All experimental procedures comply with the National Health and Medical Research Council Australian Code of Practice for the care and use of animals for scientific purposes and the Association for Research in Vision and Ophthalmology's Statement for the Use of Animals in Ophthalmic and Vision Research. Animal Ethics approval was issued from the Howard Florey Institute Animal Experimentation Ethics Committee (13-068-UM).

Adult male C57BL/6J mice were used and experimentation commenced at 12 weeks of age. All mice were obtained from the Animal Resources Centre (Canning Vale, WA, Australia) and housed with unrestricted access to normal chow (Barastoc, Melbourne, VIC, Australia) and water at the Melbourne Brain Centre (Kenneth Myer Building, Parkville, VIC, Australia). Room temperature was maintained at 21°C and lighting on a 12-h light/12-h dark cycle (maximum 50 lux inside the cage).

Animals were anesthetized with isoflurane (IsoFlo, Abbott, North Chicago, IL) for suture application. All other ocular assessments were conducted after intra-peritoneal injection of ketamine:xylazine (80:10 mg/kg, intra-peritoneal, Troy Laboratory, Glendenning, NSW, Australia). Corneal anesthesia and mydriasis were achieved using topical Alcaine (0.5% proxymetacaine, Alcon Laboratories, Sydney, NSW, Australia) and Mydriacyl (0.5% tropicamide, Alcon Laboratories), respectively. Body temperature was maintained at 37.5 ± 0.5°C using a heating pad.

Forty mice were included in the 12 week trial, comprising an ocular hypertension (OHT) group (n = 23) and a sham group (suture removed after 2 days, n = 17). In all animals, one eye was randomly chosen to be the treated eye, and the contralateral eye served as a within animal untreated control.

The circumlimbal suture method was adapted from our previous work in rats (Liu et al., 2015, 2016). Briefly, under anesthesia (isoflurane: 4% induction and 1.5% maintenance at 1.5 L/min) a 10/0 nylon suture is threaded underneath the conjunctiva at 4–5 evenly spaced locations around the globe (anchor points) preventing suture movement. In order to improve long-term stability at each anchor point, the suture was threaded underneath the conjunctiva as far as possible, taking care to ensure that the eye was not perforated. In doing so, the suture was buried under the conjunctiva 1 mm behind the limbus, evenly applying pressure onto the eyeball (Figure 1 insert). To adjust tension around the eye a slipknot was used. Throughout the procedure the cornea was hydrated with normal saline. In the OHT group, the suture remained in place for the duration of the experiment (12 weeks).

The sham surgical procedure was similar to the OHT group; with the exception that the suture was removed at day 2 post-surgery. This time point was chosen for suture removal as a circumlimbal suture in rats (Liu et al., 2015) produced an initial moderate IOP spike lasting for up to 2 days. By removing the suture at day 2 (Sham group) we could evaluate the effect of the initial IOP spike. The difference between the OHT and sham groups should then reflect the effect of the period of chronic IOP elevation only, i.e., ocular hypertension.

IOP was monitored in each eye over the 12 weeks using a rebound tonometer (Tonolab, iCare, Helsinki, Finland), and the average of 10 readings was returned as the IOP at each time point. Baseline IOP was the average of daily readings made over the 5 days before surgery. At the time of surgery, IOP was measured immediately after tightening the suture (2 min), as well as at 1 and 3 h after surgery. IOP was monitored daily for the first 3 post-operative days and then weekly until week 12. Except for the 2-min time point immediately following surgery, IOP was measured in awake animals without topical anesthesia. To minimize diurnal fluctuation, IOP measurement was conducted at the same time of day and under the same light conditions (150 lux at the bench top).

In both OHT and sham groups, structural and functional assessments were performed 12 weeks after surgery. Retinal function and structure (OHT n = 23 and sham n = 17) were measured in both eyes of all animals, and subgroups were randomly allocated for histology (OHT n = 12, sham n = 8 animal) and PCR (OHT n = 9, sham n = 9 animals).

Dark-adapted full-field ERG was utilized to examine retinal function as described previously (Bui and Fortune, 2004; Lim et al., 2014; Liu et al., 2015). Mice were dark-adapted overnight (12 h) prior to recording (Behn et al., 2003), and light exposure was minimized during set-up to ensure maximal retinal sensitivity and optimize ganglion cell-specific scotopic threshold response (STR) measurement (Bui and Fortune, 2004). A pair of custom-made chlorided silver (99.9%, A&E Metal Merchants, Sydney, NSW, AU) active and ring-shaped reference electrodes, connected to platinum leads (F-E-30, Grass Telefactor, West Warwick, RI), were placed on the central cornea and sclera, respectively. A stainless steel needle electrode (F-E2-30, Grass Telefactor) was inserted subcutaneously into the tail to act as the ground. Responses were measured from both eyes at the same time.

Calibrated (IL1700, International Light Technologies, Peabody, MA) light stimuli were delivered across a range of energies from −6.35 to 2.07 log cd·s/m² by an array of 8 white light emitting diodes (LEDs, 8 Watt Luxeon LED, Philips Lumileds Lighting Company, San Jose, CA) and a dim LED (0.1 Watt Luxeon LED, Philips Lumileds Lighting Company) embedded inside a Ganzfeld sphere (Photometric Solutions International, Oakleigh, VIC, Australia). Signals were acquired with Scope™ software (ADInstruments Pty Ltd., Bella Vista, NSW, Australia) at a 4-kHz sampling rate with pre-amplifier (P511, Grass Telefactor, Natus, Pleasantown, CA) and hardware band-pass filter settings of 0.3–1000 Hz (−3 dB). Signals were digitized (ML785 Powerlab 8SP, ADInstruments) and saved for post-hoc processing.

The very first electronegative component of the ERG waveform (the a-wave) in response to the brightest stimuli was modeled using a delayed-Gaussian function to expose the P3 component (Lamb and Pugh, 1992; Hood and Birch, 1994). The maximal amplitude (designated RmP3) returned by this model reflects photoreceptor function (number, outer segment length, or density of non-specific cationic channels). The second and largest component of the ERG waveform (the b-wave) is the combination of the photoreceptoral P3 and a P2 component, which is known to reflect ON-bipolar cell activity. We subtract the P3 model from the raw ERG waveform to reveal the putative P2. The amplitude of each P2 waveform was plotted as a function of intensity and modeled with classical agonist-receptor characteristics using a hyperbolic curve (Fulton and Rushton, 1978). This model returns a maximum amplitude, which provides an index of bipolar cell integrity. The small high frequency wavelets that overlie the P2 leading edge, known as oscillatory potentials (OPs) and can be isolated from the P2 using a digital band-pass filter (50–180 Hz, −3 dB), to provide an index of amacrine cell derived inner retinal inhibitory pathways. Finally, the positive STR (pSTR), which can be measured near absolute light threshold is a good indicator of RGC function in rodents (Saszik et al., 2002; Bui and Fortune, 2004). To improve the signal to noise ratio, peak pSTR amplitude and its implicit time were averaged from parameters isolated from waveforms collected at −4.9, −5.01, and −5.31 log cd·s/m².

Immediately after ERG recordings spectral domain optical coherence tomography (SD-OCT, Envisu R2200 VHR, Bioptigen Inc., Durham, NC) was utilized to image structural changes in living eyes. Both the anterior segment and cross sections of the posterior retina through the center of the optic nerve were evaluated as described previously (Chinnery et al., 2015).

Anesthetized mice were placed on a rodent alignment stage and a drop of ocular lubricant (Systane, Alcon Laboratories) was applied to improve tear film optics for imaging. Once the central apex was aligned with the objective lens, volumetric 4 x 4 mm rectangular scans (1000 A-scans per B-scans, 100 horizontal B-scans evenly spaced in the vertical dimension) of the anterior segment were captured using an 18-mm telecentric lens (Envisu R2200, Bioptigen Inc.). From en face images pupil size (PS) was determined by the average of horizontal (180°) and vertical (90°) width. Anterior chamber depth (ACD) and trabecular meshwork-iris angle (TIA) were quantified in B-scans that were centered at the corneal apex.

Retinal OCT images were captured using a mouse objective (Envisu R2200, Bioptigen Inc.). A 2.5 × 2.5 mm volume scan was acquired with a depth of 1.7 mm (1000 A-scans per B-scans, 200 horizontal B-scans evenly spaced in the vertical dimension). Four B-scans that crossed the optic nerve head (ONH) were analyzed using FIJI software (https://fiji.sc/) available in the public domain. In each B-scan the inner limiting membrane, retinal nerve fiber layer (RNFL), inner plexiform layer and Bruch's membrane were manually segmented by a masked observer. From these segments, RNFL thickness and total retinal thickness (TRT) excluding the ONH region were averaged over a distance of 100–800 μm on either side of the ONH to return an average thickness for each layer. In addition, Bruch's membrane opening (BMO) width was measured along with the minimum rim thickness (MRT), which is the shortest distance from Bruch's membrane opening to the inner limiting membrane (surface of the retina) and comprises retinal nerve fibers as they turn to exit the eye. The final parameters for each eye represent the average of measurements made for both temporal and nasal retina across all 4 B-scans.

After acquiring retinal OCT images, retinal blood flow was determined using an annular Doppler blood flow B-scan (1000 A-scans, Bioptien Inc.), located 1 mm from the center of the ONH. Regions showing a significant Doppler shift are indicated by red and blue pixels. Red pixels represent arterial flow (toward the objective), whereas blue pixels indicate venous flow (away from the objective). Thus, counting the number of pixels provides a surrogate measure of blood flow and arterial or venular size. Both red and blue signals were extracted using a thresholding algorithm available with FIJI software (https://fiji.sc/) and summed to return blood flow in arbitrary units (au).

In addition, fundus photos were acquired using the Micron III small animal imaging platform (Phoenix Research Labs, Pleasanton, CA, USA) to ensure that the fundus was unaffected by 12 weeks of experimentation.

Following ERG and OCT measurement, in subset of animals from each group (OHT n = 12, and sham n = 8), eyes were removed and a mark was created on the superior limbus using a surgical cautery iron to retain orientation of the eye. Whole eyes were fixed in 4% paraformaldehyde in 0.1 M phosphate buffer (PB) for 1 h at room temperature. Retinae were dissected in flat mount, the superior quadrant marked with a radial cut, washed with PB, incubated with 20 mM EDTA for 1 h at 37°C then blocked in PB containing 3% BSA, 0.3% triton X-100 and 5% goat serum. Retinae were incubated overnight at 4°C with an antibody to RNA-binding protein with multiple splicing (RBPMS) to identify ganglion cells (1:1000, Invitrogen, Thermo Fisher Scientific, Carlsbad, USA). RBPMS has been shown to be a robust marker of RGCs in mice (Rodriguez et al., 2014). Retinae were washed in PB and subsequently incubated with goat anti-rabbit Alexa Fluor-647 (1:500; Invitrogen) for 1 h at room temperature. Flat mounts were counterstained (Hoechst; 1:1000, Roche Applied Science, Mannheim, Germany), washed and cover slipped photoreceptor side down.

Images were acquired using a confocal (Leica SP8, Leica Microsystems, Wetzlar, Germany) microscope (40x) from central (0.3 mm from optic nerve) and peripheral (0.3 mm from the retinal-ciliary body border) eccentricities in all four retinal quadrants (superior, temporal, nasal, and inferior). A total of eight confocal z-stacks were collected for each eye (Leica SP8, Leica Microsystems). Z-series through the ganglion cell layer were collected at 1 μm steps using a 512 × 512 scan format (290.62 × 290.62 μm). Ganglion cell densities were manually counted by a masked observer using FIJI software.

A subgroup of animals (OHT n = 9 and sham groups n = 9) were terminally anesthetized, and the eye was immediately enucleated, the retina isolated from the posterior eyecup, and stored at −80°C until use. Quantitative real-time polymerase chain reaction (qRT-PCR) was performed as described previously (Jobling et al., 2013; Ho et al., 2015). Briefly, total RNA was isolated from retina using commercial spin columns (RNeasy, Qiagen, Valencia, CA) incorporating DNase I treatment to remove genomic contamination. Retinal samples and dilutions of external gene-specific standards were reverse transcribed (250 ng, Tetro, Bioline, London, UK) and subsequently diluted to 5 ng/μL for quantitative PCR.

Using specific primers (Table 1) gene expression of several potentially pressure sensitive receptors was quantified on a Rotorgene V3000 (Corbett Research, Mortlake, NSW, Australia) using a SYBR-green based reaction mixture (SensiFast, Bioline, Alexandria, NSW, Australia). Negative controls and four-point standard curves were performed on every run. All standards and samples were amplified in triplicate and absolute gene copy number was calculated with reference to standard curves (Rotorgene V6.1 software; Corbett Research). Gene copy number was expressed normalized to the housekeeping gene hypoxanthine-guanine phosphoribosyltransferase (Hprt).

Statistical analysis was conducted using Prism 6 software (GraphPad Software Inc., USA). Group data are given as mean ± SEM. Results for treated eyes (OHT and sham) are expressed relative to their own untreated contralateral control eyes. Normality (Shapiro-Wilk test) and homogeneity of variance (Bartlett's test) were established before employing paired t-tests and ANOVA. Comparisons between treated and control eyes were undertaken using paired t-tests. Comparison of control groups (C-OHT vs. C-Sham) was undertaken using unpaired t-test. Two-way repeated measures ANOVA was used to compare IOP between groups (OHT, sham, C-OHT, C-Sham) across the 12-weeks of experimentation.

There was a moderate attrition rate associated with suture surgery. A 50% success rate (n = 40) resulted at day 2 from a total of 81 mice that underwent suture surgery. The most common cause of complication was hyphema (n = 29, 71%), followed by suture breakage, slippage or conjunctival tear (n = 12, 29%).

As previously reported for rats (Liu et al., 2015, 2016), the circumlimbal suture approach when applied to mouse eyes produced a similar pattern of IOP elevation. Over 5 days of baseline measurement, IOP in the ocular hypertension (OHT) group was stable at ~15 mmHg (Figure 1A). Suture application resulted in immediate IOP elevation, to 32.5 ± 2.5 mmHg at 2 min and 38.7 ± 2.2 mmHg at 1 h. On day 3 it returned to 26.0 ± 1.4 mmHg and by week 2 it had settled to 19.2 ± 1.1 mmHg (Figure 1A). From week 2 onwards IOP in the OHT eye was stable and remained ~5 mmHg higher than the untreated contralateral eye (19.5 ± 0.8 mmHg vs. 13.8 ± 0.5 mmHg, 41% increase, two-way RM ANOVA from weeks 2 to 12, interaction: p = 0.21, treatment: p < 0.01). Over these 12 weeks, IOP in the untreated contralateral eye did not change significantly (one-way ANOVA from weeks 2 to 12, p = 0.10).

Figure 1B shows that the sham group also had a significant IOP spike (35.3 ± 4.7 mmHg at 2 min, 38.0 ± 2.5 mmHg at 1 h), similar to that observed in the OHT group. Suture removal on day 2 returned IOP in sham eyes to normal levels (day 2, 17.5 ± 1.0 mmHg; day 3, 14.3 ± 0.6 mmHg) that were not significantly different from the untreated contralateral control eye for the remainder of the treatment period (15.2 ± 0.6 mmHg vs. 15.1 ± 0.5 mmHg, two-way ANOVA, interaction: p = 0.92, treatment: p = 0.97, time: p = 0.18). Additionally, IOP in sham control eyes was similar to that of the untreated control eyes of the OHT group (13.8 ± 0.2 mmHg vs. 14.4 ± 0.3 mmHg, p = 0.08).

Given the location of the suture it was important to rule out damage to the anterior segment given such damage is not found in primary open angle glaucoma. Figure 2 show representative images of pupil diameter acquired during anterior segment imaging for a control eye (C-OHT, Figures 2A,D), the contralateral treated eye (OHT, Figures 2B,E) and a sham operated eye (Sham, Figures 2C,F). After 12 weeks of chronic IOP there was no significant difference in pupil diameter between eyes (OHT 1926 ± 35 μm vs. C-OHT 1896 ± 22 μm, p = 0.39; Figure 2G). This was also the case for the sham group (sham and contralateral control eye, C-sham, p = 0.68). Additionally the anterior segment showed no significant change following suture application. Anterior chamber angle was not different (Figure 2H, p = 0.15) between OHT (Figure 2D, 35° ± 2°) and control eyes (Figure 2E, 31° ± 2°). Likewise, anterior chamber depth (ACD) was similar between the two eyes in the OHT group (Figure 2I, OHT 442.9 ± 12.7 μm vs. C-OHT 434 ± 5.5 μm, p = 0.46). There was no change in the anterior chamber angle (TIA) or anterior chamber depth (ACD) of the sham group (Figure 2H, TIA sham 32° ± 2° vs. TIA C-Sham 31° ± 2°, p = 0.70; Figure 2I, ACD sham 447 ± 8.3 μm vs. ACD C-Sham 435 ± 6.7 μm, p = 0.33).

To ensure that the suture did not induce retinal ischemia, we assessed perfusion of the retinal vasculature via fundus photos and Doppler OCT. Representative fundus images in Figure 3 show that the retina was perfused in an OHT eye as well as its contralateral control eye (Figures 3A,C). For the same eyes, retinal blood flow was also examined using a Doppler annular scan. Consistent with fundus photography, robust arterial (red signal) and venous (blue signal) flow could be detected in eyes after 12-weeks of chronic IOP (Figures 3B,D). Blood flow quantification revealed no difference in total (Figure 3E, OHT 3548 ± 241 vs. C-OHT 3360 ± 258, p = 0.24), arterial (p = 0.28) or venous blood flow (p = 0.30) between OHT and contralateral control eyes. Similarly, sham eyes were not different from their controls (total 2757 ± 128 vs. 2707 ± 177, p = 0.8; arterial, p = 0.86; venous, p = 0.78). Total retinal blood flow as well as arterial and venous blood flow was higher in the OHT group compared with the sham group (one-way ANOVA, p < 0.01).

Next we considered the effect of chronic IOP elevation on the function of the major retinal neuronal classes using the ERG by comparing OHT and sham waveforms in Figures 4A,B respectively. The waveforms elicited using the dimmest light levels reflect ganglion cell dominated responses, known as the positive scotopic threshold response (pSTR). At moderate flash energies between −4.87 up to −1.61 log cd·s/m² rod ON-bipolar cell responses dominate. At the brightest flashes, the response comprises both rod and cone contributions, with outer retinal function (photoreceptoral a-wave and ON-bipolar cell b-wave) being readily discernable. Although waveforms from sham treated and C-Sham eyes showed no obvious differences, there were differences evident in waveform obtained from the OHT group.

The P3 amplitude was significantly reduced in OHT eyes (−512 ± 26 μV vs. C-OHT −624 ± 31 μV, −15 ± 4%, p < 0.01; Figure 5A), whereas photoreceptor sensitivity was not significantly affected (OHT 2.65 ± 0.05 log m²cd⁻¹s⁻³ vs. C-OHT 2.63 ± 0.04 log m²cd⁻¹s⁻³, p = 0.81, Figure 5B). ON-bipolar cell P2 amplitude was also significantly reduced after 12 weeks of chronic IOP elevation (OHT 719 ± 41 μV vs. C-OHT 863 ± 49 μV, 15 ± 4%, p < 0.05; Figure 5C), whereas P2 sensitivity was not significantly affected (OHT 3.30 ± 0.11 log m²cd⁻¹s⁻³ vs. C-OHT 3.26 ± 0.12 log m²cd⁻¹s⁻³, p = 0.82, Figure 5D). OP amplitude (OHT, 104 ± 7 μV vs. C-OHT 114 ± 7 μV, p = 0.1; Figure 5E) was unaffected by 12 weeks of chronic IOP elevation, nd implicit time was significantly delayed (OHT, 43.7 ± 1.9 ms vs. C-OHT 42.3 ± 1.6 ms, p = 0.03; Figure 5F). The largest effect of chronic IOP elevation on ERG responses was found in the RGC pSTR, which was reduced by 19 ± 6% in OHT eyes compared to control eyes (OHT, 11.9 ± 1 μV vs. C-OHT 16 ± 1.6 μV, p < 0.01, Figure 5G). There was no difference in pSTR implicit time (Figure 5H). The sham group showed no statistical difference between sham eyes and control eyes for any of the measured parameters (all p > 0.05) and control eyes in the OHT group (C-OHT) were similar to the C-Sham eyes in all parameters (all p > 0.05), with the exception that OP implicit time was slower (p = 0.05).

Retinal structure characterized using SD-OCT is shown in Figure 6A (control eye) and Figure 6B (OHT eye) highlighting key parameters for our study. Group comparisons shows that the RNFL thickness (9.6 ± 0.5 μm vs. 12.3 ± 0.9 μm, p < 0.0.1; Figure 6C) and minimal rim thickness (MRT; 143.5 ± 2.5 μm vs. C-OHT 150.4 ± 2.3 μm, p = 0.04, Figure 6D) were significantly reduced in OHT eyes compared with control eyes. Neither total retinal thickness (TRT, Figure 6E), nor Bruch's membrane opening (BMO, Figure 6F) was significantly affected. There was no difference in any of the OCT parameters between treated and control eyes in the sham group (all p > 0.05), whilst comparisons of the untreated control eyes in the two groups (C-OHT vs. C-Sham) showed differences in TRT (P = 0.04) and MRT (P = 0.03).

In addition to the in vivo structural assessments, retinal immunohistochemistry and cell counts were used to estimate the effect of chronic IOP elevation on RGC density. RBPMS immunolabeling was assessed in contralateral control and OHT eyes (Figures 7A,B respectively) after 12-weeks of IOP elevation, and showed a reduction in RGC density in all quadrants (superior: 2944 ± 123 vs. 3531 ± 122 cells/mm², p < 0.01, temporal: 3354 ± 90 vs. 3678 ± 137 cells/mm², p < 0.01, inferior: 3397 ± 143 vs. 3772 ± 137 cells/mm², p = 0.02, nasal: 3221 ± 163 vs. 3592 ± 150 cells/mm², p < 0.05). In contrast, there was no difference in RGC density in sham operated eyes compared with their controls (all p > 0.05; Figures 7C–F). When comparing contralateral control eyes across treatment groups, there was no alteration in RGC density in the temporal, inferior and nasal quadrants (all p > 0.05), however C-OHT eyes had higher RGC density in the superior quadrant compared to C-sham eyes (difference of 565 ± 215 cells/mm², p = 0.02).

To understand the underlying mechanisms of IOP elevation induced retinal injury, gene expression of several pressure or cell membrane stretch sensitive channels was investigated, including components of the purinergic system (P2X7 and A3 receptors, EctoATPase-1), pannexin 1 (Panx1) and transient receptor potential cation channel subfamily V member 4 (Trpv4) channels. Figure 8 shows gene expression for P2x7, Trpv4, Panx1, Adora3, and Entpd1 (Panels A–E, respectively). There was no significant difference in the gene expression of P2x7 (Figure 8A), Trpv4 (Figure 8B), Panx1 (Figure 8C), or Adora3 (Figure 8D) between OHT and their control eyes (p > 0.05), while Entpd1 gene expression was significantly increased in OHT eyes (Figure 8E, 0.31 ± 0.01 vs. 0.28 ± 0.01, p = 0.02). The balance between the receptors, P2X7 and A3, has been suggested to be an indicator altered purinergic signaling, however, when the ratio of gene expression was investigated, no effect was evident (Figure 8F, p = 0.45). For all genes assessed, there was no difference between treated and control eyes in the sham group, nor between C-OHT and C-Sham eyes (all p > 0.05).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.