ARTICLE pubs.acs.org/jpr
Quantitative iTRAQ Analysis of Retinal Ganglion Cell Degeneration after Optic Nerve Crush Mark Magharious,† Philippe M. D’Onofrio,‡ Adam Hollander,‡ Peihong Zhu,§ Jian Chen,§ and Paulo D. Koeberle*,†,‡ †
Graduate Department of Rehabilitation Science, University of Toronto, Canada Division of Anatomy, Department of Surgery, University of Toronto, Canada § Ontario Cancer Biomarker Network, Toronto, Canada ‡
ABSTRACT: Retinal ganglion cells (RGCs) are central nervous system (CNS) neurons that transmit visual information from the retina to the brain. Apoptotic RGC degeneration causes visual impairment that can be modeled by optic nerve crush. Neuronal apoptosis is also a salient feature of CNS trauma, ischemia (stroke), and diseases of the CNS such as Alzheimer’s, Parkinson’s, multiple sclerosis, and amyotrophic lateral sclerosis. Optic nerve crush induces the apoptotic cell death of ∼70% of RGCs within the first 14 days after injury. This model is particularly attractive for studying adult neuron apoptosis because the time-course of RGC death is well established and axon regeneration within the myelinated optic nerve can be concurrently evaluated. Here, we performed a large scale iTRAQ proteomic study to identify and quantify proteins of the rat retina at 1, 3, 4, 7, 14, and 21 days after optic nerve crush. In total, 337 proteins were identified, and 110 were differentially regulated after injury. Of these, 58 proteins were upregulated (>1.3), 46 were downregulated (500 μm. A total of four equally spaced sections through the width of each optic nerve were examined and quantified using a Leica DM LFSA microscope (20 objective) with an Andor iXon 885þ camera, with EM gain applied. The total number of regenerating axons per section in each bin was then averaged, and statistical analysis was performed by ANOVA and Tukey’s post hoc test. Retinal Processing for 8-plex iTRAQ Proteomics
iTRAQ proteomics was performed using an 8-plex procedure so that all experimental samples were processed simultaneously in the same iTRAQ run, together with the normal-unlesioned retina sample that was used as the baseline standard for comparison. Animals were anesthetized with isoflurane and killed by cervical dislocation at 1, 3, 4, 7, 14, or 21 days after crush (n = 8 for each time-point). An additional group of normal animals (n = 8) was used as the baseline unlesioned-normal for iTRAQ analysis. We did not use the contralateral eyes of experimental animals as “normals” for comparison because both microglia and astrocytes in the contralateral-unlesioned eye are activated and proliferate following unilateral optic nerve crush,61 and we wanted to ensure that these potential confounding effects did not affect our “normal” group. All protein quantifications for experimental samples were performed relative to the unlesioned-normal sample. Eyes were enucleated, and the cornea and lens were carefully removed in ice cold PBS. The live retinas were then detached from the sclera of the eye cup, placed in a 1.5 mL centrifuge tube, rapidly frozen on dry ice, and maintained at 80 °C until processing. iTRAQ was performed at the Ontario Cancer Biomarker Network facilities in Toronto, ON, Canada. Frozen retinas were thawed on ice, and 250 μL of sonication buffer was added to each retina (Thermo Scientific). The sonication buffer was made up with RIPA buffer (Thermo Scientific) and Protease Inhibitor cocktail I (Calbiochem). Each sample was alternately sonicated and cooled 5 times. Retinal lysates from the same time-point (n = 8) were then combined and incubated at 4 °C for 30 min. Thus, each time-point in the iTRAQ runs represented a sum of eight retinal samples in order to minimize variations due to experimental conditions or intra-animal variability. The normal retina standard sample was similarly a sum of eight normal retinas. The entire iTRAQ analysis via mass spectroscopy was then run
ARTICLE
in duplicate. The combined sonicated samples were spun at 14 000 rpm for 30 min. Detergents were then removed from the samples using a removal kit (Norgen Biotek), and protein concentrations were measured using a Bio-Rad DC protein assay kit. iTRAQ labeling and trypsin digestion were executed according to vendor’s instructions. The 8-Plex iTRAQ labeling kit was purchased from Applied Biosystems, and trypsin was purchased from Promega. A total of 30 μg of protein from each time point was used for labeling. iTRAQ-labeled and trypsin-digested samples were then pooled and subjected to strong cation exchange (SCX) fractionation. Sample Fractionation and LC/MS/MS Analysis
Strong cation exchange HPLC fractionation was conducted on a 2.1 mm ID 150 mm L. Biobasic SCX column (Thermo). The fractions were collected every 2 min during the course of linear gradient elution from solvent A to 50% solvent B over 30 min. Solvent A was 25 mM ammonia formate with 25% acetonitrile with pH value adjusted to 3.0. Solvent B was prepared by adding KCl into solvent A to a concentration of 0.35 M. The iTRAQ labeled samples were pooled and diluted to 2 mL with solvent A before being injected into the SCX column. SCX fractions were zip tipped before being injected into the Nano-LC system. A Proxeon Easy Nano-LC system was used to introduce the sample into the mass spectrometer. An Agilent Zorbax 300SB-C18 (5 0.3 mm) apparatus was used as the trap column. The analytical column was packed in house with a New Objective PicoTip emitter (120 0.075 mm with tip opening size of 15 μm). Column packing material (C18 stable bond, 5 μm, 300 Å) was purchased from Agilent. Flow rate was set at 300 nL/min. A linear LC gradient profile was used to elute peptides from the column. The gradient started with 6% solvent B and was ramped up to 35% B in 99 min. Solvent B was then increased to 65% within 4 min followed by an increase to 95% in 3 min. The 95% B condition was held for 3 min before it was ramped back down to the initial solvent condition. Solvent A contained 2% acetonitrile in water with 0.1% formic acid, while solvent B contained 2% water in acetonitrile with 0.1% formic acid. The LC gradient used is listed below: time
duration
%B
0:00
N/A
6
99:00
99:00
35
103:00
04:00
65
106:00
03:00
95
109:00
03:00
95
109:05
00:05
4
125:00
15:55
4
A hybrid quadrupole/time of flight MS (QStarElite from AB Sciex) instrument with nanospray ion source was used to collect data using Analyst QS 2.0 controlling software (AB Sciex). Each acquisition cycle consisted of one survey scan from 400 to 1200 m/z followed by three product ion scans for the three most intense MS peaks. The MS/MS scans from 100 to 1500 m/z were recorded. The spray voltage was set at 2.1 kV with declustering voltage set at 65 V. The quadrupole resolution was set at “low” during the MS/MS spectra acquisition. Smart IDA features were switched on for automated collision energy and automated accumulation time with 2 s maximum. Ions selected for MS/ MS fragmentation were put into dynamic exclusion list for 180 s. ProteinPilot version 3.0 (AB Sciex) was used for processing acquired iTRAQ data files. The protein identification was performed 3346
dx.doi.org/10.1021/pr2004055 |J. Proteome Res. 2011, 10, 3344–3362
Journal of Proteome Research
ARTICLE
Figure 1. Relative differences in optic nerve crush protein distribution according to cellular location. PANTHER analysis of iTRAQ data showing the relative difference (observed vs expected) in protein distribution by cell location, compared to the entire R. norvegicus proteome. The graphs depict the log-fractional difference between the number of proteins that were identified by iTRAQ (observed) and the normal distribution (expected) for the rat proteome [calculated as (number of observed proteins in a category number of expected proteins)/number of expected proteins]. The positive bars indicate categories that are over-represented in each of the iTRAQ lists, while the negative bars indicate under-represented categories. Three lists were analyzed: (A) the entire set of 337 proteins that was identified after crush, (B) the 58 proteins that were upregulated at a minimum of one time-point, and (C) the 46 proteins that were downregulated at a minimum of one time-point.
by searching raw MS data against the Swissprot-Uniprot protein database (January, 2010), with species restriction to Rattus norvegicus. Default search parameters were used except for the following variable settings: specifying trypsin as the digestion enzyme, fixed modification of methylthio on cysteine residue and iTRAQ 8Plex on N-terminal and C-terminal lysine, biological modification with emphasis on phosphorylation as the variable modification setting. Mass tolerances for precursor and fragments were default values for ProteinPilot that cannot be defined by the users. Cut-off score value for accepting protein identification for ProteinPilot was a ProteoScore of 1.3 (95% confidence). Based on our past experience, the false discovery rate at this cutoff level is at 5% or less. ProteinPilot also provided relative quantification for the iTRAQ analysis. A 1.3-fold change (either up or down regulation) was
used as a cut off value to identify proteins that were differentially regulated across the different time points after crush. Only proteins with 95% confidence in identification were reported. PANTHER analysis was performed on all identified proteins in order to compare the relative amounts of different subclasses of proteins to the known R. norvegicus proteome.
’ RESULTS Characterization of the Retinal Proteome after Optic Nerve Crush
The optic nerve crush model offers a unique opportunity to uncover proteins that are involved in RGC apoptosis/regenerative-failure at key points along a well-established injury time line. 3347
dx.doi.org/10.1021/pr2004055 |J. Proteome Res. 2011, 10, 3344–3362
Journal of Proteome Research
ARTICLE
Figure 2. Relative differences in optic nerve crush protein distribution according to molecular function. PANTHER analysis of iTRAQ data showing the relative difference (observed vs expected) in protein distribution by molecular function, compared to the entire R. norvegicus proteome. The graphs depict the log-fractional difference between the number of proteins that were identified by iTRAQ (observed) and the normal distribution (expected) for the rat proteome [calculated as (number of observed proteins in a category number of expected proteins)/number of expected proteins]. The positive bars indicate categories that are over-represented in each of the iTRAQ lists, while the negative bars indicate under-represented categories. Three lists were analyzed: (A) the entire set of 337 proteins that was identified after crush, (B) the 58 proteins that were upregulated at a minimum of one time-point (blue bars), and the 46 proteins that were downregulated at a minimum of one time-point (red bars).
When the optic nerve is cut, RGC apoptosis is delayed until ∼4 days after injury, the reasons for which remain unclear. After 14 days, the rate of cell death decreases markedly and the surviving cells live for extended periods of time. We studied the retinal proteome in order to reveal proteins that may be involved at different stages of degeneration: early proteome alterations that precede apoptosis and regulate the early retraction of injured axons (1 day), events that immediately precede the induction of apoptosis and occur during axon sprouting (3 days), the initiation of the rapid phase of apoptosis (4 days), the peak of RGC apoptosis and the time-point when surviving RGCs are actively regenerating axons (7 days), and the end of the rapid phase of cell death where regenerative failure is apparent (14 and 21 days).
iTRAQ analysis of normal retinas and injured retinas at six different time-points after optic nerve crush resulted in the identification and quantitation of 337 proteins with 95% confidence. Of these proteins, 110 were differentially regulated after optic nerve crush. A total of 58 proteins were upregulated (>1.3 relative to normal), whereas 46 proteins were downregulated (
