Background
Glaucoma is the second leading cause of blindness worldwide behind cataracts. The only modifiable risk factor for glaucoma is intraocular pressure (IOP). However, there are patients with elevated IOP who do not develop glaucoma and approximately one third of glaucoma patients have normal-range IOP. All current medical and surgical treatments focus on lowering IOP. This approach can slow the progression of disease, even in many normal-tension glaucoma patients, but often does not stop it entirely at least in part due to poor patient compliance in administering IOP lowering eye drops. As a result of the above complexities, many laboratories are focused on understanding the molecular events that underlie axon degeneration in glaucoma with the hope of intervening in the disease process in an IOP-independent manner.
Glaucoma is an axonopathy meaning that the retinal ganglion cell (RGC) axons in the optic nerve undergo pathologic changes first, followed by death of the cell bodies (for review see [
1]). Several hypotheses have been posited regarding the mechanisms that induce axon degeneration in glaucoma including mechanical damage at the optic nerve head [
2‐
5], hypoxia and oxidative stress (for reviews see [
6‐
8]), and neuroinflammation (for review see [
9,
10]). Thus, an attractive neuroprotective candidate would affect more than one of these processes. One example is erythropoietin (EPO), a type I cytokine that blocks apoptosis, preserves the blood-brain barrier, counteracts oxidative stress, and limits glial reactivity (for review see [
11,
12]). Recent studies suggest that EPO can modulate neuroinflammation independently of its ability to block cell death (for review see [
11]).
We previously demonstrated that recombinant adeno-associated virus (rAAV)-mediated gene delivery of EPO or EPO.R76E, a form of EPO with attenuated erythropoietic activity, prior to onset of elevated IOP preserved the RGC axons, cell bodies, and vision in the well-characterized DBA/2J mouse model of glaucoma [
13]. Others have also shown that injections of EPO protein are effective in glaucoma models [
14,
15]. More recently, we demonstrated that rAAV.EpoR76E is also effective in an induced model of glaucoma suggesting that the protection was not specific to the particular mouse model [
16]. Further, therapeutic benefit was achieved in both models when treatment was delayed until the onset of elevated IOP although to a lesser extent than when rAAV.EpoR76E was provided earlier [
16].
The goal of this study was to exploit the efficacy of EPO in blocking glaucoma pathogenesis in order to investigate the mechanisms responsible for this axonopathy in the DBA/2J model. We delivered rAAV.EpoR76E into 5-month-old mice in order to yield maximal gene expression when IOP starts to elevate in the DBA/2J [
17]. We assessed therapeutic efficacy at 6 and 8 months of age, the age of onset of axon degeneration and prior to RGC death in this model [
18]. We investigated the effect of EPO.R76E on neuroinflammation and oxidative stress by quantifying numbers and morphology of microglial cells, levels of pro-inflammatory chemokines and cytokines, and expression of oxidative stress-related proteins.
Methods
Animals
DBA/2J mice were obtained from Jackson Laboratories (Bar Harbor, ME). Mice were group-housed, maintained on a 12-h light-dark cycle, and provided food and water ad libitum. All experiments were approved by the Institutional Animal Care and Use Committee of Vanderbilt University and were in accordance with the ARVO Statement for the Use of Animals in Vision and Ophthalmic research, protocol number M/12/131. Animals were injected in the quadriceps with 1 × 109 gc of rAAV2/8.CMV.EpoR76E (rAAV.EpoR76E) or rAAV2/8.CMV.eGFP (rAAV.eGFP). All viral vectors were produced by the University of Pennsylvania Vector Core (Philadelphia, PA). Mice were injected at 5 months of age and euthanized at 6 or 8 months of age.
Fluorescein angiography (FA)
Anesthetized mice were injected with 0.1 ml of fluorescein (AK-Fluor-10 %; Akom, Inc; Lake Forest, IL). Eyes were dilated with 1 % tropicamide and 2.5 % phenylephrine and lubricated with 2.5 % methylcellulose (Goniovisc Eye Care and Cure, Inc., Tuscon, AZ). Retinal fluorescein was imaged on a Micron IV imaging system (Phoenix Research Labs, Pleasanton, CA).
Flash visual evoked potential (fVEP)
The fVEPs were performed according to previously published methods [
13,
16]. Briefly, animals were dark-adapted overnight and anesthetized with an intraperitoneal injection of ketamine/xylazine/urethane (25/6/600 mg/kg). Eyes were dilated with 1 % tropicamide and then moistened with lubricating eye drops. The fur over the visual cortex was removed, and platinum needle electrodes (Grass Technologies, Warwick, RI) were inserted subdermally over the left and right visual cortex. Electrodes were placed at the snout and flank for the reference and ground, respectively. Animals were placed on a warm platform under a Ganzfeld dome (Diagnosys LLC, Lowell, MA). Two hundred flashes of white light at an intensity of 1.0 cd · sec/m
2 were presented to each animal at a flash frequency of 1 Hz with a 500 ms inner sweep delay. Data collection and analysis was performed by a masked investigator.
Anterograde transport analyses
Anterograde axon transport was assessed using previously published methodology [
19]. Briefly, anesthetized mice (3 % isoflurane) received single, bilateral, intravitreal injections of 2 μL of 1 % cholera toxin subunit B (CTB) conjugated to AlexaFluor-594 (Life Technologies, Carlsbad, CA) in phosphate-buffered saline (PBS) 48 h prior to collection. Brains were post-fixed in 4 % paraformaldehyde (PFA) for at least 48 h, the cortex was removed, and the remaining tissue was cryoprotected in 30 % sucrose for 48 h. Fifty-micron-thick coronal sections through the superior colliculus (SC) were obtained using a sliding microtome. Sections were mounted in Fluoromount-G (SouthernBiotech, Birmingham, AL), imaged on a Nikon Eclipse Ti epifluorescence microscope (Melville, NY), and CTB signal density in the SC was quantified using ImagePro Analyzer 7.0 software (Media Cybernetics Rockville, MD) as previously described [
19]. Briefly, each SC section was divided into several bins from medial to lateral, and average CTB signal density within each bin was calculated by dividing the area of pixels above background by total pixel area measured. A colorimetric scale ranging from blue (0 % transport) to red (100 % transport) was used to indicate average CTB signal density for each bin. Each section was adjoined to the next in a rostral to caudal fashion and then oriented to generate a retinotopic map of the SC. Areas of transport deficit were subtracted from total SC area to calculate percent intact transport. Fluorescence quantification was performed by a masked investigator.
Immunohistochemistry
For immunohistochemistry, eyes were post-fixed in 4 % PFA for at least 2 h and then whole retinas were isolated. A subset were embedded in 5 % agarose, and 70-μm-thick cross sections were collected on a vibratome. Cross sections or retinal flat mounts were incubated overnight at 4 °C in 5 % normal donkey serum in 0.1 % Triton X-100, 0.5 % BSA, 0.1 % Azide, and PBS (PBTA). Sections or flat-mounted retinas were then incubated overnight at 4 °C in primary antibody in PBTA, rinsed with PBTA, and then incubated overnight at 4 °C in secondary antibody in PBTA. Primary antibodies used: goat anti-Iba-1 (1:100; Abcam, Cambridge, UK), rabbit anti-Ki-67 (1:100; Thermo Fisher Scientific, Waltham, MA), and chicken anti-H-ferritin (1:50; Abcam, Cambridge, UK). Appropriate secondary antibodies conjugated to AlexaFluors were obtained from Life Technologies (1:200; Carlsbad, CA). Whole retinas were flat-mounted RGC side up. Both whole retinas and sections were mounted in Fluoromount-G (SouthernBiotech, Birmingham, AL), and 20-μm-thick z-stack images were collected on an Olympus FV-1000 confocal microscope (Center Valley, PA) using a ×60 oil immersion lens. Imaging on the confocal microscope was performed through the use of the Vanderbilt University Medical Center Cell Imaging Shared Resource.
The number of IBA1+ and IBA1 and Ki67+ cells were quantified in 27 high-power views of four retinas per treatment condition. A total of 207 and 168 IBA-1 positive cells were assessed from retinas of mice injected with rAAV.eGFP or rAAV.EpoR76E, respectively. Immunofluorescence of H-ferritin was quantified according to previously published methods [
20]. Briefly, ×60 magnification confocal micrographs at a 10-μm depth within the tissue were used for quantification. Fluorescence intensity from the ganglion cell layer to the outer plexiform layer was quantified and averaged using ImageJ software. The intensity was normalized based on total area of retina assessed.
Microglial quantification and morphometrics
Flat-mounted retinas immunolabeled with Iba-1 were imaged through the inner plexiform layer (IPL) beginning just below the RGC layer in five central and five peripheral areas. All microglial analyses were conducted by a masked investigator using ImagePro Analyzer 7.0 software. In each z-stack image, microglia number and soma area were quantified. Ramification number for each microglial cell was determined by counting intersections of glial processes with a 10 × 10 μm grid overlay.
Multiplex ELISA
Retinas were sonicated and run in singlet on the Milliplex MAP Mouse Cytokine/Chemokine Magnetic Bead Panel I 25-plex assay per manufacturer’s protocol (EMD Millipore, Billerica, MA). Plates were read on a Luminex MAGPIX with xPONENT software (Thermo Fisher Scientific, Waltham, MA) using settings stated in the manufacturer’s protocol. Multiplex ELISA plates were read and analyzed through the use of the Vanderbilt Hormone Assay and Analytical Services Core.
Quantitative reverse transcription-PCR
Retinas were homogenized, and RNA was extracted using a Qiagen RNeasy kit (Valencia, CA). RNA concentration and purity was measured on a spectrophotometer. A first-strand complementary DNA (cDNA) library was synthesized from 250 ng of RNA from each sample using the Superscript III First-Strand synthesis system (Invitrogen, Waltham, MA) and oligo-dT
20 primers. Quantitative PCR (qPCR) was performed using Power SYBR green master mix (Applied Biosystems, Waltham, MA). All primer sequences were obtained from previous studies; we assessed the following: iNOS, arginase 1, IL-1β, IL-6, TNFα, TGFβ, IL-4, and IL-10 [
21‐
24]. All qPCR was performed in duplicate using the Applied Biosciences 7300 real-time PCR system (Waltham, MA). The assay was performed in duplicate on 13 retinas per condition. The amplification threshold was set using system software. Relative changes in gene expression were determined using glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as an internal control [
25]. There was no significant change between samples in GAPDH levels.
Oxidative stress PCR array
The same cDNA as was used for the quantitative PCR analysis was also used in an RT2 Profiler mouse oxidative stress PCR array kit per manufacturer’s instructions (Qiagen, Valencia, CA). The cDNA from each group was pooled to have sufficient material for the assay.
Statistical analysis
All statistical analyses were performed using GraphPad Prism software (La Jolla, CA). A two-way ANOVA with a Bonferroni post hoc test was performed for the hematocrit time course. A one-way ANOVA with a Bonferroni post hoc test (α = 0.05) was used to analyze anterograde transport data, microglia morphometric data, ON quantification data, and fVEP latencies. Percent baseline fVEP amplitudes and results from the multiplex ELISA were compared using an unpaired Student’s t test (α = 0.05). A one-way ANOVA and Dunnett’s multiple comparisons post hoc test (α = 0.05) were used to analyze the qPCR results. Means and standard deviation were calculated for each data set.
Discussion
We detected a deficit in vision at 6 months of age, prior to a quantitative decrease in axon transport or axon degeneration, suggesting that the fVEP may be the earliest indicator of disease progression in this well-recognized progressive model of pigment dispersion glaucoma. The time course assessed in this study also revealed that delivery of rAAV.EpoR76E at the earliest sign of elevated IOP (5 months) preserved axon transport as soon as 1 month later. The serotype of rAAV that we used in this study results in gene expression approximately 3 weeks after injection [
17], so it is impressive that preservation of vision was already detectable at 6-months, after only 1 week of peak gene expression. This treatment effect is also long-lasting [
13,
16], making it ideal for glaucoma, a slow, progressive disease. Finally, treatment with EPO.R76E is safer than wild-type EPO because it does not cause a dangerous rise in hematocrit [
13,
26,
27,
41].
The blood-retina barrier is disrupted both in patients and in the DBA/2J model [
42‐
44], and the DBA/2J mouse demonstrates enhanced transendothelial monocyte migration into the optic nerve head that contributes to disease pathogenesis [
44]. Further, others have shown that EPO preserves the blood-brain barrier in models of brain injury by blocking endothelial cell death and astrocyte hypertrophy (for review see [
11]), and we have previously shown that EPO.R76E retains these activities in in vitro assays [
28,
45]. Despite this, the blood-retina barrier appeared equally disrupted in both rAAV.eGFP- and rAAV.EpoR76E-treated groups in this study. Thus, the decrease in number of microglial cells by rAAV.EpoR76E was neither due to inhibition of proliferation nor from preservation of the blood-retina barrier. Rather, our results show that treatment with rAAV.EpoR76E resulted in lower levels of several molecules known to promote the recruitment and infiltration of Th1 cells from the periphery: IL-12p40 (part of IL-23) [
46], CCL4, and CCL5 (for review see [
47]), with the decrease in CCL4 being particularly dramatic.
In addition to decreasing the number of microglial cells, treatment with rAAV.EpoR76E modulated the microglial response to glaucoma in the DBA/2J. EPO is anti-apoptotic and thus could limit neuroinflammation by blocking cell death (for review see [
12]). However, this is unlikely to explain our results since the analyses were performed at a time-point prior to the onset of significant RGC death [
18]. The decrease in levels of pro-inflammatory cytokines/chemokines in retinas from rAAV.EpoR76E-treated mice could be due to the presence of fewer microglia. Alternatively, it could reflect decreased production from macroglial cells (astrocytes and Müller cells) or the remaining microglial cells. The morphometric analysis shows that treatment with rAAV.EpoR76E did not completely block microglial reactivity. We posit that this is an advantage of EPO therapy since in other models, blocking the microglial response entirely can worsen neurodegeneration while modulating it is neuroprotective (for review see [
48]).
The lack of difference in TNFα levels between the two treatment groups was unexpected considering EPO has been reported to decrease levels of TNFα in the lipopolysaccharide model of ocular inflammation [
49,
50]. It is possible that glaucoma might induce TNFα production through a different signaling pathway than that activated by lipopolysaccharide [
49,
50]. The lack of correlation between microglia number and TNFα levels suggests that the majority of TNFα in the glaucomatous retina may be produced by reactive macroglia rather than microglia. This is supported by a previous study that demonstrated production of TNFα from primary retinal astrocytes and Müller cells under stress conditions [
51]. This, in turn, may suggest that systemic gene delivery of EPO.R76E had little to no effect on the reactivity of these macroglial cells. We have previously demonstrated decreased glial reactivity after intraocular, but not systemic, delivery of EPO suggesting a possible dose-dependent effect (our unpublished data) [
45]. In addition, the detection of neuroprotection by EPO.R76E despite a lack of decrease in TNFα levels suggests that EPO.R76E acts downstream or independent of the TNFα pathway.
TNFα can activate the caspase cell death cascade, induce mitochondrial dysfunction, and cause oxidative damage [
51,
52]. Since the current study was performed prior to initiation of cell death in this model, that mechanism is less relevant [
18]. EPO can decrease oxidative stress, at least in part, through activation of Nrf-2 and the ARE (for review see [
11]) [
34]. We detected increased expression of several antioxidant enzymes and proteins in the retinas of rAAV.EpoR76E-treated mice including those involved in mitochondrial function (Nqo1, Nudt1, Gpx). Therefore, independent of modulation of neuroinflammation, EPO.R76E may protect against glaucomatous neurodegeneration by directly counteracting oxidative stress. If this is the major pathway for neuroprotection by EPO.R76E, then increasing expression of Nrf2 may be just as effective. In support of this approach, others have shown that chemically induced activation of Nrf2 protects RGCs in the optic nerve crush model of RGC death [
53].
Finally, hypoxia has been implicated as a major contributor to glaucoma pathogenesis [
8]. The hypothesis states that damage to the microvasculature at the optic nerve head leads to hypoxic events. This hypothesis is supported by studies showing an increase in levels of HIF1α in the vitreous of glaucoma patients and in animal models [
54,
55]. A major target of HIF1α is EPO and both hypoxic pre-conditioning and systemic treatment with EPO are protective to retinal neurons [
56,
57]. EPO expression is also increased in glaucoma patients and in a rat model of ocular hypertension suggesting that it may represent an endogenous protective mechanism [
58‐
62]. Systemic exogenous EPO can increase tissue oxygen delivery as a result of increased erythropoiesis. We detected an increase in H-ferritin immunofluorescence in the retinas of rAAV.EpoR76E-treated mice suggesting that although erythropoiesis was significantly attenuated as compared to that induced by wild-type EPO, it was still sufficient to increase blood flow to the retina.
Competing interests
TSR is a co-inventor in a pending US patent application (13/979,451) and international patent application (PCT/2012/021247) regarding neuroprotective use of EPO.R76E. No commercialization has occurred. No other authors have conflicts of interest to disclose.
Authors’ contributions
JHB carried out the confocal microscopy, morphometric, axon transport, and visual function studies and edited the manuscript. WSB assisted with the implementation of the molecular studies, performed the longitudinal hematocrit experiment and the fluorescein angiography, and edited the manuscript. JRB performed the molecular studies. TSR conceived and designed the study and wrote the manuscript. All authors read and approved the final manuscript.