Transcript Genomic studies of schizophrenia

Genomic studies of schizophrenia: mapping
madness?
Mike Owen, Department of Psychological Medicine, Wales
College of Medicine, Cardiff University, UK.
Genetic epidemiology of
schizophrenia
Familial, λs =10
 Individual differences in liability are largely genetic

– Heritability 0.6-0.9

Non-genetic factors also important
 Multi-locus model consistently supported by analysis of
family data
 Genes with population λs > 3 unlikely
 Number of susceptibility loci, degree of risk conferred by
each and degree of interaction all unknown
Finding genes for schizophrenia

Linkage
Disease gene
D18S21

Chromosomal abnormalities

Association

Convergent genomics
6p24-p22
8p22-p21
1q21-q22
5q21-q33
6q21-25
1q42
1
2
9
10
3
4
5
6
7
8
10p15-p11
Genome
wide
significant
10q 25-26
Multiple
positives
13q32-q34
11
12
13
14
15
16
17p11-q25
22q11-q22
17
18
19
20
21
22
Y
X
Published linkage
data 2006
Cardiff schizophrenia sib-pair
genome screen
VCFS
CNP
4.5
4
3.5
LOD
3
2.5
2
1.5
1
0.5
0
0
1
2 500
3
4 10005
6
71500 8
9
2000
10 11
12 132500
14 15 16 17 3000
18 19 20 21 223500
cM
Chromosome
351 microsatellite markers (10.3 cM) in 354 affected sib-pairs
179 (UK), 134 (Sweden), 41 (USA)
Williams et al. Am J Hum Genet 2003
Linkage studies of
schizophrenia 2006

Data consistent with
observed recurrence
risks in relatives

Meta-analysis suggests
some consensus

Many underpowered
studies on uncertain
value

Larger studies needed
for more genes, stronger
evidence, interactions
and better location

800-1000 ASPs optimal

A number of groups are
now seeking genes in
currently positive regions
Positional genetics and
schizophrenia




–
–
–
Positional Candidates
NRG1, 8p 1
DTNBP1, 6p 2
G72 (DAOA), 13q 3
Multiple positive
studies as well as
negative
No clear pattern of
associated alleles or
haplotypes
Risk variants not
identified
1. Stefansson et al. 2002; 2. Straub et al. 2002; 3. Chumakov et al. 2002;
Positional candidate genes for schizophrenia
6p24-p22
1q21-q22
DRD3
NRG1
RGS4
5q21-q33
6q21-25
1q42
2
1
8p22-p21
DTNBP1
3
4
5
6
7
8
14
15
16
Y
X
10p15-p11
HTR2A
10q 25-26
10
9
13q32-q34
G72/G30
DAO
11
12
13
17p11-q25
22q11-q22
17
18
19
20
21
22
NRG1 association with
schizophrenia
Tosato, Dazzan and Collier, Schiz. Bull. 2005.
DTNBP1: complex pattern of
association findings in schizophrenia
Williams, O’Donovan and Owen, Schiz Bull, 2005,
DTNBP1 and schizophrenia







Multiple studies suggest that variation in DTNBP is associated
with schizophrenia (11/14 positive).
Protection may be mediated by IQ.
No individual SNPs or haplotypes have consistently been
implicated in susceptibility.
No systematic study aimed at detecting all common genetic
variation.
More studies needed to identify risk variants.
Reduced expression of message and protein in schizophrenia –
cause or compensation?
Cis-acting elements regulate DTNBP expression.
– Can we relate specific haplotypes to gene expression?
DTNBP1 risk haplotype is associated
with reduced expression




Allele ratios at SNP rs1047631,
stratified by heterozygosity for the
defined schizophrenia risk
haplotype.
Schizophrenia risk haplotype tags
one or more cis-acting variants that
result in a relative reduction in
DTNBP1 mRNA expression in
human cerebral cortex.
Further analyses suggest that risk
haplotypes identified in other
Caucasian samples also index
lower DTNBP1 expression.
Ties risk haplotypes to altered
function
Suggests explanation for some of
the discrepancies between studies
1.2
1
0.8
Ratio A/G

0.6
0.4
0.2
Genomic
cDNA:
risk het
cDNA:
non risk
0
Bray et al, Human Molecular Genetics, 2005.
Chromosomal abnormalities and schizophrenia
6p24-p22
1q21-q22
DRD3
NRG1
RGS4
5q21-q33
6q21-25
1q42
1
8p22-p21
DTNBP1
2
3
4
5
6
7
8
14
15
16
Y
X
10p15-p11
HTR2A
13q32-q34
G72/G30
DAO
9
10
11
12
13
PRODH
COMT
22q11-q22
17
18
19
20
21
22
VCFS and Schizophrenia

25+% of adults with VCFS
develop schizophrenia
(Murphy et al 2000)

Does a gene on 22q11
predispose to schizophrenia in
the general population?

Some claims (COMT,
PRODH, ZDHHC8, ARVCF)
but none convincing
DISC1 and schizophrenia (Porteous,
Millar, Blackwood, St Clair et al)




Schizophrenia Associated t(1:11)
Translocation
High density of genes around ch#1
breakpoint
•
•
•
DISC1: unknown function; N globular domain; C -helical domain with coiled
coil domains; widely expressed, adult > fetal
DISC2: unknown function; RNA >15kb but no ORF; putative anti-sense RNA
regulator of DISC1; widely expressed, heart >> other
TRAX: translin related protein X; putative role in single strand DNA repair and
RNA translocation; intergenic splicing to DISC1

Possibility of position effect
Evidence for haplotype association
with SZ
Associations with visual and
working memory deficits
t (1,16) disrupts PDE4B (binding
partner of DISC1) in family with
psychosis (Millar et al 2005)
DISC1 complex protein associated
with numerous cytoskeletal
proteins involved in centrosomal
and microtubule function, and with
cell migration, neurite outgrowth,
and membrane trafficking of
receptors and possibly
mitochondrial function.
Candidate genes for schizophrenia



NRG1 and DTNBP1: multiple positive studies.
DAOA(G72): some positives for both schizophrenia and bipolar disorder
DISC1: highly promising.
Effect sizes small-moderate (OR 1.3-2)
RGS4: some positives but support weakening.
COMT, PRODH, ZDHHC8, PPP3CC, CAPON, CHRNA7, TRAR4 and
others: not yet convincing.
Cannot be explained by variants with manifest functional consequences.


Remain cautious until risk variants identified.
Inconsistencies between studies.




– Presumably effects on expression, splicing etc.
–
–
–
–
Lack of power
Genotyping error
Stratification
Incomplete genetic information from indirect studies
» NB dysbindin
– Susceptibility variants on different backgrounds??
– Allelic heterogeneity/complexity
– Heterogeneity: differences in case definition and ascertainment

It’s epidemiology now!
Emil Kraepelin, 1896, splits psychosis.

“crystallized dementia
praecox and manicdepressive illness from
an amorphous mass of
madness” (Brockington &
Leff, 1979).

Organic
Functional

– Dementia Praecox
– Manic-depressive insanity
Deconstructing Psychosis:
Challenges to the Kraepelinian
Dichotomy.
No “point of rarity”
Risk factors in common (Murray et al 2004)
Familial co-occurrence of SZP, SA and BP
Cardno et al twin study
Overlapping linkage regions
•
•
•
•
•
•

13q, 22q, 6q
New genetic studies confirm this and suggest
association with clinical syndromes.
Studying candidate genes across the
Kraepelinian divide: Dysbindin.

No association between
BP and the Cardiff
haplotype in DTNBP1.
Suggestive evidence for
association with BP with
predominant psychosis.
1.45
1.4
1.35
Odds ratio

1.3
1.25
1.2
1.15
1.1
1.05
1
0
20
40
60
P dimension
Upward trend: p = 0.014
Raybould et al, Biological Psychiatry, 57: 696-701, 2005.
80
100
Studying candidate genes across the
Kraepelinian divide: Neuregulin.
1.8
1.6
1.5
Odds ratio
1.4
1.3
1.2
1.1
BP-IP & SZP -M
Phenotype
BP-NIP & SZP-NM
SZP-M
SZP-NM
SZP
BP-IP
BP-NIP
1
BP
NRG1 HAPICE confers risk
to illness with both
schizophrenia and mood
features.
 Effect size of NRG1
HAPICE increases with
preponderance of moodincongruent psychotic
symptoms (sign test
p=0.002).
2.2
2
Odds ratio (OR)

Odds ratio
1.7
1.8
1.6
OR
1.4
1.2
1
0
Green et al, Archives of General Psychiatry, 2005.
10
20
I dimension
30
40
Studying candidate genes across the
Kraepelinian divide: G72 (DAOA).
 Some
positive replications of G72 but no clear
associated alleles or haplotypes.
 Independent
support for association with Bipolar
disorder in several studies.
– G72 probably strongest candidate gene for BP.
 Is
this a gene for both disorders?
DAOA (G72) in schizophrenia and
bipolar disorder

Significant whole gene association in BP (n=706,
p=0.045) but not SZ (n=709) vs controls (n=1416).
 Significant whole gene association in “Mood” (n= 1153,
p=0.0086) and in schizophrenia-mood (n=112, p=0.02)
but not psychosis (n=818).
 DAOA is probably a susceptibility locus for mood
disorder rather than schizophrenia per se.
 Extent to which association seen in schizophrenia
depends upon clinical characteristics of sample.
(Williams et al Archives of General Psychiatry, 2006).
DISC1 is associated with broad
psychosis and mood phenotype.

t(1:11) segregates with
Sz, BP and UP.
 Linkage to SA
– Hamshere et al 2005.

Evidence for haplotype
association
– Hennah et al 2003
– Hodgkinson et al 2004
» Schizophrenia, SA and BP
– Thomson et al 2005
» SZ and BP.
Using genetics to dissect psychosis
Craddock, O’Donovan and Owen, Schizophrenia Bulletin, 2006.
Do genetic findings in psychosis
point to a common mechanism?

The genes most clearly
implicated (NRG1, DTNBP1,
G72) all code for proteins that
potentially impact, directly or
indirectly, on the function of
glutamate synapses.
Harrison and Owen, Lancet,
2003.

But caution required!
– proteins implicated poorly
understood
– multiple processes implicated for
NRG1 and DTNBP1
White matter abnormalities in
schizophrenia (Mt. Sinai Conte Center)

Imaging studies
– Defects of connectivity
– DTI

Multiple gene expression studies in
postmortem schizophrenia brains
have found significant reduction of
expression of myelin and
oligodendrocyte related genes
(e.g. Hakak et al., 2001)
 Quantitative anatomical analyses
have demonstrated decreased
oligodendrocyte numbers in
prefrontal cortex (Hof et al 2002,
2003).

Cause or effect?






CNP is marker for
oligodendrocytes
Message and protein show
reduced brain expression in
schizophrenia
Located at 17q21.2
rs2070106 is associated with CNP
expression (P=0.001).
the lower-expressing allele was
significantly associated with
schizophrenia (P=.04) in a casecontrol sample.
All affected individuals in a linked
pedigree were homozygous for the
lower-expression allele (P=.03).
Archives of General Psychiatry (2006) 63: 18-24.
Oligodendrocyte Lineage Transcription Factor 2
(OLIG2): a master regulator of all stages of
oligodendrocyte lineage.

Basic helix–loop–helix
transcription factor central
to oligodendrocyte
development
 Down-regulated in
schizophrenia (Tchakev et al,
2003; Katsel et al, 2005; Iwamoto
et al, 2005)

Centrality in OL allows for
a primary change
responsible for several
others (parsimony)
OLIG2 associated with schizophrenia: non
redundant (r2<0.9) positives (submitted).
Meff = 9 (all 16 pooled SNPs)
Experiment-wide corrections
p = 0.013 (primary 14 genes)
p = 0.038 (all 44 genes)
gene corrected p = 0.0009
CONSERVATIVE
Conclusions





Some highly promising findings (NRG1, DTNBP1, G72/DAOA,
DISC1)
Need to establish risk nucleotides/mechanisms
» And this might not be easy without functional readout
(endophenotypic, animal, cellular)
» Needs more detailed studies, collaboration, re-sequencing
» Expect allelic heterogeneity, effects of ascertainment
Refining the associated phenotype by iteration (Symptoms, Course
and outcome, Cognitive function, Imaging). NB samples.
Curation of association data (it’s epidemiology)
– Quality standards
– Meta-analysis
Need to support genetic associations with biology
– DTNBP1 and NRG1 expression studies
Future

There are more significant linkages to account for.
– Few if any exhaustive fine mapping studies
 WGAs
– Depend on CDCV hypothesis
– Need very large samples
– Sample characteristics will be crucial
– Data handling and statistical challenges are huge
– WTCCC





CNV analysis
Candidate pathway analysis (GxG)
Inclusion of E in genetic studies (GxE)
WGS (CDRV)
Understanding gene regulation (identifying regulatory sequences,
miRNA, chromatin effects etc)
Acknowledgements













N Williams
N Norton
H Williams
N Bray
A Preece
J Wilkinson
S Dwyer
Elaine Green
Rachel Raybould
Detelina Grozeva
T Peirce
B Glaser
L Carroll





M O’Donovan
N Craddock
G Kirov
I Jones
M Hamshere, P Holmans. S
Macgregor, V Moskvina,