Corneal Immune Cell Alterations and Tau Pathology in a Mouse Model of Alzheimer’s Disease Authors: Di Zhang, Duntao Huang, Junju He, Yong Shen, Linbin Dai, Lei Shi Published: 2026 ABSTRACT PurposeEarly diagnosis of Alzheimer’s disease (AD) remains a formidable challenge. Although retinal markers are extensively studied, the cornea—the eye’s primary transparent window—presents a potential but poorly understood platform for biomarker discovery. This study systematically characterized AD-associated corneal pathological evolution using the 5 × FAD mouse model.MethodsCorneas from wild-type (WT) and 5 × FAD mice (ages 3, 5, 7, and 10 months; both sexes included) were analyzed using multimodal approaches: whole-mount immunostaining with 3D reconstruction, immunoblotting, RNA sequencing, and mass spectrometry proteomics.ResultsAlthough local human APP production was absent, a defined temporal cascade emerged selectively in the peripheral cornea. Dendritic cell (DC) morphological simplification (reduced perimeter and branching) occurred as early as 3 months, preceding significant DC expansion at 5 months and CD3⁺ T cell infiltration at 7 months. Tau hyperphosphorylation (Thr181) surfaced at 7 months, coinciding with heightened GSK3β activity (reduced pGSK3β Ser9). By 10 months, significant peripheral subbasal nerve plexus and superficial nerve terminals loss were observed, with residual fibers exhibiting markedly elevated phosphorylated tau burden. Multi-omics at 7 months validated this pro-inflammatory state, identifying upregulated adhesion molecules (ICAM1 and ITGB2) and the T cell chemoattractant IL-16.ConclusionsCorneal neuro-immune remodeling—manifesting as early DC dysfunction followed by T cell recruitment and tau pathology—precedes overt nerve degeneration. These findings suggest the cornea as a potential, noninvasively accessible site for monitoring AD progression. Alzheimer's disease (AD) is a progressive neurodegenerative disorder and a leading cause of dementia, marked clinically by worsening cognitive impairment and pathologically by extracellular amyloid-β (Aβ) plaques, intracellular neurofibrillary tangles (NFTs), neuroinflammation, and neuronal loss.1,2 Aβ is generated through amyloid precursor protein (APP) processing by β-secretases (BACE1) and γ-secretases, whereas NFTs are composed of hyperphosphorylated tau that disengages from microtubules, disrupts axonal transport, and contributes to neuronal loss and death.3–7 Neuroinflammation is increasingly viewed as an active component of AD pathogenesis rather than a secondary bystander.8 Microglial activation can be detected early in the disease trajectory and may be driven by both Aβ and tau pathology, creating a feed-forward inflammatory milieu that further aggravates protein aggregation and neurodegeneration.9–12 Although anti-amyloid immunotherapies have recently entered clinical practice, therapeutic benefit remains limited and treatment eligibility is generally restricted to early-stage patients, underscoring the unmet need for minimally invasive approaches that can detect and track disease-related changes before widespread neurodegeneration.13,14 Given its optical accessibility and shared features with the central nervous system, the eye has attracted attention as a source of potential AD biomarkers. The retina, in particular, has been reported to exhibit Aβ deposition, pathological tau species, neuroinflammatory changes, and neurodegenerative alterations in patients and experimental models.15–20 In practice, however, retinal imaging quality may be compromised by factors common in older individuals, including restricted pupil dilation and age-related ocular comorbidities.21,22 The cornea offers a complementary opportunity: it is transparent, avascular, and densely innervated, enabling high-resolution, noninvasive assessment with in vivo confocal microscopy (IVCM).23 The cornea also contains a structured immune network—including dendritic cells (DCs), macrophages, and T cells—that participates in tissue homeostasis and responds rapidly to perturbation.24,25 Emerging data suggest that corneal DC morphology and density may be altered in tauopathy models and in individuals with mild cognitive impairment,26,27 and that corneal nerve loss and reduced corneal sensitivity have been reported in AD and related conditions.28–30 Despite these observations, it remains unclear how corneal immune alterations relate temporally to tau-associated changes and nerve degeneration, and whether neuro-immune remodeling in the cornea follows an ordered trajectory during AD progression. To address this gap, we used the 5 × FAD mouse model, which develops robust Aβ pathology and exhibits a well-described temporal progression of amyloid accumulation, neuroinflammatory responses, and later neurodegenerative changes.31–33 We examined animals at 3, 5, 7, and 10 months of age to capture early, intermediate, and later stages of pathology. Across these time points, we systematically assessed corneal Aβ-related signals, tau phosphorylation, immune cell dynamics (with emphasis on DCs and T cells), and corneal nerve integrity. By defining the timing and regional distribution of these corneal changes, this study aims to evaluate whether the cornea can serve as a practical, noninvasively accessible tissue for monitoring AD-associated neuro-immune remodeling. Methods Animals All procedures involving wild-type (WT) and 5 × FAD mice (both sexes, approximately equal representation) were approved by the Institutional Animal Ethics Committee (202310311734000448035) and conducted in compliance with institutional guidelines. Mice were maintained under specific pathogen-free conditions (26°C, 12-hour light/dark cycle) with ad libitum access to standard diet and sterile water. Due to limited sample sizes, sex-stratified analyses were not performed, which is acknowledged as a study limitation. Sample sizes are specified in figure legends. Western Blotting Brain, retina, cornea, and trigeminal ganglion tissues were dissected on ice and lysed in RIPA buffer with protease/phosphatase inhibitors (Thermo Scientific). After homogenization (freeze grinder, 3 minutes) and sonication (3 minutes), samples were centrifuged at 12,000×g for 30 minutes at 4°C. The supernatant (detergent-soluble fraction) was collected; this speed isolates soluble tau species (predominant in the cornea), whereas insoluble aggregate analysis requires 100,000×g centrifugation. Protein concentration was determined by BCA assay (Vazyme). Ten micrograms protein per sample were separated by SDS-PAGE, transferred to PVDF membranes, blocked with 5% skim milk, and incubated with primary antibodies overnight at 4°C, followed by HRP-conjugated secondary antibodies. Bands were visualized by chemiluminescence and quantified using ImageJ software; signals were normalized to α-Tubulin or GAPDH. For 10-month corneal samples, additional normalization to βIII-Tubulin (TUJ1) assessed pathology in residual fibers. The following primary antibodies were used: rabbit anti-Tau (CST, #46687), rabbit anti-Phospho-Tau (Thr217) (ABclonal, #AP1373), rabbit anti-phospho-Tau (Thr205) (CST, #49561), mouse anti-Phospho-Tau (Ser202 and Thr205), AT8 (Invitrogen, #MN1020), rabbit anti-Phospho-Tau (Thr181) (CST, #12885), rabbit anti-Phospho-Tau (Thr231) (CST, #71429), rabbit anti-TUJ1 (Sigma, #T2200), rabbit anti-Phospho-GSK-3β (Ser9) (CST, #5558), rabbit anti-GSK-3β (CST, #12456), rabbit anti-GSK3β-Y216 (ABclonal, #AP1530), rabbit anti-CDK5 (CST, #14145), rabbit anti-APP (CST, #29765), rabbit anti-BACE1 (CST, #5606), rabbit anti-GAPDH (Proteintech, #10494-1-AP), and mouse anti-α-Tubulin (Proteintech, #66031-1-Ig). The secondary antibodies used were: anti-rabbit IgG (CST, #7074), and anti-mouse IgG (CST, #7076). Frozen Section Immunofluorescence Staining Mice were anesthetized with tribromoethanol (15–20 µL/g, intraperitoneally [IP]) and perfused transcardially with ice-cold PBS. Eyeballs were enucleated, fixed in 4% PFA (4°C, 12 hours), the corneas and retinas were dissected at the limbus to remove uveal tissue. After post-fixation (4% PFA, 2.5 hours, room temperature), tissues were dehydrated in 30% sucrose (4°C, overnight), embedded in optimum cutting temperature (O.C.T.) compound, and sectioned at 16 µm using a cryostat. Sections were air-dried (37°C, 2 hours), blocked/permeabilized with 10% normal goat serum and 0.3% Triton X-100 in PBS (1.5 hours at room temperature), and incubated with primary antibodies (4°C, overnight). After washing, sections were incubated with fluorophore-conjugated secondary antibodies and DAPI (1.5 hours at room temperature), mounted with 50% glycerol, and imaged on a Carl Zeiss LSM800 confocal microscope (40 × oil objective; identical settings within cohorts). Whole Cornea Immunofluorescence Staining and 3D Reconstruction Freshly isolated corneas were fixed in methanol (−20°C, 1 hour), treated with 20 mM EDTA in PBS (37°C, 40 minutes) to enhance permeability,34,35 then blocked/permeabilized (10% normal goat serum, 0.3% Triton X-100 in PBS; 1.5 hours, room temperature) and incubated with primary antibodies (4°C, overnight, with gentle shaking). After washing, the samples were incubated with fluorophore-conjugated secondary antibodies and DAPI (1.5 hours, room temperature), flattened with 4 radial incisions, and mounted epithelial-side up in 50% glycerol. Imaging used a Carl Zeiss LSM800 confocal microscope (20 × objective). Z-stacks (1 µm intervals) were acquired for one central and four peripheral fields. The 3D reconstruction used Imaris software; nerve fiber volume and surface area were quantified for subbasal nerve plexus (SBNP) and superficial nerve terminals (SNTs). For immune cells, maximum intensity projections were generated, and cells were counted manually using ImageJ software. DCs were defined as CD45⁺ cells with dendritic morphology (≥2 primary dendrites), verified by CD11c/major histocompatibility complex class II (MHC class II) co-staining; T cells were CD45⁺CD3⁺ round cells. DC morphology (area, perimeter, branch number, and total dendrite length) was quantified using the AnalyzeSkeleton plugin after binarization.36 The following antibodies were used for immunofluorescence: rabbit anti-Tau (CST, #46687), mouse anti-TUJ1 (Biolegend, #801201), rabbit anti-TUJ1 (Sigma, #T2200), rabbit anti-CD3 (Novus Biologicals, #NB600-1441), rat anti-CD45 (BD Pharmingen, #553076), PE anti-mouse MHC class II (Biolegend, #107607), PE anti-mouse CD11c (Biolegend, #117307), rabbit anti-Aβ (D54D2; CST, #8243), anti-rat IgG Alexa Fluor 647 (CST, #4418), anti-rabbit IgG F(ab')₂ Fragment Alexa Fluor 488 (CST, #4412), and anti-rabbit IgG F(ab')₂ Fragment Alexa Fluor 647 (CST, #4414). Paraffin Sections Immunohistochemical Staining Paraffin sections (4 µm) were deparaffinized in xylene and rehydrated in graded ethanol. Antigen retrieval was performed in EDTA buffer (pH 6.0) at 90°C for 15 minutes. Endogenous peroxidase was quenched with 3% H₂O₂ for 10 minutes. Sections were blocked with 10% normal goat serum, then incubated with rabbit anti-hAPP antibody (CST, #29765) overnight at 4°C. After washing, sections were incubated with biotinylated goat anti-rabbit IgG (Vector Laboratories, #BA-1000-1.5) for 1 hour, followed by ABC reagent (Vector, Germany). Signal was developed with DAB substrate; hematoxylin was used for counterstaining. Sections were dehydrated, cleared, and mounted. Images were acquired on a TissueFAXS microscope (TissueGnostics, Austria, 20 × objective). RNA Sequencing and Proteomics Corneas from 7-month-old mice (n = 5/group for RNA-seq; and n = 3–5/group for proteomics, both eyes pooled per animal) were processed. RNA-seq: total RNA extracted with TRIzol (Invitrogen), libraries constructed using Optimal Dual-mode mRNA Library Prep Kit (BGI), sequenced on DNBSEQ-G400 (PE150). Reads aligned to GRCm39 using HISAT2; differential expression used DESeq2 (|log₂FC| ≥ 1, adjusted P ≤ 0.05). Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment used phyper in R (false discovery rate [FDR] < 0.05). Proteomics: samples lysed in SDT buffer, digested with trypsin, and analyzed on timsTOF HT (Bruker) in DIA mode. Data processed with DIA-NN; differentially expressed proteins defined as fold change >1.5 and P < 0.05. Statistical Analysis All analyses used GraphPad Prism (version 10.1.2). Data are mean ± SEM. Outliers beyond mean ± 2 SD were excluded. Normality was assessed by Shapiro–Wilk test; homogeneity of variances by F test. For normal data, unpaired two-tailed Student’s t-test was applied. Non-normal data were analyzed by Mann–Whitney U test. Significance: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Results Human APP Is Undetectable in the Cornea of 5 × FAD Mice We first verified expected disease progression in WT and 5 × FAD mice. Immunofluorescence demonstrated an age-dependent increase in Aβ plaque burden in the hippocampus and cortex, with minimal plaques at 3 months, clear increases at 5 and 7 months, and further accumulation by 10 months (Supplementary Figs. S1A–C). In ocular tissues, Aβ plaques were detected in the retina at 7 and 10 months but were consistently absent in the cornea at all ages examined (Supplementary Fig. S1D). We next assessed whether local APP-related changes occurred in the cornea. Immunohistochemistry showed robust human APP (hAPP) staining in 5 × FAD brains (Fig. 1A). In the same eye sections, hAPP signal was detected in retinal ganglion cells but was not detectable in the cornea (Fig. 1B). Immunoblotting confirmed strong hAPP overexpression in the brain and retina of 5 × FAD mice, with no detectable hAPP in corneal lysates (Figs. 1C–H). β-Site APP cleaving enzyme 1 (BACE1) levels were increased in 5 × FAD brain but were unchanged in retina and cornea (see Figs. 1C–H). Collectively, these findings indicate that hAPP overexpression and BACE1 upregulation are not detectable in the cornea of 5 × FAD mice. Expression of hAPP in the brain, retina, and cornea of 7-month-old WT and 5 x FAD mice. (A) Immunohistochemical staining of hAPP in CA1 and cortical regions of brain. (B) Immunohistochemical staining of hAPP in the retina and cornea from the same eye section. (C, D) Western blot images and quantification of hAPP and BACE1 in brain tissue. (E, F) Western blot images and quantification of hAPP and BACE1 in retina tissue. (G, H) Western blot images and quantification of hAPP and BACE1 in cornea tissue (n = 3 mice per group). Data are mean ± SEM. **P < 0.01 by t-test. Dynamic, Age-Dependent Changes in Corneal Tau Phosphorylation in 5 × FAD Mice We next characterized tau localization and phosphorylation in the cornea. Co-staining with the neuronal marker TUJ indicated that tau signal localized predominantly to corneal nerve fibers (Fig. 2A). Immunoblotting of total tau and phosphorylated tau (pTau) epitopes across 3, 5, 7, and 10 months revealed a nonlinear pattern in 5 × FAD corneas (Figs. 2B–M). At 3 months, total tau and pTau levels were comparable between genotypes (see Figs. 2B, 2C). At 5 months, pTau (Thr205), pTau (Thr217), and pTau (Thr231) were significantly reduced in 5 × FAD corneas (see Figs. 2E, 2F). By 7 months, pTau (Thr181) became significantly increased in 5 × FAD mice (see Figs. 2H, 2I). At 10 months, AT8, pTau (Thr217), and pTau (Thr231) were significantly reduced in 5 × FAD corneal lysates, accompanied by a marked reduction in TUJ1 (see Figs. 2K, 2L), consistent with loss of corneal nerve fibers. Dynamic changes in tau-related proteins in corneas of WT and 5 × FAD mice. (A) Immunofluorescent localization of tau in corneal cross-sections, and co-localization of tau with nerve fiber marker TUJ1. Scale bar = 50 µm. (B–M) Representative immunoblots and quantification of total tau, pTau epitopes, TUJ1, and tau-related kinases in corneas at 3 months B and C 5 months E and F, 7 months H and I, and 10 months K and L. Quantification of tau, pTau, and kinase levels normalized to TUJ1 at 3 months D, 5 months G, 7 months J, and 10 months M (n = 5–6 per group). Data are mean ± SEM. *P < 0.05, **P < 0.01 by t-test and Mann-Whitney U test. We then examined tau-related kinases, including glycogen synthase kinase 3β (GSK3β) and cyclin-dependent kinase 5 (CDK5). At 3 and 5 months, total kinase levels and phosphorylation states did not differ between genotypes (see Figs. 2C, 2F). At 7 months, inhibitory phosphorylation of GSK3β at Ser9 was significantly reduced in 5 × FAD corneas, whereas phosphorylation at Tyr216 did not differ (see Fig. 2I). CDK5 expression remained unchanged across time points (see Figs. 2C, 2F, 2I, 2L). Given that tau is primarily localized to corneal nerve fibers, we normalized total tau, pTau, and kinase levels to TUJ1 to estimate phosphorylation burden relative to remaining nerve fibers. After TUJ1 normalization, the results at 3 and 7 months remained consistent with our initial observations (see Figs. 2D, 2J). However, at 10 months, this analysis revealed that the residual fibers in 5 × FAD corneas showed significantly increased total tau and multiple pTau epitopes (AT8, Thr181, Thr205, and Thr231) compared with WT, alongside increased GSK3β abundance relative to TUJ1 (see Fig. 2M). These findings suggest that surviving axons retain heightened kinase levels and/or activation-associated phosphorylation signatures. Collectively, this delineates a nonlinear trajectory for corneal tau pathology: early transient reduction at 5 months, pathological elevation at 7 months coincident with GSK3β activation, and by 10 months, severe neurodegeneration with persistently high phosphorylation burden in residual axons. In brain tissue, tau-related signals exhibited age-related deterioration, with no notable elevation at 3 months of age; however, a progressive increase in tau phosphorylation became evident starting from 5 months (Supplementary Fig. S2). In trigeminal ganglia collected at 7 and 10 months, total tau, pTau epitopes, and TUJ1 levels were comparable between WT and 5 × FAD mice (Supplementary Figs. S3A–S3D), making retrograde propagation less likely as the primary driver of corneal tau pathology, although it cannot be fully excluded. Peripheral Corneal Nerve Fibers are Selectively Reduced at 10 Months in 5 × FAD Mice We quantified corneal nerve architecture using whole-mount TUJ1 staining with three-dimensional reconstruction. At 3, 5, and 7 months, there were no genotype-dependent differences in area or volume of the SBNP or SNT in either central or peripheral cornea (Supplementary Fig. S4). By 10 months, 5 × FAD mice exhibited significant reductions in both SBNP and SNT area and volume in the peripheral cornea (Figs. 3A–C). In contrast, central corneal nerve parameters remained unchanged at this age (Figs. 3D–F). Thus, measurable nerve loss emerged at 10 months and was restricted to the peripheral cornea. Three-dimensional analysis of corneal nerve fibers at 10 months. (A) Representative images of TUJ1 staining in peripheral cornea with 3D reconstruction. (B, C) Quantification of SBNP and SNT area and volume in peripheral cornea. (D) Representative images in central cornea. (E, F) Quantification in central cornea. Scale bar = 50 µm (n = 3–7 mice per group). Data are mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001 by t-test and Mann-Whitney U test. Peripheral Corneal Dendritic Cells Show Early Morphological Simplification and Later Expansion in 5 × FAD Mice We next assessed immune cell changes in the cornea over time. In the peripheral cornea, total CD45+ cell density was unchanged at 3 and 5 months but was significantly increased in 5 × FAD mice at 7 months (Figs. 4A, 4C). No genotype-dependent differences were observed in the central cornea at any time point (Figs. 4B, 4D). Co-staining identified two epithelial immune populations: dendritic-shaped CD45+CD11c+MHC class II+ DCs, and smaller, round CD45+CD3+ T cells (Supplementary Figs. S5A–S5C). Density and morphology of corneal dendritic cells in WT and 5 × FAD. (A, B) Representative images of CD45+ cells in peripheral and central cornea at 3, 5, and 7 months. (C, D) Quantification of CD45+ in peripheral and central cornea at 3, 5, and 7 months. (E, F) Quantification of DCs in peripheral and central cornea at 3, 5, and 7 months. (G, I, K) Representative images of DCs in peripheral cornea. (H, J, L) Morphological parameters: cell area, perimeter, branch number, total dendrite length. Scale bar = 50 µm (n = 5–6 per group). Data are mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 by t-test and Mann-Whitney U test. Peripheral DC density was comparable at 3 months but increased significantly at 5 and 7 months in 5 × FAD mice (Fig. 4E), whereas central DC counts remained low and did not differ between genotypes (Fig. 4F). We then quantified DC morphology in the peripheral epithelium (Figs. 4G–L). Whereas the DC area was unchanged, DC perimeter was significantly reduced at 3, 5, and 7 months. Branch number showed a downward trend, and total dendrite length was significantly shorter at 7 months. Notably, the reduction in DC perimeter was already evident at 3 months, preceding the increase in DC density (5 months) and the increase in pTau (Thr181; 7 months). Peripheral Corneal T Cell Accumulation and Multi-Omics Evidence of Inflammatory Remodeling at 7 Months We also quantified corneal T cells (Figs. 5A–C). CD45+CD3+ T cell numbers were comparable between genotypes at 3 and 5 months in both peripheral and central regions (Figs. 5D, 5E). At 7 months, T cell numbers increased significantly in the peripheral cornea of 5 × FAD mice, with no corresponding change in the central cornea (Fig. 5F). This increase followed the rise in DC density (significant by 5 months), supporting the notion that early DC changes may contribute to subsequent T cell recruitment. T cell quantification and transcriptomic analysis at 7 months. (A–F) Representative images and quantification of CD45+CD3+ T cells in peripheral and central cornea at 3, 5, and 7 months. (G) Number of differentially expressed genes (DEGs). (H) Volcano plot (red: upregulated; blue: downregulated; threshold: |log₂FC| ≥ 1, adjusted P ≤ 0.05). (I) Heatmap of DEGs. (J) KEGG pathway enrichment of DEGs. (K) Volcano plot of cytokine-related genes. (L) KEGG enrichment of cytokine-related DEGs. Scale bar = 50 µm (n = 5–6 per group). Data are mean ± SEM. ****P < 0.0001 by t-test and Mann-Whitney U test. To define associated molecular changes, we performed RNA sequencing (RNA-seq) on corneas from 7-month-old WT and 5 × FAD mice. We identified 269 upregulated and 149 downregulated differentially expressed genes (DEGs) in 5 × FAD corneas (|log₂FC| ≥ 1, adjusted P ≤ 0.05; Figs. 5G–I). KEGG enrichment analysis highlighted immune-related pathways, including cytokine signaling, chemokine pathways, antigen processing and presentation, as well as tumor necrosis factor (TNF) signaling (Fig. 5J). Differentially expressed cytokine/chemokine genes included Jak1, Tgfbr1, and Ptprf (Fig. 5K), and cytokine-related DEGs showed enrichment for T cell-associated pathways (Fig. 5L). Complementary mass spectrometry-based proteomics at 7 months identified 224 differentially expressed proteins (DEPs; fold change >1.5, P < 0.05), including 154 upregulated and 70 downregulated proteins; 85 proteins were detected only in 5 × FAD corneas (Figs. 6A–C). KEGG analysis of DEPs again implicated immune and inflammatory processes, including leukocyte transendothelial migration and interleukin-17 (IL-17) signaling, as well as extracellular matrix-receptor interaction (Fig. 6D). Myeloid-associated proteins (S100A8, S100A9, myeloperoxidase [MPO], and lysozyme 2 [LYZ2]) were significantly increased (Fig. 6E). Adhesion/migration-related proteins, including intercellular adhesion molecule 1 (ICAM1) and integrin beta 2 (ITGB2) were also elevated (see Fig. 6E). In addition, interleukin-16 (IL-16) and the MHC class II protein H2-EB1 were detected exclusively in 5 × FAD corneas. Together, transcriptomic and proteomic profiling at 7 months demonstrates an immune-enriched, pro-inflammatory signature in 5 × FAD corneas coincident with peripheral T cell accumulation. Proteomic analysis of corneas from 7-month-old WT and 5 × FAD mice. (A) Numbers of differentially expressed proteins (DEPs). (B) Volcano plot (red: upregulated; blue: downregulated; threshold: fold change >1.5, P < 0.05). (C) Heatmap of DEPs. (D) KEGG pathway enrichment of DEPs. (E) Quantification of selected immune-related proteins (n = 3–5 mice per group). Discussion In this work, we mapped AD-associated corneal pathology across four ages in WT and 5 × FAD mice and found a temporally ordered, region-selective pattern of change. The phenotype was largely confined to the peripheral cornea, where DC morphology shifted early, followed by immune-cell expansion, tau phosphorylation changes, and—only later—measurable nerve loss. Specifically, peripheral DCs showed morphological simplification (reduced perimeter and dendritic complexity) by 3 months, before DC density increased (5 months), CD3⁺ T cells accumulated (7 months), and pTau (Thr181) rose (7 months); overt loss of peripheral SBNP and SNT was detected at 10 months. A key finding was that hAPP was undetectable in corneal tissue in 5 × FAD mice. This result is consistent with the transgenic design (neuronal promoter activity) and corneal neuroanatomy: the cornea is innervated predominantly by trigeminal sensory axons, but it does not contain neuronal somata.37–40 In this setting, the absence of local hAPP expression argues against a corneal-intrinsic Aβ production mechanism in this model. Therefore, the corneal neuroimmune alterations observed in 5 × FAD mice may primarily reflect the indirect effects of brain-derived, Aβ pathology-driven signaling on the ocular surface, rather than the direct consequences of local Aβ generation. This interpretation offers a new perspective on the potential origins of ocular surface changes in patients with AD. In humans, corneal nerves also originate from the trigeminal ganglion, and APP expression in human corneal nerves remains uncharacterized. Nevertheless, corneal nerve fiber loss and immune cell changes reported in patients with AD27,28 suggest that peripheral ocular surface alterations can emerge downstream of, or in parallel with, central AD-like pathology, without requiring local Aβ production. Corneal tau pathology evolved in a stage-dependent, nonlinear manner—a pattern that diverges from the more monotonic progression typically observed in brain tissue.41 At 5 months, corneal lysates showed reduced levels of multiple pTau epitopes (Thr205, Thr217, and Thr231). By 7 months, a distinct shift occurred: pTau (Thr181) increased, coinciding with reduced inhibitory phosphorylation of GSK3β at Ser9, a marker of its disinhibition. This temporal correlation suggests that the Thr181 site may be preferentially responsive to GSK3β activity. At 10 months, however, several pTau signals decreased alongside a marked reduction in the neuronal marker TUJ1, consistent with advanced nerve loss. Because a shrinking axonal compartment can dilute tau-related readouts in bulk tissue, we interpret this late-stage decline primarily as a consequence of neurodegeneration. Supporting this view, normalizing tau and kinase signals to TUJ1 revealed that residual nerve fibers carried a higher phosphorylation burden at 10 months. Although these correlative data link shifts in kinase state to tau phosphorylation, the precise mechanisms remain to be established. GSK3β is known to phosphorylate tau at multiple sites, including Thr181, Ser202, Thr205, Thr217, and Thr231.42,43 CDK5 can also target residues such as Ser404, Thr212, Ser214, Thr231, and Thr205, although tau phosphorylated by CDK5 aggregates more slowly. Beyond kinase activation, abnormal phosphorylation at specific sites (e.g. Ser262 and the Thr231/Ser235 cluster) can reduce tau’s microtubule-binding affinity, leading to its dissociation and loss of function.44 Future studies will need to directly assess kinase/phosphatase balance and cell-type specificity to establish causality. Another striking feature was the consistent restriction of both immune alterations and nerve degeneration to the peripheral cornea. This regional specificity likely reflects multiple factors. First, the limbal region contains a rich vascular network supporting immune cell recruitment.45 Second, under homeostasis, peripheral cornea harbors higher densities of both immune cells and nerve fibers,46 with more pronounced neuro-immune crosstalk.24,34 Third, proximity to conjunctiva may expose the peripheral cornea to circulating factors amplifying local responses. These findings have important clinical implications: IVCM monitoring should prioritize peripheral corneal regions to maximize sensitivity. Neuroinflammation is increasingly recognized as a central AD driver.10 Here, we observed increased DC numbers and pronounced morphological changes in 5 × FAD corneas. Critically, DC morphological alterations were detectable as early as 3 months, preceding both numerical DC increases and tau pathology. Similar early morphological changes have been reported in tau transgenic (rTg4510) mice,26 although density changes differed between models, possibly reflecting distinct pathogenic drivers (Aβ-driven versus tau-driven). Morphological simplification of DCs is thought to reflect an activated or stressed state, as observed in inflammatory conditions.27 Human studies in patients with mild cognitive impairment have documented enlarged DCs and trends toward reduced nerve fiber density,27 although small sample sizes necessitate validation. The consistency of early DC changes across models highlights their potential as early AD indicators. We also report increased T cell numbers in the corneal epithelium of an AD mouse model. This increase occurred selectively in peripheral cornea at 7 months, temporally following the rise in DC density. Corneal T cells have been implicated in nerve damage in diabetic and dry eye disease, where T cell infiltration correlates with reduced nerve density.47–50 Although significant nerve loss was not observed until 10 months, elevated T cell levels at 7 months coincided with progressive tau phosphorylation, suggesting a mechanistic link. Emerging evidence implicates T cell infiltration in AD brains,51–54 where T cells associate with pTau burden and contribute to neurodegeneration.55,56 T cell depletion in tauopathy models reduces pTau levels and improves cognitive function.53,56 Mechanistically, CD8⁺ T cells expressing granzyme K (Gzmk) identified in AD brains57,58 can activate protease-activated receptor-1 (PAR1), promoting inflammatory cytokine secretion and enhancing senescence-associated secretory phenotype gene expression.59 Gzmk-mediated PAR1 activation may also facilitate T cell migration and disrupt neuronal Ca²⁺ homeostasis, contributing to tau phosphorylation.59,60 Additionally, T cells are a major source of IFN-γ, which has been shown to directly induce tau phosphorylation, whereas administration of anti–IFN-γ antibodies attenuates neurodegeneration and reduces tau phosphorylation.56 Our multi-omics data provide molecular correlates for these processes in cornea. Upregulation of adhesion molecules (ICAM1 and ITGB2), T cell chemoattractant IL-16, and MHC class II molecules indicates an environment conducive to T cell recruitment and antigen presentation. Exclusive presence of MHC II in 5 × FAD corneas suggests enhanced antigen-presenting capacity from activated DCs. Together, these data are consistent with a model in which early DC alterations may facilitate subsequent T cell accumulation, which may in turn associate with tau phosphorylation changes and later nerve loss. Causality will require functional perturbation experiments (e.g. immune-cell depletion or cytokine blockade). Several limitations should be acknowledged. First, we focused on a single AD model (5 × FAD), which is primarily Aβ-driven; future work in tauopathy models (e.g. rTg4510) will clarify generalizability. Second, we did not perform functional experiments (e.g. immune-cell depletion, adoptive transfer, or cytokine blockade) to test whether DCs or T cells are necessary for tau changes or nerve degeneration. Third, sex-stratified analyses to explore potential sex-specific effects were precluded by limited sample sizes. Finally, translation to humans will require correlative studies linking corneal immune/nerve metrics with clinical staging and cognitive status, ideally using IVCM-based protocols optimized for peripheral sampling. In summary, the 5 × FAD cornea exhibits a defined sequence of AD-related changes: beginning with DC morphological alterations at 3 months, followed by T cell infiltration and tau hyperphosphorylation at 7 months, and culminating in peripheral nerve loss by 10 months. The early emergence of these morphological changes suggests DC activation as a candidate sentinel event in disease progression. Multi-omics data suggest potential antigen presentation and T cell recruitment mechanisms that parallel the T cell-driven pathology observed in the AD brain, although functional studies are needed to confirm this link. That the cornea, an accessible tissue amenable to noninvasive imaging, could serve as a peripheral window for early AD detection is an idea worth pursuing. Supplementary Material Supplement 1