Rescue of Glaucomatous Neurodegeneration by Differentially Modulating Neuronal Endoplasmic Reticulum Stress Molecules

Abstract

Axon injury is an early event in neurodegenerative diseases that often leads to retrograde neuronal cell death and progressive permanent loss of vital neuronal functions. The connection of these two obviously sequential degenerative events, however, is elusive. Deciphering the upstream signals that trigger the neurodegeneration cascades in both neuronal soma and axon would be a key step toward developing the effective neuroprotectants that are greatly needed in the clinic. We showed previously that optic nerve injury-induced neuronal endoplasmic reticulum (ER) stress plays an important role in retinal ganglion cell (RGC) death. Using two in vivo mouse models of optic neuropathies (traumatic optic nerve injury and glaucoma) and adeno-associated virus–mediated RGC-specific gene targeting, we now show that differential manipulation of unfolded protein response pathways in opposite directions—inhibition of eukaryotic translation initiation factor 2α-C/EBP homologous protein and activation of X-box binding protein 1—promotes both RGC axons and somata survival and preserves visual function. Our results indicate that axon injury-induced neuronal ER stress plays an important role in both axon degeneration and neuron soma death. Neuronal ER stress is therefore a promising therapeutic target for glaucoma and potentially other types of neurodegeneration.

SIGNIFICANCE STATEMENT Neuron soma and axon degeneration have distinct molecular mechanisms although they are clearly connected after axon injury. We previously demonstrated that axon injury induces neuronal endoplasmic reticulum (ER) stress and that manipulation of ER stress molecules synergistically promotes neuron cell body survival. Here we investigated the possibility that ER stress also plays a role in axon degeneration and whether ER stress modulation preserves neuronal function in neurodegenerative diseases. Our results suggest that neuronal ER stress is a general mechanism of degeneration for both neuronal cell body and axon, and that therapeutic targeting of ER stress produces significant functional recovery.

Introduction

CNS axonopathy is a common characteristic not only of acute neural injuries, such as stroke, traumatic brain injury and spinal cord injury, but also of chronic neurodegenerative diseases, including Alzheimer's disease (AD), Parkinson's disease (PD), amyotrophic lateral sclerosis (ALS), hereditary spastic paraplegia, multiple sclerosis (MS), and glaucoma (Coleman and Perry, 2002; Raff et al., 2002; Wang et al., 2012; Conforti et al., 2014). Axonopathies result in axon degeneration, which is followed by retrograde neuronal soma atrophy or death. Preventing axon degeneration is, therefore, critical for minimizing the severe consequences of CNS axonopathies and preserving neuronal function. Extensive study of the slow Wallerian degeneration protein (Wld^s) suggests that signaling pathways that regulate axon degeneration are very different from the apoptosis pathways that regulate neuronal soma degeneration (Whitmore et al., 2005; Calkins and Horner, 2012; Wang et al., 2012; Howell et al., 2013; Conforti et al., 2014). In models of glaucoma, for example, Wld^s delays retinal ganglion cell (RGC) axon degeneration, but has no effect on degeneration of RGC somata (Beirowski et al., 2008), and apoptotic molecule Bax deletion protects RGC somata, but does not prevent axon degeneration (Libby et al., 2005).

The endoplasmic reticulum (ER) is the organelle responsible for synthesis and proper folding of proteins and exerts essential quality control functions. If the ER is overwhelmed by unfolded and misfolded proteins beyond its handling capacity, ER stress develops. The cell then responds to ER stress by activating characteristic signal transduction pathways, which in aggregate are called the unfolded protein response (UPR; Ron and Walter, 2007; Wang and Kaufman, 2012). Three distinct ER-resident proteins sense stress and initiate the UPR pathways: protein kinase RNA-like ER kinase (PERK), inositol-requiring protein-1 (IRE1α), and activating transcription factor-6 (ATF6). PERK phosphorylates and inactivates eukaryotic translation initiation factor 2α (eIF2α) to attenuate general mRNA translation. However, phosphorylated eIF2α (eIF2α-P) induces the expression of a proapoptotic molecule, CCAAT/enhancer-binding protein homologous protein (CHOP), by selectively activating translation of ATF4. IRE1α, a bifunctional enzyme that contains both a Ser/Thr kinase domain and an RNase domain, initiates another UPR pathway. Its RNase activity mediates the splicing of X-box binding protein 1 (XBP-1) mRNA to generate an active (spliced) form of the transcription factor, XBP-1s. XBP-1s targets a set of ER chaperon proteins that may increase ER protein-folding capacity and facilitate degradation of misfolded proteins, which may protect against the detrimental effect of ER stress (Lin et al., 2007). ATF6 is cleaved sequentially by site 1 protease and site 2 protease in Golgi to generate an active transcription factor ATF6 fragment (ATF6f). ATF6 is generally considered cytoprotective, although its downstream effectors have not been completely identified.

ER stress has been linked with many neurodegenerative disorders (Ron and Walter, 2007; Wang and Kaufman, 2012), but appreciation of the clinical importance of neuronal ER stress has just begun (Pennuto et al., 2008; Saxena et al., 2009; Hu et al., 2012; Moreno et al., 2012; Li et al., 2013; Ma et al., 2013b). Previous studies from our laboratory revealed that both acute traumatic injury of the optic nerve (ON) and chronic elevation of intraocular pressure (IOP) induce neuronal ER stress (Hu et al., 2012). We also showed that modulation of two key downstream molecules of ER stress (CHOP and XBP-1) in opposite directions significantly protected the RGC soma in both ON crush model and glaucoma model (Hu et al., 2012). Although neuronal soma and axon degeneration are active autonomous processes with distinct molecular mechanisms, they are clearly connected. In the present study, we demonstrate in two optic neuropathies (traumatic ON injury and glaucoma) that manipulation of the UPR pathways protects both RGC axon and soma and preserves visual function. These results indicate that ER stress is the common upstream signaling mechanism for both neuronal axon and soma degeneration and suggest that targeting ER stress molecules is a promising therapeutic strategy for neuroprotection.

Materials and Methods

Mice.

CHOP KO and C57BL/6 WT mice were purchased from The Jackson Laboratory and kept as separated colonies. eIF2α A/A;fTg mice were described previously (Back et al., 2009). All mice had a C57BL/6 background, and either sex was randomly used in experiments. For all surgical and treatment comparisons, control and treatment groups were prepared together in single cohorts and the experiment repeated at least twice. All experimental procedures were performed in compliance with animal protocols approved by the Institutional Animal Care and Use Committee at Temple University School of Medicine.

Intravitreal injection of adeno-associated virus, ON crush, and induction of chronic IOP elevation.

The procedures have been described previously (Hu et al., 2012; Yang et al., 2014; Miao et al., 2016). The adeno-associated virus (AAV) titers used for this study were in the range of 1.5–2.5 × 10¹² genome copy (GC)/ml determined by real-time PCR. AAVs were injected intravitreally 2 weeks before ON crush or microbead injection, except in delayed experiments. Four copies of different mouse CHOP RNAi sequences identified from the RNAi Consortium (5′-ATTTCATCTGAGGACAGGACC-3′; 5′-CATAGAACTCTGACTGGAATC-3′; 5′-TTCCGTTTCCTAGTTCTTCCT-3′; 5′-CGATTTCCTGCTTGAGCCGCT-3′) with modified miR-155 stem-loops and GFP were driven by the ubiquitin C promoter in an AAV backbone (Chung et al., 2006), a gift from Dr. K. Park. ON crush was performed 2 weeks following AAV injection ∼0.5 mm behind the eyeball. Elevation of IOP was induced unilaterally in anesthetized 8- to 9-week-old mice by anterior chamber injection of fluorescent polystyrene microspheres with a uniform diameter of 10 μm (F8830, Life Technologies). The IOP of both eyes was monitored twice weekly until 8 weeks after microbead injection using the TonoLab tonometer (Colonial Medical Supply) according to product instructions.

Immunostaining and RGC counting.

The procedures have been described previously (Hu et al., 2012; Yang et al., 2014; Miao et al., 2016). Antibodies used were mouse or rabbit neuronal class ß-III tubulin (clone Tuj1, 1:200 dilution; Covance), CHOP antibody (MA1-250, 1:100 dilution; Thermo Fisher Scientific), and rat HA (clone 3F10, 1:200 dilution; Roche). For RGC counting, six to nine fields were randomly sampled from peripheral regions of each whole-mount retina preparation to estimate RGC survival. The investigators who counted the cells were blinded to the treatment of the samples. The percentage of RGC survival was calculated as the ratio of surviving RGC numbers in injured eyes compared to contralateral uninjured eyes.

Quantification of surviving axons.

Transverse semithin (1 μm) and ultrathin (90 nm) sections of ON were cut on an ultramicrotome (EM UC7; Leica). For the ON crush model, ON sections were collected 2 mm distal to the eye (∼1.5 mm distal to the crush site). For the glaucoma model, ON sections were collected 1 mm distal to the eye. The semithin sections were stained with 1% paraphenylenediamine (PPD) in methanol/isopropanol (1:1; Smith, 2002) and were viewed through a 100× lens of a Nikon microscope with a total amplification of 1000×. Sequential images were taken to cover the entire area of the ON without overlap. An area of 21.6 × 29 μm was cropped from the center of each image, and the surviving axons within the designated area were counted. After counting all the images taken from a single nerve, the mean of the surviving axon number was calculated for each ON. The mean of the surviving axon number in the injured ON was compared to that in the contralateral control ON to yield a percentage of axon survival value. The ultrathin sections were imaged by an electron microscope (JEM-1011; JEOL) at 12,000×.

Western blot.

Retinas were dissected out from ice-cold PBS-perfused eyes and homogenized (two retinas per sample) and lysed in 100 μl RIPA buffer (50 mm Tris HCl, pH 8.0, 150 mm NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 5 mm sodium pyrophosphate, 10 mm sodium fluoride, 1 mm sodium orthovanadate, protease inhibitor cocktail) on ice for 30 min. The homogenates were centrifuged at 12,000 × g for 20 min, and supernatants were subjected to electrophoresis with 10% SDS-PAGE. For ON, the nerve segments (3 mm from eyeball) were dissected and homogenized (two ON per sample) in 80 μl urea/SDS buffer (50 mm Tris HCl, pH 6.8, 8 m urea, 10% SDS, 10 mm sodium EDTA, and 50 mm DTT). Homogenates were then heated at 95°C for 10 min, cooled on ice for 10 min, and heated back to 95°C for another 10 min; finally, the homogenates were centrifuged at 12,000 × g for 20 min, and supernatants were subjected to electrophoresis with 10% SDS-PAGE. After gel transference, nitrocellulose membranes were blocked with Odyssey blocking buffer (LI-COR) for 1 h before incubation with primary antibody at 4°C overnight. After washing three times for 10 min each with PBS, the membranes were incubated with secondary antibodies (IRDye 680RD goat-anti-mouse IgG or IRDye 800CW goat-anti-rabbit IgG; LI-COR) at room temperature for 1 h. The membranes were then washed three times for 10 min each with PBS and scanned using the Odyssey CLx imaging system (LI-COR). The images were analyzed with Image Studio (LI-COR). The primary antibodies used were mouse anti-HA (clone 16B12, 1:2000; Covance), mouse anti-β-Actin (clone AC-15, 1:2000; Sigma), mouse anti-CHOP (MA1-250, 1:200 dilution; Thermo Fisher Scientific), mouse anti-Neurofilament M (NF-M; clone 3H11, 1:1000; Biolegend), and rabbit anti-binding immunoglobulin protein (BiP) (ab32618, 1:200; Abcam).

Visual evoked potential measurements.

Mice were implanted with three stainless-steel screws (Amazon, item #B0038QOTZU) in the skull at 2 mm rostral to bregma (reference electrode) and 2.5 mm horizontal to lambda (overlying the primary visual cortex, for the active electrodes). Mice recovered for 1 week before the first visual evoked potential (VEP) measurement. The mice were kept in the dark for 5–10 min before VEP measurement. The pupils were dilated by applying 1% tropicamide sterile ophthalmic solution (Akorn). Mice were placed in the Ganzfeld of the VEP instrument (RETI-port; Roland Consult) in a dark room, and body temperature was maintained at 37°C using a feedback-controlled heating pad (Physitemp Instruments). The VEP was recorded in response to a series of white light flashes of graded intensity from 0.030 cd · s/m² to 3.0 cd · s/m². The flashes were presented at a frequency of 1 Hz, and 30 flashes were averaged for a response. The first positive peak in the VEP waveform was designated as P1, and the first negative peak as N1. The latency of P1 and N1 were measured 1 week before microbead injection as a baseline and at the end points of 7 weeks postinjection (WPI) for the glaucoma model. The differences between the baseline and end point recordings under different stimulation intensities were calculated and referred to as ΔP1 latency and ΔN1 latency.

Statistics.

Data are presented as means ± SEM. We used Student's t test for two-group comparisons, one-way ANOVA with Bonferroni's post hoc test for multiple comparisons (three or more groups) with one variance, and two-way ANOVA with Bonferroni's post hoc test for multiple comparisons (three or more groups) with two variances.

Results

CHOP deletion and XBP-1 activation synergistically promote RGC axon survival after ON acute injuries

To determine whether manipulation of ER stress prevents axon degeneration, we quantified RGC axon survival in semithin transverse sections of ON at 5 d postcrush (dpc). As shown in Figure 1A, axon cluster disruption and filamentous astrocyte proliferation were obvious in WT mice, and only 32% of axons survived (Fig. 1B). XBP-1s overexpression alone did not promote axon survival, and CHOP KO had only a modest, but significant, protection effect. XBP-1s overexpression in combination with CHOP KO, however, showed significantly higher axon survival (58%) and relatively intact axon clusters with less astrocyte proliferation. EM studies confirmed axon protection by ER stress manipulation: regular arrangement of axon clusters with homogenous axoplasm and smooth, uniform myelin sheaths in CHOP KO mice with XBP-1s overexpression sharply contrasted with vacuolization and dense accumulation of axoplasm and irregular myelin sheaths in WT mice (Fig. 1C). As another test of axon protection, we measured the levels of neurofilament medium chain (NF-M), one of the major axonal cytoskeleton components, in ON after crush. NF-M levels were significantly increased in CHOP KO mice with XBP-1s overexpression compared to WT mice after ON crush, but not in single treated mice (Fig. 1D). We confirmed that AAV-XBP-1s injection increased the retinal level of HA-tagged XBP-1s and its downstream molecule BiP (Fig. 1E). Thus, CHOP deletion and XBP-1 activation synergistically promote RGC axon survival after ON crush.

Figure 1.

CHOP deletion and XBP-1 activation synergistically promote RGC axon survival at 5 dpc after ON injuries. A, Representative light microscope images of semithin transverse sections of ON with PPD staining without crush (contralateral naive) or 5 dpc in WT and CHOP KO mice with or without XBP-1s overexpression. B, Quantification of surviving axons in ON, represented as percentage of surviving axons in the injured ON, compared to the intact contralateral ON. n = 7. C, Representative electron microscope images of transverse sections of ON (12,000× magnification). D, Left, Western blot of ON lysates from biological duplicates/group showing expression levels of NF-M and actin. Right, Quantification of NF-M expression normalized to actin. n = 6. E, Top, Western blot of retina lysates from three biological replicates showing expression levels of BiP, actin, and HA-tagged XBP-1s in WT mice with AAV-GFP or AAV-XBP-1s injection. Bottom, Quantification of BiP expression normalized to actin. n = 3. F, Left, Representative light microscope images of semithin transverse sections of ON with PPD staining at 5 dpc after complete ON cut. Right, Quantification of surviving axons in ON, represented as percentage of surviving axons in the injured ON, compared to the intact contralateral ON. n = 7. Data are presented as means ± SEM. Scale bars: 10 μm. *p < 0.05; ***p < 0.001 (B, D, one-way ANOVA with Bonferroni's post hoc test; E, F, Student's t test).

Because similar UPR manipulation also promotes RGC soma survival (Hu et al., 2012), the axon protection effect may be secondary to neuroprotection of the RGC cell body and/or due to a local effect directly on the axon. We then studied the effects of ON cut, which completely separates the RGC somata in retina from their axons in the ON and therefore interrupts communication between them. Survival of RGC somata and axons should therefore be independent of each other. Axon survival was lower after ON cut than after crush, but CHOP deletion together with XBP-1s overexpression increased axon survival significantly compared to WT control (Fig. 1F), suggesting that ER stress manipulation acts, at least in part, by an axonal autonomous neuroprotection effect. We detected similar axon protection in CHOP KO mice with XBP-1s overexpression as early as at 3 dpc (Fig. 2A,B). However, Western blot analysis did not detect significant degradation of NF-M in ON at 3 dpc (Fig. 2C), suggesting the low sensitivity of this assay.

Figure 2.

CHOP deletion and XBP-1 activation synergistically promote RGC axon survival at 3 dpc after ON crush. A, Representative light microscope images of semithin transverse sections of ON with PPD staining without crush (contralateral naive) or 3 dpc. Scale bar, 10 μm. B, Quantification of surviving axons in ON, represented as percentage of surviving axons in the injured ON, compared to the intact contralateral ON. n = 7. ***p < 0.001 (one-way ANOVA with Bonferroni's post hoc test). C, Left, Western blot of ON lysates from three biological replicates per group showing expression levels of NF-M and actin. Right, Quantification of NF-M expression normalized to actin. n = 3. Data are presented as means ± SEM.

ER stress manipulation prevents glaucomatous RGC axon degeneration

Glaucoma causes progressive RGC axon and soma loss and is the most common cause of irreversible blindness (Quigley and Broman, 2006). Since ER stress manipulation is neuroprotective for RGC soma in the mouse glaucoma model (Hu et al., 2012) and protects RGC axons following ON crush, we tested whether it also protected glaucomatous axons. We injected microbeads into the anterior chamber of the left eyes of adult mice to elevate IOP by blocking aqueous outflow (Fig. 3A); the right eyes received a sham injection and served as control. Histology of the anterior segment of the eyeball showed accumulation of microbeads at the corner of the anterior chamber (Fig. 3B). The IOP of the mouse eyes was measured once before and twice weekly after microbead injection (Fig. 3C). We examined axon survival only in mice with sustained high IOP (≥7 mmHg difference between the microbead-injected and contralateral sham-injected eyes). In WT mice, ∼71% of RGC axons were present at 8 weeks after microbead injection. XBP-1 activation alone did not increase glaucomatous axon survival, whereas surviving axons were significantly increased in CHOP KO mice with or without AAV-XBP-1s (Fig. 3D,E). The protective effect was not due to differences in IOP levels, because microbead injection elevated IOP similarly in all experimental groups (Fig. 3C).

Figure 3.

ER stress manipulation prevents glaucomatous RGC axon degeneration. A, Injection of microbeads into the anterior chamber. B, Paraffin section of mouse eye shows microbead accumulation. C, IOP measurements of different treatment groups. D, Representative light microscope images of semithin sections of ON with PPD staining at 8 WPI. Scale bar, 10 μm. E, Quantification of surviving axons in ON. Data are presented as means ± SEM. WT, n = 9; AAV-XBP-1s, n = 12; CHOP KO, n = 7; CHOP KO + AAV-XBP-1s, n = 11. *p < 0.05; ***p < 0.001 (one-way ANOVA with Bonferroni's post hoc test).

ER stress manipulation preserves visual function in mouse glaucoma model

To determine the potential clinical importance of ER stress manipulation, we measured the VEP, a gross electrical signal recorded from the visual cortex after flash stimulation of the eye (Ridder and Nusinowitz, 2006). Because the VEP signal depends on the integrity of the entire visual pathway, including RGC somata and ON axons, it can serve as readout of the physiological function of both structures. We recorded the VEP from mice 1 week before microbead injection (baseline) and at 7 WPI (end point). Because the variation of P1–N1 amplitude was too large to be used in this model (data not shown), we focused on the more stable N1 and P1 latencies. N1 and P1 latencies were significantly prolonged in the glaucomatous eyes of WT mice; these changes were much smaller in mice with XBP-1 activation, CHOP deletion, or combined treatment, indicating preservation of visual function by ER stress manipulation (Fig. 4). It is worth noting that although XBP-1s alone did not significantly promote glaucomatous axon survival, it did improve VEP responses compared to controls, suggesting that the RGC somata protected by XBP-1s contributed directly to VEP signals.

Figure 4.

ER stress manipulation preserves visual function after IOP elevation. A, Representative waveforms showing the VEP elicited by a series of stimulus light intensities. Black line, Baseline, 1 week before injection of microbeads. Red line, 7 WPI. B, The differences in flash VEP responses between 1 week before and 7 weeks after microbead injection are represented as ΔN1 latency and ΔP1 latency in response to a series of varying stimulus light intensities. Data are presented as means ± SEM. WT, n = 21; AAV-XBP-1s, n = 12; CHOP KO, n = 6; CHOP KO + AAV-XBP-1s, n = 19. *p < 0.05; ***p < 0.001 (two-way ANOVA with Bonferroni's post hoc test).

Blocking eIF2α phosphorylation and activating XBP-1 promote neuroprotection of RGC somata and axons after ON crush

To confirm that the PERK-eIF2α-CHOP branch of UPR plays a detrimental role after axon injury, we conducted a similar analysis to test whether blocking the phosphorylation of eIF2α also provides neuroprotection. Since the homozygous eIF2α S51A mice are postnatal lethal (Scheuner et al., 2001), we used the eIF2α A/A;fTg transgenic mouse (Back et al., 2009), in which the homozygous eIF2α S51A mutant (which cannot be phosphorylated) replaces both alleles of endogenous WT eIF2α, and one copy of floxed WT eIF2α is ubiquitously expressed to overcome the postnatal lethality of homozygous eIF2α S51A. AAV-Cre injection in eIF2α A/A;fTg mice will delete the only copy of WT eIF2α, which is accompanied by expression of GFP as an indicator and effects the temporally controlled expression solely of the unphosphorylated eIF2α S51A mutant specifically in RGCs. We have confirmed that CHOP expression induced by ON crush was significantly decreased by removing the WT eIF2α and only expressing the eIF2α S51A mutant in RGCs (Fig. 5A). We also confirmed that the injection of the mixture of AAV-Cre and AAV-XBP-1s achieved high efficiency of WT eIF2α deletion and XBP-1s overexpression in RGCs (Fig. 5B), which resulted in both RGC soma and axon protection after ON crush. Blocking eIF2α-P alone, however, yielded better RGC soma protection and worse RGC axon protection than CHOP deletion. In addition, although eIF2α-P inhibition with XBP-1 activation achieved similar RGC soma protection, axon protection after ON crush was much worse than that provided by CHOP deletion with XBP-1 activation (compare Fig. 1A,B, Fig. 5C,D). These findings indicate that the neuroprotective effect of eIF2α-P inhibition was not exerted solely through CHOP inhibition. Similar to CHOP KO + XBP-1s mice, RGC soma protection was much better than axon protection in mice with eIF2α A/A + XBP-1s, which indicates that ER stress plays a more dominant role in neuron soma death than in axon degeneration.

Figure 5.

Blocking eIF2α phosphorylation and activating XBP-1 promote neuroprotection of RGC somata and axons after ON crush. A, Representative confocal images of whole-mount retinas labeled with GFP, CHOP, Tuj1, and merged images. B, Representative confocal images of whole-mount retinas 2 weeks after intravitreous injection of AAV-Cre + AAV-XBP-1s-HA in eIF2α A/A;fTg mice showing a majority of Tuj1 positive RGCs (red) are also GFP (Cre expression) and HA (XBP-1s-HA expression) positive. C, Top, Whole-mount retinas showing surviving Tuj1 positive (red) and AAV-Cre positive (GFP) RGCs at 14 dpc. Bottom, Semithin sections of ON with PPD staining at 5 dpc. D, Quantification of surviving RGCs at 14 dpc (WT, n = 9; eIF2α A/A, n = 12; eIF2α A/A + AAV-XBP-1s, n = 8) and surviving axons in ON at 5 dpc (WT, n = 7; eIF2α A/A, n = 10; eIF2αA/A + AAV-XBP-1s, n = 9). Scale bars: A, B, C, top, 20 μm; C, bottom, 10 μm. Data are presented as means ± SEM. *p < 0.05; ***p < 0.001 (one-way ANOVA with Bonferroni's post hoc test).

Blocking eIF2α phosphorylation and activating XBP-1 promote neuroprotection and preserve visual function in glaucoma model

Next we tested the combination of eIF2α-P inhibition and XBP-1s overexpression in the glaucoma model, which confirmed the neuroprotective effect of manipulating these two UPR pathways on glaucomatous RGC somata and axons (Fig. 6A,B). VEP analysis showed striking visual function protection as well, evidenced by significantly shorter P1 and N1 latencies in the eIF2α A/A glaucomatous eyes with or without AAV-XBP-1s injection than in WT mice (Fig. 6C,D). Similar to XBP-1s alone, the eIF2α A/A mutant did not significantly promote glaucomatous axon survival, but achieved better VEP, suggesting the importance of RGC soma protection for preserving visual function. Together, these results show that blocking CHOP upstream molecule eIF2α together with activating XBP-1 protects glaucomatous RGC somata and axons and, more importantly, preserves visual function.

Figure 6.

Blocking eIF2α phosphorylation and activating XBP-1 promotes neuroprotection and preserves visual function in glaucoma model. A, Top, Whole-mount retinas showing surviving Tuj1 positive (red) and AAV-Cre positive (GFP) RGCs at 8 WPI (glaucoma model). Bottom, Semithin sections of ON with PPD staining at 8 WPI (glaucoma model). Scale bars: Top, 20 μm; bottom, 10 μm. B, Quantification of surviving RGCs (WT, n = 12; eIF2α A/A, n = 10; eIF2αA/A + AAV-XBP-1s, n = 10) and surviving axons in ON at 8 WPI (WT, n = 7; eIF2α A/A, n = 5; eIF2αA/A + AAV-XBP-1s, n = 9). *p < 0.05; **p < 0.01 (one-way ANOVA with Bonferroni's post hoc test). C, Representative waveforms showing the VEP elicited by a series of stimulus light intensities. Black line, Baseline, 1 week before injection of microbeads. Red line, 7 WPI. D, The differences in flash VEP responses, represented as ΔN1 and ΔP1 latencies with a series of stimulus light intensities (WT, n = 21; eIF2α A/A, n = 10; eIF2α A/A + AAV-XBP-1s, n = 11). Data are presented as means ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001 (two-way ANOVA with Bonferroni's post hoc test).

Delayed ER stress manipulation also provides neuroprotection in glaucoma

To simulate the clinical setting, we tested whether manipulating UPR only after IOP elevation also protects RGC somata and axons. We used two strategies to manipulate UPR after IOP elevation: (1) AAV-mediated CHOP shRNA expression to knock down CHOP levels and (2) AAV-Cre-mediated blocking of eIF2α-P. We first confirmed that microRNA-based AAV-CHOP shRNA-GFP knocks down CHOP levels in cultured cells and retinas (Fig. 7A–C). We next injected AAV-CHOP shRNA + AAV-XBP-1s or AAV-scramble shRNA + AAV-GFP (control) in WT mice at 1 WPI, when IOP reaches its peak. Since AAV-mediated gene expression in RGCs normally peaks at least 2 weeks after infection (Hu et al., 2012; Boye et al., 2013), CHOP is likely to be inhibited and XBP-1s expressed in RGCs 2 weeks after IOP elevation. Importantly, this delayed ER stress modulation significantly protected RGC somata and axons (Fig. 7D,E). The treated eyes showed significantly less prolonged N1 latency and P1 latency than the control eyes (Fig. 7F). Next we injected eIF2α A/A;fTg mice with AAV-Cre + AAV-XBP-1s at 1 WPI. Delayed inhibition of eIF2α-P and XBP-1 activation significantly protected RGC somata and axons and significantly spared visual function (Fig. 7G–I). Consistent with previous reports (Lin et al., 2007; Saxena et al., 2009; Hu et al., 2012; Moreno et al., 2012; Ma et al., 2013b), our results indicate that the PERK-eIF2α-CHOP branch of ER stress is toxic to injured neurons and that combined inhibition of this pathway and activation of the XBP-1 pathway is a promising therapeutic approach for glaucoma.

Figure 7.

Delayed ER stress manipulation provides neuroprotection in glaucoma. A, Western blot of CHOP expression in mouse 3T3 cells with AAV-Scramble shRNA-GFP or AAV-CHOP shRNA-GFP infection. Tg, Thapsigargin (ER stress inducer). B, CHOP expression in whole-mount retinas at 3 dpc. Scale bar, 20 μm. C, Top, Western blot of retina lysates from three biological replicates per group showing expression levels of CHOP and actin in WT naive mice or 5 dpc of WTs with or without AAV-CHOP shRNA injection. Bottom, Quantification of CHOP expression normalized to actin. n = 3. D, G, Top, Whole-mount retinas showing surviving Tuj1 positive (red) and GFP positive RGCs at 8 WPI. Bottom, Semithin sections of ON with PPD staining at 8 WPI. AAVs were injected intravitreally at 1 WPI. E, H, Top, Quantification of surviving RGCs. Bottom, Quantification of surviving axons in ON. F, I, The differences in the flash VEP responses are represented as ΔN1 latency and ΔP1 latency elicited by a series of stimulus light intensities. All quantification data are presented as means ± SEM (AAV-Scramble shRNA + AAV-GFP, n = 6; AAV-CHOP shRNA-GFP + AAV-XBP-1s, n = 10; eIF2α A/A;fTg mice with AAV-Cre + AAV-XBP-1s, n = 6; WT mice with AAV-GFP, n = 21). *p < 0.05; **p < 0.01; ***p < 0.001 (Student's t test was used for two-group comparisons and two-way ANOVA with Bonferroni's post hoc test was used for flash VEP data analysis). Scale bars: B, D, G, top, 20 μm; D, G, bottom, 10 μm.

Discussion

Although neuroprotectants have long been sought, none has been found. Only deciphering the key upstream signals that trigger the degeneration cascade in both neuronal soma and axon will fulfill this significant unmet clinical need and lead to innovative and efficient neuroprotective treatments. By exploiting the anatomical and technical advantages of the RGC/ON system for studies of CNS traumatic injury (ON crush) and neurodegenerative diseases (glaucoma), we demonstrated that PERK-eIF2α-CHOP inhibition and XBP-1 activation preserves the structure and function of both RGC somata and axons. This result indicates the critical importance of ER stress in the pathophysiology of CNS axonopathies. Our findings that these two arms of UPR play opposite roles in neurodegeneration is consistent with the emerging theme that IRE1α-XBP-1activation occurs transiently in ER stress and is cell protective, whereas PERK-eIF2α-CHOP activation persists during chronic ER stress and triggers cell death (Rutkowski et al., 2006; Lin et al., 2007, 2009). CHOP has been associated with apoptosis downstream of ER stress through downregulating antiapoptotic Bcl2 (McCullough et al., 2001), upregulating proapoptotic BH-3-only molecules Bim (Puthalakath et al., 2007) and p53 upregulated modulator of apoptosis (Galehdar et al., 2010), increasing protein synthesis (Han et al., 2013), and stimulating death receptor 5 (Lu et al., 2014). Inhibition of CHOP is beneficial in multiple disease models (Zinszner et al., 1998; Oyadomari et al., 2002; Silva et al., 2005; Song et al., 2008; Hu et al., 2012). Moreover, CHOP KO is protective for injured motor neuron axons (Pennuto et al., 2008; Penas et al., 2011), and that an ER stress-protective agent attenuates axonal degeneration (Saxena et al., 2009). Thus, axon injury-induced neuronal ER stress could be a common mechanism for both neuronal soma and axon degeneration in a broad range of neurological diseases (Li et al., 2013). Although how ER stress is involved in axonopathies is unknown, UPR may play an intra-axonal role. Interestingly, XBP-1 splicing has been shown to occur initially in neurites and to translocate to the nucleus later (Hayashi et al., 2007), and elevated levels of eIF2α-P and ATF4 have been detected in the axons of cultured neurons treated with Aβ1–42 (Baleriola et al., 2014). However, the interaction between RGC cell body and axon certainly makes an important contribution to the decision between death and survival, and the current injury models did not allow us to distinguish definitively the neuronal soma-autonomous from the axon-autonomous effects. The injury signals transported back to the neuronal soma from the injured axon may be critical; blocking them may protect both the neuronal soma and axon. We therefore propose that the UPR signaling molecules may have similar functions in response to axon injury as the previously identified dual leucine zipper kinase (DLK; Welsbie et al., 2013; Fernandes et al., 2014).

A paradox of the PERK- eIF2α-CHOP pathway is that although it can counteract ER stress and enable the cell to achieve a new homeostasis and survive, it also leads to cell death if ER stress is prolonged (Rutkowski et al., 2006). Phosphorylation of eIF2α inhibits cap-dependent mRNA translation and therefore attenuates ER stress by reducing the protein workload of ER. Thus, reduction of translation reduction by the PERK-eIF2α pathway enables pancreatic β cells and other cell types to adapt to ER stress and survive (Malhotra and Kaufman, 2007; Back et al., 2009). Similarly, sustained eIF2α-P by inhibition of eIF2α-P phosphatase promotes motor neuron survival in mouse models of chronic neurodegenerative diseases (Zhu et al., 2008; Saxena et al., 2009; Das et al., 2015). However, suppression of protein translation by eIF2α-P is not always protective, and emerging evidence indicates that blocking eIF2α-P is actually neuroprotective. Elevated levels of eIF2α-P and axonally synthesized ATF4 have been detected in the brains of AD mice and patients and linked to neuronal cell death (Baleriola et al., 2014). Activation of the PERK/eIF2α branch may also be responsible for synaptic defects and neuronal degeneration in prion disease (Moreno et al., 2012, 2013) and AD (Ma et al., 2013b). In addition, downregulation of eIF2α-P by a PERK inhibitor or derepression of protein translation by small molecules results in potent neuroprotection (Axten et al., 2012; Ma et al., 2013b; Halliday et al., 2015). In the present study, we demonstrated that genetic blocking of eIF2α-P significantly increases RGC survival, indicating the similarity between optic neuropathies and other neurodegenerative diseases. The dissimilar neuroprotective effects of CHOP deletion and inhibition of eIF2α-P on RGC soma and axon signify that eIF2α-P plays a role in neurodegeneration that is CHOP independent. It will be of great interest to decipher the downstream mechanism by which eIF2α acts independently of blocking CHOP to contribute to neuroprotection. Because suppression of protein translation by eIF2α-P may block expression of cell survival proteins (Allagnat et al., 2011), resumption of protein synthesis after blocking eIF2α-P may enhance cell survival. Interestingly, eIF2α-P downregulates the translation of X-linked inhibitor of apoptosis (XIAP) in a CHOP-independent manner, which contributes to cell death induced by chronic ER stress (Hiramatsu et al., 2014). Future experiments will determine whether XIAP is upregulated in RGCs of eIF2α A/A mutant mice and, more intriguingly, whether CHOP deletion and eIF2α-P block act synergistically to promote neuroprotection.

In addition to Wld^s (Beirowski et al., 2008, 2009; Howell et al., 2013; Zhu et al., 2013; Conforti et al., 2014), several endogenous molecules have been found previously to promote axonal degeneration, including SARM (sterile α-motif-containing and armadillo-motif containing protein; Osterloh et al., 2012; Gerdts et al., 2013; Yang et al., 2015), PHR1 (an E3 ubiquitin ligase; Babetto et al., 2013), DLK (Welsbie et al., 2013), and the mitogen-activated protein kinase cascade (Miller et al., 2009; Yang et al., 2015). Inhibiting these molecules significantly enhances axon protection. It will be extremely interesting to investigate the potential cross talk between these pathways and the UPR pathways and to determine whether targeting them together has a synergistic neuroprotective effect.

The mechanism that induces ER stress in axonopathies is unknown. Influx of extracellular calcium and release of calcium from intra-axonal smooth ER are important early events after axon injury that correlate with initiation of axon degeneration (Wang et al., 2012; Howell et al., 2013), possibly by promoting cytoskeleton disintegration (Ma et al., 2013a) and mitochondrial dysfunction (Barrientos et al., 2011). Disturbance of calcium homeostasis is actually a major inducer of ER stress (Ron and Walter, 2007; Wang and Kaufman, 2012; Li et al., 2013). Thus, it is possible that intra-axonal calcium fluctuation early after axon injury induces ER stress, which then further increases intra-axonal calcium. Interestingly, calcium released from the ER can be taken up by mitochondria (Villegas et al., 2014), causing mitochondria dysfunction and axon degeneration (Barrientos et al., 2011; Court and Coleman, 2012; Villegas et al., 2014). Although this evidence places ER dysfunction upstream of mitochondria dysfunction, the reverse is also possible because redox changes and ATP depletion caused by mitochondria failure can induce ER stress. It is critical to determine which is the earlier event after axon injury, to establish the most promising clinical intervention target for neuroprotection.

Although axon degeneration is thought to be an active process of self-destruction that is mechanistically distinct from the process of apoptosis in the neuronal soma (Coleman and Perry, 2002; Raff et al., 2002; Whitmore et al., 2005; Saxena and Caroni, 2007; Wang et al., 2012), the two processes are sequentially connected. The protection of both neuronal soma and axon by ER stress manipulation indicates that axon injury-induced neuronal ER stress serves as a general upstream signal for both neuron apoptosis and axon autonomous degeneration. Optic neuropathies are likely to have mechanisms in common with other neurodegenerative diseases, such as ALS, AD, PD, and MS (Kersten et al., 2014). Our study therefore advances understanding of a mechanism essential for neuronal degeneration and provides information critical for the clinical application of ER stress modulation to preserve neural function in a wide range of acute and chronic neurodegenerative disorders.