Genetics of Nervous System Disease
NeuroReport 8, 683–686 (1997)
1
WE examined a deletion/insertion promoter polymorphism of the serotonin transporter gene, which confers
an ~40% reduction in expression of the protein, in 196
subjects with late onset Alzheimer’s disease (AD) and 271
controls. The frequency of the 484 bp low activity allele
was elevated in the subjects with AD (p = 0.004), and an
excess of the low activity genotype (30%) was also found
in comparison with the controls (20%) (x2 = 7.16; p =
0.03). This association was unrelated to the age of the
subjects or controls, or to e4 alleles of the ApoE gene.
The odds ratio for the effect of the homozygous low
activity genotype was 1.7 (95% CI 1.08–2.67), with a
population attributable risk of 33% (95% CI 5–54%).
These findings indicate that the low activity allele of the
serotonin transporter is a risk factor for late onset AD.
Allelic functional
variation of serotonin
transporter expression is
a susceptibility factor for
late onset Alzheimer’s
disease
Tao Li,1,6 Clive Holmes,2,4
Pak C. Sham,3 Homero Vallada,1
Joe Birkett,1 George Kirov,1
K. Peter Lesch,7 John Powell,1,2
Simon Lovestone4 and David Collier1,5,CA
1
1
Section of Molecular Genetics, Department of
Psychological Medicine; 2Department of
Neuroscience; 3Section of Genetic
Epidemiology, Department of Psychological
Medicine; 4Section of Old Age Psychiatry;
5
Department of Neuropathology, The Institute
of Psychiatry, De Crespigny Park, Denmark Hill,
London SE5 8AF, UK; 6Department of
Psychiatry Research, West China University of
Medical Science, Chengdu, Sichuan, China
610041; 7Department of Psychiatry, University
of Würzburg Füchsleinstr. 15, D-97080
Würzburg, Germany
1
Key words: Allelic association; Dementia; Depression;
Gene; 5-HTT; Late onset Alzheimer’s disease; SERT
1
1
p
Introduction
A number of studies have determined a major role
for genetic factors in the development of early onset
familial Alzheimer’s disease (AD).1 These include the
finding of highly penetrant autosomal dominant
mutations in the gene coding for amyloid-associated
protein (APP) and, more recently, in the presenilin
1 (PS1) and presenilin 2 genes (PS2) with age at onset
typically in the fourth and fifth decades. In contrast,
late onset AD, defined as age at onset after 60 years,
is a common complex disorder with a multifactorial
and polygenic aetiology and accounts for up to 95%
of all cases of AD. Carriers of the e4 allele of the
ApoE gene2 have increased relative risk of developing
the disease.
Although ApoE allele accounts for a large proportion of the genetic variance for late onset AD, it is
© Rapid Science Publishers
CA
Corresponding Author
probable that additional genetic risk factors of
moderate or minor effect exist. Both post-mortem
and biopsy studies indicate that significant changes
occur in the serotonergic system, including a loss of
neurones and formation of neurofibrillary tangles and
senile plaques in the dorsal and medial raphe.3
Reductions in the number of serotonin transporter
(5-HTT) reuptake sites have been demonstrated in
the dorsal raphe nucleus, entorhinal cortex,
hippocampus4–6 and platelets of subjects with late
onset AD,7–9 the latter indicating that this effect is
not secondary to the disease process.
We analysed two genetic variants of the 5-HTT
gene (SLC6A4), a deletion/insertion polymorphism
in the promoter (5-HTTLPR) and a VNTR in intron
2 (5-HTT-VNTR), in 271 control subjects (including
a subset of old-age controls) and 196 patients with
late onset AD. The 5-HTTLPR is associated with
Vol 8 No 3 10 February 1997
683
T. Li et al
1
11
reduced transcription of the 5-HTT gene, leading to
an ~40% reduction in the density of 5-HTT reuptake sites.10,11 The 5-HTTLPR deletion/insertion
polymorphism also results in different transcriptional
modulation of the 5-HTT gene promoter by cAMP
and protein kinase C.12 Since the two polymorphisms
have also been implicated as risk factors for affective
disorders11,13 we also analysed the genotypes and
allele frequencies with respect to depressive symptomatology in late onset AD.
Subjects and Methods
11
11
11
11
1p
Clinical sample: All individuals with AD who were
over 60 years old at the onset of their illness and
whose next of kin consented to venepuncture and
genetic testing were selected from the Camberwell
Dementia Case Register.14 Of these subjects, 131
fulfilled the National Institute of Neurological and
Communicative Disorders and Stroke and Alzheimer
Disease and Related Disorders (NINCDS-ADRDA)
diagnostic criteria15 for probable AD and 65 fulfilled
criteria for possible AD. The mean age for this group
was 82.5 years (s.d. 6.7). These subjects did not differ
from the entire group (528 subjects) on the register
fulfilling these same criteria in terms of age at interview, age at onset, duration of illness, sex distribution or family history. The next of kin or main carer
was interviewed in all cases using the CAMDEX
informant questionnaire16 with additional questions
from the Present Behavioural Examination.17 Five
questions pertaining to depressive symptomatology
were asked of the informant and the responses noted
as symptom present or absent during the subject’s
illness. A positive response to any one of these questions was considered as evidence for depressive symptomatology in the subject with AD. Subjects were
also analysed for depressive symptomatology using
the Cornell scale.18 The control group of 271
Caucasian subjects from SE England was derived
from sequential attendees at a SE London GP surgery
who had no history of major psychiatric or neurological illness (n = 104), healthy volunteers at the
Clinical Age Research Unit (CARU), King’s College
Hospital (n = 71), patients attending the department
of haematology at King’s College, London (n = 42),
and population controls collected by the Institute of
Psychiatry from a variety of sources (n = 54). The
subset of elderly controls from the Camberwell
area of London comprised 112 subjects either from
attendees at the GP surgery (n = 41) or from CARU
(n = 71) and had a mean age of 70.4 (s.d. 8.5) years.
These controls subjects had no history of major
mental or neurological illness as determined by a
general health screen and case note examination by
684 Vol 8 No 3 10 February 1997
an experienced physician or psychiatrist.
Genotyping: DNA was prepared from lymphocytes
by a standard procedure (Nucleon II kit, Scotlab, UK)
and diluted to a stock concentration of 25 ng ml–1
for polymerase chain reaction (PCR). The VNTR in
intron 2 the serotonin transporter were analysed as
described.13 Primer sequences were GTCAGTATCAACAGGCTGCGAG and TGTTCCTAGTCTTACGCCAGTG. Amplification products consisted of three alleles of 250 base pairs (9 repeats), 267
base pairs (10 repeats) and 300 base pairs (12 repeats).
The 5-HTTLPR polymorphism was amplified as
described11,13 with primer sequences GGCGTTGCCGCTCTGAATTGC and GAGGGACTGAGCTGGACAACCAC of the human 5-HTT regulatory
region which generated a 484/528 bp fragment. Alleles
were designated as 484 bp (short form, low activity)
and 528 bp (long form, high activity). ApoE genotyping was carried out using one-stage PCR followed
by restriction enzyme digestion.19
Statistical analysis: Statistical analysis consisted of
x2 tests for Hardy-Weinberg equilibrium in the
patient and control groups, x2 tests for homogeneity
of genotypic and allelic frequencies between patients
and control groups, and x2 tests of homogeneity of
genotypic and allelic frequencies between subjects
with AD and controls. Odds ratios (OR) with 95%
confidence intervals (CI) were estimated for the
effects of high-risk genotypes and alleles. Attributable risk (AR) was calculated assuming a 20% population exposure to high-risk genotype and a 10%
prevalence of AD over age 65 years. Analysis for
interaction between the 5-HTTLPR and the e4 allele
of ApoE was performed using the EH program.20
Results
Results of genotype-wise and allele-wise analysis
of the 5-HTT-VNTR and the 5-HTTLPR polymorphisms in patients with late onset AD and controls
are shown in table 1. A significant increase in the
frequency of the reduced activity allele (484 bp)
(x2 = 7.37; p = 0.004) and a significant excess of the
484 bp genotype was detected (x2 = 7.16; p = 0.03).
Calculation of OR indicated that subjects with the
484 bp genotype were 1.7 times more likely to
develop AD (95% CI = 1.08–2.67), with a population
attributable risk (AR) of 33% (95 CI 5–54%). Since
our case and control samples were not matched for
age, a potential confounding variable would occur if
there was preferential survival of subjects with the
low activity allele of the 5-HTTLPR to old age.
Serotonin reuptake activity, however, appears robust
to ageing21 and furthermore, in a subset of 112
Low activity of the 5-HTT gene and Alzheimer’s disease
Table 1. Allele and genotype frequencies for the 5/HTT and 5-HTTLPR in AD patients and controls
5-HTT-VTNR
10/10
All controls
AD
Probable AD
Possible AD
AD, depressed
AD, not depressed
1
1
1
1
58
37
27
10
16
21
5-HTTLPR
10/12
12/12
10
5-HTT VTNR
AD vs controls by genotype, x2 = 2.14, p = 0.34
AD vs controls by allele (one-tailed) x2 = 35 (n.s.)
Probable AD vs possible AD by genotype, x2 = 0.60 (n.s.)
Probable AD vs possible AD by allele (two-tailed) x2 = 0.63
(n.s.)
Depressed vs not depressed by allele (one-tailed), x2 = 2.2,
p = 0.07
Depressed vs not depressed by genotype x2 = 2.65, p = 0.27
controls of average age >60 years (s.d. 8.5, mean 70.4
years) with no history of major neurological or
mental illness, the genotype (484 bp, 21%; 484 bp/528
bp, 46%; 528 bp, 33%) and allele frequencies (484
bp, 45%; 528 bp, 55%) were very similar to those of
the total sample of control subjects. In addition we
are confident that our finding is not an artefact of
population stratification in the control sample since
allele and genotype frequencies in a separate sample
of 885 (European Caucasian and American Caucasian
population controls11 were very similar (42% for the
low activity allele and 18% for the homozygous low
activity genotype) to our control sample. There were
no significant differences between patients with late
onset AD and controls for the 5-HTT-VNTR polymorphism, even if the AD sample was divided into
probable and possible (Table 1).
All genotypes were in Hardy-Weinberg equilibrium and there were no differences between probable and possible late onset AD genotypes for both
markers. No significant association was found
between the ApoE e4 allele and the 5-HTTLPR polymorphism (x2 = 0.27), indicating that they act independently (Table 2). There is recent evidence
that the specificity of autopsy-confirmed diagnoses
in clinical patients with AD who carry an ApoE
e4 allele can be improved from < 85% to > 99%.22
p
ApoE allele
–e4
–e4
+e4
+e4
5-HTTLPR allele
528
484
528
484
484/484
484/528
(0.21) 126 (0.47) 87 (0.32) 242 (0.45) 300 (0.55) 52 (0.20) 122
(0.23) 64 (0.40) 61 (0.38) 138 (0.43) 186 (0.57) *59 (0.30)* 90
(0.25) 43 (0.39) 40 (0.36) 97 (0.44) 123 (0.56) 36 (0.28) 64
(0.19) 21 (0.41) 21 (0.40) 41 (0.39) 63 (0.61) 23 (0.35) 26
(0.21) 27 (0.35) 34 (0.44) 59 (0.38) 95 (0.62) 31 (0.30) 37
(0.25) 37 (0.43) 27 (0.32) 79 (0.46) 91 (0.54) 26 (0.28) 53
Table 2. Analysis of the interaction between e4 alleles of
ApoE and the 5-HTTLPR
1
12
bp
bp
bp
bp
Haplotype frequency
Expected
Observed
0.33
0.37
0.14
0.16
0.32
0.38
0.15
0.15
(0.48)
(0.46)
(0.49)
(0.40)
(0.46)
(0.49)
528/528
83
47
31
16
27
20
(0.32)
(0.24)
(0.24)
(0.25)
(0.24)
(0.24)
484
226
208
136
72
103
105
528
(0.44) 288
(0.53)** 184
(0.52) 126
(0.55)
58
(0.53)
91
(0.53)
93
(0.56)
(0.47)
(0.48)
(0.45)
(0.47)
(0.47)
5-HTTPLR
*AD vs controls by genotype, x2 = 7.16, p = 0.03, OR = 1.7
(95% CI = 1.08–2.67)
**AD vs controls by allele (one-tailed) x2 = 7.37, p = 0.004
Probable AD vs possible AD by genotype, x2 = 1.66 (n.s.)
Probable AD vs possible AD by allele (two-tailed) x2 = 0.42 (n.s.)
Depressed vs not depressed by genotype (one-tailed), x2 = 0
(n.s.)
Depressed vs not depressed by allele x2 = 4.7, p = 0.1
Consequently we examined the frequencies of the
5-HTTLPR in ApoE e4-positive cases of late onset
AD only. In this group (n = 60) the frequency of
the 484 bp low activity allele (54%) and genotype
(30%) were essentially identical to the frequencies
found in the overall sample, indicating that the association we detect does not come from a misdiagnosed
subgroup of the sample.
We next analysed the genotype and allele frequencies for association with depressive symptomatology
in late onset AD. There were no significant differences in genotype frequencies between patients with
late onset AD or without depressive symptomatology
as determined by the depression questionnaire,
although there was a non-significant excess of the
484 bp genotype in depressed patients with AD (x2
= 4.7; p = 0.1). Analysis by allele gave no significance
for the 5-HTTLPR relative to depression in LOAD
but borderline significance was detected for depressed
vs not depressed for allele 12 of the 5-HTT-VNTR
(x2 = 22.2; p = 0.07). Analysis using the Cornell scale
for depression in demented subjects gave similar
indices for depression for alleles of the 5-HTTLPR
(484 bp allele, 4.2 ± 4.5; 528 bp allele, 4.2 ± 4.7) or
the VNTR (allele 12, 4.2 ± 4.5; allele 10, 4.6 ± 4.9)
indicating that neither of these polymorphisms are
clearly associated with depressive symptomatology in
late onset AD.
Discussion
We propose that the low activity form of the
5-HTTLPR is a genetic susceptibility factor for late
onset AD. This finding is supported by biochemical
evidence for reduced serotonin reuptake in patients
with late onset AD4–9 and the expression of 5-HTT
in regions of the brain implicated in the pathoVol 8 No 3 10 February 1997
685
T. Li et al
1
11
11
11
11
11
physiology of AD. Although inherited low 5-HTT
activity does not appear to substantially influence
depressive symptomatology in late onset AD, it is
possible that it pleiotropically increases susceptibility
first to affective disorder and then to the later development of late onset AD. This is consistent with the
elevated family history of depression in patients with
late onset AD23 and their increased history of
premorbid depressive illness.24 Depressive symptomatology within late onset AD may have a separate
cause which may be related to neurodegeneration,
although one study has shown an association with
premorbid psychiatric illness.25
The loss of 5-HTT sites, located on 5-HT neurones
and nerve terminals in AD corresponds closely to
the level of raphe cell reduction observed.7 The loss
of presynaptic serotonergic markers, however, including markers of the 5-HTT site, has been most
consistent in the temporal cortex and hippocampus
with inconsistent results or non-significant findings
in other areas.26 The widespread innervation from the
raphe nucleus to these areas indicated that serotonergic loss is retrograde and thus disruption of the
5-HTT site would be expected to precede neuronal
cell body loss. While it is conceivable that an inheritable disruption of the 5-HTT site might instigate
retrograde degeneration over a period of time it is
difficult to see why this should only occur in neurones innervating specific areas such as the temporal
cortex or hippocampus. One possibility is that a
subset of subjects with late onset AD who inherit the
low activity allele do in fact have widespread disruption of 5-HTT binding and that another, as yet
unknown, factor explains localized losses in other
cases. In favour of this hypothesis is one study that
has shown that while patients with AD appear to
show selective 5-HTT site loss in the temporal cortex,
the subgroup of patients with depression showed
losses in all of the areas examined.6
The 5-HTTLPR deletion/insertion polymorphism
not only results in an altered basal level of expression of the gene but in different transcriptional modulation of the 5-HTT gene promoter by cAMP and
protein kinase C,12 indicating that regulation of the
gene is also altered. An attractive hypothesis is that
reduced activity and altered regulation of 5-HTT
affects the common metabolic APP-bA4 pathway by
altering the ratio of secretory APPs to amyloidogenic
bA4.1,27 It is possible to speculate that differences in
the regulation of 5-HTT affects the expression of
serotonin receptors, leading to 5-HT2c and 5-HT2amediated changes in the regulation of amyloidogenic
vs secretory processing of APP.28
1p
686 Vol 8 No 3 10 February 1997
Conclusion
We found an excess of a low activity promoter
polymorphism genotype (30%) of the 5-HTT
gene in patients with late onset AD (n = 196) when
compared with normal controls (n = 271) (20%)
(x2 = 7.16; p = 0.03). The increase in low activity
genotype was unrelated to age or to ApoE e4 genotype. The OR for the homozygous low activity
genotype was 1.7 with a population attributable risk
(AR) of 33%. These findings indicate that the low
activity allele of the serotonin transporter is a genetic
risk factor for late onset AD. Although we found no
association between polymorphisms of the 5-HTT
gene and depressive symptomatology in late onset
AD, in contrast to our previously observed association of this gene with affective disorders, the elevated
history of depression seen in patients with late onset
AD may indicate that the 5-HTT gene is a risk factor
for both types of disorder.
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
Sandbrink R, Hartman T, Masters CL et al. Mol Psychiatry 1, 27–40 (1996).
Corder EH, Saunders AM, Strittmatter WJ et al. Science 263, 921–923 (1993).
Halliday GM, McCann HL, Pamphett R et al. Acta Neuropathol 84, 638–650
(1992).
Bowen DM, Allen SJ, Benton JS et al. J Neurochem 41, 266–272 (1983).
Tejani-Butt SM, Yang J and Pawlyk AC. NeuroReport 6, 1207–1210 (1995).
Chen CP, Alder JT, Bowen DM et al. J Neurochem 66, 1592–1598 (1996).
Koren P, Diver-Haber A, Adunsky A et al. J Gerontol 48, B93–96 (1993).
Inestrosa NC, Alarcon R, Arriagada J et al. Neurosci Lett 163, 8–10 (1993).
Kumar AM, Sevush S, Kumar M et al. Neuropsychobiology 32, 9–12 (1995).
Heils A, Teufel A, Petri S et al. J Neurochem 6, 2612–2624 (1996).
Collier DA, Stober G, Li T et al. Mol Psychiatry 1, 453–460 (1996).
Lesch KP, Heils A and Riederer P. J Mol Med 74, 365–378 (1996).
Collier DA, Arranz MJ, Sham P et al. NeuroReport 7, 1675–1679 (1996).
Holmes C. Int J Ger Psychiatry 11, 369–375 (1996).
McKhann G, Drachman D, Folstein M et al. Neurology 34, 939–944 (1984).
Roth M, Huppert FA, Tyme E et al. CAMDEX: The Cambrige Examination
for Mental Disorders of the Elderly. Cambridge: Cambridge University Press,
1988.
Hope T and Fairburn CG. Psychol Med 22, 223–230 (1992).
Alexopoulos GS, Abrams RC, Young RC et al. Biol Psychiatry 23, 271–84
(1988).
Wenham PR, Price WH and Blundell G. Lancet 37, 1158–1159 (1991).
Terwilliger J and Ott J, eds. Handbook of Genetic Linkage. Baltimore: Johns
Hopkins University Press, 1994.
Andersson A, Sundman I and Marcusson J. Gerontology 38, 127-132 (1992).
Saunders AM, Hewlett C, Welsh-Bohmer KA et al. Lancet 348, 90–93 (1996).
Pearlson GD, Ross CA, Lohr D et al. Am J Psychiatry 147, 452–456 (1990).
Jorm AF, van Duijn CM, Chandra V et al. Int J Epidemiol 20, S43–47 (1991).
Rovner BW, Broadhead J, Spencer M et al. Am J Psychiatry 147, 452–456
(1990).
Francis PT, Sims NR, Procter AW et al. J Neurochem 60, 1589–1604 (1993).
Hardy J and Allsop D. Trends Phamacol Sci 12, 383–388 (1991).
Nitsch RM, Deng M, Growdon JH et al. J Biol Chem 271, 4188–4194 (1996).
ACKNOWLEDGEMENTS: We are grateful to the Psychiatry Research Trust (T.L.,
C.H., S.L.) SmithKline Beecham Pharmaceuticals (D.A.C.) and the Stanley
Foundation (H.P.V.). We are grateful to the HGMP Resource Centre, Hinxton,
UK for the provision of oligonucleotide primers, access to geonome data and
computing resources. We are also grateful to the Clinical Age Research Unit,
King’s College Hospital, London, for old age control samples, to Carsten Russ
for the preparation of DNA and to Gareth Roberts for helpful discussion of the
manuscript.
Received 1 October 1996;
accepted 21 October 1996