OUP user menu

Parkinson's disease: the first common neurological disease due to auto-intoxication?

A.C. Williams, L.S. Cartwright, D.B. Ramsden
DOI: http://dx.doi.org/10.1093/qjmed/hci027 215-226 First published online: 22 February 2005


Parkinson's disease may be a disease of autointoxication. N-methylated pyridines (e.g. MPP+) are well-established dopaminergic toxins, and the xenobiotic enzyme nicotinamide N-methyltransferase (NNMT) can convert pyridines such as 4-phenylpyridine into MPP+, using S-adenosyl methionine (SAM) as the methyl donor. NNMT has recently been shown to be present in the human brain, a necessity for neurotoxicity, because charged compounds cannot cross the blood-brain barrier. Moreover, it is present in increased concentration in parkinsonian brain. This increase may be part genetic predisposition, and part induction, by excessive exposure to its substrates (particularly nicotinamide) or stress. Elevated enzymic activity would increase MPP+-like compounds such as N-methyl nicotinamide at the same time as decreasing intraneuronal nicotinamide, a neuroprotectant at several levels, creating multiple hits, because Complex 1 would be poisoned and be starved of its major substrate NADH. Developing xenobiotic enzyme inhibitors of NNMT for individuals, or dietary modification for the whole population, could be an important change in thinking on primary and secondary prevention.


In 1887, a medical student in Strasbourg (W. His) gave a dog a quantity of the exogenous compound pyridine, and found it was predominantly excreted as its N-methyl derivative.1 Subsequent studies of xenobiotic metabolism and pharmacogenetics have yielded many examples of detoxification of foreign chemicals. Among such reactions, methyl conjugation is a well established pathway in the metabolism of endogenous agents, including hormones, neurotransmitters, vitamins and drugs. The xenobiotic metabolizing system must be an important defence mechanism, particularly as most xenobiotics are too small to be recognized by the immune system.

However, like the immune system, while normally protective, this system can also cause rather than prevent toxicity, by converting protoxins to toxins. In 1952, Peters coined the term ‘lethal synthesis from enzymic error', in his Croonian lecture to the Royal Society. He was referring to the neurotoxicity of fluoroacetic acid, a protoxin that crosses the blood brain barrier easily, and after conversion to fluorocitric acid impedes mitochondrial energy production due to blockade of the Krebs cycle,2 but many examples of this can be found from the conversion of prodrugs to drugs to toxic metabolites.

Here we suggest that the basis for idiopathic Parkinson's disease (PD) is a lethal synthesis from, with a failure to preserve, vitamin B3. There is evidence that genetic and environmental factors are important in this common disease,3–6 but until now, nobody has convincingly linked a specific gene with specified diet-derived substrates, let alone a vitamin. We propose a class of ecogenetic diseases of auto intoxication, and predict that chronic poisoning by transformed chemicals may be a common mechanism for conditions linked with ageing. Furthermore, we suggest that this hypothesis gives immediate ideas on primary prevention by dietary manipulation, reducing availability of the methyl donor, or the development of xenobiotic enzyme inhibitors (XEIs), a new class of therapeutic agents.

Nicotinamide N-methyltransferase (NNMT)

The enzyme that His was studying in 1887 was NNMT. This enzyme N-methylates pyridines, in particular nicotinamide to N-methyl nicotinamide, its major metabolite. (Figures 1 and 2). The remainder goes on to form NAD(H) and NADP(H), and therefore is vital to a large number of metabolic reactions. NADP(H) is generally associated with different roles from those of NAD(H),7 and is believed to serve primarily in oxidative defence or reductive metabolism.8 In the brain, a major oxidative defence mechanism consists of peroxide removal using GSH peroxidase. It has been demonstrated that GSH detoxification pathways may ultimately depend on the availability of NADPH reducing equivalents to replenish GSH as a cofactor for GSSG reductase. NAD(H), on the other hand, is believed to serve primarily in the energetic production of ATP. NAD+ depletion is considered a critical factor in precipitating cell death during oxidative stress due to compromised energetics. Of particular interest to PD, NADH is integral to: (i) the normal function of complex 1 of the mitochondrial chain (known to be defective in MPTP Parkinsonism9 and the idiopathic condition10,,11); (ii) the formation of tetrahydrobiopterin (a co-factor for tyrosine hydroxylase and therefore important in the production of dopamine and known to be deficient in PD12) and reduced glutathione (also known to be deficient in early stages of PD13,,14). Nicotinamide provides cytoprotection through pathways that involve poly (ADP-ribose) polymerase (PARP), Akt, mitochondrial membrane potential and cysteine protease activity. Nicotinamide also maintains cellular integrity and prevents cellular removal by maintaining DNA integrity and membrane phosphatidylserine (PS) asymmetry. It affects DNA degradation through a series of cellular pathways that involve PARP, Akt, forkhead transcription factor, mitochondrial membrane polarization, cytochrome c, and inhibition of caspase-1, caspase-3 and caspase-8. In addition, cellular membrane asymmetry, which prevents PS externalization and protects against microglial and cytokine activation with subsequent phagocytic destruction of cells, is maintained principally through activation of caspase-1 and caspase-8.15–26 Many of these pathways have been invoked for MPTP poisoning and PD.27–31 Nicotinamide protects against the toxic effects of L-DOPA in cell culture, as illustrated in Figure 3, and MPTP toxicity.32–3939 Experimental nicotinamide deficiency, produced by the antimetabolite 6-aminonicotinamide, causes dopamine deficiency, loss of cells in the pars compacta, and parkinsonism with a good response to dopamine agonists. Toxicity is based on the lethal synthesis of analogues of NADH and NADPH.40

Figure 1.

Formation of MPP+ from either MPTP by MAO-B, or methylation of 4-phenylpyridine by NNMT (top). Methylation of nicotinamide by NNMT (bottom).

Figure 2.

Pathways of synthesis of NADH, the essential cofactor for mitochondrial Complex 1, from nicotinamide, nicotinate and tryptophan. The enzymes involved in the nicotinamide pathway are listed below. Also shown schematically is the relationship between Complex 1 and Complex 5 of the mitochondrial oxidative phosphorylation pathway, the major source of ATP. The hypothesis proposed is that high levels of NNMT diminish the availability of nicotinamide for NADH synthesis at the same time that Complex 1 is inhibited, both resulting in diminished ATP synthesis. The tryptophan pathway occurs only to a very limited extent in brain, and therefore would not be able to compensate, and the nicotinate pathway would be affected by levels of NNMT because of the shuttling between NADH and nicotinamide. 1, purine-nucleoside phosphorylase [E.C.]; 2, nicotinamide ribonucleoside kinase [E.C.]; 3, nicotinamide mononucleotide adenylyltransferase [E.C.]; 4, NAD glycohydrolase (NADase CD38) [E.C.]; 5, nucleotide pyrophosphatase (Autotaxin) [E.C.]; 6, 5′ nucleotidase (CD73) [E.C.]. Alternative names shown in italics.

Figure 3.

Nicotinamide protection against the cytotoxic effects of L-DOPA. TE671 cells exposed to increasing concentrations of L-DOPA for 24 h in the presence or absence of 0.1 mM nicotinamide. Cytotoxicity determined by assay of LDH release. Nicotinamide produced a statistically significant reduction in cytotoxicity across the range of L-DOPA concentration (p<0.0001). Results previously presented in abstract form (Cartwright L, Williams AC, Ramsden DB. Nicotinamide: neuroprotective effects against dopamine and MPP+. Proceedings of The Global College of Neuroprotection and Neuroregeneration, Annual Conference 2004, Zermatt).

Increased activity of NNMT therefore would lead to cellular nicotinamide deficiency with complex results, as seen clinically in pellagra, but with reasons to believe damage would occur to dopaminergic systems, and to the cells ability to withstand an attack from dopaminergic toxins such as MPP+ and therapy with Levodopa. The latter, when combined with a decarboxylase inhibitor (vitamin B6 antagonists) will reduce the amount of nicotinamide reaching cells. Given that NNMT is partly under genetic control, there may be an evolutionary hunter/gatherer style link between the gene and a particular environment that is high in nicotinamide, as would happen with an omnivorous or carnivorous diet (NNMT barely exists in herbivores).41 This may provide some homeostatic control, but would not necessarily avoid long-latency toxicity. NNMT has been cloned, and its structure is known.42–44 It may possibly be polymorphic, but the structural basis of the polymorphism or duplications is unknown, and does not appear to be within the coding region of the gene. Alternatively the genetic control may be post-translational. The protein is inducible by its own substrates, and by other factors, including stress. Twenty-five per cent of the general population are high expressors of NNMT.

Toxicity of N-methyl compounds

A single case45 followed by a small epidemic of parkinsonism46 in young heroin injectors was found to be caused by MPTP, a N-methylated pyridine. The similarity with idiopathic PD is striking at the clinical, pathological and biochemical levels and has led to excellent animal models. Most of the biochemical and physiological abnormalities found in the last 20 years from the deficit of complex 1 and the involvement of the subthalamic nucleus in PD were first discovered in the MPTP model, and it has been extensively used for drug and surgical developments. It was soon found that compounds of similar structure that did not have the N-methyl moiety were not toxic to any degree.47,,48 MPTP is a protoxin converted by MAO-B to MPP+.49 MPP+ and similarly charged compounds cannot cross the blood brain barrier. If this conversion occurs within the brain, MPP+ may be taken up into dopaminergic cells and concentrated via the dopamine active transport system and further concentrated in mitochondria where it poisons complex 1.50,,51 Inhibition of MAO-B prevents MPTP toxicity, which is also relatively less toxic by the oral route because MAO-B in liver converts it to MPP+, which then cannot cross the blood-brain barrier. Interestingly, inhibition of MAO-B does not affect the natural history of Parkinson's disease. However, an alternative and direct pathway converts pyridines, including nicotinamide and 4-phenyl pyridine, directly to MPP+ and similar compounds This route is catalysed by NNMT (Figure 1).52 If this happened in vivo, MAO-B inhibition would not obviate the generation of toxins. Several N-methylated compounds and/or their N-methylating enzymes (probably NNMT) have been proposed as being involved in Parkinson's disease.53–64 However these findings have not been generally accepted, in part because it was unclear whether the relevant enzyme occurred in the CNS of mammals.65

Nicotinamide metabolism in parkinsonian patients

An initial study in Man66–69 followed the Wilhelm His protocol in dogs, and demonstrated that patients with Parkinson's disease excreted high quantities of N-methyl nicotinamide, compared with controls or patients with other neurodegenerative disease.70 However, at the time, the brain was not thought to contain NNMT, and therefore there were problems with the hypothesis, because increased methylation peripherally would reduce available nicotinamide but the charged N-methyl compounds would not cross the blood-brain barrier.71 However, NNMT is found in the brain, and indeed is within neurones including dopaminergic neurones of the substantia nigra (Figure 4).72 It has a regional distribution, presumably reflecting nicotinamide requirements.73 Low nicotinamide levels are toxic, but so are high levels,74,,75 so tight control may be necessary. Intra-neuronal nicotinamide levels may be a link between degeneration when too low, and carcinogenesis or developmental problems when too high via PARP depletion,76 such as might be achieved by high dietary intake of nicotinamide and low methylator status from genetic pre-disposition plus an inhibitor of NNMT such as nicotine. N-methylnicotinamide may have acquired a normally beneficial role as it inhibits the export of choline77 (also a charged N-methylated compound) from the brain, and therefore boosts acetylcholine levels. This may help developing cognition, and even delay the effects of cholinergic degeneration although it would not be pharmacologically helpful in patients who already had PD. On the other hand, N-methyl nicotinamide has MPP+-like toxicity,78–80 and choline as a methyl donor would facilitate further methylation. NNMT levels are high in the spinal cord and in some parts of the cortex, suggesting that such regions prefer lower nicotinamide and higher N-methyl nicotinamide, and vice versa for regions such as substantia nigra with low levels of the enzyme. In Parkinson's disease brain, levels of this enzyme are high (Figure 5).73,81,,82 They were not raised in the substantia nigra, but as a neuronal protein, the destruction from the disease process makes this hardly surprising; levels might have been high before the destruction took place. In any case, N-methyl compounds could diffuse from other parts of the brain and get taken up by dopaminergic cells, as happens with MPTP poisoning.

Figure 4.

A Fluorescent microscopy of control human substantia nigra with FITC-labelled anti-NNMT, showing presence of NNMT in dopaminergic neurones (cell bodies and projections). B Conventional microscopy of control human cerebellum. Section stained with anti-NNMT and peroxidase-conjugated second antibody. Brown indicates presence of NNMT; blue is due to counterstain. Light positive (brown) staining indicates low levels of NNMT in granular layer neurones. C Conventional microscopy of parkinsonian human cerebellum. Section stained with anti-NNMT and peroxidase-conjugated second antibody. Brown indicates presence of NNMT; blue is due to counterstain. Heavy positive (brown) staining indicates high levels of NNMT in granular layer neurones. D, E In situ hybridization with control human cerebellum, using antisense NNMT probe. Low levels of NNMT mRNA evidenced by low levels of bright purple staining. E shows Purkinje cells in addition to granular layer neurones. F In situ hybridization with parkinsonian human cerebellum using antisense NNMT probe. High level of NNMT mRNA evidenced by high levels of bright purple staining.

Figure 5.

NNMT expression in cerebellar granular layer neurones. A Frequency of control subjects with low, intermediate or high expression. B Frequency of parkinsonian subjects with low, intermediate or high expression. C Comparison of expression in control and parkinsonian subjects; means are shown as solid symbols. Reprinted from Neuroscience Letters, Parsons RB, Smith SW, Waring RH, Williams AC, Ramsden DB, High expression of nicotinamide N-methyltransferase in patients with idiopathic Parkinson's disease, 342:13–16, 2003, with permission from Elsevier.

S-adenosyl methionine (SAM)

SAM is the methyl donor for this and most other important methylation reactions, including that of DNA and RNA. One study showed decreased CSF levels in untreated Parkinson's disease,83 and this was lowered further by treatment.84 The former is presumably due to increased consumption perhaps from N-methylation and the latter from methylation of dopamine by catechol-O-methyl transferase (COMT). Forcing dopamine catabolism down the methylation path by blocking decarboxylation and MAO-B, particularly in patients with the high methylator COMT polymorphism, would consume SAM, reducing the amount available for other methylation reactions. This may explain some of the beneficial effects of Levodopa, which have always been complex and on several timescales not necessarily all explained by dopamine replacement. Nor are side-effects such as dyskinesia fully explained, and it is possible that a de-methylation/re-methylation injury is to blame, which is the reason given when infants who are vitamin-B12-deficient and are treated with this indirect methyl donor develop dyskinesia. Blocking COMT using inhibitors now widely available may therefore have unforeseen consequences, by boosting SAM, which can cross the blood-brain barrier. This is particularly true for Tolcapone, which unlike Entacapone inhibits COMT centrally, and therefore would increase SAM directly in the brain.85,,86 This central effect might explain its better immediate symptomatic response, but may cause longer-term toxicity. Arguing strongly for the methylation hypothesis, increasing SAM in striatum in animal models causes MPP+-like toxicity.87 Trials comparing disease progression with SAM-reducing, SAM-neutral and SAM-increasing strategies would be of interest. Agents that reduce its synthesis from methionine or increase its consumption or catabolism toward cysteine might need to be developed.

Known environmental risk factors

Would this fit with the other known risk factors for Parkinson's disease? Age is the greatest risk factor, and slow poisoning is therefore an attractive mechanism. There may be some brakes on the age increase, or it would be exponential, e.g. NNMT activity may decline with age.70 The strongest environmental factors are nicotine and caffeine, which are both protective.88–95 No convincing mechanism for this has been described, but both are N-methylated compounds, which seems an extraordinary coincidence (Figure 6). As Paracelsus first predicted, causes and cures may be closely related structurally. Nicotine is already N-methylated but has another potential site, and is a substrate (and therefore may act as a inhibitor) of NNMT. Caffeine is already N-methylated at three sites, but given that demethylation is its normal catabolic event, it may get re-methylated by NNMT and also then act as a substrate, or it may be an inhibitor in its own right.

Figure 6.

Structural formulae of nicotine and caffeine.

Studies on dietary nicotinamide intake in Parkinson's are confusing.96,,97 It has been proposed that high nicotinamide in diet is protective, but this has been disputed and is confounded by the fact that coffee has a high nicotinamide component. Parkinson's disease is rare in alcoholics who are at risk of nicotinamide deficiency. Pellagra is commoner in women, probably because oestrogen suppresses tryptophan metabolism, which is an important source of NADH in liver. Also, their dietary intake of nicotinamide is lower than in men.98 Parkinson's disease, on the other hand, is commoner in men.97 Parkinson's disease is also rare in regions such as Africa where pellagra still occurs, even though Black Americans get PD at the same rate as White Americans within a few generations.99 The epidemiological data on nicotinamide dietary dose is therefore conflicting, and could fit with a ratio of pyridine intake that is skewed toward protoxic rather than protectant compounds. At an individual level, any safe intake may well differ between high and low methylators, and there may be a window with toxicity at the extremes. Thus the epidemiology may not be explicable until we can either phenotype or genotype patients at the same time as making a complex assessment of dietary intake of most of the relevant compounds. Clearly, high methylators may need less protoxin to cause damage. They also may prove to have different smoking or coffee-drinking habits, as the increased methylation may be either unpleasant and lead to avoidance, or need a higher dose to get the pleasant and addictive effects. Indeed, the interaction between nicotine and NNMT should be investigated separately to understand its role in addiction.

Genetic risk factors

There has been immense recent interest in the genetics of the rare Mendelian versions of Parkinson's disease. These discoveries have highlighted the role of proteins such as α-synuclein, Parkin, UCHL1, DJ1 and PINK 1.100–107 The mechanism in these families is not known, but the lead hypothesis is that they interfere with the ubiquitin-proteasome pathway, or with other components of cellular protective mechanisms, including Complex 1 and free radical defence mechanism pathways that had already been implicated in Parkinson's disease. Ubiquitination is important in the removal of already damaged proteins, and is ATP-dependent, so could be impaired secondary to energy failure or as a primary event in these families. What has not been explained is what threatens the cell in the first place. Even in these families, a MPP+ like compound could be the trigger, although needing a lower dose of toxin than in idiopathic Parkinson's disease, perhaps explaining their earlier onset of clinical disease. Certainly α-synuclein knockout mice are resistant to MPP+,108,,109 and inhibition of Complex 1 makes cells sensitive to MPP+,110,,111 as are models of PARKIN mutations;112 one would predict the same with DJ1 and PINK 1 equivalents.113 Toxic exposure can cause upregulation of α-synuclein.114 Cause and effect can easily get confused here, and potential feedback loops from a network of biochemical changes may well be present, emphasizing the need to move proximally for the greatest chance of successful therapeutic intervention. Parkinson's disease may have a myriad of rare causes and common risk and protective factors, but they may all be affecting a single pathway. The methylation hypothesis is inclusive, in that N-methylation may be at the head of a toxico-biochemical pathway that later includes these proteins and mechanisms of cell death.


Hypersusceptibility to chemicals may be a cause of degenerative disease such as Parkinson's disease, and may be happening at the xenobiotic enzyme level.115–117 Both high exposure to protoxic pyridines relative to protectant compounds, and genetic predisposition in the form of high N-methylating capacity, are necessary to cause disease. Consequently, most patients will not have a family history. We suggest that a toxic brew of a variety of N-methyl compounds is a likely scenario, defining at least at first the selective death of dopaminergic neurones. Details of the mix would vary between different individuals, depending on the degree and nature of the exposure. This may explain phenotypic variation, including toxicity outside traditional MPP+ territory, as some of the toxins may not be so selective.118 A logical strategy would be either to reduce environmental exposure to protoxin or to increase exposure to protectants, either in the whole population or in individuals at risk. However, that may not be simple, as a glance at Figure 2 demonstrates. The case of nicotinamide-NNMT is an example of a two-edged sword, because cells may benefit from more of the protectant, nicotinamide, but be damaged by increased levels of toxic N-methylnicotinamide.

High exposure to its substrates, particularly nicotinamide to which the Western population has become increasingly exposed may induce the enzyme, as will stress. This could be the first of several increasingly toxic vicious circles that result from the intra-neuronal deficiency of the vitamin or its downstream direct and indirect products, such as NADH and ATP combined with toxicity from several N-methylated compounds. If true, we are dealing with upstream events and switching them off would not just affect one part of a late cascade of biochemical events or be purely symptomatic, or be only aimed at the dopaminergic damage. Increased irreversible catabolism of nicotinamide may lead to the cellular deficiency of the vitamin and of products, which would also affect non-dopaminergic cells as seen in pellagra. Here dementia and depression are the predominant features, but autonomic, pontine, anosmic and sleep disturbances are seen, and remarkably, also parkinsonism in 10–20% of cases.119 One nicotinamide anti-metabolite, 3-acetyl pyridine, causes atypical parkinsonism, and has been suggested as a model of olivopontocerebellar atrophy.120 In view of the fact that dementia and some of these other atypical features of Parkinson's disease occur in these circumstances, one wonders whether the clinical pathological spectrum of parkinsonism, and not just classic Parkinson's disease, is a hybrid of pyridinium ion poisoning and pellagra.

Altering nicotinamide in the diet as a potential protective manoeuvre may work, but the dose may need to be individualized, and N-methyl compounds would still be produced, including N-methylated nicotinamide. Western societies who suffer more from PD may now have too much nicotinamide in their diet overall, and this may need to be addressed at the population level. Strategies to reduce methylation by removing the co-factor SAM using existing drugs or the development of SAM-depleting agents would be a novel avenue to explore, but would affect other necessary methylation reactions. Inhibition of the enzyme NNMT may be more incisive, as it would boost intracellular nicotinamide and reduce or stop the production of any N-methyl compounds. If the enzyme has been induced over a lifetime by excessive exposure to its substrates or other factors, a reduction in such stimuli may not reverse its overactivity quickly or completely. Competitive inhibitors could be developed, because they are often structurally related to the known substrates. Two lead compounds that may be protective are nicotine and caffeine. Neither may be practical, because they have their own toxicity. Nevertheless, they, along with the structures of nicotinamide or 4-phenyl pyridine may provide clues to the development of a safe pyridine, whether natural or synthetic, leading to attempts at primary and secondary prevention based on a strong hypothesis supported by both biochemical and epidemiological evidence. Symptomatic therapies have obvious limitations, but so does targeting downstream biochemical events, as has been learnt, often painfully, from cancer and inflammatory diseases. XEIs alone or dietary or SAM manipulation may prove to be a revolutionary new class of preventive therapeutic agents, not just for Parkinson's disease. One example of such a possibility is overactive xenobiotic sulphur methylation pathway, which has been implicated in motor neurone disease.121,,122 Given that metal123 or pesticide exposure124,,125 may be a risk factor for MND, and S-methylation of some compounds increases their toxicity, a very similar ecogenetic scenario can be postulated. MPTP was discovered by a freak event in drug users; MND may never have had the equivalent lucky break. One can make similar arguments for some other diseases, including cancer and processes linked with increasing age. Indeed, more varied diets may inadvertently reduce xenobiotic load, perhaps explaining the recent increase in longevity not wholly due to improvements in healthcare. Dietary restriction has long been known to increase life-span and reduce age related disease in experimental animals, and is attributed to lower caloric intake,126 but might also be related to lower daily xenobiotic exposure.


View Abstract