26 Neurobiology and the genetics of suicide

Besides serotonin dysfunction, which was the main focus for about three decades, many other aspects of brain neurobiology have now been shown to be involved in the causality of suicidal behaviour. This chapter attempts to provide a broad overview of the entire range of studies performed in the area of neurobiology of suicide. The investigated involvement of genetics in each presently known neurobiological alteration is likewise presented. Although, the complexities and challenges in this field may sometimes seem overwhelming, this overview shows that the knowledge in this area is constantly being increased and refined in its details, and small breakthroughs occur constantly. Thus, it appears that if sufficient time and resources are dedicated to this problem, a critical mass of understanding will be reached, enabling the development of entirely novel tools for prevention of suicide.

The causes of suicide appear to be heterogeneous and complex, and no simple explanations of the phenomenon exists. Risk of suicide-related behaviour is considered to be determined by a complex interplay of sociocultural factors, traumatic life experiences, psychiatric history, personality traits and genetic vulnerability, as depicted in a stress–vulnerability model (Figure 26.1) (Mann and Arango 1992; Wasserman 2001; Carballo et al. 2008). Suicidal individuals show remarkable neurobiological alterations, which are often shared with other psychiatric conditions, depression, and other behaviours/traits, such as increased anger/hostility, impulsivity, anxiety and impaired behavioural inhibition. However, a substantial proportion of such individuals have likely never considered taking their own lives nor commit such an act. Therefore, it is further likely that suicidal individuals must carry some type of neurobiological alteration of these expressions, which specify the increasing vulnerability towards suicidal acts. Indeed, an intrinsic predisposition towards suicidal behaviour has been demonstrated by evidence for involvement of genetic components, by using family, adoption and twin studies. The progress in understanding the involvement of neurobiology and genetics in suicide has been reviewed elsewhere (Kamali et al. 2001; Wasserman 2001; Mann 2003; van Heeringen 2003; Balazic and Marusic 2005; Brent and Mann 2005; Bondy et al. 2006; Mann et al. 2006; Mann and Currier 2007; Rujescu et al. 2007; Voracek and Loibl 2007; Wasserman et al. 2007b; Brezo et al. 2008; Carballo et al. 2008; Currier and Mann 2008; and in the 1997 December issue, no. 29, volume 836, of the Annals of the New York Academy of Sciences).

Adoption, twin and family studies indicate that suicidal acts have a genetic contribution in terms of cause or diathesis, which is independent of the heritability of major psychiatric disorders (Schulsinger et al. 1979; Roy et al. 1991). Aggregation of the risk estimates from 21 family-study reports, with a total of nearly 25,000 suicidal subjects and close relatives (Baldessarini and Hennen 2004), showed that even a cautious estimate of the overall pooled risk ratio, based on a meta-analytic method, was 2.86.

Twin studies compare the concordance for suicidal behaviour in monozygotic twins, which share 100 per cent of their genes, and dizygotic twins, which share 50 per cent of their genes on average. This allows separating effects due to shared environment from genetic factors. Roy et al. (1991) examined 62 monozygotic and 114 dizygotic twin pairs, and reported concordance rates for suicide of 11.3 per cent and 1.8 per cent, respectively. A recent study, based on this data, suggested heritability for completed suicide of about 43 per cent by using a logistic regression model (McGuffin et al. 2001). Similar results have been found for suicide attempts. An Australian twin study with a total of nearly 6000 twins estimated that genetic factors account for approximately 45 per cent of the variance in suicidal thoughts and behaviour (Statham et al. 1998). The co-twin of monozygotic pairs, where the other twin had a history of suicide attempts, was shown to have an elevated risk for suicidal behaviour. Even after controlling for contributing risk factors like depression and other psychiatric disorders, the authors observed a highly significant 3.9-fold increased risk. A more recent review of twin-studies published between 1812 and 2006 (Voracek and Loibl 2007), confirmed the influence of genetic components (with heritability of 30–55 per cent) in suicide behaviour, largely independent of the inheritance of other psychiatric disorders.

Adoption studies are less commonly performed. Three classical studies used the same Danish health statistics register (Kety et al. 1971; Schulsinger et al. 1979; Wender et al. 1986). The investigation of Schulsinger et al. identified 57 suicide victims among early adopted Danish citizens and defined them as index cases. The biological relatives of these index cases showed a sixfold higher suicide rate (4.46 per cent) than the biological relatives of a matched, non-suicidal adoptee control group (0.74 per cent). Furthermore, there was no suicide among the adoptive relatives of the index cases.

For suicide-related behaviour, the number of susceptibility genes, the risk conferred by each gene, as well as the degree of interaction(s) between the genes is unknown. It is likely that a high degree of complexity exists (including redundant, complementing, silencing or augmenting effects/interactions), involving different genes, as well as interactions between genes and environment (stress, xenobiotics and drugs), which will further vary during the different developmental stages of the human being (Kendler 2005; Lin et al. 2007). Thus, finding genes that predispose to suicide-related behaviour presents a major challenge, given that the neuropathology of suicide, as well as neurobiological mechanisms in general, are far from being understood. The search for susceptibility genes of suicide is presently confronted by using two different strategies: either to look at all genes at once (‘genome-wide’) in a less detailed manner, or to directly perform detailed studies of biologically plausible ‘candidate’ genes, sometimes as a secondary follow-up of an indicative result from a genome-wide study. Candidate genes are usually identified following the experimental investigation of certain proteins in particular neurobiological pathways, studied in human subjects (by post mortem, brain scans or other means), animal models (including use of transgenic animals) or tissue cultures (using various lineages of nerve cells).

In the population–genetic approaches, one may study the co-segregation of certain genetic variants/mutations with the trait/disorder of interest in families (‘linkage’ studies), or to investigate if there are differences in the occurrence of certain genetic variants between non-related groups of affected, suicidal individuals, in comparison to unaffected, healthy controls (‘association’ studies). However, the progress of many population–genetic studies is often hampered due the problems of phenotypic and genetic heterogeneity, as well as under-powered investigations caused by too small sample sizes, often resulting in inconsistencies, and even statistical artefacts. As these problems are being addressed, and complemented with experimental measures, a psychobiological understanding is emerging.

The inclusion and definition of the genetic components involved in suicidality are of great interest when specifying the individuals having the clinical manifestations of, for example, depression, increased anger, impulsivity, anxiety, and/or behavioural disinhibition, who may be at increased risk of performing suicidal acts. The suicide-specific neurobiological alterations involved are likely to be assessed by using ‘endophenotypes’, which involves any measurable neurobiological parameters whose outcome/state are more directly linked with any underlying genetic alterations, and which act on the pathway of development of suicidal vulnerability. Thus, the genetic variables must be determined as they form the basal layer of information, which could influence all secondary variables/endophenotypes, and the sum of these effects would shape the individual's susceptibility to suicidality. From a pharmacogenomic perspective, it is of importance to have knowledge about how the individual's genetic variation relates to a drug in question (e.g. SSRI, lithium or other drug treatments), in context to its functionality, enabling better evaluation of drug efficacy and development of diagnostic tools by, for example, explaining side-effects and treatment failures and by improving individualized drug treatments. In the future, knowledge about unfavourable/‘dysfunctional’ genetic variants will likely bring about new, highly specific treatment possibilities with, for example, inhibitory ribonucleic acid (RNA), so-called ‘gene therapy’ (Sah 2006).

The serotonin system projects through many areas of the brain, and is considered to be important in mood and behaviour. Serotonin (5-hydroxytryptamine, or 5-HT) is the major monoaminergic neurotransmitter, first identified in blood in 1948 (Rapport et al. 1948). It is synthesized in cells from the precursor molecule L-tryptophan, an essential amino acid, catalyzed by enzymes tryptophan hydroxylase (TPH) and aromatic L-amino acid/DOPA decarboxylase (DDC). 5-HT has a long history of being widely implicated in the causality of depression, originating from the ‘monoamine hypothesis’, which proposed the involvement of a deficiency in monoamines, including 5-HT, in depression (Schildkraut 1965). There is also much evidence concerning the involvement of the 5-HT system in suicide, independently of other diagnoses. This line of research, which now spans over more than four decades, was initiated by the observations of lowered 5-HT and 5-HT metabolite (5-hydroxyindoleacetic acid—5-HIAA) levels in various brain regions of suicide victims (Shaw et al. 1967; Pare et al. 1969; Birkmayer and Riederer 1975), as well as lowered 5-HIAA in the cerebrospinal fluid (CSF) in relation to depression and violent suicide attempts (Åsberg et al. 1976a, b). Later, blunted prolactin response to fenfluramine (which acts by inducing serotonin release), altered binding/levels of serotonin receptors, and changes in the 5-HT transporter (5-HTT) were also observed (Åsberg 1997; Mann et al. 2001). Brain regions shown to be affected in relation to suicide and serotonin are mainly, among others, the brain stem (dorsal raphe nuclei)/limbic system and prefrontal cortex (PFC), with the former being related to stress responses and the latter with cognitive and behavioural restraint, with rich innervations in-between. These neurobiological observations relate well with the frequent occurrence of impulsive–aggressive behaviours among suicidal individuals, as suicide is often deemed to be an impulsive act and/or an aggression which is directed inwards. The field has then progressed further by dissecting the roles and involvement of the various components of the serotonergic system.

Tryptophan is an essential amino acid and a rate-limiting source-molecule for serotonin synthesis. At the population level, it has been observed that a high intake diet of tryptophan, which may conteract any diet-induced serotonin deficiencies, is associated with lowered levels of suicide in the industrialized populations (Voracek and Tran 2007). Conversely, depletion of serotonin by a low-tryptophan diet, reverse the effects of antidepressant treatment with selective serotonin reuptake inhibitors (SSRIs) (Delgado et al. 1990). The enzyme tryptophan hydroxylase (TPH) is involved in the biosynthesis of serotonin, converting L-tryptophan, in a rate-limiting step, into 5-hydroxytryptophan, which is a substrate for subsequent conversion into serotonin by DDC. Inhibition of TPH by parachlorophenylalanine will, similarly to low-tryptophan diets, also result in serotonin depletion. Thus, available tryptophan and TPH are important for serotonergic function. To explain the inter-individual variations in the response to such treatments, in terms of effects on the serotonergic system, as well as in terms of psychopathology, polymorphisms in the TPH gene(s) have been studied.

Initially, many studies were focused on the gene TPH1, which was for a long time the only known TPH-encoding gene. An early meta-analysis by Lalovic and Turecki (2002), did not find an association of the commonly studied intron 7 A218C single nucleotide polymorphism (SNP) with suicidal behaviour. A further meta-analysis by Rujescu et al. (2003b) summarized the results of seven studies and found a higher frequency of the A218C allele in patients with suicidal behaviour, strongly suggesting that this TPH polymorphism is associated with suicidal behaviour among at least Caucasians. A subsequent meta-analysis included nine studies and confirmed the association between the A218C polymorphism and suicidal behaviour, using both the fixed effect method and the random effect method (Bellivier et al. 2004). The most recent meta-analysis included a total of 22 studies, and examined the involvement of the A779C, A218C and A6526G polymorphisms, employing several strategies to maintain the power and robustness in the analysis (Li and He 2006). This study, which investigated all reports published between 1997 and July 2005, revealed strong cumulative evidence of association for the A779C and A218C polymorphisms among international populations (Li and He 2006). The lessons learned for other genes/polymorphisms, which often show inconsistent findings in initial studies, may be that it takes considerable time before sufficient sample size is gathered (in the thousands to tens of thousands), which carry enough statistical power required for the detection of the weak, single-locus effects believed to be involved in suicidal behaviour, as have often been observed for other complex/somatic disorders.

Almost two decades after the initial identification of the TPH1 gene (Grenett et al. 1987; Ledley et al. 1987), a second gene encoding for TPH (TPH2) was discovered in mice, and later in humans, which was shown to be highly restricted to the brain regions (Walther et al. 2003; Zhang et al. 2004). It suddenly appeared that some of the roles previously implicated for the TPH1 gene in brain serotonin and psychiatric disorders, may have been of an artefactual nature, being e.g. secondary to TPH2 activity, but recent studies demonstrate that there is actually a duality in the brain serotonin system, involving the two different TPH genes, with equal amounts of expression in regions such as the frontal cortex, hippocampus, hypothalamus and amygdala, and differential expression in the dorsal raphe (Nakamura et al. 2006; Zill et al. 2007; Abumaria et al. 2008). Nevertheless, attention has mostly shifted to TPH2, recently shown to have elevated expression levels in brainstem dorsal and median raphe nuclei of depressed suicides (Bach-Mizrachi et al. 2008), the region where most forebrain serotonin is being produced, which may reflect a homeostatic response to deficient serotonin levels. Breidenthal et al. (2004) screened the coding and exon-flanking intronic sequence of the TPH2 gene and identified several genetic variants that might serve as markers for association studies. Zill et al. (2004) could find an association of SNPs and a haplotype with completed suicide, whereas De Luca et al. (2004) could not find any association. However, a comprehensive investigation performed a linkage analysis in 1798 subjects from four different populations, and detected significant haplotype linkage of TPH2 to suicide attempt and major depression (Zhou et al. 2005). Another single marker and haplotype analysis was conducted in a large, family-based sample of patients with bipolar affective disorder. The authors detected significant association of a haplotype with both suicide attempts and bipolar affective disorder (Lopez et al. 2007). Interestingly, a new, truncated isoform of TPH2 has recently been identified and characterized, along with genetic variants, which likely cause this structural alteration and were associated with suicide and depression (Haghighi et al. 2008). In summary, the results on TPH2 are promising and further studies will better clarify the role of this gene in several facets of suicidal behaviour.

Besides the enzymes involved in serotonin synthesis (TPH and DDC), the main catabolic enzyme of serotonin (and noradrenaline, dopamine) was identified (Hare 1928), now called monoamine oxidase (MAO). MAO has been implicated in various forms of psychopathology, including suicidality, ever since the formulation of the monoamine hypothesis and the advent of MAO-inhibiting antidepressants (Schildkraut 1965). Early on, a twin study showed that inter-individual variation in the enzyme activity might be genetically determined, promoting the development of schizophrenia (Wyatt et al. 1973), and such differences were later suggested to be involved also in suicidality (Buchsbaum et al. 1977). Early knowledge about the biochemical characterization of two major MAO isoforms (A and B), likely helped to form the idea that heritable, genetic alterations might be causing the differences between the two isoforms, e.g. differences in substrate specificity, inhibitor sensitivity and tissue distribution, and that such alterations might be sufficient for causing psychobiological and behavioural consequences (Breakefield and Edelstein 1980; Murphy and Kalin 1980).

With the rapid development in DNA technologies, it was later possible to show that, as with TPH, there were two genes responsible for the major MAO isoforms (MAOA and MAOB), identified to be located on the X-chromosome (Breakefield and Edelstein 1980; Pintar et al. 1981; Ozelius et al. 1988; Sims et al. 1989) and to be expressed in the brain (Shih et al. 1990). Interestingly, loss of MAOA gene function is related to increased aggression (Brunner et al. 1993; Cases et al. 1995), a behaviour/trait which has long been considered to be of importance in explaining the suicide diathesis (i.e. the ‘anger out—anger in’ hypothesis), as well as in relation to an increased noradrenalin and cortisol levels, ideas originating from work previously done in the 1950s (Funkenstein et al. 1954; Schildkraut 1965; Ostroff et al. 1982). Curiously, the failures to demonstrate local/regional effects in the PFC in relation to MAO activity and suicidality (Mann and Stanley 1984), may have served as a stimuli for research on other serotonergic components.

The gene coding for the monoamine oxidase A (MAOA) has been shown to contain a variable number of tandem repeats (VNTR) polymorphism in the promoter region. The 30 base pair (bp) repeated sequence is present in 3, 3.5, 4 or 5 copies, and alleles with 3.5 or 4 copies are transcribed 2–10 times more efficiently than those with 3 or 5 copies (Sabol et al. 1998). Studies investigating the possible association of this MAOA–uVNTR polymorphism and suicidal behaviour have yielded inconsistent results. Several groups found no significant differences in genotype or allele distribution between subjects with suicidal behaviour and comparison groups (Kunugi et al. 1999; Ono et al. 2002; Huang et al. 2004b), whereas one investigation delivered a significant association (Ho et al. 2000). Ho and colleagues examined this VNTR and the Fnu4H1 polymorphism in a sample of patients suffering from bipolar affective disorder. They found the VNTR variant to be associated with a history of suicide attempts, especially in females. The Fnu4H1 restriction fragment length polymorphism (RFLP) only showed significant differences in allele frequencies for female subjects, but not in the total sample. Another study showed a strong association between the high activity-related EcoRV allele and depressed suicide in male subjects, but not in females or the total sample (Du et al. 2002). As discussed above, strong lines of evidence also indicate that the MAOA gene is involved in impulsive/aggressive behaviours (af Klinteberg et al. 1987; Klinteberg et al. 1987). In this context, Courtet et al. (2005a) showed that genetic variation was associated with violent methods of suicide attempt among men. The dimorphic nature of the results is not surprising, due to the location of the MAO genes on the X chromosome, and probably represents an important explanation of the higher incidence of completed suicide among men (Du et al. 2002). Gender differences may also be caused, because MAOA gene expression is controlled not only genetically, but also epigenetically in females, but not in males (Pinsonneault et al. 2006). Future studies will likely better clarify the role of genetic variation in relation to suicidality, probably in the context of other genes.

The serotonin transporter (5-HTT, SLC6A4) is located on the presynaptic membrane of serotonergic neurons and is another key serotonergic regulator, which acts by removing the serotonin released into the synaptic cleft. Carrier-facilitated 5-HT transport into, and release from the pre-synaptic neuron, which are essential for the fine tuning of serotonergic neurotransmission, directing the magnitude and duration of postsynaptic neural responses. Alterations (reduction) of 5-HTT binding in suicidal individuals, appeared related mainly to the PFC, and were originally observed as a reduction of pre-synaptic binding of antidepressant imipramine (Stanley et al. 1982). Genetic variation in this gene has been, and continues to be, massively studied in the context of various forms of psychopathology (Serretti et al. 2006). The amounts of 5-HTT present at the pre-synaptic neuron is regulated in part at the transcriptional level, involving a now well-studied polymorphism in the promoter region, located about 1 kilobase upstream of the transcription start site, often termed as the 5-HTT linked promoter region (5-HTTLPR) (Heils et al. 1996). 5-HTTLPR is located in a repetitive sequence, which is polymorphic by insertion/deletion of 44 basepairs, resulting in short (S) and long (L) alleles, respectively, whereby the S-allele mediates reduced 5-HTT transcription (Heils et al. 1996). This results in reduced serotonin uptake and reduced serotonin responsivity at the nerve synapse, as well as reduced levels of 5-HIAA in the CSF.

A number of studies have been performed in relation to suicide and suicide-related characteristics, with some degree of inconsistency between the separate/independent studies (Courtet et al. 2005b; Currier and Mann 2008). The most studied polymorphism is 5-HTTLPR, which has been shown to affect emotional regulation in different contexts, e.g. in the presence of gene–gene interactions (Ebstein et al. 1998) and environmental stress factors (Caspi et al. 2003), which is likely to be of importance for the eminence of suicidal behaviour. An initial meta-analysis conducted by Anguelova et al. (2003) included 12 studies investigating the 5-HTT promoter polymorphism. The study sample contained 10 Caucasian populations, one US population and one Chinese sample. In the pooled sample, with a total number of 1168 suicide completers/attempters and 1371 controls, a significant association of the S allele with suicidal behaviour was found (Anguelova et al. 2003). A second meta-analysis, including 18 studies with 1521 suicide attempters or completers and 2429 controls, delivered different results (Lin and Tsai 2004). In contrast to the investigation of Anguelova et al. (2003), Lin and Tsai (2004) found no overall association of 5-HTTLPR alleles with suicidal behaviour in a subsequent meta-analysis. This was also true if only the 15 studies with subjects of Caucasian origin were examined. The authors also compared the allelic and genotype distribution between 190 violent suicide attempters or completers and 733 normal control subjects. Here, they did observe a significant association of the S allele with violent suicidal behaviour, mainly characterized by the use of highly lethal and violent methods, such as hanging or shooting, but not with non-violent suicide (Lin and Tsai 2004). Lin and Tsai concluded that violent suicidal subjects might be a relatively homogenous group, and that patients carrying the S allele are likely to act more impulsive and aggressive, which is the concept of fundamental theory in suicide research (Brent and Mann 2005), as also discussed above. Similarly, we have also showed a higher occurrence of the S allele among suicide attempters with a high medical damage score of ? 2, as evaluated by using the Medical Damage Rating Scale (Beck et al. 1975; Wasserman et al. 2007a). In a meta-analysis covering all 38 studies up to January 2006, with a total of 3096 cases and 5936 controls, the results showed consistent and strong associations, both among all studies as a whole and among subgrouped studies (Li and He 2007). The L allele was lowered in frequency among suicidal individuals at allelic and genotypic levels in European and Asian populations, but ethnicity possibly affected gender differences (a dichotomy used in only three studies). The results were no longer specifically confined to violent suicide attempters, and suicidality subgroups (i.e attempter and/or completers, compared with non-attempters and/or healthy individuals) did not bias the analysis significantly. Furthermore, the results supported the notion that the relation to suicide was independent of psychiatric diagnoses.

Interestingly, further allelic variation in 5-HTTLPR has now been identified, namely a SNP in the inserted L allele, whereby reduced 5-HTT transcription was also further linked with a (new) subvariant of the L allele (L_G), in addition to the S allele (Hu et al. 2006). Incorporation of this expanded level of information has been used for implicating a role for the 5-HTTLPR in childhood trauma-related suicidality (Roy et al. 2007), depression (Zalsman et al. 2006; Frodl et al. 2008) and childhood aggression (Beitchman et al. 2006). More studies will clarify how this additional variant of the 5-HTTLPR polymorphism relates to suicidal behaviour.

5-HT receptor genes are also classical candidate genes for suicidal behaviour. All functions related to 5-HT on mood and behaviour, are exerted through action on the receptors. They represent an apparent target for local regulation of 5-HT neurotransmission at the specific synapse, by alterations in amounts of receptor sites, affinities and levels of intracellular signal propagation. Such functional variations could well be explained by genetic determinants, which may alter expression levels and receptor structure. While genes of the 5-HTR_1A, 5-HTR_1B, 5-HTR_2A, 5-HTR_1D, 5-HTR_1E, 5-HTR_1F, 5-HTR_2C, 5-HTR_5A and 5-HTR₆ have been studied genetically so far, there is a general lack of consistency in the findings to date (Bondy et al. 2006; Rujescu et al. 2007). Among these, the synaptic 5-HTR_1A and 5-HTR_2A receptors have shown the most extensive links to suicidality, based on a variety of neurobiological measurements (Albert and Lemonde 2004; Norton and Owen 2005), among which the post-synaptic 5-HTR_2A was the first to be implicated with suicide, by demonstration of increased binding in the PFC (Stanley and Mann 1983).

5-HTR_1A is located on both pre- and post-synaptic sides of the synapses and has been shown to be involved in neurobiology of depression and anxiety. It performs major tasks in regulating the functions of 5-HT system (Albert and Lemonde 2004; Drago et al. 2007). When being pre-synaptic (autoreceptor), they mediate a short feedback regulation loop of importance in the raphe nuclei in controlling the release of brain 5-HT, involving also gamma-aminobutyric acid (GABAergic) cells, via 5-HTR_2A/2C. Post-synaptic 5-HTR_1A is present in both limbic (hippocampal) and cortical medial prefrontal cortex (mPFC) regions, with a variety of roles in the coordinated functions of these brain regions (Albert and Lemonde 2004; Drago et al. 2007). In relation to suicide and the 5-HT_1A gene, Lemonde et al. (2003) examined a functional C-1019G SNP in the promoter region of the 5-HT_1A gene and found the G allele to be significantly over-represented in a depressed, suicide completer group. However, this type of relationship was not observed in a similarly designed study (Huang et al. 2004a), which nevertheless observed increased 5-HT_1A-binding in PFC of suicides, and associations to other clinical diagnoses. More recently, we showed that the G allele was associated and linked among suicide attempters exposed to high levels of stressful life events, but not among suicide attempters in general, a finding which is congruent with possible action of a gene–environment interaction at this polymorphism (Wasserman et al. 2006b). Moreover, it was discussed, and to some degree shown, that the discrepant results among the three studies may have been expected due to the sample heterogeneities, a common problem in the comparisons between different population-based studies. The investigation of other 5-HTR_1A polymorphisms, Pro16Leu and Gly272Asp, revealed no association with suicidal behaviour in Japanese subjects (Nishiguchi et al. 2002). This lack of association with suicidality for the Pro16Leu polymorphism was replicated by a second Japanese group (Ohtani et al. 2004).

A second gene, the gene for the 5-HTR_1B receptor, was studied by New et al. (2001). They found an association of the G861C SNP with a history of suicide attempts in a subsample of 90 Caucasian patients with personality disorders, but this association did not reach statistical significance in the total sample of 145 patients (New et al. 2001). Ten other studies in different populations did not report any implication of gene variants of the 5-HTR_1B receptor in the susceptibility to suicidal behaviour (Huang et al. 1999; Nishiguchi et al. 2001; Arango et al. 2003; Huang et al. 2003; Pooley et al. 2003; Rujescu et al. 2003c; Turecki et al. 2003; Hong et al. 2004; Stefulj et al. 2004b; Tsai et al. 2004).

Most studies show that the 5-HTR_2A receptor binding sites are elevated in the PFC of suicide victims (Arango et al. 1997). This is also of interest in view of the functional interactions with 5-HTR_1A, particularly this region. Interestingly, 5-HTR_2A was recently shown to have decreased signal transduction in violent suicide attempts (Malone et al. 2007). Most genetic studies investigated the common C102T SNP. A meta-analysis regarding this variant pooled nine studies with 596 suicide completers or attempters and 1003 healthy controls showed the lack of association of this particular 5-HTR_2A polymorphism with suicidal behaviour (Anguelova et al. 2003).

Turecki and colleagues examined variations in seven different 5-HTRs (1B, 1Dα, 1E, 1F, 2C, 5A and 6), in a sample of 106 suicide completers and 120 controls, and found no significant association for any of these receptors with suicide (Turecki et al. 2003). Two recent studies examined the 5-HTR2C polymorphism Cys23Ser and found no significant association between this variant and deliberate self-harm or suicidal behaviour, respectively (Pooley et al. 2003; Stefulj et al. 2004a). Overall, while these results do not provide evidence for a major effect of the examined 5-HT receptor variants on the susceptibility of suicidal behaviour, the studies also had small sample sizes, which may have carried insufficient power for detection of genetic effects, and at least 12 known 5-HT receptors remain to be tested.

In summary, the most interesting genes from a neurobiological point of view, remain 5-HTR_1A and 5-HTR_2A. Future genetics studies should perhaps involve other genes involved in modulation of various function(s) of these receptors, and perhaps novel polymorphisms.

Our present molecular understanding of neurotransmission emanates from the works which followed the discovery of adrenalin, about a hundred years ago (Bennett 2000). Tyrosine hydroxylase (TH) is the rate-limiting enzyme in the conversion of tyrosine, into the array of catecholamine neurotransmitters of dopamine (DA), noradrenaline (NA) and adrenaline, a biochemical subgroup of the monoamines (Levitt et al. 1965). The implication of TH in suicidality comes from the important roles of this entire system in responses to stress. Decreased NA in the brainstem and increased alpha2-adrenergic receptor densities were observed, indicative of a deficient NA system (Ordway et al. 1994). Various studies reported different observations concerning the alterations of TH-levels in the locus coeruleus (LC) of suicide victims. Biegon et al. (1992) described reduced TH immunoreactivity in the LC of suicide completers, while Ordway et al. (1994) reported elevated amounts of tyrosine hydroxylase in the LC of suicide victims, and Baumann et al. (1999) found unaltered levels in depressed suicide patients. Variable expression levels of TH may be controlled by polymorphic variants. Polymeropoulos et al. (1991) described a penta-allelic short tandem repeat in the first intron of the TH gene. We also examined the allelic distribution of this polymorphism and reported a tendency for a low incidence of the TH-K1 allele among suicide attempters, compared to the controls (Persson et al. 1997). Furthermore, we found a significant association between the TH-K3 allele in a subgroup of patients with adjustment disorders and attempted suicide. Another study also observed tendencies for transmission distortion at several alleles in this polymorphism, in relation to severe suicide behaviour in bipolar disorder (De Luca et al. 2008). Interestingly, a previous study observed significantly lower levels of 3-methoxy-4-hydroxyphenylglycol (MHPG), the main metabolite of NA, in TH-K3 allele carriers (Jönsson et al. 1996). In contrast, Giegling et al. (2008) instead investigated two SNPs in the TH-gene among 167 suicide completers, and found no significant associations (Giegling et al. 2008). However, this study was not comparable to the previous ones, since e.g. different polymorphisms were studied, and further clarification is needed.

The catechol-O-methyltransferase (COMT) is a major enzyme involved in the inactivation of the catecholamines dopamine and noradrenalin. Lachman et al. (1996) described a common functional polymorphism in the COMT gene, in which valine (Val) at codon 158 is replaced with methionine (Met). Homozygotes for 158Val (high-activity H allele) have three- to fourfold higher levels of enzyme activity than 158Met homozygotes (low-activity L allele). The Val/Met genotype results in an intermediate COMT activity (Weinshilboum et al. 1999).

Several studies investigated the relationship between the described COMT genotype and violent and/or suicidal behaviour. An initial study found no differences in allele distribution between schizophrenic patients and controls, but a subsample of extremely violent patients was found to be more often homozygous for the L allele (Strous et al. 1997). Other investigations replicated the association of the L allele with aggressive behaviours in schizophrenic patients (Kotler et al. 1999; Strous et al. 2003; Volavka et al. 2004). The association appeared to be expressed particularly among males (Lachman et al. 1998; Nolan et al. 2000), but this effect has not yet been systematically studied. However, other studies failed to show an association of this polymorphism with violent behaviour (Wei and Hemmings 1999; Liou et al. 2001; Zammit et al. 2004) or obtained contrary results (Jones et al. 2001). An investigation of this polymorphism in a sample of German suicide attempters and German control subjects without a lifetime history of psychiatric disorders, measured an-ger-related traits in both groups (Rujescu et al. 2003a). The genotype or allele frequencies did not differ significantly between controls and suicide attempters, but the L allele was over-represented in violent suicide attempters. In addition, a multivariate effect of the COMT genotype on anger-related traits was observed. LL-carriers expressed their anger more outwardly, whereas HH-carriers expressed it more inwardly and reported more state anger, as assessed by the self-report questionnaire. Nolan et al. (2000) examined the genotype distribution of the 158Val/Met polymorphism in a sample of Finnish and US schizophrenic and schizoaffective patients, and found the low-activity allele to be more frequent in male subjects with a history of violent suicide attempts, but not in females. Similar results have been reported in a Japanese sample (Ono et al. 2004). In this case, the high-activity Val/Val genotype occurred significantly less frequently in male suicide completers compared with male controls. The authors concluded that the Val/Val genotype may be a protective factor against suicide in males, which implies that the low-activity Met allele increases the suicide risk. One study failed to detect an association between suicidal behaviour and COMT genotype frequencies in patients who were considered to be at high risk of suicide versus controls, but the possible association was not analysed by gender in this investigation (Russ et al. 2000). The study of Ono et al. (2004) mentioned above also failed to show significant differences in genotype distribution if the results for both genders were combined. The observed sexual dimorphism could be a result of the modulation of neurotransmission and neuronal excitability of catecholaminergic systems by estrogen in females (Balthazart et al. 1996). The results with COMT are thus reminiscent of those observed with MAOA, and both genes are along the same biological pathway, since both COMT and MAO are needed for DA and NA catabolism.

The identification of dopamine (DA) as a neurotransmitter in the brain in the late 1950s, was followed by better understandings of the chemical synapse, the development of the first SSRI as well as treatment of Parkinson's disease with L-DOPA. The DA-system is abnormal in depression, and CSF studies have provided support for a possible involvement of the dopaminergic system in suicidal behaviour. Some investigations showed a correlation between low levels of the dopamine metabolite, homovanillic acid (HVA), and suicidal behaviour (Roy et al. 1986), although other groups failed to detect significant differences. Recently, an inverse correlation between mental energy and dopamine transporters (DAT) in the basal ganglia was observed among suicide attempters by brain imaging (Ryding et al. 2006). Furthermore, depressed patients with a history of suicide attempt showed a diminished growth hormone (GH) response to DA-agonist apomorphine, compared to depressed patients without such a history (Pitchot et al. 1992). This blunted response was also found in non-depressed male suicide attempters compared to non-depressed controls (Pitchot et al. 2001).

The genetic investigations have so far been focused on two of the DA-receptors. A SNP in the 3′-UTR of exon 8 (E8) in the DA-receptor D2 gene (DRD2) was investigated in alcoholics and non-alcoholic controls (Finckh et al. 1997). The E8 A/A genotype was found to be associated with an increased number of suicide attempts, and with increased anxiety and depression scores in the alcoholic group. The DRD2 gene further contains a functionally relevant –141C insertion/deletion polymorphism upstream to exon 1. Ho et al. (2000) observed no association of this DRD2 promoter polymorphism with a history of suicidal behaviour in a sample of unipolar and bipolar patients. Another group found the –141C deletion to be over-represented in alcoholics with suicidality compared to controls, although this association did not remain significant after Bonferroni correction (Johann et al. 2005). Two studies examined the association of a 48 bp repeat polymorphism in the DA-receptor D4 gene (DRD4) with suicide attempts (Persson et al. 1999; Zalsman et al. 2004). Both groups did not find any evidence for an implication of this DRD4 polymorphism in suicidal behaviour. More studies are required.

The observed increase of TH and alpha2-adrenergic receptor densities could be indicative of NA depletion, compensatory to increased NA release. This hypothesis is important with regard to the relation between the NA system and stress response, as severe anxiety or agitation are associated with NA over-activity, higher suicide risk and over-activity of the hypothalamic–pituitary–adrenal (HPA) axis (Ordway 1997; Mann 2003). Post-mortem studies reported fewer NA neurons in the LC (Arango et al. 1996), high NA and reduced alpha2-adrenergic receptor in the PFC (Arango et al. 1993) and increased brainstem levels of TH (Ordway et al. 1994). Furthermore, lowered levels of 3-methoxy-4-hydroxy-phenylglycol (MHPG, a metabolite of NA) has been observed in suicide attempters (Pandey and Dwivedi 2007). Besides these neurobiological abnormalities, the genetics of the NA system in suicide has not been studied to a large extent (Ordway 1997; Pandey and Dwivedi 2007). Sequeira et al. (2004) investigated four variants in the alpha 2A adrenergic receptor gene; three of them were located in the promoter region and showed no differences in allele or genotype distribution. The fourth polymorphism (N251K) was functional, leading to an asparagine to lysine amino acid change. The rare N251K allele was only present in three suicide cases, two homozygous and one heterozygous. The two N251K homozygous subjects were depressed, while no clinical information was available for the heterozygous individual. This result could suggest a possible implication of this variant in the susceptibility to suicide or depression (Sequeira et al. 2004). Further studies are required.

The development of neurons in the growing, and adult nervous system and synaptic plasticity is controlled, in part, by members of the neurotrophin family. Neurotrophic factors are secreted by developing neuronal cells and protect them from apoptosis; those neurons that obtain sufficient amounts of neurotrophins survive. The neurotrophic hypothesis of depression suggests that stress increases the susceptibility to depressive illness via increased HPA axis activation, in turn decreasing neurotrophic factors that are required for hippocampal neuronal survival and function. It is also likely that such interruptions may contribute to the developing of vulnerability for suicide, maybe in relation to early childhood adversity. Indeed, levels of neurotrophins are lowered in suicide. Dwivedi et al. (2003) examined the post-mortem brains of 27 suicide victims and 21 non-psychiatric control subjects, and found a significant reduction of messenger ribonucleic acid (mRNA) levels of brain-derived neurotrophic factor (BDNF) and tyrosine kinase B, its receptor, in both PFC and hippocampus of suicide subjects. The reduction of BDNF mRNA expression was accompanied by a decrease in protein levels of BDNF, suggesting that this neurotrophin may play a role in the etiology of suicidal behaviour. In support of this finding, other studies have further indicated lowered BDNF in suicide, measured in serum, plasma and brain (Dawood et al. 2007; Deveci et al. 2007; Kim et al. 2007b). Less is known about the genetics. Hong et al. (2003) examined a functional BDNF Va166Met polymorphism in a Chinese sample, and found no association of this SNP with mood disorders, age at onset or suicidal behaviour. Interestingly, there is a study by Lang et al. (2005), which describes an association of this BDNF polymorphism with anxiety-related personality traits, in a sample of 343 unrelated subjects of German descent. A study by Perroud et al. (2008) showed further compelling evidence for a genetic effect of BNDF along the lines discussed, i.e. an interaction between BDNF Va166Met and childhood trauma, affecting violent suicide attempts in adulthood. Another genetic component of neurotrophins has also been examined, the low-affinity neurotrophin receptor p75NTR. Kunugi et al. (2004) studied a common missense S205L mutation in p75NTR, whereby the minor allele was under-represented among sucidal, depressed Japansese patients. However, this was subsequently not observed in a sample of suicide attempters with child-onset mood disorders in Canada (McGregor et al. 2007). In summary, neurotrophins represent an interesting new pathway in suicidal causality, which remains to be further explored.

There is evidence that low serum cholesterol levels are associated with a higher risk of violent and aggressive behaviour, although there are some contradictions (Golomb 1998). A disordered cholesterol metabolism may contribute to a serotonergic deficit in the central nervous system, and thus, increase the susceptibility for suicidal behaviour (Brunner and Bronisch 1999). Convincing evidence for the association of violence and low cholesterol levels comes from studies in non-human primates. In juvenile monkeys, dietary cholesterol lowering inhibits the central serotonergic activity (Kaplan et al. 1994). In addition, the low cholesterol-diet monkeys behave more aggressively. An investigation in humans reported a significant correlation between plasma serotonin and cholesterol levels, a population of 100 men with low cholesterol, although this was not observed in a reference population with normal cholesterol levels (Steegmans et al. 1996). Terao et al. (1997) found a significant positive correlation between serum cholesterol and prolactin response to the serotonin agonist m-chlorophenylpiperazine in ten healthy volunteers (Terao et al. 1997). The relationship between a depletion of brain serotonergic activity and impulsive, aggressive behaviour has been demonstrated repeatedly (Coccaro 1989; Lesch and Merschdorf 2000).

Cholesterol is a main component of the lipid bilayer of neural membranes, and experimental reduction of the level of membrane cholesterol was reported to decrease the activity of the serotonin transporter due to a loss of substrate affinity and a reduction of the maximal transport rate (Scanlon et al. 2001). Pucadyil and Chattopadhyay et al. (2004) showed that cholesterol modulates specific ligand binding and binding affinity of the 5-HTR_1A receptor from bovine hippocampus.

An association study of suicidal behaviour with polymorphisms of genes involved in the cholesterol biosynthesis or transport was conducted by Lalovic et al. (2004). The 3-hydroxy-3-methylglutaryl CoA reductase (HMGCR) and the 7-dehydrocholesterol reductase (7-DHCR) are enzymes catalyzing important steps in cholesterol biosynthesis. Mutations in the DHCR7 gene have been shown to cause the Smith–Lemli–Opitz syndrome. Patients affected by this autosomal recessive disorder have abnormal low levels of cholesterol, mental retardation and display aggressive, self-injurious behaviour (Tierney et al. 2001). Variants in the genes coding for the enzyme lipoprotein lipase (LPL) and the low-density lipoprotein receptor (LDLR) are associated with low levels of cholesterol (Gudnason et al. 1998; Nicklas et al. 2000). Apolipoprotein E (apoE) is involved in the transport of cholesterol and lipids throughout the circulation. Lalovic et al. (2004) investigated variants in these five genes in a sample of 305 Caucasian male subjects of French-Canadian origin, 145 suicide completers and 160 controls with no history of suicidal behaviour and no major psychiatric diagnoses. They found no differences in allele or genotype frequencies between the suicides and controls for the HMGCR, DHCR7, LPL and LDLR polymorphisms. The examination of the APOE gene by characterizing the allelic forms E2, E3 and E4, using nomenclature as described by Zannis et al. (1982), also showed no relationship between suicidal behaviour and allele or genotype distribution. The authors also tested the possible association of the polymorphisms examined and violent and impulsive behaviours. Measures of impulsive and aggressive behaviours were available for a subset of 42 suicide cases, and no significant relationship was found between these traits or any of the genes studied. The ATP-binding cassette (ABC) transporter is a less known component, which may be involved in human cholesterol homeostasis, and variants in its gene have been associated with the aggressive traits, but not suicide per se (Rujescu et al. 2000; Gietl et al. 2007). These genetic studies have not yet found evidence of a major role of the investigated polymorphism in the etiology of suicide, although this may very well be due to a lack of sufficient statistical power.

The implication of cholesterol in suicidal behaviour might be due to other factors, and the biological relationship remains puzzling. One clue into the causal pathway may come from the many interactions between blood lipids and RORalpha (orphan Rev-erb nuclear receptor), which has key roles in neuroendocrine homeostasis and circardian regulation (Ramakrishnan and Muscat 2006), and which is a biological target of lithium treatment (Yin et al. 2006). Furthermore, it is also interesting to note that cholesterol is the biochemical precursor for cortisol synthesis.

Hans Selye pioneered the scientific field of stress and hormones in the 1930s (Selye 1936), particularly that of corticosteroid effects on brain function (general adaptation syndrome—GAS), inspired by interaction with the works of Walther B Cannon on adrenal secretion of adrenalin in response to emotions, i.e. sympathetic systems adaptations, defining the terms ‘fight or flight’ response and ‘homeostasis’ (Cannon and de la Paz 1911). GAS was described to have three stages, namely an initial alarm phase (fight or flight), an adaptation phase, building resistance to the stress upon survival of the initial phase, and if the stress duration was sufficiently long/chronic, a third exhaustion stage similar to ‘ageing’ (Selye 1936). This pathway of stress-response and -modulation is nowadays defined as the hypothalamic–pituitary–adrenal (HPA) axis, being receptive to externally or internally perceived stress stimulation, by physical, psychological or inflammatory means, whereby the homeostasis of secreted steroids is altered, affecting e.g. behavioural, cognitive, autonomic, psychological and immunologic functions.

Upon affective or sensory input stimuli, the HPA-axis is activated in a sequential manner (Figure 26.2); (i) release of corticotropin-releasing hormone (CRH) from the paraventricular nucleus (PVN) of the hypothalamus, which is transported by portal vessels to the corticotrophs in the anterior pituitary; (ii) binding to CRH receptor 1 (CRHR1), resulting in preferential release of adrenocorticotropic hormone (ACTH), derived from the pro-opiomelanocortin (POMC)-prehormone, into blood circulation; (iii) stimulation of the adrenal gland cortex to secrete cortisol. Cortisol acts on two types of receptors, glucocorticoid (GR) and mineralocorticoid (MR), throughout many parts the brain, with GRs having the broadest effects, being distributed in the limbic system, frontal cortex, brainstem, pituitary and hypothalamus. A functional response in the feedback regulation at all levels, through the balanced activation of GRs and MRs by cortisol and on CRH-secretion from the PVN by ACTH, is essential for returning to a normal state. Furthermore, CRH itself acts heavily on other neurosystems, directly through hypothalamic–central amygdala connections (CeA), affecting e.g. all monoaminergic systems, in response to stressful situations, e.g. by suppression of 5-HTR_1A in the dorsal raphe nuclei (Meijer and de Kloet 1998) or tryptophan-/5HT-depletion and neurotoxicity, due to activation of the indoleamine 2, 3-dioxygenase pathway, as well as glutamatergic hyperfunction (Leonard 2005; Muller and Schwarz 2007). Thus, the HPA axis/CRH system may act as a central modulator of emotions (amygdala) and memory (hippocampus), with consequences for subsequent executive processing of, and behavioural response to, these parameters (PFC).

An overactive HPA axis is a consistent neurobiological indicator of depression (Roy et al. 1987; Hasler et al. 2004; Bale 2005; Nemeroff and Vale 2005; Swaab et al. 2005), and was the first biological risk factor to be implied in the causality of suicide (Bunney and Fawcett 1965). The HPA axis has been shown to be affected neurobiologically in relation to suicidality, at the level of CRH (Arato et al. 1989; Traskman-Bendz et al. 1992; Brunner et al. 2001; Austin et al. 2003; Merali et al. 2006) and its receptors (Nemeroff et al. 1988; Hiroi et al. 2001; Merali et al. 2004), or at the levels of other HPA components (Lopez et al. 1992; Pitchot et al. 1995; Dumser et al. 1998; Pitchot et al. 2005; Kozicz et al. 2008). Most studies have confirmed that an overactive HPA axis, manifested as e.g. hypercortisolism or failure to suppress cortisol after GR stimulation in the dexamethasone suppression test (DST), often precedes suicide (Coryell and Schlesser 2001; Young 2005; Mann and Currier 2007). There is a high degree of prognostic value of predicting suicide among depressed, by using such measures of HPA responsiveness, particularly when combined with measures of 5-HT activity, i.e. CSF 5-HIAA or cholesterol (Mann et al. 2006; Coryell and Schlesser 2007; Mann and Currier 2007). However, there have also been observations of reductions in HPA axis activity among suicide attempters (Pfennig et al. 2005). As observed initially (Selye 1936), a prolonged state of HPA hyperactivity, e.g. by exposure to chronic stress, is pathological to brain development and function. Less well studied, but nevertheless of equal importance, is that prolonged HPA hypoactivity may have equally detrimental consequences (Raison and Miller 2003). In summary, the HPA axis is dysregulated in suicide (as well as in other psychopathologies), with consequences for cognition, emotions and memory, and a better understanding of these dysregulations will likely come from studying variation in (new/further) key participants of this system.

One such key component is the CRHR1 gene, which is involved in both HPA activation and direct central effects in the CeA. Consequently, antagonists active against CRHR1 are being developed as a novel type of antidepressants (Nielsen 2006), which are likely have applications in the treatment of suicidality. From a pharmacogenomic perspective, it is of importance to have access to the level of individual variation in this gene in context to its functionality, which can be needed for evaluation of treatment efficacy as well as being used in diagnostic tools. Knowledge about dysfunctional CRHR1 variants may also bring about new, highly specific treatment possibilities with inhibitory RNAs (Sah 2006). Surprisingly, little is known about the influence of subtle, ‘natural’ genetic variations in the human CRHR1 gene in these contexts, since only a few studies have been reported to date. Polymorphisms in human CRHR1 have been shown to be associated with depression, treatment efficiency of depression and levels of alcohol intake, in the contexts of stressful life events (Licinio et al. 2004; Liu et al. 2006, 2007; Papiol et al. 2007; Bradley et al. 2008; Wasserman et al. 2008a). So far only our group has investigated the CRHR1 gene in relation to suicide (Wasserman et al. 2008a). We demonstrated association and linkage of an SNP to suicide attempt, among whom most males were depressed. The finding was also confined to individuals who had been exposed to low levels of lifetime stress, and it was proposed that these individuals may carry variants of CRHR1, which were more likely to produce a dysregulated HPA/CRH system (Wasserman et al. 2008a). A subsequent follow-up investigation, with more SNPs in the CRHR1 gene, confirmed and expanded these findings to several non-correlated SNPs (Wasserman et al. 2008b). Interestingly, these studies emanated after we had reported associations between genetic variants in the TBX19 gene and neurotic personality traits. The transcription factor TBX19 is along the same biological pathways as CRHR1, i.e. the HPA axis, since they are both involved in the activation of the POMC gene, of key importance for ACTH production, in the corticotrophs of the anterior pituitary (Wasserman et al. 2006a).

As indicated, the 5-HT system is highly intertwined with the HPA axis (Sullivan Hanley and Van de Kar 2003). Slight changes in one system affect the other, and these interactions will probably be more accounted for in future suicide research. It is interesting that cortisol deficiencies can cause antisocial aggression behaviour in rats upon stress exposure, with loss of the normal 5-HT-related aggression-control in mPFC (Kim and Haller 2007). It is likely that measurement of cortisol levels/HPA-responsiveness is a too crude measurement, towards understanding the full neurocognitive effects, as this also involves CeA and mPFC activities. One common pathological denominator of both HPA hyper- and hypoactivity is however the hippocampal processing of emotion and memories, as both conditions impede long-term potentiation (LTP)/plasticity, involving e.g. neurotrophins such as BDNF. Hippocampus, which is innervated with both CeA (fear) and medial PFC (behavioural inhibition), is highly affected by cortisol, being the part of the brain with highest amounts of GRs, and with biochemistry and gene expression altered in suicide brains (Sequeira et al. 2007). Furthermore, the stress-response related to traumatic memories, which often involves suicidality, has been studied by the investigation of hippocampal LTP processes, in relation to CeA, mPFC and HPA activity (Diamond et al. 2007). Such a systemic neurobiological model should also be implemented in studies of suicidality, in which the HPA axis is given a more central importance.

Genetic population studies. As described, complex neural circuitry with many interacting neurotransmitters and their effectors, each involving polygenic influences, needs to be considered when delineating the genetics of the complex phenotype of suicidal behaviour. Ultimately, all genes expressed in the central nervous system (CNS) represent potential candidate genes for suicidal behaviour, and ideally, identification of genetic vulnerability factors should involve a comprehensive survey of the entire human genome. One such attempt is by using ‘genome-wide’ approaches, whereby the goal is to look at all genetic variation in all genes at a time.

The first genome-wide linkage study of suicidal behaviour used the COGA (Collaborative Study on the Genetics of Alcoholism) sample of alcohol-dependent subjects and their biological relatives (Hesselbrock et al. 2004). Chromosome 2 showed a maximum LOD score of 4.2 for the qualitative phenotype ‘ever tried suicide’ near marker D2S1790. In addition, Hesselbrock et al. (2004) created a ‘suicidality index’ regarding lifetime suicidal thoughts and behaviour and examined this second phenotype. This investigation yielded modest evidence for linkage of suicidality to chromosomes 1 and 3, but these results did not reach statistical significance.

Zubenko et al. (2004) performed a genome-wide linkage survey for suicidal behaviour in 81 families with mood disorders. Significant evidence of linkage was found at cytogenetic location 8p22–p21 for all phenotypes. The marker DXS1047 at location Xq25–26.1 showed significant LOD scores for affected relative pairs with recurrent MDD. The cytogenetic locations for the other markers that reached genome-wide adjusted levels of significance are 2p12 (peak at marker D2S1777), 5q31–q33, 6q12 and 11q25 (Zubenko et al. 2004).

Interestingly, these two studies actually showed concordant signals for a region of chromosome 2, namely 2p12 (with markers being 6.5 mb apart), and this was replicated in a third, independent study of 162 bipolar pedigrees, including suicide attempt as a covariate (Willour et al. 2007). This region now needs to be dissected in a more detailed manner, in order to identify the gene(s) which influence suicidality. 2p12 contains 170 known genes, including the interesting candidates TACR1 (the substance P receptor) and TGOLN2 (a trans-golgi network protein), genes which have been experimentally implicated in depression/anxiety/aggression or completed suicide, respectively (De Felipe et al. 1998; McLean 2005; Sequeira et al. 2006).

The genome-wide approaches used so far only represent an approximation of full genome coverage, and the method has relied heavily on indirect associations between genetic variants through linkage disequilibrium, using e.g. ‘tag-SNPs’. There is tremendous progress in this field, but it is also becoming clear, thanks to the efforts of the HapMap project, that much novel genetic variation remains to be included in the future analyses, in addition to the previous markers. Even when true genome-wide coverage will be available by cheap and fast DNA-sequencing of entire human genomes, the candidate genes approach will be viable due to the need of detailed and focused studies of subsets of genes and proteins. Nevertheless, despite the initial flaws, current genome-wide approaches still have huge potential in the primary indication of new, previously unknown neurobiological pathways, due to its screening/hypothesis-free nature. This approach is only likely to increase in importance, as a tool for progressing our understanding of psychobiological mechanisms in the future.

Gene expression studies. The use of microarray technology, has also made it possible to quantify the abundance of many different mRNAs at once. Once again, while the technology might not be entirely genome-wide in its coverage, it has potential to find complex multipoint patterns/changes in mRNA expression levels. It should also be noted that steady state levels of mRNA is quite a crude measurement, which may not always represent the actual levels of gene expression, due to a variety of other factors, e.g. expression control at levels of mRNA transport or mRNA translation. Furthermore, the tendency of RNA to be degraded easily, as well as with differential rates of decay, e.g. in post-mortem suicide brains (Gwadry et al. 2005), itself presents a technological challenge microarray studies.

Sibille et al. assayed the PFC of depressed suicide victims with an array of 22,000 annotated genes, but found no significant alterations in mRNA levels compared to matched, non-suicidal subjects (Sibille et al. 2004). A Japanese study showed increased levels of 14–3–3 epsilon gene brain expression, a protein implicated in neurogenesis, together with association results for three polymorphisms in this gene, among suicide victims (Yanagi et al. 2005). In one of the most extensive studies to date, differential expression was demonstrated among several genes in the cortical brain regions of suicide victims (Sequeira et al. 2006). Among these genes, detailed analysis was performed of the SSAT gene, with verification using reverse transcription polymerase chain reaction (RT-PCR), as well as by quantification of the protein product, with additional evidence of polymorphic association with suicide (Sequeira et al. 2006). Since the identified SSAT-gene has key roles in the metabolism of polyamines, potent neuromodulators of the stress-response linked to the HPA axis (Gilad and Gilad 2003), the results demonstrate the potential to identify new psychobiological pathways in the causality of suicidal behaviour. The research group further extended the study to the limbic system, which provided confirmatory evidence of a global alteration of GABAergic neurotransmission (Sequeira et al. 2007). Further on, Kim et al. (2007a) has evaluated post-mortem microarray data on the PFC, implicating differential expression among 70 genes in relation to suicide completion, among which, results with PLSCR4 and EMX2 seemed independent of other diagnoses, which may provide entry points for novel pathways. Tochigi et al. (2008) also studied differential gene expression in the PFC, and one of the comparisons found that suicide victims had differential expression of CAD and ATP1A3 genes, as was observed in relation to major depression. This summary shows that experimental gene expression screening at the level of the genome has already showed significant potential for future suicide research, and more is likely to follow, particularly as the methods of this study are being refined and improved (Ernst et al. 2008).

Many more systems are further implicated with suicidality, either directly or through correlated clinical measures.

Modulations of anxiety, as well as certain behaviour, e.g. motor activity and exploration, are the effects of cannabinoids, which is indicated as another pathway involved in suicide (Vinod and Hungund 2006). It involves activation of the HPA axis/CRH system, possibly by multiple routes, e.g. modulation of GABAergic/cholecystokinin (CCK)—containing neurons, glutamatergic transmission, opioid receptors and 5-HT/5-HTR_1A (Viveros et al. 2005). Increased cannabinoid receptor (CB1) binding has been observed in the PFC of depressed suicide victims (Hungund et al. 2004). The genetics of the endocannabinoid system remain to be studied in the context of suicide.

Furthermore, neuropeptide CCK is, by itself, not only implicated in anxiety modulation, but also in suicide, as CCK was elevated in the PFC and CSF in relation to suicide (Harro et al. 1992; Bachus et al. 1997; Lofberg et al. 1998). Moreover, this may be influenced genetically, as promoter polymorphisms in CCK were associated with male suicide behaviour (Shindo and Yoshioka 2005).

Similarly, substantial evidence has accumulated for the involvement of gamma-aminobutyric acid (GABA) system, the principal inhibitory neurotransmitter, as well as the glutamatergic system, in depression and suicidality (Manchon et al. 1987; Pandey et al. 1997; Sundman-Eriksson and Allard 2002; Merali et al. 2004; Choudary et al. 2005; Marazziti et al. 2005; Zhu et al. 2006). GABA is synthesized from glutamate by enzyme glutamate decarboxylase (GAD), and upon release in the synapse, active on two receptors types,A and B. GABA is then catabolized after pre-synaptic reuptake,by GABA transaminase. This metabolic cycle thus involves a balance with glutamatergic transmission, which increases in depression, interactively with 5-HT system/HPA axis modulations (Muller and Schwarz 2007). The indicated components specific for suicidality have so far implicated GABA-A receptor subunits (Merali et al. 2004), but the further mechanisms and genetic involvements remain to be resolved.

The gender differences, related to both depression and suicide, are commonly hypothesized to be affected, in part, by the differential actions of HPA-secreted oestrogen (Fourestie et al. 1986; Leibenluft 1999). Interestingly, Tsai et al. (2003) found support for such possible involvement, by examination of the oestrogen alpha receptor (ESR1) gene in relation to suicide and depression. More studies in this context are expected.

Another psychobiological pathway involves the angiotensin I-converting enzyme (ACE), which is a peptidase involved in e.g. producing bioactive receptor-ligand angiotensin II, as well as in degrading substance P, a 5-HT modulator in the dorsal raphe. Reduced depression has been observed, both in relations to antagonists against ACE, or in ACE-deleted mice mutants. Inter-individual variations in this system have been observed in relation to psychopathologies (Arregui et al. 1980; Callreus et al. 2007), and polymorphic variation in the ACE gene has been associated with suicide in one study (Hishimoto et al. 2006), but not in another (Hong et al. 2002).

It has also been observed that among individuals with the Wolfram syndrome, who have homozygous (or are heterozygous carriers of) loss-of-function mutations in the WSF1 gene, often suffered from severe depression, anxiety, impulsive aggression and attempted suicide. Certain genetic variants within this gene may have involvement in suicide (Sequeira et al. 2003; Aluclu et al. 2006), whereas other may not (Crawford et al. 2002). Interestingly, novel mouse models will help to investigate this psychobiological pathway further (Kato et al. 2008).

Finally, preliminary data has further showed association and linkage of genetic variants of novel genes relevant for signal transduction, a sodium channel gene (SCN8A) and a gene for one of the vesicle-associated membrane proteins (VAMP4), in suicide attempter families (Wasserman et al. 2005). The role of these genes remains to be investigated further, with implications for possible cognitive/affective effects.

More systemic approaches are needed when interpreting ‘local’ alterations, since e.g. the prefrontal cortex–hippocampus–central amygdala–HPA tend to function as one operational unit, by the effects of many functional interactions wherein each part contributes in its unique manner. Furthermore, more/novel (biological) endophenotypes, which better reflect these interactions, may be informative (novel, ‘common denominators’), such as long-term potentiation in hippocampus/BDNF levels and other subreceptor intracellular processes. Novel psychobiological pathways identified may shed light on the interacting processes related to stress response, as identified by e.g. genome-wide approaches. The temporal aspects of stress exposure, i.e. duration and time point in development, are also of crucial importance in the shaping of the suicidal brain; therefore, this aspect of environmental exposure must be continuously included. In all this, the influence of the individual's genetic set-up must be determined at all stages of discovery, since it is of major importance for the function of all neurobiological process. To summarize, it seems that suicidal vulnerability appears through changes in the brain's neurobiology, which occur in its development, being influenced by environmental exposures and the genetic set-up of the individual, and that such changes may produce life-lasting psychobiological alterations of brain structure and function, by e.g. epigenetic mechanisms (McGowan et al. 2008). To this end, progress in the field is continuing to evolve rapidly and knowledge emanate with greater detail from the original discoveries described here, enabling the development of better tools for prevention, diagnosis and treatment of suicidality in the future.