**Background:** Dry eye syndrome (DES) is a common ocular condition characterized by tear film instability and inadequate tear production, leading to tissue damage. Existing in vitro models fail to accurately replicate the complex microenvironment of the ocular surface, and animal models have interspecies limitations. Organ-on-a-chip (OoC) technology offers a promising approach by enabling controlled application of external stimuli such as air flow. This study aimed to develop an air–liquid interface (AL) within a corneal epithelium-on-a-chip (CEpOC) to model evaporative DES and investigate the effects of diclofenac (DCF), a non-steroidal anti-inflammatory drug with controversial efficacy in DES.
**Methods:** The CEpOC device was modified by adding an open reservoir (4 mm diameter) at the outlet and integrating a syringe pump with an air compressor tube to generate AL cycles. Human corneal epithelial cells (HCE-T) were cultured in the device for 7 days to form a barrier, confirmed by immunofluorescence staining of ZO-1, F-actin, CK12, and P-gp. Cells were then subjected to AL stimulus cycles (10 s per cycle: 1 s air, 1 s liquid, 8 s pause) for 24 h, with or without 50 μM DCF. Cell viability was assessed using Calcein AM staining. Transcriptomic analysis was performed via RNA-seq, and extracellular metabolites were analyzed using non-targeted LCMS at 0, 3, 6, and 24 h.
**Key Results:** Cell viability showed no significant decrease under AL or AL_DCF conditions, though cell area slightly decreased (not statistically significant). RNA-seq identified 1148 differentially expressed genes (DEGs) in AL vs. static control (ST) (417 upregulated, 731 downregulated) and 1139 DEGs in AL_DCF vs. ST (310 upregulated, 829 downregulated). GO pathway analysis revealed upregulation of developmental processes, cell differentiation, and cell adhesion, and downregulation of tissue development, epithelial development, keratinization, and cornification in both AL and AL_DCF. Key DES-related changes included upregulation of PIEZO2 (mechanosensing), IL-6, CXCL2, CCL5, and ECM remodeling genes (COL3A1, COL1A2, MMP1, MMP2), and downregulation of OCLN (tight junction), MUC16, MUC20, MUC4, MUC2, and MUC6 (mucins). DCF did not reverse IL-6 overexpression or mucin/occludin downregulation. Metabolomic analysis identified 121 annotated metabolites. VIP scores highlighted methyl-2-oxovaleric acid, 3-methyl-2-oxobutanoic acid, palmitoyl-carnitine, myristoyl-carnitine, lauroyl-carnitine, and decanoyl-carnitine as key metabolites. Under AL, methyl-2-oxovaleric acid and 3-methyl-2-oxobutanoic acid showed apical AUCs of 111.8±10 and 124.8±14, respectively; these increased with AL_DCF to 151.6±24 and 159±30. Lauroyl-carnitine and decanoyl-carnitine showed basolateral AUCs of 52.49±13 and 55.63±25 under AL, with slight increases under AL_DCF. Gene expression analysis showed upregulation of ACAT1, SLC16A7, CPT1A, and SLC25A20 under AL and AL_DCF.
**Clinical Implications:** This CEpOC model successfully recapitulates key features of evaporative DES, including inflammation, disrupted epithelial structure, and altered metabolism. The findings suggest that AL stimulation induces early-stage DES-like changes, and DCF does not effectively mitigate these effects at the transcriptomic or metabolomic level, potentially explaining its controversial clinical outcomes. The model provides a platform for studying DES mechanisms and screening therapeutics, though limitations include the lack of immune cells and serum-free conditions needed to model later disease stages.