Transcript XVth World Congress for Psychiatric Genetics

HUMAN
GENETICS
NIJMEGEN
Genome-wide scan of copy number variation
in Attention deficit hyperactivity disorder (ADHD)
1,2
Franke ,
1
Hehir-Kwa ,
1,2
AriasVasquez ,
1
Veltman ,
1
Bloemen ,
3
Lasky-Su ,
Barbara
Jayne
Alejandro
Joris
Bart
J.
P.
Asherson4, M. Gill5, J. Sergeant6, R. Ebstein7, A. Rothenberger8, HC. Steinhausen9, T. Banaschewski10, R.
11
12
13
14
2
15
Oades , E. Sonuga-Barke , A. Miranda , H. Royers , J. Buitelaar , S.V. Faraone , for the IMAGE consortium
Departments of 1Human Genetics and 2Psychiatry, Radboud University Nijmegen
Medical Centre, Nijmegen, The Netherlands; 3Boston, MA, USA; 4London, UK;
5Dublin, Ireland; 6Amsterdam, NL; 7Jerusalem, Israel; 8Göttingen, Germany; 9Zurich,
Switzerland; 10Mannheim and 11Essen, Germany; 12Southampton, UK; 13Valencia,
Spain; 14Ghent, Belgium; 15SUNY Upstate Medical University, Syracuse, NY, USA
correspondence to: [email protected]
Abstract
Recent data suggest that copy number variants (CNVs) can contribute to complex disease susceptibility. The relative impact of CNVs compared to single nucleotide polymorphisms (SNPs) on one of the
processes underlying disease vulnerability, variable gene expression, has been estimated at 18% (1). The involvement of CNVs in ADHD etiology has not been investigated. Within the International Multisite ADHD Genetics (IMAGE) study, sponsored by the Genetic Association Information Network (GAIN), a whole genome association study investigating over 500.000 SNPs has been carried out on 949
European Caucasian parent-child trios with offspring meeting the DSM-IV combined-type criteria for ADHD. Families were collected in the Netherlands, Ireland, UK, Germany, Belgium, Switzerland,
Spain and Israel. Using the intensity data from the SNP analysis carried out at Perlegen Sciences, copy number information is extracted for each individual. Pilot data are presented. In the next analysis
steps, we plan to identify known and new CNVs in the patients and their parents. By comparing parents with offspring we will investigate which CNVs are inherited, which are de novo. For inherited CNVs
a TDT-based association study will be carried out. For those that occur de novo in the patients, we will investigate the gene content to find out if the CNV carries genes that can explain the presence of
ADHD in the particular patient.
In conclusion: The Perlegen SNP-microarrays used in the GAIN projects contain copy number information. The CNV analysis can potentially identify new candidate genes and new risk alleles for ADHD.
Introduction
Preliminary Results and Future Plans
Copy number variants (CNVs) comprise a newly identified type of structural variation within the
Raw Log2 intensity ratios of the sample versus a reference pool were calculated using the QC
human genome. CNVs include insertions, deletions and duplications and encompass relatively
unfiltered allele 1 and 2 intensity data. For this the intensity of the two different alleles is
large genomic segments of 1kb to several Mb in size. Detection of CNVs can be achieved by
combined to give an average intensity for each SNP. This is compared to the average intensity
karyotyping, (array) CGH and SNP microarrays (reviewed in (2)), depending on their size (3).
in a reference (pool), calculated also across both alleles. These Log2 intensity ratios are plotted
CNVs are very common in the human genome (around 12% of the genome) and their
frequencies can reach high percentages in the population. Depending on their location, they can
include whole genes and influence their regulation due to dosage effects (1). For many of the
rare CNVs there is a clear link with human disease (4-6); they often occur de novo in monogenic
for each SNP according to their chromosomal position (Figure 1). The plots indicate the
presence of both loss and gain CNVs. The signal to noise in these plots will be further improved
to allow the detection of smaller CNVs, and a statistical algorithm (Hidden Markov Model) will be
incorporated to automatically detect all CNVs present.
disorders. This link with disease is less clear for the polymorphisms within the CNV category
(those with frequencies above 1% in the population). However, a number of studies have
Chromosome 14
already found evidence that common CNVs can be risk factors for disease, e.g. in autoimmune
Chromosome
Locus in
CNV
database
Method used for
detection
Gains
or
losses
conditions inlcuding glomerulonephritis, Crohn’s disease, systemic lupus erythematosus,
• Affymetrix 500K SNP
polyangiitis and Wegener’s granulomatosis (7-10) and in HIV susceptibility/progression (11).
14: 19.28-19.5
Mb
In the current study we set out to study the role of CNVs in the etiology of ADHD. This highly
Locus 2884
heritable disorder is the most common neuropsychiatric disorder in children (seen in 3-5% of
school-aged children). It also has a prevalence of at least 1% in the adult population. ADHD is
22: 21.35-21.56
Mb
characterized by early-onset, age-inappropriate and persistent hyperactive, inattentive and
Locus 3746
array
• Affymetrix 100K SNP
array
• ROMA
• Resequencing trace
mapping
• Array-CGH
• ROMA
• Computational
Fosmid End Mapping
Gains
and
losses
Gains
and
losses
impulsive behavior.
Experimental design
Within the IMAGE project, 1400 European Caucasian families with at least one child aged 5-19
Figure 1: Example of the detection of known
Chromosome 22
affected with combined subtype ADHD have been recruited from 8 countries (Netherlands, UK,
CNVs on chromosomes 14 and 22 in the
Ireland, Germany, Israel, Spain, Switzerland, Belgium). Diagnosis of ADHD was based on the
same ADHD patient from the GAIN project
PACS (Parental Account of Childhood Symptom) combined with the teacher-rated Conners’
(phs000016). The arrowheads indicate a
copy number gain on chromosome 14 and a
ADHD subscale (CTRS-R:L) and situational pervasiveness (see dbGAP website for more
copy number loss on chromosome 22. The
information: http://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?id=phs000016).
table above shows that these CNVs have
been reported before in healthy individuals,
As part of the Genetic Association Information Network (GAIN (12)), a total of 949 child-parent
as registered in the ‘Database of Genomic
triads (2835 samples) were genotyped on a newly developed SNP-array by Perlegen Sciences.
Variants’
SNP selection for the Perlegen array was guided by a linkage disequilibrium (LD) analysis of the
(http://projects.tcag.ca/variation)
(13).
integrated Phase I and Phase II HapMap data, using a minimum pairwise r2 threshold of 0.8, for
SNPs with a minor allele frequency (MAF) of 0.05 or higher. The set comprises almost 600,000
The following analyses will be carried out after extraction of the copy number information in all
SNPs, yielding nearly complete coverage of the European Caucasian (CEU) map. In addition to
patient-parent trios:
the LD-based SNP selection, 20,000 non-synonymous SNPs culled from the HapMap, Celera’s
•
exon resequencing data, and Perlegen data were included. For Caucasian samples, the
average r2 between a SNP in the HapMap and a working assay is 0.94. The percentage of
SNPs in the HapMap with r2 > 0.8 with a Perlegen SNP is 93%.
•
References
(1) Stranger et al., Science. 2007 Feb 9;315(5813):848-53.
(2) Feuk, Carson, Scherer. Nat Rev Genet. 2006 Feb;7(2):85-97.
(3) Hehir-Kwa et al., DNA Res. 2007 Feb 28;14(1):1-11.
(4) Vissers et al., Nat Genet. 2004 Sep;36(9):955-7.
(5) de Vries et al., Am J Hum Genet. 2005 Oct;77(4):606-16.
(6) Koolen et al., Nat Genet. 2006 Sep;38(9):999-1001.
(7) Aitman et al., Nature. 2006 Feb 16;439(7078):851-5.
(8) Fellermann et al., Am J Hum Genet. 2006 Sep;79(3):439-48.
(9) Fanciulli et al., Nat Genet. 2007 Jun;39(6):721-3.
(10) Yang et al., Am J Hum Genet. 2007 Jun;80(6):1037-54.
(11) Gonzalez et al., Science. 2005 Mar 4;307(5714):1434-40.
(12) The GAIN Collaborative Research Group et al., Nat Genet.
2007 Sep;39(9):1045-1051.
(13) Redon et al., Nature. 2006 Nov 23;444(7118):444-54.
•
Common, inherited CNVs (≥ 5% allelic frequency):
•
•
•
Association study with ADHD using TDT and qTDT.
Are the CNVs tagged by SNPs (LD / r2 pattern)?
Are the findings from the SNP WGAS and the CNV analysis comparable?
Rare, inherited CNVs:
•
A sum score of CNVs will be created and transmission to children will be tested
De novo CNVs:
•
•
•
What genes present in the CNV?
Are the genes possibly related to ADHD?
Are the CNVs present in regions found in linkage/association studies in ADHD?
Websites: www.humangenetics.nl; www.ncmls.eu