|Home | About | Journals | Submit | Contact Us | Français|
12/15-Lipoxygenase (12/15-LOX) is an important mediator of brain injury following experimental stroke in rodents. It contributes to neuronal death, but the underlying mechanism remains unclear. We demonstrate here that in neuronal HT22 cells subjected to glutamate-induced oxidative stress, 12/15-LOX damages mitochondria, and this represents the committed step that condemns the cell to die. Importantly these events, including breakdown of the mitochondrial membrane potential, the production of reactive oxygen species (ROS), and cytochrome c release, can all be replicated by incubation of 12/15-LOX with mitochondria in vitro, without the need to add other cytosolic factors. Proteasome activity is required downstream of mitochondrial damage to complete the cell death cascade, but proteasome inhibition is only partially protective. These findings position 12/15-LOX as the central executioner in an oxidative stress-related neuronal death program.
Oxidative stress is an important feature of several neurodegenerative processes, including stroke, Parkinson’s Disease, and Alzheimer’s(Lo et al. 2005; Lin and Beal 2006). Neurons are particularly susceptible to oxidative stress-related injury, suggesting that preventing oxidative damage should be neuroprotective. Early studies of an oxidative stress model in cultured neuronal cells, oxidative glutamate toxicity, demonstrated a role for the lipid-oxidizing enzyme, 12/15-lipoxygenase (12/15-LOX; EC 22.214.171.124; the product of the ALOX15 gene, also known as leukocyte-type 12-LOX)(Li et al. 1997). Here, applying high concentrations of glutamate to the cells leads to a glutamate receptor-independent drop in glutathione levels, followed by cell death(Murphy et al. 1990; Ratan et al. 2002). The physiological relevance of this model relies not so much on the challenge by glutamate, in which very high (millimolar) concentrations of glutamate are employed, but in the subsequent loss of glutathione, which is characteristic of many acute and chronic brain disorders. Similar results are achieved when alternate forms of glutathione depletion are used, e.g. inhibition of glutathione synthetase with buthionine sulfoximine(Li et al. 1997). These studies employed the mouse hippocampus-derived neuronal cell line HT22 as well as immature primary neurons, and HT22 cells are now being used as a screening tool for novel neuroprotective reagents(Maher et al. 2007; van Leyen et al. 2008). Besides neurons, oligodendroglial and brain endothelial cells are also subject to 12/15-LOX mediated cytotoxicity(Wang et al. 2004; Jin et al. 2008). Thus inhibition of 12/15-LOX may protect a variety of cell types in the brain. The trigger for 12/15-LOX to become neurotoxic is the depletion of the intracellular antioxidant glutathione, and recently, glutathione peroxidase 4 (Gpx4) has been shown to regulate the cytotoxic effects of 12/15-LOX(Seiler et al. 2008). Beyond these findings however, the actual mechanism by which 12/15-LOX damages neural cells has remained unknown. Besides 12/15-LOX itself, we have also focused here on the proteasome, which we have previously shown to be involved in oxidative glutamate toxicity (van Leyen et al. 2005). That study had left unclear if proteasome activity is required at an early or late point in the cell death cascade.
The damaging effects of 12/15-LOX have often been attributed to its soluble metabolites, which include the oxidized polyunsaturated fatty acids 12- and 15-hydroxy-eicosatetraenoic acid (12- and 15-HETE) and 12- and 15-hydroperoxy-eicosatetraenoic acid (12- and 15-HPETE)(Loscalzo 2008). We instead hypothesized that a mechanism, Programmed Organelle Degradation, by which 12/15-LOX contributes to the physiological breakdown of mitochondria in maturing red blood cells (van Leyen et al. 1998), may be causing irreversible damage to neuronal cells. Since the enzyme has been shown through immunohistochemistry, metabolic profiling, knockout mouse studies, and work with pharmacological inhibitors to be relevant to several neurodegenerative diseases(Pratico et al. 2004; Khanna et al. 2005; Yao et al. 2005; van Leyen et al. 2006), it appeared crucial to elucidate the mechanism of action by which 12/15-LOX damages neural cells. We show here that 12/15-LOX amplifies oxidative stress by attacking mitochondria, leading to cytochrome c release and production of reactive oxygen species (ROS). Inhibition of 12/15-LOX provides robust protection against cell death through glutathione depletion in HT22 cells, even without restoring glutathione levels.
HT22 cells were cultured in DMEM containing 10% fetal bovine serum and penicillin / streptomycin (all media from Invitrogen, Carlsbad, CA), treated as indicated, and the percentage of lactate dehydrogenase (LDH) determined using a cytotoxicity detection kit (Roche, Indianapolis, IN) as described(van Leyen et al. 2005). To clarify which pathways are involved in oxidative glutamate toxicity, we used the following agents in this study: Baicalein (Cayman Chemicals, Ann Arbor, MI) inhibits both the platelet-type 12-lipoxygenase, as well as 12/15-LOX, but does not inhibit the leukotriene-generating 5-LOX. It is also a fairly strong antioxidant. AA-861 (Biomol, Plymouth Meeting, PA) inhibits both 12/15-LOX and 5-LOX, but does not have antioxidant activity. The specific proteasome inhibitor epoxomicin (Biomol, Plymouth Meeting, PA) has been shown to be protective in this model before (as have two other, structurally unrelated proteasome inhibitors), but without further examination of the mechanism involved (van Leyen et al, 2005). Indomethacin inhibits both cyclooxygenases 1 and 2. Primary neurons were prepared from mouse embryos at E16, either from ALOX15-/- mice(Sun and Funk 1996) obtained from Jackson laboratories and bred in our animal facility, or the corresponding wild-type C57Bl6J mice, and treated after 24 hours in culture(Ratan et al. 2002). These experiments were performed following an institutionally approved protocol in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. We have previously confirmed the knockout phenotype of these mice by immunohistochemistry with a 12/15-LOX specific antibody(van Leyen et al. 2006). Glutathione levels were determined as described in(Sebastia et al. 2003), briefly: HT22 cells seeded in 96-well plates were treated as indicated for 12 hours in a total volume of 100 μl, and incubated another 30 minutes in the dark after addition of monochlorobimane (mBCl; Sigma, St. Louis, MO) reagent to a final concentration of 40 μM. Fluorescence emission was measured at 460 nm, with excitation at 360 nm. Results were averaged from at least 3 independent experiments, as indicated in the figure legends. In all cases, results are given as mean, error bars represent SEM. Multi-group comparisons were carried out by One-way ANOVA followed by Tukey-Kramer HSD comparison.
To analyze breakdown of the mitochondrial membrane potential by FACS analysis, HT22 cells were incubated for 24 hours in the presence or absence of 5 mM glutamate and either 10 μM baicalein, 5 μM AA-861, or 50 nM epoxomicin. Cells were then trypsinized, washed twice with PBS, resuspended in 500 μl PBS and stained with the mitochondrial dye JC-1 (Molecular Probes/Invitrogen) according to the manufacturer’s protocol. As control for complete breakdown of the mitochondrial membrane potential, untreated cells were incubated for 5 minutes with 50 μM carbonyl cyanide 3-chlorophenylhydrazone (CCCP, Sigma, St. Louis, MO). Intact cells were gated by Forward / Side Scatter, 10000 cells / sample were counted, and red fluorescence (585 nm) was graphed versus green fluorescence (527 nm). Alternately, the cells were stained for 15 minutes at room temperature in the dark in the presence of 500 nM tetramethylrhodamine ethylester (TMRE, Invitrogen), washed once, and fluorescence was measured in the red channel. To measure the membrane potential of isolated mitochondria, a mitochondrial fraction isolated from HT22 cells as described(van Leyen et al. 2005) was incubated with or without purified 12/15-LOX (500 μg/ml)(van Leyen et al. 1998) for 2 hours at 37 °C and subjected to FACS analysis as described for the HT22 cells. To measure oxidative stress, HT22 cells, treated as described above, were incubated for 20 minutes at room temperature with 100 μM 2,7-Dichlorodihydrofluorescein diacetate (Cayman Chemicals, Ann Arbor, MI), washed twice with PBS and resuspended in 500 μl PBS prior to measuring fluorescence at 527 nm.
Cytosolic and mitochondrial fractions were isolated from HT22 cells as described(van Leyen et al. 2005). Cytosolic fractions of HT22 cells treated as indicated, representing 20 μg protein per lane, were separated on 4-12% NuPAGE gels (Invitrogen), blotted to nitrocellulose membranes and probed with an antibody to cytochrome c (Chemicon/Millipore, Billerica, MA). In our hands, cytochrome c migrated as a tetramer at an apparent molecular weight of 56 kDa in this gel system. It was also detected at this apparent molecular weight using two other commercially available antibodies to cytochrome c (data not shown). An antibody to β-actin (Sigma, St. Louis, MO) was used as protein loading control. To determine cytochrome c release from HT22 mitochondria, the mitochondrial fraction from untreated cells was incubated with or without 12/15-LOX (500 μg/ml) for 2 hours at 37 °C and re-isolated by centrifugation. Mitochondrial pellet and supernatant fractions were then separated on 4-12% NuPAGE gels and probed for cytochrome c as described above.
12/15-LOX, mitochondria, and ER membranes were isolated as described(van Leyen et al. 1998). 12/15-LOX (500 μg/ml) was incubated with organelles (1 mg/ml) at 37°C for 2.5 hours in a buffer containing 250 mM sucrose, 50 mM HEPES/KOH pH 7.5, 50 mM KCl and 5 mM MgCl2 in a final volume of 20 μl. DCF reagent was prepared according to (Keller et al. 2004): Fifteen μl 2,7-Dichlorodihydrofluorescein diacetate were hydrolysed in the dark with 60 μl 0.01 N NaOH (oxygen free) for 30 minutes at 37°C, and the reaction was stopped by adding 300 μl 25 mM potassium phosphate pH 7.5 (oxygen free), resulting in a final Dihydro-DCF concentration of 0.2 mM. Twenty μl of LOX / organelle solutions were incubated in the dark with 25 μl DCF reagent and 55 μl buffer for 20 minutes at room temperature and read with an excitation wavelength of 485 nm and an emission wavelength of 530 nm. Results were averaged from 3 independent experiments.
To analyze LOX-dependent cell death, we subjected mouse neuronal HT22 cells to glutamate, which leads to blockage of the Xc− antiporter in the plasma membrane and a subsequent decline of glutathione levels(Murphy et al. 1990; Ratan et al. 2002). After 24 hours of glutamate exposure, most cells were dead, as indicated by release of intracellular lactate dehydrogenase (LDH). The LOX inhibitors baicalein and AA-861 prevented cell death, as did the proteasome inhibitor epoxomicin (Figure 1a). Besides lipoxygenases, the second major class of arachidonic acid-metabolizing enzymes are the cyclooxygenases COX-1 and COX-2, known as mediators of excitotoxicity. To rule out cyclooxygenase involvement in oxidative glutamate toxicity, we used indomethacin, an inhibitor of both COX-1 and COX-2. Indomethacin was not protective, indicating that cyclooxygenases are not required for cell death in this model.
To determine whether the 12/15-LOX isoform is responsible for oxidative stress-related neuronal cell death, we treated neurons derived from 12/15-LOX knockout mice (ALOX15-/-)(Sun and Funk 1996) or the corresponding wild-type neurons with glutamate after one day in vitro (DIV). Primary neurons at this stage are also subject to oxidative glutamate toxicity, rather than excitotoxicity(Murphy et al. 1990; Ratan et al. 2002). The 12/15-LOX knockout neurons were significantly protected against glutamate challenge, suggesting that 12/15-LOX is the major LOX isoform responsible for neuronal cell death in this model (Figure 1b) and confirming earlier reports of 12/15-LOX involvement in oxidative glutamate toxicity(Li et al. 1997; Khanna et al. 2003).
One way of preventing cell death in this model would be to restore glutathione levels. Indeed, increasing the amount of the anti-apoptotic protein BCL2, which protects against oxidative glutamate toxicity, has been reported to increase the level of glutathione, so that oxidative stress does not occur(Sahin et al. 2006). To show that this is not the protective mechanism here, we measured glutathione levels in HT22 cells. Following 8 hours exposure to 5 mM glutamate, glutathione levels were reduced to about 10% of the initial value (Figure 1c). Importantly, neither baicalein nor epoxomicin restored the glutathione levels, suggesting their protective effect must lie downstream of glutathione depletion.
HT22 cells experience oxidative stress after prolonged exposure to glutamate, leading to an increased fluorescent signal for the oxidative stress marker 2,7-Dichlorofluorescein (DCF; Figure 1d). Surprisingly, although LOX inhibition could not restore glutathione levels, both baicalein and AA-861 were able to reduce oxidative stress (Figure 1e and 1f, respectively). In the case of baicalein, oxidative stress could also be reduced through the antioxidant activity of baicalein, rather than through lipoxygenase inhibition. Importantly, the LOX inhibitor AA-861, which is not an antioxidant(Wang et al. 2004), also decreases DCF fluorescence (Figure 1f), demonstrating that 12/15-LOX activity contributes to the generation of oxidative stress. In contrast, inhibition of the proteasome, while protecting against cell death (Figure 1a), did not prevent the oxidative stress associated with glutamate treatment (Figure 1g). Therefore, the protective effects of LOX inhibition appear to be more complete than those of proteasome inhibition.
Our previous studies had suggested mitochondria might be involved in this cell death pathway(van Leyen et al. 2005). Using the dye JC-1 to measure the integrity of the mitochondrial membrane potential (MMP) and the uncoupler of oxidative phosphorylation CCCP as a positive control, we found a significant shift to increased green fluorescence as indicator of a compromised MMP (Figure 2a-b). A similar shift was detected when exogenous glutamate was applied (Figure 2c), confirming a breakdown of the mitochondrial membrane potential following glutamate treatment. The membrane potential was restored by co-incubation with the two LOX inhibitors (Figures 2d-e). In contrast, the proteasome inhibitor epoxomicin was not able to restore the MMP (Figure 2f-g). To confirm this result, we employed another widely-used membrane potential-dependent dye, TMRE. Here, loss of the mitochondrial membrane potential leads to a loss of fluorescence. Similar to the results seen with JC-1, baicalein was able to prevent the glutamate-induced loss of membrane potential, (Figure 3a-b), while epoxomicin was not (Figure 3c). To determine the consequences of this LOX-dependent loss of membrane potential, we studied release of the mitochondrial protein cytochrome c into the cytosol. Indeed, cytochrome c was increased in the cytosolic fraction following glutamate treatment, and this was reduced by LOX inhibition (Figure 3e). Again, proteasome inhibition was not sufficient to prevent cytochrome c release. Taken together, these results suggest that LOX inhibition protects against mitochondrial damage, while proteasome inhibition cannot protect the mitochondria.
The cell culture experiments demonstrated that LOX activity in HT22 cells leads to mitochondrial damage, but left open the question of whether the effect is direct or indirect. Because LOX inhibition was able to prevent mitochondrial damage in glutamate-treated HT22 cells, we hypothesized that 12/15-LOX might directly damage mitochondria. This would be in keeping with the enzyme’s function in maturing red blood cells, where 12/15-LOX contributes to removal of mitochondria(Rapoport and Schewe 1986; Grüllich et al. 2001). Following this rationale, we isolated mitochondria from healthy HT22 cells and incubated them with 12/15-LOX in vitro. FACS analysis of the mitochondria showed that, similar to the positive control CCCP, incubation of the mitochondria-enriched fraction of HT22 cells with purified 12/15-LOX led to a disruption of the MMP (Figure 4a-c). Likewise, cytochrome c was released from mitochondria incubated with 12/15-LOX (Figure 4d). Therefore, 12/15-LOX can directly lead to an increased permeability of mitochondria, along with cytochrome c release.
Reactive Oxygen Species (ROS) can be generated via several different pathways in cells subjected to oxidative stress, including lipid peroxidation and superoxide generation(Chan 2005). Given the apparent requirement for LOX activity to generate ROS in HT22 cells and its ability to directly damage mitochondria, we investigated whether or not 12/15-LOX itself could be the source of intracellular ROS. We therefore incubated isolated mitochondria in the presence or absence of 12/15-LOX and measured the generation of ROS as the formation of DCF. We detected a robust increase in DCF fluorescence when both mitochondria and 12/15-LOX were present (Figure 4e). In contrast, incubation of 12/15-LOX with ER membranes gave only a slight increase in ROS, suggesting that the interaction of 12/15-LOX with mitochondria may be the major source of ROS in HT22 cells.
It has been well established that 12/15-LOX has a role in neuronal cell death involving oxidative stress, but its mechanism of action remained unclear. We have shown here that in neuronal HT22 cells 12/15-LOX attacks the cells’ mitochondria, and this is the committed step of the oxidative glutamate toxicity pathway. While soluble LOX metabolites, such as HETEs and HPETEs, could also contribute to the cell death process, they may not be absolutely required (see Supplementary Figure S1). Instead, a direct attack of 12/15-LOX on mitochondria leads not just to the release of pro-apoptotic molecules like cytochrome c, but also generates ROS, an important characteristic of oxidative glutamate toxicity and many other forms of oxidative stress. Thus, the low level of oxidative stress attained through the loss of glutathione is amplified by the intracellular attack of 12/15-LOX on the cells’ mitochondria. One caveat to these findings concerns the concentration of 12/15-LOX needed to permeabilize mitochondria. In previous studies with purified 12/15-LOX we have found 125 μg/ml to be sufficient to permeabilize ER vesicles (van Leyen et al. 1998). We have here used a higher concentration, 500 μg/ml, to achieve clear results, and we do not know if these concentrations of 12/15-LOX are reached in HT22 cells or primary neurons undergoing oxidative glutamate toxicity. However, the purified enzyme at low protein concentrations, in striking contrast to the high protein density inside the cell, is known to undergo rapid inactivation, drastically limiting its activity in vitro (Rapoport and Schewe 1986). In the intracellular milieu, far lower concentrations of 12/15-LOX may be sufficient to cause mitochondrial damage. This damage to mitochondria is reminiscent of the role 12/15-LOX plays in red blood cell precursors, where mitochondria are eliminated as part of the physiological process of maturation(Rapoport and Schewe 1986; van Leyen et al. 1998; Grüllich et al. 2001). In many ways, these findings parallel the continuum mode of cell death that has been observed in various forms of neurotoxicity, where phenotypically, a mixture between apoptotic and necrotic features is observed(Ankarcrona et al. 1995; Portera-Cailliau et al. 1997).
The current study suggests the following picture of 12/15-LOX mediated neuronal cell death. When the intracellular pool of glutathione is depleted in HT22 cells following exposure to extracellular glutamate, the normally cytosolic 12/15-LOX attacks mitochondria, which release cytochrome c. At the same time, ROS are formed through the interaction of 12/15-LOX with mitochondria, leading to further cell damage. Blocking LOX activity may be sufficient to keep neuronal cells healthy, even if their glutathione levels cannot be directly restored. The quality of protection through 12/15-LOX inhibition is illustrated by the robust proliferation seen in HT22 cells treated with both glutamate and LOX inhibitor (see Supplementary Figure S2).
There are several limitations to our study. One, we do not know if signaling events in addition to glutathione depletion are required to activate 12/15-LOX. This appears very possible, as inhibiting the phosphorylation of extracellular signal-regulated kinase (ERK) also protects HT22 cells against oxidative glutamate toxicity(Stanciu et al. 2000); moreover, 12/15-LOX itself has been reported to be tyrosine phosphorylated by the kinase c-src(Khanna et al. 2005). Two, although this study demonstrates that the attack of 12/15-LOX on mitochondria is most likely the centerpiece of cell death through glutathione depletion, additional factors must contribute to the cells’ demise. The proteasome is clearly involved at a late stage, because proteasome inhibition can reduce cell death, but cannot restore mitochondrial integrity or cell proliferation (Supplementary Figure S2). Although cytochrome c is released to the cytosol, it may not have a direct role in the cell death process, because our previous studies have demonstrated that caspase-3 is not activated(van Leyen et al. 2005). In contrast, apoptosis-inducing factor AIF may be involved (manuscript in preparation). Finally, to investigate this 12/15-LOX dependent cell death mechanism we have employed a cell line derived from the mouse hippocampus. Clearly, rapid proliferation is a feature that sets HT22 cells apart from the neurons they are derived from. And the glutamate concentration of 5 mM is high. So is this a cell death pathway that is specific for HT22 cells, or does it have more general relevance for neuronal cell death? There are several indications that this may be the case. Firstly, immature primary neurons are subject to the same type of oxidative glutamate toxicity(Murphy et al. 1990; Li et al. 1997; Ratan et al. 2002; Canals et al. 2003; de Bernardo et al. 2004; van Leyen et al. 2005), and we show here that 12/15-LOX knockout neurons are less susceptible (see Figure 1b). Secondly, glutamate concentrations (1 mM for HT22 cells, around 500 μM for immature primary neurons) that are more typical for acute cerebrovascular injury also lead to significant cell death at 24 hours(Sagara et al. 2002). We here chose the 5 mM glutamate concentration to maximize cell injury, and thus to be able to clearly detect protective effects of LOX inhibition. Thirdly, besides glutamate, other methods to lower the intracellular glutathione levels, e.g. buthionine sulfoximine or homocysteate administration, cause a very similar type of cell death(Murphy et al. 1990; Li et al. 1997). In this sense, we are essentially investigating a glutathione depletion model. This suggests that diseases in which a drop in glutathione levels leads to increased oxidative stress, including stroke, Parkinson’s, or Alzheimer’s, may initiate a LOX-mediated form of neuronal cell death.
Intriguingly, besides the damaging effects of 12/15-LOX documented here, several metabolites of 12/15-LOX including neuroprotectin D1 and the resolvins, have neuroprotective and anti-inflammatory qualities(Marcheselli et al. 2003; Bazan 2008). Extensive animal studies, some of which have already commenced(Khanna et al. 2005; van Leyen et al. 2006; Lapchak et al. 2007; Jin et al. 2008), will be needed to determine the long-term benefit of targeting 12/15-LOX for treatment of stroke and other neurodegenerative diseases. The results shown here should provide a solid mechanistic rationale for the utility of 12/15-LOX inhibition to prevent mitochondrial damage in oxidative stress-related brain injury. In summary, we have demonstrated here that 12/15-LOX can directly damage mitochondria in oxidatively stressed HT22 cells, leading to a novel form of cell death in which 12/15-LOX is the central executioner.
Support through grants from the NIH (R01NS049430 to K.v.L., R01NS53560 and P01NS555104 to E.H.L.), and a Scientist Development Grant from the American Heart Association (to K.v.L.), is gratefully acknowledged. The authors declare no conflict of interest.