The TrkAIII Oncoprotein Inhibits Mitochondrial Free Radical ROS-Induced Death of SH-SY5Y Neuroblastoma Cells by Augmenting SOD2 Expression and Activity at the Mitochondria, within the Context of a Tumour Stem Cell-like Phenotype

The developmental and stress-regulated alternative TrkAIII splice variant of the NGF receptor TrkA is expressed by advanced stage human neuroblastomas (NBs), correlates with worse outcome in high TrkA expressing unfavourable tumours and exhibits oncogenic activity in NB models. In the present study, we report that constitutive TrkAIII expression in human SH-SY5Y NB cells inhibits Rotenone, Paraquat and LY83583-induced mitochondrial free radical reactive oxygen species (ROS)-mediated death by stimulating SOD2 expression, increasing mitochondrial SOD2 activity and attenuating mitochondrial free radical ROS production, in association with increased mitochondrial capacity to produce H2O2, within the context of a more tumour stem cell-like phenotype. This effect can be reversed by the specific TrkA tyrosine kinase inhibitor GW441756, by the multi-kinase TrkA inhibitors K252a, CEP-701 and Gö6976, which inhibit SOD2 expression, and by siRNA knockdown of SOD2 expression, which restores the sensitivity of TrkAIII expressing SH-SY5Y cells to Rotenone, Paraquat and LY83583-induced mitochondrial free radical ROS production and ROS-mediated death. The data implicate the novel TrkAIII/SOD2 axis in promoting NB resistance to mitochondrial free radical-mediated death and staminality, and suggest that the combined use of TrkAIII and/or SOD2 inhibitors together with agents that induce mitochondrial free radical ROS-mediated death could provide a therapeutic advantage that may also target the stem cell niche in high TrkA expressing unfavourable NB.


Introduction
The alternative TrkAIII splice variant (UniProtKB/Swiss-Prot: P04629-4) of the NGF receptor TrkA (NCBI: NM_0010122331.1; GenBank: AB019488.2; UniProtKB/Swiss-Prot: P04629) is expressed by advanced stage human neuroblastoma (NB), is associated with poor outcome in high TrkA expressing unfavourable tumours and exhibits oncogenic activity in NB models [1][2][3][4][5][6][7]. Alternative TrkAIII splicing is stress-regulated, providing a mechanism through which tumour suppressing signals from fully spliced TrkA receptors can be converted to oncogenic signals from the alternative spliced TrkAIII variant within the tumour microenvironment. We consider this to potentially represent the conservation and pathological subversion of a physiological developmental and stress-regulated, neural stem/progenitor cell stress-protection mechanism [1,8].
Alternative TrkAIII splicing is characterised by exon 6,7 and 9 skipping and produces a TrkAIII protein that is devoid of the extracellular D4 Ig-like domain and related N-glycosylation sites required for cell surface receptor expression and prevention of ligand-independent activation [9,10]. Unlike cell surface TrkAI (exon 9 excluded) and TrkAII (exon 9 included) splice variants [11], TrkAIII is not expressed at the cell surface but is retained within the intracellular membrane compartment, within which it exhibits spontaneous, ligand-independent activation [1][2][3]. This, results in chronic signal transduction through the IP3k/Akt/NF-kB but not Ras/MAPK pathway, which differs to activated cell surface TrkA receptors that signal also through Ras/MAPK [1,[12][13][14][15]. In contrast to TrkA activated at the NB cell surface, intracellular TrkAIII activity in NB cells does not inhibit proliferation nor induce neuronal differentiation but promotes an undifferentiated stem cell-like phenotype that exhibits increased tumourigenic and metastatic behaviour [1,4]. TrkAIII exerts its ''oncogenic'' activity in NB cells by: protective IP3K/Akt/NF-kB signalling; induction of a pro-angiogenic pattern of gene expression; interacting with the centrosome, promoting centrosome amplification, peri-nuclear microtubule assembly and genetic instability; increasing the level of sister chromatid exchange; and modulating the unfolded protein response, pre-conditioning and adapting cells to stress [1][2][3][4][5].
Mitochondrial reactive oxygen species (ROS) also regulate stress adaptation, cellular differentiation, and chronological lifespan and play important roles in tumour pathogenesis and metastatic progression [16][17][18]. The superoxide free radical is produced during oxidative phosphorylation by single electron reduction of O 2 , leaks from respiratory chain complexes I and III and is detoxified to the non-free radical ROS H 2 O 2 by mitochondrial superoxide dismutases (SODs), optimising physiological function [16][17][18]. Free-radical ROS do not penetrate cellular membranes but react locally and are detoxified by appropriately localised SODs. In contrast to superoxide, the non-free radical ROS H 2 O 2 penetrates cellular membranes, acts as an extra-mitochondrial effector and is detoxified by appropriately localised catalase, glutathione peroxidase and peroxiredoxin antioxidants [17,18]. If not tightly regulated, both free radical and non-free radical ROS cause oxidative damage to mitochondrial proteins, lipids and DNA, with fatal consequences [16,[19][20][21]. The unbridled accumulation of mitochondrial ROS represents a major mecha- pcDNA, TrkA and TrkAIII SH-SY5Y cell lines. B) Histogram depicting differences in SOD2 densitometric ratio to a-tubulin between pcDNA, TrkA and TrkAIII SH-SY5Y cells. Results are displayed as the mean6 s.d. fold difference in SOD2 levels, as a densitometric ratio to a-tubulin, with respect to pcDNA controls adjusted to the arbitrary value of 16s.d., in three independent experiments. C) Representative ethidium bromide stained agarose gels demonstrating the relative levels of total (4.2 kb plus 1.5 kb) SOD2 and GAPDH RT-PCR products from selected pcDNA, TrkA and TrkAIII SH-SY5Y cell lines. D) Histogram depicting differences in SOD2 levels, as a densitometric ratio to GAPDH, in pcDNA, TrkA and TrkAIII SH-SY5Y cells. Results are displayed as the mean6s.d fold difference in SOD2 levels, as a densitometric ratio to GAPDH, with respect to pcDNA controls adjusted to the arbitrary value of 16s.d., in 3 independent experiments performed in duplicate. E) Representative ethidium bromide stained agarose gels demonstrating relative differences in 4.2 kb SOD2, 1.5 kb SOD2, catalase, glutathione peroxidase and GAPDH RT-PCR products in selected pcDNA, TrkA and TrkAIII SH-SY5Y cell lines. F) Histogram depicting differences in 4.2 kb SOD2, 1.5 kb SOD2, SOD1, catalase and glutathione peroxidase levels, as a densitometric ratio to GAPDH, in pcDNA, TrkA and TrkAIII SH-SY5Y cell lines. Results are displayed as the mean6s.d fold difference in densitometric ratio with resect to pcDNA SH-SY5Y controls, adjusted to the arbitrary value of 16s.d. for each ratio, in 3 independent experiments performed in duplicate. In all histograms mean values are provided above each column and asterisks denote statistical significance by t-test comparison of TrkAIII transfectants to both pcDNA and TrkA transfectants). doi:10.1371/journal.pone.0094568.g001 nism of action for many chemotherapeutic agents, cytotoxic compounds and ionising radiation [20,22], and mechanisms that attenuate the production of mitochondrial ROS promote therapeutic resistance in cancer [23][24][25][26][27][28][29][30][31].
Considering the roles of SOD2 in cellular differentiation, survival, the nervous system, and in tumour pathogenesis and progression, we have assessed the influence of the TrkAIII oncogene upon SOD2 expression, activity and function in human SH-SY5Y NB cells.

Cell Extraction and Western Blotting
Cells were extracted in lysis buffer (PBS containing 0.5% sodium deoxycholate, 1% NP40, 0.1% SDS, 1 mM sodium orthovanadate, 1 mM PMSF, 1 mg/ml of pepstatin A and Aprotinin) and protein concentrations were calculated by Bradford protein concentration assay (Sigma-Aldrich). Samples were mixed with reducing SDS-PAGE sample buffer and subjected to reducing SDS-PAGE/Western blotting. Briefly, proteins separated by reducing SDS-PAGE, were trans-blotted onto Hybond C+ nitrocellulose membranes by electrophoresis (Amersham Int. UK) and the membranes subsequently air-dried. Non-specific protein binding-site on membranes were blocked by incubation for 2 hours in 5% non-fat milk in TBS-T prior to incubation with primary antibodies, at recommended dilutions, for 2-16 hours at 4uC. Membranes were then washed in TBS-T, incubated with secondary HRP-conjugated antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA) diluted in blocking solution and immunoreactive species detected by chemiluminescence reaction, as directed by the manufacturer (Amersham Int).

RNA Purification and RT-PCR
RT reactions were performed on total RNAs (1 mg), purified using RNA-easy Plus, as described by the manufacturer (Qiagen), using the Moloney Murine Leukemia virus RT kit, as detailed by the manufacturer (LifeTechnologies, Inc, Paisley, UK). RT reactions were subjected to PCR using the following primers: For each primer set, PCRs were performed on reverse transcription reactions serially diluted from 1 to 1:1000. Reactions below saturation were compared by densitometric analysis of Jpeg images of ethidium bromide stained gels, using ImageJ64 software [60].

Purification of Mitochondria
Mitochondria were purified using a Focus Sub cell Mitochondrial isolation kit, as described by the manufacturer (G-Biosciences, St. Louis, MO). Briefly, cells were harvested in ice cold PBS by scraping, centrifuged at 12006g for 5 minutes, at 4uC, 500 ml of ice cold Buffer I, containing 1x protease inhibitor cocktail (Sigma), was added to the pellet and cells were disrupted by 10 passages through a 20 gauge needle. 250 ml of Buffer II was then added to the homogenate and samples were then centrifuged at 1,2006g for 5 minutes, at 4uC. Supernatants were transferred to fresh tubes, centrifuged at 15,0006g for 10 minutes, at 4uC and the resulting mitochondrial-rich pellet was washed with 500 ml of Buffer II, centrifuged at 15,0006g for 10 minutes at 4uC and re-suspended in mitochondrial storage buffer (250 mM mannitol, 5-mM HEPES (pH 7.4). Crude mitochondrial preparations in 250 mM mannitol, 5 mM HEPES (pH 7.4) and 0.5 mM EGTA were then separated by Percoll density gradient, as previously described [61]. Briefly, crude preparations were layered onto a 30% Percoll gradient, in the same buffer, and ultra centrifuged at 90,0006g for 40 minutes, at 4uC. Ultrapure mitochondria were collected from a band located 2/3rds from the top, transferred to a fresh tube, diluted 1:10 in 250 mM mannitol, 5 mM HEPES (pH 7.4) containing 0.5 mM EGTA and re-centrifuged at 15,0006g for 10 minutes, at 4uC. Ultrapure mitochondrial pellets were resuspended in the desired volume of cold 250 mM mannitol, 5-mM HEPES (pH 7.4) containing 0.5 mM EGTA immediately prior to experimentation.

Mitochondrial H 2 0 2 Assay
Mitochondrial hydrogen peroxidase (H 2 O 2 ) production was measured in a fluorescence-based assay using Amplex Red (Nacetyl-3, 7-dihydroxyphenoxazine) reagent, as previously described [62]. Briefly, purified mitochondria (60 mg protein) were incubated for 10 minutes, at 37uC, in 100 ml of PBS containing 50 mU/ml of horseradish peroxidase and 40 mM Amplex Red. Fluorescence levels were monitored up to the point of stabilization (approximately 10 minutes) in a Perkin Elmer Fluoro-Count Multiwell plate reader (excitation wavelength 530 nm, emission wavelength 590 nm). Upon signal stabilization, succinate in PBS was added to a final concentration of 10 mM, reactions incubated at 37uC and fluorescence monitored at 20-minute intervals.

Cell Death Assay
Cell death assays were performed using a modification of previously described methods [63,64]. Briefly, cells were incubated with appropriate reagents at concentrations and for the times indicated in the results section and figure legends, washed once in Ca 2+ free PBS, detached with ice cold PBS containing 1 mM

Mitosox-Red Fluorescence Detection of Free Radical ROS
Cells grown on glass chamber slide (Nunc) were treated with appropriate reagents at concentrations and for the times indicated in the results section and figure legends. Following incubation, Mitosox red solution (Invitrogen) was added to a final concentration of 2.5 mM and incubated for 10 minutes, at 37uC. Cells were then washed three times in pre-warmed PBS, mounted with a coverslip and examined immediately under a Zeiss ''Axioplan-2'' fluorescence microscope. Representative fields were digitally photographed under identical exposure conditions and fluorescence levels quantified from jpeg images, using ImageJ64 software [60].

SOD Zymography
Native SOD zymography was performed as previously described [65]. Briefly, cells harvested in cold Ca 2+ Mg 2+ -free PBS were pelleted by centrifugation at 2006g for 5 minutes, at 4uC, lysed and mitochondria ultra-purified, as described in the materials and methods section. Purified mitochondria were sonicated at 40% maximal power in a vibracell sonicator, protein concentrations determined by Bradford protein concentration assay (Sigma-Aldrich), and 150 mg of mitochondrial proteins were mixed with an equal volume of native PAGE loading buffer (1.5 M Tris Stacking gel buffer, pH 6.8, 5% bromophenol blue and 50% glycerol). Pre-prepared native 12% PAGE gels (pH 8.8), with 5% PAGE stacking gels (pH 6.8), containing sucrose and 0.004% Riboflavin-5-phopsphate to replace APS and Temed, were pre-run at 40 mA for 1 hour at 4uC to remove all traces of Temed and APS, that interfere with SOD activity, and were stored overnight at 4uC. Samples were applied to pre-prepared gels and subjected to an initial pre-electrophoresis run for three hour at 40 mA, at 8uC, to remove all remaining traces of APS and TEMED. The electrophoresis buffer was replaced and gels run for a further 2-3 hours at 40 mA at 4uC, until the bromophenol blue had reached the bottom of the gel, and then for a further 2 hours. Following electrophoresis, gels were incubated in freshly prepared SOD staining buffer (2.43 mM Nitro Blue Tetrazolium, 28 mM Temed and 0.14 mM Riboflavin in 50 mM phosphate buffer, pH 7.8) for 20 minutes in the dark, at room temperature, with shaking. Following incubation, gels were washed twice in deionized H 2 0 and exposed to UV light for 15-120 minutes, at room temperature, taking care not to let the gels dry out. Upon appearance of clear achromatic bands (SOD activity) within a dark background, gels were washed a further three times in deionized H 2 0 and left for 18-24 hours under ambient light, at room temperature, for further band development. Two achromatic bands were detected, SOD2 representing the higher and SOD1 the lower band.

Tumour Spheroid Growth
Stable transfected SH-SY5Y cell lines were grown as tumour spheroids, as previously reported for tumour and neural stem cells spheroid growth [66]. Briefly, cells were plated in upright T25 culture flasks at a density of 1610 5 cells/ml in 3 ml of serum-free neuro-sphere culture medium 1:1 DMEM/Ham's F12 (Sigma-Aldrich), containing 0.6% Glucose, 1x N-2 and 1x B-27 nutrient supplements (Life technologies); 20 ng/ml Epidermal Growth Factor (Sigma-Aldrich), 40 ng/ml Fibroblast Growth Factor (Sigma-Aldrich), 1% glutamine and Penicillin-Streptomycin (Sigma-Aldrich). Growth factors were added every 2 days; medium was changed every 4 days and spheroid growth monitored for a period of 20 days. In SOD2 knockdown tumour spheroid assays, TrkAIII SH-SY5Y cells were transiently transfected with 25 nM control siRNA or SOD2 specific siRNA for 48 hours, as described above. At 48 hours, siRNA transfected cells were plated in neuro sphere growth medium and spheroid growth monitored for 7 and 10 days.
For the evaluation of irregular sphere volumes, digital phase contrast micrographs were made of 20 day-old tumour spheres and the three greatest spheroid dimensions (referred to as a, b and c) were used to calculate estimates of volume, using the equations for the volume of a sphere ([4 p/3]r3), where the radius r is the average of a, b and c, and the volume of an ellipsoid V = (abcp)/6. Spherical volume calculations were based on the average radius across all three axes.

Statistical Analysis
Data were analysed statistically by Student's t-test and One-Way ANOVA. Statistical significance was associated with probabilities of #0.05. provided above each histogram column and asterisks denote statistical significance by t-test comparison of TrkAIII SH-SY5Y cells to either pcDNA or TrkA SH-SY5Y cells. B) Digital phase-contrast micrographs demonstrating differences in the relative tumour spheroid sizes in TrkA and TrkAIII SH-SY5Y cells cultured for 20 days in the absence (Control) or presence of 100 nM or 1 mM GW441756 (bar = 100 mm), plus a histogram showing the difference in TrkA and TrkAIII SH-SY5Y tumour spheroid volumes, cultured as described above, displayed as the mean+s.d. volume (mm 3 ) of 50 individual tumour spheres per cell line, in three independent experiments. Mean values are provided above each histogram column and asterisks denote statistical significance by t-test comparison of GW441756-treated and untreated TrkAIII SH-SY5Y cells. C) Ethidium bromide stained agarose gels demonstrating the relative levels of 4.2 kb SOD2, 1.5 kb SOD2 and GAPDH RT-PCR products in 20-day pcDNA, TrkA and TrkAIII SH-SY5Y spheroids, plus a histogram depicting the relative densitometric differences in 4.2 kb SOD2 and 1.5 kb SOD2 mRNA expression. Results are displayed as mean+ s.d. densitometric ratio to GAPDH, in three independent experiments. Mean values are provided above each histogram column and asterisks denote statistical significance by t-test comparison of TrkAIII to both pcDNA and TrkA SH-SY5Y cells. D) Ethidium bromide stained agarose gels demonstrating differences in CD133, CD117, SOX-2, Nestin, Nanog and GAPDH RT-PCR products in 20-day pcDNA, TrkA and TrkAIII SH-SY5Y spheroids, plus histograms depicting fold change in (I), CD117, (II) SOX-2, (III) Nestin, (IV) Nanog and (V) CD133 expression in 20-day tumour spheroids. Results are displayed as mean6s.d. fold change compared to pcDNA SH-SY5Y cells, given the arbitrary value of 1 as a densitometric ratio to GAPDH expression, in three independent experiments. Mean values are provided above each histogram column and asterisks denote statistical significance by t-test comparison of TrkAIII to pcDNA SH-SY5Y cells. doi:10.1371/journal.pone.0094568.g008

TrkAIII Stimulates SOD2 Expression and Mitochondrial SOD2 Activity in SH-SY5Y Cells
The effect of constitutive TrkAIII expression upon SOD2 expression and activity was compared to that of TrkA in human SH-SY5Y NB cells by Western blotting, RT-PCR and zymogram.
Western blot comparisons of total cell extract proteins (20 mg) from duplicate pcDNA control, TrkA and TrkAIII stable transfected SH-SY5Y NB cell lines revealed that TrkAIII transfectants expressed higher levels of SOD2, relative to atubulin, than both pcDNA control and TrkA SH-SY5Y transfectants ( Fig. 1A and B). Densitometric quantification of experiments, indicated that TrkAIII transfectants exhibited a significant 3.160.6 fold increase (t-test, p,0.0001, degrees of freedom (df) = 10) in the SOD2 densitometric ratio to a-tubulin, when compared to pcDNA controls and a significant 2.1760.44 fold increase (t-test, p = 0.024, df = 10) in the SOD2 densitometric ratio to a-tubulin, when compared to TrkA transfectants (data not displayed). TrkA transfectants did not exhibit a significant difference in the SOD2 densitometric ratio to a-tubulin, when compared to pcDNA controls ( Fig. 1A and B). The levels of catalase in total cell extracts did not significantly differ between pcDNA, TrkA and TrkAIII SH-SY5Y transfectants ( Fig. 1A and  B).
In comparative semi-quantitative RT-PCR experiments, TrkAIII SH-SY5Y cells (selected clone A) exhibited a significant 2.160.56 fold (t-test, p = 0.02, df = 10) increase in the total (4.2 kb plus 1.5 kb) SOD2 mRNA levels, as a densitometric ratio to GAPDH, compared to pcDNA SH-SY5Y transfectants. The total SOD2 mRNA densitometric ratio to GAPDH in TrkA transfectants did not differ significantly to pcDNA controls ( Fig. 1C and  D).
In comparative semi-quantitative RT-PCR experiments, using primers specific for 4.2 kb and 1.5 kb SOD2 mRNAs, TrkAIII SH-SY5Y cells exhibited a significant 2.260.4 fold (t-test, p = 0.023, df = 10) increase in 4.2 kb SOD2 but not in 1.5 kb SOD2 levels, as a densitometric ratio to GAPDH (t-test, p = 0.12, df = 10), when compared to pcDNA SH-SY5Y controls ( Fig. 1E and F). TrkA SH-SY5Y cells did not exhibit significant differences in either 4.2 kb or 1.5 kb SOD2 densitometric ratio to GAPDH, when compared to pcDNA controls ( Fig. 1E and F). No significant differences were detected in SOD1, glutathione peroxidase and catalase levels, as a densitometric ratio to GAPDH, between pcDNA, TrkA and TrkAIII SH-SY5Y cells (Fig. 1E and F).
Densitometric analysis of Western blots, containing equal protein concentrations (20 mg) of ultra purified mitochondrial extracts from pcDNA, TrkA and TrkAIII SH-SY5Y cells, revealed a significant 4.860.96 fold increase (t-test, p = 0.004, df = 4) in SOD2 levels, as a densitometric ratio to Hsp60 used a mitochondrial protein loading control, in mitochondria from TrkAIII SH-SY5Y compared to pcDNA SH-SY5Y cells ( Fig. 2A  and B). SOD2 levels in mitochondria from TrkA SH-SY5Y cells were not significantly elevated over those detected in mitochondria from pcDNA SH-SY5Y cells (1.6560.44 fold, t-test, p = 0.2, df = 4) ( Fig. 2A and B). TrkAIII stimulation of mitochondrial SOD2 antioxidant expression was compared to that of the mitochondrial loading control Hsp60 and also the mitochondrial antioxidant thioredoxin-2 (Trx-2) [67]. Neither Hsp60 nor Trx-2 expression differed significantly in purified mitochondrial extracts from pcDNA, TrkA or TrkAIII SH-SY5Y cells ( Fig. 2A and B).
Form these data, we conclude that TrkAIII expression and spontaneous activation in SH-SY5Y cells promotes SOD2 mRNA and protein expression, augments mitochondrial SOD2 but not Trx2 levels and increases mitochondrial SOD2 activity.

TrkA, IP3K and NF-kB Inhibitors Reduce SOD2 Expression in TrkAIII Transfectants
In order to further characterise signalling pathways involved in TrkAIII promotion of SOD2 expression, experiments were performed under TrkA, IP3K, MAPK and NF-kB inhibitory conditions. Wild type (wt) TrkAIII was also compared to tyrosine kinase dead (kd) Y670/674/675F-mutated TrkAIII.
Densitometric comparison of Western blots, containing equal protein concentrations (20 mg) of total cell extracts from untreated (Con) and NGF (100 ng/ml)-treated TrkA SH-SY5Y cells, revealed a significant increase in SOD2 levels, as a densitometric ratio to a-tubulin, at 30 minutes (1.860.23 fold increase, t-test, siRNA-treated controls (lower histogram), in three independent experiments. Mean values are presented above each column and asterisks denote significant differences by Student's t-test. doi:10.1371/journal.pone.0094568.g009  (Fig. 3F).
From these data, we conclude that TrkAIII tyrosine kinase activity is required to stimulate SOD2 expression in SH-SY5Y cells, that this involves IP3K and NF-kB but not MAPK.  Since mitochondrial SOD2 activity regulates free radical ROS accumulation, mitochondrial free radical ROS production in response to the free radical ROS inducers Rotenone [56], Paraquat [57] and LY83583 [58] was compared in pcDNA control, TrkA and TrkAIII SH-SY5Y cells, in a Mitosox-based assay.
In inhibitor studies, densitometric analysis of free radical ROS levels adjusted to densitometric units per 100 cells, revealed no significant increase in TrkAIII SH-SY5Y cells following 27 or 48 hours treatment with either 1 mM CEP-701, 1 mM K252a, 1 mM Gö6976 or 100 nM GW441756, alone ( Fig. 5B and C, results displayed for 48 hours treatment only).
As described above, treatment of TrkAIII SH-SY5Y cells with the free radical ROS inducers Rotenone, Paraquat and LY83583 did not significantly increase the ROS levels compared to mocktreated controls. However, TrkAIII SH-SY5Y cells produced significant levels of ROS in response to these ROS inducers when cells were pre-treated with the TrkA tyrosine kinase inhibitors CEP-701, K252a, Gö6976 or GW441756. These tyrosine kinase inhibitors by themselves did not affect ROS levels in TrkAIII SH-SY5Y cells (Fig. 5C and 5D). From these data, we conclude that TrkAIII tyrosine kinase activity is necessary to inhibit mitochondrial ROS production induced by Rotenone, Paraquat or LY83583.

SOD2 siRNA Restores the Sensitivity of TrkAIII SH-SY5Y Cells to Rotenone, Paraquat and LY83583-induced Mitosox-reactive Mitochondrial Free-radical ROS Production
In order to confirm SOD2 involvement in the reduced sensitivity of TrkAIII SH-SY5Y cells to agents that induce mitochondrial free radical ROS, the induction of mitochondrial ROS was assessed in TrkAIII SH-SY5Y cells exhibiting siRNA knockdown of SOD2 expression.
In Western blots, whole cell extracts from TrkAIII SH-SY5Y cells treated for 48 hours SOD2-specific but not negative control siRNA duplexes exhibited a significant reduction in SOD2 compared to a-tubulin protein levels, used as a loading control (Fig. 6A).
SiRNA knockdown of SOD2 expression in TrkAIII SH-SY5Y cells, but not transfection with control siRNA, resulted in the production of significant levels of ROS in response to Rotenone, Paraquat and LY83538 (Figs. 6B and 6C). From these results we conclude that elevated SOD2 expression in TrkAIII SH-SY5Y cells is largely responsible for attenuating Rotenone, Paraquat, and LY83583-induced mitochondrial ROS production.
From these data, we conclude that TrkAIII SH-SY5Y cells produce higher levels of H 2 0 2 compared to controls in response to succinate, presumably due to elevated SOD2 activity.
From these data, we conclude that both TrkAIII tyrosine kinase activity and SOD2 promote the formation of larger SH-SY5Y cell tumour spheroids/spheroid aggregates, in association with increased SOD-2, CD117, Nestin, SOX-2 and Nanog mRNA expression, and that TrkAIII tyrosine kinase activity but not SOD2 promotes Nanog, Nestin and SOX2 neural stem cell marker expression in TrkAIII SH-SY5Y tumour spheroids.
As described above, treatment of TrkAIII SH-SY5Y cells with the free radical ROS inducers Rotenone and Paraquat alone did not induce significant TrkAIII SH-SY5Y cell death and LY83583 induced only a subtle but significant increase in TrkAIII SH-SY5Y cell death compared to mock-treated controls. In contrast, Rotenone, Paraquat and LY83583 induced a marked significant increase in TrkAIII SH-SY5Y cell death following pre-treatment with the TrkA tyrosine kinase inhibitors CEP-701, K252a, Gö6976 or GW441756. CEP-701, K252a and Gö6976 alone did not induced more than 25% TrkAIII SH-SY5Y cell death, and GW441756 alone did not induce significant levels of TrkAIII SH-SY5Y cell death (Figs. 10C and 10D).
From these data, we conclude that TrkAIII tyrosine kinase activity is required to inhibit Rotenone, Paraquat or LY83583induced cell death.

SiRNA SOD2 Knockdown Re-sensitizes TrkAIII SH-SY5Y Cells to Agents that Induce ROS-Mediated Cell Death
In order to confirm SOD2 involvement in TrkAIII SH-SY5Y cell resistance to agents that induce ROS-mediated cell death, TrkAIII SH-SY5Y cell sensitivity to Rotenone, Paraquat and LY83583 induced death was assessed following siRNA knockdown of SOD2 expression and compared to TrkAIII SH-SY5Y cells treated with control siRNA.
SiRNA knockdown of SOD2 expression in TrkAIII SH-SY5Y cells restored sensitivity to Rotenone, Paraquat and LY83538induced death compared to TrkAIII SH-SY5Y cells treated with control siRNA (Figs. 11A and 11B). From these results, we conclude that elevated SOD2 expression in TrkAIII SH-SY5Y cells is largely responsible for attenuating Rotenone, Paraquat, and LY83583-induced death.

Discussion
In the present study, we report that constitutive expression of the TrkAIII oncogene in human SH-SY5Y NB cells inhibits mitochondrial free radical reactive oxygen species (ROS)-mediated death induced by Rotenone, Paraquat and LY83583, by stimulating SOD2 expression, increasing mitochondrial SOD2 activity and attenuating mitochondrial free radical ROS production. These effects associate with increased mitochondrial capacity to produce H 2 O 2 , occur within the context of a more tumour stem cell-like phenotype, and can be reversed by the specific TrkA tyrosine kinase inhibitor GW441756 [51], by the multi-kinase TrkA inhibitors CEP-701 [59], K252a [49] and Gö6976 [50] and by siRNA knockdown of SOD2 expression, all of which restore the sensitivity of TrkAIII SH-SY5Y cells to Rotenone, Paraquat and LY83583-induced mitochondrial free radical accumulation and free radical-mediated death.
TrkAIII stimulated the expression of SOD2 protein and mRNA in SH-SY5Y cells without altering the expression of antioxidants SOD1, catalase, glutathione peroxidase or thioredoxin-2, indicating relative specificity of action for the antioxidant SOD2. The specific TrkA tyrosine kinase inhibitor GW441756 [51] and the multi-kinase TrkA inhibitors CEP-701, K252a and Gö6976 [49,50,59,70], all of which inhibited TrkAIII activity [this study, 1], inhibited the expression of SOD2 mRNA and protein in TrkAIII SH-SY5Y but not TrkA SH-SY5Y cells. Furthermore, SOD2 expression in SH-SY5Y cells was not stimulated by the expression of Y670/674/675F mutated kinase dead TrkAIII, confirming that stimulation of SOD2 expression was dependent upon TrkAIII tyrosine kinase activity. TrkAIII stimulated the expression of the 4.2 kb relative to 1.5 kb SOD2 mRNA species, within the context of a proliferating, undifferentiated phenotype, suggesting stimulation of SOD2 transcription and supporting a previous report that transcription of the 4.2 kb SOD2 mRNA species predominates in un-differentiated, proliferating cells [34]. SOD2 expression was inhibited by the IP3K inhibitor LY294002 [52], and by the NF-k B inhibitors PDTC [54] and Bay 11-7082 [55] in TrkAIII SH-SY5Y cells but not in TrkA SH-SY5Y cells, but was not inhibited by the MAPK inhibitor PD098059 [53]. This implicates IP3K and NF-kB but not MAPKs in TrkAIII stimulation of SOD2 expression, supports previous observations that spontaneously active TrkAIII signals through IP3K/Akt/NF-kB but not Ras/MAPK [1] and that SOD2 expression is induced by NF-kB in NB and other cell lines [25,71] but differs to reports that SOD2 expression is stimulated by the TrkA ligand NGF through rac7, Ki-Ras and CREB transcription factor [14,15,47]. Here, we confirm that NGF stimulates SOD2 expression [14,15,47] in TrkA SH-SY5Y cells (this study), adding to observations that NGF induces Ras/MAPK signalling and neuronal differentiation in TrkA SH-SY5Y cells [1], and suggesting that SOD2 expression in SH-SY5Y cells could be regulated differently by spontaneously active intracellular TrkAIII and NGF-activated cell surface TrkA receptors. We do not exclude, however, that mechanisms in addition or alternative to NF-kB -mediated SOD2 transcription may regulate SOD2 expression in TrkAIII SH-SY5Y cells, since 4.2 kb SOD2 mRNA is also regulated by small inhibitory RNAs of Alu/7SL origin [35] and by SPB1 binding protein [36]. It is interesting that the IP3K inhibitor LY294002 also inhibited TrkAIII phosphorylation. This was detected following 24 hour but not 3-hour treatments, suggesting an indirect rather than direct inhibitory mechanism and implicating IP3K in the maintenance of constitutive of TrkAIII phosphorylation. We are further investigating this observation.
TrkAIII stimulation of SOD2 expression in SH-SY5Y cells was associated with a more stem cell-like phenotype and not with neuronal differentiation. Previously, SOD2 has been directly implicated in NGF/TrkA-induced neuronal differentiation [14,15,47], corroborated by our own observations that NGF stimulates SOD2 expression in TrkA SH-SY5Y cells (this study) in association with neuronal differentiation [1,12,13]. This suggests that SOD2 involvement in neuronal differentiation must depend upon additional factors not activated by TrkAIII, which would include Ras/MAPK/Erk signalling induced by NGF-activated TrkA but not spontaneously active TrkAIII in SH-SY5Y cells [1]. In support this, activation of the NGF/TrkA/SOD2/H 2 0 2 axis within the context of ki-Ras/MAPK/Erk signalling has been reported to promote neuronal differentiation by prolonging Erk1/ 2 transcriptional activity, augmenting Akt activation and inducing cytoskeletal complexes between microtubules, mitochondria and ki-Ras containing plasma membranes, in a coordinated mechanism of microtubule assembly and assembly promoting factors [15,47,72]. In the absence of Ras/MAPK/Erk signalling, the TrkAIII/SOD2/H 2 0 2 axis may also regulate the assembly of cytoskeletal components important for maintaining cells in an undifferentiated stem cell-like state. In support of this, TrkAIII interacts with the interphase centrosome and promotes microtubule assembly from centrosomal microtubule organising centre, promoting formation of short peri-nuclear microtubule arrays, similar to those that maintain cells in an undifferentiated state [3,5,73].
TrkAIII promotion and maintenance of an undifferentiated SH-SY5Y phenotype [1,5], is further supported by the increased capacity of TrkAIII SH-SY5Y cells to grow as large tumour spheroids in vitro, which is considered to represent cancer stem celllike behaviour in NB cell lines [66]. The increased size of TrkAIII SH-SY5Y spheroids did not only associate with increased expression of 4.2 kb SOD2 mRNA but also with increased expression of the neural stem cell markers CD117, SOX2, Nestin and Nanog. The possibility that TrkAIII may not only maintain SH-SY5Y cells in an undifferentiated state but may also promote staminality was supported by the reduction of Nanog, Nestin and SOX2 mRNA expression in TrkAIII SH-SY5Y tumour spheroids following incubation with the specific TrkA tyrosine kinase inhibitor GW441756 [51]. This directly implicates TrkAIII in the regulation of SOX2, Nestin and Nanog involvement in NB cancer stem cell biology [74], is consistent with reports that cancer stem cells express higher levels of SOD2 [29,45,46] and is, therefore, of relevance to regulation of the cancer stem cell niche in high TrkA expressing unfavourable NBs.
The capacity to grow as tumour spheroids exhibited by control, TrkA and TrkAIII SH-SY5Y cells was completely inhibited by the multi-kinase TrkA inhibitors CEP-701 and K252a [49,59,70]. In contrast, the TrkA specific inhibitor GW441756 [51], which inhibited TrkAIII tyrosine phosphorylation and reduced SOD2 expression in TrkAIII SH-SY5Y cells, inhibited spheroid growth exclusively in TrkAIII SH-SY5Y cells reducing spheroid volumes to those formed by TrkA SH-SY5Y cells. These data implicate kinases other than TrkA in SH-SY5Y cell spheroid growth and confirm that TrkAIII promotes the formation of larger SH-SY5Y spheroids. The fact that TrkAIII does not stimulate SH-SY5Y cell proliferation [1], combined with observations of spheroid aggregation by control, TrkA and TrkAIII SH-SY5Y cells, suggests that the formation of larger spheroids by TrkAIII SH-SY5Y cells most likely reflects the increased stress-resistance exhibited by TrkAIII SH-SY5Y cells [this study, 1, 2], rather than increased proliferation and/or spheroid-aggregation. Within the context of large tumour spheroid-aggregates, stress would result from reduced oxygen and nutrient diffusion [75], which would be expected to regulate spheroid size on the basis of relative resistance to stress. SiRNA knockdown of SOD2 expression significantly reduced mean TrkAIII SH-SY5Y tumour spheroid volume but, unlike GW441756, did not reduce Nanog, Nestin, SOX2 or CD117 expression in tumor spheroids, implicating SOD2 in the promotion of SH-SY5Y tumour spheroid growth and survival but not staminality. The TrkAIII/SOD2 axis may, therefore, act in an analogous but opposite way to the NGF/TrkA/SOD2 axis, by driving NB staminality and survival rather than neuronal differentiation and survival [1,4,[12][13][14][15]47,76].
The increased levels of mitochondrial SOD2 protein and activity in TrkAIII SH-SY5Y cells is consistent with the mitochondrial translocation sequence present within SOD2 and confirms that TrkAIII does not inhibit SOD2 activity, which is negatively regulated by tyrosine nitration, serine/threonine phosphorylation and lysine acetylation [17,18,37]. Consistent with high mitochondrial SOD2 levels, TrkAIII SH-SY5Y cells exhibited attenuation of mitochondrial free radical ROS accumulation induced by the mitochondrial free radical ROS inducing agents Rotenone [56], Paraquat [57] or LY83583 [58]. Furthermore, highly purified mitochondria from TrkAIII SH-SY5Y cells also exhibited increased capacity to produce H 2 0 2 in response to succinate, also consistent with increased mitochondrial SOD2 activity. The sensitivity of TrkAIII SH-SY5Y cells to Rotenone, Paraquat and LY83583-induced mitochondrial ROS accumulation could be restored by the specific TrkA inhibitor GW441756, the multi-kinase TrkA inhibitors CEP-701, K252a and Gö6976, and also by siRNA knockdown of SOD2 expression, confirming that resistance to Rotenone, Paraquat and LY83583-induced mitochondrial ROS accumulation was TrkAIII tyrosine kinasemediated and SOD2 dependent.
Since the unbridled accumulation of both free radical and nonfree radical ROS causes oxidative damage and cellular death [16,21,77,78], we compared the relative sensitivity of pcDNA, TrkA and TrkAIII SH-SY5Y cells to Rotenone, LY83583 and Paraquat-induced ROS-induced death. TrkAIII SH-SY5Y cells were significantly more resistant to Rotenone, LY83583 and Paraquat-induced death than either pcDNA SH-SY5Y or TrkA SH-SY5Y cells, indicating that mitochondrial free radical ROS rather than non-free radical H 2 0 2 were largely responsible for cell death under these conditions. The sensitivity of TrkAIII SH-SY5Y cells to Rotenone, Paraquat and LY83583-induced death was also restored by the specific TrkA inhibitor GW441756, the mutikinase TrkA inhibitors CEP-701, K252a and Gö6976 and by siRNA knockdown of SOD2 expression. These observations extend our previous report that TrkAIII protects SH-SY5Y cells against doxorubicin-induced death [1], also mediated by mitochondrial free radical ROS [22], and confirm that TrkAIII protection of SH-SY5Y cells against Rotenone, LY83583 and Paraquat-induced death results from the attenuation of mitochondrial free radical ROS accumulation, is TrkAIII tyrosine kinasemediated and SOD2 dependent.
The increased capacity of mitochondria from TrkAIII SH-SY5Y cells to produce H 2 0 2 , indicates that SH-SY5Y cells have an adequate capacity to detoxify H 2 0 2 , which is supported by the high levels of the H 2 0 2 detoxifiers catalase and glutathione peroxidase expressed by SH-SY5Y cells. Mitochondrial H 2 0 2 has been implicated in malignant progression, augments genetic instability by increasing sister chromatid exchange and chromosomal translocation [79,80], increases MMP-expression and MMPdependent invasion and promotes metastatic dissemination [28,[40][41][42]. TrkAIII also augments genetic instability by increasing sister chromatid exchange and chromosomal instability, increases MMP-9 expression and promotes metastatic growth [1][2][3][4]. We are currently investigating whether these oncogenic effects of TrkAIII are also mediated by an increase in mitochondrial SOD2-mediated H 2 0 2 production.
In conclusion, our study shows that TrkAIII promotes SH-SY5Y NB cell resistance to agents that induce mitochondrial free radical ROS-mediated death by up-regulating SOD2 expression, increasing mitochondrial SOD2 activity and attenuating the accumulation of mitochondrial free radical ROS, in association with up-regulated mitochondrial capacity to produce H 2 0 2 and within the context of a more tumour stem cell-like phenotype. This novel TrkAIII/SOD2 axis function suggest that the combined use of TrkAIII and/or SOD2 inhibitors, together with radiotherapy or chemotherapeutic agents that induce mitochondrial free radical ROS-mediated death, could provide a therapeutic advantage that may not only enhance tumour cytotoxicity but may also reduce malignant behaviour and target the stem cell niche in high TrkA expressing unfavourable NBs.