A Cell-Permeable Phosphine-Borane Complex Delays Retinal Ganglion Cell Death After Axonal Injury Through Activation of The Pro-Survival extracellular signal-regulated kinases 1/2 Pathway

Abstract

The reactive oxygen species superoxide has been recognized as a critical signal triggering retinal ganglion cell (RGC) death after axonal injury. Although the downstream targets of superoxide are unknown, chemical reduction of oxidized sulfhydryls has been shown to be neuroprotective for injured RGCs. Based on this, we developed novel phosphine-borane complex compounds that are cell permeable and highly stable. Here, we report that our lead compound, bis (3-propionic acid methyl ester) phenylphosphine borane complex 1 (PB1), promotes RGC survival in rat models of optic nerve axotomy and in experimental glaucoma. PB1-mediated RGC neuroprotection did not correlate with inhibition of stress-activated protein kinase signaling, including ASK1, JNK or p38. Instead, PB1 led to a striking increase in retinal BDNF levels and downstream activation of the ERK1/2 pathway. Pharmacological inhibition of ERK1/2 entirely blocked RGC neuroprotection induced by PB1. We conclude that PB1 protects damaged RGCs through activation of pro-survival signals. These data support a potential cross-talk between redox homeostasis and neurotrophin-related pathways leading to RGC survival after axonal injury.

Introduction

Axonal injury is a common cause of neuronal death in the central nervous system (CNS) of adult mammals and is the primary damaging event in most optic nerve diseases, including glaucoma. A crucial element in the pathophysiology of optic neuropathies is the death of retinal ganglion cells (RGCs), the neurons that convey visual information from the retina to the brain. The signals leading to RGC loss in glaucoma are not well understood (Nickells 2007, Wax & Tezel 2002), but several mechanisms have been proposed, including neurotrophic factor deprivation, mechanical compression, excitotoxicity, reactive astrocytosis and induction of pro-apoptotic pathways (Mansour-Robaey et al. 1994, Cui & Harvey 1995, Pearson & Thompson 1993, Carpenter et al. 1986, Shen et al. 1999, Yoles et al. 1997, Stys et al. 1990, Kiryu-Seo et al. 2000, Kikuchi et al. 2000). The relationship between these processes is complex and it is likely that more than one signal leads to RGC death induced by axonal damage.

The hypothesis that neurotrophin deprivation contributes to RGC death after axonal injury has received considerable attention because a lack of target-derived brain-derived neurotrophic factor (BDNF) or nerve growth factor (NGF) leads to apoptotic death of developing RGCs (Chau et al. 1992, Nurcombe & Bennett 1981, Rabacchi et al. 1994, Thoenen et al. 1987). Although the role of neurotrophins in the maintenance of adult RGCs is less clear, there is substantial evidence showing that administration of exogenous BDNF promotes robust RGC survival in a variety of optic nerve injury paradigms (Mey & Thanos 1993, Mansour-Robaey et al. 1994, Peinado-Ramon et al. 1996, Di Polo et al. 1998, Klöcker et al. 2000, Chen & Weber 2001). Upon binding of BDNF to its cognate receptor TrkB, multiple signaling pathways are activated including the extracellular signal-regulated kinases 1/2 (ERK1/2) and the phosphatidylinositol-3 kinase (PI3K)/Akt pathways (Kaplan & Miller 2000). Endogenous activation of ERK1/2 and PI3K has been reported in RGCs in response to BDNF and other protective agents, and pharmacological inhibition of these molecules effectively blocks their survival effect (Cheng et al. 2002, Diem et al. 2001, Kermer et al. 2000, Schallenberg et al. 2009). Furthermore, we previously showed that viral vector-mediated stimulation of ERK1/2 was sufficient to protect RGCs from death induced by axotomy or ocular hypertension (Pernet et al. 2005, Zhou et al. 2005).

Oxidative signaling, caused by the imbalance between the production of reactive oxygen species (ROS) and their elimination by antioxidants, has been recognized as another central contributor to neuronal injury and death. ROS can modulate protein function by altering redox states leading to cysteine sulfhydryl oxidation. Oxidative cross-linking creates new disulfide bonds causing protein conformational changes and subsequent activation of cell death signals (Carugo et al. 2003, Park and Raines, 2001). Consistent with this, RGC viability has been shown to depend on the intracellular sulfhydryl redox state, with survival observed under mildly reducing conditions and increased death rates induced by sulfhydryl oxidation (Castagne & Clarke 1996, Castagne et al. 1999, Geiger et al. 2002, Swanson et al. 2005).

We recently demonstrated that ROS superoxide is a key signal triggered by axonal injury leading to RGC apoptosis. Using live imaging, we showed that there is a marked elevation of superoxide in RGCs soon after optic nerve axotomy, and that a decrease in intracellular superoxide levels delays RGC death in vivo (Kanamori et al. 2010). Based on this, we hypothesized that reduction of oxidized sulfhydryls on critical proteins might attenuate the activation of death pathways that influence the fate of RGCs after injury. To test this, we developed reducing agents using a borane-protected phosphine backbone (Schlieve et al. 2006). Here we characterize a leading compound, bis (3-propionic acid methyl ester) phenylphosphine borane reducing complex 1 (PB1), and show that PB1 promotes RGC protection in rat paradigms of optic nerve injury. We demonstrate that, rather than inhibiting cell death pathways, PB1 leads to increased retinal levels of BDNF and that PB1-mediated RGC neuroprotection requires activation of ERK 1/2 in vivo. Our data support the conclusion that the reducing agent PB1 protects injured RGCs through activation of pro-survival pathways, and suggest a potential cross-talk between intracellular redox regulation and activation of neurotrophin-related neuroprotective signals in retinal neurons.

Materials and Methods

Experimental Animals

All procedures were carried out in accordance with the Animal Research: Reporting In Vivo Experiments (ARRIVE) and the Canadian Council on Animal Care guidelines. The optic nerve axotomy model, a paradigm of acute axonal damage and RGC death, was carried out in adult Sprague-Dawley rats (Charles River, 180-200 g). The experimental glaucoma model, induced by ocular hypertension (OHT) surgery, was performed in retired breeder Brown Norway rats (Charles River, Canada; 300-400 g). Brown Norway rats were used for the experimental glaucoma model because they have a larger eye suitable for the OHT surgical procedure (Johnson et al. 1996, Morrison et al. 1997). The number of animals used in each experiment (n) is indicated above the bar in the corresponding graph.

RGC Retrograde Labeling

For quantification of neuronal survival, RGCs were retrogradely labeled with Fluorogold (2%, Fluorochrome, Englewood, CO) or DiI (3%, 1,1′-dioctadecyl-3,3,3′,3′-tetramethyl-indocarbocyanine perchlorate; Molecular Probes, Junction City, OR). Tracers were dissolved in 0.9% NaCl containing 10% dimethyl sulfoxide (DMSO) and 0.5% Triton X-100. The superior colliculus was exposed bilaterally and a small piece of gelfoam (Pharmacia and Upjohn Inc., Mississauga, ON) soaked in tracer was applied to the surface. Seven days is the earliest time for detection of the entire RGC population after application of retrograde tracers to the rat superior colliculus (Vidal-Sanz et al. 1988). Thus, to ensure that all RGCs were fully labeled prior to axonal injury, axotomy or OHT surgery were performed at 7 days after tracer application.

Optic Nerve Injury Paradigms

i) Optic nerve axotomy

Animals were deeply anesthetized (2% isoflurane, 0.8 liter/min) and the left optic nerve was carefully exposed within the dura and transected ∼1 mm posterior to the globe. This procedure avoided injury to the ophthalmic artery and its branches. Fundus examination was performed immediately after axotomy and 3-5 days later to check the integrity of the retinal circulation after surgery. Animals showing signs of compromised blood supply were excluded from the study.

ii) Ocular hypertension (Morrison model)

Animals were anesthetized by intraperitoneal injection of 1 ml/kg of standard rat cocktail (100 mg/ml ketamine, 20 mg/ml xylazine, 10 mg/ml acepromazine). Unilateral elevation of intraocular pressure (IOP) was induced by a single injection of hypertonic saline solution (1.85 M NaCl) into an episcleral vein as previously described (Morrison et al. 1997), a procedure called OHT surgery. A plastic ring was applied to the ocular equator to confine the injection to the limbal plexus. Animals were kept in a room with constant low fluorescent light (40-100 lux) to stabilize circadian IOP variations (Moore et al. 1996, Jia et al. 2000b). IOP was measured using a calibrated tonometer (TonoPen XL, Medtronic Solan, Jacksonville, FL) in awake animals to avoid the anesthetic-induced reduction of IOP (Jia et al. 2000a)

Phosphine-Borane Complex Synthesis

Phosphine-borane complex 1 (PB1) was synthesized according to previously published methods (Schlieve et al. 2006). Briefly, the intermediate bis (3-propionic acid methyl ester) phenylphosphine (Phosphine 1) was produced by adding potassium hydroxide to phenylphosphine dissolved in acetonitrile, cooling to 0°C, and then slowly adding methyl acrylate, maintaining the temperature below 35°C. The reaction product was heated at 50°C for 8 h, washed and dried over MgSO₄, then concentrated and purified by distillation as a clear liquid. Phosphine 1 was dissolved in tetrahydrofuran (THF) and cooled to 0°C. Borane-THF was slowly added and allowed to react. The solvent was then removed under reduced pressure and the residue purified by flash chromatography producing PB1.

In Vivo Drug Delivery

PB1 (150 μM) or the mitogen activated protein (MAP) kinase kinase 1 (MEK1) inhibitor PD98059 (200 μM, Sigma, Oakville, ON) were dissolved in phosphate buffered saline (PBS) containing 0.1% DMSO (vehicle). PB1, PD98059 or vehicle were injected into the vitreous chamber of the injured eye using a Hamilton syringe fitted with a 32-gauge glass microneedle. We selected a PB1 concentration of 150 μM administered in a 4 μl volume, which yields an estimated final intravitreal concentration of 10 μM (approximate vitreous volume in rats: 60 μl), based on our previous in vitro study showing that this amount is an effective neuroprotective dose (Schlieve et al. 2006). The sclera was exposed and the tip of the needle was inserted at a 45° angle through the sclera and retina into the vitreous space using a posterior approach. This route of administration avoided injury to the iris or lens, which can promote RGC survival (Leon et al. 2000, Mansour-Robaey et al. 1994). The injection was performed within ∼30 sec, after which the needle was gently removed. Some animals received two consecutive injections of PB1 and PD98059 or vehicle, through the same injection site, with a delay of 20 min between each injection. Surgical glue (Indermill, Tyco Health Care, Mansfield, MA) was used to seal the injection site.

Quantification of RGC soma and axons

Quantification of RGC bodies or axons was performed in duplicate by an observer masked to the treatment assignments. For RGC density counts, rats were deeply anesthetized and perfused transcardially with 4% paraformaldehyde (PFA) and both eyes were immediately enucleated. Retinas were dissected and flat-mounted on a glass slide with the ganglion cell layer side up. RGCs were counted in three square areas at distances of 1, 2 and 3 mm from the optic disc in each of the four retinal quadrants (superior, inferior, nasal and temporal) for a total of 12 retinal areas encompassing a total area of 1 mm². For axon counts, animals received a transcardial injection of heparin (1000 U/kg) and sodium nitroprusside (10 mg/kg), followed by perfusion with 2% PFA and 2.5% glutaraldehyde. Optic nerves were dissected, fixed in 2% osmium tetroxide, and embedded in Epon resin. Semi-thin sections (0.7-μm thick) were cut on a microtome (Reichert, Vienna, Austria) and stained with 1% toluidine blue. RGC axons were counted at 1 mm from the optic nerve head in five non-overlapping areas (center, peripheral dorsal and peripheral ventral) encompassing a total area of 5.5 mm² per nerve. The total area per optic nerve cross-section was measured using Northern Eclipse image analysis software (Empix Imaging, Toronto, ON), and this value was used to estimate the total number of axons per optic nerve.

Western blot analysis

Whole fresh retinas (n=4 per condition) were rapidly dissected and homogenized with an electric pestle (Kontes, Vineland, NJ) in ice-cold lysis buffer: 50 mM Tris (pH 7.4), 1 mM EDTA, 150 mM NaCl, 1% NP-40, 5 mM Na fluoride, 0.25% Na deoxycholate and 2 mM NaVO₃ supplemented with protease and phosphatase inhibitors. Retinal extracts (60–150 μg) were resolved on 10-15% SDS polyacrylamide gels and transferred to nitrocellulose membranes (Bio-Rad Life Science, Hercules, CA, USA). Non-specific binding was blocked by incubation in 10 mM Tris (pH 8.0), 150 mM NaCl, 0.2% Tween 20 (TBST), and 5% bovine serum albumin (Fisher Scientific, Fair Lawn, NJ) for 1 h at room temperature (20°C). Membranes were then incubated with the following primary antibodies: BDNF (1 μg/ml, Promega, Madison, WI), phospho-ERK1/2 (Thr185/Tyr187, 1 μg/ml, Invitrogen-BioSource, Carlsbad, CA), ERK1/2 (1 μg/ml, Invitrogen-BioSource), phospho-Akt (Thr308, 0.14 μg/ml, Cell Signaling, Danvers, MA), Akt (0.2 μg/ml, Cell Signaling), phospho-ASK1 (Thr838, 0.5 μg/ml, Cell Signaling), ASK1 (0.5 μg/ml, Cell Signaling), phospho-JNK (Thr183/Tyr185, 0.8 μg/ml, Cell Signaling), JNK (0.4 μg/ml, Cell Signaling), phospho-p38 (Thr180/Tyr182, 0.3 μg/ml, Cell Signaling), p38 (0.2 μg/ml, Cell Signaling), or ß-actin (0.5 μg/ml, Sigma). Blots were washed in TBST and incubated in the following peroxidase-linked secondary antibodies: anti-mouse or anti-rabbit (0.5 μg/ml, GE Healthcare, Little Chalfont Bucks, UK) or anti-chicken (0.5 μg/ml, Promega). Blots were developed using chemiluminescence reagents (ECL or Plus-ECL, Perkin Elmer Life and Analytical Sciences, Woodbridge, ON) and exposed to autoradiographic film (X-OMAT; Eastman Kodak, Rochester, NY). Densitometric analysis was performed using Scion Image software (Scion Corporation, Frederick, MD) on scanned autoradiographic films obtained from a series of 3 independent western blots each carried out using retinal samples from distinct experimental groups. The densitometric values obtained for BDNF were normalized with respect to their ß-actin loading controls in the same blot to obtain the final ratios. The densitometric values for phosphorylated (active) proteins were normalized with respect to their loading (non-phosphorylated) controls in the same blot to obtain the final phosphorylated/total protein ratios.

Statistical analysis

Data analysis and statistics were carried out using GraphPad InStat software (GraphPad Software Inc., San Diego, CA) by performing one-way analyses of variance (ANOVA) followed by Bonferroni multiple comparison post-hoc testing.

Results

Intraocular Delivery of the Phosphine-Borane Compound PB1 Protects RGCs from Axotomy-Induced Death

PB1, an analogue of tris (2-carboxyethyl) phosphine (TCEP), was designed to contain a borane-protected phosphine group to prevent oxidation, thus enhancing the stability of the molecule (Schlieve et al. 2006). The phenyl group in PB1 is non-polar and increases the cell permeability of this compound. Once inside the cell, the methyl esters are cleaved by intracellular esterases resulting in an anionic molecule that is less likely to exit the cytosol (Fig.1). We previously demonstrated that PB1-mediated inhibition of sulfhydryl oxidation protects early postnatal, acutely axotomized RGCs in vitro (Schlieve et al. 2006), but the role of PB1 on the survival of adult RGCs in vivo was not established.

FIGURE 1. Chemical Structure of Bis (3-Propionic Acid Methyl Ester) Phenylphosphine Borane Reducing Complex 1 (PB1).

The borane group protects the phosphine from oxidation increasing chemical stability during storage and before administration. The non-polarity of the phosphine-borane and the phenyl group contribute to the ability of PB1 to readily cross cell membranes. The methyl esters are cleaved by extracellular amines and/or intracellular esterases, resulting in an anionic molecule that is unlikely to exit the cytosol and thus forms a strong intracellular concentration gradient.

To investigate this, we first examined PB1-induced RGC survival following axotomy of the optic nerve, an injury modality that leads to rapid apoptotic RGC death (Berkelaar et al. 1994). Eyes that received an intraocular injection of PB1 showed robust RGC neuroprotection compared to control eyes injected with vehicle (Fig. 2A-C, Table 1). Previous studies, including ours, have demonstrated that virtually all RGCs survive for 4 to 5 days after axotomy and then die rapidly: the RGC population is reduced to ∼50% by day 7 and to ∼10% by day 14 post-lesion (Berkelaar et al. 1994, Cheng et al. 2002, Mansour-Robaey et al. 1994). Figure 2D shows that in PB1-treated eyes, 66% of RGCs survived at one week after axotomy (1,434 ± 37 RGCs/mm², mean ± S.E.M., n=6) compared to only 47% remaining in vehicle-treated eyes (1,011 ± 37 RGCs/mm², n=4) (ANOVA, P < 0.001). This neuroprotective effect was still substantial at 2 weeks after axotomy following PB1 treatment at the time of axotomy and one week later, accounting for 25% of RGC survival (533 ± 90 RGCs/mm², n=4) compared to 11% survival afforded by vehicle (239 ± 25 RGCs/mm², n=4) (ANOVA, P < 0.01). These data indicate that the reducing agent PB1 promotes adult RGC neuroprotection following acute optic nerve injury. Microglia and macrophages, which may have incorporated Fluorogold after phagocytosis of dying retinal ganglion cells, were excluded from our analysis based on well-established morphological criteria (Kacza & Seeger 1997, Thanos 1991). Microglia were identified by their invariably smaller cell size, visible process ramifications, and lack of axons (Figs. 2E, 2F) as previously described by us (Lebrun-Julien et al. 2009).

FIGURE 2. The Phosphine-Borane Compound PB1 Protects RGCs from Axotomy-Induced Death.

Retinal flat mounts show Fluorogold-labeled RGCs from non-injured, non-treated eyes (A, Intact) and axotomized PB1-treated (B) or vehicle-treated (C) eyes. Scale bars (A-C) =100 μm. (D) Quantitative analysis of RGC survival following injection of PB1 (solid bars) or vehicle (hatched bars) at the time of axotomy (n=4-6 rats/group) (ANOVA, ***: P<0.001). Animals examined at two weeks received an injection at the time of axotomy and a week later. The density of RGCs in intact, non-injured Sprague-Dawley rat retinas is shown as reference (open bar, 100%, n=8). Data are expressed as RGCs/mm² (mean ± S.E.M). (E, F) Microglia and macrophages (arrowheads) that may have incorporated Fluorogold after phagocytosis of dying retinal ganglion cells (arrows) were excluded from our analysis based on their distinct morphology. Scale bars: E=100 μm, F=10 μm.

Table 1. PB1-induced RGC soma and axonal survival in axotomy and ocular hypertension models.

Modality of optic nerve damage	Time after injury	Treatment	RGCs/mm2 (Mean ± S.E.M.)	RGC axons (Mean ± S.E.M.)
Axotomy	1 week	PB1	1434 ± 37 (n=6)	-
		Vehicle	1011 ± 37 (n=4)	-
	2 weeks	PB1	533 ± 90 (n=4)	-
		Vehicle	239 ± 25 (n=4)	-
Ocular hypertension	3 weeks	PB1	1484 ± 36 (n=6)	70058 ± 4547 (n=9)
		Vehicle	1072 ± 64 (n=6)	55997 ± 4531 (n=6)
	5 weeks	PB1	687 ± 35 (n=6)	311136 ± 5132 (n=7)
		Vehicle	598 ± 77 (n=6)	304158 ± 4673 (n=6)

PB1 Protects RGC Soma and Axons in Experimental Glaucoma

To determine if PB1 was able to promote RGC survival in a paradigm of optic nerve injury resembling glaucomatous pathophysiology, we tested its neuroprotective effect in a rat ocular hypertension (OHT) model. Gradual increase of eye pressure and progressive death of RGCs are observed in this model, with an excellent linear correlation between IOP increase and RGC loss (Chauhan et al. 2002, Johnson et al. 1996, Morrison et al. 1997). Inner retinal atrophy, optic nerve degeneration, and optic nerve head remodeling in this model are similar to those seen in human glaucoma, therefore this model is considered a premier in vivo paradigm of this optic neuropathy. PB1 was injected intravitreally two weeks after OHT surgery to allow for IOP stabilization and RGC survival was examined at 3 or 5 weeks after OHT. Analysis of DiI-positive RGCs in retinal whole mounts showed that PB1 led to higher neuronal densities in glaucomatous eyes compared to control eyes at 3 weeks after OHT (Figure 3A-C, Table 1). Quantitative analysis of RGC neuroprotection demonstrated that 82% of RGCs survived in the presence of PB1 (1,484 ± 36 RGCs/mm², mean ± S.E.M., n=6) compared to 59% in control eyes treated with vehicle (1,072 ± 64 RGCs/mm², n=6) (Fig. 3D, ANOVA, p < 0.001). The mean sustained IOP elevation in PB1- and vehicle-treated eyes was similar, allowing for a reliable comparison between these groups.

FIGURE 3. PB1 Protects RGC Soma in Experimental Glaucoma.

Retinal flat mounts show DiI-labeled RGCs from non-injured, non-treated eyes (A, Intact) and glaucomatous PB1-treated (B) or vehicle-treated (C) eyes. Scale bars=100 μm. (D) Quantitative analysis of RGC survival following injection of PB1 (solid bars) or vehicle (hatched bars) at three weeks after ocular hypertension surgery (OHT) (n=6 rats/group) (ANOVA, ***: P<0.001). The density of RGCs in intact, non-injured Brown Norway rat retinas is shown as reference (open bar, 100%, n=6). Data are expressed as RGC densities (RGCs/mm², mean ± S.E.M).

Glaucoma is characterized by the degeneration of RGC axons in the optic nerve followed by the progressive loss of cell bodies (Quigley 1999, Schwartz et al. 1999), hence we also investigated the effect of PB1 on RGC axonal protection following ocular hypertensive damage. Analysis of axons in optic nerves treated with PB1 at 3 weeks after OHT demonstrated a higher number of RGC axons with normal morphology compared to vehicle-treated optic nerves, which featured extensive disarray of fascicular organization and degradation of myelin sheaths (Figure 4A-C). Axonal quantification in optic nerve cross sections showed that PB1 protected a significant number of RGC axons from glaucomatous damage (69% = 70,058 ± 4,547 axons, n=10) compared to vehicle-treated controls (55%= 55,997 ± 4,531 axons, n=6) (Fig. 4D, ANOVA, p < 0.001). Although a slight trend in RGC soma and axon protection was observed at 5 weeks after OHT (Table 1), this effect was not statistically significant suggesting that the biological activity of a single dose of PB1 has a limited duration in vivo. Collectively, these results indicate that PB1 attenuates the loss of both RGC soma and axons in experimental glaucoma.

FIGURE 4. PB1 Attenuates Axonal Loss in Experimental Glaucoma.

Cross-sections of optic nerve segments from non-injured eyes (A, Intact) and glaucomatous eyes treated with PB1 (B) or vehicle (C) at 3 weeks after ocular hypertension surgery (OHT). PB1-treated eyes displayed a larger number of axonal fibers with normal morphology compared to vehicle-treated control eyes, which showed extensive axon degeneration. Scale bars= 20 μm. (D) Quantitative analysis of RGC axons in optic nerves after treatment with PB1 (solid bar), or vehicle (hatched bar) (n=6-10 rats/group) (ANOVA, ***: P<0.001). The number of axons in the non-injured Brown Norway rat optic nerve is shown as reference (open bar, 100%, n=9). Data are expressed as the total number of RGC axons per optic nerve (mean ± S.E.M.).

PB1-Mediated RGC Neuroprotection Requires Activation of the Extracellular Signal-Regulated Kinases 1/2 Pathway

Oxidative stress has been linked to the activation of stress-activated protein kinase (SAPK) signaling and subsequent cell death (Cross & Templeton 2004, Sumbayev & Yasinska 2005). To gain mechanistic insight into how PB1 promoted RGC neuroprotection in vivo, we asked whether PB1 leads to inhibition of pro-apoptotic pathways. We chose the axotomy model for these experiments because the onset of RGC death in this injury paradigm is extremely consistent, starting at 4-5 days after optic nerve lesion (Berkelaar et al. 1994). This predictable time-course of RGC loss allowed us to examine protein changes prior to neuronal death (24 hrs), which are more likely to influence RGC fate. Furthermore, a well-defined burst of superoxide occurs within 24 hrs of optic nerve axotomy (Kanamori et al. 2010).

We first examined the activation of retinal Apoptosis Stimulating Kinase 1 (ASK1), a SAPK and mitogen-activated protein kinase kinase kinase (MAPKKK) family member, which is activated by ROS and has been shown to mediate RGC death (Harada et al. 2006, Harada et al. 2010). ASK1 is normally bound to reduced thioredoxin, a protein disulfide oxidoreductase that prevents ASK1 autophosphorylation. Oxidation of cysteine thiols in thioredoxin results in its dissociation from ASK1, triggering ASK1 autophosphorylation and downstream stimulation of c-Jun NH2-terminal kinase (JNK) and p38 death signaling (Hatai et al. 2000, Ichijo et al. 1997, Saitoh et al. 1998). If PB1 exerted RGC neuroprotection via the regulation of ASK1, a decrease in phosphorylated ASK1 (P-ASK1) following PB1 treatment would be expected. Western blot analysis demonstrated low but detectable levels of phosphorylated ASK1 in intact (non-injured, non-treated) retinas (Fig. 5A). An increase in phospho-ASK1 was observed in control, axotomized eyes treated with vehicle, however, PB1 failed to significantly reduce the levels of activated ASK1. Consistent with this, the levels of ASK1 downstream effectors JNK (P-JNK, Fig. 5B) or p38 (P-p38, Fig. 5C) were not affected by PB1. These results suggest that PB1-mediated RGC neuroprotection does not involve the ASK1 pathway.

FIGURE 5. The Pro-Apoptotic ASK1 Signaling Pathway Is Not Regulated by PB1.

Western blots of total retinal extracts probed with antibodies that selectively recognize phosphorylated (active) ASK1, JNK or p38. Protein samples were collected from non-injured, non-treated eyes (Intact) or axotomized eyes treated with PB1 or vehicle and collected at 24 hrs post-lesion. (A) An injury-induced increase in phospho-ASK1 (P-ASK1) was observed in control, axotomized eyes treated with vehicle. PB1 failed to significantly reduce the levels of active ASK1 after axotomy. The levels of ASK1 downstream effectors P-JNK (B) or P-p38 (C) were not affected by PB1. The densitometric values are the ratio of phospho-proteins normalized to their loading (non-phosphorylated) controls in the same blot for intact (open bars), PB1-treated (solid bars) or vehicle-treated (hatched bars) eyes (n=4/group) (ANOVA, ***: P<0.001).

An alternative possibility is that PB1 results in the stimulation of pro-survival signals required for RGC viability after injury. To test this hypothesis, we investigated the levels of BDNF and its downstream effectors ERK1/2, Akt and CREB in axotomized retinas exposed to PB1 or vehicle. In control axotomized eyes treated with vehicle there was a slight, but significant, increase in BDNF compared to intact eyes (Fig. 6A), which is consistent with previous reports showing a ∼50% increase in retinal BDNF mRNA after axotomy (Gao et al. 1997, Hirsch et al. 2000). Surprisingly, PB1 led to a 4-fold increase (200%) in BDNF protein levels after axotomy compared to intact retinas. Consistent with this, PB1 produced a robust activation of the BDNF effector ERK1/2 (P-ERK1/2, Fig. 6B) while Akt and the transcription factor CREB remained unchanged (P-Akt and P-CREB, Figs. 6C, D). Intraocular administration of PB1 at 2 weeks after OHT also resulted in enhanced ERK1/2 activation (Fig. 6E) suggesting that PB1 promotes RGC survival through activation of this pathway after acute and chronic optic nerve injury.

FIGURE 6. PB1 Increases Retinal BDNF and Activates ERK1/2.

(A) PB1 led to a 4-fold increase in BDNF protein levels after axotomy compared to intact retinas. (B) Robust activation of the BDNF effector ERK1/2 was observed in PB1-treated axotomized eyes (P-ERK1/2), while Akt (P-Akt, C) and CREB (P-CREB, D) remained unchanged. (E) Intraocular administration of PB1 at 2 weeks after OHT also resulted in enhanced retinal ERK1/2 activation. The densitometric values are the ratio of phospho-proteins normalized to their loading (non-phosphorylated) controls in the same blot, or ß-action in the case of BDNF, for intact (open bars), PB1-treated (solid bars) or vehicle-treated (hatched bars) eyes (n=4-5/group) (ANOVA, *p < 0.05, ***: P<0.001).

To establish whether ERK1/2 signaling was involved in PB1-mediated survival of axotomized RGCs, we co-injected PB1 with PD98059, a pharmacological inhibitor of MEK1, the obligate upstream activator of ERK1/2 (Dudley et al., 1995). We previously established that the optimal dose of PD98059 to selectively inhibit retinal Erk1/2 in vivo without affecting other pathways, including Akt, is 200 μM (16.7 μM intravitreal concentration) (Cheng et al. 2002). Figure 7 shows that co-administration of PB1 and PD98059 resulted in complete inhibition of the survival effect produced by PB1, characterized by low RGC densities similar to those found in vehicle-treated retinas, at 1 week after optic nerve transection. Together, these findings demonstrate that the ERK1/2 pathway is essential for PB1-mediated survival of injured adult RGCs in vivo.

FIGURE 7. PB1-Mediated RGC Neuroprotection Requires Activation of ERK1/2.

(A-C) Retinal flat mounts show that co-administration of PB1 and the MEK1 inhibitor PD98059, injected intravitreally at the time of optic nerve transection, resulted in inhibition of the survival effect produced by PB1 at 1 week after optic nerve transection. (D) Quantitative analysis of Fluorogold-labeled neurons in eyes treated with PB1 and PD98059 (gray bar) showed that RGC density at 1 week post-lesion was similar to that found in control retinas treated with vehicle (hatched bar) (n=4-6/group) (ANOVA, ***: P<0.001). Data are expressed as RGCs/mm² (mean ± S.E.M).

Discussion

The generation of an intracellular superoxide burst is a critical molecular event underlying RGC death after axonal injury (Geiger et al. 2002, Kanamori et al. 2010, Lieven et al. 2006, Nguyen et al. 2003, Swanson et al. 2005). Superoxide increases dramatically in RGCs at the single-cell level, soon after optic nerve axotomy, and precedes RGC apoptosis (Kanamori et al. 2010). Human glaucomatous retinas contain high levels of the lipid peroxidation indicator 4-hydroxy-2-nonenal (HNE), which leads to protein modification induced by superoxide (Tezel et al. 2010). Administration of pegylated superoxide dismutase-1 (SOD), which catalyzes the dismutation of superoxide into oxygen and hydrogen peroxide (H₂O₂), attenuates RGC death (Kanamori et al. 2010, Schlieve et al. 2006) supporting the idea that interfering with superoxide generation might be beneficial. However, the translation of a protein-based therapy that requires intracellular delivery is considerably more challenging than a small molecule approach. In this study, we characterized the neuroprotective role and mechanism of action of PB1, a small reducing compound with several advantages including good cell permeability, the ability to form a high intracellular concentration gradient, and stability.

Our data demonstrate that intraocular delivery of PB1 promotes RGC survival in vivo following traumatic optic nerve injury (axotomy) and ocular hypertension damage (experimental glaucoma). The finding that PB1-mediated neuroprotection was observed in these distinct injury paradigms, despite the fact that the RGC response to different types of lesion may vary widely, suggests that PB1 regulates a conserved pathway and underlines its translational potential to human disease. Glaucoma has been defined as an axogenic disease, characterized first by the degeneration of RGC axons in the optic nerve followed by the progressive loss of cell bodies (Schwartz et al. 1999). In the experimental glaucoma model, we performed quantitative analysis of the neuroprotective effect of PB1 on two major RGC compartments: soma and axons. Consistent with the idea that the primary site of degeneration in glaucoma is at the level of the axon, we found that all eyes had more pronounced axonal loss than cell body loss. However, intraocular injection of PB1 protected a similar proportion of RGC soma and axons within the optic nerve at 3 weeks after OHT. The ability to protect all RGC compartments following hypertension damage is paramount for the preservation of neuronal function and vision; hence it is an important attribute of PB1. Interestingly, functional studies in macaque monkeys subjected to experimental glaucoma demonstrated that only subtle visual field defects are detected despite massive loss (>50%) of RGCs, whereas vision loss increases dramatically with more advanced glaucoma (Harwerth et al. 1999). Therefore, structural protection of a proportion of RGC soma and axons, as afforded by PB1, might be sufficient to preserve functional vision. The lack of significant soma or axon protection at 5 weeks after OHT suggests that a single dose of PB1 confers limited biological activity in vivo. A priority of future studies will be to devise sustained delivery strategies, such as PB1 coupled to nanoparticles, to achieve long-term neuroprotection.

What are the molecular mechanisms underlying PB1-mediated RGC survival? Evidence from studies on cell death inhibition induced by manipulation of the mitochondrial electron transport chain is consistent with PB1 acting externally to the mitochondrial matrix (Seidler et al. 2010). Phosphines might scavenge superoxide directly, but our studies with both borane-protected phosphines and deprotected PB1 have ruled out significant superoxide scavenging (Niemuth et al, unpublished data). The redox system can regulate the function of proteins involved in cell death and survival by modifying gene expression, posttranslational modifications (e.g. phosphorylation) and stability. In most cases, superoxide stimulates stress-activated protein kinase (SAPK) signaling and cell apoptosis (Sumbayev & Yasinska 2005). Therefore, we hypothesized that PB1 might promote survival through inhibition of pro-apoptotic pathways. ASK1, a crucial redox sensor for initiation of the SAPK signaling cascade, leads to JNK and p38 stimulation and subsequent cell death (Kyriakis & Avruch 2001). Contrary to our expectations, PB1 did not reduce the levels of phosphorylated (active) ASK1, JNK or p38 in axotomized retinas. Thus, we conclude that the regulation of the SAPK cell death signaling pathway is not a target for PB1-induced neuroprotection.

We then considered an alternative scenario involving PB1-induced modification of RGC survival pathways. PB1 stimulated a robust increase of retinal BDNF levels that was several-fold higher than that observed in control axotomized eyes. Emerging data supports a tight redox regulation of transcription factors that encode cell survival proteins (Trachootham et al. 2008). The transcriptional regulation of BDNF is complex and often depends on activity-driven events that involve Ca⁺²-responsive elements and cAMP-responsive elements (CRE) required for promoter transactivation (Shieh et al. 1998, Tao et al. 1998). The cAMP response element binding protein (CREB) transcription factor is of interest because, while it regulates BDNF gene expression, it also responds to BDNF by stimulating the transcription of pro-survival molecules such as Bcl-2 (Bonni et al. 1995, Finkbeiner et al. 1997, Wilson et al. 1996). Moreover, CREB plays a role in the regulation of ROS detoxification (Herzig et al. 2001, Krönke et al. 2003, Lee et al. 2009) and it is susceptible to redox regulation (Bedogni et al. 2003). PB1 failed to increase CREB activation, suggesting that other mechanisms including CREB-independent transcription, stability, subcellular localization, and translational events may underlie PB1-induced BDNF upregulation.

BDNF binds to its signaling receptor TrkB, which is abundantly expressed by adult RGCs (Jelsma et al. 1993, Pérez & Caminos 1995, Rickman & Brecha 1995), and activates the pro-survival ERK1/2 and Akt pathways. Our data demonstrate that ERK1/2, but not Akt, was activated following PB1 administration. This finding is consistent with our previous observation that combined BDNF and TrkB upregulation promoted RGC survival exclusively via ERK1/2, while Akt was not involved (Cheng et al. 2002). It is possible that endogenous BDNF leads to differential activation of downstream pathways depending on the redox status of the cell. In PB1-treated retinas, BDNF might selectively use the ERK1/2 pathway to promote RGC neuroprotection. In addition, PB1 might directly activate upstream molecules that converge on ERK1/2. This latter possibility is supported by the fact that autophosphorylation and activation of tyrosine kinase receptors, such as TrkB, can occur by direct thiol modification of the receptor (Chen et al. 1998). Similarly, the activity of Ras, an upstream activator of ERK1/2, is modulated by redox regulation (Lander et al. 1996, Mallis et al. 2001). Nonetheless, the complete inhibition of RGC survival exerted by PB1 in the presence of PD98059 strongly supports our hypothesis that ERK1/2 activity is essential for PB1-mediated RGC neuroprotection in vivo.

In summary, we demonstrate that PB1, a novel phosphine-borane complex, promotes RGC neuroprotection in vivo through activation of the ERK1/2 pathway. BDNF is a potent anti-apoptotic factor for RGCs, but its clinical application has been hampered due to pleiotropic effects leading to non-specific signaling, potential toxicity and low diffusion rates (Barinaga 1994, Verrall 1994). The identification of small molecule compounds that mimic some of the beneficial effects of BDNF, such as PB1, is of clinical interest. Our study offers the interesting and unexpected possibility that redox homeostasis in RGCs can converge on neurotrophin-related pathways to promote survival after axonal injury.