NAT gene polymorphisms and susceptibility to Alzheimer's disease: identification of a novel NAT1 allelic variant

Background Alzheimer's disease is multifactorial, having environmental, toxicological and genetic risk factors. Impaired folate and homocysteine metabolism has been hypothesised to increase risk. In addition to its xenobiotic-metabolising capacity, human arylamine N-acetyltransferase type-1 (NAT1) acetylates the folate catabolite para-aminobenzoylglutamate and is implicated in folate metabolism. The purpose of this study was to determine whether polymorphisms in the human NAT genes influence susceptibility to Alzheimer's disease. Methods Elderly individuals with and without Alzheimer's disease were genotyped at the polymorphic NAT1 (147 cases; 111 controls) and NAT2 (45 cases; 63 controls) loci by polymerase chain reaction-restriction fragment length polymorphism, and the genotype and allele frequencies were compared using the chi-squared test. Results Although a trend towards fast NAT2 acetylator-associated Alzheimer's disease susceptibility was indicated and the NAT1*10/1*10 genotype was observed only in cases of Alzheimer's disease (6/147, 4.1%), no significant difference in the frequency of NAT2 (p = 0.835) or NAT1 (p = 0.371) genotypes was observed between cases and controls. In addition, a novel NAT1 variant, NAT1*11B, was identified. Conclusions These results suggest that genetic polymorphisms in NAT1 and NAT2 do not influence susceptibility to Alzheimer's disease, although the increase in frequency of the NAT1*10 allele in Alzheimer's disease is worthy of further investigation. Due to its similarity with the NAT1*11A allele, NAT1*11B is likely to encode an enzyme with reduced NAT1 activity.

low concordance rate of AD observed in monozygotic twins [7,8] suggests that environmental factors may also influence AD susceptibility. This may be the case particularly for later-onset 'sporadic' cases of the disease, which account for the majority (approximately 90%) of AD cases but show only modest familial clustering [6].
Many putative environmental risk factors for AD have been proposed. In particular, low blood levels of folate and elevated serum total homocysteine have been associated with increased risk of AD [9][10][11]. Differences in the level of serum homocysteine and folate were not associated with increasing duration of the symptoms of AD [10], which suggests that the observed level of these biochemical markers was not due to progression of the disease. Therefore, impaired folate and homocysteine (onecarbon) metabolism has been hypothesised as a risk factor in AD [10]. Such altered folate and homocysteine metabolism could arise as a result of genetic mutations in enzymes of folate metabolism.
The association between impaired folate and homocysteine metabolism and AD is particularly interesting in view of the putative endogenous role of the human phase II xenobiotic-metabolising enzyme arylamine N-acetyltransferase type-1 (NAT1)(and its murine homologue NAT2), in folate catabolism [12][13][14][15], and our previous demonstration of the expression of murine NAT2 in particular cell types, such as the cytoplasm and dendrites of the Purkinjie cells of the cerebellum in the adult mouse brain [16]. There has as yet been no description of the pattern of human NAT1 expression in the adult brain, although human NAT1 (the human equivalent of murine NAT2) activity has been identified in the brain early in embryonic development [17]. Human NAT1 is polymorphic [18] and inter-individual variation in NAT1 activity [19] may modulate individual folate levels. NAT1 may therefore be a potential low penetrance gene which can modify individual risk of AD.
Polymorphisms in the arylamine N-acetyltransferase type-2 (NAT2) gene, which encodes the phase II xenobioticmetabolising isozyme NAT2, have been linked with increased susceptibility to multifactorial neurodegenerative disorders such as Parkinson's disease [20,21]. In this case the impaired ability of the individual to handle environmental xenobiotics or neurotoxins acting on the brain has been hypothesised to contribute to the development of the disease. In a similar manner human NAT2, and in addition human NAT1, in their more traditionally recognized role as phase II xenobiotic-metabolising enzymes [22], may also be modulators of AD risk as a result of chemical insult. An association study carried out by Rocha et al., [23] has previously indicated that human NAT2 may be a potential low penetrance gene in AD pathogen-esis. In view of these findings we undertook a study to genotype an elderly group of individuals with and without a history of AD for major alleles at the NAT1 and NAT2 loci, to investigate the role of NAT1 and NAT2 polymorphisms in AD susceptibility.

Study populations
Genomic DNA from 148 elderly Caucasian individuals with AD and 90 elderly Caucasian individuals without AD, from the Oxford Project to Investigate Memory and Ageing (OPTIMA) cohort, was available for investigation. OPTIMA is a longitudinal study of normal, non-institutionalised elderly volunteers with good cognitive function, and elderly patients with memory problems from the Oxfordshire community (UK), and has been described in detail elsewhere [10,24]. Briefly, the OPTIMA study follows the AD course of an individual, from initial diagnosis to post mortem confirmation. Every year each subject is examined by neurological and neuropsychological tests, brain scans and biochemical analysis of blood and cerebrospinal fluid. Cognitive evaluation is undertaken using the Cambridge Examination for Mental Disorders of the Elderly [25]. This includes the Cambridge Cognitive Examination (CAMCOG), a neuropsychological test which includes elements of the Mini-Mental State Examination and assesses a broad range of cognitive functions, such as memory, language, attention, perception, praxis and thinking. A cut-off value of <80/107 CAMCOG points discriminates between demented and normal subjects. Clinical diagnosis is made according to the National Institute of Neurological and Communicative Disorders and Stroke-Alzheimer's Disease and Related Disorders Association criteria [1]. Post mortem, the brain is examined by histochemical and biochemical methods to confirm disease pathology using Consortium to Establish a Registry for Alzheimer's Disease criteria [26]. In this study the AD case individuals from the OPTIMA cohort included: 101 living patients with a clinical diagnosis of probable or possible AD, 30 patients with histologically confirmed pure AD with the absence of any additional disease pathology, 16 patients with histologically confirmed AD with the coexistence of vascular or Parkinson's disease pathology and 1 patient who demonstrated clinical symptoms of AD which were not confirmed post mortem. The non-demented control individuals from the OPTIMA cohort included: 83 living subjects with no clinically diagnosed symptoms of AD, 4 subjects with the absence of AD confirmed post mortem and 3 subjects who demonstrated no symptoms of AD which was not confirmed post mortem. The OPTIMA study had ethical approval from the central Oxford and Psychiatric Research Ethics Committees and informed consent was obtained in writing from all subjects.
Genomic DNA from 22 elderly volunteers from the Foresight Challenge cohort was also available for investigation. Foresight Challenge is a cohort of normal, noninstitutionalised elderly volunteers with good cognitive function from the Oxfordshire community (UK)(nonblack and Caucasian), recruited for a three-year longitudinal study aimed at further defining early markers or predictors of cognitive impairment and their relationship to the subsequent development of dementia [27]. Participants were excluded if they: (a) scored ≤80/107 points on the CAMCOG or ≤24/30 points on the Mini-Mental State Examinations [25] at the initial screening visit, (b) reported significant progressive subjective memory complaints, (c) lived in institutional care, or (d) were unable to complete the Cambridge Examination for Mental Disorders of the Elderly [25]. For the purpose of this study the individuals in the case and control groups were regarded as homogeneous groups. Subjects with other causes of dementia were excluded from the study.

Materials
All chemicals were purchased from Sigma-Aldrich Company Ltd., Merck Ltd. or BDH Laboratory Supplies, UK. Molecular biology reagents were purchased from Promega, Roche Molecular Biochemicals and New England Biolabs Inc., UK. Oligonucleotides were synthesized by Sigma-Genosys Ltd., UK.

DNA extraction
Genomic DNA from each AD case (n = 148) and control (n = 90) subject in the OPTIMA cohort, and from each control subject (n = 22) in the Foresight Challenge cohort was prepared from the buffy coat fraction of EDTA blood [28] using the Wizard ® Genomic DNA Purification Kit (Promega, UK), and was stored at 4°C in TE buffer (10 mM Tris-HCl (pH 7.5), 1 mM Na 2 EDTA).

Statistical analysis
NAT allele and genotype frequencies for the AD case and control groups were calculated and the NAT genotype distribution for AD case and control groups was tested for Hardy-Weinberg equilibrium. Any deviations from this equilibrium were assessed using a chi-squared test. NAT allele and genotype distributions of the AD case and control groups were compared using the chi-squared test, and p values of less than 0.05 were considered to be significant. The strength of association between inheritance of a particular NAT allele or genotype and AD was assessed by calculation of crude odds ratios (OR) and 95% confidence intervals (CI).

Results
The NAT1 genotype of each individual in the Alzheimer's case group (n = 147) and the control group (n = 111) was determined and the frequency of NAT1 alleles and NAT1 genotypes is illustrated in Table 1 and Table 2 respectively. The distribution of alleles demonstrates that NAT1*4 is the most common allele in both groups (75-79%), whilst NAT1*10 is the next most frequent (15-19%). The same pattern has been observed in all Caucasian populations that have been studied [32]. The genotypes observed were determined to be in Hardy-Weinberg equilibrium. The number and frequency (in brackets) of NAT1 alleles in 147 AD cases and 111 controls is shown. The OR of AD to non-disease is given with 95% CI for each NAT1 allele (with NAT1*4 as the reference group). The number and frequency (in brackets) of NAT1 genotypes in 147 AD cases and 111 controls is shown. Expected genotype frequency was calculated based on the allele frequency in the control group (Table 1). 'Others' includes additional NAT1 genotypes containing the alleles 1*4, 1*3, 1*10, 1*11A and 1*14A expected at low frequency which were not observed in these AD cases and controls. The OR of AD to non-disease is given with 95% CI for each NAT1 genotype (with NAT1*4/NAT1*4 as the reference group).
Although a chi-squared comparison of the two groups being investigated demonstrates that the allele and genotype frequency of the two groups does not differ significantly (χ 2 = 1.84, p = 0.765 and χ 2 = 6.48, p = 0.371 respectively) the number of NAT1*10 alleles is slightly elevated in the Alzheimer's patient population. Comparing the NAT1*10 homozygote genotype frequency, there are 6 out of 147 Alzheimer's cases with this genotype, whilst there are no control individuals with this genotype out of 111 investigated. Although the numbers are small, it is a finding worthy of further investigation, especially since the NAT1*10 allele may have an effect on the level of NAT1 expression [33].
NAT2 genotyping was carried out on a random subgroup of the AD cases (n = 45) and of the controls (n = 63). The frequency of NAT2 alleles and NAT2 genotypes (which were in Hardy-Weinberg equilibrium) is illustrated in Table 3 and Table 4 respectively. NAT2*5B is the most common allele in both groups (39-42%), whilst NAT2*6A is the next most frequent allele (31-36%), as seen in other Caucasian populations that have been studied [32]. A direct chi-squared comparison of the NAT2 allele and genotype frequencies in the Alzheimer's case and control groups indicates that they are not significantly different (χ 2 = 1.95, p = 0.857 and χ 2 = 5.76, p = 0.835 respectively). In a Caucasian population, an individual's NAT2 phenotype can be predicted quite accurately (in more than 95% of cases) by their NAT2 genotype [31]. Therefore the Alzheimer's cases and control individuals were classified as fast or slow acetylators and compared (Table 5). Although a small increase in the frequency of the NAT2 fast acetylator phenotype was observed in the AD cases (38%) compared with the controls (35%), which is in agreement with a similar study carried out in a Portuguese sporadic AD population by Rocha et al. [23], once again no significant difference between the Alzheimer's case and control group was observed (χ 2 = 0.09, p = 0.761). Other studies have also demonstrated a lack of association between the NAT2 acetylator phenotype and risk of AD including those by Steventon et al. [34] and Ladero et al. [35]. However, in these cases it has been suggested that administration of therapeutic agents such as tacrine, an acetylcholinesterase inhibitor used in the treatment of AD, which is thought to be metabolised by CYP1A2 [36,37] may have affected the acetylation of sulphamethazine which was used to phenotype NAT2.

Discussion
It is hypothesized that variation in NAT1 activity may alter the risk of AD (which has been associated with low blood folate levels [9][10][11]), via the postulated role of NAT1 in folate metabolism. Together, the results of the present study suggest that genetic polymorphisms in NAT1 and NAT2 do not influence susceptibility to AD. It should be noted that the study presented here is preliminary and its statistical power is limited due to the relatively small number of samples analysed, particularly the study of NAT2 genotype and risk of AD. Therefore much larger case-control investigations, using more highly automated detection methods such as the LightCycler real-time PCR methods of Blömeke et al. [38] and Wikman et al. [39] will help to validate the results shown here. In order to detect a 1.5 fold increased risk of AD associated with the NAT1*10 allele, approximately 650 cases and 650 controls would need to be studied to give 80% power of achieving significance at the 5% level. Similarly, in order to detect a 1.5 fold increased risk of AD associated with the NAT2*4 allele, approximately 520 cases and 520 controls would need to be studied.
It is hypothesised that expression of a high activity NAT2 enzyme might increase the metabolic activation of environmental compounds (including neurotoxins), contributing to the neuronal tissue degeneration characteristic of AD. However, as no environmental compounds or neurotoxins currently suspected of contributing to the development of AD are known to be metabolised by the NAT enzymes, the identification of endogenous NAT substrates The number and frequency (in brackets) of NAT2 alleles in 45 AD cases and 63 controls is shown. The OR of AD to non-disease is given with 95% CI for each NAT2 allele (with NAT2*4 as the reference group).
or NAT substrates which may be precursors of neurotoxic derivatives, will be important to strengthen the hypothesis that NAT1 and NAT2 may be risk factors for AD as a result of chemical insult.
At present, the suggestion for an endogenous role for human NAT1 has focused on the ability of human NAT1 to acetylate the folate catabolite para-aminobenzoylglutamate [12,13]. However, it is possible that NAT1 may play a role in the metabolism of other, as yet unidentified arylamines. Experiments in which the murine gene equivalent to human NAT1 (murine NAT2) has been knocked out indicate that superficially the mice are well [40]. However, in view of the distribution of murine NAT2 in the nervous system [16], it is important that subtle tests involving behaviour are carried out and the histology of the nervous system is studied in order to determine the role of murine NAT2 in the nervous system and identify whether there might be compensating factors at play in the genetically modified mice. In addition, mice overexpressing the human NAT1 gene show developmental abnormalities [41].
Whilst identifying the NAT1 alleles, a pattern was observed in one individual that was not consistent with any known NAT1 genotype. Only individuals with the NAT1*4 allele generate restriction fragments of 176 bp and 125 bp following PCR amplification of the region corresponding to nucleotides 769 to 1113, and digestion with the restriction enzyme Mbo II. This allele (NAT1*4) is present in the individual indicated by the open arrow in Figure 1(a). However, most other known NAT1 alleles (in which the primer N1113 does not generate an additional Mbo II restriction site) generate fragments of 176 bp and 144 bp, apart from NAT1*11, in which the corresponding fragment is 135 bp due to a 9 base pair deletion. In the case of the individual identified by the open arrow, an additional band of 116 bp was identified which was not The number and frequency (in brackets) of NAT2 genotypes in 45 AD cases and 63 controls is shown. Expected genotype frequency was calculated based on the allele frequency in the control group (Table 3). 'Others' includes additional NAT2 genotypes containing the alleles 2*4, 2*5A, 2*5B, 2*5C, 2*6A and 2*7B expected at low frequency which were not observed in these AD cases and controls. The OR of AD to non-disease is given with 95% CI for each NAT2 genotype (with NAT2*4/NAT2*4 as the reference group). The number and frequency (in brackets) of fast and slow NAT2 acetylators in 45 AD cases and 63 controls is shown. Based on their genotype, AD cases and controls were classified as either 'fast' or 'slow' NAT2 acetylators. Individuals homozygous or heterozygous for the NAT2*4 allele were assigned fast NAT2 acetylators, individuals who carried two 'slow' NAT2 alleles (NAT2*5A, NAT2*5B, NAT2*5C, NAT2*6A and NAT2*7B) in any combination were assigned slow NAT2 acetylators [31]. The OR of AD to non-disease is given with 95% CI for each NAT2 phenotype (with NAT2 fast acetylator as the reference group).
consistent with any of the then-known alleles. To confirm whether this restriction pattern was due to the presence of a novel allele, the NAT1 alleles (a 1.6 Kb region) from the individual were cloned and sequenced. As a result, a novel allele was identified and defined as NAT1*11B (with C at 1095) due to its high sequence similarity with NAT1*11A (with A at 1095) (Figure 1(b)). NAT1*11B contains six mutations with respect to the NAT1*4 sequence: C-344T, A-40T, G445A (V149I), G459A, T640G (S214A), 9 bp deletion in the nucleotide region 1065-1090. The functional consequences of this allele are likely to be similar to that of NAT1*11A where an association with low enzymic activity has been proposed as a result of the amino acid changes V149I and S214A [42,43]. Other changes in the novel allele compared with NAT1*4 are outside the coding region (Figure 1(b)) and their effects have not yet been Identification of a novel NAT1 allele established. Although this allele was identified in an AD patient, it is unlikely to play a significant role in the pathogenesis of AD since it was found in only one individual.

Conclusions
In conclusion, the study presented here, which to our knowledge includes the first known investigation of the association between NAT1 genotype and susceptibility to AD, suggests that genetic polymorphisms in NAT1 and NAT2 do not influence susceptibility to AD. However, the inclusion of SNP analysis at positions 1088 and 1095 in a multi-variant analysis would be warranted based on the present studies since there is an indication that NAT 1*10 is elevated in AD.