Current Eye Research

Received on July 25, 1996; revised on November 20, 1996 and accepted on January 10, 1997

© Oxford University Press

Adenovirus infection of the cornea causes histopathologic changes in

the lacrimal gland

Richard L. Wood, Melvin D. Trousdale, Douglas Stevenson, Ana Maria Azzarolo and Austin K. Mircheff

Departments of Cell and Neurobiology, Ophthalmology and Physiology and Biophysics, University of Southern California School

of Medicine and Doheny Eye Institute, Los Angeles, CA, USA

Abstract

Purpose.

To explore the effects on the lacrimal gland of adeno-virus infection of the cornea.

Methods.

Rabbit corneas were inoculated with human adeno-viruses Ad5, Ad14, or a rabbit adapted form of Ad 5, and insome instances booster inoculations were given. Sections oflacrimal glands removed 21–59 days post-inoculation wereimmunostained using antibodies against rabbit Class I andClass II MHC molecules, CD4, CD8, CD18, and rabbit thymiclymphocyte antigen (RTLA). Relative numbers of positivelystained cells were quantified with a Metamorph image analysissystem.

Results.

RTLA and CD18 antigens were expressed on manyinterstitial cells in the normal lacrimal gland, but few expressedCD4 or CD8. The number of RTLA+ cells increased by 60–100% after inoculation of Ad5 and after boosting, and CD18+cells increased from 33–100% after inoculation of Ad5 andafter boosting. Booster inoculations also caused focal lympho-cytic infiltration. MHC Class I was expressed on interstitialcells and duct epithelium, but not acinar cells, and there was nodetectable difference after viral infection. In controls, MHCClass II was localized to a population of interstitial cells and afew acinar cells. A single inoculation of the Ad5 virus did notresult in an increase in the total number of MHC Class II-positive cells at 21 days, but inoculation with the rabbit-adapted Ad 5 and booster inoculations caused a 30% increase.

Conclusions.

Ad5 and rabbit-adapted Ad5 infection of the cor-nea induce lymphocytic infiltration in the lacrimal gland, andthe effect is enhanced by boosting. There is also an increase inexpression of MHC Class II after inoculation with rabbit-adapted Ad5 and with booster inoculations. Curr. Eye Res. 16:459–466, 1997.

Key words:

adenovirus; autoimmunity; cornea; immunohis-tochemistry; lacrimal gland

Introduction

We elsewhere (1, 2) have presented the hypothesis that pro-gressive lymphocytic infiltration of the lacrimal gland stromaand accompanying atrophy of secretory tissue might be trig-gered when lacrimal epithelial cells are induced to expressClass II MHC molecules aberrantly. It has recently been estab-lished that lacrimal acinar cells generate an extensive mem-brane recycling traffic between endosomal compartments andtheir plasma membranes (3, 4). A turnover flux of Class IIMHC molecules is superimposed on this recycling traffic, withinternalized molecules degraded and replaced from an intracel-lular reserve of molecules (5, 6). Preliminary experimentsdemonstrate that lacrimal acinar cells contain cathepsin B (6, 7),one of the proteases that mediates antigen processing in profes-sional antigen-presenting cells. The lacrimal gland also con-tains a number of mediators which might, plausibly, functionas accessory signals for T helper cell activation in response toClass II MHC-dependent autoantigen presentation: these in-clude interleukin-2 (8), a 65 kDa lymphocyte proliferationpotentiating factor (9), prolactin (or a prolactin-like molecule)(10, 11), and a spectrum of sensory and autonomic neuropep-tides (12–14). Thus, Class II MHC molecule-expressing lacri-mal epithelial cells would seem to possess the machinery, andto exist in an environment, in which they could trigger localautoimmune responses by processing and presenting autoanti-gens in much the same way that professional antigen-presentingcells process and present foreign antigens (15).

Among the fundamental issues posed by this hypothesis isthe question of what, if any, factors may induce lacrimal epi-thelial cells to begin expressing Class II molecules. Acinarcells in young, healthy rats and rabbits do not express Class IImolecules at levels detectable by immunohistochemistry,although they readily begin to do so when they are isolated and

Correspondence: Dr. Richard L. Wood, Department of Cell and Neurobiology,University of Southern California School of Medicine, 1333 San Pablo St., LosAngeles, CA 90033, USA

Cur

r E

ye R

es D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

ichi

gan

Uni

vers

ity o

n 11

/01/

14Fo

r pe

rson

al u

se o

nly.

460

R. L. Wood

et al.

placed in a serum-free, hormone-supplemented primary culturemedium (16). A survey of epithelial Class II molecule expres-sion in cadaver donor lacrimal glands revealed a broad range ofresults, from cases with no detectable Class II+ epithelial cellsto cases in which virtually all epithelial cells were Class II+. Inthat study there appeared to be a positive correlation betweenthe number of infiltrating lymphocytes and the number ofClass II+ epithelial cells (16). As Bottazzo and collaboratorshave noted (17, 18), this correlation could reflect the initiationof T cell activation and proliferation by Class II+ epithelialcells. However, since interferon-gamma is a potent inducer ofClass II molecule expression, the reciprocal cause and effectrelationship is also possible. That is, T-lymphocytes may havebeen responsible for the increased numbers of Class II+ epithe-lial cells. This relationship does, however, suggest the possibil-ity of a positive feed-back loop, in which a local inflammatoryresponse induces epithelial cells to express Class II molecules,and the subsequent presentation of autoantigens in a milieurich in accessory signals then triggers additional T helper cellactivation.

Recently, a rabbit model has been introduced for the study ofpotentially vision-threatening adenovirus infections (19, 20).The availability of this model affords an opportunity to learnwhether the inflammatory responses to ocular virus infectionsinduce lymphocytic infiltration and expression of Class IIMHC molecules by lacrimal acinar cells. We have induced cor-neal adenovirus infections by intrastromal inoculation. We findthat, within 3 weeks of infection, there is lymphocytic infiltra-tion in the lacrimal gland which is further enhanced by boosterinoculation of virus. After booster inoculations, there is also anincreased number of interstitial and acinar cells expressingMHC Class II molecules.

Materials and methods

Materials

New Zealand white rabbits (1.5–2 kg) were obtained from Irishfarms (Norco, CA). They were used in accord with the Declara-tion of Helsinki, the Guiding Principals for the Care and Useof Animals by the NIH, and the ARVO Resolution on the Use ofAnimals in Ophthalmic Research. Mouse monoclonal antibod-ies against rabbit MHC Class I (MAb 73.2), MHC Class II(MAb 45-3), CD4 (MAb KEN-4), CD8 (MAb 12.C7) andCD18 (MAb L13/64) molecules were obtained from Spring Val-ley Laboratories (Sykesville, MD). A goat polyclonal antibodyto rabbit thymic lymphocyte antigen (RTLA) was obtained fromAccurate Chemical and Scientific Corp. (Westbury, NY). Bioti-nylated goat-anti-mouse antiserum was obtained from Chemi-con International, Inc. (Temucula, CA), and biotinylated donkey-anti-goat antiserum was obtained from The Binding Site Ltd.(San Diego, CA). The Vectastain Elite ABC kit obtained fromVector Laboratories (Burlingame, CA) was used for final local-ization. Immu-mount was obtained from Shandon/Lipshaw(Pittsburgh, PA), and Brite floor polish (S. C. Johnson & Son,Inc., Racine, WI) was purchased at a local supermarket. Otherreagents were obtained from standard suppliers.

Methods

Adenovirus inoculation

Ad 5 McEwen strain was provided by Y. G. Gordon, M.D., Eyeand Ear Institute of Pittsburgh, PA. Ad 14, 1091-VR, and hostculture cells, A549 derived from a human lung carcinoma des-ignated CCL185, were purchased from the American TypeCulture Collection, Rockville, MD. In addition, a form of theAd 5 virus was recovered from a culture of Ad

5-

infected rab-bit corneal epithelial cells which maintained its proliferativecapability longer than the original Ad 5 virus, and when re-inoculated into corneal stroma elicited a more extensiveimmune cell infiltration than Ad 5. We designated this virus asrabbit-adapted Ad 5. Ad 14 virus is incapable of proliferationin rabbit corneal epithelial cells or keratocytes, but like Ad 5and the rabbit-adapted Ad 5, induced seroconversion and cor-neal subepithelial opacities containing immune cell infiltrates.The general procedures and reagents associated with cell cul-tures, virus production, purification, animal inoculation andvirus isolation from ocular specimens were as previouslydescribed (21, 22).

Our procedures for propagating virus, inoculation of infec-tious and inactive forms into the cornea, and monitoring ofclinical effects are described in detail in a recent publication(22). Eyes were examined for preinoculation defects and eachvirus or vehicle alone was administered under general andlocal anesthesia. The virus was inoculated intrastromally toform five focal blebs (10

m

l per bleb) using a dice pattern, fol-lowed by scarification of the corneal epithelium around theblebs and topical inoculation with 50

m

l as described by Gor-don

et al.

(19). Total volume of the inoculum was 10

6

plaque-forming units in 100

m

l per eye. Sham-infected eyes received100

m

l of 0.01 M Tris-HCl buffer, pH 8.0. Three hours afterinoculation, eyes were irrigated with sterile saline to removeunadsorbed virus and swabbed to determine whether residualvirus was present. Eyes were swabbed every other day for viralisolation and culture, followed by biomicroscopic examinationby an uninformed observer using slit lamp illumination. Therewas no evidence of viral shedding after 10–11 days. Stromaledema was induced by all inoculations. The edema began tosubside by 13 days post-inoculation, this coinciding with theappearance of subepithelial opacities. The latter were mostpronounced after inoculation of rabbit-adapted Ad 5 and Ad 5,but occurred after Ad 14 inoculation as well. Sham inocula-tions did not produce opacities.

Sixteen animals were used for study of the lacrimal gland,11 inoculated with virus and 5 control (2 sham inoculated, 3untreated). Four animals were inoculated with Ad 5, 3 animalswith rabbit-adapted Ad 5, and 4 animals inoculated with Ad 14,Ad 5 or rabbit-adapted Ad 5 and given booster inoculationswith Ad 5 or rabbit-adapted Ad 5. Table 1 summarizes theexperimental protocols. In all cases only one lacrimal gland peranimal was used for the immunohistochemical studies reportedhere.

The main (inferior) lacrimal glands from each eye wereremoved at the time of sacrifice (21, 28 or in one instance, 57days). As indicated above, in 4 animals a second inoculation of

Cur

r E

ye R

es D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

ichi

gan

Uni

vers

ity o

n 11

/01/

14Fo

r pe

rson

al u

se o

nly.

Ocular adenovirus-induced lacrimal histopathology

461

virus was given 21 or 42 days after the initial inoculation andthe animals were sacrificed 7 or 14 days later. Glands werebisected, and one half was immersed in OCT and frozen in liq-uid nitrogen for cryostat sectioning, while the other half wasused for viral culturing and plaque assays as previouslydescribed (22).

Immunohistochemistry

Cryostat sections were fixed in cold acetone for 5 min, blockedwith 10% normal goat serum or 5% BSA, and incubated atroom temperature for 60 min or overnight with the primaryantibodies diluted 1:100 to 1:20,000 in 50 mM Tris buffer con-taining 150 mM NaCl plus 0.1% BSA. Sections were thenrinsed in dilution medium and incubated 60 min with goat-anti-mouse-biotin 1:750, or donkey anti-goat-biotin 1:500 pre-pared in the same medium. After rinsing again, sections werequenched in 0.3% H

2

O

2

in 40% methanol for 15–30 min, incu-bated in ABC reagent for 30 min, rinsed 10 min, and developed3 min in a solution of 0.05% diaminobenzidine containing0.03% H

2

O

2

. The sections were rinsed in tap water andmounted in Immu-mount or ‘Brite’ floor polish for viewingand photography with an Olympus Vanox microscope. Immu-nohistochemistry controls consisted of the substitution of non-immune IgG for the primary antibody, use of an irrelevantprimary antibody, absence of primary antibody, and the ab-sence of both primary and secondary antibodies (ABC reactionalone).

Quantitation

Slides stained for RTLA, CD18, MHC Class I and MHC ClassII molecules were examined with a Nikon Microphot-FXAmicroscope equipped with a Metamorph

®

image analysis sys-tem. With the 10X plano objective used, the area for each fieldanalyzed was approximately 245,000

m

m

2

, and each measure-ment consisted of an average of at least 5 separate fields oneach section. For each set of measurements, a detection thresh-old was selected by matching the microscopic image with theMetamorph monitor image. The same individual made all ofthe measurements, and samples were coded so the measurerhad no knowledge of experimental conditions at the time ofmeasurement. The data were analyzed by single factor analysisof variance (ANOVA) and Duncan’s New Multiple Range Test(23) using a p-value of 0.05 to indicate a significant differencebetween experimentals and controls. Sections stained for CD4and CD8 maintained low numbers of positively stained cellswith all experimental procedures and the Metamorph system ofquantitation became impractical. To illustrate trends, entiresections (approximately 1 cm

2

) were surveyed manually andscored according to the following scale:

2

= no positive cellsin most fields, only rarely stained individual cells;

±

= up to 5positive cells in some fields of some sections; + = up to 10 pos-itive cells in several fields of all sections; ++ = 10 or morepositive cells in most fields in all sections. Because lympho-cytic foci around ducts and blood vessels are characteristic ofthe lacrimal gland in Sjögren’s patients, the presence or ab-sence of lymphocytic foci was noted and scored as follows:

2

=

none observed;

±

= 1–2 foci in some sections, up to 50 positivecells per focus; + = 2–4 foci in every section, up to 50 cells perfocus; ++ = 5 or more foci in every section, up to 100 + cellsper focus.

Results

Normal lacrimal tissue

Lacrimal glands from untreated or sham inoculated animalscontained an extensive resident population of lymphocytes andplasma cells. Most of these interstitial cells were distributeddiffusely (Fig. 1A), but occasional small lymphocytic fociwere observed. The majority of these cells expressed the RTLAantigen, indicating that they were T-lymphocytes (Fig. 1B), buta significant population of cells expressed CD18 antigen,which occurs predominantly on B-lymphocytes, plasma cells,macrophages, and other bone marrow-derived cells (Fig. 1C).The number of cells expressing the CD18 antigen was approxi-mately one-third that for RTLA antigen. Surprisingly, few ofthe resident lymphoid cells expressed either CD4 or CD8 anti-gen (Fig. 1D, E arrowheads), although there was a normal pop-ulation of lymphoid cells in the corneal infiltrates thatexpressed these antigens (22), and splenic nodules from thesame animals had normal numbers of CD4+ and CD8+ cells(Fig. 1F, G). MHC Class I antigen was observed on many inter-stitial cells and all lacrimal gland duct cells, but not the lacri-mal acinar cells (Fig. 1H). MHC Class II antigen was ex-pressed on some interstitial cells, and a few acinar cells in mostpreparations (Fig. 1I).

Inoculated animals

The general pattern of immunostaining in lacrimal glands ofanimals receiving virus inoculation in the corneal stroma wasthe same as for normal animals, but both the histochemicalstaining and the quantitative data indicated that there was agreater number of RTLA+ cells in the Ad 5 and boosted exper-imental groups (Figs. 2A and 3). Interestingly, there was nosignificant change in the number of RTLA+ cells after inocula-tion of the rabbit-adapted Ad 5 virus alone, but boosting withthe same virus elicited a significant increase (Fig. 3). The num-ber of CD18+ cells also increased significantly after the boost-ing protocol (Fig. 3). The intent of this part of the study was toexamine the effect of boosting in general, not to analyze indetail the various protocols for boosting. Therefore, theboosted experimental groups were pooled for the quantitativeanalysis (see Table 1). The numbers of cells expressing CD4and CD8 antigens remained low even after boosting, but someapparent increase was noted (Fig. 2C, D and Table 2). Therewas also an increase in the number and sizes of focal infiltrates(Table 1), and these infiltrates contained significant numbers ofCD4+ cells (Fig. 2C).

In viewing the slides directly, the total amount of staining forMHC Class II molecules appeared to be increased in the lacri-mal gland in all experimental groups, but the quantitative datarevealed that the apparent increase was not statistically signifi-

Cur

r E

ye R

es D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

ichi

gan

Uni

vers

ity o

n 11

/01/

14Fo

r pe

rson

al u

se o

nly.

462

R. L. Wood

et al.

cant in the Ad 5-inoculated group. However both the rabbit-adapted Ad 5 and boosted groups had a statistically significantincrease in expression of MHC Class II antigen (Fig. 3). Theexpression of MHC Class II antigen by acinar cells was some-what variable, but in all cases the number of acinar cells stain-ing positively appeared greater in inoculated animals than innormal controls, and was greatest in animals given boosterinoculations (Fig. 2E). A distinction between the staining ofacinar cells versus interstitial cells could not be accomplishedwith the Metamorph quantitation method, because it is basedon

total area of positive staining

after establishing a thresholddetection value above background.

Attempts to detect virus in lacrimal tissue

In view of the immunohistochemical evidence of changes inthe lacrimal gland after adenovirus infection of the cornea,attempts were made to isolate adenovirus from lacrimal tissue.Shedding of infectious virus was detectable in the cornea in allAd 5-, rabbit-adapted Ad 5- and booster-inoculated animals forapproximately 7 days after inoculation. No infectious viruswas detected in Ad 14-inoculated animals. Virus was notdetected in explanted lacrimal gland from any of the animals21 days post-inoculation. In addition, in other experiments inour laboratories lacrimal gland tissue has been explanted at 3, 5and 7 days after inoculation of adenovirus in the cornealstroma, and no virus has ever been detected. Two of four eyesfrom animals given secondary inoculations were virus culture-positive (from corneal swabs) until day 5 after the second inoc-ulation, but neither corneas nor lacrimal gland explants haddetectable infectious virus at the time of sacrifice (7 or 14 dayspost-inoculation).

Discussion

At least three major new observations have emerged from thisstudy. First, there is a large resident population of lymphocytesexpressing the RTLA antigen in the lacrimal gland of rabbits,the majority of which do not express CD4 or CD8. Large num-

Figure 1. Cryostat sections of lacrimal gland (A–E and H–I) andspleen (F–G) from normal rabbit. (A) numerous interstitial cells areseen with methylene blue staining; (B) many interstitial cells areRTLA+; (C) some cells are CD18+; (D and E) very few cells stainfor CD4 or CD8 (there is background staining of secretory granulesin acinar cells with anti-CD4); (F and G) CD4 and CD8 staining inthe spleen showing normal distribution; (H) duct cells and interstitialcells stain for MHC Class I; (I) a few interstitial and acinar cells stainfor MHC Class II. Staining with methylene blue highlights nucleiand the immunostains delineate whole cells because they recognizesurface antigens. All micrographs are at the same magnification.Bar = 50 mm.

Figure 2. Lacrimal gland from rabbit inoculated in the cornealstroma with adenovirus, boosted after 21 days and sacrificed at 28days. Increased numbers of interstitial cells stain for RTLA (A),CD18 (B), CD4 (C) and CD8 (D). (There is background staining ofacinar cell secretory granules with the anti-CD4 antibody.) There isalso increased staining by interstitial and acinar cells for MHC ClassII (E) and a greater frequency of focal infiltrations of lymphoid cellscompared to controls (compare with Fig. 1). All micrographs at samemagnification. Bar = 50 mm.

Cur

r E

ye R

es D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

ichi

gan

Uni

vers

ity o

n 11

/01/

14Fo

r pe

rson

al u

se o

nly.

Ocular adenovirus-induced lacrimal histopathology

463

bers of double negative T cells are known to appear in themammalian thymus during early embryonic development, andmost of these express the CD4 or CD8 antigens with matura-tion (15). However, isolated populations of double negative Tcells have been reported to occur in association with the epi-dermis and gut epithelium in some adult experimental animals.In the mouse, such cells have been shown to express T cellreceptors (TCR) containing gamma/delta subunits rather thanthe more common alpha/beta subunits (24). Rabbits arereported to have an unusually large number of circulating Tcells expressing the gamma/delta TCR (about 23%) as com-pared to humans and other laboratory animals (2–5%) (25).

Unfortunately, antibodies to either the alpha/beta or gamma/delta TCRs of rabbits are not available commercially. A mono-clonal antibody against human gamma/delta TCR showed nostaining in rabbit lacrimal gland, but this could be due to lackof cross-species reactivity, and therefore, we have been unableto determine which, if either, receptor is expressed by the dou-ble negative T cells observed in the rabbit lacrimal gland. How-ever, it should be pointed out that the exact roles in immunereactions of gamma/delta T cells or double negative alpha/betaT cells are still largely a matter of conjecture. Interestingly, theMRL/lpr mouse, a strain which spontaneously develops auto-immune disease involving lacrimal gland inflammatory lesions,exhibits CD4+ T cells in the lacrimal gland but a prevalence ofdouble negative T cells in lymph nodes (26). The presence of alarge population of double negative T cells in the lacrimalgland of the rabbit appears to be unusual, and whichever TCRthey express, this experimental system may prove useful forfurther elucidation of factors important in immunoregulation inthe lacrimal gland.

Second, we find it intriguing that the numbers of T and Bcells in the lacrimal gland are increased by adenovirus infec-tion of the cornea. The source of these cells in the lacrimalgland is uncertain at this time. It seems most likely that addi-tional lymphocytes are simply recruited from the circulationafter stimulated cells have migrated to regional lymph nodesand undergone proliferation, but the possibility that the localpopulation of cells in the lacrimal gland is induced to prolifer-ate

in situ

should also be considered. The fact that the CD4+and CD8+ T cell phenotypes are present in substantial numbersin the corneal infiltrates, but are rare in the lacrimal glandargues against a simple migration of activated T cells from thecornea to the lacrimal gland. The numbers of T cells expressingCD4 and CD8 antigens do appear to increase modestly in thelacrimal gland after viral infection of the cornea, particularly inlymphocytic foci, but the predominant interstitial cell popula-

Figure 3. Summary of quantitative data for RTLA, CD18 andMHC Class II staining. The error bars indicate ± SEM. For RTLA,the values for Ad 5 and boosted treatments are significantly differentfrom controls (p = < 0.05). For CD18, the value for the boostedgroup is significantly different from controls (p = < 0.05). For MHCClass II, the values for rabbit-adapted Ad 5 and boosted animals aresignificantly different from controls (p = < 0.05).

Table 1.

# Animals Treatment

Control2 Sham inoculated–21 d3 Untreated

Experimental4 Ad 5–21 d3 Rabbit-adapted Ad 5–21 d4 Boosted

1 Rabbit-adapted Ad 5–21 d Rabbit-adapted Ad 5–7 d

1 Rabbit-adapted Ad 5–40 dAd 5–17 d

1 Ad 14–21 d Rabbit-adapted Ad 5–7 d

1 Ad 14–28 dRabbit-adapted Ad 5–7 d

Cur

r E

ye R

es D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

ichi

gan

Uni

vers

ity o

n 11

/01/

14Fo

r pe

rson

al u

se o

nly.

464

R. L. Wood

et al.

tion during the time course of our studies continues to be dou-ble negative T cells and cells expressing the CD18 antigen.Some T cells are known to express the CD18 antigen, but the dis-tribution pattern and morphological characteristics of theCD18+ cells observed in the lacrimal gland indicate that theyare predominantly macrophages and plasma cells. Observa-tions on conventional sections reveal significant numbers ofplasma cells in normal rabbit lacrimal gland, consistent withthe known presence of polymeric IgA receptors on lacrimalacinar cells and of secretory IgA in the lacrimal fluid.

A third new observation emanating from these studies is thatthe number of cells expressing MHC Class II moleculesincreases in the lacrimal gland as a result of viral infection ofthe cornea. The increased MHC Class II expression is due to anincrease in numbers of both positive epithelial cells and posi-tive interstitial cells (i.e. macrophages, B cells and activated Tcells). It is already known that injury to the rabbit cornea caninduce extremely rapid alterations in the expression of earlyresponse genes in the lacrimal gland (27) and that expressionof MHC Class II molecules by rabbit lacrimal acinar cellsoccurs within hours after the cells are isolated and placed inprimary culture (16). Although our evidence indicates thatthere is no obvious change in expression of MHC Class II mol-ecules by lacrimal epithelial cells 21 days after a single inocu-lation of Ad 5 into the cornea, a change is clearly evident afterinoculation of the rabbit-adapted Ad 5 and by 7 days after abooster inoculation, even if the first inoculation was with Ad14, which does not proliferate in rabbit tissues. Since we exam-ined some samples at 3, 4, and 5 days post-inoculation andthey were also negative (data not shown), it seems likely thatthe induction of expression of MHC Class II molecules by lac-rimal epithelial cells in this experimental system requireschronic stimulation, and/or a secondary exposure to viral anti-gen. Both of these conditions presumably would sustain a pro-longed increase in the population of inflammatory cells.

Several different mechanisms could be invoked to explainhow lacrimal epithelial cells might be induced to express MHCClass II antigen in response to viral infection in the cornea.

First, it is conceivable that inoculated virus could spreaddirectly to the lacrimal gland and initiate a local inflammatoryreaction at that site as well as in the cornea. We have no directevidence that the inoculated virus spreads to the lacrimalgland, and we were unable to demonstrate replicating viruswithin the lacrimal gland in any of our experimental animals.Other experiments in our laboratories, as well, have failed todetect the virus in lacrimal gland explants at any time periodafter corneal inoculation. Thus, we do not believe that directspread of the virus is the explanation, although we cannottotally eliminate this possibility. Second, it is conceivable thatadenovirus antigens generated in the corneal epithelium orstroma might be processed by macrophages or dendritic cellslocally or in regional lymph nodes, and activated lymphocytescould proliferate and find their way to the lacrimal gland viathe circulation. Activated lymphocytes in turn might second-arily induce expression of the MHC Class II molecules by aci-nar cells or other antigen presenting cells within the lacrimalgland. This remains a viable possibility which needs furtherinvestigation. Third, inflammation in the cornea might initiate aneural reflex loop similar to what occurs with corneal wound-ing (27). The reflex loop would presumably involve neu-rotransmitters interacting directly with epithelial cells or im-mune cells in the lacrimal gland, release of local modulatorssuch as interferon gamma, and a subsequent activation of tran-scription of mRNA for MHC Class II molecules. The failure toinduce significant alterations in the population of interstitialcells or expression of MHC II by the lacrimal acinar cells aftersham inoculation in the cornea indicates that, if a neural reflexloop is involved, it is not due simply to the mechanical stimula-tion associated with the inoculations. The possibility of a spe-cific response to viral infection of the cornea via sustained acti-vation of a neural reflex loop by chronic corneal inflammationalso needs further investigation.

A question not addressed in the present study is the role ofgender in the regulation of MHC Class II expression and lym-phocytic infiltration in the lacrimal gland. It is well recognizedthat clinically apparent autoimmune disease of the lacrimalglands is much more frequent in women than in men. The ani-mals used in this study were mostly juvenile males, and noneshowed symptoms of dry eye during the course of these exper-iments. Future studies will compare the effects of viral infec-tions in male and female rabbits, and coupled with manipula-tion of the hormonal environment to simulate conditionstypical of mature and postmenopausal women, we anticipatethat additional insights into the pathogenesis of autoimmuneprocesses in the lacrimal gland will emerge.

In summary, these studies have established the existence of alarge population of double negative T cells in the lacrimalglands of rabbits and indicate that there are one or more path-ways linking adenovirus infection in the cornea with anincrease in numbers of immune cells and an increased expres-sion of MHC Class II molecules in the lacrimal gland. It is notclear whether the lymphocytes infiltrating the lacrimal glandsafter corneal infection with adenovirus are reactive againstviral antigens or autoantigens. Nevertheless, any kind of stimu-lus causing increased expression of MHC Class II molecules

Table 2.

CD8 and CD4 staining of cells and lymphocyte fociin the lacrimal gland

Treatment

Control Ad 5 Adapted Boosted

Cells CD4+

2

+ + ++CD8+

±

+ ++ ++Foci CD4+

2

+ + ++CD8+

2 2 2

±

Cells:

2

= no positive cells in most fields, only rare when present;

±

= up to 5positive cells in some fields of some sections; + = up to 10 positive cells inseveral fields of all sections; ++ = 10 or more positive cells in most fields inall sections. Foci:

2

= none observed;

±

= 1–2 per slide, up to 50 positive cellsper infiltrate, in some sections; + = 2–4 per slide, up to 50 cells per infiltrate,in all sections; ++ = 5 or more per slide, up to 100 + cells per infiltrate in allsections.

Cur

r E

ye R

es D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

ichi

gan

Uni

vers

ity o

n 11

/01/

14Fo

r pe

rson

al u

se o

nly.

Ocular adenovirus-induced lacrimal histopathology

465

could plausibly increase the possibility of activation of CD4+T cells reactive against autoantigens. As we have argued else-where (2), whether or not CD4+ T cell-antigen receptor stimu-lation progresses to autoimmune disease must depend on anumber of factors, including the spectrum of accessory signalsin the local milieu, and the individual’s ability to generateappropriate suppresser responses. In the present experimentsviral infection of the cornea had a minimal effect on numbersof CD4+ and CD8+ cells in the lacrimal gland, but repeatedinfection, which is known to occur in humans, would be pre-dicted to elicit a much more vigorous immune response. Sincelymphocytic infiltration with increased numbers of CD4+ cellsis characteristic of Sjögren’s Syndrome, and there is indicationthat the numbers of lymphocytic foci and their content ofCD4+ cells are increased in number after viral infection of thecornea in our studies on the rabbit lacrimal gland, we believethat this experimental system has the potential to serve as auseful model for further studies on the kinds of accessory sig-nals that might be involved in the generation of dry eye condi-tions and autoimmune disease in the lacrimal gland.

Acknowledgments

The authors thank Ms. Aileen Kuda, Ms. Francie Yarber, andMs. Barbara Platler for skilled technical assistance. They alsothank Dr. Dwight W. Warren for helpful discussions during thecourse of the work and Dr. Cheryl Craft, Mary D. Allen Profes-sor of Cell and Neurobiology for the use of the MetamorphImage Analysis System. This study was supported by NIH grantsEY 10550, EY 05801, EY 09417, EY 03040 and EY 09405.

References

1. Mircheff, A. K., Gierow, J. P., Wood, R. L., Akashi, R. H.and Hofman, F. M. (1994) Hypothesis for autoantigen pre-sentation and T cell activation. In

The Lacrimal Gland,Tear Film, and Dry Eye Syndromes

(Ed. Sullivan, D. A.).Pp. 651–654. Plenum Press, New York.

2. Mircheff, A. K., Gierow, J. P. and Wood, R. L. (1994)Autoimmunity of the lacrimal gland

. Internat. Ophthal-mol. Clinics

,

34,

1–18.3. Lambert, R. W., Maves, C. A., Gierow, J. P., Wood, R. L.

and Mircheff, A. K. (1993) Plasma membrane internaliza-tion and recycling in rabbit lacrimal acinar cells.

Invest.Ophthalmol. Vis. Sci.

34,

305–316.4. Gierow, J. P., Lambert, R. W. and Mircheff, A. K. (1995)

Fluid phase endocytosis by isolated rabbit lacrimal glandacinar cells.

Exp. Eye Res.

60,

511–525.5. Mircheff, A. K., Gierow, J. P. and Wood, R. L. (1994) Traf-

fic of major histocompatibility complex Class II moleculesin rabbit lacrimal gland acinar cells.

Invest. Ophthalmol.Vis. Sci.

35,

3943–3951.6. Yang, T., Gierow, J. P., Rafisolyman, S., Wood, R. L. and

Mircheff, A. K. (1994) Cathepsin B and D in lacrimal aci-nar cells: possible implications for autoantigen processing.(Abstract).

Invest. Ophthalmol. Vis. Sci.

35 (Suppl.),

1791.

7. Yang, T., Gierow, J. P., Bekmezian, A., Norian, J. M.,Rafisolyman, S., Wood, R. L. and Mircheff, A. K. (1995)MHC Class II molecules and cathepsins in endosomalcompartments of lacrimal gland acinar cells. (Abstract).

Invest. Ophthalmol. Vis. Sci.

36 (Suppl.),

S990.8. Stepkowski, S. M., Li, T. and Franklin, R. (1993) Interleu-

kin-2-like molecule produced by acinar cells in lacrimalgland regulate local immune response. (Abstract).

Invest.Ophthalmol. Vis. Sci.

34 (Suppl.),

1486.9. Liu, S. H., Zhou, D.-H. and Franklin, R. M. (1993) Lacri-

mal gland-derived lymphocyte proliferation potentiatingfactor.

Invest. Ophthalmol. Vis. Sci.

34,

650–657.10. Frey, W. H., II, Nelson, D., Frick, M. L. and Elde, R. P.

(1986) Prolactin immunoreactivity in human tears and lac-rimal gland: possible implications for tear production. In

The Preocular Tear Film in Health Disease and ContactLens Wear

(Ed. Holly, F. J.). Pp. 798–806. Dry Eye Insti-tute, Lubbock, TX.

11. Mircheff, A. K., Warren, D. W., Wood, R. L., Tortoriello,P. J. and Kaswan, R. L. (1992) Prolactin localization, bind-ing, and effects on peroxidase release in rat exorbital lacri-mal gland.

Invest. Ophthalmol. Vis. Sci.

33,

641–650.12. Nikkinen, A., Lehtosalo, J. I., Uusitalo, H., Alkama, A.

and Panula, P. (1984) The lacrimal glands of the rat andguinea pig are innervated by nerve fibers containing im-munoreactivities for substance P and vasoactive intestinalpolypeptide.

Histochemistry,

81,

23–27.13. Gibbins, I. L. (1990) Target-related patterns of co-existence

of neuropeptide Y, vasoactive intestinal peptide, enkepha-lin, and substance P in cranial parasympathetic neuronsinnervating the facial skin and exocrine glands of guineapigs.

Neuroscience

,

38,

541–560.14. Matsumato, Y., Tanabe, T., Ueda, S. and Kawata, M.

(1992) Immunohistochemical and enzyme histochemicalstudies of peptidergic, aminergic, and cholinergic innerva-tion of the lacrimal gland of the monkey (

Macaca fus-cata

).

J. Autonon. Nerv. Syst.

37,

207–214.15. Abbas, A. K., Lichtman, A. H. and Pober, J. S

.

(1994

)

In

Cellular and Molecular Immunology

, 2nd ed. Pp. 167–185. W. B. Saunders Company, Philadelphia.

16. Mircheff, A. K., Gierow, J. P., Lambert, R. W., Lee, L. M.,Akashi, R. H. and Hofman, F. M. (1991) Class II antigenexpression by lacrimal epithelial cells: an updated workinghypothesis for antigen presentation by epithelial cells

.Invest. Ophthalmol. Vis. Sci.

32,

2302–2310.17. Bottazzo, G. F., Pujol-Borrell, R. and Hanafusa, T. (1983)

Role of aberrant HLA-DR expression in induction ofendocrine autoimmunity.

Lancet

,

2,

1115–1118.18. Bottazzo, G. F., Todd, I., Mirakian, R., Belfiore, A. and

Pujol-Borrell, J. (1986) Organ-specific autoimmunity: a1986 overview

. Immunol. Rev.

94,

137–169.19. Gordon, Y. J., Romanowski, E. and Araullo-Cruz, T.

(1992) An ocular model of adenovirus type 5 infection inthe NZ rabbit.

Invest. Ophthalmol. Vis. Sci.

33,

574–580.20. Gordon, Y. J., Romanowski, E., Arullo-Cruz, T. and De

Clercq, E. (1992) Pretreatment with topical 0.1% (S)-1-(3-hydroxy-2-phosphonylmethoxypropyl) cytosine inhibits ad-

Cur

r E

ye R

es D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

ichi

gan

Uni

vers

ity o

n 11

/01/

14Fo

r pe

rson

al u

se o

nly.

466

R. L. Wood

et al.

enovirus type 5 replication in the New Zealand rabbit ocu-lar model.

Cornea

,

11,

529–533.21. Tsai, J. C., Garlinghouse, G., McDonnell, P. J. and Trous-

dale, M. D. (1992) An experimental animal model of adeno-virus-induced ocular disease: the cotton rat.

Arch. Oph-thalmol. 110, 1167–1170.

22. Trousdale, M. D., Nóbrega, R., Wood, R. L., Stevenson,D., Klein, D. and McDonnell, P. J. (1995) Studies of adeno-virus-induced eye disease in the rabbit. Invest. Ophthalmol.Vis. Sci. 36, 2740–2748.

23. Steel, R. G. D. and Torrie, J. H. (1960) Principles and Pro-cedures of Statistics. P. 99. McGraw-Hill, New York.

24. Hass, W. (1993) Gamma/delta cells. Ann. Rev. Immunol.11, 637–685.

25. Sawasdikosol, S., Hague, B. F., Zhao, T. M., Bowers, F. S.,Simpson, R. M., Robinson, M. and Kindt, T. J. (1993)Selection of rabbit CD4–CD8– T cell receptor-gamma/delta cells by in vitro transformation with human T lym-photropic virus-I. J. Exp. Med. 178, 1337–1345.

26. Jabs, D. H., Lee, B., Burek, C. L., Saboori, A. M. andPrendergast, R. A. (1996) Cyclosporine therapy suppressesocular and lacrimal gland disease in MRL/Mp-lpr/lprmice. Invest. Ophthalmol. Vis. Sci. 37, 377–383.

27. Cripps, M. H., Thompson, W. and Beuerman, R. W. (1994)Vasoactive intestinal peptide induces primary responsegene expression in lacrimal gland. In The Lacrimal Gland,Tear Film, and Dry Eye Syndromes (Ed. Sullivan, D. A.).Pp. 225–230. Plenum Press, New York.

Cur

r E

ye R

es D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

ichi

gan

Uni

vers

ity o

n 11

/01/

14Fo

r pe

rson

al u

se o

nly.