Cytosolic caspases mediate mislocalised SOD2 depletion in an in vitro model of chronic prion infection

The transmissible spongiform encephalopathies (TSEs; also known as prion diseases) encompass a group of fatal neurodegenerative diseases that differ from other dementias, such as Alzheimer’s disease, because of their transmissible nature. The causative agent is composed wholly or largely of misfolded conformers of the prion protein (PrP) (Prusiner, 1982; Weissmann et al., 1994). These misfolded conformers (PrP^Sc) form a template for the cellular isoform of PrP (PrP^C) to misfold and in this way propagate themselves, resulting in the transmissibility of the TSEs.

Similar to other neurodegenerative diseases, prion disease in mice and humans is associated with markers of oxidative stress in the brain (Wong et al., 2001; Freixes et al., 2006), which increase concurrently with disease-associated PrP burden (Brazier et al., 2006). The oxidative stress has been attributed to an increased production of reactive oxygen species (ROS) due to metal ion dyshomeostasis and resulting redox activity (reviewed in Singh et al., 2010), changed redox signalling through NADPH oxidase (Schneider et al., 2003; Mouillet-Richard et al., 2007) and alterations in nitric oxide synthase (Park et al., 2011). PrP itself has been linked with an antioxidant function; possibly due to an inherent superoxide dismutase (SOD)-like activity (Brown et al., 1999) or by modulation of protective signal transduction pathways (Mouillet-Richard et al., 2007; Rachidi et al., 2003). Aside from a potential PrP SOD-like function, the cell has two further intracellular SODs: SOD1 (CuZnSOD) and SOD2 (MnSOD). SOD2 is localised within the mitochondria, whereas SOD1 has a more ubiquitous localisation (Kawamata and Manfredi, 2010). Reduced SOD activity is reported in prion protein knockout mice and cell cultures (Brown et al., 1997; Brown and Besinger, 1998; Klamt et al., 2001; Sakudo et al., 2005), and a loss of SOD function is one proposed mechanism of prion disease pathogenesis. Alterations in activity and redistribution to mitochondria is reported for SOD1 mutants associated with genetic amyotrophic lateral sclerosis (ALS) in humans and mice (Carrì and Cozzolino, 2011; Goldsteins et al., 2008) indicating that mislocalisation and altered activity of the SOD enzymes can have deleterious consequences for neurons.

Our previous work defined four stages of oxidative response in cultured cells exposed to infectious prions: acute, adaptive, chronic and terminal (Haigh et al., 2011). In these phases the oxidative state of the cell changes from oxidative stress to adaptational response to ‘normal’ oxidative capacity to loss of normal oxidative metabolism and increased ROS preceding death. The terminal subpopulation of cells accounts for ∼6% of the chronically infected population and these cells show markers of apoptosis [phosphatidylserine (PS) externalisation] that correlate with detrimental oxidative stress, suggesting a possible failure of antioxidant compensatory responses. This apoptotic response is associated with the formation of aggregates of highly peroxidised lipid and activated caspases to which PrP is targeted. In this terminal phase of infection, the cells seem to be unable to maintain a defensive response to the increasing ROS burden, with a higher density of these aggregates appearing in apoptotic cells. We therefore sought to further investigate the role of the cellular antioxidant defence mechanisms in chronic and terminal prion infection in vitro. Primarily, we aimed to define the failure(s) in the adaptational response to chronic infection that result in the infected cells progressing to apoptosis. Our results suggest that chronic prion infection is associated with SOD2 mislocalisation, for which increased SOD1 activity compensates. The mislocalised SOD2 is directed to redox-active aggregates in the cytosol where it is degraded by caspases, resulting in a reduction of SOD2 protein levels. Eventually, mitochondria fail to cope with the lack of SOD2, which manifests as increased superoxide production, and ultimately mitochondrial numbers decline such that the cell can no longer maintain viability, resulting in the activation of apoptosis.

moRK13 cells [rabbit kidney epithelial (RK13) cells expressing full-length murine PrP] readily propagate infectious prions and the line remains viable indefinitely, allowing the study of chronic cellular infection. Our previous work has shown that chronically infected cells do not show a difference in the overall production of intracellular ROS; however, increased ROS production is seen in a subpopulation of infected cells expressing apoptotic markers when separated from the predominant population (Haigh et al., 2011). To investigate the possibility that the greater population was maintaining normal ROS levels by increased antioxidant defences, we looked at intracellular total SOD activity in the infected and mock equal-passage moRK13 populations. All experimental data presented herein are generated using the moRK13 cell model unless otherwise stated. The SOD activity assay showed an increased activity in the infected cells (Fig. 1A); this activity was approximately double that seen in the mock cells. Confirming that the results were not a cell-specific effect, we also assayed differentiated murine neural stem cells (NSCs) for changes in total SOD activity after 5 days of exposure to infectious mouse brain homogenate (Fig. 1B). Increased SOD activity was found in these cultures, which displayed a PrP expression dependence. The NSCs were treated for only 5 days with infectious homogenate owing to the pronounced toxicity to these cells (Fig. 1C–E) (Haigh et al., 2011). In order to determine whether the increase in total SOD activity in the moRK13 cells was predominantly due to SOD1 or SOD2 or both, we used pharmacological intervention to inactivate each of the SODs individually. The doubled SOD activity in the infected cells is maintained when SOD2 is removed by precipitation out of the lysate but is lost when SOD1 is inactivated, indicating that SOD1 is contributing the increased SOD activity (Fig. 1F).

SOD1 and SOD2 protein levels within the mock and infected moRK13 cells were analysed by western blotting. Protein detection indicated that SOD2 levels were significantly decreased by around 50% in the infected cells (Fig. 1G), but SOD1 levels remained unchanged. When compared with the detected protein levels of each SOD, SOD2 activity was not changed relative to its expression levels. SOD1 activity was significantly increased but a concurrent increase in protein expression was not detected, indicating a possible allosteric modulation of SOD1 function (supplementary material Fig. S1).

SOD2 is mitochondrially localised. With the protein levels of SOD2 decreased in the infected moRK13 cells as compared with the mock cells we sought to determine whether the mitochondria of these cells were experiencing greater oxidative stress. To measure the levels of superoxide within the mitochondria, MitoSOX fluorescent sensor probe was incubated in the live moRK13 cells and intensity levels determined. No significant difference between the mock and infected cells in mitochondrial superoxide production could be detected (Fig. 2A). This is consistent with our previously obtained results in these cells, which showed no difference in ROS production or evidence of oxidative stress between the mock and infected cells (Haigh et al., 2011). However, given the significant decrease in SOD2 protein levels, we queried whether other cellular changes might account for the apparent lack of change in mitochondrial superoxide; for example, rather than being due to maintenance of homeostasis, it might be a result of decreased numbers of stressed mitochondria resulting in a ‘normal’ fluorescent signal (i.e. a lower quantity of mitochondria with a higher fluorescent signal). We therefore utilised two further mitochondrial probes: MitoTracker and Mito-RFP. The mitochondrial importation of both of these probes might be altered by the mitochondrial state of health (probe metabolism or active transport of probe), so both were considered together. No difference was found between the mock and infected moRK13 cells for either probe (Fig. 2B,C). As a further indication of mitochondrial stress, cellular metabolism of formazan was considered. Formazan metabolism occurs in both the mitochondria and the cytosol, so two inhibitors (rotenone, which indicates mitochondrial contribution by inhibiting NAD-linked substrate oxidation at the oxygen side of NADH dehydrogenase and DPI, an inhibitor of NADPH oxidase enzymes in the cytosol) were used to differentiate the mitochondrial contribution. Neither the cellular rate of formazan metabolism nor the mitochondrial contribution was altered in the infected moRK13 cells (Fig. 2D,E). Finally, we considered the mitochondrially localised anti-apoptotic protein Bcl-2, which has previously been linked with PrP binding and toxicity (Kurschner and Morgan, 1995; Rambold et al., 2006). This protein was decreased significantly in the infected moRK13 cells (Fig. 2F). Overall, it would seem that mitochondrial numbers, superoxide levels and function are maintained during chronic prion infection but with selective loss of SOD2 and Bcl-2.

In our previous study ROS levels were only found to be increased in the apoptotic subpopulation of infected moRK13 cells (Haigh et al., 2011). This population was only a small proportion of the entire culture population and changes in such a small number of cells might easily be missed; therefore, it was possible that SOD-activity was increased to compensate for the ROS insult in the dying cells only. Mock and infected moRK13 cells were separated based upon PS externalisation, a marker for apoptosis, and the SOD activity assays repeated. The opposite response to that hypothesised was observed, with increased SOD activity only observed in the nonapoptotic population of infected cells but not in the cells separated by their externalised PS and therefore committed to apoptotic cell death (Fig. 3A).

Because the increased total SOD activity was not maintained in the apoptotic infected cells, it was plausible that, in the absence of appropriate antioxidant control, the mitochondrial superoxide production could be increased. Therefore, we tested the superoxide production in these cells using the MitoSOX probe. Although mitochondrial superoxide was increased in both populations of apoptotic moRK13 cells (mock and infected), the infected apoptotic cells showed significantly higher levels than the non-infected apoptotic cells (Fig. 3B). Examining these cells with the MitoTracker and Mito-RFP probes indicated that the apoptotic infected cells had lower numbers of mitochondria, as determined by lower fluorescence levels (Fig. 3C,D). Relatively, in the pro-apoptotic infected cells, fewer mitochondria were generating considerably more superoxide.

Various pathways by which mitochondria can be lost are reported within the literature. Mitochondria can be extruded from cells by being engulfed in vesicles that are similar to autophagic vacuoles, with release into the surrounding milieu (Lyamzaev et al., 2008; Nakajima et al., 2008). To look for evidence of mitochondrial extrusion as the cause of their loss in the pro-apoptotic infected cells, conditioned media from the whole, unseparated moRK13 cell populations was western blotted for SOD2 and Bcl-2 as evidence that mitochondria were located outside of the cell. Although significant signal was seen for SOD2 in both the mock and infected conditioned media, and this was greater for the infected cells, there was no detectable Bcl-2, suggesting that mitochondria were not being extruded but SOD2 was being expelled from the cell by some other means (Fig. 4A).

Another mechanism by which mitochondria are lost in disease is through mitoptosis, considered the mitochondrial equivalent of apoptosis (Mijaljica et al., 2010; Tinari et al., 2007). In mitoptosis, mitochondria can fragment, losing their outer membrane and releasing their contents into the cytosol, thus losing their functionality, or suffer fragmentation of the inner membrane leaving only a dysfunctional outer husk (Mijaljica et al., 2010; Tinari et al., 2007). Mitochondrial preparations of the whole, unseparated moRK13 cell populations were examined to look for evidence of SOD2 in the cytosol of infected cells as a result of leaky mitochondria. Western blotting showed that SOD2 was increased relative to total cellular SOD2 detection in the cytosol of the infected cells compared with their mock counterparts (Fig. 4B). Bcl-2 blotting confirmed minimal cross-contamination of the cytosolic fraction with the mitochondrial fraction (Fig. 4B), suggesting a selective relocalisation of SOD2.

Western blotting indicated a mislocalisation of SOD2 into the cytosol of infected cells, so immunofluorescence was used to image the location of both SOD1 and SOD2 in the cytosol of mock and infected moRK13 cells. Acknowledging that a change of location can be part of an apoptotic pathway, co-staining with a rhodamine-labelled pan-caspase inhibitor (SR-VAD-FMK) was also performed. Both SOD1 and SOD2 localise with the previously described (Haigh et al., 2011) caspase-positive aggregates within the infected cells but SOD2 shows a more pronounced colocalisation, with decreased levels of SOD2 elsewhere throughout the cell (Fig. 4C). SOD2 colocalisation with caspase-positive aggregates was also observed in cultures of differentiated wild-type NSCs exposed to infectious inoculum for 5 days (Fig. 4D,E). Immunofluorescent staining shows that cleaved (active forms of) caspases 3 and 9 localise with these aggregates but that pro-caspase 3 and caspase 6 show very little localisation (supplementary material Fig. S2). To test the hypothesis that SOD2 was being degraded by mislocalisation to the cytosolic caspase aggregates, the caspase inhibitor was used to block caspase activity for up to 48 hours. This treatment has previously been shown to increase the viability of these cells (Haigh et al., 2011). At 24 hours of caspase inhibitor treatment, the level of SOD2 protein within the cell was significantly increased (Fig. 4F), with detected levels in the infected cells restored to match those measured in the mock cells. Bcl-2 is also cleaved by caspase 3 to a Bax-like pro-apoptotic form (Cheng et al., 1997). Therefore, we also treated the mock and infected cells with the caspase inhibitor and also a radical scavenger, in case ROS production was also influencing Bcl-2 loss, at the 24-hour time point to see whether Bcl-2 also increased following treatment. No significant differences in Bcl-2 levels were observed in the treated infected cells compared with the untreated (supplementary material Fig. S3).

Previously, we have shown unchanged ROS levels in overall populations of infected cells compared with their mock counterparts during chronic prion infection; however, the cells in the acute or apoptotic terminal phase had higher ROS levels than the mock control cells (Haigh et al., 2011). Prompted by the observation that overall ROS levels were ‘normal’ during chronic infection but the apoptotic subpopulation showed oxidative stress, we investigated the cellular antioxidant defence pathways responsible for compensation to the chronic infection and the adaptational failure(s) associated with progression to apoptosis. Regulation of, or influence on, cellular SOD activity by PrP has been an important ongoing theme in prion infection and therefore SOD1 and SOD2 were the candidate molecules chosen for examination. SOD1 activity was increased and SOD2 protein levels decreased in chronically infected cells. SOD2 downregulation has previously been associated with prion infection in vivo (Choi et al., 1998; Lee et al., 1999; Park et al., 2011). Although these studies did not report an increase in SOD1 activity, regulation of SOD1 activity by PrP has also been reported (Brown et al., 1997; Sakudo et al., 2005) and SOD1 knockout has been shown to significantly reduce disease incubation time in murine models of prion disease (Akhtar et al., 2013). In the current system, the increased SOD1 activity might explain how the majority of moRK13 cells are able to remain viable in culture, despite reduced SOD2 levels, while propagating infection over long periods of time. The upregulated SOD response to infectious inoculum was also seen in the NSC cultures. The NSCs show an acute toxic response of increased ROS and decreased viability within 24 hours of exposure to infectious PrP (Fig. 1) (Haigh et al., 2011) and therefore were assayed after 5 days to allow a recovery period after the acute insult. Although it is not possible to discern whether it is de novo PrP^Sc propagation or PrP^Sc exposure that causes toxicity, those cells that survive to 5 days post-exposure have increased their antioxidant defences to compensate.

Given that SOD1 activity was increased in chronically infected cells, the subpopulation of apoptotic cells were separated from nonapoptotic cells to assess the activity in this apoptotic population, which has previously been shown to have increased ROS production (Haigh et al., 2011). The apoptotic population is relatively small compared with the non-apoptotic population, but still constitutes ∼6% of infected cells, double the 3% of mock cells (Haigh et al., 2011), and so the responses of these cells can easily be overlooked because detection of these changes are overwhelmed by the remaining population. The total SOD activity in the infected apoptotic cells was not different from the levels seen in the mock cells (either separated or not); therefore, the overall increase in SOD1 activity must be due to the non-apoptotic population. This seems to indicate a relative decline in SOD activity in the infected apoptotic cells, thereby allowing detrimental ROS production within these cells, which contributes to their demise.

Mitochondrial respiration is one of the major cellular processes that produce ROS. Because levels of the mitochondrially localised SOD2 were decreased in chronically infected cells, mitochondrial representation within the cell was considered, but no difference was seen between the overall mock and infected populations. This lack of change is consistent with the investigations of Sisková et al., who found no prion-disease-associated changes in mitochondrial density or expression of mitochondrial proteins (Sisková et al., 2010). The same study did find changes to the mitochondrial inner membrane morphology and reduced cytochrome c oxidase activity, and similar observations were also made by others (Choi et al., 1998; Lee et al., 1999).

The pronounced loss of mitochondria in the pro-apoptotic prion-infected cells might indicate a primary contribution to cell death as part of the apoptotic response. Mitoptosis, the pathways of mitochondrial death, manifests in various ways (Lyamzaev et al., 2008; Nakajima et al., 2008). Although we saw evidence of SOD2 in conditioned media, there was no evidence for mitochondrial extrusion, indicating that the cytosolic SOD2 was apparently independently being expelled from the cell. In the absence of other intracellular molecules, such as tubulin, this was probably a specific event as opposed to non-specific cell lysis or fragmentation. Colocalisation with active caspases in the previously reported oxidised aggregates (Haigh et al., 2011) indicated that SOD2 might be degraded when located outside the mitochondria. No evidence of mitochondria has ever been observed in these aggregates, additionally suggesting that the SOD2 is selectively mislocalised. Mislocalisation might be a result of mis-targeting of SOD2 to the cytosol following synthesis or could be due to loss from failing mitochondria. Furthermore, in the current study and as previously reported (Haigh et al., 2011), in cells in which the caspase reactive aggregates are observed there is no apparent depletion of mitochondria, indicating that the SOD2 relocalisation to the aggregates occurs while the mitochondria are still present and intact. Park et al. found reduced mitochondria in an in vivo model of prion disease, which occurred concurrently with a decrease in SOD2 (Park et al., 2011). Electron microscopy comparing infected and uninfected mouse brain tissue found evidence of loss of internal structures of the mitochondria. The authors correlated this with increased endothelial nitric oxide synthase (eNOS) and thus with increased intra-mitochondrial oxidative stress. This is consistent with our findings, but does not explain the relatively selective loss of the SOD2 protein from the mitochondria at a stage before mitochondrial function is apparently significantly compromised and the presence of this organelle depleted.

A decline in SOD2 protein level and function does not seem to be a prion-disease-specific pathway. Similar decreases are reported in other conditions of chronic oxidative stress such as heart failure, in which mRNA levels are increased but protein levels decreased (Holley et al., 2011; Sam et al., 2005), and osteoarthritis, in which SOD2 protein decreases precede erosion of cartilage (Scott et al., 2010). A paradigm attempting to recapitulate a generic model of chronic low-level oxidative stress using hydrogen peroxide exposure did not replicate a reduction in protein levels of SOD2 in the moRK13 cells (supplementary material Fig. S4), but rather an increase in SOD2 protein was instead observed. This might imply that the reduction in SOD2 protein levels in the chronic phase of prion infection requires much longer periods of time to manifest, or is caused by a specific ROS or reactive nitrogen species or localisation of these species, or that other dysfunctional pathways are also contributing to SOD2 protein level reduction. Further investigation is required to resolve these intricacies.

Caspase degradation of SOD2 has been reported as part of the extrinsic apoptotic pathways, with two caspase 3 cleavage sites identified within human SOD2 and recombinant caspase 3 shown to cleave recombinant SOD2 (Pardo et al., 2006). Activation of CD95 (also known as Fas) apoptotic pathways has been shown to redistribute PrP^C to mitochondria in lymphoid lineage cells (Mattei et al., 2011). Li et al. found that cytosolic PrP disrupts the proper formation of the cytoskeleton (Li et al., 2011). Together, the altered cytoskeleton and the direct PrP (or the disease-associated PrP^Sc isoforms) action on mitochondria might serve to mis-target SOD2 to the cytosol, permitting its incorporation into the caspase reactive aggregates and its loss into the surrounding cellular environment. Caspase 3 has also been shown to cleave Bcl-2 (Cheng et al., 1997) and could represent a common pathway, linking the decreases of SOD2 and Bcl-2. Interestingly, the resulting peptide formed by caspase 3 cleavage of Bcl-2 has Bax-like death effector function (Cheng et al., 1997) and might provide an explanation as to why prion-infected mice that overexpress Bcl-2 succumb to disease earlier than wild-type mice (Steele et al., 2007), apparently contrary to the expected anti-apoptotic function of Bcl-2. Although a pancaspase inhibitor failed to restore Bcl-2 levels in our cell system (supplementary material Fig. S3), we cannot rule out that prion-infected cells downregulate protein expression of Bcl-2 to protect against deleterious consequences of caspase cleavage producing the pro-apoptotic form.

Evidence of caspase activation has been detected in brain tissue obtained post-mortem from human Creutzfeldt-Jakob disease (CJD) patients (Jesionek-Kupnicka et al., 1997; Puig and Ferrer, 2001) and can be observed in the brains of live mice in the pre-terminal phases of prion disease (Lawson et al., 2010), with caspase-reactive aggregates observed inside neurons using the same active caspase label as in the current study (Drew et al., 2011). Furthermore, in GT1 and N2a cell models of prion infection, aggresomes associated with increased caspases 3 and 8 activity are formed when the proteosome is inhibited (Kristiansen et al., 2005). The targeting of SOD2 to similar aggregates, with a loss of up to 50% of normal activity in the chronically infected cells, before any detectable decline in mitochondrial numbers is indicative of this event forming part of the toxic pathways that ultimately contributes to mitochondrial loss (as a result of superoxide accumulation) and cellular demise.

SOD2 knockout mice develop a neonatal lethal cardiomyopathy and metabolic acidosis within the first week of life; aside from the heart, post-mitotic tissues with high metabolic requirements including the brain are the most severely affected tissues (Lebovitz et al., 1996). Interestingly, if the mice are kept alive beyond the expected 1 week using antioxidant therapy, they develop an oxidative-stress-mediated spongiform encephalopathy (Melov et al., 1998). Complexes I–IV of the electron transport chain were found to have diminished activity and protein levels, which was pronounced for complex II, directly correlating with the dose of antioxidant used to prolong the life of the mouse (Hinerfeld et al., 2004). PrP knockout mice show increased superoxide production from complex I (Paterson et al., 2008). During prion disease, increased superoxide due to the conversion of PrP^C to PrP^Sc, coupled with a loss of SOD2 and therefore its antioxidant activity, might begin a cycle of heightened oxidative damage within mitochondria, eventuating in their loss. SOD2 mimetics have been proposed as potential therapeutics for prion disease (Fukuuchi et al., 2006). One such mimetic demonstrated an ability to extend the lifespan of prion-infected mice with reduced vacuolation within the hippocampus in early and terminal disease (Brazier et al., 2008), and therefore might indicate that a restoration of SOD2 function is crucial for cellular survival in prion disease. A schematic of potential cell death pathways enhanced by prion infection is shown in Fig. 5.

A further factor in this scenario is the heightened SOD1 activity in the non-apoptotic infected cells. As stated, this could represent a compensatory response that allows prolonged survival of cells in the same manner as using antioxidant therapy in SOD2 knockout mice. However, increased SOD1 activity might itself contribute to a loss of cellular ROS homeostasis with a potential toxic gain of function inadvertently contributing to cellular decline. In a murine model of ALS, in which mice express a mutant form of SOD1 displaying a toxic gain of function, the absence of PrP leads to decreased levels of Bcl-2 and vacuolation of motor neurons (Steinacker et al., 2010). These observations might indicate that an overactivity of SOD1 caused by or coinciding with a disturbance of a PrP regulatory function could be responsible for characteristic prion pathology and cellular demise. We did not observe any increased localisation of SOD1 in mitochondria of infected cells (data not shown), but cannot exclude that a gain of function is adversely affecting this organelle.

The mechanisms by which SOD2 is relocated from mitochondria and those responsible for the decline in mitochondrial numbers during terminal prion infection still remain to be revealed. The relocalisation might be a deliberate redirection by cellular targeting mechanisms to compensate for increased cytosolic oxidative flux or might be a side effect of the disease process. Overall, SOD2 mislocalisation seems to be an integral part of SOD2, mitochondria and ultimately cell loss in terminal prion infection.

In summary, elaborating on previous work, we believe that the present results support the contention that chronic prion infection is associated with relatively early loss of SOD2 from mitochondria, while organelle numbers and functionality are generally maintained. SOD2 is mislocalised to the cytosol where it is catabolised at least in part by caspases and also apparently undergoes increased selective extrusion from the cell through unclear mechanisms; overall cellular ROS homeostasis is maintained by compensatory increased SOD1 activity. Eventually, in a subset of chronically prion-infected cells, a deleterious cycle leads to initiation of apoptosis that is associated with prominent loss of mitochondria and heightened superoxide generation in the remaining dysfunctional mitochondria, culminating in the death of these cells.

Rabbit kidney epithelial (RK13) cells expressing full-length murine PrP (moRK13) were generated as described previously (Haigh et al., 2009). Cells were routinely cultured in Dulbecco’s modified Eagle’s medium (DMEM; Invitrogen, Melbourne, VIC, Australia) supplemented with 10% (v/v) foetal bovine serum (Lonza, Australia), 50 U/ml penicillin/50 μg/ml streptomycin solution (Sigma, Melbourne, VIC, Australia) and 2.5 μg/ml puromycin (Sigma). Cell cultures were maintained at 37°C with 5% CO₂ in a humidified incubator.

Neural stem cells (NSCs) were harvested from c57BL/6×SV129 mice of prnp^−/− (prion gene knockout) (Büeler et al., 1993), wild type or Tga20 (transgenic prion overexpression of approximately eightfold at the protein level) (Fischer et al., 1996) genetic background, as described in Haigh et al. (Haigh et al., 2011). NSCs were maintained in Neurocult complete proliferation medium (Stem Cell Technologies, Melbourne, VIC, Australia) at 37°C with 5% CO₂ in a humidified incubator. Differentiation was induced by transferral of cells into Neurocult complete differentiation medium (Stem Cell Technologies) and plating on a poly-D-lysine matrix (Sigma-Aldrich).

Cells were infected with the M1000 mouse adapted, human GSS-derived strain of prions or ‘mock-infected’ with normal brain homogenate (Lawson et al., 2008; Lewis et al., 2009) and permitted to propagate infection until the infected population was uniformly laden with protease-resistant PrP (PrP^Res; approximately >passage 50), a marker of propagating prions, as determined by cell blot assay (Fig. 1A). These cells are referred to as ‘mock’ or ‘infected’ throughout. De novo infections with M1000 were prepared by centrifugal collection of M1000-infected and mock-infected cells (300 g for 5 minutes), followed by washing in Dulbecco’s phosphate buffered saline (dPBS; Invitrogen) and centrifugation as before. Lysate pellets were weighed and a 10% wet pellet weight/volume homogenate made in dPBS. Naïve cells were overlaid with 0.01 or 0.1% (w/v) of the mock or M1000-infected homogenised lysate. Cells were returned to the incubator until they were ready to be passaged. Long-term infection (empirically determined by cell blotting, see below) is defined as greater than passage 50. New infections were routinely established to ensure cell drift did not bias results.

Samples were prepared by washing cells in dPBS, followed by lysis with ice-cold 0.1 M Tris/HCl, pH 7.4 containing 0.5% (v/v) Triton X-100, and 0.1 mg/ml of Complete Mini, EDTA free, Protease Inhibitor Tablets (Roche, Melbourne, VIC, Australia). Cell lysates were briefly centrifuged to separate membrane and nuclear debris and the supernatant collected for use in the assay. Determination of SOD activity was performed as described by the Superoxide Dismutase Activity Colorimetric Assay Kit protocol (Abcam, Sapphire Biosciences, Sydney, NSW, Australia).

SOD1 and SOD2 activity discrimination utilised chloroform precipitation of SOD1 (SOD2 does not survive this treatment) and inclusion of diethyldithiocarbamate (DDC; Sigma) in the sample buffer to deactivate SOD1. For chloroform precipitation, 96 μl of ice-cold 37.5/62.5 (v/v) chloroform/ethanol solution was added to 60 μl of cell lysate. Samples were centrifuged for 10 minutes at 1000 g at 4°C and the aqueous phase was removed and stored at 4°C until use. SOD1 inhibition was achieved by the addition of DDC to 1 mM final concentration. Remaining SOD activity in these samples was measured as above.

Cells were transferred into serum-free OptiMEM overnight (16 hours) and incubated under normal conditions. Media was collected and centrifuged at 150 g for 3 minutes to remove any cell carcasses. The media was then transferred to a fresh tube and mixed with four equivalent volumes of ice-cold methanol then stored at −20°C overnight. The precipitated protein was collected by centrifugation at 3273 g for 1 hour. Protein pellets were solubilised in 1× NuPAGE gel loading buffer (Invitrogen) with 5% (v/v) β-mercaptoethanol.

Mitochondria and cysolic preparations were separated by washing cells briefly in dPBS, then collecting the cell suspension in phosphate buffered saline (PBS) by centrifugation at 335 g for 5 minutes. The cell pellet was resuspended in hypotonic buffer (10 mM NaCl, 1.5 mM CaCl₂, 10 mM Tris, pH 7.5) and lysed on ice by needle aspiration 20 times through an 18-gauge needle. One quarter sample volume of 5× sucrose solution (1.65 M sucrose, 30 mM EDTA, 43 mM Tris, pH 7.5) was added to the homogenate and the sample mixed. Nuclei and unbroken cells were separated by centrifugation at 600 g for 10 minutes. The supernatant was transferred to a fresh tube and centrifuged for 10 minutes at 10,000 g. The supernatant (cytosolic fraction) was collected and the pellet (mitochondria) resuspended in 1× sucrose solution.

Cells were lysed in RIPA buffer [50 mM Tris-HCl pH 7.4, 150 mM NaCl, 0.1% (w/v) SDS, 0.5% (w/v) sodium deoxycholate, 1% (v/v) NP-40] at 37°C for 20 minutes. Samples were electrophoresed using the NuPAGE system (Invitrogen), transferred onto PVDF (Millipore) membranes and blocked as described previously (Haigh et al., 2009) with the exception of PrP blots, which used 2% (w/v) ECL Advance buffer with ECL Advance developer (GE Healthcare, Sydney, NSW, Australia). Protein of interest detection was made using the following dilutions of primary antibody: 1 in 2000 SOD1 (Abcam); 1 in 5000 SOD2 (Abcam); 1 in 1000 Bcl-2 (GE Healthcare); 1 in 5000 EP1802Y against PrP (Abcam) and 1 in 10,000 β-tubulin (Sigma-Aldrich) in PBS with 0.1% (v/v) Tween-20 (PBS-t). Secondary antibodies (Dako, Melbourne, VIC, Australia) were used at 1 in 5000 anti-rabbit (SOD1, SOD2), 1 in 10,000 anti-rabbit (PrP), 1 in 2000 anti-mouse (Bcl-2) and 1 in 10,000 anti-mouse (β-tubulin). ECL-plus (GE Healthcare) was used to visualise bands. Following blotting, total protein levels on gels were assessed by Coomassie brilliant blue staining [0.1% (w/v) Coomassie brilliant blue in 50% (v/v) methanol and 7% (v/v) acetic acid] for 2 minutes at room temperature with agitation. Membranes were rinsed briefly in destain [50% (v/v) methanol and 7% (v/v) acetic acid], then washed in destain for 10 minutes before drying and image capture.

The cell blot assay is a variation of that developed by Bosque and Prusiner (Bosque and Prusiner, 2000) as described by Lewis et al. (Lewis et al., 2009). PrP was detected using 1 in 10,000 dilution of ICSM18 monoclonal antibody (murine epitope 142–152; D-Gen, UK) in 1% milk PBS-t. Anti-mouse HRP secondary antibody (GE Healthcare) was used at 1 in 10,000 dilution and blots were visualised as above.

Cells were labelled with 5 μM MitoSOX fluorescent indicator probe (Invitrogen) in normal media for 10 minutes under standard incubator conditions. Media was then replaced with fresh, phenolred-free OptiMEM (Invitrogen) for the duration of the experiment.

Labelling with MitoTracker Green (Invitrogen) was performed with 1 μl MitoTracker in normal media for 30 minutes under standard culture conditions before transfer into phenol-red-free OptiMEM.

CellLights BacMAM 2.0 Mito-RFP (Invitrogen) at 50 particles per cell was included in normal culture media for no less than 24 hours before cells were to be used.

5 μl of one-solution MTS reagent (Promega, Sydney, NSW, Australia) was added per 100 μl of culture media in a 96-well plate and absorbency change measured at 492 nm for 2 hours at 37°C. Rates of change were determined from the linear portion of the curve.

A 600 μM stock solution of SR-VAD-FMK (Immunochemistry Technologies, Sapphire Biosciences) was prepared in sterile PBS (pH 7.4; Invitrogen) containing 20% (v/v) high-quality sterile-filtered DMSO (Sigma-Aldrich). Cells were plated 24 hours before the start of the assay. 0.6 μM of the SR-VAD-FMK solution was added to Opti-MEM^® reduced serum, phenol-red-free media and cells were incubated for 30 minutes under normal growth conditions. The cells were then washed and incubated in fresh Opti-MEM for the duration of the assay.

Cells were washed in Mg²⁺/Ca²⁺-free dPBS and then suspended using trypsin and gently triturated to remove cell clumps. Magnetic separation was performed as described in the magnetic-activated cell sorting (MACS) protocol (Miltenyi Biotec, Sydney, NSW, Australia). Briefly, cells were centrifuged for 10 minutes at 300 g, resuspended in 80 μl of binding buffer (Miltenyi Biotec) and 20 μl of MACS Annexin V Microbeads (Miltenyi Biotec) added. After 15 minutes incubation at 4°C, the cells were washed in 10 equivalent volumes of binding buffer, centrifuged for 10 minutes at 300 g and resuspended in 500 μl of binding buffer. MS columns (Miltenyi Biotec) were equilibrated with 500 μl of binding buffer. Labelled cells were applied to the column in a magnetic field and washed twice with 1 ml of binding buffer. The columns were removed from the magnet and cells eluted in 1 ml of binding buffer. These samples were then centrifuged for 5 minutes at 300 g and resuspended in OptiMEM media. Both annexin-V-labelled and unlabelled fractions were transferred into fresh media for use in ensuing experiments, which were carried out immediately to avoid loss of viability of the annexin-V-positive fraction.

Cells were grown in chambered coverslips (Nunc, Invitrogen) in normal media. Transfections or dye incubations were carried out before fixing. Cells were fixed in 4% (v/v) paraformaldehyde/PBS for 30 minutes and then permeabilised in PBS with 0.1% (v/v) Triton X-100 for 10 minutes. Coverslips were blocked in PBS with 10% FBS (v/v) and 1% (w/v) BSA for 30 minutes. Primary antibodies were diluted in PBS-FBS as appropriate (SOD1, 1 in 500; SOD2, 1 in 250, Abcam; neurofilament-L, 1 in 50, Stem Cell Technologies) and applied overnight at 4°C. Secondary Alexa-Fluor-488/647-conjugated antibodies (Invitrogen) were diluted 1 in 250 in PBS-FBS and were applied for 1 hour at room temperature. Coverslips were coated with Prolong-Gold mounting media (Invitrogen). Fluorophore-stained coverslips were protected from light at all times.

Luminescent signals of bands on western blots were captured using a Las-3000 intelligent dark box (FujiFilm, Berthold, VIC, Australia) and the intensity quantified, after the subtraction of background, by ImageJ 1.38x. Statistical analyses were carried out using GraphPad Prism 4 or Minitab15 statistical software. Graphs represent the mean and standard error of the mean (s.e.m.) of four independent experiments unless otherwise stated. Primary statistical tests are stated in the text; for one-way ANOVA analysis Tukey’s secondary test was applied and Bonferroni’s secondary testing was used for twoway ANOVA. Microscope images were captured using a Nikon Eclipse TE2000-E epi-fluorescence microscope (Nikon-Roper Scientific, Coherent Scientific, Adelaide, SA, Australia). All image parameters were maintained throughout an experiment, across all conditions, and the channels of probes sensitive to photo-oxidation were collected before other channels. For imaging data the ‘n’ represents independent experiments, which each included capturing sufficient fields to analyse >30 cells per experiment. Cell intensity analysis was done using Nikon NIS-elements 3.0 software with regions of interest used to gate individual cells.

The authors acknowledge Prof. Andrew Hill for the kind gift of the pIRES vector containing the PrP open reading frames and Dr Simon Drew for the SR-VAD-FMK reagent, and also thank Dr Victoria Lawson and Dr Marcus Brazier for helpful discussions.