Intrinsically photosensitive retinal ganglion cells are resistant to N-methyl-D-aspartic acid excitotoxicity

Animals

Animals were treated in accordance with the regulations of the Cantonal Veterinary Authority of Zürich and with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research. The 129S6/SvEvTac (129S6) mice (Taconic, Hudson, NY), rd10 mice (Jackson Laboratory, Bar Harbor, ME), and CD1 mice (Charles River Laboratories, Wilmington, MA) were housed in a 12 h:12 h light-dark cycle with access to food and water ad libitum. During the light cycle, the light level was maintained at 60 lux.

Intravitreal injections

Injections were always performed between 9 and 11 AM to control for circadian variations in gene expression levels. Mice were anesthetized with a subcutaneous injection (2 μl per gram bodyweight) of a mixture of 510 μl ketamine and 60 μl xylazine. Withdrawal response to paw pinch was tested after 10 min, and when sufficiently immobilized, the animals were prepared for surgery. A surgical needle was used to create a sling for the upper eyelid, and the cornea and conjunctiva were anesthetized locally with oxybuprocaine 0.4% eye drops (Théa Pharma, Schaffhausen, Switzerland). A small area of conjunctiva 1 mm distal to the limbus was dissected away to reveal the sclera, and a sclerostomy was made with a 30-gauge needle. A 30-gauge blunt-end needle on a 10 μl Hamilton syringe was then used to inject 1 μl of 40 mM NMDA (Sigma-Aldrich, St. Louis, MO) in 0.1 M phosphate buffered saline (PBS: 2.622 g of NaH₂PO₄xH₂O; 11.5 g Na₂HPO₄; 8 g NaCl; 0.2 g KCl per 1000 ml H₂O) under light microscopy visualization through the pupil. The fellow eye was injected with PBS alone. Alternately, 1 μl of a mixture of 40 mM NMDA / 1 mM wortmannin (Sigma-Aldrich) in 10% dimethyl sulfoxide (DMSO, Sigma-Aldrich) and PBS or 1 µl of a mixture of 40 mM NMDA / 10 mg/ml AG-490 (LC Laboratories, Woburn, MA) in 50% DMSO and PBS were coinjected. Here, the fellow eye was injected with 40 mM NMDA in 10% DMSO and PBS, with 1 mM wortmannin in 10% DMSO and PBS or with 10 mg/ml AG-490 in 50% DMSO and PBS, respectively. The injection needle was left in place for 20 s before being slowly withdrawn. Care was taken to avoid injury to the lens or retina. After injection, the cornea was dabbed with a cotton swab and coated with a lubricating eye gel (Lacrinorm, Bausch and Lomb, Schenkon, Switzerland). Mice recovered from anesthesia on a heating pad in dimmed light conditions with frequent monitoring and were assessed daily after injection for signs of infection.

Morphology and quantification of retinal ganglion cells

At 6 days post injection, the eyes were enucleated and fixed overnight in 4% (w/v) paraformaldehyde in PBS. After a washing step with PBS, the eyes were dehydrated in a series of increasing ethanol concentrations, washed in xylene, and fixed in paraffin. Semithin sagittal sections (500 nm) bisecting the optic nerve were prepared and stained with hematoxylin and eosin. Sections were analyzed with light microscopy, and cell bodies in the ganglion cell layer were counted from periphery to periphery in two sections per eye and averaged. A total of three eyes (n=3) were analyzed per condition. Erythrocytes and endothelial cells were excluded from counting.

RNA isolation and semiquantitative real time–polymerase chain reaction

Mice were sacrificed at 6 h, 24 h, 48 h, or 6 days post injection. Retinas were isolated through a corneal incision and immediately frozen in liquid nitrogen. Total RNA was extracted using an RNA isolation kit (RNeasy; Qiagen, Hilden, Germany) including a DNase treatment step. Retinas from eyes injected with AG-490 were isolated, and RNA and protein were simultaneously prepared from the same retina: Retinas were homogenized in 200 µl H₂O by sonication (10 cycles; 0.3 s ON (30% output) and 0.7 s OFF) at 4 °C. Immediately after homogenization, 50 µl were added to 450 µl lysis/binding buffer from the High Pure RNA Isolation Kit (Roche Diagnostics, Mannheim, Germany). RNA isolation was performed using the same kit according to the manufacturer’s recommendations. 140 µl of the homogenate were added to 16 µl of 1M Tris-HCl (pH 7.6), and protein concentrations were determined using Bradford reagent and a bovine serum albumin standard. Proteins were used for western blotting as described below.

Reverse transcription was performed using oligo(dT) and M-MLV reverse transcriptase (Promega, Madison, WI). Semiquantitative real-time PCR was used to analyze gene expression of samples in duplicate or triplicate. This was performed with specific primer pairs (Table 1) spanning an exon-exon junction in the RNA of the gene in question, a polymerase ready mix (LightCycler 480 SYBR Green I Master Mix; Roche Diagnostics, Indianapolis, IN), and a thermocycler (LightCycler; Roche Diagnostics). Signals were normalized to β-actin (Actb) and relative expression was calculated with the comparative threshold cycle (ΔΔCT) method using a control sample for calibration [23].

Immunofluorescence on sagittal sections

Treated 129S6 wild-type mice were sacrificed 6 days following injection. Eyes were enucleated and fixed overnight in 4% (w/v) paraformaldehyde in PBS. After the cornea and lens were removed, eyecups were postfixed in 4% paraformaldehyde for an additional 2 h before being transferred to 30% sucrose in 0.1 M PBS at 4 °C for 4–12 h. The eyes were then embedded in tissue-freezing medium (Leica Microsystems Nussloch GmbH, Nussloch, Germany) and frozen in a 2-methylbutane bath cooled by liquid nitrogen. Retinal sagittal sections (12 µm) were cut, placed on slides, and incubated with a blocking solution (3% normal goat serum, 0.3% Triton X-100 in 0.1 M PBS) for 1 h at room temperature (RT). For protein detection, sections were incubated at 4 °C overnight with mouse anti-BRN3A (1:100, cat no. MAB1585, Millipore, Billerica, MA) diluted in blocking solution. After three washes with PBS, slides were incubated with a secondary antibody coupled to Cy3 for 1 h at room temperature, washed, counterstained with 4',6-diamidino-2-phenylindole (DAPI), and mounted with antifade medium (10% (v/v) Mowiol 4–88; Calbiochem, San Diego, CA), in 100 mM Tris (pH 8.5), 25% glycerol (w/v), and 0.1% 1,4-diazabicyclo (2.2.2) octane. Immunofluorescent staining was analyzed with a digitalized microscope (AxioVision; Carl Zeiss Meditec, Dublin, CA).

Western blots

Wild-type 129S6 mice were sacrificed at 6 h or 24 h following injection and the retinas isolated and snap frozen as described above. Retinas were sonified in 0.1 M Tris/HCl (pH 8.0) and analyzed for protein content using Bradford reagent. Protein extracts were mixed with sodium dodecylsulfate sample buffer and incubated for 10 min at 75 °C. Equivalent amounts of proteins were separated with sodium dodecylsulfate–polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes. Membranes were blocked in 5% milk (Bio-Rad, Hercules, CA) in TBST (10 mM Tris/HCl [pH 8.0], 150 mM NaCl, and 0.05% Tween-20) for 1 h at room temperature before being incubated overnight at 4 °C in the same 5% milk solution containing the respective primary antibody. The primary antibodies used were as follows: rabbit anti-STAT3 (1:1000, no. 9132, Cell Signaling Technology, Beverly, MA), rabbit anti-pSTAT3 (1:500, no. 9131, Cell Signaling Technology), rabbit anti-STAT1 (1:1000, no. 9172, Cell Signaling Technology), rabbit anti-pSTAT1 (1:1000, no. 9171, Cell Signaling Technology), rabbit anti-caspase 1 (1:10,000, generous gift from Peter Vandenabeele, Ghent University, Belgium), mouse anti-glial fibrillary acidic protein (1:500, no. G-3893, Sigma-Aldrich), rabbit anti-pAKT_Ser473 (1:2500, no. 9271, Cell Signaling Technology), rabbit anti-AKT (1:2500, no. 9272, Cell Signaling Technology), and mouse anti-β-actin (1:5000, no. A5441, Sigma-Aldrich). Detection was with horseradish peroxidase–conjugated secondary antibodies, and proteins were visualized using a detection kit (Western Lightning Plus-ECL, PerkinElmer Life Sciences, Boston, MA).

Immunofluorescence and quantification of cells in retinal flat mounts

Six days after intravitreal injection, 129S6 wild-type mice were deeply anesthetized with 200 μl ketamine / 60 μl xylazine administered intraperitoneally before being perfused with 10 ml of PBS followed by 20 ml of 4% paraformaldehyde in PBS. Eyes were enucleated, incubated for 5 min in 2% paraformaldehyde in PBS, and transferred to PBS. The eyes were cut along the ora serrata and the cornea and lens were removed. The retina was then dissected from the sclera and flattened by making four radial cuts yielding a cloverleaf shape [24]. Before immunofluorescence analysis, retinal flat mounts were incubated for 1 h (at RT) in 4% paraformaldehyde in PBS and blocked with PBS containing 3% fetal bovine serum and 0.3% Triton X-100 for 1 h (at RT). They were then incubated with the appropriate primary antibodies for 48 h: mouse anti-BRN3A (1:100, cat no. MAB1585, Millipore), rabbit anti-OPN4 (1:500, generous gift from Dr. Ignacio Provencio, Charlottesville, VA), and mouse anti-NR1 (1:500, clone N308/48, NeuroMab c/o Antibodies Inc., Davis, CA). Flat mounts were washed 3 times for 10 min each in PBS and then incubated for 2 h with the respective secondary antibody (anti-mouse-Cy2, anti-rabbit-Cy3, 1:500). They were washed again with PBS before being mounted with antifade medium (Mowiol 4-88 Reagent, Sigma-Aldrich). Immunofluorescent staining was analyzed with a digitalized microscope (AxioVision, Carl Zeiss Meditec).

Quantification of BRN3A-positive cells was performed by counting labeled cells in eight 380 by 610 μm microscopic fields per retina. Fields were located at 700 μm and 1700 μm from the optic nerve head in each retinal quadrant. The cell counts of all eight fields were averaged and extrapolated to the number of cells per mm² using the measured total retina area. As there are far fewer OPN4-positive cells in the retina, these were quantified by counting the total number of labeled cells per whole retina and then converting to cells per mm² as above.

Statistical analyses

Statistical analyses were performed using Prism4 software. Statistical differences of means were calculated using one-way (if three or more experimental groups and one variable) or two-way (if two or more variables) analysis of variance (ANOVA) followed by Bonferroni post-hoc testing. A two-tailed unpaired Student t test was used only when only two experimental groups and one variable were present. P values less than 0.05 were considered significant.

Ganglion cell death after intravitreal injection of N-methyl-D-aspartic acid

We confirmed loss of cells in the ganglion cell layer with light microscopy of sagittal retinal sections at 6 days after intravitreal injection of NMDA (Figure 1A-C), and with immunofluorescence staining for BRN3A (Figure 1D-F). BRN3A is a POU-domain transcription factor expressed in thalamocortical and collicular projecting RGCs. BRN3A is frequently used as an RGC marker, as a decrease in Brn3a mRNA levels correlates with loss of ganglion cells [25-27]. NMDA-treated retinas showed reduced cell density in the GCL (Figure 1G) and probably the INL (not quantified). No difference was observed between PBS-treated and uninjected retinas; they appeared essentially normal. As in previously published studies [11,28-30], we observed a loss of about two-thirds of cells in the ganglion cell layer after NMDA was injected compared to PBS (p<0.001). As already shown by others, this effect was dose dependent [11] (data not shown). Although we did not differentiate between ganglion cells and displaced amacrine cells in the ganglion cell layer, NMDA treatment leads to significant loss of both types of cells in the inner retina, and a loss of cells in the ganglion cell layer strongly correlates with axonal loss in the optic nerve [18].

Expression of Opn4 is not affected by N-methyl-D-aspartic acid injection

To test the sensitivity of the melanopsin-expressing subset of ganglion cells to NMDA toxicity, we analyzed expression of Brn3a and Opn4 mRNA via semiquantitative real-time PCR in wild-type mice at 6 h, 24 h, 48 h, and 6 days after intravitreal injection of NMDA (Figure 2). As expected, expression of Brn3a was strongly reduced starting at 24 h after treatment. Although apoptosis starts as early as 6 h after NMDA injection [28], the decrease in Brn3a mRNA expression at this early time point was not yet statistically significant. In contrast to Brn3a, levels of Opn4 mRNA were unchanged at all four time points after NMDA injection, suggesting either that Opn4-expressing RGCs were resistant against NMDA toxicity or that the surviving cells increased expression as a compensatory reaction. Since Opn4 is expressed in a circadian pattern [31,32], NMDA-treated and control mice of a particular time group were always treated in parallel and analyzed at the same time of day.

OPN4-positive ganglion cells are resistant to N-methyl-D-aspartic acid–induced excitotoxic cell death

To distinguish between resistance against NMDA toxicity and a compensatory upregulation of Opn4 in surviving RGCs, we costained flat mounted retinas of NMDA- and PBS-injected mice for BRN3A and OPN4 (Figure 3A-H). We observed markedly fewer BRN3A-positive cells in NMDA-treated retinas (Figure 3A,C,G) compared to the control retinas (Figure 3B,D,H), but no obvious difference in the number of OPN4-positive cells between the two treatment groups (Figure 3A,E,G,B,F,H). Quantification of BRN3A- and OPN4-positive cells confirmed the mRNA expression data, showing a significantly reduced number of BRN3A-positive cells in the retinas of the NMDA-treated mice (1188±834.6 cells/mm² for NMDA versus 2567±306 cells/mm² for PBS; mean±SD; Figure 3I) while the number of OPN4-positive cells did not change (37.1±2.3 cells/mm² for NMDA versus 35.5±3.3 cells/mm² for PBS; Figure 3J). Thus, Opn4 RNA levels were maintained after NMDA treatment not because of a compensatory upregulation of gene expression but because of the resistance of OPN4-positive ipRGCs to NMDA excitotoxicity.

Intrinsically photosensitive retinal ganglion cell resistance to N-methyl-D-aspartic acid toxicity is independent of genetic background, pigmentation, and the presence of photoreceptor cells

To determine whether the survival of OPN4-positive ipRGCs after NMDA treatment was a phenomenon isolated to the particular strain of wild-type mice used (129S6), we also analyzed Brn3a and Opn4 expression in NMDA-treated albino CD1 mice. Again, NMDA treatment significantly reduced Brn3a but not Opn4 expression (Figure 4A). This observation suggests that the survival of ipRGCs after NMDA is a general phenomenon and is not due to differences in pigmentation or genetic background.

To determine whether ipRGC resistance to NMDA toxicity depended on the presence of regulated glutamate release from bipolar cells and thus on phototransduction-initiated signaling from photoreceptor cells, we injected NMDA in rd10 mice. The rd10 mouse carries a missense mutation in exon 13 of the β-subunit of cyclic guanosine monophosphate phosphodiesterase, and exhibits degeneration of rod and cone photoreceptors beginning at PND16 with almost complete degeneration by PND60 [33]. We injected the eyes of rd10 mice at PND20 when the retinas contained functional photoreceptors and at 6 months of age, when the rd10 mice had virtually no photoreceptor function left. We confirmed degeneration of photoreceptors with semiquantitative real-time PCR for rod (Gnat1) and cone (Gnat2) transducin, which were expressed at the expected levels [34] in each age group (data not shown). Six days after injection, we compared Brn3a and Opn4 expression in young and old rd10 mice. As in the wild-type mice, we observed a significant reduction in Brn3a but not Opn4 mRNA expression in both age groups (Figure 4B,C). This suggests that neither NMDA-induced toxicity to “traditional” ganglion cells nor survival of ipRGCs depended on signaling from photoreceptor cells.

Endogenous rescue and stress pathways are activated after intravitreal N-methyl-D-aspartic acid injection

The JAK/STAT pathway is an endogenous survival signaling pathway activated in response to various inner and outer retinal insults such as photoreceptor injury [16,35] and ganglion cell death after intraocular hypertension [17]. To test a potential role of this signaling mechanism in NMDA-induced excitotoxicity, we analyzed the mRNA levels of several members of the JAK/STAT pathway at various time points after intravitreal NMDA injection (Figure 5A). We found that the Lif and Clc mRNA levels were significantly increased by a factor of 5 and 3.5, respectively, at 6 h after injection. Edn2 and Fgf2 mRNA expression peaked at 24 h, with approximately ten- and threefold greater expression levels compared to the PBS-injected retinas. This was followed by an increase in Stat3 and Gfap expression, which peaked at 48 h. STAT3 is known to have antiapoptotic effects via activation of the suppressor of cytokine signaling (SOCS) family of proteins and the Bcl-2 family [19]. Glial fibrillary acidic protein (Gfap) is a marker for activated Müller glial cells. Several genes encoding proapoptotic proteins also increased expression after NMDA injection: Levels of Stat1 mRNA were significantly increased at 24 h, and caspase-1 (Casp-1) mRNA was threefold and fourfold elevated compared to controls at 24 h and 48 h, respectively. In contrast, monocyte chemotactic protein-1 (Mcp-1), a cytokine involved in recruiting white blood cells to sites of infection or inflammation [36], was similarly expressed in the NMDA- and PBS-treated retinas, even though a tendency for increased expression was detected in NMDA retinas at 24 h after injection.

Activation of some of these molecules after NMDA injection was also detectable at the protein level with western blotting (Figure 5B). At 24 h after injection, we found strongly elevated levels of phospho-STAT3, STAT3, phospho-STAT1, and STAT1 in the NMDA-treated retinas compared to the PBS-injected controls. In addition, expression of glial fibrillary acidic protein and the proform of CASP1 was also increased, although somewhat less robustly than the proteins mentioned above.

Intrinsically photosensitive retinal ganglion cell survival after N-methyl-D-aspartic acid injection is independent of phosphatidylinositol 3-kinase/AKT or STAT3 signaling

In models of optic nerve transection and ocular hypertension, the PI3K/AKT pathway was implicated in enhanced survival of ipRGCs [5]. To test whether this pathway may also contribute to the resistance of ipRGCs against NMDA toxicity, we coinjected NMDA with wortmannin (WM), an inhibitor of PI3Ks, and compared the mRNA levels of Brn3a and Opn4 to retinas treated with NMDA or WM alone (Figure 6A). Although Brn3a levels were decreased with NMDA and NMDA plus WM injections as expected, Opn4 remained at control levels even in the presence of the inhibitor. To confirm the inhibitory action of WM on AKT activation, we tested levels of p-AKT_Ser473 with western blotting. At 6 h after injection, p-AKT_Ser473 levels were high in NMDA (as previously established [14]), but not in NMDA plus WM injected retinas, indicating that the inhibitor did indeed function as expected (Figure 6B).

Injection of NMDA activated JAK/STAT signaling in the retina (Figure 5), and expression of a constitutively active form of STAT3 protected retinal ganglion cells against ischemia reperfusion in vivo and glutamate toxicity in vitro [19]. We coinjected eyes with NMDA and AG-490, an inhibitor of JAK2, to test whether activation of the JAK/STAT pathway is essential for ipRGC survival in vivo. Coinjection of NMDA with AG-490 reduced phosphorylation of STAT3 compared to injection of NMDA alone suggesting that AG-490 inhibited JAK2 signaling (Figure 6D). However, inhibition of JAK2 did not influence expression of Brn3a and Opn4 after NMDA injection as indicated by the respective RNA levels at 48 h after injection (Figure 6C).

In summary, these results suggest that PI3K/AKT and STAT3 signaling may not be crucial factors in the survival of ipRGCs after NMDA injection.

To verify expression of NMDA receptors on ipRGCs, we treated retinal flat mounts of wild-type mice with anti-NMDAR1 and anti-OPN4 antibodies and analyzed the resulting staining in the GCL. NMDAR1 was widely expressed in cell bodies but not the nuclei of the cells in the GCL (Figure 7A,D). As shown before (Figure 3), OPN4-positive cells were rare but easily detectable (Figure 7B,E). Merged images suggest that OPN4-positive cells also express NMDAR1 subunits (Figure 7C,F).