Approval for the study was obtained from the local Ethics Research Board at the Faculty of Medicine and Dentistry (University of Alberta) and all adult subjects gave their informed consent according to the Helsinki protocol. Blood eosinophils from atopic donors were purified as previously described 19 . Briefly, venous blood (100 ml) was collected in tubes with heparin. Red cells were sedimented using Dextran 6% (Sigma-Aldrich Canada Ltd. Oakville, Ontario, Canada). Granulocytes were separated from mononuclear cells by centrifugation on Ficoll Paque. Eosinophils were separated from neutrophils by CD16 immunomagnetic negative selection using a magnetic cell sorter (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany). Purity of eosinophil preparations was always greater than 98%, the contaminating cells being neutrophils and/or lymphocytes.

SCC9 (American Type Culture Collection (ATCC), Manassas, Virginia) were seeded at 5 × 10⁴ cells/ml in 12 well plates 24h prior addition of eosinophils at a range of eosinophil-SCC9 concentrations. Plates were cultured until confluence of SCC9 cells was reached in control wells. Eosinophils were then removed and adherent cells were fixed with 4% paraformaldehyde (Sigma-Aldrich Canada Ltd. Oakville, Ontario, Canada) for 20 minutes and stained with crystal violet 1% (Sigma-Aldrich Canada Ltd. Oakville, Ontario, Canada). After a PBS wash, cells were lysed with 15% acetic acid and absorbance (550 nm) was measured. Absorbance was in the linear range for 2 × 10² to 2 × 10⁶ eosinophils/well.

The transmigration of eosinophils through the basement membrane components was evaluated in 24-well Biocoat Matrigel Invasion Chambers® (Becton Dickinson, Bedford, MA) as previously described 20 . Before the assay, SCC-9 cells were allowed to grow to confluence in the lower chambers, washed and fresh media replaced before the addition of 0.5 × 10⁶ eosinophils in the upper chamber. PGD₂ synthase inhibitor HQL-79 (Cayman Chemical, Ann Arbor, MI.) (0.1-100 μM) was added to lower chambers to block the production of PGD₂. HQL-79 is an orally available selective inhibitor of hematopoietic prostaglandin synthase specific with an IC50 of 100 uM for PGD2 but marginally affect the production of other prostanoids 21 22 . The percentage of transmigration was calculated by dividing the number of cells in the lower chamber of the Matrigel Invasion Chamber® by total number of cells in the lower (migrated) and upper chamber (non-migrated). The proportion of cells recovered was always >95% of the number of added cells and viability always above 98% as determined by annexin V/PI assay (n = 3).

Measurement of EPO release has been previously described 19 23 . Briefly, eosinophil suspensions (5 × 10⁵ cells/ml HBSS + CaCl₂ 1.6 mM, 0.1% gelatin) were distributed in 96-well plate. Cells were incubated for 30 min at 37°C. Peroxidase substrate solution for the measurement of EPO was prepared by adding 800 μL of 5 mM O-phenylenediamine HCl (OPD) (Sigma-Aldrich Canada Ltd. Oakville, Ontario, Canada) to 4 ml of 1M Tris buffer (pH 8.0) and 1.25 μL of 30% hydrogen peroxide (Sigma). Distilled water was added to a total volume of 10 ml. The OPD solution was added to each well of eosinophils, incubated for 2 minutes before the reaction was stopped with the addition of 4M H₂SO₄ (Sigma-Aldrich Canada Ltd. Oakville, Ontario, Canada). Each experiment was done in triplicate and absorbance reading at 490 nm wavelength was done for this colorimetric assay (Softmax, 490 nm wavelength).

Detection of PGD₂ in media after transmigration was achieved using by EAI according to manufacturer instructions (Prostaglandin D₂ EIA Kit, Cayman chemical, Ann Arbor, MI).

The apoptosis/viability assay was performed using the BD Bioscience Annexin-V-A488 detection kit according to manufacturer instructions. Acquisition was performed using BD FACS-CANTO flow cytometer. Viable cells were double negatives for ToPRO-3 and Annexin-V-A488.

Eosinophils were incubated in the presence of anti-CRTH-2-PE (MB16, RatIgG2a, Miltenyi Biotec, Auburn, CA) or matched isotype control antibody. Direct staining was used to detect the presence of specific surface binding with BD FACScanto flow cytometer.

All results are expressed as mean ± standard error of mean. Comparison between the groups was made using analysis of variance (ANOVA; Statview 5.0, SAS Institute, Cary, NC). A p-value <0.05 was considered significant.

Paraffin embedded oral squamous carcinomas specimens were obtained post-surgically from a cohort of 21 subjects. Evidence of eosinophil infiltration was observed around the tumor mass. Figure 1A depicts a specimen with massive eosinophil infiltration in connective tissue. In contrast, very few eosinophils were detected in the tumour cellular compartment (Figure 1B).

To examine the capacity of OSC to elicit eosinophil migration we used a migration system that uses an artificial cell basement membrane (Matrigel™) to separate compartment. Eosinophils did not migrate spontaneously through the lower chamber in the absence of SCC9 cells (7 ± 2%, n = 20). In comparison, the confluent monolayer of SCC9 cells induced migration toward OSC through matrigel (Figure 2, 32 ± 5% n = 17, p < 0.05). To determine whether PGD₂ played a role in this migration, a PGD₂ synthase inhibitor (HQL-79) was used. Before migration, the OSC monolayer was washed with fresh media to ensure only de novo generation of PGD₂ was affected by the HQL-79 inhibitor. A dose response effect of HQL-79 on SCC9 induced eosinophil migration was observed. At a maximal concentration (100 μM) HQL-79 reduced the migration to a level similar to background spontaneous migration (11 ± 7%, n = 9, p < 0.05). The positive control, CCL-11 (eotaxin-1), induced migration at a level of 43 ± 6% (n = 20, p < 0.05). HQL-79 did not impair CCL-11-induced migration or reduced eosinophil viability as determined by flow cytometry (data not shown). The secretion PGD₂ was also confirmed by enzyme-linked immunosorbent assay (EIA) in SCC9 confluent wells after transmigration, a mean concentration 1.5 ± 0.4 nM of PGD₂ was detected in the 4 experiments tested. The addition of HQL-79 (100 μM) reduced the PGD₂ concentration below detection limit.

Figure 3A shows a representative confluent layer of OSC after 5 days of culture. In contrast, adding eosinophils prevented OSC from reaching confluence (Figure 3B). In an attempt to potentiate eosinophil survival and function and potentially decrease tumor growth, we added IL-2 (100 ng/ml), IL5 or a combination to the culture media. Figure 3C depicts the mean results of 5 independent co-culture assays. Maximum growth inhibition (60 ± 7%, n = 5, p < 0.05) was achieved in co-culture with a 25:1 ratio of eosinophil: SCC9. The growth of OSC was unaffected by treatment with IL-5 alone (10 ng/ml; SCC9 + IL-5). In conditions were SCC9 + IL-5 + eosinophils (25:1) were co-incubated inhibition reached only 18 ± 4% (n =5, p < 0.05). The addition of IL-5 and IL-2 increased SCC9 growth inhibition when co-incubated with eosinophils ratio of 10:1 (25 ± 6%, n = 5) and 25:1 (39 ± 8%, n = 5, p < 0.05).

Eosinophil peroxidase (EPO) is a potent cytotoxic protein stored by eosinophils. We observed two forms of EPO release: first from necrotic non-IL-5-treated eosinophils and those stimulated by IL-2. In contrast, negligible EPO activity was detected in the supernatant of IL-5 treated co-culture of cells associated with the minimum growth inhibition (Figure 4A). Eosinophils viability in co-culture medium was measured immediately after control OSC reached confluence (after 5 days). Eosinophil viability was reduced under medium alone conditions but remained well above 80% in medium containing IL-5 (Figure 4B). In IL-2 + IL-5-treated groups, eosinophil viability remained unchanged indicating that the EPO activity measured was unlikely to be attributable to necrosis. Since IL-5 alone did not induce significant EPO release or OSC growth inhibition, IL-2 plus IL-5 appeared to induce EPO release without exerting an effect on eosinophil viability thereby contributing to OSC growth inhibition. These results indicated that OSC growth inhibition in untreated medium might be a result of eosinophil necrosis, cell membrane disruption and inevitable toxic mediator release by cytolysis. Maximum EPO activity correlated with maximum growth inhibition. These results suggest that cytotoxic mediator release may be a mechanism by which eosinophil inhibit the growth of OSC.