1. Strategies for investigating the genetics of psychiatric disorders
Margit Burmeister, Ph.D.
Molecular & Behavioral Neuroscience Institute (formerly Mental Health Research Institute)
Department of Psychiatry
Department of Human Genetics
Neuroscience Program
Bioinformatics Program
University of Michigan, USA

2. Overview Heritability Terminology Linkage analysis
Association studies – candidate genes
Whole genome association studies
Phenotypes
Gene x environment interaction

3. Familiality is not heritability
Bipolar Disorder, Depression, Schizophrenia and Alcoholism run in families
But: Going to Medical School, being Catholic and speaking Mandarin also run in families.
Heritable due to genes?
Familial due to environment?
Adoption: risk of child more influenced by biological parents (genetic) or adoptive (environment) parents?
Twins: Monozygotic (MZ, identical) twins share 100% of genes, dizygotic (DZ, fraternal) twins share 50% of genes.

4. Twin concordance rates
(From: Plomin et al. 1994, Science 264: )

5. Locus (plural: Loci) polymorphic Allele
From Latin: Place
The position on a chromosome
of a gene or a genetic marker or a SNP
A locus that comes in several forms is called
polymorphic
Allele
one of several forms of a locus
We inherit one allele from each parent

6. Genetic marker
DNA segment with known location on a chromosome whose inheritance can be followed
DNA segments near each other on a chromosome tend to be inherited together
so markers are an indirect way to track inheritance of an unknown gene
Eye color and bristle length are used as easily recognizable genetic marker in Drosophila crosses
Molecular markers are now used in human genetics
Restriction fragment length polymorphisms (RFLPs).
minisatellites, often also called VNTR markers
microsatellite or STRP markers, e.g. CA/GT repeats
SNPs: Single nucleotide polymorphism

7. Linkage analysis
Hypothetical marker with alleles A, B, C and D is is linked to the disease.
In the first pedigree, allele A segregates with the disease, but in the second pedigree, allele B segregates with the disease. The indicated recombination event leads to allele C segregating with the disorder in the right branch.

8. What is linkage?
Linkage is caused by loci (e.g. the risk gene and a genetic marker) being close to each other on a chromosome.
The recombination fraction θ is the probability that, in any meiosis, there will be a recombination between them.
θ=0.5 Mendelian segregation – no linkage
θ=0.0 very tight linkage – no recombination.

9. How sure are we of linkage?
LOD score! Logarithm of Odds
(Probability of a particular family constellation
if the marker and disease locus are linked, with 10% recombination between them)
LOD (q=0.1) = log
(Probability of the same family constellation of markers and disease if the marker is unlinked to the disease locus)

10. When is linkage significant? How many families are needed?
LOD score of odds of 1000:1 are equivalent to p= 0.05
With pedigree at hand, the best possible and a likely LOD score can be calculated.
For a Mendelian disease, only 10 informative meioses are needed to get a LOD score of 3.03 – single families can be enough to find linkage

11. Family with recessive ataxia: SCASI+* + * * * * * * * * * * * *
In this single family with a rare recessive disorder, we found linkage to 1p36 with a LOD score of Unaffected; Affected; * DNA available

12. Complex Disorders
Psychiatric disorders are complex - unlike Mendelian disorders, where a single mutation is both necessary and sufficient to bring about the disorder. Complexity can be due to
genetic heterogeneity: many different risk genes
small effect of each genetic factor: risk allele only increases risk by a small fraction
diagnostic uncertainty: is a depressed or alcoholic subject in a Bipolar pedigree affected or not? What about a BP II subject?
interaction with the environment

13. LOD score for complex traits – in multipoint analysis
(Probability of this particular data constellation
if a locus that increases risk for the disorder is in a particular chromosomal region)
LOD = log
if there is no locus that increases disease risk
in this particular chromosomal region)

14. COGA Genome Linkage Scan (Max. Drinks)
Saccone et al. (2000) Am J Med Genet 96:632-7.

15. Summary linkage/LOD score
looks at chromosomal location, with the help of informative genetic markers
powerful for simple Mendelian inheritance – no need for large samples
assumes you know the genetic model
no assumptions about the biology involved
Assumes one or just a few genes are involved
Families easily combined – add LOD scores
More powerful if less heterogeneous subtypes can be defined (e.g. Bipolar Disorder with psychosis, antisocial alcoholism; early onset MDD)

16. How do I follow up my positive linkage finding that showed 3 great POSITIONAL candidate genes in the region linked to the disease?
How do I test whether a new BIOLOGICAL candidate gene is involved in alcoholism?

17. Association Cases with Alcoholism Control sample ALDH2: 1/1: 283 (83%)
1/2: 57 (17%)
2/2: 0
ALDH2: 1/1: 304 (56%)
1/2: 218 (40%)
2/2: 23 ( 4%)
ALDH2*2 allele significantly protects against alcoholism
in this sample from China (Chen et al., Am J Hum Genet. 1999, ).

18. IN RESPONSE TO DRINKING ALCOHOL
FLUSHING SYMPTOMS
IN RESPONSE TO DRINKING ALCOHOL
Blood Flow to Face
Blood Flow to Body
Feel Itchy
Feel Dizzy
Feel Tired
Feel Anxious
Pounding Head
Have Sweats
Increased Heart Rate
Feel Nauseous
These symptoms are experienced by those heterozygote for
the ALDH2*2 allele, and are severe in homozygotes

19. Pharmacogenetic association
Within a disease population, pharmacogenetic studies ask about drug side effects and/or response:
drop outs versus drug continuation
“responders” versus non-responders
Greer and Schatzberg, 2003
Am J Psychiatry 160:
Zanardi et al., 2001
Biol. Psych. 50:

20. Confounding of association by ethnicity (stratification)
cases
controls
Example:
alcoholics in San Francisco.
Asians (stripes) will be underrepresented among cases.
ANY genetic marker more common among Asians will also
be overrepresented in the controls – e.g. HLA.

21. Linkage disequilibrium (LD)
Linkage disequilibrium shows the history of the mutation.
starting chromosomes that exist in population at one point
A,a B,b C,c and D,d are different alleles at four different nearby loci:
A b c D
A b C d
a b C d
New mutation arises on one particular chromosome:
a B C d

22. Haplotype Combination of specific SNPs on chromosome I: A,c,D
II: A,C,d
A C d b
III: a,C,d
a C d b
a B C d
We might find association of a disorder with haplotype III
because of linkage disequilibrium with the (untyped) allele B!

23. Applying the data: There are >10 Million SNPs in the human genome.
Because of LD (linkage disequilibrium), typing 300, ,000 SNPs may give us some information about all the >10 Million SNPs
Single SNPs as well as haplotypes can be tested for association with disease

24. COGA Genome Linkage Scan (Max. Drinks)
Saccone et al. (2000) Am J Med Genet 96:632-7.

25. cM/Mb bp
SNPs
haplotypes
Drawn by Jeffrey Long from Edenberg HJ, et al. (2004) Am J Hum Genet 74:705-14

26. Linkage - versus - Association
Needs families (e.g. sibpairs) in which the disease segregates
powerful when there are just a few predisposing genes
Works well even when many different mutations or alleles in those few genes increase risk for the disease
Requires no hypothesis about the nature of the defect in the disorder
usually requires a model of the pattern of inheritance
Needs unrelated cases and unrelated controls
Assumes that only one or few common variants in each gene are risk factors
Powerful even when there are many different genes, each with a small effect, as long as the variant in each gene is common
Until recently, required having a limited number of genes in mind

27. 2005 and beyond: Whole Genome Association Studies
cases
controls
Extract DNA – hybridize to whole genome chip
Construct haplotypes
Calculate which of ,000 SNPs and ??? haplotypes is more frequent in cases than controls
Currently: $1,000 per individual – planned: $100

29. Evidence from Twin Studies for involvement of genes in depression
Risk of Major Depression Onset by Genotype and Severe Life Event 5 10 15 20
MZ Co-Twin Aff
MZ Co-twin Unaff
DZ Co-Twin Aff
DZ Co-twin Unaff
Probability of Onset of Major Depression
Redrawn from
Kendler et al., 1995
Am. J. Psychiatry
152: 833
Absence
Presence
of severe life event

30. Genes, Stress and Depression
Genetic Vulnerabilities
(Personality, Physiology, …?)
Early experiences
(stress, rearing,..?)
Vulnerable Phenotype
Exercise,
antidepressants
psychotherapy
Daily stress
Traumatic event
Life event
Depression

31. High Neuroticism score is a risk factor for depression
Endophenotypes or intermediate traits mark the genetic predisposition to a disorder, less dependent on the additional external triggers or vulnerabilities.
Neuroticism is a psychological domain, measured with the established NEO-PI. Sample questions:
I often feel tense and jittery (anxiety facet)
Sometimes I feel completely worthless (depression)
When I’m under a great deal of stress, sometimes I feel like I’m going to pieces (vulnerability)
Twin and high risk studies suggest that a high Neuroticism score is a trait marker for depression
At least in women, about 55% of the genetic liability to depression is shared with Neuroticism

32. Serotonin Transporter Promoter Variant 5-HTTLPR
5-HTT target of SSRIs
S-short and L- long form of promoter:
The L form has higher promoter activity
Both are quite common in Caucasians

33. 5-HTTLPR is associated with Neuroticism
Total variance
explained: 1.8%
From: Sen et al. 2004, Biol Psychiatry. 55:244-9.

34. Genes, Stress and Depression
Genetic Vulnerabilities
(Personality, Physiology, …?)
Early experiences
(stress, rearing,..?)
Vulnerable Phenotype
Exercise,
antidepressants
psychotherapy
Daily stress
Traumatic event
Life event
Depression

35. Is there an effect of childhood maltreatment on depression risk?
Caspi et al., Science 2003

36. Genotype x life events interact with respect to 5-HTTLPR genotype and depression (Caspi et al., 2003: Science 301:386-9)
symptoms
MDD episode
suicidality
MDD per informant

37. Conclusions
Twin and adoption studies suggest genes are involved in nearly all psychiatric disorders
Psychiatric disorders are complex
Linkage studies are a powerful means to find where genes are located
Subtypes of psychiatric disorders that decrease heterogeneity will increase power of linkage analysis
Association studies can help identify common genetic risk factors even with very small effect size
Association studies are now feasible for the whole genome, not just candidate genes
Research into intermediate traits/endophenotypes can help identify genetic risk variants
Genes and environment interact