Although imaging genetics studies in PTSD are lacking, a few studies have examined candidate gene associations with the behavioral phenotype of PTSD [11-17]. These studies are consistent with evidence of a genetically mediated vulnerability to PTSD in the context of traumatic stress exposure. Individuals with a family history of PTSD have a 3 to 5 fold relative risk of developing PTSD [18-20], and twin studies suggest the heritability for PTSD is over 30% [21,22]. Growing evidence suggests that the serotonin transporter gene linked polymorphic region (5-HTTLPR) is associated with risk for PTSD. The frequency of 5-HTTLPR short homozygotes was greater in PTSD patients relative to healthy control subjects who were not recruited for the presence of trauma exposure [11]. Recent studies have begun to examine the role of gene-environment interplay in PTSD risk mechanisms [23]. In a study of highly traumatized Rwandan refugees, 5-HTTLPR genotype predicted PTSD risk; individuals homozygous for the short allele were at high risk for developing PTSD regardless of levels of trauma exposure, whereas the other genotypes exhibited a dose-response relationship of the number of lifetime trauma events with risk for PTSD [14]. In hurricane survivors, the expression levels of the serotonin transporter gene were associated with PTSD, but only in the setting of high exposure to stress and low social support [12]. A recent study showed the 5-HTTLPR genotype alone did not predict PTSD, but rather interacted with childhood adversity and adult traumatic events to increase the risk of PTSD, particularly with high levels of exposure to both trauma types [15].

However, the "common disease common variants" hypothesis suggests multiple genes and gene variants (common variants) are likely to influence risk for PTSD (common disease) [24], and therefore these initial intriguing associations are likely only a small part of the story. The effect of individual gene variants may be more precisely characterized by examining phenotypes closer to the biological activity of the gene in the context of PTSD [25,26]. Imaging genetics is one approach that is gaining interest in the assessment of the genetic modulation of neural activity associated with specific behavioral phenotypes [1,26,27]. For instance, variants of the 5-HTTLPR gene have been associated with differential activation in the amygdala [28-35], a region of the brain associated with fear learning [36] and shown to be hyperactive during emotion processing in PTSD [37,38]. For example, patients with PTSD exhibited greater activation for trauma-relevant pictures in the amygdala and ventrolateral prefrontal cortex (PFC), when compared to trauma-exposed controls [37,39-41]. Furthermore, SLC6A4 variants have been linked to alterations in prefrontal activation during cognitive processing [28]. The inferior PFC, particularly the ventrolateral PFC, plays an important role in the cognitive processing of emotionally salient information [42-47]. Genetic mechanisms appear to influence serotonergic pathways related to human fear conditioning [48]. Fear conditioning models have been applied to prominent PTSD symptoms (e.g., hypervigilance and exaggerated fear response to cues of the traumatic event) and proposed to inform neuroimaging and genetics investigations of PTSD (reviewed in [49,50]).

Our goal was to investigate the link between common variants of the serotonin transporter gene (including 5-HTTLPR) and known functional brain differences in PTSD. Our hypotheses followed from the previously demonstrated role of this candidate gene in modulating neural activity in emotion processing regions [51] coupled with the findings of genetic influences in PTSD [11-14] and other anxiety disorders [52,53].

Until recently, the 5-HTTLPR was analyzed as functionally biallelic with Long (L, 16 repeats) and Short (S, 14 repeats) alleles where the S allele leads to lower expression of mRNA and reduced serotonin transporter in membranes. More recently a functionally triallelic classification includes an A/G single nucleotide polymorphism (SNP), rs25531, that is observed predominantly within the L allele (L_A and L_G alleles)[54,55]. In light of reported effects of intragenic SNPs on transcriptional activity, it is important to evaluate not only these promoter polymorphisms, but common sequence variation across the entire gene for association with altered brain function in PTSD. We predicted that variation in the following functional polymorphisms would modulate neural activity in the amygdala and the ventrolateral PFC in patients with PTSD: 5-HTTLPR/rs25531 previously associated with severity of PTSD, rs140701 previously associated with panic disorder [52], and rs16965628 previously associated with obsessive compulsive disorder (OCD) in haplotype analysis [53] and recently reported to exert the greatest relative effect of common SLC6A4 variants on serotonin transporter gene expression in human cell lines [56]. Across all variants, we hypothesized that those previously associated with PTSD or other anxiety disorders and/or a relative decrease in 5-HTT expression would predict increased amygdala and ventrolateral PFC activation [37,40,41,57,58]. Moreover, consistent with GxE models we hypothesized that any genotype-related differences in brain function would be most pronounced in patients with PTSD.

We found the rs16965628 alleles were significantly associated with 5-HTTLPR alleles wherein the G allele of the SNP co-segregated with the S allele of 5-HTTLPR, which has been implicated in PTSD [11-15]. Interestingly, rs16965628 has been reported to exert the greatest relative effect amongst common variants on the level of serotonin transporter gene expression in human cell lines [56]. Its role was elaborated by assaying allelic imbalance in cell lines genotyped by the HapMap consortium. These investigators examined 55 SNPs in the 100 kb window around SLC6A4 to assess their influence on gene transcription. They found that aside from 5-HTTLPR, two SNPs rs16965828 and rs2020933 that are located in the first intron of the gene and highly correlated with each other, made the greatest contribution to the variation in serotonin transporter gene expression [56]. The present results suggest that rs16965628 accounts for a substantial difference in distractor-related activation of the ventrolateral PFC between PTSD and trauma-exposed control groups. The ventrolateral PFC is known to have an important role in the cognitive control and processing of emotionally salient information [42,43,45,46,68,69]. Previously, we reported greater activation in the PTSD group compared to the trauma-exposed control group in ventrolateral PFC during cognitive tasks such as working memory and executive processing [37,39,58]. Findings from our previous investigations [42,45] suggest an engagement of this region both in general emotion processing and in coping with emotional distraction. The observed intermediate phenotype of increased ventrolateral PFC activation during the distraction delay period appears to be related to the GG gentotpye of rs16965628 in patients with PTSD, which shows both increased emotional reactivity and a need for greater allocation of resources to maintain working memory performance in the face of emotional distraction. The observation of a significant association of a SLC6A4 SNP and PTSD (an association that has not previously been reported with a diagnostic phenotype) underscores that an intermediate phenotype approach may be more sensitive and powerful than behavioral measures given that neural circuitry is more proximal to gene effects than to behavior. The findings also highlight the potential value of intermediate phenotypes identified by imaging genetics for the discovery of associations between gene variants and disease.

Since rs16965628 has not been described in the imaging genetics literature, we consider our results in the context of closely associated 5HTTLPR [56]. A limited number of studies have examined the role of 5HTTLPR on the ventrolateral PFC, as most studies have focused on the amygdala. Surguladze and colleagues [34] reported that the S/S group showed greater functional connectivity between the right fusiform gyrus and the right ventrolateral PFC in response to fearful faces. Structural morphology of the ventrolateral PFC is associated with emotion-cognition interaction in carriers of the short allele of 5HTTLPR [70] who exhibit lower 5HT_1A receptor density throughout the cortex [71]. In tasks of social cognition, 5HTTLPR modulates ventrolateral PFC [72]. This evidence considered together with increased ventrolateral activation in PTSD associated with emotion-cognition studies [37,57,58] and conventional symptom provocation studies [40,41,73] places the ventrolateral PFC at the nexus between 5HTTLPR and PTSD. Thus, increased vulnerability to PTSD and other disorders associated with 5HTTLPR genotype may be mediated through ventrolateral PFC engagement.

We find evidence at the trend level that 5-HTTLPR differentially modulates left amygdala activation in S allele carriers with PTSD. Specifically, S allele carriers with PTSD tended toward greater left amygdala activation in response to combat (relative to non-combat) distractors presented during the working memory delay period than trauma exposed controls. However this left amygdala activation difference was not observed between PTSD and trauma exposed control groups with the LL genotype. This finding is related to three lines of evidence showing that (i) 5-HTTLPR modulates threat-related amygdala activity in healthy normal subjects [28-35,74], (ii) heightened task-related amygdala activation in PTSD [37,38], and (iii) 5-HTTLPR may constitute a vulnerability for developing PTSD in the setting of trauma exposure [11]. Whereas initial reports in healthy subjects showed a 5-HTTLPR effect only in the right amygdala [30,31], subsequent reports extended this finding to the left amygdala [28,33]. The overall balance of neuroimaging data in PTSD from the past decade demonstrate greater amygdala activation in PTSD compared to controls [37,38,41,75-81]. These findings are consistent with the amygdala playing a central role in regulating responses to trauma reminders and cues [82]. Indeed, SLC6A4 has been implicated in PTSD, initially with data from biallelic 5-HTTLPR [11], and in more recent follow-up studies with triallelic 5-HTTLPR [11-15]. However, the present results concerning the role of 5-HTTLPR must be considered preliminary given the paucity of S as well as L homozygote's in our sample.

We did not observe the hypothesized association between 5-HTTLPR and right amygdala activation as previously reported in numerous imaging genetics investigations of healthy participants [28-34]. Several explanations may account for this difference. First, our sample contained the confounding effect of race. Second, the behavioral task in most of the prior studies consisted of viewing of fearful faces. This differs from the present working memory task with trauma related distractors that is designed to probe emotion-cognition interactions. Third, previous studies included healthy individuals that were not identified on the basis of trauma exposure and the present study does not compare trauma-exposed participants to healthy, non-exposed subjects. Finally, the threat-related nature of the previous task stimuli may elicit amygdala activation that has a unique association with 5-HTTLPR that is not specific to our working memory task where combat-related images are used as distractors.

We did not find any associations with triallelic 5-HTTLPR as might be suggested by several recent reports of association to PTSD diagnosis in the setting of high lifetime trauma exposure [11-15]. Our data was limited in assessing the effects of triallelic 5-HTTLPR on neural activity due to a lack of subjects possessing the L_AL_A gentoype. Reports of triallelic 5-HTTLPR generally show an interaction effect with the level of lifetime trauma exposure on diagnosis of PTSD whereas the present study was designed to match for level of trauma exposure between the PTSD and control group. Moreover, we are not aware of studies showing effect of triallelic 5-HTTLPR on brain function particularly as further modulated by exposure to childhood trauma. However in the overall group, we found increased left and right amygdala activity and fusiform gyrus activity associated with L_AL_A. These findings are consistent with results of emotion tasks eliciting greater amygdala activation that is differentially affected by the L_AL_A genotype in a normative sample [83] and in major depression [29].

While early imaging genetics studies of 5-HTTLPR assessed only amygdala activity, some recent studies in healthy subjects utilized cognitive attention and emotion processing tasks to show not just modulatation of amygdala, but also frontal cortical activation including the anterior cingulate, dorsolateral PFC, intraparietal sulcus, insula, and other regions [28,84]. We extend these findings by showing that rs16965628, the first tagging SNP downstream of 5-HTTLPR, modulates task ventrolateral PFC activation in PTSD associated with maintaining information in working memory while being distracted by combat pictures. Our findings support the supposition that fMRI data provides us with an intermediate phenotype that is closer to the function of proteins expressed by the candidate gene than a clinical entity. Thus, the definition of a precise intermediate phenotype that is closely linked to the biological function of gene expression is imperative. Core features of PTSD include hypervigilance and re-experiencing symptoms associated with the processing of emotional cues likely to be irrelevant to ongoing task demands, resulting in distractibility and compromised task performance. Emotional stimuli are known to influence behavioral performance on experimental tasks requiring cognitive processing [42,44-47], and therefore brain systems mediating cognitive control of emotion are relevant to PTSD [85]. While imaging phenotypes may be closer to the action of genes compared to behavioral or clinical phenotypes, it is certain that the imaging phenotypes employed in the present study are imprecise and are downstream manifestations of multiple gene systems working together to produce a complex ensemble of brain activity [27].

Based on preliminary nature of our results, the role of rs16965628 in PTSD deserves further investigation. While this SNP has not previously been described in association with PTSD, nor does data from our sample support an association of this SNP with PTSD as a diagnostic phenotype, there is currently insufficient information available to characterize the role of this SNP in PTSD or other anxiety disorders. Given the previous association of 5-HTTLPR with PTSD, a role for rs169656258 as a modifier is consistent with a number of other disorders, most notably cystic fibrosis, a monogenic disease determined by mutations of the cystic fibrosis transmembrane conductance regulator (CFTR) gene [86]. In cystic fibrosis, other genes are required to explain the clinical heterogeneity with the extent of liver [87] and lung [88] involvement not explained by CFTR alone. Conceived as a modifier SNP, the present results suggest that rs16965628 predicts brain activity related to the disruption of cognitive control by emotional or traumatic information in the ventrolateral prefrontal cortex. This type of model is certainly one that deserves to be investigated in PTSD where it is likely that multiple genes might predict onset of PTSD and other genes or SNPs within the causal genes might modify the variability of PTSD in concert with environmental exposures such as lifetime trauma.

It is important to consider the present findings in the broader context of neuroimaging and genetics findings observed in related neuropsychiatric disorders, particularly major depression and anxiety disorders. There is increasing evidence that a common set of underlying mechanisms are operating in depression and PTSD that may explain their shared diathesis [89]. Recent meta-analyses showed consistent patterns of amygdala hyperactivation in major depression [90], social phobia, specific phobia, and PTSD [91]. However, PTSD shows divergent findings when compared to the other anxiety disorders in the rostral anterior cingulate and ventral and dorsal medial prefrontal regions [91]. In these regions specific phobia and social phobia fail to show differences, while the PTSD literature contains evidence of both greater activation [[76], Lanius, 2002 #1016, Morey, 2008 #625, Pannu Hayes, 2009 #654] and lower activation [[92], Shin, 2005 #741, Bremner, 1999 #291, [93]] in ventromedial prefrontal cortex, which may be influenced by a variety of factors including illness chronicity, emotional versus trauma-specific stimuli, and others. Differences in ventrolateral PFC activity have been consistently demonstrated in both PTSD and depression. A meta-analysis of neuroimaging studies using emotional stimuli in depression found increased inferior frontal gyrus and left amygdala activation in response to negative emotional images [94].