Reactive Oxygen Species Regulate Prosurvival ERK1/2 Signaling and bFGF Expression in Gliosis within the Retina

Abstract

Purpose.: Gliosis is the response of glial cells within retinal tissue to injury. It can be beneficial in the short term, but if the response is extended it can lead to scar formation, which contributes to blindness. Phosphorylation of extracellular signal regulated kinase 1/2 (ERK1/2) is considered to be a hallmark event of gliosis, but the factors involved throughout its associated signaling pathway remain poorly understood, particularly in the retina. Because reactive oxygen species (ROS) can inhibit phosphatases, thereby altering the phosphorylation of proteins, this study tested the hypothesis that ROS regulate the phosphorylation of ERK1/2 (pERK1/2) in gliosis.

Methods.: Increases in pERK1/2 were detected using Western blotting and immunofluorescence in three models of retinal stress, specifically the in vivo light induction, the rd1 disease, and the ex vivo retinal explant models. Explanted murine retinas were used to identify the signaling partners of pERK1/2 via Western blotting, in conjunction with inhibitors. The effect of this pathway on cell death was measured with terminal dUTP nick end labeling.

Results.: It was demonstrated that several inhibitors of ROS greatly reduce the levels of pERK1/2 in the somata of Müller cells and furthermore decrease two other downstream signaling events: the phosphorylation of STAT3 and the upregulation of basic fibroblast growth factor. Using the specific inhibitor of ERK1/2, UO126, the resultant outcomes of this signaling pathway were determined to contribute significantly to cell survival.

Conclusions.: The novel finding of this study that ROS contribute to a prosurvival signaling pathway in retinal Müller cell gliosis indicates that some degree of caution should be used when considering antioxidants as therapeutics.

Introduction

Gliosis is the response of glial cells within neuronal tissue, including the retina, to any insult. It is characterized by the activation of glial cells, but the final outcome may be beneficial or detrimental to the surrounding tissue depending on the time scale involved. ^1,2 In gliosis, retinal glial cells (Müller cells, astrocytes, and microglia) are activated. Early events in gliosis involve the upregulation of several key cytokines, which initially promote cell survival. This upregulation, however, can have adverse effects by triggering proliferation of glial cells, which ultimately results in scarring. Although scarring is beneficial to maintaining the structural integrity of the retina, it is detrimental to neuronal function. For this reason, a full understanding of the molecular events of gliosis is important to prevent scar tissue formation in the brain and retina after injury and has the potential to identify novel therapeutic targets and strategies for a wide variety of retinal disorders. 

A key characteristic of gliosis is the increased phosphorylation of extracellular signal-regulated kinase 1/2 (ERK1/2). ^2,3 This happens in several models of retinal disease/damage. ^4,5 ERK1/2 are part of the mitogen-activated kinase (MAPK) family of proteins, which transduce signals from the cell surface to the nucleus. When ERK1/2 are activated through phosphorylation on threonine and tyrosine residues, they in turn can activate transcription factors by phosphorylating them directly or they can phosphorylate other target proteins. They mainly play a role in cellular differentiation, proliferation, and survival, but there is increasing evidence that they can also initiate prodeath pathways. ⁶ Although an increase in pERK1/2 is seen as a marker of gliosis, the mechanisms by which it is regulated remain poorly understood, and knowledge of the specific signaling outcomes of this phosphorylation event in gliosis within the retina is limited. 

Reactive oxygen species (ROS) have only recently been recognized as intracellular signaling molecules in their own right, whereas in the past they were simply thought of as destructive molecules formed as a byproduct of respiration. ⁷ Their capacity to modify signaling pathways comes from their ability to oxidize key cysteine residues in the active sites of phosphatases, thereby inhibiting them, which results in alterations of the phosphorylation status of other proteins. ⁸ Because the phosphorylation status of ERK1/2 changes with gliosis, it is possible that ROS play a role in these events. There is indirect evidence in the literature that ROS play a role in gliosis (e.g., antioxidants prevent gliosis in certain models ^9,10 and deletion of particular genes increases oxidative stress while reducing gliosis ¹¹ ), but there are very few, if any, studies that examine the signaling role of ROS in gliosis. Therefore, the precise role of ROS, specifically in the possible regulation of pERK1/2, and its impact on signaling pathways in gliosis remain poorly elucidated. 

This study on gliosis within the retina has two main aims. The first, broad aim was to understand more fully the regulation and sequence of the molecular signaling pathways of this stereotypical response to neuronal injury. The second, more specific, hypothesis tested was that ROS regulate pERK1/2 signaling events in gliosis. The retinal explant system was chosen as the model for most of the experiments presented herein, because it demonstrated a similar increase in pERK1/2 as that in other in vivo retinal models used here, but was more conducive to experimental manipulation. During the course of this study, a final question arose as to the function of pERK1/2 in gliosis, due to the dual beneficial/detrimental nature ascribed to both of these events in the literature. Through Western blotting and immunofluorescence studies, we demonstrate the novel finding that ROS play an important role in the pERK1/2, phospho-signal transducer and activator of transcription 3 (pSTAT3), basic fibroblast growth factor (bFGF) signaling pathway, and that this pathway is of beneficial nature to the retina. 

Materials and Methods

Animals/Ethics Statement

Mice were supplied by the Biological Services Unit, University College Cork, Ireland. All animal experiments were carried out in accordance with the Association for Research in Vision and Ophthalmology statement for the Use of Animals in Ophthalmic and Vision Research and approved by the University College Cork Animal Experimentation Ethics Committee. C57BL/6 mice were used at postnatal day (P) 15 for explanting and as age-matched controls against the C3H/HEN (rd1) mice. Light-induction experiments used adult Balb/c mice. 

Reagents

All chemicals were purchased from Sigma-Aldrich (Wicklow, Ireland) unless otherwise stated. Inhibitors were included in the culture medium of the explants for the durations indicated in the figure legends, at the following concentrations: 10 μM UO126 (Cell Signaling Technology, Beverly, MA), 10 μM diphenyliodonium (DPI), 10 mM N-acetyl cysteine (NAC), 100 μM AG490 (Enzo Life Sciences, Exeter, UK), and 4 mM apocynin (Calbiochem/Merck, Nottingham, UK). 

Light Induction

Following the protocol detailed in Donovan et al., ¹² Balb/c mice were reared in dim conditions (less than 10 lux, 12-hour on/off cycle). For 18 hours prior to an experiment, they were dark adapted, and then their pupils dilated using 5% cyclopentolate. They were placed in the test chamber, where they were exposed to 5000 lux of cool, white light for 2 hours. Animals were given recovery time of up to 24 hours in the dark. Animals were euthanized at various time points through this protocol to allow examination of a complete time course of the events occurring in the retina. 

Explanting

Retinal explants were prepared following Caffe et al. ¹³ Briefly, C57BL/6 mice at the age of P15 were culled through cervical dislocation. The lens, anterior segment, vitreous body, retinal pigment epithelium, and sclera were removed from enucleated eyes, before the retina was mounted flat with photoreceptor side down on top of a nitrocellulose insert (Millipore, Billerica, MA) in six-well plates (Sarstedt AG & Co., Wexford, Ireland). All explants were cultured in R16 medium (recipe from P. A. Ekstrom, Wallenberg, Retina Center, Lund University, Lund, Sweden) for the times stated in the figures, ranging from 0 to 48 hours. 

Western Blotting

Following treatments, retinal explants were washed with ice-cold PBS followed by lysis as described previously in Groeger et al. ¹⁴ All debris was removed by centrifugation and protein concentration was quantified using a commercial protein assay (Bio-Rad, Hemel Hempstead, UK) using bovine serum albumin as a standard. Equivalent amounts of protein were resolved using denaturing sodium dodecyl sulfate–polyacrylamide gel electrophoresis, followed by transfer to nitrocellulose membranes (Schleicher & Schuell, Whatman, Dassel, Germany). After blocking membranes with 5% (w/v) nonfat dry milk in Tris-buffered saline/0.1% Tween-20 for 1 hour at 18 to 22°C, they were incubated at 4°C overnight with the appropriate dilution of primary antibody (see the following text). After three 5-minute washes with Tris-buffered saline/0.1%Tween-20, blots were incubated with the corresponding peroxidase-conjugated secondary antibody (dilution 1:1000; Dako, Glostrup, Denmark) for 1 hour. They were then washed again and developed with enhanced chemiluminescence reagent (Thermoscientific, Northumberland, UK). 

The primary antibodies used were pERK1/2, ERK1/2, pSTAT3, STAT3, and c-fos (all from Cell Signaling Technology, Beverly, MA, and all used at 1:1000), and bFGF (Millipore, Carrigtwohill Co., Cork, Ireland, used at 1:500). Antibodies used to confirm equal loading were glyceraldehyde-3-phosphate dehydrogenase (GAPDH; Advanced Immunochemical, Long Beach, CA) or α-tubulin (Sigma-Aldrich; both at 1:5000). 

Immunofluorescence

Enucleated eyes or retinal explants were fixed in 10% neutral buffered formalin overnight, followed by cryoprotection in 25% sucrose overnight, both at 4°C. Sections of 7 μm were cut using a cryostat (Leica CM1950; Leica Co., Meath, Ireland), and incubated in a blocking buffer (10% normal goat serum containing 0.3% Triton X-100) for 1 hour at 20 to 22°C. Subsequently, they were incubated with one or two of the following primary antibodies overnight at 4°C: anti-p-ERK1/2, anti-c-fos, anti-phosphoSTAT3 (all 1:100; Cell Signaling Technology), antiglutamine synthetase (GS), or anti-bFGF (both 1:100; Millipore). Antibody binding was detected by using conjugated secondary antibodies (Alexa Fluor-594 or Alexa Fluor-488; Invitrogen, Carlsbad, CA) at a dilution of 1:200 for 1 hour at 20 to 22°C. Hoechst 33342 (1 μg/mL; Sigma-Aldrich) was included with the secondary antibody incubation to counterstain the nuclei. Sections were washed again, with PBS, mounted, and imaged with a fluorescence microscope (Leica DM LB2; Leica Co.). 

Terminal dUTP Nick-End Labeling (TUNEL)

Retinal explants were fixed, cryoprotected, and sectioned as described in the previous section. Sections were incubated with terminal deoxynucleotidyl transferase (MSC, Dublin, Ireland) and fluorescein-12-dUTP (Roche, Lewes, UK) according to manufacturer's instructions at 37°C for 1 hour and nuclei counterstained with Hoechst. Sections were mounted and viewed under a fluorescence microscope (Leica DM LB2). 

The numbers of TUNEL positive cells were counted. For each treatment and time point used, at least three animals were used and three fields (×40 magnification) per section of at least three different sections were evaluated. Error bars represent the SEM values. 

Statistics

ImageJ (1.40g; developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/download.html) was used to perform densitometric analysis of Western blots. Values in all graphs represented the mean ± SEM. For experiments where multiple comparisons were made, a commercial software program (SPSS 15.0.1 for Windows; SPSS, Dublin, Ireland) was used to perform ANOVA to check for any overall statistical significance, followed by the Tukey post hoc test when appropriate, and significant differences (P < 0.05) are indicated on the graphs with asterisks (*). Throughout, each experiment was repeated a minimum three times. 

Results

We initially examined the phosphorylation status of ERK1/2 in two models of retinal degeneration. The acute model of light-induced retinal degeneration, where cell death is detectable from 6 hours post light induction, ¹² showed an upregulation in pERK1/2, as detectable by Western blotting at 3 hours post light induction, which was sustained at 6 hours, and decreased at further time points (Figs. 1A, 1B). In the rd1 mouse model of retinitis pigmentosa, which is due to a mutation in the rod phosphodiesterase 6 gene, where cell death peaks at P12–P14, ^15,16 there was an increase in pERK1/2 at P11, P12, and P13, over and above that found in age-matched C57BL/6 control animals (Figs. 1C, 1D). These changes were also detectable by immunofluorescence, in which the staining pattern of pERK1/2 mainly localized to the inner nuclear layer (INL), and formed the typical striatal pattern associated with activated Müller cells (Figs. 1B, 1D). Due to the many cell types present in the INL of the retina, it was important to address whether this upregulation in pERK1/2 did indeed colocalize to the Müller cell subpopulation within the retina. Costaining was performed using an antibody against pERK1/2 combined with an antibody against glutamine synthetase (GS), a marker commonly used for Müller cells. For both of the models, one time point was chosen where pERK1/2 was high according to the Western blots. Figure 1E demonstrates that cells high in pERK1/2 were also labeled with GS. Although there were also numbers of GS positive cells that did not express pERK1/2 highly, this can be explained by the heterogeneity of responses found in Müller cells. ¹ Therefore this increase in pERK1/2 is localized to Müller cells, demonstrating that both of these models were undergoing a gliotic response. 

Figure 1. 

 

pERK1/2 increases in two models of retinal injury and localizes to Müller cells. (A) Western blot showing the increase in pERK1/2 levels in the light induction (LI) model. (B) This is also visible using immunofluorescence staining of the retina (pERK1/2: red, Hoechst: blue). (C) Western blot analysis comparing the levels of pERK1/2 in the rd1 mouse against c57 controls at postnatal (P) days 11, 12, and 13. (D) Immunofluorescence staining of the same aged retinas (pERK1/2: red, Hoechst: blue). (E) In both models of retinal injury (LI: upper panels; rd1: lower panels), 7-μm sections were stained with pERK1/2 (red), glutamine synthetase (GS, Müller cell marker, green), and Hoechst (blue). Yellow staining in the merge and zoom panels indicates colocalization, thereby demonstrating pERK1/2 is present in Müller cells. Scale bars: (B), (D), (E): 50 μm. Densitometric analysis of the Western blots was undertaken and the graphs represent the mean ± SEM for three independent experiments. ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer.

A similar increase in pERK1/2 was detected in retinas that were explanted onto membranes, with a strong signal present from 0.5 to 6 hours postexplanting, decreasing below basal levels at 48 hours (Figs. 2A, 2B). This increase in pERK1/2 was also localized to the INL, and to the Müller cells in particular (Fig. 2C). The C57BL/6 mouse strain used for explanting and as the control against rd1 (Fig. 1D) consistently demonstrated high fluorescence of pERK1/2 in the outer plexiform layer (compare Fig. 1D with Fig. 2B). We believe this to be strain specific because it is absent in the rd1 and the Balb/c mice (Figs. 1B, 1D) and it does not change following explanting. The increase in a Müller cell staining pattern of pERK1/2 is evident in each of these three models through the striatal pattern of pERK1/2 in both the ONL and INL of these models (see zoom images of Figs. 1E, 2C). It is noteworthy that in each of these three models glial fibrillary acidic protein was also upregulated but at a later time point than pERK1/2 upregulation (data not shown). These results confirm that the phosphorylation of ERK1/2 is a common, and probably universal, event in gliosis within the retina. 

Figure 2. 

 

Retinal explants demonstrate a similar increase in pERK1/2. (A) Western blot showing the changes in pERK1/2 in retinas postexplanting. (B) This corresponds with an increase in immunofluorescence staining of pERK1/2 in explants over the control retina. (C) This pERK1/2 (red) staining partially colocalizes with glutamine synthetase (GS, green). Scale bars: 50 μm; blue staining is Hoechst, labeling all nuclei. Densitometric analysis of the Western blots was undertaken and the graph represents the mean ± SEM for three independent experiments.

Another marker of gliosis is the upregulation of bFGF. ¹ There is limited evidence in the literature that links pERK1/2 to this event, ¹⁷ but several studies link the phosphorylation of STAT3 to increased expression of bFGF. ^18–21 In addition, c-fos is another transcription factor that has also been linked to bFGF expression. ²² We therefore sought to delineate the pathway(s) involving pERK1/2 and bFGF in gliosis. All of the investigated signaling proteins showed a distinctive temporal pattern, either in their phosphorylation or expression (Fig. 3A). There was a notable increase in c-fos expression at 1 hour, which was transient in nature, as it decreased toward basal levels by 3 hours postexplanting. The slight upward shift in the c-fos band is likely indicative of its phosphorylation. ^23,24 More sustained patterns were seen with bFGF upregulation, which was very slight at 0.5 hour and pronounced at 24 and 48 hours, and with pSTAT3, which was increased at 1 hour and continued to be detected at high levels up to 48 hours. These antibodies were suitable for immunofluorescence staining and each of them demonstrated a significant expression in the INL (Fig. 3B), implying that the response is gliotic in nature, and probably related to Müller cells. Notably, bFGF is also highly expressed in the outer nuclear layer (ONL), as detected by immunofluorescence, at 0 and 24 hours postexplanting (Fig. 3B), but there was no bFGF detectable at the 0 hour time point by Western blotting (Fig. 3A). This is likely due to inherent differences in the sensitivities of these techniques and because photoreceptors are known to produce bFGF. ²⁵ Overall, therefore, retinal stress induced by explanting resulted in two of the common features of gliosis, that is, the phosphorylation of ERK1/2 and the upregulation of bFGF, in conjunction with other signaling events. 

Figure 3. 

 

The time course of other signaling events and their localizations within retinal explants. (A) Western blot analysis demonstrates the changes in bFGF, c-fos, and pSTAT3 signaling up to 48 hours postexplanting. (B) Immunofluorescence images show that these changes are in the inner nuclear layer of the retina, where Müller cells reside. Scale bars: 50 μm; blue staining is Hoechst, labeling all nuclei. Densitometric analysis of the Western blots was undertaken and the graphs represent the mean ± SEM for three independent experiments.

Having demonstrated that retinal explants have a similar gliotic response to animal models of gliosis, specifically in terms of their transient upregulation of pERK1/2 in Müller cells, we next sought to determine if ROS play a role in the regulation of this key phosphorylation event using explanting as our model system. Three inhibitors of ROS were used in the retinal explant system. NAC is a glutathione precursor, and so although not a direct scavenger of ROS, it acts as an antioxidant by altering the redox status of the cell. ²⁶ DPI inhibits all flavo-containing proteins, including NADPH oxidase (Nox) enzymes, a potential source of ROS. ²⁷ Apocynin can act as a specific Nox inhibitor at low concentrations, but at the concentration used here is likely acting as a general antioxidant. ²⁸ Doses were selected based on those typically used in previously published studies. All three notably decreased the levels of pERK1/2 at early time points (Fig. 4). These time points were chosen for this particular experiment because they offered the best compromise between allowing the inhibitors to be effective, while still measuring the phosphorylation of ERK1/2 when the control would be at a relatively high level. The decrease in pERK1/2 was matched by a marked reduction in the expression of bFGF, at 24 hours for apocynin and DPI (Figs. 5A, 5C, respectively), and at 48 hours for NAC (Fig. 5B). The different treatment times necessary to see inhibition of bFGF production are most likely due to the efficacy of these different drugs. It is notable that the bFGF antibody used throughout this study normally detected three isoforms of bFGF, but the expression levels of these isoforms varied frequently. Overall, these results strongly indicate that ROS are playing a role in two of the major gliotic events: the phosphorylation of ERK1/2 and increase in bFGF expression. 

Figure 4. 

 

Reactive oxygen species play a role in regulating the phosphorylation of ERK1/2 at early time points. (A) At 0.5 hour postexplanting, Apocynin (Apo) reduces the levels of detectable pERK1/2 and c-fos. (B) At 6 hours postexplanting, pERK1/2 was still decreased with NAC (B) and DPI (C). α-Tubulin and GAPDH were used to determine equal protein in the different blots. Densitometric analysis of the Western blots was undertaken and the graphs represent the mean ± SEM for three independent experiments.

Figure 5. 

 

Reducing ROS results in decreases in bFGF and pSTAT3 at later time points. Western blot analysis shows that apocynin (Apo) at 24 hours (A), NAC at 48 hours (B), and DPI at 24 hours (C) reduce the levels of bFGF expressed by retinal explants. GAPDH was used to determine equal protein in the different blots. Densitometric analysis of the Western blots was undertaken and the graphs represent the mean ± SEM for three independent experiments.

To dissect further the role of pERK1/2 in gliosis, a specific inhibitor of this MAPK pathway was used. UO126 inhibits the upstream activators of ERK1/2, that is, MEK1 and MEK2. ²⁹ Treatment of explants with UO126 decreased pERK1/2 in this explant model, as expected (Figs. 6A, 6B). It also considerably reduced c-fos at 1 hour (Fig. 6A), pSTAT3 at 24 hours (Fig. 6B), and bFGF at 48 hours (Fig. 6B). AG490, an inhibitor of JAK-mediated pathways, was also used. At 24 hours, AG490 had no effect on pERK1/2, but a substantial effect on both pSTAT3 and bFGF (Fig. 6C), proving that pSTAT3 lies downstream of pERK1/2. 

Figure 6. 

 

Inhibition of pERK1/2 through the use of UO126 results in decreased c-fos, bFGF, and pSTAT3 expression. (A) Western blots to show that UO126 (UO) inhibits pERK1/2 levels at 1 hour postexplanting, which resulted in a corresponding reduction in c-fos. (B) At 24 and 48 hours postexplanting, UO also decreased pSTAT3 and bFGF, respectively. (C) The JAK/STAT inhibitor, AG490, decreased pSTAT3 and bFGF, but had no effect on pERK1/2 at 24 hours postexplanting. In all of these, GAPDH or α-tubulin was used as a loading control. Densitometric analysis of the Western blots was undertaken and the graphs represent the mean ± SEM for three independent experiments.

Finally, we wanted to establish the effect of pERK1/2 inhibition on retinal survival, given that its phosphorylation can trigger prosurvival and prodeath pathways. ⁶ Cell death was examined using TUNEL. We previously demonstrated no significant cell death, as detected via TUNEL positive cells, with UO126 at 24 hours in the ONL, ³⁰ so here we extended the time point to 48 hours and included the entire retina (Figs. 7A–D). TUNEL positive cells were counted in each of the retinal layers and there was an increase in cell death at 48 hours with UO126 treatment when compared with untreated controls, but the differences were larger in the INL and ganglion cell layer compared with the ONL (Figs. 7B–D), similar to the results reported by Nakazawa et al. ³¹ with ERK1-deficient mice. This shows that the increase in pERK1/2 in gliosis contributes to retinal cell survival, in general. 

Figure 7. 

 

Inhibiting pERK1/2 increases cell death in retinal explants. (A) Representative images of TUNEL staining from explants, untreated controls (Unt), and treated with UO126 (UO). Scale bar: 50 μm. Counts of TUNEL positive cells from the outer nuclear layer (ONL; [B]), the inner nuclear layer (INL; [C]), and the ganglion cell layer (GCL; [D]). The graphs represent the mean ± SEM for three independent experiments. * indicates where significant differences lay between samples, as determined using one-way ANOVAs followed by Tukey post hoc tests, P < 0.05.

Discussion

The phosphorylation of ERK1/2 is a key signaling event that marks gliosis in several systems. Although it is recognized as important, there is relatively little information as to how this phosphorylation is regulated and which specific pathways are affected. In this study, we establish for the first time that ROS contribute to the regulation of pERK1/2 during gliosis within the retina through the use of the general antioxidant, NAC, and the more specific Nox inhibitor, DPI. When ROS were inhibited, there was reduced phosphorylation of ERK1/2 as well as decreased expression of bFGF. Two transcription factors (c-fos and STAT3) were shown to be possible intermediaries in this pathway (Fig. 8). Importantly, these molecular signaling events were demonstrated to play a role in cell survival (Fig. 7). 

Figure 8. 

 

A schematic diagram to illustrate the order of the pathway from explanting to bFGF upregulation in a glial cell. This sequence was determined from the experiments undertaken in this study. Dashed lines indicate the probable movement of the transcription factors (c-fos and pSTAT3) into the nucleus, where they initiate transcription of bFGF. There is high probability that the bFGF diffuses from the glial cells throughout the retina to promote cell survival elsewhere. Throughout, arrows do not necessarily mean direct interactions between the illustrated molecules, but indicate the overall direction of the pathway.

Gliosis is the response of glial cells within a neuronal tissue to injury and is studied using many models. A previous study demonstrated the similarity between explanting and retinal ischemic-induced gliosis in that both show a comparable decrease in the expression of inwardly rectifying potassium currents and a change in osmotic swelling characteristics of Müller cells. ³² We have added to this by establishing that retinal explanting also replicates gliosis in terms of an increase in pERK1/2 and its expression pattern (Figs. 1, 2). Retinal explanting is, therefore, a useful model of gliosis. It is not surprising that explanting results in gliosis, because it is well recognized that mechanical injury causes gliosis ³³ and the act of explanting is a severe mechanical injury. Mechanical injury has previously been shown to activate ERK1/2, c-fos, and bFGF. ³⁴ Because this injury type is an important activator of gliosis in the retina, we have used explanting as a model system to examine signaling pathways involved in gliosis, specifically the regulation of pERK1/2 by ROS and the downstream effectors of this pathway. 

ROS are generated in stretch models of gliosis, ^35,36 and explanting mimics this as the retina is stretched from its cup shape to being flattened onto a membrane. Through the use of three inhibitors, we demonstrated that by decreasing ROS levels, the phosphorylation of ERK1/2 and related downstream signaling events were also reduced (Figs. 4, 5). Although there is some evidence in the literature demonstrating a correlation between ROS and gliosis, ^9,10 very few examine the signaling pathways linking the two. In other systems, ROS have been shown to lie upstream of pERK1/2, such as when an astrocytic cell line was treated with interleukin 1-beta, ³⁷ but whether this applied to gliosis within the retina was unknown. Phosphatases are known to contribute to the phosphorylation status of ERK1/2, such as protein phosphatase 2A ^38,39 and Shp2, ⁴⁰ both of which can be reversibly oxidized and inhibited by ROS. ^41–43 Therefore, it is probable that the effect of ROS on the phosphorylation status of ERK1/2 was due to the oxidation and inhibition of at least one phosphatase. 

Although not the focus of this study, we add to the knowledge that the ROS associated with gliosis may originate from Nox proteins. The family of Nox proteins generates ROS in response to several different stimuli, which result in prosurvival and prodeath signals, depending on the duration and concentration of the ROS produced. ²⁷ DPI, which inhibits all flavocytochromes including Nox enzymes, reduced pERK1/2 levels and the downstream signaling pathway (Figs. 4, 5). Alone this is not proof that ROS involved in gliosis are generated from Nox, but when taken with other studies from the literature, is highly suggestive that Nox would play a role in the signaling pathways of neuronal injury. ^44,45  

Downstream of pERK1/2, there is only limited evidence that pERK1/2 directly leads to increased bFGF expression, ¹⁷ but plentiful evidence that pSTAT3 plays such a role, ^18–21 and, therefore, both were examined in this study. As the signaling pathway from pERK1/2 to bFGF was dissected, it was surprising to find that pSTAT3 was downstream of pERK1/2 (Fig. 6B). In general these two pathways, JAK/STAT and MAPK/ERK, are seen as separate. However, ERK1/2 can directly phosphorylate STAT3 at its serine 727 site. ⁴⁶ STAT3 can be phosphorylated on two sites, tyrosine 705 and serine 727, both of which play a role in the transcriptional activity of STAT3, although there is debate as to whether the serine site promotes or inhibits this activity. ^46–48 Data presented in Figure 6 suggest pERK1/2 is likely to positively regulate STAT3 transcription activity in gliosis, given that it led to increased bFGF production. c-fos was also examined because it is an immediate early transcription factor that increases when the retina is stretched, ³⁴ and could also be playing a role in this pathway. It also proved to be downstream of pERK1/2 (Fig. 6). Through the series of experiments presented here, we determined that both c-fos and STAT3 are intermediaries in the pERK1/2 to bFGF signaling pathway (Fig. 8). 

Because pERK1/2 led to increased bFGF expression, it was unsurprising to find that pERK1/2 is responsible for cell survival in this system (Figs. 7A–D) and this correlates well with a study by Nakazawa et al., ³¹ where mice deficient in ERK1 displayed more retinal cell death in the INL and GCL upon treatment with N-methyl-d-aspartate, compared with control mice. In a recent review of the role of ERK1/2 in neuronal cell death, it was concluded that the duration of the pERK1/2 response is important (i.e., transient phosphorylation triggers prosurvival signaling, whereas extended activity results in prodeath signaling ⁶ ). Explanting provides another example where the transient upregulation of pERK1/2 promotes cell survival. ROS regulation of this prosurvival pERK1/2 pathway is somewhat contradictory to studies where the use of antioxidants has been shown to protect photoreceptors (e.g., in the rd1 model). ^49,50 However, it is now recognized that ROS, like gliosis and pERK1/2, can have beneficial as well as detrimental effects. ⁷ The apparent contradiction between these two sets of results may possibly be explained by the duration of response. In the rd1 model, there is a sustained increase in ROS levels, ⁵¹ resulting in detrimental oxidative stress, ⁵⁰ and pERK1/2 is increased over 3 days at least (P11–P13, Fig. 1C), but explanting results in the temporary phosphorylation of ERK1/2 (Fig. 2). To date, however, antioxidants have proven to be a valid target for therapeutic development, ^9,10 but this study proves that ROS have beneficial effects in the retina and so antioxidants should be used with caution. 

Acknowledgments

The authors thank the technical team at the Biological Services Unit for help with animal work. 

Footnotes

 Supported in part by Fighting Blindness Ireland, the Health Research Board of Ireland, and Science Foundation Ireland.

Footnotes

 Disclosure: G. Groeger, None; F. Doonan, None; T.G. Cotter, None; M. Donovan, None