Differential Effects of Experimental Glaucoma on Intrinsically Photosensitive Retinal Ganglion Cells in Mice

Laser-photocoagulation to induce chronic OHT in mice

Adult (3–14 months) male and female wild-type (WT) C57BL/6 mice were used in this study. Ocular hypertension (OHT) was generated by laser photocoagulation of the trabecular meshwork of the eyes in 2–3 months old mice as described previously (H. Chen et al., 2015; Feng, Chen, Suyeoka, & Liu, 2013; Feng et al., 2016). In brief, mice were anesthetized by intraperitoneal injection of 80 mg/kg ketamine (Kataset, Zoetis; NADA #043–304) and 4 mg/kg xylazine (AnaSed, Akorn; NADA#139–236). Eyes were dilated with topical application of 2.5% phenylephrine hydrochloride ophthalmic solution (Akorn, NDC #17478-201-15) and 1% Tropicamide Ophthalmic solution (Akron, NDC # 17478-102-12), and the aqueous humor was aspirated from the anterior eye by a glass micropipette (World Precision Instruments Inc, Sarasota, FL). Laser illumination was delivered unilaterally or bilaterally (Figure. 1) by the PASCAL Synthesis 532 system (Topcon, Oakland, NJ). Six groups of five 200 μm (diameter) laser spots with zero spacing were applied perpendicular to the corneal limbus to circumscribe the trabecular meshwork at a laser power of 120 mW and exposure time of 20 ms. This procedure took 10–15 minutes for each eye. A drop of VIGAMOX (Moxifloxacine Hydrochloride Ophthalmic solution 0.5%) was applied to prevent infection, and lubricants were applied (Puralube Vet ointment, NDC # 17033-211-38) to prevent drying and formation of a cataract. The animals then were kept on a heating pad until fully responsive.

IOP measurement and the optomotor test

Intraocular pressure (IOP) was measured in awake mice using a Tonolab rebound tonometer (iCare, Raleigh, NC, USA) as previously described (Feng, Chen, et al., 2013; Thomson et al., 2020). In brief, each mouse was placed into a soft plastic cone holder and restrained on a platform. Averages from three sets of measurements per eye were recorded. All IOP measurements were obtained under ambient lighting between 10:00 AM and 1:30 PM.

Both visual acuity and contrast sensitivity were assessed by the optomotor test (Rangarajan et al., 2011; Thomson et al., 2020) (PhenoSysqOMR, PhenoSys GmbH, Berlin, Germany). In the optomotor test, the freely-moving animal was placed on a stationary round platform in the middle of four LCD monitors (Thomson et al., 2020). After 1–3 minutes of adaptation with a gray screen, a sine wave grating will be made to move across all 4 monitors. The animal was presented with moving gratings which were alternating clockwise and counter-clockwise for 10 seconds in each direction with varied spatial frequencies from 0.05 to 0.50 cycles/degree (c/d). The movement of the animal’s head in-concert with the drifting grating was scored as “seen”; the highest spatial frequency “seen” was defined as the animal’s visual acuity. The two eyes of individual mice were examined separately by reversing the drifting grating direction (i.e., a clockwise drifting grating was used to identify the visual function of the left eye, and a counterclockwise drifting grating for the right eye (H. Chen et al., 2015; Douglas et al., 2005; Feng, Chen, et al., 2013; Feng et al., 2016). Contrast sensitivity was measured at two preselected frequencies: 0.103 and 0.192 c/d, each with contrasts from 1 to 0.1, defined as the Michelson contrast from the screen’s luminance (maximum − minimum)/(maximum + minimum). The contrast threshold for each eye is defined as the lowest contrast that elicits responses at the pre-fixed frequency, and contrast sensitivity is the reciprocal of the threshold (Prusky, Alam, Beekman, & Douglas, 2004).

Immunohistochemistry and confocal imaging

Mice were euthanized with 600 mg/kg euthasol (Euthasol, Virbac ANADA, # 200–071) and perfused with 4% paraformaldehyde (PFA) (ChemCruz, sc-281692). Eye cups were dissected, post-fixed in PFA for 30 minutes, washed with phosphate buffered saline containing Triton-X detergent (PBST, 0.5% Triton X-100), and then blocked for 1 hour in blocking buffer (1% BSA and 10% normal donkey serum, 0.5% Triton X-100 (Sigma-Aldrich, St. Louis, MO, USA). Primary antibodies, diluted using blocking buffer, included a rabbit polyclonal anti-melanopsin (1:2000) (Panda et al., 2002), mouse anti-SMI-32 (Biolegend, 801702/01, 1:250, RRID: AB_2715852), mouse anti-Brn3a (Millipore, MAB1585, 1:125, RRID: AB_94166) and rabbit anti-rbpms (Abcam, ab194213, 1:500). Secondary antibodies, including donkey anti-mouse immunoglobulin G conjugated to Alexa Fluor 594 dye (Invitrogen A-21203, RRID: AB_141633), donkey anti-mouse immunoglobulin G conjugated to Alexa Fluor 647 dye (Invitrogen A-31571, RRID: AB_162542) and donkey anti-rabbit immunoglobulin G conjugated to Alexa Fluor 488 dye (Invitrogen A-21206, RRID: AB_2535792), were also diluted at 1:1000 in blocking buffer and incubated overnight at 4°C. After immunostaining, retinas were flat-mounted and cut into four quadrants: temporal, nasal, inferior and superior. For cryosection, fixed eye cups were cryoprotected overnight in 30% sucrose solution and embedded in OCT medium (Sakura Finetek, Torrance, CA, USA). Blocks were sectioned by cryostat at 15–20 μm and counterstained with DAPI (Vectashield H-1200, Vector Laboratories, Inc. Burlingame, CA, USA).

Confocal images were taken using a Zeiss LSM800 confocal microscope (Zeiss, Thornwood, NY). For flat-mounted retinas, Z-stack images covering the depth of the retina from the inner nuclear layer (INL) to the ganglion cell layer (GCL; approximately 50–80 μm) were acquired. Lower magnification (5×) pictures were captured for each leaflet of the retina using the tiling/stitch function in Zen (Zen 3.2; Oberkochen, Germany, http://www.zeiss.com/microscopy/en_us/products/microscope-software/zen.html#introduction, RRID: SCR_013672). For cell counting, individual images were captured at 20×, covering an area of 0.102 mm². To distinguish the M1 cells in the GCL from the displaced M1s in the INL, individual z-stack images covering an area of 0.0255 mm² were taken using a 40× water immersion objective. 3-D reconstruction and vertical view images were done and acquired by Imaris (Imaris 9.6, Bitplane Inc. Concord, MA, http://www.bitplane.com/imaris/imaris, RRID: SCR_007370).

Antibody characterization

Primary antibodies are listed in Table 1. The primary polyclonal antiserum (UF006) to melanopsin was raised in rabbit against a synthetic peptide consisting of the 15 N-terminal amino acids of mouse melanopsin (Genbank accession NP_038915) conjugated to keyhole limpet hemocyanin (KLH). Specificity of this antiserum has been confirmed in control studies showing a dose-dependent loss of immunoreactivity by pre-absorption with the immunogen and by the lack of immunoreactivity in the retinas of melanopsin-null mice (Panda et al., 2002). The lack of immunoreactivity in melanopsin knockout mice was confirmed using the immunoperoxidase method (Panda et al., 2002). We also validated the antibody using the Opn4^cre/cre mice (Ecker et al., 2010) crossed with a Synaptophysin-tdTomato cre reporter line, Ai34(RCL-Syp/tdT)-D (PubMed ID: J:170755). We found some of tdTomato+ cells are strongly SMI-32 positive and weakly melanopsin positive (data not shown), confirming their identity as M4 cells.

The mouse anti-Brn3a antibody (Millipore, #MAB1585, RRID: AB_94166) was generated against amino acids 186–224 of Brn-3a fused to the T7 gene 10 protein. According to the manufacturer, it does not recognize either Brn3b or Brn3c, nor stains tissues of Brn-3a knockout mice. This antibody labels RGCs in mouse retina (Voinescu, Kay, & Sanes, 2009), as seen in this study. We also co-labeled Brn3a with rbpms, another general RGC marker, and found that all Brn3a-positive cells were rbpms-positive (Figure 3a).

The mouse anti-SMI-32 antibody (Biolegend, #801702, RRID: AB_2715852) specifically recognizes a 200 kD non-phosphorylated epitope in neurofilament H, tested in various mammalian species. Specifically, in retina it labels some retinal ganglion cell bodies, dendrites and axons (Feng, Zhao, et al., 2013).

The rabbit anti-rbpms antibody (Abcam, ab194213) was generated against a synthetic peptide within rat rbpms (N-terminal) conjugated to KLH. In western blots, the rbpms antibody detects the expected 22 kDa band from rat heart lysate (manufacturer’s specification). In mouse retina, it specifically labels RGCs (Thomson et al., 2020).

Cell quantification

For general RGC and ipRGC cell counting, mouse retinas were immunostained with anti-rbpms, anti-Brn3a, or anti-melanopsin antibodies. For each retina, at least 24 en face z-stack images covering the depth from the GCL to the INL were captured (roughly 50–80 μm of total depth). Six images were acquired for each quadrant to ensure broad coverage of the entire retina (Fig. 1A). A rectangle covering no less than 0.025mm² area was randomly drawn on the images and Brn3a-, rbpms-, or melanopsin-positive cells within the rectangle were manually counted in Zen (Zen 3.2; Oberkochen, Germany).

To quantify M4 cells, mouse retinas were double immunostained with anti-SMI-32 and anti-melanopsin antibodies. Images were acquired as described above, and cells were classified as M4 when they met the following three criteria: 1) a cell soma size larger than 200 μm²2) faintly positive label for melanopsin, and 3) strongly positive label for SMI-32.

M1 in the GCL and displaced M1 cells in the INL were counted by taking images at the focal plane of the GCL and INL, respectively. For each retina, a total of 36 en face z-stack images were taken, with six images for each quadrant. Three-dimensional reconstruction was done using Imaris software (Imaris 9.6, Bitplane Inc. Concord, MA) and melanopsin-positive cells with dendrites reaching only the off sub-lamina of the IPL were counted as M1 cells. Although M6 ipRGCS also have dendrites reaching the off sub-lamina of the IPL, they are rarely labeled by immunofluorescence, even with amplification, and therefore are unlikely be misidentified as M1 (Quattrochi et al., 2019). Similarly, cells with at least one dendrite terminating in the on sub-lamina of IPL, potentially M3 cells, were not included. Cell density for each retina was calculated using the total number of cells counted divided by the total area.

Circadian behavioral analysis

Animals were housed in individual cages with running wheels in light tight boxes under a 12 hour light:dark (LD) schedule. Fluorescent lights (100 μW/cm²) were used for illumination, and food and water were provided ad libitum. Wheel running was monitored and analyzed with the ClockLab collection and analysis software suite (Actimetrics, Wilmette, IL). After at least 14 days of entrainment to the LD cycle, the dark period was advanced by 6 hours and wheel running was monitored for the following 15 days. The duration of re-entrainment was defined as the number of days required to shift activity onset by 6 hours followed by two consecutive days of activity onset or offset within this range. Activity onset was determined by the ClockLab software and adjusted manually to avoid masking effect. Entrainment was scored manually by a blind observer.

Circadian phase shifting using low intensity light was performed in a similar manner, except during the 12-hour light period, the intensity of the fluorescent source was adjusted to 1.5 μW/cm² by wrapping the fluorescent light bulbs with layers of pliable, semitranslucent neutral density filters.

Animals were also exposed to constant darkness (DD), and the free-running period was calculated according to the onset of activity across at least 10 days in constant darkness. The circadian period was calculated by 24 hours minus the slope of the least square fitted line of the onsets.

All results were expressed as the mean ± SEM. Non-parametric Mann-Whitney tests were performed to compare two groups of samples. All animal procedures were approved by the Institutional Animal Care and Use Committee at University of Virginia and conformed to the guidelines on Use of Animals from the National Institute of Health (NIH).

The general RGC population suffers a significant loss following chronic ocular hypertension.

We performed laser photocoagulation in adult C57BL/6 mice to induce OHT as described previously (H. Chen et al., 2015; Feng, Chen, et al., 2013; Feng et al., 2016) (Figure 1). For unilaterally lasered mice, the right eye was lasered and the left untreated eye of the same mouse was used as a control (Figure 1a). We also generated bilaterally lasered mice in order to induce chronic OHT in both eyes for circadian behavioral tests (Figure 1b). We tracked the intraocular pressures (IOP) every 2 weeks for the first two months and then every month up to one year, and simultaneously tracked visual acuity of each eye. The unilaterally lasered mice with treated eyes that exhibited consistently elevated IOP for more than three months and showed declining visual acuity were used for cell quantification (Figure 1a), while bilaterally lasered mice were used for circadian behavioral tests (Figure 1b).

The changes in IOP and visual acuity were tracked over time (Figure 2). For the first 3 months post-laser photocoagulation, the IOP of the lasered eye was elevated more than 25% than that of the control eye (Figure 2a). Three months post-laser photocoagulation, the IOP of the lasered eye of the unilaterally lasered mice was elevated to 19.15 ± 0.84 mmHg (n=7), significantly higher than that of the non-lasered control eyes (15.31± 0.29 mmHg, n=7, p<0.01, Mann-Whitney test, Figure 2a). The IOP of the lasered eye gradually dropped and was not significantly different from the control eyes by eight months post-laser surgery (controls: 14.6±0.5 mmHg, n=4; OHT: 18.2± 3.7 mmHg, n=6, p=0.2, Mann-Whitney test). The IOP was also elevated in each eye of the bilaterally lasered mice for more than three months (left panel, Figure 2b) and eventually dropped at eight months post-laser photocoagulation (left eye OHT: 15.94±1.18 mmHg, right eye OHT: 15.77±1.54 mmHg, p=0.87, Mann-Whitney test, Figure 2b).

Chronic IOP elevation induced a continuous decrease in visual acuity (Figure 2a). The acuity of the OHT eyes dropped from 0.35±0.02 c/d (n=11) to 0.28±0.02 c/d (n=6) at eight months post-laser photocoagulation. At the same time, the control eyes’ acuity remained within normal range at 0.39±0.01 c/d (n=4) eight months post-laser surgery, significantly higher than the OHT eyes’ (p=0.01, Mann-Whitney test) (H. Chen et al., 2015; Douglas et al., 2005; Feng, Chen, et al., 2013; Feng et al., 2016; Prusky et al., 2004).

Next, RGC loss following chronic IOP elevation was quantified. The general RGC population was immunolabeled using antibodies to Brn3a (Xiang et al., 1995) and rbpms (Rodriguez, de Sevilla Müller, & Brecha, 2014). Previous studies suggested that rbpms is expressed exclusively in most if not all RGCs, while Brn3a is expressed in some RGCs (Feng, Zhao, et al., 2013; Pan, Yang, Feng, & Gan, 2005; Rodriguez et al., 2014). We confirmed the labeling pattern in wild type mice (Figure 3a). It was estimated that about 59% of the cells in the GCL are displaced amacrines and 41% are RGCs (Jeon, Strettoi, & Masland, 1998). Indeed, we found that about 48.2% of all cells in the GCL were rbpms-positive RGCs; in addition, all Brn3a-positive cells were rbpms-positive, while about 87% of rbpms-positive cells were Brn3a-positive (Figrue 3a). At six months post-laser photocoagulation, the densities of Brn3a-positive cells were quantified and compared between OHT and control eyes (Figure 3b, c). We found that Brn3a-positive RGC density was significantly reduced by 25.6% in OHT eyes compared to controls (controls: 3859.7±156.8 /mm², n=5; OHT: 2871.7±94.4 /mm², n=3, p=0.04, Mann-Whitney test). At 8-months post-laser photocoagulation, rbpms-positive RGC density was also significantly reduced by 32.6% in OHT eyes compared to the controls (controls: 3942.5±100.4 /mm², n=3; OHT: 2658.1±216.0 /mm², n=4, p=0.03, Mann-Whitney test) (Figure 3d, e). Together our results showed that 6 to 8 months of chronic IOP elevation induced approximately 25–33% RGC loss, consistent with our previous findings (H. Chen et al., 2015; Feng, Zhao, et al., 2013).

Significant loss of ipRGCs occurs following chronic ocular hypertension.

Previously studies, including ours, showed that the RGC degeneration is cell type-specific (Duan et al., 2015; Feng, Zhao, et al., 2013). For instance, ipRGCs have been shown to be relatively resistant to chronic OHT or acute optic nerve damage (Duan et al., 2015; R. S. Li et al., 2006). Therefore, we next examined whether ipRGC survival differs from survival of the general RGC population in our chronic model of ocular hypertension. Melanopsin antisera was used to label the majority of ipRGCs, M1 through M4, and possibly some weakly melanopsin-positive M5 and M6 (Figure 3f, g) (Quattrochi et al., 2019; Stabio et al., 2018). We have included all cells immunolabeled with the anti-melanopsin antisera in the GCL and the INL for general ipRGC quantification. Surprisingly, at 8 to 9 months post-laser surgery, we observed a 16.5% reduction in the total ipRGC population in the OHT eyes compared to the control eyes (controls: 219.6±3.0 /mm², n=16; OHT: 183.4±4.5 /mm², n=11, p<0.001, Mann-Whitney test) (Figure 3f, h). We conclude that, like the general RGC population, the ipRGC population also suffers significant degeneration following the chronic hypertension insult, albeit to a lesser extent. Because the ipRGCs are not a homogeneous group and can be further classified into several subtypes, from M1 to M6, each with distinct morphology and function (Do, 2019), we next examined how different ipRGC subtypes survive in the context of chronic OHT.

M4 ipRGCs are susceptible to chronic IOP elevation.

Combining immunohistochemistry and three-dimensional reconstruction of ipRGC morphology, we characterized and quantified several ipRGC subtypes. To identify the M4 ipRGCs, we double-stained retinas with antibodies against melanopsin and the SMI-32 neurofilament epitope (Figure 4a). We compared the survival rate of M4 ipRGCs at eight to nine months post-laser photocoagulation. We found that M4 ipRGCs were significantly reduced by 25.7% (controls: 52.2±2.0 /mm², n=10; OHT: 38.8±2.7 /mm², n= 10, p<0.01, Mann-Whitney test) (Figure 4b), similar to the general ipRGC population.

Because M4 ipRGCs are thought to be involved in image-forming vision, especially in contrast sensitivity (Schmidt et al., 2014), we tested the contrast sensitivity of the OHT eyes at the spatial frequencies of 0.103 and 0.192 c/d. We found that the contrast sensitivities were significantly reduced (0.192: p<0.01, and 0.103: p<0.001 in Mann-Whitney test) in OHT eyes compared to control eyes (Figure 4c). At 0.103 c/d, the average contrast sensitivity for control eyes was 6.09±0.56, while for the OHT eyes it was 3.15±0.20 (n=12 in each group). At 0.192 c/d, the average contrast sensitivity for the control eyes was 5.50 ± 0.35, while in the OHT eyes it was only 3.63±0.31. This reduction in contrast sensitivity is consistent with the significant loss of general RGCs and M4 ipRGCs.

M1 ipRGCs are resistant to chronic IOP elevation.

Based on 3-D reconstruction of confocal z-stack images of ipRGCs, we identified common M1 ipRGCs with dendrites reaching the OFF sub-lamina of the IPL (Figure 5a). A small group of M1 cells also have perikarya located in the INL and thus are called displaced M1 ipRGCs (Figure 5b) (Berson et al., 2010; Duda et al., 2020; Ecker et al., 2010). We quantified the common M1 ipRGCs with cell bodies in the GCL and the displaced ipRGCs separately. Interestingly, we found that the density of M1 ipRGCs in the GCL did not change significantly at 8 months post-IOP elevation compared to controls (controls: 54.4±3.1 /mm², n=3; OHT: 56.4±1.1 /mm², n=4 p=0.86, Mann-Whitney test) (Figure 5c). Similarly, we found no significant change in the density of displaced M1 ipRGCs (OHT: 6.7±1 /mm², n=12; controls: 7.0±0.7 /mm², n=16, p=0.66, Mann-Whitney test) (Figure 5d). Taken together, the M1 ipRGC population suffered little, if any, cell loss following chronic ocular hypertension.

Circadian re-entrainment and circadian rhythmicity remain largely intact in OHT mice.

Given that M1 ipRGCs are known to innervate the master circadian clock in the hypothalamic suprachiasmatic nucleus (SCN) (Baver et al., 2008; S. K. Chen et al., 2011; Hattar, Liao, Takao, Berson, & Yau, 2002; Jain, Ravindran, & Dhingra, 2012), and M1 ipRGCs are preferentially spared from damage in OHT mice, we tested the hypothesis that the circadian behaviors may remain largely unaffected in glaucomatous mice. We performed bilateral laser photocoagulation on both eyes of C57BL/6 mice, and then tracked the IOP, visual acuity, and contrast sensitivity for up to one year. Mice that showed sustained IOP elevation for more than three months for both eyes (Figure 2b) were used for circadian studies. Out of more than 40 mice upon which we performed bilateral laser surgeries, six met this criterion for circadian testing (Figure 1b).

At seven months post-laser photocoagulation, six bilaterally lasered mice and six age-matched control mice were transferred to individual cages equipped with running wheels and maintained in a 12-hour light:dark (LD) schedule. The mice were allowed to adapt for 10 days in the running wheel cages. We observed clear onsets of activity immediately after lights-off in both control and bilaterally lasered mice (Figure 6a). We then advanced the dark phase of the LD cycle by six hours and monitored the wheel running behavior for 15 days. The entrainment was scored by a blind observer where re-entrainment was quantified as the number of days required to achieve a stable phase of entrainment to the advanced light cycle followed by two consecutive days of similar phase. Surprisingly, we found that the bilaterally lasered mice entrained significantly faster than control mice (controls: 8.8±0.5 days, n=6; OHT: 6.7±0.7 days, n=6, p=0.049, Mann-Whitney test) (Figure 6a, b). To confirm this result, we performed another round of phase advance and found similar results (controls: 7.5±0.4 days, n=6; OHT: 5.8±0.5 days, n=6, p=0.04, Mann-Whitney test).To test whether sensitivity of photoentrainment was altered, we performed a shift where the light intensity was reduced 60-fold from 100 μW/cm² to 1.5 μW/cm² (Figure 6c). We found no significant difference in the number of days required for re-entrainment between control and the bilaterally lasered mice (controls: 8.8±0.5 days, n=6; OHT: 7.8±0.7 days, n=6, p=0.22, Mann-Whitney test) (Figure 6d).

We next tested whether the difference observed in re-entrainment rate is caused by a difference in circadian periods. The same animals were maintained under constant darkness for at least 2 weeks and the stable free-running period was calculated across at least 10 days for each mouse (Figure 6e). We found no significant difference in circadian period between the control and bilaterally lasered mice (controls: 23.6±0.02 hours, n=6; OHT: 23.6±0.02 hours, n=6, p=0.81, Mann-Whitney test) (Figure 6f).

After the circadian experiments (about 11 months post-laser surgery), we examined the visual acuity (Figure 7a) and contrast sensitivity (Figure 7b) of the experimental and control animals. The OHT mice exhibited a significantly reduced contrast sensitivity: At 0.103 c/d, the contrast sensitivity of OHT mice (n=12 eyes) was 3.06±0.18, and that of controls (n=12 eyes) was 5.56±0.49 (p<0.001, Mann-Whitney test). At 0.192 c/d, the contrast sensitivity of OHT mice (n=12 eyes) was 3.28±0.18, and that of controls (n=12 eyes) was 5.53±0.69 (p<0.001, Mann-Whitney test). Similarly, the visual acuity of OHT mice was significantly reduced compared to controls, which remained high (controls: 0.36 c/d, n=12; OHT: 0.28±0.01 c/d, n=12; p<0.001, Mann-Whitney test).

Finally, we analyzed the overall morphology of the retinas (Figure 7c). The laminar structure was largely normal in OHT eyes. As expected, we observed a decrease in RGC density as determined by rbpms labeling in the OHT retina (Figure 7d). We quantified two OHT retinas and the overall density of the rbpms-positive cells in the OHT eyes was approximately 1674 cells/mm² at twelve months post laser photocoagulation, a 54.8% reduction compared to the control retina (3700 cells/mm², n=1). Similarly, the density of melanopsin-positive ipRGCs was also reduced (Figure 7e) to 182.5 cells/mm² (n=2), 13.5% lower than the control retina (211 cells/mm², n=1).

Characterization of general RGC population and ipRGC subtypes

Rbpms is a general RGC marker that is expressed by most, if not all, RGCs in mouse retina (Rodriguez et al., 2014). We found that the anti-rbpms antibody labeled about half of all cells in the GCL, s previously reported (Jeon et al., 1998). The other general RGC marker we used in this study, Brn3a, is present in 87% of all rbpms-positive RGCs (Figure 3a). We further assessed the general ipRGC population identified by the melanopsin antibody, which immunolabeled M1 through M4 (and possibly some M5) (Quattrochi et al., 2019; Stabio et al., 2018). We calculated a density of 219.6 cells/mm², which can be extrapolated to about 2800 cells per retina. This number agrees with a previous report of an estimated 2600 ipRGCs (Berson et al., 2010).

A total of six different ipRGC subtypes have been characterized, each with distinct morphology, physiology, and functions (Do, 2019; Duda et al., 2020). In the control retina, we calculated a density of 52.2 cells/mm² of M4 cells by anti-SMI32 and anti-melanopsin antibodies. In one study by Schmidt and colleagues (Schmidt et al., 2014), an average of 571.6 total M4 cells were found in mouse retinas, which translates to about 45 cells/mm². Similar results were confirmed by another study where a total of 529 M4 cells were observed (Sonoda, Okabe, & Schmidt, 2020). We further classified displaced M1 cells by their specific dendritic lamination in the OFF sublamina of the IPL, similar to the M1 ipRGCs with perikarya in the GCL. We have calculated densities of 54.4 cells/mm² and 6.7 cells/mm² for M1 cells and displaced M1 cells, respectively. The total number of M1 cells is 54.4+6.7 = 61.1 cells/ mm², which agrees with the previous finding by Berson and colleagues that the density of the total M1 cells was 63 cells/mm² (Berson et al., 2010). Together, our data show that these markers can assess accurately the general RGC population and ipRGC subtypes, thus enabling us to further examine their survivals in mice with chronic ocular hypertension.

Subtype-dependent ipRGC degeneration in OHT mice

Glaucoma is characterized by progressive RGC cell degeneration and death (H. Chen et al., 2015; Feng, Zhao, et al., 2013; Quigley, 2016). We observed a 25.6% cell loss of Brn3a-positive RGCs and a 32.6% reduction in rbpms-positive RGC density 6 to 8 months post-laser photocoagulation. Many studies have shown that ipRGCs are resistant to various disease insults (H. Chen, Liu, & Tian, 2014; Duan et al., 2015; Feng, Zhao, et al., 2013; Tran et al., 2019). For example, in a chronic OHT model in rats, no substantial ipRGCs loss or change in dendritic arbor structure could be found after 3 months of sustained elevation of IOP, while a significant number of superior colliculus projecting RGCs degenerated (R. S. Li et al., 2006). In a mouse optic nerve crush model, ipRGCs also have been found to preferentially survive relative to other RGC types (Duan et al., 2015; Tran et al., 2019). Our results show that with chronic IOP elevation the general ipRGCs (all M1-M4, possibly some M5 and M6) density was reduced 16.5%, which is less than the 30% loss of the general RGC population.

The survival of ipRGCs may not be uniform among different subtypes. At 14 days post-optic nerve crush injury, only a small number of M2 cells survived, but over 70% of M1 cells and over 80% of alpha RGCs labeled in a Kcng4-YFP transgenic line, including the M4 ipRGCs, survived (Duan et al., 2015). At 60-days post optic nerve transection in rats, M1 cells comprised about 80% of all surviving RGCs (Perez de Sevilla Muller, Sargoy, Rodriguez, & Brecha, 2014). We also found no significant loss in M1 cells and displaced M1 cells in OHT mice, however, M4 cells suffered 25.7% loss. Other ipRGC subtypes, including M2s and possibly some M5 and M6s, also suffered about 23% cell loss (table1). This inconsistency may be explained by the different methods applied for RGC subtype categorization, as well as the differential survival of each subtype in different animal models. For example, Tran et al (2019) quantified RGCs from single cell samples dissociated from whole retinas, which provides a comprehensive profile of RGC survival, but may also cause disproportionate representation of RGC types (Tran et al., 2019). Additionally, optic nerve crush injury or transection is an acute model of RGC damage, analyzed within a couple weeks of crush, while chronic OHT induces slow and progressive RGC loss over months. Therefore, RGCs may respond differently to different types of disease insult (McKinnon, Schlamp, & Nickells, 2009). Interestingly, melanopsin overexpression in RGCs promoted axonal regeneration by activating the mammalian target of rapamycin (mTOR) pathway (S. Li et al., 2016). Taken together, we found that ipRGCs survived well with chronic OHT and this survival is subtype dependent. More work is needed to better understand the underlying molecular mechanisms of protection, degeneration, and regeneration which, in turn, will offer insights to develop neuroprotective treatments of glaucoma.

Behavioral consequences of chronic OHT on circadian entrainments and visual acuity

As discussed, ipRGCs are involved in a broad range of image forming and non-image forming functions (Do, 2019; Duda et al., 2020; Lucas, Allen, Milosavljevic, Storchi, & Woelders, 2020). Thus, we further investigated how the subtype-specific loss of ipRGCs affects visual behaviors in mice suffering chronic OHT insult. It has been shown that M4 cells project to dLGN (Estevez et al., 2012) and contribute to contrast sensitivity aspects of image forming vision with and without rods/cones inputs (Ecker et al., 2010; Schmidt et al., 2014; Sonoda et al., 2020).

Specifically, following ablation of several ipRGC subtypes including the M4s, contrast sensitivity was significantly reduced at various spatial frequencies (Schmidt et al., 2014). The authors also observed that spatial frequencies at which significant deficits were found in these mice matched the spatial frequencies (0.05 and 0.09 c/d) to which M4s respond most strongly in isolated retina (Estevez et al., 2012), suggesting that M4s could contribute to contrast sensitivity. In the current study, we found a significant decrease in contrast sensitivity, which is also correlated with the significant loss of M4 cells in OHT eyes. Similarly, we also observed that our OHT mice exhibited greater reduction of contrast sensitivity at 0.103 c/d than 0.192 c/d (Figure 4c). Yet we cannot exclude that degeneration of other non-M4 RGCs may contribute to the change in contrast sensitivity in the OHT mice (Baden et al., 2016; Khani & Gollisch, 2017). M1 cells, on the other hand, project to the suprachiasmatic nucleus (SCN), the master clock that regulates circadian photoentrainment, a non-image forming function (Baver et al., 2008; Do, 2019; Vadnie & McClung, 2017). With little M1 loss under chronic OHT, we found no delay in circadian re-entrainment between non-ocular hypertensive controls and OHT mice.

Note that when mice were subjected to a phase advance of a relatively bright LD cycle, the OHT mice established a stable phase of entrainment significantly faster than the age-matched controls. In a similar study using a rat OHT model, the authors found the opposite result where it took significantly more time for the OHT rat to re-entrain after phase advance (Drouyer et al., 2008). They also observed a significant reduction in RGC terminals in the SCN. These differences could be attributed to the intensity and duration of IOP elevation which may induce a different severity of RGC cell death. For example, IOP had been doubled for over 4 months in the rat glaucoma model, which led to a 50–70% reduction of RGC axon terminals in a majority of visual and non-visual centers in the brain (Drouyer et al., 2008). By contrast, our mouse model exhibited a mild IOP elevation (about 20–30% increase) for a longer duration (from 3 months to up to a year).

The accelerated re-entrainment observed in mice with elevated IOP relative to normotensive controls was unexpected. We hypothesize that even a slight reduction of the direct photic input into the hypothalamus may cause an increase in the desynchrony among the cellular oscillators of the SCN. Such desynchrony speeds circadian entrainment in response to a shifted LD cycle. For example, the administration of vasoactive intestinal peptide (VIP) induces desynchrony or “phase tumbling” among SCN oscillator cells. If VIP is given prior to a shift in the LD cycle, the number of days required to re-entrain is halved compared to mice receiving vehicle injections(An et al., 2013). Although we did not observe a reduction in M1 cells in our bilaterally lasered mice, it remains possible that the synaptic input from M1 cells to the SCN was damaged under prolonged elevated IOP. In addition, the expression level of melanopsin as well as melanopsin activity could also affect re-entrainment rates. Melanopsin knockout mice exhibit reduced phase-shifting responses (Panda et al., 2002). The lack of C-terminal phosphorylation sites of the melanopsin molecue may also lead to faster re-entrainment (Somasundaram et al., 2017).Alternatively, we observed a decrease in M4 cells which project to the intergeniculate leaflet (IGL) (Ecker et al., 2010). The IGL, in turn, encodes irradiance and sends a projection to the SCN via the geniculohypothalamic tract (Morin, 2013; Yuan et al., 2018). Therefore, the degeneration of M4 cells could affect the IGL’s modulation of the retinal input to the SCN and indirectly impact its behavioral output(Hanna, Walmsley, Pienaar, Howarth, & Brown, 2017).

The disruption of the circadian system observed in animal glaucoma models are also observed in patients. Primary open angle glaucoma (POAG) and primary angle-closure glaucoma (PACG) patients often suffer mood and sleep disorders, and the severity is correlated with visual field impairment (Ciulla et al., 2020; Wang, Zhang, Ding, & Wang, 2013). Sleep efficiency assessed by polysomnographic recordings was significantly lowered (Gracitelli et al., 2015), and blood melatonin levels are higher in early morning in POAG patients (Ma, Shen, Shen, Qi, & Sun, 2018). It will be of great interest to further understand the ipRGC-mediated neural circuits that account for the behavioral changes suffered by glaucoma patients. Such an understanding will lead to improved patient care and treatment as glaucoma develops and progresses.