Profiling IOP-Responsive Genes in the Trabecular Meshwork and Optic Nerve Head in a Rat Model of Controlled Elevation of Intraocular Pressure

Abstract

Purpose: The rat controlled elevation of intraocular pressure (CEI) model allows study of in vivo responses to short-term exposure to defined intraocular pressures (IOP). In this study, we used NanoString technology to investigate in vivo IOP-related gene responses in the trabecular meshwork (TM) and optic nerve head (ONH) simultaneously from the same animals.

Methods: Male and female rats (N = 35) were subjected to CEI for 8 hours at pressures simulating mean, daytime normotensive rat IOP (CEI-20), or 2.5× IOP (CEI-50). Naïve animals that received no anesthesia or surgical interventions served as controls. Immediately after CEI, TM and ONH tissues were dissected, RNA was isolated, and samples were analyzed with a NanoString panel containing 770 genes. Postprocessing, raw count data were uploaded to ROSALIND for differential gene expression analyses.

Results: For the TM, 45 IOP-related genes were significant in the CEI-50 versus CEI-20 and CEI-50 versus naïve comparisons, with 15 genes common to both comparisons. Bioinformatics analysis identified Notch and transforming growth factor beta (TGFβ) pathways to be the most up- and downregulated Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways, respectively. For ONH, 22 significantly differentially regulated genes were identified in the CEI-50 versus naïve comparison. Pathway analysis identified defense response and immune response as two significantly upregulated biological process pathways.

Conclusions: This study demonstrated the ability to assay short-term IOP-responsive genes in both TM and ONH tissues simultaneously. In the TM, downregulation of TGFβ pathway genes suggests that TM responses may reduce TGFβ-induced extracellular matrix synthesis. For ONH, the initial response to short-term elevated IOP may be protective.

Intraocular pressure (IOP) is regulated by the trabecular meshwork (TM), a tissue located in the angle of the anterior chamber. Normotensive pressures are required to maintain the shape of the eye globe and allow consistent focus of light on the retina. Elevated IOP occurs due to blockage of the aqueous outflow channels in the TM and is a major risk factor for glaucoma.¹ TM cells detect IOP elevation, and in response they alter gene transcription and induce other homeostatic mechanisms to adjust IOP back to normal ranges.^1–3 With IOP elevation, retinal ganglion cells die, likely due to initial injury of their axons at the level of the optic nerve head (ONH).^4,5 Identifying which genes respond to elevated IOP will be important to developing new IOP-lowering therapies that protect the TM and ONH. 

Different approaches have been used to study the effects of elevated IOP on TM gene expression. Static or cyclic mechanical stretch of cultured TM cells produces an approximately 10% stretch, mimicking what TM cells experience when IOP is elevated in vivo.^6–8 This model has been used in many microarray and quantitative polymerase chain reaction (qPCR) studies to evaluate which genes are likely to respond to elevated IOP. However, the main criticism of this technique is that the TM cells are taken out of their native environment and are grown on a plastic or silicone membrane. Ex vivo perfusion of anterior segments is also commonly used.^2,9–13 In this organ culture method, the whole globe is bisected; the iris, ciliary body, and lens are removed; and the resulting anterior segment, containing cornea, TM, and sclera, is clamped into a perfusion apparatus.⁹ Serum-free media are perfused at either 1× pressure (8 mmHg) or 2× pressure (16 mmHg) to mimic the elevated IOP experienced by glaucoma patients. Constant flow systems are also used, where IOP is continuously monitored. However, because the globe is dissected, anterior segments are not subject to normal IOP fluctuations and aqueous dynamics found in vivo. Animal models are necessary to understand better the role of cellular responses to IOP in the living eye. 

Numerous animal models of ocular hypertension are used in glaucoma research. These include the DBA/2J mouse and transgenic mouse models, such as mice overexpressing the human Y437H myocilin mutation.^14,15 IOP can also be elevated by intracameral injection of viral vectors, which transduce TM cells to overexpress bioactive molecules such as transforming growth factor beta (TGFβ),^16,17 gremlin,¹⁸ secreted frizzled-related protein-1 (SFRP1),¹⁹ Dickkopf-related protein-1 (DKK1),²⁰ or CD44.²¹ Other methods to increase IOP are injection of polystyrene microbeads into the anterior chamber, injecting hypertonic saline into Schlemm's canal via episcleral veins, or by laser photocoagulation of TM.^22–25 Most of these models were primarily developed to study mechanisms of pressure-induced optic nerve damage and generally compromise TM tissue structure and function. Thus, although they are useful for studying neuropathology, they cannot be used to study the effects of IOP on TM genes. For practical reasons, IOP is monitored sporadically in these models, generally only once or twice per week, which likely misses pressure spikes or IOP fluctuations that contribute to axonal injury. Thus, the actual IOP is not fully known, which can profoundly affect interpretation of results. 

An additional animal model of IOP, controlled elevation of IOP (CEI), has been recently described.^26,27 Ocular hypertension is achieved by cannulating the anterior chamber and raising a reservoir of fluid to deliver IOP at a known pressure for a desired duration. IOP is monitored in these anesthetized CEI rats, along with several physiologic parameters including blood pressure and oxygen saturation. We recently used this model to demonstrate the sequence of ONH cellular responses at 0, 1, 2, 3, 7, and 10 days following a single, 8-hour pressure stimulus,²⁷ and we identified several genes and pathways in common between CEI and those previously reported in two different chronic models of ocular hypertension.²⁷ For example, immediately following a short-term CEI, we found an upregulation in defense-related genes, several interleukin 6 (IL-6)-type cytokines, and the Jak-Stat pathway. This was followed by upregulation in cell proliferation (3–7 days following CEI), downregulation in axonal-related genes (3–10 days following CEI), and upregulation in immune-related genes (10 days following CEI). These cellular events induced by short-term IOP exposures are similar to those found in chronic models of hypertension. For example, upregulation in the Jak-Stat pathway and increased cellular proliferation were reported in the hypertonic saline chronic model of ocular hypertension, as well as in the feline congenital glaucoma model.^28,29 Thus, the CEI model encapsulates cellular events that occur in chronic glaucoma, but with the added benefit of being able to identify sequential events that lead to axonal degeneration. With its ability to control the duration and extent of IOP elevation, this model can ultimately be used to test the effect of inhibition or enhancement of specific genes and pathways on axonal survival. 

Although the CEI model was specifically designed to evaluate ONH cellular events, we have now adapted it to help us also understand TM responses to discrete pressure insults. From the perspective of TM research, this is the only model that allows study of an undamaged TM exposed to elevated IOP. The cannula tip resides anterior to the iris, and elevated IOP does not cause angle closure but induces a stretch/distortion of TM cells similar to that found in hypertensive glaucoma patients. Furthermore, we have recently described a method to microdissect TM from rat eyes and isolated sufficient RNA for gene expression analysis.³⁰ This produced relatively pure TM that was not substantially contaminated with ciliary body tissue, as defined by qPCR using biomarkers of the ciliary body.³⁰ With these advances, we are now able to investigate IOP-related gene response in the TM in a live rodent model for the first time. 

In this study, we performed CEI at 50 mmHg for 8 hours and used NanoString technology to compare TM gene expression changes to two control groups. We also analyzed gene expression changes in the ONH, providing the first study, to our knowledge, to simultaneously investigate IOP-related changes in the two tissues that are primarily affected by glaucoma within the same animal. 

Methods

Animals

A total of 35 retired Brown Norway breeder rats (Charles River Laboratories, Wilmington, MA, USA) were used for this study. They were 6 to 9 months old with an approximately equal number of male and female animals distributed among experimental groups. Rats were kept in a standard 12-hour light and 12-hour dark cycle with ad libitum access to food and water. Animal protocols were approved by the Institutional Animal Care and Use Committee at the Oregon Health & Science University, were performed in accordance with the National Research Council's Guide for the Care and Use of Laboratory Animals, and adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Rat CEI Model

CEI was performed in anesthetized rats as previously described, with constant monitoring of body temperature, O₂ saturation, heart rate, and blood pressure with a MouseOx oximeter (Starr Life Sciences, Oakmont, PA, USA) and a CODA monitor (Kent Scientific Corporation, Torrington, CT, USA).²⁶ Briefly, one eye of each rat was cannulated with a 1-inch polyurethane tube (0.010 inch OD/0.005 inch ID; Instech Laboratories, Plymouth Meeting, PA, USA). The tip of the cannula was positioned anterior to the iris and TM (Fig. 1A), under microscopic visualization, to ensure that the results would be associated with IOP-related stretching of the TM rather than angle closure and inflammatory responses from iris contact with the TM. The cannula was connected to larger Tygon tubing (Component Supply Company, Fort Meade, FL, USA) attached to a reservoir with sterile balanced salt solution (BSS Plus; Alcon Laboratories, Fort Worth, TX, USA) and a pressure transducer. The height of the reservoir determines the extent of pressure elevation, which was independently confirmed via a second cannula and transducer to accurately measure IOP in the eye.²⁶ 

Figure 1.

 

(A) Rat eye cannulation. Arrows point to the small tube inserted through the cornea, by which BSS Plus fluid is delivered at a precise IOP. (B) Graph shows the mean (and SD) IOP of each animal over the 8-hour duration of the experiment in the CEI-20 normotensive control group and the CEI-50 hypertensive experimental group. (C) Heatmap and dendrogram of ONH and TM genes from naïve samples showing differences in gene expression in the two tissues.

Previous work has shown that rat daytime mean IOP is ∼20 mmHg.^31–37 Therefore, we chose 50 mmHg (CEI-50), which is 2.5× daytime IOP, for the pressure challenge group (n = 12). This IOP is consistent with the magnitude of pressure challenge used in human ex vivo studies.^2,8,13 One control group received the normotensive daytime mean IOP of 20 mmHg (CEI-20; n = 11). An 8-hour CEI pressure challenge was chosen because >6 hours of pressure challenge is required to produce an IOP homeostatic response in ex vivo perfused human anterior segments.^2,3,12,38 Importantly, we have found that Brown Norway rats can tolerate 8 hours of general anesthesia, with no measurable impact on their physiology.²⁶ A naïve control group of rats (n = 12), with no anesthetic or surgical manipulations, was also included. During CEI, IOP was monitored every 30 minutes by a TONOLAB tonometer (iCare, Vantaa, Finland) to detect leakage or cannula failure. Mean TONOLAB IOPs during CEI-20 and CEI-50 are presented in Figure 1B. To prevent eyes from drying out, alternating applications of topical artificial tears or 0.5% proparacaine hydrochloride (Akorn, Lake Forest, IL, USA) were applied every 15 minutes to both eyes. Physiological parameters (O₂ saturation, heart rate, temperature, and blood pressure) were monitored throughout and remained stable throughout CEI. At the end of CEI, IOP was reduced to 20 mmHg for 5 minutes, the cannula was removed, and the needle track was self-sealed. Rats were then euthanized by an overdose of 5% isoflurane anesthesia followed by decapitation, and their eyes were immediately enucleated for TM and ONH microdissection. 

TM and ONH Microdissection and RNA Isolation

TM and ONH dissection occurred as described in previous publications.^27,30,39 Following dissection, the tissues were placed in individual tubes with 200 µL Invitrogen TRIzol (Thermo Fisher Scientific, Waltham, MA, USA), and RNA was isolated as described.³⁰ The NanoDrop One spectrophotometer (Thermo Fisher Scientific) was used to measure RNA concentration. RNA was isolated, and the mean ± SD RNA concentrations were determined to be 51 ± 10.07 ng/TM (range, 31.3–60.4) and 39.8 ±16.97 ng/ONH (range, 23.4–65.3). 

RNA Amplification and Reporter Probe Hybridization

Our samples had low RNA concentrations, so a low RNA input reagent kit was used to amplify the target following the manufacturer's protocol (NanoString Technologies, Seattle, WA, USA). Briefly, 4 µL of RNA was mixed with 1 µL RT Master Mix, placed in a thermocycler, and then converted to cDNA using the following protocol: 25°C for 10 minutes, 42°C for 60 minutes, 85°C for 5 minutes, and then placed on ice. A multiplexed target enrichment (MTE) was then performed for all samples as follows: Each cDNA sample was mixed with 1.5 µL 5X dT Amp Master Mix (NanoString Technologies) and 1 µL low input primers. Enrichment was performed with the following parameters: 95°C for 10 minutes, 18 cycles of 95°C for 15 seconds, and 60°C for 4 minutes. To ensure successful RNA amplification, 1 µL MTE reaction was quantitated by Qubit 3.0 Fluorometer analysis at the Oregon Health & Science University Gene Profiling Core Facility. The results showed an average of 149.2 ± 3.74 ng/µL, ranging from 122 to 163.5 ng/µL (n = 12). The remaining volume of MTE reactions was heated to 95°C for 2 minutes, snap-cooled on ice, and mixed with 8 µL of hybridization master mix containing 5 µL hybridization buffer and 3 µL reporter probe CodeSets. After mixing, 2 µL capture probe sets in hybridization buffer was added to each sample. The reporter CodeSets contained half of the target-specific sequence and six fluorescent labeled RNA segments (barcodes) of four different colors. Unique combinations of these barcodes allow up to 972 target genes to be assayed at one time. The capture probe set contained the other half of the target-specific sequence and was labeled with biotin. The capture and reporter probes were hybridized to gene targets in each of our TM and ONH samples. Tubes were placed in a preheated 65°C thermocycler for 24 hours. Hybridization reactions were then immediately transferred to the nCounter SPRINT Profiler instrument (NanoString Technologies). 

nCounter SPRINT Profiler Analysis

Because predesigned rat-specific NanoString panels are not available, the hybridized target–probe complexes were immobilized on Mus musculus PanCancer IO 360 cartridges (NanoString Technologies). These contain 770 gene targets related to immune response (eight pathways), tumor microenvironment (four pathways), and tumor biology (four pathways). Internal reference genes are also included. Each cartridge can process 12 samples at one time, and five cartridges were used for these studies (see Supplemental Table S1). TM samples were randomized over three of these cartridges, and ONH samples were randomized over an additional three cartridges (i.e., one cartridge contained both TM and ONH samples). Cartridges were placed in the nCounter SPRINT Profiler instrument, and hybridized samples were injected into the cartridge. The instrument follows a series of automated steps that include immobilization on a streptavidin-coated surface, flowing fluids over the surface to align the probes, and then data counting of the individual barcodes by a fluorescence microscope. Multiple fields of view are imaged, and the number of fluorescent barcodes is counted. One barcode is equivalent to one RNA molecule. 

ROSALIND and Bioinformatics Analysis

Raw data files (*.RCC) for each experiment were analyzed by ROSALIND bioinformatics software (San Diego, CA, USA). For TM, three comparisons were made: CEI-50 versus CEI-20, CEI-50 versus naïve, and CEI-20 versus naïve. For ONH, only CEI-50 versus naïve samples were compared, as we found no significant gene responses in the CEI-20 versus naïve comparison in our previous RNA sequencing (RNA-seq) study.²⁷ Samples were excluded if they had low image quality (<0.8) or high binding density (>1), or if housekeeping genes had zero counts. The final number of samples, derived from 27 rats, included in the bioinformatic analyses were as follows: for TM samples, n = 6 for CEI-50, n = 8 for CEI-20, and n = 4 for naïve; for ONH samples, n = 5 for CEI-50 and n = 7 for naïve (Supplementary Table S1 provides details of sample and sex distributions per group). Digital counts were normalized to 20 internal housekeeping genes on the panel. Background thresholds are calculated in the software as 97.5th percentile of the negative controls and are removed from further analysis if more than half the samples do not meet this threshold. If this occurs, the log₂ fold change (FC) is reported as 0 and the level of significance as 1. Differentially expressed genes were identified using a cutoff of ≥1.5 or ≤ −1.5 FC and were considered significant for P < 0.05. Because RNA concentrations were low for each individual sample, technical replicates could not be performed. To investigate pathways, differentially expressed gene lists were uploaded to ShinyGO 0.77 (South Dakota State University, Brookings, SD, USA). 

Quantitative PCR

TaqMan qPCR analysis was performed for a subset of the differentially expressed genes identified by NanoString technology. Briefly, 13 predesigned, rat-specific primers were purchased (Thermo Fisher Scientific; see Supplementary Table S8), as well as primers for 18S and glyceraldehyde 3-phosphate dehydrogenase (GAPDH), which were used as housekeeping genes for TM and ONH, respectively. Rat TM RNA was converted into cDNA using SuperScript III Reverse Transcriptase (Thermo Fisher Scientific). cDNA was pre-amplified with pooled primers for the target genes. PCR was amplified using the following conditions: 95°C for 10 minutes, 18 cycles of 95°C for 15 seconds, 60°C for 4 minutes, and finally 99°C for 10 minutes. The pre-amplified cDNA was diluted 1:8, and 4.5 µL of this diluted sample was added in each PCR reaction. Quantitative PCR was performed on a QuantStudio thermocycler (Applied Biosystems, Waltham, MA, USA). Results were analyzed by QuantStudio 3 software, using a geometric mean to calculate ΔΔCt, and results were exported to Excel (Microsoft, Redmond, WA, USA). Results were statistically analyzed using an unpaired t-test in Prism 10 (GraphPad, Boston, MA, USA). 

Results

NanoString technology was used to investigate in vivo IOP-related gene responses in the TM and ONH from the same animals. Initially, gene expression between TM and ONH samples from naïve animals were compared to validate that the technology was able to detect differences between the two naïve tissues. Not surprisingly, naïve TM samples clustered together and into a group distinct from naïve ONH samples (Fig. 1C). These findings support the idea that genes are differentially regulated in naïve TM and ONH tissues. We will now address significantly regulated genes in the TM and ONH following CEI. 

TM Gene Analyses by NanoString

Genes significantly altered in the CEI-50 versus CEI-20 comparison are shown in Table 1. Of 750 genes on the panel, four TM genes were significantly upregulated and 26 were significantly downregulated, as shown in the volcano plot in Figure 2A. The log₂ normalized expression for the four upregulated genes (Cblc, Jag2, Pdzk1ip1, Prom1) and four selected downregulated genes (Vegfa, Relb, Irf1, and Inhba) are also shown (Fig. 2B). Pathway analysis of all 30 significantly differentially regulated genes identified defense response, immune response, and response to other organism as the top three significantly regulated Gene Ontology (GO) biological process pathways in response to short-term IOP exposure (Fig. 2C). 

Table 1.

 

All TM Genes Significantly Differentially Expressed in CEI-50 Versus CEI-20 (FC >1.5 or <−1.5; P < 0.05)

 

Figure 2.

 

Bioinformatic analysis of NanoString data for the CEI-50 versus CEI-20 TM groups. (A) Volcano plot showing all genes on the cartridge. Significantly upregulated genes are shown in red, and downregulated genes are shown in blue. (B) Normalized expression levels of selected up- and downregulated genes in CEI-50 (n = 6; gray) and CEI-20 (n = 8; white) are shown. (C) Gene enrichment analysis (ShinyGO 0.77) shows the top 10 most significant gene ontology terms and pathways.

In the CEI-50 versus naïve comparison, there were 19 upregulated and 62 significantly downregulated TM genes, of which the top 20 are shown in Table 2 (a complete list of the 62 genes can be found in Supplemental Table S3). A volcano plot shows the distribution of all of the genes in the CEI-50 versus naïve comparison (Fig. 3A). The log₂ normalized expressions for four randomly selected upregulated (Mmp13, Spry4, Stc1, and Tnfaip6) and downregulated (Aldoa, Mfge8, Rock1, and Tgfbr1) genes are also shown (Fig. 3B). Pathway analysis of all 81 significantly differentially regulated genes identified regulation of cell population proliferation, cell population proliferation, and cell death as the top three of 10 significantly differentially regulated GO biological process pathways (Fig. 3C). Differentially regulated genes in the two hypertensive comparisons (CEI-50 vs. CEI-20; n = 30; CEI-50 vs. naïve, n = 81) were compiled into a single list (n = 111). Genes that appeared twice were removed, leaving behind a total of 95 unique genes. This list includes genes that are both IOP responsive and related to cannulation or anesthesia. To identify TM gene responses to elevated IOP alone, we next identified significantly differentially regulated genes in the CEI-20 versus naïve comparison. 

Table 2.

 

All Upregulated and 20 Downregulated TM Genes Significantly Differentially Expressed in CEI-50 Versus Naïve (FC >1.5 or <−1.5; P < 0.05)

 

Figure 3.

 

Bioinformatic analysis of NanoString data for the CEI-50 versus naïve TM groups. (A) Volcano plot showing all genes on the cartridge. Significantly upregulated genes are shown in red, and downregulated genes are shown in blue. (B) Normalized expression levels of selected up- and downregulated genes in CEI-50 (n = 6; gray) and naïve (n = 4; white) are shown. (C) Gene enrichment analysis (ShinyGO 0.77) shows the top 10 most significant gene ontology terms and pathways.

In the CEI-20 versus naïve comparison, there were 56 significantly differentially regulated TM genes, with 23 being upregulated and 33 downregulated (Fig. 4). Table 3 shows 20 upregulated and 20 downregulated genes. A complete list of the 56 genes can be found in Supplementary Table S4. These 56 genes were deemed to be anterior chamber cannulation related. Pathway analysis identified cell activation, leukocyte activation, immune system development, and inflammatory response genes as the top GO biological process pathways, a finding that is consistent with our hypothesis that this comparison would yield genes differentially regulated in response to cannulation or anesthesia. 

Figure 4.

 

Bioinformatic analysis of NanoString data for the CEI-20 versus naïve TM groups. (A) Volcano plot showing all genes on the cartridge. Significantly upregulated genes are shown in red, and downregulated genes are shown in blue. (B) Gene enrichment analysis (ShinyGO 0.77) shows the top 10 most significant gene ontology terms and pathways.

Table 3.

 

Twenty Upregulated and 20 Downregulated TM Genes Significantly Differentially Expressed in CEI-20 Versus Naïve (FC >1.5 or <−1.5; P < 0.05)

 

Table 4.

 

All IOP-Related TM Genes (n = 45). These TM Genes Were Significantly Differentially Expressed in CEI-50 Versus CEI-20 and CEI-50 Versus Naïve Comparisons, But Without Those Genes Common to the CEI-20 Versus Naïve Comparison

 

To identify IOP-responsive TM genes, we compared the 95 unique genes identified in the two hypertensive comparisons (CEI-50 vs. CEI-20 and CEI-50 vs. naïve) to the list of 56 cannulation-related genes (CEI-20 vs. naïve). This yielded a total of 151 genes, with 45 of these being only significant in the hypertensive comparisons. We therefore considered there to be 45 IOP-responsive genes (Table 4; Fig. 5). We analyzed the GO cellular component pathways with these 45 IOP-responsive genes. Of the top 10 pathways identified, genes related to the extracellular matrix (ECM; extracellular space and extracellular region) and plasma membrane (membrane raft and membrane microdomain) were significantly differentially regulated (Fig. 5A). Genes identified as being ECM related were Angptl4, Ccl5, Ctss, Cxcl3, Ifi35, Il1a, Inhba, Mfge8, Prom1, Serpinh1, and Wnt7b. Genes associated with the plasma membrane were Aldoa, Aplnr, Ccr9, Cd28, Ctnnb1, Fcgr2b, Ifitm2, Jag2, Jak2, Map3k12, Oas3, Tlr7, and Zap70. 

Figure 5.

 

Pathway analysis of 45 IOP-related TM genes. (A) ShinyGO 0.77 analysis shows the top 10 most significantly affected GO cellular component pathways. (B) KEGG pathway analysis of the four upregulated and 11 downregulated genes that were common to the CEI-50 versus CEI-20 and the CEI-50 versus naïve groups.

Of these 45 IOP-related genes, only 15 genes were differentially regulated in the same direction in both hypertensive comparisons. Four genes were upregulated (Cblc, Jag2, Pdxk1IP1, and Prom1), and 11 were downregulated (Aldoa, Angpl4, Atf3, Bcl2l1, Eif2b4, Ifitm2, Inhba, Irf1, Mfge8, Psmb9, and Tgfbr1) (Fig. 5B). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis showed that Notch signaling and ubiquitin-mediated proteolysis were affected by upregulated genes, and the TGFβ pathway was significantly affected by downregulated genes (Fig. 5B). Normalized counts for all 750 genes for each sample in the TM groups are found in Supplemental Table S5. 

ONH Gene Analysis by NanoString

We also analyzed ONH RNA from tissue dissected from the same group of hypertensive rats described above. In the CEI-50 versus naïve comparison, we identified 22 significantly differentially regulated genes (Table 5). Several of these genes (Angptl4, Cebpb, Edn1, Myd88, Nfil3, Slc2a1, and Srebf1) were differentially regulated in the same direction between this NanoString study and our published ONH RNA-seq study (Table 6).²⁷ Pathway analysis identified immune response, response to cytokine, and response to IL-1 as examples of significantly affected biological process pathways immediately after CEI (Fig. 6). 

Table 5.

 

All ONH Genes Differentially Expressed in CEI-50 Versus Naïve (FC >1.5 or <–1.5; P < 0.05)

 

Table 6.

 

ONH Genes Identified by Both NanoString and RNA-Seq Following an 8-Hour Pressure Exposure

 

Figure 6.

 

Bioinformatic analysis of NanoString data for the CEI-50 versus naïve ONH groups. (A) Volcano plot showing all genes on the cartridge. Significantly upregulated genes are shown in red, and downregulated genes are shown in blue. (B) Normalized expression levels of selected up- and downregulated genes in CEI-50 (n = 5; gray) and naïve (n = 7; white) are shown. (C) Gene enrichment analysis (ShinyGO 0.77) shows the top 10 most significant gene ontology terms and pathways.

Discussion

Elevated IOP is a known risk factor for glaucoma. IOP becomes elevated due to obstructed aqueous humor outflow pathway in the anterior chamber leading to optic nerve degeneration in the posterior pole. Yet, most animal models of glaucoma cause damage to the TM, causing IOP to become elevated, but they render the TM unusable for functional analyses. To overcome this, we used the CEI rat model to study short-term in vivo TM gene responses to elevated IOP because the cannulation tube does not damage the TM. This study identified endogenous TM IOP-responsive genes in vivo. Furthermore, we report TM and ONH gene responses to short-term IOP elevations from the same cohort of animals. 

In response to a pressure challenge of 50 mmHg (approximately 2.5× mean rat daytime IOP), we detected 45 IOP-responsive genes in the TM. GO pathway analysis highlighted 13 genes in ECM-related pathways that were significantly differentially regulated. This is consistent with ex vivo perfusion studies and cell mechanical stretch experiments, which have led to the hypothesis that ECM remodeling adjusts the TM outflow resistance as a pressure-induced homeostatic response.^1,2,6,8 

We compared the 45 IOP-related genes identified in this in vivo rat IOP study to those genes identified in pressure-challenged human perfusion culture studies^2,12,13 or with TM cells subject to mechanical stretch (Table 7).^6,7 Although there are differences in species studied (rat, human, pig), experimental design (pressure level and duration), and assay method (microarray, PCR arrays, NanoString), ANGPTL4 was detected among all studies. ANGPTL4 is a secreted, matricellular protein, which can bind cell surface integrins and fibronectin. Interestingly, interaction of ANGPTL4 with fibronectin delays its proteolytic degradation by matrix metalloproteinases (MMPs).⁴⁰ In the rat TM, Angptl4 was downregulated, implying that proteolytic sites in fibronectin may become exposed and more readily available for IOP-induced MMP degradation. In addition, studies with retinal vascular endothelial cells have implicated ANGPTL4 in activating the Rho/ROCK signaling pathway, which controls actin cytoskeleton contractility.⁴¹ This suggests that reduced Angptl4 gene expression would suppress Rho/ROCK activity, leading to relaxation of TM cells, increased aqueous outflow, and a reduction in IOP. 

Table 7.

 

Comparison of 45 Rat TM IOP-Related Genes Compared to Genes Identified as Being Significantly Altered in Pressure-Challenged Human Perfusion Cultures or TM Cell Mechanical Stretch

 

Several secreted molecules were significantly differentially regulated in the TM, including upregulation of Jag2 and downregulation of Mfge8 and Wnt7b. Jagged-2 (Jag2) is a ligand for the Notch family of receptors.⁴² Notch signaling is important for cellular communication between adjacent cells and may play important functional roles in mechanotransduction.⁴³ Upon binding of the JAG2 ligand, Notch receptor is proteolytically cleaved both extracellularly and intracellularly, with the intracellular fragment translocating to the nucleus to activate or inhibit transcription. Thus, Notch may act as a “mechanostat” that maintains the stasis of tissues subject to mechanoforces.⁴³ Upregulation of Jag2 is consistent with this paradigm. Milk fat globule–EGF factor 8 (Mfge8), also known as lactadherin, is a secreted glycoprotein that binds to integrin β3 and β5 chains and is dexamethasone responsive in TM cells.⁴⁴ It functions as an “eat me” signal, being upregulated on apoptotic or injured cells, thereby labeling them for phagocytosis. Mfge8 is also an abundant component of TM-derived exosomes.⁴⁵ Wnt7b is a short-range signaling protein involved in canonical Wnt signaling via Frizzled receptors. The canonical Wnt signaling pathway regulates IOP and ECM synthesis in the TM.^20,46 Downregulation of Wnt ligands, such as Wnt7b in this study, results in intracellular degradation of β-catenin, which is therefore unable to translocate to the nucleus and transcription of Wnt target genes is therefore suppressed.⁴⁷ Interestingly, the β-catenin (Ctnn1) gene is also downregulated in the IOP-challenged rat TM. 

Several transmembrane receptors were identified as significantly altered by pressure challenge, including Fcgr2b. Clustering of the Fcgr2b receptor induces phagocytosis and synthesis of inflammatory cytokines.⁴⁸ Downregulation of Fcgr2b in pressure-challenged TM suggests that these cellular functions are suppressed during IOP homeostasis. Only one of the 12 integrins on the panel was differentially regulated in the hypertensive group comparisons. This was Itgae (–4.75 FC; –2.25 log₂ FC in the CEI-50 vs. naïve comparison). Itga6 was significantly different in the CEI-20 versus naïve comparison. Integrins are transmembrane proteins that detect changes to the biomechanical environment, such as stretch/distortion resulting from elevated IOP.⁴⁹ Despite the lack of transcriptional changes at this single time point to a short-term IOP challenge, mechanical stretch induces integrin proteins to undergo conformational changes that lead to structural activation, which propagates biomechanical signals intracellularly. Thus, even if integrin gene expression is unchanged, it is likely that integrin-mediated signaling occurs in response to pressure challenge. 

Analysis of genes common to both hypertensive comparisons revealed that the TGFβ pathway was significantly differentially regulated. The TGFβ pathway has been closely studied in relation to glaucoma since the first report of increased levels of TGFβ2 in the aqueous humor of patients with primary open-angle glaucoma (POAG).⁵⁰ Increased TGFβ2 induces ECM protein synthesis, thereby contributing to the fibrotic-like matrix characteristic of the glaucomatous TM.⁵¹ Of the TGFβ pathway genes on the panel in this study, Pdzk1-interacting protein 1 (Pdzk1ip1) was upregulated, whereas TGFβ-receptor 1 (Tgfbr1) and inhibin beta A (Inhba) were significantly downregulated in both TM hypertensive comparisons. Pdzk1ip1 interacts with intracellular target of TGFβ signaling, SMAD4, thereby preventing translocation of SMADs from the cytosol to the nucleus. Thus, Pdzk1ip1 is an antagonist of TGFβ signaling. Inhba encodes activin beta, a ligand for the TGFβR1. When activin engages its receptor, SMAD2/3 proteins become phosphorylated, initiating SMAD4 translocation to the nucleus, which then modulates the expression of many target genes.⁵² Reduced synthesis of a TGFβ ligand and receptor, with concomitant upregulation of a TGFβ antagonist, suggests that a short-term elevated IOP prevents activation of this signaling pathway, thereby preventing ECM overproduction. Thus, our results suggest that an in vivo 8-hour pressure challenge induces protective pathways that induce homeostasis, which may initially limit TGFβ-induced TM fibrosis. However, this short-term 8-hour pressure likely does not model the chronic fibrosis that is observed in glaucoma. 

In the CEI-20 versus naïve comparison, 56 TM genes identified were deemed to be cannulation related. A previous study showed that transcorneal needle puncture significantly lowered IOP at 24 hours following puncture (Schuman D, et al., IOVS 2023;61:ARVO E-Abstract 2048). We would not be able to detect similar IOP decreases caused by needle puncture placement, as we precisely controlled the pressure challenge. Transcorneal needle puncture also significantly increased the macrophage density of distal outflow vessels (Schuman D, et al., IOVS 2023;61:ARVO E-Abstract 2048). Our cannulation procedure is likely to induce similar inflammation-related events in the anterior chamber and perhaps into the distal outflow pathway. When the 56 genes were analyzed, IL-17, TNFα, and NOD-like KEGG pathways and the GO biological process inflammatory response were significantly affected. These findings are consistent with inflammation. However, several of these TM cannulation-related genes can directly affect IOP. For example, Rock1 is a major effector of the actin cytoskeleton and a target for Rho kinase inhibitors, the newest class of glaucoma drugs.⁵³ Downregulation of Rock1 in this comparison is consistent with TM relaxation and increased aqueous outflow. Another gene affected in the CEI-20 versus naïve comparison was Vegfa, which was upregulated. Vascular endothelial growth factor A (VEGFA) protein is secreted by TM cells in response to mechanical stretch and application of exogenous VEGFA in enucleated mouse eyes increased outflow facility.⁵⁴ Thus, although we discounted these genes from our IOP-related genes, our somewhat simplistic conservative approach of subtracting one set of genes from another gene list may have also discarded some potentially IOP-responsive genes that play an IOP-regulating role in vivo. 

In a previous study, we exposed animals to a CEI of 20 or 60 mmHg for 8 hours and analyzed the ONH transcriptome by RNA-seq.²⁷ ONHs were collected at 0 hour and 1 to 10 days following CEI and compared to ONH from naïve animals. Following CEI of 20 mmHg, there were only 18 significantly differentially regulated ONH genes at 0 hour and none at the other time points. These 18 genes did not cluster into any significant gene category and were not significant following a CEI of 60 mmHg. By comparison, we identified 1354 significantly differentially regulated genes immediately following CEI (0 hr) of 60 mmHg, the same time point used in the current study. These findings support that ONH gene responses following a CEI of 60 mmHg are responses to elevated IOP and not a byproduct of anesthesia or anterior chamber cannulation performed during CEI. The top gene ontology category immediately following CEI (0 hr) of 60 mmHg was defense response. These 0-hour time point genes likely play an early role in glaucomatous neurodegeneration that may be important regulators of downstream cellular events. For example, we previously identified that cellular proliferation of astrocytes is an early cellular event following chronic and transient elevations of IOP.^28,39,55 From our RNA-seq study, we were able to identify several key genes that reflect signaling or regulation of ONH cellular proliferation at 0 hour. Importantly, the highest number of significantly differentially regulated ONH genes occurred immediately following CEI (0 hr), relative to the other time points analyzed (1–10 days). Therefore, we decided to re-analyze this early time point using NanoString technology but following a CEI of 50 mmHg, 10 mmHg lower that our previous RNA-seq study. 

In spite of the lower pressure exposure, we were able to identify 22 significantly differentially regulated ONH genes by NanoString. More than half of these 22 genes are involved with an immune response (C7, Cebpb, Clec4e, Edn1, Fasl, H2-DMa, Irf4, Myd88, Nfil3, Snca, Srebf1, Tnfrsf11a, and Tnfsf8). Despite the different levels of IOP, there were seven genes in common between the current NanoString data and our published RNA-seq findings (Table 6).²⁷ Three genes (Angptl4, Cebpb, and Slc2a1) have previously been reported to be upregulated and secreted by cortical astrocytes in Alzheimer's and Parkinson's disease.^56–60 In addition, endothelin-1 (Edn1) is significantly elevated in glaucomatous optic nerves, where it induces astrocyte proliferation,⁶¹ and negatively affects blood flow and retinal ganglion cell viability.^62,63 It is important to highlight that out of the 750 genes in the NanoString panel, only 109 of them were significant in our previous RNA-seq study. In spite of the lower number of significantly altered ONH genes identified by NanoString, the most prominent gene category remained related to an immune/defense response, as previously identified by our RNA-seq study. The lower number of significantly differentially regulated genes identified by NanoString would have been influenced by technological differences, the lower IOP level used in the current study, and/or selecting a panel that was more specific for TM cellular events. In this study, RNA was extracted from an ONH region that extended from Bruch's membrane to 400 µm posterior, a region that is enriched with Sox2⁺ nuclei, an astrocytic marker, suggesting that astrocytes are the predominant cell type in this ONH region.²⁸ Therefore, it is likely that these genes are also expressed/secreted by ONH astrocytes, as well. Although we did not directly demonstrate the involvement of astrocytes in this study, these findings are consistent with astrocytes playing a key role in the early response to elevated pressure in rats. Additional studies are needed to investigate whether similar responses occur in human tissue. 

A major advantage of our approach is the ability to assay gene responses in both TM and ONH simultaneously. As expected, the heatmap in Figure 1, generated from naïve datasets, shows that the two tissues express unique gene sets. This is not surprising because, during development, the retina and optic nerve arise from neuroectoderm but TM develops from mesenchymal neural crest cells.⁶⁴ Two genes, SerpinH1 and Angptl4, were differentially regulated by elevated IOP in both TM and ONH. SerpinH1 was downregulated in both tissues. SerpinH1 (heat shock protein 47) is an important chaperone for fibrillar collagen biosynthesis. It has a role in fibrosis in several tissues, and its downregulation is consistent with decreased expression of fibrotic-related genes.⁶⁵ On the other hand, Angptl4 was upregulated in ONH but downregulated in TM, suggesting that it may induce different downstream effectors in each tissue. 

This study compared directly the responses of the TM and ONH within the same animal to a given level of IOP. It is extremely interesting that, immediately following an 8-hour pressure exposure, the TM demonstrates a range of responses encompassing several different pathways, many of which are consistent with published in vitro work. The robust nature of this response so soon after a pressure challenge is entirely consistent with the notion that maintaining homeostasis is a fundamental function of the TM. This is important for overall eye stability, as well as possibly protecting the TM itself. Although relatively fewer differentially regulated genes were identified in the ONH, they represent pathways that largely mirror the defense response identified at 0 hour in our earlier RNA-seq study. As with the TM, it is reasonable to hypothesize that these early gene responses may also be protective. We believe that the CEI model, which reveals a sequence of cellular events following a single pressure challenge, can be used to test this hypothesis and allow us to better understand the relationship of these early ONH changes to axonal injury. 

There are several limitations of this study. First, the panel contained only 750 genes and did not include several known IOP-responsive genes important for remodeling TM ECM, such as MMP2, MMP3, and MMP14.^8,66 In addition, many genes identified in the rat ONH RNA-seq study were not on the NanoString panel, and the panel was designed to study cancer. Thus, our study is likely missing some key genes that respond to IOP in vivo. Also, the predesigned NanoString panel is designed for mouse genes and uses mouse primers to amplify the low-concentration RNA samples. Although there is generally good homology between rat and mouse DNA sequences, we found that 17% of these mouse genes were less than 80% identical to the target rat gene, possibly impacting the detection of rat RNA on the panel. Nevertheless, we detected significant changes to 45 IOP-related genes in the TM, and 22 differentially expressed genes in the ONH. TaqMan qPCR analysis using rat-specific primers showed results similar to those of the NanoString for several TM and ONH genes (Supplementary Fig. S1), increasing our confidence that the gene changes detected by NanoString are not artifactual. An additional limitation is that we used P values to identify significantly differentially regulated genes, not an adjusted P value. However, the agreement among our NanoString data, qPCR, and previous RNA-seq study provides additional evidence that our findings are true representations of TM and ONH responses to elevated IOP. An additional limitation is the use of a rodent model to study these changes. Although the rat TM has a single-lumen Schlemm's canal (SC), the TM only has a few beams compared to the extensive fenestrated beams of human TM. In the posterior chamber, rodents lack a lamina cribrosa and instead have a glia lamina that is enriched with astrocytes. However, this is the most cost-effective model to study IOP-related changes in vivo. Also, even though the cannula did not touch the TM, our study did detect differential expression of some immune genes suggesting occurrence of inflammatory events. However, there were no visible signs of inflammation during the 8-hour experimental time frame and some of the “inflammatory” genes are also expressed by the TM, so it is difficult to ascertain whether this is an actual immune response or due to the TM responding to the IOP challenge. Animals cannot tolerate longer anesthesia, so 8 hours was the maximum exposure duration that we assayed. Finally, another aspect to consider is that circadian and/or IOP fluctuations were not considered in this study. However, we incorporated fluctuations into a variation of the CEI model that we refer to as the pulse-train CEI (PT-CEI) model. In the PT-CEI model, IOP is elevated for 1 hour and then lowered for 5 minutes. This is repeated to cumulatively deliver 8 hours of elevated IOP. We published ONH gene expression responses using our standard CEI model³⁶ and the PT-CEI model²⁷ and showed that both approaches yield qualitatively similar ONH gene responses. 

In summary, this study is the first, to our knowledge, to report TM and ONH IOP-responsive genes in the same animals. Our results indicate that, in the TM, genes related to ECM synthesis and actin cytoskeleton are predominantly differentially regulated in vivo, confirming in vitro and ex vivo studies that have implicated these pathways in IOP homeostasis. This study additional confirmed the early upregulation in immune/defense response–related genes, as previously identified by our RNA-seq study. The CEI model provides an excellent animal model system to study short-term IOP elevation–responsive genes in both TM and ONH, tissues that are relevant to glaucoma pathology. 

Acknowledgments

Supported by grants from the National Eye Institute, National Institutes of Health (R21 EY033073, R01 EY019643, and R01 EY032590 to KEK; R01 EY010145-17S1 to DCL; R01 EY010145 to JCM; P30 EY010572 to Oregon Health & Science University), by the Malcolm M. Marquis, MD, Endowed Fund for Innovation, and by unrestricted departmental funding from Research to Prevent Blindness. 

Disclosure: D.C. Lozano, None; Y.-F. Yang, None; W.O. Cepurna, None; B.F. Smoody, None; E. Ing, None; J.C. Morrison, None; K.E. Keller, None