Acute neurological injury, such as that arising from stroke and trauma, is a major cause of morbidity and mortality worldwide (Kraus 1996; Taylor et al. 1996; Matchar et al. 1997; Harkey et al. 2003). Although the precise mechanisms leading to neuronal damage in these conditions are likely to differ, mitochondrial dysfunction is emerging as a major common pathway on the road to cell death. It is generally accepted that, following brain ischemia and trauma, large quantities of the excitatory amino acid (EAA) neurotransmitter glutamate are released into the synapse, activating both NMDA and non-NMDA receptors (Zipfel et al. 2000). Pathological stimulation of NMDA receptors in particular leads to an increased flux of Ca2+ into the cytoplasm (Zipfel et al. 2000) and the activation of numerous lipases and phosphatases. In an apparent attempt to prevent these potentially lethal events, Ca2+ is sequestered by mitochondria, which serve as high capacity ‘sinks’ (Budd and Nicholls 1996; Khodorov et al. 1996; White and Reynolds 1997; Peng and Greenamyre 1998). Perhaps as a consequence of Ca2+ uptake, stimulation of NMDA receptors also results in the formation of reactive oxygen species (ROS) by the mitochondria (Dugan et al. 1995; Reynolds and Hastings 1995). The accumulation of high levels of Ca2+ and formation of ROS that overwhelm antioxidant defenses (i.e. oxidative stress) can destabilize mitochondria (Dykens 1997), which, depending on the severity, can lead to either apoptotic or necrotic cell death (Ankarcrona et al. 1995). Thus, interventions directed at maintaining mitochondrial homeostasis may provide a novel means by which to attenuate acute central nervous system injury.
Mitochondrial dysfunction, resulting from the disruption of calcium homeostasis and the generation of toxic reactive oxygen species, is a central process leading to neuronal injury and death following acute CNS insults. Interventions aimed at preventing disturbances in mitochondrial function have therefore become targets of intense investigation. Mitochondrial uncoupling is a condition in which electron transport is disconnected from the production of ATP. As a consequence, there is a decrease in the mitochondrial membrane potential, which can temporarily decrease calcium influx and attenuate free radical formation. The potential use of pharmacological agents with uncoupling properties may provide a novel therapeutic approach for the treatment of acute neuronal injury.
brain mitochondrial carrier protein-1
excitatory amino acid
reactive oxygen species
Mitochondria are responsible for > 90% cellular oxygen consumption and the vast preponderance of ATP production (Nicholls and Ferguson 1992). During electron transport, energy is extracted by mitochondrial complexes I, III and IV and used to transfer protons out of the matrix into the intermembrane space, thus establishing a proton gradient. The predominant result of this gradient is the establishment of the mitochondrial membrane potential (ΔΨm), which provides the driving force for the influx of protons through the channel of ATP synthase resulting in the formation of ATP (Mitchell and Moyle 1967). The term uncoupling refers to a condition in which a ‘leak’ of protons back into the matrix bypasses ATP synthase and electron transport becomes functionally disconnected from the phosphorylation of ADP to ATP. As a consequence of this short-circuit, there is a reduction in ΔΨm that may have important ramifications under certain conditions.
Over the years, a number of pharmacological agents that facilitate the flow of protons back into the mitochondrial matrix (i.e. protonophores) have been identified (Skulachev 1998). Probably the most thoroughly investigated uncoupling agent is 2,4-dinitrophenol (DNP) (Loomis 1948). Initially used in the manufacturing of munitions during World War I, clinical studies in the early 1930s revealed DNP to be a highly effective weight loss agent (Tainter et al. 1935). At least 100 000 persons were estimated to have taken DNP for weight loss. Unfortunately, chronic ingestion of DNP was associated with a number of untoward effects. Most frequently observed were skin rashes and cataracts (Boardman 1935; Horner et al. 1935; Tainter et al. 1935), however, rare cases of hepatic dysfunction, blood dyscrasias, cardiac arrhythmias and gastroenteritis were also reported (Sidel 1934; Tainter et al. 1934; Simkins 1937). The nascent Food and Drug Administration removed DNP from the market in 1938.
Actions of mitochondrial uncouplers on neuronal survival
While complete and/or prolonged uncoupling is typically deleterious to normal mitochondrial and ultimately cellular function, there is accumulating evidence that transient or partial uncoupling can protect against EAA-induced (i.e. excitotoxic) insults. In cell culture, for instance, exposure of forebrain neurons to the uncoupling agent FCCP [carbonyl cyanide p-(trifluoromethoxy)phenyl-hydrazone] during a 10-min incubation with glutamate resulted in a greater than 50% reduction of cell death at 24 h (Stout et al. 1998). In this study, the concentration of FCCP employed was estimated to cause a 60% reduction in ΔΨm, and while a 10 min exposure to this amount of FCCP alone resulted in no significant cell loss, a 24-h exposure proved toxic. These findings are consistent with the notion that long-term inhibition of mitochondrial function may have dire consequences. More importantly, they suggest that even a relatively large reduction in ΔΨm, if transient, can be tolerated. Due to the non-linear response of fluorescent dyes, it is possible that the degree of depolarization reported in this study may not accurately reflect the actual magnitude of change induced by FCCP. In a more recent study, partial uncoupling of cortical neurons with DNP reduced neuronal death induced by oxygen and glucose deprivation, in an in vitro model of cerebral ischemia (Mattiasson et al. 2003), lending further credence to the notion that uncoupling may be beneficial. Not all studies, however, have demonstrated neuroprotection against excitotoxicity using uncouplers (Dugan et al. 1995; Sengpiel et al. 1998), possibly resulting from differences in the cell lines and concentrations of uncouplers used.
Additional evidence that mitochondrial uncoupling can be neuroprotective comes from in vivo studies. For example, it has recently been shown that treatment of animals with systemically administered DNP up to 3 h after intraparenchymal administration of the NMDA agonist quinolinic acid attenuated the average striatal lesion volume by 25% compared to vehicle-treated control animals (Maragos et al. 2003). Although hyperthermia might have been expected to occur following treatment with DNP, the dose used in this study did not significantly affect the core temperature compared to vehicle-treated animals, consistent with only partial uncoupling of the mitochondria. Animals treated with the DNP analogue 2,4,6-trinitrophenol, which uncouples submitochondrial particles but is unable to penetrate intact mitochondria (and, hence, is unable to uncouple them) (Hanstein and Hatefi 1974), conferred no neuroprotection suggesting that the beneficial effect of DNP was likely related to its ability to uncouple mitochondrial respiration. Studies have also demonstrated that rats administered DNP 1 h after reperfusion developed 40% smaller infarct volumes following 2 h of focal cerebral ischemia (Korde et al. unpublished observations). As with QA-injected animals, administration of DNP to animals rendered ischemic had no effect on core temperature or any other physiological parameters compared to vehicle-treated animals. Lastly, ∼50% reduction in tissue loss was seen in the cortex of animals treated with either DNP or FCCP 5 min after traumatic brain injury (Sullivan et al. in press) and 30% increase in tissue sparing in the spinal cord of animals treated with DNP 5 min prior to injury (Jin et al. 2001). Although the observation that two structurally distinct uncoupling agents were effective suggests that uncoupling is indeed occurring in vivo, currently available methods do not allow the precise determination of the degree to which mitochondria are depolarized in the live animal.
Investigations of endogenous ‘uncoupling proteins’ (UCPs) have also provided support for a beneficial role of mitochondrial uncoupling. UCPs form a family of mitochondrial anion transporters that may be responsible for basal proton leak, which in the rat accounts for 10–27% of resting oxygen consumption, depending on the organ studied (Rolfe and Brand 1997). First described in brown adipose tissue, UCP-1 is responsible for heat generation in the newborn and may be involved in the normal response to cold stress in the adult (Bouillaud et al. 1985; Nicholls 2001). UCP-3 is found exclusively in muscle (Boss et al. 1997; Vidal-Puig et al. 1997). UCP-2, which is distributed in numerous organs, has been identified in the nervous system (Fleury et al. 1997), as have two brain-specific proteins, UCP-4 (Mao et al. 1999; Yu et al. 2000) and brain mitochondrial carrier protein-1 (BMCP-1) (Sanchis et al. 1998; Yu et al. 2000; Kim-Han et al. 2001). In brain, regions expressing high levels of UCP-2 show decreased energy coupling efficiency (reflected as a decrease in the respiratory control ratio) and a reduction in ΔΨm (Diano et al. 2003) while neurons expressing UCP-4 and BMCP-1 showed a reduction in ΔΨm and an increase in state 4 respiration (Mao et al. 1999; Kim-Han et al. 2001), all consistent with an uncoupling function. In one of these studies, exposure of mitochondria overexpressing UCP-5 to linoleic acid further enhanced state 4 respiration while addition of bovine serum albumin, which sequesters free fatty acids, prevented this enhancement (Kim-Han et al. 2001). These findings support the observations of Nicholls who first described activation of UCP-1 by free fatty acids, albeit at much lower concentrations (Nicholls and Locke 1984) and are in line with the observations that neonatal rats fed fat-rich diets showed an increase in UCP-2 expression and a reduction in kainic acid seizure-induced hippocampal damage (Sullivan et al. 2000). Overexpression of UCP-2 in mice is also neuroprotective against injury resulting from cerebral ischemia and traumatic brain injury (Mattiasson et al. 2003) as well as against pilocarpine-seizure-induced cell death (Diano et al. 2003). Thus, there are compelling data from both pharmacological and molecular biological studies indicating a potential neuroprotective role for mitochondrial uncoupling in models of acute injury.
Possible neuroprotective mechanisms of mitochondrial uncoupling
Several mechanisms might contribute to the neuroprotective efficacy of mitochondrial uncoupling. Uncoupling agents might diminish the ability of mitochondria to buffer cytoplasmic Ca2+. Although this might seem paradoxical at first, high levels of mitochondrial Ca2+ can destabilize mitochondrial function via activation of the permeability transition (Ankarcrona et al. 1995; White and Reynolds 1996; Dubinsky and Levi 1998; Kristian et al. 2000; Kristian et al. 2002) and the increase in mitochondrial-generated ROS formation (Dykens 1994; Sousa et al. 2003; Starkov and Fiskum 2003), both which can promote cell death. As ΔΨm is the driving force used to promote cytoplasmic Ca2+ entry into mitochondria via the calcium uniporter (Nicholls and Ferguson 1992; Gunter et al. 1994; White and Reynolds 1997), treatment with uncoupling agents, by reducing ΔΨm, would be predicted to decrease mitochondrial Ca2+ uptake and prevent cell death. In support of this, our laboratory has revealed that increases in mitochondrial Ca2+, as a consequence of quinolinic acid-induced NMDA receptor activation, are reduced by almost 40% in animals treated with DNP compared to controls (Korde et al. 2003). We have also shown, in a model of traumatic brain injury, that the ability of mitochondria to sequester Ca2+ was reduced by ∼ 30% in animals treated with DNP (Sullivan et al. in press). Additionally, cortical neurons transiently and partially depolarized with FCCP (Stout et al. 1998) and cerebellar granule cells in which ΔΨm was abolished (Khodorov et al. 2002) maintained higher levels of cytoplasmic Ca2+ following exposure to glutamate compared to non-depolarized cells. Although other mechanisms have been postulated to account for enhanced cytoplasmic Ca2+ during uncoupling (Castilho et al. 1998; Stout et al. 1998) the results of these studies indicate that high levels of cytoplasmic Ca2+ may be better tolerated than previously thought and that there exists a delicate balance in the cycling of cytoplasmic and mitochondrial Ca2+.
In addition to preventing mitochondrial Ca2+ levels from achieving potentially lethal levels, uncouplers such as DNP may also lead to a reduction in the mitochondrial production of ROS which, in addition to activating the permeability transition pore (Dykens 1997), may also enhance the release of ‘store-operated Ca2+ entry’ into the cytoplasm (Suzuki et al. 2003) and in so doing, create a vicious cycle. Several groups have recently demonstrated that the formation of ROS in different cell types, including neurons, is dependent on ΔΨm and increases logarithmically above ∼140 mV (i.e. the state 3→4 transition) when fueled with substrates for either complex II (Korshunov et al. 1997; Liu 1997) or complex I (Starkov and Fiskum 2003), the principle sites of electron entry into the transport chain. Thus, only small reductions in ΔΨm may be necessary to bring about a profound reduction in ROS formation (Votyakova and Reynolds 2001; Starkov and Fiskum 2003). Indeed, we have recently shown that the administration of DNP attenuates indices of ROS formation following quinolinic acid injections into the striatum (Korde et al. 2003) and in experimental models of cerebral ischemia (Korde et al. unpublished observations), traumatic brain (Sullivan et al. in press) and spinal cord injuries (Jin et al. 2001). In contrast, and providing further support for this relationship between ΔΨm and ROS formation, inhibition of UCP-2 with GDP resulted in a rise of ΔΨm and a parallel increase in hydrogen peroxide production (Negre-Salvayre et al. 1997).
Several factors may be responsible for the association between ROS formation and ΔΨm. By altering the equilibrium of protons across the inner mitochondrial membrane, DNP may affect the redox state of the mitochondria (Skulachev 1996; Liu 1997; Starkov and Fiskum 2003). This could shift the normally long-lived, reduced electron-carrying intermediates, such as cytochrome b1 and the pyridine NAD(P)H which transfer electrons to molecular oxygen, to the oxidized form. Due to the diminished longevity and enhanced stability of these oxidized intermediates, they are less likely than their reduced counterparts to donate electrons to oxygen and so would be less likely to generate superoxide. Uncouplers, which stimulate oxygen consumption, may also decrease mitochondrial ROS formation by reducing concentrations of molecular oxygen, thereby limiting its availability to act as an electron acceptor (Skulachev 1996). Lastly, as the generation of ROS has been linked to mitochondrial Ca2+ influx (Dykens 1994; Sousa et al. 2003; Starkov and Fiskum 2003), attenuation of ROS formation may be a direct consequence of decreased Ca2+ uptake resulting from a reduction in ΔΨm caused by the uncoupling agent.
There are mounting data that the disruption of mitochondrial homeostasis represents a common pathway in cell death resulting from acute CNS insults involving excitotoxic mechanisms. Increases in mitochondrial calcium and ROS formation are two events that are intimately related and believed to activate both necrotic and apoptotic cascades. The concept of ‘mild uncoupling’ has recently been postulated to favorably alter the mitochondrial milieu in a way that provides protection when the organelle is placed under excessive stress. Evidence from both in vitro and in vivo studies indicates that transient exposure to pharmacological uncouplers is neuroprotective against excitotoxic cell death. Importantly, current data indicate that pharmacological uncoupling is efficacious, even when administered after the insult; an attractive property if agents with uncoupling properties were to be useful in the clinical setting.
As mentioned above, long-term exposure to DNP may have harmful side-effects. Whether the toxic effects of DNP are due to the chemical nature of the agent itself or to its uncoupling properties is not known. Although DNP was removed from the market over six decades ago, a critical reappraisal of the toxicological profile of this compound following a single dose might be warranted. In addition to the side-effects associated with chronic ingestion, DNP has a relatively steep dose–toxicity profile (Harper et al. 2001), with potentially lethal pyrexia occurring at doses only two to three times that used for weight loss and in our small animal experiments. As prolonged and large reductions in neuronal ATP would be predictably lethal to neurons, we believe that agents which act transiently and only mildly uncouple mitochondria will be required to provide safe and optimum neuroprotection in acute CNS injury. It is interesting to note that in mice undergoing chronic alternate day dietary restriction, there were increased levels of the endogenous uncoupling protein UCP-4 in both the cerebral cortex and hippocampus (Mattson and Liu 2003). As dietary restriction has been shown to be neuroprotective in animals models of Alzheimer's and Parkinson's diseases (Mattson 2003), this suggests that pharmacological uncoupling may have a molecular counterpart and that safe, non-toxic agents with mild uncoupling properties might also provide neuroprotection in chronic neurodegenerative disorders.
Although our current understanding of the mechanisms by which uncoupling confers neuroprotection may involve improved mitochondrial calcium homeostasis and decreased oxidative stress, other potential mechanisms could be involved. For instance, does uncoupling increase antioxidant defense systems or by altering the mitochondria redox potential, inhibit apoptotic cascades? Alternatively, could uncoupling transiently decrease ATP levels and, in so doing, prevent the activation of ATP-dependent apoptotic pathways? We do not believe this to be the case, as administration of the complex II inhibitor, malonate, which depletes cellular ATP stores, augments rather than attenuates NMDA receptor-mediated striatal injury (Greene and Greenamyre 1995). Conversely, could uncoupling activate endogenous anti-apoptotic pathways? Lastly, could treatment with uncouplers, by altering ROS formation, affect signal transcription pathways? A critical and detailed examination of the molecular events associated with mitochondrial uncoupling should reveal new insights into the regulation of both ‘cell death’ and ‘cell survival’ cascades.
The authors thank Daret St. Clair and John T. Slevin for critically reviewing the manuscript. This work was supported by NIH grants NS01941 and NS42111 to WFM.