---

Abstract

A correlation between protein damage and death, but not DNA damage, was found among cells from robust and standard bacterial and invertebrate species. However, the bottleneck in DNA repair efficacy appears more at the level of proteome damage than DNA damage. Here we present a comparative study of SW480 cells derived from a primary colon adenocarcinoma and SW620 metastatic cells derived from the same primary tumor and provide evidence that correlation between death and proteome damage extends to human tumor cells. A higher resistance of SW620 cells, compared to SW480, to killing by UVC light correlates with reduced levels of incurred irreversible oxidative protein damage (carbonylation) related to lower levels of ROS and of proteins intrinsically susceptible to oxidative damage due to imperfect folding. This study provides a concept for sensitizing tumor cells to cancer therapies and assessment of cancer cell fitness.

---

Keywords

Protein oxidation; Cell death; Primary tumor; Metastasis

---

Abbreviations

PTMs: Post-Translational Modifications; ROS: Reactive Oxygen Species; PC: Protein Carbonylation; DMEM: Dulbecco’s Modified Eagle Medium; FBS: Fetal Bovine Serum; PBS: Phosphate Buffered Saline; DNPH: 2,4-Dinitrophenyl Hydrazine; HRP: Horseradish Peroxidase; H2DCFDA: Dichloro Fluorescein Diacetate; DCF: 2’, 7’-Dichlorofluorescein; SDS-PAGE: SDS-Polyacrylamide Gel Electrophoresis; UTCDTT buffer: Urea Thiourea CHAPS DTT buffer; CHAPS: (3-[(3-Cholamidopropyl) Dimethylamonio]-1- propanesulphonate); DTT: Dithiotreitol; RMI: Relative Modification Index; MS: Mass Spectrometry; RSLC: Rapid Separation Liquid Chromatographic system; ACN: Acetonitrile; TFA: Trifluoroacetic Acid; FA: Formic Acid; AGC: Automatic Gain Control; GO: Gene Otology; TGM2: Protein-Glutamine Gamma-Glutamyl transferase 2; PRP19: Pre-mRNA-Processing Factor 19; TRIC: T-complex protein 1 subunit eta; OLA1: Obg-like ATPase 1; DX39B: Spliceosome RNA Helicase DDX39B; PGK1: Phospho Glycerate Kinase 1; IF5A1: Eukaryotic Translation Initiation factor 5A-1; ATPB: Mitochondrial ATP Synthase subunit beta; ENOA: Alpha Enolase; ALDOA: Fructose-Biphosphate Aldolase A; PGK1: Phospho Glicerate-Kinase 1; PRDX2: Peroxiredoxin 2; GDIR1: rho GDP-dissociation inhibitor 1; SERC: Phosphoserine Amino transferase; NPM: Nucleophosmin; OAT: Ornithine Amino transferase, PRS8: 26S Proteasome (regulatory subunit 8); SBSN: Suprabasin; PTPA: Serine/threonine- Protein Phosphatase 2A; CRC: Colorectal Carcinoma; TCA: Tricarboxylic Acid Cycle; PPP: Pentose Phosphate Pathway; TALDO: Transaldolase

---

Introduction

Resistance of tumor cells to diverse therapies is a major hurdle to curing cancer. A recent finding that prokaryotic and eukaryotic cell death correlates with proteome, rather than genome, damage [1,2] led us to seek correlations between tumor cell resilience and proteome damage.

Functional homeostasis requires conformational integrity and stability of cellular proteins that can be influenced by physiological post-translational modifications (PTMs) as well as toxic PTMs, such as those caused by reactive oxygen species (ROS). Oxidative modifications of protein side chains are stable and deleterious to protein function. After the selective breakdown of damaged proteins has occurred, their residual levels can (i) serve as cumulative markers for incurred toxic oxidative stress [3] and (ii) allow for anticipation of severe phenotypic consequences of persistent proteome damage [4]. Irreversible ROS-induced protein damage (carbonylation) accumulates in human and animal cells exponentially with age [5] and may be the root cause of cellular morbidity [4] and death [1,2].

Increase in protein carbonylation (PC) levels can result from: a decrease in the antioxidant defense system, a reduced ability of cells to remove damaged proteins and an increase in protein susceptibility to oxidative modification [6]. Studies on differential proteome resistance to oxidative stress caused by UVC light and by ionizing radiation have shown the same correlation between cell survival and protein carbonylation in bacteria [1] and invertebrates [2].

These findings motivated us to investigate whether the correlation between cell mortality and the extent of protein carbonylation holds also for two tumor cell lines: primary colon adenocarcinoma, SW480, and its derived metastasis, SW620.

Here, we show that the cell line derived from the primary colon adenocarcinoma, SW480, is more susceptible to UVC-induced oxidative protein damage, compared to its metastasis, SW620. The resistance of SW620 cells correlates with lower levels of ROS and of inherently misfolded proteins. Proteomic analyses show increased expression of proteins that promote cell radiation resistance and reveal reduced levels of stress and protein damage in the metastatic cell line correlating with the observed increased resistance and survival. A deeper insight into the patterns of the cellular proteome damage could define biomarkers-predictors of cancer cell fitness before and after therapeutic intervention. Our data suggest that targeting damage to cancer cell proteome could synergize with cytotoxic effects of many standard anti-cancer therapies and stand itself as an effective therapy.

Adenocarcinoma cell lines SW480 and SW620 were cultured in high glucose Dulbecco’s Modified Eagle Medium (DMEM 4500 mg/L) supplemented with 10% of fetal bovine serum (FBS) and 1% of antibiotic (penicillin/streptomycin). Cells were grown at 37 °C in humidified incubator with 5% CO2 and passaged at 80% confluence by 0.05% Trypsin EDTA.

Cells were irradiated after reaching 80% of confluence. Prior to irradiation, the media was removed by aspiration and washed twice with 2 ml of phosphate buffered saline (PBS) for removal of dead cells. For irradiation cells were covered with 5 ml of PBS. UVC irradiation of cells was performed at 37 °C to final doses of 100, 1000, 2000 and 3000 J/m2. After irradiation, cells were scraped from Petri dish, centrifuged 5 min at 2000 rpm and frozen at -80 °C in order to measure the amount of protein damage accumulated during the irradiation.

To measure the amount of protein damage after the recovery from radiation, PBS was removed by aspiration and replaced by supplemented DMEM for 18 h incubation.

The viable cells were counted after 18 h using Trypan blue and a haemocytometer.

Cell pellets were resuspended in PBS containing 1% nonionic detergent Triton X-114 Surfact-Amps (Thermo Fisher Scientific, Rockford, IL, USA) and a mixture of protease inhibitors (Thermo Fisher Scientific, Rockford, IL, USA). Cells were lysed using a mechanical homogenizer and glass beads (425-600 μm). Cell lysates were centrifuged for 20 min at 10000 × g. Samples were then supplemented with 10 mg/100 mL lipid removal agent (13360-U, Sigma Aldrich-Supelco, Bellefonte, PA, USA), incubated for 1 h at room temperature with shaking and pelleted for 15 min at 10000 × g. The amount of proteins in the supernatant was measured by the Micro BCA™ Protein Assay Reagent Kit (Thermo Fisher Scientific- Pierce Biotechnology, Rockford, IL, USA). Protein extracts diluted to 10 μg/mL were loaded into Maxisorp wells (Thermo Scientific Nunc, Rockford, IL, USA) and incubated overnight at 4 °C to allow proteins to adsorb to the surface.

Adsorbed proteins were derivatized using carbonyl reagent 2,4-dinitrophenylhydrazine (DNPH) at the concentration of 12 μg/ mL. Derivatization of adsorbed proteins was followed by detection of derivatized dinitrophenol (DNP)-carbonyl by a rabbit anti-DNP primary antibody (Sigma Aldrich, St Louis, MO, USA) and goat antirabbit secondary antibody conjugated to HRP (Jackson Immuno Research, West Grove, PA, USA). Stocks of antibodies were prepared at ~1 μg/μL and used at 1:7000 dilutions. Subsequent incubation with enzyme substrate 3, 3, 5, 5-tetramethylbenzidine resulted in a colored product that was quantified using a SunriseTM (Tecan, Männedorf, Switzerland) microplate reader with maximum absorbance at 450 nm.

Assessment of ROSFor analysis of basal intracellular ROS, cells were collected, washed with PBS and incubated in the dark with 10 μM 2’,7’-dichlorofluorescein diacetate (H2DCFDA, Sigma) for 30 min at 37 °C and subsequently analyzed on FACS Calibur flow cytometer. In order to measure ROS after exposure to UVC irradiation, 1.5×105 cells per well were plated in 12 well plates. The following day cells were washed three times with PBS and covered with 0.5 ml of 10 10 μM H2DCFCA in PBS. The cells were irradiated with indicated doses of UVC light (0, 100, 1000, 2000 and 3000 Jm-2), collected and resuspended in 10 10 μM H2DCFCA in PBS.

The fluorescence of 10000 cells resulting from the intracellular conversion of non-fluorescent H2DCFDA into fluorescent 2’, 7’-dichlorofluorescein (DCF) was measured in FL1 channel. The collected data was analyzed using FlowJo software version 7.2.5 for Microsoft (Tree Star, San Carlos, CA, USA) to determine the mean green fluorescence intensity after the each treatment. The fluorescence after the staining with H2DCFDA for each cell line was corrected for the auto-fluorescence of the corresponding cell line. The results are expressed as mean fluorescence of 10000 cells of each cell line.

Protein extracts (25 μg) were resolved by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) using 10% polyacrylamide gels and blotted on nitrocellulose membrane (0.45 μm, Thermo Scientific). The membranes were blocked for 1 h with 5% nonfat milk (Carl Roth GmbH, Karlsruhe, Germany) in TBS containing 0.1% Triton X-100 (Sigma Aldrich, St Louis, MO, USA) and probed with primary antibodies diluted in blocking buffer over night at 4 °C with shaking.

The following primary antibodies were used: rabbit anti-HSP70 (1:20000, Abcam, Cambridge, MA, USA), rabbit anti-HSP90 (1:200, Abcam, Cambridge, MA, USA), rabbit anti-HSP105 (1:20000, Abcam, Cambridge, MA, USA), rabbit anti-MnSOD (1:2000, Enzo Life Sciences, Farmingdale, NY, USA), rabbit anti-Cu/ZnSOD (1:2000, Enzo Life Sciences, Farmingdale, NY, USA) and mouse anti-GAPDH (1:600, Millipore, St. Luis, MO, USA). The recognized proteins were detected using horseradish peroxidase-labeled anti-rabbit IgG (1:20000, Jackson Immunoresearch, West Grove, PA, USA) or antimouse IgG (1:10000, Jackson Immunoresearch, West Grove, PA, USA) produced in goat [7]. Western blots were developed using enhanced chemiluminiscence (Luminol, SantaCruz Biotechnology, Santa Cruz, CA, USA) and quantified using ImageJ.

Cell lines were treated with UVC irradiation killing 90% of the population: 1835 J/m2 for SW480 and 4000 J/m2 for SW620. Irradiation was performed as above and cell pellet quadruplicates were stored at -80 °C.

Protein extraction was performed on pellets containing 5×106 cells by using 150 μL of UTCDTT buffer containing 8 M urea, 2 M thiourea, 4% CHAPS (3[(3-cholamidopropyl)dimethylamonio]- 1-propanesulphonate) and 10 mM dithiotreitol (DTT). Samples were homogenized and sonicated for 10 s (20 kHz). After overnight incubation on Thermomix (Eppendof, Wien, Austria) at 1200 rpm and 4 °C, cell lysates were centrifuged at 15000 × g for 2 h at 4 °C and protein concentration was determined using Bradford assay [9].

Each sample (100 μg of protein extract) was derivatized with 5 μL of 12 mM CF647DI Hydrazide fluorescent dye (Biotium Inc, Hayward, CA, USA) for 1 h at 25 °C. To remove the unreacted dye, proteins were precipitated with 20% TCA and 0.2% deoxycholate for 15 min at 4 °C. The pellet was washed once with 20% DMF in acetone and stored for 1 h at -20 °C. Subsequently, the pellet was washed for the second time with 100% acetone and stored for 30 min at -20 °C. Dry pellets were solubilized in UTC buffer Protein concentration was determined using Bradford assay [9].

Each sample (40 μg of previously extracted proteins labeled with CF647DI Hydrazide) was labeled with 0.8 μL of 400 pM Cy3 NHS ester fluorescent dye (Lumiprobe, Hannover, Germany). The internal standard containing equal amounts of every analyzed sample (8 times 40 μg of protein extract) was labeled with 10.4 μL of 400 pM Cy2 NHS ester fluorescent dye (Lumiprobe, Hannover, Germany).

ImmobilineTM DryStrips (24 cm, GE Healthcare, Buckinghamshire, UK) were rehydrated for 7 h at room temperature using rehydration buffer containing 8 M urea, 2 M thiourea, 4% CHAPS, 0.0068% DeStreak reagent, and 0.0024% IPG Buffer pH 3-11 NL. Samples were loaded on strips pH 3-11 NL by the sample Cup-loading method. An Ettan IPGphor3 (GE Healthcare, Buckinghamshire, UK) was used to perform isoelectric focusing for a total of 44 350VH. An equilibration step was carried out in buffers containing 6 M urea, 2% SDS, 30% glycerol, and 50 mM Tris-HCl, pH 8.6, together with 1% m/w DTT or 4.7% m/w iodoacetamide, respectively, for reduction and alkylation of disulfide bridges. Strips were then rinsed with SDS electrophoresis buffer (containing 25 mM Tris, 192 mM glycine and 0.2% SDS, pH 8.6) and set onto gradient SDS-PAGE gels 8-18% acrylamide-bis acrylamide. Electrophoresis was performed with two Ettan Dalt 6 Electrophoresis Unit (GE Healthcare, Buckinghamshire, UK) at 15 °C, protected from light with 2.5 W/gel overnight. Gels were scanned by a Typhoon FLA 9500 scanner (GE Healthcare, Buckinghamshire,UK), with excitation/emission wavelengths corresponding to each fluorescent dye (473/510 nm for Cy2 NHS, 532/575 nm for Cy3 NHS and 635/665 nm for CF647DI hydrazide).

Data and statistical analysis was carried out with the SameSpots software (Totallab, Newcastle upon Tyne, UK). Spots of interest were selected according to two criteria: (i) if significant difference in expression and/or carbonylation of protein spots between the cell lines was confirmed (p < 0.05, T-test), and (ii) if the fold change varied for at least 50%. Protein carbonylation signal was normalized to the protein expression signal within each spot (CF647DI spot volume/ Cy3 NHS spot volume) and the relative modification index (RMI) was expressed as ratio of the CF647DI signal corrected by the normalized Cy3 NHS expression signal between SW620 and SW480 cell lines. A p value for each carbonylation signal comparison was evaluated (T-test). Hierarchical clustering and PCA was performed using DANTE R package for R (http://omics.pnl.gov/software/danter).

The protein spots showing significant difference in expression and/or oxidation levels between SW480 and SW620 cell lines were hand-picked by using OneTouch 2D gel spot-picker, 1.5 mm in diameter (Gel Company, San Francisco, CA, USA) and the Screen Picker (Proteomic Consult, Kampenhout, Belgium).

The spots of interest were sent for analysis to 3P5 Proteomic Facility, Université Paris Descartes, Sorbonne Paris Cité. In-gel digestion was carried out with trypsin, according to the published procedure with minor adjustments [10]. Mass spectrometry (MS) and MS/MS Orbitrap analyses were performed using an Ultimate 3000 Rapid Separation Liquid Chromatographic (RSLC) system (Thermo Fisher Scientific, Rockford, IL, USA) online with a hybrid LTQ-Orbitrap-Velosmass spectrometer (Thermo Fisher Scientific, Rockford, IL, USA). Briefly, peptides were dissolved in 4 μL of 10% acetonitrile (ACN)-0.1% trifluoroacetic acid (TFA). The peptides were then loaded and washed on a C18 reverse phase precolumn (3 μm particle size, 100 Å pore size, 150 μm i.d., 0.5 cm length). The loading buffer contained 98% H2O, 2% ACN and 0.1% TFA. Peptides were then separated on a C18 reverse phase resin (2 μm particle size, 100 Å pore size, 75 μm i.d.,15 cm length) with a 16 min gradientfrom 100% A (0.1% formic acid (FA) and 100% H2O) to 50% B (80% ACN, 0.085% FA and 20% H2O). The Linear Trap Quadrupole Orbitrap mass spectrometer acquired data throughout the elution process and operated in a data dependent scheme with full MS scans acquired with the Orbitrap, followed by up to 20 LTQ MS/MS CID spectra on the most abundant ions detected in the MS scan. Mass spectrometer settings were: full MS (automatic gain control (AGC): 1×106, resolution: 6×104, m/z range 400-2000, maximum ion injection time: 500 ms) and MS/MS (AGC: 5×103, maximum injection time: 20 ms, minimum signal threshold: 500, isolation width: 2 Da, dynamic exclusion time setting: 30 s). The fragmentation was permitted for precursors with a charge state of 2,3,4 and above. For the spectral processing, the software Proteome Discoverer 1.4.0.288 was used. The threshold of signal-to-noise for extraction values was 3. Database searches were carried out using Mascot version 2.4 (MatrixScience,London, UK) on “homo sapiens” proteins (20273 sequences) from the SwissProt databank [11] containing 547 599 sequences (195 014 757 residues) (February 2015). The search parameters were as follows: carbamidomethylationas a variable modification for cysteins, and oxidation as a variable modificationfor methionines. Up to 1 missed tryptic cleavage was tolerated and mass accuracy tolerance levels of 5 ppm for precursors and 0.56 Dafor fragments were used for all tryptic mass searches. Positive identification was based on a Mascot score above the significance level (i.e. 5%). The reported proteins were always those with the highest number of peptide matches.

Gene ontology analysis was performed using the Panther database (http://www.pantherdb.org/) [12].

UVC irradiation kills both, SW480 cells derived from a primary human adenocarcinoma and SW620 derived from its metastasis. However, SW480 cell line displays greater sensitivity than SW620: 50% survival of SW480 cells is observed at 1300 Jm-2, while that of SW620 is at 2100 Jm-2. There is a significant difference in survival between SW480 and SW620 cell lines at each UVC dose (P≤0.0001 for all dose rates, except for the highest 3000 Jm-2 dose, P≤0.05). Decreased survival is correlated with an increase in protein carbonylation (PC) measured immediately following UVC irradiation. Consistent with the survival decline, SW480 cells show an increase in PC at lower doses, relative to SW620 (Figure 1A).

After the recovery period of 18 h we observed a decrease in PC at 100 and 1000 Jm-2 back to the PC level before irradiation, in both cell lines. Although the level of residual PC remains significantly elevated at higher doses (at 2000 Jm-2 P≤0.0001 for SW480 and P≤0.001 for SW620, at 3000 Jm-2 P≤0.0001 for both), probably due to the prevalent mortality in the population, one can see that PC is higher in SW480 cells at 3000 Jm-2 compared to SW620 (P≤0.01) (Figure 1B). Since the PC level 18 h after irradiation does not correlate with mortality of SW480 and SW620 cells, we can conclude that the PC detected immediately after irradiation is the one related to cell survival presumably by being determinant of the capacity for cell recovery from UVC damage.

  SW480 cells display higher ROS level

To explore the cause of differential oxidative proteome damage in SW480 and SW620 cell lines, both spontaneous and UVC-induced, we refer to more advanced studies with bacterial cells where it was shown that PC is affected by the (i) level of ROS, (ii) efficacy of elimination of oxidized proteins by selective proteolysis and (iii) characteristic intrinsic susceptibility of each protein molecule to ROS effects [13]. The latter is based on the fact that proteins possess native structures assuring durability of their function by an evolved resistance to oxidation [4]. Therefore, mistakes in biosynthesis or folding and polymorphic mutations can break the resistance to oxidation [14,15]. To explore origin of the observed differential sensitivity to proteome oxidative damage of the two cell lines, we measured ROS levels before UVC exposure. Indeed, the results show 49% higher basal ROS level in SW480 cells than in SW620 (P<0.0001) (Figure 2A).

 In order to examine the effect of UVC irradiation on proteome of SW480 and SW620 cell lines in more detail, we measured ROS levels immediately after UVC exposure. The H2DCFDA probe was present during irradiation in order to scavenge ROS immediately upon their formation. A gradual increase in ROS level is detected with the increase in UVC dose (Figure 2B). UVC-induced ROS levels were higher for all doses in SW480 cells. More precisely, the results show 50% higher ROS level in SW480 cells than in SW620 at 100 Jm-2 (P≤0.05), 85% at 1000 Jm-2 (P≤0.05), 67% at 2000 Jm-2 (P≤0.05) and 96% at 3000 Jm-2 (P≤0.05), respectively.

Identification of proteins susceptible to PC

To further study the relationship between PC and mortality in these cell lines, we undertook a detailed proteomic analysis of differences in expression levels as well as carbonylation at the level of individual protein spots (Figures 3A and 3B). We compared the two cell lines at 90% UVC-induced mortality, to identify proteins that are most susceptible to PC and could potentially serve as targets in anti-cancer treatments. We aimed also to find the cause of increased resistance of metastatic cells in the differential protein expression patterns. All gels were aligned using Internal Standard Image and alignment was applied to the rest of the images for spot identification. A total of 953 spots were detected. We detected the total of 35 proteins significantly up-regulated upon UVC irradiation, 13 in the SW480 and 22 in the SW620 metastatic cell line (Figures 3C and 3D). Surprisingly, among these 22 up-regulated proteins, 19 were more carbonylated in SW620 and only 3 in SW480 cell line (Figures 4A and 4B, Tables 1A and 1B).

    Using PANTHER database, we grouped significantly upregulated and carbonylated protein spots according to their gene ontology (GO) classification: protein class (Supplementary Figures S 1A-1D), molecular function (Supplementary Figures S 2A-2D), biological process (Supplementary Figures S 3A-3D), cell component (Supplementary Figures S 4A-4C) and pathway (Supplementary Figures S 5A-5C).

Although we observed significantly up-regulated proteins associated with molecular function in both cell lines, we detected a larger number of up-regulated proteins in SW620 cells relative to SW480. In particular, we identified 7 up-regulated proteins with catalytic activity (GO:0003824) in SW480 (Supplementary Table S 1A) and 13 in SW620 cell line (Supplementary Table S 1B). Also, two proteins were significantly carbonylated in SW480 cell line, proteinglutamine gamma-glutamyltransferase 2 (TGM2) and pre-mRNAprocessing factor 19 (PRP19). TGM2 has a role in progression of several types of cancers [16] and PRP19 is known for its multiple roles in pre-splicing, ubiquitination and apoptosis [17]. Moreover, we found 7 up-regulated proteins involved in binding (GO:0005488) in SW620 (Supplementary Table S 1B) compared to 3 in SW480 (Supplementary Table S 1A).

According to the biological process, we found 12 up-regulated proteins connected with metabolic activities (GO:0008152) in SW480 (Supplementary Table S 2A) and 17 in SW620 cells (Supplementary Table S 2B). In addition to TGM2 and PRP19, we detected increased protein carbonylation in one of the chaperonin proteins essential for protein folding: the T-complex protein 1 subunit eta (TRIC) in SW480 [18].

Four proteins associated with metabolic function in SW620 cell line (Supplementary Table S 2B) were up-regulated without changes in carbonylation: obg-like ATPase 1 (OLA1), spliceosome RNA helicase DDX39B (DX39B), phosphoglycerate kinase 1 (PGK1), eukaryotic translation initiation factor 5A-1 (IF5A1). These proteins are reported to be up-regulated in tumor cells [19-22].

We found two proteins associated with cellular pathways to be up-regulated in SW480 cells: PRP19, as a part of mRNA splicing pathway (P00058) and mitochondrial ATP synthase subunit beta (ATPB) involved in an ATP synthesis (P02721) (Supplementary Table S3A). In the metastatic cell line, we identified in total 6 up regulated proteins involved in 7 different pathways (Supplementary Table S3B). Alpha enolase (ENOA), fructose-biphosphate aldolase A (ALDOA) and phosphoglicerate-kinase 1 (PGK1) are up-regulated in glycolysis pathway (P00024). The latter protein is not oxidatively modified and previous studies show that increased expression of PGK1 in colon cancer is associated with metastasis [22] and could promote radioresistance [23].

SW480 cells have increased expression of HSP105 chaperone

Furthermore, we sought to measure the levels of several main groups of protein chaperones, often indicative of the cellular proteome quality and sensitivity to oxidation [4,6]. In addition, chaperones have multiple roles in the resistance of cancer cells to various anticancer treatments [24-26].

In this context, we determined basal expression levels of HSP70, HSP90 and HSP105 in both cell lines. Interestingly, we found an increased expression of HSP105 in SW480 cells. On the other hand, we did not detect a significant difference of HSP70 and HSP90 expression, as well as the levels of ROS scavenging enzymes MnSOD and Cu/ZnSOD ((Figures 5A and 5B)).