Adhesion Molecules and Relationship to Leukocyte Levels in Allergic Eye Disease

Annette S. Bacon,1'5 James I. McGill,5 David F. Anderson,13 Susan Baddeley,1 Susan L Lightman,2 and Stephen T. Holgate1

PURPOSE. TO evaluate the conjunctival expression of leukocyte cell adhesion molecules (CAMs) and their relationship to leukocyte patterns on the microvasculature in the different clinical subtypes of allergic eye disease. METHODS. Immunohistochemical analysis, using appropriate monoclonal antibodies, was applied to glycolmethacrylate-embedded biopsies of bulbar and tarsal conjunctival tissue. The proportion of total blood vessels expressing a particular CAM was derived and related to individual cell types identified by cell-specific markers, such as mast cells, eosinophils, neutrophils, T cells, and macrophages. Statistical analysis was used to correlate adhesion molecule expression and, ulti- mately, cell type.

RESULTS. There was a basal expression of intercellular adhesion molecule-1 (ICAM-1) (21% bulbar, 18% tarsal), E-selectin (15% bulbar, 21% tarsal), and vascular cell adhesion molecule-1 (VCAM-1) (13% bulbar and tarsal) in normal controls. In seasonal and perennial (bulbar and tarsal conjunctival) allergic tissue, ICAM-1 and E-selectin were expressed in 40% to 78% of vessels; in chronic disease, they were expressed in 45% to 80% of vessels; and in vernal giant papillae, they were expressed in as many as 90% of vessels. There was also increased expression of endothelial VCAM-1 in all forms of allergic eye disease; the greatest values were found in vernal giant papillae (64%). Biopsies taken in winter from seasonal sufferers demonstrated a marked reduction in levels of all three CAMs compared with those taken in the pollen season. This is almost consistent with values found in normal conjunctiva. Positive correlations were found between the levels of ICAM-1 and E-selectin expression and the degree of granulocyte and lymphocyte infiltration, although VCAM-1 expression correlated most closely with eosinophil numbers. CONCLUSIONS. Increased levels of cell adhesion molecules on the microvasculature and the factors that regulate them are likely to be responsible for the infiltration of cells bearing their ligands and may perpetuate inflammation in the chronic forms of allergic eye disease. (Invest Ophthalmol Vis Set. 1998;39:322-330)

llergic diseases1 of the conjunctiva are a common and ening keratopathy (atopic keratoconjunctivitis [AKC]) or lid heterogeneous group of disorders about which rela- changes (atopic blepharoconjunctivitis [ABC]). Giant papillary A tively little is known. Many atopic persons experience conjunctivitis (GPC) occurs in response to various types of self-limiting and often relatively mild seasonal (seasonal allergic ocular foreign body and has cellular features indicative of a conjunctivitis [SAC]) or perennial symptoms with acute exac- hypersensitivity reaction.1 erbations (perennial allergic conjunctivitis [PAC]). A small Cellular infiltrates characterize allergic conjunctivitis but number of others are more severely affected. Young patients have been reported to differ in quality and intensity between with active vernal keratoconjunctivitis (VKC) may demonstrate the two ends of the disease spectrum. SAC and PAC resemble gross lid involvement with giant papillae formation and the models of a type 1 hypersensitivity response, with well-docu- production of copious stringy mucus; acute pain and photo- mented mast cell degranulation occurring as a prelude to phobia accompany the formation of central corneal ulcers. In symptoms and the finding of high tear and serum levels of total older patients, chronic atopic conjunctivitis, usually in associ- and allergen-specific immunoglobulin E (IgE).2 In contrast, ation with facial eczema, is also characterized by sight-threat- there is increasing evidence that VKC and AKC may result from a more complex immunologic response involving predominantly T lymphocytes, mast cells, and eosinophils.3"5 These From the 'First Medicine Department and the Southampton Eye chronic conjunctival responses are more commonly associated Unit, Southampton General Hospital, Southampton, United Kingdom; 2 with other major atopic disorders, such as asthma and eczema, and the Moorfields Eye Hospital and Institute of Ophthalmology, but the basis for disease severity and chronicity is poorly London, United Kingdom. Supported by the Medical Research Council (MRC) of the United understood. Kingdom grant G90052985B and the Sir Jules Thorne Trust. More is known about the processes involved in selectively Submitted for publication March 4, 1997; revised August 6, 1997; attracting different populations of inflammatory cells to in- accepted November 3, 1997. flamed tissue. Under appropriate inflammatory conditions, Proprietary interest category: N. cells migrate at microvenular sites and begin rolling on the Reprint requests: James I. McGill, Consultant Ophthalmic Sur- 6 geon, Southampton Eye Unit, Southampton General Hospital, Tremona endothelium, as a result of adhesion molecule-ligand bonds Road, Southampton, United Kingdom SOl6 6YD. that form between two resistant shearing forces.7 First, this

Downloaded from jov.arvojournals.org on 09/25/2021 IOVS, February 1998, Vol. 39, No. 2 Adhesion Molecules in Allergic Conjunctivitis 323

process involves P-selectin,8 a preformed glycoprotein released upper tarsus, and at least one other major atopic disorder. The from Weibel-Palade bodies on the endothelial cell surface VKC group was restricted to patients younger than 16 years within minutes of contact with histamine or leukotrienes from who had itching and mucus discharge, hyperemia, infiltration, activated mast cells.9 The ligand for P- and E-selectin is the papillae of any size, keratopathy, and limbal inflammation. A sialyl Lewis-X family of molecules predominantly expressed on diagnosis of GPC could be made only if the patient had a neutrophils, eosinophils, and monocytes.8 Later E-selectin is history of contact lens wear or exposure to a suture or pros- produced as a result of cytokine-induced gene transcription thesis, conjunctival hyperemia, and large or giant papillae. 10 and protein synthesis. E-selectin (CD62E) is expressed on All patients had active disease requiring treatment, but, for endothelial cells 4 to 6 hours after stimulation with interleu- the purpose of this study, none had been administered topical 1 kin-1/3 (EL-1/3), tumor necrosis factor-a (TNFo!), or interferon-y. ' corticosteroids for at least 3 months, sodium cromoglycate for Intercellular adhesion molecule-1 (ICAM-1) (CD54), a 1 month (except two patients with AKC who had discontinued member of the immunoglobulin superfamily, is expressed with this drug for 1 week only), and either oral corticosteroids or increased intensity on the luminal surface of the vascular en- H^antihistamines for 3 months. All patients gave their in- dothelium 6 to 28 hours after interaction with IL-1/3, TNFa, or formed consent, and the study was approved by the ethical interferon-y.1 '"'3 Its two cellular ligands are the integrins committees of the Moorfields Eye Hospital and the Southamp- LFA-1 (CDlla/CD18) and Mac-1 (CDllb/CD18), present on ton University Hospitals and followed the Declaration of Hel- granulocytes, monocytes, and lymphocytes.14 Subjects with sinki. The history and examination were recorded according to ongoing allergic skin disease express increased amounts of a detailed protocol described previously.26 ICAM-1 and E-selectin on the skin microvasculature.15 After Biopsies were taken in season from 14 patients with SAC exposure to allergen, atopic persons may also be induced to (mean age, 41.8 ± 6.2 years), and 7 subjects (mean age, 26 ± express ICAM-1 and E-selectin on the endothelial surface of the 4.8 years) were biopsied out of season. There were 8 patients 15 16 microvasculature of skin, bronchial mucosa, and conjunc- with PAC (mean age, 35.5 ± 7.3 years), 12 with AKC (mean 17 tiva. age, 38.9 ± 14.1 years), 10 with ABC (mean age, 46.2 ± 16.2 Vascular-cell adhesion molecule-1 VCAM-1 (CD 106) is an years), 10 with VKC (mean age, 21.3 ± 5.4 years), and 8 with additional member of the immunoglobulin superfamily. It is GPC (mean age, 430 ± 99 years). induced late (24-28 hours) after the interaction of TNFa and Controls consisted of nonatopic patients, 12 of whom IL-4 or IL-13 with endothelial cells18 and after allergen chal- were undergoing routine cataract surgery under general anes- lenge is expressed in bronchi by 24 hours.19 Its integrin ligand thetic. These patients were administered only two to four very late antigen-4 (CD29d/CD49) is predominantly expressed drops of 0.5% chloramphenicol prophylaxis before surgery, on T lymphocytes, eosinophils, and basophils.20 Patients with and the conjunctiva was uninflamed. Five controls volunteered perennial rhinitis express increased levels of ICAM-1 and for biopsy under local anesthetic in the absence of other ocular VCAM-1,21 whereas those with steady state asthma express problems. E-selectin, ICAM-1, and VCAM-1 on the bronchial microvasculature.19 Conjunctival Biopsy The differential expression of cell adhesion molecules Tissue was biopsied from the central one third of the upper (CAMs) under various inflammatory conditions is thought to be tarsal conjunctiva, using a standard 3-mm trephine, under plain responsible for the preferential adhesion of different popula- lignocaine anesthetic. If there were giant papillae, one or two tions of cells, such as lymphocyte homing22 or the selective were shaved from the surface. A small bulbar conjunctival deployment of memory T cells and effector leukocytes in specimen was taken with scissors from high in the upper inflammation.23 To test the hypothesis that the expression of fornix. Tissue was removed from control subjects under gen- different adhesion molecules may dictate the pathogenesis of eral anesthetic at the beginning of the surgical procedure as variants of allergic eye disease, we have studied the expression described above. of ICAM-1, E-selectin, and VCAM-1 and their ligands in bulbar and tarsal conjunctival tissue by immunohistochemical analysis and have examined their levels of expression in relation to the Tissue Processing and Immunohistochemical numbers of infiltrating cells present. Analysis The biopsy material was immediately transferred into acetone containing the enzyme inhibitors 20 mM iodoacetamide and 2 mM phenylmethylsulfonyl fluoride at — 20°C. It was then pro- METHODS cessed into glycolmethacrylate resin and stored at — 20°C. Sequential sections were cut at 2 jam, so that the same vessels Subjects could be identified for each antibody. Immunostaining was Patients were recruited and assigned to subgroups according performed using the method described by Britten et al.27 to well-defined clinical criteria.24'25 Those included in the SAC Briefly, the glycolmethacrylate sections were floated onto am- or PAC groups had symptoms for at least 2 years. They were monia water (one part ammonia in 500 parts reverse osmosis either predominantly seasonal (hay fever) or perennial, and water). After 60 to 90 seconds, sections were fixed to poly-i.- they included itching, conjunctival hyperemia, minimal infil- lysine-coated slides and were dried for 3 hours at room tem- trate, and small papillae, as well as positive skin-prick test perature. Endogenous peroxidase activity was inhibited by results, raised tear, or serum IgE or eosinophils on conjunctival incubation of the sections in 0.1% sodium azide and 0.003% cytology. Others were assigned to the AKC or ABC groups hydrogen peroxide solution for 30 minutes. Nonspecific bind- depending on the presence or absence, respectively, of corneal ing was blocked by incubating the sections with tissue culture involvement. All the patients were older than 16 years, and medium for 30 minutes, and antibodies were applied overnight their symptoms were itching and burning, large papillae on the at room temperature. Biotinylated second-stage antibodies

Downloaded from jov.arvojournals.org on 09/25/2021 324 Bacon et al. IOVS, February 1998, Vol. 39, No. 2

were applied and visualized with the application of strepta- The levels of E-selectin, ICAM-1, and VCAM-1 expressed in vidin-biotin-peroxidase complexes (DAKO, Buckingham- the bulbar and tarsal tissue in each disease group were com- shire, UK) with the chromogen aminoethylcarbazole. The pared with the numbers of infiltrating tryptase-containing mast slides were counterstained with hematoxylin and mounted in cells, elastase-containing neutrophils, and ECP-containing eo- glycerol jelly. sinophils. Macrophages were identified by their cell-surface The details of mouse monoclonal antibodies used in the receptors CD68 and CD3, and T cells were identified by their immunostaining of conjunctival biopsies are as follows: cell-surface receptors CD4 and CD8. Relationships between ICAM-1, a gift from Rothleim (dilution 1/1000; British Biotech- CAM expression and individual leukocyte numbers were nology); E-selectin (dilution 1/1000; British Biotechnology); sought using the Spearman rank correlation test. After analyz- Serotec (dilution 1/1000); VCAM-1 (dilution 1/1000; British ing the groups separately, SAC and PAC were analyzed to- Biotechnology) and Serotec (dilution 1/1000); Ulex lectin (vas- gether to represent the more acute (self-limiting) allergic eye cular endothelial adhesion molecule-1; dilution 1/200; Sigma disorders, because SAC is accompanied by acute seasonal Chemical, St. Louis, MO); AA1, a gift from A. Walls of symptoms and PAC is a perennial disorder with acute seasonal Southampton (mast cell tryptase; dilution 1/100); neutrophil exacerbations. AKC and ABC together represent the more elastase (dilution 1/1000; Dakopatts, Glostrup, Denmark); eo- chronic (destructive) forms of allergic eye disease. Because of sinophil cationic protein (activated; dilution 1/200; Dako- its similarity to GPC, VKC was analyzed separately and together patts); CD3 (pan T cell; dilution 1/50, Dakopatts); CD4 (help- with GPC. er-inducer T cell; dilution 1/10 [Dakopatts] and dilution 1/1000 [Becton Dickinson, Mountain View, CA]); CD8 (sup- pressor-cytotoxic T cell; dilution 1/20; Dakopatts); CD68 RESULTS (macrophages; dilution 1/100; Dakopatts); and secondary (bi- Clear immunostaining was present at the cell surface, in the otinylated mouse antihuman F(ab)2 antibody; dilution 1/200; cytoplasm, and in the vascular endothelium, and there was no Dakopatts). staining of the negative control slides (Fig. 1). Control slides— one without the primary antibody and the other with a nonspecific IgGl monoclonal antibody at the Normal Conjunctiva highest concentration of antibody used (M9269; dilution 1/10; There was no difference in the immunohistochemical appear- Sigma)—were treated identically. Positive controls were tonsil ance of the normal conjunctiva obtained at cataract surgery or and previously positive-staining conjunctival tissue. Positive from normal volunteers from whom biopsies were taken under controls using different species of antibodies were not used a local anesthetic. In the non-atopic control tissue, there was a because previous work showed that this technique has no low level of constitutive adhesion molecule expression (Table advantage over positive control from the different tissues used 1). These low levels found in normal tissue were reflected in here.28 On every occasion, at least one normal control con- the low numbers of inflammatory cells identified by cell-spe- junctival tissue was stained simultaneously with the allergic cific immunostaining. Although E-selectin expression in the specimens. bulbar conjunctiva correlated with the number of CD4+ T cells, a closer correlation was obtained in the tarsal conjunctiva Quantification of Immunostaining and Statistical with CD8+ T cells (Table 2). The level of ICAM-1 expression Analysis failed to correlate with any of the cell types examined in either Cell counts were undertaken on coded slides using a standard bulbar or tarsal normal conjunctiva, whereas VCAM-1 expres- 1-mm eyepiece graticule at a total magnification of X400 by an sion correlated in the tarsal conjunctiva only with the low observer blinded to the protocol. Five to 10 adjacent fields number of eosinophils (Table 2). lying immediately under the epithelium were counted. A quar- ter of randomly selected slides were counted by a second Allergic Conjunctiva observer, and they were recounted if the results gave more that In each of the disease groups, the percentage of vessels stain- a 5% discrepancy. Cell counts were compared among tissue ing positively for ICAM-1, E-selectin, and VCAM-1 is shown in groups using the Mann-Whitney test. A nonparametric test Table 1. When compared with normal conjunctiva, there was was used because the data were non-normally distributed. a consistent and marked upregulation of E-selectin and ICAM-1 There was a skew in distribution in the data obtained. There- in all forms of allergic conjunctivitis. fore, a nonparametic approach was required. The Bonferroni Seasonal Allergic Conjunctivitis. When compared with correction was used to allow multiple comparisons, with the P normal conjunctiva, there was a significantly increased expres- value multiplied by a factor to account for the six groups sion of ICAM-1 and E-selectin. By contrast, out of season there compared. P < 0.05 was considered significant. was no significant difference observed in such levels when Whatever their size, vessels were counted once per field compared with biopsies obtained from the normal controls. but were again included in adjacent 2-jam sections. The total Immunostaining for VCAM-1 demonstrated a similar upregula- number of vessels in each field was identified by positive tion in bulbar tissue, although less so in tarsal tissue. In tissue immunostaining with the monoclonal antibody to Ulex euro- taken out of season, the level of VCAM-1 expression was paeus lectin, and the total number counted per square milli- similar to that observed in the control tissue (see Table 1). meter was calculated. The number of vessels staining positively Perennial Allergic Conjunctivitis. The findings in PAC for the different CAMs was also calculated per square millime- were almost identical with those of SAC. In this form of the ter and expressed as a percentage of the total number of disease, bulbar tissue showed a greater upregulation of ICAM-1, vessels, as described by Montefort et al.21 The percentage of E-selectin, and VCAM-1 than did the tarsal tissue (see Table 1). vessels immunostaining in the bulbar and tarsal tissues ob- Atopic Keratoconjunctivitis and Blepharoconjuncti- tained from each clinical group were compared with those in vitis. In AKC and ABC, when compared with normal tissue, the control group using the Mann-Whitney test. marked upregulation of endothelial ICAM-1 and E-selectin was

Downloaded from jov.arvojournals.org on 09/25/2021 IOVS, February 1998, Vol. 39, No. 2 Adhesion Molecules in Allergic Conjunctivitis 325

FIGURE 1. Photomicrographs showing immunohistochemical labeling of 2-jnm sections of conjunctiva from control subjects and those with seasonal allergic conjunctivitis. Positive staining was visualized using the red cromogen aminoethylcarbazole. (A) Ring staining of subepithelial CD4+ lymphocytes (arrows). (B) Ring staining of intraepithelial CD8+ lymphocytes (arrows). Vascular endothelium labeled for E-selectin (arrows) (C) and intracellular adhesion molecule-1 (iCAM-1) (arrows) (D). Control slides, treated identically, with the omission of the primary antibody (E) or using an unrelated inimunoglobulin Gl monoclonal antibody (F) showed no labeling. Original magnifications: (A) X555; (B) X555; (C) X555; (D) X555; (E) X35O; (F) X445.

observed, with little difference between the bulbar and tarsal junctiva. The level of expression of all three adhesion mole- conjunctiva. When compared with the normal tissue, VCAM-1 cules was significantly greater in AKC and ABC than in the was only significantly overexpressed in the AKC bulbar con- out-of-season SAC tissue, but there was little difference be-

Downloaded from jov.arvojournals.org on 09/25/2021 326 Bacon et al. IOVS, February 1998, Vol. 39, No. 2

TABLE 1. Details of Adhesion Molecule Expression in Different Allergic Eye Conditions Tarsal Conjunctiva P Value Bulbar Conjunctiva P Value

E-Selectin CAM Control *16.8 ± 4.2 15.2 ± 2.9 SAC 55.5 ± 6.3 < 0.001 56.4 ± 5.4 < 0.001 PAC 40 ± 5.1 ns 68 ± 11.5 = 0.002 AKC 638 ± 5.6 = 0.001 48.5 ± 4.2 ns ABC 45.3 ± 8.3 ns 63.5 ± 9.3 0.001 VKC 79-2 ± 4.7 < 0.001 65.8 ± 5.5 < 0.001 GPC 83-1 ± 8.5 = 0.002 51.1 ±6.0 = 0.011 ICAM-1 Control 17.9 ± 3.4 21.5 ± 4.9 SAC 59.2 ± 6.9 < 0.001 69.1 ± 5.6 < 0.001 PAC 53.6 ± 9.2 = 0.005 7.80 ± 10 = 0.001 AKC 54.4 ± 7.2 < 0.001 65 ± 7.9 < 0.001 ABC 69 ± 8.6 < 0.001 47.9 ± 9.4 ns VKC 79 ± 4.7 < 0.001 73 ± 4.9 < 0.001 GPC 85 ± 10.5 = 0.001 68 ± 8.6 = 0.009 VCAM-1 Control 13 ± 2.8 12.9 ± 4.9 SAC 23.6 ± 4.0 ns 50.1 ± 4.9 = 0.001 PAC 32.0 ± 8.6 ns 48.4 ± 10 = 0.008 AKC 16.6 ± 4.3 ns 41.3 ±5.5 ns ABC 336 ± 9.6 ns 41.3 ± 8.6 ns VKC 58 ± 7.7 = 0.001 44 ± 10.5 ns GPC 44.7 ± 14.7 ns 37.8 ± 9.1 ns ICAM-1, intercellular adhesion molecule-1; VCAM-1, vascular adhesion molecule-1; SAC, seasonal allergic conjunctivitis; PAC, perennial allergic conjunctivitis; AKC, atopic keratoconjunctivitis; ABC, atopic blepharoconjunctivitis; VKC, vernal keratoconjunctivitis; GPC, giant papillary conjunctivitis. *Results are expressed as a percentage of the number of vessels staining positively for the different CAMs per square millimeter out of the total number of vessels in each field. The Mann-Whitney test was used, with the Bonteroni test applied to allow multiple comparisons.

tween the two chronic disease groups (see Table 1). with bulbar CD8+ T cells. There was no correlation with Vernal Keratoconjunctivitis. The level of expression of endothelial VCAM-1 expression. ICAM-1 and E-selectin was at its greatest in VKC and GPC. Vernal Keratoconjunctivitis and Giant Papillary Immunoreactivity for VCAM-1 was also increased in tarsal and Conjunctivitis. The cellular infiltrate in VKC did not correlate bulbar tissue (see Table 1). with either E-selectin or ICAM-1 expression on the microves- Giant Papillae Conjunctivitis. In this disorder the level sels. However, VCAM-1 expression correlated with the eosin- of upregulation of ICAM-1 and E-selectin was also high in tarsal ophil infiltrate in bulbar and tarsal tissue. There was also a and less in bulbar tissue, and in this respect was similar to VKC. significant correlation between VCAM-1 and the neutrophil Expression of VCAM-1 was increased only in tarsal tissue (see infiltrate in the tarsal conjunctiva. Table 1). There were insufficient data points to correlate GPC findings on their own, but because the cellular infiltrate bore a Relationships between CAM Expression and Leukocyte Infiltration resemblance to VKC, the results of both were analyzed together. The addition of the GPC data made no difference Seasonal Allergic Conjunctivitis and Perennial Aller- except to emphasize the link between heavy tarsal expression gic Conjunctivitis. Because of the similarity on pathologic of VCAM-1 and the eosinophil-neutrophil infiltrate. appearance, SAC and PAC were analyzed together for the purpose of examining the relationship between the level of All Allergic Conjunctivitis Groups (Amalgamated adhesion molecule expression and degree of infiltrate of Data) each cell type; E-selectin upregulation was significantly associated with the number of infiltrating neutrophils in tarsal The expression of E-selectin and of ICAM-1 correlated with tissue. ICAM-1 expression correlated with the number of the numbers of neutrophils, eosinophils, CD3+ T cells, bulbar neutrophils and eosinophils, and VCAM-1 expression CD4+ T cells, and mast cells in tarsal and bulbar tissue and related with the number of bulbar eosinophils and with with macrophages in tarsal tissue only. There was also a tarsal CD4+ T cells (see Table 2). negative correlation between the levels of expression of + Atopic Keratoconjunctivitis and Atopic Blepharo- ICAM-1 with the numbers of CD8 T cells in tarsal tissue. conjunctivitis. Because of their similarities, AKC and ABC The expression of VCAM-1 correlated positively with the were analyzed together. E-selectin correlated with bulbar and numbers of eosinophils in bulbar and tarsal tissue of neutro- tarsal neutrophil infiltration, ICAM-1 correlated with the bulbar phils, macrophages, and CD4+ T cells in tarsal conjunctiva CD3 and CD4 counts, and there was a negative correlation only (see Table 3).

Downloaded from jov.arvojournals.org on 09/25/2021 IOVS, February 1998, Vol. 39, No. 2 Adhesion Molecules in Allergic Conjunctivitis 327

TABLE 2. The Relationships between Conjunctival Cell Adhesion Molecule Expression and Leukocyte Infiltration in Each Subtype of Allergic Eye Disease Tarsal Conjunctiva Bulbar Conjunctiva

Cell Type r2 P r2 P

Control CD8 Tcell ns ns 0.42 0.004 E-selectin CD8 Tcell 0.56 0.01* ns ns VCAM-1 Eosinophil 0.51 0.025 ns ns SAC, PAC E-selectin Neutrophils ns ns 0.56 0.004 ICAM-1 Neutrophils ns ns 0.48 0.014 Eosinophils ns ns 0.45 0.025 VCAM-1 Eosinophils ns ns 0.46 0.021 CD4 T cells 0.41 0.04 ns ns ACK, ABC E-selectin Neutrophil ns ns 0.57 -0.006 ICAM-1 CD3 ns ns 0.46 0.026 CD4 ns ns 0.47 0.022 CD5 ns ns -0.46 0.026 VKC VCAM-l Eosinophil 0.59 0.035 0.60 0.051 Neutrophil 0.71 0.007 ns ns GPC VCAM-1 Eosinophil 0.50 0.02 ns ns Neutrophil 0.66 0.001 ns ns The Mann-Whitney test was used, with the Bonferroni correction applied to allow multiple comparisons. CAM, cell adhesion molecule; SAC, seasonal allergic conjunctivitis; PAC, perennial allergic conjunctivits; VKC, vernal keratoconjunctivitis; AKC, atopic keratoconjunctivitis; ABC, atopic blepharoconjunctivitis; GPC, giant papillary conjunctivitis; ICAM-1, intercellular adhesion molecule-1; VCAM-1, vascular adhesion molecule-1.

DISCUSSION disease. For example, ICAM-1 expression in psoriasis has been clearly related to disease activity.30 AKC and ABC tissues ex- Conjunctival vessels should be added to the list of tissues hibited CAM levels that were no greater than those seen in known to constitutively express ICAM-l. The finding of basal acute SAC or subacute PAC. One possible explanation for this levels of E-selectin and VCAM-1 on conjunctival vessels, how- is that the disease severity in the patients with AKC and ABC ever, is unexpected and probably reflects the high level of may not reflect the activity of the allergic process, but other aero-antigen continuously dealt with at the conjunctival sur- factors such as lid margin infection and scarring may contrib- face. There is a sizable population of inflammatory cells in the ute an important component of these disorders. normal conjunctiva, which presumably uses these adhesion Amalgamation of all data from the different disease groups molecules to enter the tissue and help protect the wet surface allows a global perspective to be obtained between the cellular of the eye against offending environmental insults. infiltrate and the degree of expression of cell adhesion mole- The level of expression of E-selectin and ICAM-1 was cules on conjunctival microvessels. For the purposes of corre- markedly increased in acute SAC and subacute PAC and more lating the extent of cell infiltrate and adhesion molecule ex- so in the various clinical subtypes of chronic allergic conjunc- pression, it was considered reasonable to combine several tivitis. VCAM-1 expression was particularly increased in VKC. clinical disease entities because they shared common immuno- The relatively short-lived peak expression of E-selectin ob- pathologic mechanisms. Moreover, separate analyses did not served in endothelial cells in tissue culture after cytokine ex- alter the associations found. For instance, SAC and PAC repre- posure (4-6 hours)1' does not appear to be mirrored in aller- sent type-1 hypersensitivity reactions and result from cross- gic eye disease, in which elevated levels were apparent in all linkage of antigens to membrane-bound IgE, leading to mast forms of chronic and acute disease alike. This is similar to the cell degranulation. Although mast cell numbers increase in findings of increased expression of adhesion molecules in these conditions,28 there is no corresponding increase in other other subacute diseases, such as perennial rhinitis,21 and cell types31 and the degree of symptoms present correlates chronic remitting conditions, such as asthma19 and atopic with the mast cell numbers.31 In AKC and ABC, there is a dermatitis.15 E-selectin may be responsible for the homing of marked increase of mast cells, eosinophils, neutrophils, and memory T lymphocytes,23 and its overexpression on microves- macrophages32 together with a rise in CD4+ T cells.33 The sels has also been described in several types of chronic inflam- keratopathy of AKC is probably a result of the greater cell matory skin disorders,29 which are T-cell dependent. The sim- numbers in this form and the toxic effect on corneal epithe- ilarities between the levels of adhesion molecule expression in lium of their breakdown products.32 In VKC and GPC, CD4 cell the different disease groups are perhaps not surprising, be- numbers are increased to their greatest extent,32 and there is cause the subjects were all atopic persons with active-phase an extensive eosinophil infiltrate.34

Downloaded from jov.arvojournals.org on 09/25/2021 328 Bacon et al. TOYS, February 1998, Vol. 39, No. 2

TABLE 3. The Relationships between Conjunctival Cell Adhesion Molecules Expression and Leukocyte Infiltration in Amalgamated Data from Different Types of Allergic Eye Disease Bulbar Tarsal

Correlations of E-selectin Mast cells 0.44 <0.001 0.48 <0.001 Neutrophils 0.55 <0.001 0.65 <0.001 Eosinophils 0.57 <0.001 0.42 <0.001 Macrophages 0.16 ns 0.41 <0.001 CD3+ T cells 0.23 ns 0.47 <0.001 CD4+ T cells 0.39 <0.001 0.46 <0.001 CD8+ T cells -0.2 ns -0.22 ns Correlations of ICAM-1 Mast cells 0.38 <0.001 0.43 <0.001 Neutrophils 0.38 <0.001 0.45 <0.001 Eosinophils 0.54 <0.001 0.39 <0.001 Macrophages 0.19 ns 0.33 <0.001 CD3+ T cells 0.31 <0.001 0.39 <0.001 CD4+ T cells 0.36 <0.001 0.42 <0.001 CD8+ T cells -0.21 ns 0.28 0.004 Correlations of VCAM-1 Mast cells 0.18 ns 0.23 ns Neutrophils 0.28 ns 0.36 <0.001 Eosinophils 0.48 <0.001 0.44 <0.001 Macrophages 0.03 ns 0.28 0.003 CD3+ T cells 0.16 ns 0.40 <0.001 CD4+ T cells 0.22 ns 0.42 <0.001 CD8+ T cells -0.27 ns -0.05 ns *Cell counts were compared between tissue groups using the Mann-Whitney test. To allow for multiple comparison, the Bonferroni correction was used. ICAM-1, intercellular adhesion molecule-1; VCAM-1, vascular adhesion molecule-1.

In the normal conjunctiva, the failure of the extent of which is extensively expressed on this cell type.38 A combina- ICAM expression to correlate with any cell type may represent tion of the influence of ICAM-1 and VCAM-1 could explain the the low levels of the cell adhesion molecules present. Allergic ingress of CD3+ and CD4+ T cells, although it is difficult to conjunctivitis is characterized by infiltration with acute and explain how ICAM-1 correlated negatively with the level of the chronic inflammatory cells. In SAC and PAC, the high granulo- CD8+ T cells, because this cell also expresses the sialyl cyte counts were similar, probably reflecting a mast cell-initi- Lewis-X oligosaccharide ligand LFA-1. ated cell recruitment process, which may even explain the The histologic picture of VKC is dominated by CD4+ T positive correlation we found between bulbar E-selectin and cells and eosinophils, and the latter has beeen implicated in the the mast cell count. Others729 have shown that the neutrophil pathogenesis of corneal damage with its serious clinical con- infiltrate appears to relate most directly to the level of expres- sequences.32 VCAM-1, which is regulated by TNFa and IL-4, is sion of E-selectin and ICAM-1,713 whereas the eosinophii infil- associated with the level of the eosinophii infiltrate and is the trate has been found to relate to ICAM-16'8 and VCAM-1 ex- CAM most closely linked to chronic eosinophii responses in pression.35 Immunohistochemical studies demonstrate that the other allergic disorders.8 The neutrophil infiltrate associated mast cell in SAC and PAC contains preformed cytokines, such with the often ulcerated or sloughing surface of giant papillae as TNFa and IL-4,36 which are known to be important in the may be secondary and, as such, may be only indirectly linked NF-KB-mediated upregulation of E-selectin and ICAM-1 to the expression of VCAM-1. (TNFa)3738 and VCAM-1 (TNFa and IL-438). Cross-linking of The ligands for E-selectin and for ICAM-1 are ubiquitous as IgE bound to human mast cells not only releases preformed evidenced by the close association between the expression of cytokines along with histamine and other granule mediators, it these molecules and the number of infiltrating inflammatory induces the generation of newly formed cytokines, which are cell types. We found a correlation between the mast cell count secreted for as long as 48 hours after cell activation.39 With and the expression of E-selectin and ICAM-1, suggesting a intermittent and/or continued allergic stimulation of the con- cause-and-effect relationship. In acute and chronic allergic eye junctival mast cells, a mechanism is provided for selective diseases, the release of pro-inflammatory cytokines from acti- leukocyte recruitment and subsequent ongoing vasoactive me- vated mast cells36 not only induces increased adhesion mole- diator release. cule expression,37 but, with intermittent or continued allergic AKC and ABC exhibited a mixed cellular infiltrate, with mast cell activation, it activates the generation of newly formed E-selectin probably contributing mostly to the neutrophil pop- products that are secreted as long as 48 hours after cell acti- ulation by interacting with sialyl Lewis-X oligosaccharide, vation.38 A mechanism is thus provided for selective leukocyte

Downloaded from jov.arvojournals.org on 09/25/2021 JOVS, February 1998, Vol. 39, No. 2 Adhesion Molecules in Allergic Conjunctivitis 329

recruitment and subsequent ongoing mediator release. That 8. Bevilacqua MP. Endothelial-leukocyte adhesion molecules. Annu the mast cell may be the cause rather than the result of heavy Rev Immunol. 1993;11:767- 804. adhesion molecule expression may explain this link.39 In addi- 9. McEver RP, Beckstead JH, Moore KL, Marshall-Carlson L, Bainton DF. GMP-140, a platelet alpha-granule membrane protein is also tion to activating the recruitment of eosinophils, VCAM-1 is 20 synthesized by vascular endothelial cells and is located in Weibel- actively engaged in T cell recruitment. In severe chronic Palade bodies./ C/m Invest. 1989;84:92-99. allergic eye disorders, especially VKC and GPC, T cells of the 10. Montgomery KF, Osborn L, Hession C, et al. Activation of endo- Th-2 subtype are central to the high eosinophil response re- thelial-leukocyte adhesion molecule 1 (ELAM-1) gene transcrip- lated to the release of cytokines found to be encoded on the tion. Proc Natl Acad Sci USA. 1991;88:6523-6527. IL-4 gene cluster on chromosome 5, which includes IL-3, -4, -5, 11. Montefort S, Gratzion L, Golding D, et al. Bronchial and lung evidence for leucocyte infiltration and upregulation of leucocytes- and -13 and granulocyte macrophage-colony-stimulating fac- endothelial cell adhesion molecule 6 hours after local allergen challenge of sensitized asthmatic airway. / Clin Immunol, 1994; That conjunctival inflammation can be perpetuated by the 93:1411-1421. continued high expression of CAMs is an attractive theory. SAC 12. Bevilacqua MP, Stengelin S, Gimbrone MA, Seed B. Endothelial clinically regresses completely out of season, and the state of leucocyte adhesion molecule 1: an inducible receptor for neutrophils related to complement regulator)' proteins and lectins. Sci- apparent normality is mirrored by the loss of CAM expression. ence. 1989;243:1160-1165. In this respect, it would be interesting to assess the effect of 13. PoberJS, Cotran RS. The role of endothelial cells in inflammation. commonly used antiallergic ocular drugs on the levels of adhe- Transplantation. 1990;50:537-544. sion molecule expression in the different forms of allergic eye 14. Kishimoto TK, Larson RS, Corbi AL, et al. The leucocyte integrins. disease. Recent trials of nedocromil sodium have indicated a Adv Immunol. 1989;46:l49-182. powerful inhibitory effect on eosinophil activation and migra- 15. Kyan-Aung U, Haskard DO, Poston RN, Thornhill MH, Lee TH. tion,42'43 and it has been postulated that cytokine generation Endothelial leucocyte adhesion molecule-1 and intercellular adhesion molecule-1 mediate the adhesion of eosinophils to endothelial and the expression of leukocyte-specific adhesion molecules cells in vitro and are expressed by endothelium in allergic cutane- are affected by this drug.44 It may even be possible to design ous inflammation in vivo./Immunol. 1991;l46:521-528. studies that indicate whether there is likely to be a relapse 16. Leung DY, Pober JS, Cotran RS. Expression of endothelial-leuco- when drugs are withdrawn; in patients treated with cyclospo- cyte adhesion molecule-1 in elicited late phase allergic reactions. rin A who have a relapse of psoriasis, the amount of vascular J Clin Invest. 1991;87:1805-1809. endothelial adhesion molecules remains high.30 In animal mod- 17. Ciprandi G, Buscaglia S, Pesce G, Villaggio B, Bagnasio M, Ca- nonica GW. Allergic subjects express intercellular adhesion mole- els of disease, monoclonal antibodies to adhesion proteins on cule-1 (ICAM-1 or CD54) on epithelial cells of conjunctiva after cells or vessels have been used to prevent tissue edema, in- allergen challenge. / Allergy Clin Immunol, 1993;91:783-792. flammation, and cell death.45 The functional activity of ICAM-1 18. Schleimer RP, Sterbinsky SA, Kaiser J, et al. 1L4 induces adherence has been blocked by a specific antibody in primates resulting in of human eosinophils and basophils but not neutrophils to endo- a reduction in eosinophil recruitment to allergen-provoked thelium. J Immunol. 1992; 148:1086-1092. airway inflammation.46'47 In the eye, the early results of treat- 19- Bentley AM, Durham SR, Robinson DS, et al. Expression of the endothelial and leucocyte adhesion molecules ICAM-1, E-selectin ing uveitis with antiadhesion molecule monoclonal antibodies 48 and VCAM-1 in the bronchial mucosa in steady state asthma and have met with some success. Having demonstrated the up- allergen induced asthma. Thorax. 1992;47:852. regulation of ICAM-1, E-selectin, and VCAM-1 in allergic eye 20. Elices MJ, Osborn L, Takeda Y, et al. VCAM-1 on activated endo- disease, we have established the background from which to thelium interacts with the leukocyte integrin VLA-4 at a site dis- select these modes of treatment as they become available to tinct from the VLA-4/fibronectin binding site. Cell. 1990;60:577- 584. clinical practice. 21. Montefort S, Feather IH, Wilson SJ, et al. The expression of leukocyte-endothelial adhesion molecules is increased in perennial allergic rhinitis. AmJ Respir Cell Mol Biol. 1992;7:393-398. Acknowledgments 22. Picker LJ, Kishimoto TK, Smith CW, Warnock-Butcher EC. ELAM-1 is an adhesion molecule for skin-homing T cells. Nature. 1991; The authors thank R. Buckley for his kind permission to include his 349:796-799. patients in the study, and they thank Lynn Key for typing the manu- 23- Shimizu Y, Shaw S, Graber N, et al. Activation-independent binding script. of human memory T cells to adhesion molecule ELAM-1. Nature. 1991;349:799-802. References 24. Allansmith MR, Ross MN. Ocular allergy. Clin Allergy. 1988; 18:1- 13- 1. Allansmith MR, Korb DR, Greiner JV. Giant papillary conjunctivitis 25. Friedlaender MH. Conjunctivitis of allergic origin; clinical presen- induced by hard or soft contact lens wear. Ophthalmology. 1978; tation and differential diagnosis. Surv Ophthalmol. 1993;38:S105- 85:766-778. S114. 2. Allansmith MR, Ross RN. Ocular allergy and mast cell stabilisers. 26. Bailey CS, Buckley RJ. Nedocromil sodium in contact-lens-associ- Surv Ophthalmol. 1986;30:229-244. ated papillary conjunctivitis. Eye. 1993;7:S29-S33- 3. Foster CS, Rice BA, Dutt JE. Immunopathology of atopic kerato- 27. Britten KM, Howarth PH, Roche WR. Immunohistochemistry on conjunctivitis. Ophthalmology. 1991 ;98:1190-1196. resin sections: a comparison of resin embedding technique for 4. Abu el Asrar AM, Geboes K, Missotten L, Emarah MH, Maugdal PC, small mucosal biopsies. Biotech Histochem. 1993;68:271-280. Desmet V. Cytological and immunohistochemical study of the 28. Baddeley SM, Bacon AS, McGill JI, Lightman SC, Holgate ST, Roche limbal form of vernal keratoconjunctivitis by the replica technique. WR. Mast cell distribution and neutral phase expression in acute Br J Ophthalmol. 1987;71:867-872. and chronic allergic conjunctivitis. Clin Exp Allergy. 1995;25:4l- 5. Metz DM. T Cells in Atopic Eye Disease. London: London 50. University; 1995. DPhil Thesis. 29. Groves RW, Allen MH, Barker JN, et al. Endothelial leukocyte 6. Ley K, Gaehtgens P, Fennie C, Singer MS, Lasky LA, Rosen SD. adhesion molecule-1 (ELAM-1) expression in cutaneous inflamma- Lectin-like cell adhesion molecule-1 mediates leucocyte rolling in tion. BrJDermatol. 1991;124:117-123. mesenteric venules in vivo. Blood. 1991;77:2553-2555. 30. Horrocks C, Duncan JI, Oliver AM, Thompson AW. Adhesion 7. Springer TA. Adhesion receptors to the immune system. Nature. molecule expression in psoriatic skin lesions and the influence of 1990;346:425-434. cyclosporin A. Clin Exp Immunol. 1991 ;84:157-162.

Downloaded from jov.arvojournals.org on 09/25/2021 330 Bacon et al. IOVS, February 1998, Vol. 39, No. 2

31. Anderson DF, MacLeod JDA, Baddeley SM, et al. Seasonal allergic 40. Kay AB, Sun Ying, Varney V, et al. Messenger RNA expression of conjunctivitis is accompanied by changes in mast cell numbers in the cytokine gene cluster, interleukin 3 QL-3), IL-4, 11-5, and gran- the absence of leukocyte infiltration. Clin Exp Allergy. 1997;27: ulocyte/macrophage colony-stimulating factor, in allergen-induced 1060-1066. late phase cutaneous reactions in atopic subjects. / Exp Med, 32. Bacon AS, Tuft SJ, Metz DM, et al. The origin of keratopathy in 1991;173:775-778. chronic allergic eye disease: a histopathological study. Eye. 1993; 41. McKenzie ANJ, Li X, Largaespada D, et al. Structural comparison 7:S21-S25. and chromosomal localisation of the human and mouse IL-13 33. Metz DP, Bacon AS, Holgate ST, et al. Phenotypic characterization genes. / Immunol. 1993;15O:5436. of T cells infiltrating the conjunctiva in allergic eye disease. / 42. Bruinjnzeel PLB, Warringa RAJ, Kok PTM, Kreukniet J, Wardlaw J, Allergy Clin Immunol. 1996;98:686-696. Kurlat L, Kay AB. Inhibition of neutrophil- and eosinophil-induced 34. Trocme SD, Kephart GM, Allansmith MR, Bourne WM, Gleich GJ. chemotaxis by nedocromil sodium cromoglycate. Br JPharmacol. Conjunctival deposition of eosinophil major basic protein in vernal 1990;99:798-802. keratoconjunctivitis and contact lens-associated giant papillary 43. Moqbel R, Cromovet O, Walsh GM, et al. The effects of conjunctivitis. Am J Ophthalmol. 1989;108:57-63. nedocromil sodium (Tilade) on activation of human eosino- 35. Okawara Y, Yamanchi K, Maruyama N, et al. In situ expression of phils, neutrophils and histamine release from mast cells. Al- the cell adhesion molecules in bronchial tissues from asthmatics lergy. 1988;43:268-276. with air flow limitation: In vitro evidence of VCAM-l/VLA-4 interaction in selective eosinophil infiltration. Am J Respir Cell Mol 44. Twentyman OP, Sams VR, Holgate ST. Albuterol and nedocromil Biol. 1995;12:4-12. sodium affect airway and leukocyte responses to allergen. Am Rev Respir Dis. 1993;l47:l425-l430. 36. MacLeod JDA, Anderson DF, Baddeley SM, McGill JI, Holgate ST, Roch WR. Immunolocalisation of cytokines to mast cells in normal 45. Carlos TM, Harlan JM. Membrane proteins involved in phagocyte and allergic conjunctiva. Clin Exp Allergy. 1997;27:1328-1334. adherence to endothelium. Immunol Rev. 1990;l 14:5-28. 37. Bradding P. Human mast cell cytokines. Clin Exp Allergy. 1996; 46. Wegner CD, Gundel RH, Reilly P, Haynes N, Letts LG. Control of 26:13-19. inflammatory process by adhesion glycoproteins. Agents Actions 38. Thornhill MH, Wellicome SM, Mahiouz DL, Lanchbury JSS, Kyan- Suppl. 1993;4l:S47-S57. Aung V, Haskard D. Tumor necrosis factor combines with IL-4 or 47. Gundel RH, Wegner CD, Torcellini CA, Letts LG. The role of IFN-gamma to selectively enhance endothelial cell adhesiveness intercellular adhesion molecules in chronic airways inflammation. for T cells. J Immunol. 1991; 146:592-598. Clin Exp Allergy. 1992;22:569-575. 39. Wershil BK, Wang Z-S, Gordon JR, Galli SJ. Recruitment of neutro- 48. Rosenbaum JT, Boney RS. Efficacy of antibodies to adhesion mol- phils during IgE-dependent cutaneous late phase reactions in the ecules, CDlla or CD11B, in a rabbit model of uveitis. Curr Eye mouse is mast cell-dependent. / Clin Invest. 1991 ;87:446-453. Res. 1993;12:827-831.

Downloaded from jov.arvojournals.org on 09/25/2021