Adaptive Changes in the Neuronal Proteome: Mitochondrial Energy Production, Endoplasmic Reticulum Stress, and Ribosomal Dysfunction in the Cellular Response to Metabolic Stress

All standard growth media used was obtained from Invitrogen (Paisley, UK), unless otherwise stated, and all procedures were conducted in compliance with all relevant legislation in the United Kingdom. Cells were routinely maintained in standard 75 cm² cell culture-treated flasks (Greiner Bio One, Stonehouse, UK) within a continuous flow incubator at 37°C, with 5% CO₂ in fully humidified air. All growth media and solutions were prewarmed to 37°C before contact with cells, unless otherwise stated. The acetonitrile and water used in the LC-MS buffer solutions were of HPLC grade (Fisher, Loughborough, UK). Formic acid was of 98% to 100% purity (Merck, Darmstadt, Germany) and trifluoroacetic acid was 99% sequencing grade (Sigma-Aldrich, Gillingham, UK). All other chemicals used in the processing of the protein samples were of reagent grade or better. The trypsin used for protein digestion was from Promega (Southampton, UK).

SH-SY5Y cells (ATCC, Manassas, VA, USA) were cultured in Dulbecco's Modified Eagle Medium: Nutrient Mixture F12 (DMEM/F12) supplemented with 2 mmol/L L-Glutamine, 100 U/mL Penicillin, 100 mg/mL Streptomycin, and fetal bovine serum. Cells were routinely maintained at 37°C, 5% CO₂, with culture medium replaced biweekly. Oxygen glucose deprivation (for 3, 6, 9, 12, and 18 hour durations) was achieved through incubation of the cells at 37°C in 95% N₂:5% CO₂. Aglycemia was induced by replacing standard culture medium with glucose-free DMEM supplemented with fetal bovine serum, after a glucose-free wash step. Control cells underwent the same glucose-free wash step, followed by incubation with glucose containing media (4,500 mg/L glucose). The osmolality of the DMEM solutions used was within the range of 300 to 340 mOsm/kg for the no glucose solutions, and 320 to 360 mOsm/kg for the glucose containing media. All media were equilibrated overnight before experimentation in either normoxic or anoxic chambers. Oxygen content in the media was recorded in preliminary studies using a fiber optic dual oxygen-temperature sensing probe (OxyLite 2000 E Series, Oxford Optronix, Oxford, UK). A rapid oxygen depletion phase was recorded, with oxygen partial pressure falling exponentially from 140 to 7 mm Hg over the initial 3 hours of OGD. Oxygen depletion plateaued and remained severely hypoxic between 4 and 18 hours with a recorded average partial pressure of 2 mm Hg.

Cells were seeded onto poly-D-lysine coated flat-bottomed 96-well plates at a density of 125,000 cells per well overnight. After OGD, glucose-free DMEM was replaced with 200 μL glucose-containing DMEM and cell viability was assessed using 20 μL of CellTiter 96 AQueous One Solution Cell Proliferation Assay (MTS) (Promega, Madison, USA) as per the manufacturer's guidelines. Cells were incubated for 2 hours in glucose containing, normoxic media (37°C, 5% CO₂) to allow residual mitochondrial function after OGD to be assessed using a Dynex MRX plate reader (Dynex Technologies Ltd., Worthing, UK) at 490λ.

Trypan blue staining was carried out to assess the extent of cell death occurring after each OGD time point. Cells were collected immediately at the end of each time point by trypsinization, centrifugation (200 g, 7 minutes) and resuspension in glucose-containing DMEM. Cells were incubated with 0.1% trypan blue as per the manufacturer's instructions (Invitrogen). A hemocytometer was used to count stained and unstained cells. The percentage of stained cells in each OGD and time-matched control sample was calculated.

SH-SY5Y cells were treated with 6 hours OGD (n = 6 independent samples per group) and 18 hours OGD (n = 6), with a time-matched control for each sample. After OGD or ‘control’ incubations, SH-SY5Y cells were immediately scraped and centrifuged (200 g, 7 minutes). Cell pellets were washed in Hanks Buffered Saline Solution and recentrifuged. Washed cell pellets were resuspended in 1 ml Hanks Buffered Saline Solution and centrifuged (6,000 g, 3 minutes). Supernatants were removed and the pellets snap frozen for temporary storage at −80°C. Cell pellets were defrosted and lysed in 100 μL 8 M urea using a cordless motor pellet pestle (Sigma-Aldrich, Dorset, UK). Samples were then centrifuged (14,000 g, 15 minutes) and supernatants recovered. Total protein content was determined using the Pierce BCA Protein Assay Kit (Thermo-Fisher, Loughborough, UK). Protein extract (140 μg) from OGD and control samples were digested as previously described.¹⁴

Label-free intensity-based LC-MS was used to identify proteins differentially expressed after 6 and 18 hours OGD relative to time-matched controls. Capillary LC-MS data were acquired on an online system consisting of a micro-pump (1200 binary HPLC system, Agilent, Edinburgh, UK) coupled to a hybrid LTQ-Orbitrap XL instrument controlled through Xcalibur 2.0.7 (Thermo-Fisher). Peptides were reconstituted in 20 μL of loading buffer before injection and 8 μL was loaded. The HPLC separation was provided by a gradient between buffer A (97.5% H₂O, 2.5% acetonitrile, 0.1% formic acid) and buffer B (90% acetonitrile, 10% H₂O, 0.025% trifluoroacetic acid, 0.1% formic acid). The peptide mixture was separated on a 140-minute gradient. The mass spectrometer was operated in ‘data-dependent mode’, with a single MS scan (400 to 2,000 m/z) followed by MS/MS scans in the linear ion trap on the five most abundant ions and excluded for 120 seconds.

All LC-MS runs were analyzed using a label-free intensity-based approach with Progenesis (version 4.0 Nonlinear Dynamics, Newcastle, UK), under randomized conditions. The MS/MS data were searched using MASCOT versions 2.3 (Matrix Science Ltd, London, UK) against a human subset of the NCBI database (downloaded on 12 January 2011) with 34,281 sequences using a maximum missed-cut of 2. Variable methionine oxidation, STY phosphorylation, protein N-terminal acetylation, and fixed cysteine carbamidomethylation were used in all searches.

LC-MS label-free quantification was performed using Progenesis 4.0 (Nonlinear Dynamics). Protein conflict (peptides shared between different proteins) was solved as followed: conflict resulting from multiple sequence assignment to the same peak; we kept only the sequence having the highest score. Conflict resulting from same peptide sequences assigned to different proteins, the assignment was singly attributed to the protein that had the highest number of peptides. Regarding the label-free quantitation, the total number of Features (i.e., intensity signal at a given retention time and m/z) was reduced to MSMS peaks with charge of 2, 3, or 4+ and we only kept the five most intense MSMS spectra per ‘Feature’ . The subset of multicharged ions (2+, 3+, and 4+) was extracted from each LC-MS run and the ions intensities summed for normalization. Protein quantitation was performed as follows; for a specific protein, the associated unique peptide ions were summed to generate an abundance value. The measured protein abundances were transformed using an ArcSinH function (as the method of detection can generate a significant amount of near zero measurement for which a log transform is not ideal). The within-group means were calculated to determine the fold change and the transformed data were then used to calculate the P values using one-way analysis of variance. Significant changes in protein expression (P < 0.01) between control and OGD experimental groups were determined for each time point (6 and 18 hours) using Student's t-test.

All 958 proteins identified by LC-MS by at least two or more peptides were uploaded to the Database for Annotation, Visualization and Integrated Discovery (DAVID) (http://david.abcc.ncifcrf.gov¹⁵) to determine subcellular localization based on gene ontology (GO). Protein clusters founded on subcellular localization with DAVID scores >2 were used. Proteins significantly altered (P < 0.01) after 18 hours OGD were also assessed by DAVID according to subcellular localization.

Protein identifiers of all the proteins significantly altered after 18 hours OGD (P < 0.01) were uploaded with their corresponding fold change values to Ingenuity Pathway Analysis software (IPA; http://www.ingenuity.com). Interactomes were algorithmically generated based on direct relationships (physical interactions and associations) between eligible proteins. The interactomes are color coded with green nodes representing proteins that were downregulated after 18 hours OGD and red protein nodes representing proteins that were upregulated after 18 hours OGD. The shading of the interactome nodes is positively correlated with the magnitude of fold change. The IPA generates a score for each interactome, which is a putative measure of probability (see Deighton et al¹⁶ for critical analysis).

SH-SY5Y cells were exposed to OGD using the same paradigm as the LC-MS study. Samples were collected immediately after 3, 9, and 18 hours of OGD and fixed for a minimum of 20 hours in aldehyde fixative (4% paraformaldehyde and 5% glutaradehyde in 0.08 M sodium cacodylate buffer, pH 7.2). Once fixed, cells were centrifuged at 100 g, the supernatant removed and replaced with a 3% solution of agar (type VII, low gelling temperature; Sigma-Aldrich, Gillingham, UK). The gels were set at 4°C for 5 minutes before being cut into 1 mm³ blocks and processed for electron microscopy (EM). Processing, embedding, and staining were performed, as described previously.¹⁷ Briefly, the blocks were cut into 70 nm thick sections, collected on 300 mesh copper grids, stained with Reynold's lead citrate and contrasted with uranyl acetate.

The EM sections were evaluated on a JEOL CX-100 II transmission electron microscope (Tokyo, Japan). The evaluator was blinded to the treatment condition of all samples. Cells were chosen from grid squares in a predetermined pattern. Fifty cells were assessed from each treatment. For each cell, the condition of mitochondria and rough ER was assessed. Mitochondria were graded as ‘normal’ (cristae clearly visible across >50% of the matrix), ‘swollen’ (with a proportion of the matrix lost) and ‘abnormal’ (with most/ the entire matrix lost). The ER in each cell was graded similarly, i.e., ‘normal’ (where ER outer membranes ran parallel to each other), ‘swollen’ (where modest swelling of ER outer membrane was present), or ‘abnormal’ where ER displayed extensive membrane swelling and distortions. A Chi-Squared 3 × 2 matrix was used to test the null hypothesis that OGD would have no effect on the ultrastructure of organelles.

Normal mitochondrial function was significantly reduced in SH-SY5Y cells after OGD relative to time-matched control cells (P < 0.0001, at each time point analyzed), with longer durations of OGD inducing more severe mitochondrial dysfunction (Figure 1A). Viability of SH-SY5Y cells treated with OGD was quantified using Trypan Blue staining (Figure 1B). Minimal cell death was detected after OGD, with 91% of cells remaining viable after 18 hours OGD.

Ultrastructural morphologic changes in subcellular organelles were analyzed relative to time-matched controls by EM after OGD. With increasing OGD duration, the number of cells displaying abnormal ER was significantly increased (Chi-Square, P < 0.01) (Figures 2A and 2C). SH-SY5Y cells also showed changes in mitochondrial morphology with increasing OGD duration (Figures 2B and 2D). It should be noted that even after 18 hours OGD exposure, only 26% of cells displayed abnormal ER, and only 34% had abnormal mitochondria.

A total of 958 proteins were identified across all samples with two or more peptides by LC-MS (Supplementary Table 1). Minimal protein changes were observed after 6 hours OGD, with only 14 proteins significantly altered in expression level (P < 0.01). After 18 hours OGD, 193 proteins were significantly altered in expression level (P < 0.01) with 130 proteins increased and 63 proteins decreased. For 765 proteins, no significant difference was observed between control and OGD samples after 18 hours OGD.

Subcellular distribution of all the 958 proteins identified by LC-MS was examined using DAVID software. Proteins were detected from across many subcellular compartments, including the nuclear lumen, cytoskeleton, mitochondria, ER, golgi apparatus, ribosomes, and vesicles (Figure 3A).

The ascribed subcellular locations of the significantly altered proteins with OGD (18 hours) indicate that organelles are not uniformly affected by the metabolic challenge. None of the 197 detected nuclear lumen proteins were significantly altered after OGD, while 41% of the 56 detected ER proteins and 32% of the 118 detected mitochondrial proteins were significantly altered (Figure 3B). The significantly increased and decreased proteins after OGD (18 hours) are listed by subcellular location in Table 1 and Table 2, respectively.

Functional protein–protein interactions for the 193 proteins altered with 18 hours OGD were defined using IPA software. IPA-generated 10 protein interactomes (Supplementary Table 2) and three high-scoring interactomes (IPA scores >40) are discussed in context below. These interactomes highlight a shift in mitochondrial energy production, enhanced ER stress, and ribosomal dysfunction.

In all, 38 mitochondrial proteins (out of 118 detected; assigned using the GO term mitochondrion GO:0005739) were significantly altered after OGD (18 hours): 29 were increased and 9 were decreased. VDAC1 and TOM22, proteins involved in mediating the movement of metabolites across the mitochondrial membrane, were significantly reduced after OGD. UQCRC1 and UQCRC2, integral components of complex III of the electron transport chain (ETC), were also significantly reduced (Figure 4A). Coordinated changes in other regions of the ETC were observed after 18 hours OGD treatment: ETC complex V proteins ATP5J and ATP5A1, and ETC complex I proteins NDUFS3 and NDUFA5 were all significantly increased (P < 0.01).

Numerous mitochondrial proteins that have a role in lipid metabolism were also found to be altered after 18 hours OGD. ACADM, the enzyme involved in the first step of fatty acid beta oxidation was significantly decreased (P < 0.01); and ECH1, HSD17B10 and ACAA2, enzymes downstream in the beta-oxidation process were significantly increased. ACAT1, a mitochondrial protein involved in ketone metabolism was also significantly increased.

The highest scoring protein interactome generated by IPA features some of the alterations in energy production described above, namely increases in Complex V proteins (including ATP5J and ATP5A1) and decreases in Complex III proteins (UQCRC1 and UQCRC2) (Figure 4B). This Energy Production Interactome also highlights interactions between altered components of the ETC and other cellular machinery. Both UQCRC1 and UQCRC2 interact with RTN4, a protein predominantly associated with regulating ER function that was significantly upregulated after OGD (Figure 4A). UQCRC2 is also shown to interact with upregulated RUVBL2, a cytoskeletal protein involved in numerous cellular activities including transcription and cellular transformation (Figure 4A). The Energy Production Interactome also contains a cluster of ribosomal proteins that are distinct but intersect with the ribosomal interactome described below.

In all, 23 ER proteins (out of 56 detected; assigned using the GO term Endoplasmic reticulum GO:0005783) were significantly altered after 18 hours OGD: 19 were increased and 4 were decreased. A number of proteins centrally involved in the UPR were significantly increased, including, HSPA5, HSP90B1, PDIA3, and PDIA4 (P < 0.01) (Figure 5A). Other proteins including CALR, UGGT1, ERP29, ERP44, and VCP that are integral to ER stress response and protein folding were significantly increased.

Multiple proteins that have a role in protein degradation were significantly altered after OGD (18 hours). Protein subunits of the 20S and 19S core structures of the 26S proteasome (PSMA5, PSMC3, PSMC4, PSMD1, PSMD2, PSMD6, and PSMD11) were increased. These proteasomal proteins, along with proteins involved in the UPR and heat-shock response after OGD dominate the high scoring, Protein Folding and Degradation Interactome generated by IPA (Figure 5B).

In all, 11 ribosomal proteins (out of 82 detected; assigned using the GO term Ribosome GO:0005840) were significantly altered after 18 hours OGD: 1 was increased and 10 were decreased. The decreased ribosomal proteins were associated with both the 60S (RPL38, RPL23, RPL35, and RPL13) and 40S (RPS1, RPS14, RPS23, RPS9, and RPS17) ribosomal subunits (Figure 5C). These downregulated ribosomal proteins feature prominently in another of the high scoring protein interactomes generated by IPA (Figure 5D). An additionally important feature of this Ribosomal Interactome is the cluster of upregulated peroxidases (PRDX1, PRDX2, and PRDX5) (Figure 5C) involved in redox homeostasis and protecting neurons from oxidative insults (for a review, see Bell and Hardingham, 2011¹⁸).