Aberrant prostaglandin synthase 2 expression defines an antigen-presenting cell defect for insulin-dependent diabetes mellitus

Abstract

Prostaglandins (PGs) are lipid molecules that profoundly affect cellular processes including inflammation and immune response. Pathways contributing to PG output are highly regulated in antigen-presenting cells such as macrophages and monocytes, which produce large quantities of these molecules upon activation. In this report, we demonstrate aberrant constitutive expression of the normally inducible cyclooxygenase PG synthase 2 (PGS(2)/ COX-2) in nonactivated monocytes of humans with insulin-dependent diabetes mellitus (IDDM) and those with islet autoantibodies at increased risk of developing this disease. Constitutive PGS(2) appears to characterize a high risk for diabetes as it correlates with and predicts a low first-phase insulin response in autoantibody-positive subjects. Abnormal PGS(2) expression in at-risk subjects affected immune response in vitro, as the presence of a specific PGS(2) inhibitor, NS398, significantly increased IL-2 receptor alpha-chain (CD25) expression on phytohemagglutinin-stimulated T cells. The effect of PGS(2) on CD25 expression was most profound in subjects expressing both DR04 and DQbeta0302 high-risk alleles, suggesting that this cyclooxygenase interacts with diabetes-associated MHC class II antigens to limit T-cell activation. These results indicate that constitutive PGS(2) expression in monocytes defines an antigen-presenting cell defect affecting immune response, and that this expression is a novel cell-associated risk marker for IDDM.

Figure 1

Flow cytometric analysis of PGS₂ expression in CD14⁺ monocytes from freshly isolated human PBMCs. Broken line indicates control isotype fluorescence; solid line indicates anti-PGS₂ fluorescence. (a) An example of PGS₂ expression in CD14⁺ monocytes of freshly isolated PBMCs from a healthy control. (b) PGS₂ expression in LPS-activated (5 mg/mL for 24 hours) CD14⁺ monocytes from the same individual as in a. (c) PGS₂ expression in CD14⁺ monocytes of freshly isolated PBMCs from a subject at high risk for IDDM. Please note that the scales of events (y-axis) differ between the control and subject panels.

Figure 2

Percentage of CD14⁺ monocytes expressing PGS₂ is significantly higher in subjects at risk for IDDM compared with controls. Data from blood samples taken from subjects at risk for IDDM, relatives, and controls. The first 3 data sets (Controls, Relatives, All Subjects) are data taken from all participants in the study. “All Subjects” data are divided into groups based on risk (i.e., low-, moderate-, and high-risk IDDM subgroups; established diabetics [IDDM]; and ICA⁺ individuals with clinically established AI). The percentages of PGS₂⁺/CD14⁺ monocytes from relatives and all subjects were significantly different from the normal control group, as assessed by weighted Tobit analysis (P = 0.004). Solid lines indicate group means; the dotted line represents the control group mean plus 2 SDs, the level used to define high-level PGS₂ expression (9.9%).

Figure 3

PGS₂ expression in CD14⁺ monocytes correlates with FPIR, a clinical criterion for IDDM risk assessment. PBMCs from subjects at low, moderate, and high risk for IDDM were analyzed for PGS₂ expression in CD14⁺ monocytes by flow cytometry. The “Developed IDDM” designation in this data set indicates 4 high-risk-group individuals who developed diabetes during the period of observation. The correlation of maximum PGS₂ levels obtained during the observation period with insulin secretory capacity (FPIR) was determined using Prism 2.01 software (Graph Pad Software for Science Inc., San Diego, California, USA). A significant inverse correlation between insulin secretion and PGS₂ expression was found (P = 0.0201; n = 46 total subjects), with best fit being a hyperbolic curve (r² = 0.83).

Figure 4

ROC curves derived from PGS₂/CD14 expression and FPIR comparison. ROC analysis was used to examine the property of positive PGS₂ expression (>9.9%) as a predictor of low FPIR. The analysis includes a correction for the variation in time between PGS₂ and FPIR analyses. The resulting curve was generated with a c-statistic (area under the curve) of 0.9. An example of the ability of PGS₂ expression to predict a low FPIR is given in the inset table, with a decision threshold set at 0.81 (*).

Figure 5

NS398 inhibition of PGS₂ enzymatic activity affects T-cell expression of CD25 during PHA activation. PBMCs were cultured in the presence of PHA (5 μg/mL) with or without NS398 (5 μM) for 24 hours. The expression of CD25 on CD3⁺ T cells was determined by dual-color flow cytometry. “All subjects tested” includes ICA⁺ diabetics, ICA⁺ moderate-risk subjects, ICA⁺ high-risk subjects, and those classified as AI. Baseline values for the percentage of CD25⁺/CD3⁺ cells in freshly isolated PBMCs were not significantly different between the groups assayed (controls: 4.61 ± SD 4.6, n = 12; other HLA subjects: 6.2 ± SD 6.7, n = 10; and DR04/DQ0302: 4.5 ± SD 3.9, n = 14). The data are presented as a fold increase, calculated as the percent of CD3⁺ T cells expressing CD25 when activated with PHA in the presence of NS398, relative to CD25 expression on CD3⁺ T cells activated with PHA alone. For individuals with multiple sample runs, the fold increase of each assay run was averaged to give a mean fold increase shown here. Statistical analysis demonstrated a significant difference between controls and all subjects (P = 0.04; Student’s t test). When HLA data were available, the CD25 data were also analyzed by this parameter. Enhancement of T-cell CD25 expression was most prevalent in subjects expressing both DR04 and DQβ0302 (P = 0.012; 1-way ANOVA), when compared with other subjects expressing HLA DR alleles 16, 03, 01, 07, 13a, and 08 and DQβ alleles 05, 0201, 0303, 0604, and 04.