*2.2. The GC Receptors*

The HPA axis—through the secretion of cortisol—is the primary hormonal mediator of stress responses (Box 1). GC actions are mediated by two types of receptors: the mineralocorticoid receptor (MR) and the GR. The MR has a 6–10-fold higher affinity for binding GCs than the GR [55]. The MR is basally occupied, while the GR is only activated during periods of stress and during the circadian peak of the HPA axis activity. The MR is important in terms of the time of day of HPA axis activity and the regulation of basal circadian and ultradian rhythms. Whereas the GR is widely found in the brain, the MR is restricted to the limbic areas [7]. The capability to cope with a TE depends on the balance of these two receptors in the regulation of the stress response [56] as low GC concentrations enhance MR homodimerisation, while GC stress levels induce MR–GR heterodimerisation [57]. The MR has a role in the appraisal of the threat and initiation of the response (Figure 1) [56].

The GR has three domains: the N-terminal transactivating domain (NTD), the DNA binding domain (DBD) (which binds to DNA GC response elements (GRE)), and the C-terminal ligand-binding domain (LBD) [58]. The NTD is the most variable domain, due to alternative translation initiation. There are many GR isoforms which depend on alternative splicing at several points of the *NR3C1* gene. Alternative translation initiation and post-translational modifications add to the functional pool of GC signalling diversity—which has tissue and cell specificity [59,60]. Epigenetic mechanisms which regulate gene transcription and expression have differential functional effects depending both on the gene's region where they occur and on the SNPs' makeup [61]. In addition, the location of the SNPs has specific influences on gene function [62].

The unbound GR is not alone in the cell's cytoplasm, as it is associated with a multiprotein complex, including chaperone proteins and immunophilins [63]. Upon GC binding, this complex conformation changes and the GR becomes hyperphosphorylated and translocates to the nucleus, where it regulates the transcription of various genes directly and indirectly [46]. The GR binds to GRE as a homodimer directing transactivation of target genes. GC-induced gene expression is frequently cell type-specific, which has been shown to be dependent on the accessibility of the GR-binding site. In turn, this is determined by the chromatin structure and DNA methylation, as well as histone acetylation and methylation (HDAC2) (Figure 1) [58]. Bound-GRs also act as monomers, by interacting with other transcription factors, such as activator protein 1 (AP-1) and nuclear factor k (NF-kB), which represses their transcriptional activity and results in less pro-inflammatory effects (Figure 1) [64].

Furthermore, GCs have fast acting non-genomic actions which are mediated by cytosolic GRs and mGRs and use the activity of various kinases, thus adding greater complexity to GC signalling [65]. The effects are strongly dependent on both cell type and the cellular context. The existence of mGRs in the neuronal cells of the hypothalamus, basolateral amygdala, and hippocampus [42] plays an important role in rapid energy mobilisation and memory consolidation under stress, as well as rapidly preparing the cell for the ultimate genomic effects [66,67].

Chaperones and co-chaperones influence GR affinity for GCs. In particular, the co-chaperone of the heat shock protein (hsp)-90 *FKBP5* gene is subject to the transcriptional influence of the GR upon cortisol binding, leading to an increase in its expression, which is associated with GC resistance, serving as an intracellular feedback control of GC actions [68,69].

### *2.3. GR Gene and Other Genes that Regulate the HPA Axis*

The most-studied gene related to the HPA axis is the GR *NR3C1* gene, which is located on chromosome 5q31-32 and is a member of the nuclear hormone receptor superfamily of ligand-activated transcription factors [70]. The human GR gene has nine exons, whereas exon 9 alternative splicing generates two GR isoforms: GRα and GRβ. The first exon is in effect a set of nine exons, which are referred to as A–J (excluding "G"), which are not translated and result in different mRNA, as they are independently controlled by a unique promoter which is located immediately upstream and thus influences the translation start site [71,72]. In the 5- untranslated region of the *NR3C1* gene, there is a cytosine–phosphate–guanine dinucleotides (CpG) island which includes seven of the first exon's set of nine exons: 1D, 1J, 1E, 1B, 1F, 1C and 1H [71] and is a target of DNA differential methylation studies.

Several functional SNPs (Table 1) of the *NR3C1* gene have been extensively studied, namely: *TthIII*I, ER22/23EK, N363S, B*cl*I, and GR-9β [73].

**Table 1.** Single nucleotide polymorphisms (SNPs) of the *NR3C1*, *FKBP5*, and *CRHR1* genes which have been associated with post-traumatic stress disorder (PTSD) or which have altered sensitivity of the hypothalamus–pituitary–adrenal axis.


\* Studied specifically for GR sensitivity or inferred from functional research. † Only with the ER22/23EK; ↑: GR hypersensitivity; ↓: GR resistance; ?: unknown. ‡ Aliases. ¥ Also known as. Abbreviation: dbSNP ID: Single Nucleotide Polymorphism Database identification. Adapted by permission from [10].

The *TthIII*I SNP (rs10052957) is a C to T change in the promoter of the GR gene, and is only functional when associated with ER22/23EK [74].

The ER22/23EK polymorphism (rs6189/rs6190) is located in the transactivation domain and results in altered translation to the GR protein: glutamic acid-arginine (E-R) to glutamic acid-lysine (E-K) [75]. This SNP has been associated with relative GC resistance [76,77], which in turn has been attributed to the reduced transactivating capacity of the GR [78].

The N363S SNP (rs6195; alias rs56149945) results from one nucleotide substitution in codon 363 of exon 2, and the subsequent alteration from asparagine (N) to serine (S) [75]. This SNP is associated with increased sensitivity to GCs in vivo, which was manifested by significantly enhanced suppression of serum cortisol levels after a low dose of dexamethasone [79], which is probably due to increased transactivating capacity [10,76].

The *Bcl*I intronic SNP (rs41423247) which is situated downstream of exon 2, results from a C to G substitution at nucleotide 646, and has been associated with increased sensitivity to GCs [80,81]. This SNP was associated with emotional memory performance in healthy individuals [82], although its action mechanism is not known [10,73].

The 9β (also referred to as A3669G) SNP (rs6198) which consists of a naturally-occurring A to G substitution in the 3- UTR of exon 9β[73], is associated with increased expression and stability of the GRβ isoform of the GR [62], and with GC resistance, probably owing to decreased transrepression [10,83,84].

Other SNPs of the GR gene are associated with PTSD and depression, such as: rs258747 [18,85], rs10482612 [86], rs6191, rs33388 [87], rs6196, and rs10482605 [88].

The *FKBP5* gene, which is located in the human chromosome 6p21.31 [89], has also been extensively studied as it encodes a co-chaperone of hsp-90 in the GR molecular complex, which influences cortisol binding and subsequent conformational changes in the GR and the subsequent translocation to the nucleus [10]. FKBP5 is implicated in the feedback regulation of the GC response, as GCs increase *FKBP5* expression, which in turn decreases GR affinity for GCs. This is considered to be the ultra-short negative feedback loop for GR activity [90,91].

*FKBP5* overexpression is associated with GC resistance [68,69]. This increase in GC resistance has not been associated with higher plasma cortisol levels [92]. Four functional SNPs (Table 1) in the *FKBP5* gene (rs9296158, rs3800373, rs1360780, and rs9470080) have been identified which are associated with GR resistance, in normal, mainly Afro-descent individuals (Figure 2) [93].

**Figure 2.** Representation of the epigenetic modification in *FKBP5*. The single-nucleotide polymorphism rs1360780 which is close to a functional GRE in intron 2 constitutes the genetic predisposition for an increased *FKBP5* transcriptional response to stress. Demethylation with higher cortisol levels is possibly restricted to certain developmental periods (e.g., childhood). The exposure to childhood trauma leads to an increased activation of *FKBP5* due to a reduction in DNA methylation in SNP risk allele carriers. Adapted by permission from [Nature] [94].

The corticotrophin-releasing hormone receptor 1 (*CRHR1*) and 2 (*CRHR2*) genes have also been studied. In addition, variants for*CRHR1* (Table 1), rs110402, rs242924, rs7209436, rs12944712, rs12938031, and rs4792887 have been studied [93,95,96] and some of them demonstrate G × E interactions [97,98].

### *2.4. Epigenetic Regulation of the HPA Axis*

Epigenetics refers to modifiable, potentially heritable, non-nucleotide changes to DNA transcription which are essential for normal cell di fferentiation and neurogenesis [99]. These changes can both be stable over time and dynamically responsive to environmental challenges such as nutritional, pharmacological, physical, and psychosocial changes, providing experience-dependent DNA modulation. Three main mechanisms have been identified for epigenetic gene transcription regulation and expression. The first is the DNA methylation of the cytosine pyrimidine ring of CpG sites, which suppresses gene transcription. It should be noted that these dinucleotides are especially numerous in the promoter regions (CpG islands). However, DNA methylation of the gene body can have the opposite e ffect [100]. The second mechanism consists of post-translational modifications of histone proteins, such as methylation, phosphorylation, and acetylation, which alter DNA that bind to regulatory proteins as well as the availability of chromatin for transcriptional activity. The third mechanism comprises RNAs (including siRNAs, miRNAs, and piRNAs) and long noncoding RNAs, which also regulate gene expression [99,101–104].

These modifications constitute the epigenome, which can record long-lasting influences from the environment and can be inherited [105,106]. These influences from the environment can occur at specific sensitive periods of development, including exposures in adulthood. DNA methylation is the most studied epigenetic mechanism of genes involved in HPA axis regulation, such as *NR3C1* and *FKBP5*. In particular, methylation of *NR3C1* is thought to underlie the programming of the HPA axis function in response to environmental exposures, such as childhood abuse [107,108]. Furthermore, higher cortisol levels demethylate *FKBP5* intron 7 in carriers of the SNP rs1360780 T risk allele possibly in specific developmental periods such as childhood, increasing *FKBP5* expression and GR resistance [94].

A review of the methylation e ffects on the GR gene concluded that early life adversity has been repeatedly shown to be associated with hypermethylation of the non-coding first exons [71], which could impair HPA axis functioning, specifically with the respective decreased GR expression, and predispose the exposed subjects to several psychiatric disorders.

The noncoding RNAs epigenetic mechanism has also been shown to influence the regulation of the HPA axis in response to stress. The miRNA miR-320a interacts with the *FKBP5* SNP rs3800373 C risk allele, leading to increased *FKBP5* expression and GR resistance [109]. Other studies of noncoding RNAs influences on the HPA axis regulation are reviewed elsewhere [110].

In sum, the HPA axis response to stress challenges has multiple levels of regulation which are influenced by variability in genes (SNPs) and epigenome. These e ffects can further expand environmental influences on genetic expression and consequently on endophenotypes and phenotypes [14]. This means that nature and nurture interactions have endless possibilities. This phenomenon can also be viewed from a therapeutic point-of-view, as these interactions can use pharmacotherapy and psychotherapy to environmentally manipulate these genes—by gene, by epigenome, and by environment interactions.

### **3. HPA Axis and PTSD**

As the new DSM-5 classification rightfully implies, PTSD is a disorder of the stress adaptation system, which is initiated by the influence of the TE on a susceptible person [111]. Accordingly, HPA axis dysfunction has been consistently pointed out to constitute the main pathophysiologic mechanism involved in the development and maintenance of PTSD [54,112,113].

Although PTSD only develops after exposure to TEs, other risk factors exist which increase the probability of a TE influencing the development of PTSD [4,6]. Some of these risk factors also influence the HPA axis function, examples being childhood adversity, parental PTSD, or the occurrence of prior TE or PTSD [9,28]. Exposure to these TEs generates anticipatory responses of the HPA axis, which in turn are moderated by SNPs and the epigenome [7,54,114]. Another important moderator between HPA axis functioning and PTSD is the biological sex, even when controlling for confounders. It is well known that women have a greater risk of developing PTSD than men [115,116]. Interestingly, female victims of sexual assault with lower hippocampal volumes showed increased risk of developing PTSD [117].

The most studied regulator of the HPA axis in PTSD is the GR [10]. Patients with PTSD show reduced GC signalling, which has been shown to be associated with increased GR responsiveness or sensitivity [113]. Indeed, hypersensitivity of the GR in PTSD patients has been found in several studies [111,118–121] and has been specifically attributed to decreased GC signalling [113]. In addition, studies which assess learning and memory also argue that PTSD patients have increased CNS GC hypersensitivity [122]. Indeed, cortisol plays an important role in coping with the emotional memories of trauma by enhancing consolidation and by decreasing retrieval and working memory [42].

Accordingly, lower cortisol levels in saliva, blood, or urine have been found in two metanalyses [118,123] in specific samples and assessment moments of the day, although lower cortisol levels were also found to be associated with trauma exposure, especially in the afternoon and evening. Another metanalysis, which included only saliva samples, showed that morning cortisol levels were significantly lower in PTSD patients when compared to controls [124]. Accordingly, a recent metanalysis found lower concentrations of 24-h urinary cortisol in PTSD patients when compared to controls [125]. Furthermore, neurobiological correlates of PTSD symptom clusters have also been found in some studies to show a negative association between cortisol levels and specific symptom clusters, such as intrusion, avoidance, and numbing e.g., [126,127].

On the other hand, CRH overactivity has been pointed out to be a possible model of HPA axis dysfunction in PTSD [54]. Central spinal fluid CRH levels have been found to be increased in patients with PTSD [128–130]. This finding is consistent with decreased GC signalling in PTSD, as GCs are potent inhibitors of CRH release [131]. Consistent with these HPA axis alterations is the concept that PTSD is a fear disorder, characterised by an excessively strong learning of trauma-related fear, or a failure over time to extinguish memories of previous trauma-related fear experiences [132].

Whether these HPA axis alterations are a cause or a consequence of PTSD development, or whether they are TE exposure-related has been a matter of debate in the scientific community [54]. HPA axis-related pre-trauma altered functions have been found to be more consistently associated with PTSD. High pre-deployment GR number, low *FKBP5* mRNA expression, and high glucocorticoid-induced leucine zipper (GILZ) gene mRNA expression are all independently associated with increased risk for a high level of PTSD symptoms six months after deployment to a combat zone [133]. Accordingly, lower hair cortisol concentrations prior to deployment were associated with post-deployment PTSD symptoms related to new stressors [134]. Furthermore, di fferent clinical trajectories after experiencing a TE are also predicted by cortisol secretion as a response to experimental stress challenges, showing lower saliva cortisol change. When assessed prior to exposure to TEs, these trajectories are less-favourable on the whole [134,135]. With regards to peritraumatic HPA axis-related risk factors, lower cortisol levels [121,136] have been found to predict PTSD development, although a metanalysis only found that heart rate was correlated with posterior PTSD symptoms, among the other biological markers studied, including cortisol measured immediately after TE [137].

Imaging studies also link PTSD pathophysiology with the HPA axis. As an example, lower levels of expression of *FKBP5* were associated with smaller hippocampus and medial orbitofrontal cortex in PTSD patients when compared to non-PTSD controls [138]. A systematic review has shown that lower hippocampal volumes are associated with PTSD [139]. Excessive activation of the amygdala on exposure to threat stimuli has also been demonstrated in patients with PTSD [140].

### *3.1. PTSD and GC Regulation Genes SNPs*

Genome wide association studies have failed to find associations between HPA axis-related genes and PTSD e.g., [141]. However, a recent meta-analysis found that SNPs in *NR3C1* and *FKBP5* genes (rs258747 in *NR3C1* and rs9296158 in *FKBP5*) are significantly associated with PTSD [18]. It has been hypothesised that some of the SNPs of the *NR3C1* gene associated with PTSD could be in linkage disequilibrium with previously-identified functional SNPs [85]. With regards *FKBP5*, the risk allele

of the SNP rs9296158 has been previously associated with GR resistance in normal Afro-descent individuals and with GR hypersensitivity in PTSD patients (Figure 2) [93].

Although no main e ffects had been found for any of the previously-identified functional GR gene SNPs in PTSD, the severity of PTSD symptoms and basal plasma cortisol levels were negatively correlated within war veterans with PTSD who are homozygous for the B*cl*I SNP. In this same subgroup of PTSD patients, the authors also found a tendency for an increased response to a test of peripheral GC sensitivity which correlated with higher PTSD symptoms [142]. In cardiac surgery patients, homozygous carriers of the B*cl*I SNP variant allele showed significantly lower preoperative plasma cortisol levels and more traumatic memories six months after surgery and intensive care unit treatment [143]. A significant interaction e ffect was found of haplotype B*cl*I carrier state and childhood trauma on the pre-deployment GR number [133]. In this study, childhood trauma and a pre-deployment high GR number both predicted the subsequent development of a high level of PTSD symptoms. The higher GR number in the PTSD group was maintained after one and six months [10,144].

With regards *FKBP5*, the four functional SNPs which have been identified as being associated with GR resistance in normal, mainly Afro-descent individuals (rs9296158, rs3800373, rs1360780, and rs9470080) have all shown hypersensitivity to GC in patients with PTSD (Figure 2) [10,93]. This finding suggests that other factors influence the functionality of these SNPs. Interestingly, in this study, these SNPs significantly interacted with the severity of child abuse to predict levels of adult PTSD symptoms [93]. In a study on the genetics of substance dependence which also screened for PTSD, the *FKBP5* rs9470080 genotype (TT) was shown to moderate PTSD risk in interaction with childhood abuse in Afro–American participants, but not in the case of European–Americans [145]. Afro–Americans who were carriers of the TT genotype, but with no experience of childhood adversity, demonstrate the lowest risk for PTSD, but inversely they had the highest level if they had experienced childhood adversity. This result suggests that childhood adversity changes *FKBP5* gene functionality, as was the case in the study by Binder et al. [93]. A study of gene expression in 15 patients with PTSD, comparing them with 20 participants without PTSD, all of whom had been exposed to the World Trade Centre attacks, found that *FKBP5* expression was significantly reduced in current PTSD which was predicted by cortisol when entered with PTSD symptom severity in a regression analysis [146]. This study was extended to include SNP analyses of the *FKBP5* gene, as well as a sample of recovered PTSD patients and it was found that any of the four PTSD risk-related SNPs was a negative predictor of *FKBP5* expression, which in turn was related to lower levels of plasma cortisol and higher PTSD symptom severity [147]. Low pre-deployment levels of *FKBP5* mRNA expression were independently associated with the increased risk of high PTSD symptoms when assessed six months after deployment [148]. However, the study found no associations between *FKBP5* SNPs rs1360780 and rs3800373 haplotypes and pre-deployment *FKBP5* mRNA expression or GR number. On the other hand, in a sample of 412 chronic pain outpatients, the *FKBP5* gene SNP rs9470080 was associated with lifetime PTSD and it was found that participants without the risk allele had decreased PTSD risk, even in the presence of high levels of previous trauma exposure [149]. The interaction of genotype and PTSD symptoms associated with specific gene expression patterns suggested the existence of di fferent biological PTSD endophenotypes, which are determined by functional SNPs in the *FKBP5* gene [150]. In this study, PTSD patients carrying the risk allele of the *FKBP5* gene SNP rs9296158 showed GR super-sensitivity with the dexamethasone suppression test, whereas those without the risk allele had lower baseline serum cortisol concentrations. PTSD carriers of this SNP risk allele had lower *FKBP5* mRNA expression, whilst those without PTSD had increased expression. However, in a study of 3890 US service members deployed to Iraq and Afghanistan, the carriers of the most frequent haplotype AGCC tended to have probable PTSD. Furthermore, each wild type variant SNP was associated with probable PTSD. In this study, the authors considered that *FKBP5* could be a risk factor for PTSD. No interactions with lifetime stressful events were found [151].

Klengel et al. [14] found that this diversity is determined by developmental impacts, such as early childhood adversities, which occurred by way of epigenetic mechanisms in the *FKBP5* SNP rs1360780

(Figure 2). This study showed that being a carrier of this *FKBP5* SNP risk allele, together with being exposed to early childhood adversities, both lead to *FKBP5* intron 7 demethylation. Furthermore, a recent meta-analysis found strong evidence of interactions between *FKBP5* genotypes in three SNPs (namely risk allele carriers of the rs1360780, rs3800373, or rs9470080) and early-life stress, which the authors considered could constitute significant risk factors for stress-related disorders such as PTSD [152]. Unfortunately, rs9296158 *FKBP5* SNP, which has been found to be associated with PTSD in a recent metanalysis [18], was not included.

The first study of the associations between *CRHR1* gene and PTSD in adults found that the SNPs rs12938031 and rs4792887 major alleles were associated with post-hurricane PTSD symptoms and that the former was also associated with PTSD diagnosis [96]. A longitudinal study of pediatric injury patients found an association of the *CRHR1* gene SNP rs12944712 with acute PTSD symptoms and their trajectory over time. Results from this study were considered preliminary, as the sample size was small [95]. One study identified that two SNPs (rs8192496 and rs2190242) in the *CRHR2* gene reduced PTSD risk among trauma-exposed female veterans. The minor allele of these SNPs was associated with reduced risk and severity of PTSD symptoms, although these findings have not been replicated [153].

The studies reviewed in this section show that GC regulation genes' SNPs have an important role in PTSD risk and development. Studies have a greater focus on *FKBP5* gene variation, and on the whole, the literature shows that SNP alleles which are associated with HPA axis hypersensitivity are associated with PTSD. G × E interactions have also been shown, particularly in cases with experience of childhood adversity.
