A Novel Murine Model for Lupus-Like Ocular Chronic Graft-Versus-Host Disease

Abstract

Purpose

Lupus-like chronic graft-versus-host disease (cGVHD) has been previously described, but the ocular findings have not been elucidated. Recipient mice in a lupus-like cGVHD model manifested notable and persistent ocular surface phenotypes. Herein, we further explored immunopathogenic mechanisms underlying these ocular phenotypes.

Methods

A previously described lupus-like cGVHD model was established by intraperitoneal injection of splenocytes from bm12 mice into C57BL/6J mice. Systemic findings were evaluated for the presence of splenomegaly, proteinuria, and autoantibodies. Comprehensive evaluations were conducted on ocular manifestations and immunopathological features in this model.

Results

The lupus-like cGVHD model was successfully constructed 2 weeks post-transplantation. The recipient mice developed lupus-like phenotypes, including splenomegaly, proteinuria, and increased autoantibodies, and their ocular presentations included corneal epithelial defects and decreased tear secretion. Histological analysis revealed a reduction in corneal nerve fiber density and corneal endothelial cells, along with conjunctival fibrosis and loss of goblet cells. Moreover, cGVHD induced progressive aggravation of immune cell infiltration and fibrosis in the lacrimal glands. RNA-Sequencing (RNA-seq) results of the lacrimal glands demonstrated that the differentially expressed genes (DEGs) between the control and cGVHD groups were associated with GVHD pathways. Immune infiltration analysis using RNA-seq and flow cytometry confirmed that CD8⁺ T lymphocytes predominantly constituted the inflammatory infiltrating cells within the lacrimal glands.

Conclusions

This lupus-like cGVHD model (bm12→C57BL/6J) exhibited persistent ocular surface manifestations, characterized by immune infiltration of CD8⁺ T lymphocytes in the lacrimal glands. Thus, this ocular cGVHD model may be used to explore the underlying mechanisms and discover novel therapeutic interventions.

Graft-versus-host disease (GVHD) is a highly intricate disorder of allogeneic hematopoietic stem cell transplantation (allo-HSCT) characterized by the involvement of multiple organs.^1,2 With the progressive extension of patients' life span after allo-HSCT, chronic GVHD (cGVHD) has emerged as one of the primary causes of morbidity and mortality in allo-HSCT patients.³ Ocular cGVHD develops in over 60% of allo-HSCT recipients,^4,5 leading to a remarkably diminished quality of life and visual acuity.⁶ These ocular manifestations primarily include conjunctivitis, dry eye disease (DED), meibomian gland dysfunction, keratitis, corneal ulcers, and corneal vascularization.⁷ Similar to the pathogenesis of cGVHD damaging other organ systems, T lymphocyte infiltration-derived immunoinflammatory cascade plays a vital role in damaging ocular tissues, including the lacrimal gland,⁸ conjunctiva,⁹ meibomian gland,¹⁰ cornea,^11,12 retina, and optic nerve.¹³ The current clinical treatments include artificial tears, corticosteroids, nonspecific immunosuppressants, or surgical interventions with unsatisfactory outcomes.¹⁴ Given the challenges in obtaining ocular tissues from patients with cGVHD, it is imperative to conduct in-depth investigations into the immune mechanisms underlying ocular cGVHD using experimental animal models.

Recently, preclinical murine models of allo-HSCT have made significant contributions to the comprehension of cGVHD pathogenesis.¹⁵ Nevertheless, animal models with ocular involvement of cGVHD are limited.^12,16–18 Most of the current models of ocular GVHD come from xenogeneic cell transplants among mice with different genetic backgrounds, including B10.D2→BALB/c,^19–28 C57BL/6→C3-SW,^18,29–31 129S2/SvPasCrl→C57BL/6,³² and C57BL/6→BALB/c.^33,34 These models have different pretreatment methods, including total-body irradiation (TBI) or chemotherapy treatment. The rapid demise of the host mice from multiple organ failures induced by robust rejection reactions precludes the application of this method to studies requiring long-term observation. To ensure accurate ophthalmic observation of cGVHD, maximizing the animals' survival duration is necessary. In addition, a unified mouse background is indispensable for mitigating substantial genetic disparities that may impact the outcomes. The murine models of cGVHD can be broadly categorized into sclerodermatous models, lupus-like models, and thymic dysfunction models.³⁵ The sequence of events leading to the development of cGVHD has primarily been defined through different mouse models.³⁵ Compared to other cGVHD models, the lupus-like cGVHD model is characterized by splenomegaly, lymphadenopathy, nephritis, and increased autoantibodies.³⁶ Most studies involving ocular cGVHD are sclerodermatous models, whereas few studies have utilized the lupus-like cGVHD model to evaluate ocular phenotypes and explore its potential mechanisms comprehensively.

As early as 1989, Appleby et al. have advocated the use of chronic GVHD (bm12→C57BL/6J) as a model for systemic lupus erythematosus (SLE) and verified the satisfactory reproducibility of disease induction in this model.³⁷ This classical lupus-like GVHD model is established by transferring lymphocytes from B6(C)-H2-Ab1^bm12/KhEgJ (bm12) mice, which are genetically identical to C57BL/6 mice except for 3 amino acid substitutions on major histocompatibility complex (MHC) class II, into C57BL/6 mice.³⁶ Perry D et al. summarized the characteristics of bm12 inducible GVHD model and emphasized that it took a single injection of donor cells to induce a lupus-like syndrome.³⁸ Based on these advantages, the bm12 inducible model has been extensively used to investigate the pathogenesis of SLE.^39–41 The severity of the cGVHD model generally relies on the extent of mismatch in MHC I, MHC II, and minor histocompatibility antigen (miHA).³⁵ Because the genetic background of the host and donor mice is B6 mice with minor MHC-II mismatch, the recipient mice (bm12→C57BL/6J) exhibit ameliorated systemic phenotypes, leading to prolonged survival. However, there have been no investigations into the ocular phenotypes and underlying pathogenesis of the bm12 inducible cGVHD model.

Herein, the well-established lupus-like cGVHD model was used to investigate ocular alterations and elucidate the immune pathophysiology driving the progression of ocular cGVHD. First, systemic and ocular phenotypes were thoroughly analyzed and assessed in the recipient mice. Subsequently, we performed histological examinations of ocular tissues, including the cornea, conjunctiva, and lacrimal gland. Furthermore, the immune infiltration of the lacrimal gland was analyzed through a combination of sequencing and flow cytometry.

Methods

Establishment of a Lupus-Like cGVHD Murine Model

Animal experiments were in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and approved by the Institutional Animal Care and Use Committee of Sun Yat-Sen University, Guangzhou, China. All animal procedures on mice were carried out in accordance with standard ethical guidelines. C57BL/6J were obtained from GemPharmatech (Jiangsu, China) and B6(C)-H2-Ab1^bm12/KhEgJ mice (bm12, H-2^b) were purchased from the Jackson Laboratory (Bar Harbor, ME, USA). The 6 to 8-week-old mice were used to establish the previously described cGVHD model.^36,41 The C57BL/6J allogeneic recipient mice were injected intraperitoneally with 3 × 10⁷ splenocytes in 250 µL phosphate-buffered saline (PBS; Gibco, Grand Island, NY, USA) containing 1% fetal bovine serum (FBS) (Gibco) from bm12 donor mice and were subsequently raised under normal conditions (allogeneic transplantation). The C57BL/6J syngeneic control mice were injected intraperitoneally with 3 × 10⁷ splenocytes 250 µL PBS containing 1% FBS from C57BL/6J donor mice (syngeneic transplantation). Blood, spleens, ocular tissues, and urine samples were obtained from experimental mice the day before the transplant and biweekly after that.

Systemic Evaluation in the Lupus-Like cGVHD Model

Daily systemic assessment was conducted. The body weights and spleen weights were recorded weekly following splenocyte grafting. Proteinuria was evaluated using semi-quantitative dipstick analysis (Bayer Diagnostics, Basingstoke, UK). Serum anti-dsDNA autoantibodies were assessed by ELISA, as previously described (Sigma, St. Louis, MO, USA).⁴¹

Ocular Evaluation in the Lupus-Like cGVHD Model

Corneal Fluorescein Staining Assessment

Corneal fluorescein staining was used to assess the extent of corneal epithelial defects.⁴² Then, 2 µL of 2.5% fluorescein sodium dissolved in normal saline was dropped into the conjunctival sac of the mice and any excess sodium fluorescein was carefully absorbed using cotton swabs before photography. Then, the corneas were observed and photographed using a slit-lamp biomicroscope under cobalt blue light. The percentage of the fluorescein-stained area in the cornea was analyzed and calculated by ImageJ software. The corneal fluorescein score (CFS) value was assessed using a grading scale ranging from 0 to 4: 0 = absent; 1 = slightly punctate staining with <30 spots; 2 = punctate staining with >30 spots but not diffuse; 3 = severe diffuse staining but no positive plaques; and 4 = positive fluorescein plaques. The scores obtained from the four domains were aggregated to derive a composite score (ranging from 0 to 16 points).¹⁹

Measurement of Corneal Sensitivity

As previously described, the Cochet-Bonnet esthesiometer (Lunean, France) was used to determine corneal sensitivity in the central cornea.⁴³ A monofilament ranging from 6.0 to 0.5 cm in length was applied to stimulate the central cornea of mice without anesthesia, eliciting a blink reflex. Subsequently, the length of the monofilament was progressively reduced in increments of 0.5 cm, starting from an initial length of 6.0 cm. A positive response was recorded for each length when two or more blinks were observed within three attempts.

Measurement of Tear Production

Tear production using cotton thread test was assessed in general anesthetized mice using phenol red cotton threads (Jingming, Tianjin, China), as previously described.⁴⁴ The phenol red cotton threads were carefully grasped with forceps and gently positioned in the lower palpebral conjunctival sac, approximately one-third of the distance from the lateral canthus for 30 seconds. Afterward, the length of the wetted thread was precisely measured using a vernier caliper. The average values obtained from bilateral eyes were utilized for subsequent analysis.

Color Fundus Photograph and Fundus Fluorescein Angiograms

The color fundus photography and fundus fluorescein angiograms (FFAs) were performed in anesthetized mice using a Phoenix Micron IV retinal imaging microscope (Phoenix, Pleasanton, CA, USA) at week 10 after transplantation. After mydriasis, the mice were gently positioned on the bracket and utilized carboxymethylcellulose ointment as a medium. The optical lens was carefully maneuvered toward the cornea along the visual axis. The fundus color photographs were captured, and FFA images were acquired 5 minutes following the intraperitoneal injection of fluorescein sodium.

Histological Analysis of Ocular Tissues

Ocular Tissue Isolation, Fixation, and Staining

Mice were humanely euthanized, and the enucleated eyes, conjunctiva, and extraorbital lacrimal glands were isolated under a stereomicroscope. It should be noted that the eyeballs were sectioned sagittally through the optic nerve. Corneal scissors were used to cut the fur from the outer canthus to the ear to expose the lacrimal glands. These ocular tissues were fixed for paraffin-embedded or frozen sections for different purposes. The tissue fixation, dehydration, and embedding were conducted in accordance with the methods described in the reference 45. Afterward, these sections were processed for hematoxylin-eosin (H&E) staining, Masson staining, periodic acid-schiff (PAS) staining, and Oil Red O staining (Servicebio, Wuhan, China). Images were scanned by KFBio scanner (KFBIO, KF-PRO-020, Ningbo, China).

Corneal Nerve Fiber Staining

The corneal nerve fiber was stained using a standardized method, as previously described.⁴⁶ Whole corneas were fixed in 4% PFA for 30 minutes and washed by PBS. After blocking with PBS containing 5% bovine serum albumin (BSA) and 0.3% triton (Sigma), the corneas were incubated with neuron-specific marker β-III tubulin antibody (1:50; ab52623; Abcam, Waltham, MA, USA) at 4°C overnight. The following day, the secondary antibody was incubated for 60 minutes at room temperature (RT) under light-protected conditions. Images of the whole corneas were captured and analyzed using an Olympus microscope.

Immunofluorescence and Immunohistochemistry Staining

The immunofluorescence staining and immunohistochemistry staining were performed as previously described protocols.⁴⁵ The tissue sections were blocked with PBS containing 3% BSA and 0.3% triton for 60 minutes at RT and incubated with primary antibodies at 4°C overnight. For immunofluorescence staining, the sections were incubated with appropriate fluorescent secondary antibodies for 60 minutes at RT, and the nuclei were stained using DAPI for 10 minutes. For immunohistochemistry staining, the sections were incubated with Horseradish peroxidase- (HRP)-conjugated secondary antibodies, followed by chromogenic detection using DAB. Finally, fluorescence sections were taken under an Olympus fluorescence microscope, and immunohistochemistry sections were scanned using KFBio scanner. The primary antibodies used in immunohistochemistry and immunofluorescence assays were as follows: α-SMA (1:100; GB111364; Servicebio), E-Cadherin (1:100; GB12083; Servicebio), CD4 (1:100; GB13064-2; Servicebio), CD8 (1:100; GB13429; Servicebio), NLRP3 (1:100; A5652; ABclonal, Wuhan, China), and iNOS (1:100; GB11119; Servicebio). Isotype control antibodies were used as negative control (Supplementary Figs. S1, S2).

RNA-Seq Analysis

The methods of RNA sequencing and bioinformatic analysis have been referred to in previous literature.⁴⁷ The lacrimal glands were isolated and snap-frozen in liquid nitrogen. RNA extraction was performed using the Qiagen RNeasy Plus Mini Kit following the manufacturer's instructions. The transcriptome sequencing and analysis of the bilateral lacrimal glands in the control (n = 4 mice) and cGVHD (n = 4 mice) groups at week 10 post-transplantation were conducted by OE Biotech Co., Ltd. (Shanghai, China) with the Illumina HiSeq sequencing platform. After quality control, the clean reads were aligned to the mouse reference genome using Hisat2. The read counts of each gene were acquired from HTSeq-count and subsequently normalized by the fragments per kilobase million (FPKM) method for gene expression quantification. Q value < 0.05 and fold change >2 or <0.5 were used as the threshold values to identify the differentially expressed genes (DEGs). To assess the functions of DEGs in the control and cGVHD groups, Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway, and Wiki pathway enrichment analyses of DEGs were performed using R statistical software. The GO analysis was performed at three levels: biological process (BP), cellular component (CC), and molecular function (MF). For immune infiltration analysis, the Ensembl database was utilized to convert mouse gene names into their human counterparts. The immune-cell proportions in the lacrimal glands between the control and cGVHD groups were calculated using the CIBERSORT algorithm.

Flow Cytometric Analysis

Multichannel flow cytometry was used to evaluate the infiltration and subtypes of immune cells. The lacrimal gland was first isolated and then dissected into multiple small fragments using microscopic scissors. Subsequently, the tissue fragments were enzymatically dissociated into single-cell suspensions using RPMI 1640 medium (Gibco) supplemented with collagenase type D and DNase I (Sigma) for 1 hour at 37°C in a shaking water bath. Following complete digestion, the suspension was filtered using a BD Falcon 40 µm strainer, subjected to centrifugation, and subsequently the supernatant was decanted. The resulting pellet was resuspended in fluorescence-activated cell sorting (FACS) buffer containing 1% FBS. The cells were incubated with CD45, CD3, NK1.1, CD19, CD8, and CD4 antibodies for 30 minutes at 4°C under light-protected conditions. Then, the cells were rinsed and treated with BD stabilizing fixative. The stained cells were detected the next day on a flow cytometer (Invitrogen Attune NxT), and the results were analyzed using FlowJo Software. The antibodies used in flow cytometry were as follows: CD45-eF450 (48-0451-82; eBioscience, Waltham, MA, USA), CD3e-FITC (100306; Biolegend, San Diego, CA, USA), CD8a-PE (100707; Biolegend), CD4-PEcy5 (15-0042-82; eBioscience), NK1.1-APC (108710; Biolegend), and CD19- PEcy7 (25-0193-82; eBioscience). Unstained cells, Full Minus One (FMO), and full panel controls were performed (Supplementary Fig. S3).

Statistical Analysis

All experiments and measurements were performed in triplicate to ensure reproducibility. The results were presented as the mean ± standard deviation (SD). Statistical analysis was conducted using GraphPad Prism Software (version 9.5.0; GraphPad Software Inc., San Diego, CA, USA). A two-tailed Student's t-test was used to compare the control and cGVHD groups. Statistical significance was defined as * P < 0.05, ** P < 0.01, and *** P < 0.001.

Results

Systemic Manifestations in the Lupus-Like cGVHD Model

The workflow of establishing the lupus-like GVHD model was presented in Figure 1A. The lupus-like cGVHD murine model was successfully established 2 weeks post splenocyte grafting and lasted to the 10th week, as evidenced by splenomegaly (Figs. 1B, 1C), proteinuria (Fig. 1D), and elevated autoantibodies in recipient mice (Fig. 1E). The lupus-like cGVHD did not elicit manifestations such as weight loss (Fig. 1F), alopecia, or diarrhea observed in other models.^18,48–50 The survival rate observed until the end point in the cGVHD model was 100%.

Figure 1.

Systemic manifestations in the lupus-like cGVHD model. (A) Experimental design and workflow of establishing the lupus-like cGVHD model by transferring spleen lymphocytes from bm12 mice into the C57BL/6J mice. The workflow was drawn with www.biorender.com. (B) Gross appearances of the spleens and (C) the ratios of spleen/body weight (mg/g) in the control and cGVHD groups at week 2, week 6, and week 10 (n = 5 mice). (D) The percentage of mice with proteinuria > 1⁺ at week 2, week 6, and week 10 (n = 12 mice). (E) Serum anti-dsDNA autoantibody titers at week 2, week 6, and week 10 (n = 5 mice). (F) Body weight changes over time after transplantation (n = 12 mice). Data were presented as mean ± SD. ns = not significant. * P < 0.05, ** P < 0.01, *** P < 0.001.

Ocular Phenotypes of the Lupus-Like cGVHD Model

Subsequently, we focused on the ocular cGVHD phenotypes by assessing corneal fluorescein staining, quantifying corneal sensitivity, and evaluating tear production. As shown in Figure 2A, the corneas of mice in the cGVHD group started appearing punctate staining at the second week and gradually progressed to diffuse plaque staining by the 10th week. Accordingly, the CFS values in the cGVHD group were significantly higher compared to the control group (Fig. 2B). Moreover, regressive loss of corneal sensitivity in the cGVHD group was demonstrated in Figure 2C. After 2 weeks of transplantation, a significant decrease in tear volume in the cGVHD group was observed, and this decrease persisted until the tested end point of 10 weeks (Figs. 2D, 2E). Additionally, color fundus photography, FFA, and H&E staining were used for the fundus examinations. All the results revealed that the fundus of mice in the cGVHD group appeared normal, and no discernible changes were detected after transplantation (Figs. 3A, 3B). Thus, the lupus-like cGVHD mainly led to progressively aggravate the ocular surface phenotypes.

Figure 2.

Ocular surface phenotypes in the lupus-like cGVHD model. The evaluation of the ocular surface was conducted at week 2, week 6, and week 10 following transplantation (n = 3 mice). (A) Representative photographs of corneal fluorescein sodium staining under slit lamp and (B) scoring of corneal fluorescein. (C) Central corneal sensitivity was measured with Cochet–Bonnet esthesiometry. (D) Tear secretion was measured using phenol red cotton threads, with damp threads indicated by a red coloration. (E) Tear volume was quantitated as the length (mm) of the red-colored threads. Data were presented as mean ± SD. * P < 0.05, ** P < 0.01, *** P < 0.001.

Figure 3.

Effects of lupus-like cGVHD on the retina. The evaluation of the retina was conducted at week 10 following transplantation (n = 3 mice). (A) Representative photographs of the fundus and FFA were obtained at week 10 of follow-up. (B) H&E staining of the retina across the optic nerve (left; Bar = 100 µm) and higher-magnification images (right; Bar = 50 µm) at week 10 post-transplantation.

Histological Changes of the Cornea in the Lupus-Like cGVHD Model

To further confirm the corneal sensation loss in the progression of cGVHD, the corneal nerve fiber density of the cGVHD mice was evaluated by whole mount staining of β-III Tubulin. As anticipated, the density of corneal nerve fiber in the cGVHD group decreased significantly, particularly in the central cornea (Fig. 4A). However, any apparent changes were not observed in the average central corneal thickness or the corneal structure throughout the progression of GVHD by H&E staining (Fig. 4B). In comparison to the control group, the cGVHD group exhibited a significant decrease in corneal endothelial cell count (Figs. 4C, 4D), which was consistent with findings reported in a previous study.⁵¹ Furthermore, through TUNEL assay staining, we observed a substantial presence of apoptotic squamous epithelial cells within the outermost layer of the corneal epithelium, which aligns with our findings of corneal epithelial defects (Supplementary Fig. S4A). Consequently, lupus-like cGVHD induced a reduction in the corneal nerve fiber density, increased the presence of apoptotic squamous epithelial cells, and loss of corneal endothelial cells.

Figure 4.

Histological changes of the cornea in the lupus-like cGVHD model. The cornea underwent histological analysis at week 2, week 6, and week 10 following transplantation (n = 5 mice). (A) Immunofluorescence staining of corneal fiber nerves stained with β-III Tubulin in corneal whole mounts (Bar = 500 µm) and higher-magnification images of the central cornea (Bar = 100 µm). (B) H&E staining was performed to assess the histological changes in the central cornea (Bar = 50 µm). (C) Higher-magnification images of the central cornea were obtained to quantitatively assess the number of corneal endothelial cells (blue arrows; Bar = 25 µm). (D) Quantification of corneal endothelial cell density in two groups. Data were presented as mean ± SD. ns = not significant. * P < 0.05, ** P < 0.01, *** P < 0.001.

Histological Changes of the Conjunctiva in the Lupus-Like cGVHD Model

Considering the susceptibility of the conjunctiva to inflammation caused by cGVHD, we also examined alterations in conjunctival tissues in our experimental model.¹¹ First, H&E staining of the conjunctiva revealed the presence of vacuolization in goblet cells and a decrease in cell count (Fig. 5A), which was further confirmed by PAS staining (Figs. 5B, 5C). The TUNEL assay staining revealed a notable occurrence of apoptosis in conjunctival goblet cells (see Supplementary Fig. S4B). Second, Masson staining revealed a pronounced increase in collagen deposition surrounding the conjunctiva (Fig. 5D). Overall, the lupus-like cGVHD led to the loss and apoptosis of conjunctival goblet cells, as well as the development of conjunctival fibrosis.

Figure 5.

Histological changes of the conjunctiva in the lupus-like cGVHD model. The conjunctiva underwent histological analysis 10 weeks post-transplantation (n = 5 mice). (A) H&E staining was conducted to evaluate the structural alterations of the conjunctiva (Bar = 50 µm). (B) PAS staining of the conjunctiva (Bar = 50 µm). (C) Quantification analysis of PAS-positive goblet cells. (D) Masson staining of the conjunctiva (Bar = 50 µm). Data were presented as mean ± SD. * P < 0.05, ** P < 0.01, *** P < 0.001.

Histological Analysis of the Lacrimal Gland in the Lupus-Like cGVHD Model

The extraorbital lacrimal gland is one of the profoundly impacted organs in cGVHD.⁵² Gross photographs showed that cGVHD induced lacrimal gland atrophy 6 and 10 weeks post splenocyte grafting (Fig. 6A). Correspondingly, the ratios of bilateral lacrimal glands/body weight of the cGVHD group were markedly lower than those of the control group (Fig. 6B). To evaluate the severity of lacrimal gland injuries, sectioning and staining were subsequently performed on the lacrimal glands. The H&E staining results revealed a progressive increase in inflammatory cell infiltration within the lacrimal glands of the cGVHD mice compared to the control group over time (Fig. 6C). The H&E staining showed significant apoptosis of lacrimal glands, which was confirmed by TUNEL staining (see Supplementary Fig. S4C). Masson staining showed that the extent of fibrotic collagen deposition in the lacrimal glands of the cGVHD mice was significantly elevated compared to that observed in healthy controls (Figs. 6D, 6E). Accumulation of lipid droplets was observed among the acinar cells in the cGVHD group using Oil Red O staining, particularly at sites with immune-cell infiltration (Fig. 6F). Progressive destruction of the lacrimal glands in the cGVHD group, characterized by atrophy of the acini and ducts, was observed in both H&E and Masson staining (see Figs. 6C, 6D). To further investigate the extent of fibrosis in the lacrimal glands, we performed immunofluorescence staining (Fig. 6G) to assess the expression of α-smooth muscle actin (α-SMA) and E-cadherin. Increased expression of α-SMA and reduced expression of E-cadherin indicated that lacrimal gland epithelia underwent the epithelial-mesenchymal transition (EMT) process, which was also confirmed in the human GVHD model.⁵³ In addition, cGVHD induced NOD-like receptor thermal protein domain associated protein 3 (NLRP3) inflammasome activation in the lacrimal gland (Supplementary Fig. S5). In summary, the results demonstrated that lipid accumulation, fibrosis, apoptosis, and immune responses were primary pathological characteristics observed within the lupus-like cGVHD model's lacrimal glands, ultimately contributing to reduced tear production and secretion.

Figure 6.

Histological analysis of the lacrimal gland in the lupus-like cGVHD model. The lacrimal gland underwent histological analysis at week 2, week 6, and week 10 following transplantation (n = 5 mice). (A) Gross appearances of the lacrimal glands and (B) the ratios of bilateral lacrimal glands/body weight (mg/g) in 2 groups at week 2, week 6, and week 10 post-transplantation. (C) H&E staining of the lacrimal glands (Bar = 100 µm) and higher-magnification images (Bar = 25 µm) revealed that immune cell infiltration commenced as early as 2 weeks post-transplantation in the lacrimal glands and exhibited a progressive increased trend every week thereafter. (D) Masson staining of the lacrimal glands (Bar = 100 µm). (E) Blue fibrotic areas by Masson staining were measured using ImageJ. (F) Oil red O staining of the lacrimal glands at week 10 post-transplantation (Bar = 50 µm). (G) Immunofluorescence staining of E-cadherin and α-SMA (Bar = 100 µm) and higher-magnification images (Bar = 20 µm) in the lacrimal glands at week 10 post-transplantation. Data were presented as mean ± SD. ns = not significant. * P < 0.05, ** P < 0.01, *** P < 0.001.

RNA-Seq Analysis of Gene Expression Profiles in the Lacrimal Glands

Our study revealed significant infiltration and damage specifically in the lacrimal gland, prompting us to conduct RNA-seq analysis to investigate alterations in gene expression profiles during the progression of cGVHD. The sequencing data revealed that the cGVHD group exhibited 1590 upregulated genes and 1551 downregulated genes in contrast to the control group (Fig. 7A). Enrichment analysis using GO (BP/CC/MF), KEGG pathways, and Wiki pathways was conducted to predict the potential functions of DEGs. GO analysis indicated that DEGs were enriched in the cellular response to positive regulation of T cell activation (BP), immune response (BP), extracellular region (CC), external side of the plasma membrane (CC), protein binding (MF), and transmembrane signaling receptor activity (MF; Fig. 7B). KEGG pathway analysis revealed that DEGs were enriched in GVHD, antigen processing and presentation, and Th1 and Th2 cell differentiation (Fig. 7C). Wiki enrichment analysis demonstrated that DEGs were enriched in the inflammatory response, oxidative damage response, lung fibrosis, and apoptosis (Fig. 7D). GSEA further confirmed that significant activation of adaptive immune response (GO), positive regulation of T cell activation (GO), allograft rejection (KEGG), GVHD (KEGG), type II interferon (IFN-γ) signaling (Wiki), and complement activation (Wiki) in the lacrimal glands of the cGVHD group (Fig. 7E). Genes contributing to the top enrichment terms of GO, KEGG, and Wiki analyses by chord plots in Supplementary Figures S6A–C. In summary, RNA-seq results revealed a cascade of immune-related events in the lacrimal glands that had undergone cGVHD, highly consistent with the pathogenesis of GVHD proved by previous studies.

Figure 7.

RNA-seq analysis of the lacrimal glands. RNA-seq analysis was used to compare the genetic changes of the lacrimal glands between the control (n = 4 mice) and cGVHD groups at week 10 post-transplantation (n = 4 mice). (A) Histogram of differently expressed gene numbers. (B) GO enrichment analysis in biological process, cellular component, and molecular function. (C) KEGG pathway enrichment analysis. (D) Wiki pathway enrichment analysis. (E) GSEA analysis based on the GO gene set, the KEGG gene set, and the Wiki gene set. (F) Comparisons of immune-cell proportions between the cGVHD and control lacrimal glands by using the CIBERSORT algorithm. (G, H) Among these immune cell subsets, the proportions of CD8⁺ T cells and M1 macrophages were significantly increased in the cGVHD group. Data were presented as mean ± SD.

To further elucidate the subtypes of immune infiltration, the CIBERSORT algorithm was used to calculate individual relative proportions of infiltrated immune cells (Fig. 7F). The CIBERSORT algorithm further illustrated that significantly elevated ratios were observed for CD8⁺ T cells (Fig. 7G) and proinflammatory M1 macrophages (Fig. 7H) in the lacrimal glands from the cGVHD group compared to those from the control group among these immune cells. Therefore, these results preliminarily suggested a dominance of CD8⁺ T cell infiltration accompanied by upregulation of M1-type pro-inflammatory macrophages.

Inflammatory Cell Infiltration in the Lacrimal Glands

To validate the transcriptomic findings, flow cytometric analysis was conducted to investigate the infiltration of inflammatory cells in lacrimal glands. The flow cytometric analysis included live, single, and CD45⁺ cells for assessing inflammatory cell populations. Initially, there was a slight increase in the percentage of CD45⁺ cells observed in the cGVHD group compared to the control group (Figs. 8A, 8B). Furthermore, the cGVHD group presented an elevated proportion of CD3⁺ T lymphocytes compared to controls (see Figs. 8A, 8C). Subsequently, CD3⁺ T lymphocytes were further subdivided into CD4⁺ T lymphocyte cells and CD8⁺ T lymphocyte cells (see Figs. 8A, 8D). Consistent with the RNA-seq results, CD8⁺ T cells accounted for a higher proportion among T lymphocytes, and the corresponding ratio of CD4⁺ T cells to CD8⁺ T cells was significantly down-regulated within the lacrimal glands in the cGVHD group (see Figs. 8A, 8E). The extent of CD8⁺ and CD4⁺ T cell infiltration in tissue sections was further evaluated through immunohistochemistry experiments. The findings verified a significant infiltration of CD8⁺ and CD4⁺ T cells along the acini and ducts of the lacrimal glands in the cGVHD group, and the infiltration of CD8⁺ T cells was more pronounced compared to that of CD4⁺ T cells (see Figs. 8F, 8G). Collectively, alloreactive CD8⁺ T cells played a dominant role in mediating the damage to the lacrimal gland following allogeneic transplantation.

Figure 8.

Inflammatory cell infiltration of the lacrimal glands. The infiltration of inflammatory cells in the lacrimal glands was analyzed by multichannel flow cytometric analysis at week 10 after transplantation (n = 3 mice). (A) Gating strategy and representative flow cytometry plots to analyze inflammatory cells in the lacrimal glands. (B) The quantitative proportion of CD45⁺ cells among total live cells. (C) The quantitative proportion of T cells among CD19^- cells. (D) The percent composition of T cell subsets presented a higher CD8⁺ T cell composition in the cGVHD group. (E) The ratios of CD4⁺/CD8⁺ cells were significantly lower in the cGVHD group compared to the control group. (F) Immunohistochemical staining of CD8 in the lacrimal glands (Bar = 50 µm). (G) Immunohistochemical staining of CD4 in the lacrimal glands (Bar = 50 µm). Data were presented as mean ± SD. * P < 0.05, ** P < 0.01, *** P < 0.001.

Discussion

Herein, our study comprehensively assessed a well-established cGVHD model characterized by the recipients' lupus-like phenotypes with extended survival. In terms of the construction method, both recipient and donor mice were derived from C57BL/6J background mice and did not require TBI, so the persistent ocular phenotypic abnormalities could be purely attributable to the immunologic changes from the transplantation in this model.⁵⁴ Additionally, the observed systemic manifestations in mice closely resembled lupus syndrome, including splenomegaly, proteinuria, and elevated levels of autoantibodies, with a 100% survival rate. Furthermore, the ocular phenotypes observed in these mice were as noted in the patients with ocular cGVHD, characterized by persistent corneal sensitivity loss, corneal epithelial defects, and reduced tear secretion.⁷ Finally, the observed pathological changes in ocular tissues, such as infiltration of immune cells, cellular apoptosis, fibrosis, and other phenomena, are consistent with the progressive nature of cGVHD.^55,56 Therefore, this model serves as a valuable tool for investigating the pathogenic mechanisms of ocular cGVHD.

In this cGVHD model, ocular findings occurred on the ocular surface, although the fundus exhibited near-normal characteristics, consistent with clinical symptoms in patients with ocular cGVHD.⁵⁷ As shown in Figure 2, the cGVHD model showed that ocular epithelial defects and decreased tear production were the most prominent ocular surface phenotypes. Consistently, the main histological features in the cornea included a reduction in corneal nerve fiber density (see Fig. 4A)^31,58 and corneal endothelial cell count (see Figs. 4C, 4D).^51,59 Corneal neurosensory abnormalities can result in a decline in blink reflex, thereby inducing corneal epithelial defects and delaying epithelial healing.^60,61 The prevalence of neurotrophic keratopathy in patients with ocular cGVHD was reported to be 14%.⁶¹ Abnormal activation of complement C3/CD4⁺ T-cell axis can mediate corneal sensory neuropathy.³¹ The corneal endothelial cell number reduction was associated with increased expression of the proinflammatory marker NK1R in an acute GVHD (aGVHD) model.⁵¹ Regarding decreased tear secretion, progressive inflammatory infiltration, fibrosis, and aberrant lipid accumulation in lacrimal glands contributed to reduced tear production and secretion (see Figs. 6C–F). Generally, the interplay among these factors is also evident. Corneal nerve abnormalities may arise as a consequence of diminished tear secretion and augmented levels of inflammatory cytokines within the tears. In turn, decreased corneal sensitivity would further exacerbate dry eye.

In contrast to aGVHD, cGVHD is a multifaceted autoimmune disorder characterized by the involvement of both donor and recipient B and T cells, leading to diverse clinical manifestations associated with immune complex deposition and systemic fibrosis, affecting multiple organs such as the lungs, skin, and gastrointestinal tract.^55,62–64 In our study, progressive fibrosis and immune cell infiltration of the lacrimal glands were evident in the cGVHD group (see Figs. 6C, 6D). Numerous studies have demonstrated that immunological mediators released from immune cells lead to fibrosis and fibrotic disorders.^52,65–67 Therefore, these results prompted us to investigate immune response and immune-infiltrated subtypes in the lacrimal glands.

To further elucidate the genetic and functional alterations in the lacrimal gland in this model, transcriptome sequencing analysis was conducted. Additionally, immune infiltration analysis using transcriptome sequencing can facilitate the identification of immune-infiltrated cells within the lacrimal gland. First, RNA-seq analysis confirmed that DEGs were significantly enriched in pathways related to GVHD, immune rejection, and T lymphocyte activation, providing robust evidence for the immune response storm in ocular cGVHD (see Figs. 7B–E). Moreover, these results indicated a severe T cell-mediated immune response in the lacrimal gland. Indeed, both CD8⁺ and CD4⁺ T cells are essential in most MHC-mismatched GVHD models; some studies exhibit a CD4-dominated pattern, while others demonstrate a CD8-dominated pattern.⁶⁴ Utilizing the CIBERSORT algorithm (see Fig. 7G), flow cytometry (see Fig. 8A), and immunohistochemical analysis (see Fig. 8F), it was revealed that T-lymphocyte subtypes in the lacrimal glands primarily consisted of CD8⁺ T lymphocytes in our model. The periductal area of the lacrimal gland was infiltrated by CD4⁺ T cells and cytotoxic CD8⁺ T cells, resulting in tissue damage. The CIBERSORT algorithm also revealed a significant increase in proinflammatory M1 macrophages, and we further identified a small number of infiltrating M1 macrophage cells expressing iNOS within the lacrimal glands in the cGVHD group (Supplementary Fig. S7). Furthermore, Wiki analysis revealed that DEGs in the cGVHD group were significantly enriched in fibrosis and apoptosis-related pathways (see Fig. 7D). This finding was consistent with Masson and TUNEL assay staining observed in ocular tissues, verifying that the pathological changes in our model were consistent with patients with cGVHD. Additionally, DEGs were enriched in the IFN-γ signaling pathway (see Figs. 7D, 7E), and a heatmap of the IFN signature presented that genes, including Stat1 and Cxcl9, were significantly upregulated in the cGVHD group (see Supplementary Fig. S6D). The cytokine IFN-γ, secreted by activated T cells, has been demonstrated to exert direct effects on the pathology of cGVHD and can induce polarization of macrophages towards a proinflammatory M1-like state.^68,69 To sum up, the findings suggested that the activation, expansion, and migration of CD8⁺ T cells was the main cause of inflammation and fibrosis in the lacrimal glands.

The immune infiltration of CD8⁺ T cells was also found in the cornea and conjunctiva. In contrast to a pronounced infiltration of immune cells in the lacrimal glands of the cGVHD group, only a few CD8⁺ T cells were detected in the corneal limbus (Supplementary Fig. S8) and conjunctiva (Supplementary Fig. S9). The conjunctiva, as a mucosal tissue, possesses an immune protection mechanism that limits the infiltration of immune cells and the propagation of inflammatory responses.^70,71 The special immune tolerance mechanism of the conjunctiva, could explain the lower degree of inflammatory cell infiltration seen in our murine model, as compared to their abundance in the lacrimal gland. Ocular surface health highly depends on the homeostasis of the tear film. The biochemical analysis of tears revealed significantly elevated levels of inflammatory cytokines in ocular GVHD eyes compared to control eyes, as reported by previous studies.^72,73 The decrease in tear volume concurrently led to an elevation in tear osmolarity in a GVHD murine model.⁷⁴ Conjunctival goblet cells are highly sensitive to cytokines and can undergo metaplasia or cell death in response to even small amounts of IFN-γ, disrupting mucin secretion.⁷⁵ Our sequencing results from the lacrimal gland further support these findings, revealing significant activation of the IFN-γ-related pathway. Based on these results, we speculated that the damage to the cornea and conjunctiva might be secondary to the exacerbation of the immune inflammatory response in the lacrimal glands. Reinforced by the findings, we will further utilize this model to identify potential therapeutic avenues for ocular cGVHD by suppressing immune responses in the lacrimal gland.

In conclusion, the current lupus-like cGVHD murine model has ocular findings consistent with ocular cGVHD, characterized by measurable clinical features, such as decreased tear production, corneal and conjunctival findings, histo- and immuno-pathological features, characterized by CD8⁺ T cell-dependent lacrimal gland damages, leading to ocular surface disease found in patients with ocular cGVHD. The bm12 inducible model is a rapid method to establish the ocular cGVHD murine model, facilitating further investigation into the immunopathological mechanisms, novel therapeutic targets, and evaluation of the efficacy of drugs.