Behavior Genetics, Vol. 36, No. 1, January 2006 (Ó 2006) DOI: 10.1007/s10519-005-9000-4

New Approaches to the Genetic Analysis of Neuroticism and Anxiety Jan Fullerton1,2 Received 3 Feb. 2005—Final 23 Jun. 2005

The completion of the human genome project and the complementary genome projects for other species has broadened the scope for novel bioinformatic approaches to quantitative trait locus (QTL) identification. A key issue for quantitative trait nucleotide (QTN) identification is progressing from a large QTL peak, spanning perhaps 50 cM and many hundreds of genes, to a gene or nucleotide variant which is responsible for that QTL effect. The complementary use of mouse models to dissect large syntenic loci in humans is a powerful method for reducing QTL intervals to the order of 1 Mb. This paper presents an overview of the approaches used in our laboratory to ultra-fine map QTLs for anxiety-related traits, and to identify quantitative trait genes (QTG). As new genetic techniques and statistical approaches arise, we are getting closer to identifying those long sought after QTNs. KEY WORDS: Anxiety; behaviour; human; mapping; mouse; neuroticism; outbred; personality; quantitative trait; quantitative trait gene.

neuroticism, anxiety and depression are inter-related phenotypes, and that successful identification of the underlying genetic variants contributing to any one of these phenotypes may have implications for understanding the biology of personality and two of the most common psychiatric illnesses. The heritability of neuroticism is similar to other quantitative phenotypes that have been subject to genetic linkage analysis—additive effects account for around 30% and non-additive effects around 15% of the variance (Lake et al., 2000)—suggesting that a family based method for gene detection in neuroticism might be worth undertaking. However, such approaches have so far had limited success in both psychiatric genetics and for other complex traits. This is thought to be because the phenotypic effect attributable to individual loci influencing variation in most complex phenotypes is small, and hence very large sample sizes will be required for any genome analysis to be effective. Fortunately it is possible to collect very large samples for the analysis of neuroticism: self-report questionnaires can be used which are reliable and

MAPPING PERSONALITY: LINKAGE ANALYSIS OF NEUROTICISM There is considerable evidence to suggest that the personality trait of neuroticism is heritable: results from large scale twin and family studies indicate that about 40% of the variance is attributable to genetic variation (Lake et al., 2000). Neuroticism manifests at one extreme as anxiety, depression, moodiness, low self-esteem, and diffidence (Cloninger, 1994; Digman, 1990; Dreary and Matthews, 1993; Loehlin and Nichols, 1976; Zuckerman et al., 1988), and genetic susceptibility to neuroticism is shared with two other common psychiatric disorders, major depression and generalised anxiety disorder (Duggan et al., 1990, 1995; Hettema et al., 2004; Hirschfeld et al., 1989; Kendler et al., 2004; Roy, 1990; Zeiss and Lewinsohn, 1988). These phenotypic overlaps suggest that 1

2

The Wellcome Trust Centre for Human Genetics, The University of Oxford, Roosevelt Drive, OX37BN, UK. To whom correspondence should be addressed at The Wellcome Trust Centre for Human Genetics, The University of Oxford, Roosevelt Drive, OX37BN, UK. e-mail: [email protected]

147 0001-8244/06/0100-0147/0 Ó 2006 Springer Science+Business Media, Inc.

148

Fullerton 4.74 at 105 cM), and 13 (logP 3.81 at 64 cM), which exceeded our empirically derived 5% significance level of logP 3.8. The linkage analysis also revealed evidence for sex specific loci. While heritability of N is no different in males and females, the sibling correlation of N score in opposite sex pairs is less than that for same sex pairs (0.157 versus 0.185) suggesting that loci contributing to neuroticism is different in males and females. To test this, we separated our sibling-pairs into two sex specific groups, female–female pairs and male–male pairs, and repeated our genome-wide IBD regression analysis, the results of which are shown in Fig. 1. Our analysis showed a difference in the genetic architecture of N in males and females, such that some loci appear almost entirely sex specific, for example chromosomes 1 and 13 in females, and chromosomes 7 and 8 in males. Additionally in the single sex analyses we detected two loci on chromosome 1; while this had been apparent in the joint analysis, it was accentuated when males and females were analysed separately. Since these loci do not occur on the X chromosome, the mechanism that determines sex specification is unclear; at one level, we can explain the effect as an example of gene by environment interaction: the expression of some genes will vary according to their hormonal, cellular or anatomical environment. However, the molecular basis of sex-specific action remains unknown, possibly reflecting epigenetic modification of DNA. Our study is only one of a number of attempts to map neuroticism or anxiety-related traits. Encouragingly, there seems to be some consistency in the

cheap to administer. We sent the Eysenck personality questionnaire (EPQ) by mail to over 200,000 residents of South-West England (Eysenck and Eysenck, 1975). Of more than 88,000 responses, we ascertained 34,580 sibpairs in 20,427 independent sibships (Martin et al., 2000), and collected DNA by mail-out via buccal mouth swab. Test–retest correlations 18 months after the first EPQ questionnaire was completed (n=1618) was 0.932 for N, showing that neuroticism is a relatively robust phenotype with consistent measurement (Martin et al., 2000). In order to obtain a maximally informative sample for genetic linkage analysis we selected the sibling pairs through an extreme selection strategy (Risch and Merikangas, 1996; Risch and Zhang, 1995). Three groups were identified, those where siblings both scored concordantly high or low on the neuroticism scale (N) were put in the high and low concordant groups (174 and 205 pairs respectively), and those whose high ranking scores were discordant (one sib scoring extremely high and the other scoring extremely low) were put in the discordant group (182 pairs). Linkage mapping of neuroticism was performed using a regression based method (Fullerton et al., 2003). First, genome-wide identity-by-descent (IBD) probabilities were calculated for each sibling-pair. The squared differences and sums of the sibling-pair phenotypes were combined and regressed onto the IBDs as proposed by Visscher and Hopper (2001). Linkage analysis, expressed as the negative logarithm of the probability of linkage, revealed peaks on chromosomes 1 (logP 3.95 at 126 cM), 4 (logP 3.84 at 176 cM), 7 (logP 3.90 at 42 cM), 12 (logP 5 1

2

3

4

5

6

7

8

9

10

11

12 13

14 15 16 17

18 19 20 2122

4

-logP

3

2

1

0 0

500

1000

1500

2000

2500

3000

Distance (cM)

Fig. 1. Genome-wide linkage analysis for individual variation in neuroticism in female–female (red line) and male–male (black line) sibling pairs. The )logP values (vertical axis) for the Visscher–Hopper regression are shown. The cumulative centimorgan distance across the genome is displayed at the bottom, and chromosome numbers are given at the top.

New Approaches to the Genetic Analysis location of some quantitative trait locus (QTLs) detected, as evidenced in Table I. Of particular interest, there are multiple linkages to chromosome 1, providing consistent evidence for at least one neuroticism-related locus on this chromosome. This finding is important because of the overlap with studies in rodents that is described later. Since the publication of the human linkage data in 2003, we have collected 40 additional extreme scoring families for sibling-pair analysis, and a cohort of 1000 unrelated individuals for association testing. We have also added additional microsatellite markers around our major linkage peaks to refine our QTL intervals; however the 95% confidence intervals of QTLs we are mapping still encompass over 40 cM and many hundreds of genes. USING MOUSE MODELS TO IDENTIFY HUMAN CANDIDATE GENES Linkage mapping is only the first stage in a long and currently very arduous path towards gene identification. In order to accelerate the speed of gene discovery, we have taken advantage of the relationship between human neuroticism and its rodent equivalent, sometimes called emotionality. While anthropomorphic validation of animal models for human disease will always be debated (Flint, 2004; Gray, 1982, 1987; Gray and McNaughton, 2000), from a genetic perspective the use of animal models for their relative genetic simplicity and applicability to breeding manipulation over many generations makes them a powerful resource for dissecting QTLs, and a key step in progressing from a large QTL interval to the ever elusive quantitative trait nucleotide (QTN). While the gene responsible for the particular QTL effects detected in rodent studies may not be identical to those detected in the human studies, it is likely that some anxiety-related genes will be identical across species, because anxiety is an evolutionarily ancient behavioural trait which is common to all vertebrates. Since mice and humans share many orthologous genes which map to blocks of chromosomal synteny, the use of high resolution mouse mapping studies should more rapidly identify candidate genes for human behavioural traits. Intriguingly, as I will show, mapping studies for anxiety-related phenotypes in rodents have also implicated regions syntenic with human chromosome 1 (Caldarone et al., 1997; Fernandez-Teruel et al., 2002; Flint et al., 1995; Gershenfeld et al., 1997),

149 presenting provocative evidence for an evolutionarily conserved anxiety-related gene on this chromosome. Hence there is a good rationale for fine mapping anxiety QTLs in mice for regions syntenic with human chromosome 1. Crosses between inbred strains of laboratory rodents have been successful in detecting loci for quantitative traits for behaviour as they are capable of detecting small genetic effects. The drawback of the method is its poor resolution. Our laboratory started mapping QTLs for emotionality using an F2 intercross between 2 inbred mouse strains, derived from animals selected for high and low open field activity (Flint et al., 1995). The most extreme scoring 10% from 879 F2 progeny were genotyped, and QTLs were detected on mouse chromosomes 1 (LOD 13.4), 4 (LOD 6.2), 12 (LOD 4.3) and 15 (LOD 11.0). However the 95% confidence intervals for these QTLs spanned at least 40 cM (Fig. 2), containing many hundreds of positional candidate genes. A similar study in rats, using a cross between the Roman high and low avoidance strains, detected a locus on rat chromosome 5 that contributed to variation in a number of measures of emotionality (Fernandez-Teruel et al., 2002). This locus is not homologous to any loci yet found in the mouse, but given the limited diversity sampled for these inbred mapping experiments (just two chromosomes in each case) it is not surprising that there was no overlap between the results of the two experiments. However, there was overlap between the rodent and human experiments: the locus on rat chromosome 5 and the locus on mouse chromosome 1 coincide with the two loci detected on chromosome 1 in humans. In order to increase mapping resolution in the study of mouse QTLs, and thus get closer to identifying genes for analysis in the human sample, we decided to map emotionality QTLs in genetically heterogeneous mice (McClearn et al., 1970). The heterogeneous stock we used was derived from an eight way cross of A, AKR, BALB/c, C3H, C57BL/6, DBA, I and RIII inbred strains. At the time of our experiment, the HS had been maintained for 60 generations from a breeding population of 40 mating pairs (to reduce potential for allele fixation). Consequently enough recombinants were present to provide a very high degree of mapping resolution: HS chromosomes are a fine-grained genetic mosaic of the founder strains, with an average distance between recombinants of 1/60 or 1.7 cM. In our first mapping experiment with HS mice, 751 animals were phenotyped for emotionality in an

1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 3 3 4 4 6 6 7 7 7 7 8 8 10 10 10 11 11 11 11 11 12 12

Depression or alcoholism RMD/mood disorder Comorbid alcoholism and depression

MDD-RE or anxiety EPQshort—Neuroticism

MDD-RE or anxiety EPQ-Neuroticism EPQ-Neuroticism

EPQshort—Neuroticism

MDD-RE or anxiety EPQ-Neuroticism Depression TPQ—Novelty Seeking

Harm avoidance TPQ—Harm avoidance

TPQ–Novelty seeking Anxiety susceptibility Major depression

Panic disorder/agrophobia Recurrent major depression EPQ-Neuroticism TQP—Harm avoidance EPQshort—Neuroticism TPQ—Novelty seeking EPQshort—Neuroticism

Chrm

Mood disorder EPQ-Neuroticism G EPQ-Neuroticism G Simple phobia Depression or alcoholism EPQ-Neuroticism EPQshort—Neuroticism Anxiety susceptibility EPQ-Neuroticism Panic disorder/agoraphobia

Phenotype

Not given

Not given 163–173 34–63

100–113 Not given

Not given 201–216

18–40 37–55 43–70 42–90 57–89 92–120 100–150 113–150 133–143 188–229 181–226 243–268

Interval

5 2 99 114 132 15 45

95 148 149

23 60

102–137 15–24 39–51

0–11 0–20 94–103

77–100 Not given Not given

45–64

0.1 Not given 42 31–54 150 184 180–190

147

151 168 47

105 105

100 205 248

35 45 64 80 80 106 120 126 137 203 221 258

cM

112 pedigrees—611 affected Selected: 560 sibpairs 226 sibpairs 758 sibpairs

201 ASPs

20 pedigrees 81 families (375 sibpairs) Selected: 560 sibpairs 758 sibpairs 201 ASPs 758 sibpairs 201 ASPs

758 sibpairs Pedigree—29 affecteds 81 families (375 sibpairs)

QMFLINK GENEHUNTER GENEHUNTER GENIBD Visscher–Hopper SOLAR Merlin-regress QMFLINK Merlin-regress

Camp et al. (2005) Neale et al. (2005)

Nurnberger et al. (2001) Zubenko et al. (2002) Nurnberger et al. (2001)

Zubenko et al. (2003) Fullerton et al. (2003) Nash et al. (2004) Nash et al. (2004) Nash et al. (2004) Gelernter et al. (2003) Nurnberger et al. (2001) Fullerton et al. (2003) Neale et al. (2005) Smoller et al. (2001) Nash et al. (2004) Gelernter et al. (2001)

Study

Gelernter et al. (2001) Zubenko et al. (2003) Fullerton et al. (2003) Cloninger et al. (1998) Neale et al. (2005) Curtis (2004) Neale et al. (2005)

Curtis (2004) Smoller et al. (2001) Zubenko et al. (2003)

Cloninger et al. (1998) Dina et al. (2005)

Camp et al. (2005) Fullerton et al. (2003) Nurnberger et al. (2001) Curtis (2004)

Neale et al. (2005)

112 pedigrees—males only coded affected Camp et al. (2005) Selected: 560 sibpairs Fullerton et al. (2003) Selected: 601 pairs+86 male pairs Nash et al. (2004)

112 pedigrees—611 affected 201 ASPs

740 sibpairs 170 female pairs 107 sibpairs

2.3d* 2.4* 5.4** 2.0 4.2** 2.6a 1.6 1.96a 3.1d** 2.85a

Sample 81 families (610 sibpairs) 249 Female sibpairs Selected: 601 sibpairs Selected: 601 sibpairs 254 female pairs 14 families (incl 57 affected) 740 sibpairs Selected: 560 sibpairs 201 ASPs Pedigree—29 affecteds 254 female pairs Families, 153 individuals

SOLAR 758 sibpairs Maximum likelihood binomial 377 sibpairs

MCLINK Visscher–Hopper ASPEX QMFLINK

Merlin-regress

MCLINK Visscher–Hopper Merlin-regress

MCLINK Merlin-regress

GENIBD ASPEX

GENIBD Visscher–Hopper Merlin-regress Merlin-regress Merlin-regress GENEHUNTER/ALLEGRO ASPEX Visscher–Hopper Merlin-regress GENEHUNTER Merlin-regress GENEHUNTER

Analysis method

3.2** 2.7**

2.9* 3.9a 2.9* 2.1d*

2.6 2.8a** 2.7* +3.9** 1.95a

3.8* 1.95a

3.3 6.9** 2.2*

3.6** 3.3a** 2.2* 1.6 2.6* 1.9c 4.6** 2.8a 2.52a* 2.1b 1.5 2.0

LOD

Table I. Linkage Analysis of Psychiatric Disorders and Traits Genetically Related to Neuroticism

150 Fullerton

151 Linkage results are shown for each phenotype, with chromosome number (chrm), peak QTL position in centimorgans (cM), QTL peak interval in centimorgans (interval), analysis method used to determine linkage, and the sample size used for each study. Linkages are described as LOD score unless otherwise indicated: alogP, bNPL, cZlr, dMALOD. The significance of the linkage result is indicated as *suggestive or **significant. Phenotypes determined by either the Eysenck- or tri-personality questionnaire are indicated as EPQ or TPQ respectively. The G phenotype represents as a composite index for liability to depression and anxiety.

Camp et al. (2005) Cloninger et al. (1998) 21 pedigrees—96 affected 758 sibpairs MCLINK SOLAR 3.75* 1.6 68–85 Not given 18 18 MDD-RE and anxiety Harm avoidance

73 109

Nash et al. (2004) Curtis (2004) Curtis (2004) 86 male pairs 758 sibpairs 758 sibpairs Merlin-regress QMFLINK QMFLINK 2.1* 3.4d** 1.7d 0–13 50–66 50–55 17 17 17 G TPQ—Novelty seeking TPQ—Reward dependence

6 58 52

Gelernter et al. (2004) 17 families (incl 56 affected) Allegro 2.2* 40–94 16 Social phobia

71

12 12 12 12 13 14 15 15 Panic/agoraphobia G Major depression EPQ-Neuroticism EPQ-Neuroticism Simple phobia MDD-RE or anxiety Recurrent early onset depression

66 90 97 116 64 36 98 103

60–70 81–102 85–104 101–132 43–76 8–45 Not given 99–110

4.9a** 1.8* 6.1** 4.7a** 3.8a** 3.2** 2.9 3.7**

GENEHUNTER Merlin-regress MCLINK Visscher–Hopper Visscher–Hopper GENEHUNTER/ALLEGRO MCLINK Allegro

Pedigree—29 affecteds 254 female pairs 32 pedigrees, affected males Selected: 560 sibpairs Selected: 560 sibpairs 14 pedigrees (incl 57 affected) 112 pedigrees—males only coded affected 685 informative relative pairs

Smoller et al. (2001) Nash et al. (2004) Abkevich et al. (2003) Fullerton et al. (2003) Fullerton et al. (2003) Gelernter et al. (2003) Camp et al. (2005) Holmans et al. (2004)

New Approaches to the Genetic Analysis

open field arena. Markers spaced at approximately 1 cM intervals across the 95% confidence interval of the QTL on chromosome 1 were used for mapping (Caldarone et al., 1997; Flint et al., 1995; Gershenfeld et al., 1997). Single marker analysis of variance (ANOVA) revealed a QTL peak spanning an interval of 0.8 cM (Talbot et al., 1999). When additional markers were used in an attempt to resolve the QTL into the smallest possible interval we found a surprisingly complex picture: physically adjacent markers varied considerably in the strength of association with the phenotype. In some cases this could range from completely nonsignificant to highly significant within an interval of only a few hundred kilobases (Fig. 3a, green line). Differences in the strength of association detected at tightly linked markers can be explained by the differences in the relationship between marker and QTL alleles. For example, Fig. 3b shows the founder strain of origin of marker alleles for the two diallelic markers that are almost identical in terms of position and heterozygosity, but which show contrasting association with the QTL alleles. Single marker association mapping is an inefficient method when the allele frequencies of the variant and the QTL do not coincide and when contrasting QTL alleles are in linkage disequilibrium with the same marker allele, as illustrated in Fig. 4. In this example the population from which all chromosomes descend consisted of four inbred strains. Inbred strains A and C contain a QTL allele that increases the phenotype, and strains B and D contain an allele that decreases the phenotype. A polymorphic marker, m, in complete linkage disequilibrium with the QTL possessing four distinct alleles (labeled 1,2,3,4 in Fig. 4) will show evidence of association. Consider what happens with a diallelic marker n that, in the ancestral population, has one allele 1 physically linked to strains A, B and C and the other allele 0 linked to strain D. Evidence of association will now be reduced since the marker fails to partition the genotypes into two equally contrasting groups. Things are even worse at markers p and q where both alleles 0,1 are linked to both increaser and decreaser QTL alleles; there is no detectable association at all. But notice that the haplotypes defined by the markers p+q distinguish all four strains and so are a surrogate for the efficient marker m. Therefore multipoint haplotype-based mapping is generally more powerful than singlepoint. Heterogeneous stocks (HS) are mosaics of eight inbred founder strains, and QTL mapping in these

152

Fullerton 16 14

LOD score

12 10 8 6 4 2 0 0

20

40

60

80

100

120

Distance (cM) Fig. 2. QTL analysis of open field activity on mouse chromosome 1. The LOD score (Y axis) was generated by the MAPMAPKER-QTL program, and plotted against marker position in centimorgans along chromosome 1.

stocks is best approached by reconstructing the ancestral haplotype mosaic. However, the strain distribution patterns of neighbouring markers rarely distinguish between all founder strains so instead we compute the probability that an animal is descended from a given pair of strains at a particular locus. A

useful statistical model of the genome mosaic of a diploid HS animal is a pair of hidden Markov models, in which the hidden states of the model are the ancestral strains, and the observed data the marker genotypes. A dynamic programming algorithm is then used to compute the probabilities of descent

Fig. 3. QTL analysis of emotionality (EMO) in HS animals on mouse chromosome 1. (a) Single marker (green) and HAPPY (red) analyses of mouse chromosome 1, with 1% significance threshold for HAPPY shown (blue). The P values for both analysis methods are shown on the yaxis, at marker positions along the chromosome in centimorgans (cM) on the x-axis. (b) Comparison of adjacent diallelic markers showing conflicting results for single marker association. Shown in the first three columns are the chromosomal position, significance of single marker ANOVA represented as a negative logarithm of the P value, base pair (bp) size of alleles detected in HS chromosomes. The founder strain origin of the two alleles for both markers is shown in the fourth column, as determined by genotyping DNA from the eight founder strain mice and grouping them by allele size.

New Approaches to the Genetic Analysis

QTL

153

m

n

p

q

1

1

1

0

2

1

1

1

3

1

0

1

4

0

0

0

(a) (b)

(c) (d) Fig. 4. The effect of allele distribution on the likelihood of detecting association. The figure shows a QTL with an increaser and a decreaser allele (indicated by up and down arrows), and a series of nearby markers m, n, p and q along a chromosome (blue line). Marker m will distinguish between the high and low QTL alleles, and therefore will be the most informative single marker for the QTL.

using all available information (Mott et al., 2000). Rather than reconstructing haplotypes from the available genotypes using pedigree information, as with traditional haplotype reconstruction programs, the origin of marker alleles in the current generation is inferred at each marker by comparison with progenitor strains, and the probabilities of progenitor strain origin calculated on the basis of the strain distribution pattern of adjacent markers. This method has been incorporated into a software package, HAPPY, which was used to compute the data presented in Fig. 3a (red line). The HAPPY software is freely available from http://www.well.ox.ac.uk/ rmott/HAPPY. Haplotype analysis in the HS animals showed that we were at the maximum resolving power of the HS across this QTL with the markers we had genotyped, and that no additional recombinants would be identified by genotyping additional markers at a higher density. As a consequence, further genetic mapping in the HS animals would not yield any additional positional information for the QTN. Physical mapping and sequencing of our 0.8 cM interval revealed our region spanned 4.78 Mb of DNA, and contained 17 pseudogenes and 9 genes, whose descriptions are given in Table II. Sequencing the coding and promoter regions of all genes in the 8 inbred strains, and a selection of 12 HS and 12 MF1 animals revealed 7 coding variants in three genes; Glrx2, SSA2 and RGS1, which led to conservative amino acid substitutions or were silent, but none of the variants could explain the detected

genetic effect based on the strain distribution pattern of the coding variants. So we were forced to further refine our QTL using an even higher resolution mapping strategy. Outbred and wild mouse populations offer a higher level of resolution for QTL mapping, as the populations are very old and recombination between loci has been allowed to accumulate over hundreds of generations. The problem of using outbred stocks for QTL analysis is that the progenitor strain origin of the stock is generally unknown and pedigree information is scarce. However, if we could determine the polymorphism patterns in an outbred mouse stock and relate them to inbred strain haplotypes, we could treat the descendants as if they were descended from the inbred strains, and use the ancestral haplotype reconstruction method described above (Mott et al., 2000). We sequenced 62 kb of coding and promoter regions for the 9 genes in the QTL in 12 MF1 animals, and found the variant patterns were almost identical to those found in the HS founders, suggesting that the MF1 mice are closely related to the eight inbred strains. To investigate this idea further, we genotyped 42 SNPs across the QTL in 729 MF1 animals and constructed haplotypes assuming they are unrelated firstly using PHASE2 to determine basic haplotype structure. We discovered that a relatively small number of haplotypes were represented in the MF1 population, and that 14 haplotypes represented 95% of the total chromosome complement in the MF1 (Yalcin et al., 2004). To test the idea that the MF1 could be derived from the eight inbred strains, we reconstructed the haplotypes as if they were descended from inbred strains using a dynamic programming algorithm, using the known haplotype structure for each inbred strain. Two features of importance emerged from this analysis, which is represented in Fig. 5. First, it can be seen that the haplotypes consist of long blocks of DNA derived from individual inbred strains, consistent with the presence of regions of unrecombined ancestral haplotype. If the MF1 were unrelated to the inbred strain progenitors, we would expect to see a random pattern in the strain of origin of each variant from this analysis; but in contrast to this expectation, we see blocks of haplotype consistent with the theory that the chromosomal segments may originate from individual inbred strains. The second point to emerge from this analysis is that just four of the possible eight strains can account for all the haplotypic variation in the MF1; C3H, AKR, C57BL/6J, and I. That

Regulates signalling through the G-protein coupled receptors Regulates signalling through the G-protein coupled receptors Regulates signalling through the G-protein coupled receptors Unknown—has 53% homology with BMP/retinoic acid inducible neural specific protein and deleted in bladder cancer

RGS2 (Regulator of G protein Signalling 2)

RGS13 (Regulator of G protein Signalling 13) RGS18 (Regulator of G protein Signalling 18)

B830045N13Rik

Bone, brain, colon, eye, heart, kidney, liver, lymph node, muscle, pancreas, pituitary gland, spleen, stomach, thymus, embryo, adult Bone marrow, colon, liver, lymph node, stomach, thymus, embryo, adult Bone, brain, colon, eye, kidney, liver, lung, lymph node, pituitary gland, spleen, thymus, embryo, adult Brain, lymph node, ovary, thymus, adult *Brain, bone, heart, liver, lymph node, spleen, stomach, embryo, adult Brain, colon, eye, heart, liver, lung, muscle, pancreas, pituitary gland, spleen, thymus, embryo, adult

Brain, colon, eye, heart, kidney, liver, lung, lymph node, muscle, spleen, thymus, embryo, adult Bone marrow, brain, eye, liver, lung, pancreas, pituitary gland, thymus, embryo, adult

Brain, eye, heart, lung, mammary gland, spleen, embryo, adult

Tissue expression

Data on gene location and function were obtained through the ensembl database, or where unavailable, from summaries of PubMed entries. Tissue expression data was largely obtained from the Unigene database, where ESTs from the listed genes were detected in various tissue expression libraries. In the interests of summarising the data, some libraries listed have been compressed e.g., libraries of different embryonic stages were grouped into a single category Ôembryo’. * Unigene describes RGS18 as not being expressed in the brain; however in situ hybridisation and RT-PCR data from our laboratory shows low level expression of RGS18 in the brain.

Uchl5 (ubiquitin thiolesterase L5)

SSA2 (SS-A/Ro nucleoprotein)

Glrx2 (glutaredoxin 2)

RGS1 (Regulator of G protein Signalling 1)

Function Tumor suppressor gene which inhibits cell proliferation. Implicated in sporadic parathyroid cancer and hyperparathyroidism-jaw tumor syndrome Protects against oxidative stress by catalyzing reduction of proteins with glutathione Central component of the Ro ribonucleoprotein complex, functions in a quality-control pathway for 5S rRNA biogenesis and in resistance to environmental stress Post-translational modification of proteins by covalent attachment of ubiquitin and consequent degradation by the proteosome Regulates signalling through the G-protein coupled receptors

Hrpt2 (parafibromin)

Gene

Table II. Description of Function and Tissue Expression of Known Genes in the QTL Region

154 Fullerton

New Approaches to the Genetic Analysis these haplotypes could be derived with minimal recombination compared to the number required for the random haplotype reconstruction, provides supporting evidence that the MF1 can accurately be treated as if derived from the inbred strain progenitors. With the knowledge of the underlying haplotypic structure of the MF1, we attempted to map QTLs to ultra-high resolution in these mice using probabilistic ancestral haplotype reconstruction. We applied the program HAPPY to search for QTLs using unphased genotypes. The results of this analysis dissected our single 0.8 cM or 4.78 Mb QTL detected in the HS animals into three independent QTL effects (Fig. 6). Interestingly, single marker association of the same genotypes used in the HAPPY analysis revealed three markers across the whole interval which reached suggestive significance (Fig. 6, red dots), compared to the highly significant peaks generated through the more powerful method of probabilistic ancestral reconstruction (Fig. 6, black line). Of the three QTLs detected, only the QTL at 0.7 Mb overlaps with known genes, both of which belong to the regulator of G-protein signalling family; firstly rgs2, which is expressed widely in the brain and other tissues, and has been shown previously to modify rodent behaviour and synaptic development in a knockout experiment (Oliveira-Dos-Santos et al., 2000), and secondly rgs13 which is not entirely contained within the 95% confidence interval for the QTL and is expressed at low levels in the brain. FUNCTIONAL COMPLEMENTATION TO IDENTIFY QUANTITATIVE TRAIT GENES (QTGs) We conducted a complementation experiment using an rgs2 knockout mutant to determine if QTNs affecting rgs2 were responsible for the QTL at

155 0.7 Mb. This involved performing breeding crosses between the rgs2 mutant line (obtained from OliveiraDos-Santos and colleagues) (Oliveira-Dos-Santos et al., 2000), and inbred strains where the QTL allele is derived from a high anxiety line (e.g. C57BL/6J) followed by a low anxiety line (e.g. A/J or C3H/HeJ). As a background test for inbred strain effects, we also conducted the crosses with a wild type strain of a similar isogenic background as the rgs2 mutant mouse. The results showed a significant interaction between the natural allelic variation (i.e. high and low EMO strains) and the presence or absence of the functional rgs2 gene, (p=0.009), implicating rgs2 as a gene involved in the QTL effect (Yalcin et al., 2004). The other two QTLs detected from the MF1 progenitor mapping do not contain any known expressed sequence and lie over 100 kb from the closest known gene. Functionally important elements have been identified up to 1 Mb from their cognate gene (Higgs et al., 1990; Lettice et al., 2002; Nobrega et al., 2003), so these QTLs likely represent functional elements which regulate the expression of a nearby gene, possibly rgs18 and B830045N13RIK, or even rgs2 itself. To investigate the possibility of long range enhancer elements, we investigated the homology of this mouse genomic region with those of human and Fugu rubipes, as functional elements are likely to be conserved across species. We identified 567 conserved non-coding sequences (CNS) which were of greater than 70% homology and over 100 bp in length, but did not match any expressed sequences. Sequencing the CNS regions (a total of 42,795 bp) in the eight inbred HS progenitor strains provided an additional 91 SNPs (and 7 microsatellites and 4 insertion/deletions), which we included in our final analysis of 1700 SNPs covering the 4.8 Mb interval.

Fig. 5. MF1 haplotypes across the QTL showing strain of origin of SNP variants. The founder strain of origin of each SNP haplotype is depicted by coloured blocks. Each colour is derived from a single inbred founder strain.

156

Fullerton

Fig. 6. Single marker and HAPPY QTL mapping in MF1 mice. The horizontal scale is in megabases, corresponding to a region from 64.4 to 65 cM on mouse chromosome 1, and the vertical scale is the negative logarithm of the P-value (logP). This analysis was carried out assuming four progenitor strains. Single point ANOVA results are shown as red circles. Multiple marker HAPPY results are indicated by the solid black line. The orientation of known genes is shown by arrows. The 95% confidence intervals of the three QTLs are shown by horizontal lines, numbered 1, 2 and 3. The horizontal dotted line represents the Bonferroni corrected 5% significance threshold.

Using a modification of the HAPPY dynamic programming progenitor strain mapping algorithm, we determined the probability that each SNP variant is the QTN responsible for the detected genetic effect (Yalcin et al., 2005). The method combines information about the genetic action of the QTL, obtained by genotyping and phenotyping the HS and MF1 animals, with the strain distribution patterns among the founders of the HS, obtained by sequencing. The algorithm then assigns a probability to each SNP based on its likelihood of being a QTN. We are currently investigating the highest ranking SNPs from this analysis for functionality; firstly using tests for DNA-protein binding via Electrophoretic Mobility Shift Assay, and secondly testing the effect of these variants on gene expression by cloning the surrounding DNA into promoter driven expression vectors and testing for alterations in gene expression. Preliminary investigation has identified SNP variants with differential promoter expression between the high and low anxiety strain constructs,

providing evidence that the SNPs have an effect on gene expression. While our data are preliminary, we are optimistic that this approach will elucidate a number of functional variants which can then be investigated as potential QTN in both mouse models and as targets for human association. Of course the question of how to definitively prove a SNP is a QTN is a topic under heated debate (Abiola et al., 2003), particularly in the area of behavioural traits. USING MOUSE MAPPING TO IDENTIFY HUMAN QTN As anxiety is an evolutionarily ancient behavioural trait, which exhibits itself in all vertebrates, we expect that at least part of the biological basis for anxiety will overlap with that of other species. So we could expect that loci which contribute to variation in anxiety in mice, which we have identified in our high resolution mapping studies, could also contribute to variation in some anxiety-related traits in humans,

New Approaches to the Genetic Analysis such as neuroticism or depression. While the phenotypes we are comparing are different, the underlying genetic basis of these disorders is likely to overlap, so we have used clues from our high resolution mapping strategies in the mouse to identify potential candidate SNPs for analysis in our human cohort. The 4.78 Mb region of mouse chromosome 1 is syntenic in inverse orientation with a region spanning 186.5–190 Mb at 201–203 cM on human chromosome 1. Our human linkage analysis for neuroticism revealed a QTL on the q-arm of human chromosome 1 at 113–144 cM, which is centromeric to the mouse syntenic region. It may be that the locus we have fine mapped in the mouse is different from that which is responsible for the large chromosome 1q QTL peak in our human study, or the mouse loci may have a more moderate effect in humans and therefore have not been detected with high significance in our human linkage analysis. On the other hand, the exact QTL position is dependant on the accuracy of the genetic map used in the analysis, and while we have used a high resolution sex averaged genetic map which combines physical and genetic mapping data (Kong et al., 2004), the differences in recombination rates between males and females may give a biased map position of QTL location, particularly as our human linkage analysis shows sex specific loci. Preliminary data for the association analysis is presented (Fig. 7) for a selection of our most extreme scoring families used in the original neuroticism study (n=320) and from a sample of unrelated extreme scoring individuals for EPQ-N whose DNA has been collected as a follow-up sample for association testing (n=700). We are detecting association at about the 5% significance level (logP 2.9) with the neuroticism phenotype for a handful of SNPs which are in the syntenic regions of the mouse QTLs. Interestingly, when we separate the samples into single sex groups and test for association, we observe some sex specific effects, as we might expect from our single sex analysis of the linkage data (see Fig. 1) which shows a different molecular architecture of N in males and females. The full set of data from the human association study will soon be published (Willis-Owen et al.—in preparation), with full analysis of the total sample; including over 500 families and 1000 unrelated individuals with EPQ-N scores, and additional phenotypes including major depression and recurrent major depression. Also, as functional candidate SNPs are revealed through the expression analysis of mouse promoter and enhancer constructs, additional SNPs

157 will be investigated for association in the human cohort. DISCUSSION In the post-genome era, with our increasing understanding of the complexity of the genetic architecture of the human genome and the molecular complexity underlying common diseases, our approaches to identify QTGs are becoming more ambitious. With the completion of the human and mouse genome projects, we now have access to an extraordinary amount of information about the similarities and differences between the two genomes, in addition to the genetic location and frequency of hundreds of thousands of SNPs which can be used for fine scale association and linkage mapping of complex traits. The international HapMap consortium is expected to produce a linkage disequilibrium map across the human genome at a resolution of 1 SNP per 5 kb. While this might appear to be a sufficient marker density for mapping purposes, preliminary analysis of chromosome structure at even higher resolution shows that we will still not understand the full complexity of the underlying DNA structure even if we sample SNPs every 500 bp (Ke et al., 2004). This has huge implications on the potential study design of association studies and suggests we need to know more about the fine detailed structure of human chromosomes to implement efficient genomewide genetic association studies. Encouragingly though, the picture emerging from the field through linkage replication studies is beginning to show some consistency in loci influencing anxiety-related traits (Table I). Multiple linkages have been reported for chromosome 1, implying at least two linkage regions at around 70 and 120 cM which have been detected with significant or suggestive linkage in more than two studies (Fullerton et al., 2003; Gelernter et al., 2003; Nash et al., 2004; Neale et al., 2005; Nurnberger et al., 2001). Under the same criteria, consistent findings have implicated loci on chromosomes 7 (160 cM) (Curtis, 2004; Nurnberger et al., 2001), 8 (50 cM) (Cloninger et al., 1998; Dina et al., 2005), 10 (150 cM) (Smoller et al., 2001; Zubenko et al., 2003), and 12 (100 cM) (Abkevich et al., 2003; Fullerton et al., 2003; Nash et al., 2004). Other loci for which there is reasonable support, with a significant or suggestive finding from one study and an additional positive trend from a second study, are chromosomes 3 (around 105 cM) (Camp et al., 2005;

158

Fullerton

Neale et al., 2005), 11 (pter, 120 cM) (Cloninger et al., 1998; Fullerton et al., 2003; Gelernter et al., 2001; Neale et al., 2005; Zubenko et al., 2003), 15 (100 cM) (Camp et al., 2005; Holmans et al., 2004), 17 (55 cM) (Curtis, 2004; Nash et al., 2004), and 18 (80 cM) (Camp et al., 2005; Cloninger et al., 1998). While no single study to date has identified all of these possible QTLs, the joining together of data from many studies through large multi-site collaborative efforts (Farmer et al., 2004), and meta-analysis of data from smaller individual groups (Levinson, 2005) may prove to be the best approach for confirming real linkage signals. While there appears to be some consistency of linkage to some chromosomal regions, the issue of whether a finding can be considered a positive replication if two QTL peaks map to approximately the same interval is a critical question. Simulation studies have shown that the location estimates for linkage peaks can vary, depending on the strength of the linkage signal, which can be affected by heterogene5 (a) 4.5 4 3.5 3 L 2.5 o 2 g 1.5 1 0.5 0 187 187.5 188

188.5 189

189.5 190

190.5

ity, penetrance and sample size (Roberts et al., 1999). Roberts illustrates a similar problem in schizophrenia research, where seven different investigations show linkage to the p-arm of chromosome 6, however it remains unclear as to whether these linkages are detecting the same or different QTL effects. As with the field of anxiety and depression, confirmation of replication will not be possible until QTGs are identified, and the nucleotide variants responsible are finally characterised. Additionally, there is the issue of whether two findings can be considered replications if they were identified using different, but related, phenotypes. Considering the evidence of familial clustering and co-morbidity of depression, neuroticism and anxiety disorders (Kessler et al., 2003; Khan et al., 2005), it seems possible to suggest that the genetic factors influencing these disorders overlap, and so overlapping linkage findings should be expected among different related phenotypes. Whether the linkage findings in humans will in general corroborate those 5 (c) 4.5 4 3.5 3 L 2.5 o 2 g 1.5 1 0.5 0 187 187.5 188

188.5 189 Position (Mb)

189.5

190

190.5

189.5 190

190.5

Position (Mb)

Position (Mb)

5 4.5 (b) 4 3.5 3 L 2.5 o 2 g 1.5 1 0.5 0 187 187.5 188

188.5 189

189.5 190

190.5

5 (d) 4.5 4 3.5 3 L 2.5 o 2 g 1.5 1 0.5 0 187 187.5 188

188.5

189

Position (Mb)

Fig. 7. Single marker ANOVA for human syntenic markers for mouse candidate SNPs. The human megabase positions according to ensembl build 35 are shown on the x-axis, and the negative logarithm of the P value of association is shown on the y-axis. Panels a and c show data from 320 individuals from extreme scoring families from original neuroticism study. Panels b and d show data from follow-up sample of 700 unrelated individuals with extreme N scores. Panels a and b show combined data for both sexes (purple diamonds), and panels c and d show data separated into female only (red circles) and male only (blue squares). The human syntenic positions corresponding to mouse QTL 95% confidence intervals are indicated by the coloured horizontal lines; QTL1 (light blue), QTL2 (purple) and QTL3 (brown).

New Approaches to the Genetic Analysis commonly used behavioural phenotypes in model organisms will depend on the similarity of the phenotype and genetic architecture of the disorder. Our laboratory has shown that a cross-species approach to elucidating QTN using the mouse as a model organism is an efficient way of reducing some of the confounding difficulties surrounding human QTN identification; such as heterogeneity, sample size and unknown genetic or haplotypic background of the QTN. I have discussed our experience of trying to map the genetic basis of a human personality trait, neuroticism, which we have used as a surrogate phenotype for two common psychiatric illnesses, depression and generalised anxiety. I have shown that although the increased power afforded by the use of large sample sizes can be obtained through the selection of phenotypic extremes, the results of the human linkage analysis with an extremely large dataset still did not provide chromosomal intervals small enough to allow us easily to identify candidate genes. In contrast, genetic analysis of mice is a much more robust procedure and can now proceed to the point of identifying individual genes. The advantages of working with animal models for genetic analysis are great; breeding experiments can be carried out with mice that are impossible in human genetics. We have shown that such experiments can be used to obtain extremely high resolution mapping, even for loci that contribute as little as 5% to the phenotypic variance. Because of the genetic relationship of commercially available outbred mouse populations to some inbred strains, we have been able to refine a mouse QTL from an interval of 40 cM detected through an F2 cross, to at least three individual loci comprising 0.7, 1.7 and 2.5 Mb of DNA by ultra-fine mapping using progenitor strain distribution patterns. Although there are difficulties in choosing the appropriate experimental design and analysis, as has been discussed at length elsewhere (Flint et al., 2005). Through this analysis, we have successfully identified rgs2 as a QTG which is responsible for a small proportion of the genetic variance in anxiety in mice. Translating the mouse mapping data to help in the elucidation of human QTN is difficult. We have started to see whether we can apply the findings in mouse to our human data set, by testing some SNP variants in RGS2 and in some of the orthologous conserved non-coding regions implicated from the mouse study for association in the human neuroticism sample, with minimal or suggestive evidence for association.

159 However the ability of a marker to detect association depends on the degree of linkage disequilibrium with the QTN, whether QTN and marker have similar allele frequencies, and, as illustrated from our progenitor strain distribution analysis represented in Fig. 4, whether the marker allele has the same allelic pattern as the QTN. Before applying descent mapping to our outbred mice data, our single marker association across the mouse QTL region was largely non-significant (Fig. 6), suggesting that this locus had no genetic effect on the phenotype; a similar picture to the association data presented in our human study (Fig. 7). However when we identified the correct strain distribution pattern of our mouse QTL effect and applied descent mapping to the same genotypes, we increased power by extracting more information about the chromosomal structure and were able to identify multiple QTLs with individual effects. So far it has not been possible to use a similar strategy in human genetics, because without access to information about the genetic constitution of our ancestors, we do not know the relationship between QTL and marker alleles. However, studies in coalescent theory for human haplotypic variation may shed some light on the ‘‘progenitor haplotypes’’ of human populations across small segments of DNA (Durrant et al., 2004), and may elucidate additional fine structural information which could prove useful in the design of human association study strategies. While the human association data presented here are preliminary, and we cannot be certain that genes identified using animal models will translate into candidates for human traits, it does seem as if an approach informed by clues gained from genetic analysis in the mouse will benefit attempts to find QTGs for complex human traits.

REFERENCES Abiola, O., Angel, J. M., Avner, P., Bachmanov, A. A., Belknap, J. K., Bennett, B., Blankenhorn, E. P., and Blizard, D. A., (2003). The nature and identification of quantitative trait loci: a community’s view. Nat. Rev. Genet. 4(11):911–916. Abkevich, V., Camp, N. J., Hensel, C. H., Neff, C. D., Russell, D. L., Hughes, D. C., Plenk, A. M., and Lowry, M. R., (2003). Predisposition locus for major depression at chromosome 12q22–12q23.2. Am. J. Hum. Genet. 73(6):1271–1281. Caldarone, B., Saavedra, C., Tartaglia, K., Wehner, J. M., Dudek, B. C., and Flaherty, L. (1997). Quantitative trait loci analysis affecting contextual conditioning in mice. Nat. Genet. 17(3):335–337. Camp, N. J., Lowry, M. R., Richards, R. L., Plenk, A. M., Carter, C., Hensel, C. H., Abkevich, V., Skolnick, M. H., Shattuck, D., Rowe, K. G., Hughes, D. C., and Cannon-Albright, L. A.

160 (2005). Genome-wide linkage analyses of extended Utah pedigrees identifies loci that influence recurrent, early-onset major depression and anxiety disorders. Am. J. Med. Genet. B Neuropsychiatr. Genet. 135(1):85–93. Cloninger, C. R. (1994). Temperament and personality. Curr. Opin. Neurobiol. 4:266–273. Cloninger, C. R., Van Eerdewegh, P., Goate, A., Edenberg, H. J., Blangero, J., Hesselbrock, V., Reich, T., and Nurnberger, J. Jr., (1998). Anxiety proneness linked to epistatic loci in genome scan of human personality traits. Am. J. Med. Genet. 81(4):313–317. Curtis, D. (2004). Re-analysis of collaborative study on the genetics of alcoholism pedigrees suggests the presence of loci influencing novelty-seeking near D12S391 and D17S1299. Psychiatr. Genet. 14(3):151–155. Digman, J. M. (1990). Personality structure: emergence of the 5factor model. Ann. Rev. Psychol. 41:417–440. Dina, C., Nemanov, L., Gritsenko, I., Rosolio, N., Osher, Y., Heresco-Levy, U., Sariashvilli, E., Bachner-Melman, R., Zohar, A. H., Benjamin, J., Belmaker, R. H., and Ebstein, R. P. (2005). Fine mapping of a region on chromosome 8p gives evidence for a QTL contributing to individual differences in an anxiety-related personality trait: TPQ harm avoidance. Am. J. Med. Genet. B Neuropsychiatr. Genet. 132(1):104–108. Dreary, I. J., and Matthews, G. (1993). Personality traits are alive and well. The Psychologist 299–311. Duggan, C., Sham, P., Lee, A., Minne, C., and Murray, R. (1995). Neuroticism: a vulnerability marker for depression evidence from a family study. J. Affect Disord. 35(3):139–143. Duggan, C. F., Lee, A. S., and Murray, R. M. (1990). Does personality predict long-term outcome in depression? Br. J. Psychiatry 157:19–24. Durrant, C., Zondervan, K. T., Cardon, L. R., Hunt, S., Deloukas, P., and Morris, A. P. (2004). Linkage disequilibrium mapping via cladistic analysis of single-nucleotide polymorphism haplotypes. Am. J. Hum. Genet. 75(1):35–43. Eysenck, H., and Eysenck, S. (1975). Manual of the Eysenck Personality Questionnaire. San Diego, CA: Educational and Industrial Testing Service. Farmer, A., Breen, G., Brewster, S., Craddock, N., Gill, M., Korszun, A., Maier, W., and Middleton, L., (2004). The Depression Network (DeNT) Study: methodology and sociodemographic characteristics of the first 470 affected sibling pairs from a large multi-site linkage genetic study. BMC Psychiatry 4(1):42. Fernandez-Teruel, A., Escorihuela, R. M., Gray, J. A., Aguilar, R., Gil, L., Gimenez-Llort, L., Tobena, A., Bhomra, A., Nicod, A., Mott, R., Driscoll, P., Dawson, G. R., and Flint, J. (2002). A quantitative trait locus influencing anxiety in the laboratory rat. Genome Res. 12(4):618–626. Flint, J. (2004). The genetic basis of neuroticism. Neurosci. Behav. Rev. 28:307–316. Flint, J., Corley, R., DeFries, J. C., Fulker, D. W., Gray, J. A., Miller, S., and Collins, A. C. (1995). A simple genetic basis for a complex psychological trait in laboratory mice. Science 269(5229):1432–1435. Flint, J., Valdar, W., Shifman, S., and Mott, R. (2005). Strategies for mapping and cloning quantitative trait genes in rodents. Nat. Rev. Genet. 6(4):271–286. Fullerton, J., Cubin, M., Tiwari, H., Wang, C., Bomhra, A., Davidson, S., Miller, S., Fairburn, C., Goodwin, G., Neale, M. C., Fiddy, S., Mott, R., Allison, D. B., and Flint, J. (2003). Linkage analysis of extremely discordant and concordant sibling pairs identifies quantitative-trait loci that influence variation in the human personality trait neuroticism. Am. J. Hum. Genet. 72(4):879–890. Gelernter, J., Bonvicini, K., Page, G., Woods, S. W., Goddard, A. W., Kruger, S., Pauls, D .L., and Goodson, S. (2001). Linkage

Fullerton genome scan for loci predisposing to panic disorder or agoraphobia. Am. J. Med. Genet. 105(6):548–557. Gelernter, J., Page, G. P., Bonvicini, K., Woods, S. W., Pauls, D. L., and Kruger, S. (2003). A chromosome 14 risk locus for simple phobia: results from a genomewide linkage scan. Mol. Psychiatry 8(1):71–82. Gelernter, J., Page, G. P., Stein, M. B., and Woods, S. W. (2004). Genome-wide linkage scan for loci predisposing to social phobia: evidence for a chromosome 16 risk locus. Am. J. Psychiatry 161(1):59–66. Gershenfeld, H. K., Neumann, P. E., Mathis, C., Crawley, J. N., Li, X., and Paul, S. M. (1997). Mapping quantitative trait loci for open-field behavior in mice. Behav. Genet. 27(3):201–210. Gray, J. A. (1982). The Neuropsychology of Anxiety: An Enquiry into the Function of the Septo-hippocampal System. Oxford: Oxford University Press. Gray, J. A. (1987). The Psychology of Fear and Stress. Cambridge: Cambridge University Press. Gray, J. A., and McNaughton, N. (2000). The Neuropsychology of Anxiety. Oxford: Oxford University Press. Hettema, J. M., Prescott, C. A., and Kendler, K. S. (2004). Genetic and environmental sources of covariation between generalized anxiety disorder and neuroticism. Am. J. Psychiatry 161(9):1581–1587. Higgs, D. R., Wood, W. G., Jarman, A. P., Sharpe, J., Lida, J., Pretorius, I. M., and Ayyub, H. (1990). A major positive regulatory region located far upstream of the human alphaglobin gene locus. Genes Dev. 4(9):1588–1601. Hirschfeld, R. M., Klerman, G. L., Lavori, P., Keller, M. B., Griffith, P., and Coryell, W. (1989). Premorbid personality assessments of first onset of major depression. Arch. Gen. Psychiatry 46(4):345–350. Holmans, P., Zubenko, G. S., Crowe, R. R., DePaulo, J. R. Jr., Scheftner, W. A., Weissman, M. M., Zubenko, W. N., and Boutelle, S., (2004). Genomewide significant linkage to recurrent, early-onset major depressive disorder on chromosome 15q. Am. J. Hum. Genet. 74(6):1154–1167. Ke, X., Hunt, S., Tapper, W., Lawrence, R., Stavrides, G., Ghori, J., Whittaker, P., Collins, A., Morris, A. P., Bentley, D., Cardon, L. R., and Deloukas, P. (2004). The impact of SNP density on fine-scale patterns of linkage disequilibrium. Hum. Mol. Genet. 13(6):577–588. Kendler, K. S., Kuhn, J., and Prescott, C. A. (2004). The interrelationship of neuroticism, sex, and stressful life events in the prediction of episodes of major depression. Am. J. Psychiatry 161(4):631–636. Kessler, R. C., Bergland, P., Demler, O., Jin, R., Koretz, D., Merikangas, K., Rush, A. J., Walters, E. E., and Wang, P. S. (2003). The epidimiology of major depressive disorder: results from the National Comorbidity Survey Replication (NCS-R). JAMA 289:3095–3105. Khan, A. A., Jacobson, K. C., Gardner, C. O., Prescott, C. A., and Kendler, K. S. (2005). Personality and comorbidity of common psychiatric disorders. Br. J. Psychiatry 186:190–196. Kong, X., Murphy, K., Raj, T., He, C., White, P. S., and Matise, T. C. (2004). A combined linkage-physical map of the human genome. Am. J. Hum. Genet. 75(6):1143–1148. Lake, R. I., Eaves, L. J., Maes, H. H., Heath, A. C., and Martin, N. G. (2000). Further evidence against the environmental transmission of individual differences in neuroticism from a collaborative study of 45,850 twins and relatives on two continents. Behav. Genet. 30(3):223–233. Lettice, L. A., Horikoshi, T., Heaney, S. J., van Baren, M. J., van der Linde, H. C., Breedveld, G. J., Joosse, M., and Akarsu, N., (2002). Disruption of a long-range cis-acting regulator for Shh causes preaxial polydactyly. Proc. Natl. Acad. Sci. USA 99(11):7548–7553. Levinson, D. F. (2005). Meta-analysis in psychiatric genetics. Curr. Psychiatry Rep. 7(2):143–151.

New Approaches to the Genetic Analysis Loehlin, J. C., and Nichols, R. C. (1976). Heredity, Environment and Personality. Austin: University of Texas Press. Martin, N., Goodwin, G., Fairburn, C., Wilson, R., Allison, D., Cardon, L. R., and Flint, J. (2000). A population-based study of personality in 34,000 sib-pairs. Twin Res. 3(4):310–315. McClearn, G., Wilson, J., and Meredith, W. (1970). Contributions to Behaviour-Genetic Analysis: The Mouse as a Prototype. New York: Appleton Cenury Crofts. Mott, R., Talbot, C. J., Turri, M. G., Collins, A. C., and Flint, J. (2000). A method for fine mapping quantitative trait loci in outbred animal stocks. Proc. Natl. Acad. Sci. USA 97(23):12649–12654. Nash, M. W., Huezo-Diaz, P., Williamson, R. J., Sterne, A., Purcell, S., Hoda, F., Cherny, S. S., and Abecasis, G. R., (2004). Genome-wide linkage analysis of a composite index of neuroticism and mood-related scales in extreme selected sibships. Hum. Mol. Genet. 13(19):2173–2182. Neale, B. M., Sullivan, P. F., and Kendler, K. S. (2005). A genome scan of neuroticism in nicotine dependent smokers. Am. J. Med. Genet. B Neuropsychiatr. Genet. 132(1):65–69. Nobrega, M. A., Ovcharenko, I., Afzal, V., and Rubin, E. M. (2003). Scanning human gene deserts for long-range enhancers. Science 302(5644):413. Nurnberger, J. I. Jr., Foroud, T., Flury, L., Su, J., Meyer, E. T., Hu, K., Crowe, R., Edenberg, H., Goate, A., Bierut, L., Reich, T., Schuckit, M., and Reich, W. (2001). Evidence for a locus on chromosome 1 that influences vulnerability to alcoholism and affective disorder. Am. J. Psychiatry 158(5):718–724. Oliveira-Dos-Santos, A. J., Matsumoto, G., Snow, B. E., Bai, D., Houston, F. P., Whishaw, I. Q., Mariathasan, S., Sasaki, T., Wakeham, A., Ohashi, P. S., Roder, J. C., Barnes, C. A., Siderovski, D. P., and Penninger, J. M. (2000). Regulation of T cell activation, anxiety, and male aggression by RGS2. Proc. Natl. Acad. Sci. USA 97(22):12272–12277. Risch, N., and Merikangas, K. (1996). The future of genetic studies of complex human diseases. Science 273(5281):1516–1517. Risch, N., and Zhang, H. (1995). Extreme discordant sib pairs for mapping quantitative trait loci in humans. Science 268(5217):1584–1589. Roberts, S. B., MacLean, C. J., Neale, M. C., Eaves, L. J., and Kendler, K. S. (1999). Replication of linkage studies of complex traits: an examination of variation in location estimates. Am. J. Hum. Genet. 65:876–884.

161 Roy, A. (1990). Personality variables in depressed patients and normal controls. Neuropsychobiology 23(3):119–123. Smoller, J. W., Acierno, J. S. Jr., Rosenbaum, J. F., Biederman, J., Pollack, M. H., Meminger, S., Pava, J. A., Chadwick, L. H., White, C., Bulzacchelli, M., and Slaugenhaupt, S. A. (2001). Targeted genome screen of panic disorder and anxiety disorder proneness using homology to murine QTL regions. Am. J. Med. Genet. 105(2):195–206. Talbot, C. J., Nicod, A., Cherny, S. S., Fulker, D. W., Collins, A. C., and Flint, J. (1999). High-resolution mapping of quantitative trait loci in outbred mice. Nat. Genet. 21(3):305–308. Visscher, P. M., and Hopper, J. L. (2001). Power of regression and maximum likelihood methods to map QTL from sib-pair and DZ twin data. Ann. Hum. Genet. 65(Pt 6):583–601. Yalcin, B., Flint, J., and Mott, R. (2005). Using progenitor strain information to identify quantitative trait nucleotides in outbred mice. Genetics 171(2):673–681. Yalcin, B., Willis-Owen, S. A., Fullerton, J., Meesaq, A., Deacon, R. M., Rawlins, J. N., Copley, R. R., Morris, A. P., Flint, J., and Mott, R. (2004). Genetic dissection of a behavioral quantitative trait locus shows that Rgs2 modulates anxiety in mice. Nat. Genet. 36(11):1197–1202. Zeiss, A. M., and Lewinsohn, P. M. (1988). Enduring deficits after remissions of depression: a test of the scar hypothesis. Behav. Res. Ther. 26(2):151–158. Zubenko, G. S., Hughes, H. B. 3rd., Maher, B. S., Stiffler, J. S., Zubenko, W. N., and Marazita, M. L. (2002). Genetic linkage of region containing the CREB1 gene to depressive disorders in women from families with recurrent, early-onset, major depression. Am. J. Med. Genet. 114(8):980–987. Zubenko, G. S., Maher, B., Hughes, H. B. 3rd., Zubenko, W. N., Stiffler, J. S., Kaplan, B. B., and Marazita, M. L. (2003). Genome-wide linkage survey for genetic loci that influence the development of depressive disorders in families with recurrent, early-onset, major depression. Am. J. Med. Genet. B Neuropsychiatr. Genet. 123(1):1–18. Zuckerman, M., Kuhlman, D. M., and Camac, C. (1988). What lies beyond E and N? Factor analysis of scales believed to measure basic dimensions of personality. J. Pers. Soc. Psychol. 54:96– 107.

Edited by Dorret Boomsma