Title: Autism is a common complex neurological disorder with a high heritability 90% 1'
1Linkage and Candidate Gene Studies of Autism in
EU Populations
R. Holt1, G. Barnby1, E. Maestrini2, E.
Bacchelli2, D. Brocklebank1, A. J. Bailey3, A. P.
Monaco1 EU Autism MOLGEN Consortium
1 Wellcome Trust Centre for Human Genetics,
Roosevelt Drive, Oxford, OX3 7BN, UK, 2
Dipartimento di Biologia, Università di Bologna,
Via Selmi 3, 40126 Bologna, Italy, 3 University
Department of Psychiatry, Park Hospital for
Children, Oxford, UK
Results
Introduction
- Linkage
- 379 SNPs and 1209 samples passed quality
control with a mean genotyping success rate of
gt0.99. - Singlepoint analysis (CHECK!) was performed on
affacted sib pairs from a total of 397 families
when EU and AGP results were combined. - Results were analysed both with and without the
Finnish samples (32 families), as they are from
an isolated population and may, therefore, have
different susceptibility loci. - Analysis of all results found weak evidence of
linkage on chromosomes 3 and 6 (LOD scores of
1.64 and 1.57, respectively). - With Finnish samples excluded, evidence for
linkage increased on chromosome 2 (LOD 1.87),
but decreased on chromosomes 3 and 6 (LOD 1.08
and 1.17, respectively). - Figures 1 A C show the linkage pattern found
for these three loci. - LOD scores were not significant when linkage
disequilibrium was taken into account (r2 of 0.5
and 0.3). - Parent-of-origin analysis showed that the
linkage on chromosomes 2 and 3, in particular, is
due to maternal alleles. - The analysis of haplotypes is in progress.
- Autism is a common complex neurological disorder
with a high heritability (90) 1. - Attempts to identify the underlying causal
variants have met with limited success. - Linkage studies have identified possible
susceptibility loci on the majority of
chromosomes 2. - Association has been reported for a large number
of candidate genes, although replication has
proved difficult. - We have utilised three approaches to attempt to
find or replicate autism susceptibility loci
using European Union (EU) populations - Linkage meta-analysis of seven previously
implicated regions by combining data from an
Autism Genome Project (AGP) study 3 with that of
newly generated genotypes for additional samples.
- A search for extended haplotypes in regions of
linkage using trios from two isolated European
populations. - Association analysis of seven previously reported
candidate genes.
Materials and Methods
- SNP Selection
- Separate Illumina GoldenGate 384 SNP arrays
were designed for both the linkage/haplotype and
association studies. - SNPs with a MAF gt0.05 were chosen from the
HapMap CEU to tag variation within the regions
investigated. - SNPs included on the linkage array were chosen
from those on the Affymetrix 10K SNP array, used
by the AGP, to allow the combination of our data
and that of the AGP. - Additional SNPs of interest from published
literature were also included on the association
array. - Table 1 shows the distribution of SNPs for the
various loci covered by each array.
A
Figure 1 Pairwise linkage results for chromosomes
2 (A), 3 (B) and 6 (C). Graphs plot lod score
against position along the chromosome in cM.
Green lines indicate the results for all
families, blue lines are the result when the
Finnish families have been excluded.
C
B
Table 1 Distribution of SNPs on the two
genotyping arrays
- Association
- The FST test showed there is no significant
population structure in the combined data set. - 357 SNPs and 1127 samples passed quality
control with a mean genotyping success rate
gt0.99. - The two strongest associations were for rs362780
(RELN) and rs2518261 (GRIK2) (Table 3) (Figure 2)
(382 families). - Weak association found in PRKCB1 for the IMGSAC
population (279 families) (rs11074601, P
0.00596). - No associations were significant after Bon
Ferroni correction for multiple testing. - Analysis of the results for ASMT is ongoing.
- Samples
- Samples used were either
- Genomic DNA with a concentration of
60-100ng/µl. - Whole genome amplified (WGA) DNA.
- The sample populations used in this study were
IMGSAC (European and North American), PARIS
(French and Swedish samples), Finnish and Dutch. - Table 2 details the distribution of samples
genotyped for each array. - Replication of three SNPs from the candidate
gene array was performed in 215 trios from the
Dutch population using WGA DNA.
Figure 2 TDT meta-analysis of candidate gene
genotyping. 1 rs362780 (RELN) P 0.0016, 2
rs2518261 (GRIK2) P 0.0017, 3 rs11074601
(PRKCB1) P 0.017.
Table 2 Number of families run per
array (Multiplex MPX)
SLC6A4
NOSTRIN
GRIK2
RELN
PRKCB1
SHANK3
2
1
3
- Genotyping and Analysis
- Genotyping was performed using standard
Illumina protocols. - Replication of three association array SNPs was
performed using Sequenom iPlex genotyping
(standard protocols). - Genotypes were called in BeadStudio, with some
manual adjustment of clusters. Poor clustering
SNPs were excluded. - Mendelian errors were identified and removed
using PedCheck. Samples with a genotype call rate
of lt80 or with gt10 Mendelian inheritance errors
were also excluded. - Prior to analysis, the genotypes from the
linkage array where combined with those for an
additional 362 IMGSAC and 49 PARIS families
genotyped for the same SNPs by the AGP. - Linkage analysis was performed using Merlin.
Parent-of-origin analysis was performed with the
sex_split option of ASPEX. - Association analysis was performed using the
transmission disequilibrium test with the STATA
package. The data was analysed both by combining
the results across all populations genotyped and
by performing a meta-analysis using weighted odds
ratios. - Association to autism and the broader phenotype
of autism spectrum disorders (ASDs) was tested. - Stratification was analysed using an FST test
on 105 SNPs (including 30 specifically genotyped
for the purpose table 1) from the association
array.
Table 3 Top association results for combined and
meta-analyses.
Conclusions
- We were unable to identify significant regions
of linkage, either when our results were combined
with those of samples from the same populations
previously genotyped by the AGP, nor when they
were analysed separately (data not shown). Weak
evidence for linkage was found on chromosomes 2,
3 and 6. - Haplotype analysis in two founder populations
is ongoing. - We found weak evidence for association for
single SNPs in RELN, GRIK2 and PRKCB1. This weak
association failed to reach significance when
accounting for multiple testing. - Replication of the three most significant
association results is currently being performed
in a Dutch population. - We have, therefore, been unable to narrow the
known autism linkage regions. Our results do,
however, support the plausibility of RELN
containing autism susceptibility variants. - Increasingly, the importance of rare variants
of large effect 4, including copy number variants
5, are being recognised in autism susceptibility.
Therefore, it is perhaps not surprising that
there is continued difficulty in identifying
susceptibility alleles using methods designed to
find common alleles of moderate effect, such as
association studies.
References
- Rutter, M. Genetic studies of autism from the
1970s into the millennium. J Abnorm Child
Psychol. 28(1), 3-14 (2000). - Bacchelli, E., Maestrini, E. Autism spectrum
disorder molecular genetic advances. Am J Med
Genet C Semin Med Genet. 142C(1), 13-23 (2006). - Szatmari, P. et al. Mapping autism risk loci
using genetic linkage and chromosomal
rearrangements. Nat Genet. 39(3), 319-28 (2007). - Durand, C. M. Mutations in the gene encoding the
synaptic scaffolding protein SHANK3 are
associated with autism spectrum disorders. Nat
Genet. 39(1), 25-7 (2007). - Jacquemont, M. L., et al. Array-based comparative
genomic hybridisation identifies high frequency
of cryptic chromosomal rearrangements in patients
with syndromic autism spectrum disorders. J Med
Genet. 43(11) 843-9 (2006).
Acknowledgments
Our thanks go to Chris Allan for performing the
genotyping. This work has been funded by the EU
Sixth Framework Programme.