Transcript One of the first COMT fMRI studies

Combining functional genetics and fMRI
The ‘candidate gene’ ‘intermediate phenotype’ approach
-What it basically comes down to:-
-Looking at the relationship between a particular polymorphism
in a gene and brain activity during performance of a given task.
-Why do it?
- An interesting approach to studying individual differences in
neural and cognitive function.
- In particular, some common genetic polymorphisms lead to 3-5
fold changes in the expression or activity of proteins involved in
neurotransmitter metabolism or uptake – studying the influence
of these polymorphisms can provide a convergent approach to the
use of pharmacological manipulations to understanding the
impact of these neurotransmitters upon brain function.
• Background:-20,000 genes + play a role in the development, plasticity &
maintenance of the CNS.
• 6 million + polymorphisms contribute to genetic variability
between individuals
• One estimate - approx 5 polymorphisms per gene are
‘functional’ – resulting in change in expression or behaviour of
proteins
• ‘Alleles’ – variants of genes resulting from polymorphisms
• Traditional approaches:• Allelic association studies
• These look to see whether certain groups (e.g. individuals with a
given psychiatric diagnosis) have an increased frequency of a
particular allele compared to controls.
• There seem to be a lot of failures to replicate
• This has recently led Weinberger and colleagues to propose the
‘candidate gene’/ ‘intermediate phenotype’ approach;
- premise – there is a more direct relationship between genetic
variation and brain function than between genetic variation and
psychiatric diagnosis.
Principles for (successfully) conducting a candidate gene/
intermediate phenotype study
1) Chose your genetic polymorphism of interest with care
- is it functional?
- is there a reason to believe it may impact upon activity in the
region of interest / affect the cognitive function of interest?
2) Chose your imaging task with care
- Is there evidence for the cognitive process being tapped
being heritable?
- Does the task have high reliability?
- Does it reliably (across individuals) activate the region of
interest?
Are there individual differences in task performance and/or
extent of neural activation?
Types of polymorphism:VNTRs Variable Number of Tandem Repeats (a sequence of x
nucleotide bases is repeated a variable number of times)
If these occur in the promotor region these can influence the
expression (i.e. alter the amount of mRNA produced)
A good example is the 5HTT-LPR : a VNTR polymorphism in
the promotor region of the serotonin transporter gene
People either have 2 short alleles (20%), one short and one long
(50%) or two long alleles (30%)
The l allele is linked to higher concentration of 5-HTT mRNA
and greater 5-HT reuptake (as shown in cultured human
lymphoblast cell lines).
In vivo SPECT imaging also shows differences in 5HTT binding
levels.
The S allele appears to be dominant
fMRI findings:There are now a number of studies suggesting that individuals
with one or more 5HTT-LPR short alleles show an increased
amygdala response to threat-related stimuli (e.g Hariri 2002)
SNPs – Single nucleotide polymorphisms
If these occur in the coding region of the gene they can lead to
changes that alter the function of the protein
A good example is the COMTval158met polymorphism
Here, a G to A mutation causes a substitution of Val by Met.
(GTG to ATG result s in Val to Met). The Met allele is unstable
and has1/4 of the activity of the Val allele.
A regionally selective effect of COMT in PFC may result from
dopamine transporters being expressed in low abundance and
not within synapses
People either have 2 val alleles, one val and one met, or two met
alleles
The alleles are codominant, heterozygous individuals having
enzyme activity that is midway between homozygous
individuals
fMRI findings:There are now a number of studies suggesting that the number of
val alleles possessed is positively correlated with the extent of
PFC activation observed during working memory and other
prefrontal tasks (e.g. Egan et al. 2001, Bishop et al. 2003)
Other common polymorphisms
DAT (dopamine transporter)
9 vs 10 VNTR in an untranslated region of the gene
It is not clear whether this is functional or not.
However, Spect imaging has shown 9/10 repeat heterozygous
individuals have reduced DAT availability compared to 10/10
homozygous individuals (Heinz et al. 2000)
This could be due to
altered transcription or
translation.
Or it could be due to
linkage disequilibrium
with another polymorphism accounting
for the difference
between groups.
Other common polymorphisms
ApoE Є2, Є3, Є4 alleles
Є4 allele associated with increased risk for Alzheimer’s Disease
ApoE has a stimulatory effect on neurite outgrowth promoting
structural changes in response to brain injury and
neurodegeneration.
This is held to be reduced in the case of Є4
Figures taken from Raber et al. 2004
Polymorphisms not necessarily associated with variation in function in
the general population …
FOX P2
A rare mutation has been associated with severe speech & language
impairment (in the KE family)
A number of more common polymorphisms in the Fox P2 gene have
been identified.
However, so far none of them have been found to account for
variability in performance on tests of speech or language
The functionality of most of these polymorphisms has not been
established
Issues
•Interpreting the data with/ without behavioural findings
•It is often hard to get effects of genotype upon cognitive
performance without sample sizes in the 100s. But null behavioural
results cause difficulties in interpreting fMRI findings…
•The Efficiency Hypothesis
• In a number of studies on COMT, Weinberger and colleagues
have reported an effect of genotype at the brain level either with no
effect on task performance (e.g Egan et al 2000) or an effect in the
opposite direction (increasing neural activity associated with poorer
performance) .
•They have argued that this is because the COMT met allele is
associated with enhanced neural efficiency (equivalent or enhanced
performance together with less PFC activity).
•However, not all studies find results that fit with the neural
efficiency hypothesis (e.g. Fan et al 2003)
Issues
•Interpreting the data:- U shapes, dominance, interactions ….
• We also have to contend with possible nonlinearities in the
influence of our genetic variable upon neural or cognitive function.
•For example, it has been argued that COMT activity (via its
impact on PFC DA levels) may have a U shaped relationship with
PFC function. So, even though the number of COMT val 158
alleles possessed may have a linear effect upon COMT activity this
may not carry through to the neural or cognitive level.
• While the COMT met & val alleles are thought to be co-dominant
(heterozygous individuals having intermediate levels of COMT
activity) with other polymorphisms there is the issue of dominance.
Issues
•Dominance - with the 5HTT-LPR polymorphism, the s allele
appears to be dominant with cell line and SPECT studies giving
similar finding for s/l and s/s cases vs l/l cases.
• This shows how it is important to know about dominance for
analysing your data – here we might want to compare l/l/
individuals against a group of l/s and s/s individuals.
•However, to complicate things further still …when susceptibility
to depression is considered, there is some suggestion that this
dominance effect may not be so clear with l/s individuals
showing intermediate patterns for some measures of risk.
• Interactions – we may wish to consider whether two or more
polymorphisms of interest interact or have additive effects (e.g.
different polymorphisms impacting on DA metabolism/ uptake)
Issues
•Multiple comparisons (avoiding false positives)
• Either in a single study, or across studies, association studies
typically end up looking at the association between many
different polymorphisms and the phenotype of interest (e.g.
diagnosis of Schizophrenia).
•As a result, the case is often made for doing multiple
comparisons corrections for all the polymorphisms studied OR
even all the polymorphisms that could conceivably be examined.
• While proponents of the candidate gene / intermediate
phenotype model do not suggest applying the latter approach to
corrections in the case of genomic imaging studies, not all
reviewers may be of the same opinion …
Issues
• Linkage disequilibrium
This refers to a correlation between a specific allele at one locus
and another allele at a second locus. It may mean that any effect
observed is not actually due to the polymorphism under
consideration. This is particularly problematic when there is no
evidence for the functionality of the polymorphism being
investigated.
•Functionality issues
•The above seems to suggest sticking to polymorphisms of
known functionality …
•However it is not necessarily clear that we understand all the
ways in which polymorphisms may influence protein expression
or activity, and many functionality studies still remain to be done.
Practical Issues
• Genotyping prior to or post the imaging study:•Prior :pros - best for matching samples
cons - large numbers of volunteers need to be screened to
nnnnn get equal groups when one of the alleles is rarer than the
ccccccother
• arrangements for genotyping – attempts to set it up in
aCambridge via MRC Epidemiology and the BCNC
• Ethnic stratification: Certain alleles may be more or less
common in different ethnic groups. Hence, if your genotype
groups are not ethnically equivalent, apparent genetic effects
could be due to differences linked to ethnicity for other reasons.
Differences in genotype groups in IQ, gender, age etc may also
make it hard to detect genetic effects.
What running a study entails …
• Recruiting a lot of potential volunteers
•Screening them for one or more polymorphisms by taking a
cheek swab or a blood sample
• Getting these analysed (may take time …..)
• Deciding how you are going to split your group according to the
polymorphism of interest – issues if dominance
(e.g. for the 5HTT polymorphism you may want to look at l/l
individuals vs l/s and s/s individuals; while for COMT you may
want to do a correlation against the number of val alleles
possessed)
• Scanning an appropriate (?) number of individuals with each
allele type.
• Conducting a standard SPM analysis with genotype entered as a
covariate at the random effects level (or modelling 2nd order
terms/ non linear effects)
If you’re still interested …
further reading
• Hariri AR, Weinberger DR. Imaging genomics. Br Med Bull. 2003;65:25970. A good basic review of the methods
• Egan MF, et al. Effect of COMT Val108/158 Met genotype on frontal lobe
function and risk for schizophrenia. Proc Natl Acad Sci U S A. 2001 Jun
5;98(12):6917-22. Epub 2001 May 29. One of the first COMT fMRI studies
• Bilder RM, Volavka J, Lachman HM, Grace AA. The catechol-Omethyltransferase polymorphism: relations to the tonic-phasic dopamine
hypothesis and neuropsychiatric phenotypes. Neuropsychopharmacology. 2004
Nov;29(11):1943-61. Review. A more complex take on the COMT story
• Hariri AR, Mattay VS, Tessitore A, Kolachana B, Fera F, Goldman D, Egan
MF, Weinberger DR. Serotonin transporter genetic variation and the response
of the human amygdala. Science. 2002 The first 5HTT fMRI paper
•Raber J, Huang Y, Ashford JW. ApoE genotype accounts for the vast majority
of AD risk and AD pathology. Neurobiol Aging. 2004 May-Jun;25(5):641-50.
Review. A paper reviewing the evidence for the link between ApoE and risk for
Alzheimers