Mechanistic Insights into Glucocorticoid-Induced Ocular Hypertension Using Differences in Mouse Strain Responsiveness

Abstract

Glucocorticoids (GCs) are widely prescribed anti-inflammatory agents. Unfortunately, many people experience negative adverse effects associated with long term GC therapy, developing GC-induced ocular hypertension (GC-OHT), which can lead to secondary glaucoma. Approximately 40% of the treated individuals are susceptible to GC-OHT. Seventy years since this discovery, the molecular mechanisms underlying GC-OHT remain unclear. We previously developed a mouse model of GC-OHT delivering the potent GC dexamethasone and observed strain-specific disparities in the development of GC-OHT. We now compare phenotypic and transcriptomic differences between five genetically distinct inbred mouse strains to identify biomarkers of GC susceptibility, and to better understand the molecular mechanisms of GC-OHT. Like humans, mouse strains differ in their ability to develop GC-OHT. Phenotypic characterization revealed that C57BL/6J and C3H/HeJ mice are GC responders and more susceptible to develop GC-OHT. Dexamethasone treatment in these strains led to elevated intraocular pressure compared with the GC nonresponder strains DBA/2J.Gpnmb⁺, 129P3/J, and BALB/cJ. Transcriptomic analysis of responder and nonresponder mouse strains revealed novel trabecular meshwork biomarkers of GC-OHT susceptibility involving enrichment of molecular pathways unique to this response. The present study identifies putative mechanisms underlying GC-OHT and provides insight into the pathogenesis of the clinically similar but more prevalent primary open-angle glaucoma.

Graphical Abstract

Glucocorticoids (GCs) are one of the most prescribed classes of anti-inflammatory and immunosuppressive therapies. Currently, an estimated 1% to 3% of the worldwide population is prescribed glucocorticoid therapy annually.1, 2, 3, 4 Since their discovery in the 1940s, synthetic GCs have been a mainstay for treating immunologic and inflammatory disorders, reflecting their increased potencies and specificity highlighting their clinical impact and increased use over the years. Despite their numerous therapeutic benefits, the adverse effects associated with chronic GC use collectively contribute to a significant health and economic burden.^2,5

Elevated intraocular pressure (IOP) is a major ocular adverse effect associated with GC therapy, leading to glucocorticoid-induced ocular hypertension (GC-OHT). GC-induced adverse effects are observed only in a subset of patients undergoing GC therapy. Approximately, 30% to 40% of individuals treated with GCs are susceptible to GC-OHT and are GC responders, whereas the remaining are GC nonresponders.6, 7, 8 In these susceptible individuals (GC responders), GC-induced insults to the iridocorneal angle tissues cause increased resistance to aqueous humor outflow and elevated IOP, which can lead to a secondary form of iatrogenic open-angle glaucoma known as glucocorticoid-induced glaucoma (GIG). The pathology of GIG involves ocular hypertension-induced neuropathic damage to the optic nerve axons, which results in progressive vision loss. Interestingly, the prevalence of GC-OHT is considerably higher (90%) in individuals experiencing primary open-angle glaucoma (POAG), the most prevalent form of primary glaucoma.⁹ Furthermore, first-degree relatives of patients with POAG are more likely to be GC responders (approximately 70% to 80%).^10,11 This indicates a potential genetic influence on GC-OHT susceptibility and pathogenesis.

Multiple studies have shown that GC-OHT response in human is heterogeneous, and only a subset of the population (30% to 40%) experiences IOP elevation as an adverse effect of long-term GC therapy.7, 8, 9^,12, 13, 14 Relatives of GC responders have a greater probability of being GC responders.10, 11, 12^,14 Patients with glaucoma and their first-degree relatives are also more likely to be GC responders.¹⁰ Taken together, there appears to be a strong genetic component influencing GC-OHT response in humans.^11,14 Despite this genetic association, studying the GC-OHT response in humans is logistically difficult. Several studies have attempted to identify risk alleles associated with GC-OHT in humans.15, 16, 17, 18 In addition, studies using in vitro and ex vivo models have obvious limitations.

The clinical presentation and disease pathology of GIG closely resembles that of POAG.^5,19,20 Increased stiffening of the aqueous humor outflow tissues occurs in both diseases. In GIG, this stiffening is associated with GC-induced deposition of excess extracellular matrix (ECM) and cytoskeletal remodeling,21, 22, 23, 24 with similar changes occurring in POAG. Given its iatrogenic nature, several animal models of GC-OHT have been developed to study disease mechanisms pertaining to both GIG and POAG.²⁵A reproducible model of GC-OHT in C57BL/6J (B6) mice using periocular delivery of the potent GC dexamethasone (DEX) to elevate IOP was previously developed.26, 27, 28 This and other similar models have been used to study glaucoma pathophysiology. Recently, our group observed a strain-specific variance in GC-OHT susceptibility. Mouse strain background can affect the severity of glaucomatous ocular phenotypes.²⁹ Several studies in mice have shown differences in baseline IOP between genetically distinct inbred mouse strains, indicating a role of genetics in baseline IOP regulation.^30,31

Clinical evidence shows that severity of GC response varies between individual humans. However, it remains unclear whether GC response rate in mice, across different genetic backgrounds, is homogeneous or similar to the heterogeneous response reported in humans. To address this knowledge gap, this study reports a strain survey involving phenotypic and transcriptomic comparisons of five genetically distinct inbred mouse strains: B6, C3H/HeJ, DBA/2J.Gpnmb⁺ (D2.Gpnmb⁺), 129P3/J, and BALB/cJ. In agreement with earlier studies,^30,32 differences in baseline IOPs were observed across several genetically distinct mouse strains. Furthermore, GC-OHT susceptibility was assessed by measuring IOP in mice receiving weekly periocular injections of DEX. Among the five strains analyzed, B6 and C3H/HeJ developed significant DEX-induced ocular hypertension (ie, DEX responder strains). D2.Gpnmb⁺, 129P3/J, and BALB/cJ mice did not develop GC-OHT (ie, nonresponder strains). The GC-OHT response across these five mouse strains is variable as it is clinically in humans (approximately 30% to 40% response rate observed in normal humans).7, 8, 9^,12 Transcriptional analysis was further performed in the trabecular meshwork (TM) tissue of responder and nonresponder mouse strains to assess mechanisms associated with GC-OHT susceptibility. There was significant down-regulation of immune-related genes in responder mouse strains compared with nonresponder mouse strains. Furthermore, pathway analysis of differentially expressed genes revealed significant enrichment of immune regulatory pathways in responder strains, confirming a stronger GC response in these strains.

Materials and Methods

Animal Husbandry

The present study examined both male and female mice. Similar findings have been reported in both sexes in terms of GC-OHT response.19, 20, 21 Genetically distinct parental mouse strains, C57BL/6J, DBA/2J.Gpnmb⁺, C3H/HeJ, 129P3/J, and BALB/cJ, used in this study were obtained from The Jackson Laboratory (Bar Harbor, ME). All mice were 3 to 4 months old at the start of experiments. All animal studies and care were performed in compliance with the Association for Research in Vision and Ophthalmology Statement of the Use of Animals in Ophthalmic and Vision Research and the University of North Texas Health Science Center Institutional Animal Care and Use Committee (IACUC) regulations (approved protocol: IACUC2023-0027). Mice were housed under controlled temperature (21°C to 26°C) and humidity (40% to 70%), with a 12-hour light/12-hour dark cycle (lights on at 7:00 am). Food and water were provided ad libitum. The number of animals used in each experiment is indicated in the corresponding figure legends.

Mouse Model of Dexamethasone-Induced Ocular Hypertension

DEX suspension was prepared by mixing 10 mg of micronized DEX (DE121; Spectrum Chemicals, New Brunswick, NJ) in 1 mL of vehicle suspension (Perrone Pharmacy, Fort Worth, TX). Ingredients and preparation of vehicle suspension were described previously.19, 20, 21 A uniform suspension with desired DEX particle size was achieved by mixing the suspension along with two stainless steel 5-mm beads (Qiagen, Valencia, CA) in a TissueLyser LT (Qiagen, Germantown, MD) for 10 minutes at 50 oscillations/second and further rotated overnight at 4°C until use. Isoflurane anesthetized mice [isoflurane (2.5%); oxygen (0.8 L/minute)] were weekly injected bilaterally with 20 μL/eye of either vehicle (both eyes) or freshly made DEX (ie, 200 μg) suspension (both eyes) via the periocular route using a 32-gauge needle attached to a 100-μL volume glass microsyringe (Hamilton Company, Reno, NV). There were separate groups of vehicle- and DEX-treated mice. The injection site for the right eye was the inferior fornix, and the left eye was the superior fornix of the mouse eye. The injection sites varied between eyes for ease of injection by a right-handed individual.

Mouse IOP Measurement

Baseline IOPs were measured on multiple strains of naïve animals as well as on dexamethasone- or vehicle-injected animals under isoflurane anesthetized conditions. For isoflurane anesthesia, the mouse was placed in the induction chamber and allowed to inhale a mixture of 2.5% isoflurane and 0.8 L/minute oxygen for 1.5 to 2 minutes, or until a deep plane of anesthesia was achieved, indicated by a lack of righting reflex and slowed breathing. Once the desired plane of anesthesia was achieved, the mouse was then immediately removed from the induction chamber and moved to an upright height adjustable stand, where the mouse received a maintenance dose of isoflurane via a nasal cone. IOPs were measured using a TonoLab rebound tonometer (ICare, Vantaa, Finland)³³ stably positioned upright using a metal clamp stand. This position allowed perpendicular placement of the probe to the central cornea without unwanted movement of animal during measurements, leading to improved accuracy. Each acceptable tonometer reading consisted of six simultaneous button presses involving six individual readings. The final, sixth button press displayed an average of four individual readings while excluding the highest and the lowest reading. The mean of seven such IOP readings per eye was recorded.

Mouse TM Isolation

Following 4 weeks of periocular dexamethasone or equivalent vehicle treatment, mice were euthanized for ocular enucleation and isolation of TM rim tissue. Each mouse was individually euthanized using an IACUC-approved protocol involving CO₂ asphyxiation followed by cervical dislocation. Immediately after a mouse was euthanized, ocular enucleation and TM tissue isolation were performed. Isolation of TM rim tissues was performed in aseptic conditions. Extraocular tissues were first removed with care using extrafine forceps and scissors. The globe was then punctured approximately 1 mm posterior to the limbus using an ocular stab blade. Using the stab puncture, the globe was hemisected along the posterior limbus and divided into the anterior segment (containing the TM, sclera, ciliary body, iris, lens, and cornea) and the posterior segment (containing the retina). The anterior segment was further cleaned by removing the iris and ciliary body using the extrafine forceps and scissors. The remaining anterior corneoscleral tissue was folded in a semicircle (like a taco). This allowed removal of the cornea using a 4-mm trephine, leaving the TM tissue and the underlying sclera intact. This TM rim tissue was then washed with 1× phosphate-buffered saline to remove unwanted cellular or tissue debris. The TM rims from both eyes were then pooled and snap frozen using liquid nitrogen and stored in a –80°C freezer for subsequent analysis. TM tissues from both eyes of each bilaterally injected animal were pooled to isolate RNA for RNA-sequencing analysis. Therefore, for the RNA-sequencing analysis, each n represents an individual animal containing the left and the right eye TMs pooled together for sequencing.

RNA-Sequencing and Pathway Analysis

RNA from mouse TM rim tissues was isolated using RNAeasy Micro kit (Qiagen, Germantown, MD) protocol, as recommended by the manufacturer. Fibrous TM tissues were homogenized by mechanical lysis in the supplied lysis buffer using a TissueLyser LT (Qiagen, Germantown, MD). Two 5-mm stainless steel mechanical beads were added to the round-bottom 2-mL microcentrifuge tube with frozen TM rim tissues and lysed in TissueLyser at 20 oscillations/second for three cycles of 1 minute, separated by 1 minute ice incubation after each cycle. RNA quality was determined using a Bioanalyzer (Agilent 2100; Agilent Technologies, Santa Clara, CA). Ribosomal RNA depletion and library preparation were performed using a Zymo Research Library Prep kit (Zymo Research, Irvine CA). The resulting library was deep sequenced using the NovaSeq 6K S4PE150 Platform (Microbiome Research Lab at the University of Texas Southwestern, Dallas, TX). Reads, quality control, mapping, and differential expression analysis were performed using the Qiagen CLC Genomics Workbench version 23 (Qiagen, Germantown, MD). Reads were mapped to Mus musculus (house mouse) genome assembly GRCm38. Select differentially expressed (DE) mRNAs were validated at the protein level. Pathway analysis was performed on DE genes using Qiagen's Ingenuity Pathway Analysis (IPA) version 01-22-01 (Qiagen, Germantown, MD).

Western Blot Analysis

TM tissues from mouse anterior segments were carefully dissected from enucleated eyes, as previously described, and were subsequently lysed in Pierce radioimmunoprecipitation assay lysis buffer (ThermoFisher, Waltham, MA) using sonication (three cycles of 25 Hz for 3 seconds, separated by ice incubation). Protein levels in the lysates were determined using Pierce Gold bicinchoninic acid protein estimation kit (ThermoFisher). Equal protein concentrations of lysates were loaded on denaturing 4% to 12% gradient polyacrylamide ready-made gels (NuPAGE Bis-Tris gels; Life Technologies, Carlsbad, CA) for PAGE. Proteins from the gels were electrophoretically transferred onto polyvinylidene difluoride membranes. Blots were blocked with 10% nonfat milk in solution (1× Tris-buffered saline + 0.1% Tween-20; Sigma Aldrich, St. Louis, MO) for 2 hours and then incubated for 2 hours or overnight with specific primary antibodies (Table 1) at 4°C on a rotating shaker at 200 revolutions per minute. The membranes were washed thrice with Tris-buffered saline + 0.1% Tween-20 and incubated with corresponding horseradish peroxidase–conjugated secondary antibody for 2 hours. The proteins were then visualized using SuperSignal West Femto Maximum Sensitivity detection reagent (Life Technologies). Densitometric analysis was performed on immunoblots using ImageJ software (ImageJ2 Fiji version 2.9.0/1.53t; NIH, Bethesda, MD; https://imagej.net/software/fiji/downloads).

Table 1.

List of Antibodies Used for Downstream Validation

Antibody name	Catalog no.	Dilution	Vendor
JPH2 polyclonal antibody	40-5300	1:500	ThermoFisher
Triadin monoclonal antibody (GE 4.90)	MA3-927	1:500	ThermoFisher
Myoglobin polyclonal antibody	PA5-78396	1:1000	ThermoFisher
SMARCA4/BRG1 monoclonal antibody (GT2712)	MA5-31550	1:500	ThermoFisher
Elastin polyclonal antibody	PA5-76676	1:1000	ThermoFisher
LOXL1 polyclonal antibody	LOXL-101AP	1:1000	ThermoFisher
MRTFA/MKL1 antibody	21166-1-AP	1:500	ThermoFisher
P38 MAPK antibody	8690P	1:1000	Cell Signaling Technology (Danvers, MA)
Phosphorylated P38 MAPK antibody	4511P	1:1000	Cell Signaling Technology
NICD antibody	2421S	1:500	Cell Signaling Technology
COL1 antibody	NB600-408	1:1000	NovusBio (Centennial, CO)

BRG1, Brahma-related gene 1; COL1, collagen1; JPH2, junctophilin2; LOXL1, lysyl oxidase like 1; MAPK, mitogen-activated protein kinase; MKL1, megakaryoblastic leukemia 1; MRTFA, myocardin-related transcription factor; NICD, Notch intracellular domain; SMARCA4, SWI/SNF-related, matrix-associated, actin regulator of chromatin, subfamily A, member 1.

Statistical Analysis

Statistical analysis was performed using GraphPad Prism 10 (GraphPad, San Diego, CA). Data are expressed as means ± SEM. Two-group comparisons were analyzed by unpaired t-test. Multiple comparisons were analyzed by two-way analysis of variance, followed by the Bonferroni post hoc test. Significance was designated at ∗P < 0.05, ∗∗P < 0.01, and ∗∗∗P < 0.001.

Study Approval

This study in animals was approved by the IACUC at the University of North Texas Health Science Center at Fort Worth (approved protocol: IACUC2023-0027).

Data Availability

RNA-sequencing FASTQ files are available via MIAME (minimal information about a microarray experiment) compliant public database sequence read archive by the Bioproject accession number PRJNA1338568 (https://www.ncbi.nlm.nih.gov/search/all/?term=PRJNA1338568).

Results

Mouse Genetic Background Influences Susceptibility to GC-OHT

Mouse models have been widely used to understand glaucomatous disease pathology and to discover new therapeutic targets. The B6 mouse model of GC-OHT was previously used to study glaucoma-related pathology in the trabecular meshwork and outflow pathway tissues.26, 27, 28 The B6 mouse strain is a widely used strain in glaucoma disease modeling, including GC-OHT. Given that all mice share the same genetic background within the strain, variation in baseline IOP is low and DEX treatment leads to significant GC-OHT with near-complete penetrance. Previous studies have shown differences in baseline IOPs between genetically distinct mouse strains.^30,32 These inter-strain differences in mouse baseline IOPs are considered analogous to the genetic variation seen between humans. Knowing that only 30% to 40% of the individuals undergoing long-term GC therapy develop GC-OHT, it was asked whether a similar heterogeneity in GC-OHT response is present in mice. To test the genetic association of GC-OHT in mice, five genetically distinct inbred mouse strains, which include B6, D2.Gpnmb⁺, 129P3/J, BALB/cJ, and C3H/HeJ mice, were used. Baseline IOP under isoflurane anesthesia was measured in all five strains. Comparison of baseline IOP between the widely used B6 parental strain and other genetically distinct mouse strains showed statistically significant differences (Supplemental Figure S1). Following measurement of baseline IOP, these strains received weekly periocular-subconjunctival injections of DEX (200 mg/eye) or vehicle, and daytime IOPs were measured weekly in the mornings (Figure 1A). As expected in the B6 strain, DEX injections led to IOP elevation of 4.8 mmHg over the vehicle-injected eyes (Figure 1B). The average peak IOP in DEX-injected eyes was 23.3 mmHg, whereas in vehicle-injected eyes, it remained at 18.3 mmHg. The C3H/HeJ mouse strain also developed OHT when challenged with DEX (Figure 1C). In these mice, DEX injections led to an IOP elevation of 5.0 mmHg, which was comparable to the OHT response observed in B6 mice. The average peak IOP in DEX-injected eyes of C3H/HeJ mice was 18.1 mmHg, whereas vehicle-injected eyes had an average IOP of 13.1 mmHg. To quantify and compare the DEX-induced IOP response in multiple strains, the average peak IOP (Figure 1G) and the area under the curve (Figure 1H) were used. The difference in peak IOP and IOP response (area under the curve) was determined to be statistically significant (P < 0.00001) in the two GC responder strains, B6 and C3H/HeJ (Figure 1, G and H). A statistically significant DEX-induced IOP elevation was not observed in D2.Gpnmb⁺, 129P3/J, and BALB/cJ strains (Figure 1, D–H), which were referred to as the DEX or GC nonresponder strains.

Figure 1.

Differential GC-induced ocular hypertension (GC-OHT) response in genetically distinct mouse strains. A: Diagram illustrating the experimental design for modeling GC-OHT using a potent synthetic steroid, dexamethasone (DEX). Baseline intraocular pressures (IOPs) for each strain were measured. Weekly bilateral injections of DEX (200 μg) or vehicle (Veh) were administered through the subconjunctival fornix and into the periocular area of the mouse eye to induce GC-OHT. IOP was measured weekly to determine GC-OHT response. B–F: DEX-mediated effect on IOP in genetically distinct mouse strains. Two-way analysis of variance (ANOVA) was used. P < 0.0001. G: Comparison of peak IOP attained in mouse strains after DEX or Veh treatment. One-way ANOVA was used. P < 0.0001. H: Analyzing the ΔIOP response over time [area under curve (AUC)] in different mouse strains to assess susceptibility to GC-OHT and identify GC response. Each n represents an individual eye. C57BL/6J (n = 38 Veh; n = 48 DEX), DBA/2J.Gpnmb⁺ (n = 10 Veh; n = 10 DEX), BALB/cJ (n = 14 Veh; n = 16 DEX), 129P3/J (n = 10 Veh; n = 16 DEX), C3H/HeJ (n = 10 Veh; n = 14 DEX). C57BL/6J strain is a known GC responder strain and, therefore, used as a positive control against other strains for technique validation. DEX responder strains (vehicle = black bars, DEX = pink bars); GC nonresponder strains (vehicle = teal bars, DEX = purple bars). One-way ANOVA was used. Data are represented as means ± SEM (B–H). ∗∗∗∗P < 0.0001. NS, not significant; PO, periocular; TM, trabecular meshwork.

Immune Regulatory Effect of Dexamethasone Is Stronger in GC Responders

Despite clinical association of GCs with the OHT phenotype and glaucomatous pathology, the mechanisms responsible for GC-induced pathogenic damage to the eye have yet to be identified. Given that only 30% to 40% of humans and 40% of mouse strains are susceptible to GC-OHT, it was asked which mechanisms make GC responders more susceptible to OHT and glaucomatous damage. In addition, whether protective mechanisms are at play in the GC nonresponders was examined. The physiological data comparing the IOP phenotype in multiple strains show that the mouse strains B6 and C3H/HeJ are GC responders, whereas the other three strains are nonresponders. Therefore, the mouse TM tissue transcriptomes were compared between responders and nonresponders to discover the mechanistic differences in gene expression. After IOP measurements, RNA was isolated from TM tissues, and RNA sequencing was performed (Supplemental Figure S2). The Venn diagram in Figure 2A shows the numbers of DEX-induced DE genes in each mouse strain. Several differentially expressed mRNAs were unique to the DEX responders. The GC nonresponder strain BALB/cJ demonstrated the highest number of differentially expressed genes among the five strains analyzed (Figure 2A). This was surprising, given that the phenotypic examination did not reveal a DEX-induced ocular hypertensive phenotype in these mice, so it was expected that their TM transcriptome would be similar to the other two GC nonresponder strains. The experiment was independently repeated in BALB/cJ mice, and DEX-induced changes in IOP were measured via daytime conscious IOP measurement. There was no significant difference in IOP of DEX-injected BALB/cJ mice compared with vehicle-injected mice, indicating that the BALB/cJ strain is a GC nonresponder strain (Supplemental Figure S3A). The analysis further revealed that a vehicle-treated BALB/cJ TM sample contained >60% rRNA contamination, which indicates an improper removal of rRNA before sequencing (Supplemental Figure S3B). This is perhaps why the BALB/cJ strain showed a higher level of differential expression (Figure 2A), because it was highly enriched in several pathways related to RNA metabolism and translation (Supplemental Figure S3C). To address this, all RNA genes were filtered out from the data set, and pathway analysis was performed on solely the protein coding genes (Supplemental Figure S4A). This still did not alleviate the highly variable gene expression in the BALB/cJ strain. The highest-ranking pathways enriched in BALB/cJ (Supplemental Figure S4B) were also uncommon to other GC responder and GC nonresponder strains (Supplemental Figure S4C), and as a result, this strain is considered as an outlier, and the transcriptomic changes in BALB/cJ were analyzed separately.

Figure 2.

Differential expression of RNAs in GC responder (C57BL/6J and C3H/HeJ) and GC nonresponder (D2.Gpnmb⁺, 129P3/J, and BALB/cJ) mouse strains. A: Venn diagram showing differentially expressed genes between each strain (P < 0.05). B–F: Volcano plot of top 20 most significant differentially expressed protein coding genes as a result of dexamethasone (DEX) treatment in each mouse strain. C57BL/6J [n = 5 vehicle (Veh); n = 5 DEX] (B), C3H/HeJ (n = 5 Veh; n = 6 DEX) (C), DBA/2J.Gpnmb⁺ (n = 5 Veh; n = 5 DEX) (D), 129P3/J (n = 4 Veh; n = 7 DEX) (E), BALB/cJ (n = 3 Veh; n = 5 DEX) (F). Each n represents a sample from an individual mouse containing pooled trabecular meshwork tissue from left and right eyes.

As expected, several immune-related genes were differentially expressed across all strains (Figure 2, B–F). However, the number of these differentially expressed immune-related mRNAs was higher in GC responders when compared with GC nonresponders, with significant reduction in overall expression levels. This signified a greater immune regulatory effect in GC responder strains.

Dexamethasone-Induced Pathways Enriched in GC Responders Differ from Nonresponder Mouse Strains

Supplemental Table S1 contains the commonly differentially expressed genes within the GC responder strains, and Supplemental Table S2 contains the differentially expressed genes within the GC nonresponder strains. Then, the DEX-induced differentially expressed mRNAs for each strain were analyzed using Qiagen's proprietary IPA software. All DE mRNAs below the P-value cutoff of 0.05 and beyond the fold change cutoff of ±1.5 were considered for this analysis. Figure 3 shows the interrelated pathways enriched in each strain grouped into broader categories and arranged according to the adjusted –log (P value) for each pathway category. Apart from determining the adjusted P value for each pathway, the IPA also assigns the pathway an activation Z score determined by the expression pattern of each differentially expressed mRNA belonging to the pathway and expressed in the data set. Highly significant enrichment of pathways related to extracellular matrix organization and immune regulation was observed in the GC responder strains (Figure 3, A and B) compared with nonresponder strains (Figure 3, C and D). Also, an overall negative Z score was observed for pathways related to cellular growth, development, and proliferation in the GC responder group compared with the nonresponder group. Surprisingly, the GC responders were also enriched in pathways related to muscle contraction, which was not the case in the GC nonresponder strains (Figure 3).

Figure 3.

Ingenuity Pathway Analysis–mediated enrichment of dexamethasone (DEX)–induced pathways in GC responder strains (A and B) and nonresponder (C and D) mouse strains. Related pathways are grouped into a single category. Blue = inhibition of pathway, orange = activation of pathway; size of circle corresponds to number of differentially expressed genes that overlap the pathway. Expression fold change of 1.5 was used as cutoff for pathway analysis. C57BL/6J [n = 5 vehicle (Veh); n = 5 DEX], C3H/HeJ (n = 5 Veh; n = 6 DEX), DBA/2J.Gpnmb⁺ (n = 5 Veh; n = 5 DEX), 129P3/J (n = 4 Veh; n = 7 DEX). Each n represents a sample from an individual mouse containing pooled trabecular meshwork tissue from left and right eyes.

The GC nonresponder strains, on the other hand, were highly enriched in pathways related to neuronal signaling, second messenger signaling, and neurotransmitter release, indicating robust cell signaling activity at the cellular level compared with GC responders (Figure 3, C and D). Pathways related to visual phototransduction were highly enriched in all the GC nonresponders, with a high positive activation Z score. Also, differential expression of photoreceptor marker gene rhodopsin (Rho) and retinal ganglion cell marker gene POU class IV homeobox 1 (POU4F1) was observed. This uncharacteristic enrichment of vision-related pathways and genes in TM samples indicated possible retinal contamination. The small tissue volume of mouse TM and the higher number of retinal cells compared with TM cells increase the likelihood of retinal cells being inadvertently transferred during the dissection process. To address this possibility, a photoreceptor signature from a large microarray study of the eyes of the BXD strain set was used.³⁴ The photoreceptor signature was made by taking significant correlates to Rho across the data set. Then, all genes that highly correlated with photoreceptor marker Rho (1515 genes) and retinal ganglion cell marker Pou4F1 (2001 genes) were removed from the data set, as previously defined. This included even the genes that were not differentially expressed.

IPA was used to identify pathways with the highest Z score enriched within the two GC responder strains C57BL/6J and C3H/HeJ (Figure 4). Similarly, two of the GC nonresponder strains, D2.Gpnmb⁺ and 129P3/J (Figure 4), were examined. Photoreceptor genes were removed from these data in Figure 4. As a control, Supplemental Figure S5 contains the same analysis without the removal of photoreceptor signature genes, showing similar results. The heat maps in Figure 4 show the top ranked pathways enriched in GC responder and GC nonresponder strains, determined using the activation Z scores. Pathways are ranked according to descending order of the adjusted –log (P values), with the highest ranked pathways at the top. In general, the two responder strains were similar to each other in their transcriptomic profiles and distinct from the nonresponder strains (Figure 4). The two nonresponder strains were also similar to one another. The GC responder strains (B6 and C3H/HeJ) were enriched in several immune regulatory pathways, signifying the potency of immunomodulatory action of GCs in these strains. At a systems level, DEX-mediated immunomodulatory effect in GC responder strains relies on down-regulation of cell surface and plasma membrane proteins, which include classes of integrins and immune-related cell surface receptors. Downstream kinase signaling was also reduced in GC responders, which was not observed in the nonresponder strains. In the TM, glucocorticoids are also known to induce fibrotic changes involving up-regulation of extracellular matrix proteins and dysregulated ECM homeostasis.^35,36 However, in the data set analyzing long-term effects of GC in mouse strains, antifibrotic changes were observed with down-regulation of several ECM genes, including collagen and elastin. Apart from the ECM genes, down-regulation of myosin heavy chains and genes regulating intracellular calcium levels was also observed, indicating dysregulation of contractile function. In the nonresponder strains (Figure 4), DEX-induced up-regulation of genes involved in membrane receptor signaling, indicating dynamic membrane activity in these cells, was observed. To this end, the nonresponder strains were also enriched in calcium and potassium signaling pathways, with a positive activation Z score. Furthermore, a subdued immunomodulatory effect of GC was observed in nonresponder mouse strains, as demonstrated by the lack of high-scoring immune-related pathways (Figure 4).

Figure 4.

Top dexamethasone (DEX)–induced canonical pathways enriched on the basis of activation Z score in GC responder (left) and nonresponder (right) mouse strains. The Z score indicates a predicted activation or inhibition of a pathway/gene, where a negative Z value connotates an overall pathway's inhibition, and a positive Z value connotates an overall pathway's activation. The heat map shows pathways enriched after removal of genes associated with photoreceptors and retinal ganglion cells. C57BL/6J [n = 5 vehicle (Veh); n = 5 DEX], C3H/HeJ (n = 5 Veh; n = 6 DEX), DBA/2J.Gpnmb⁺ (n = 5 Veh; n = 5 DEX), 129P3/J (n = 4 Veh; n = 7 DEX). Each n represents a sample from an individual mouse containing pooled trabecular meshwork tissue from left and right eyes. GPCR, G-protein–coupled receptor; NAFLD, nonalcoholic fatty liver disease; nNOS, neuronal nitric oxide synthase; PDGF, platelet-derived growth factor; TCR, T-cell receptor; Th1, type 1 helper T cell.

Identification of Pathways Attributed to Susceptibility and Protection against GC-OHT

To identify the genes and mechanisms responsible for GC-OHT susceptibility, the two GC responder strains were grouped together (Figure 5A), and 137 differentially expressed genes, which may be involved in the GC-OHT response, were identified. These 137 DE genes were analyzed using IPA, and a list of pathways common to the C57BL/6J and C3H/HeJ strains was identified (Supplemental Table S1 and Figure 5A). Likewise, the two GC nonresponder strains, D2.Gpnmb⁺ and 129P3/J, were grouped together, and 107 DE genes common to these strains were identified (Supplemental Table S2 and Figure 5B). Given that GC nonresponder strains do not develop DEX-induced IOP elevation, it was inferred that the DE genes common to these strains may play a role in protection against GC-OHT. Furthermore, these DE genes were analyzed using IPA, and DEX-induced pathways prevalent in GC nonresponder strains were identified (Figure 5B). Furthermore, IPA identifies the set of differentially regulated genes within the data set to determine the potential upstream regulators that contribute to the observed expression pattern in each group and predict the downstream biological function that may be affected. Figure 6 displays the graphical summary highlighting the relationship between differentially expressed genes within the group, their potential upstream regulators, and the related downstream cellular and disease-related functions that are affected in GC responder (Figure 6A) and GC nonresponder (Figure 6B) strains. A side-by-side comparison of canonical pathway Z scores (Figure 7A) and P values (Figure 7B) of GC responder strains and GC nonresponder strains was performed using the Fisher exact test. Most of these pathways are shared between the two GC responder strains and are dissimilar from the GC nonresponder strains. Similarly, most of the pathways are shared between the two GC nonresponder strains and are different from the responder strains. Most pathways in the responder strains are down-regulated, whereas many other pathways are up-regulated in the nonresponder strains. Notable among the down-regulated pathways of the responders are: phagosome formation, integrin signaling, collagen degradation, actin cytoskeletal signaling, and extracellular matrix organization (among others), which may play active roles in the development of DEX-OHT (Supplemental Figure S6).

Figure 5.

A and B: Canonical pathways associated with GC-induced ocular hypertension susceptibility and protection. A: Pathways contributing to intraocular pressure (IOP) elevation in susceptible GC responder mouse strains. Inset: Venn diagram showing 137 shared dexamethasone (DEX)–inducible differentially expressed (DE) mRNAs (P < 0.05) common to GC responder mouse strains. These 137 mRNAs are differentially expressed in response to DEX treatment and are common to the two responder mouse strains (C57BL/6J and C3H/HeJ). Main image: Diagram showing overlap between the canonical pathways enriched from Ingenuity Pathway Analysis (IPA) of 137 DE genes common to responder mouse strains. B: Pathways contributing to protection from IOP elevation in GC nonresponder mouse strains. Inset: Venn diagram showing 107 shared DEX-inducible differentially expressed mRNAs (P < 0.05) common to GC nonresponder mouse strains. These 107 mRNAs are differentially expressed in response to DEX treatment and are common to the two nonresponder mouse strains (D2.Gpnmb⁺ and 129P3/J). Main image: Diagram showing overlap between the canonical pathways enriched from IPA analysis of 107 DE genes common to nonresponder mouse strains. Blue = inhibited pathway with negative Z score, orange = activated pathway with positive Z score; size of circle corresponds to number of genes that overlap the pathway. C57BL/6J [n = 5 vehicle (Veh); n = 5 DEX], C3H/HeJ (n = 5 Veh; n = 6 DEX), DBA/2J.Gpnmb⁺ (n = 5 Veh; n = 5 DEX), 129P3/J (n = 4 Veh; n = 7 DEX). Each n represents a sample from an individual mouse containing pooled trabecular meshwork tissue from left and right eyes. GPCR, G-protein–coupled receptor; PKC, protein kinase C; TCR, T-cell receptor; Th2, type 2 helper T cell.

Figure 6.

Graphical summary representing the in silico Ingenuity Pathway Analysis performed on 137 genes common between two GC-responder strains (A) and 107 genes common between two nonresponder strains (B). Dex, dexamethasone; MRTFA, myocardin-related transcription factor; TNF, tumor necrosis factor; Veh, vehicle.

Figure 7.

Comparison of enriched pathways between GC responder and nonresponder mouse strains. A and B: Comparison of activation Z scores (A) and Benjamini-Hochberg–corrected P values (B) for dexamethasone (DEX)–induced canonical pathways between GC responder mouse strains (C57BL/6J and C3H/HeJ) and GC nonresponder mouse strains (D2.Gpnmb⁺ and 129P3/J). Pathways are vertically arranged by Z scores (A) and by P value (B), with a dot representing P value below the insignificance threshold (absolute value, 1.3). Expression fold change of 1.5 was used as cutoff for pathway analysis. C57BL/6J [n = 5 vehicle (Veh); n = 5 DEX], C3H/HeJ (n = 5 Veh; n = 6 DEX), DBA/2J.Gpnmb⁺ (n = 5 Veh; n = 5 DEX), 129P3/J (n = 4 Veh; n = 7 DEX). Each n represents a sample from an individual mouse containing pooled trabecular meshwork tissue from left and right eyes. GPCR, G-protein–coupled receptor; PKC, protein kinase C; TCR, T-cell receptor; Th2, type 2 helper T cell.

Molecular Mechanisms Involved in GC-OHT Pathology

IPA was used to analyze the DEX-induced DE genes within the data set and determine possible upstream regulators of DEX-induced IOP elevation. The IPA listed SMARCA4 (Brg1), a catalytic subunit of the SWItch/Sucrose Non-Fermentable (SWI/SNF) chromatin remodeling complex, as a putative upstream regulator within the data set, which was predicted to be inhibited (Figure 6A). TM tissue lysates were analyzed from C57BL/6J mice injected with DEX or vehicle for protein levels of SMARCA4, and reduced protein expression was observed in the DEX-treated group (Figure 8A and Supplemental Figures S7 and S8). Also, protein levels of myocardin-related transcription factor (MRTFA), another upstream regulator identified by IPA (Figure 6A) and a known binding partner of SMARCA4, were tested. Protein levels of MRTFA were reduced, but the trend was not statistically significant. A previous report in TM cells showed up-regulation of SMARCA4 and MRTFA proteins after 5 to 7 days of DEX treatment.³⁷ This was not observed in the present model, likely because of difference in treatment timeline. IPA also predicted P38 mitogen-activated protein kinase signaling to be inhibited. P38 mitogen-activated protein kinase phosphorylation was checked for in DEX-injected TM tissues, which was significantly reduced when compared with vehicle-injected controls (Figure 8B and Supplemental Figure S8). The rationale in selecting these upstream regulators for further validation is based on their associated Z scores, statistical significance, and relevance to prior publications in GC-OHT or GC pathology. Then, the IPA was used to unravel the causal relationships associated with the experimental data, which it does by expanding upstream analysis to include regulators that are not directly connected to targets in a data set but derived from relationships curated from the literature. The causal network analysis generates mechanistic hypotheses that may explain the expression changes observed in the data set. The causal network analysis results identified Notch signaling as the most prevalent network predicted within the GC responder data set. A comparison analysis between the two GC responder and the two GC nonresponder strains was performed, and all Notch-related causal networks were extracted for a side-by-side comparison between different strains. Notch-related networks solely enriched in GC responder strains were observed (Supplemental Figure S9). The causal network analysis result for the master regulator RBPJ/HAT1/Notch intracellular domain (NICD)/P300 complex (Figure 8C) was further tested at the protein level. The RBPJ/HAT1/NICD/P300 complex consists of three intermediate regulators that control the expression of the several downstream data set molecules. A significant reduction was observed in levels of cleaved NICD in the TM tissues of B6 mouse injected with DEX compared with that of control (Figure 8C and Supplemental Figure S8).

Figure 8.

Identification of putative molecular mechanisms involved in GC-induced ocular hypertension response. A: Left: Hub-and-spoke diagram of the upstream regulators SMARCA4 and myocardin-related transcription factor (MRTFA), with its target genes differentially expressed within the data set. Right: Quantification of protein levels of SMARCA4 and MRTFA in trabecular meshwork (TM) tissues of C57BL/6J mouse treated with dexamethasone (DEX) or vehicle (Veh); unpaired two-tailed t-test was used. B: Left: Diagram illustrating the P38 mitogen-activated protein kinase (MAPK) upstream regulator and its target genes. Right: Phosphorylation level of P38 MAPK against total P38 MAPK in TM tissues of DEX-treated C57BL/6J mice compared with vehicle; unpaired two-tailed t-test was used. C: Left: Causal network analysis revealed down-regulation of Notch signaling in response to DEX treatment in responders, thereby driving the immunomodulatory effect. Right: Levels of cleaved Notch were reduced in DEX-treated TM rim tissues compared with vehicle. Unpaired two-tailed t-test was used. Each n represents a sample from an individual mouse containing pooled TM tissue from left and right eyes. Data are represented as means ± SEM (A–C). n = 4 per group (A–C). ∗P < 0.05. GAPDH, glyceraldehyde-3-phosphate dehydrogenase; NS, not significant.

Apart from predicting upstream master regulators, the IPA also enables identification of putative downstream pathologic mechanisms that are at play as a result of the DE genes within the data set. Hypertrophy is one of the putative downstream pathologic mechanisms activated prominently in the responder strains. DEX has previously been associated with increased TM nuclear and cell size, but underlying molecular mechanisms have not been thoroughly investigated.³⁸ Figure 9A shows the hypertrophy as a pathologic mechanism surrounded by the associated DE genes curated from the literature. Supplemental Figure S10 shows the log fold change expression levels and P value of DE genes related to hypertrophy. Several DE genes from this association were validated at the protein level. Protein levels of elastin (ELN), junctophilin2 (JPH2), and triadin (TRDN) were significantly lower in DEX-treated TM samples (Figure 9, B–D, and Supplemental Figure S11). Also, reduced protein levels of collagen1 (COL1) and myoglobin (MB) were observed in response to DEX treatment, which did not meet the threshold of significance (Figure 9, E and F, and Supplemental Figure S11).

Figure 9.

Identification of hypertrophy as disease-associated biological function involved in GC-induced ocular hypertension response. A: Hub-and-spoke diagram of hypertrophy as a disease-associated biological function with its target genes differentially expressed within the data set. B–F: Protein levels of elastin (ELN; B), junctophilin2 (JPH2; C), triadin (TRDN; D), collagen1 (COL1; E), and myoglobin (MB; F) in trabecular meshwork (TM) rim tissues of C57BL/6J mice treated with dexamethasone (DEX) compared with vehicle (Veh)–treated control. Unpaired two-tailed t-test was used. Each n represents a sample from an individual mouse containing pooled TM tissue from left and right eyes. Data are represented as means ± SEM (B–F). n = 4 per group (B–F). ∗P < 0.05. GAPDH, glyceraldehyde-3-phosphate dehydrogenase; NS, not significant.

Discussion

GC-OHT is an adverse effect of prolonged GC therapy that can potentially lead to glaucomatous neurodegeneration. However, only 30% to 40% of individuals treated with GCs develop GC-OHT. Such heterogeneity in GC-OHT response has been known for decades, but the underlying mechanisms that make certain individuals susceptible to GC-related pathology remain elusive. Although genetic factors have been associated with GC-OHT response in humans,^10,11,14 limited numbers of positively identified patients with GC-OHT15, 16, 17, 18 make investigating the underlying pathologic mechanisms challenging. To overcome this, the previously developed mouse model of GC-OHT was used,26, 27, 28 which is also now widely adopted to investigate IOP-related pathology in the TM. This model demonstrates, for the first time, a heterogeneity in GC-OHT response in genetically distinct mouse strains. Among the five genetically distinct mouse strains tested for susceptibility to GC-OHT, two (C57BL/6J and C3H/HeJ) were found to develop DEX-OHT. This finding, in addition to the distinct baseline IOP observed in these strains, supports that genetics may play a role in determining GC-OHT response in mouse, as it has been reported in humans, This is the first study to compare GC responder and nonresponder mouse strains to selectively identify biomarkers and mechanistic pathways relevant to GC-OHT. To compare the genes and pathways that contribute to GC response or GC nonresponse, it was decided to solely focus on candidate genes unique to each data set. GCs, such as DEX, are potent anti-inflammatory agents that are involved in immune regulation. There was a strong DEX-mediated change in the transcriptome of GC responder mouse strains compared with nonresponders, which involved down-regulation of several immune-related genes. This was further confirmed by the comparison of enriched canonical pathways and their respective activity Z scores between the two groups (Figure 7). The GC responder strains were heavily enriched in immune regulatory pathways with negative Z scores. This indicated the presence of a DEX regulated immune microenvironment in the outflow pathway of GC responder mouse strains. Although such immune regulation is expected from GC treatment, the strength of the effect was higher only in the two GC responder strains (C57BL/6J and C3H/HeJ), which compelled us to consider that dysregulation in ocular immune environment may be a possible contributor to GC pathology. GC-OHT pathology in the TM, unlike POAG, is reversible in animals and in humans. This means that once GC therapy is ceased, the pathology subsides, and IOP returns to normal, which is not the case in POAG. From a functional perspective, TM cells possess several macrophage-like characteristics; they are capable of phagocytosis to maintain the health of the TM by clearing away cellular waste and debris in the aqueous humor, which is vital for proper fluid drainage and regulation of IOP.39, 40, 41, 42 Like macrophages, TM cells use mechanotransduction for sensing and responding to environmental cues, like elevated IOP, using mechanosensory cation channels (TRPV4, TRPM4, TREK1, PIEZO),43, 44, 45, 46, 47 cell-cell interacting proteins (cadherins and Notch receptors),^48,49 cell-ECM proteins (integrins), and calcium signaling.⁵⁰ These cells also possess immune-related markers (toll-like receptor 4) and secrete immune-regulatory factors (transforming growth factor-β) to modulate the tissue microenvironment.⁵¹ At the systems level, a marked reduction in immune-related surface markers and associated signaling mechanisms in GC responder strains was observed. This indicates either subtype-specific disruption of immune microenvironment or an overall reduction in immune cells within the outflow pathway. In either case, GC-mediated temporal regulation of immune cells in the TM microenvironment, which would hypothetically be reversible on discontinuation of GC therapy, and subsequent regeneration and reentry of immune cells into the outflow pathway would help revive tissue function. It is yet unclear whether depletion of immune cell populations or change in repertoire of immune cell subtypes has a role in IOP regulation. Recent reports in multiple systems have identified a variety of tissue-resident immune cells playing a role in tissue homeostasis.^52,53 In their recent preprint, Liu et al⁵⁴ demonstrated the presence of long-lived tissue-resident macrophages (embryonic origin) and short-lived monocyte-derived macrophages in the anterior chamber that may have a role in maintaining IOP homeostasis. This niche of resident immune cells is likely maintained by cues received from the surrounding cells from tissues of the proximal and distal outflow pathways in the form of growth factors and cytokines. If depleted due to stress, these resident immune cells can be replenished using self-regeneration or via replacement by monocyte-derived immune cells.^52,53 This regeneration or replacement of immune cells may contribute to the reversible nature of GC-OHT pathology after discontinuation of GC therapy. A more thorough characterization at the single-cell level is needed to confirm the role of GC-OHT in altering the resident immune cell repertoire.

From a systems-level perspective, it was observed that DEX is decreasing immune-related gene expression and regulating immune-associated pathways in GC responders. At the cellular level, the in silico analysis shows several different downstream processes being affected by DEX in the GC responder mouse strains. Down-regulation of ECM proteins and regulatory factors was seen. Cell-cell and cell-ECM signaling is altered, with levels of cell-surface receptors and G-protein–coupled receptor activity being reduced, likely contributing to reduced cell activation. Also, down-regulation of genes responsible for calcium homeostasis and actin cytoskeleton signaling, which would affect immune cell mobility within the outflow tissues, was observed. Interestingly, DEX-induced down-regulation of junctophilin2 (JPH2) and triadin (TRDN) was observed at the mRNA and protein levels (Figure 9), indicating a possible dysregulation of plasma membrane to endoplasmic reticulum junctional complexes and intracellular stored calcium release.55, 56, 57, 58 Downstream, the in silico analysis also showed enrichment of actin cytoskeleton and nuclear cytoskeleton signaling pathways, both with negative activation Z scores. These changes may have implication in regulating outflow facility. Actin-disrupting agents increase aqueous outflow, and the actin cytoskeleton is rearranged into cross-linked actin networks in TM cells and TM tissues that have been treated with DEX^23,59, 60, 61, 62 as well as TM cells and tissues derived from POAG donor eyes.²¹

Multiple processes, such as ECM deposition, focal adhesion, collagen, and transforming growth factor-β signaling, have been described to be involved in the pathogenesis of GC-OHT.^35,36 The GC responder mRNA data set did not completely align with previous reports. Unlike previous reports, no significant changes in fibrotic ECM markers were observed, except for the levels of collagen and elastin, which were reduced. However, changes in ECM organization, collagen degradation, actin cytoskeletal signaling, and integrin interactions, and decreased phagocytosis have all been previously reported as DEX-induced changes in the TM, often associated with decreased aqueous humor outflow.^59,63, 64, 65, 66, 67, 68, 69, 70 Results did not find that the earlier described DEX responder gene Angptl7⁷¹ to be differentially expressed in the data set, and it therefore appears that Angptl7 is not differentially expressed at this later time point. These differences from previous studies may be due to the prolonged timeline of DEX treatment in the present study (4 weeks versus 7 to 14 days). It is possible that this trend could be because of ongoing changes in response to GC treatment, with TM trying to balance the matrix metalloproteinase and tissue inhibitor of metalloproteinase ratios. Apart from the ECM genes, putative upstream regulators were also tested for the DE genes identified in the data set. SMARCA4, a core subunit of the SWI/SNF chromatin remodeling complex, is essential for development and homeostasis of various organs. SMARCA4 was identified as a strong upstream regulator influencing the data set of DE genes (Figure 6) in GC responder strains. Given its role in regulating chromatin accessibility and transcription, SMARCA4 may play an important role in orchestrating several processes associated with GC response. A previous study in primary TM cells reported GC-mediated up-regulation of SMARCA4 after 5 to 7 days of DEX treatment, which the authors concluded played a role in downstream actin cytoskeleton remodeling and cell adhesion.³⁷ However, the in silico analysis using IPA predicted SMARCA4 activity levels to be reduced after 4 weeks of DEX treatment, which was validated and confirmed at the protein level in the TM tissues of C57BL/6J mouse (Figure 8A). The in silico analysis also predicted reduced activity of MRTFA, a SMARCA4-interacting transcription factor, as a potential upstream regulator (Figure 8A).

Furthermore, Notch signaling, which involves cleavage of NICD and the formation of transcription activation complexes, was one of the most highly scored causal networks influencing the expression of gene in the GC responder data set. It was shown that NICD protein expression is attenuated in response to GC treatment in C57BL/6J mice. Although the role of Notch signaling in IOP homeostasis is unclear, a previous study in human primary TM cells has reported diminished expression of Notch signaling molecules with increase in cell-substrate stiffness resembling glaucomatous TM, an effect that also correlated with reduced phagocytosis.⁴⁸ GC treatment also increased TM substrate stiffness.³¹ Notch signaling plays a crucial role in tissue homeostasis by regulating activation, differentiation, proliferation, and survival of constituent cells.⁷² We agree with the sentiments expressed by the previous study⁴⁸ that more in-depth analysis of Notch signaling is required to understand its potential role in TM pathophysiology. It is interesting to note that a recent preprint on genotyping GC responders and nonresponders in humans has identified single-nucleotide polymorphisms within the Notch signaling pathway as risk alleles for the development of GC-OHT.⁷³

Although DEX-induced regulatory changes in GC nonresponders were diminished, comparable differential expression of genes was still observed. A total of 107 DE genes common between the two nonresponder strains (D2.Gpnmb⁺ and 129P3/J) were identified, and these may be responsible for protection against GC-OHT. IPA of this subset of genes had enriched pathways that were different from GC responder strains in function and activity, with most enriched pathways having positive activation Z scores. Given the lack of enrichment in immune-related pathways, it was asked if there are genetic differences between the glucocorticoid receptors of responder and nonresponder mouse strains. To answer this, the RNA-sequencing reads were mapped to mouse genome assembly 38, and low-frequency variant detection was performed for identifying variants that may possibly hinder interaction of DEX with its receptor. Fisher comparison was used between GC nonresponder strains and GC responder strains to detect variants that were exclusive to GC nonresponders. We did not find any variants for the GC binding region but instead found variants in the N-terminal domain (adjusted P < 0.01). Further analysis revealed that the variant is in the CAG repeat region of the N-terminal domain, which is susceptible to repeat expansion, and the length of which has been shown to vary between mouse strains. Previous studies in rodents have associated the number of CAG repeats with varying states of glucocorticoid receptor (GR) activation.74, 75, 76, 77, 78, 79 It is yet unknown whether the difference in CAG repeat length has a role in severity of GC-OHT response. This question is being pursued by using the BXD recombinant mouse model,⁸⁰ which represents inbred recombinant mouse strains derived from crossing the parental strains C57BL/6J (GC responder strain) and DBA/2J (GC nonresponder strain). Therefore, it was hypothesized that by determining which parental strain contributes the GR gene (Nr3c1) to the recombinant BXD stain, it can predict the likelihood of GC-OHT response in that BXD strain. The preliminary data involving these recombinant BXD mouse strains that carry the GR gene from either of the two parental strains (C57BL/6J or DBA/2J) do not appear to correlate with the responder status of the recombinant strain.

Study Limitations

There were no phenotypic differences between males and females of each strain in their IOP responses to DEX treatment, as previously reported.19, 20, 21 This work compared DEX-induced changes in TM gene expression between males and females of each strain. Interestingly, apparent differences in gene expression between males and females in response to DEX were seen; however, the small numbers of male and female mice within each group prevented the generation of statistically meaningful data. Additional studies with greater numbers of male and female mice are warranted to better document and examine this potential sexual dimorphism in TM transcription in response to DEX exposure.

This study differs in several ways from previous studies investigating GC-induced pathology in the TM. In the prolonged GC-OHT mouse model, no DEX-induced up-regulation of fibrotic mRNA markers in the TM was observed, as previously reported by us and others in human, mouse, and bovine tissues.^26,81, 82, 83, 84, 85 This may be due to the protracted timeline of the treatment, type of model, difference in method of tissue isolation, and marker detection. Usually the in vitro cell-based models and ex vivo human and bovine eye perfusion models are used for up to 14 days of DEX treatment. In the in vivo mouse model, we extended the treatment time to 4 weeks. This provides a long-term perspective on GC-induced disease pathology that closely resembles the clinical timeline.^8,13 In fact, the DEX treatment lasted for 4 weeks, at which point no up-regulation of many traditional fibrotic markers within the data set was observed. This incongruency with previous reports may be because GC-induced elevation of IOP in mouse and human reaches a peak and then plateaus after 2 to 3 weeks of DEX treatment. This stabilization in IOP may be a form of regulatory mechanism to limit ECM deposition in the TM and to facilitate establishment of a new state of pathologic tissue homeostasis. For example, reduced activity of SMARCA4 (Figure 8), which is a chromatin remodeling protein that is associated with fibrosis in TM cells, was observed.³⁷ It is also acknowledged that there are obvious drawbacks in using mouse eyes for this study. Mouse eyes are small and complicate the dissection of the TM, leading to inadvertent collection of the underlying cells from tissues of Schlemm's canal, outflow vessels, and sclera. Apart from the low tissue yield from mouse eyes, the small number of cells in the TM compared with other ocular tissues, like the retina, increases the chances of sample contamination with RNA from other cell types. Several predefined retinal ganglion cell– and photoreceptor-specific mRNAs differentially expressed in the data set were observed, which were removed before pathway analysis.⁸⁶ Most reported studies rely on isolated primary TM cells, which excludes the effect of other outflow pathway tissues working synchronously with the TM. Although tissues from ex vivo GC perfused human eyes contain multiple tissues/cell types, they are generically heterogeneous and are relatively less fresh (because of the 24- to 48-hour delay between time of death and collection) than an in vivo system like the mouse eye. Therefore, despite the drawbacks, the model provides a comprehensive systems-level image of the putative mechanisms underlying GC-OHT.

Conclusion

The present results show, for the first time, a phenotypic difference in the development of GC-OHT among genetically distinct mouse strains. It was demonstrated that two mouse strains (C57BL/6J and C3H/HeJ) of the five analyzed are susceptible to GCs and develop elevated IOP (GC-OHT) in response to prolonged treatment with the potent GC dexamethasone. This is significant because this difference in GC response has been previously reported in humans with a portion of the population being GC responders (approximately 40%) and the rest considered GC nonresponders (approximately 60%), and we see similar responder rates in mice, although additional strains need to be tested. We further leverage the mouse strain-specific phenotypic differences in GC response and the power of mouse transcriptomics to gain mechanistic insights into GC-OHT pathology. To our knowledge, this is also the first study to isolate and perform transcriptomic analysis on GC-treated mouse outflow tissues. The transcriptomics evidence suggests that GC responder strains share common pathways and mechanisms that contribute to development of GC-OHT, which are much different from the nonresponder strains. We further report a significant inhibition of immune/inflammatory mechanisms in GC responder strains compared with GC nonresponder strains. Given the clinical and mechanistic resemblance of GC-OHT to the more prevalent disease of POAG, this work is of broader impact, with potential in unravelling molecular mechanisms underlying POAG OHT. This study lays the foundation for future investigations to determine the role of GC-induced immunomodulation in OHT pathology in this form of iatrogenic secondary open-angle glaucoma.

Disclosure Statement

G.C.P. is an employee of Regeneron Pharmaceuticals.

Supplemental Data

Supplemental Figure S1

Baseline intraocular pressure (IOP) in genetically distinct mouse strains. IOP of each strain is compared with C57BL/6J, a known responder strain. One-way analysis of variance with Bonferroni post hoc analysis for multiple comparisons was used. Data are given as means ± SEM. n = 41 eyes (C57BL/6J); n = 30 eyes (C3H/HeJ); n = 20 eyes (DBA/2J.Gpnmb⁺ and BALB/cJ); n = 26 eyes (129P3/J). ∗∗∗∗P < 0.0001.

Supplemental Figure S2

Schematic illustrating trabecular meshwork (TM)/scleral rim tissue isolation and RNA-sequencing (RNASeq) analysis pipeline. DE, differentially expressed; QC, quality control.

Supplemental Figure S3

Despite being a phenotypic nonresponder, the RNA-sequencing profile of BALB/cJ mouse strain is different from any other mouse strain. This may be perhaps due to possible rRNA contamination. A: Conscious intraocular pressure (IOP) measurement performed in BALB/cJ mouse injected with dexamethasone (DEX; 10 mg/mL) or equivalent vehicle (Veh) for 4 weeks. B: Overlap of top DEX-induced enriched pathways in BALB/cJ mouse strain with the two remaining GC nonresponder mouse strains. DBA/2J.Gpnmb⁺ (n = 5 Veh; n = 5 DEX), 129P3/J (n = 4 Veh; n = 7 DEX), BALB/cJ (n = 3 Veh; n = 5 DEX). Each n represents a sample from an individual mouse containing pooled trabecular meshwork tissue from left and right eyes. C: Individual pathways enriched as a result of DEX treatment in BALB/cJ mouse strain. n = 10 eyes per group (A). GCPR, G-protein–coupled receptor; ns, not significant.

Supplemental Figure S4

Pathway analysis on differentially expressed (DE) gene data sets of mouse strains filtered to include only protein coding genes. A: Venn diagram of DE genes in each strain shows the highest number of dexamethasone (Dex)–induced DE genes in the BALB/cJ strain. B: Top 10 most significant Dex-induced pathways enriched in BALB/cJ strain. C: Graph comparing Z score (left) and P value (right) of the top 10 most significant pathways enriched in BALB/cJ strains with other strains analyzed. Top 10 BALB/cJ pathways were not enriched in any other strains. Expression fold change of 1.5 was used as cutoff for pathway analysis. C57BL/6J [n = 5 vehicle (Veh); n = 5 DEX], C3H/HeJ (n = 5 Veh; n = 6 DEX), DBA/2J.Gpnmb⁺ (n = 5 Veh; n = 5 DEX), 129P3/J (n = 4 Veh; n = 7 DEX), BALB/cJ (n = 3 Veh; n = 5 DEX). Each n represents a sample from an individual mouse containing pooled trabecular meshwork tissue from left and right eyes.

Supplemental Figure S5

Top dexamethasone (DEX)–induced canonical pathways enriched on the basis of activation Z score in GC responder (left) and nonresponder (right) mouse strains. The Z score indicates a predicted activation or inhibition of a pathway/gene, where a negative Z value connotates an overall pathway's inhibition, and a positive Z value connotates an overall pathway's activation. Heat map shows pathways enriched without removal of genes associated with photoreceptors and retinal ganglion cells. Expression fold change of 1.5 was used as cutoff for pathway analysis. C57BL/6J [n = 5 vehicle (Veh); n = 5 DEX], C3H/HeJ (n = 5 Veh; n = 6 DEX), DBA/2J.Gpnmb⁺ (n = 5 Veh; n = 5 DEX), 129P3/J (n = 4 Veh; n = 7 DEX). Each n represents a sample from an individual mouse containing pooled trabecular meshwork tissue from left and right eyes. GPCR, G-protein–coupled receptor; nNOS, neuronal nitric oxide synthase; PDGF, platelet-derived growth factor; TCR, T-cell receptor; Th1, type 1 helper T cell; Th2, type 2 helper T cell.

Supplemental Figure S6

Pathways relevant to glaucomatous pathology enriched by top differentially expressed (DE) genes overlapping in two GC responder strains (C57BL/6J and C3H/HeJ). Pathway analysis on 137 DE genes overlapping in the two GC responder strains was used to enrich key pathways relevant to glaucomatous disease with Z scores >2. C57BL/6J [n = 5 vehicle (Veh); n = 5 dexamethasone (DEX)], C3H/HeJ (n = 5 Veh; n = 6 DEX). Each n represents a sample from an individual mouse containing pooled trabecular meshwork tissue from left and right eyes.

Supplemental Figure S7

Hub-and-spokes network diagram representing the analyzed upstream regulators SMARCA4 and myocardin-related transcription factor (MRTFA) with their differentially expressed targets. Left: The pathways enriched by the differential expression (DE) of genes in the network are shown. Right: The cellular processes affected by the DE genes from the network are shown. The color blue indicates predicted inhibition, and orange indicates predicted activation, of upstream regulator, pathway, or cellular processes. Green and red are used for down-regulation and up-regulation of DE genes, respectively. Expression fold change of 1.5 was used as cutoff for pathway analysis. Upstream regulators were identifying by performing pathway analysis on 137 DE genes overlapping in the two GC responder strains. C57BL/6J [n = 5 vehicle (Veh); n = 5 dexamethasone (DEX)], C3H/HeJ (n = 5 Veh; n = 6 DEX). Each n represents a sample from an individual mouse containing pooled trabecular meshwork tissue from left and right eyes.

Supplemental Figure S8

Immunoblots representing protein expression of key upstream regulators relevant to GC-induced ocular hypertension response [SMARCA4, myocardin-related transcription factor (MRTFA), P38 mitogen-activated protein kinase (MAPK), and c-NOTCH) in C57BL/6J mouse. Each n represents a sample from an individual mouse containing pooled trabecular meshwork tissue from left and right eyes. n = 4 per group. DEX, dexamethasone; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; VEH, vehicle.

Supplemental Figure S9

Notch signaling is a highly enriched causal pathway in GC responder strains compared with GC nonresponder strains. A: Notch-related causal pathways were filtered and arranged according to the Z score, and their enrichment was compared with GC nonresponder strains. B: Notch-related causal pathways were filtered and arranged according to the P value, and their enrichment was compared with GC nonresponder strains. Expression fold change of 1.5 was used as cutoff for pathway analysis. Comparison analysis was performed on 137 differentially expressed (DE) genes overlapping in GC responder strains and 107 DE genes overlapping in GC nonresponder strains. C57BL/6J [n = 5 vehicle (Veh); n = 5 dexamethasone (DEX)], C3H/HeJ (n = 5 Veh; n = 6 DEX), DBA/2J.Gpnmb⁺ (n = 5 Veh; n = 5 DEX), 129P3/J (n = 4 Veh; n = 7 DEX). Each n represents a sample from an individual mouse containing pooled trabecular meshwork tissue from left and right eyes.

Supplemental Figure S10

Hypertrophy is one of the biological disease processes activated in GC responder strains but not in GC nonresponder strains. Left: Heat map representing log fold change (FC) of differentially expressed (DE) genes contributing to hypertrophy in GC responder strains. Right: Heat map representing P value of DE genes contributing to hypertrophy in GC responder strains. These genes were not differentially expressed at significance levels in GC nonresponder strains. Expression FC of 1.5 was used as cutoff for pathway analysis. Comparison of disease and biological processes was identified by performing pathway analysis on 137 DE genes overlapping in the two GC responder strains and 107 DE genes from GC nonresponder strains, followed by subsequent comparison analysis. C57BL/6J [n = 5 vehicle (Veh); n = 5 dexamethasone (DEX)], C3H/HeJ (n = 5 Veh; n = 6 DEX), DBA/2J.Gpnmb⁺ (n = 5 Veh; n = 5 DEX), 129P3/J (n = 4 Veh; n = 7 DEX). Each n represents a sample from an individual mouse containing pooled trabecular meshwork tissue from left and right eyes.

Supplemental Figure S11

Immunoblots representing protein expression of key differentially expressed genes relevant to hypertrophy, a cellular process relevant to GC-induced ocular hypertension pathology [junctophilin2 (JPH2), triadin (TRDN), elastin (ELN), collagen1 (COL1), and myoglobin (MB)] in C57BL/6J mouse. Two bands 5 kDa apart were observed for JPH2, the smaller of which is likely a cleaved form [N-terminal fragment of the JPH2 protein (JPH2-NT)]. Each n represents a sample from an individual mouse containing pooled trabecular meshwork tissue from left and right eyes. n = 4 per group. DEX, dexamethasone; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; VEH, vehicle.