Access to this full-text is provided by Wiley.
Content available from Alzheimer's & Dementia
This content is subject to copyright. Terms and conditions apply.
Received: 31 January 2024 Revised: 6 June 2024 Accepted: 18 June 2024
DOI: 10.1002/alz.14128
RESEARCH ARTICLE
Whole-genome sequencing study in Koreans identifies novel
loci for Alzheimer’s disease
Moonil Kang1John J. Farrell1Congcong Zhu1Hyeonseul Park2Sarang Kang3
Eun Hyun Seo3,4Kyu Yeong Choi3,5Gyungah R. Jun1,6,7,8Sungho Won9,10,11
Jungsoo Gim2,3,12,13 KunHoLee
2,3,12,14 Lindsay A. Farrer1,6,7,8,15,16
1Department of Medicine (Biomedical Genetics), Boston University Chobanian & Avedisian School of Medicine, Boston, Massachusetts, USA
2Department of Integrative Biological Sciences, Chosun University, Gwangju, Republic of Korea
3Gwangju Alzheimer’s and Related Dementia (GARD) Cohort Research Center, Chosun University, Dong-gu, Gwangju, Republic of Korea
4Premedical Science, College of Medicine, Chosun University, Dong-gu, Gwangju, Republic of Korea
5Kolab Inc., Dong-gu, Gwangju, Republic of Korea
6Department of Ophthalmology, Boston University Chobanian & Avedisian School of Medicine, Boston, Massachusetts, USA
7Department of Biostatistics, Boston University School of Public Health, Boston, Massachusetts, USA
8Alzheimer’s Disease Research Center, Boston University Chobanian & Avedisian School of Medicine, Boston, Massachusetts, USA
9Institute of Health and Environment, Seoul National University, Gwanak-gu, Seoul, Republic of Korea
10Department of Public Health Sciences, Graduate School of Public Health, Seoul National University, Gwanak-gu, Seoul, Republic of Korea
11RexSoft Corps, Gwanak-gu, Seoul, Republic of Korea
12Department of Biomedical Science, Chosun University, Dong-gu, Gwangju, Republic of Korea
13Well-ageing Medicare Institute, Chosun University, Dong-gu, Gwangju, Republic of Korea
14Korea Brain Research Institute, Dong-gu, Daegu, Republic of Korea
15Department of Neurology, Boston University Chobanian & Avedisian School of Medicine, Boston, Massachusetts, USA
16Department of Epidemiology, Boston University School of Public Health, Boston, Massachusetts, USA
Correspondence
Kun Ho Lee, BK21 FOUR, Department of
Integrative Biological Sciences, Chosun
University, 2, Chosundae 4-gil, Dong-gu,
Gwangju 61452, Republic of Korea.
Email: leekho@chosun.ac.kr
Lindsay A. Farrer, Department of Medicine,
Biomedical Genetics E200, Boston University
Chobanian & Avedisian School of Medicine, 72
East Concord St., Boston, MA 02118, USA.
Email: farrer@bu.edu
Funding information
NIA, Grant/AwardNumbers: U01-AG062602,
R01-AG048927, U01-AG032984,
U01-AG058654, U54-AG052427,
U19-AG068753, U01-AG081230,
Abstract
INTRODUCTION: The genetic basis of Alzheimer’s disease (AD) in Koreans is poorly
understood.
METHODS: We performed an AD genome-wide association study using whole-
genome sequence data from 3540 Koreans (1583 AD cases, 1957 controls) and
single-nucleotide polymorphism array data from 2978 Japanese (1336 AD cases, 1642
controls). Significant findings were evaluated by pathway enrichment and differential
gene expression analysis in brain tissue from controls and AD cases with and without
dementia prior to death.
RESULTS: We identified genome-wide significant associations with APOE in the
total sample and ROCK2 (rs76484417, p=2.71×10−8) among APOE ε4 non-carriers.
A study-wide significant association was found with aggregated rare variants in
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial- NoDerivs License, which permits use and distribution in any
medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.
© 2024 The Author(s). Alzheimer’s & Dementia published by Wiley Periodicals LLC on behalf of Alzheimer’s Association.
8246 wileyonlinelibrary.com/journal/alz Alzheimer’s Dement. 2024;20:8246–8262.
KANG ET AL.8247
P30-AG072978, P30AG10161, R01AG15819,
R01AG17917, R01AG30146, R01AG36042,
RC2AG036547, R01AG36836, R01AG48015,
RF1AG57473, U01AG32984, U01AG46152,
U01AG46161, U01AG61356; Illinois
Department of Public Health; Translational
Genomics Research Institute; KBRI Basic
Research Program, Grant/AwardNumber:
24-BR-03-05
MICALL1 (MICAL like 1) (p=9.04×10−7). Several novel AD-associated genes, including
ROCK2 and MICALL1, were differentially expressed in AD cases compared to controls
(p<3.33×10−3). ROCK2 was also differentially expressed between AD cases with and
without dementia (p=1.34×10−4).
DISCUSSION: Our results provide insight into genetic mechanisms leading to AD and
cognitive resilience in East Asians.
KEYWORDS
Alzheimer’s disease, differential gene expression, gene-based association, genome-wide associa-
tion study, pathway enrichment, whole-genome sequencing
Highlights
∙Novel genome-wide significant associations for AD identified with ROCK2 and
MICALL1.
∙ROCK2 and MICALL1 are differentially expressed between AD cases and controls in
the brain.
∙This is the largest whole-genome-sequence study of AD in an East Asian population.
1INTRODUCTION
The idea that Alzheimer’s disease (AD) has a strong genetic background
is supported by accumulated evidence from familial aggregation and
twin studies.1–4 Although late-onset AD tends to be more multifacto-
rial with genetic and non-genetic components than early-onset forms
developing before age 65 years old,5–7 AD heritability is high, rang-
ing from 60% to 80%.8APOE genotype is a major contributor to the
heritability—the presence of the ε4 allele may account for over half of
the genetic effect in AD.9–11 Currently, APOE and more than 75 loci
have been identified in genome-wide association studies (GWASs) con-
ducted by large consortia.12–17 However, only a portion of the under-
lying genetic basis of AD was explained by the GWAS findings,18–21
and most large-scale AD studies target primarily individuals of Euro-
pean ancestry. Compared to studies of outbred European groups,22
novel AD variants have also been detected in studies of homogeneous
or genetically admixed groups, including several thousand or fewer
samples.23–27
South Korea, one of the most rapidly aging populations in the world,
has a high incidence (21.0 cases per 1000 person-years) and preva-
lence (estimated between 4.2% and 9.0%) of AD among individuals
aged 65 years or older.28,29 As reported bythe Korean Statistical Infor-
mation Service (KOSIS), the rate of the Korean elderly was 7.2% in
2000 (aging society), 14.3% in 2018 (aged society), and expected to
be 20.6% by 2025 (super-aged society). If the current aging growth
rate continues, 19.0 million people (40.1%) of the Korean population
will be 65 years or older by 2050. Like other East Asian countries, the
number of AD cases in Korea is expected to increase enormously in a
few decades.30 On the other hand, the genetic profile of Koreans has
remained distinct from other East Asians,31–34 even considering some
assimilation with their neighboring populations, such as Chinese and
Japanese. Unfortunately, only a few GWASs using variants genotyped
by microarrays, imputed with population reference panels, or called
by next-generation sequencing have included Koreans,35–37 so a large
portion of the genetic basis of AD still needs to be unveiled.
Here, we performed whole-genome sequencing (WGS) of Korean
AD cases and cognitively normal controls over age 60 years and tested
associations of AD with individual variants and aggregated gene-
based rare variants. We also combined the results of single-variant
association tests with GWAS results from two Japanese cohorts by
meta-analysis to validate our findings and identify additional associa-
tions.
2METHODS
2.1 Subject ascertainment and diagnostic
procedures
The Korean sample included participants of the Gwangju Alzheimer’s
and Related Dementias (GARD) Study. In brief, the primary GARD
sample included individuals living in local senior citizens centers in
the southwestern Korean city of Gwangju. These individuals have
been followed longitudinally since 2010. During the same period,
additional persons with AD or mild cognitive impairment (MCI) were
recruited from Chosun University Hospital and its affiliated hospitals
in Gwangju and elsewhere in Korea. All participants completed med-
ical and neuropsychological exams, including the Korean Mini-Mental
State Examination (K-MMSE)38 and the Seoul Neuropsychological
Screening Battery (SNSB), a standardized comprehensive cognitive
exam widely used for Koreans.39,40 A dementia review panel consist-
ing of neurologists and psychiatrists at Chosun University Hospital
8248 KANG ET AL.
and Chonnam National University Hospital determined the latest diag-
nostic status for each participant. The review panel used the updated
guidelines for AD dementia from the National Institute on Aging and
the Alzheimer’s Association (NIA-AA) workgroup41 – subjects who
met the criteria for probable AD dementia were classified into the
AD group in this study. Further details regarding the study design,
recruitment, and examination were provided elsewhere.42
2.2 Whole-genome sequencing and alignment
A total of 3887 blood samples were sent to DNA Link Inc. (Seoul,
Republic of Korea), which extracted genomic DNA from leukocytes in
whole blood using the QuickGene DNA kit S from KURABO Inc. (Osaka,
Japan). WGS was performed to an average depth of 30×with a read
length of 150 bp using the NovaSeq 6000 system from Illumina Inc.
(San Diego, CA, USA), and raw reads were aligned to the human ref-
erence genome version of the Genome Research Consortium human
build 38 (GRCh38) with the Burrow-Wheeler Aligner (BWA) tool for
longer read sequences (BWA-MEM) version 0.7.15 algorithm43,44 and
stored in the compressed reference-oriented alignment map (CRAM)
file format. The BWA-MEM parameter-K 100000000 was used to
achieve deterministic alignment results and parameter-Y to configure
soft-clipping rather than the default hard-clipping of supplementary
alignments. Duplicate reads were marked with the Genome Analysis
Toolkit (GATK) version 4.1.8.1.
2.3 Sample quality control (QC)
To identify potential duplications or sample swaps, we compared
CRAM files for 2937 individuals with single-nucleotide polymorphism
(SNP) data that were generated for those subjects using the Affymetrix
Axiom KORV1.0-96 Array (Santa Clara, CA, USA), a SNP microarray
customized for Koreans.45 Concordance within and between sam-
ples was checked using CrosscheckFingerprints in GATK,46,47 which
enables the comparison of SNP microarray data with alignment map
files before variant calling. Phenotypic and genetic gender information
paired with the CRAM files was also compared with the estimated gen-
der based on genotypes for X and Y chromosome SNPs. The identity
of gender-mismatched samples was confirmed using phenotypic and
APOE genotype data maintained in a separate file.
2.4 Sequence data processing and QC pipeline
We applied the DRAGEN-GATK (release 4.2.6.1) pipeline, a joint vari-
ant calling workflow developed jointly by the Broad Institute and
Illumina Inc. that links features of Dynamic Read Analysis for Genomics
(DRAGEN)48 and GATK.49 The DRAGEN-GATK workflow aligns geno-
types and applies QC filters to SNPs and insertion/deletions (INDELs).
In the first step, the DragSTR tool was applied to each CRAM file to
RESEARCH IN CONTEXT
1. Systematic review: The authors reviewed the liter-
ature reported in traditional (eg, PubMed, published
abstracts) as well as preprinted (eg, medRxiv) sources on
Alzheimer’s disease (AD) genetic association studies and
functional studies of top-ranked genes.
2. Interpretation: We identified a genome-wide significant
association of AD risk in Koreans and Japanese with vari-
ants in ROCK2 among APOE ε4 carriers and MICALL1
(MICAL like 1) in the total study sample. Both of these
novel loci are functionally related to AD pathogenesis
and were differentially expressed between pathologically
confirmed AD cases and controls, particularly for compar-
isons with the subgroup of AD cases who were cognitively
impaired prior to death. Highly suggestive associations
were observed with 14 other novel loci.
3. Future directions: Follow-up studies should confirm the
GWAS findings in larger cohorts. Future research could
also investigate AD-related mechanisms underlying these
associations, as well as the connection of ROCK2 and
MICALL1 to cognitive resilience.
model INDEL errors, which may represent polymerase chain reaction
(PCR)-induced stutters in short tandem repeat (STR) regions. Next, the
HaplotypeCaller tool (with dragen mode turned on) was applied using
the DragSTR model and other DRAGEN features to calculate genotype
calls for INDELs with greater accuracy than GATK.
Next, we applied the GATK site-specific variant quality score recal-
ibration (VQSR) approach to evaluate each genomic position and pass
or filter alleles at that position, regardless of how many alternate alleles
were detected. The allele-specific filtering workflow was then used to
examine each allele separately in the QC annotation, recalibration, and
filtering steps. In this mode, the VariantRecalibrator was implemented
to build the statistical model based on data for each allele rather than
each site. This approach allows the more accurate filtering of false
positive variants (often INDELs) at multi-allelic sites.
In the first step of the allele-specific workflow, QC annotation
was performed with the HaplotypeCaller, and sample genomic vari-
ant call format (gVCF) files were reblocked using the ReblockGVCF
tool with genotype quality (GQ) bands of 10, 20, 30, and 40. Joint
genotyping of the complete set of reblocked gVCFs was performed
using the GenotypeGVCFs tool. Next, SNP and INDEL recalibra-
tion models were trained and applied using GATK VQSR with the
allele-specific mode enabled and standard GATK training resources
(HapMap, Omni, 1000 Genomes, and Mills INDELs). SNP and INDEL
alleles with allele-specific variant quality scores less than 99.7% and
99.0% truth-set sensitivity, respectively, were excluded from analy-
ses. Variants that were monomorphic or that had a read depth <10, a
KANG ET AL.8249
GQ score <20, a missing call rate >20%, or a heterozygous allele bal-
ance estimate <0.25 or >0.75 were also excluded. Quality-controlled
project-level variant call format (pVCF) files used for subsequent anal-
yses included 49,074,372 SNPs, 2,488,596 insertions, and 3,390,751
deletions. APOE genotypes were determined using the apoe-genotyper
tool (https://github.com/Biomedical-Genetics/apoe-genotyper)based
on rs429358 and rs7412 genotypes in the pVCF.
2.5 Principal component analysis and
determination of sample relatedness
SNPs were selected to estimate the kinship coefficient and principal
components (PCs) of ancestry if they had a minor allele frequency
(MAF) >0.05, missing call rate <0.01, and Hardy–Weinberg equilib-
rium (HWE) pvalue >1.0×10−6. SNPs that were pruned based on a
linkage disequilibrium (LD) threshold of 0.1 were used to estimate
the kinship coefficient between samples and PCs with the assistance
of the GENESIS R/Bioconductor package,50 including the KING,51
PC-AiR,52 and PC-Relate53 programs. After excluding 24 duplicate
samples (kinship coefficient >0.49), one gender discordant sample,
individuals who did not meet criteria for AD case or control (includ-
ing 63 MCI and nine classified as having another dementia), and 111
individuals less than age 60 years, 3540 subjects remained for further
analysis.
2.6 Genome-wide association tests
Association of AD with 12,170,786 SNPs and 2,370,352 INDELs that
have a minor allele count (MAC) ≥10 was tested using logistic regres-
sion models, including covariates for age, sex, and the first five PCs.
To account for relatedness among individuals, we built an empiri-
cal genetic relationship matrix (GRM) that was incorporated as a
random effect. Single-variant association tests were also conducted
separately in subgroups stratified by APOE ε4 status or sex. Analy-
ses were performed using GENESIS50 and a genome-wide significance
(GWS) threshold of p=5.0×10−8. The gene-based association was
evaluated for aggregated variants that had MAF <0.01 and were
predicted to have a functional impact. This set included 41,003,908
SNPs and 3,961,251 INDELs, which were categorized into four groups
based on their putative impact determined by SnpEff54 – high-impact
(n=25,057), moderate-impact (n=326,398), low-impact (n=231,743)
variants, and modifiers (n=44,381,961). Two sets of analyses were
conducted using variants with high or moderate impact (20,220 genes)
and high impact only (11,673 genes). Logistic regression models
adjusted for age, sex, GRM, and the first five PCs were evaluated using
the SAIGE-GENE+55,56 R package, implementing an optimal sequence
kernel association test (SKAT-O).57 Each gene-based test was con-
ducted three times by applying MAF cutoffs of <0.0015, 0.003, and
0.01, respectively. Results obtained for multiple MAF cutoffs for each
gene-based method were combined using the Cauchy distribution.58
Study-wide significance (SWS) thresholds of 2.47×10−6for genes with
high or moderate impact and 4.28×10−6for those with high impact
only were applied.
2.7 Enhanced discovery of genetic associations in
an enlarged East Asian dataset
To increase the power for discovery and extension of findings to
other East Asian populations, we combined single-variant test results
for the Korean sample with those obtained from analyses of GWAS
data obtained from 2978 participants of the Japanese Genetic Study
Consortium for Alzheimer’s Disease (JGSCAD).59,60 Details about the
ascertainment and evaluation of AD for the JGSCAD participants were
reported previously.25,59,60 GWAS data for this cohort were included in
previous studies.25,61–63 Subjects in this dataset were genotyped using
the Affymetrix SNP Array 6.0 (Santa Clara, CA, USA) for 1753 sub-
jects (JGSCAD 1) and the Illumina Asian Screening Array version 1.0
(San Diego, CA, USA) for 1225 subjects (JGSCAD 2). A much larger set
of variants was derived by genotype imputation that was performed
using the Trans-Omics for Precision Medicine (TOPMed) reference
panel aligned to GRCh38.64,65 APOE genotypes of the Japanese sub-
jects were determined by constructing haplotypes of the genotyped
variants rs429358 and rs7412. After filtering variants with high impu-
tation quality (r2>0.8), MAF >0.01, missing call rate <0.05, and
HWE p>1.0×10−6, 5,032,643 variants (4,703,376 SNPs and 329,267
INDELs) for JGSCAD 1 and 3,755,677 variants (3,510,239 SNPs and
245,438 INDELs) for JGSCAD 2 remained for analysis. Individual vari-
ant association tests were performed using the same methods and
models described earlier. Results from the Korean and two Japanese
cohorts were combined by meta-analysis using the inverse variance
weighted approach implemented in METAL.66
2.8 Fine-mapping analyses
We performed fine-mapping to localize more precisely the top-ranked
association signals (p<1.0×10−6). Target regions included variants
within 500 kb upstream and downstream of the top variant at each
locus. LD among SNPs was estimated from the Korean WGS data using
PLINK.67,68 We applied a Bayesian method for selecting true causal
variants from a cluster of multiple highly correlated variants using the
sum-of-single-effects (SuSiE) model69,70. This method calculates each
SNP’s posterior inclusion probability (PIP), sorts SNPs into descend-
ing order of their PIPs, and purifies credible sets of variants possibly
underlying the association at each target region by summing their PIPs
until the sum exceeds a specific threshold that provides enough power
in various simulations71. We set the coverage of credible sets to 95%,
meaning that the probability of containing a causal variant in the cred-
ible sets is 95%. We used GWAS summary statistics from each data
stratum corresponding to the top variant at each target region for
fine-mapping.
8250 KANG ET AL.
2.9 Differential expression and pathway
enrichment analyses
We conducted differential gene expression (DGE) analyses for AD
focused on genes containing or adjacent to variants showing associa-
tion at a significant threshold of p=1.0×10−6in the total or stratified
samples for the Korean GWAS or East Asian meta-analyses. These
analyses incorporated the GSE118553 dataset72 obtained from the
Gene Expression Omnibus (GEO)73 website (https://www.ncbi.nlm.
nih.gov/geo) and data generated from Religious Orders Study and
Rush Memory and Aging Project (ROSMAP) participants.74,75 The
GSE118553 dataset included gene expression levels measured in the
entorhinal, temporal, frontal cortex, and cerebellum in brain tissue
from 27 cognitively normal, 33 asymptomatic AD, and 52 AD sub-
jects. The ROSMAP dataset included gene expression levels measured
in the dorsolateral prefrontal cortex from 232 cognitively normal
and 344 AD subjects. Further details of these datasets are described
elsewhere.72,74 We compared gene expression levels in cognitively
normal subjects with AD cases using all subjects within each dataset
and among carriers and non-carriers of the APOE ε4 allele in the
ROSMAP dataset. DGE results were regarded as significant if the
pvalue adjusted for a false discovery rate (FDR) derived by using
the Benjamini–Hochberg method76 was less than a SWS threshold of
p=3.33×10−3.
We also constructed biological pathways using the Ingenuity Path-
way Analysis software (QIAGEN Inc.)77 that were seeded with genes
containing variants associated with AD risk emerging from the total
or stratified samples for the Korean GWAS or meta-analysis of all East
Asian datasets at a significance threshold of p<1.0×10−4. Each canon-
ical pathway’s enrichment pvalues were adjusted for the FDR, and the
significance threshold was set at 0.001.
3RESULTS
In all three East Asian cohorts, AD cases had a lower censored age
(p<.001) and a higher proportion of females (p=.056 for Koreans
and p<.001 for Japanese) compared to subjects with normal cognition
(Table 1). These patterns were consistent among both APOE ε4 carriers
and non-carriers (Table S1).
3.1 Association of AD with individual variants
There was little evidence of genomic inflation (λgc =0.979 to 1.044)
in the total or stratified GWAS of the Korean and Japanese datasets
analyzed individually or combined (Figures S1 to S3). GWS associa-
tions in the combined East Asian datasets were observed with the
APOE isoform SNP rs429358 (MAF =0.20, odds ratio [OR] =2.22,
p=4.31×10−48) and many other variants in the APOE region (Tables 2
and S2). Several variants in the SORL1 region were also associated
with AD at the suggestive significance level (p<1.0×10−6), including
rs80256323 (MAF =0.23, OR =0.77, p=6.28×10−7) in the combined
datasets and rs35751378 (MAF =0.24, OR =0.75, p=5.52×10−7)in
the Korean dataset only (Table 2), and these findings were supported by
associations with adjacent variants (Figure 1A). We identified sugges-
tive associations with four novel AD loci, including MSX1 (rs16836361,
MAF =0.40, OR =0.81, p=5.07×10−7)andCOL9A1 (rs10755539,
MAF =0.02, OR =2.01, p=1.05×10−7) in the combined datasets
(Table 2,Figures1B and C)andLINC01340 (rs202068267, MAF =0.03,
OR =0.46, p=2.06×10−7)andTTC8 (rs17125031, MAF =0.38,
OR =0.78, p=7.13×10−7) in the Korean dataset only (Table 2,Figures
S4A and S4B), which were supported by adjacent variants.
Among APOE ε4 non-carriers, we identified a novel GWS associ-
ation with a ROCK2 variant (rs76484417, MAF =0.06, OR =1.82,
p=2.71×10−8) that was modestly supported by many weak LD vari-
ants spanning ROCK2 and an immediate neighboring gene, SLC66A3
(Table 3and Figure S5A). This association was also primarily female-
specific (OR =1.67, p=7.48×10−7)(Table4and Figure S5B). Further
evaluation revealed a significant association with rs76484417 among
female ε4 non-carriers (OR =2.11, p=1.08×10−7) and to a lesser
extent in male ε4 non-carriers (OR =1.49, p=1.83×10−2)(Table
S3). Additional novel suggestive associations were observed among
APOE ε4 non-carriers that were supported primarily or exclusively
by the Korean sample (Table 3), including LINC02479 (rs1380276,
MAF =0.34, OR =1.35, p=1.58×10−7,Figure1D), LRCH1 (rs1216859,
MAF =0.41, OR =1.41, p=1.93×10−7,FigureS4C), and BCL11A
(rs117361511, MAF =0.01, OR =4.77, p=2.99×10−7,FigureS6A).
One novel suggestive association was observed with an MCTP2 vari-
ant among APOE ε4 carriers in the Korean (rs116991563, MAF =0.03,
OR =3.01, p=7.58×10−7,FigureS6B) but not in Japanese (p>.3)
datasets. Associations with the top LINC02479 and LRCH1 variants
were well supported by many neighboring variants (Figures 1D and
S4C). A suggestive association was observed with a variant in ARMH3
(rs7086516, MAF =0.37, OR =0.71, p=1.71×10−7,Figure1E)inmales
only, and several female-specific associations were also found in addi-
tion to ROCK2 (Table 4), including DLGAP2 (rs74366463, MAF =0.05,
OR =1.89, p=7.11×10−7,Figure1F), TBC1D32 (rs60434653,
MAF =0.17, OR =1.53, p=6.15×10−7,FigureS4D), CTNNA3
(rs577389261, MAF =0.01, OR =8.95, p=9.41×10−7,FigureS4E),
EXD2 (rs10710360, MAF =0.04, OR =2.39, p=2.61×10−7,Figure
S4F), and GRM3 (rs76741056, MAF =0.03, OR =0.40, p=1.28×10−7,
Figure S6C). The findings for TBC1D32,CTNNA3,andEXD2 were
observed in the Korean sample only, noting that the associated variants
at these loci were either too rare or not well imputed in the Japanese
GWAS datasets. In contrast, the results for DLGAP2 and GRM3 were
supported by the Korean and Japanese datasets. The associations with
ARMH3,TBC1D32,CTNNA3,andEXD2 were supported by evidence
from neighboring variants (Figures 1and S4).
3.2 Localization of association signals
Three loci (ROCK2,LINC01340,andEXD2) achieved the desired level
of 95% coverage in credible sets. A GWS ROCK2 variant, rs76484417,
was included in the 95% credible sets among East Asian APOE ε4
KANG ET AL.8251
TAB L E 1 Subject characteristics.
GARD (n=3540) JGSCAD 1 (n=1753) JGSCAD 2 (n=1225)
AD Control Pvalue AD Control Pvalue AD Control pvalue
Participants, n(%) 1583 (44.7) 1957 (55.3) <10−15 898 (51.2) 855 (48.8) 1.56×10−1438 (35.8) 787 (64.2) <10−15
Age, mean ±SD
(years)
75.1 ±6.7 76.0 ±5.1 1.07×10−573.0 ±4.3 76.9 ±6.0 <10−15 74.5 ±5.2 76.7 ±6.7 4.59×10−10
Female, n(%) 914 (57.7) 1066 (54.5) 5.58×10−2653 (72.7) 474 (55.4) 6.53×10−14 305 (69.6) 370 (47.0) 3.76×10−14
APOE ε4 carrier, n(%)810 (51.2) 535 (27.3) <10−15 511 (56.9) 144 (16.8) <10−15 249 (56.8) 137 (17.4) <10−15
APOE genotype, n(%)
ε2/ε23 (0.2) 9 (0.5) <10−15 0 (0.0) 1 (0.1) <10−15 0 (0.0) 0 (0.0) <10−15
ε2/ε386 (5.4) 189 (9.7) 18 (2.0) 66 (7.7) 12 (2.7) 67 (8.5)
ε3/ε3684 (43.2) 1224(62.5) 369 (41.1) 644 (75.3) 177 (40.4) 583 (74.1)
ε2/ε429 (1.8) 32 (1.6) 12 (1.3) 7 (0.8) 8 (1.8) 9 (1.1)
ε3/ε4645 (40.7) 485(24.8) 408 (45.4) 134 (15.7) 189 (43.2) 124 (15.8)
ε4/ε4136 (8.6) 18 (0.9) 91 (10.1) 3 (0.4) 52 (11.9) 4 (0.5)
Abbreviations: AD, Alzheimer’s disease; GARD, Gwangju Alzheimer’s and Related Dementias; JGSCAD,Japanese Genetic Study Consortium for Alzheimer’s
Disease.
FIGURE 1 LocusZoom plots showing association of AD with top-ranked loci in East Asian cohorts, supported primarily by Korean sample. (A)
SORL1 in total sample, (B) MSX1 in total sample, (C) COL9A1 in total sample, (D) LINC02479 among APOE ε4 non-carriers, (E) ARMH3 among males,
(F) DLGAP2 among females. SNPs are color-coded according to their correlation with top-ranked SNPs in region. Dotted line represents
genome-wide significance threshold. Solid blue line indicates recombination rate among SNPs at chromosomal position. Approximate location,
transcription direction, and coding portions (exons represented by vertical bars) of genes in region are shown below plots. AD, Alzheimer’s disease;
cM, centimorgan; Mb, megabase; SNP, single-nucleotide polymorphism.
8252 KANG ET AL.
TAB L E 2 Genome-wide significant (p<5.0×10−8) or suggestive (p<1.0×10−6) associations for AD risk in total sample.
Individual datasets Combined
Chr Position Variant ID A1/A2 Locus Dataset
MAF
AD
MAF
Control
OR
(95% CI) pvalue
OR
(95% CI) pvalue
4 4889006 rs16836361 G/T MSX1/
LOC101928306
GARD 0.33 0.39 0.78 (0.71 to 0.86) 9.23×10−70.81
(0.74 to 0.88)
5.07×10−7
JGSCAD 1 0.47 0.48 0.88 (0.76 to 1.03) 1.13×10−1
JGSCAD 2 NA NA NA NA
597377397 rs202068267 CT/C LINC01340/
RIOK2
GARD 0.02 0.04 0.46 (0.35 to 0.62) 2.06×10−7NA NA
JGSCAD 1 NA NA NA NA
JGSCAD 2 NA NA NA NA
6 70268694 rs10755539 C/T COL9A1 GARD 0.03 0.02 2.13 (1.57 to 2.88) 1.48×10−62.01
(1.55 to 2.60)
1.05×10−7
JGSCAD 1 0.03 0.02 1.76 (1.11 to 2.79) 1.80×10−2
JGSCAD 2 NA NA NA NA
11 121563217 rs80256323 A/C SORL1 GARD 0.21 0.25 0.76 (0.68 to 0.85) 2.15×10−60.77
(0.70 to 0.86)
6.28×10−7
JGSCAD 1 NA NA NA NA
JGSCAD 2 0.19 0.22 0.83 (0.67 to 1.03) 9.41×10−2
11 121580918 rs35751378 CTT/CT SORL1 GARD 0.22 0.27 0.75 (0.67 to 0.84) 5.52×10−7NA NA
JGSCAD 1 NA NA NA NA
JGSCAD 2 NA NA NA NA
14 88945188 rs17125031 G/C TTC8/
FOXN3
GARD 0.35 0.41 0.78 (0.70 to 0.86) 7.13×10−7NA NA
JGSCAD 1 NA NA NA NA
JGSCAD 2 NA NA NA NA
19 44834017 rs142518277 G/GACAA NECTIN2/
BCAM
GARD 0.07 0.03 2.21 (1.78 to 2.74) 1.19×10−12 2.06
(1.67 to 2.52)
5.36×10−12
JGSCAD 1 NA NA NA NA
JGSCAD 2 0.03 0.02 1.32 (0.76 to 2.28) 3.32×10−1
19 44908684 rs429358 C/T APOE GARD 0.30 0.14 2.55 (2.27 to 2.88) 3.83×10−52 2.22
(1.99 to 2.47)
4.31×10−48
JGSCAD 1 NA NA NA NA
JGSCAD 2 0.21 0.15 1.39 (1.11 to 1.73) 4.44×10−3
Note: Genomewide significant and suggestive findings are highlighted in bold.
Abbreviations: A1, minor allele; A2, reference allele; AD, Alzheimer’s disease; GARD, Gwangju Alzheimer’s and Related Dementias; JGSCAD, Japanese Genetic Study Consortium for Alzheimer’s Disease; MAF,
minor allele frequency; NA; not available; OR, odds ratio.
KANG ET AL.8253
TAB L E 3 Genome-wide significant (p<5.0×10−8) or suggestive (p<1.0×10−6) associations for AD risk by APOE ε4 status.
Individual datasets Combined
Chr Position Variant ID A1/A2 Locus APOE ε4 Dataset
MAF
AD
MAF
control OR (95% CI) pvalue
OR
(95% CI) pvalue
2 11262602 rs76484417 T/C ROCK2 Non-
carrier
GARD 0.06 0.04 1.57 (1.17 to 2.11) 3.09×10−31.82
(1.47 to 2.25)
2.71×10−8
JGSCAD 1 0.10 0.05 2.37 (1.63 to 3.46) 7.92×10−6
JGSCAD 2 0.09 0.06 1.70 (1.05 to 2.76) 3.25×10−2
Carrier GARD 0.05 0.04 1.09 (0.75 to 1.57) 6.61×10−11.05
(0.80 to 1.38)
7.30×10−1
JGSCAD 1 0.08 0.06 1.27 (0.76 to 2.13) 3.74×10−1
JGSCAD 2 0.06 0.08 0.73 (0.39 to 1.36) 3.20×10−1
260605771 rs117361511 A/G BCL11A/
PAP OLG
Non-
carrier
GARD 0.02 0.01 4.77 (2.64 to 8.60) 2.99×10−7NA NA
JGSCAD 1 NA NA NA NA
JGSCAD 2 NA NA NA NA
Carrier GARD 0.01 0.01 1.01 (0.51 to 1.97) 9.85×10−1NA NA
JGSCAD 1 NA NA NA NA
JGSCAD 2 NA NA NA NA
4 130594431 rs1380276 C/T LINC02479/
SNHG27
Non-
carrier
GARD 0.38 0.31 1.38 (1.21 to 1.57) 2.46×10−61.35
(1.20 to 1.50)
1.58×10−7
JGSCAD 1 0.38 0.33 1.27 (1.04 to 1.55) 2.00×10−2
JGSCAD 2 NA NA NA NA
Carrier GARD 0.34 0.35 0.94 (0.79 to 1.11) 4.79×10−10.92
(0.80 to 1.07)
2.91×10−1
JGSCAD 1 0.31 0.34 0.88 (0.66 to 1.18) 3.94×10−1
JGSCAD 2 NA NA NA NA
13 46657776 rs1216859 G/A LRCH1 Non-
carrier
GARD 0.46 0.38 1.41 (1.24 to 1.61) 1.93×10−71.24
(1.12 to 1.37)
3.21×10−5
JGSCAD 1 0.42 0.38 1.14 (0.93 to 1.40) 2.02×10−1
JGSCAD 2 0.37 0.40 0.86 (0.67 to 1.10) 2.32×10−1
Carrier GARD 0.40 0.42 0.94 (0.79 to 1.10) 4.33×10−10.94
(0.82 to 1.08)
4.06×10−1
JGSCAD 1 0.39 0.38 1.15 (0.86 to 1.53) 3.39×10−1
JGSCAD 2 0.36 0.44 0.77 (0.55 to 1.07) 1.22×10−1
15 94568744 rs116991563 G/A MCTP2/
LOC440311
Non-
carrier
GARD 0.03 0.03 0.81 (0.57 to 1.15) 2.49×10−10.77
(0.57 to 1.04)
8.90×10−2
JGSCAD 1 0.01 0.02 0.37 (0.17 to 0.79) 1.11×10−2
JGSCAD 2 0.03 0.02 1.26 (0.57 to 2.76) 5.71×10−1
Carrier GARD 0.05 0.01 3.01 (1.96 to 4.62) 7.58×10−72.34
(1.61 to 3.40)
7.21×10−6
JGSCAD 1 0.02 0.01 1.67 (0.63 to 4.45) 3.08×10−1
JGSCAD 2 0.02 0.02 0.58 (0.18 to 1.90) 3.72×10−1
Note: Genomewide and suggestive findings are highlighted in bold.
Abbreviations: A1, minor allele; A2, reference allele; AD, Alzheimer’s disease; GARD, Gwangju Alzheimer’s and Related Dementias; JGSCAD, Japanese Genetic Study Consortium for Alzheimer’s Disease; MAF,
minor allele frequency; NA, not available; OR, odds ratio.
8254 KANG ET AL.
TAB L E 4 Genome-wide significant (p<5.0×10−8) or suggestive (p<1.0×10−6) associations for AD by sex.
Individual datasets Combined
Chr Position Variant ID A1/A2 Locus Sex Dataset
MAF
AD
MAF
Control
OR
(95% CI) pvalue
OR
(95% CI) pvalue
2 11262602 rs76484417 T/C ROCK2 Male GARD 0.05 0.04 1.08 (0.77 to 1.51) 6.70×10−11.14
(0.88 to 1.46)
3.16×10−1
JGSCAD 1 0.08 0.06 1.38 (0.85 to 2.24) 1.92×10−1
JGSCAD 2 0.06 0.07 1.00 (0.55 to 1.80) 9.99×10−1
Female GARD 0.06 0.04 1.58 (1.17 to 2.13) 3.04×10−31.67
(1.36 to 2.04)
7.48×10−7
JGSCAD 1 0.09 0.05 2.01 (1.43 to 2.82) 7.09×10−5
JGSCAD 2 0.08 0.06 1.35 (0.87 to 2.11) 1.84×10−1
6120553561 rs60434653 GAAAC/G TBC1D32/
MIR3144
Male GARD 0.17 0.17 1.00 (0.82 to 1.21) 9.76×10−1NA NA
JGSCAD 1 NA NA NA NA
JGSCAD 2 NA NA NA NA
Female GARD 0.21 0.14 1.53 (1.29 to 1.81) 6.15×10−7NA NA
JGSCAD 1 NA NA NA NA
JGSCAD 2 NA NA NA NA
7 86537844 rs76741056 T/G GRM3/
LINC00972
Male GARD 0.03 0.03 0.76 (0.50 to 1.17) 2.22×10−10.83
(0.56 to 1.24)
3.61×10−1
JGSCAD 1 NA NA NA NA
JGSCAD 2 0.02 0.01 1.57 (0.49 to 5.06) 4.49×10−1
Female GARD 0.02 0.04 0.44 (0.30 to 0.65) 4.77×10−50.40
(0.28 to 0.56)
1.28×10−7
JGSCAD 1 NA NA NA NA
JGSCAD 2 0.01 0.04 0.29 (0.14 to 0.57) 3.90×10−4
81435574 rs74366463 T/C DLGAP2 Male GARD 0.06 0.06 0.97 (0.71 to 1.32) 8.35×10−10.98
(0.73 to 1.32)
9.05×10−1
JGSCAD 1 NA NA NA NA
JGSCAD 2 0.03 0.03 1.11 (0.46 to 2.67) 8.10×10−1
Female GARD 0.08 0.04 1.89 (1.45 to 2.48) 3.75×10−61.89
(1.47 to 2.44)
7.11×10−7
JGSCAD 1 NA NA NA NA
JGSCAD 2 0.03 0.02 1.90 (0.93 to 3.89) 8.05×10−2
10 67560169 rs577389261 T/C CTNNA3 Male GARD 0.01 0.01 1.10 (0.50 to 2.43) 8.16×10−1NA NA
JGSCAD 1 NA NA NA NA
JGSCAD 2 NA NA NA NA
Female GARD 0.01 0.00 8.95 (3.74 to 21.40) 9.41×10−7NA NA
JGSCAD 1 NA NA NA NA
JGSCAD 2 NA NA NA NA
(Continues)
KANG ET AL.8255
TAB L E 4 (Continued)
Individual datasets Combined
Chr Position Variant ID A1/A2 Locus Sex Dataset
MAF
AD
MAF
Control
OR
(95% CI) pvalue
OR
(95% CI) pvalue
10 102045319 rs7086516 T/ C ARMH3 Male GARD 0.31 0.39 0.71 (0.61 to 0.82) 6.68×10−60.71
(0.62 to 0.81)
1.71×10−7
JGSCAD 1 0.37 0.43 0.71 (0.55 to 0.92) 8.39×10−3
JGSCAD 2 NA NA NA NA
Female GARD 0.35 0.35 1.01 (0.89 to 1.16) 8.51×10−11.02
(0.91 to 1.13)
7.62×10−1
JGSCAD 1 0.44 0.44 1.02 (0.86 to 1.22) 8.03×10−1
JGSCAD 2 NA NA NA NA
14 69226773 rs10710360 GTT/GT EXD2 Male GARD 0.03 0.04 0.82 (0.55 to 1.21) 3.32×10−1NA NA
JGSCAD 1 NA NA NA NA
JGSCAD 2 NA NA NA NA
Female GARD 0.05 0.02 2.39 (1.72 to 3.32) 2.61×10−7NA NA
JGSCAD 1 NA NA NA NA
JGSCAD 2 NA NA NA NA
19 44730118 rs62117161 G/A BCL3/
CEACAM16
Male GARD 0.19 0.14 1.38 (1.14 to 1.66) 1.30×10−31.36
(1.17 to 1.60)
9.30×10−5
JGSCAD 1 0.17 0.13 1.43 (1.01 to 2.04) 4.38×10−2
JGSCAD 2 0.16 0.12 1.23 (0.82 to 1.84) 3.21×10−1
Female GARD 0.18 0.15 1.23 (1.04 to 1.47) 1.80×10−21.45
(1.27 to 1.65)
2.37×10−8
JGSCAD 1 0.21 0.11 2.10 (1.66 to 2.65) 8.96×10−10
JGSCAD 2 0.18 0.14 1.36 (1.00 to 1.85) 5.06×10−2
19 44834017 rs142518277 G/GACAA NECTIN2/
BCAM
Male GARD 0.09 0.03 2.58 (1.88 to 3.53) 7.17×10−92.53
(1.87 to 3.44)
1.83×10−9
JGSCAD 1 NA NA NA NA
JGSCAD 2 0.03 0.01 2.13 (0.80 to 5.67) 1.32×10−1
Female GARD 0.07 0.03 1.92 (1.42 to 2.59) 2.09×10−51.72
(1.30 to 2.27)
1.38×10−4
JGSCAD 1 NA NA NA NA
JGSCAD 2 0.03 0.03 1.01 (0.52 to 1.95) 9.79×10−1
19 44908684 rs429358 C/T APOE Male GARD 0.28 0.13 2.50 (2.08 to 3.02) 2.71×10−21 2.23
(1.88 to 2.64)
1.20×10−20
JGSCAD 1 NA NA NA NA
JGSCAD 2 0.20 0.13 1.40 (0.96 to 2.02) 7.96×10−2
Female GARD 0.31 0.15 2.59 (2.22 to 3.04) 3.19×10−32 2.21
(1.92 to 2.55)
8.15×10−29
JGSCAD 1 NA NA NA NA
JGSCAD 2 0.21 0.17 1.35 (1.02 to 1.79) 3.56×10−2
Note: Genomewide and suggestive findings are highlighted in bold.
Abbreviations: A1, minor allele; A2, reference allele; AD, Alzheimer’s disease; GARD, Gwangju Alzheimer’s and Related Dementias; JGSCAD, Japanese Genetic Study Consortium for Alzheimer’s Disease; MAF,
minor allele frequency; NA, not available; OR, odds ratio.
8256 KANG ET AL.
TAB L E 5 MICALL1 variants contributing to the gene-based association with AD.
Chr Position Variant ID A1/A2 Function
MAF
total
MAC
AD
MAC
control OR (95% CI) pvalue
22 37917746 rs751411410 G/C Exonic 0.002 2 13 0.30 (0.09 to 0.93) 1.86×10−2
37922210 rs750390382 T/C Exonic 0.003 2 17 0.26 (0.10 to 0.68) 3.15×10−3
37924666 rs201690599 G/T Exonic 0.003 5 16 0.41 (0.15 to 1.13) 4.19×10−2
37927476 rs181831287 A/G Exonic 0.007 25 23 1.37 (0.78 to 2.42) 2.74×10−1
37940792 rs201526262 T/C Exonic 0.004 3 22 0.29 (0.13 to 0.64) 1.10×10−3
Abbreviations: A1, minor allele; A2, reference allele; AD, Alzheimer’s disease; MAF, minor allele frequency; MAC, minor allele count; OR, odds ratio.
non-carriers (MAF =0.06, OR =1.82, p=2.71×10−8,PIP=0.995)
and East Asian females (MAF =0.06, OR =1.67, p=7.48×10−7,
PIP =0.952) (Table S4). The 95% credible sets generated from Korean
samples included a LINC01340 variant (rs202068267, MAF =0.03,
OR =0.46, p=2.06×10−7,PIP=0.957) from the total sample and an
EXD2 variant (rs10710360, MAF =0.04, OR =2.39, p=2.61×10−7,
PIP =0.955) from females (Table S4). On the other hand, the 90% cred-
ible sets generated from East Asian females included a DLGAP2 variant
(rs74366463, MAF =0.05, OR =1.89, p=7.11×10−7,PIP=0.938).
A well-established APOE variant, rs429358, was also included in the
90% credible sets with PIP =0.913 to 0.932 among the total sample
of Koreans and East Asians.
3.3 Gene-based associations
Gene-based analyses that included ultra-rare (MAC ≤10) and rare
(MAF <0.01) variants in the Korean dataset with high or mod-
erate impact identified a SWS association with MICALL1 (MICAL
like 1) (p=9.04×10−7)(TableS5a). Except for one variant located
at 37,927,476 bp, all MICALL1 variants included in this test have
MAF <0.005andwereprotective(OR=0.26 to 0.41) and nominally
significant (p<.05) (Table 5). Notably, gene-based tests restricted to
only high-impact variants did not reveal any SWS associations with AD
(Table S5B).
3.4 Differentially expressed genes and enriched
pathways
Nine of the top-ranked AD-associated genes, including ROCK2 and
MICALL1, were significantly (p<3.33×10−3) differentially expressed
between pathologically confirmed AD cases and controls, particularly
for comparisons with the subgroup of AD cases who showed clinical
symptoms of AD prior to death (Figure 2and Table S6). In the entorhi-
nal cortex, expression of MSX1,TTC8,andCOL9A1 (Figure 2A–C)was
significantly higher, and expression of GRM3,BCL11A,andCTNNA3
(Figure 2D–F) was significantly lower in symptomatic AD cases com-
pared to asymptomatic AD cases and controls. In the temporal cortex,
expression of ROCK2 was significantly lower in symptomatic AD
cases than in asymptomatic AD cases (p=1.34×10−4) and controls
(p=1.45×10−3), and expression of LRCH1 was progressively lower
in asymptomatic and symptomatic AD cases compared to controls
(Figure 2G–H). Four additional genes were differentially expressed
at a nominal significance level (Table S6). Pathway analyses that were
seeded with genes containing AD-associated (p<1.0×10−4) variants
emerging from analyses of the total or stratified Korean GWASor com-
bined East Asian datasets revealed no significant (p<.001) canonical
pathways. However, several neuronal and cardiac pathways were sig-
nificantly enriched (Table S7), and the top-ranked pathways included
neuropathic pain signaling in dorsal horn neurons, calcium signaling,
neurexins, neuroligins, and glutaminergic receptor signaling.
4DISCUSSION
4.1 Discovery of novel associations for AD in East
Asians
Our East Asian AD GWAS that included WGS data obtained from more
than 3500 Koreans and genome-wide imputed SNP data from two
Japanese cohorts identified GWS associations with APOE in the total
sample and with one novel locus, ROCK2, among persons lacking the
APOE ε4 allele. Suggestive associations with SNPs in SORL1, another
well-established AD locus, were observed in the Korean or combined
East Asian datasets. A SWS association was found with aggregated rare
high- and moderate-impact variants in another novel gene, MICALL1.
Suggestive associations were also observed with variants in 14 novel
loci (MSX1,BCL11A,ARMH3,CTNNA3,YTHDF1P1,GRM3,LINC02479,
LRCH1,COL9A1,TTC8,MCTP2,DLGAP2,TBC1D32,andEXD2). Nine of
these emerged from the Korean sample only because the associated
variants were either too rare or had a poor imputation quality in the
Japanese GWAS datasets. To our knowledge, this is the largest WGS
study of AD in an East Asian population.
4.2 Biological relevance of novel loci to AD
We showed that the two novel GWS loci identified in this study, ROCK2
and MICALL1, were differentially expressed in cortical tissue between
pathologically confirmed AD cases and controls, particularly for com-
parisons with the subgroup of AD cases who showed clinical symptoms
KANG ET AL.8257
FIGURE 2 Box plots showing expression level of AD-associated genes in multiple brain regions in pathologically diagnosed AD cases who were
cognitively impaired (SymAD) or cognitively healthy (AsymAD) prior to death and in individuals with no AD pathology (Control). Genes shown are
those that were differentially expressed in AD cases and controls at a study-wide significance level. (A) MSX1,(B)TTC8,(C)COL9A1,(D)GRM3,(E)
BCL11A,(F)CTNNA3,(G)ROCK2,(H)LRCH1. Units on y-axis represent log2(expression). *p<.05, **p<.01, ***p<.001. AD, Alzheimer’s disease.
8258 KANG ET AL.
of AD prior to death. Comparable expression between controls and
AD cases who were cognitively normal at the time of death suggests
that these loci may have a role in cognitive resilience. Pathway analysis
seeded with top-ranked genes in the GWAS identified multiple signifi-
cant pathways implicated in AD including neuropathic pain,78 calcium
signaling,79 neurexins and neuroligins,80 and glutaminergic receptor
signaling.81,82
The protein encoded by ROCK2, Rho-associated coiled-coil con-
taining protein kinase 2, is a negative regulator of Parkin-dependent
mitophagy that is involved in maintaining neural cells and reducing
the accumulation of dysfunctional mitochondria.83 ROCK2 is one of
the two homologous isomers of ROCK, which is a regulator of the
actin cytoskeleton and a critical component of neuronal signaling
pathways.84 It is expressed in the cerebral cortex and hippocam-
pal neurons, whereas ROCK1 is expressed primarily in tissues other
than the brain. Recent studies suggest that ROCK is a potential ther-
apeutic target for AD.85 MICALL1 is involved in endosomal protein
sorting, an essential mechanism for recycling cargo proteins related
to neurodegenerative diseases, including AD.86 For example, several
members of the sortilin family of sorting proteins, including SORL1,
SORCS1,andSORCS2, have been linked to intracellular trafficking of
amyloid beta (Aβ).24,87–89 Loss of phospholipase D3 (PLD3) function
results in MICALL1 reduction, which could worsen endosomal protein
sorting defeat and lead to altered amyloid precursor protein (APP)
processing.90,91
Five of the suggestive loci are also functionally related to AD
pathology and clinical course. MSX1 (msh homeobox 1) encodes a
transcription factor involved in brain development92 and may also
regulate astrocytosis and have a role in the proliferation and differ-
entiation of oligodendrocytes in hippocampal fimbria.93,94 Aweighted
gene co-expression network analysis identified a module including
MSX1 that is associated with neurofibrillary tangles (NFTs) and cog-
nitive impairment.95 BCL11A (BCL11 transcription factor A) mediates
seasonal factors impacting cognitive plasticity and cerebrospinal fluid
Aβ42 levels.96 CTNNA3 encodes alpha-T catenin that binds to beta-
catenin, which interacts with presenilin-1.97 Alpha-T catenin is also
associated with plasma Aβ42 levels in late-onset AD families.98 A
candidate gene study conducted in the JGSCAD 1 dataset identified
CTNNA3 as associated with AD among females.60 YTHDF1 (YTH N6-
methyladenosine RNA binding protein F1) contributes to m6A mod-
ification of activity-regulated cytoskeleton-associated protein (ARC)
and causes the Aβ-induced reduced expression of ARC,99 which has
a critical role in the synaptic plasticity of memory consolidation.100
Altered expression of GRM3 (glutamate metabotropic receptor 3)
affects contextual fear memory deficits and reduces hippocampal
neuronal excitability and intrinsic plasticity in the 5XFAD mouse
model.101 The role of ARMH3,LINC02479,LRCH1,COL9A1,TTC8,
MCTP2,DLGAP2,TBC1D32,andEXD2 in AD is also not clear at
this time.
Many of the top-ranked genes (ROCK2,MICALL1,MSX1,GRM3,
BCL11A,CTNNA3,LRCH1,TTC8,andCOL9A1) emerging from the
GWAS were significantly differentially expressed in brain tissue from
AD cases compared to cognitively normal controls, particularly in the
entorhinal cortex, where AD-related histological changes begin,102–104
supporting the conclusion that they may have a role in AD. Changes
in expression for several of these genes, including ROCK2,werepro-
gressively higher or lower in asymptomatic and symptomatic AD cases
compared to controls, suggesting that they have a more direct role in
the development of AD pathology than in cognitive impairment. This
may explain why some of the novel associations identified in this study
were not observed in previous much larger GWAS including primarily
clinically defined AD cases, a portion of whom will not meet the criteria
for a pathological diagnosis of AD. Results from the biological path-
way and gene network analyses also link AD to our GWAS findings.
The majority of the nominally enriched pathways involve the brain and
cardiac vasculature, including calcium signaling105 and glutaminergic
receptor signaling.106
4.3 Comparison of findings with prior AD GWAS
in East Asians
Our findings do not replicate associations identified in GWASs
for AD or AD-related magnetic resonance imaging (MRI) traits
conducted in East Asian populations including Koreans,42,107
Japanese,25,63,108 and Chinese109,110 (Table S8). A previous GWAS
that included a much smaller portion of the GARD cohort and the
JGSCAD 1 dataset identified GWS associations with variants in LRIG1
(rs2280575, OR =0.54, p=1.51×10−8)andCACNA1A (rs189753894,
OR =1.79, p=2.49×10−8) among APOE ε4 non-carriers.42 However,
rs189753894 failed QC in the GARD WGS dataset and was filtered
out of the Japanese datasets because it was not well imputed using
the TOPMed reference panel. TOPMed is much larger and ethnically
diverse than the Haplotype Reference Consortium panel that was
used for imputation in the Kang et al. study.42 A GWAS for AD-related
MRI traits conducted in the GARD cohort including 1356 AD cases,
2184 MCI cases, and 2030 controls identified GWS association of
a rare SHARPIN missense variant (rs77359862) with an entorhinal
thickness (p=5.0×10−9) and hippocampal volume (p=5.1×10−12).111
Although the association of this variant with AD did not quite reach
statistical significance in our study (p=8.05×10−2), another SHARPIN
variant (rs34173062) that is frequent in European populations but
absent in East Asians was significantly associated with AD in large
European ancestry samples evaluated using multiple approaches.17,112
A Japanese study reported GWS associations with a variant in FAM47E
(rs920608, OR =0.65, p=5.34×10−9),63 and a large study in China
including 3913 AD cases and 7593 controls found GWS associations
with variants in LINC02325 (rs234434, OR =1.71, p=1.35×10−67),
LINC01950 (rs6859823, OR =0.74, p=2.49×10−23), RHOBTB3
(rs3777215, OR =0.69, p=3.07×10−19), and CHODL (rs2255835,
OR =1.23, p=4.81×10−9).110 Although varied genetic backgrounds
and heterogeneity might account for the lack of overlap in findings
between our study and those conducted in Chinese samples,109,110 it
is noteworthy that none of the novel loci identified in these studies
were among the top-ranked findings in large AD GWASs in Euro-
pean ancestry17 and other populations,113–115 and nearly all of the
KANG ET AL.8259
top-ranked findings in the other Korean107 and Japanese25,63,108
samples were not GWS and, thus, possibly false positives (Table S8).
4.4 Study limitations
Our study has several limitations. Similar to previous AD GWAS of
non-White cohorts, including East Asians,25,42 63,107–110 Caribbean
Hispanics,114,115 and African Americans,113,116 the Korean WGS sam-
ple had modest power to detect associations with small effect variants
(ie, ORs less than 1.2). Also, we could not obtain results for about half
of the novel findings emerging from the Korean cohort in two Japanese
genome-wide array datasets in which many of the top-ranked variants
were absent or not well imputed. Similarly, most of the top-ranked
variants in the East Asian cohorts in this study were not replicated
in WGS datasets for other ethnic groups in the Alzheimer’s Disease
Sequencing Project (Table S9).117 Failure to replicate in samples from
other populations may be due to their rare frequency or limited power,
especially in stratified subgroups. Also, the functional variants under-
lying the association signals may have different patterns of LD or are
absent. However, despite limited replication in independent datasets,
our novel ROCK2 findings were supported by evidence in multiple East
Asian datasets, and nearly one-half of our top-ranked findings were val-
idated by DGE analysis conducted in brain tissue from AD cases and
controls.
5CONCLUSIONS
We identified GWS or SWS associations for AD with two novel loci,
ROCK2 and MICALL1, and suggestive associations with 14 additional
loci not previously linked to AD by GWASs, as well as associations
with two previously established AD loci (APOE and SORL1). Our
results also provide some insight into genetic mechanisms leading
to AD-associated dementia, which are not entirely coincident with
those leading to AD-related pathology. Ongoing recruitment efforts to
recruit Korean and other East Asian AD cohorts for WGS studies of
AD118,119 will provide additional opportunities to replicate and extend
our findings.
ACKNOWLEDGMENTS
GWAS data for the JGSCAD cohort were obtained from the
Alzheimer’s Disease Genetics Consortium, which is funded by
National Institute on Aging (NIA) grant U01-AG032984. Gene expres-
sion data were provided by the Rush Alzheimer’s Disease Center,
Rush University Medical Center, Chicago, IL, USA. Data collection
was supported through funding by NIA grants P30AG10161 (ROS),
R01AG15819 (ROSMAP; genomics and RNAseq), R01AG17917
(MAP), R01AG30146, R01AG36042 (5hC methylation, ATACseq),
RC2AG036547 (H3K9A), R01AG36836 (RNAseq), R01AG48015
(monocyte RNAseq), RF1AG57473 (single-nucleus RNAseq),
U01AG32984 (genomic and whole-exome sequencing), U01AG46152
(ROSMAP AMP-AD, targeted proteomics), U01AG46161 (TMT
proteomics), U01AG61356 (whole-genome sequencing, targeted
proteomics, ROSMAP AMP-AD), the Illinois Department of Pub-
lic Health (ROSMAP), and the Translational Genomics Research
Institute (genomic). Additional phenotypic data can be requested
at www.radc.rush.edu. This study was supported by NIA grants
U01-AG062602, R01-AG048927, U01-AG032984, U01-AG058654,
U54-AG052427, U19-AG068753, U01-AG081230, P30-AG072978.
This study was supported by the KBRI Basic Research Program
through the Korea Brain Research Institute, funded by the Ministry of
Science and ICT of the Korean government (24-BR-03-05).
CONFLICT OF INTEREST STATEMENT
J.J.F. and G.R.J. received support from NIH grants. L.A.F. received sup-
port from NIH grants and an honorarium for serving as a journal editor.
None of the other authors have conflicts of interest to disclose. Author
disclosures are available in the Supporting Information.
DATA AVAILABILITY STATEMENT
ADGC GWAS data, ADSP WGS data, and summarized results are avail-
able from the National Institute on Aging Genetics of Alzheimer’s
Disease Storage site (https://www.niagads.org).
CONSENT STATEMENT
All human subjects provided informed consent.
REFERENCES
1. Pericak-Vance MA, Yamaoka LH, Haynes CS, et al. Genetic linkage
studies in Alzheimer’s disease families. Exp Neurol. 1988;102:271-
279.
2. Farrer LA, O’Sullivan DM, Cupples LA, Growdon JH, Myers RH.
Assessment of genetic risk for Alzheimer’s disease among first-
degree relatives. Ann Neurol. 1989;25:485-493.
3. Breitner JC, Gatz M, Bergem AL, et al. Use of twin cohorts for
research in Alzheimer’s disease. Neurology. 1993;43:261-267.
4. Bergem AL, Engedal K, Kringlen E. The role of heredity in late-onset
Alzheimer disease and vascular dementia. A twin study. Arch Gen
Psychiatry. 1997;54:264-270.
5. Farrer LA, Myers RH, Connor L, Cupples LA, Growdon JH. Segrega-
tion analysis reveals evidence of a major gene for Alzheimer disease.
Am J Hum Genet. 1991;48:1026-1033.
6. van Duijn CM, Farrer LA, Cupples LA, Hofman A. Genetic transmis-
sion of Alzheimer’s disease among families in a Dutch population
based study. J Med Genet. 1993;30:640-646.
7. Rao VS, van Duijn CM, Connor-Lacke L, Cupples LA, Growdon
JH, Farrer LA. Multiple etiologies for Alzheimer disease are
revealed by segregation analysis. Am J Hum Genet. 1994;55:
991-1000.
8. Gatz M, Reynolds CA, Fratiglioni L, et al. Role of genes and envi-
ronments for explaining Alzheimer disease. Arch Gen Psychiatry.
2006;63:168-174.
9. Saunders AM, Strittmatter WJ, Schmechel D, et al. Association
of apolipoprotein E allele ε4 with late-onset familial and sporadic
Alzheimer’s disease. Neurology. 1993;43:1467-1472.
10. Nalbantoglu J, Gilfix BM, Bertrand P, et al. Predictive value of
apolipoprotein e genotyping in Alzheimer’s disease: results of an
autopsy series and an analysis of several combined studies. Ann
Neurol. 1994;36:889-895.
11. Farrer LA, Cupples LA, Haines JL, et al. Effects of age, sex, and eth-
nicity on the association between apolipoprotein E genotype and
8260 KANG ET AL.
Alzheimer disease. a meta-analysis. Apoe and Alzheimer disease
meta analysis consortium. JAMA. 1997;278:1349-1356.
12. Lambert JC, Heath S, Even G, et al. Genome-wide association study
identifies variants at Clu and CR1 associated with Alzheimer’s
disease. Nat Genet. 2009;41:1094-1099.
13. Lambert JC, Ibrahim-Verbaas CA, Harold D, et al. Meta-analysis
of 74,046 individuals identifies 11 new susceptibility loci for
Alzheimer’s disease. Nat Genet. 2013;45:1452-1458.
14. Jansen IE, Savage JE, Watanabe K, et al. Genome-wide meta-analysis
identifies new loci and functional pathways influencing Alzheimer’s
disease risk. Nat Genet. 2019;51:404-413.
15. Kunkle BW, Grenier-Boley B, Sims R, et al. Genetic meta-analysis of
diagnosed Alzheimer’s disease identifies new risk loci and implicates
Aβ, tau, immunity and lipid processing. Nat Genet. 2019;51:414-430.
16. Wightman DP, Jansen IE, Savage JE, et al. A genome-wide associ-
ation study with 1,126,563 individuals identifies new risk loci for
Alzheimer’s disease. Nat Genet. 2021;53:1276-1282.
17. Bellenguez C, Küçükali F, Jansen IE, et al. New insights into the
genetic etiology of Alzheimer’s disease and related dementias. Nat
Genet. 2022;54:412-436.
18. Lee SH, Harold D, Nyholt DR, et al, ANZGene Consortium, Inter-
national Endogene Consortium, Genetic and Environmental Risk for
Alzheimer’s disease Consortium. Estimation and partitioning of poly-
genic variation captured by common SNPs for Alzheimer’s disease,
multiple sclerosis and endometriosis. Hum Mol Genet. 2013;22:832-
841.
19. Ridge PG, Mukherjee S, Crane PK, Kauwe JS, Alzheimer’s Disease
Genetics Consortium. Alzheimer’s disease: analyzing the missing
heritability. PLoS One. 2013;8:e79771.
20. Brainstorm Consortium, Anttila V, Bulik-Sullivan B, Finucane HK,
et al. Analysis of shared heritability in common disorders of the brain.
Science. 2018;360:eaap8757.
21. Holland D, Frei O, Desikan R, et al. The genetic architecture of
human complex phenotypes is modulated by linkage disequilibrium
and heterozygosity. Genetics. 2021;217:iyaa046.
22. Li D, Farrell JJ, Mez J, et al. Novel loci for Alzheimer’s disease
identified by a genome-wide association study in Ashkenazi Jews.
Alzheimers Dement. 2023;19:5550-5562.
23. Meng Y, Baldwin CT, Bowirrat A, et al. Association of polymorphisms
in the angiotensin-converting enzyme gene with Alzheimer disease in
an Israeli Arab community. Am J Hum Genet. 2006;78:871-877.
24. Rogaeva E, Meng Y, Lee JH, et al. The neuronal sortilin-related recep-
tor SORL1 is genetically associated with Alzheimer disease. Nat
Genet. 2007;39:168-177.
25. Miyashita A, Koike A, Jun G, et al. SORL1 is genetically associ-
ated with late-onset Alzheimer’s disease in Japanese, Koreans And
Caucasians. PLoS One. 2013;8:e58618.
26. Ghani M, Reitz C, Cheng R, et al. Association of long runs of homozy-
gosity with Alzheimer disease among African American individuals.
JAMA Neurol. 2015;72:1313-1323.
27. Barral S, Cheng R, Reitz C, et al. Linkage analyses in Caribbean His-
panic families identify novel loci associated with familial late-onset
Alzheimer’s disease. Alzheimers Dement. 2015;11:1397-1406.
28. Lee JY, Chang SM, Jang HS, et al. Illiteracy and the incidence
of Alzheimer’s disease in the Yonchon county survey, Korea. Int
Psychogeriatr. 2008;20:976-985.
29. Cho MJ, Lee JY, Kim BS, Lee HW, Sohn JH. Prevalence of the
major mental disorders among the Korean elderly. J Korean Med Sci.
2011;26:1-10.
30. Ferri CP, Prince M, Brayne C, et al. Global prevalence of dementia: a
Delphi consensus study. Lancet. 2005;366:2112-2117.
31. Jin HJ, Kwak KD, Hammer MF, et al. Y-chromosomal DNA hap-
logroups and their implications for the dual origins of the Koreans.
Hum Genet. 2003;114:27-35.
32. Jin HJ, Kwak KD, Hong SB, et al. Forensic genetic analysis
of mitochondrial DNA hypervariable region I/II sequences: an
expanded Korean population database. Forensic Sci Int. 2006;158:
125-130.
33. Yang JH, Sohn YH, Ko SY, Choi SE, Kim MH, Oh HB. Anthropological
analysis of Koreans using HLA class II diversity among East Asians.
Tissue Antigens. 2010;76:282-288.
34. Liu X, Lu D, Saw WY, et al. Characterising private and shared signa-
tures of positive selection in 37 Asian populations. Eur J Hum Genet.
2017;25:499-508.
35. Kim KW, Jhoo JH, Lee KU, et al. Association between apolipoprotein
e polymorphism and Alzheimer’s disease in Koreans. Neurosci Lett.
1999;277:145-148.
36. Chung SJ, Lee JH, Kim SY, et al. Association of GWAS top hits with
late-onset Alzheimer disease in Korean population. Alzheimer Dis
Assoc Disord. 2013;27:250-257.
37. Choi KY, Lee JJ, Gunasekaran TI, et al. APOE promoter
polymorphism-219t/G is an effect modifier of the influence of
APOE ε4 on Alzheimer’s disease risk in a multiracial sample. J Clin
Med. 2019;8:1236.
38. Kang Y, Na DL, Hahn S. A validity study on the Korean Mini-Mental
State Examination (K-MMSE) in dementia patients. J Korean Neurol
Assoc. 1997;15:300-308.
39. Kang Y, Na DL, Hahn S. Seoul Neuropsychological Screening Battery
(SNSB). Human Brain Research & Consulting Co; 2003.
40. Ahn HJ, Chin J, Park A, et al. Seoul Neuropsychological Screening
Battery-Dementia version (SNSB-D): a useful tool for assessing and
monitoring cognitive impairments in dementia patients. J Korean Med
Sci. 2010;25:1071-1076.
41. McKhann GM, Knopman DS, Chertkow H, et al. The diagnosis of
dementia due to Alzheimer’s disease: recommendations from the
National Institute On Aging-Alzheimer’s Association workgroups on
diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement.
2011;7:263-269.
42. Kang S, Gim J, Lee J, et al. Potential novel genes for late-
onset Alzheimer’s disease in East-Asian descent identified by
APOE-stratified genome-wide association study. J Alzheimers Dis.
2021;82:1451-1460.
43. Li H, Durbin R. Fast and accurate short read alignment with burrows-
wheeler transform. Bioinformatics. 2009;25:1754-1760.
44. Li H. Aligning sequence reads, clone sequences and assembly contigs
with BWA-MEM. arXiv: Genomics. 2013. doi:10.6084/m9.figshare.
963153.v1
45. Jee YH, Lee SJ, Jung KJ, Jee SH. Alcohol intake and serum glucose
levels from the perspective of a Mendelian randomization design: the
KCPS-II biobank. PLoS One. 2016;11:e0162930.
46. McKenna A, Hanna M, Banks E, et al. The Genome Analysis Toolkit: a
MapReduce framework for analyzing next-generationDNA sequenc-
ing data. Genome Res. 2010;20:1297-1303.
47. Javed N, Farjoun Y, Fennell TJ, Epstein CB, Bernstein BE, Shoresh
N. Detecting sample swaps in diverse NGS data types using linkage
disequilibrium. Nat Commun. 2020;11:3697.
48. Behera S, Catreux S, Rossi M, et al. Comprehensive and accurate
genome analysis at scale using DRAGEN accelerated algorithms.
bioRxiv. 2024. doi:10.1101/2024.01.02.573821
49. DePristo MA, Banks E, Poplin R, et al. A framework for variation dis-
covery and genotyping using next-generation DNA sequencing data.
Nat Genet. 2011;43:491-498.
50. Gogarten SM, Sofer T, Chen H, et al. Genetic association test-
ing using the GENESIS R/Bioconductor package. Bioinformatics.
2019;35:5346-5348.
51. Manichaikul A, Mychaleckyj JC, Rich SS, Daly K, Sale M, Chen WM.
Robust relationship inference in genome-wide association studies.
Bioinformatics. 2010;26:2867-2873.
KANG ET AL.8261
52. Conomos MP, Miller MB, Thornton TA. Robust inference of popula-
tion structure for ancestry prediction and correction of stratification
in the presence of relatedness. Genet Epidemiol. 2015;39:276-293.
53. Conomos MP, Reiner AP, Weir BS, Thornton TA. Model-free estima-
tion of recent genetic relatedness. Am J Hum Genet. 2016;98:127-
148.
54. Cingolani P, Platts A, le Wang L, et al. A program for annotating and
predicting the effects of single nucleotide polymorphisms, SnpEff:
SNPs in the genome of Drosophila melanogaster strain w1118; iso-2;
iso-3. Fly. 2012;6:80-92.
55. Zhou W, Zhao Z, Nielsen JB, et al. Scalable generalized linear mixed
model for region-based association tests in large biobanks and
cohorts. Nat Genet. 2020;52:634-639.
56. Zhou W, Bi W, Zhao Z, et al. SAIGE-GENE+improves the efficiency
and accuracy of set-based rare variant association tests. Nat Genet.
2022;54:1466-1469.
57. Lee S, Wu MC, Lin X. Optimal tests for rare variant effects in
sequencing association studies. Biostatistics. 2012;13:762-775.
58. Liu Y, Xie J. Cauchy combination test: a powerful test with analytic
p-value calculation under arbitrary dependency structures. JAmStat
Assoc. 2020;115:393-402.
59. Kuwano R, Miyashita A, Arai H, et al. Dynamin-binding protein
gene on chromosome 10q is associated with late-onset Alzheimer’s
disease. Hum Mol Genet. 2006;15:2170-2182.
60. Miyashita A, Arai H, Asada T, et al. Genetic association of ctnna3
with late-onset Alzheimer’s disease in females. Hum Mol Genet.
2007;16:2854-2869.
61. Jun GR, Chung J, Mez J, et al. Transethnic genome-wide scan identi-
fies novel Alzheimer’s disease loci. Alzheimers Dement. 2017;13:727-
738.
62. Asanomi Y, Shigemizu D, Miyashita A, et al. A rare functional variant
of SHARPIN attenuates the inflammatory response and associates
with increased risk of late-onset Alzheimer’s disease. Mol Med.
2019;25:20.
63. Shigemizu D, Mitsumori R, Akiyama S, et al. Ethnic and trans-
ethnic genome-wide association studies identify new loci influ-
encing Japanese Alzheimer’s disease risk. Transl Psychiatry. 2021;
11:151.
64. Das S, Forer L, Schönherr S, et al. Next-generation genotype imputa-
tion service and methods. Nat Genet. 2016;48:1284-1287.
65. Taliun D, Harris DN, Kessler MD, et al. Sequencing of 53,831 diverse
genomes from the NHLBI TOPMed program. Nature. 2021;590:
290-299.
66. Willer CJ, Li Y, Abecasis GR. Metal: fast and efficient meta-analysis of
genomewide association scans. Bioinformatics. 2010;26:2190-2191.
67. Purcell S, Neale B, Todd-Brown K, et al. PLINK: a tool set for whole-
genome association and population-based linkage analyses. Am J
Hum Genet. 2007;81:559-575.
68. Chang CC, Chow CC, Tellier LC, Vattikuti S, Purcell SM, Lee JJ.
Second-generation PLINK: rising to the challenge of larger and richer
datasets. Gigascience. 2015;4:7.
69. Wang G, Sarkar A, Carbonetto P, Stephens M. A simple new
approach to variable selection in regression, with application to
genetic fine mapping. J R Stat Soc Series B Stat Methodol. 2020;82:
1273-1300.
70. Zou Y, Carbonetto P, Wang G, Stephens M. Fine-mapping from
summary data with the “Sum Of Single Effects” model. PLoS Genet.
2022;18:e1010299.
71. Hutchinson A, Watson H, Wallace C. Improving the coverage of
credible sets in Bayesian genetic fine-mapping. PLoS Comput Biol.
2020;16:e1007829.
72. Patel H, Hodges AK, Curtis C, et al. Transcriptomic analysis of prob-
able asymptomatic and symptomatic alzheimer brains. Brain Behav
Immun. 2019;80:644-656.
73. Barrett T, Wilhite SE, Ledoux P, et al. NCBI GEO: archive for func-
tional genomics data sets—update. Nucleic Acids Res. 2013;41:D991-
D995.
74. De Jager PL, Ma Y, McCabe C, et al. A multi-omic atlas of the human
frontal cortex for aging and Alzheimer’s disease research. Sci Data.
2018;5:180142.
75. Panitch R, Hu J, Xia W, et al. Blood and brain transcriptome analy-
sis reveals APOE genotype-mediated and immune-related pathways
involved in Alzheimer disease. Alzheimers Res Ther. 2022;14:30.
76. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a prac-
tical and powerful approach to multiple testing. J Royal Statistical Soc:
Series B (Methodological). 2018;57:289-300.
77. Krämer A, Green J, Pollard J Jr, Tugendreich S. Causal analy-
sis approaches in Ingenuity Pathway Analysis. Bioinformatics.
2014;30:523-530.
78. Cao S, Fisher DW, Yu T, Dong H. The link between chronic pain and
Alzheimer’s disease. J Neuroinflammation. 2019;16:204.
79. Miller MR, Lee YF, Kastanenka KV. Calcium sensor yellow Cameleon
3.6 as a tool to support the calcium hypothesis of Alzheimer’s disease.
Alzheimers Dement. 2023;19:4196-4203.
80. Hishimoto A, Pletnikova O, Lang DL, Troncoso JC, Egan JM, Liu
QR. Neurexin 3 transmembrane and soluble isoform expression and
splicing haplotype are associated with neuron inflammasome and
Alzheimer’s disease. Alzheimers Res Ther. 2019;11:28.
81. Wang R, Reddy PH. Role of glutamate and NMDA receptors in
Alzheimer’s disease. J Alzheimers Dis. 2017;57:1041-1048.
82. Arias-Aragón F, Tristán-Clavijo E, Martínez-Gallego I, et al. A
Neuroligin-1 mutation associated with Alzheimer’s disease produces
memory and age-dependent impairments in hippocampal plasticity.
iScience. 2023;26:106868.
83. Mani S, Jindal D, Chopra H, et al. ROCK2 inhibition: a futuris-
tic approach for the management of Alzheimer’s disease. Neurosci
Biobehav Rev. 2022;142:104871.
84. Liu J, Gao HY, Wang XF. The role of the Rho/ROCK signaling path-
way in inhibiting axonal regeneration in the central nervous system.
Neural Regen Res. 2015;10:1892-1896.
85. Moskal N, Riccio V, Bashkurov M, et al. ROCK inhibitors upregu-
late the neuroprotective Parkin-mediated mitophagy pathway. Nat
Commun. 2020;11:88.
86. Fecto F, Esengul YT, Siddique T. Protein recycling pathways in
neurodegenerative diseases. Alzheimers Res Ther. 2014;6:13.
87. Reitz C, Tokuhiro S, Clark LN, et al. SORCS1 alters amyloid precur-
sor protein processing and variants may increase Alzheimer’s disease
risk. Ann Neurol. 2011;69:47-64.
88. Reitz C, Tosto G, VardarajanB, et al. Independent and epistatic effects
of variants in VPS10-d receptors on Alzheimer disease risk and pro-
cessing of the amyloid precursor protein (APP). Transl Psychiatry.
2013;3:e256.
89. Mishra S, Knupp A, Szabo MP, et al. The Alzheimer’s gene SORL1 is
a regulator of endosomal traffic and recycling in human neurons. Cell
Mol Life Sci. 2022;79:162.
90. Mukadam AS, Breusegem SY, Seaman MNJ. Analysis of novel
endosome-to-Golgi retrieval genes reveals a role for PLD3 in reg-
ulating endosomal protein sorting and amyloid precursor protein
processing. Cell Mol Life Sci. 2018;75:2613-2625.
91. Van Acker ZP, Bretou M, Annaert W. Endo-lysosomal dysregulations
and late-onset Alzheimer’s disease: impact of genetic risk factors. Mol
Neurodegener. 2019;14:20.
92. Bach A, Lallemand Y, Nicola MA, et al. Msx1 is required for dorsal
diencephalon patterning. Development. 2003;130:4025-4036.
93. Ramos C, Martinez A, Robert B, Soriano E. Msx1 expression in
the adult mouse brain: characterization of populations of beta-
galactosidase-positive cells in the hippocampus and fimbria. Neuro-
science. 2004;127:893-900.
8262 KANG ET AL.
94. Roybon L, Hjalt T, Christophersen NS, Li JY, Brundin P. Effects on
differentiation of embryonic ventral midbrain progenitors by Lmx1a,
Msx1, Ngn2, and Pitx3. J Neurosci. 2008;28:3644-3656.
95. Liang JW, Fang ZY, Huang Y, et al. Application of weighted gene co-
expression network analysis to explore the key genes in Alzheimer’s
disease. J Alzheimers Dis. 2018;65:1353-1364.
96. Lim ASP, Gaiteri C, Yu L, et al. Seasonal plasticity of cognition and
related biological measures in adults with and without Alzheimer
disease: analysis of multiple cohorts. PLoS Med. 2018;15:e1002647.
97. Zhang Z, Hartmann H, Do VM, et al. Destabilization of beta-catenin
by mutations in presenilin-1 potentiates neuronal apoptosis. Nature.
1998;395:698-702.
98. Ertekin-Taner N, Ronald J, Asahara H, et al. Fine mapping of the alpha-
T catenin gene to a quantitative trait locus on chromosome 10 in late-
onset Alzheimer’s disease pedigrees. Hum Mol Genet. 2003;12:3133-
3143.
99. Xu C, Huang H, Zhang M, et al. Methyltransferase-like 3 rescues the
amyloid-beta protein-induced reduction of activity-regulated
cytoskeleton associated protein expression via YTHDF1-
dependent N6-methyladenosine modification. Front Aging Neurosci.
2022;14:890134.
100. Zhang Z, Wang M, Xie D, et al. METTL3-mediated N6-
methyladenosine mRNA modification enhances long-term memory
consolidation. Cell Res. 2018;28:1050-1061.
101. Neuner SM, Wilmott LA, Hoffmann BR, Mozhui K, Kaczorowski CC.
Hippocampal proteomics defines pathways associated with memory
decline and resilience in normal aging and Alzheimer’s disease mouse
models. Behav Brain Res. 2017;322:288-298.
102. Braak H, Braak E. Neuropathological stageing of Alzheimer-related
changes. Acta Neuropathol. 1991;82:239-259.
103. Braak H, Braak E. The human entorhinal cortex: normal morphol-
ogy and lamina-specific pathology in various diseases. Neurosci Res.
1992;15:6-31.
104. Zhou M, Zhang F, Zhao L, Qian J, Dong C. Entorhinal cortex: a good
biomarker of mild cognitive impairment and mild Alzheimer’s disease.
Rev Neurosci. 2016;27:185-195.
105. Pchitskaya E, Popugaeva E, Bezprozvanny I. Calcium signaling and
molecular mechanisms underlying neurodegenerative diseases. Cell
Calcium. 2018;70:87-94.
106. Gasiorowska A, Wydrych M, Drapich P, et al. The biology and patho-
biology of glutamatergic, cholinergic, and dopaminergic signaling in
the aging brain. Front Aging Neurosci.2021;13:654931.
107. Park JH, Park I, Youm EM, et al. Novel Alzheimer’s disease risk
variants identified based on whole-genome sequencing of APOE ε4
carriers. TranslPsychiatry. 2021;11:296.
108. Hirano A, Ohara T, Takahashi A, et al. A genome-wide association
study of late-onset Alzheimer’s disease in a Japanese population.
Psychiatr Genet. 2015;25:139-146.
109. Zhou X, Chen Y, Mok KY, et al. Identification of genetic risk fac-
tors in the Chinese population implicates a role of immune system
in Alzheimer’s disease pathogenesis. Proc Natl Acad Sci U S A.
2018;115:1697-1706.
110. Jia L, Li F, Wei C, et al. Prediction of Alzheimer’s disease using
multi-variants from a Chinese genome-wide association study. Brain.
2021;144:924-937.
111. Park JY, Lee D, Lee JJ, et al. A missense variant in SHARPIN medi-
ates Alzheimer’s disease-specific brain damages. Transl Psychiatry.
2021;11:590.
112. Lancour D, Naj A, Mayeux R, et al. One for all and all for one: improv-
ing replication of genetic studies through network diffusion. PLoS
Genet. 2018;14:e1007306.
113. Kunkle BW, Schmidt M, Klein HU, et al. Novel Alzheimer dis-
ease risk loci and pathways in African American individuals using
the African genome resources panel: a meta-analysis. JAMA Neuro l.
2021;78:102-113.
114. VardarajanBN, Zhang Y, Lee JH, et al. Coding mutations in SORl1 and
Alzheimer disease. Ann Neurol. 2015;77:215-227.
115. Tosto G, Fu H, Vardarajan BN, et al. F-box/LRR-repeat protein 7
is genetically associated with Alzheimer’s disease. Ann Clin Transl
Neurol. 2015;2:810-820.
116. Reitz C, Mayeux R. Genetics of Alzheimer’s disease in Caribbean
Hispanic and African American populations. Biol Psychiatry.
2014;75:534-541.
117. Xue D, Bush WS, Renton AE, et al. Large-scale sequencing stud-
ies expand the known genetic architecture of Alzheimer’s disease.
Alzheimers Dement. 2021;13:e12255.
118. Byun MS, Yi D, Lee JH, et al. Korean brain aging study for the
early diagnosis and prediction of Alzheimer’s disease: methodology
and baseline sample characteristics. Psychiatry Investig.2017;14:851-
863.
119. Ho PC, Yu WH, Tee BL, et al. Asian Cohort for Alzheimer’s Dis-
ease (ACAD) pilot study on genetic and non-genetic risk factors
for Alzheimer’s disease among Asian Americans and Canadians.
Alzheimers Dement. 2024;20:2058-2071.
SUPPORTING INFORMATION
Additional supporting information can be found online in the Support-
ing Information section at the end of this article.
How to cite this article: Kang M, Farrell JJ, Zhu C, et al.
Whole-genome sequencing study in Koreans identifies novel
loci for Alzheimer’s disease. Alzheimer’s Dement.
2024;20:8246–8262. https://doi.org/10.1002/alz.14128
