Associations between prefrontal γ-aminobutyric acid concentration and the tryptophan hydroxylase isoform 2 gene, a panic disorder risk allele in women

Abstract

Interactions between the central serotonergic and γ-aminobutyric acid (GABA) systems play key roles in the prefrontal cortical regulation of emotion and cognition and in the pathophysiology and pharmacotherapy of highly prevalent psychiatric disorders. The goal of this study was to test the effects of common variants of the tryptophan hydroxylase isoform 2 (TPH2) gene on GABA concentration in the prefrontal cortex (PFC) using magnetic resonance spectroscopy. In this study involving 64 individuals, we examined the associations between prefrontal cortical GABA concentration and 12 single nucleotide polymorphisms (SNPs) spanning the TPH2 gene, including rs4570625 (–703 G/T SNP), a potentially functional TPH2 polymorphism that has been associated with decreased TPH2 mRNA expression and panic disorder. Our results revealed a significant association between increased GABA concentration in the PFC and the T-allele frequencies of 2 TPH2 SNPs, namely, rs4570625 (of –703 G/T) and rs2129575 (p ≤ 0.0004) and the C-allele frequency of 1 TPH2 SNP, namely, rs1386491 (p = 0.0003) in female subjects. We concluded that rs4570625 (–703 G/T), rs2129575, and rs1386491 play a significant role in GABAergic neurotransmission and may contribute to the sex-specific dysfunction of the GABAergic system in the PFC.

Introduction

The neurotransmitter serotonin (5-hydroxytryptamine, 5-HT) is one of the key modulators of the brains circuits for emotion and adaptation to stress. Associations have been found between different neuropsychiatric disorders and the genes that modulate serotonergic neurotransmission, such as the TPH2 gene, which encodes for tryptophan hydroxylase isoform 2, an enzyme exclusively expressed in the brain (Maron et al., 2007, Gutknecht et al., 2009, Roche and McKeon, 2009). TPH, which is the rate-limiting enzyme in 5-HT synthesis, catalyzes the hydroxilation of tryptophan. One of the major target structures of the central serotonergic system is the prefrontal cortex (PFC). Serotonin reuptake inhibitors (SSRIs) can significantly increase brain γ-aminobutyric acid (GABA) concentration (Bhagwagar et al., 2004); this suggests that altering serotonin concentration will have a direct impact on GABA concentration. However it should be noted that Bhagwagar et al. (2004) observed GABA concentration in the occipital cortex. Evidence regarding possible effects of SSRIs on GABA levels in the PFC is still missing. Glutamatergic pyramidal neurons and GABA interneurons are the 2 major neuronal populations in the PFC. GABAergic interneurons exert powerful inhibitory control over excitatory efferent projections from the PFC (Feng et al., 2001). The main effect of physiologically released 5-HT on these PFC projections is inhibitory (Puig et al., 2005). Serotonergic axons synapse predominantly on interneurons (Smiley and Goldman-Rakic, 1996), and 5-HT can strongly modulate GABA-mediated inhibitory neurotransmission (Ciranna, 2006). The psychiatric conditions associated with impaired serotonergic neurotransmission are also associated with a dysfunctional GABA system. Schizophrenia, depression, and anxiety have been associated with reduced cortical GABA concentration (Akbarian et al., 1995, Sanacora et al., 1999, Goddard et al., 2001, Hasler et al., 2007a, Hasler et al., 2010a). Specifically, anxiety disorders have been associated with low GABA concentration in the PFC (Goddard et al., 2001, Hasler, 2010, Hasler et al., 2010b). Moreover, the GABA system is the target of novel antidepressants and mood-stabilizing treatments (Krystal et al., 2002). Serotonin reuptake inhibitors that produce antidepressant and anxiolytic effects have been shown to increase the firing rates of GABA interneurons (Zhong and Yan, 2011) and, as mentioned above, increase brain GABA concentration (Bhagwagar et al., 2004). Taken together, these data suggest that the 5-HT–GABA interactions within the PFC play a key role in the prefrontal cortical regulation of emotion and cognition and in the pathophysiology and pharmacotherapy of highly prevalent psychiatric disorders.

Human variants of TPH2 have been related to anxiety-related traits (Waider et al., 2011), amygdala reactivity (Brown et al., 2005, Canli et al., 2005), and psychiatric disorders (Walitza et al., 2005, Mossner et al., 2006b, Kim et al., 2009). However, the molecular pathways that are affected by the TPH2 variants and that result in the risk of psychiatric disorders have not yet been elucidated. Low cortical GABA concentrations have been proposed as the most promising imaging endophenotype in mood and anxiety disorders (Hasler and Northoff, 2011).

We hypothesized that variations within the TPH2 gene significantly influence prefrontal cortical GABA concentration, as determined by magnetic resonance spectroscopy (MRS). We examined the associations between GABA concentration in the medial PFC and 12 single nucleotide polymorphisms (SNPs) spanning the TPH2 gene, including rs4570625 (–703 G/T SNP). Moreover, based on sex differences in 5-HT (5-HT transporter and 5-HT_1A and 5-HT_2A receptors) and brain GABA concentrations (Cosgrove et al., 2007), we examined the associations in male and female subjects separately. Kim et al. (2009) showed a significant association of rs4570625 with panic disorder (PD) in women. This SNP is potentially functional given recent evidence that the G allele is associated with decreased TPH2 mRNA expression relative to the T allele (Lin et al., 2007, Chen et al., 2008). The samples were genotyped with a haplotype capture array developed and described previously (Hodgkinson et al., 2008). Conceivably, our results may help to elucidate the molecular mechanisms that influence the serotonergic regulation of GABAergic transmission within the PFC.

Methods

Participants

Volunteers were recruited through newspaper advertisements and posters displayed on the National Institutes of Health-Bethesda campus under a protocol that was approved by the Institutional Review Board of the National Institute of Mental Health. All participants were evaluated for psychiatric diagnoses by an unstructured clinical interview with a psychiatrist as well as with a Structured Clinical Interview of the Diagnostic and Statistical Manual of Mental Disorders-IV and the Structured Interview Guides for the Hamilton Depression Rating Scale and Montgomery–Åsberg Depression Rating Scale. The clinical evaluation involved a physical examination, electrocardiography, and laboratory tests, including kidney and liver function tests, hematologic profiling, thyroid function test, urinalysis, and toxicology (drug screen). Exclusion criteria were current medical or neurological disorders, pregnancy, smoking and/or substance abuse. Patients were not under psychotropic medication for at least 4 weeks prior to scanning. All participants gave informed consent, and the study was approved by the Combined Neuroscience Institutional Review Board at the National Institutes of Health (Hasler et al., 2007b).

The study population consisted of 64 individuals from the following 4 diagnostic groups: healthy controls (N = 20; 15 women; age [mean ± SD], 36.9 ± 13.8 years), individuals with a current episode of major depressive disorder (N = 19; 12 women; age, 31.5 ± 9 years), individuals with remitted major depressive disorder (N = 16; 12 women; age, 40.8 ± 11.7 years), and individuals with PD (N = 9; 8 women; age, 33.8 ± 12.8 years). In individuals with PD, mean score for the Panic Disorder Severity Scale (PDSS) was 7.15 ± 3.16 [mean ± SD], means score for the Panic Symptom Scale (PSS) was 18.5 ± 6.02 [mean ± SD] and mean score for the Hamilton anxiety scale (HAMAS) was 10.08 ± 6.42 [mean ± SD].

Magnetic Resonance Spectroscopy

We used in vivo proton MRS to assess the cerebral GABA concentration. Participants underwent scanning in a single session with a 3-T whole-body scanner and a transmit-receive head coil (General Electric Medical Systems, Milwaukee, WI) that was capable of providing a homogeneous radio frequency field and spectroscopic measurements of the PFC.

Proton MRS spectra were acquired in a voxel (3 × 3 × 2 cm³) positioned with the posterior edge at 1 mm anterior to the rostrum of the corpus callosum and centered on the midline in the horizontal plane and on the bicommissural line in the sagittal plane. This voxel included portions of the perigenual anterior cingulate cortex and the adjacent frontal polar cortex (i.e., portions of Brodmann areas 24, 32, and 10) (Hasler et al., 2010a).

The unedited part of an interleaved position resolved spectroscopy sequence-based J editing method (Sailasuta et al., 2001, Hasler et al., 2007b, Hasler et al., 2010a) was used to measure the concentrations of the N-acetylaspartate, choline, and creatine metabolites, and the spectrum resulting from the subtraction of the unedited and the edited spectra was used to specifically measure the GABA concentration. Echo time for this sequence was set to 68 ms, and a single scan consisted of 1024 averages at a repetition time of 1.5 s for a total scan time of 25.6 min. The concentrations of GABA, which were referenced to the concentration of creatine, were expressed as mmol/L (mM). In order to reference the spectra to a quantitative standard, the creatine concentration was set to 7.1 mM, which is the average creatine concentration in the gray and white matter reported in the literature. This conventional creatine referencing method, which is well suited for this study because of the extremely low turnover rate of total creatine in the brain (Wyss and Kaddurah-Daouk, 2000), has been validated and described previously (Shen et al., 2002, Sanacora et al., 2003). In the current study, only the GABA-to-creatine ratios were measured. Since the brain metabolism of creatine is unrelated to tryptophan hydroxylase or variations in the TPH2 gene, we believe that it is safe to assume that creatine is not affected by variations in TPH2 SNPs. Interleaved acquisition and careful in vivo and in vitro quality control procedures were employed in order to ensure the quality of the data. No subtraction errors were detected between the edited and non-edited scans.

Genotyping

Peripheral blood was drawn from all 64 participants, and DNA was extracted with standard procedures. Samples were genotyped with an addiction array, which was developed by Hodgkinson and colleagues (Hodgkinson et al., 2008). This array was developed in order to include a panel of markers that are able to provide full haplotype information for the candidate genes of interest in addictive and mood disorders (Hodgkinson et al., 2008). Our array included the following 12 SNPs that span the TPH2 gene: rs4570625, rs10748185, rs2129575, rs1386488, rs1843809, rs1386496, rs2171363, rs1386491, rs6582078, rs1352250, rs1487275, and rs1386483. For all 64 participants, information was available on ancestry informative markers (AIMs). The ancestry of each subject was evaluated separately in the context of the worldwide genetic factor structure. As a result, the AIMs were derived in a consistent way in order to reflect any of the major worldwide ancestry factors. We used 186 AIMs to evaluate these individuals for the population stratification of GABA concentration. Each AIM was a SNP of a known allele frequency in HapMap reference populations, and a 6-factor solution was used to estimate the ancestry with Structure version 2.0 (Falush et al., 2003), as described previously (Zhou and Wang, 2008).

Statistics

Statistical analyses were performed with the Statistical Package for the Social Sciences (SPSS) for Mac (version 19.0) (IBM Corporation, Armonk, NY). Allele frequencies are provided in Table 1. We calculated pairwise linkage disequilibrium (LD) between the SNPs with the SNPAnalyzer 1.2 (Istech Corp., Goyang City, Korea). In order to describe LD, we used the absolute value of D′. D′ values and statistical significance are provided in Table 2. Multiple linear regressions for the number of minor alleles were performed in order to examine the association between GABA concentration and genotype at each SNP. An additive genetic model was tested by recoding the 3 SNP genotypes as 0 for the homozygote of the common allele, 1 for the heterozygote, and 2 for the homozygote of the uncommon allele. In order to keep type-I error at a rate less than 5%, we computed the Sidak-corrected significance threshold for multiple testing of SNPs in LD with SNPSpD (Šidák, 1967, Nyholt, 2004).

Table 1.

Descriptive statistics of the 12 SNPs.

	SNP	Chromosome	Gene	Location	Polymorphism	Allele Frequency
SNP1	rs4570625	12	TPH2	70618190	G/T	0.73/0.27
SNP2	rs10748185	12	TPH2	70622122	A/G	0.42/0.58
SNP3	rs2129575	12	TPH2	70626340	G/T	0.71/0.29
SNP4	rs1386488	12	TPH2	70630885	A/C	0.79/0.21
SNP5	rs1843809	12	TPH2	70634965	G/T	0.18/0.82
SNP6	rs1386496	12	TPH2	70637057	C/T	0.16/0.84
SNP7	rs2171363	12	TPH2	70646531	C/T	0.47/0.53
SNP8	rs1386491	12	TPH2	70648645	C/G	0.22/0.78
SNP9	rs6582078	12	TPH2	70661158	G/T	0.45/0.55
SNP10	rs1352250	12	TPH2	70684051	A/G	0.52/0.48
SNP11	rs1487275	12	TPH2	70696559	G/T	0.27/0.73
SNP12	rs1386483	12	TPH2	70698761	A/G	0.6/0.4

SNP, single nuclear polymorphism

Table 2.

D′ and p values for all combinations of TPH2 SNPs

	SNP1	SNP2	SNP3	SNP4	SNP5	SNP6	SNP7	SNP8	SNP9	SNP10	SNP11	SNP12
SNP1		<0.0001	<0.0001	0.218	0.173	0.172	<0.0001	<0.0001	<0.0001	<0.0001	0.332	<0.0001
SNP2	1		<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001
SNP3	0.913	0.889		0.368	0.24	0.098	<0.0001	<0.0001	<0.0001	<0.0001	0.031	<0.0001
SNP4	0.344	0.84	0.239		<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	0.017
SNP5	0.421	0.83	0.343	0.827		<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001
SNP6	0.458	0.772	0.52	0.933	1		<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	0.03
SNP7	1	0.846	1	0.743	1	1		<0.0001	<0.0001	<0.0001	<0.0001	<0.0001
SNP8	0.785	1	0.889	1	1	1	1		<0.0001	<0.0001	0.041	0.871
SNP9	0.782	0.842	0.876	0.814	0.767	0.869	0.957	1		<0.0001	<0.0001	<0.0001
SNP10	0.839	0.761	0.932	0.588	0.744	0.68	0.884	0.896	0.892		<0.0001	<0.0001
SNP11	0.086	0.503	0.189	0.325	0.492	0.484	0.798	0.201	0.645	0.928		<0.0001
SNP12	0.459	0.471	0.403	0.311	0.449	0.346	0.791	0.033	0.471	0.781	0.831

SNP, single nuclear polymorphism

$${\alpha }_p=1-{(1-{\alpha }_e)}^{1/M_{\mathit{effLi}}}$$ (i)

The number of independent SNPs (M_effLi) was 6, and the M_effLi-Sidak correction was 0.008. M_effLi was calculated using the following equation:

$$\begin{array}{l}{M}_{\mathit{effLi}}=\sum _{i=1}^{M}f(\mid {\lambda }_i\mid )\\ f(x)=I(x\ge 1)+(x-\lfloor x\rfloor ),x\ge 0\end{array}$$ (ii)

λ_i is the eigenvalue of c identical tests, I(x ≥ 1) is the indicator function and ⌊ x ⌋ is the floor function (for detailed information see Li and Ji, 2005). Possible effects of gender, age, diagnosis, and ancestry on GABA level were calculated with a stepwise linear regression.

Results

We found significant associations between prefrontal GABA concentration and allele frequencies for the rs4570625, rs2129575, and rs1386491 SNPs (p ≤ 0.008; Table 3). Rs4570625 (G/T polymorphism), rs2129575 (G/T polymorphism), and rs1386492 (C/G polymorphism) significantly predicted GABA concentration. GABA levels increased with the number of T or C alleles.

Table 3.

Association analysis of polymorphisms within TPH2 and prefrontal GABA levels. The effects of 7 SNPs were tested in 64 individuals by the linear regression of minor allele occurrences on the magnetic resonance spectroscopy-determined in vivo GABA concentrations. Significant association of the 3 SNPs, rs4570625, rs2129575, and rs1386491, with the GABA levels survived the Bonferroni correction for multiple testing (bold).

Chr/Gene	SNP	Linear regression with GABA concentration (N = 64)
	rs4570625	B = 0.007; P = 0.002; r2 = 0.14
	rs10748185	B = 0.002; P = 0.21; r2 = 0.025
	rs2129575	B = 0.007; P = 0.001; r2 = 0.159
	rs1386488	B = −0.002; P = 0.45; r2 = 0.009
TPH2	rs1843809	B = −0.001; P = 0.76; r2 = 0.002
	rs1386496	B = −0.001; P = 0.59; r2 = 0.005
	rs2171363	B = −0.001; P = 0.65; r2 = 0.003
	rs1386491	B = 0.006; P = 0.008; r2 = 0.109
	rs6582078	B = 0.003; P = 0.13; r2 = 0.036
	rs1352250	B = −0.003; P = 0.15; r2 = 0.032
	rs1487275	B = 0.000; P = 0.86; r2 = 0.000
	rs1386483	B = 0.000; P = 0.86; r2 = 0.000

SNP, single nuclear polymorphism; GABA, γ-aminobutyric acid

There were no significant associations between gender, age, and psychiatric disorder with GABA concentration (p > 0.47, Figure 1). Additionally, none of the ancestry factors predicted GABA level (p > 0.1).

Figure 1.

Relationships between prefrontal cortical GABA concentrations and a) gender, b) psychiatric diagnosis, and c) age. None of these variables was associated significantly with GABA levels.

Number of participants per genotype for rs4570625, rs2129575 and rs1386492 is presented in Table 4. Due to little number of participants with the homozygote of the uncommon allele, participants were divided into two groups based on genotype for both rs4570625 and rs2129575 for further analysis (see also Canli et al., 2005). One group consisted of either one or two copies of the T-allele (T-carriers) and the other group consisted of non-T-carriers. Similar groups were built for rs1386492. One group consisted of either one or two copies of the C-allele (C-carriers) and the other group consisted of non-C-carriers.

Table 4.

Number of participants depending on genotype and gender for rs4570625, rs2129575 and rs1386491.

SNP	Gender	GG	GT	TT
rs4570625	male	7	10	0
	female	24	22	1
	total	31	32	1
rs2129575	male	6	11	0
	female	24	20	3
	total	30	31	3
		GG	CG	CC
rs1386491	male	8	7	2
	female	31	15	1
	total	39	22	3

SNP, single nuclear polymorphism

Given the important sex differences in the central serotonin and GABA systems (Hasler and Northoff, 2011), we tested the associations in male and female participants separately. In an attempt to examine the possible interactions of gender and genotype, we used the SPSS macro MODPROBE developed by (Hayes and Matthes, 2009). The dependent variable was GABA level, the focal predictor was the respective genotype (rs4570625, rs2129575, and rs1386491), and the moderator variable was gender.

The overall fit of the first model (rs4570625) was highly significant (R² = 0.2, F = 5.06, df1 = 3, df2 = 60, p = 0.003). Including the interaction (b = 0.013, SE = 0.005, t = 2.59, p = 0.012) between genotype and gender increased R² significantly (ΔR² = 0.1, F = 6.69, p = 0.01). The moderator variable, gender, did not predict GABA level (b = −0.005, SE = 0.004, t = −1.23, p = 0.22). There was a tendency for genotype to predict GABA level (b = −0.017, SE = 0.009, t = −1.83, p = 0.07). Interestingly, the conditional effects of the focal predictor revealed a highly significant effect of genotype on GABA level in female participants (b = 0.01, SE = 0.003, t = 3.72, p = 0.0004) but not in male participants (b = −0.004, SE = 0.005, t = −0.8, p = 0.43). The T allele of rs4570625 was associated with higher prefrontal GABA concentration in women (Fig. 2).

Figure 2.

Association between GABA levels and the single nucleotide polymorphism rs4570625 that was analyzed separately for male and female subjects. The association between genotype and GABA level became significant for female subjects (p = 0.0004) but not for male subjects (p = 0.43).

The overall fit of the second model (rs2129575) was highly significant (R² = 0.26, F = 6.96, df1 = 3, df2 = 60, p = 0.0004). Including the significant interaction (b = 0.016, SE = 0.005, t = 3.18, p = 0.002) between genotype and gender significantly improved R² (ΔR² = 0.12, F = 10.09, p = 0.002). Genotype (b = −0.021, SE = 0.009, t = −2.32, p = 0.023) significantly predicted GABA level but gender (b = −0.007, SE = 0.004, t = −1.67, p = 0.1) did not. The conditional effects of the focal predictor depending on the values of the moderator variable revealed that the effects for genotype became significant for female participants (b = 0.01, SE = 0.003, t = 4.33, p = 0.0001) but not for male participants (b = −0.005, SE = 0.004, t = −1.17, p = 0.25). The T allele of rs2129575 was associated with increased prefrontal GABA concentration in women (Fig. 3).

Figure 3.

Association between GABA levels and the single nucleotide polymorphism rs2129575 that was analyzed separately for male and female subjects. The association between genotype and GABA level became significant for female subjects (p = 0.0001) but not for male subjects (p = 0.25).

The overall fit of the third model (rs1386491) was highly significant (R² = 0.21, F = 5.16, df1 = 3, df2 = 60, p = 0.003). Including the interaction (b = 0.01, SE = 0.005, t = 2.41, p = 0.004) between genotype and gender significantly improved R² (ΔR² = 0.08, F = 5.83, p = 0.02). Genotype (b = −0.015, SE = 0.009, t = −1.57, p = 0.12) and gender (b = −0.003, SE = 0.004, T = −0.7, p = 0.49) had no effect on prefrontal GABA concentration. The conditional effects of the focal predictor genotype depending on the values of the moderator variable, gender, revealed that the effects for genotype became significant for female participants (b = 0.01, SE = 0.003, t = 3.82, p = 0.0003) but not for male participants (b = −0.002, SE = 0.004, t = −0.44, p = 0.66) (Fig. 4). The C allele of rs1386491 was associated with higher prefrontal GABA concentration in female participants.

Figure 4.

Association between GABA levels and the single nucleotide polymorphism rs1386491 that was analyzed separately for male and female subjects. The association between genotype and GABA level became significant for female subjects (p = 0.0003) but not for male subjects (p = 0.66).

Discussion

To our knowledge, this is the first assessment of associations between genetic variations within the brain-specific serotonin-synthesizing enzyme gene TPH2 and prefrontal cortical GABA concentrations. Our results revealed a significant association between the T-allele frequencies of 2 TPH2 SNPs [rs4570625 (–703 G/T) and rs2129575] and the C-allele frequency of 1 SNP (rs1386491) and increased GABA concentration in the PFC in female subjects. Given the evidence for an association between –703 G/T and TPH2 mRNA expression, the findings of this study suggest that this polymorphism exerts a functional effect on GABAergic neurotransmission in the PFC and that this may contribute to deficits in GABAergic system function that have been reported in patients with PD and other highly prevalent psychiatric conditions.

The T allele of the SNP rs4570625 (-703G/T) has been associated with different psychopathological disorders, such as schizophrenia (Fang et al., 2011), suicidal behavior in major depression (Yoon and Kim, 2009), personality disorder, and attention deficit/hyperactivity disorder (Jacob et al., 2010). Furthermore, the T allele of the SNP rs4570625 (–703 G/T) has also been associated with low anxiety levels (Gutknecht et al., 2007, Reuter et al., 2007). Functional MR imaging studies have shown relatively increased hemodynamic activity in the amygdala in response to negatively and positively valenced emotional stimuli in T-allele carriers than in G-allele homozygous individuals (Brown et al., 2005, Canli et al., 2005). It is conceivable that this differential hemodynamic responsiveness of the amygdala to emotional stimuli is related to the differential GABA concentrations in the PFC. Studies in humans and experimental animals have shown that neural activity in some medial PFC regions reduces outflow from the central nucleus of the amygdala (Quirk et al., 2003, Phelps et al., 2004, Sotres-Bayon and Quirk, 2010). The intrasynaptic GABA concentration is regulated in part by the intracellular GABA concentration (Jackson et al., 2000). Therefore, higher MRS-determined GABA concentrations in both intracellular and extracellular compartments would most likely be associated with increased GABAergic transmission within the medial PFC. If so, this higher inhibitory transmission within the medial PFC might reduce efferent prefrontal cortical transmission to the amygdala, thereby diminishing the cortical modulation of the amygdalar response to emotional stimuli and potentially producing effects like those observed by (Brown et al., 2005, Canli et al., 2005).

The T allele of rs4570625 (–703 G/T) is associated with increased TPH2 levels in the raphe nuclei where it catalyzes the rate-limiting step in 5-HT synthesis (Lim et al., 2007). Thus, the T allele’s presence may result in increased serotonergic neurotransmission to the PFC, which receives serotonergic projections from the dorsal raphe nucleus. Serotonergic axons synapse predominantly on GABAergic interneurons in the PFC (Smiley and Goldman-Rakic, 1996), and the higher 5-HT transmission associated with the administration of selective 5-HT reuptake inhibitors increases both the firing rates of GABAergic interneurons (Zhong and Yan, 2011) and the brain GABA concentrations (Bhagwagar et al., 2004). Thus, the association between the T allele and the increased TPH2 levels in the raphe nuclei suggests a mechanism by which this variant may result in the higher GABA concentrations found herein.

Several studies have shown that genetic variations in TPH2 polymorphisms are associated with a number of psychiatric disorders, such as major depression, suicide, anxiety, and attention deficit/hyperactivity disorders (De Luca et al., 2004, De Luca et al., 2005, Mouri et al., 2009, Must et al., 2009). It has been hypothesized that dysfunction within the serotonergic system plays a major role in anxiety disorders and that both selective 5-HT reuptake inhibitors and monoamine oxidase inhibitors show antipanic, anxiolytic, and antidepressant effects in such disorders. Unfortunately, we did not assess anxiety levels in subjects with past or present major depressive disorder and healthy controls. However, it’s important to note that Hasler et al. (2009) did not find any relationship between PFC GABA concentration and current anxiety levels.

A recent study could show that genetic variations (including rs2129575) in the TPH2 gene may partly account for variations in 5-HT synthesis in the orbitofrontal cortex (Booij et al., 2012). Jacobsen et al. (2012) reported reduced basal and stimulated levels of extracellular 5-HT and increased frontal 5-HT_2A receptors in TPH2 R439H knockin mice. However, they did not find changes in basal extracellular prefrontal GABA levels. Interestingly, a recent study in mice found that brain 5-HT deficiency, resulting from TPH2 inactivation, differentially affects GABAergic systems in limbic regions (Waider et al., 2012). Moreover, genetic studies have revealed an association between variations in the TPH2 gene and PD. Although Mössner et al. (2006a) found no association between the 2 SNPs [rs4570625 (–703 G/T) and rs4565946] of the TPH2 gene and PD in a population from Germany, Kim et al. (2009) found a significantly lower frequency of the T allele of the rs4570625 (–703 G/T) polymorphism in PD patients than in healthy controls. Both these previous studies focused on a single SNP and did not examine the effects of genetic variations along the entire length of the TPH2 gene.

The present study revealed that the C allele of rs1386491 was associated with higher prefrontal GABA concentration. Only a few studies have examined the association between psychiatric disorders and rs1386491. One study reported an association between the TPH2 risk haplotype GGTG (including rs1386491) and borderline personality disorder, suicidal behavior, and aggression score (Perez-Rodriguez et al., 2010). The risk haplotype GGTG (rs2171363, rs1386491, rs6582078, and rs1352250) was first identified by Zhou and colleagues (Zhou et al., 2005). Other studies could not show an association between the TPH2 gene (including rs1386491) and suicidal behavior (Mouri et al., 2009, Must et al., 2009).

One limitation of the present study needs to be mentioned. We did not find any differences in GABA concentration between the control subjects and the other diagnostic groups, which may be because of the small sample sizes of each diagnostic group (Hasler et al., 2009). Nevertheless, the results showed a strong relationship between variations of the TPH2 gene and prefrontal GABA concentration. This may underline the importance of examining molecular mechanisms rather than behavioral phenotypes, given the relatively long road from genes and gene products to psychopathology and the complex and heterogeneous genetic underpinnings of psychological traits (Hasler and Northoff, 2011).

Recent studies have suggested sex differences in the GABA systems and in the neurotransmission of 5-HT (Cosgrove et al., 2007). These molecular differences have also been linked to the pathophysiology of anxiety disorders, such as PD. PD is more prevalent in women than in men (Weissman et al., 1997, Grant et al., 2006). While male PD patients showed abnormal 5-HT transporter availability in various brain regions, there was no such abnormality in female PD patients (Maron et al., 2011). Furthermore, treatment with the 5-HT reuptake inhibitor sertraline showed superior efficacy in women than in men with PD (Clayton et al., 2006). Finally, previous studies have found important sex differences in the association between TPH2 variants and PD (Maron et al., 2007, Kim et al., 2009). In particular, Kim et al. (2009) have found a significant association with rs4570625 in the female subgroup only. These results are in line with our results that showed strong associations between the 3 TPH2 SNPs (–703 G/T [rs4570625], rs2126575, and rs1386491) with prefrontal GABA concentration in female study participants.

One downside to the present study was the relatively low number of male subjects, which may have been a result of the heterogeneous mix of patient groups and healthy subjects. However, it is important to keep in mind that the association between the TPH2 gene and prefrontal GABA level became stronger when male subjects were excluded and that the descriptive data for male subjects goes in the reverse direction than for female subjects. A power analysis revealed that for rs4570625 a minimum of 297 male participants would have been required to have a 80% chance to detect a significant effect of genotype on GABA. Furthermore, for rs2129575 a minimum of 187 and for rs1386491 a minimum of 1263 male subjects would have been required. Hence, we cannot exclude that a higher number of male participants might have led to a reversed main effect in males or may have canceled out. However, it seems unlikely that a larger male sample size would have revealed a main effect for male subjects in the same direction as for female subjects. The results provide evidence for a sex-specific interconnection between the serotonergic and GABAergic systems.

Taken together, the present data provided evidence for a specific molecular mechanism involving the interconnection between the serotonergic and the GABAergic systems in the PFC, which may play important roles in the pathophysiology of highly prevalent psychiatric conditions. Moreover, our results, when combined with previous evidence, suggested that these molecular mechanisms contribute to the sex differences in the epidemiology and clinical characteristics of stress-related psychiatric disorders.