A Murine Model of HSV-1-Induced Corneal Endotheliitis: Primary Infection, Latency, and Recurrence

Abstract

Purpose

HSV-induced corneal endotheliitis (HSV endotheliitis) is a sight-threatening ocular disease, with poor underlying its pathogenesis due to the lack of suitable animal models, we therefore aimed to develop a murine HSV endotheliitis model.

Methods

Herpes simplex virus type 1 (HSV-1) was injected into the anterior chamber (AC) of C57BL/6 mice to establish primary and latent infection. Recurrence was induced at least 5 weeks post-infection via ultraviolet B (UVB) corneal exposure. Clinical manifestations were dynamically monitored. The histopathology and molecular changes were assessed using transmission electron microscopy, immunofluorescence (IF) staining, hematoxylin and eosin (H&E) staining and multiple cytokines and chemokines analysis. HSV-1 load in corneas and trigeminal ganglia (TG) was detected via plaque assay, quantitative real-time PCR and IF staining.

Results

Compared with controls, the primary HSV-1-infected mice exhibited characteristic manifestations of viral endotheliitis, including corneal and iris edema, AC inflammation, keratic precipitates (KPs), elevated intraocular pressure, and corneal endothelial cell damage and loss. Early after infection, viral loads were elevated both in corneas and TGs. By day 28, however, a latency-associated transcript (LAT) in TGs increased markedly, while HSV-1 titers became undetectable, indicating viral latency. Following UVB corneal exposure, the recurrent mice showed significant stromal edema, increased KPs, extensive endothelial cell loss, and elevated viral loads, confirming the successful induction of recurrence.

Conclusions

We successfully established a murine model of primary infection, latency and recurrence of HSV endotheliitis, providing a reliable platform for investigating pathological mechanisms and potential treatments.

Herpes simplex virus-induced corneal endotheliitis (HSV endotheliitis), a severe subtype of herpetic keratitis, is clinically characterized by corneal edema, keratic precipitates (KPs) and progressive loss of corneal endothelial (CoEndo) cells, often leading to severe endothelial dysfunction.¹ Globally, an estimated 1.5 million individuals suffer from ocular HSV infections annually.² Approximately 25% of these patients progress to recurrent stromal keratitis or endotheliitis, contributing to 6% of the corneal blindness cases worldwide.^3–5 Among herpesviridae infections, HSV endotheliitis accounts for 60% to 80% of viral endotheliitis cases in immunocompetent populations and is the most vision-threatening subtype, with 32% of affected eyes progressing to legal blindness within 5 years of diagnosis.⁶ Consequently, HSV endotheliitis poses a critical global health burden due to its high prevalence, therapeutic challenges, and severe vision-threatening consequences.^7,8

Unlike epithelial and stromal keratitis, corneal endotheliitis directly targets CoEndo cells. These cells are pivotal for maintaining corneal transparency via their “pump-leak” mechanism, which regulates stromal dehydration.^9,10 Once damaged, these postmitotic cells have limited regenerative capacity; their loss leads to irreversible corneal edema, opacification, and permanent visual impairment.^11–13 Critically, 30% to 40% of patients eventually require penetrating keratoplasty, yet the 5-year graft survival rate remains only 58% to 62% due to viral reactivation and immune rejection.^14,15

Current antiviral therapies, such as acyclovir, suppress acute infection in 89% to 95% of cases but fail to eliminate viral latency or prevent CoEndo loss.^16,17 Chronic or recurrent HSV endotheliitis is associated with progressive CoEndo loss; longitudinal in vivo confocal studies estimate an annual endothelial cell density decline of approximately 10%, far exceeding the physiologic age-related loss (approximately 0.6% per year).^1,9 These therapeutic limitations underscore critical gaps in understanding HSV-endothelial interactions, including viral entry mechanisms (nectin-1 versus herpesvirus entry mediator [HVEM]) and immune privilege modulation.^18,19 Thus, comprehensive elucidation of HSV endotheliitis pathogenesis is critical for advancing therapeutic strategies, especially the mechanism of latency and recurrence.

Although observational reports and case series are abundant, the fundamental research into HSV endotheliitis remains limited, including its pathological dynamics, mechanisms, and optimal treatments.²⁰ Notably, recurrent HSV endotheliitis is clinically more prevalent and destructive than primary infection,^21,22 but its recurrence progress and underlying mechanism also remain poorly characterized. Therefore, establishing a reliable model recapitulating HSV endotheliitis is of great necessity to advance investigation into disease pathogenesis and therapeutic management.

In this study, we established an optimized protocol for murine HSV endotheliitis, which recapitulates key clinical and pathological manifestations observed in human disease. The typical endotheliitis manifestations after infection included corneal and iris edema, KPs, elevated intraocular pressure (IOP), and CoEndo loss. We further evaluated the latency and recurrence following intracameral inoculation. At early time points post-infection, viral loads increase in both corneas and trigeminal ganglia (TGs). By day 28 post-infection, latency-associated transcript (LAT) expression in TGs increased significantly whereas HSV-1 titers became undetectable, indicative of latent infection. Following UVB corneal exposure, recurrent disease manifested by stromal edema, increased KPs, extensive CoEndo loss, and elevated viral loads in both corneas and TGs. Together, these findings provide a tractable model for dissecting disease pathogenesis and evaluating therapeutic strategies.

Materials and Methods

Animals

Healthy male C57BL/6 mice, aged 6 to 8 weeks and weighing 18 to 22g with normal eye development, were procured from Beijing Vital River Laboratory Animal Technology Co., Ltd. (Beijing, People's Republic of China). All procedures were conducted in accordance with the ARVO Statement and were approved by the Committee for the Use of Animals in Ophthalmic and Vision Research at the Shandong Eye Institute (SDSYKYJS No. 20231217).

Virus Strain and Titer Determination

The HSV-eGFP (H129-G4 HSV-1) strain with cytomegalovirus (CMV) promoter-driven EGFP expression was procured from Shanghai Genechem Co., Ltd. whereas the HSV-1 (McKrae strain) was obtained from Eye Institute of Shandong First Medical University. Viral titers were quantified using a standard plaque assay on confluent Vero cell monolayers in 24-well plates. Briefly, viral samples were obtained by swabbing the murine eyes with sterile cotton swabs at indicated time points. The prepared Vero monolayer cells in 24-well plates were then infected with the collected viral solutions through serial dilutions. Following an incubation period of 3 or 4 days, the cells were fixed and stained with crystal violet (Beyotime). Finally, plaques were counted and graphed, and the virus titers were calculated.

Infection of Mouse Corneal Endothelial Cells With HSV-eGFP

Primary mouse corneal endothelial cells (MCECs) were isolated and cultured as previously described.²³ Briefly, Descemet’s membranes with adherent MCECs were stripped from murine corneas, digested with 1 mg/mL collagenase A (37°C for 15 minutes), and suspended in DMEM containing 10% FBS, penicillin (50 U/mL), and streptomycin (50 µg/mL). Confluent cells were washed with Ca²⁺/Mg²⁺-free PBS, trypsinized with 0.25% trypsin-EDTA (5 minutes at 37°C), and second-passage cells were used for experiments. Confluent MCECs seeded onto coverslips in 48-well plates were infected with HSV-eGFP (10^4 PFU/5 µL, multiplicity of infection [MOI] = 1) or mock-treated with DMEM. Infected cells were monitored by confocal microscopy (Leica TCS SP8) at 0, 24, 48, and 72 hpi to visualize viral replication dynamics.

Establishment of HSV-Induced Corneal Endotheliitis Model

Specific pathogen-free C57BL/6 mice (6–8 weeks) were randomized into three groups (n = 10/group): (1) HSV (intracameral 5 µL HSV-1 suspension, 2 × 10^1 PFU/mL); (2) UV-HSV (UV-inactivated virus); and (3) UV-MOCK (BBS Plus). Mice received 0.6% sodium pentobarbital intraperitoneal (IP) anesthesia prior to aseptic anterior chamber inoculation using a 36-gauge syringe. The novel intracameral injection method uses the posterior ciliary sulcus of the iris as the entry point; the needle first accesses the posterior chamber and then traverses the pupil to enter the anterior chamber, thereby minimizing the risk of viral leakage. This approach also eliminates the need for air or viscoelastic injection and avoids corneal puncture.²⁴ Postoperative monitoring (days 1–5 post-infection, dpi) included daily clinical scoring (corneal opacity/iris dilation), rebound tonometry for IOP, and multimodal imaging (spatial light modulator/anterior segment optical coherence tomography [SLM/AS-OCT]).

A separate cohort (n = 54) underwent terminal sampling on days 0, 1, 2, 3, 4, 5, 7, 14, and 28 post-infection (dpi; n = 6 per group per time point). Enucleated eyes were processed for histopathology: fixation in 4% paraformaldehyde, paraffin embedding, and hematoxylin and eosin (H&E) staining. Ultrastructural analysis: glutaraldehyde/osmium tetroxide fixation for electron microscopy. Immunofluorescence: antigen detection in corneal/iris tissues via frozen sections. Trigeminal ganglia from HSV/UV-HSV groups were analyzed by immunofluorescence (HSV antigen) and RT-PCR (viral gene expression), with triplicate testing for reliability.

Slit-Lamp Examination and Ocular Imaging

SLM was performed to assess corneal edema, KPs, anterior chamber inflammation, and iris/pupil changes. AS-OCT (Optovue) measured central corneal thickness (CCT) and visualized anterior chamber inflammation/pupil dynamics. In vivo confocal microscopy (IVCM; Heidelberg Engineering) performed serial endothelial scans to detect lesion progression.

Intraocular Pressure Measurement Protocol

IOP was measured using a rebound tonometer (Icare Finland) at prespecified time points. Mice were restrained with head stabilization, and central corneal measurements were obtained with the eyelids gently retracted if necessary. For each eye, six consecutive readings were acquired in triplicate, and the mean IOP was used for analysis.

Clinical Scoring of Corneal Endotheliitis

The severity of corneal endotheliitis was evaluated based on corneal opacity as previously described.^25,26 The criteria of scoring system were as follows: 0 = transparent; 1 = mildly cloudy (a slightly cloudy area identifiable; iris and pupil clearly visible); 2 = moderate opacity (cloudy area easily identified; partial visibility of iris and pupil details); 3 = severe opacity (details of the iris and pupil obscured by opacity, but iris and pupil distinguishable); 4 = most severe opacity (iris and pupil completely obscured by opacity); and 5 = corneal rupture. All measurements and scoring were performed in a blinded manner by two trained ophthalmic examiners.

Quantitative Real - Time Polymerase Chain Reaction

Corneal tissues and aqueous humor were collected at appropriate sampling time points post-infection. Total RNA was extracted using the TransGen Biotech RNA isolation kit, followed by reverse transcription into cDNA with the Vazyme cDNA synthesis kit. The qRT-PCR was performed on an ABI Prism 7500 system using ChamQ Universal SYBR qPCR Mix (Vazyme). Relative gene expression was normalized to GAPDH and calculated via the 2^−ΔΔCt method. Primer sequences are listed in Supplementary Table S1.

Immunofluorescence Staining

Eyeball and trigeminal ganglion tissues were cryosectioned at 7 µm (CM 1950 cryostat; Leica) after embedding in optimum cutting temperature compound (Sakura). Sections were fixed in 4% PFA (Sigma) and permeabilized with 0.3% Triton X-100 (Solarbio). Sections were incubated with anti–HSV-1 (1:1000; ab9533; Abcam) for TGs and anti-CD45 (1:200; ab23910; Abcam) for ocular tissues. After overnight incubation at 4°C, the sections were stained with Alexa Fluor 488-conjugated goat anti-rabbit IgG (1:500; ab150077; Abcam) for 2 hours in the dark. Nuclei were counterstained with DAPI, and images were acquired with an Echo Revolve fluorescence microscope.

TUNEL Assay

TUNEL assay was performed using a One-step TUNEL Apoptosis Assay Kit (KTA2011, Abbkine) according to the manufacturer’s instructions. Samples were counterstained with DAPI, and fluorescence images were captured with a confocal microscope.

Electron Microscopy Sample Preparation and Observation

Corneal samples were fixed in 2.5% glutaraldehyde (phosphate buffer, pH 7.4) at 4°C overnight, followed by 1% OsO₄ post-fixation at 4°C for 2 hours. Specimens underwent gradient acetone dehydration (30%–100%), infiltration with 1:1 acetone/epoxy resin (Epon 812), and embedding in pure resin. After polymerization (37°C, 24 hours; 60°C, 48 hours), 50-nm ultrathin sections were double-stained with uranyl acetate and lead citrate. Observations were performed on a JEOL JEM-1400Flash TEM (Japan) at 80 kV, and images were acquired at representative magnifications.

Quantitative Analysis of Cytokines and Chemokines in the Aqueous Humor

The levels of IL-10, IL-2, IL-4, IL-5, IL-1α, IL-1β, CXCL1, TNF-α, IL-12P70, IL-17A, GM-CSF, IFN-γ, IL-6, CCL2, and CCL5 in aqueous humor from different groups were detected by multiplex secretome analysis (XMPLEX Mouse Macrophage 15-Plex Panel) according to the manufacturer’s instructions with the assistance of SXM Biotechnology Co., Ltd. (Wuhan, People's Republic of China).

Induction of Recurrent HSV Endotheliitis Following Latent Infection

The recurrent HSV endotheliitis (UV-rHSK) model is primarily induced by reactivating latent HSV-1 using UVB irradiation, one of the most effective inducers of reactivation. Briefly, the latently infected mice were first established by intracameral inoculation of the HSV-1 McKrae strain and co-treatment with acyclovir and a glucocorticoid. The latent mice were then exposed to UVB light at a dose of 250 mJ/cm² (peak wavelength 302 nm) using a transilluminator.^27,28 The UV-MOCK group consisted of non-infected mice receiving the same UVB irradiation condition. The L-HSK group included latently infected mice that did not undergo UVB exposure.

Statistical Analysis

Data are presented as mean ± SD. Experiments were repeated greater than or equal to three times for reproducibility. Normally distributed continuous variables were analyzed using unpaired Student's t-test (2 groups) or 1-way ANOVA with post hoc tests (multiple groups). Clinical scores underwent 2-way ANOVA. Statistical significance was defined as P < 0.05.

Results

Pathological Characterizations of CoEndo Cells Following HSV-1 Infection In Vitro

We first examined the effect of HSV-1 on human CoEndo cells. As shown in Figure 1A, human corneal endothelial cells (HCECs) infected with HSV-1 (MOI = 1) underwent progressive morphological alterations: cytoplasmic shrinkage and granular darkening without detachment at 8 hpi), followed at 16 hpi by cytopathic effect (CPE) with syncytia and endosome-like structures. By 32 hpi, >80% of cells had undergone necrotic death, as visualized by phase-contrast microscopy. Transmission electron microscopy (TEM) revealed ultrastructural pathology in infected human CoEndo cells, including viral particles undergoing nuclear egress (Fig. 1B–a, b) and characteristic viral inclusion bodies (Fig. 1B–c, d).

Figure 1.

Infection of endothelial cells with HSV in vitro. (A) The dynamic morphological changes of human CoEndo cells after infection with HSV-1 observed under a microscope. (B) TEM view of human CoEndo cells after viral infection. Viral particles and inclusion bodies were observed. (C) Dynamic observation of CPE in MCECs after inoculation with HSV-eGFP (MOI = 2 × 10^4) at 8, 16, 24, and 32 hpi. Green-fluorescence positive MCECs probably represented the viral replication. (D) Viral titer analysis of HSV-1 in cell supernatant at different times (n = 6 per group). *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant.

Longitudinally tracking using HSV-eGFP showed eGFP-positive MCECs foci at 8 hpi (Fig. 1C). Notably, between 16 and 32 hpi, eGFP-positive cell counts plateaued and declined despite rising viral titers (Fig. 1D), indicating lytic death-medicated fluorescent signal loss. Collectively, these data demonstrate conserved HSV-1 tropism for CoEndo cells across species, featuring rapid CPE progression, syncytia formation, and necrotic cell death.

Clinical Outcomes Following Intracameral HSV-1 Inoculation in Mice

We subsequently evaluated murine corneal endotheliitis induced by intracameral HSV-1 injection to define an optimal inoculum. As shown in Supplementary Figure S1, titers ≥2 × 10^2 PFU/mL caused rapid and severe endotheliitis, precluding detailed observation. At 2 × 10^1 PFU/mL, mice spontaneously developed progressive endotheliitis, which was suitable for modeling. This concentration was therefore selected to establish HSV endotheliitis.

On day 3 post-inoculation with 2 × 10^1 PFU/mL HSV-1 (HSV group), severe endotheliitis was confirmed by SLM and AS-OCT, manifested by corneal stromal edema and neovascularization (Fig. 2A), and inflammatory cell exudation, KP formation, and iris edema (Fig. 2B). HSV-infected mice exhibited significantly greater pathology than UV-inactivated HSV (UV-HSV) and mock controls, featuring the disrupted anterior segment architecture and increased CCT (see Figs. 2B, 2C). Clinical scores in the HSV group were significantly higher than those in the UV-HSV group (P < 0.001). The IOP peaked at 30.17 ± 1.94 millimeters of mercury (mm Hg) on day 5 post-infection in HSV-infected mice, whereas UV-HSV group and mock controls maintained normal IOP throughout the observation period (Fig. 2E).

Figure 2.

The overview characteristics of HSV-1 corneal endotheliitis model of mice. (A) SLM views and (B) AS-OCT images of the anterior segment of the eyes in different groups at 1, 3, and 5 dpi. (C) Bar graphs showing the corneal thickness in different groups at 1, 2, 3, 4, and 5 dpi (n = 6 mice per group). *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. (D) Ocular clinical scores on post-infection days 1, 3, and 5. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. (E) The dynamic observation of IOP in different groups at 1, 2, 3, 4, and 5 dpi. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. HSV group (active viral inoculation), UV-HSV group (UVB-inactivated viral exposure), and UV-MOCK group (BBS Plus). (F) HSV-1 antigen (green) in corneal tissues examined by IF staining, nuclei counterstained with DAPI (blue). Scale bar = 100 µm; n = 3 mice per group. (G) Viral titer analysis of HSV-1 in corneal samples (n = 6 mice per group). HSV, the corneas infected with HSV-1; ACV: the HSV-1 infected corneas co-treated with acyclovir. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. (H) Quantitative RT-PCR analysis of HSV-1-related genes’ (gB, gK, and ICP0) expression (n = 6 mice per group). HSV, the AC of mice infected with active HSV-1; UV-HSV, the AC of mice inoculated with UVB-inactivated HSV-1; UV-MOCK (BBS Plus), the AC of mice inoculated with PBS. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant.

We next quantified corneal viral burden. Through IF, more HSV-1 protein expression in corneal tissues from HSV group was observed when compared to UV-HSV group and mock controls (Fig. 2F). Elevated viral titers in HSV-inoculated corneas were determined, but abrogated by acyclovir treatment (Fig. 2G). The upregulated HSV-related genes (gB, gK, and ICP0) in the HSV group were detected by quantitative PCR, which were consistent with protein and viral culture data (Fig. 2H). Collectively, intracameral HSV-1 inoculation recapitulates human corneal endotheliitis hallmarks, including endothelial damage, inflammatory infiltration, and viral persistence.

Clinical Manifestations of Corneal Endotheliitis After Intracameral HSV-1 Injection

To characterize pathological alterations in HSV-1-infected corneas, we first assessed disease processes using SLM. As shown in Figure 3A, several key clinical manifestations of murine corneal endotheliitis were presented. Central corneal edema (Fig. 3A–a), anterior chamber (AC) angle obstruction by inflammatory cells (Fig. 3A–b) were observed within 1 to 2 dpi, as well as KPs on corneal endothelium (Fig. 3A–c). As disease progressed, limbal neovascularization was frequently observed on 2 to 5 dpi (Figs. 3A–d, e). Moreover, diffuse corneal edema and opacity were also observed in infected corneas on 4 to 5 dpi (Fig. 3A–f).

Figure 3.

The clinical manifestations of HSV corneal endotheliitis in mice by SLM and AS-OCT. (A) The SLM views of clinical manifestations. (a) Central corneal edema (white dotted line circle); (b) blockage of inflammatory cell in the angle of AC; (c) pigmented KPs; (d) and (e) limbal neovascularization of cornea; and (f) the diffused corneal edema and opacity. Original magnification = × 40. (B) AS-OCT views of clinical manifestations. (a) Large number of inflammatory cells in the AC; (b) pigmented KPs; (c) blockage of inflammatory cell in the angle of AC; and (d) the diffused corneal edema.

Through AS-OCT, we further revealed the typical phenotypes of corneal endotheliitis, characterized by more inflammatory cell infiltration into AC (Fig. 3B–a), pigmented KPs adhesion to corneal endothelium (Fig. 3B–b), and AC angle obstruction by inflammatory cells (Fig. 3B–c). Additionally, progressive corneal edema and thickening on 4 to 5 dpi were observed (Fig. 3B–d).

Histopathological Alterations Following Intracameral HSV-1 Inoculation

Complementing clinical observations, we characterized histopathologic changes in ocular tissues post-HSV-1 inoculation. H&E staining revealed significant alterations in HSV-1-infected corneas versus mock controls, including disrupted architectural integrity with dense inflammatory infiltration, enhanced corneal neovascularization, corneal endothelial-adherent inflammatory cells, and CoEndo cell swelling (Fig. 4A). The main pathology of the AC angle post-infection included profound inflammatory cell accumulation, architectural disruption of the AC angle, iris edema, and iridocorneal angle disorganization (see Fig. 4A), which probably contributed to the IOP elevation and fluctuation. Mild retinal pathology was also observed, manifested as swelling in the outer nuclear layer (ONL) and ganglion cell layer (GCL; see Fig. 4A).

Figure 4.

The histopathological alterations and strong inflammatory response in AC segment after intracameral inoculation of HSV-1. (A) Pathological changes evaluated by H&E. (a) Normal corneal tissue (mock); (b) normal AC angle (mock); (c) normal retina tissue (mock); (d) infected corneal tissue (black arrow: swelling of corneal endothelial cells); (e) infected AC angle, characterized by anterior synechia and numerous inflammatory cells and swelling of the iris; (f) infected retina tissue, manifested with swelling of the outer nuclear layer (ONL) and ganglion cell layer (GCL). (B) Left: IF staining of CD45⁺ immunocytes (green) in corneal tissues and AC angle, nuclei counterstained with DAPI (blue). Right: Histogram showing the mean IF density under different conditions. Scale bar = 100 µm (n = 6 mice per group). *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. (C) Quantitative RT-PCR analysis of inflammatory cytokines (Il-1b, Il-6, Tnf-a, Ifn-a, and Ifn-γ) in corneal tissues (n = 4 mice per group). *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. (D) Multiple cytokines and chemokines of AC were detected with the XMPLEX Mouse Macrophage 15-Plex Panel, mOCK group: BSS Plus, HSK group: HSV-infected endotheliitis (n = 3 per group). *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant.

IF staining demonstrated more infiltration of CD45⁺ immunocytes in corneal tissues and AC angles of infected mice when compared with uninfected controls (Fig. 4B). The qPCR analysis revealed a time-dependent upregulation of proinflammatory cytokines (Il-1β, Il-6, Tnf-α, Ifn-α, and Ifn-γ) in infected corneal tissues within 48 hpi (Fig. 4C). Compared to MOCK (BSS Plus) group, aqueous humor from HSV-infected (HSK) mice at 3 dpi showed a dramatic increase of pro-inflammatory cytokines and chemokines, including TNF-α, IFN-γ, CXCL1, IL-6, IL-1β, and CCL5 (Fig. 4D). These findings indicated an augmented inflammatory cascade in the aqueous humor following HSV-1 infection.

Temporal Pathological Dynamics of CoEndo Cells Following Intracameral HSV-1 Infection

To investigate in vivo temporal dynamics, recombinant HSV-1/eGFP virus was injected into AC. As presented in Figure 5A, viral foci (green) were localized to discrete zones with ZO-1⁺ endothelial cells (red) losing hexagonal morphology at 12 hpi, whereas distant endothelia remained normal. By 24 hpi, viral spread was accompanied by regional endothelial damage, with significant morphological disruption and monolayer detachment (see Fig. 5A). Widespread endothelial compromise was further evident in infected corneas at 48 hpi, characterized by severe cellular hypertrophy, nuclear pyknosis, and near-total loss of hexagonal architecture. Viral burden inversely correlated with declining ZO-1 intensity (see Fig. 5A). By 72 hpi, near-complete endothelial denudation occurred, with residual cells showing necrosis and diffuse viral fluorescence across corneal endothelium (see Fig. 5A). Quantitative analysis confirmed logarithmic viral increase (Fig. 5B) and precipitous CoEndo cell decline (Fig. 5C).

Figure 5.

The pathological characterization of CoEndo cells after intracameral inoculation of HSV-1. (A) Top, Dynamic observation of replicating HSV-1 (green) at 12, 24, 48, and 72 hpi, bottom, the morphological and density change of mouse CoEndo cells (red) at 12, 24, 48, and 72 hpi. Scale bar = 200 µm. (B) Histogram showing the mean IF density at different times post-infection (n = 6 mice per group). *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. (C) Histogram showing the density of CoEndo cells at different times post-infection (n = 6 mice per group). *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. (D) The corneal damage observed by IVCM. (a, b, c) Normal corneal epithelium, stroma, and endothelium; (d) diffused inflammatory cells among the stromal cells; (e, f, g) Enlarged intercellular gaps and dendritic cells among the stromal cells (red arrow); (h) the aggregation of inflammatory cells on the endothelial surface; (i) the edema of CoEndo cells with unclear boundaries (bar = 50 um). (E) The microstructural change of corneas after intracameral inoculation examined using TEM. (a, b) Normal endothelial cells full of normal mitochondria; (c, d) vacuole formation and detachment of endothelial cells; (e) severe swelling of mitochondrial cristae (red arrow); (f) mitochondrial autophagy (red arrow); (g) nuclear condensation and apoptosis (red arrow); (h) abnormally increased deformed mitochondria (red arrow).

We further evaluated the histopathologic alterations of corneal tissues using IVCM and TEM. As illustrated in Figure 5D, the normal corneas showed regular epithelium (a), aligned stromal fibers (b), and characteristic hexagonal endothelia (c). On the contrary, in infected corneas, the exhibited epithelial edema and increased inflammatory infiltrations (Fig. 5D–d), stromal swelling, fiber separation, and substantial increase of dendritic cells (Fig. 5D–e). Notably, the infected endothelium not only showed adhesion of inflammatory cells and KP formation, but also exhibited cellular edema and ill-defined boundary loss (Fig. 5D, f–i).

Through TEM analysis, the infected corneas showed remarkable ultrastructural damage in CoEndo cells when compared with the uninfected corneas, including intercellular vacuolization (Fig. 5E–c) and partial Descemet’s membrane detachment (Fig. 5E–d). The mitochondria were also damaged, characterized by swollen and widened cristae (Fig. 5E–e), mitophagy (Fig. 5E–f), nuclear necrosis (Fig. 5E-g), and deformed mitochondrial proliferation (Fig. 5E–h). These findings demonstrated profound HSV-1-induced endothelial damage, elucidating infection-mediated corneal endothelial dysfunction.

HSV-1 Latency in TG Following AC Infection

Herpes simplex keratitis exhibits viral latency as a key pathogenic feature.^5,29,30 After primary intracameral infection, HSV-1 undergoes retrograde transport via corneal nerves to establish latency in TGs. We characterize this process following the experimental procedures in Figure 6A.

Figure 6.

The latent properties of HSV-1 in TGs after intracameral inoculation. (A) The workflow of investigating the viral latency in TGs. (B) IF staining of HSV-1 antigen (green) in TGs, nuclei counterstained with DAPI (blue). Scale bar = 100 µm. (C) The IF intensity of HSV-1 in TGs under different time points shown by histogram. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. (D) Viral titer of HSV-1 in TGs determined by PFU counting. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. (E) Quantitative RT-PCR analysis of latency-associated transcript (LAT) in TGs under different time points. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. (F) Quantitative RT-PCR analysis of HSV-related gene in TGs under different time points. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant.

IF staining detected HSV-1 antigen in TGs at 1-week post-inoculation (wpi), with no significant signals observed at 2 wpi or 4 wpi (Fig. 6B). This antigen clearance correlated with viral titer dynamics, with TG viral loads detectable at 1 wpi and undetectable by 2 to 4 wpi (Fig. 6D). The qPCR analysis confirmed robust lytic gene expression (gB, gK, and ICP0) at 1 wpi, with subsequent downregulation (Fig. 6F). Conversely, LAT expression significantly increased by 1 wpi, peaked at 2 wpi, and remained elevated throughout 28 dpi (Fig. 6E).

By contrast, ocular inflammation and CCT resolved progressively after acute infection (Supplementary Fig. S2A). Corneal HSV lytic gene (gB, gK, and ICP0) expression remained low post-acute phase (7–28 dpi vs. 3 dpi; Supplementary Fig. S2B). Unlike TG tissues, LAT was undetectable in corneas at all time points (Supplementary Fig. S2C). These data demonstrated that intracameral HSV-1 infection mediates retrograde viral transport to TGs where latency is established and maintained.

Characterization of UVB-Induced Recurrent HSV Endotheliitis After Latency

Recurrent HSV endotheliitis causes more rapid and severe vision loss.^1,22 To model recurrence, we followed the protocol in Figure 7A. After primary infection resolution, corneal transparency and viral clearance, latency was established (Fig. 7B). After corneal UVB exposure, the manifestations of recurrent HSV endotheliitis was examined using SLM and AS-OCT. When compared with UV-MOCK group (equivalent UVB irradiation alone) and latent control, the recurrent group (UV-rHSK) showed evident localized stromal edema (Fig. 7C) and endothelial KPs (Fig. 7D) on day 3 post-UVB exposure, confirming reactivation. The disease progression in the recurrent stage was more rapid than in primary infection, featured by endothelial failure within 3 days (versus 5 days in primary infection; see Fig. 7C), and exacerbated corneal edema and AC inflammation (see Fig. 7D). No significant pathological changes were observed in both the UV-MOCK group and the latent group.

Figure 7.

The clinical and viral characterization of UVB-induced recurrent HSV endotheliitis following latency. (A) The workflow of recurrence of HSV endotheliitis following latency. (B) The comparison of primary infection and latency in corneal opacity, AS-OCT findings, and viral titer. (C) SLM views of recurrent HSV endotheliitis induced by UVB irradiation in different groups at 1 to 7 dpi. (D) AS-OCT images of recurrent HSV endotheliitis induced by UVB irradiation in different groups at 1 to 7 dpi. (E) HSV-1 antigen (green) in corneal and trigeminal ganglion tissues examined by IF staining, nuclei counterstained with DAPI (blue). Scale bar = 100 µm; n = 3 mice per group. (F) Quantitative RT-PCR analysis of HSV-1-related gene (gB, gK, and ICP0) expression in different groups, n = 3 mice per group. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. (G) Quantitative RT-PCR analysis of latency-associated transcript (LAT) in TGs in different groups, n = 3 mice per group. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. (H) The viral titer of HSV-1 in corneas determined by PFU counting. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. (I) Top, The dynamic observation of the replicated HSV-1 (green) at 1, 2, 3, 4, and 5 dpi; bottom, the morphological and density change of mouse CoEndo cells (red) at 1, 2, 3, 4, and 5 dpi. Scale bar = 200 µm. (J) Histogram showing the mean IF density at different times post-infection, n = 4 mice per group. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. (K) Histogram showing the density of CoEndo cells at different times post-infection, n = 6 mice per group. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant.

IF staining showed increased HSV-1 antigen in corneal tissues and TGs of UV-rHSK group compared with the UV-MOCK and the latent groups on day 5 post-recurrence (Fig. 7E). Viral titers analysis revealed the exclusively elevated viral load in recurrent corneas on day 3 post-reactivation (Fig. 7F). The qPCR analysis confirmed the significant increase of viral genes (gB, gK, and ICP0) in corneal tissues only from the recurrent group (Fig. 7G). Additionally, we also detected the increased expression of LAT in recurrent TGs on day 5 post-reactivation (Fig. 7H). These results further confirmed the reactivation and recurrence after corneal UVB exposure.

We subsequently characterized the pathology of CoEndo cells following recurrent HSV-1 infection (Fig. 7I–K). On day 2 post-UVB exposure, the corneal endothelium showed minimal HSV-1 (green) fluorescence and slight endothelial. On days 3 to 4 post-exposure, corneal endothelium exhibited extensive viral proliferation and thereby massive CoEndo cell loss with morphological disruption and monolayer detachment. However, on day 5, near-total CoEndo cells were infected and sharply decreased, along with reduced viral signaling due to massive cell detachment and loss.

Discussion

Corneal endotheliitis, predominantly of viral origin, clinically manifests with corneal edema, KPs, and mild AC inflammation.^22,31–33 Numerous studies have identified herpesviruses as primary pathogens, including CMV, varicella zoster virus (VZV), and HSV-1.^6,34–37 Although clinical features and treatment responses suggest viral etiology, the pathogenesis remains incompletely understood due to the scarcity of validated animal models.

In this study, we used a novel intracameral inoculation protocol to establish a murine HSV-1 corneal endotheliitis model.²⁴ Our technique involves limbal needle insertion into the ciliary sulcus posterior to the iris, contrasting with prior methods involving direct corneal puncture in mice³⁸ and rabbits.³⁹ This approach minimizes ocular trauma while eliminating the requirement for intraocular air or viscoelastic, thereby enhancing procedural precision and safety. By delivering HSV-1 directly to CoEndo cells while bypassing epithelial and stromal compartments, we achieve endothelial-specific pathology, and overcome limitations of earlier models that confounded stromal keratitis with endothelial injury.³⁹ The established model recapitulates human HSV endotheliitis hallmarks, including corneal edema, AC inflammation, KP formation, CoEndo cell loss, and iris involvement, thereby providing a reliable and tractable platform for mechanistic studies and therapeutic discovery.

Although HSV-1 is a major cause of viral corneal endotheliitis,^5,37,40,41 its direct effects on CoEndo cells remain poorly understood. In vitro, CoEndo cells following HSV-1 infection exhibited canonical CPE, including syncytia formation and endosome-like structures, with viral inclusion bodies visualized by TEM. In vivo, HSV-1–inoculated mice replicated the key features of human viral endotheliitis, such as corneal edema, AC inflammation, KPs, CoEndo cell loss, and iris edema. TEM further revealed mitochondrial damage in infected CoEndo cells, suggesting an underlying mechanism for endothelial dysfunction.^42,43 These findings demonstrated that the AC inflammation was also an important contributor to CoEndo loss. Further research is therefore warranted to investigate the respective and synergistic roles of direct HSV-1 cytopathology and the resultant inflammation in driving CoEndo loss.

Elevated IOP was observed in mice following live HSV-1 inoculation, correlating with the severity of corneal endotheliitis. Pathological mechanisms contributing to IOP elevation include aggravated AC inflammation, inflammatory cell accumulation obstructing the iridocorneal angle, and narrowing of the trabecular meshwork, collectively disrupting the blood-aqueous barrier and mechanically impeding aqueous outflow.^44,45 Trabeculitis, identified in other viral models (e.g., CMV and Zika virus), likely exacerbates outflow resistance and IOP elevation.³⁴ Although our study characterized iris and aqueous humor alterations, the associations between trabecular meshwork injury and HSV endotheliitis require further investigation.

Corneal edema, a cardinal feature of HSV-1 infection, was corroborated by IVCM, which revealed endothelial damage (e.g., increased “black holes,” and enlarged intercellular gaps in corneal endothelium) and inflammatory infiltration, consistent with human herpetic endotheliitis.¹ Pathologically, endothelial destruction results from direct viral cytotoxicity and intraocular inflammation, including proinflammatory cytokines and immune cell recruitment.¹ TEM analysis highlighted mitochondrial damage in infected CoEndo cells, a critical observation given that endothelial mitochondria supply ATP for Na+/K+-ATPase pumps,⁴⁶ which maintain corneal transparency. Notably, mitochondrial dysfunction is implicated in diseases such as Fuchs endothelial corneal dystrophy (FECD) and bullous keratopathy,^43,47 indicating a common mechanism underlying HSV-1-induced edema.

Beyond acute endotheliitis, we investigated the establishment of HSV-1 latency following primary infection. During the acute phase, intracameral inoculation resulted in elevated viral loads in TGs and robust expression of lytic genes (gB, gK, and ICP0). In contrast, viral titers in TGs became undetectable by 2 and 4 wpi, coinciding with marked upregulation of LAT—a key marker of viral latency.^29,48 As a neurotropic virus, HSV-1 typically infects corneal sensory nerves and undergoes retrograde transport to TGs, where it establishes lifelong latency.^49,50 Our findings indicate that intracameral HSV-1 likely reach TGs via corneal or nasociliary nerves, facilitating latent persistence. Reactivation of latent HSV-1 in TGs could enable viral reentry into the cornea, causing recurrent herpes simplex keratitis (HSK), which clinically presents as epithelial, stromal, endothelial, or mixed phenotypes.^5,51,52 Given the high recurrence rate of herpetic endotheliitis,⁵³ decoding viral endotheliitis latency mechanisms is essential to advance therapeutic development.

In addition to the pathological feature of primary infection, our study further delineated the UVB-induced recurrence in a murine model of HSV endotheliitis. The established recurrence recapitulates a key aspect of human herpetic endotheliitis: recurrent episodes are more aggressive and destructive than primary infections.^22,54 We observed that recurrent endotheliitis progressed from initial signs (such as localized stromal edema and KPs) to complete endothelial dysfunction within just 3 days, compared with 5 days in primary infection. This accelerated progression aligns with clinical reports of rapid vision loss in recurrent human cases.^1,5 Elevated viral titers, upregulation of lytic genes (gB, gK, and ICP0), and increased expression of LAT in TGs during recurrence confirmed that UVB effectively reactivates virus from latency. Notably, recurrent infection led to massive detachment of CoEndo cells, highlighting viral reactivation accelerates endothelial destruction. We speculate that the rapid progression probably stem from the pre-existing immune priming and residual endothelial damage from the primary infection, which lower the threshold for viral reactivation and tissue destruction. Further studies are required to investigate the underlying mechanism causing the rapid and severe pathology, especially the epigenetic modification in CoEndo cells.

Our study also has several limitations. First, although we established the UVB-induced recurrence mode, it cannot fully recapitulate recurrent herpetic endotheliitis in humans, particularly cases triggered by immunosuppression and feeling fluctuation. Second, the damage and rapid loss of CoEndo cells were observed during primary infection and recurrent herpetic endotheliitis; the underlying mechanisms causing endothelial damage remain incompletely elucidated, specifically during recurrent stage. Third, to better characterize the immune profile and causal gene regulatory networks in primary and recurrent herpetic endotheliitis, single-cell multi-omics is required. Finally, although the involvement of anterior chamber–associated immune deviation (ACAID) in HSV-associated endotheliitis in rabbit models has been reported,³⁹ but its exact roles in primary and recurrent endotheliitis remains to be further explored.

VZV, another alpha-herpesvirus that causes corneal endotheliitis, shares similar pathogenic features with HSV-1,³⁶ including the establishment of latency in the TG and reactivation triggered by immune suppression or UVB exposure. Nevertheless, the pathogenesis of VZV-induced endotheliitis remains poorly understood due to the lack of a suitable animal model, particularly the mechanisms concerning its latency and recurrence. Notably, UVB exposure has been identified as a reliable inducer of VZV reactivation in experimental settings.⁵⁵ Thus, the methodological approach described herein could be adapted to develop a comprehensive VZV endotheliitis model that encompasses primary infection, latency, and recurrence.⁵⁶ Further optimization will be necessary to refine this model, including adjustments to the VZV inoculation titer, UVB exposure dosage, and observation duration. The successful establishment of such a model would facilitate in-depth investigations into virus-endothelial cell interactions, enable preclinical therapeutic evaluation, and thereby advance our understanding of the pathogenesis of herpetic ocular diseases while accelerating the development of targeted therapies.

Conclusions

We have developed a murine model of HSV-1-induced corneal endotheliitis that closely recapitulates the key clinical manifestations of the human disease, including corneal edema, AC inflammation, KP, and CoEndo loss. This model reliably reproduces both primary and latent viral persistence, and reactivation achieved via UVB exposure, thereby offering a robust platform for investigating the pathogenesis of HSV corneal endotheliitis (especially latency and reactivation) and its treatment.