Apoptotic Engulfment Pathway and Schizophrenia
Xiangning Chen1,2*, Cuie Sun1, Qi Chen1, F. Anthony O’Neill3, Dermot Walsh4, Ayman H. Fanous5,
Kodavali V. Chowdari6, Vishwajit L. Nimgaonkar6, Adrian Scott7, Sibylle G. Schwab7, Dieter B.
Wildenauer8, Ronglin Che9, Wei Tang9, Yongyong Shi9, Lin He9, Xiong-jian Luo10, Bing Su10, Todd L.
Edwards11, Zhongming Zhao12, Kenneth S. Kendler1,2
1Department of Psychiatry and Virginia Institute for Psychiatric and Behavior Genetics, Virginia Commonwealth University, Richmond, Virginia, United States of America,
2Department of Human and Molecular Genetics, Virginia Commonwealth University, Richmond, Virginia, United States of America, 3The Department of Psychiatry, The
Queens University, Belfast, Northern Ireland, 4The Health Research Board, Dublin, Ireland, 5Washington VA Medical Center, Washington, D. C., and Georgetown University
School of Medicine, Washington, D. C., United States of America, 6Department of Psychiatry, University of Pittsburgh School of Medicine and Graduate School of Public
Health, Pittsburgh, Pennsylvania, United States of America, 7Western Australian Institute for Medical Research, Centre for Medical Research, University of Western
Australia, Perth, Australia, 8Centre for Clinical Research in Neuropsychiatry, School of Psychiatry and Clinical Neurosciences, University of Western Australia, Perth,
Australia, 9Bio-X Center, Shanghai Jiao Tong University, Shanghai, People’s Republic of China, 10State Key Laboratory of Genetic Resources and Evolution, Kunming
Institute of Zoology, Chinese Academy of Sciences, Kunming, People’s Republic of China, 11Miami institute of human genetics, University of Miami, Miami, Florida, United
States of America, 12Departments of Biomedical Informatics and Psychiatry, Vanderbilt University School of Medicine, Nashville, Tennessee, United States of America
Background: Apoptosis has been speculated to be involved in schizophrenia. In a previously study, we reported the
association of the MEGF10 gene with the disease. In this study, we followed the apoptotic engulfment pathway involving
the MEGF10, GULP1, ABCA1 and ABCA7 genes and tested their association with the disease.
Methodology/Principal Findings: Ten, eleven and five SNPs were genotyped in the GULP1, ABCA1 and ABCA7 genes
respectively for the ISHDSF and ICCSS samples. In all 3 genes, we observed nominally significant associations. Rs2004888 at
GULP1 was significant in both ISHDSF and ICCSS samples (p=0.0083 and 0.0437 respectively). We sought replication in
independent samples for this marker and found highly significant association (p=0.0003) in 3 Caucasian replication
samples. But it was not significant in the 2 Chinese replication samples. In addition, we found a significant 2-marker
(rs2242436 * rs3858075) interaction between the ABCA1 and ABCA7 genes in the ISHDSF sample (p=0.0022) and a 3-marker
interaction (rs246896 * rs4522565 * rs3858075) amongst the MEGF10, GULP1 and ABCA1 genes in the ICCSS sample
(p=0.0120). Rs3858075 in the ABCA1 gene was involved in both 2- and 3-marker interactions in the two samples.
Conclusions/Significance: From these data, we concluded that the GULP1 gene and the apoptotic engulfment pathway are
involved in schizophrenia in subjects of European ancestry and multiple genes in the pathway may interactively increase the
risks to the disease.
Citation: Chen X, Sun C, Chen Q, O’Neill FA, Walsh D, et al. (2009) Apoptotic Engulfment Pathway and Schizophrenia. PLoS ONE 4(9): e6875. doi:10.1371/
Editor: Hitoshi Okazawa, Tokyo Medical and Dental University, Japan
Received April 3, 2009; Accepted July 14, 2009; Published September 1, 2009
Copyright: ? 2009 Chen et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This study is supported by an Independent Investigator Award and research grant 07R-1770 to XC from NARSAD (http://www.narsad.org/) and the
Stanley Medical Research Institute (http://www.stanleyresearch.org/dnn/), by grant RO1MH41953 to KSK from National Institute of Mental Health (http://www.
nimh.nih.gov/) and by a Young Investigator Award to ZZ from NARSAD. The funding agencies had no role in study design, data collection and analysis, decision to
publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: email@example.com
Schizophrenia is a complex psychiatric disorder with strong
genetic influences. Recently, many genes have been identified as
potential susceptibility candidates [8,10,55,56]. In a recent study,
we reported the association of MEGF10 with schizophrenia in our
Irish family and case-control samples . MEGF10 is the human
ortholog of the draper (DRPR) and cell death abnormal 1 (CED-1)
gene in Drosophila melanogaster and Caenorhabditis elegans respectively
. In both Drosophila and C. elegans, DRPR and CED-1 are key
components of phagocytosis functioning in the clearance of
apoptotic cells [23,67]. In multicellular organisms, phagocytosis
is a crucial process in development and in the innate immune
system [17,26]. In addition to its role in the clearance of apoptotic
cells, DRPR plays a critical role in axon pruning and degenerating
neurons [1,30]. In C. elegans, CED-1 mutations cause abnormal
axon patterns and commissure branching .
In C. elegans, two parallel pathways are involved in the clearance
of apoptotic cells. One involves the CED-2, CED-5 and CED-12
genes, which activate GTPase CED-10 and trigger rearrangement
of cytoskeleton during phagocytosis. The other involves the CED-
1, CED-6 and CED-7 genes. CED-1 is a membrane protein that
senses cell death signals and recognizes neighboring cell corpses. It
clusters at the phagocytic cups and initiates pseudopod extension.
CED-6 is an adaptor protein for CED-1, helping to control the
delivery of vesicles to phagocytic cups and phagosomes. CED-7
PLoS ONE | www.plosone.org1September 2009 | Volume 4 | Issue 9 | e6875
works with CED-1 in recognizing engulfment signals of cell
corpses. The detail functions of these genes are reviewed recently
The apoptotic engulfment pathways are evolutionarily con-
served and it is believed that they function similarly in higher
organisms, including humans . In the last several years,
orthologs of several components of these pathways have been
identified in mammals [28,39,63]. In humans, MEGF10 and
GULP1 have been determined to be orthologs of CED-1 and CED-
6 respectively [16,28,59]. For CED-7, two human genes, ABCA1
and ABCA7, are proposed as orthologs [16,19,39]. Based on our
study of the MEGF10 gene and its function in apoptosis, we
hypothesize that the apoptotic engulfment pathways are involved
in the etiology of schizophrenia. To test the hypothesis, we
conducted association studies of GULP1, ABCA1 and ABCA7,
located at chromosomes 2, 9 and 19 respectively, with the Irish
study of high density schizophrenia families (ISHDSF) and Irish
case-control study of schizophrenia (ICCSS) samples and followed
up these analyses with targeted replication in multiple independent
samples. In this article, we report the results from these studies.
The GULP1 gene.
CED pathway we analyzed in this study. As in our previous studies
[4–9], the ISHDSF sample was used as a screening sample and the
ICCSS was used to follow up as a replication sample. For the 10
SNPs typed in the ISHDSF sample, multiple markers reached
nominal significance by the pedigree disequilibrium test (PDT)
 (Table 1). Subsequently, we typed the same SNPs in the
ICCSS sample. Two SNPs, rs2004888 and rs4522565 were
nominally significant, and another SNP, rs10469735, was
significant at a trend level. Rs2004888 was the only marker that
was significant in both ISHDSF and ICCSS sample, and the
associated allele, T, was the same in the two samples. Overall, the
10 SNPs typed in GULP1 gene have similar frequencies and LD
structure in both ISHDSF and ICCSS samples (see Table S1,
Figure S1 in supplementary materials).
The GULP1 gene was the first gene in the
The ABCA1 and ABCA7 genes.
the ABCA1 and ABCA7 genes to test whether these genes in the
phagocytosis pathway may impact on risk for schizophrenia.
Rs4149324 in ABCA1 was nominally associated with the disease in
the ICCSS sample (p=0.0258) and rs2242436 in ABCA7 was
significant in the ISHDSF sample (p=0.0013) (supplementary
Table S2). Like the GULP1 gene, markers in ABCA1 and ABCA7
genes have similar frequencies and LD structure in the two Irish
samples (Table S1 and Figure S1).
We typed 11 and 5 SNPs for
Gene-gene interaction in the CED pathway
To test whether genes interact to increase risk for schizophrenia,
we examined gene-gene interaction by the multifactor dimension-
ality reduction pedigree disequilibrium test (MDR-PDT)  and
the multifactor dimensionality reduction method (MDR)  for
the ISHDSF and ICCSS samples respectively. In both samples, 36
SNPs typed in the MEGF10 (10 markers), GULP1 (10 markers),
ABCA1 (11 markers) and ABCA7 (5 markers) gene were included in
the analyses. We limited our search to 2- and 3-marker
interactions. In the MDR-PDT analyses of the ISHDSF sample,
we found a significant 2-marker interaction (p=0.011): SNPs
rs3858075 in the ABCA1 gene and rs2242436 in the ABCA7 gene
interact with each other to increase risk of the disease. In the
model, 3 genotype combinations of rs3858075 and rs2242436 (C/
C-G/G, C/T-G/G and T/T-G/A) were overrepresented in the
affected individuals (Figure 1). The overall OR of these groups
were 26.2, 95% CI 3.2–210.8, p=0.002 (Table 2). In the analyses
of the ICCSS sample, we found a significant 3-marker interaction
(p=0.006). These 3 markers come from the GULP1 (rs4522565),
ABCA1 (rs3858075) and MEGF10 (rs246896) genes. In this model,
the most abundant risk genotype group is T/T-C/C-C/T for
markers rs4522565-3858075-rs246896 (Figure 1). None of these
markers had significant main effects in our regression analysis. In
contrast, the 3-marker interaction term was significant with an OR
of 1.31, 95% CI 1.03–1.7, p=0.0120 (Table 2). SNP rs3858075 in
the ABCA1 was involved in both the 2-marker and 3-marker
interactions observed in the two samples. No gene-gene interaction
analyses were conducted for other samples since they were typed
only for rs2004888.
Table 1. Single marker association analyses of the GULP1 gene.
AlleleZ P*Trio-TTrio-NT AffSib UnafSib
Freq Ctrl FreqOR
rs6718697A 2.100.0356 140134 1195 1149 0.857 1.090.13950.889 0.8721.18
rs9808557C 2.340.0193 126 120 1120 10720.816 1.100.13810.854 0.8351.16
rs10469735T1.850.0649 144 13711751134 0.8531.090.06590.8890.8681.22
rs2004888T2.640.0083140132 11961143 0.8561.11 0.04370.889 0.866 1.25
rs6753371A1.99 0.046886 79654620 0.5111.15 0.45240.4960.4821.06
rs4413123G2.190.0282 8982 7216730.547 1.160.74090.463 0.4571.02
rs4522565T 0.430.6700136 13411421134 0.8301.02 0.00620.8730.8381.33
rs6714454A1.770.0761434215 0.0284.810.9040 0.020 0.0201.03
rs7595327G 0.820.4134131 1291075 1065 0.7921.03 0.12780.833 0.8121.16
rs8273C 0.66 0.5090134134 114211380.8721.00 0.7725 0.871 0.8750.97
Trio-T, trio transmitted; Trio-NT, trio not transmitted; Affsib, affected sib; UnafSib, unaffected sib; Freq, frequency; Ctrl, control; OR, odds ratio; OTR, odds of transmission
*P values #0.05 are in bold.
GULP1 and Schizophrenia
PLoS ONE | www.plosone.org2 September 2009 | Volume 4 | Issue 9 | e6875
Replication of rs2004888
Based on the results of the GULP1 gene, we selected rs2004888
for replication in other samples. In addition to the ICCSS, we
typed this marker in the German, Pittsburgh, Shanghai and
Kunming samples. As shown in Table 3, of the 5 replication
samples, only ICCSS and Kunming samples were nominally
significant, but the associated alleles were different in these two
samples. When the 3 Caucasian replication samples were
combined (3139 subjects, 1378 affected), the result was very
significant (p=0.0003). For the two Chinese samples, the Shanghai
sample had the same allele overrepresented in the affected
individuals as the Caucasian samples. But the Kunming sample
was in the opposite direction. When the Chinese samples
combined (cases =1458, controls =1740), the results were not
significant. When all 5 replication samples were combined (6131
(p=0.0057, Table 3) due to the Chinese samples. Given our
limited testing in these genes, the p value for the Caucasian
replication samples survived Bonferroni correction for experiment-
wide significance. But when the 2 Chinese samples were added,
while the overall p value was nominally significant, it would not
survive experiment-wide correction. We verified these results by
meta-analysis. When all samples were included in the meta-
analyses, rs2004888 was significantly associated with the disease
(OR=1.09, 95% CI 1.02–1.16, p=0.011). While the meta-
analysis results for all replication samples were not significant
(OR=1.06, 95% CI 0.98–1.13, p=0.152), the results for
Figure 1. MDR-PDT and MDR analysis of the MEGF10, GULP1, ABCA1 and ABCA7 genes in the ISHDSF (A) and ICCSS (B) samples. For
each cell, the count of affected individuals was plotted as the left bar, the count of unaffected was plotted as the right bar. The risk genotype
combinations, defined as ratio of affected/unaffected in each cell, were highlighted.
Table 2. Effect estimates of the gene-gene interaction.
SampleRegression Term Odds Ratio95% CI P-value
ISHDSFrs2242436 0.079 0.011–0.622 0.0159
rs2242436* rs3858075 26.1683.249–210.758 0.0022
rs246896 0.9470.813–1.103 0.4580
Table 3. Replication of rs2004888*.
Sample Case Cnt (Freq)Ctrl Cnt (Freq)T NT OR (OTR)P
ICCSS1380 (0.889) 1319 (0.866)–– 1.25 0.0437
Pittsburgh–– 331 (0.836) 325 (0.821)(1.11) 0.4383
German–– 457 (0.856)443 (0.830)(1.22)0.2265
Shanghai1693 (0.928) 1662 (0.915)–– 1.22 0.1300
Kunming885 (0.922) 1458 (0.943)––0.71 0.0364
Caucasian samples1380 (0.889)1319 (0.866)788 (0.847)768 (0.826) 1.25 (1.17)0.0003
Chinese samples 2578 (0.926)3120 (0.928)–– 0.97 0.7625
All replication samples3958 (0.913) 4439 (0.909)788 (0.847) 768 (0.826)1.05 (1.17) 0.0057
All samples3958 (0.913) 4439 (0.909)928 (0.855)1345 (0.846)1.05 (1.07)0.0030
*Cnt, count; ctrl, control; Freq, frequency; T, transmitted; NT, not transmitted; OTR, odds of transmission ratio.
GULP1 and Schizophrenia
PLoS ONE | www.plosone.org3September 2009 | Volume 4 | Issue 9 | e6875
Caucasian samples were significant (OR=1.15, 95% CI 1.03–
1.29, p=0.0140). These results were consistent with the combined
While there is compelling evidence that genetic factors are
involved in the etiology of schizophrenia, the involvement of
specific genes and variants remains elusive. Most recent efforts to
identify the risk genes focus on genome-wide association followed
by targeted replications. Using this strategy, some genes have been
identified as promising candidates in very large samples [42,43].
Our group has taken a different approach, focusing on the clearly
defined biological pathways and interactions. Using this approach,
we identified that both IL3 and CSF2RB are involved in the disease
[4,6], and independent studies [27,38,60] provide further support
that CSF2RB, CSF2RA and IL3RA, the common and interleukin
specific subunits of IL3 and CSF2 receptors, may also play an
etiological role. Collectively, these studies support a role for the
IL3/CSF2 pathway in schizophrenia. This study follows the same
strategy and extends our finding of MEGF10 involvement in
schizophrenia . MEGF10 is the human ortholog of CED-1, a
key component of apoptotic engulfment pathway in C. elegans.
Studies have shown that this phagocytic pathway is evolutionarily
conserved, and multiple orthologs of CED genes have been
identified in Drosophila , mouse , rat  and human
[28,39,62]. Furthermore, protein products of these ortholog genes
perform similar biological functions in the same pathway
[17,23,26,28,39,58,59,62,63,68,69]. To test whether other CED
orthologs in human are involved in schizophrenia, we conducted
association analyses of the GULP1, ABCA1 and ABCA7 genes that
are the human orthologs of the CED-6 and CED-7 genes. In this
study, we found nominal associations in all 3 genes. Most
importantly, rs2004888, a marker in the GULP1 gene, was
consistently significant (p=0.0003) in combined Caucasian
replication samples (3139 subjects, 1378 affected). In addition,
we found significant gene-gene interactions in both our ISHDSF
and ICCSS samples that involve all known human orthologs
(MEGF10, GULP1, ABCA1 and ABCA7) of the CED-1, CED-6 and
CED-7 pathway. Based on these data, we conclude that GULP1
and the CED-1, CED-6 and CED-7 pathway are involved in
schizophrenia. This finding is significant that it opens the door for
focused biological studies of the role of these genes in
There may be ethnic heterogeneity at the rs2004888 locus. Four
of the 6 samples used in this study are Caucasian and these
samples identify the same risk allele in both family and case-
control samples. Collectively, these 4 Caucasian samples produced
a highly significant association (p=0.0001, data not shown). In
contrast, the combined Chinese samples were not significant. We
compared the allele frequency between the Caucasian and
Chinese samples, while the major alleles are the same in both
Caucasian and Chinese samples, its frequency in the Caucasian
sample is significantly lower than that of the Chinese sample
(0.878 vs. 0.927). Since the two Chinese samples were not typed
for other SNPs in this region, we compared the haplotype
structure of GULP1 between Caucasian and Chinese populations
using the SNPs typed in the dbSNP database. In both populations,
rs2004888 is in high LD with other markers in this region.
However, the frequency of the undertransmitted haplotype in the
Caucasian population is significantly higher than that in the
Chinese population (0.142 vs. 0.039). These differences in LD
structure may be responsible for the differences in association
signal observed at this locus.
We have noticed that the OR at rs2004888 is close to some
promising candidates recently identified from genome-wide
association studies [42,43]. The ZNF804A gene (rs1344706) has
an OR of 1.12 and rs17101921 (near FGFR2 gene) has an OR of
1.17. At rs2004888, we have an OR of 1.25 for the Caucasian
case-control samples and an odds of transmission ratio of 1.17 in
the family samples. Consistent with these other studies, our results
clearly suggest that large samples are necessary to identify these
risk genes. This may explain why some individual samples did not
produce significant results.
The statistical interactions observed in this study are supported
by physical contact and biological modulation. In one study,
ABCA1 protein is found to physically interact with MEGF10
protein and functionally modulates the MEGF10-mediated
engulfment of apoptotic cells . In addition to phagocytosis,
both ABCA1 and ABCA7 are involved in lipid metabolism and
homeostasis , and there is evidence that these genes are
regulated in a coordinated fashion . We also notice that
multiple polymorphisms in the ABCA1 gene have been found to be
associated with Alzheimer’s disease [11,21,49,61], although the
relationship between the Alzheimer disease and schizophrenia
At this time, we don’t understand the mechanism how the
apoptotic engulfment pathway contributes to schizophrenia. Based
on the functions of this pathway, we can speculate how the
apoptotic clearance and engulfment pathway might be involved in
the disease. In the development of the central nervous system,
programmed cell death, i.e. apoptosis, is evident in shaping the
neuronal circuitries and functions [25,31,40]. Rapid clearance of
cell corpses is essential for maintaining tissue homeostasis and
preventing the release of potentially cytotoxic or antigenic
molecules from dying cells. Defects in cell corpse clearance are
closely associated with autoimmune and inflammatory responses
in C. elegans . There are reports that there are dysfunctions in
autoimmune system and elevated amount of inflammatory
cytokines in schizophrenia patients [20,41,45,52], suggesting that
autoimmune and inflammatory cytokines may be involved in the
disease. The defects in the apoptotic pathway may be a factor
contributing to dysfunctional autoimmune system and elevated
inflammatory cytokines observed in schizophrenia patients.
Specific to the genes, mutations in CED-1, CED-6 and CED-7
results in abnormal commisure patterns  and axon pruning
[1,30] in C. elegans and Drosophila. Since the genes are highly
conserved, it would not be a surprise that mutations in their
human orthologs (i.e. MEGF10, GULP1, ABCA1 and ABCA7)
would cause similar abnormal projections and synaptic connec-
In this study, we provide genetic evidence that the GULP1 gene
and the apoptotic engulfment pathway may be involved in
schizophrenia. Since the engulfment pathway is evolutionarily
conserved, many genetic and functional studies in model
organisms can provide new insights to understand the pathophys-
iology of schizophrenia. Following the guidance of these animal
model studies, it would be of great interest to examine the
functions of these genes in humans and to determine how the
dysfunction of these genes may lead to the disease.
Materials and Methods
the principles expressed in the Declaration of Helsinki. The study
was approved by Institutional Review Boards/Ethics Committees
at Virginia Commonwealth University (ISHDSF and ICCSS),
This study was conducted according to
GULP1 and Schizophrenia
PLoS ONE | www.plosone.org4September 2009 | Volume 4 | Issue 9 | e6875
University of Pittsburgh (the Pittsburgh sample), Western
Australian University (the German sample), the Zoology Institute
of Chinese Academy of Science (Kunming sample) and Jiaotong
University (Shanghai sample). All patients provided written
informed consent for the collection of samples and subsequent
The ISHDSF sample.
The Irish study of high density
schizophrenia families (ISHDSF) was collected in Northern
Ireland, United Kingdomand
Phenotypes were assessed using DSM-III-R. The diagnoses were
originally classified a hierarchy of 10 categories reflecting the
probable genetic relationship of these syndromes to classic
schizophrenia. In this study, we used the narrow disease
definition, which include only categories D1 and D2 in the
original classification. In this narrow definition, only patients met
schizoaffective disorder and simple schizophrenia were classified
as affected. The sample contained 273 pedigrees and about 1350
subjects had DNA sample for genotyping. Of them, 515 were
diagnosed as affected using definition described above. Detailed
descriptions of the sample have been published previously .
The Irish case-control study of
schizophrenia (ICCSS) sample was collected in the same
geographic regions as that of the ISHDSF sample. The affected
subjects were selected from in-patient and out-patient psychiatric
facilities in the Republic of Ireland and Northern Ireland. Subjects
were eligible for inclusion if they had a diagnosis of schizophrenia
or poor-outcome schizoaffective disorder by DSM-III-R criteria,
and the diagnosis was confirmed by a blind expert diagnostic
review. Controls, selected from several sources, including blood
donation centers, were included if they denied a lifetime history of
schizophrenia. However, the controls were not screened by
clinicians. Both cases and controls were included only if they
reported all four grandparents as being born in Ireland or the
United Kingdom. Using these criteria, a total of 1417 subjects (625
affected subjects and 792 controls) were included in this study.
The Germanfamily sample.
consisted of two subsamples, 79 affected sib pairs with parents
and 125 proband-parents trios. Of the 79 sib pairs, 54 were used in
the linkage study where overlapping linkage peak was found
between the ISHDSF and German sib samples [53,57]. The trios
were selected from a larger trio sample  with positive family
history (defined as at least one first or second degree relative of the
proband meeting the DSM IV criteria of schizophrenia or
Pittsburgh family sample.
contained 247 nuclear families (case and parents) with a total of
729 subjects with DNA for genotyping. Subjects were recruited
from inpatients and outpatients facilities within a 500 mile radius
of Pittsburgh and met DSM IV criteria of schizophrenia or
schizoaffective disorder. All probands self-reported as with
Caucasian ancestry .
The Shanghai case control sample.
schizophrenia and 946 control individuals. All subjects were of
Han Chinese origin. A clinical interview was administered by two
independent senior psychiatrists to all patients according to the
criteria of the DSM-IV. All patients were policlinic and recruited
from the Shanghai Mental Health Center, East China. The
healthy controls were drawn from the general population in the
East China. None had a history of psychotic disorders. Participants
were fully informed of, and gave written consent for, the genetics
analysis, which was reviewed and approved by the Shanghai
Ethics Committee of Human Genetic Resources.
The Shanghai case
The Kunming case control sample.
control sample consisted of 516 schizophrenia patients and 794
normal controls. All affected subjects were self-reported as Han
Chinese and were recruited from Yunnan Mental Health
Hospital, Kunming city, Yunnan Province, China. The patients
were diagnosed by two trained psychiatrists independently and
met the criteria of the Diagnostic and Statistical Manual of Mental
Disorders, 4th edition (DSM-IV) for schizophrenia. Subjects with
substance-induced psychotic disorders, learning disabilities, head
injuries and other symptomatic psychoses were excluded. The
controls were recruited from the same province, and all were self-
reported as Han Chinese.
The Kunming case
Marker selection and genotyping
We used the HapMap data (http://www.hapmap.org/) and the
available assays developed by Applied BioSystems to assist in our
selection of markers. For each gene, we decided to cover the
transcribed genomic region plus 50 kb on both ends. We
downloaded the Caucasian HapMap data from HapMap website,
and analyzed the LD structure with the HaploView program .
We selected SNPs (r2.=0.80) tagging major haplotypes (with
frequency .=5%) for each gene. SNPs tagging minor haplotypes
(those with frequencies less than 5%) were not considered. The
main reasons we considered only major haplotypes were twofold.
First, minor haplotypes with low frequencies are normally less
reliable because computational inference usually does not do a
very good job when the allele frequencies are low. Second, given
the sample size of our Irish samples, it is not very likely that we
have sufficient power to detect these low frequency haplotypes.
Based on these criteria, we obtained 10 SNPs for GULP1, 11 SNPs
for ABCA7 and 5 SNPs for ABCA7.
Five samples used in this study (ISHDSF, ICCSS, the German,
the Pittsburgh and the Shanghai samples) were typed with the
TaqMan method . The assays used were either validated
assays or custom designed assays developed by Applied BioSys-
tems Corporation (Foster city, CA). Genotypes were scored with a
semi-automated procedures developed in our lab  or with
software from the commercial providers. The Kunming sample
was typed with single base extension with capillary electrophoresis.
The PCR primers used were TTTTGGATTCGGCGGAT-
TAGG and CTGGAAGTTCGCTCCTGGGTC, and the exten-
sion primer used was TTTTTTTTTTTTACCTTACCGCCCC
TCGGGATATCAGCTTCT. All markers typed were checked
for deviation from the Hardy-Weinberg Equilibrium (HWE) and
Mendelian errors by the PEDSTATS program . Detail
information of these typed markers in the ISHDSF and ICCSS
was listed in Table S1 in the supplementary materials.
Single marker association tests.
UNPHASED program (version 2.4, PDTPHASE module) 
to analyze the ISHDSF sample. In these analyses, both vertical
and horizontal transmissions were included. The p values reported
were based on weighing all families equally (the ave option in the
program). For the other 5 individual samples and the combined
samples, the newer version of the UNPHASED program (version
3.11), which was designed to analyze case-control samples, family
samples or combined case-control and family samples , was
used to analyze single marker associations. In this analysis,
ethnicity was used as a covariate. Meta-analyses were conducted
using the Comprehensive Meta-Analysis software 2.0 from Biostat
(Englewood, NJ, USA) (www.meta-analysis.com/). We used the
HaploView program (version 4.0)  to estimate pairwise LD and
We used the pedigree
implemented (PDT) as inthe
GULP1 and Schizophrenia
PLoS ONE | www.plosone.org5 September 2009 | Volume 4 | Issue 9 | e6875
to illustrate haplotype blocks. The haplotype blocks were
partitioned by the confidence interval algorithm .
reduction pedigree disequilibrium test (MDR-PDT)  was
used to explore multi-locus associations in the ISHDSF sample.
The MDR-PDT was a within-family measure of indirect or direct
association between genotype and disease. As described previously
, the PDT statistic  functioned within the framework of the
MDR algorithm. Genotypes were classified as high and low-risk by
comparing the PDT statistic to a threshold of 0, where positive
statistics indicate evidence for association at that genotype. The
MDR-PDT statistic was then calculated for the pooled high-risk
genotypes for each set of loci. The models were ordered and
evaluated by MDR-PDT statistics. A permutation test was applied
to estimate the significance of the result, which inherently adjusted
for the size of the search performed. The permutation test
consisted of randomizing status for offspring, holding the
proportion of affected individuals constant across permutations,
calculating the statistic, and repeating many times to estimate the
distribution of the null hypothesis. The test based on the
permutation procedure would have the correct type I error rate,
even for sparse data. This validity was due to all contingency table
cells from each permutation containing the same number of
observations as those from the unpermuted data. In this study, we
limited our search to include only 2- and 3-locus interactions. To
estimate the degree of effect modification between the MDR-PDT
model SNPs, we fitted conditional logistic regression models with
adjustment for residual correlation among affected offspring due to
linkage . For model fitting, the genotypes were specified as
high or low-risk, denoted as exposed (1) or unexposed (0)
respectively, by MDR-PDT analysis of individual loci.
For the ICCSS sample, we used the multifactor dimensionality
reduction method (MDR) [37,47,48] to examine potential gene-
gene interactions. MDR was a nonparametric method that
performs an exhaustive search of all possible interactions and
maps the data into a single dimension relevant to association.
Similar to the MDR-PDT procedure, significance was evaluated
via permutation testing, which inherently adjusted for the multiple
comparisons from the search performed. As in the family sample,
we limited our search to 2- and 3-locus interactions. The effect of
interaction was estimated by generalized linear regression model
as implemented in the SPSS software (SPSS for Windows, version
ICCSS samples for GULP1, ABCA1 and ABCA7 zgenes.
Found at: doi:10.1371/journal.pone.0006875.s001 (0.51 MB
A comparison of LD between the ISHDSF and
Found at: doi:10.1371/journal.pone.0006875.s002 (0.07 MB
Found at: doi:10.1371/journal.pone.0006875.s003 (0.06 MB
Single marker association analyses in the ABCA1 and
We are grateful to the patients and their families for participating in this
study. The Northern Ireland Blood Transfusion Service assisted with
collection of control sample.
Conceived and designed the experiments: XC. Performed the experiments:
CS QC AS SGS RC XjL. Analyzed the data: XC TLE. Contributed
reagents/materials/analysis tools: AO DW AHF KVC VLN AS SGS
DBW RC WT YS LH XjL BS ZZ KSK. Wrote the paper: XC.
1. Awasaki T, Tatsumi R, Takahashi K, Arai K, Nakanishi Y, et al. (2006)
Essential role of the apoptotic cell engulfment genes draper and ced-6 in
programmed axon pruning during Drosophila metamorphosis. Neuron 50:
2. Banerjee H, Hawkins Z, Johnson T, Eley S, Alikhan A, et al. (2003)
Identification of a mouse orthologue of the CED-6 gene of Caenorhabditis
elegans. Plasmid 49: 30–33.
3. Barrett JC, Fry B, Maller J, Daly MJ (2005) Haploview: analysis and
visualization of LD and haplotype maps. Bioinformatics 21: 263–265.
4. Chen Q, Wang X, O’Neill FA, Walsh D, Fanous A, et al. (2007) Association
study of CSF2RB with schizophrenia in Irish family and case - control samples.
Mol Psychiatry 13: 930–938.
5. Chen Q, Wang X, O’Neill FA, Walsh D, Kendler KS, et al. (2008) Is the
histidine triad nucleotide-binding protein 1 (HINT1) gene a candidate for
schizophrenia? Schizophr Res 106: 200–207.
6. Chen X, Wang X, Hossain S, O’Neill AF, Walsh D, et al. (2007) Interleukin 3
and schizophrenia: The impact of sex and family history. Mol Psychiatry 12:
7. Chen X, Wang X, Chen Q, Williamson V, van den OE, et al. (2008) MEGF10
association with schizophrenia. Biol Psychiatry 63: 441–448.
8. Chen X, Wang X, Hossain S, O’Neill FA, Walsh D, et al. (2006) Haplotypes
spanning SPEC2, PDZ-G EF2 and ACSL6 genes are associated with
schizophrenia. Hum Mol Genet 15: 3329–3342.
9. Chen X, Wang X, Sun C, Chen Q, O’Neill FA, et al. (2008) FBXL21
association with schizophrenia in Irish family and case-control samples.
Am J Med Genet B Neuropsychiatr Genet 147B: 1231–1237.
10. Chowdari KV, Mirnics K, Semwal P, Wood J, Lawrence E, et al. (2002)
Association and linkage analyses of RGS4 polymorphisms in schizophrenia.
Hum Mol Genet 11: 1373–1380.
11. Chu LW, Li Y, Li Z, Tang AY, Cheung BM, et al. (2007) A novel intronic
polymorphism of ABCA1 gene reveals risk for sporadic Alzheimer’s disease in
Chinese. Am J Med Genet B Neuropsychiatr Genet 144B: 1007–1013.
12. Conradt B, Xue D (2005) Programmed cell death. WormBook. pp 1–13.
13. Dudbridge F (2003) Pedigree disequilibrium tests for multilocus haplotypes.
Genet Epidemiol 25: 115–121.
14. Dudbridge F (2008) Likelihood-based association analysis for nuclear families
and unrelated subjects with missing genotype data. Hum Hered 66: 87–98.
15. Gabriel SB, Schaffner SF, Nguyen H, Moore JM, Roy J, et al. (2002) The
structure of haplotype blocks in the human genome. Science 296: 2225–2229.
16. Hamon Y, Trompier D, Ma Z, Venegas V, Pophillat M, et al. (2006)
Cooperation between Engulfment Receptors: The Case of ABCA1 and
MEGF10. PLoS ONE 1: e120-.
17. Haskins KA, Russell JF, Gaddis N, Dressman HK, Aballay A (2008) Unfolded
protein response genes regulated by CED-1 are required for Caenorhabditis
elegans innate immunity. Dev Cell 15: 87–97.
18. Hayashi M, be-Dohmae S, Okazaki M, Ueda K, Yokoyama S (2005)
Heterogeneity of high density lipoprotein generated by ABCA1 and ABCA7.
J Lipid Res 46: 1703–1711.
19. Jehle AW, Gardai SJ, Li S, Linsel-Nitschke P, Morimoto K, et al. (2006) ATP-
binding cassette transporter A7 enhances phagocytosis of apoptotic cells and
associated ERK signaling in macrophages. J Cell Biol 174: 547–556.
20. Jones AL, Mowry BJ, Pender MP, Greer JM (2005) Immune dysregulation and
self-reactivity in schizophrenia: do some cases of schizophrenia have an
autoimmune basis? Immunol Cell Biol 83: 9–17.
21. Katzov H, Chalmers K, Palmgren J, Andreasen N, Johansson B, et al. (2004)
Genetic variants of ABCA1 modify Alzheimer disease risk and quantitative traits
related to beta-amyloid metabolism. Hum Mutat 23: 358–367.
22. Kendler KS, Myers JM, O’Neill FA, Martin R, Murphy B, et al. (2000) Clinical
features of schizophrenia and linkage to chromosomes 5q, 6p, 8p, and 10p in the
Irish Study of High-Density Schizophrenia Families. Am J Psychiatry 157:
23. Kinchen JM, Cabello J, Klingele D, Wong K, Feichtinger R, et al. (2005) Two
pathways converge at CED-10 to mediate actin rearrangement and corpse
removal in C. elegans. Nature 434: 93–99.
24. Kinchen JM, Ravichandran KS (2007) Journey to the grave: signaling events
regulating removal of apoptotic cells. J Cell Sci 120: 2143–2149.
25. Kurant E, Axelrod S, Leaman D, Gaul U (2008) Six-microns-under acts
upstream of Draper in the glial phagocytosis of apoptotic neurons. Cell 133:
GULP1 and Schizophrenia
PLoS ONE | www.plosone.org6 September 2009 | Volume 4 | Issue 9 | e6875
26. Lamitina T, Cherry S (2008) Dangerous liaisons: the apoptotic engulfment
receptor CED-1 links innate immunity to the unfolded protein response. Dev
Cell 15: 3–4.
27. Lencz T, Morgan TV, Athanasiou M, Dain B, Reed CR, et al. (2007)
Converging evidence for a pseudoautosomal cytokine receptor gene locus in
schizophrenia. Mol Psychiatry 12: 572–580.
28. Liu QA, Hengartner MO (1999) Human CED-6 encodes a functional
homologue of the Caenorhabditis elegans engulfment protein CED-6. Curr
Biol 9: 1347–1350.
29. Livak KJ (1999) Allelic discrimination using fluorogenic probes and the 59
nuclease assay. Genet Anal 14: 143–149.
30. MacDonald JM, Beach MG, Porpiglia E, Sheehan AE, Watts RJ, et al. (2006)
The Drosophila cell corpse engulfment receptor Draper mediates glial clearance
of severed axons. Neuron 50: 869–881.
31. Mallat M, Marin-Teva JL, Cheret C (2005) Phagocytosis in the developing
CNS: more than clearing the corpses. Curr Opin Neurobiol 15: 101–107.
32. Mangahas PM, Zhou Z (2005) Clearance of apoptotic cells in Caenorhabditis
elegans. Semin Cell Dev Biol 16: 295–306.
33. Martin ER, Bass MP, Gilbert JR, Pericak-Vance MA, Hauser ER (2003)
Genotype-based association test for general pedigrees: the genotype-PDT. Genet
Epidemiol 25: 203–213.
34. Martin ER, Monks SA, Warren LL, Kaplan NL (2000) A test for linkage and
association in general pedigrees: the pedigree disequilibrium test. Am J Hum
Genet 67: 146–154.
35. Martin ER, Ritchie MD, Hahn L, Kang S, Moore JH (2006) A novel method to
identify gene-gene effects in nuclear families: the MDR-PDT. Genet Epidemiol
36. Martins-Silva C, Ferreira LT, Cyr M, Koenen J, Fernandes DR, et al. (2006) A
rat homologue of CED-6 is expressed in neurons and interacts with clathrin.
Brain Res 1119: 1–12.
37. Moore JH (2004) Computational analysis of gene-gene interactions using
multifactor dimensionality reduction. Expert Rev Mol Diagn 4: 795–803.
38. Moskvina V, Craddock N, Holmans P, Nikolov I, Pahwa JS, et al. (2008) Gene-
wide analyses of genome-wide association data sets: evidence for multiple
common risk alleles for schizophrenia and bipolar disorder and for overlap in
genetic risk. Mol Psychiatry -.
39. Moynault A, Luciani MF, Chimini G (1998) ABC1, the mammalian homologue
of the engulfment gene ced-7, is required during phagocytosis of both necrotic
and apoptotic cells. Biochem Soc Trans 26: 629–635.
40. Napoli I, Neumann H (2008) Microglial clearance function in health and
disease. Neuroscience -.
41. Nunes SO, Matsuo T, Kaminami MS, Watanabe MA, Reiche EM, et al. (2006)
An autoimmune or an inflammatory process in patients with schizophrenia,
schizoaffective disorder, and in their biological relatives. Schizophr Res-.
42. O’Donovan MC, Craddock N, Norton N, Williams H, Peirce T, et al. (2008)
Identification of loci associated with schizophrenia by genome-wide association
and follow-up. Nat Genet 40: 1053–1055.
43. O’Donovan MC, Norton N, Williams H, Peirce T, Moskvina V, et al. (2009)
Analysis of 10 independent samples provides evidence for association between
schizophrenia and a SNP flanking fibroblast growth factor receptor 2. Mol
Psychiatry 14: 30–36.
44. Petryshen TL, Middleton FA, Tahl AR, Rockwell GN, Purcell S, et al. (2005)
Genetic investigation of chromosome 5q GABAA receptor subunit genes in
schizophrenia. Mol Psychiatry 10: 1074–88, 1057.
45. Potvin S, Stip E, Sepehry AA, Gendron A, Bah R, et al. (2008) Inflammatory
cytokine alterations in schizophrenia: a systematic quantitative review. Biol
Psychiatry 63: 801–808.
46. Ritchie MD, Hahn LW, Moore JH (2003) Power of multifactor dimensionality
reduction for detecting gene-gene interactions in the presence of genotyping
error, missing data, phenocopy, and genetic heterogeneity. Genet Epidemiol 24:
47. Ritchie MD, Hahn LW, Moore JH (2003) Power of multifactor dimensionality
reduction for detecting gene-gene interactions in the presence of genotyping
error, missing data, phenocopy, and genetic heterogeneity. Genet Epidemiol 24:
48. Ritchie MD, Hahn LW, Roodi N, Bailey LR, Dupont WD, et al. (2001)
Multifactor-dimensionality reduction reveals high-order interactions among
estrogen-metabolism genes in sporadic breast cancer. Am J Hum Genet 69:
49. Rodriguez-Rodriguez E, Mateo I, Llorca J, Sanchez-Quintana C, Infante J, et
al. (2007) Association of genetic variants of ABCA1 with Alzheimer’s disease risk.
Am J Med Genet B Neuropsychiatr Genet 144B: 964–968.
50. Savill J, Dransfield I, Gregory C, Haslett C (2002) A blast from the past:
clearance of apoptotic cells regulates immune responses. Nat Rev Immunol 2:
51. Schmitz C, Kinge P, Hutter H (2007) Axon guidance genes identified in a large-
scale RNAi screen using the RNAi-hypersensitive Caenorhabditis elegans strain
nre-1(hd20) lin-15b(hd126). Proc Natl Acad Sci U S A 104: 834–839.
52. Schmitz T, Chew LJ (2008) Cytokines and myelination in the central nervous
system. ScientificWorldJournal 8: 1119–1147.
53. Schwab SG, Eckstein GN, Hallmayer J, Lerer B, Albus M, et al. (1997) Evidence
suggestive of a locus on chromosome 5q31 contributing to susceptibility for
schizophrenia in German and Israeli families by multipoint affected sib-pair
linkage analysis. Mol Psychiatry 2: 156–160.
54. Siegmund KD, Langholz B, Kraft P, Thomas DC (2000) Testing linkage
disequilibrium in sibships. Am J Hum Genet 67: 244–248.
55. Stefansson H, Sigurdsson E, Steinthorsdottir V, Bjornsdottir S, Sigmundsson T,
et al. (2002) Neuregulin 1 and susceptibility to schizophrenia. Am J Hum Genet
56. Straub RE, Jiang Y, MacLean CJ, Ma Y, Webb BT, et al. (2002) Genetic
variation in the 6p22.3 gene DTNBP1, the human ortholog of the mouse
dysbindin gene, is associated with schizophrenia. Am J Hum Genet 71: 337–348.
57. Straub RE, MacLean CJ, O’Neill FA, Walsh D, Kendler KS (1997) Support for
a possible schizophrenia vulnerability locus in region 5q22-31 in Irish families.
Mol Psychiatry 2: 148–155.
58. Su HP, Brugnera E, Van CW, Smits E, Hengartner M, et al. (2000)
Identification and characterization of a dimerization domain in CED-6, an
adapter protein involved in engulfment of apoptotic cells. J Biol Chem 275:
59. Su HP, Nakada-Tsukui K, Tosello-Trampont AC, Li Y, Bu G, et al. (2002)
Interaction of CED-6/GULP, an adapter protein involved in engulfment of
apoptotic cells with CED-1 and CD91/low density lipoprotein receptor-related
protein (LRP). J Biol Chem 277: 11772–11779.
60. Sun S, Wang F, Wei J, Cao LY, Wu GY, et al. (2008) Association between
interleukin-3 receptor alpha polymorphism and schizophrenia in the Chinese
population. Neurosci Lett 440: 35–37.
61. Sundar PD, Feingold E, Minster RL, DeKosky ST, Kamboh MI (2007) Gender-
specific association of ATP-binding cassette transporter 1 (ABCA1) polymor-
phisms with the risk of late-onset Alzheimer’s disease. Neurobiol Aging 28:
62. Suzuki E, Nakayama M (2007) MEGF10 is a mammalian ortholog of CED-1
that interacts with clathrin assembly protein complex 2 medium chain and
induces large vacuole formation. Exp Cell Res 313: 3729–3742.
63. Suzuki E, Nakayama M (2007) The mammalian Ced-1 ortholog MEGF10/
KIAA1780 displays a novel adhesion pattern. Exp Cell Res 313: 2451–2464.
64. Takahashi K, Kimura Y, Nagata K, Yamamoto A, Matsuo M, et al. (2005) ABC
proteins: key molecules for lipid homeostasis. Med Mol Morphol 38: 2–12.
65. van den Oord EJ, Jiang Y, Riley BP, Kendler KS, Chen X (2003) FP-TDI SNP
scoring by manual and statistical procedures: a study of error rates and types.
Biotechniques 34: 610–20, 622.
66. Wigginton JE, Cutler DJ, Abecasis GR (2005) A note on exact tests of Hardy-
Weinberg equilibrium. Am J Hum Genet 76: 887–893.
67. Yu X, Lu N, Zhou Z (2008) Phagocytic receptor CED-1 initiates a signaling
pathway for degrading engulfed apoptotic cells. PLoS Biol 6: e61-.
68. Yu X, Odera S, Chuang CH, Lu N, Zhou Z (2006) C. elegans Dynamin
mediates the signaling of phagocytic receptor CED-1 for the engulfment and
degradation of apoptotic cells. Dev Cell 10: 743–757.
69. Zhou Z, Hartwieg E, Horvitz HR (2001) CED-1 is a transmembrane receptor
that mediates cell corpse engulfment in C. elegans. Cell 104: 43–56.
GULP1 and Schizophrenia
PLoS ONE | www.plosone.org7 September 2009 | Volume 4 | Issue 9 | e6875