peroxidasehistochemistry, 17 weeks gestation). MHC class II reactivity (arrows), visualized using immuno-gold histochemistry, revealed dendritic cells similar to those of peroxidase-labeled corneas(E) in the corneal epithelium and stroma (19 weeks gestation) and rounded cells with shortprocesses (F) in mid-stroma and posterior stroma (15 weeks gestation). Conjunctival controlsat 17 weeks gestation (immunogold histochemistry) displayed MHC class Il-positive LC (arrows)with similar morphologies to immunoreactive cells of the fetal cornea (G). An absence ofstaining was evident in control corneas in which a nonspecific IgGl isotype control wassubstituted for the primary antibody (16 weeks gestation) (H). Scale bar = 50 fim.


Dendritic Cells and Macrophages in Fetal Cornea

A «* -* v 1 B

649

* » ? * ** « * •

#



50-

3 40-

I 30-

f 20

i 10

12 15 18 20

• CLIISUP• CL II DEEP• S22SUP0 S22DEEP

12 15 18 20

Age (weeks' gestation)

FIGURE 3. Average field and soma diameters of MHC classII and macrophage antigen S22-positive cells in human fetalcornea. Average diameters were estimated from eight cor-neal wholemounts prepared using immunogold histochem-istry (12, 15, 18, and 20 weeks gestation) in superficial (ante-rior) cornea and deep (mid and posterior) cornea.

Topography in Wholemounts

Immunoreactive cells were distributed across the fullcorneal area and throughout the conjunctiva in allpreparations (Fig. 6).

Peroxidase. In peroxidase-reacted, corneas anti-MHC class II and anti-S22 labeled cells were visualizedat two planes of focus in the superficial and deepcornea (Figs. 2A to 2D and 6). Greater numbers ofcells were evident in the superficial cornea than inthe deep cornea. Quantitative data from corneas inwhich MHC class II (n = 50) and S22 (n = 8) immuno-reactivity were visualized with peroxidase are summa-rized in Table 1 and illustrated in Figure 7. The aver-age density of MHC class II immunoreactive cells in-creased from 2 cells/mm" at 10 weeks gestation to 245cells/mm2 at 25 weeks gestation; the total numbers ofMHC class II immunoreactive cells were 10 at 10 weeksgestation and 7627 at 25 weeks gestation. The averagedensity of S22 immunoreactive cells increased from14 cells/mm*' at 12 weeks gestation to 86 cells/mm2

at 18 weeks gestation; the total numbers of immunore-active cells increased from 33 at 12 weeks gestation to1456 at 18 weeks gestation.

Total densities (superficial plus deep counts) anddistributions of anti-class Il-positive peroxidase-re-acted cells are shown in five representative maps cov-ering the range of fetal ages studied (Fig. 8). Similarly,four representative maps displaying the distributionsof peroxidase-reacted anti-S22-positive cells are shownin Figure 9. These distribution maps indicate that for

all specimens, the densities of immunoreactive cellswere higher in the central cornea than in the periph-eral cornea, and they increased with gestational age.Further, the densities of anti-S22-positive cells weregenerally less than half of anti-class Il-positive cells atsimilar ages.

Immunogold. Immunogold labeling for MHC classII and macrophage S22 reactivity in 10 corneas re-vealed positive cells of both dendritic and roundedform (Figs. 2E, 2F). Using immunogold techniques,positive cells were distributed throughout the cornealepithelium and stroma in a nonlaminar fashion.Many of these immunogold-labeled cells (S22-posi-tive and MHC class Il-positive), present in the stroma,had a few short processes and a rounded morphology(Fig. 2F). In contrast to corneas prepared using per-oxidase techniques, immunogold-labeled corneasdisplayed greater numbers of immunoreactive cellsin the deep cornea than in the superficial cornea(not illustrated). Quantitative analyses indicated that

*"r" »

•^.i. ->

•0 J

• V

<*.'N ...,•>-^,

f .

.-., \K

B K

FIGURE 4. Electron micrographs of immunoreactive cells inhuman fetal cornea prepared using immunogold histo-chemistry. (A) An MHC class Il-positive cell in the anteriorstroma displays distinct labeling (arrows) of the cell surface(16 weeks gestation). (B) A macrophage (S22) immunoreac-tive cell in the stroma (20 weeks gestation) displays distinctcytoplasmic labeling (arroivs). Unlabeled keratocytes (K) arealso visible. N = nucleus. Scale bar = 5 nm.


Dendritic Cells and Macrophages in Fetal Cornea

35-

30"

0 J r

n-9

n-5

TTTT7 l

n-8

f10 12 14 16 18 20 22 24 26


FIGURE 5. Mean corneal area, estimated from all specimensanalyzed (n = 68) at each gestational age using peroxidaseand immunogold histochemistry. A sixfold increase in aver-age corneal area was evident during the period of develop-ment studied. Error bars = standard deviations.

greater numbers of immunoreactive cells were pres-ent in corneas processed using immunogold histo-chemistry compared with avidin-biotin-peroxidasehistochemistry.

Data from the 10 specimens prepared using im-munogold histochemistry are presented in Table 2.The average density of MHC class II immunoreactivecells was 40 cells/mm2 at 12 weeks gestation and 181cells/mm2 at 15 weeks gestation, and the total num-bers of MHC class II immunoreactive cells were 104cells at 12 weeks gestation and 1956 cells at 15 weeksgestation. The average density of S22 immunoreac-tive cells was 16 cells/mm2 at 12 weeks gestation and156 cells/mm2 at 18 weeks gestation, and total num-bers of immunoreactive cells were 63 at 12 weeksgestation and 2650 cells at 18 weeks gestation. Gener-ally, the total numbers and average densities of im-munogold labeled S22-positive cells were 50% or lessthan those of MHC class II-positive cells at similarages.

DISCUSSION

Morphology

The present study indicates that MHC class II and S22immunoreactive cells are present in the epitheliumand stroma of the normal human fetal cornea. Previ-ous studies have described isolated MHC class II anti-gen-bearing cells in the corneal stroma of adult guineapigs3 and humans."131416-2028 Rounded MHC class II-positive cells, assumed to be of lymphoid origin, havebeen reported in the peripheral stroma of guinea pigcornea.3 Others have identified MHC class II-positive

651

cells of dendritic morphology in corneal stroma asinterstitial DC,"1'8 rare MHC class II antigen-bearingkeratocytes,17 passenger cells,11 "spindle cells" (possi-ble precursors of epithelial DC)14 or "indeterminatecells." 28 However, many reports have suggested thatMHC class II-positive cells in the corneal stroma maybe LC." 1718-20-28 Langerhans cells have been shown toexpress high levels of MHC class II antigens.321"11

MHC class II-positive cells in the adult human periph-eral cornea and limbus have also been shown to con-tain Birbeck granules, ultrastructural features consid-ered specific for LC.832 MHC class II-positive "indeter-minate cells" of the corneal stroma have beenreported to be morphologically similar to LC exceptfor the absence of Birbeck granules.28 Andersson etalM and Karas et al,34 however, reported that examina-tion of epidermal serial sections in combination with

FIGURE 6. Photomicrographs of a human fetal cornealwholemount (19 weeks gestation) reacted for MHC class IIreactivity (peroxidase histochemistry). Immunoreaciive cellswere distributed evenly in the full corneal area and wereevident at planes of focus through the corneal epithelium(A) and the posterior stroma (B). Variation in cell morphol-ogies was apparent: (A) anteriorly immunoreactive cells dis-played dendritic morphology; (B) posteriorly immunoreac-tive cells were dendritic, elongated, or rounded with shortprocesses. Scale bar = 200 /xm.



"E300-E

Class ll-positiveS22-posiliv9 I

10 12 14 16 18 20 22 24 26


FIGURE 7. Average densities of MHC class II and macrophageantigen (S22) immunoreactive cells estimated from countsof the full cornea] area in 58 corneal wholemounts (peroxi-dase histochemistry). Average densities increased steadilyduring the period of development studied. S22-positive cellswere consistently present in numbers comprising 30% to50% of the MHC class II-positive cells.

tilting of the sections using a goniometer stage alwaysreveals Birbeck granules in the indeterminate cells.The presence of Birbeck granules in the so-called in-determinate cells must be interpreted as evidence ofa relationship with LC. The ultrastructural identifica-tion of Birbeck granules as the definitive feature ofLC remains contentious; confident identification ofBirbeck granules necessitates meticulous serial sec-tioning of tissue, and such a task is difficult and im-practical in any extensive study. No attempt was madein the present study to identify Birbeck granules be-cause all electron microscopic specimens had beenfixed and stained for optimal immunohistochemicalreactivity, and, as such, ultrastructural preservationwas inadequate for confident identification of thesesmall subcellular granules. The present study indicatesthat the majority of the MHC class II-positive epithelialand stromal cells are morphologically and phenotypi-cally consistent with previous descriptions of LC.32'33

We observed two morphologically distinct popula-tions of MHC class II and S22-positive cells throughoutthe developing cornea: large, dendritic cells andsmaller, rounded cells; a larger proportion of thesmaller, rounded cells were present in deep stroma.These observations suggest that there is a heteroge-neous population of MHC class II immunoreactivecells in the developing human cornea. Although mac-rophages are expected to express both MHC class IIand macrophage (S22) antigens, dendritic LC are ex-pected to be MHC class II positive but macrophage(S22) antigen negative.2 Our data suggest that in thedeveloping human cornea, macrophages representapproximately 30% to 50% of the population of MHCclass II immunoreactive cells. The remaining MHCclass II immunoreactive cells, we suggest, representdendritic LC. Dendritic cells can be distinguished

from macrophages by their morphology, more potentaccessory cell function, minimal phagocytic capacity,and lower levels of lysosomal enzymes.36 The observa-tion that dendritic morphology is not indicative of cellorigin or function37 is confirmed by the results of thepresent study. Cells of rounded and dendritic mor-phologies are evident in human fetal cornea labeledwith either MHC class II or anti-S22 antibodies, sug-gesting that corneal LC and macrophages can occurin "dendritic" and rounded forms.

Comparison of ImmunohistochemicalTechniquesAvidin-biotin-peroxidase and immunogold tech-niques indicated a greater number of MHC class II-positive cells than S22-positive cells in corneas of simi-lar ages. However, in specimens of similar ages andincubated in the same primary antibodies, the immu-nogold technique consistently labeled more cells thanthe avidin-biotin-peroxidase technique. Further-more, differences in apparent lamination of immuno-reactive cells, prepared using the peroxidase or immu-nogold techniques, suggest better penetration of thetissue by the immunogold F(ab' )2 secondary antibodycompared with the biotinylated (IgG) secondary anti-body used for the peroxidase labeling.

Comparison of corneal areas from the two tech-niques suggests that dehydration of corneas in theperoxidase method produced a shrinkage of 10% to20%. Densities of positive cells from peroxidase-re-acted corneas are, therefore, slightly overestimated.Taking account of the improved penetration usingthe immunogold procedure and approximately 15%shrinkage of peroxidase-labeled specimens, we esti-mate that the peroxidase-labeled specimens may un-derestimate actual numbers of LC and macrophagesby approximately 40% to 50%. Despite these discrep-ancies, both methods reveal similar observations incell morphologies, general topographies, and overallpopulations (see below).

TopographyLeukocyte lineage cells have been reported to migratefrom the bone marrow as monocyte-like cells to pe-ripheral structures through the circulating blood.31'38

The results of the present study indicate that DC andmacrophages are normal constituents of the humancornea from as early as 10 weeks gestation and furthersuggest a continued influx of MHC class II and S22immunoreactive cells up to, and most likely beyond,25 weeks gestation. These findings are consistent witha study of LC ontogeny in human fetal skin whichreports MHC class II-positive LC to be present duringearly gestation (6 to 7 weeks gestation) .39 Fetal corneasare relatively thin, and the wholemount techniqueallows direct analysis of immunoreactive cells in all



layers. Previous studies have used corneal sections orflatmounts of corneal epithelium (separated from theunderlying stroma) to examine LC distribution inadults and infants.8'91519 Nucleoside phosphatases,such as adenosine triphosphatase (ATPase), adeno-sine diphosphatase, and inosine diphosphatase, havebeen used as markers of retinal microglia, macro-phages, and epidermal LC.37'40 Quantification of AT-Pase-positive LC in one human fetal cornea (23 weeksgestation) reported densities considerably lower thanthose described in the present study (48 LC/mm2 incentral cornea and 75 LC/mm2 in peripheral cor-nea).15 We estimate an LC density in the central cor-neal epithelium (and immediately adjacent anteriorstroma) in specimens of 20 to 25 weeks gestation of118 to 241 LC/mm2, which is greater than those re-ported by Chandler et al.15 The higher LC densitiesobserved in the present study do not simply reflectdifferences in technique but more likely arise fromdifficulties in comparing data from a single cornea15

with data averaged from a number of corneas (n =11, 20 to 25 weeks gestation, present study).

Studies of adult corneas in guinea pig, hamster,mouse, and man have established that LC are presentin the epithelium of the conjunctiva and peripheralthird of the cornea but are absent from the centralcornea.341'42 In contrast, LC have been reported to bepresent in the epithelium of the central fetal and in-fant cornea.15 We observed immunoreactive cells inthe full corneal area distributed throughout the epi-thelium and stroma and concentrated in the centralcornea at all gestational ages. The increase in averagedensity of immunoreactive cells with increasing gesta-tional age observed in the present study indicates thatMHC class II and S22 immunoreactive cells are notsimply included pro rata in relation to an increasein corneal area. The presence of maximum averagedensities of MHC class II immunoreactive cells cen-trally at 25 weeks gestation (present study) and innewborn infants15 indicates that LC exit the corneawell after 25 weeks gestation and that the characteris-tic adult topography, in which central corneal LC den-sities approach 0/mm2, occurs sometime after birth.

Streamer-like dendrites have been described inguinea pig43'44 and mouse corneas,45 and it has beensuggested that these elongated cells may be a migrat-ing or activated form of LC.44 Although we cannotconclude that the streamer-like cells observed at 25weeks gestation in the present study (not illustrated)are migrating, their similarity to previously reportedstreamers suggests that, by 25 weeks gestation, LC mayhave commenced their movement away from the cen-tral concentration of LC observed during develop-ment.

We also observed that the average density of MHCclass II immunoreactive cells centrally increases 120-

fold over the period studied, whereas corneal areaincreases 6-fold. The elevated densities of LC in infantcorneas have been suggested to be the result of thesmaller eye allowing greater numbers of LC to reachthe central corneas because of limited mobility of LCfrom blood vessels.15 However, this is inconsistent withour data showing that the highest densities occur cen-trally, the region most distant from the limbal vascula-ture. The susceptibility of LC to ultraviolet radiationhas also prompted the suggestion that ultraviolet radi-ation may be responsible for the decreased LC densi-ties in the adult cornea.15 However, although acuteultraviolet radiation has been shown to reduce epider-mal and corneal LC temporarily,4'31'44 LC numbers arenot reduced with age or in response to chronic ultravi-olet exposure.31 Exposure of the infant eye to ultravio-let radiation does not appear to be the principal rea-son for clearing LC from these corneas. The absenceof LC in central adult cornea is more likely the resultof the production of factors by adult corneal epithelial

FIGURE 8. Total densities (superficial plus deep counts) anddistributions of MHC class II-positive cells in human fetalcorneas (peroxidase histochemistry) are shown in five repre-sentative maps covering the range of fetal ages studied. Im-munoreactive cells were distributed at all corneal locations,and densities increased steadily as gestation progressed, witha tendency for cells to concentrate centrally. Dot size isproportional to cell density; density range is indicated tothe left of each map. Within the outline for each specimen,the limbus is indicated by a broken circle.



12 Weeks

16 Weeks

FIGURE 9. Total densities (superficial plus deep counts) anddistributions of macrophage antigen S22-positive cells in hu-man fetal corneas (peroxidase histochemistry) are shown infour representative maps covering the range of fetal agesstudied. S22 immunoreactive cells were also distributed atall corneal locations, and densities increased steadily as ges-tation progressed, with a tendency for cells to concentratecentrally (compare with Fig. 8). Dot size is proportional tocell density, and density range is indicated to the left of eachmap. Within the oudine for each specimen, the limbus isindicated by a broken circle.

or stromal cells that inhibit the migration or reducethe life span of LC.15 Furthermore, other possible ex-planations to account for the reduced numbers of LCin the central adult cornea may include cell migrationof central LC toward the limbus, undetectable levels ofexpression of MHC class II antigens by LC, or possiblyapoptotic deletion of centrally located LC. An im-portant objective of future studies will be to determinethe precise stimulus, time course, and mechanism ofreduction of LC numbers in prenatal and postnatalcorneas.

Immunologic Significance

This study describes in detail the presence of largenumbers of DC and macrophages throughout the hu-man fetal cornea, consistent with observations in othertissues. Resident populations of DC and macrophageshave been described in normal tissues including

spleen, bone marrow, thymus, lymph node, liver,brain, lung, kidney, colon, and skin.21'3646"50 In addi-tion, heterogeneous populations of resident DC andmacrophages have been reported in the rat ciliarybody, iris,22 and choroid23 and in the human ret-ina.24'25 In the human retina, it has been estimatedthat a subpopulation of microglia express macrophageantigens.24 Functional tests indicate that in the gut andlung, macrophages may inhibit the activity of dendriticantigen-presenting cells, thereby modulating the localimmune responses.46'50 Accordingly, it seems reason-able to suggest that in most tissues, including cornea,similar functional capacities related to the occurrenceof both DC and macrophages may exist. A completepicture of the means by which tolerance to self-anti-gens is established is not presently available; however,T cells, B cells, and specialized populations of DC havebeen implicated.51 Similarly, DC in the developing hu-man cornea may have a role in establishing toleranceto corneal antigens.

A similar study of fetal epidermal LC demon-strated their presence before the full activation of theimmune system and suggested that fetal epidermal LCmay have the capacity to function as antigen-pre-senting cells but that parts of the immune system maybe too immature to cooperate.39 Because LC are pow-erful initiators of T cell-mediated immune responses,2

their presence in fetal and adult cornea, even in smallnumbers, has important implications for transplanta-tion. The transplantability of the cornea has been sug-gested to be related to the absence, or reduced num-bers, of class II-positive immunocompetent cells in thecorneal epithelium. The presence of class II-positivecells in the central stroma of adult corneas appear tobe related to the relatively low corneal graft successrate in vascularized and inflamed eyes.19 Retentioninto adulthood of a proportion of the class II-positivecells observed in human fetal corneal stroma may ex-plain the presence of some stromal class II-positivecells in normal human adult cornea and cornealgrafts. The present results confirm, therefore, the un-suitability of fetal and infant tissue for corneal trans-plantation. The significance of the existence of a tran-sient population of LC and macrophages in the fetalcorneal stroma and epithelium and the implicationsfor the immunologic status of this tissue remain to beestablished fully.

Key Words

cornea, dendritic cells, Langerhans cells, macrophages,MHC class II

Acknowledgments

The authors thank Drs. Michael J. Simcock, K. Kuah, andBernie Tuch for coordination of tissue donation, Ms. SandyBeynon for providing fetal specimens, Mrs. Tania Balind



for technical assistance, and Professor Francis A. Billson forsupport and encouragement.

References

1. Romani N, Lenz A, Glassel H, et al. Cultured humanLangerhans cells resemble lymphoid dendritic cells inphenotype and function. / Invest Dermatol. 1989;93:600-609.

2. Steinman RM. The dendritic cell system and its role inimmunogenicity. Ann Rev Immunol. 1991;9:271-296.

3. Klareskog L, Forsum U, Malmnas Tjernlund U, RaskL, Peterson PA. Expression of la antigen-like mole-cules on cells in the corneal epithelium. Invest Ophthal-mol Vis Sri. 1979; 18:310-313.

4. Wolff K, Stingl G. The Langerhans cell. /Invest Derma-tol. 1983;80(suppl):17-21.

5. Teunissen MBM. Dynamic nature and function of epi-dermal Langerhans cells in vivo and in vitro; A review,with emphasis on human Langerhans cells. HistochemJ. 1992;24:697-716.

6. Rowden G, Lewis MG, Sullivan AK la antigen expres-sion on human epidermal Langerhans cells. Nature.1977:268:247-248.

7. Klareskog L, Malmnas Tjernlund U, Forsum U, Pe-terson PA. Epidermal Langerhans cells express la anti-gens. Nature. 1977; 268:248-250.

8. Rodrigues MM, Rowden G, HackettJ, Bakos I. Langer-hans cells in the normal conjunctiva and peripheralcornea of selected species. Invest Ophthalmol Vis Sri.1981;21:759-765.

9. Gillette TE, Chandler JW, Greiner JV. Langerhanscells of the ocular surface. Ophthalmology. 1982;89:700-711.

10. Bhan AK, Fujikawa LS, Foster CS. T-cell subsets andLangerhans cells in normal and diseased conjunctiva.Am J Ophthalmol. 1982;94:205-212.

11. Fujikawa LS, Colvin RB, Bhan AK, Fuller TC, FosterCS. Expression of HLA-A/B/C and -DR locus antigenson epithelial, stromal, and endothelial cells of thehuman cornea. Cornea. 1982; 1:213-222.

12. Whitsett CF, Stulting RD. The distribution of HLAantigens on human corneal tissue. Invest OphthalmolVis Sri. 1984:25:519-524.

13. Vantrappen L, Geboes K, Missotten L, Maudgal PC,Desmet V. Lymphocytes and Langerhans cells in thenormal human cornea. Invest Ophthalmol Vis Sri.1985;26:220-225.

14. Catry L, Van den Oord J, Foets B, Missotten L. Mor-phologic and immunophenotypic heterogeneity ofcorneal dendritic cells. Graefe's Arch Clin Exp Ophthal-mol. 1991;229:182-185.

15. Chandler JW, Cummings M, Gillette TE. Presence ofLangerhans cells in the central corneas of normal hu-man infants. Invest Ophthalmol Vis Sri. 1985;26:113-116.

16. Mayer DJ, Daar AS, Casey TA, Fabre JW. Localizationof HLA-A,B,C and HLA-DR antigens in the humancornea: Practical significance for grafting techniqueand HLA typing. Transplant Proc. 1983; 15:126-129.

17. Treseler PA, Foulks GN, Sanfilippo F. The expression

of HLA antigens by cells in the human cornea. Am]Ophthalmol. 1984;98:763-772.

18. Tuft SJ, Hawkins P, McKenzieJL, Hart DNJ. The locali-sation of histocompatibility antigens in human cor-nea. Trans Ophthalmol Soc NZ. 1984;36:36-37.

19. Williams KA, Ash JK, Coster DJ. Histocompatibilityantigen and passenger cell content of normal anddiseased human cornea. Transplantation. 1985; 39:265-269.

20. Baudouin C, Fredj-Reygrobellet D, Gastaud P, LapalusP. HLA DR and DQ distribution in normal humanocular structures. Curr Eye Res. 1988; 7:903-911.

21. Franklin WA, Mason DY, Pulford K, et al. Immunohis-tological analysis of human mononuclear phagocytesand dendritic cells by using monoclonal antibodies.Lab Invest. 1986;54:322-335.

22. McMenamin PG, Holthouse I, Holt PG. Class II majorhistocompatibility complex (la) antigen-bearing den-dritic cells widiin the iris and ciliary body of the rateye: Distribution, phenotype and relation to retinalmicroglia. Immunology. 1992; 77:385-393.

23. Forrester JV, McMenamin PG, Holthouse I, LumsdenL, Liversidge J. Localization and characterization ofmajor histocompatibility complex class II-positive cellsin the posterior segment of die eye: Implications forinduction of autoimmune uveoretinitis. Invest Ophthal-mol Vis Sri. 1994;35:64-77.

24. Penfold PL, Madigan MC, Provis JM. Antibodies tohuman leucocyte antigens indicate subpopulations ofmicroglia in human retina. Vis Neurosri. 1991;7:383-388.

25. Penfold PL, Provis JM, Liew SCK. Human retinal mi-croglia express phenotypic characteristics in commonwith dendritic antigen-presenting cells. / Neuroimmu-nol. 1993;45:183-192.

26. Adams JC. Heavy metal intensification of DAB-basedHRP reaction product. / Histochem Cytochem. 1981;29:775-778.

27. Halasz P, Martin PR. A microcomputer based systemfor semi-automatic analysis of histological sections.Proc Roy Microscop Soc. 1984; 19:312.

28. Rowden G, Phillips TM, Lewis MG. la antigens onindeterminate cells of the epidermis: Immunoelec-tronmicroscopic studies of surface antigens. BrJDer-matol. 1979; 100:531-542.

29. King PD, Katz DR. Mechanisms of dendritic cell func-tion. Immunol Today. 1990;ll:206-211.

30. Jager MJ. Corneal Langerhans cells and ocular immu-nology. Reg Immunol. 1992;4:186-195.

31. Bergfelt L. Langerhans cells, immunomodulation andskin lesions: A quantitative, morphological and clini-cal study. Ada Derm Venereal (Stockh). 1993; 180(suppl):7-37.

32. Sugiura S, Matsuda H. Ultrastructure of Langerhanscell in human corneal limbus. Jpn J Ophthalmol.1970;13:197-202.

33. Andersson A, Sjoborg S, Elofsson R, Falck B. The epi-dermal indeterminate cell-a special cell type? Ada DermVenereol (Stockh). 1981;99(suppl):41-48.

34. Karas Z, WarcholJB, Kaczmarek E, Brelinska R, Jaros-zewski J. Application of graph theory to analysis of



Langerhans cell types on the basis of their cell nucleiultrastructure. Gegenbaurs Morph Jahrb. 1984; 130:755-760.

35. Dezutter-Dambuyant C, Schmitt D, Faure M, Horis-berger M, Thivolet J. Immunogold technique appliedto simultaneous identification of T6 and HLA-DR anti-gens on Langerhans cells by electron microscopy. JInvest Dermatol. 1985;84:465-468.

36. Wood GW, Turner RR, Shiurba RA, Eng L, WarnkeRA. Human dendritic cells and macrophages: In situimmunophenotypic definition of subsets that exhibitspecific morphologic and microenvironmental char-acteristics. AmJPathol. 1985; 119:73-82.

37. Stingl G. Dendritic cells of the skin. Immunodermatol-ogy. 1990; 8:673-679.

38. Katz SI, Tamaki K, Sachs DH. Epidermal Langerhanscells are derived from cells originating in bone mar-row. Nature. 1979; 282:324-326.

39. Foster CA, Holbrook KA, Farr AG. Ontogeny of Lang-erhans cells in human embryonic and fetal skin: Ex-pression of HLA-DR and OKT-6 determinants. J InvestDermatol. 1986;86:240-243.

40. Sanyal S, De Ruiter A. Inosine diphosphatase as a his-tochemical marker of retinal microvasculature, withspecial reference to transformation of microglia. CellTissue Res. 1985;241:291-297.

41. Streilein JW, Toews GB, Bergstresser PR. Corneal allo-grafts fail to express la antigens. Nature. 1979;282:326-327.

42. Bergstresser PR, Fletcher CR, Streilein JW. Surfacedensities of Langerhans cells in relation to rodent

epidermal sites with special immunologic properties.J Invest Dermatol. 1980; 74:77-80.

43. Lang RM, Friedlaender MH, Schoenrock BJ. A newmorphologic manifestation of Langerhans cells inguinea pig corneal transplants. Curr Eye Res.1981;1:161-167.

44. Kelley JG, Ohashi Y, Friedlaender MH. Langerhanscell alterations in the guinea pig cornea. Invest Oph-thalmol Vis Sti. 1985;26:1293-1296.

45. Hazlett LD. Corneal and ocular surface histochemis-try. Prog Histochem Cytochem. 1993; 25:1-60.

46. Holt PG, Degebrodt A, O'Leary C, Krska K, Plozza T.T cell activation by antigen-presenting cells from lungtissue digests: Suppression by endogenous macro-phages. Clin Exp Immunol. 1985; 62:586-593.

47. Poulter LW, Campbell DA, Munro C, Janossy G. Dis-crimination of human macrophages and dendriticcells by means of monoclonal antibodies. ScandJ Im-munol. 1986; 24:351-357.

48. Brooks CF, Moore M. Differential MHC class II expres-sion on human peripheral blood monocytes and den-dritic cells. Immunology. 1988;63:303-311.

49. Matyszak MK, Lawson LJ, Perry VH, Gordon S. Stro-mal macrophages of the choroid plexus situated at aninterface between the brain and peripheral immunesystem constitutively express major histocompatibilityclass II antigens. /Neuroimmunol. 1992;40:173-182.

50. Pavli P, Hume DA, van de Pol E, Doe WF. Dendriticcells, the major antigen-presenting cells of the humancolonic lamina propria. Immunology. 1993;78:132-141.

51. Mitchison NA. Specialization, tolerance, memory,competition, latency, and strife among T cells. AnnRev Immunol. 1992; 10:1-12.


Immunohistochemical and topographic studies of dendritic cells and ...

Documents

Transcript of Immunohistochemical and topographic studies of dendritic cells and ...