The study presented here contains a number of novel findings. First, we show that drugs that up-regulate autophagy and have been proposed for use in human neurodegenerative disease do not cause detectable increases in ROS production. However, the up-regulation of autophagy by these drugs, and, indeed basal autophagy, are still inhibited by antioxidant drugs, suggesting that some level of ROS production or redox signalling is required for the normal regulation of autophagy. We show that antioxidants have effects on different classic autophagy regulatory pathways, consistent with their ability to inhibit autophagy. Vitamin E has an effect on the mTOR pathway, while thiol-containing antioxidants, such as NAC, decrease the phosphorylation of JNK and Bcl-2. Given the central role of superoxide in autophagy and our observation that menadione has opposite effects on the phosphorylation of JNK and Bcl-2 to NAC, it is possible that NAC exerts its effects by acting downstream of superoxide. However, as the rate constant of thiol groups for the reaction with superoxide is significantly less than that of the abundant SOD, it seems unlikely that superoxide is being directly scavenged. Alternatively, the increased cell thiol content may be blocking or reversing redox changes that occur downstream of superoxide production. We also show in vivo pathological and physiological relevance for these findings including, what is to our knowledge, the first ever vertebrate model of polyglutamine disease which allows quantification of changes in autophagic flux in conjunction with levels of pathological autophagic substrates.
We believe that the data presented here may have significant application to the design of treatment for human neurodegenerative disease. The idea that oxidative stress is pathological in the context of these diseases, and that decreasing oxidative stress is therefore a likely therapeutic strategy, has become a paradigm in the field. Our data suggest that low levels of some forms of ROS or redox signals are essential for autophagy. At least in diseases caused by aggregate-prone proteins which are autophagy substrates, the potential benefits of ROS scavenging may therefore be counterbalanced by an increased load of toxic protein. This is of significance, as supplements with putative antioxidant properties are commonly taken in the general population and even more frequently in those with neurodegenerative disease. In a previously published study from our group, we found that over two-thirds of HD patients were routinely taking supplements with putative antioxidant activity, a similar figure to other published series of PD patients (37
). Furthermore, previously published series also suggest that many of these patients are unaware of any possible side effects of these supplements and do not tell their treating physician they are taking them (38
). This could be a potentially significant confounder in any trial of autophagy-enhancing drugs and indeed it is conceivable that long-term inhibition of autophagy with these agents may even exacerbate the disease course. The only clinical trial of vitamin E in HD published to date showed non-significant worsening of all primary outcome measures, while trials of vitamin E in PD have similarly failed to show striking benefit or harm (23
We do not suggest that treatments aimed at ameliorating oxidative stress may not have their place as therapy, but rather that better understanding of their interaction with other cellular processes may allow for more appropriately designed (and therefore more effective) treatment. For example, since the effects of antioxidants on autophagy appear dose-dependent, it is possible that dose may be crucial in balancing the effects of these drugs, and that higher doses may not necessarily mean greater efficacy. This may be of particular relevance as higher doses of some antioxidants have been proposed for use in HD (3
). Additionally, the mechanisms of autophagy regulation by NAC and vitamin E appear distinct (as one might expect from such different antioxidant compounds), indicating that the interaction with autophagy may vary depending on the nature of the antioxidant used. Similarly, it is possible that antioxidant treatment at a time point in the disease when autophagic defence responses have become overwhelmed, or targeting specific cellular compartments, may be more efficacious. Understanding the interaction between various agents that may modify the progression of neurodegenerative diseases has significant current relevance, as clinical trials of both autophagy-enhancing drugs (39
) and antioxidant therapies (40
) have been performed and combination treatments may appear an attractive therapeutic strategy. Our data suggest that some combinations that appear attractive on the basis of current knowledge (such as rapamycin and cystamine) may not be beneficial.
The demonstration that NAC inhibits the up-regulation of autophagy in vivo
may have clinical utility outside of neurodegeneration. NAC has a long safety record in the treatment of acetaminophen poisoning in humans where it is used intravenously at high doses, resulting in serum concentrations comparable to those used in cell and mouse models described here (41
). Some malignant tumours up-regulate autophagy as a way of surviving chemotherapy/radiotherapy and to provide nutrient recycling in the avascular tumour core. Autophagy inhibitors may be potential cancer therapies in this context (42
). NAC may be particularly attractive in this regard as our data suggest it has the unusual combination of properties of inhibiting both mTOR (which may inhibit tumour growth) and autophagy. Thus, while long-term inhibition of autophagy is likely to be deleterious, it may well be possible to use potent, reversible autophagy inhibitors safely for short periods. For these applications, NAC may be an attractive candidate therapy.
The apparently intimate relationship between ROS and autophagy is just one example of how our appreciation of ROS has moved beyond regarding them simply as a dangerous and unwanted by-product of metabolism. Enhanced understanding of the interaction between these cellular processes is likely to have application to the treatment of human diseases.