Stress-Induced Disruption of Circadian and Neuroimmune Networks in Lacrimal Glands Drives Dry Eye Pathogenesis

Abstract

Purpose

Chronic psychological stress is an important but poorly defined risk factor for dry eye disease. This study investigated how sustained stress alters circadian regulation in the lacrimal gland (LG) and the role of neuroendocrine signaling in stress-induced dry eye disease.

Methods

Male C57BL/6J mice underwent two weeks of daily restraint stress with or without β-adrenergic blockade or glucocorticoid synthesis inhibition. Tear secretion was assessed by pilocarpine-stimulated phenol red thread test. LGs were collected every three hours across 24 hours for bulk RNA sequencing, and rhythmicity was analyzed with Jonckheere-Terpstra-Kendall cycle (JTK_CYCLE) algorithm. Single-cell RNA sequencing at zeitgeber time 6 mapped stress hormone receptor expression across LG cell types. Histology, immunohistochemistry, and plasma hormone assays characterized structural, immune, metabolic, and neural changes.

Results

Stress reduced tear secretion, altered LG morphology, and activated sympathetic and hypothalamic–pituitary–adrenal axis. Transcriptomic profiling showed a marked loss of rhythmic gene expression in immune, metabolic, neural, and proliferative pathways, whereas core clock oscillators remained intact, defining circadian output decoupling. This was associated with disrupted diurnal immune trafficking, reduced neuronal activity, impaired metabolism, and diminished proliferation. Single-cell analysis revealed β-adrenergic and glucocorticoid receptor expression in epithelial, endothelial, and immune populations. Propranolol or metyrapone partially restored rhythmic transcription, glandular architecture, and tear secretion.

Conclusions

Chronic stress induces circadian output decoupling in the lacrimal gland, driving neuroimmune and metabolic dysfunction and dry eye disease pathogenesis. Partial rescue by β-adrenergic and glucocorticoid pathway inhibition highlights neuroendocrine signaling as a therapeutic target and establishes the LG as a peripheral model of stress-induced circadian disruption.

Dry eye disease is a common, multifactorial disorder of the ocular surface characterized by tear film instability, ocular discomfort, and inflammation.¹ The global prevalence of dry eye disease ranges from 5% to 50%, influenced by demographic, behavioral, and environmental factors.² In addition to well-established contributors such as aging, digital screen exposure, and air pollution,^3–8 psychological stress has emerged as an important but underexplored risk factor that exacerbates both the onset and severity of dry eye disease.^9–11 Despite increasing clinical recognition, the mechanistic link between chronic stress and lacrimal gland (LG) dysfunction remains poorly defined.

In mammals, psychological stress activates two key neuroendocrine axes-the sympathetic nervous system (SNS) and the hypothalamic-pituitary-adrenal (HPA) axis.^12,13 Persistent activation of these systems leads to sustained release of catecholamines (norepinephrine and epinephrine) and glucocorticoids, which signal through β-adrenergic and glucocorticoid receptors, respectively.¹⁴ The lacrimal gland, a highly innervated exocrine tissue expressing both receptor types,¹⁵ is likely a direct target of stress-induced neurohormonal signaling. However, how this neuroendocrine activation alters lacrimal gland structure, transcriptional dynamics, and secretory function remains largely undefined.

Concurrently, the lacrimal gland operates under strong circadian regulation. A wide array of physiological outputs-tear production, epithelial renewal, and immune cell trafficking-display robust diurnal oscillations driven by the central pacemaker in the suprachiasmatic nucleus and modulated by hormonal and autonomic cues.^16–21 Mounting evidence indicates that chronic stress can perturb both central and peripheral clocks, disrupting circadian output in multiple organs such as the liver, pancreas, and adrenal glands.^22–25 However, whether and how psychological stress affects the circadian architecture and time-dependent physiology of the lacrimal gland remains unknown.

Recent studies suggest that a variety of physiological and pathological conditions—including nutritional challenge, time-restricted feeding, and aging—can lead to a functional uncoupling between core circadian oscillators and their downstream transcriptional programs.^26–29 This state, sometimes referred to as “circadian output decoupling,” denotes a persistent phase-shift or damping of physiological rhythms (immune, metabolic, neural) while the core molecular oscillator remains intact (i.e., intact rhythmicity of Arntl, Clock and Per2 is unchanged). This concept provides a framework for distinguishing peripheral circadian dysfunction from intrinsic clock disruption and has been documented in the liver, adrenal gland, pancreas, and brain under systemic stress.^26–29

To address this gap, we used a validated mouse restraint stress (RS) model to investigate the impact of chronic psychosocial stress on the circadian transcriptome, structure, immune function, and tear secretion of the lacrimal gland. Using high-resolution time-series RNA sequencing, we found that chronic stress reprograms the circadian transcriptome of the LG, particularly impairing rhythmic expression in immune, metabolic, and neural pathways, while leaving core clock genes largely intact. This dissociation between core oscillators and downstream rhythmic outputs defines a phenomenon of circadian output decoupling. Notably, pharmacological inhibition of β-adrenergic signaling or glucocorticoid synthesis partially restored rhythmic gene expression and glandular homeostasis. Together, these findings identify the lacrimal gland as a novel peripheral site of stress-induced circadian output failure, and uncover neuroendocrine decoupling as a pathophysiological mechanism linking psychological stress to dry eye disease.

Materials and Methods

Experimental Design

Eight-week-old male C57BL/6J mice were acclimated for two weeks under a controlled 12-hour light/dark cycle and randomly assigned to four groups: normal control (NC), RS, RS + propranolol (RS + Pro), and RS + metyrapone (RS + Met) (Fig. 1A). The RS group received RS daily for 14 consecutive days, whereas the RS + Pro and RS + Met groups were additionally given pharmacological treatment. After the intervention, body weight was recorded and blood samples were collected to evaluate activation of the SNS and HPA axis by measuring plasma levels of corticosterone, epinephrine, norepinephrine, corticotropin-releasing hormone (CRH), and adrenocorticotropic hormone (ACTH) (Fig. 1B). Tear secretion was measured after pilocarpine-induced stimulation (Fig. 1B). For circadian transcriptome analysis, LGs were harvested every three hours over a 24-hour cycle for RNA extraction and RNA sequencing. Rhythmic genes were identified using the Jonckheere-Terpstra-Kendall cycle (JTK_CYCLE) algorithm, and downstream analyses included Kyoto Encyclopedia of Genes and Genomes (KEGG), gene ontology (GO), phase set enrichment analysis (PSEA), gene set enrichment analysis (GSEA), and time-series clustering (Fig. 1C). To assess stress-induced structural and functional changes, additional LG samples were collected every six hours for histological evaluation of acinar cell morphology, immune cell infiltration, and lipid droplet content (Fig. 1D). Adrenal gland and superior cervical ganglion (SCG) tissues were collected from NC and RS mice at zeitgeber time (ZT) 6 for histological evaluation and transcriptomic profiling (Fig. 1E).

Figure 1.

Schematic overview of experimental design, intervention protocols, and analytical workflow. (A) Male C57BL/6J mice were randomized into four groups: NC, RS, RS + Pro, and RS + Met. (B) Measurement of hormone levels and tear secretion. Blood samples were collected following the intervention period for measurement of circulating CRH, ACTH, corticosterone, epinephrine, and norepinephrine. (C) LGs were collected every three hours across a 24-hour period for transcriptome profiling. (D) Additional LG samples were collected every six hours for histological assessments. (E) Adrenal gland and SCG samples were collected for histological assessments and transcriptomic profiling.

Animals

Eight-week-old male C57BL/6J mice were obtained from the Model Animal Research Institute of Nanjing University (Nanjing, China) and housed under a 12-hour light/dark cycle (lights on at 7:00 AM, lights off at 7:00 PM) in a temperature- and humidity-controlled circadian chamber (Guangzhou Langge Biotechnology Co., Ltd., Guangzhou, China). ZT is used to indicate the time relative to the onset of light (ZT0 = 7:00 AM; ZT12 = 7:00 PM). Mice had as desired access to standard chow and water. All animal procedures were approved by the Animal Ethics Committee of Henan Provincial People's Hospital (approval no. HNEECA-2022-20) and conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. At the study endpoint, mice were euthanized by isoflurane anesthesia followed by cervical dislocation.

Psychological Stress Model

Psychological stress was induced using the RS paradigm as previously described.^30,31 Mice in the RS group were individually placed into 50 mL perforated conical tubes (Jet Biofil, Guangzhou, China) for two hours daily between ZT3 and ZT5 over a 14-day period. During restraint, food and water were withheld. The tubes contained 0.4 cm diameter ventilation holes to prevent hypoxia and heat accumulation while allowing limited movement. Control mice were subjected to the same two-hour period of food and water deprivation without physical restraint.

Measurement of Physiological Parameters

After the final stress session, whole blood was collected at ZT6 via retro-orbital sinus puncture under light anesthesia. Plasma levels of corticosterone (AF2061-A; Aifang Biological, Changsha, China), norepinephrine (AF2533-A; Aifang Biological), epinephrine (AF2351-A Aifang Biological), CRH (AF2509-A; Aifang Biological), and ACTH (AF2554-A; Aifang Biological) were quantified using enzyme-linked immunosorbent assay kits according to the manufacturer's instructions. All samples were assayed in duplicate, and hormone concentrations were calculated based on standard curves.

Pharmacological Intervention

Mice in the RS + Pro group received propranolol (0.5 g/L in drinking water; S4076; Selleck Chemicals, Houston, TX, USA), yielding an estimated daily intake of 60–75 mg/kg based on water consumption. Propranolol blocks peripheral β₁/β₂ receptors on the lacrimal gland and, on central penetration, dampens locus-coeruleus sympathetic drive.³² This dose is adapted from prior studies demonstrating effective β-adrenergic blockade in chronic stress models without toxicity.³³ Mice in the RS + Met group received metyrapone (S5416; Selleck Chemicals), an adrenal CYP11B1 inhibitor that selectively suppresses corticosterone synthesis without extra-adrenal activity. Metyrapone was dissolved in sterile saline solution and intraperitoneally injected at 50 mg/kg body weight, 30 minutes before each stress session during the two-week intervention.^34–36 The NC, RS, and RS + Pro groups received equivalent volumes of saline intraperitoneally.

Tissue Collection, Library Preparation, and RNA Sequencing Analysis

After psychological stress exposure, LGs were collected from the left side of euthanized mice (Groups 1–4) every three hours over a 24-hour circadian cycle (ZT0, ZT3, ZT6, ZT9, ZT12, ZT15, ZT18, ZT21).^16,37,38 Adrenal gland and SCG tissues were collected from mice in the four groups at ZT6.^39,40 Total RNA was extracted using the RNAeasy Spin Column Kit (Qiagen, Hilden, Germany), and RNA sequencing was performed in biological triplicates. Library preparation and sequencing were conducted as previously described.^16,37,38 RNA concentration was determined using a NanoDrop spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA), and mRNA was isolated using oligo(dT)-conjugated magnetic beads. First- and second-strand cDNA synthesis and library construction were performed according to standard protocols, including circularization using phi29 DNA polymerase. Libraries were sequenced on the BGISEQ-500 platform (BGI Group, Shenzhen, China) to generate approximately 20 million single-end 50-bp reads per sample.

Raw reads were filtered using SOAPnuke (v2.3; https://github.com/BGI-flexlab/SOAPnuke) and aligned to the mouse reference genome (GCF_000001635.26_GRCm38.p6) using HISAT2 (v2.0.4; http://www.ccb.jhu.edu/software/hisat).^41,42 Gene expression levels were quantified as fragments per kilobase of transcript per million mapped reads (FPKM). Differentially expressed genes (DEGs) were identified using DESeq2 (v4.3.3) with thresholds of adjusted P < 0.05 and fold change (FC) ≥ 1.20 or ≤ 0.83 between NC and RS groups.

Identification of Rhythmic Genes

Circadian rhythmic genes in LGs were identified using the JTK_CYCLE algorithm (version 3.1) implemented in R, as previously described.^37,38,43,44 Temporally ordered gene expression data, quantified as FPKM, were analyzed across multiple ZT points to detect transcripts exhibiting 24-hour rhythmicity. The algorithm estimated the circadian phase and amplitude for each gene. Genes were categorized as follows: lowly expressed (FPKM < 0.1), rhythmic (FPKM ≥ 0.1 with adjusted P < 0.05), or non-rhythmic (FPKM ≥ 0.1 with adjusted P ≥ 0.05). This classification enabled robust identification and stratification of diurnally rhythmic transcripts for downstream clustering and pathway enrichment analyses.

Functional Annotation With KEGG, GO, PSEA, and GSEA

Functional annotation of LG gene sets was performed using the KEGG, GO, PSEA, and GSEA frameworks, following established protocols.^16,18,37,38 KEGG and GO enrichment analyses were conducted using the Dr. Tom online platform (BGI Group), with statistical significance defined as a Benjamini-Hochberg adjusted Q value < 0.05. PSEA (version 1.1) was performed using canonical KEGG gene sets from the Molecular Signatures Database (c2.cp.kegg.v7.2.symbols.gmt), with Q < 0.05 considered significant. GSEA (version 3.0) was used to evaluate enrichment of biological pathways and molecular functions based on the GO biological process gene sets from the Molecular Signatures Database (c5.go.bp.v7.5.symbols.gmt), applying significance thresholds of false discovery rate (FDR) < 0.25 and |normalized enrichment score (NES)| > 1.

Time-Series Clustering Analysis and Protein-Protein Interaction Network

Temporal patterns of rhythmic gene expression in LGs were analyzed using fuzzy c-means clustering implemented via the Mfuzz package (version 2.54.0) in R.^38,45 The number of clusters was set to four for the NC, RS, RS + Pro, and RS + Met groups based on visual inspection; all other parameters were maintained at default settings. PPIN analysis was performed using the STRING database (https://string-db.org/), which integrates both experimental and curated interaction data. A medium confidence score threshold was applied, and K-means clustering was used to partition the network into three functional modules. This combined strategy enabled detailed visualization of oscillatory gene clusters and facilitated the identification of circadian- and stress-responsive subnetworks in LGs.

Immunohistochemistry of LGs, Adrenal Glands, and SCGs

Immunohistochemical analysis of LGs, adrenal glands, and SCGs was conducted to assess temporal cellular dynamics, as previously described.^{19,39,40,44,46} Tissues were collected every six hours across a 24-hour circadian cycle (ZT0, ZT6, ZT12, ZT18). Primary antibodies included anti-c-Fos (GB12069-100; Servicebio, Wuhan, China), anti-steroidogenic factor-1 (Cat No.18658-1-AP; Proteintech, Wuhan, China), anti-CD4 (GB13064-2; Servicebio), anti-CD8 (GB13429; Servicebio), anti-CD19 (GB11061-1; Servicebio), and anti-PCNA (GB11010; Servicebio) to label neuronal activation, immune subsets, and proliferating cells. Lipid droplets and nerve fibers were visualized via Oil Red O (G1016; Servicebio) staining and immunofluorescence labeling with anti-βIII Tubulin (GB12139; Servicebio) and anti-tyrosine hydroxylase (TH) antibodies (GB11181-100; Servicebio). Isotype-matched controls were used to verify staining specificity. Microscopy was performed using the Pannoramic 250/MIDI system (3DHISTECH Ltd, Budapest, Hungary). Quantification of TH⁺, c-Fos⁺, steroidogenic factor-1(SF-1)⁺, CD4⁺, CD8⁺, CD19⁺, and PCNA⁺ cells was expressed as the percentage of positive cells relative to total nuclei within defined regions, using CaseViewer software (v2.4).

Single-Cell Sample Preparation and Analysis

Single-cell RNA sequencing was performed following a previously established protocol with minor modifications.⁴⁷ To minimize circadian variability, LGs were harvested consistently at ZT6 from NC (n = 3) and RS (n = 3) mice. Tissues were minced into 1 mm² fragments and enzymatically dissociated into single-cell suspensions using Collagenase D (0.5 mg/mL) and DNase I (0.2 mg/mL). Suspensions were washed and resuspended in calcium-and magnesium-free PBS containing 0.04% BSA.

Approximately 10,000 cells per sample were loaded onto a Chromium Single Cell Controller (10x Genomics, San Francisco, CA, USA) to generate Gel Bead-in-Emulsion droplets. Reverse transcription, cDNA purification (DynaBeads MyOne Silane; Thermo Fisher Scientific), amplification (14 PCR cycles), and library construction were performed following standard protocols. Libraries were sequenced on an Illumina HiSeq X Ten platform (Illumina, San Diego, CA, USA). Raw data were processed with Cell Ranger (v7.1.0) for demultiplexing, alignment, and UMI quantification, and converted into Seurat objects (v4.3.0.1) in R.

Cells expressing fewer than 300 or more than 7500 genes or >10% mitochondrial transcripts were excluded. Doublets were removed using DoubletFinder (v2.0.3).⁴⁸ After quality control, 29,119 high-quality cells were retained. Data normalization, variable gene selection (2000 genes, "vst" method), scaling, and dimensionality reduction (RunPCA, 30 PCs) were performed in Seurat. Batch effects were corrected using Harmony (v1.0).⁴⁹ Cells were clustered at a resolution of 0.8 and visualized with t-distributed Stochastic Neighbor Embedding (t-SNE). Major cell types were annotated based on canonical marker gene expression.

Measurement of Tear Secretion

Tear secretion was evaluated at ZT6 after stress intervention, as previously described.^{19,38,44,50,51} Mice received an intraperitoneal injection of pilocarpine hydrochloride (4.5 mg/kg; MedChemExpress, Monmouth Junction, NJ, USA) dissolved in physiological saline solution. After a 10-minute acclimation period, a phenol red-impregnated thread was placed at the medial canthus of the right eye for 20 seconds. Tear volume was quantified by measuring the length of dye migration along the thread.

Statistical Analysis and Software

All statistical analyses and graphical visualizations were performed using GraphPad Prism (version 9.3.1). Quantitative data are presented as mean ± SEM. Group comparisons were conducted using independent samples t-test or one-way ANOVA with Bonferroni correction. For non-normally distributed data, the Wilcoxon rank sum test or the Kruskal–Wallis H test was used, depending on the comparison context. Statistical significance was set at P < 0.05.

Results

SNS and HPA Axis Activation Validates the Stress Model

To confirm neuroendocrine stress activation, body weight was monitored after two weeks of RS. RS mice exhibited significant weight loss versus NC, RS + Pro, and RS + Met (Fig. 2A). At ZT6, serum CRH, ACTH, corticosterone, epinephrine, and norepinephrine were significantly elevated in RS relative to NC (Figs. 2B–F); propranolol and metyrapone partially blunted these increases.

Figure 2.

Physiological, hormonal, and histological assessment of RS and pharmacological intervention in mice. (A) Body weight measurements after 14 days of RS in NC, RS, RS + Pro, and RS + Met treatment groups. (B–F) Plasma levels of CRH (B), ACTH (C), corticosterone (D), epinephrine (E), and norepinephrine (F) across groups. (G) Tear secretion volume measured in each group. (H) The weight of LGs was measured in each group. (I, J) Immunofluorescence images (I) and quantification (J) of c-Fos⁺ neurons in the SCG. Scale bars: 100 µm (left and right); 25 µm (middle). (K–M) Immunofluorescence staining of adrenal glands for TH (green) and SF-1 (purple) (K), with quantification of SF-1⁺ cell proportions (L) and TH intensity (M). Scale bars: 200 µm (left and right); 50 µm (middle). (n = 6 mice per group; statistical analysis by one-way ANOVA with Bonferroni correction. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001).

RS significantly reduced tear secretion and the weight of LGs, which was partially restored by either treatment (Figs. 2G, 2H). Histological analysis of the SCG and adrenal gland further supported stress system activation. Histology of the SCG and adrenal gland showed increased c-Fos⁺ neurons in the SCG of RS mice, reduced by both interventions (Figs. 2I, 2J). In the adrenal gland, RS induced upregulation of TH and SF-1, indicating enhanced catecholamine synthesis and steroidogenesis, respectively. Both markers were partially suppressed in the RS + Pro and RS + Met groups (Figs. 2K–M).

RNA sequencing of SCG and adrenal gland from NC and RS mice corroborated these findings. In the SCG, RS upregulated 419 genes and downregulated 263 genes (Fig. 3A; Supplementary Table S1). Many of the upregulated genes were linked to sympathetic activity and metabolic signaling (Cpt1b, Cd36, Adra1a, Adrb1/2, Ppargc1b, Fos family, Atf3, Akt2), whereas downregulated genes included interferon-related transcripts (Ifit1-3, Socs3) (Figs. 3B, 3C). GSEA revealed enrichment of pathways related to metabolism, fatty acid oxidation, and neuroactive signaling (Fig. 3D). In the adrenal gland, 1206 genes were upregulated and 1007 genes downregulated under RS conditions (Fig. 3E; Supplementary Table S2). Upregulated transcripts included stress-responsive and steroid biosynthetic genes (Fosb, Gadd45a/b, Atf4, Il1b, Cyp11b1, Hspa5, MAPK family), consistent with robust adrenal activation (Figs. 3F, 3G). GSEA further showed enrichment in corticosteroid synthesis, MAPK signaling, and cellular stress adaptation pathways (Fig. 3H). Collectively, chronic RS activated both the SNS and HPA axis, with concordant hormonal, histological, and transcriptomic signatures; propranolol and metyrapone significantly attenuated these responses.

Figure 3.

Transcriptomic analyses of the SCG and adrenal gland in NC and RS groups. (A) Volcano plot of DEGs in the SCG between NC and RS groups. (B, C) Heatmap (B) and bar graph (C) showing elevated expression of stress-responsive genes in the RS group. (D) GSEA of SCG transcriptomes comparing NC and RS groups. (E) Volcano plot of differential gene expression in the adrenal gland between NC and RS groups. (F, G) Heatmap (F) and bar graph (G) showing expression levels of representative genes in the adrenal gland across groups. (H) GSEA of adrenal gland transcriptomes comparing NC and RS groups. (n = 3 mice per group; statistical analysis performed using one-way ANOVA with Bonferroni correction. *P < 0.05; **P < 0.01; ***P < 0.001).

Stress Induces Global Transcriptomic Reprogramming in Lacrimal Glands and Is Partially Reversed by Pharmacological Intervention

To assess the impact of psychological stress and pharmacological inhibition on global gene expression in LGs, RNA sequencing was performed for NC, RS, RS + Pro, and RS + Met groups. Principal component analysis (PCA) of DEGs (FC≥1.20 or ≤0.83; adjusted P < 0.05) revealed clear group separation (Fig. 4A): RS-treated LGs formed a well-separated cluster on the left along PC1 (92.8%), whereas NC clustered on the right; both RS + Pro and RS + Met shifted toward the NC profile along PC1.

Figure 4.

RS induces global transcriptomic reprogramming in lacrimal glands. (A) PCA of LG transcriptomes from NC, RS, RS + Pro, and RS + Met groups. (B) Venn diagram showing the number of DEGs identified in pairwise comparisons between NC and each of the three experimental groups. (C) Volcano plot showing differential gene expression in LGs between NC and RS groups. (D) Heatmap of DEGs across NC, RS, RS + Pro, and RS + Met groups over a 24-hour cycle. (E) GSEA of LG transcriptomes comparing NC with RS, RS + Pro, and RS + Met groups.

DEG distributions (Venn and volcano) are shown in Figures 4B and 4C. Relative to NC, RS induced 1,044 DEGs (171 upregulated, 873 downregulated; Supplementary Table S3). In contrast, RS + Pro and RS + Met showed substantially fewer changes—305 DEGs (57 upregulated, 248 downregulated) and 252 DEGs (51 upregulated, 201 downregulated), respectively—indicating mitigation of RS-induced transcriptional perturbation (Supplementary Figs. S1A; S1B, Supplementary Table S3). Heatmaps (Fig. 4D) showed a broad inversion of NC expression patterns in RS (genes high in NC were suppressed in RS and vice versa), with partial normalization in both intervention groups (Supplementary Figs. S1C, S1D).

GSEA revealed significant enrichment of RS-induced DEGs in immune and lipid metabolism pathways (Fig. 4E), including antigen processing and presentation, antimicrobial humoral response, defense response, monocarboxylic acid catabolism, and sterol metabolism. Notably, monocarboxylic acid catabolic and sterol metabolic pathways were specifically activated in the RS group and were not enriched in RS + Pro or RS + Met (Supplementary Figs. S1E, S1F), consistent with pharmacologic suppression of stress-induced metabolic reprogramming.

Together, these results show that RS reprograms the lacrimal gland transcriptome, enhancing immune activity and lipid metabolism. Blockade of β-adrenergic or glucocorticoid signaling partially reverses these changes, restoring key regulatory pathways toward homeostasis.

Circadian Transcriptomic Reprogramming in the LG Reveals Stress-and Pharmacological Intervention-Induced Modulation

Stress-Induced Circadian Disruption and Partial Re-Entrainment by Pharmacological Intervention

We profiled time series LG transcriptomes from NC, RS, RS + Pro, and RS + Met. PCA resolved distinct group clusters (Fig. 5A): RS deviated markedly from NC; RS + Pro overlapped with NC at multiple time points; RS + Met showed limited overlap. Using JTK_CYCLE (adj P < 0.05 or < 0.01), RS displayed a significant reduction in rhythmic genes versus NC (Fig. 5B; Supplementary Table S4). Re-entrainment—circadian realignment to the light–dark cycle—was partially achieved by RS +Pro and RS + Met, each restoring distinct but overlapping rhythmic gene subsets versus NC^52,53 (Figs. 5B,5C; Supplementary Table S4).

Figure 5.

Stress disrupts circadian rhythmicity of LG gene expression and partial restoration by pharmacological intervention. (A) PCA of circadian transcriptomes from NC, RS, RS + Pro, and RS + Met groups (n = 3 mice per group). (B) Venn diagrams showing the number of cycling genes in each group identified by JTK_CYCLE analysis (adjusted P < 0.05; high expression ≥ 0.1). (C) Overlap of highly expressed rhythmic genes across NC, RS, RS + Pro, and RS + Met groups. (D–G) Heatmaps showing 24-hour expression profiles of rhythmic genes uniquely detected in NC (D), RS (E), RS + Pro (F), and RS + Met (G). (H) Violin plots showing the distribution of oscillation amplitudes (log₁₀AMP) across groups. (I–K) Waffle plots displaying phase shift profiles of rhythmic genes shared between NC and RS (I), RS + Pro (J), and RS + Met (K). (L–Q) Rose plots displaying phase distribution of rhythmic genes across NC versus RS (L, M), NC versus RS + Pro (N, O), and NC versus RS + Met (P, Q). (ZT0 = 7:00 AM; ZT12 = 7:00 PM).

Heatmaps of group specific rhythmic genes showed temporal pattern divergence. In NC, 5321 rhythmic genes peaked at ZT3, ZT6, and ZT9 (Fig. 5D). RS shifted peaks to ZT6, ZT12, and ZT15 (Fig. 5E). RS + Pro restored more NC-like peaks (ZT3, ZT6, ZT9; Fig. 5F), whereas RS + Met peaked at ZT6, ZT9, and ZT15 (Fig. 5G). RS significantly altered oscillation amplitude distributions versus NC (P < 0.0001; Fig. 5H); propranolol partially normalized amplitudes, whereas metyrapone had a modest effect. Waffle plots showed phase changes in 75.1% of overlapping genes in RS versus NC, reduced to 68.2% with propranolol and 63.5% with metyrapone (Figs. 5I–K). Rose plots detailed phase distributions (Figs. 5L–Q): NC specific genes peaked at ZT6-ZT9 (Fig. 5L), RS-specific genes shifted nocturnally (ZT16.5-ZT18; Fig. 5M); propranolol recentered peaks toward daytime, closer to NC (Figs. 5N, 5O), whereas metyrapone retained night biased peaks (Figs. 5P, 5Q). Together, psychological stress profoundly reprogrammed LG circadian rhythms, and pharmacological blockade partially re-entrained timing, with propranolol generally showing greater realignment than metyrapone across analyses.

Stress Reshapes Rhythmic Gene Clustering and Output Pathways in LGs

KEGG enrichment on rhythmic genes unique to NC and RS showed that NC specific circadian genes were enriched in pathways supporting homeostasis—Hedgehog signaling, axon guidance, ECM-receptor interaction, and focal adhesion (Fig. 6A; Supplementary Table S5). RS-specific rhythmic genes were enriched in immune, metabolic, and neural pathways, including chemokine signaling, antigen processing and presentation, Toll-like and RIG-I-like receptor signaling, and Th1/Th2/Th17 differentiation, with additional enrichment in glycan degradation and neurotrophin signaling (Fig. 6B).

Figure 6.

Functional enrichment and circadian phase distribution of unique rhythmic genes in control and stress-treated lacrimal glands. (A, B) KEGG pathway enrichment analysis of rhythmic genes detected in NC (A) and RS (B) groups. Dot plots display pathway terms (Q < 0.05), with rich ratio on the x-axis and dot size indicating gene counts. SP denotes signaling pathways. (C, D) Circadian phase distribution of rhythmic genes in NC (C) and RS (D) groups. Inner circles show the number of rhythmic genes per Zeitgeber time; outer colored lines indicate KEGG pathways enriched at each circadian phase. SP denotes signaling pathways.

PSEA revealed daytime predominance of NC-specific pathways (Fig. 6C) and nighttime peaks for RS-specific pathways (Fig. 6D). RS engaged a greater number of immune related pathways than NC (e.g., antigen presentation, RIG-I-like receptor signaling), and shared pathways such as MAPK and chemokine signaling inverted phase (day in NC, night in RS).

We applied KEGG and PSEA to rhythmic genes unique to NC, RS + Pro, and RS + Met (Supplementary Tables S6, S7). NC unique rhythmic genes were enriched for cellular architecture and growth (adhesion, signaling, cytoskeletal remodeling) with predominantly daytime phase (Supplementary Figs. S2A, S2B). RS + Pro–specific rhythmic genes yielded no statistically significant KEGG categories; PSEA identified several modestly enriched pathways peaking during the light phase (Supplementary Fig. S2C). RS + Met–specific rhythmic genes showed KEGG enrichment in fewer pathways than untreated RS, primarily cell cycle control and immune regulation (Supplementary Fig. S2D) PSEA indicated that most enriched RS + Met pathways peaked during daytime with limited nighttime activity (Supplementary Figs. S2E, S2F).

In summary, RS induces a profound temporal reorganization of circadian output in the lacrimal gland, particularly affecting immune-related pathways. Pharmacological intervention could restore lacrimal function in stressed mice by reinstating circadian expression patterns.

Stress Reshapes Cluster-Dependent Transcriptomic Map and KEGG Pathways in LGs

Time series clustering classified rhythmic genes into four oscillatory clusters by peak/trough (Supplementary Table S8). Cluster 1 (peak ZT3, trough ZT12): NC 1,292; RS 578; RS + Pro 918; RS + Met 603 (Figs. 7A, 7E, 7I, 7M). Cluster 2 (peak ZT15, trough ZT12): NC 509; RS 1259; RS + Pro 672; RS + Met 717 (Figs. 7B, 7F, 7J, 7N). Cluster 3 (peak ZT6, trough ZT18): NC 2232; RS 1042; RS + Pro 1601; RS + Met 1746 (Figs. 7C, 7G, 7K, 7O). Cluster 4 (peak ZT12, trough ZT0): NC 1286; RS 1027; RS + Pro 872; RS + Met 1338 (Figs. 7D, 7H, 7L, 7P).

Figure 7.

Rhythmic gene clustering and KEGG pathway enrichment in lacrimal glands under control, stress, and pharmacological intervention conditions. Normalized circadian expression profiles of rhythmic genes in LGs were clustered into four groups for each condition: NC (A, C, E, G), RS (B, D, F, H), RS + Pro (I, K, M, O), and RS + Met (J, L, N, P). Left panels show temporal expression patterns across Zeitgeber time (ZT0–ZT24); shaded areas indicate expression amplitude, and the dark phase is marked in gray (ZT12–ZT24). Right panels show KEGG pathway annotations for each cluster (Q < 0.05). SP denotes signaling pathways. All conditions were analyzed using the same clustering parameters.

KEGG analyses revealed cluster specific functional redistribution under stress. In NC, Cluster 4 was enriched in metabolic pathways (Fig. 7D), whereas Clusters 1 and 3 were associated with key signaling pathways (Figs. 7A, 7C). In RS, Clusters 1 and 3 shifted toward metabolic functions (Figs. 7E, 7G), and Cluster 4 toward signaling (Fig. 7H). RS Cluster 2 showed robust enrichment in immune pathways (Toll-like receptor, TNF, NOD-like receptor signaling; Fig. 7F), an enrichment absent in NC Cluster 2 (Fig. 7B). After intervention, RS + Pro and RS + Met partially realigned rhythmic clustering toward NC profiles: in RS + Pro, Clusters 1 and 3 re-engaged signaling (Figs. 7I, 7K) and Clusters 2 and 4 were enriched for metabolic pathways (Figs. 7J, 7L); RS + Met showed a similar trend (Figs. 7M–P), with Cluster 3 recovering signaling and Cluster 4 remaining metabolic, with altered amplitude and phase features.

Alterations in Core Clock Gene Expression in the LGs of RS Mice and Pharmacologically Treated RS Mice

To assess core clock gene expression, we analyzed 24-hour profiles and calculated area under the curve (AUC). Line plots revealed that RS altered the temporal expression of several core clock genes at discrete time points (Fig. 8, left). For instance, Arntl showed significant deviations at four circadian time points (Fig. 8A), suggesting altered rhythmicity. However, the 24-hour AUC for Arntl remained comparable between RS and NC groups (Fig. 8A, right), indicating that overall transcript levels were preserved. A similar pattern was observed for other clock genes such as Clock and Cry2 (Figs. 8B, 8D). These genes exhibited significant changes at one to three time points, yet their total AUC values did not significantly differ from controls. For Npas2, Nr1d1, Nr1d2, Cry1, and Rorc (Figs. 8C, 8E,8F, 8I, 8J), RS caused mild temporal shifts in expression, but no significant change in total expression was detected. Intervention with propranolol or metyrapone did not substantially alter the AUC of core clock genes compared to RS alone. However, propranolol treatment modestly attenuated the temporal dysregulation observed in Nr1d1, Cry1 and Rorc (Figs. 8E, 8I, 8J), suggesting partial stabilization of rhythmicity. Metyrapone treatment had minimal effects on oscillatory patterns and overall expression levels (Figs. 8A–J). Within the two-week RS exposure, core clock transcripts showed preserved 24-hour aggregate output despite time point specific deviations; pharmacological inhibition partially mitigated these timing disruptions, with propranolol providing greater stabilization than metyrapone.

Figure 8.

Core clock genes remain rhythmic in LGs despite stress-induced circadian output disruption. (A–J) Temporal expression profiles (left panels) and total expression quantification (right panels) for ten core circadian genes—Arntl (A), Clock (B), Npas2 (C), Cry2 (D), Nr1d1 (E), Nr1d2 (F), Per3 (G), Per2 (H), Cry1 (I), and Rorc (J)—in LGs from NC, RS, RS + Pro, and RS + Met groups. Line plots show expression levels across six ZT points over a 24-hour cycle; the dark phase (ZT12–ZT24) is shaded in gray with bar graphs showing the area under the curve as a measure of transcript abundance. Statistical annotations: *P < 0.05 for NC versus RS; ^#P < 0.05 for NC versus RS + Pro; ^&P < 0.05 for NC versus RS + Met. (n = 3 mice per group; one-way ANOVA with Bonferroni correction).

Stress-Induced Neural Dysregulation and Pharmacological Rescue in Lacrimal Glands

We identified neural process–associated DEGs between RS and NC LGs (FC ≥ 1.20 or ≤ 0.83; adjusted P < 0.05). A total of 14 upregulated and 129 downregulated DEGs were identified between the NC and RS-treated groups (Supplementary Table S9). Heatmap analysis revealed marked differences in expression profiles across the 24-hour cycle, with inverse expression patterns between NC and RS groups (Fig. 9A). Both propranolol and metyrapone treatments partially normalized these expression patterns. A volcano plot further highlights representative neural-related DEGs between NC and RS (Fig. 9B). GO enrichment (Fig. 9C) implicated processes essential for neural regulation, including nervous system development, neuron projection, axon guidance, and neuronal apoptosis. GSEA further confirmed suppression of neural-related pathways in RS-treated glands, including those involved in neuron projection guidance, axonal regeneration, and regulation of synaptic plasticity (FDR< 0.25; |NES| > 1) (Fig. 9D). Leading edge heatmaps demonstrated partial restoration toward NC levels with either treatment. Figure 9E shows the PPIN of neural DEGs, highlighting the broad neural circuits disrupted by stress.

Figure 9.

Inhibition of stress signaling restores neural activity-associated pathways in lacrimal glands. (A) Heatmap showing temporal expression patterns of neural-related DEGs in LGs from NC, RS, RS + Pro, and RS + Met groups. Samples were collected every three hours over a 24-hour cycle; expression values are z-score normalized. (B) Volcano plot of neural-related DEGs between NC and RS groups. Annotated points indicate selected genes. (C) GO enrichment analysis of neural-related DEGs from the NC versus RS comparison. (D) GSEA of neural-associated pathways based on transcriptomic comparisons. Heatmaps show expression of leading-edge genes. (E) PPIN of neural-related DEGs constructed using STRING, with clusters annotated by inferred functional categories.

In summary, RS impairs neural signaling pathways in the lacrimal gland, affecting neuronal structure, function, and repair. Pharmacological inhibition of stress pathways with propranolol or metyrapone partially mitigates these effects, supporting a neuroprotective role for stress axis blockade in preserving these findings suggests lacrimal gland neural integrity.

Pharmacological Intervention Partially Restores Metabolic Pathway Alterations in Lacrimal Glands of Mice Subjected to RS

To explore stress-induced metabolic alterations in the LGs, we identified DEGs related to metabolic processes between RS-treated and NC mice (FC ≥ 1.20 or ≤ 0.83; adjusted P < 0.05). A total of 58 upregulated and 221 downregulated DEGs were identified between the NC and RS-treated groups (Supplementary Table S10). Heatmap showed widespread inversions of expression between NC and RS; propranolol and metyrapone partially normalized these patterns (Fig. 10A). A volcano plot further highlighted key metabolic DEGs between NC and RS (Fig. 10B). GO enrichment analysis showed that DEGs were significantly enriched in pathways related to lipid metabolism, protein catabolism, and carbohydrate processing (Fig. 10C). GSEA confirmed suppression of metabolic programs in RS-treated LGs, particularly those involved in very long-chain fatty acid metabolism and collagen degradation (FDR < 0.25; |NES| > 1) (Fig. 10D). Heatmaps of leading edge genes demonstrated partial restoration toward control levels with either intervention. Figure 10E depicts the PPIN of metabolic DEGs, revealing chronic stress-induced dysregulation of lacrimal gland metabolism. We performed Oil Red O staining of LGs at ZT18 to assess lipid droplet content (Fig. 10F). Quantitative analysis revealed a significant reduction in lipid accumulation in the RS group compared to NC, which was partially restored by both propranolol and metyrapone treatments (Fig. 10G; P < 0.05). Together, these data show that chronic psychological stress disrupts multiple metabolic programs in the lacrimal gland, with partial recovery following β adrenergic or glucocorticoid pathway blockade.

Figure 10.

Inhibition of stress signaling restores metabolic homeostasis in lacrimal glands under psychosocial stress. (A) Heatmap showing temporal expression profiles of metabolic-related DEGs across NC, RS, RS + Pro, and RS + Met groups over a 24-hour period. (B) Volcano plot of metabolic-related DEGs between NC and RS groups (adjusted P < 0.05). (C) GO enrichment analysis of metabolic-related DEGs. (D) GSEA of metabolic pathways based on transcriptomic comparisons. (E) PPIN of metabolic-related DEGs, with clustering based on inferred functional categories. (F) Representative Oil Red O staining images of lipid droplets in LG sections from each group. Scale bar: 25 µm. (G) Quantification of lipid droplet content across groups. (n = 6 mice per group; one-way ANOVA with Bonferroni correction. ***P < 0.001; ****P < 0.0001).

Pharmacological Interventions Partially Restore Immune Function Disruptions in Lacrimal Glands of Mice Subjected to RS

To investigate the impact of psychological stress on lacrimal gland immune function, we identified a subset of DEGs related to immune processes between RS-treated mice and NC, using thresholds of FC ≥ 1.20 or ≤ 0.83 and adjusted P < 0.05. A total of 36 upregulated and 263 downregulated DEGs were identified between the NC and RS-treated groups (Supplementary Table S11). Across the 24-hour cycle, heatmaps revealed widespread alterations with inverse expression patterns between groups (Fig. 11A); propranolol or metyrapone partially normalized these RS induced shifts. A volcano plot further illustrated immune-related DEGs between NC and RS (Fig. 11B). GO enrichment analysis of these DEGs indicated significant downregulation of biological processes involved in immune response, antigen presentation, chemokine signaling, and regulation of interleukin-6 production (Fig. 11C). GSEA confirmed suppression of key immune pathways in RS-treated LGs, including lymphocyte and leukocyte migration, acute inflammatory response, T cell cytokine production, and B cell activation (Fig. 11D). Heatmaps of leading-edge genes from these suppressed pathways demonstrated that both propranolol and metyrapone treatments partially restored immune gene expression (Fig. 11D), suggesting therapeutic reversal of stress-induced immune dysfunction. PPINs of immune DEGs (Fig. 11E) map the diverse immunological circuits disrupted by RS in the LGs.

Figure 11.

Inhibition of stress signaling mitigates immune dysregulation in lacrimal glands under psychosocial stress. (A) Heatmap showing temporal expression patterns of immune-related DEGs in LGs from NC, RS, RS + Pro, and RS + Met groups. (B) Volcano plot of immune-related DEGs between NC and RS groups (adjusted P < 0.05). (C) GO enrichment analysis of immune-related DEGs. (D) GSEA of immune-related pathways based on transcriptomic comparisons. Right panels show leading-edge gene expression. (E) PPIN of immune-related DEGs constructed using STRING, with clustering based on functional annotations.

To validate stress-induced immunosuppression at the cellular level, we performed single-cell RNA sequencing on LGs from NC (n = 3) and RS (n = 3) mice. The cellular composition shifted markedly under stress, with reductions in multiple immune populations, including Cd4⁺ and Cd8⁺ T cells, CD79⁺ B cells, Trdc⁺ T cells, innate lymphoid cells (ILCs), and mast cells (Figs. 12A–C). Quantitative analysis further confirmed significant decreases in Cd4⁺, Cd8⁺, and CD79⁺ B cell subsets in the RS group compared to controls (Figs. 12D–L). We performed immunohistochemical staining of lacrimal glands collected at ZT6. Quantification revealed significantly reduced numbers of CD4⁺ T cells, CD8⁺ T cells, and CD19⁺ B cells in RS-treated mice, all of which were partially rescued by propranolol and metyrapone interventions (Figs. 12M–O). These results support the notion that psychological stress disrupts the magnitude of immune cell infiltration in the lacrimal gland.

Figure 12.

Single-cell transcriptomic and immunohistochemical analysis of immune cell populations in lacrimal glands from NC and treated groups. (A, B) The t-SNE plots and cell composition summaries for NC (A) and RS (B) groups, different cell populations are represented by different colors for cluster identification. (C) The proportion of each cell type in the lacrimal gland of NC group and RS group. (D, E) Spatial distribution of Cd4⁺ T cells in NC (D) and RS (E) groups. (F) Quantification of Cd4⁺ T cells, showing significant reduction in RS glands (n = 3 mice per group; *P < 0.05, Independent samples t-tests). (G, H) Spatial distribution of Cd8a⁺ T cells in NC (G) and RS (H) LGs. (I) Quantification of Cd8a⁺ T cells between groups (n = 3; *P < 0.05). (J, K) Distribution of Cd79a⁺ B cells in NC and RS glands. (L) Quantification of Cd79a⁺ B cells per section (n = 3; *P < 0.05). (M–O) Representative immunohistochemical images of LG sections collected at ZT12 showing staining for immune cell markers in NC, RS, RS + Pro, and RS + Met groups: (M; left panels) CD4⁺ T lymphocytes (scale bar = 20 µm); (N; left panels) CD8⁺ T lymphocytes (scale bar = 25 µm); (O; left panels) CD19⁺ B lymphocytes (scale bar = 20 µm). Quantification of immune cell staining, expressed as the percentage of positive cells per field: (M; right panel) CD4⁺ T cells; (N; right panel) CD8⁺ T cells; (O; right panel) CD19⁺ B cells. (n = 6 mice per group; one-way ANOVA with Bonferroni correction. **P < 0.01; ***P < 0.001; ****P < 0.0001).

Together, these findings demonstrate that chronic psychological stress suppresses both immune gene expression and cell recruitment in the lacrimal gland, likely through neuroendocrine-mediated disruption of rhythmic immune trafficking. Pharmacological inhibition of stress signaling pathways partially reverses these changes, highlighting a therapeutic avenue to restore immune homeostasis in stress-associated dry eye disease.

Pharmacological Interventions Partially Restore Cell Cycle-Related Pathway Alterations in Lacrimal Glands of Mice Subjected to RS

To assess the effects of psychological stress on cell cycle regulation in the LGs, we identified DEGs associated with cell cycle processes between RS-treated mice and NC (FC ≥ 1.20 or ≤ 0.83; adjusted P < 0.05). A total of 10 upregulated and 51 downregulated DEGs were identified between the NC and RS-treated groups (Supplementary Table S12). Heatmap visualization revealed marked transcriptional differences across the 24-hour cycle, with multiple genes inversely regulated between NC and RS groups (Fig. 13A). A volcano plot further illustrated the differential expression of cell cycle-related genes between NC and RS groups (Fig. 13B). GO enrichment analysis of the identified DEGs showed significant alterations in pathways related to cell cycle progression, arrest, and apoptosis in RS-treated LGs (Fig. 13C). GSEA confirmed suppression of key regulatory programs, including positive regulation of the cell cycle and checkpoint control pathways (FDR < 0.25; |NES| > 1) (Fig. 13D). Leading edge heatmaps demonstrated partial restoration of genes involved in mitotic progression and checkpoint fidelity with either intervention.

Figure 13.

Inhibition of stress signaling restores cell cycle-related transcriptional programs in lacrimal glands under psychosocial stress. (A) Heatmap showing temporal expression patterns of cell cycle-related DEGs in LGs from NC, RS, RS + Pro, and RS + Met groups over a 24-hour circadian cycle. Expression values are z-score normalized. (B) Volcano plot of cell cycle-related DEGs between NC and RS groups (adjusted P < 0.05). Selected genes are annotated. (C) GO enrichment analysis of cell cycle-related DEGs. (D) GSEA of pathways related to cell cycle regulation. Heatmaps show expression of leading-edge genes across groups. (E) Representative immunohistochemical images of PCNA⁺ cells in LG sections collected at ZT12. Scale bar: 20 µm. (F) Quantification of PCNA⁺ cells across groups. (n = 6 mice per group; one-way ANOVA with Bonferroni correction. *P < 0.05; ***P < 0.001).

To investigate the impact of stress on circadian dynamics of cell proliferation, we analyzed the number of PCNA⁺ cells at ZT12 across treatment groups. RS exposure significantly reduced PCNA⁺ cell abundance in LGs compared to controls, whereas both propranolol and metyrapone treatments attenuated this decline (Figs. 13E, 13F), consistent with partial restoration of proliferative capacity. Collectively, these findings demonstrate that psychological stress suppresses cell cycle activity in the lacrimal gland through transcriptional and functional pathways, whereas neuroendocrine pathway inhibition partially rescues proliferative rhythms.

Single-Cell Mapping Reveals Cell-Type-Specific Hormonal Responsiveness in the Lacrimal Gland

Previous studies have demonstrated that the mouse LG is richly innervated by sympathetic nerve fibers,^15,54 supporting its regulation by autonomic neural input. Consistent with these findings, our immunofluorescence staining revealed abundant β-III tubulin⁺/TH⁺ fibers, prominently localized around acinar units and vasculature (Figs. 14A, 14B), providing anatomical support for neurogenic control of lacrimal function. To identify potential target cells for sympathetic neurotransmitters and stress hormones, we performed single-cell RNA sequencing on dissociated LGs. Unsupervised clustering and t-SNE identified 13 transcriptionally distinct cell populations, including acinar cells, ductal (basal and luminal), myoepithelial cells, endothelial cells, and immune cell types such as B cells, T cells, ILCs, and myeloid subsets (Fig. 14C). We next examined the expression of adrenergic receptor genes to map the landscape of potential catecholamine-responsive cells. Adrb1 was selectively enriched in myoepithelial cells, basal and luminal ductal epithelial cells, Trdc⁺, γδ T cells and myeloid cells (Fig. 14D). Adrb2 showed broader distribution, with expression across endothelial cells, B cells, ILCs, γδ T cells, Cd8a⁺/Cd4⁺ T cells, and myeloid compartments (Fig. 14E), indicating a wider sphere of influence for β2-adrenergic signaling across both epithelial and immune niches. Given the central role of glucocorticoids in stress physiology, we also assessed expression of Nr3c1, which encodes the nuclear glucocorticoid receptor. Nr3c1 was detected in a variety of cell types, including ductal epithelial, endothelial, and immune subsets, but was notably absent in acinar epithelial cells, suggesting selective glucocorticoid responsiveness within the LG (Fig. 14F). Quantitative comparison of receptor expression between NC-RS is shown in Supplementary Figure S3. A significant difference in receptor expression was observed between the NC and RS groups, indicating that chronic stress substantially down-regulates receptor expression (Supplementary Fig. S3). This is consistent with some previous research results.^55,56 To visualize overall receptor distribution, we plotted average expression levels of Adrb1, Adrb2, and Nr3c1 across all 13 clusters (Fig. 14G). These analyses revealed partially overlapping but distinct expression domains for β-adrenergic and glucocorticoid receptors. Together, these results define a high-resolution molecular atlas of stress hormone receptor expression in the mouse lacrimal gland, establishing a cellular framework for neuroendocrine regulation under psychological stress.

Figure 14.

Sympathetic innervation and single-cell mapping of stress hormone receptors in the mouse lacrimal gland. (A) Immunofluorescence staining reveals sympathetic nerve fibers (β-III Tubulin⁺/TH⁺) innervating perivascular regions in the lacrimal gland (yellow arrows; scale bar = 100 µm). (B) High-magnification image showing β-III Tubulin⁺/TH⁺ fibers surrounding secretory acini (yellow arrow; scale bar: 25 µm). (C) The t-SNE projection of single-cell RNA sequencing data identifying 13 transcriptionally distinct cell clusters, including epithelial, immune, and stromal populations. (D–F) Feature plots illustrating expression of Adrb1 (D), Adrb2 (E), and Nr3c1 (F) across annotated cell clusters. (G) Line plot summarizing average expression levels of Adrb1 (orange), Adrb2 (green), and Nr3c1 (magenta) across major cell types.

Discussion

Chronic psychological stress is increasingly recognized as a contributing factor in dry eye disease, yet its cellular and molecular mechanisms remain incompletely understood. In this study, we show that sustained RS alters the circadian transcriptome and disrupts the functional homeostasis of the LG, resulting in histopathological changes and reduced tear production. Activation of the SNS and HPA axis reprograms immune, metabolic, neural, and proliferative pathways without affecting core clock gene oscillations-a phenomenon referred to as circadian output decoupling. In this study, we operationally define circadian output decoupling as a state in which the rhythmicity (adjusted P < 0.05, JTK_CYCLE) of downstream functional pathways (e.g., immune, metabolic, proliferative genes) is lost or significantly phase-shifted (>4 hours), whereas the core oscillator genes (e.g., Arntl, Clock, Cry2) retain statistically significant rhythmicity with preserved amplitude over a 24-hour period. Quantitatively, this was evidenced by >70% loss or misalignment of previously rhythmic non-clock genes, and stable AUC values of core clock gene expression. Pharmacological inhibition of β-adrenergic signaling or glucocorticoid synthesis partially mitigates these stress-induced alterations, underscoring the therapeutic potential of targeting neuroendocrine pathways. Together, these findings suggest that chronic stress may modulate peripheral exocrine gland function and highlights potential therapeutic targets for the treatment of stress-associated ocular surface disorders, as illustrated in Figure 15.

Figure 15.

Chronic psychological stress disrupts lacrimal gland homeostasis via neuroendocrine activation and circadian output decoupling. The schematic depicts the relationship between neuroendocrine signaling, core clock gene oscillation, circadian output pathways (immune, neural, metabolic, and proliferative), and tear secretion. Pharmacological interventions targeting β-adrenergic receptors (propranolol) or glucocorticoid synthesis (metyrapone) are indicated.

Chronic stress concurrently activates the SNS and HPA axis, resulting in elevated levels of circulating catecholamines and glucocorticoids.^57–59 Single-cell RNA sequencing confirmed the expression of β-adrenergic and glucocorticoid receptors across major lacrimal gland cell types, supporting the potential for direct neuroendocrine regulation. Although core clock gene oscillations remained intact, time-series RNA sequencing revealed a marked loss of circadian rhythmicity in immune, metabolic, and neural pathways, consistent with circadian output decoupling.⁶⁰ Altered immune cell trafficking, dampened neurogenic and metabolic rhythms, and reduced cellular proliferation collectively contribute to glandular dysfunction and diminished tear production—hallmark features of dry eye disease.

The secretory activity of the lacrimal gland is highly dependent on autonomic innervation.¹⁵ Stress-induced sympathetic overactivation and potential parasympathetic withdrawal likely underlie observed reductions in tear secretion.^61,62 Additionally, prolonged glucocorticoid exposure impairs lacrimal cell metabolism and may induce cell death, further contributing to glandular deterioration.^63,64 Recent translational evidence has converged on the notion that psychological stress is not merely a comorbid epiphenomenon of dry eye disease, but a causal driver that precipitates LG dysfunction via neuroendocrine pathways.⁶⁵ The lacrimal gland contains a dynamic immune microenvironment that is modulated by circadian signals.^19,38,47 Chronic stress not only suppresses immune-related gene expression but also disrupts the temporal coordination of immune cell trafficking, in line with impaired clock-regulated immune rhythms.^66,67 Dysregulated adrenergic signaling and abnormal glucocorticoid secretion likely contribute to the observed immune dysregulation within the gland, including features consistent with an immunosuppressive or “immune fatigue” state.^68–71 At the metabolic level, chronic stress markedly alters lipid metabolism, mitochondrial activity, and glucose-related pathways within the LG.^72–76 Histological evidence of lipid droplet depletion suggests localized lipolytic imbalance, which may compromise acinar cell integrity. Collectively, these observations support the notion that the lacrimal gland is a metabolically responsive organ vulnerable to neuroendocrine stress signals.

Pharmacological intervention targeting the SNS with propranolol or the HPA axis with metyrapone improved stress-induced histological and functional alterations, including partial restoration of circadian rhythmicity, immune homeostasis, and tear secretion. These results highlight the potential of neuroendocrine pathway modulation as a therapeutic strategy for stress-related dry eye disease.^77–82 However, neither treatment fully reversed circadian output decoupling, indicating that chronic stress may induce lasting alterations in circadian regulatory mechanisms despite normalization of systemic hormone levels.

Despite the strengths of this study, several limitations should be acknowledged. Although rodent models provide valuable mechanistic insights, they differ from humans in circadian organization and melatonin physiology.^83,84 In addition, the experiments were conducted exclusively in male mice; given known sex-based differences in lacrimal gland biology, future studies should include both sexes to assess potential variability.^85,86 Although circadian disruption strongly correlates with hyposecretion, causal evidence awaits circadian-rescue experiments (recovery experiment or clock-gene restoration). In our study, core clock gene output appeared preserved at the transcript level under two-week stress exposure, but it remains possible that longer-term stress paradigms (e.g., four to eight weeks) could disrupt oscillator function. Future studies using extended duration models will be essential to determine whether chronic stress eventually degrades intrinsic clock machinery. Finally, although propranolol and metyrapone yielded partial improvements, complete restoration of glandular rhythmicity may require combinatorial approaches or targeted clock-based interventions.

Future translational efforts should include the validation of circadian output markers in human tear samples and the evaluation of stress-modulatory therapies in clinical populations with dry eye disease. Modulation of the neuroendocrine-circadian interface could represent a promising therapeutic strategy for mitigating stress-associated ocular surface disorders.

Conclusion

Chronic psychological stress disrupts circadian output fidelity in the lacrimal gland, uncoupling core clock oscillators from downstream immune, neural, and metabolic functions. This output decoupling leads to structural degeneration, impaired tear secretion, and immune suppression—hallmarks of dry eye disease. Pharmacological blockade of β-adrenergic signaling and glucocorticoid synthesis partially restored rhythmic gene expression and glandular homeostasis, underscoring the therapeutic potential of targeting stress-responsive neuroendocrine pathways. These findings establish the lacrimal gland as a novel peripheral model of stress-induced circadian disruption and lay the groundwork for future translational strategies in stress-related ocular surface disease.