The pathogenesis of cell death in Parkinson's disease. Peter Jenner, PhD and C. Warren Olanow, MD, FRCPC

Transcription

1 The pathogenesis of cell death in Parkinson's disease Peter Jenner, PhD and C. Warren Olanow, MD, FRCPC From the Neurodegenerative Diseases Research Centre, School of Health and Biomedical Sciences, King's College, London, United Kingdom (Dr. Jenner), and the Departments of Neurology and Neuroscience, Mount Sinai School of Medicine, New York, New York (Dr. Olanow). Address correspondence and reprint requests to Dr. Peter Jenner, NDRC, School of Health and Biomedical Sciences, King's College, London SE1 1UL, UK; Abstract. Concepts of pathogenesis in Parkinson's disease (PD) have been based on attempts to understand the mechanisms responsible for nigral dopaminergic cell death. Pathogenesis has been proposed to involve oxidative and nitrative stress, excitotoxicity, inflammation, mitochondrial dysfunction, and altered proteolysis. These processes are considered to form a complex cascade of interrelated events that lead to neuron death by way of apoptosis. However, current views on pathogenic mechanisms in PD may not be as exact as commonly proposed. Future concepts of pathogenesis in PD need to incorporate events leading to the destruction of non-dopaminergic nuclei and to distinguish between primary factors that are responsible for disease initiation and secondary factors that contribute to disease progression. Importantly, there is a need to determine whether PD is a single illness with a common pathogenesis or a group of related illnesses with different pathogenic mechanisms. This is an essential step to understanding pathogenesis and is critical to the development of comprehensive neuroprotective approaches to treatment. The pathology and biochemistry of Parkinson's disease (PD) have been extensively investigated. It is accepted that the brunt of the pathologic change affects the dopaminergic cells of the zona compacta of the substantia nigra (SNc), with a consequent loss of striatal dopamine. 1 SNc degeneration is believed to underlie the major motor features of PD, a concept that is supported by the dramatic improvement provided by dopamine replacement therapy. 2 The focus on dopaminergic therapies in the treatment of PD explains the emphasis that has been placed on attempting to understand the mechanisms involved in nigral cell degeneration. However, the fascination with events occurring in this region does not take into account the many other pathologic changes that have been described in nerve cells of the central and peripheral nervous systems. Indeed, pathologic change in PD is extensive and includes degeneration of norepinephrine neurons of the locus coeruleus, cholinergic neurons of the nucleus basalis of Meynert, and serotonin neurons of the dorsal raphe, as well as nerve cells in the olfactory system, the pedunculopontine nucleus, the dorsal motor nucleus of the vagus, and the peripheral autonomic nervous system. 3,4 Although the consequences of these biochemical abnormalities are largely unknown, it is suspected that they might account for the motor and non-motor features of PD that do not respond to dopaminergic therapy. 5 Significantly, they are frequently ignored when concepts of the pathogenesis of PD are considered. However, the emphasis on the SNc has been challenged by the recent findings of Braak et al. 6 suggesting that the pathology of PD starts in the brainstem and olfactory regions and gradually spreads to affect other brain areas. Only when cell death occurs in the substantia nigra and motor symptoms appear is a diagnosis of PD proposed. Although this concept of the illness is controversial and is based exclusively on the distribution of Lewy bodies, it serves to explain many non-motor aspects of the illness and has evoked a reconsideration of what is PD and what are the pathogenic mechanisms that underlie its cause.

2 These issues will be returned to in the course of this review, but it is important to note that there are key features of the pathology of PD that link all affected regions of the nervous system. The hallmarks of the illness are the presence of the proteinaceous intracellular inclusions, termed Lewy bodies, in surviving cells and the occurrence of a reactive microgliosis. 7,8 Recently there has been a surge of interest in these aspects of the disease because they provide a means for attempting to understand the commonality of pathogenic mechanisms that occur in PD. They may also provide an explanation for the spectrum of disease that exists between PD and dementia of the Lewy body type and for the pathologic phenomenon of incidental Lewy body disease that occurs in 10% to 15% of individuals older than 60 years who are neurologically intact at death. Why is studying the pathogenesis of PD considered such an important issue? Symptomatic therapies are effective in the early stages of the disease. In the majority of patients, however, current treatments are associated with motor complications and with the development of potentially disabling features that cannot be adequately controlled with presently available medications. It is our inability to prevent the ultimate development of disability, despite current symptomatic therapies, that makes the search for the holy grail of a disease-modifying or neuroprotective drug so important. 9 It is clear that there are many different genetic causes of PD, and the cause(s) of the common sporadic form of the illness is not known. Defining a therapy based on blocking an etiologic event therefore appears to lie far in the future, and an intervention that is effective against one cause may not influence disease progression in patients with PD due to alternate etiologies. Understanding the pathogenic processes involved in neuron death is in itself an important scientific question, and provides an opportunity to target common pathways with agents that might slow or stop disease progression in patients with PD due to multiple different etiologies. Understanding why some individuals develop PD while others do not, and why and how the disease progresses, could provide clues to the development of a diseasemodifying intervention and is central to much current research. In addition, if the mechanism responsible for cell death can be defined in PD, it is possible that this would have therapeutic implications for other neurodegenerative illnesses as well. So why have we not found agents able to modify PD with the wealth of knowledge that we already have at our disposal? This review looks at some of the potential issues that exist and critically examines whether our current concepts of PD pathogenesis are correct or whether we have limited our opportunities by self-fulfilling thinking. An introduction to pathogenic mechanisms. We have provided many reviews of the evidence for a range of pathogenic mechanisms believed to occur in PD, and it is not the purpose of this review to extensively revisit this information (for examples, see references ). The bulk of the evidence comes from the study of postmortem PD tissues and from attempts to understand the mechanism of action of toxins that destroy nigral dopaminergic cells in model systems. More recently, gene mutations in familial PD have provided additional clues, as will be discussed later However, much of what is currently proposed is tentative and is based largely on interpretation of indirect evidence implicating specific processes derived from studies of late-stage PD brains. Such studies do not, however, permit a determination of which changes are epiphenomena and incidental and which are primary and drive the cell death process. A cynical view would be that, in some respects, the cant of accepted pathogenic processes reiterated over many years leads us to revisit self-fulfilling ideas of the primary mechanisms thought to be involved. This personal and controversial view of the current status of understanding pathogenesis is put forward as a preface to the arguments proposed below.

3 Our current thoughts about the mechanism responsible for cell death in PD are based primarily on findings in the substantia nigra pars compacta of PD patients, or on experimental studies of dopaminergic neurons in primates, rodents, or cultured dopamine neurons/cell lines. Dopaminergic cells are believed to die by apoptosis rather than necrosis, but even this basic concept is disputed There is no doubt that oxidative and nitrative stress occurs in substantia nigra in PD, with the data for the latter being stronger than for the former The source of nitrogen species (nitric oxide and peroxynitrite) is clearly related to alterations in inos activity. The origin of oxygen radicals in much less clear and is based mainly on indirect biochemical changes, such as increased iron levels and alterations in antioxidant mechanisms. 26,27 The actions of toxins, such as 6-OHDA, have been used to support oxidative stress as a primary driver of nigral cell death in PD, 28 but it may be that this is part of the self-fulfilment that has taken place in this area and that secondary involvement of these systems is certainly possible. We have shown that anatomic lesions of the median forebrain bundle in rats lead to a secondary accumulation of iron in the substantia nigra pars compacta in a pattern similar to that observed in PD. 29 There are alterations in nigral mitochondrial complex I and -ketoglutatate dehydrogenase in PD, and both toxin-based and genetic mechanisms have been proposed, although the precise cause remains unknown Here, too, the potential of secondary mitochondrial damage has not been excluded. Excitotoxicity is commonly put forward as part of the pathogenic process that occurs in PD There is some laboratory-based evidence to support this theoretical concept, because there is overactivity of the subthalamic nucleus glutamatergic output pathway in PD and increased staining for 3-nitrotyrosine in Lewy bodies. 37 However, direct evidence that excitotoxicity occurs in PD is notably lacking, and we are not aware of a single publication that robustly demonstrates its occurrence. Is this another example of the cant surrounding the pathogenesis of the illness? Getting back to the basics of the illness appears important, and studies based on understanding the genesis of Lewy bodies are therefore critical. There is good evidence that altered proteolysis contributes to nigral cell death and Lewy body formation 45,46 in both the sporadic and the familial form of PD. Patients with sporadic PD were observed to have significant reductions in levels of -subunits of the 20S proteasome, reduced compensatory activity of the PA700 and PA28 proteasomal activators, and impairment in the activity of each of the three proteasomal enzymes within the substantia nigra pars compacta compared with controls. 40,41 As detailed elsewhere in this supplement, inhibition of proteasomal function can lead to selective dopaminergic cell degeneration coupled with formation of inclusion bodies in cultured cells 47,48 and after direct injections into the supranigral region in rats. 49,50 Most exciting was the report that systemic administration of proteasomal inhibitors to rats caused progressive nigral cell degeneration with pathology in other brain regions affected in PD, associated with the presence of proteinaceous inclusions and motor deficits. 51 However, these findings have proved to be difficult to reproduce by many groups (unpublished data), although some have succeeded in at least partially confirming the original report with observations of cell loss in the substantia nigra pars compacta, locus coeruleus, and raphe nuclei, combined with the presence of -synuclein-positive inclusions and motor deficits that worsen over 8 to 12 weeks (Jenner et al. and Schapira et al., unpublished data). Whether this means that proteolytic stress is a primary universal mechanism responsible for cell death in each of the different forms of genetic and sporadic PD is a different story. When we examined nigral tissue from MPTP-treated common marmosets with stable motor deficits many months after exposure, we found changes in proteasomal enzyme activity, - and ß-subunit expression, and levels of the PA700 and PA28 regulatory caps very similar to those we found in the SNc in PD. 52 These findings raise the possibility that the changes found in patients with PD could be a consequence rather than a cause of nigral cell death. A final issue that needs to be raised at this point are the many attempts to involve dopamine metabolism in cell death in PD (see reference 53 for recent review). This is another example of cant that affects understanding of the events that occur. Although dopamine metabolism may lead to oxidative stress and, in theory, could cause the death of some dopaminergic cells, this does not

4 readily explain the pathogenesis of PD. Dopamine neurons have robust reuptake and storage systems in place that protect against the development of oxidative stress. Furthermore, most dopamine is in the striatum, which does not degenerate, and not in the SNc, which does. In addition, not all nigral dopaminergic cells degenerate, and some dopaminergic cell groups are only partially affected or not affected at all. The evidence for increased dopamine turnover in early disease has been questioned, and there is no convincing evidence that dopamine or levodopa is toxic in normal humans or PD patients. 54,55 Lastly, and most importantly, many of the brain areas affected by pathology in PD are not dopaminergic. There is little to suggest that dopamine cell death leads to degeneration of non-dopaminergic nerve cells, and there are data suggesting that non-dopaminergic cells degenerate earlier in the course of the illness. 6 Perhaps pathogenic mechanisms involved in dopaminergic cell death are different from those involved in the degeneration of non-dopaminergic neurons. This is an issue that warrants further consideration. All of the above is an illustration of how much we have uncovered but how little we understand about the events that occur during cell death in PD. Evidence supporting different pathogenic factors exists, but these factors are not all present to the same extent in all patients, and it is not clear which, if any, are primary and contribute to the initiation of the cell death process, or if pathogenic mechanisms are the same for all patients or for all forms of PD. Indeed, there is now evidence indicating that patients in the same family and with the same mutation can have different clinical and pathologic pictures, 56 and it is reasonable to consider that this may relate to different pathogenic mechanisms. Putting together the pieces. The evidence available for a range of different mechanisms thought to be involved in the pathogenesis of PD is strong, but putting them into an order that reflects the disease process is reminiscent of attempting to put together a jigsaw puzzle without a picture on the box. It is almost impossible to unravel the strands because each process appears to be intimately connected to the others. 16 For example, oxidative stress causes mitochondrial dysfunction and can impair proteasomal function. Alternatively, inhibition of mitochondrial function induces oxidative and nitrative stress and impairs protein clearance because ATP production is required for normal proteolytic activity. Impairment of proteasomal function induces oxidative and nitrative stress and inhibits mitochondrial function. And so it goes on, with each new piece of evidence for a mechanism involved in cell death turning out to be part of a single cycle or cascade of events that involves other factors that eventually lead to destruction of neurons. The postmortem evidence for pathogenic mechanisms in PD inevitably comes from studying brain tissue from individuals at the end stages of the illness and with a long history of drug treatment. Postmortem delay and a range of other factors influence the quality and reliability of biochemical measurements made, and this can be controlled for only by comparison to material from normal individuals whose brain tissue is matched for these parameters. This is not ideal because it is possible that postmortem changes occur differently in the two groups as a result of the underlying pathogenic process. For example, agonal events that induce no morphologic change in a control substantia nigra may promote cell death in the substantia nigra of a PD patient where proapoptotic signals have already been activated, thus creating a falsely increased index of the number of cells that die by apoptosis. 22 Allowance for the effects of drug treatment is made by comparison to other disease states for which dopaminergic medications are also used, notably multiple system atrophy and progressive supranuclear palsy. 57 Again, this is not ideal, but tissues from untreated patients with advanced PD are rarely available and even then do not reflect the same state of disease severity.

5 The relationship of the components of the pathogenic process to the stage of the disease is also barely studied. There is some evidence suggesting that the degree of iron accumulation and the reduction in levels of reduced glutathione (GSH) become more pronounced with disease progression. 57,58 Otherwise, we have little idea as to the point at which oxidative or nitrative stress, mitochondrial dysfunction, and failure of proteasomal activity first occur. Some data have been gained from analysis of biochemical changes in cases of incidental Lewy body disease. 13 These indicate that iron levels and mitochondrial function are normal in the early stages of nigral degeneration but that GSH levels are reduced to the same extent as in advanced PD. This is perhaps an important snapshot of the beginning of the degenerative cascade, but it presumes that incidental Lewy body disease is a preclinical stage of PD, and this remains unproven. Another attempt to relate biochemical changes associated with pathogenesis to the disease process has been the study of other illnesses in which Lewy bodies are formed, notably dementia with Lewy bodies. However, the available evidence suggests that changes in this condition do not precisely reflect events that occur in PD, although the numbers of studies are limited. 59 In any event, the response of cortical non-dopaminergic cells to a given pathogenic process is not necessarily the same as that in nigral dopaminergic neurons. We are at a point in our understanding of the mechanism of cell death in PD that suggests that, despite our accomplishments, there is still a long way to go in defining the primary pathogenic event. For example, although agents that interfere with oxidative stress, excitotoxicity, inflammation, mitochondrial dysfunction, and apoptosis have been tested in PD patients, none has been determined to provide a disease-modifying or neuroprotective effect. 60 There is also a suggestion that the pathogenesis of PD is a more widespread process than was previously believed. For example, increased protein oxidation, as measured by levels of protein carbonyls, occurs in all areas of the brain and not just in the SNc. 61 It may also be that PD is not restricted to the CNS, because there is evidence for oxidative stress and oxidative damage with reduced antioxidant defenses in blood and urine. 62,63 It should also be remembered that Lewy bodies are found in peripheral autonomic neurons in the myenteric plexus, 64 olfactory dysfunction and olfactory pathology appear to be early features of the disease, 6,65 dementia with cortical Lewy bodies is common, 66 and visual impairment is more common than in controls, possibly due to degeneration of dopaminergic neurons in the retina. 67,68 Therefore, we are not even certain that we have all the pieces of the jigsaw puzzle, let alone the ability to complete the picture. Applying pathogenic mechanisms to neuroprotective therapies. The primary objective in studying pathogenesis in PD is to facilitate the development of neuroprotective therapies that might slow or stop disease progression. This presumes that PD is a single illness, whereas the evidence suggests that it is a syndrome associated with different etiologies, which varies in complexity among individuals and involves a range of motor and nonmotor symptoms. Different signs and symptoms may commence at different points in the illness and may progress at different rates. If this is the case, then the range and degree of involvement of the varying pathogenic processes may also vary among individuals. In fact, this appears to be the case for oxidative stress and mitochondrial dysfunction. Although there are differences between control and PD tissues when data are averaged, this is not true for individual patients. This is best illustrated with respect to mitochondrial activity, which is one of the best-studied areas in PD. 69 Although there is a mean reduction of approximately 40% in complex I activity in the substantia nigra in PD, only 30% of patients have a decrease that is more than 2 SD away from the mean value for control tissues. This means that many patients with PD have levels of mitochondrial function that are perfectly normal. The same is true with respect to markers of oxidative stress. Moreover, it is not the same patients who show abnormal mitochondrial function and high levels of oxidative stress. This suggests that the cascade of events to which we so

6 resolutely adhere may not reflect reality and, at the least, are not likely to occur in a linear fashion in which change in one parameter is dependent on change in another. It might also imply that individual patients enter the pathogenic sequence at different points, thus adding to the notion of heterogeneity in pathogenesis as well as etiology and clinical pattern PD. Almost everything discussed thus far refers to events that occur in SNc and affect dopaminergic neurons. This neglects the widespread pathology occurring in PD that affects multiple brainstem, midbrain, and forebrain nuclei that are non-dopaminergic in nature, where degeneration with Lewy body formation occurs. 6,7 The assumption is made, by analogy, that all the pathogenic events that occur in the SNc also occur in these brain regions. Yet we know virtually nothing about the changes that occur in these areas because little or no analysis has yet been undertaken. It is presumed that Lewy bodies are formed by one common mechanism, 46 but it may be that different events alter protein handling in different ways in different neuron populations. All of this has bearing on our approach to neuroprotection. If PD is not a single illness but a syndrome, if different individuals are going down a different pathogenic road, and if different events are taking place in different brain regions affected in PD, maybe it is not surprising that clinical trials of neuroprotective agents routinely fail. 9,60,70 Unless a common pathogenic factor is identified, it may be that only when subgroups of patients are identified and treated with regard to the pathogenic cascade that is relevant to their specific illness are we likely to succeed. If we continue to examine large heterogeneous patient populations with therapies targeted at individual factors thought to play a role in the pathogenic cascade, we may miss benefits in individual patients and vital clues toward the development of a disease-altering intervention. Progression and pathogenesis. The focus of pathogenic studies in PD has been based on understanding the primary mechanisms responsible for dopaminergic neuron death. The progression of the illness has received less attention, and the presumption is that it is a consequence of the continuation of the primary pathogenic mechanism that makes neuronal cells become sick and die. This has led to a situation in which the question of "what is wrong with dopaminergic neurons in PD" has dominated research. An alternative view is that the neurons may be distressed by one factor but that an entirely different process impairs their ultimate survival. Whatever the events that lead to the destruction of cells in PD, they are not limited to neurons. This is clear from studies looking at the extent of change in measures of oxidative stress, mitochondrial function, and proteasomal activity in the substantia nigra pars compacta. 16 The magnitude of change in levels of iron, reduced glutathione, complex I activity, and proteasomal enzyme activity in homogenates derived from the entire substantia nigra pars compacta are always in the region of 30% to 40%. 26,27,31,41 This makes it difficult to account for these changes solely on the basis of degeneration of dopaminergic neurons, which comprise just 1% to 2% of the total cell population. A more parsimonious explanation is that these processes also occur in glial cells, as they constitute the commonest cell type. Indeed, alterations in iron levels and reduced GSH have been demonstrated in glial cells in the substantia nigra in PD. 71,72 This raises the possibility of a more generalized malaise in the SNc, to which dopaminergic neurons are highly sensitive but to which glial cells are relatively resistant. All of this raises the question of the role of glial cells in PD. It is an important issue because a reactive microgliosis and, to a lesser extent, astrocytosis are hallmark features of PD. 8,55 It is presumed that gliosis occurs after the onset of neuron cell death, although this has not been proved to be the case. Reactive microgliosis in the SNc has also been observed many months

7 after MPTP administration to primates and is thought to correlate with, and perhaps to mediate, mild progression over prolonged periods of time. 74,75 There is substantial laboratory evidence demonstrating the capacity of activated glial cells to damage dopaminergic neurons. 76,77 It therefore appears more likely that the glial response in PD may represent a compensatory protective response or a contributor to disease progression rather than a factor involved in disease initiation. However, this is not a certainty because there is evidence for peripheral inflammatory changes in PD and for leakage of the blood brain barrier that may allow monocytes to penetrate into the substantia nigra. 78,79 This raises the possibility that inflammation may be a primary factor in some cases of PD. Indeed, immunization with nigral homogenates induces a nigral degeneration in rodents, 80 and population studies show that regular intake of aspirin or other non-steroidal anti-inflammatory agents can significantly reduce the risk for development of PD. 81 Glial cells are a dual-edged sword. They have a physiologic role in supporting neuron survival by providing nutrients and trophic factors. Activated glial cells, on the other hand, can become phagocytic and cause neuron destruction. In PD, microglial cell activation is associated with inflammatory change, including an increase in the levels of inflammatory mediators (interleukins and TNF ), COX-1, and inos. 82,83 The latter may be important in increasing production of NO, which can itself be toxic, but is more likely to react with superoxide to generate the more toxic molecule peroxynitrite. This may bring oxidative and nitrative stress together as pathways leading to a common toxic mechanism. 16 Indeed, a role for peroxynitrite toxicity in PD is suggested by the presence of 3-nitrotyrosine adducts in substantia nigra and notably in Lewy bodies. 37 The role of glial cell activation has been extensively studied in experimental models. In vitro glial activation using lipopolysaccharide (LPS) causes increased release of cytokines, glutamate, NO, and oxygen radicals, with decreased formation of BDNF and GDNF Mixing activated glial cells with primary fetal ventral mesencephalon cultures leads to selective dopaminergic neuron loss, emphasizing their potential role in pathogenesis. Similarly, direct intranigral injection of LPS causes glial cell activation and death of dopaminergic neurons. 87,88 This is associated with inos induction and 3-nitrotyrosine immunoreactivity in both neurons and glial cells. The role of NO in this pathologic process is further illustrated by the ability of selective inos inhibitors to partially prevent LPS toxicity. All of this information leads to the question of whether preventing glial cell activation or inflammatory change in PD has an effect on the progression of the illness. In experimental studies, the toxic effects of 6-OHDA, MPTP, and LPS on dopaminergic neurons are prevented by a range of anti-inflammatory and immunosuppressant agents However, these findings are controversial (see, e.g., references 96 and 97 ), and initial claims that protection was related to selective inhibition of COX-2, or of cyclo-oxygenase activity in general, appear not to be true. Interestingly, dopamine agonist drugs also seem able to inhibit LPS toxicity, but without altering glial cell activation (unpublished data). What is missing is any data from studies in PD to indicate that these effects are relevant to modification of the course of the illness. Indeed, the routine use of NSAIDs does not appear to affect disease progression, contrary to its effect on disease prevention. Altering glial cell activation or its consequences in PD would seem an excellent approach to providing neuroprotection, but at present a rational therapeutic strategy is not evident. Clinical studies using minocycline or NSAIDs may reveal new leads and provide more information on the role played by glial cells in the pathogenesis of PD. Minocycline was not rejected in PD patients in a futility analysis conducted by the NIH (the NET-PD group) and will likely be further studied in a long-term simple study (but see reference 98 ).

8 Bringing sporadic and familial PD together. The study of rare families with inherited PD has led to the identification of at least 11 gene defects and six gene products that appear to underlie nigral cell degeneration (see references 8 20 and the chapter by Schapira in this supplement, pages S10 S23). These manifest as autosomal dominant or recessive disease, generally with an early age of onset. There is little doubt that this has been one of the great success stories in recent years and has significantly advanced understanding of pathogenic mechanisms in PD. For example, the lack of E3 ubiquitin/protein ligase activity of mutant forms of parkin 99 and the decreased recycling of monomeric ubiquitin in UCH-L1 mutations 100 led to the investigation into proteasomal function in sporadic PD 40,41 and ushered in the concept that proteolytic stress might be a common factor in both the familial and the sporadic form of PD. 38,78,79 However, there may be some danger in attempting to link the toxic effects of gene products in familial PD to pathogenesis in the sporadic form of the illness. Similarly, there may be an inherent flaw in expecting gene products to result in actions that we associate with sporadic disease. This has an inbuilt belief that there is only one pathogenic process that leads to all forms of PD and, as emphasized above, this may not be true. It also makes concepts of the pathogenesis of PD selffulfilling and means that we do not think outside the box when we consider how the illness arises. I will try to illustrate this by a few selected examples. The discovery of mutant forms of -synuclein stimulated a huge amount of research activity that showed these proteins to be toxic to dopaminergic cells through mechanisms that included misfolding and aggregation, impairment of proteolysis, and interactions with dopamine ,101,102 The presence of -synuclein in Lewy bodies in sporadic PD fuelled the idea that PD was a synucleinopathy. 103 The subsequent discovery that duplication of the -synuclein gene, leading to an increase in wild-type protein load, was apparently pathogenic in familial PD 104 further fanned these flames. This is an oversimplification of a huge wealth of literature, but it captures the flavor of commonly accepted sentiment about -synuclein and its relationship to pathogenesis in PD. However, if one takes a step back and looks at this question from a distance, then the outcome can be quite different. Where is the evidence that wild-type -synuclein is pathogenic in sporadic PD? Certainly it is present in Lewy bodies, but it is a very common protein and many other proteins have also been identified to be constituents of these proteinaceous inclusions. It is also found in inclusions in a range of other neurodegenerative diseases for which the pathogenic process is presumably different (see, e.g., reference 105 ). Cell death induced by proteasome inhibitors is associated with inclusion bodies that accumulate -synuclein, but there is currently no evidence that -synuclein plays a role in neurodegeneration in this model. Lewy bodies also contain ubiquitin and, on this basis, PD could equally be termed an ubiquitinopathy because mutant forms of ubiquitin are toxic when expressed in cell lines. 106 As far as we are aware, in sporadic PD there is no evidence for overexpression of wild-type -synuclein or the presence of the mutant -synucleins found in familial forms of the illness. Nitrated -synuclein protein has been detected, but other proteins are also present in nitrated forms. 107 There is some evidence to support the presence of abnormal forms of -synuclein, but the relationship to pathogenesis is unclear. 108 Expression of mutant -synuclein in substantia nigra using a viral vector produces dopaminergic cell death, but abnormal expression of many proteins induces proteolytic stress and could be toxic to cells. Perhaps most importantly, all attempts to produce nigral pathology in transgenic mice with wild-type or mutant -synuclein expression have failed, although other pathology and proteinaceous inclusions have been observed (see, e.g., references 111 and 112 ). It therefore remains to be established precisely how mutant -synuclein causes PD or whether wild-type -synuclein has anything to do with cell death in sporadic PD. Attempts have also been made to link parkin mutations to sporadic PD. 113 Parkin mutations are a common cause of the autosomal recessive young-onset form of PD, 114 and parkin is found in Lewy bodies in patients with sporadic PD. 115 Again, nitrated forms of parkin have been detected in

9 sporadic PD, 116 and viral vector delivery of parkin protects dopaminergic neurons. 117 However, parkin mutations are a rare cause of late-onset disease, most cases do not have Lewy bodies, there is no alteration in wild-type parkin expression in sporadic illness, and attempts to produce nigral pathology in parkin transgenic mice have failed. 118,119 Although parkin mutations provide a link with the loss of proteasomal activity observed in sporadic PD, it could alternatively be viewed as a different form of the illness. More recently, mutations in DJ-1, PINK1, and LRKK-2 have been described in patients with familial PD DJ-1 and PINK1 mutations appear rare, but LRRK-2 mutations are associated with as many as 7% of young-onset cases and approximately 2% of late-onset cases, some of whom have no family history and pathology no different from that found in sporadic PD. This illustrates the complexity of trying to blindly lump all PD cases into one pathogenic mechanism. The physiologic function of these proteins is not clear, but they have been respectively associated with cellular redox control, mitochondrial function, and protein kinase activity. It is hoped that the study of these proteins will advance knowledge of cellular control and cell death. However, the inevitable iteration process has already begun, with attempts to link them to sporadic disease and known pathogenic mechanisms rather than viewing these mutations as causing specific and perhaps alternative forms of what we currently call PD. Conclusions. PD is viewed as a primary degeneration of nigral dopaminergic cells of unknown cause, albeit there have been extensive investigations into etiology and pathogenesis. With respect to pathogenesis, evidence has accumulated to implicate oxidative stress, excitotoxicity, mitochondrial dysfunction, and inflammation, although what role, if any, each of these plays in the degenerative process, and how they interact with each other, remain unknown. This review has attempted to encapsulate the knowledge base that has arisen as a result of the studies performed to date but at the same time to avoid dogma and to add critical comment. There is a danger in any disease area that the apparent understanding of cause becomes somewhat blurred through cant, and it is certainly true in PD that strongly held views on pathogenesis have sometimes become truth through repetition rather than fact, and that they obscure rather than promote further investigation. Perhaps the biggest rethinking required is the question of what is PD. Is this a single illness centered on nigral pathology and motor abnormalities? Is it a group of diseases that all affect the substantia nigra but through different pathogenic mechanisms that lead to a core disorder of movement? Or are we dealing with a syndrome of complex symptomatology involving various brain nuclei of which PD, as originally described, forms just a part? Whatever the answers, we need to better understand the nature of the disease and the mechanism whereby cells die before we can develop neuroprotective therapies that slow or stop its progression. Discussion Dr. Schapira: How much of what you said do you actually believe?

10 Dr. Jenner: I haven't got a quote from anybody famous in neurology, but I'll quote my mother, who basically lived to the ripe old age of 94 and used to say the most outrageous things, and when I commented about it, she said, "Don't worry, I'm just old." So when I say all these terrible things that you disagree with, it's because I'm old. Dr. Schapira: Let's start with apoptosis there is considerable evidence suggesting that, whatever the pathogenesis, cells die by a signal-mediated apoptosis. If this is so, it provides an opportunity for neuroprotection that could work in different types of PD. Dr. Jenner: There's no doubt that there are papers showing signal activation and morphologic changes of apoptosis in PD and in models of PD. But what we put to one side is the fact that there are some papers that have not found evidence of apoptosis in PD. So I think the evidence is mixed. In addition, if you accept the papers that support apoptosis, there appear to be too many apoptotic neurons for a disease that lasts for years. Dr. Olanow: I personally think the evidence for apoptosis is very strong, and there is no evidence demonstrating necrosis. Also, in the agonal stages surrounding death, vulnerable neurons could be induced to go into apoptosis, thus giving a falsely high number of apoptotic cells. Dr. Jenner: Is apoptosis the only way cells die in PD? Dr. Olanow: That's a different question. But I don't think there's much doubt but that there is apoptotic cell death in PD. Dr. Jenner: I agree with that. But what we do is, we translate this into apoptosis being the be-all and end-all of what happens in terms of cell death in PD. Dr. Schapira: Do you see apoptotic cell death in MPTP monkeys? Dr. Jenner: No, but we look months after the insult. Dr. Olanow: That may be part of the problem. MPTP is an acute toxin that kills cells acutely. Nadine Tatton and Steve Kish reported on apoptotic cells after MPTP in mice, and you may just be waiting too long to see them. Dr. Jenner: All I'm doing is raising the doubt about whether apoptosis is what we should focus on. Dr. Olanow: That's perfectly fair. But bear in mind that there's not a shred of information supporting necrotic cell death. So you may have incomplete evidence for apoptosis, but there is no evidence for necrosis. And there are practical issues here, because if cells die by apoptosis, as Tony pointed out, this gives us an opportunity to develop a neuroprotective drug that blocks pro-apoptotic signals that might work in all forms of PD. Dr. Schapira: Let's consider oxidative stress. In PD you have increased iron levels, decreased ferritin levels, decreased levels of reduced glutathione, increased levels of superoxide dismutase. Is there any question that there is oxidative stress in PD? Dr. Jenner: I think there is evidence of oxidative stress and also evidence of oxidative damage to lipids, DNA, and proteins in postmortem studies. But where does it come from? Dr. Schapira: But don't these indices overlap with controls?

11 Dr. Jenner: They do, and in some patients the findings are much more pronounced than in others, just as you have found with complex I deficits. Dr. McNaught: Has anybody looked for markers of oxidative stress in parkin cases or in other patients with genetic forms of PD? It would be interesting if they were different than in sporadic cases. Dr. Jenner: Not to my knowledge. And I come back to the question of where the oxidative damage is coming from. I can't find any evidence for increased formation of free radicals in PD. We talk about them coming from the mitochondrial transport chain or from dopamine metabolism, but I don't think there is any evidence of an increase in PD. Dr. Poewe: There's evidence of increased free radicals in model systems Dr. Jenner: But how do you know they are relevant to PD? Dr. Olanow: Free radicals last for one-trillionth of a second. You've got a dead brain. The presence of oxidative damage is the evidence of free radical formation. Dr. Jenner: Those things happen, and we presume that they occur as a result of increased oxygen radical formation. But I don't know if that's true. We say oxidative stress occurs. But we haven't got any direct evidence that this occurs at all, none at all. What I worry about is that we take the evidence that we have, we study model systems, and then it becomes truth. But we don't actually know that this happens in PD. Dr. Olanow: What's interesting about this is that you have done most of the work on oxidative stress and you have been its major advocate. So this is an attack on yourself. I think what one can say is that there is indirect evidence of oxidative stress based on findings of increased iron, which promotes oxidative stress, decreased glutathione, which protects against oxidative stress, and markers of oxidative damage. That we have not measured free radicals directly reflects the fact that we are dealing with dead brain and can't do that. If you want us to keep an open mind, that's fair. But bear in mind that all we have learned in nuclear physics was based on this kind of indirect information. Dr. Jenner: What I'm pointing out is that much of what we see is indirect evidence, and this worries me increasingly as age comes upon me. Dr. Schapira: The observations are direct. Your mechanisms are indirect. Dr. Kieburtz: But there is something dangerous here, and I kind of agree with Peter. We make observations and create model systems to try to understand those observations, and then believe in the model systems beyond what the observations actually support. And you start to take a path based on the models and not on the observations. Dr. Jenner: Exactly. And contact with the ground becomes removed as the idea takes off. Dr. Jenner: In contrast, there is better direct evidence for nitric oxide involvement in PD. Here normal, not raised, levels of superoxide can combine with nitric oxide to form the highly toxic molecule peroxynitrite, which can decompose into the hydroxyl radical, which is a damned sight more toxic than either hydrogen peroxide or superoxide. Dr. Schapira: So you're saying that there is a little bit more direct evidence of nitric oxide involvement.

12 Dr. Jenner: Yes. You have increased levels of 3-nitrotyrosine in Lewy bodies, nitrated alphasynuclein, and so on. Dr. Olanow: I think this is the same kind of evidence you have for oxidative stress. Dr. Schapira: I must say I agree with Warren. You're talking about free radical-mediated damage versus nitric oxide-mediated damage. You're not measuring either of them directly. Besides, you are probably dealing with cascades or networks where all of these happen at some point or other. Dr. Jenner: I would agree with you. None of these pathogenic factors work in isolation. It's all interrelated. Trying to break into the circle at any one point is absolutely impossible. If you start with mitochondria, you finish up with proteasomes. If you start with free radicals, you finish up with inflammation. You cannot separate out these events. And if you want to call them a network, a cascade, whatever you call them, then you have to accept that actually they're all occurring at probably at the same time and they combine to cause neurodegeneration. It's chicken and egg. Dr. Poewe: What is the implication of this for neuroprotection? Dr. Jenner: It makes neuroprotective approaches very much more difficult to conceive, and the cocktail idea might be a much better idea than trying to approach any one part of these pathogenic mechanisms in isolation. Particularly, as I keep trying to point out, is that some of the things we're trying to prevent may not actually occur in the disease. Dr. Olanow: So let me get this straight. We don't know what really is going on, so we should try for neuroprotection with a cocktail of drugs. And should we allow all of this shoddy information that you have put forward influence us in choosing the cocktail, or should we just take a cocktail of anything? Dr. Jenner: We do not know what we should use, but we have concepts and hypotheses that we all support to varying degrees about what causes PD, and it would probably be best to put together groups of compounds based on best guesstimates. Dr. Siderowf: They do this routinely in HIV and cancer. Dr. Jenner: Absolutely. And I think we are heading in the opposite direction. Dr. Olanow: In fairness, I think everyone would be delighted to test a cocktail. The problem has been, what do we use and where. The first neuroprotective study we did was a cocktail where we evaluated deprenyl and vitamin E. So we're not averse to this. The problem is that the comments you've made so far haven't advanced our ability to design a neuroprotective study. Dr. Jenner: There is also a problem with laboratory models that have little relevance to PD. We use them to study pathogenesis and to assess putative neuroprotective drugs. An example is TCH, the propargylamine tested by Novartis. In the laboratory it did everything right it was antiapoptotic, it blocked MPTP toxicity. The only problem is that it didn't work. Dr. Olanow: Hold on a sec. There are many reasons that results in models weren't replicated in PD patients. One is choosing the dose that would provide comparable effects in animals and humans, especially for an agent like TCH that is highly protein-bound, poorly bioavailable, works in very low concentrations, and has a U-shaped effect in the laboratory. You also don't have a marker with neuroprotection in the way you do with a symptomatic drug to help you choose the dose. So in designing the study it is hard to extrapolate from the rodent or the tissue culture dish to the human.

13 Dr. Jenner: Another factor to consider is that while there is evidence of oxidative stress, mitochondrial dysfunction, excitotoxicity, and proteasomal inhibition in PD, you don't find all of these parameters changed in the nigral tissue of each patient nor any one of them in all patients. For example, we measured some parameters of oxidative stress in nigral tissue, and in one patient we'd find reduced glutathione levels but no change in anything else. In another patient we'd find evidence of oxidative stress but no change in reduced glutathione. Dr. Poewe: Did you find any where there was no abnormality? Dr. Jenner: I don't think so. Dr. Olanow: Were there any that didn't have proteasomal dysfunction? Dr. Jenner: We didn't measure this we did the study too long ago. Dr. Olanow: I think that this goes back to the point you made, Tony, that cell death occurs as a consequence of a cascade of events. If you think that this is a linear sequence of events and try to block a critical step, I think you're bound to fail for the reasons you have outlined. I think you need to consider this as a network in which a group of factors can contribute to cell death and blocking one of them isn't sufficient. Dr. Jenner: We talk about networks and cascades, but if you look at an individual patient you don't find a cascade, because one parameter is changed but others aren't. Dr. Olanow: But if you consider that there are multiple factors that determine whether a cell lives or dies, some may compensate but others may not and may lead to neurodegeneration, and they may be different in different patients. Dr. Jenner: That could be. The other interesting thing is what I call the 30 to 40 percent rule. It doesn't matter what you measure, it always changes by 30 to 40 percent. You don't have to do biochemistry; you know the answer is going to be between a 30 to 40 percent change. That's been found for iron, reduced glutathione, complex I, and proteasome activity. It's always 30 to 40 percent. And we always focus on neurons. But you can't have these sort of changes occurring just in neurons, which make up only 5 percent of cells in the substantia nigra. The most likely explanation is that the defect is occurring in both neurons and glia. You can see increased iron and decreased glutathione in glial cells. So it appears to be a much more general metabolic change in the nigra than what is generally contemplated. Dr. Schapira: I don't think anyone has ever disagreed with the idea that glia are affected, but it may be that there is a difference in the susceptibility of neurons. Dr. Jenner: That's fine. But then you have to consider the cause as something that affects both neurons and glial cells. If you want to latch on to something, there are two things we really know about PD. One is that there are Lewy bodies there, so protein handling and its role in pathogenesis are probably important. The other is that there are inflammatory changes. Dr. Schapira: You're ignoring the fact that that there are very few dopaminergic neurons remaining at the time you do your studies. Dr. Jenner: I don't disagree, but I still think it is important to consider the role of glial cell activation. We did an experiment in which we showed that the endotoxin lipopolysaccharide, which activates glial cells, leads to nigral neuron death. And there is evidence of increased nitrous oxide formation and 3-nitrotyrosine immunoreactivity, an index of peroxynitrite as is seen in PD.