These above described observations paved the way to a better explanatory model of suicide proposed by Mann et al. that involves a predisposing diathesis as a “distal risk factor” which when combined with “proximal risk factors” act as precipitants and increases the risk of suicide (Mann et al., 1999). Distal risk factors include developmental, personality, family, and genetic factors such as childhood adversities, family history of suicide, and impulsive-aggressive personality traits. Proximal risk factors include life events, stress, and psychiatric disorders including substance abuse. Interestingly, suicidal patients differ from non-suicidal patients in distal stressors and additionally, proximal risk factors might serve as triggers for suicidal behavior (Roy et al., 2009).

Mann et al. (1999) proposed a clinical model based on findings from a sample of 347 consecutive patients admitted for mood disorders, psychoses, and other diagnoses. Lifetime suicide attempts, aggressive traits and impulsivity, objective and subjective severity of psychopathology, developmental and family history, and past substance abuse including alcoholism were evaluated. Suicide attempters demonstrated higher scores on subjective depression, more lifetime episodes of major depression/psychoses, higher scores on suicidal ideation and fewer reasons for living when compared to non-attempters. Also, suicide attempters showed higher rates of lifetime aggression and impulsivity, family history of suicide, head injury, and child abuse when compared to non-attempters.

By factor analysis two state factors (psychosis and depression) and one trait factor (aggression/impulsivity) were generated. Using logistic regression, the authors demonstrated that aggression/impulsivity was strongly associated with lifetime suicide attempts whereas psychosis and depression were not significant predictors of lifetime suicide attempt (Mann et al., 1999).

Mann et al. (1999) proposed a hypothetical, predictive stress-diathesis model in which the risk for suicidal actions is not only due to a proximal stressor but also due to a distal stressor—the diathesis. To reach the threshold of suicidal behavior, both components are required. Mann's model conceptualizes diathesis as a dynamic condition of continuous character, underlining that it can vary during the lifespan and that it is not dichotomous.

Melhem et al. (2007) demonstrated that diathesis for suicide is partly heritable and has familial transmissions across offspring. Per Mann, elements of the diathesis including aggressive-impulsive traits, hopelessness, more severe suicidal ideation and cognitive inflexibility could enhance the individual's increased risk of completed suicide.

Since it has been proposed that the diathesis component of Mann's model follows (at least partially) the laws of heritability, we are suggesting that the diathesis represents genetic factors in a neurobiological model of the capability of suicide.

Early life adversities are one of the strongest predictors of suicidal behavior in adulthood (Santa Mina and Gallop, 1998). The high prevalence of adverse life events in suicidal individuals have been confirmed by several sociodemographic studies (Liu and Miller, 2014). A crucial question in these studies as well as in the theoretical models is the definition of “stressful life events” or “stress.” An encyclopedia describes these traumatic events as “rape, combat exposure, sexual, or physical abuse, partner violence or negative life events such as death of family member, divorce, loss of child custody, break up with significant other, bullying, verbal abuse, chronic medical/psychiatric problems such as physical illness, substance abuse, physical pain, feelings of hopelessness, and so on” (Figley, 2012). Stress per Mann's model could refer to psychosocial crises as well as development of a psychiatric disorder (Mann, 2003). In other publications “stress” mostly refers to early life adversities (Currier and Mann, 2008; Mann and Currier, 2010). Since this model attempts to combine two theories and since Joiner's referencing of stress covers a much broader field, the definition of the factor “stress” will be expanded to “Stress and Trauma” including life events such as combat exposure, sexual, or physical abuse over the life span. The stress component corresponds to the epigenome, life-events like the ones above mentioned conveying long-lasting marks on genes, affecting the risk of suicidal behavior over the life span (Labonté and Turecki, 2012; Turecki et al., 2012).

The heart of “The Interpersonal Theory of Suicide” is the idea that the capability to commit suicide is something that has to be acquired throughout the lifespan—being juxtaposed to the natural human survival instinct, which merely presents itself as the individual's fear of death. Joiner claims that even in suicidal individuals, who have already acquired a certain level of capability, it really is a conflict between these two contrasting factors. This sort of capability can be acquired by experiencing repetitive trauma and violence, which can take a wide range of forms, from being a combat fighter (Selby et al., 2010a; Silva et al., 2016) or experiencing proxy-violence and death (as physicians do on a regular basis; Fink-Miller, 2015a,b) or the experience of having attempted suicide or self-harming behavior in eating disorders (Holm-Denoma et al., 2008; Selby et al., 2010b) to being responsible for animal euthanasia (Platt et al., 2010, 2012). Also personality traits such as impulsivity are thought to contribute to the individual's capability. The two major features of increased level of capability are reduced fear of death and increased threshold of pain tolerance, which are thought to interact with suicidal desire (ideation) and together lead up to suicidal death (Van Orden et al., 2010). Joiner dissects suicidal desire into two cognitive factors: thwarted belongingness and perceived burdensomeness. Thwarted or failed belongingness are synonyms for social alienation and loneliness, the experience of being outside and not part of one's family and peer group. The theory argues that this feeling of isolation will contribute vastly to one's suicidal desire. Perceived burdensomeness is the individual's perceived conviction of being a burden to their peer system and family (Joiner, 2005; Van Orden et al., 2012).

With regard to the proposed model, the capability of suicide can be understood in terms of a spectrum rather than in terms of a dichotomous variable. However, in the actual clinical setting, differentiating suicide ideators and attempters from death by completed suicide is one of the most crucial tasks. Besides, there is a growing body of evidence that the transition between suicide attempters and death by completed suicide is not so seamless after all, arguing that they represent distinct entities, and that they carry first and foremost distinguishable features (Giner et al., 2013; Klonsky et al., 2016).

We understand Joiner's capability of suicide as a profoundly biological concept, comprising factors that have been scientifically assigned to genetics, epigenetics and layer of gene function and gene expression. Factors like impulsivity or individual's aggression threshold have been associated with genetic polymorphisms (Oquendo and Mann, 2000; Oquendo et al., 2006); life time experiences like childhood adversities, trauma (Yehuda et al., 2016), and combat exposure (Yehuda et al., 2013; Kaminsky et al., 2015; Sadeh et al., 2016) have been associated with epigenetic changes. Genetic influence on the individual's pain threshold has been conflicting (Eide and Hole, 1993), but an involvement of the serotonin system has been suggested in multiple publications (Lindstedt et al., 2012; Horjales-Araujo et al., 2013; Schaldemose et al., 2014). Exposure to suicide has been associated with a higher suicide risk for the exposed individuals (Nanayakkara et al., 2013; Cerel et al., 2016), but this factor has not been established from a neurobiological perspective (Griffiths and Hunter, 2014).

Suicide is a complex, multifactorial behavior phenotype and a consequence of interactions of genetics and environment. Family, twin, and adoption studies have supported the genetic risk contribution to the psychopathology of suicide. Statistically, it was estimated that the heritability for suicide approximately reached 30–55% for suicidal behavior and ideation (Voracek and Loibl, 2007). Here, we are illustrating the genetics of the capability of suicide with candidate gene studies and genome wide association studies, which contribute to the diathesis component of the Life Span Model of Suicide. In Mann's model, the diathesis is dynamic, whereas our proposed model understands the diathesis as a rather stable trait corresponding to steady polymorphisms in the genome, not only modifying the susceptibility to traumatic events and stress, but also the vulnerability toward psychiatric disorders or psychosocial crises.

Epigenetic regulation is heritable and known to influence gene function by different biochemical modifications other than altering the DNA sequence (Eccleston et al., 2007). Despite their close association with disease pathophysiology and active participation in regulating developmental pathways (Portela and Esteller, 2010; Cantone and Fisher, 2013), epigenetic modification is a relatively new concept in suicide neurobiology (El-Sayed et al., 2012).

DNA methylation based epigenetic modifications confer silencing in gene expression involving a covalent attachment of a methyl group to cytosine residues (Moore et al., 2013). This was found to be true for a wide spectrum of genes, for pathways including the γ-aminobutyric acid (GABA) neurotransmission, HPA axis related stress response system and the polyamine system of the suicide brain (Turecki, 2014a). Part of this DNA methylation process also involves the neurotropic system (Duclot and Kabbaj, 2015). A summary of these methylation-based studies is provided in Table 3.

Evidence from the GABAergic system has shown deficiency in GABAA α1 receptor subunit expression in the frontal cortex of suicide victims (Klempan et al., 2009; Poulter et al., 2010) with a corresponding change in the site-specific methylation at promoter gene for the GABAA receptor (Poulter et al., 2008). A similar downregulation of expression was observed for a set of genes (SMOX, SMS, and SAT1) from the polyamine system in various Brodmann areas (including BA 4, 8/9, and 11) of suicide completers (Fiori et al., 2011; Limon et al., 2016). Proximal promoter analysis of both SMOX and SMS genes did not find significant overall changes in methylation, except for one site (+73) on the SMOX promoter and concomitant decreased SMOX expression level in the same area (Fiori and Turecki, 2010). An overall change in methylation pattern was identified in the SAT1 promoter region of 10 suicide completers, which included three highly polymorphic sites (Fiori and Turecki, 2011). In agreement with the hypothesis, a strong negative correlation was noted between the overall promoter methylation and SAT1 gene transcription. As mentioned above, the presence of three highly polymorphic sites (rs6526342, rs928931, and rs1960264) on SAT1 promoter added an additional layer of haplotype specific gene regulation driven by DNA methylation. Amongst the three sites, rs6526342 showed methylation enrichment in the suicide group albeit no significant correlation with SAT1 expression. The frequent occurrence of a hypermethylated “C” allele in suicide subjects was considered to be associated with downregulated SAT1 expression (Fiori and Turecki, 2011). A possible explanation might be an inaccessible DNA structure affecting the binding of potential transcription inducers (NF-E2, YY1, and AP-1) with the SAT1 gene promoter (Fiori and Turecki, 2011).

Studies of additional polyamine related genes such as ornithine decarboxylase antizymes 1 and 2 (OAZ1 and OAZ2), AMD1 and arginase 2 (ARG2) show expression upregulation (Gross et al., 2013; Limon et al., 2016). Interestingly, the hyperfunctional status for all four genes was found to be associated with promoter hypomethylation. Contrary to the overall methylation status, site specific CpG hypomethylation on AMD1 (CpG9, CpG16) and ARG2 (CpG5-7 and CpG42-44) promoters were found to be highly significant in suicide victims. This hypomethylation status was further supported by the significant negative correlation between two specific CpG sites (CpG9 and 16) and AMD1 gene expression whereas CpG5-7 was found to be negatively correlated with gain of function of ARG2 gene. This was not found to be true for OAZ1 and OAZ2 genes, although site-specific as well as overall differences in methylation between controls and suicide victims were noticed for all four genes investigated (Gross et al., 2013).

An interesting observation on lower BDNF expression was noted in suicide victims (Dwivedi et al., 2003a; Banerjee et al., 2013) with concomitant change in BDNF exon IV based methylation status (Keller et al., 2010). Analysis of four CpG sites (+10, +16, +25, and+28) of BDNF exon IV in the Wernicke's area identified a significant mean methylation difference between suicide and non-suicide subjects (Keller et al., 2010). Following the same pattern, analysis of methylation difference for individual CpG sites on exon IV indicated a significant correlation between suicidal behavior and the hypermethylation status of two CpG site (+10 and +25). Moreover, lower expression of BDNF gene was noticed in those suicide subjects who had earlier shown exon IV promoter hypermethylation (Keller et al., 2010). These observations suggest that epigenetic influence on neurotrophic deficiency results in improper modulation of neural plasticity (Dwivedi, 2009; Duclot and Kabbaj, 2015), a common finding in suicide brains (Dwivedi et al., 2005).

DNA based methylation in dysregulating gene function was further evidenced from studies in astrocytes of suicide subjects (Maussion et al., 2014; Nagy et al., 2015). These studies identified three potential hypermethylated sites on 3′untranslated region (UTR) of TRKB gene in suicide victims, which showed 3′ UTR-mediated epigenetic regulation on compromised TRKB.T1 isoform expression (Maussion et al., 2014). Structural characterization of TRKB.T1 3′UTR indicated the presence of four CpG sites (CpG6–CpG9) within a span of 150 base pairs. Methylation analysis of these four CpG sites showed significant methylation enrichment in the suicide group compared to the control group; additional statistical analysis established a significant correlation between methylation and expression levels. Contrary to this, earlier reports on decreased TRKB.T1 expression has shown a different mechanism of DNA methylation mediated regulation in BA 8/9 of suicide subjects which involved two specific hypermethylated CpG sites on TRKB promoter (Ernst et al., 2009).

Addressing the epigenetic influence on suicide neurobiology remains incomplete without discussing the functional involvement of the HPA axis. In a postmortem brain expression and methylation study, hippocampal NR3C1 expression was found to be significantly decreased in 12 suicide victims. Interestingly, DNA methylation analysis (promoter NR3C1 exon 1F) was found to be closely associated with childhood adversity (McGowan et al., 2009). A recent report has identified similar site-specific DNA methylation changes related to three non-coding glucocorticoid (GR) transcript variants (1_B, I_C, and 1_H) in hippocampal regions of suicide victims with a history of childhood abuse (Labonte et al., 2012). This indicates an integral role of coordinated DNA methylation response to alter the GR functionality, resulting in HPA axis dysregulation in suicide subjects.

Taken together, the studies discussed herein suggest the functional implication of DNA methylation based epigenetic modifications in dysregulating multiple cellular pathways in brain areas primarily involved in neurocognitive and vegetative functions. These cellular abnormalities may give rise to an overall dysfunctional state in information processing and may eventually contribute to the individual's capability of suicide (Turecki, 2014b).

Postmortem findings of gene expression and gene function studies point to a hyperactivation of the HPA-system, a downregulated serotonin system and decreased levels of BDNF compared to non-psychiatric control subjects or psychiatric patients who died of reasons other than suicide. In this review, we are focusing primarily on postmortem brain studies, since the reliability and the validity of peripheral biomarkers has been questioned before (Blasco-Fontecilla et al., 2013; Niculescu et al., 2015; The Lancet, 2016) and suicide attempters might also confound the variable we are attempting to define: the capability of suicide (Table 4).

Several early studies investigated the association between the HPA-axis and suicide by autoradiographic and ligand binding techniques, describing an overexpression of CRH and a subsequent downregulation of the corresponding receptors (Nemeroff et al., 1988; Merali et al., 2006). Studies focusing on molecular gene expression in post-mortem samples of suicide victims show a similar picture: downregulated CRHR1 receptor (not CRHR2) (Merali et al., 2004), increased pro-opiomelanocortin (POMC) mRNA but no difference in GR (glucocorticoid receptor) mRNA (López et al., 1992); no difference between the expression levels of CRHR1 and CRHR2, but a difference in the CRHR1/CRHR2-ratio (Hiroi et al., 2001) compared to control subjects. More recent studies found significantly decreased GR-α protein and mRNA expression in teenage suicide victims' amygdala and PFC (Pandey et al., 2013), and significantly reduced protein and mRNA expression levels of GR and FKBP5 in the amygdala of suicide victims (Pérez-Ortiz et al., 2013).

A very recent study examined postmortem transcription levels in depressed suicide victims, depressed subjects who died from non-suicidal causes, and subjects without any psychiatric history. Interestingly, depressed suicide victims carried a distinctive transcription profile, different from prior findings: CRH mRNA was significantly increased in the suicide group, CRHR1 and GR failed to show significance but were upregulated compared to the other two groups and CRHR2 showed an insignificant downregulation (Zhao et al., 2015).

The abundant serotonin receptor 5HT_2A has been the focus of neurobiological suicide studies for a long time. The majority of publications reported an increased level of 5HT_2A receptors in the cortical area of suicide victims (Stanley and Mann, 1983; Turecki et al., 1999; Escribá et al., 2004); mRNA expression seems to be increased in the hippocampus area as well (Pandey et al., 2002). A recent study found no significant changes of monoamine-related genes (5HTA1, 5HT2A, MAOA, MAOB) in a group of MDD-suicide subjects compared to non-suicidal MDD subjects and controls without any psychiatric history (Zhao et al., 2015). Of all 14 serotonin receptors, 5HT2C is one of the less examined ones: Pandey et al. compared its mRNA and protein expression in the PFC, hippocampus, and choroid plexus between suicide victims and normal control subjects. They found no significant differences in mRNA expression levels, but higher levels of protein in the PFC of suicide victims (Pandey et al., 2006). Recent publications suggest that a highly edited isoform of the pre-mRNA of serotonin receptor 2C is overrepresented in the brains of suicide victims and is significantly correlated to gene expression levels of associated genes (Niswender et al., 2001; Lyddon et al., 2013; Di Narzo et al., 2014).

A few studies investigating the noradrenergic system in postmortem brain samples of suicide victims demonstrated increased norepinephrine enzyme tyrosine hydroxylase (TH) and increased α₂- and β₂-adrenergic receptors (Pandey and Dwivedi, 2007). As for catechol-O-methyltransferase (COMT), there was only one gene expression study that reported significantly increased COMT mRNA in cortex areas of depressed suicide victims (Du et al., 2014).

In line with functional polymorphisms found in postmortem brain samples of suicide victims, BDNF also shows corresponding changes in gene expression and gene function. Generally speaking, most of the studies were able to find suicide-specific effects, which did not correlate with any underlying condition or psychiatric disorder (Dwivedi, 2010). A significant decrease in both BDNF gene expression and corresponding protein expression has been reported in PFC and hippocampus (Dwivedi et al., 2003a). Decreased gene expression in the hippocampus and PFC but not in the entorhinal cortex have been presented by Karege et al. (2005). The expression of other neurotrophins such as NGF, NT-3, NT-4/5 has been described as decreased as well (Dwivedi et al., 2005; Karege et al., 2005). A recent study comparing depressed subjects to suicide subjects and controls: BDNF and its receptors showed a merely insignificant upregulation in the suicide group (Zhao et al., 2015). Reduced neurotrophins (BDNF and NGF) as well as reduced receptor (TRKB, TRKA) mRNA and protein expression were found in the hippocampus of a sample of suicide victims, compared with a sample of controls who died of other causes (Banerjee et al., 2013). Decreased receptor expression of TRKB in suicide victims has also been reported in several other studies in hippocampus (Dwivedi et al., 2003a) and prefrontal cortex (Dwivedi et al., 2003a; Ernst et al., 2009), incongruent results have been reported for the Wernicke area (Keller et al., 2011; Zarrilli et al., 2014). In line with the above mentioned findings, phosphoinositide 3 (PI 3)-kinase, a key enzyme in the neurotrophin pathway, was found to be decreased in PFC and hippocampus of suicide victims (both mRNA and protein levels; Dwivedi et al., 2008).

Cyclic-AMP response element binding (CREB) is an important transcription factor that interacts with promoters of genes involved in neuronal signaling (Sheng et al., 1990). Its association with BDNF in particular (Finkbeiner, 2000) makes CREB an interesting candidate for suicide studies. Earlier studies found decreased immunoreactivities of both CREB and phosphorylated CREB (Yamada et al., 2003), decreased expression of CREB in postmortem brains of drug-free MDD-patients (Yuan et al., 2010) and decreased CREB protein expression in neutrophils of drug-free MDD patients (Ren et al., 2011). Irrespective of the diagnosis, CREB mRNA and protein expression were found to be significantly decreased in the PFC and hippocampus of suicide victims, compared to subjects without any psychiatric history (Dwivedi et al., 2003b). In a sample of teenage suicide victims, these findings were replicated in the PFC only (Pandey et al., 2007).

Taken together, these findings indicate that changes in the serotonin and the neurotrophin system as well as changes in the HPA axis and CREB are contributing to the capability of suicide on the level of gene expression.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.