Ocular Hypertension Results in Hypoxia within Glia and Neurons throughout the Visual Projection

The magnitude and duration of hypoxia after ocular hypertension (OHT) has been a matter of debate due to the lack of tools to accurately report hypoxia. In this study, we established a topography of hypoxia in the visual pathway by inducing OHT in mice that express a fusion protein comprised of the oxygen-dependent degradation (ODD) domain of HIF-1α and a tamoxifen-inducible Cre recombinase (CreERT2) driven by a ubiquitous CAG promoter. After tamoxifen administration, tdTomato expression would be driven in cells that contain stabilized HIF-1α. Intraocular pressure (IOP) and visual evoked potential (VEP) were measured after OHT at 3, 14, and 28 days (d) to evaluate hypoxia induction. Immunolabeling of hypoxic cell types in the retina and optic nerve (ON) was performed, as well as retinal ganglion cell (RGC) and axon number quantification at each time point (6 h, 3 d, 14 d, 28 d). IOP elevation and VEP decrease were detected 3 d after OHT, which preceded RGC soma and axon loss at 14 and 28 d after OHT. Hypoxia was detected primarily in Müller glia in the retina, and microglia and astrocytes in the ON and optic nerve head (ONH). Hypoxia-induced factor (HIF-α) regulates the expression of glucose transporters 1 and 3 (GLUT1, 3) to support neuronal metabolic demand. Significant increases in GLUT1 and 3 proteins were observed in the retina and ON after OHT. Interestingly, neurons and endothelial cells within the superior colliculus in the brain also experienced hypoxia after OHT as determined by tdTomato expression. The highest intensity labeling for hypoxia was detected in the ONH. Initiation of OHT resulted in significant hypoxia that did not immediately resolve, with low-level hypoxia apparent out to 14 and 28 d, suggesting that continued hypoxia contributes to glaucoma progression. Restricted hypoxia in retinal neurons after OHT suggests a hypoxia management role for glia.


Introduction
Glaucoma is a neurodegenerative disease that impacts the function of RGCs in transmitting visual information from the eye to the brain. Consequently, glaucoma is the second leading cause of blindness for millions of people around the world [1]. Accumulating research effort and recent findings in models of glaucoma have shown that low oxygen availability (hypoxia) and the subsequent reoxygenation exposes cells to severe stress and protein damage in the retina and ONH [2][3][4][5][6][7].
During hypoxia, cells initiate a variety of defense responses mediated by the hypoxiainducible factors (HIFs). HIFs are heterodimeric transcription factors consisting of α and β subunits. HIF-β is a stable, oxygen-insensitive subunit and HIF-α has three oxygensensitive subunits (HIF-1α, HIF-2α, or HIF-3α) that together bind to the hypoxia response element (HRE) in the nucleus to promote the transcription of hypoxia-responsive genes [8]. HIF-3α has not been studied extensively, and it is currently unknown whether HIF-1α or

Genotyping
Ear or tail snips were collected and sent to Transnetyx for genotyping. The Cre-ERT2-ODD forward primer (TTAATCCATATTGGCAGAACGAAAACG), Cre-ERT2-ODD reverse primer (CAGGCTAAGTGCCTTCTCTACA), and tdT primers were used to detect the Cre-ERT2-ODD construct and/or the tdT. Control mice in this study were littermates that were negative for the Cre-ERT2-ODD construct.

Tamoxifen Injection
Tamoxifen (300 µL of a 10 mg/mL solution in sunflower oil) was administered interperitoneally once/day for five days to initiate transcription of cre-recombinase in 2-monthold CAG-ODD mice.

Intraocular Pressure (IOP) Measurements
Ten IOP measurements per eye were taken in lightly anesthetized mice (2.5% isoflurane) using a TonoLab rebound tonometer (iCare Finland) prior to, and then weekly after, induction of OHT. For time points shorter than 1 week, IOP was measured prior to sacrifice. IOP integral (mmHg-days exposure over baseline) was also calculated [7].

Ocular Hypertension (OHT) Model
Magnetic microbeads (1.5 µL, 8 µm diameter; COMPEL COOH-Modified, UMC4001; Bangs Laboratories) were injected into the anterior chamber of both eyes via glass pulled micropipette attached to a microinjection system, and a neodymium magnet was used to guide the microbeads into the trabecular meshwork to block the outflow of aqueous humor and increase IOP. We used the magnetic bead model successfully to elevate IOP up to four weeks. To eliminate the confounding factor of contralateral eye effects on glial activation [15], separate animals served as control.

Determination of Visual Function
The Diagnosis visual testing system (Celeris) was used to measure visual evoked potential (VEP). For both, mice were dark adapted for ≥1 h, then anesthetized with Ketamine (100 mg/kg)-Xylazine (10 mg/kg) and placed prone on a heated stage. Electrodes were placed in the mouse's cheek (reference), scalp at the base of the skull, and the base of the tail (ground). Bright luminance stimulators, with a light intensity of 1 cd*s/m 2 , were placed on each eye and the output of 600 sweeps was processed. VEP output was compared across groups by evaluating N1 (negative peak) and P2 (positive peak) amplitudes, and response amplitude (the amplitude from N1 trough to P2 peak amplitude). Terminal VEP was measured before sacrifice.

Hypoxia Detection by Pimonidazole
As previously described [7], 60 mg/kg pimonidazole hydrochloride was administered by intraperitoneal (IP) injection 90 min before animals were sacrificed to detect hypoxia. The subsequent immunofluorescence staining of tissue sections with an anti-pimonidazole antibody reveals the presence of hypoxic cells. Briefly, 4% paraformaldehyde-fixed retinas and ONs were cryosectioned, then incubated for 1 h in blocking agent (1% Triton X-100, 0.5% bovine serum albumin (BSA), 0.9% NaCl, and 5% donkey serum in 1% phosphate buffered saline (PBS); PBS-T-BSA). Slides were incubated with anti-pimonidazole antibody (FITC-conjugated mouse anti-pimonidazole, 1:200, Hypoxyprobe Green kit; Hypoxyprobe, Burlington, MA, USA) in blocking solution overnight. The following day, retinal and ON sections were rinsed in PBS and cover-slipped for imaging using a Leica DMi8 confocal microscope. Six slides (four sections/slide) were imaged. We added primary antibody against pimonidazole conjugated to FITC (1:200) to ON and retinal sections from mice that received no pimonidazole IP injection in order to affirm pimonidazole labeling specificity.

Perfusion and Tissue Preparation
Mice were killed 6 h, 3 d, 14 d, and 28 d after OHT with an overdose of sodium pentobarbital (Beuthanasia-D, 390 mg/kg, IP; Merck Animal Health, Baton Rouge, LA, USA), then perfused transcardially with 0.1 M PBS, then with 4% PFA. Retina, ON, and brain were harvested for histology and immunofluorescence.

Microtome Sectioning
Brains were coronally sectioned at 50 µm using a Leica freezing microtome (Wetzlar, Germany), then 10 representative sections were collected of the superior colliculus (SC), the primary and most distal site of the RGC projection in rodents; the hypothalamic suprachiasmatic nuclei (SCN), site of the master circadian pacemaker; and the lateral geniculate nucleus (LGN) of the thalamus, which receives inputs from both eyes and relays these messages to the primary visual cortex via the optic radiation. Sections were mounted on slides, imaged using a Zeiss AxioZoom V16 (AxioCam MRm Rev.3; Zeiss, Jena, Germany) and analyzed using ImageJ to calculate tdT mean intensity [7].

Quantification of Axons
Unbiased stereological analysis of axons in PPD-stained ON using a 100× objective was performed as previously described [7] using the optical fractionator module within StereoInvestigator (MicroBrightfield Bioscience, Williston, VT, USA). A 5 × 5 µm counting frame was utilized across roughly 40 sites. The coefficient of error (Schmitz-Hof) was maintained at 0.05 or below, ensuring sufficient sampling rate [7,18].

Protein Extraction for Capillary-Based Electrophoresis (WES)
Retina and myelinated ONs were collected in T-PER buffer with HALT protease and phosphatase inhibitors, then disrupted with a Branson Sonifier to create a protein lysate. After centrifugation for 15 min at 15,000 rpm, the supernatant was collected. Total protein concentration was measured by a Bicinchoninic Acid (BCA) assay kit. Proteins were analyzed by capillary tube-based electrophoresis immunoassay using the Wes and normalized to total protein in the sample. Wes is a Protein Simple instrument that separates proteins by an electrical charge in capillary tubes and allows binding of primary antibody then protein visualization within the capillary. Protein analysis for GLUT1 (1:50, rabbit, Novus Biologicals, Littleton, CO, USA, NB110-39113), GLUT3 (1:25, mouse, R & D systems, Minneapolis, MN, USA, NAB1415), HIF-1α (1:1000, mouse, Santa Cruz, Dallas, TX, USA, sc-13515), and HIF-2α (mouse, 1:1000, Santa Cruz, Dallas, TX, USA, sc-13596) was repeated at least three times with biological replicates [7,19].

Statistical Analysis
Statistical analysis was performed using GraphPad Prism 7 (GraphPad, La Jolla, CA, USA). Data were expressed as mean ± SEM and analyzed using unpaired, Student's t-test (two groups) when comparing control versus experimental and one-way ANOVA followed by Tukey's multiple-comparison post hoc test when making comparisons across more than two groups. The Kruskal-Wallis test was used to compare more than two groups and Dunn's multiple comparison testing when a significant Bartlett's test was detected. A p-value < 0.05 was considered statistically significant. Table 1 shows a breakdown of all of the sample numbers used in this study.   Figure 1A shows the experimental sequence of events for mice, including induction of cre recombinase expression using tamoxifen injection, visual function testing, induction of OHT, and sacrifice. There was no difference in IOP, RGC number, axon number, or VEP traces between the cre+ and cre− CAG-ODD mice, so cre+ and cre− mouse values were combined in Figure 1 panels. IOP was significantly elevated by 3 days after magnetic microbead injection to generate OHT in both cre− and cre+ CAG-ODD mice; this elevation was maintained for all mice subjected to OHT ( Figure 1B). As expected, there was a significant decrease in RGC and axon number at 14 d and 28 d after OHT compared to control mice and mice in the 6h and 3d after OHT groups ( Figure 1C,E). Figure 1D shows representative immunofluorescence for sections used to quantify RGCs.
We analyzed VEP at 3, 14, and 28 d after hypoxia. A significant decrease in terminal VEP, including N1 and P2 amplitudes, was detected 3, 14, and 28 d after OHT in both cre− and cre+ CAG-ODD mice (Figure 2A,B). A significant decrease in P2 amplitude at 3 d compared to 14 d after OHT was detected. In addition, a significant decrease in terminal response amplitude (the amplitude of the N1 trough to P2 peak) was detected at 3, 14, and 28 d after OHT in both cre− and cre+ CAG-ODD mice ( Figure 2C). Furthermore, the response amplitude at 3 d after OHT was significantly decreased compared to 28 d ( Figure 2C), suggesting some recovery of VEP 28 d after OHT. Figure 2D shows representative VEP traces taken at baseline, prior to OHT (Control), then at 3 d, 14 d, and 28 d after OHT.

Hypoxic Glia and Neurons in Retina after OHT
To determine which visual system cells showed evidence of hypoxia, we immunolabeled sagittal sections of retina from CAG-ODD mice subjected to OHT with antibodies against proteins that are specific for RGCs (RBPMS), subsets of amacrine cells (choline acetyltransferase, ChAT), and bipolar cells (PKCα) in Figure 3A-C, respectively. Although there was significant tdT labeling in the ganglion cell layer (GCL) at 6 h and 3 d after OHT ( Figure 3A), the labeling corresponded to Müller cell endfeet. By 14 and 28 d, however, there was some colocalization of RBPMS with tdT (arrowhead at 6 h, 14 d, and 28 d). We observed colocalization of ChAT with tdT after OHT ( Figure 3B). There was some colocalization of tdT with PKCα in bipolar cells at 6 h and 3 d after OHT (arrowheads, Figure 3C).
In Figure 4, immunolabeling with antibodies against proteins specific for astrocytes and Müller glia (GFAP), microglia (Iba1), or Müller glia (Vimentin) was undertaken to determine which glial cell types were exhibiting the tdT label. The majority of tdT labeling, denoting HIF-1α stabilization and thus, hypoxia, occurred in Müller glia after OHT. The hypoxia labeling was evident by 6h and remained strong at 3 d after OHT across all analyzed retinal sections. By 14 and 28 d after OHT, the tdT had dissipated to a degree, though it could still be observed in the ganglion cell layer (GCL) and the inner nuclear layer (INL), corresponding to the location for Müller glia cell bodies. Labeling for tdT also colocalized with GFAP, Vimentin (Vim) and Iba1 (see arrows in Figure   . Experimental design. Intraperitoneal injection of tamoxifen was administered to 2-month-old CAG-ODD transgenic mice. Visual evoked potential (VEP) and IOP were measured before and after 3, 14, and 28 days of OHT induction. OHT was induced by magnetic bead intracameral injection. Mice received an IP injection of 60 mg/kg pimonidazole, then were sacrificed 90 min later to compare and assess hypoxia using tdTomato (tdT) expression and pimonidazole (pimo) methods at 6 h, 3 days, 14 days, and 28 days after OHT in the retina, ON and visual centers in the brain. (B). A significant IOP elevation was detected (****, p < 0.0001) 3, 14, and 28 d after OHT. n = 16 eyes at 3 d, 14 eyes at 14 d, 12 eyes at 28 d, and 30 control eyes. (C). A significant decrease (F (4, 75) = 17.51) in RGC count per GCL was shown 14 d after OHT compared to control (%, p < 0.0001), 6 h (*, p < 0.0332), and 3 d (&, p = 0.0026). In addition, a significant decrease was detected in RGC count per µm GCL 28d after OHT compared to control (@, p < 0.0001), 6h ($, p = 0.0052), and 3d (#, p = 0.0002); n=12 eyes at 6 h, 12 eyes In addition, a significant decrease was detected in RGC count per µm GCL 28d after OHT compared to control (@, p < 0.0001), 6 h ($, p = 0.0052), and 3d (#, p = 0.0002); n = 12 eyes at 6 h, 12 eyes at 3 d, 15

Hypoxic Glia and Neurons in Retina after OHT
To determine which visual system cells showed evidence of hypoxia, we immunolabeled sagittal sections of retina from CAG-ODD mice subjected to OHT with antibodies against proteins that are specific for RGCs (RBPMS), subsets of amacrine cells (choline acetyltransferase, ChAT), and bipolar cells (PKCα) in Figure 3A-C, respectively. Although there was significant tdT labeling in the ganglion cell layer (GCL) at 6 h and 3 d after OHT (Figure 3A), the labeling corresponded to Müller cell endfeet. By 14 and 28 d, however, there was some colocalization of RBPMS with tdT (arrowhead at 6 h, 14 d, and 28 d). We observed colocalization of ChAT with tdT after OHT ( Figure 3B). There was some colocalization of tdT with PKCα in bipolar cells at 6 h and 3 d after OHT (arrowheads, Figure 3C). To compare the specificity of the CAG-ODD construct, we also injected some mice with pimonidazole, a chemical that forms covalent adducts on proteins in cells exposed to hypoxic conditions (<10 mm Hg oxygen), see Figure 4D. Pimonidazole was detected in astrocytes, Müller glia (arrows), and RGCs or amacrine cells (arrowheads) 6 h, 3 d, and 28 d after OHT ( Figure 4D), a finding that corroborates our previous study [7]. Pimonidazole was shown separately because it was otherwise masked by the intense labeling of tdT. Overall, the pimonidazole labeling was observed in the same cell types as detected by the tdT but was not as widespread. In Figure 4, immunolabeling with antibodies against proteins specific for astrocytes and Müller glia (GFAP), microglia (Iba1), or Müller glia (Vimentin) was undertaken to determine which glial cell types were exhibiting the tdT label. The majority of tdT labeling, denoting HIF-1α stabilization and thus, hypoxia, occurred in Müller glia after OHT. The hypoxia labeling was evident by 6h and remained strong at 3 d after OHT across all analyzed retinal sections. By 14 and 28 d after OHT, the tdT had dissipated to a degree, though it could still be observed in the ganglion cell layer (GCL) and the inner nuclear layer (INL), corresponding to the location for Müller glia cell bodies. Labeling for tdT also colocalized with GFAP, Vimentin (Vim) and Iba1 (see arrows in Figure 4A-C).

Strong Hypoxia in Optic Nerve (ON) Glia after OHT
Like the retina, tdT-positive cells were most prominent at 6 h and 3 d after OHT in the myelinated ON. Cells positive for tdT colocalized with GFAP (astrocytes; Figure 5A) and Iba1 (microglia; Figure 5B). Though tdT was significantly decreased at 14 and 28 d compared to 6 h and 3 d after OHT, the occasional cell with strong labeling can still be observed (see Iba1 panel at 28 d, Figure 5B). The tdT labeling at 28 d after OHT in the ON was not observed using pimonidazole [7], suggesting that the CAG-ODD hypoxia reporter strategy is more sensitive than pimonidazole labeling. Pimonidazole-positive microglia and astrocytes were evident after 6 h and 3 d of OHT ( Figure 5C,D) but did not label hypoxic astrocytes or microglia 14 and 28 d after OHT (data not shown).

Hypoxia Is Highest in the Optic Nerve Head (ONH) after OHT
The ONH showed the highest intensity of tdT expression 6 h and 3 d after OHT ( Figure 5E,F) compared to control. The 14 and 28 d time points are not shown since they are comparable to the ON at 6 h and 3 d. The tdT colocalized with both GFAP and Iba1, though it also appeared to exceed the cumulative labeling of each. Figure 6 shows the quantification of tdT labeling intensity in the ON, retina and the ONH. When compared across the 6 h and 3 d time points after OHT, the ONH had significantly stronger tdT labeling than the ON and the retina. Within the ON, tdT mean intensity was significantly greater than control at 6 h and 3 d, and 28 d after OHT. After a small dip in tdT intensity at 14 d, the hypoxia label increased, although non significantly (p = 0.064), at 28 d compared to 14 d after OHT. In the ONH, all time points showed a significant increase in tdT mean intensity compared to control. The 6 h and 3 d after OHT time points had significantly greater tdT mean intensity than 14 and 28 d after OHT. Within the retina, all time points showed a significant increase in tdT mean intensity compared to control. A significantly greater tdT mean intensity was detected 6 h and 3 d after OHT compared to 14 and 28 d after OHT. To compare the specificity of the CAG-ODD construct, we also injected some mice with pimonidazole, a chemical that forms covalent adducts on proteins in cells exposed to hypoxic conditions (<10 mm Hg oxygen), see Figure 4D. Pimonidazole was detected in astrocytes, Müller glia (arrows), and RGCs or amacrine cells (arrowheads) 6 h, 3 d, and 28 d after OHT ( Figure 4D), a finding that corroborates our previous study [7]. Pimonidazole was shown separately because it was otherwise masked by the intense labeling of tdT. Overall, the pimonidazole labeling was observed in the same cell types as detected by the tdT but was not as widespread.

Strong Hypoxia in Optic Nerve (ON) Glia after OHT
Like the retina, tdT-positive cells were most prominent at 6 h and 3 d after OHT in the myelinated ON. Cells positive for tdT colocalized with GFAP (astrocytes; Figure 5A) and Iba1 (microglia; Figure 5B). Though tdT was significantly decreased at 14 and 28d

Hypoxia in Brain Visual Nuclei after OHT
We examined retino-recipient areas of the brain, including the suprachiasmatic nucleus (SCN), lateral geniculate nucleus (LGN), and superior colliculus (SC), for tdT labeling after OHT. A significant increase in tdT expression was evident at 6 h and 3 d after OHT in the SCN (Figure 7A,D) and LGN ( Figure 7B,E). Similarly, at 6 h and 3 d after OHT, there was a significant increase in tdT labeling in the SC compared to control, 14 d, and 28 d ( Figure 7C,F). We quantified the intensity of tdT labeling, finding significantly higher tdT in the LGN 6h after OHT compared to the SC at both 6 h and 3 d after OHT ( Figure 7G).

11, 888 11 of 22
was not observed using pimonidazole [7], suggesting that the CAG-ODD hypoxia reporter strategy is more sensitive than pimonidazole labeling. Pimonidazole-positive microglia and astrocytes were evident after 6 h and 3 d of OHT ( Figure 5C,D) but did not label hypoxic astrocytes or microglia 14 and 28 d after OHT (data not shown).  0.064), at 28 d compared to 14 d after OHT. In the ONH, all time points showed a significant increase in tdT mean intensity compared to control. The 6 h and 3 d after OHT time points had significantly greater tdT mean intensity than 14 and 28 d after OHT. Within the retina, all time points showed a significant increase in tdT mean intensity compared to control. A significantly greater tdT mean intensity was detected 6 h and 3 d after OHT compared to 14 and 28 d after OHT.

Hypoxia in Brain Visual Nuclei after OHT
We examined retino-recipient areas of the brain, including the suprachiasmatic nucleus (SCN), lateral geniculate nucleus (LGN), and superior colliculus (SC), for tdT labeling after OHT. A significant increase in tdT expression was evident at 6 h and 3 d after OHT in the SCN ( Figure 7A,D) and LGN ( Figure 7B,E). Similarly, at 6 h and 3 d after OHT, there was a significant increase in tdT labeling in the SC compared to control, 14 d, and 28 d ( Figure 7C,F). We quantified the intensity of tdT labeling, finding significantly higher tdT in the LGN 6h after OHT compared to the SC at both 6 h and 3 d after OHT ( Figure  7G).   LGNs at 6 h, 4 LGNs at 3 d, 4 LGNs at 14 d, 5 LGNs The superior colliculus contains the RGC synapses furthest removed from the retina, so we determined revealing the cell types that were positive for the tdT hypoxia reporter there. The tdT label was observed in the collicular vasculature, surrounded by GFAPpositive astrocyte endfeet ( Figure 8A), or alone ( Figure 8B). Figure 8C-E show a vessel at high magnification labeled with endothelial marker CD-31; in Figure 8D, a yellow dotted line shows the placement of the colocalization plot line, with results in Figure 8E. The CD-31 immunolabel is colocalized with the tdTomato hypoxia reporter, as shown by the coincidence of green (CD-31) and magenta (tdT) labeling peaks ( Figure 8E).
We used NeuroTrace to label neurons in the superior colliculus sections, finding at 6 h and 3 d after OHT that neurons were positive for tdT label (Figure 9, arrowheads). Hypoxia reporter label was barely detectable by 14 and 28 d after OHT (Figure 9, 14 d and 28 d panels). No tdT label was observed in the non-OHT (Control) superior colliculus.

Significant Changes in GLUT1 and GLUT3 Levels after OHT
While GLUT1 protein significantly increased 28 d after OHT in ON compared to 6 h and 14 d after OHT, it significantly increased in the retina 3 d after OHT compared to control, 6 h, and 28 d ( Figure 10A,B). GLUT3 protein, on the other hand, significantly increased in the ON 3 d after OHT compared to 14 d; there was no change in GLUT3 of the retina at any time point ( Figure 10C,D). See supplementary data of original electrophoresis blots, see Figure S1.
Mice that were cre-recombinase negative (cre−) yielded results comparable to control; there was no tdT expression after OHT. Therefore, we did not show cre− animal results. 8E. The CD-31 immunolabel is colocalized with the tdTomato hypoxia reporter, as shown by the coincidence of green (CD-31) and magenta (tdT) labeling peaks ( Figure 8E). We used NeuroTrace to label neurons in the superior colliculus sections, finding at 6 h and 3 d after OHT that neurons were positive for tdT label (Figure 9, arrowheads). Hypoxia reporter label was barely detectable by 14 and 28 d after OHT (Figure 9, 14 d and 28 d panels). No tdT label was observed in the non-OHT (Control) superior colliculus.

Significant Changes in GLUT1 and GLUT3 Levels after OHT
While GLUT1 protein significantly increased 28 d after OHT in ON compar and 14 d after OHT, it significantly increased in the retina 3 d after OHT com Figure 9. A timecourse of superior colliculus sections after OHT. Superior colliculus was stained using NeuroTrace (green) and DAPI (blue); tdTomato (tdT, magenta) reporter indicates cells subjected to hypoxia. Left panels were imaged at 20× magnification (scale bar = 50 µm) and center panels at 60× magnification (scale bar = 20 µm). Right panels are the tdTomato fluorescence alone, in grayscale. In the control superior colliculus, no tdTomato label (right panel) was observed. tdTomato labeling was most widespread at 6 h and 3 d after OHT, including colocalization with NeuroTrace-positive cells (arrowheads). The dotted line in the 6 h panel is the outline of the 60× center panel. By 14 d, tdTomato labeling was no longer evident in the colliculus. In the 28 d panels, all are at 20× magnification, with the center showing GFAP labeling and the right the tdT. Five brains per group and ten sections per brain were analyzed from the 4 time points (6 h, 3 d, 14 d, 28 d).

Discussion
We determined the timeline of hypoxia after OHT in the visual pathway using a transgenic hypoxia reporter mouse. Hypoxia commenced with high intensity tdT expression at 6 h and 3 d after OHT in the entire visual pathway: Retina, ON, ONH, and visual centers in the brain. The highest intensity of hypoxia reporter labeling was observed in the ONH. By 14 d and 28 d after OHT, the hypoxia reporter was significantly decreased compared to 6 h and 3 d in the visual pathway; however, it did not return to control levels. Hypoxic glia (microglia, astrocytes, Müller glia) and, to a much lesser extent, neurons (RGCs, amacrine cells, bipolar cells), were evident in the retina. In the ON and ONH, hypoxic GFAP-positive astrocytes and Iba1-positive microglia were observed. Müller glia were the cells most hypoxic after OHT, as measured by tdT labeling in the retina. In the superior colliculus, endothelial cells and neurons were observed to be hypoxic. Overall, hypoxia was more evident in glial cells compared to neurons, suggesting a hypoxia management role for glia.
We observed hypoxia in the retina, ON, and superior colliculus within 6h of OHT using the microbead model. Thus, the hypoxia-driven tdTomato indicates that ocular blood flow was reduced within 6 h of OHT induction. The early timing of this hypoxia suggests an initiating IOP spike with bead injection since the IOP measured at 3 d was 16 mm Hg, only 4 mm Hg higher than baseline. Due to the relatively slow onset and chronic nature of primary open angle glaucoma in human patients, there is an expectation that hypoxia/ischemia, and thus, IOP spiking, does not reliably model glaucoma. However, data from patients and non-human primates indicates IOP can spike up to 10 mm Hg from eyeblink, up to 90 mm Hg from squeezing one's eyelids shut [20], or even as a result of acute stress [21]. These IOP changes, though relatively short-lived, are nonetheless capable of reducing ocular bloodflow to a degree that can lead to intermittent hypoxia/ischemia [22]. We also do not yet understand the impact of potentially thousands of IOP spikes daily, particularly for individuals susceptible to developing glaucoma. Significantly, the instability of oxygen supply to the retina or optic nerve through changes in IOP may further exacerbate glaucoma-associated optic neuropathy in individuals with disturbed autoregulation of ocular blood flow [23], a leading hypothesis for the mechanism of normal tension glaucoma. Additional work is necessary to establish how often and to what degree hypoxia contributes to glaucomatous neurodegeneration, but the evidence for IOP spiking in patients and in animal models suggests that it is of value to examine its impact. In research from the Pardue lab, a significant decline in the b-wave amplitude of the flash electroretinogram indicated whether a retina had been exposed to ischemia [24]. We evaluated VEP in our study, not flash ERG, so cannot apply this yardstick to our data. The early onset of hypoxia (6 h) does suggest an IOP spike with OHT induction. However, the maintenance of tdTomato labeling at 3 d, beyond the point at which the tdTomato protein would have turned over [25], suggests that hypoxia induction extended beyond the induction of OHT.
Visual function measured by VEP significantly decreased 3 d after OHT, which coincided with initial IOP elevation. However, IOP remained elevated through the 28 d of OHT while the VEP response amplitude gradually increased, though it never achieved control levels. The improvement of visual function seen 28 d after OHT may be the result of compensation from functional RGCs. The minimum number of RGCs needed to achieve visual function is still unknown. Interestingly, Park et al. have found that after induced IOP elevation in rats, presynaptic and postsynaptic vesicle proteins increased between RGCs and bipolar cells. This finding indicates that the loss of RGCs may drive new synaptic contact formation between RGCs and bipolar cells [26], potentially relieving the initial loss of synapses. In addition, the significant increase in GLUT1 level 28 d after OHT in the ON could have increased glycolysis and preserved some axons, which contributed to better visual function.
These findings confirm the impact of IOP elevation on initiating damage to RGCs and their axons but also suggest a potential functional adaptation to elevated IOP. The greatest decrement in visual function corresponded to the most intense hypoxia reporter labeling (3 d after OHT), suggesting that OHT-induced hypoxia contributes significantly to visual system dysfunction. These alterations occur prior to the significant declines in RGC and axon numbers at 14 and 28 d after OHT, which further supports that early hypoxia can drive glaucoma progression.

Hypoxic Glia and Neurons in Glaucomatous Retina
The OHT retina showed hypoxic microglia, astrocytes, Müller glia, bipolar cells, amacrine cells, and RGCs, but primarily Müller glia. In all cases, cells showed the highest labeling early (6 h and 3 d) after OHT but continued with lesser intensity at 14 and 28 d after OHT. The visual structures evaluated always showed minor hypoxia in glial cells, even out to 28 d, never returning to control levels. Reporter expression is a result of stabilization of the ODD domain of HIF protein. Interestingly, our previous publication found a significant increase in HIF-1α protein at 28 d compared to 14 d in the retina [7], despite our observation here that tdT reporter labeling remained relatively flat at 14 and 28 d, though higher than control. One explanation could be the masking of individual cells' hypoxia response in whole retina lysate that was used to measure HIF-1α in the previous publication [7]. In support of that, it was more often the case that we observed stand-out individual cell tdT labeling at 28 d than at 14 d (for example, microglia in the ON as shown in Figure 5B). Small numbers of individual cells with intense tdT labeling would not have appreciably altered total retinal HIF-1α levels.
Here, we found some evidence of hypoxia in RGCs, bipolar cells, and amacrine cells. There have been few cell-specific examinations of hypoxia in the retina during glaucoma. In an oxygen-induced retinopathy mouse model, HIF-1α staining was prominent in cells across the inner nuclear layer (INL) and ganglion cell layer (GCL), whereas HIF-2α was highly restricted to Müller glia and astrocytes [10]. In addition, HIF-2α was expressed at a higher level than HIF-1α within the hypoxic inner retina. We have observed activation of HIF-1α but no change in HIF-2α protein in the retina and ON after OHT [7]; however, we did not use immunolabeling to distinguish among cell types at the time. Further investigation is warranted to unravel the cellular specificity of HIF-1α and HIF-2α.
Since Müller glia span the entire retina, they may be exposed to stress more dramatically than other retinal cells. Müller glia became activated as early as 2-3 days after OHT in rats as determined by upregulation of GFAP; the activation was sustained for several months [27,28]. The dramatic HIF stabilization that occurred in Müller glia in our study suggests that these cells are quite sensitive to oxygen levels. During hypoxia, metabolic activity is converted from aerobic respiration, oxidative phosphorylation (OxPhos), to anaerobic glycolysis by hypoxia-activated, HIF-1-mediated transcriptional activity, which suppresses mitochondrial aerobic metabolic processes, including the tricarboxylic acid cycle and OxPhos [29][30][31]. Müller glia have a reputation for being glycolytic cells because they secrete lactate [32]. Müller glia are also capable of maintaining ATP production during complex III inhibition; this was interpreted to mean Müller glia are preferentially glycolytic as well [33]. However, Müller glia have significant mitochondria in their basal processes [34], and low levels of pyruvate kinase [35] that suggest an impaired ability for glycolysis. There is evidence that Müller glia utilize lactate released from photoreceptors to generate ATP through mitochondria [35]. The sensitivity of Müller glia for oxygen. as shown through tdT labeling, would support a Müller glial need for oxygen and potentially a preference for OxPhos. It is not clear if HIF activation could propel Müller glia into a state more supportive of neuronal survival because hypoxic Müller glia and neurons undergo a HIF-activated transition to glycolysis that would increase competition for the available glucose [32,36].
Reduced oxygenation during hypoxia stabilizes HIF-α, which promotes the upregulation of glucose transporters (GLUTs) to induce glycolysis and maintain ATP availability, enabling metabolic adaptation to limited oxygen [11]. We found a significant increase in glut1 (astrocytic) transcripts [7], and GLUT1 protein 3 d after OHT in the retina compared to control, 6 h, and 14 d. In ON, GLUT1 protein was low 14 d after OHT, timing that matches our detection of no tdT expression 14 d after OHT in ON. However, GLUT1 in ON at 28 d was higher than at any other time point. This raises the possibility that, in the ON, alterations to improve glucose transport can be delayed when compared to the timing of the impetus for that increase (hypoxia). In contrast to the astrocyte-associated glucose transporter, Glut3 (neuronal) mRNA, but not protein, significantly increased in glaucomatous DBA/2J ON [37]. Similarly, GLUT3 protein did not change at all after OHT in the retina. However, in the ON, GLUT3 significantly increased 3 d after OHT (compared to 14 d), a time frame that would suggest that the upregulation was a response to the hypoxia occurring there. IOP elevation compromises blood flow to the visual pathway and that leads to oxygen and glucose deprivation. This glucose deprivation will drive the use of glycogen stores in cells, but glycogen would be depleted after prolonged hypoxia, therefore leading to metabolic vulnerability. Indeed, GLUT could increase at 3 d in the ON, but if glucose is not also available, then there will be no glycolysis nor subsequent ATP production. There were fewer hypoxic RGCs compared to glia early after OHT (6 h and 3 d). However, tdT expression seems to be more localized in cells, specially RGCs, later after OHT (14 and 28 d), which is coincident with RGC degeneration. This suggests that hypoxia may contribute to RGC degeneration.

Hypoxic Astrocytes and Microglia in Glaucomatous ON with the Highest Hypoxia in ONH
Surprisingly, our current and recently [7] reported data for the ON showed an absence of hypoxic cells at 14d but, although nonsignificant, their presence 28 d after OHT ( Figure 5B). One cell type, microglia, stood out for late tdT labeling in the ON. We show hypoxic microglia at each of the time points analyzed, and in the retina, ONH, and ON. RGC death is induced by microglial-derived pro-inflammatory cytokines during neonatal hypoxia [38]. It is likely that the hypoxia response in microglia shown here also contributes to inflammation and facilitates RGC loss and dysfunction as a result of neuronalmicroglia interaction. Indeed, microglia make brief and direct contact with neuronal synapses [39]. Reduced neuronal activity reduces microglia-neuron contact frequency. Prolonged microglia-synapse contacts are observed after transient cerebral ischemia followed by the disappearance of synapses, suggesting microglia contribution to synaptic disconnection [40]. Kaur et al. showed that hypoxia-induced microglia activation in the developing periventricular white matter, cerebellum and the retina. Activated microglia released TNF-α and IL-1β, which was coupled with NO, iron, and ROS accumulation, which collectively resulted in neuroinflammation and death of oligodendrocytes, Purkinje neurons and the RGCs [41].
Mitochondrial density is higher in the unmyelinated axons in the prelaminar and laminar optic nerve to meet the high energy demand for electrical conduction [42]. Therefore, we expect that the ONH may be more susceptible to the mitochondrial dysfunction [43,44] and metabolic vulnerability [37] observed in glaucoma. Our ONH findings support that the ONH has the greatest sensitivity toward IOP elevation because of the significant hypoxia response recorded in the astrocytes of the glial lamina and the microglia. After laser photocoagulation of the trabecular meshwork, Chidlow et al. reported that the ONH is the pivotal site of RGC injury, as protein accumulation was detected within ONH axons 8 h after OHT. In addition, axonal cytoskeletal damage was detected after 3 d and later time points following OHT [45].

Hypoxia in Visual Centers of the Glaucomatous Brain
The glaucomatous visual centers in the brain (SCN, LGN, SC) showed a significant increase in hypoxia 6 h and 3 d after OHT. Hypoxia was evident in these visual centers at the same time points hypoxia was most significantly detected in the retina, ON, and ONH. Since the visual centers do not experience IOP increase directly, the tdT reporter expression observed there may be the result of oxygen depletion through the visual pathway, possibly through changes in bloodflow. RGC projections may have become a sink for any available oxygen with the advent of hypoxia. In support of this, we observed that tdT mean intensity peaked first in the LGN, followed by the SC, the most distal portion of the RGC axon projection. Although visual dysfunction and increased HIF-1α stabilization persist, hypoxia was not as profound 14 and 28 d after OHT because visual centers were devoid of tdT expression. At both 6 h and 3 d, endothelial cells and NeuroTrace-positive neurons in the superior colliculus carried the tdT label.
Despite ramified microglia being dependent on OxPhos [46,47], we observed no stabilization of HIF in microglia in the retinorecipient regions in the brain.

The CAG-ODD Reporter System
We found that tdT expression upon cre-recombinase activation was a more sensitive approach to detecting hypoxia after OHT than an intraperitoneal injection of pimonidazole, the compound that forms covalent adducts in cells that have a partial pressure of oxygen less than 10 mm Hg. There were more cell types and time points in which the CAG-ODD reporter was observed compared to pimonidazole. For example, significant tdT labeling intensity was observed at 14 d in the retina and ONH, whereas no pimonidazole labeling was observed in these structures at 14 d. Additionally, there was tdT expression in hypoxic microglia 28 d after OHT in ON while no pimondazole-positive cells were detected. These results, compared to our recent findings using pimonidazole, indicate that tdT expression is a more sensitive method [7]. Importantly, tdT is a short-lived protein, degrading within 24 h [25]. Relatively fast tdT turnover confirms that tdT labeling we observed at 3, 14, and 28 d after OHT is a result of new reporter expression from recent hypoxia and does not represent the leftover tdT label. Despite the ubiquity and strength of the CAG reporter, its expression will nevertheless vary across cell types, which could conceivably affect reporter expression.
This study provides evidence for pre-degenerative hypoxia in the visual pathway that may drive glaucoma progression; therefore, hypoxia represents a potential therapeutic target. Hypoxia and visual dysfunction following IOP elevation were evident in the first 6 h and out to 28 d. The impact of events downstream of hypoxia response, such as metabolic disruption and glial activation, are likely prime contributors to pathology in the retina, ONH, ON, and visual centers of the brain. The profound hypoxia response in Müller glia suggests greater attention should be paid to their mechanisms of hypoxia management.