Human Molecular Genetics

INTRODUCTION

The butyrylcholinesterase (BCHE) gene contains two common polymorphic variants. One of these polymorphisms, the K-variant (BCHE-K) arises from a G->A substitution at nucleotide 1615 and causes an Ala539Thr missense substitution which reduces the catalytic activity of BCHE enzyme (1). Recently, the BCHE-K variant has been reported to show allelic association with Alzheimer disease (AD) in subjects who are also carriers of the [epsis]4 allele of apolipoprotein E (APOE), especially in subjects over the age of 75 years (2). We have re-examined the frequency of BCHE-K in subjects with late onset sporadic AD or with familial AD (FAD). We have also examined the segregation of two informative polymorphic markers located on chromosome 3 flanking the BCHE locus.

RESULTS

The genotype at the BCHE locus was determined using two independent methods. Initially, we screened for the presence of the BCHE-K variant using the previously published PCR-based amplification-created restriction site method (2,3), and a simpler method based upon allele-specific oligonucleotides (ASOs) to detect the wild-type and BCHE-K alleles. A total of 353 Caucasian subjects were genotyped for both BCHE and for APOE. While the APOE [epsis]4 allele frequency in subjects with AD showed the expected increase in the [epsis]4 allele frequency relative to control subjects (Table 1), there was no difference in the BCHE-K allele frequency between control subjects (n = 165) and subjects with either sporadic AD (n = 138) or unrelated subjects with FAD (n = 50) (Table 1). There was also no significant difference in BCHE-K allele frequency between cases and controls even when each subgroup was stratified either by age (using the mean censoring age of 70 years as the cut-off) or by APOE genotype (Table 2). Power calculations, however, indicate that using the sample sizes in Table 1, we would have at least an 80% power to detect a difference of 7-17% in BCHE-K allele frequency (depending on the actual population frequency) at the 5% significance level. To test for the possibility that an effect of BCHE-K is masked or confounded with APOE genotype, we performed multiple logistic regression analysis including BCHE and APOE genotypes as predictors of disease outcome, adjusting for age and sex (4). While APOE [epsis]4 was found to confer a significant 4.6-fold increase in risk for AD [95% confidence interval (CI) 2.4-8.6], the odds ratio of 1.6 obtained for BCHE-K was not significantly different from one (95% CI: 0.86-3.0). Our data therefore do not confirm an association between BCHE-K and AD, nor do they suggest a synergistic interaction between BCHE-K and the [epsis]4 allele of APOE.

Table 1. Age, sex, APOE and BCHE allele frequencies in AD, FAD and control subjects

Group	No.	Mean	F:M	APOE allele frequency (%)			BCHE allele frequency (%)
		age	ratio	[epsis]2	[epsis]3	[epsis]4	W	K
Alzheimer	138	75.4	1:8	2.3	66.3	31.4	79.7	20.3
FAD	50	67.2	2.2	2.1	48.9	49.0	76.0	18.8
Control	165	63.4	0.8	7.5	78.5	14.0	81.2	18.8

Table 2. BCHE-K allele frequencies in AD and control subjects

Group	[le]70 years		>70 years		APOE [epsis]4+		APOE [epsis]4-
	No.	f(K)	No.	f(K)	No.	f(K)	No.	f(K)
AD	32	28.1	104	18.3	76	19.1	56	30.4
Controls	95	22.1	55	13.6	41	17.1	113	19.9

To examine the possibility that another nearby gene on chromosome 3 might be associated with increased risk for FAD, we examined the segregation of two informative microsatellite markers (D3S1744 and D3S1763) in 56 FAD pedigrees. These markers have been genetically mapped in the order D3S1744-17.7cM-D3S1763 (data obtained from the Genome Database). The BCHE gene has been physically mapped to a radiation hybrid and yeast contig (WC3.20) as D3S3668, which places BCHE at least 1 cM centromeric of D3S1763. Maximum likelihood calculations excluded linkage in our FAD pedigrees to the entire interval between D3S1744 and D3S1763 regardless of whether an affecteds-only model or age- and sex-dependent penetrance model was used to handle disease status information in asymptomatic relatives. No evidence was found for linkage in the presence of heterogeneity (P > 0.5), a result which is consistent with the observation that most of the individual family scores were negative, and none exceeded z = +0.5. Stratification of the data set into groups of families positive or negative for APOE [epsis]4 and into `age of onset' groups did not suggest the possibility of linkage in any subset even after allowing for linkage heterogeneity within these groups. Repetition of these analyses using non-parametric methods also failed to detect significant evidence for allele sharing in affected relatives (maximum NPL statistic = 1.25, P = 0.25).

DISCUSSION

While our data confirm the well-established association between APOE [epsis]4 and AD (5), our results do not support the notion that the K-variant of BCHE is a risk factor for AD either independently or in association with APOE [epsis]4. The explanation for the disparity between our data and the previous report is not entirely clear. It is of note, however, that the BCHE-K allele frequency observed in our sporadic AD cases (0.20) and our unrelated FAD cases (0.19) is very similar to that observed in AD cases in the previous report (0.17) (2). In contrast, the BCHE-K frequency in control subjects in our study (0.19) was considerably higher than that in the control group of the previous report (0.09) (2). It is unlikely that this difference reflects differing ethnic origins of our control and AD populations because the majority of controls were either unaffected spouses of AD subjects or were normal controls drawn from the same communities as the AD patients. Furthermore, two other studies of normal subjects have shown BCHE-K frequencies similar to those reported here (0.13-0.18) (1,3).

One potential explanation for the failure to discover an allelic association between BCHE and AD is that the BCHE-K allele might be in weak linkage disequilibrium with a causative sequence change in another nearby gene. However, the segregation data in our FAD pedigrees using nearby anonymous, informative, microsatellite markers does not support this hypothesis because neither the maximum likelihood nor the non-parametric analyses detect any significant evidence for allele sharing amongst affected relatives. Cumulatively, our data therefore argue for caution in the application of the BCHE-K variant in identification of elderly subjects who are at high risk of developing AD.

MATERIALS AND METHODS

A total of 138 subjects with sporadic AD diagnosed using NINCDS/ADRDA criteria and/or CERAD neuropathological criteria (128 probable AD; 10 definite AD) were collected from clinics specializing in memory disorders in Toronto or Miami (6-8). The 165 normal control subjects who had no evidence of neurologic disease were collected from the same sources, and were generally either spouses of AD-affected subjects, or were subjects specially recruited from the same communities to serve as normal controls.

Fifty six FAD pedigrees with samples from at least three living AD-affected family members, and usually with AD in at least two contiguous generations, were collected from Caucasian families living in Europe or North America. Mutations in PS1 (9), PS2 (10,11) or APP (12-14) previously have been excluded in these pedigrees.

Genotypes at APOE were determined as previously described (5). The genotype at BCHE was determined using the method previously described (3), and also by ASO methods. Briefly, exon 4 of BCHE was amplified by PCR using the primers 1478, 5[prime]-TCA GTT AAT GAA ACA GAT AAA AAT, and 1479, 5[prime]-TAA GTT AAA GAT GTG AGG AAT CAA in a reaction volume of 10 µl containing 100 ng of genomic DNA, 10 pmol of each primer 250 µM dNTPs, 1.5 mM MgCl₂ and 0.5 U of Taq polymerase, for 35 cycles of 94°C for 20 s, 54°C for 20 s and 72°C for 20 s. The 331 bp PCR product was denatured in 100 µl of 0.4 M NaOH, 25 mM EDTA, and slot-blotted to duplicate Hybond-N⁺ nylon membranes. The ASOs 1480, 5[prime]-TAT TGA TGA AGC AGA ATG GGA (wild-type), and 1481, 5[prime]-TAT TGA TGA AAC AGA ATG GGA (BCHE-K) were end labelled and hybridized at 45°C for 2 h in hybridization buffer (5× Denhardt's, 5× SSC, 0.5% SDS), washed to a final stringency of 2× SSC, 0.1% SDS at 58°C, and then exposed to autoradiographic film.

The primer sequences and PCR conditions for the anonymous markers D3S1744 and D3S1763 were obtained from the Genome Database.

A [chi]² test was used to compare BCHE allele frequencies between unrelated AD, FAD and control cases. The Fisher's exact test was used to compare APOE genotypes. The influence of BCHE and APOE genotypes, age group and sex on the odds of developing AD was assessed using logistic regression procedures (4). To accommodate polychotomous classification of the genetic loci in regression analysis, indicator variables were constructed representing two BCHE genotype classes (K/W and K/K) and two APOE genotype classes (2/2 or 2/3; and 2/4, 3/4 or 4/4). These variables took on the value of `1' if the subject had the corresponding genotype, and `0' otherwise. According to this scheme, BCHE-W/W and APOE-3/3 genotypes were considered as the referents. Interaction between BCHE, APOE, age and sex was evaluated by deriving product terms for each genotype with each other, age group and sex. Models were evaluated using the LOGISTIC procedure in SAS (15). Comparisons of the relative fit of hierarchical models were carried out by computing the difference in the -2 ln likelihoods for the models, which follows a [chi]² distribution.

The hypothesis for the existence of an AD gene adjacent to BCHE was evaluated by multilocus linkage analysis in 56 FAD pedigrees. Linkage to the interval between D3S1744 and D3S1763 was tested using the GENEHUNTER program (16). Marker allele frequencies were obtained from the Genome Database. For parametric analyses, AD was modelled as an autosomal dominant trait with a mutant allele frequency of 0.001. Replication using a disease allele frequency of 0.01 gave similar results. Age- and sex-dependent penetrance was defined as a step function based on 17 age intervals derived from censored data analyses for early-onset (family mean onset age <65 years) and late-onset (family mean onset age >65 years) families (17,18). A second set of analyses were carried out by assigning a constant low penetrance (0.02) to all unaffected at-risk individuals. This represents a conservative `affecteds-only' model where unaffected individuals provide minimal information with regard to the disease but are important for linkage phase determination with respect to marker data. The possibility of linkage in the presence of heterogeneity was assessed using Smith's admixture test implemented in the HOMOG program (19). Comparison of the null hypothesis of no linkage (H₀) and the alternate hypothesis of linkage with heterogeneity (H_a; [alpha] <1) for the location scores was carried out as a likelihood ratio test with P values calculated from the asymptotic [chi]² distribution with two degrees of freedom. Families having fewer than two informative meioses were excluded from the test. Linkage data were also evaluated using a non-parametric approach implemented in GENEHUNTER.

ACKNOWLEDGEMENTS

This work was supported through grants from the Medical Research Council of Canada, The Canadian Genetic Diseases Network, The Alzheimer Association of Ontario, The Howard Hughes Medical Research Foundation, the EJLB Foundation, the National Institutes of Health (AG09029), the Peter Burgess Fellowship (E.A.R.), the Alzheimer Society of Canada Fellowship (G.L.), the Helen B. Hunter Fellowship (G.Y.), the National Institutes of Health (T32-AG00115) (S.P.) and the Medical Research Council of Canada (A.P.).