We used a multidisciplinary approach to investigate the temporal, intercellular and intracellular signaling that immediately follow optic nerve injury. Our hypothesis was that there are cellular events in different cells in the retina very early after optic nerve injury. Our intent was not to investigate all pathways or any one pathway in-depth, but to find multiple signaling pathways, that would be representative of sequential changes. Our data provide a temporal, sequential framework of early events within the first 6 hrs after optic nerve injury. Previous studies have investigated changes in selected protein phosphorylation [1] or gene expression [2, 3] at one day to several weeks after optic nerve injury. Thus, our data provides the first observation of responses in the neural retina as early as 30 min after axonal injury.
As answers to the questions that were raised in the Introduction of this paper, we believe that:
1. The soma of the RGC "senses" that its axon has been injured within 30 min. This interpretation of our data is based on the dramatic de-activation of the phosphorylation state of ERK1 in the Muller cells within 30 min. Muller cells and astrocytes express activated ERK-1 in the retina, and these cells express higher levels in retinas from glaucomatous donors [24]. Muller cells wrap around the somas of RGCs and have many points of contact. We hypothesize that within 30 min, the somas of the RGCs signal the Muller cells, which lead to loss of the activation/phosphorylation state of ERK1 in the Muller cells. If there are changes in Muller cell activity by 30 min post axonal injury, the likely source of the signals to affect these changes is the somas of the RGCs with damaged axons. Thus, the somas of the RGCs have presumably "sensed" that a cellular event has occurred.
2. The somas of the RGCs probably signal that a catastrophic event has occurred to many neurons and glia in the retina within 30 min. Surprisingly, 30 min after acute injury, when Muller cells have lost ERK-1 activation/phosphorylation, ERK-1 activation/phosphorylation simultaneously appears in the outer plexiform layer, the location of the photoreceptor synapses, in the inner nuclear layer and in the inner plexiform layer.
3. Cell death signals are apparent within 6 hrs following injury to the RGC axon. These death signals include: an increase in TNFα production and changes in phosphorylation of related TNFα pathway proteins (CASP8AP2 (Flash), TTRAP, SDCCAG3, JNK, CARD9, and DAP2IP). DABIP2 (AIP1) [25, 26] is directly involved in signaling from the TNFα receptors to activate JNK, while SDCCAG3 is involved in receptor trafficking [27]. These signals may not be sufficient to induce cell death at 6 hrs but may be part of the early events that lead to programmed cell death. The lack of activation of the NFκB survival pathway is consistent with changes in the phosphorylation of other TNFα pathway components, (eg TTRAP, CARD9, CASP8AP2) that can negatively regulate NFκB activation [28–30]. Finally, protein kinase MAST2, which is upregulated 6 hrs post optic nerve crush (Table 3), interacts with TRAF6 in the TNFα pathway so as to decrease activation of NFκB [31].
4. There are nuclear events causing new protein synthesis within 6 hrs following injury to the RGC axon. Two new proteins, for example, BAX and AIFM3, are associated with the initation of programmed cell death; whereas, another, RTN4, sequesters the antiapoptotic BCL-2 protein [32]. The decrease in histone methylation (Figure 5) and the upregulation of transcription factors (Table 3) at 6 hrs after optic nerve crush are consistent with changes in transcriptional activity. Furthermore, phosphorylation of H2A at Ser-139 is associated with release of H2A from the nucleus and apoptosis [18].
The increase in expression of Socs3 can be related to JAK1 activation [33] and cytokine induced degeneration. For example, Socs3 is also upregulated in the neural retina upon light-induced injury [34]. In these studies, the activity of one or more of the IL-6 family of cytokines (IL-6, CNTF, NP, LIF, CLC) was the stimulus for Socs3 upregulation. We did not detect upregulation of any single member of the IL-6 family in our microarray data. Thus, as with light-induced injury, multiple cytokines likely lead to the increase of Socs3 expression.
The increased expression of Bax and other genes associated with programmed cell death is consistent with the beginnings of a pro-apoptotic program that eventually translocates BAX [5, 6] and the related proapoptotic protein BIM [4, 35] to the mitochondria to induce death in RGCs. Upregulation of BAX protein has been shown to persist after optic nerve crush [1, 6]. Although a detailed promoter analysis has not been reported, BAX upregulation has been linked to JNK activation [36] that we observed within 6 hrs of optic nerve injury. Bax knockout mice are more resistant to RGC cell death after optic nerve crush, but not to degeneration induced by glutamate excitotoxicity [37]. RGC cells in Bim knockout mice are also protected from optic nerve axotomy-induced death [4]. The proapoptotic activity of Bim is negatively regulated by ERK-1 phosphorylation, while phosphorylation by JNK enhances Bim activity possibly by dissociation from intracellular sequestration [38]. Phosphorylation of BIM by ERK-1 causes its degradation by the proteosome [39] so that the regional differences we observe in ERK-1 and JNK activation (Figs 3 and 4) could affect Bim levels in various cell types.
In addition, we note that there is limited survival signaling in the retina immediately after optic nerve injury. Previous studies have shown that survival signaling by IGF-1 through the phosphoinositide-Akt pathway begins to decrease within two days after optic nerve crush [1]. The loss of IGF-1 signaling may be due to the upregulation of Socs3 (Table 3) which is known to antagonize this pathway [40] and interacts directly with the IGF-1 receptor [41]. The changes in glutamate receptor phosphorylation that we observed after optic nerve crush suggests that altered Ca2+ signaling is part of the degenerative process. Brain-derived neurotrophic factor (BDNF) is an important trophic factor for RGC cells and has been shown to be neuroprotective in RGC injury paradigms (Reviewed in [42]). However, the upregulation of Camk2 and related Ca2+ signaling (Tables 2 and 3) antagonizes the trophic activity of BDNF [43]. Thus, application of BDNF, IGF-1, and related factors [44] may be of only short-term benefit after optic nerve injury.
Based on our data we offer the following hypothesis: The soma of the RGC "senses" that its axon is damaged within 30 min and signals the Muller cells, which, we believe, signal the entire retina that a catastrophic event has occurred. Furthermore, within 6 hrs of damage to the optic nerve, death signals are present in the retina that will ultimately lead to RGC degeneration. The temporal rapidity with which these events occur suggest that attempting to interfere with programmed cell death at a later time may be fruitless and, perhaps, not possible
Experimental Procedures
Animal model
Retinas were obtained from male C57BL/6J mice. All experiments were performed in accordance with the U.S. Public Health Service Policy on Humane Care and Use of Laboratory Animals, the National Institutes of Health Guide for the Care and Use of Laboratory Animals, the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and institutional, federal, and state guidelines regarding animal research. Groups of 5–9 mice were subjected to optic nerve crush. Axons of the optic nerve were crushed with fine forceps for 10 sec, 2 mm posterior to the globe, under direct visualization, within an intact meningeal sheath. By performing the optic nerve crush 2 mm posterior to the globe, the injury is distal to the entry of the ophthalmic artery into the optic nerve. Thus, care is taken to not disturb the retinal blood supply. Optic nerve crush has been widely used by many laboratories and is well documented in the literature [45, 46]. At the desired times (30, 60 and 6 hrs after crush) eyes were enucleated and neural retina removed and frozen at -80°C. Controls (0 min) were contralateral eyes that had not been injured from the same animals in each group.
Preparation of retinal extracts
Retinal tissue (55–100 mg) was homogenized in 150 μl of lysis buffer (20 mM Tris-HCl, pH 7.4, containing 2 mM EDTA, 150 mM NaCl, 1% (v/v) NP-40, multiple protease inhibitor cocktail (Roche Applied Science, Indianapolis, IN) and 1 mM sodium orthovanadate). A motorized pestle was used in 3 × 20 sec bursts on ice. The mixture was centrifuged at 15,800 × g for 20 min at 4°C. The supernatant was removed and the pellet re-extracted with 50 μl of lysis buffer with mixing by up and down action of a pipette. The mixture was centrifuged again and the supernatants combined. Protein concentrations were measured (BCA reagent, Pierce Chemical Co.).
Strategy
The initial screening for changes in phosphorylated proteins was done using affinity capture methods coupled to mass spectrometry for protein identification. These methods included anti-phosphotyrosine beads for enrichment of tyrosine phosphorylated proteins followed by separation of the captured material using one-dimensional gel electrophoresis. We used metal-ion chelate chromatography of phosphopeptides obtained from proteins that were not captured by anti-phospho-tyrosine antibodies. We did these experiments at multiple points post injury to attempt to capture a broad spectrum of events in cellular signaling. Although the method used was only semi-quantitative, it lends itself to detection of changes in multiple phosphoproteins for each experimental time point.
Isolation of phosphoproteins/peptides
Tyrosine-phosphorylated proteins were isolated by immunocapture with anti-phosphotyrosine antibody 4G10 conjugated to agarose beads (Millipore Billerica, MA). This antibody has been used previously to characterize tyrosine phosphoryated proteins in stimulated cell systems [47] and for quantitative phosphoprotein detection [48]. Lysate containing 1 mg of protein was diluted 1:10 with lysis buffer without NP-40 detergent and applied to 50 μl of a slurry of 4G10-conjugate. The antibody binding reaction was incubated at 4°C with gentle rocking for 16 to 24 hr. Beads were pelleted by centrifugation (2000 rpm, 3 min). The supernatant was removed and saved for subsequent digestion and isolation of additional phosphopeptides (see below). The beads were washed two times with 50 μl of lysis buffer without NP40 and the washings combined with the original supernatant. The beads were washed with lysis buffer (500 μl) without NP-40 and the supernatants discarded. Proteins were eluted from the beads by applying 50 μl of SDS-PAGE sample buffer (Invitrogen) and heating to 95°C for 10 min. After brief centrifugation, the supernatants were removed and applied to individual lanes of a 4–12% polyacrylamide gel (Biorad Criterion, Hercules, CA) and electrophoresed at constant voltage. Gels were stained with Simply Blue stain (Invitrogen, Carlsbad, CA) and de-stained in water.
In gel digestion of phosphotyrosine antibody-captured proteins
Each gel lane was cut into 10 bands (4 mm × 6 mm) and further chopped into ~1 mm pieces and transferred to 1.5 ml Eppendorf tubes. Gel pieces were washed with 50 mM ammonium bicarbonate, 50% acetonitrile solution (3 × 15 min), and then in 100% acetonitrile. After removal of the solvent and drying in a Speed Vac (Thermo-Fisher, Pittsburgh, PA) concentrator, gels were rehydrated with 70–80 μl of 50 mM ammonium bicarbonate containing 0.01% (w/v) trypsin (Promega, Madison, WI). After incubation at 37°C (16–24 hr), the reactions were stopped by adding 1 volume of 5% (v/v) trifluoroacetic acid. The supernatants were removed and gel pieces further extracted twice with 100 μl of 0.1% trifluoroacetic acid/60% acetonitrile for 30 min. Combined extracts were then evaporated to dryness with a Speed Vac concentrator. The residues were dissolved in 20 μl of 0.1% formic acid/10% v/v acetonitrile).
Isolation of additional phosphopeptides from retinal extracts
The flow-through or non-bound fraction from the antiphosphotyrosine capture step (above) was denatured by addition of an equal volume of 6 M guanidine hydrochloride solution. Protein disulfides were reduced with triscarboxyl-ethylphosphine (Sigma-Aldrich, St. Louis, MO) (12.5 mM) at room temperature for 1 hr. To each sample, iodoacetamide (in acetonitrile) was added to a final concentration of 25 mM and the reactions incubated in the dark for 1 hr. The solution was then transferred to a dialysis cassette (10,000 MWCO) and dialyzed against 50 mM ammonium bicarbonate (1 L) at 4°C. The dialysis buffer was changed 3–4 times over 24 hr. The retained fraction was then concentrated in the Speed Vac to 0.5 ml and then trypsin was added to a final concentration of 0.01% and incubated at 37°C for 20 hr. The reactions were stopped by adding 10 μl of acetic acid. The reactions were dried on a Speed Vac concentrator and re-dissolved in 200 μl of 0.1% formic acid, 10% acetonitrile. The OD280 of each solution was measured after 100-fold dilution with water. A volume equivalent to 150 OD280 units of each sample was then diluted to 200 μl with 5% acetic acid and applied to a Ga-conjugated phosphopeptide isolation cartridge (Pierce, Rockford, IL) that had been rehydrated as per the manufacturer's instructions. The flow-through from the cartridge was re-applied and then washed with 200 μl of 0.1% acetic acid, 2 × 100 μl of 0.1% formic acid/10% acetonitrile. Peptides were eluted with 2 × 75 μl applications of 0.1 M ammonia in 10% methanol. One hundred μl of 5% acetic acid was then added to the eluates and the samples evaporated to dryness on the Speed Vac concentrator. The residue was re-dissolved in 25 μl of 0.1% formic acid/10% acetonitrile.
Mass Spectrometry
Peptides from the gel bands of tyrosine phosphorylated proteins were separated on a nano-flow column (0.075 × 50 mm, C18 Agilent Santa Clara, CA). The column was eluted at 0.3 μl/min with a gradient of 20 to 65% (v/v) acetonitrile in 0.1% formic acid. Peptides from the phosphopeptide isolation cartridge were separated using automated two-dimensional chromatography on the Agilent 1100 chromatography system. In the first dimension, phosphopeptides were injected onto a sulfated-ion exchange column (PolyLC 0.3 mm × 50 mm (Columbia, MD). Peptides were then eluted with steps of 0.05, 0.1, 0.2 and 0.4 M ammonium acetate onto a nanoflow column (0.075 mm × 150 mm, Agilent). The column was eluted at 0.3 μl/min with a gradient of 5–65% acetonitrile in 0.1% formic acid. Peptides were detected using an Agilent 1100 XCT ion-trap LC-MS system as described previously [49]. Each sample from the gel band isolation was run twice.
Data Analysis
Peptide mass and fragmentation data were filtered and database searching done using Spectrum Mill software (Agilent) as described previously [49]. We used the mouse International Protein Index (IPI) database for searching the mass spectrometry data generated from each sample. From each sample run, a curated list of hits was obtained. This list was based upon the database score and the quality of mass spectral data. For comparisons, we used only those hits that were identified at least twice in each sample. The MS-data intensity values for each peptide were then averaged over the three time points (3 0 min, 60 min, 6 hrs) and divided by averaged control values. The final list of identified proteins is summarized in Additional file 1. This spreadsheet contains the sequences of phosphopeptides identified by mass spectrometry data. Also included is the analysis for the presence of phosporylation sites using Phosphosite [50].
Immunohistochemistry
Enucleated eyes were fixed in 2% wt/vol paraformaldehyde in 0.01 M phosphate buffered saline (PBS; pH 7.4) at 4°C overnight. Six animals were used for each group in all immunohistochemistry experiments. Immunohistochemistry was performed on paraffin sagittal sections of retina for pJNK (Cell signaling, Danvers, MA) using the Vectastain Elite ABC kit (Vector Laboratories, Burlingame, CA) and diaminobenzidine as a substrate. As a negative control, sections were treated in the same manner, except that incubation with primary antibody was omitted. The sections were treated for equal time in DAB reagent and photographed at the same time. Phosphorylated GluR1 (Ser-845, Zymed, Billerica, MA), phosphorylated ERK-1 (Cell Signaling) SOCS3 (Santa Cruz P-19, Santa Cruz, CA), BAX (Santa Cruz H103), and glutamine synthase (Chemicon-Millipore) were detected using Rhodamine-labeled secondary antibody (Vector Labs). Nuclei were stained with DAPI. Sections were washed, mounted and viewed under a fluorescence microscope (Olympus, AX70). As a negative control, sections were treated in the same manner, except that incubation with primary antibody was omitted.
Isolation of retinal ganglion cell layer by laser capture microscopy (LCM)
The LCM system that we used (Veritas Microdissection System, Molecular Devices, Sunnyvale, CA) obviated the need for tissue dehydration prior to microdissection. This enabled us to isolate quiescent retinal ganglion cells directly from 8-μM frozen sections of mouse eyes, thereby increasing the yield and quality of RNA (see below). Frozen sections, mounted on special membrane-coated slides (P.A.L.M. Microbeam), which facilitated the capture of cells, were briefly (1 min) stained with hematoxylin (HistoGene™; Arcturus, Mountain View, CA). In addition to visualizing the retinal structures, this staining procedure also removed the OCT mounting medium. Labeled sections were tracked with intergral light microscope using a 20× objective. Retinal ganglion cells to be isolated were outlined with a "light" pen or cursor on a monitor screen. Such an outline defined the area that would be cut and catapulted intact into a Capsure Macro LCM cap (Molecular Devices, Sunnyvale, CA). In this manner, approximately 6000 cells from the ganglion cell layer were isolated from each eye.
Total RNA Isolation and cRNA Amplification
Total cellular RNA from LCM-captured cells was isolated and purified (Pico-Pure™; Arcturus). Samples of the total starting RNA were analyzed by capillary electrophoresis (Agilent Technologies, Palo Alto, CA) to assess the degree of purification. Approximately 60 ng of total cellular RNA could be extracted from 6000 cells from the ganglion cell layer that were isolated by LCM. When this RNA was contrasted with commercially prepared total RNA from mouse liver using picogram chips and a Bioanalyzer (Agilent Technologies), sharp bands corresponding to the 18 S and 28 S RNA were observed for all samples RNA quality was further assessed by calculating the RNA integrity number, which is based on a proprietary Agilent Technologies algorithm. Total RNA from the isolated cells was subjected to cRNA amplification. Briefly, 1.5 rounds of cRNA amplification were accomplished using a Ribo-Amp® OA RNA amplification protocol (Arcturus). First strand cDNA was generated by reverse transcription using the total RNA. After the second strand cDNA was synthesized, a T7 RNA polymerase-driven cRNA synthesis was performed to obtain the first round of cRNA amplification. A second double strand cDNA synthesis was performed followed by a second round of cRNA amplification. A BioArray™ HighYield™ RNA transcript labeling protocol (T7) (Enzo Life Sciences, Famingdale, NY) was employed for the second round of amplification to biotinylate the cRNAs. To examine the reaction quality, an aliquot from the first stand cDNA synthesis in the first round cRNA amplification and another from the second strand cDNA synthesis in the second round amplification sample were removed for real-time PCR analysis.
Microarray Analysis
Amplified RNA (8 μg) was hybridized to Affymetrix (Santa Clara, CA) Mu430 v2.0 chips and processed as recommended by the manufacturer. Three chips each were used for the control and 6 hrs crush samples. For each dataset, invariant set normalization was performed using the PM/MM model for calculating signal intensities in dChip 2006 [51]. Thresholds for selecting significant genes were set at a relative fold-difference of > 1.2, absolute intensity difference between sample and baseline > 80, and paired t-test value of p <0.05. Genes meeting all of these criteria were considered as significantly different. This resulted in a list of 239 differentially expressed genes (160 upregulated and 79 down regulated, Additional file 2). Microarray data have been deposited in the GEO database with the series accession number GSE11862.
ELISA
ELISA assays for phosphoproteins were done with sandwich ELISA kits (Cell Signaling) following the manufacturer's instructions. These were used to detect phosphoJNK (Thr183, Tyr185) and phospho IκB (Ser32) in soluble tissue extracts. Briefly, extracts were diluted with the assay buffer to the desired total protein concentration (0.25, 0.5, 1.0 mg/ml) to 100 μl and applied in duplicate to the wells of the ELISA plate containing the capture antibody. Controls without added lysate were included in all assays. The plates were covered and incubated at 4°C for 12–16 hrs to allow binding of the target protein to the plate. Wells were then washed 4 times with the wash buffer supplied in the kit. The wells were then covered and incubated with the antiphosphoprotein antibody and incubated for 1 hr at 37°C. The plates were washed again and then incubated with horse radish peroxidase (HRP) linked detector antibody for 30 min at 37°C. The plate was washed again and then incubated with HRP substrate solution for 30 min at room temperature. Stop solution was added and the absorbance of converted substrate read at 450 nm in a plate reader.
Mouse TNFα was measured in soluble tissue extracts using an ELISA kit (Pierce). Briefly, tissue extracts were diluted to 0.25 or 0.5 mg/mL and 50 μl applied in duplicate to the ELISA plate. TNFα standards over the range of 35 to1225 pg/ml were measured in duplicate along with the samples. Then 50 μl of biotinylated antiTNFα antibody was added to all of the wells and the plate covered and incubated for 2 hr at room temperature. The plate was washed 5 times and then the wells were incubated with 100 μl of HRP-streptavidin solution for 30 min at room temperature. The reaction was stopped by adding an equal volume of acidic (2N HCl) stop solution. The absorbance was then read at 450 nm in a plate reader.