Evidence for Genetic Linkage of
Alzheimer’s Disease to
Lars Bertram,1Deborah Blacker,2,3Kristina Mullin,1
Devon Keeney,1Jennifer Jones,1Sanjay Basu,1Stephen Yhu,1
Melvin G. McInnis,4Rodney C. P. Go,5Konstantinos Vekrellis,6
Dennis J. Selkoe,6Aleister J. Saunders,1Rudolph E. Tanzi1*
Recent studies suggest that insulin-degrading enzyme (IDE) in neurons and
microglia degrades A?, the principal component of ?-amyloid and one of the
neuropathological hallmarks of Alzheimer’s disease (AD). We performed para-
metric and nonparametric linkage analyses of seven genetic markers on chro-
mosome 10q, six of which map near the IDE gene, in 435 multiplex AD families.
These analyses revealed significant evidence of linkage for adjacent markers
(D10S1671, D10S583, D10S1710, and D10S566), which was most pronounced
in late-onset families. Furthermore, we found evidence for allele-specific as-
sociation between the putative disease locus and marker D10S583, which has
recently been located within 195 kilobases of the IDE gene.
The deposition and aggregation of ?-amyloid
(A?) in various regions of the brain is one of
the key neuropathological hallmarks of AD.
Consequently, agents that can inhibit and/or
reverse these processes are attractive candi-
date genes for AD. Recent data suggest a
principal role for IDE in the degradation and
clearance of A? secreted by microglial cells
and neurons (1). We performed genetic link-
age analyses with six genetic markers close to
the presumed location of the IDE gene on
chromosome 10q23-q25 in 1426 subjects
from 435 multiplex AD families (2, 3). In
addition, we genotyped marker D10S1225,
which is located 32 to 47 cM proximal to this
region and lies closest to a linkage peak
identified in a recent whole-genome screen
(4, 5) using an overlapping set of families. To
test for genetic linkage, we performed para-
metric [FASTLINK (6, 7)] and nonparamet-
ric [GENEHUNTER-PLUS (8) and ASM (9,
10)] analyses in the sample as a whole and in
subsets stratified by onset age and APOE
Under a dominant model, we found sig-
nificant evidence for linkage around marker
D10S583 (Zmax? 3.3) (Table 1) in the full
sample and for D10S1671 in the late-onset
sample (Zmax? 3.4). Results were similar
under a recessive model, with a maximum
lod score (logarithm of the odds ratio for
linkage) of 3.8 for marker D10S1671 in the
late-onset sample [Web table 1 (12)]. Al-
though linkage was generally more pro-
nounced in families without the APOE ε4/4
genotype, none of the markers had lod
scores ?3 in this stratum [Web table 2
(12)]. Two-point nonparametric linkage
results were consistent with parametric
findings, yielding the strongest signals
for markers D10S1671, D10S583, and
D10S1710 in late-onset families [Web table
3 (12)]. Finally, multipoint nonparametric
analyses (13) generated maximum Z scores
for the likelihood ratio (Zlr) of 1.9 (P ?
0.029, full sample), 2.1 (P ? 0.02, late-
onset), and 2.15 (P ? 0.016, APOE ε4/4-
negative) at marker D10S1710, which lies
between the two markers with the strong-
est two-point signals [Web table 4 (12)].
None of the analyses yielded significant
findings for marker D10S1225, located ?40
cM proximal to the linkage peak reported
here, in contrast to previous reports in an
overlapping sample of National Institute of
Mental Health (NIMH) families (4, 5). In an
effort to understand this difference, we divid-
ed our sample into two groups according to
whether or not all sampled affected individ-
uals were included in the previous reports
(14). Although the distal linkage peak was
more pronounced in families included in the
previous studies (n ? 188), the linkage signal
at marker D10S1225 did not increase [Web
table 5 (12)]. These discrepancies could be
due to a number of factors, including sam-
pling issues [sibling pairs (4, 5) versus full
families used here], inclusion criteria (diag-
nostic and age-of-onset cutoffs), stratification
procedures, and analytic methods.
During the course of our investigation,
public sequence data became available show-
ing that marker D10S583 and IDE are located
on the same ?195-kb bacterial artificial chro-
mosome (15), leading us to test this marker
for allelic association with the disease. As
determined with the Family-Based Associa-
tion Test program (FBAT) (16, 17), the mul-
tiallelic test on all 11 alleles was not signifi-
cant (P ? 0.15), but the diallelic test revealed
significant association of the 211–base pair
allele with AD (nominal P ? 0.004, Bonfer-
roni corrected P ? 0.04). These preliminary
findings suggest that there may be linkage
disequilibrium between D10S583 and the pu-
tative AD locus on chromosome 10q.
Overall, the findings reported here indi-
cate an AD gene on the long arm of chromo-
some 10. It remains unclear whether the peak
reported here between D10S583 (115 cM)
and D10S1671 (127 cM) and the more prox-
imal peak at D10S1225 (81 cM) reported
previously (4, 5) represent linkage to one or
two underlying loci. Recent reports have sug-
gested that chance variation of the location
estimates obtained from linkage studies in
complex diseases can cover as much as 20 to
30 cM or more, even in relatively large sam-
ples (18, 19). The identification of the puta-
tive AD gene(s) on chromosome 10q will
require more detailed studies of linkage dis-
equilibrium to narrow the region of interest as
well as a thorough assessment of candidate
References and Notes
1. K. Vekrellis et al., J. Neurosci. 20, 1657 (2000).
2. Subjects were collected as part of the NIMH Genetics
Initiative following a standardized protocol applying
NINCDS/ADRDA (National Institute of Neurological
and Communicative Disorders and Stroke/Alzhei-
mer’s Disease and Related Disorders Association) cri-
teria for the diagnosis of AD (3). Only families in
which all sampled affected individuals had onset ages
?50 years were included (n ? 435 families, n ?
1426 subjects, mean age of onset ? 72.5 ? 7.7 years,
1Genetics and Aging Unit, Department of Neurology,
2Gerontology Research Unit, Department of Psychia-
try, Massachusetts General Hospital, Harvard Medical
School, Charlestown, MA 02129, USA.3Department
of Epidemiology, Harvard School of Public Health,
Boston, MA 02115, USA.4Department of Psychiatry,
Johns Hopkins University Medical Institutions, Balti-
more, MD 21287, USA.5Department of Epidemiology,
School of Public Health, University of Alabama, Bir-
mingham, AL 35294, USA.
Diseases, Harvard Medical School and Brigham and
Women’s Hospital, Boston, MA 02115, USA.
6Center for Neurologic
*To whom correspondence should be addressed at
Genetics and Aging Unit, Massachusetts General Hos-
pital–East, 149 13th Street, Charlestown, MA 02129,
USA. E-mail: email@example.com
Table 1. Autosomal-dominant model, maximum
two-point parametric lod scores (Zmax), and re-
combination fractions (?). Families were consid-
ered “late-onset” if all sampled affected individu-
als had onset ages ?65 years. Marker locations are
in Kosambi cM according to the Marshfield map.
Values in bold indicate significant linkage.
R E P O R T S
22 DECEMBER 2000VOL 290SCIENCEwww.sciencemag.org
range 50 to 97 years). Additional information on the
procedures is available as supplemental material (12).
3. D. Blacker et al., Neurology 48, 139 (1997).
4. P. Kehoe et al., Hum. Mol. Genet. 8, 237 (1999).
5. A. J. Myers et al., Neurobiol. Aging 21 (suppl.), S103
6. R. W. Cottingham Jr., R. M. Idury, A. A. Schaffer, Am. J.
Hum. Genet. 53, 252 (1993).
7. A. A. Schaffer, S. K. Gupta, K. Shriram, R. W. Cotting-
ham Jr., Hum. Hered. 44, 225 (1994).
8. L. Kruglyak, M. J. Daly, M. P. Reeve-Daly, E. S. Lander,
Am. J. Hum. Genet. 58, 1347 (1996).
9. A. Kong, N. J. Cox, Am. J. Hum. Genet. 61, 1179
10. Version 1.0, applying the exponential model.
11. Disease gene frequencies were set to 0.01 in the dom-
inant model and to 0.05 in the recessive model. Pen-
etrance in affected individuals corresponded to pheno-
copy rates of 5% for definite AD (n ? 278), 10% for
probable AD (n ? 645), and 14% for possible AD (n ?
65). Families were classified as “late-onset” when all
sampled affected individuals had onset ages ?65 years,
as “APOE ?4/4-positive” if at least one affected individ-
ual had the ?4/4 genotype, and as “APOE ?4/4-nega-
affected individuals only.
12. Supplementary data are available at Science Online
13. Intermarker distances are according to marker loca-
tions from Marshfield, except for interval D10S566 to
D10S1671, where 0.7 cM was used according to the
MAP-O-MAT program (http://linkage.rockefeller.edu:
14. The list of individuals included in Myers et al. (5) was
supplied by P. Holmans.
15. GenBank accession number AL356128.
16. D. Rabinowitz, N. Laird, Hum. Hered. 50, 211 (2000).
17. Analyses are based on estimated empirical variances
(to account for the presence of linkage) (20) as
implemented in FBAT (version 1.0, 1999) under an
additive disease model.
18. S. B. Roberts, C. J. MacLean, M. C. Neale, L. J. Eaves,
K. S. Kendler, Am. J. Hum. Genet. 65, 876 (1999).
19. E. R. Hauser, M. Boehnke, Am. J. Hum. Genet. 61,
20. S. L. Lake, D. Blacker, N. M. Laird, Am. J. Hum. Genet.
67, 1515 (2000).
21. We thank T. Moscarillo for analytical assistance.
We thank all families for their participation and the
staff at NIMH and at all three sites for assistance
with all aspects of the project. The NIMH study
involves three research teams in collaboration with
NIMH extramural staff. The principal investigators
(PIs) and co-PIs at each site are as follows: Massa-
chusetts General Hospital: M. S. Albert, R.E.T., and
D.B.; Johns Hopkins University: S. Bassett, G. A.
Chase, and M. F. Folstein; and the University of
Alabama: R.C.P.G. and L. E. Harrell. This work was
sponsored by grants from the NIMH and National
Institute on Aging (NIA) (Alzheimer’s Disease Re-
search Center). L.B. is a fellow of the Deutsche
Forschungsgemeinschaft. A.J.S. is an NIA–National
Research Service Awards recipient.
31 August 2000; accepted 16 November 2000
Linkage of Plasma A?42 to a
Quantitative Locus on
Chromosome 10 in Late-Onset
Alzheimer’s Disease Pedigrees
Nilufer Ertekin-Taner,1Neill Graff-Radford,1Linda H. Younkin,1
Christopher Eckman,1Matthew Baker,1Jennifer Adamson,1
James Ronald,1John Blangero,2Michael Hutton,1*
Steven G. Younkin1
Plasma A?42 (amyloid ?42 peptide) is invariably elevated in early-onset fa-
milial Alzheimer’s disease (AD), and it is also increased in the first-degree
relatives of patients with typical late-onset AD (LOAD). To detect LOAD loci
that increase A?42, we used plasma A?42 as a surrogate trait and performed
linkage analysis on extended AD pedigrees identified through a LOAD patient
a maximal lod score of 3.93 at 81 centimorgans close to D10S1225. Remark-
ably, linkage to the same region was obtained independently in a genome-wide
screen of LOAD sibling pairs. These results provide strong evidence for a novel
LOAD locus on chromosome 10 that acts to increase A?.
The autosomal dominant mutations that
cause early-onset familial AD all increase
A?42 in plasma and brain (1–6). Compared
to age-matched controls, plasma A?42 is
also elevated in the cognitively normal
first-degree relatives and extended families
of patients with typical LOAD (7). To as-
sess the genetic component affecting plas-
ma A?42 levels, we collected 10 LOAD
pedigrees [see Web table 1 for family de-
scription and ascertainment scheme (8)],
used a sandwich enzyme-linked immu-
nosorbent assay (5) to measure plasma
A?42, estimated the heritability of plasma
A?42 using the variance component meth-
od implemented in SOLAR (9), and found
it to be 64.8 ? 15.5% (P ? 0.0001; n ?
Given the association of elevated plas-
ma A?42 with AD, the substantial herita-
bility of this quantitative trait in our LOAD
pedigrees, and the recent successful linkage
of genetic loci to quantitative traits associ-
ated with complex diseases (10–12), we
decided to search for LOAD genes by per-
forming linkage analysis in our LOAD
families using plasma A?42 as a surrogate
trait. Using a traditional affected sibling
pair approach, Kehoe et al. (13) performed
a genome-wide screen for LOAD loci that
identified regions on chromosomes 1, 5, 9,
10, and 19 with multipoint lod (logarithm
of odds for “linkage/no linkage”) scores
(MLSs) ?1. Reasoning that these regions
might contain genes linked to AD because
they elevate A?42, we tested each region
for linkage to plasma A?42.
In previous searches for genes governing
quantitative traits, the power to identify
gene(s) with strong effect (major genes) has
been increased by performing linkage analy-
sis on families ascertained using probands
with extreme values for the quantitative trait
in question (10, 12). For this reason, we
focused our analysis on five families that had
an AD proband with extremely high plasma
A? (top 10% of AD patients). When robust
MLSs for these five families were calculated
using SOLAR (9, 14), the region on chromo-
some 10 gave a maximum MLS of 3.93
(Fig. 1) at 81 centimorgans (cM) between
D10S1227 and D10S1211 (empirical P value
by simulation ? 0.0001). In all other regions,
which were tested both in the extreme fami-
lies and the entire group, the maximum MLS
was ?0.5 [see Web tables 2 and 3 for details
of the analysis (8)]. Because we examined
only 10 families and deliberately weighted
our collection with pedigrees ascertained via
an AD proband with high A? (top 10%), our
results [Web tables 2 and 3 (8)] cannot be
used to evaluate the contribution of the chro-
mosome 10 locus to AD in general.
Here, we focused on A?42 because of
its close association with AD, but we also
performed linkage analysis on the five
“extreme” families using plasma A?40 as
the quantitative trait. In this analysis, the
maximum MLS obtained for the chromo-
some 10 region was 1.36 (point-wise P value
?0.006). All other regions gave maximum
MLSs ? 0.3. This result suggests that the
locus on chromosome 10 may influence both
A?40 and A?42.
There are no obvious candidate genes in
the chomosome 10 region that we identified
(1-lod support interval of ?8 cM), but the
gene for insulin-degrading enzyme (IDE),
which is 30 cM distal to our peak, is consid-
ered by Bertram et al. in their accompanying
1Mayo Clinic Jacksonville, Jacksonville, FL 32224, USA.
2Southwest Foundation for Biomedical Research, San
Antonio, TX 78245, USA.
*To whom correspondence should be addressed. E-
R E P O R T S
www.sciencemag.org SCIENCEVOL 290 22 DECEMBER 2000