Genome-wide expression profiling of lymphoblastoid cell lines distinguishes different forms of autism and reveals shared pathways.

Abstract

Autism is a heterogeneous condition that is likely to result from the combined effects of multiple genetic factors interacting with environmental factors. Given its complexity, the study of autism associated with Mendelian single gene disorders or known chromosomal etiologies provides an important perspective. We used microarray analysis to compare the mRNA expression profile in lymphoblastoid cells from males with autism due to a fragile X mutation (FMR1-FM), or a 15q11-q13 duplication (dup(15q)), and non-autistic controls. Gene expression profiles clearly distinguished autism from controls and separated individuals with autism based on their genetic etiology. We identified 68 genes that were dysregulated in common between autism with FMR1-FM and dup(15q). We also identified a potential molecular link between FMR1-FM and dup(15q), the cytoplasmic FMR1 interacting protein 1 (CYFIP1), which was up-regulated in dup(15q) patients. We were able to confirm this link in vitro by showing common regulation of two other dysregulated genes, JAKMIP1 and GPR155, downstream of FMR1 or CYFIP1. We also confirmed the reduction of the Jakmip1 protein in Fmr1 knock-out mice, demonstrating in vivo relevance. Finally, we showed independent confirmation of roles for JAKMIP1 and GPR155 in autism spectrum disorders (ASDs) by showing their differential expression in male sib pairs discordant for idiopathic ASD. These results provide evidence that blood derived lymphoblastoid cells gene expression is likely to be useful for identifying etiological subsets of autism and exploring its pathophysiology.

Full text

Page 1

Genome-wide expression profiling of
lymphoblastoid cell lines distinguishes different
forms of autism and reveals shared pathways{
Yuhei Nishimura1,2,3, Christa L. Martin7, Araceli Vazquez Lopez2, Sarah J. Spence1,4,5,
Ana Isabel Alvarez-Retuerto4, Marian Sigman4,6, Corinna Steindler8, Sandra Pellegrini8,
N. Carolyn Schanen9, Stephen T. Warren7and Daniel H. Geschwind1,2,3,4,*
1Center for Autism Research and Treatment,2Program in Neurogenetics, Department of Neurology,3Center for
Neurobehavioral Genetics,4Department of Psychiatry,5Department of Pediatrics,6Department of Psychology, Semel
Institute for Neuroscience and Human Behavior, David Geffen School of Medicine, University of California Los
Angeles, Los Angeles, CA 90095, USA,7Department of Human Genetics, Emory University, Atlanta, GA 30322, USA,
8Unite de Signalisation des Cytokines, Institut Pasteur, Paris 75724, France and9Center for Pediatric Research,
Nemours Biomedical Research, Alfred I. duPont Hospital for Children, Wilmington, DE 19803, USA
Received February 5, 2007; Revised and Accepted March 20, 2007
Autism is a heterogeneous condition that is likely to result from the combined effects of multiple genetic fac-
tors interacting with environmental factors. Given its complexity, the study of autism associated with
Mendelian single gene disorders or known chromosomal etiologies provides an important perspective. We
used microarray analysis to compare the mRNA expression profile in lymphoblastoid cells from males
with autism due to a fragile X mutation (FMR1-FM), or a 15q11–q13 duplication (dup(15q)), and non-autistic
controls. Gene expression profiles clearly distinguished autism from controls and separated individuals with
autism based on their genetic etiology. We identified 68 genes that were dysregulated in common between
autism with FMR1-FM and dup(15q). We also identified a potential molecular link between FMR1-FM and
dup(15q), the cytoplasmic FMR1 interacting protein 1 (CYFIP1), which was up-regulated in dup(15q) patients.
We were able to confirm this link in vitro by showing common regulation of two other dysregulated genes,
JAKMIP1 and GPR155, downstream of FMR1 or CYFIP1. We also confirmed the reduction of the Jakmip1
protein in Fmr1 knock-out mice, demonstrating in vivo relevance. Finally, we showed independent confir-
mation of roles for JAKMIP1 and GPR155 in autism spectrum disorders (ASDs) by showing their differential
expression in male sib pairs discordant for idiopathic ASD. These results provide evidence that blood
derived lymphoblastoid cells gene expression is likely to be useful for identifying etiological subsets of
autism and exploring its pathophysiology.
INTRODUCTION
Genetic factors are significant determinants of autism spec-
trum disorders (ASDs) pathophysiology (1–4). Yet, identifi-
cation of causal genes has been hampered by genetic and
phenotypic heterogeneity (1–18). Thus, it seems reasonable
to accelerate the gene discovery process by using combi-
nations of experimental approaches, such as the study of
‘single gene’ or more simple causes, such as chromosomal
copy number imbalances whose phenotypes include ASD
(1–4). One such disorder is fragile X syndrome (FXS)
(1–4,19), which is caused by an expansion of the trinucleotide
repetitive sequence (CGG)n in the promoter region of the
fragile X mental retardation 1 (FMR1) gene located at
Xq27.3 (20). This mutation causes a significant deficit of the
FMR1 protein (FMRP) and a phenotype including cognitive
# The Author 2007. Published by Oxford University Press. All rights reserved.
For Permissions, please email: journals.permissions@oxfordjournals.org
{The microarray expression data from this study have been submitted to GEO under accession number GSE7329.
*To whom correspondence should be addressed at: 710 Westwood Plaza, Los Angeles, CA 90095-1769, USA. Tel: þ1 3102066814; Fax: þ1
3102672401; Email: dhg@ucla.edu
Human Molecular Genetics, 2007, Vol. 16, No. 14
doi:10.1093/hmg/ddm116
Advance Access published on May 21, 2007
1682–1698
at Universiti Teknologi Malaysia on May 18, 2011
hmg.oxfordjournals.org
Downloaded from

Page 2

impairment and other behavioral abnormalities that overlap
with ASD. The prevalence of ASD among FXS cases has
been estimated at 15–33% (21,22) and ?1–3% of those
with autism and no obvious physical features of FXS are
found to have FMR1-FM (19,23).
Another disorder associated with susceptibility for ASD is
a maternally inherited duplication of 15q11–q13 (dup(15q))
(1–4,24). Multiple repeat elements within the region mediate
a variety of rearrangements, including interstitial dupli-
cations, interstitial triplications and supernumerary isodi-
centric marker chromosomes (25). Dup(15q) occurs with
an estimated frequency of 1:600 children with developmental
delay (26) and is the most common copy number variation
causing ASD (3,24). Several lines of evidence (24,27)
suggest that dysregulation of non-imprinted genes in the
duplicated region may contribute to the autistic phenotype
observed in dup(15q).
Therefore, we reasoned that the identification of genes
whose expression is dysregulated by both FMR1-FM and
dup(15q) may be relevant to ASD, since the two genetic
abnormalities represent cases where single mutations, either
a trinucleotide repeat or copy number variation (28), cause
ASD. We also wanted to examine, as a proof of principle,
whether lymphoblast gene expression profiles identified by
microarrayscoulddifferentiate
‘simple’ causes of autism from each other and controls. This
would provide a basis for further application of these
methods in idiopathic autism, where more multigenic inheri-
tance and environmental influences may be involved (1–4).
Recently, several studies have suggested that lymphoblas-
toid cells can be used to detect biologically plausible corre-
lations between candidate
diseases,includingRettsyndrome
X-linked mental retardation (30), bipolar disorder (31),
FXS (32) and dup(15q) (27). In the present study, we inves-
tigated whether gene expression profiles of lymphoblastoid
cells could be used to: (i) differentiate autistic subjects
who were ascertained and diagnosed as having ASD in the
Autism Genetic Resource Exchange (AGRE) (33) repository
into etiological categories (FMR1-FM and dup(15q)) and (ii)
identify common genes and pathways shared by FMR1-FM
and dup(15q) that might be relevant to autism pathophysiol-
ogy. Here, we demonstrate that gene expression profiles
were able to clearly distinguish individuals based on their
etiology. We also identified 68 genes dysregulated in both
autism with FMR1-FM and dup(15q). Interestingly, we
identified a molecular connection between FMR1-FM and
dup(15q), CYFIP1, which was significantly induced in
dup(15q) and is known to antagonize certain aspects of
FMRP function (34). We further demonstrated that the
expression of Janus kinase and microtubule interacting
protein 1 (JAKMIP1) and G protein-coupled receptor 155
(GPR155) were commonly dysregulated by the reduction
of FMR1 or induction of CYFIP1 in vitro. The expression
of Jakmip1 was also dysregulated in the brain of the Fmr1
knock-out mouse. Finally, we were able to show that
JAKMIP1 and GPR155 were dysregulated in males with
ASDs relative to their non-affected siblings, providing inde-
pendent confirmation to suggest these genes are associated
with ASD.
thesesingle mutation
genesand neuropsychiatric
(29),non-specific
RESULTS
Hierarchical clustering and principal component analysis
distinguished individuals based on genetic etiology
We analyzed the whole-genome mRNA expression profile in
lymphoblastoid cells from 15 autistic males (8 autistic males
with FMR1-FM and 7 autistic males with dup(15q)) and 15
non-autistic controlmales from
Material, Table S1) using Agilent Whole Genome Human
Microarrays. Overall, of 41 000 probes analyzed, 31 044
probes, representing 23 822 genes, were expressed in the lym-
phoblastoid cells. To find genes that were differentially
expressed across the three subject groups, the expression
profile of the lymphoblastoid cells was subjected to Analysis
of Variance (ANOVA) (35). The ANOVA identified 293
probes (277 genes) below a defined false discovery rate
(FDR) threshold of 5% (Supplementary Material, Table S2).
It has been shown that the expression of FMR1 is decreased
in lymphoblastoid cells with FMR1-FM (36) and that the
expression of UBE3A is increased in lymphoblastoid cells
with dup(15q) (37). Concordant with these reports, FMR1 and
UBE3A were among the 293 differentially expressed probes,
providing independent controls for the microarray analysis.
As shown in Figure 1A and B, hierarchical clustering using
the 293 probes clearly classified individuals based on their
genotype. The 293 probes were also subjected to principal
component analysis (PCA). As shown in Figure 1C, three
dominant PCA components that contained 70% of the var-
iance in the data matrix clearly separated individuals based
on genetic etiology. In this plot, the first principal component
axis accounted for 56% of the variance in the data set and
clearly separated autism with FMR1-FM and dup(15q) from
controls, whereas the second principal component (PC2)
accounted for 10% of the variance and segregated autism
with FMR1-FM from autism with dup(15q). The top 10
genes contributing to PC2 include FMR1, UBE3A, CYFIP1,
non-imprinted inPrader-Willi/Angelman
(NIPA2) and hect domain and RLD 2 (HERC2). The latter
four genes are all located in 15q11–q13. These results
suggest that the selective reduction of FMR1 and the selective
induction of the four genes located in 15q11–q13 differen-
tiated autism with FMR1-FM from autism with dup(15q).
These data provide a critical proof of principle that the gene
expression profile of lymphoblastoid cells is useful in the
study of autism.
AGRE (Supplementary
syndrome2
Microarray analyses revealed the significant overlap of
FMR1-FM and dup(15q)
To identify the set of the most robustly differentially expressed
genes in each group, we applied three different statistical
methods, ANOVA, Significant Analysis of Microarray
(SAM) (38) and Rank Product Analysis (RankProd) (39).
SAM is a modified t-test statistic, whereas RankProd is a non-
parametric statistic that detects items that are consistently
highly ranked in a number of lists. SAM identified 5139
probes and 1281 probes as significant (FDR , 5%) in
autism with FMR1-FM and dup(15q), respectively (Fig. 2A
and B, Supplementary Material, Tables S3 and S4). RankProd
Human Molecular Genetics, 2007, Vol. 16, No. 141683
at Universiti Teknologi Malaysia on May 18, 2011
hmg.oxfordjournals.org
Downloaded from

Page 3

identified 2281 probes and 1444 probes as significant
(FDR , 5%) in autism with FMR1-FM and dup(15q), respect-
ively (Fig. 2A and B, Supplementary Material, Tables S5 and
S6). The combination of ANOVA, SAM and RankProd ident-
ified 146 probes (120 genes) in autism with FMR1-FM and 97
probes (80 genes) in autism with dup(15q) (Fig. 2C). Eighty-
three probes representing 68 genes were dysregulated in both
autism with FMR1-FM and with dup(15q) (Table 1). This
degree of overlap was highly significant (hypergeometric prob-
ability, P ¼ 1.2?102153). Fifty-two genes and 12 genes were
selectively dysregulated in either autism with FMR1-FM and
autism with dup(15q), respectively (Table 2), with fold
changes ranging between 0.54 and 1.98 fold.
Quantitative real-time PCR confirmed the differential
expression identified by the microarray analysis
To validate the differential expression identified by microarray
analysis using independent methods, we performed quantitat-
ive real-time PCR (qRTPCR) analysis of 19 genes chosen as
a cross-section using the same samples used in the microarray
analysis. The qRTPCR confirmed that 17 of the 19 genes were
differentially expressed as expected by the microarray analysis
(Fig. 3A–C). There was an overall highly significant corre-
lation between microarray and qRTPCR results (Pearson cor-
relation, r ¼ 0.57, P , 0.0001). Fold changes observed with
qRTPCR were typically higher that the smaller fold change
observed on the arrays.
CYFIP1 was one of the genes selectively induced in autism
with dup(15q). Because CYFIP1 is known to antagonize
FMRP (34), we reasoned that the induction of CYFIP1 in
dup(15q) might explain some of the significant overlap
between autismwith FMR1-FM
JAKMIP1, also known as MARLIN-1, was significantly
induced in autism with FMR1-FM and had a positive trend in
autism with dup(15q) (P ¼ 0.062), suggesting that JAKMIP1
could represent a commonly dysregulated pathway. In fact,
RankProd identified JAKMIP1 as a significantly up-regulated
gene in dup(15q) by microarray analysis (Supplementary
Material, Table S6). This gene is a particularly biologically
important candidate, given its putative role in GABABreceptor
expression (40) and microtubule networks (41).
and withdup(15q).
Figure 2. Differentially expressed probes identified by three different statisti-
cal methods, ANOVA, SAM and RankProd. Venn diagram showing the
number of probes identified as differentially expressed between (A) autism
with FMR1-FM (n ¼ 8) and control (n ¼ 15) and (B) autism with dup(15q)
(n ¼ 7) and control (n ¼ 15). (C) Overlap of the differentially expressed
probes (genes) in autism with FMR1-FM and dup(15q).
Figure 1. Hierarchical clustering and PCA differentiate individuals based on
their etiology. ANOVA identified 293 probes with significantly different
expression between autism with FMR1-FM (n ¼ 8), autism with dup(15q)
(n ¼ 7) and control (n ¼ 15). The probes were subjected to hierarchical clus-
tering and PCA. (A) Hierarchical clustering of the 30 individuals and genes.
Each row represents an individual and each column represents one of the
293 probes. A pseudo-colored representation of the relative intensity is
shown, such that a red color indicates high expression and green color low
expression, with the scale shown below. Relative distance of each probe (hori-
zontal axis) and individuals (vertical axis) are also demonstrated. (B) Enlarge-
ment of the hierarchical clustering dendrogram of the sample in (A). All eight
autism with FMR1-FM, seven autism with dup(15q) and 15 controls correctly
clustered within their etiological categories. The scale shows the Spearman
rank correlation coefficient used to construct the dendrogram. (C) PCA of
the expression profile of the 293 probes from 30 individuals. Shown here
are three principal components. Autism with FMR1-FM are depicted as
orange, autism with dup(15q) as magenta and control as cyan. The individuals
are clustered according to their genetic etiologies.
1684Human Molecular Genetics, 2007, Vol. 16, No. 14
at Universiti Teknologi Malaysia on May 18, 2011
hmg.oxfordjournals.org
Downloaded from

Page 4

Table 1. Genes dysregulated in both autism with FMR1-FM and dup(15q)
SymbolGene product Refseq ID
P-valuea
FC (FM/C)b
FC (dup/C)c
Gene lociAutism locid
Reference
G1P2
TNFRSF8
RIMS3
FAM46C
HIST2H3C
C1orf61
SLAMF1
SELL
DYRK3
RRM2
RSNL2
FAM82A
HNRPLL
SLC20A1
BBS5
GPR155
WNT6
CMKOR1
LZTFL1
GCET2
GHSR
AREG
TCF7
NRG2
NDFIP1
NR3C1
NID67
LCP2
TXNDC5
PHACTR1
PAQR8
CHST12
PSCD3
C1GALT1
RASA4
HIG2
CLDN23
TRAM1
DENND3
TRAF1
KIAA0649
VIM
PTPLA
RASSF4
ACADSB
APBB1
PRICKLE1
SAV1
ACTN1
JDP2
Interferon alpha-inducible protein (clone IFI-15K)
Tumor necrosis factor receptor superfamily member 8
Regulating synaptic membrane exocytosis 3
Family with sequence similarity 46 member C
Histone 2 H3c
Chromosome 1 open reading frame 61
Signaling lymphocytic activation molecule family member 1
Selectin L (lymphocyte adhesion molecule 1)
Dual-specificity tyrosine-phosphorylation regulated kinase 3
Ribonucleotide reductase M2 polypeptide
Restin-like 2
Family with sequence similarity 82 member A
Heterogeneous nuclear ribonucleoprotein L-like
Solute carrier family 20 (phosphate transporter) member 1
Bardet–Biedl syndrome 5
G protein-coupled receptor 155
Wingless-type MMTV integration site family member 6
Chemokine orphan receptor 1
Leucine zipper transcription factor-like 1
Germinal center expressed transcript 2
Growth hormone secretagogue receptor
Amphiregulin (schwannoma-derived growth factor)
Transcription factor 7 (T-cell specific HMG-box)
Neuregulin 2
Nedd4 family interacting protein 1
Nuclear receptor subfamily 3 group C member 1
Putative small membrane protein NID67
Lymphocyte cytosolic protein 2
Thioredoxin domain containing 5
Phosphatase and actin regulator 1
Progestin and adipoQ receptor family member VIII
Carbohydrate (chondroitin 4) sulfotransferase 12
Pleckstrin homology Sec7 and coiled-coil domains 3
Core 1 synthase galactosyltransferase 1
RAS p21 protein activator 4
Hypoxia-inducible protein 2
Claudin 23
Translocation associated membrane protein 1
KIAA0870 protein
TNF receptor-associated factor 1
KIAA0649
Vimentin
Protein tyrosine phosphatase-like member a
Ras association (RalGDS/AF-6) domain family 4
Acyl-Coenzyme A dehydrogenase short/branched chain
Amyloid beta precursor protein-binding family B member 1
Prickle-like 1 (Drosophila)
Salvador homolog 1 (Drosophila)
Actinin alpha 1
Jun dimerization protein 2
NM_005101
NM_001243
NM_014747
NM_017709
NM_021059
NM_006365
NM_003037
NM_000655
NM_003582
NM_001034
NM_024692
NM_144713
NM_138394
NM_005415
NM_152384
NM_001033045
NM_006522
NM_020311
NM_020347
NM_152785
NM_004122
NM_001657
NM_003202
NM_004883
NM_030571
NM_001018077
NM_032947
NM_005565
NM_030810
NM_030948
NM_133367
NM_018641
NM_004227
NM_020156
NM_006989
NM_013332
NM_194284
NM_014294
NM_014957
NM_005658
NM_014811
NM_003380
NM_014241
NM_032023
NM_001609
NM_001164
NM_153026
NM_021818
NM_001102
NM_130469
2.2E-06
1.5E-08
1.9E-06
5.3E-06
2.3E-06
7.5E-06
3.4E-06
1.5E-06
9.5E-06
4.4E-07
1.4E-07
4.1E-07
2.8E-06
1.0E-07
7.0E-06
1.0E-05
1.0E-05
6.5E-08
3.4E-06
5.8E-09
2.9E-06
9.0E-07
1.3E-06
1.3E-05
1.7E-08
2.8E-07
7.3E-07
3.4E-07
5.9E-06
1.5E-07
1.6E-07
1.8E-07
7.5E-06
3.9E-06
5.0E-06
2.9E-08
2.3E-07
7.1E-06
1.1E-05
1.4E-05
4.8E-07
3.0E-06
1.1E-06
1.6E-06
3.0E-08
1.2E-06
3.3E-07
1.6E-07
4.3E-07
6.5E-06
0.79
1.23
0.66
0.76
0.77
1.13
1.19
0.70
1.21
0.82
1.23
1.47
1.35
1.13
1.11
0.82
0.86
1.40
1.20
1.50
1.29
0.77
1.18
1.36
0.86
1.13
1.31
1.16
0.82
1.26
1.15
0.81
1.13
0.88
1.17
1.24
0.79
0.88
1.16
1.17
0.66
0.67
0.72
1.25
1.20
1.26
0.82
1.15
0.70
1.19
0.85
1.27
0.75
0.82
0.84
1.20
1.17
0.79
1.18
0.81
1.13
1.35
1.37
1.14
1.16
0.83
0.87
1.57
1.23
1.52
1.26
0.81
1.44
1.57
0.87
1.14
1.21
1.16
0.81
1.23
1.13
0.87
1.27
0.86
1.16
1.12
0.82
0.87
1.22
1.21
0.82
0.75
0.74
1.39
1.13
1.23
0.81
1.15
0.76
1.20
1p36.33
1p36.22
1p22.2
1p12
1q21.3
1q22
1q23.3
1q24.2
1q32.1
2p25.1
2p23.2
2p22.2
2p22.1
2q13
2q31.1
2q31.1
2q35
2q37.3
3p21.3
3q13.2
3q26.31
4q13.3
5q31.1
5q31.2
5q31.3
5q31.3
5q33.1
5q35.1
6p24.3
6p24.1
6p12.1
7p22
7p22.1
7p14–p13
7q22.1
7q32.2
8p23.1
8q13.3
8q24.3
9q33–q34
9q34.3
10p13
10p14–p13
10q11.21
10q25–q26
11p15
12q12
14q13–q23
14q24.1–q24.2
14q24.3
1p36.23–1q36.1312
1p13–1q12 17
2q31.1–2q33.3 6,8
5q14.3–5q31.2 12
7q22.1–7q31.317
Continued
Human Molecular Genetics, 2007, Vol. 16, No. 14
1685
at Universiti Teknologi Malaysia on May 18, 2011hmg.oxfordjournals.orgDownloaded from

Page 5

Functional annotation revealed pathway dysregulation
In an attempt to uncover common fractions among the
differentially expressed genes, we classified genes into gene
ontology groups using DAVID (42). Table 3 shows the top
three clusters identified by DAVID using the 68 genes dysre-
gulated in autism with FMR1-FM and dup(15q) or the 52
genes selectively dysregulated in autism with FMR1-FM.
The number of genes selectively dysregulated in autism with
dup(15q) was too small to analyze using functional annotation
clustering.
Genes related to cell communication (P ¼ 7.6?1026) and
signal transduction (P ¼ 2.2 ? 1025) were most significantly
enriched in the 68 genes commonly dysregulated in autism
withFMR1-FManddup(15q).Genesrelatedtoimmuneresponse
(P ¼ 3.7 ? 1023) and defense response (P ¼ 7.3 ? 1023) were
also enriched in this gene set. Genes related to chaperone
(P ¼ 2.6 ? 1022) and protein folding (P ¼ 3.2 ? 1022) were
enriched in the 52 genes selectively dysregulated in autism
with FMR1-FM.Genes
(P ¼ 1.2 ? 1022) and mRNA metabolism (P ¼ 2.1 ? 1022)
were also enriched in this gene set, consistent with the FMRP
protein’s function as an RNA binding protein important in
regulatory translation (43). Chaperones and folding proteins
are commonly found to operate co-translationally, providing
a potential link with FMRPfunction.
To provide a more refined functional classification of genes,
we used Ingenuity Pathway Analysis (IPA) (44), a powerful
tool for investigating the biological pathways represented by
the genes commonly dysregulated in autism with FMR1-FM
and dup(15q). IPA uses known protein–protein and gene–
gene interactions that have been culled into a curated database
and associates the list of differentially expressed genes with
biological networks. IPA identified three statistically signifi-
cant networks, each containing at least 10 genes (Table 4,
Supplementary Material, Fig. S1). Principal functions associ-
ated with these networks were cell cycle (P ¼ 5.2 ? 1028),
cellular movement (P ¼ 1.3 ? 1028) and cell-to-cell signaling
and interaction (P ¼ 4.3 ? 1028). The ‘cell-to-cell signaling
and interaction’ was consistent with ‘cell communication’ and
‘signal transduction’ categories identified by DAVID. The
identification of the ‘molecular transport’ pathway containing
JAKMIP1 was particularly salient, given this gene’s known
role in GABAR trafficking within neurons (41). There were
also other important genes in this pathway, including PSCD3,
an ADP-ribosylation factor of unknown CNS function, and
ACTN1, a cytoskeletal anchoring protein. JAKMIP1 may act
alongwiththesegenesinthesegregationofsignalingcomplexes
involved in neural transmission.
relatedto RNA binding
Effect of downregulation of FMR1 and up-regulation of
CYFIP1 in a neuronal cell on the expression of the
dysregulated genes identified in lymphoblastoid cells
Although we identified dysregulated genes in autism with
FMR1-FM and dup(15q) using lymphoblastoid cells, we
were interested in whether the expression of these genes
would also be dependent on FMR1 and CYFIP1 in neuronal
cells. To examine the effect of FMR1 and CYFIP1 in neuronal
cells, we used the well-characterized human neuronal cell line
SH-SY5Y (45). FMR1 and CYFIP1 dependence in SH-SY5Y
Table 1. Continued
Symbol
Gene product
Refseq ID
P-valuea
FC (FM/C)b
FC (dup/C)c
Gene loci
Autism locid
Reference
BATF
Basic leucine zipper transcription factor ATF-like
NM_006399
8.1E-06
1.11
1.17
14q24.3
KIAA1370
Hypothetical protein FLJ10980
NM_019600
2.0E-08
0.80
0.86
15q21.2–q21.3
SV2B
Synaptic vesicle glycoprotein 2B
NM_014848
2.3E-07
1.51
1.42
15q26.1
CIITA
MHC class II transactivator
NM_000246
7.0E-07
1.21
1.19
16p13
16p13.2–16p13.13
5
SNN
Stannin
NM_003498
4.9E-07
1.19
1.21
16p13
16p13
17
KIAA0251
KIAA0251
NM_015027
4.1E-06
0.78
0.84
16p13.11
PRKCB1
Protein kinase C beta 1
NM_212535
6.3E-06
1.21
1.21
16p11.2
16p11.2
15
IL4R
Interleukin 4 receptor
NM_000418
1.8E-07
1.13
1.15
16p12.1
IL21R
Interleukin 21 receptor
NM_021798
1.5E-07
1.14
1.13
16p11
CCL17
Chemokine (C-C motif) ligand 17
NM_002987
3.6E-06
1.38
1.58
16q13
CCL22
Chemokine (C-C motif) ligand 22
NM_002990
1.5E-05
1.41
1.42
16q13
FLJ35773
Hypothetical protein FLJ35773
NM_152599
1.9E-06
1.23
1.16
17p13.1
RAB11FIP4
RAB11 family interacting protein 4 (class II)
NM_032932
1.3E-05
1.11
1.17
17q11.2
17q11.2–17q12
11,12
RASL10B
RAS-like family 10 member B
NM_033315
1.2E-08
0.88
0.84
17q12
17q11.2–17q12
12
ARHGAP23
Rho GTPase activating protein 23
XM_290799
4.2E-07
0.71
0.74
17q12
17q11.2–17q12
12
PITPNC1
Phosphatidylinositol transfer protein cytoplasmic 1
NM_012417
4.7E-06
1.23
1.41
17q24.2
SLC16A6
Solute carrier family 16 member 6
NM_004694
3.7E-08
0.84
0.85
17q24.2
MIA
Melanoma inhibitory activity
NM_006533
2.3E-06
1.36
1.29
19q13.2
19q13.11–19q13.33
5
aP-value was calculated by one-way ANOVA using controls (N ¼ 15), autism with FMR1-FM (N ¼ 8) and autism with dup(15q) (N ¼ 7).
bFC was calculated between mean values of controls (N ¼ 15) and autism with FMR1-FM (N ¼ 8).
cFC was calculated between mean values of controls (N ¼ 15) and autism with dup(15q) (N ¼ 7).
dAutism loci identified by other genetic studies were shown with references.
1686Human Molecular Genetics, 2007, Vol. 16, No. 14
at Universiti Teknologi Malaysia on May 18, 2011
hmg.oxfordjournals.org
Downloaded from

Page 6

Table 2. Genes selectively dysregulated in autism with FMR1-FM or dup(15q)
SymbolGene productRefseq ID
P-valuea
FC (FM/C)b
Gene loci
Genes selectively dysregulated in autism with FMR1-FM
EDG1
Endothelial differentiation sphingolipid
G-protein-coupled receptor 1
Leucine rich repeat containing 8 family member D
Adenylate kinase 2
tRNA isopentenyltransferase 1
Presenilin 2 (Alzheimer disease 4)
Chromosome 1 open reading frame 153
Protein kinase interferon-inducible dsRNA
dependent activator
Kruppel-like factor 7 (ubiquitous)
Adaptor protein containing pH PTB
domain and leucine zipper motif 1
Kalirin RhoGEF kinase
Zinc finger CCHC domain containing 4
Janus kinase and microtubule interacting protein 1
2-aminoadipic 6-semialdehyde dehydrogenase
Scavenger receptor class B member 2
Alpha-kinase 1
Activated RNA polymerase II transcription
cofactor 4
Mannosidase alpha class 2A member 1
Heterogeneous nuclear ribonucleoprotein A0
DnaJ (Hsp40) homolog subfamily C member 18
Histone 1 H3d
Histone 1 H3 h
Cell division cycle 40 homolog (yeast)
SEC63-like (S. cerevisiae)
RAB guanine nucleotide exchange factor (GEF) 1
Stromal antigen 3
Steroidogenic acute regulator
Tripartite motif-containing 14
KIAA0368
Selenophosphate synthetase 1
Annexin A11
Regulator of G-protein signaling 10
Chromosome 10 open reading frame 137
Hypothetical protein FLJ39441
Baculoviral IAP repeat-containing 3
Heat shock protein 90 kDa beta (Grp94) member 1
Reed–Steinberg cell-expressed intermediate
filament-associated protein
Hypothetical protein MGC7036
Guanine nucleotide binding protein (G
protein) gamma 2
Sel-1 suppressor of lin-12-like (C. elegans)
Hypothetical protein FLJ10980
Membrane-bound transcription factor protease site 1
Myosin XVIIIA
Zinc finger protein 36 C3H
type homolog (mouse)
Ferritin light polypeptide
Zinc finger protein 160
hsp70-interacting protein
Cystatin C (amyloid angiopathy and
cerebral hemorrhage)
Cystatin D
Chromosome 20 open reading frame 35
Hypothetical protein DKFZp434O0213
Oncostatin M
Fragile X mental retardation 1
NM_0014007.7E-060.841p21
LRRC8D
AK2
TRIT1
PSEN2
FAM89A
PRKRA
NM_018103
NM_001625
NM_017646
NM_000447
NM_198552
NM_003690
3.7E-06
5.4E-06
3.6E-07
1.2E-06
4.1E-06
1.1E-06
0.88
1.11
1.15
0.86
1.23
1.14
1p22.2
1p34
1p35.3–p34.1
1q31–q42
1q42.2
2q31.2
KLF7
APPL
NM_003709
NM_012096
5.9E-06
1.7E-06
1.26
1.21
2q32
3p21.1–p14.3
KALRN
ZCCHC4
JAKMIP1
AASDH
SCARB2
ALPK1
SUB1
NM_001024660
XM_376310
NM_144720
NM_181806
NM_005506
NM_025144
NM_006713
7.8E-06
1.4E-05
5.5E-06
8.2E-06
5.7E-06
4.8E-06
3.8E-07
1.40
1.13
1.98
1.15
0.78
1.22
0.88
3q21.1–q21.2
4p15.2
4p16.1
4q12
4q21.1
4q25
5p13.3
MAN2A1
HNRPA0
DNAJC18
HIST1H3D
HIST1H3H
CDC40
SEC63
RABGEF1
STAG3
STAR
TRIM14
KIAA0368
SEPHS1
ANXA11
RGS10
C10orf137
SPTY2D1
BIRC3
HSP90B1
RSN
NM_002372
NM_006805
NM_152686
NM_003530
NM_003536
NM_015891
NM_007214
NM_014504
NM_012447
NM_000349
NM_014788
XM_001129450
NM_012247
NM_001157
NM_001005339
NM_015608
NM_194285
NM_001165
NM_003299
NM_002956
2.1E-06
4.5E-06
3.7E-06
1.5E-06
6.5E-06
1.3E-06
6.0E-06
7.3E-06
7.0E-06
4.2E-06
5.1E-06
8.4E-06
5.7E-06
9.6E-06
9.0E-06
2.3E-06
2.7E-06
1.1E-05
2.7E-06
8.3E-06
0.74
1.12
1.20
0.78
0.83
1.12
1.11
1.12
1.39
1.32
1.15
1.11
0.84
0.87
1.12
0.88
0.87
1.19
0.88
1.18
5q21–q22
5q31
5q31.2
6p21.3
6p22–p21.3
6q21
6q21
7q11.21
7q22.1
8p11.2
9q22.33
9q31.3
10p14
10q23
10q25
10q26.13–q26.2
11p15.1
11q22
12q24.2–q24.3
12q24.3
MGC7036
GNG2
NM_145058
NM_053064
6.0E-06
1.2E-05
0.88
1.15
12q24.31
14q21
SEL1L
KIAA1370
MBTPS1
MYO18A
ZFP36
NM_005065
NM_019600
NM_003791
NM_078471
NM_003407
1.8E-07
4.8E-07
1.7E-09
8.3E-06
7.0E-06
0.88
0.85
0.88
0.86
0.87
14q24.3–q31
15q21.2–q21.3
16q24
17q11.2
19q13.1
FTL
ZNF160
HSPBP1
CST3
NM_000146
NM_033288
NM_012267
NM_000099
1.0E-07
6.9E-06
3.2E-06
1.1E-08
0.88
1.13
0.87
0.78
19q13.3–q13.4
19q13.41
19q13.42
20p11.21
CST5
C20orf35
LRP5L
OSM
FMR1
NM_001900
NM_018478
NM_182492
NM_020530
NM_002024
3.7E-07
6.2E-06
5.2E-06
1.2E-05
4.1E-15
0.83
0.87
1.11
0.86
0.54
20p11.21
20q13.12
22q11.23
22q12.2
Xq27.3
Genes selectively dysregulated in autism with dup(15q)
FC(dup/C)c
0.87
1.16
0.87
BRP44
CIB4
PTMA
Brain protein 44
Calcium and integrin binding family member 4
Prothymosin alpha
NM_015415
NM_001029881
NM_002823
1.6E-07
4.0E-06
8.4E-07
1q24
2p23.3
2q35–q36
Continued
Human Molecular Genetics, 2007, Vol. 16, No. 141687
at Universiti Teknologi Malaysia on May 18, 2011
hmg.oxfordjournals.org
Downloaded from

Page 7

cells was assessed using a short hairpin RNA (shRNA) to
reduce the expression of FMR1 and a plasmid expression
vector to induce the expression of CYFIP1, respectively. As
shown in Figure 4A, the expression of FMR1 was reduced
to ?60% of its normal level in SH-SY5Y cells stably expres-
sing FMR1 shRNAs, whereas the expression of CYFIP1 was
significantly induced (11-fold) in SH-SY5Y cells stably trans-
fected with the CYFIP1 plasmid.
We were able to further demonstrate the effect of downregu-
lation of FMR1 and up-regulation of CYFIP1 on the expression
of two key downstream genes (Fig. 4B). In SH-SY5Y cells
transfected with FMR1 shRNA, the expression of JAKMIP1
and GPR155 was significantly reduced and induced, respect-
ively. In SH-SY5Y cells over-expressing CYFIP1, the
expression of JAKMIP1 and GPR155 was also reduced and
induced, respectively. These findings demonstrated that the
expression of JAKMIP1 and GPR155 was also dependent on
FMR1 and CYFIP1 in SH-SY5Y cells and that reduction of
FMR1 and induction of CYFIP1 can share common down-
stream effects on the expression of JAKMIP1 and GPR155.
The expression of JAKMIP1 protein was dependent on
FMR1 and CYFIP1
We next validated the effect of FMR1 or CYFIP1 on the protein
expression of JAKMIP1 in the central nervous system (CNS).
We examined the expression of the JAKMIP1 protein in the
cortex of Fmr1 knock-out (KO) and wild-type (WT) mice and
SH-SY5Y cells transfected with the CYFIP1 over-expression
plasmid. The expression of Jakmip1 was reduced in the cortex
ofFmr1KOmice(Fig.5A)andSH-SY5Ycellsover-expressing
CYFIP1 (Fig. 5B). These results confirmed the in vitro findings
thattheexpressionofJakmip1wasdependentonFmr1inmouse
brain, suggesting that at least some of the changes observed in
lymphoblastoid cells reflect similar changes in the CNS.
The expression of JAKMIP1 and GPR155 was significantly
different between 27 male sib pairs discordant for
idiopathic ASD
To determine the potential generalizability of these findings to
idiopathic autism, we examined whether the expression of
JAKMIP1 and GPR155 was also dysregulated in lymphoblas-
toid cells from idiopathic ASD cases. We selected 27 male sib
pairs discordant for ASD from AGRE (Supplementary
Material, Table S1). The 27 males with ASD did not have
FMR1-FM or dup(15q) and had surrogate IQ markers
(Raven’s progressive matrices) .70. As shown in Figure 6,
the expression of JAKMIP1 and GPR155 was significantly
dysregulated in the 27 males with ASD when compared with
their sibs without ASD. These results show that the dysregula-
tion of JAKMIP1 and GPR155 is associated with ASD. The
lack of general intellectual disability in this ASD group also
shows that these dysregulations are not simply due to a non-
specific cognitiveimpairment
observed in FXS and dup(15q). However, in both in vitro
(SH-SY5Y cells) and in vivo (brain) CNS tissues, the directions
of JAKMIP1 and GPR155 regulation were opposite to that
observed in lymphoblastoid cells. The differences may reflect
many facts, including immortalization or alternative regulatory
signaling pathways in different tissues. However, these data are
consistent between FMR1-FM and dup(15q) and indicate that
expression of JAKMIP1 and GPR155 is regulated by both
FMR1 and CYFIP1 levels, although differently between
neural tissues and lymphoblastoid cells, providing potential
common signaling pathways dysregulated in ASD (Fig. 7).
or intellectualdisability
DISCUSSION
Autism is a heterogeneous condition and likely results from
the combined effects of multiple genetic changes including
copy number variations and single nucleotide polymorphisms,
interacting with environmental factors (1–4). Classification of
autistic patients on the basis of genotypic and phenotypic
information is one effective way to identify more homo-
geneous subgroups and hasten the identification of genes
underlying autism (1–4). Approximately 3% of autistic chil-
dren have either FMR1-FM or dup(15q); these patients com-
prise homogeneous populations for investigation.
In this study, we performed global mRNA expression profil-
ing inmales with autism carryingeither FMR1-FM ordup(15q)
and in control males. We found that these autistic individuals
can be differentiated based on their genetic etiologies based
on lymphoblast gene expression profiles. Interestingly, this
analysis also revealed a common gene expression signature
Table 2. Continued
SymbolGene productRefseq ID
P-valuea
FC (FM/C)b
Gene loci
FYTTD1
PSIP1
RCL1
PPM1A
CYFIP1
NIPA2
GOLGA8E
UBE3A
ICAM1
Forty-two-three domain containing 1
PC4 and SFRS1 interacting protein 1
RNA terminal phosphate cyclase-like 1
Protein phosphatase 1A magnesium-dependent alpha isoform
cytoplasmic FMR1 interacting protein 1
Non-imprinted in Prader-Willi/Angelman syndrome 2
Golgi autoantigen golgin family member
Ubiquitin protein ligase E3A
Intercellular adhesion molecule 1 human
rhinovirus receptor
NM_001011537
NM_021144
NM_005772
NM_021003
NM_014608
NM_001008860
NM_001012423
NM_130838
NM_000201
4.8E-06
8.8E-09
1.5E-07
5.8E-06
1.1E-09
9.7E-08
1.9E-06
1.6E-09
3.1E-06
1.14
1.18
1.12
1.12
1.23
1.14
1.17
1.18
1.13
3q29
9p22.3
9p24.1–p23
14q23.1
15q11.2
15q11.2
15q11.2
15q11.2–15q12
19p13.3–p13.2
aP-value was calculated by one-way ANOVA using controls (N ¼ 15), autism with FMR1-FM (N ¼ 8) and autism with dup(15q) (N ¼ 7).
bFC was calculated between mean values of controls (N ¼ 15) and autism with FMR1-FM (N ¼ 8).
cFC was calculated between mean values of controls (N ¼ 15) and autism with dup(15q) (N ¼ 7).
1688Human Molecular Genetics, 2007, Vol. 16, No. 14
at Universiti Teknologi Malaysia on May 18, 2011
hmg.oxfordjournals.org
Downloaded from

Page 8

across these two distinct genetic conditions that was signifi-
cantly different from control profiles. We used the intersection
of three different statistical tests to identify the most robustly
differentially expressed genes (35,38,39), and the qRTPCR
data confirmed this gene selection strategy.
Gene expression profiles of lymphoblastoid cells
carrying the FMR1-FM
We identified 120 genes differentially expressed in FMR1-FM
carriers compared with controls. Among these genes, NR3C1
Figure 3. Confirmation of the differential gene expression by qRTPCR. Total RNA was extracted from lymphoblastoid cells with FMR1-FM, dup(15q) or
control and qRTPCR was performed to confirm the differential expression identified by microarray analysis. (A) Genes specifically dysregulated in autism
with FMR1-FM or dup(15q). (B) Genes up-regulated in both autism with FMR1-FM and dup(15q). (C) Genes downregulated in both autism with
FMR1-FM and dup(15q). Results represent means+SD of each group. The mean of the value of control subjects was set as 1. P-value was calculated by
Mann–Whitney U test using control (N ¼ 15) versus autism with FMR1-FM (N ¼ 8) or autism with dup(15q) (N ¼ 7).?P , 0.05,??P , 0.01,???P , 0.001.
Human Molecular Genetics, 2007, Vol. 16, No. 141689
at Universiti Teknologi Malaysia on May 18, 2011
hmg.oxfordjournals.org
Downloaded from

Page 9

and VIM were previously identified as target RNAs of FMRP
(46), although the mRNA expression changes of these genes in
FMR1-FM have not been reported. Brown et al. (32) pre-
viously identified 144 genes as differentially expressed in lym-
phoblastswith FMR1-FM
lymphoblastoid cells and pooled normal lymphoblastoid
cells. Because there was no overlap except for FMR1
using pooledfragileX
between these 144 genes and the 120 genes identified here
with our most stringent analyses using ANOVA, SAM and
RankProd, we used the larger gene list identified by either
SAM and/or RankProd to compare with the 144 genes ident-
ified by Brown et al. We found that 13 genes were shared
in these gene lists, including iduronate 2-sulfatase (IDS),
hairy and enhancer of split 1 (HES1) and immunoglobulin
Table 3. Functional annotation clustering using the 68 genes dysregulated in both autism with FMR1-FM and dup(15q) or the 52 genes selectively dysregulated in
autism with FMR1-FM
CategoryGene ontology
P-valuea
Top 3 clusters enriched in the 68 genes [autism with FMR1-FM and dup(15q)]Hits in the 68 genes
Cluster 1
GO_BP
GO_BP
GO_BP
Cluster 2
GO_BP
GO_BP
GO_BP
GO_BP
GO_BP
GO_BP
GO_BP
GO_BP
Cluster 3
SP_PIR
SP_PIR
GO_MF
UP_SEQ
GO_CC
GO_CC
KEGG_PATHWAY
GO_CC
SP_PIR
SP_PIR
SP_PIR
UP_SEQ
UP_SEQ
UP_SEQ
GO_CC
Enrichment scoreb: 3.53
Cell communication
Signal transduction
Cellular process
Enrichment scoreb: 1.47
Immune response
Defense response
Response to biotic stimulus
Response to pest, pathogen or parasite
Response to other organism
Response to stress
Response to stimulus
Organismal physiological process
Enrichment scoreb: 1.33
Membrane
Transmembrane
Signal transducer activity
Transmembrane region
Integral to membrane
Intrinsic to membrane
Cytokine–cytokine receptor interaction
Membrane
Signal
Transmembrane protein
Glycoprotein
Disulfide bond
Signal peptide
Glycosylation site:N-linked (GlcNAc)
Cell
27
25
47
7.6E-06
2.2E-05
1.5E-04
11
11
11
6
6
8
13
11
3.7E-03
7.3E-03
9.8E-03
3.8E-02
4.8E-02
6.9E-02
1.6E-01
2.9E-01
21
20
19
20
21
21
5
25
11
5
13
13
11
13
41
4.1E-04
9.1E-04
5.4E-03
1.9E-02
4.7E-02
4.8E-02
5.0E-02
7.1E-02
8.8E-02
1.1E-01
1.2E-01
1.4E-01
4.8E-01
5.0E-01
7.5E-01
Top 3 clusters enriched in the 52 genes (autism with FMR1-FM) Hits in the 52 genes
Cluster 1
SP_PIR
GO_BP
GO_MF
Cluster 2
GO_MF
GO_BP
GO_BP
GO_BP
GO_BP
GO_CC
Cluster 3
GO_CC
GO_CC
GO_CC
GO_CC
SP_PIR
GO_MF
Enrichment scoreb: 1.4
Chaperone
Protein folding
Unfolded protein binding
Enrichment scoreb: 1.36
RNA binding
mRNA metabolism
RNA metabolism
RNA processing
mRNA processing
Ribonucleoprotein complex
Enrichment scoreb: 1.31
Membrane-enclosed lumen
Organelle lumen
Nucleoplasm
Nuclear lumen
Calcium
Calcium ion binding
3
4
3
2.6E-02
3.2E-02
7.7E-02
6
4
5
4
3
4
1.2E-02
2.1E-02
2.7E-02
6.3E-02
9.5E-02
1.7E-02
6
6
4
4
3
3
6.0E-03
6.0E-03
2.6E-02
6.2E-02
3.6E-02
6.7E-01
GO, gene ontology; BP, biological process; MF; molecular function; CC, cellular component; SP, SWISS-PROT; PIR, Protein Information Resources.
KEGG, Kyoto Encyclopedia of Genes and Genomes.
aP-value was calculated by Fisher Exact test.
bEnrichment score was the negative log of geometric mean of each member’s P-values in the cluster.
1690 Human Molecular Genetics, 2007, Vol. 16, No. 14
at Universiti Teknologi Malaysia on May 18, 2011
hmg.oxfordjournals.org
Downloaded from

Page 10

superfamily, member 3 (IGSF3) as up-regulated genes and
CDK2-associated protein 2 (CDK2AP2), ubiquitin specific
peptidase 8 (USP8), MAX-like protein X (MLX), ribosomal
protein S5 (RPS5), C-terminal binding protein 1 (CTBP1),
spleen tyrosine kinase (SYK), F-box protein 6 (FBXO6),
mitogen-activatedprotein
(MAP3K11), sorting nexin 15 (SNX15) and CD44 antigen
(CD44) as downregulated genes. Although these genes have
kinasekinase kinase 11
Table 4. Gene networks identified by IPA using the genes dysregulated in both autism with FMR1-FM and dup(15q)
Genes in networka
Top functions (P-valueb)
ACTB, APBB1, CTNNB1, CYP27A1, DBT, GAD1, GHSR, HK2, IL8RA
ISG15, JDP2, LAD1, LEP, MT1L, NDFIP1, NEDD4, NR3C1, NRG2
PHACTR1, PHKG1, PLEC1, PODXL, PTENRB1, RRM2, RRM2B,
SLC20A1, SNN, TCF7, TNF, TNFRSF8, TOB2, TP53, TRAF1, TRAM1
Cell cycle (P ¼ 5.2E-8)
Cancer (P ¼ 2.3E-7)
Gastrointestinal disease (P ¼ 2.3E-7)
ACTN1, ADORA2A, ADRBK2, CCBP2, CCL17, CCL22, CCR4, CIITA,
CLEC11A, CMKOR1, DOK2, EGF, GCET2, IER3, IGFALS, IL13, IL13RA1
IL13RA2, IL1B, IL21R, IL4R, JAK1, JAKMIP1, MIA, NR4A3, PIP3-E,
PLA2G2A,PLA2G4A,PLB1,PSCD3,SLAMF1,SLC16A6,SPHK1,TYK2,ZYX
Cellular movement (P ¼ 1.3E-8)
Lipid metabolism (P ¼ 1.6E-8)
Molecular transport (P ¼ 1.6E-8)
ANXA6, AREG, BATF, CD53, CEACAM1, D830050J10RIK, DYRK3, EPO,
HIF1A,HIG2,HIST2H3C,HRAS,IL2,IL6,LCP2,LDB3,MAZ,MSN,MYOD1
PODXL, PRKCB1, RASA4, RASGRP3, RASSF4, RRAS, SCGB1A1, SELL,
SELPLG, SNAP23, SV2B, TXK, TYK2, UPP1, VIM, WNT6
Cell-to-cell signaling and interaction (P ¼ 4.3E-8)
Cell cycle (P ¼ 5.7E-7)
Cancer (P ¼ 1.2E-6)
aGenes identified as differentially expressed by microarray analysis is shown in bold.
bP-value was calculated using the right-tailed Fisher’s Exact Test.
Figure 4. JAKMIP1 and GPR155 are dysregulated by reduction of FMR1 and induction of CYFIP1. SH-SY5Y cells were stably transfected with (i) vector expres-
sing shRNA for control, (ii) vector expressing shRNA for FMR1, (iii) empty expression vector or (iv) expression vector for CYFIP1. Total RNA was extracted
from each and qRTPCR was performed to validate the effect of FMR1 and CYFIP1 on the expression of JAKMIP1 and GPR155. (A) The expression of FMR1 is
significantly reduced in SH-SY5Y cells expressing shRNA for FMR1, whereas the expression of CYFIP1 was significantly induced in SY5Y cells over-expressing
CYFIP1. (B) The expression of JAKMIP1 is significantly reduced in SH-SY5Y cells expressing shRNA for FMR1 and over-expressing CYFIP1. The expression of
GPR155 is significantly induced in SH-SY5Y cells expressing shRNA for FMR1 and over-expressing CYFIP1. Results represent means+SD of each group. The
mean of the value of each control was set as 1. Significance was calculated by the Mann–Whitney U test using SH-SY5Y cells expressing shRNA for control
(N ¼ 4) versus shRNA for FMR1 (N ¼ 4) or empty expression vector (N ¼ 8) versus expression vector for CYFIP1 (N ¼ 7).?P , 0.05,??P , 0.01.
Human Molecular Genetics, 2007, Vol. 16, No. 141691
at Universiti Teknologi Malaysia on May 18, 2011
hmg.oxfordjournals.org
Downloaded from

Page 11

not been reported as associated with FMR1 or autism, HES1
was associated with attention-deficit hyperactive disorder
(47), which is a symptom frequently seen in FXS (48) and
overlapping with ASD (49). The relatively low overlap
between the two gene lists could be due to the difference of
clinical features of individuals (autism versus not specific
for autism), experimental design (each individual versus
pooled), microarray platforms (Agilent versus Affymetrix)
and the statistical analysis used to find the differential
expression between groups. The initial study (32) whose
primary aim was to find FMRP ligand mRNPs was relatively
underpowered to detect overall differences in gene expression
and our study used very conservative statistical criteria.
However, this core set of genes provides an interesting gene
list for further investigation.
Gene expression profiles of lymphoblastoid cells
with dup(15q)
We identified 80 genes differentially expressed in dup(15q)
carriers compared with controls. Among these genes, four
genes located in 15q11–q13 (the region of duplication)
UBE3A, CYFIP1, NIPA2 and GOLGA8F were all induced.
It is important to note that five other genes located in the
duplicatedregion,tubulin
protein 5 (TUBGCP5), HERC2, HERC2 pseudogene 2
(HERC2P2), NIPA1 and ATP10A were also identified as
up-regulated genes by at least one of the three different stat-
istical analyses (Supplementary Material, Table S7). Five
other genes in the duplicated region, gamma-aminobutyric
acid A receptor (GABR) beta 3 (GABRB3), GABR alpha 5
(GABRA5), GABR gamma 3 (GABRG3), oculocutaneous
albinism II (OCA2) and necdin homolog (NDN), were not
expressed at detectable levels in the lymphoblastoid cells.
It is important to emphasize that the 15q11–q13 region is
subject to paternal imprinting. Three paternally imprinted
genes, makorin ring finger protein 3 (MKRN3), MAGE-like
2(MAGEL2) andSNRPN
(SNURF)-small nuclear ribonucleoprotein polypeptide N
(SNRPN) were expressed in the lymphoblasts, but showed
no significant changes relative to controls. These data are
gammacomplex associated
upstream readingframe
consistent with the fact that the duplicated region was mater-
nally derived in all seven cases analyzed in this study. So,
overall, these findings suggest that the genes located in
the duplicated region were globally up-regulated except for
the paternally imprinted genes. Global up-regulation due to
gene-dosage has also been reported in Down syndrome
(50,51).
Baron et al. (27) identified 81 known genes as differentially
expressed in lymphoblastoid cells with dup(15q) (seven indi-
viduals) when compared with controls (eight individuals)
using the Affymetrix platform. They identified up-regulation
of UBE3A, NIPA1, NIPA2 and HERC2, findings consistent
with our results. We used the gene list identified by SAM
and/or RankProd to compare with the 81 genes identified by
Baron et al. and identified 11 other genes shared in the two
gene lists, a significant overlap (the hypergeometric prob-
ability is 0.001). These genes were abhydrolase domain con-
taining 6 (ABHD6), potassium
member 1 (KCNK1), hypothetical protein KIAA1147 and
zinc finger, DHHC domain containing 14 (ZDHCC14) as
up-regulated andRho GTPase
(ARHGAP25), clone LOC387882,
12-hydroxydehydrogenase (LTB4DH), clone MGC27165,
PFTAIRE protein kinase 1 (PFTK1), zinc finger protein 43
(ZNF43) and ring finger protein 41 (RNF41) as downregu-
lated. The relationships between these genes and autism
remain unknown.Again,
FMR1-FM, these genes represent a set of independently repli-
cated genes between two studies.
channel, subfamily K,
activatingprotein 25
B4 leukotriene
similarto thestudiesof
Significant overlap of dysregulated genes in autism with
FMR1-FM and dup(15q)
We identified 68 genes that were dysregulated in both autism
with FMR1-FM and with dup(15q), a very significant result
(the hypergeometricprobability
1.2?102153). However, we cannot formally exclude the possi-
bility that some of the 68 common dysregulated genes might
be related to common pathways between FMR-FM and
dup(15q) unrelated to ASD. Microarray analysis using lym-
phoblastoid cells with FMR1-FM or dup(15q), but without
ASD is needed to exclude the possibility, as was done in tuber-
ous sclerosis cases with and without autism (51).
We found that the expression of CYFIP1 was significantly
induced in autism with dup(15q). CYFIP1 has been shown
to antagonize FMRP in the eye and nervous system of Droso-
phila (34). In FXS, the absence of FMRP, a binding partner to
CYFIP1, results in excess free CYFIP1. Similarly, excess free
CYFIP1 may be the outcome of dup(15q). Thus, antagoniza-
tion of FMRP by over-expression of CYFIP1 and/or alternate
actions of excess CYFIP1 may be common mechanistic links
between FMR1-FM and dup(15q).
ofthisoverlapis
Effect of FMR1 and CYFIP1 on the commonly
dysregulated genes in SH-SY5Y and mouse brain
We validated the effect of downregulation of FMR1 in
SH-SY5Y cells and mouse brain and up-regulation of
CYFIP1 in SH-SY5Y cells on the expression of the commonly
dysregulated genes identified in patient lymphoblastoid cells.
Figure 5. The expression of JAKMIP1 was dependent on FMR1 and CYFIP1
in mouse cortex and SH-SY5Y cells. Proteins were extracted from cortices of
Fmr1 WT or KO mice (A), or SH-SY5Y cells transfected with empty vector or
CYFIP1 cDNA (B). Western blotting was performed to validate the effect of
the reduction of FMR1 or induction of CYFIP1 on the expression of JAKMIP1
protein. The protein expression of Jakmip1 was reduced in cortex of Fmr1-KO
mice (A) as well as SH-SY5Y cells transfected with shRNA for FMR1 (data
not shown) and SH-SY5Y over-expressing CYFIP1 (B). Data shown in (A)
and (B) were the representative of two independent experiments.
1692 Human Molecular Genetics, 2007, Vol. 16, No. 14
at Universiti Teknologi Malaysia on May 18, 2011
hmg.oxfordjournals.org
Downloaded from

Page 12

We demonstrated that the expression of JAKIMIP1 and
GPR155 was dysregulated by reduction of FMR1 and induc-
tion of CYFIP1 in SH-SY5Y cells. The Jakmip1 protein was
also dysregulated by knock-out of Fmr1 in mouse brain. Inter-
estingly, the direction of changes observed in both of these
genes was opposite in neural tissues (SH-SY5Y cells and
brain) and lymphoblastoid cells. Such differences between
brain and blood cells have been previously observed in other
signaling pathways (31,52). It is likely that it is not the
precise direction observed in lymphoblastoid cells that is
most important, but the common dysregulation of JAKMIP1
and GPR155 downstream of these single gene defects,
which is observed in idiopathic ASD.
JAKMIP1isassociatedwithJanuskinases(41),microtubules
(41) and GABRB receptors (40). The expression levels of
JAKMIP1 affect the intracellular levels of the GABRBrecep-
tor (40). Because the GABRBreceptor could interact with the
metabotropic glutamate receptor 1 (mGluR1) and increase the
glutamate sensitivity of mGluR1 (53), JAKMIP1 might affect
mGluR1 signaling through GABRBreceptors. It is important
to note that mGluR signaling is exaggerated in Fmr1 knock-
out mice (54) and that glutamergic and GABAergic systems
have been reported to be abnormal in autism (55). Jakmip1
is highly expressed throughout the mouse brain, especially
in hippocampus, where GABRBreceptors and mGluR1 are
also highly expressed (56). Although the function of Gpr155
is unknown, it is highly expressed in the limbic system in
mouse brain (56), suggesting that Gpr155 might have func-
tions relevant to the limbic system.
It is also interesting to consider how the reduction of FMR1
and the induction of CYFIP1 might regulate the expression of
JAKMIP1 and GPR155. G-quadruplex motifs in RNA have
been shown to play significant roles in FMRP binding (57).
Using QGRS mapper (58), we found that human and mouse
JAKMIP1 each had two of the G-quadruplex (G2N2–4G2N2–4
G2N2–4G2) and that human and mouse GPR155 had five and
one of the G-quadruplex, respectively (data not shown).
FMRP can also bind target RNAs through non-coding RNAs
(59) or microRNAs (60). Using miRBase (61), we found puta-
tive microRNA binding sites in human and mouse JAKIMIP1
and GPR155 (data not shown). Further studies are required to
clarify the functional importance of JAKMIP1 and GPR155
in autism and the mechanism of regulation of these genes by
FMR1 and CYFIP1. In this regard, the potential link with
neuronal transmission is intriguing.
Figure 7. Molecular convergence of FMR1-FM, dup(15q) and idiopathic
ASD. The mRNA expression profile in lymphoblastoid cells from autism
with FMR1-FM or dup(15q) and control were compared using microarray
analysis. Sixty-eightgeneswere
FMR1-FM and dup(15q). Induction of CYFIP1 in dup(15q) is a potential mol-
ecular link between FMR1-FM and dup(15q). Among the dysregulated genes,
JAKMIP1 and GPR155 were further analyzed to confirm the causal relation-
ship between CYFIP1 and FMR1 expression and their expression in neural
cells or tissue and to validate the dysregulation of these genes in lymphoblas-
toid cells from subjects with idiopathic ASD.
dysregulated in bothautism with
Figure 6. JAKMIP1 and GPR155 are dysregulated in the ASD proband in discordant male sib pairs. Total RNA was extracted from lymphoblastoid cells of 27
male sib pairs discordant for ASD and qRTPCR were performed to confirm the differential expression of JAKMIP1 and GPR155. Results represent means+ SD
of each group. The mean of the value of control subjects was set as 1. P-value was calculated by Wilcoxon rank-sum test using control (N ¼ 27) versus ASD
(N ¼ 27).?P , 0.05.
Human Molecular Genetics, 2007, Vol. 16, No. 14 1693
at Universiti Teknologi Malaysia on May 18, 2011
hmg.oxfordjournals.org
Downloaded from

Page 13

The expression of JAKMIP1 and GPR155 was also
dysregulated in 27 males with idiopathic ASD
The findings in autism with FMR1-FM and dup(15q) suggest
that JAKMIP1 and GPR155 may be involved more generally
in idiopathic ASD, since their dysregulation is observed in
neural cells and brain. We tested whether dysregulation of
these genes were more generalizable in an independent
sample of idiopathic ASD cases. To attempt to reduce the het-
erogeneity of idiopathic ASD and extend these findings
beyond those with mental retardation or intellectual disability,
we used an IQ surrogate based on Raven’s Progressive
Matrices, which is highly correlated with IQ defined by
other measures (62). We selected 27 ASD males with an IQ
score of more than 70. These data demonstrated that the
expression of JAKMIP1 and GPR155 was significantly dysre-
gulated in lymphoblastoid cells from idiopathic ASD when
compared with controls. These results based on independent
data on lymphoblastoid cell gene expression from ASD sub-
jects with FMR1-FM, or dup(15q), as well as idiopathic
ASD, suggest that JAKMIP1 and GPR155 may be useful as bio-
markers for ASD. In this sample overall, JAKMIP1 appeared to
be more robustly dysregulated than GPR155, although the
expression differences appeared heterogeneous. How the dysre-
gulation of these genes relates to distinct autism subtypes will
also be important to determine in the future.
The mechanism for the opposite regulation of JAKMIP1
and GPR155 in lymphobastoid cells and neural cells remains
unclear. There are several previous reports of genes showing
the opposite expression between lymphoblastoid cells and
brains in neuropsychiatric disease. One example is inositol
monophosphatase 2 (IMPA2), which has been identified as a
plausible locus for bipolar disorder (63–65). The expression
of IMPA2 was reduced and induced in lymphoblastoid cells
and brains, respectively, in patients with bipolar disorder
(66). A genetic association between IMPA2 promoter poly-
morphism and bipolar disorder has been confirmed (67,68).
In this regard, it is notable that GPR155 is located on
2q31.1, 300 kb from D2S2188, which has shown strong
linkage to autism in studies by two independent groups
(6,69). Association analyses for GPR155 and JAKMIP1 are
ongoing using the large AGRE cohort. These data provide
the first identification and independent validation of the roles
of JAKMIP1 and GPR155 dysregulation in ASD (Fig. 7).
Further work is needed to understand the functional conse-
quences of these changes in the developing brain and to
assess the general utility of these and other genes as potential
biomarkers.
MATERIALS AND METHODS
Individuals and lymphoblastoid cells analyzed in this study
We have analyzed individuals diagnosed with ASD using stan-
dard validated measures, including the Autism Diagnostic
Interview (ADI-R) (70) and Autism Diagnostic Observation
Schedule (ADOS) (71). Eight males with FMR1-FM and
three males with dup(15q) were drawn from AGRE (33)
(http://www.agre.org/). An additional
dup(15q) were obtained from N.C.S. Twenty-seven males
without autism, FMR1-FM or dup(15q) were drawn from the
four maleswith
AGRE for controls. In addition, another 27 males with idio-
pathic ASD and with unaffected male siblings were chosen
from AGRE for a comparison sample (Supplementary
Material, Table S1). Surrogate IQ scores (using the Raven Pro-
gressive Matrices) were available. FMR1-FM and dup(15q)
were examined by PCR and fluorescence in situ hybridization,
respectively. The 15q11–q13 duplicated regions in the seven
males analyzed in this study were all maternally derived.
We also used 14 other individuals from AGRE for common
reference (pool) in microarray analysis (data not shown).
Lymphoblastoid cell lines (human Epstein-Barr virus trans-
formed lymphocytes) from these individuals were available
from AGRE.
The lymphoblastoid cells of the subjects were grown in
RPMI 1640 medium with 2 mM L-glutamine and 25 mM
HEPES (Invitrogen, Carlsbad, CA, USA), 10% fetal bovine
serum, and 1 ? Antibiotic-Antimycotic solution (Invitrogen)
at 378C in a humidified 5% CO2chamber. Cells were grown
to a density of 6 ? 105/ml. Special attention was given to
maintain all the cell lines in the same conditions to minimize
environmental variation.
Microarray experiments
A total of 9 ? 106of lymphoblastoid cells were seeded in a
T-75 flask in 30 ml of fresh medium. After 24 h, total RNA
was extracted from the cells using an RNeasy Mini Kit with
DNase treatment (Qiagen, Valencia, CA, USA) according to
the manufacturer’s protocol. The RNA quantity and quality
were measured by ND-100 (Nanodrop, Wilmington, DE,
USA) and 2100 Bioanalyzer (Agilent, Santa Clara, CA,
USA), respectively.
Target preparation was performed using a Low RNA Input
Fluorescent Linear Amplification Kit (Agilent) according to
the manufacturer’s protocol. We extracted total RNA from
lymphoblastoid cells from each individual and made target
and labeled it with Cy5 fluorescence. We also made reference
target by using pooled total RNA from the 14 individuals for
reference and labeled it with Cy3 fluorescence. The generated
targets were mixed and subjected to hybridization to the
Whole Human Genome Array G4112A (Agilent) according
to the manufacturer’s protocol. Scanning of the microarrays
were done by a DNA microarray scanner (Agilent).
Scanner output image files were normalized and filtered
using Feature Extraction Software v8.5 (Agilent). Normaliza-
tion was performed so that overall intensity ratio of Cy5 to
Cy3 was equal to one. Probes with signal-to-noise ratio
.2.7 in both Cy3 and Cy5 in at least 14 of 15 controls
were used for further analysis.
Statistical analysis of microarray data
ANOVA was performed by MeV3.1 (72). P-values were cal-
culated based on 1000 permutations. Hierarchical clustering
using Spearman’s rank correlation with average linkage clus-
tering was performed by MeV3.1. PCA was performed using
GeneSpring GX7.3 (Agilent). SAM (38) and RankProd (39)
were performed using Bioconductor (73) packages Siggene
and RankProd, respectively. For cross-validation in SAM
and RankProd, respectively, 100 and 1000 permutations
1694 Human Molecular Genetics, 2007, Vol. 16, No. 14
at Universiti Teknologi Malaysia on May 18, 2011
hmg.oxfordjournals.org
Downloaded from

Page 14

were performed. We used three different statistical tests to
conservativelyidentifythe
expressed genes. Numerous feature selection methods have
been applied to the identification of differentially expressed
genes in microarray data (74). The genes commonly identified
by ANOVA, SAM and RankProd are likely to be differentially
expressed, given the relative robustness of these statistical
approaches (35,38,39,74). Functional Annotation Clustering
was performed by DAVID (42) with medium classification
stringency. The clustering algorithm is based on the hypoth-
esis that similar annotations should have similar gene
members. The Functional Annotation Clustering uses two
different statistics to measure the degree of the common
genes between two annotations and to classify the groups
with similar annotations. The Group Enrichment Score is the
geometric mean (in 2log scale) of a member’s P-values in
a corresponding annotation cluster. IPA was used to find sig-
nificant pathways related to the genes commonly dysregulated
in autism with FMR1-FM and dup(15q). The Ingenuity
Pathway Knowledge Base builds gene networks based upon
known protein and gene interactions (44). IPA determines a
statistical score for each network according to the probability
of the network given in the gene list. The Ingenuity Pathway
Knowledge Base provides pathways with biological function
based upon the scientific literature. The significance value
associated with Functions and Pathways measures how likely
it is that genes from the data set file participate in that biologi-
cal function. The significance was expressed as a P-value,
which is calculated using the right-tailed Fisher’s exact test.
mostrobustly differentially
qRTPCR analysis
One microgram of total RNA was used to make cDNA by
SuperScript III First-Strand Synthesis SuperMix (Invitrogen).
The qRTPCR was done by ABI Prism 7900 (Applied Biosys-
tems, Foster City, CA, USA) using Platinum SYBR Green
qPCR SuperMix UDG with ROX (Invitrogen). Thermal
cycling consisted of an initial step at 508C for 2 min followed
by another step at 958C for 2 min and 50 cycles of 958C
for 15 s and 608C for 30 s. The qRTPCR was performed
for 16 genes. The primers used in this study are shown
in Supplementary Material, Table
(Hs00327005_m1,Applied
measure JAKMIP1 expression in lymphoblastoid cells. Data
were normalized by the quantity of hypoxanthine phosphori-
bosyltransferase 1 (HPRT1). HPRT1 was selected rather than
beta-actin,glyceraldehyde-3-phosphate
other possible internal controls because it was shown to be
the most stable RNA species from the lymphoblastoid cell
lines. This allowed us to account for the variability in the con-
version efficiency of the reverse transcription reaction.
S8. TaqMan probe
Biosystems)wasused to
dehydrogenaseor
Transduction of retroviral shRNAs
To construct retrovirus vectors expressing shRNAs, oligonu-
cleotides encoding stem-loop shRNAs for FMR1 (Supplemen-
tary Material, Table S8) and negative control were ligated into
the BamHI and EcoRI site of the pSIREN-RetroQ (BD Clon-
tech, Mountain View, CA, USA). PT67 cells (BD Clontech)
were transfected for retrovirus production. A total of
6 ? 106of SH-SY5Y cells were seeded in a T-75 flask in
20 ml of fresh medium of DMEM (Invitrogen) with 10%
FBS. After 1 day, SH-SY5Y cells were infected with retro-
viruses in the presence of 5 mg/ml of polybrene. After 2
days, the SH-SY5Y cells were treated with 10 mg/ml of puro-
mycin (Sigma, St Louis, MO, USA). Cells that survived after 4
weeks were collected, and this population of cells was used for
further experiments. Total RNA was extracted from the cells
using RNeasy Mini Kit with DNase treatment (Qiagen)
according to the manufacturer’s protocol. We compared
SH-SY5Y cells expressing FMR1 shRNA (n ¼ 4) and
SH-SY5Y cells expressing shRNAs for negative control
(n ¼ 4) to examine the effect of reduction of FMR1 on the
expression of JAKMIP1 and GPR155.
Transfection of CYFIP1
The human CYFIP1 coding region (amino acid 1–1254)
obtained by PCR using IMAGE clone 10625411 (ATCC,
Manassas, VA, USA) was subcloned into the EcoRV and
NotI sites of the plasmid vector pIRES-neo3 (BD Clontech).
The sequence of the construct was confirmed by automated
DNA sequencing.
A total of 6?106of SH-SY5Y cells were seeded in a T-75
flask in 20 ml of fresh medium of DMEM (Invitrogen) with
10% FBS. After 1 day, SH-SY5Y cells were transfected
with 120 ml of lipofectamine 2000 (Invitrogen) diluted with
3 ml of OptiMEM (Invitrogen) and 24 mg of plasmid (pIRES-
CYFIP1 or pIRES) diluted with 3 ml of OptiMEM (Invitro-
gen). After 5 min at room temperature, they were combined
and incubated for 20 min. The reaction mixture was added
with 16 ml of DMEM with 10% FBS. The cell culture
medium was replaced by this solution. After 2 days, the
SH-SY5Y cells were treated with 500 mg/ml of G418 (Invitro-
gen). Cells that survived after 3 weeks were collected, and this
population of cells was used for further experiments. Total
RNA was extracted from the cells using RNeasy Mini Kit
with DNase treatment (Qiagen) according to the manufac-
turer’s protocol. We compared SH-SY5Y cells stably trans-
fected with the expression vector for CYFIP1 (n ¼ 7) and
SH-SY5Y cells transfected with the empty expression vector
(n ¼ 8) to examine the effect of induction of CYFIP1 on the
expression of JAKMIP1 and GPR155. Protein was also
extracted using Cellytic M (Sigma) with proteinase inhibitors
(Sigma) according to the manufacturer’s protocol.
Animals and tissue collection
WT and Fmr1 KO mice were raised at the Emory University
animal facility and treated in accordance with National Insti-
tue of Health regulations and under approval of the Emory
University Institutional Animal Care and Use Committee.
WT and Fmr1 KO littermates were produced by breeding het-
erozygous females with FMR1 KO males in a congenic back-
ground of C57BL/6. The genotype of each animal was
confirmed by PCR. For tissue collection, cortices were dis-
sected and protein was isolated using Cellytic M (Sigma)
with proteinase inhibitors (Sigma) according to the manufac-
turer’s protocol.
Human Molecular Genetics, 2007, Vol. 16, No. 141695
at Universiti Teknologi Malaysia on May 18, 2011
hmg.oxfordjournals.org
Downloaded from

Page 15

Immunoblot analysis
Proteins extracted from SH-SY5Y cells or cortices of Fmr1
WT and KO mice were subjected to SDS–PAGE using
NuPAGE Novex 4–20% Bis–Tris gel and MOPS buffer (Invi-
trogen) according to the manufacturer’s protocol. After elec-
trophoresis, gels were electroblotted onto PVDF membranes
(Millipore, Bedford, MA, USA). After electroblotting, mem-
branes were blocked in SuperBlock blocking buffer (Pierce
Biotechnology,Rockford, IL,
probed in the blocking solution at 48C overnight with the fol-
lowing antibodies: FMRP (Chemicon, Temecula, CA, USA),
CYFIP1 (Upstate, Temecula, CA, USA), JAKMIP1 (41) or
GAPDH. Membranes were washed 3? in PBS supplemented
with 0.05% Tween 20 (PBS-T) and incubated with the appro-
priate horseradish peroxidase-conjugated secondary antibody
in the blocking solution for 1 h at room temperature. Mem-
branes were again washed 3? in PBS-T and developed with
SuperSignal West Pico Chemiluminescent (Pierce Biotechnol-
ogy). Membranes were stripped by Restore Western Blot
Stripping Buffer (Pierce Biotechnology) and used for different
antibodies.
USA).Membranes were
SUPPLEMENTARY MATERIAL
Supplementary Material is available at HMG Online.
ACKNOWLEDGEMENTS
We gratefully acknowledge the resources provided by the
AGRE consortium and sincerely thank the AGRE families
and the IsoDicentric 15 Exchange Advocacy and Support
(IDEAS) group who have participated in the study. AGRE is
a program of the Cure Autism Now foundation (CAN) and
is supported, in part, by grant MH64547 from the National
Institute of Mental Health (NIMH) to D.H.G. (PI). This
work was supported by NIMH grant MH64547 (to D.H.G.),
The Gassin Family Foundation (to Y.N. and D.H.G.), The
Boler Company Foundation (to Y.N. and D.H.G.), NICHD
grant HD020S21 (to S.T.W.) and NIH U19-HD35470 CPEA
grant (to N.C.S., Marian Sigman, PI). Y.N. is a recipient of
a Young Investigator Award from CAN. We are also grateful
to Giovanni Coppola and Brett Abrahams for valuable discus-
sions; Gena Konopka and Michelle Stofa for critical readings
of the manuscript; Jeffrey Gregg, Stephenie Liu, Barb Malone
and Jen Driscoll for providing lymphoblastoid cells with
dup(15q): Tamika Malone for providing brains of FMR1-KO
and WT mice; and Lauren Kawaguchi for her help as labora-
tory manager.
Conflict of Interest statement. D.H.G. provides consulting
service to AGRE, a non-profit arm of Cure Autism Now, as
Chief Scientific Officer.
APPENDIX
The AGRE Consortium: D.H.G., M.D., Ph.D., UCLA, Los
Angeles, CA; Maja Bucan, Ph.D., University of Pennsylvania,
Philadelphia, PA; W. Ted Brown, M.D., Ph.D., F.A.C.M.G.,
N.Y.S. Institute for Basic Research in Developmental
Disabilities, Staten Island, NY; Rita M. Cantor, Ph.D.,
UCLA School of Medicine, Los Angeles, CA; John
N. Constantino, M.D., Washington University School of
Medicine, St Louis, MO; T. Conrad Gilliam, Ph.D., University
of Chicago, Chicago, IL; Martha Herbert, M.D., Ph.D.,
Harvard Medical School, Boston, MA; Clara Lajonchere,
Ph.D, CureAutismNow,
H. Ledbetter, Ph.D., Emory University, Atlanta, GA; Christa
Lese-Martin, Ph.D., Emory University, Atlanta, GA; Janet
Miller, J.D., Ph.D., Cure Autism Now, Los Angeles, CA;
Stanley F. Nelson, M.D., UCLA School of Medicine, Los
Angeles, CA; Gerard D. Schellenberg, Ph.D., University of
Washington, Seattle, WA; Carol A. Samango-Sprouse,
Ed.D., George Washington University, Washington, D.C.;
S.J.S., M.D., Ph.D., NIMH, Bethesda, MD; Matthew State,
M.D., Ph.D., Yale University, New Haven, CT; Rudolph
E. Tanzi, Ph.D., Massachusetts General Hospital, Boston, MA.
Los Angeles,CA; David
REFERENCES
1. Folstein, S.E. and Rosen-Sheidley, B. (2001) Genetics of autism: complex
aetiology for a heterogeneous disorder. Nat. Rev. Genet., 2, 943–955.
2. Belmonte, M.K., Cook, E.H., Jr., Anderson, G.M., Rubenstein, J.L.,
Greenough, W.T., Beckel-Mitchener, A., Courchesne, E., Boulanger,
L.M., Powell, S.B., Levitt, P.R. et al. (2004) Autism as a disorder of
neural information processing: directions for research and targets for
therapy. Mol. Psychiatry, 9, 646–663.
3. Veenstra-Vanderweele, J., Christian, S.L. and Cook, E.H., Jr. (2004)
Autism as a paradigmatic complex genetic disorder. Annu. Rev. Genomics
Hum. Genet., 5, 379–405.
4. Muhle, R., Trentacoste, S.V. and Rapin, I. (2004) The genetics of autism.
Pediatrics, 113, e472–e486.
5. Liu, J., Nyholt, D.R., Magnussen, P., Parano, E., Pavone, P., Geschwind,
D., Lord, C., Iversen, P., Hoh, J., Ott, J. et al. (2001) A genomewide
screen for autism susceptibility loci. Am. J. Hum. Genet., 69, 327–340.
6. IMGSAC (2001) A genomewide screen for autism: strong evidence for
linkage to chromosomes 2q, 7q, and 16p. Am. J. Hum. Genet., 69,
570–581.
7. IMGSAC (2001) Further characterization of the autism susceptibility
locus AUTS1 on chromosome 7q. Hum. Mol. Genet., 10, 973–982.
8. Buxbaum, J.D., Silverman, J.M., Smith, C.J., Kilifarski, M., Reichert, J.,
Hollander, E., Lawlor, B.A., Fitzgerald, M., Greenberg, D.A. and Davis,
K.L. (2001) Evidence for a susceptibility gene for autism on chromosome
2 and for genetic heterogeneity. Am. J. Hum. Genet., 68, 1514–1520.
9. Alarcon, M., Cantor, R.M., Liu, J., Gilliam, T.C. and Geschwind, D.H.
(2002) Evidence for a language quantitative trait locus on chromosome 7q
in multiplex autism families. Am. J. Hum. Genet., 70, 60–71.
10. Yonan, A.L., Alarcon, M., Cheng, R., Magnusson, P.K., Spence, S.J.,
Palmer, A.A., Grunn, A., Juo, S.H., Terwilliger, J.D., Liu, J. et al. (2003)
A genomewide screen of 345 families for autism-susceptibility loci.
Am. J. Hum. Genet., 73, 886–897.
11. Stone, J.L., Merriman, B., Cantor, R.M., Yonan, A.L., Gilliam, T.C.,
Geschwind, D.H. and Nelson, S.F. (2004) Evidence for sex-specific
risk alleles in autism spectrum disorder. Am. J. Hum. Genet., 75,
1117–1123.
12. Alarcon, M., Yonan, A.L., Gilliam, T.C., Cantor, R.M. and Geschwind,
D.H. (2005) Quantitative genome scan and Ordered-Subsets Analysis of
autism endophenotypes support language QTLs. Mol. Psychiatry, 10,
747–757.
13. Cantor, R.M., Kono, N., Duvall, J.A., Alvarez-Retuerto, A., Stone, J.L.,
Alarcon, M., Nelson, S.F. and Geschwind, D.H. (2005) Replication of
autism linkage: fine-mapping peak at 17q21. Am. J. Hum. Genet., 76,
1050–1056.
14. McCauley, J.L., Li, C., Jiang, L., Olson, L.M., Crockett, G., Gainer, K.,
Folstein, S.E., Haines, J.L. and Sutcliffe, J.S. (2005) Genome-wide and
Ordered-Subset linkage analyses provide support for autism loci on 17q
and 19p with evidence of phenotypic and interlocus genetic correlates.
BMC Med. Genet., 6, 1.
1696Human Molecular Genetics, 2007, Vol. 16, No. 14
at Universiti Teknologi Malaysia on May 18, 2011
hmg.oxfordjournals.org
Downloaded from