Oxytocin has received much attention as a pro-social and anxiolytic neuropeptide (1). Its role in trust, social attachment, anxiety, fear conditioning, and amygdala responsiveness to emotional stimuli has been shown (1,2). In human studies, the G-allele of a common variant (rs53576) in the oxytocin receptor gene (OXTR) has been associated with protective properties such as reduced stress response, and higher receptiveness for social support (3-5). In contrast, recent studies suggest a detrimental role of the rs53576 G-allele in the context of childhood maltreatment (6,7). To further elucidate the role of OXTR, gene x maltreatment (GxE) interaction on brain structure and function was investigated, based on the assumptions that (A) the previously described effect of reduced gray matter volumes in limbic areas after childhood maltreatment (8,9) is stronger in GG homozygotes than in A-allele carriers, and (B) maltreatment related amygdala responsiveness to emotional facial expressions (9-12) is stronger in G-allele than in A-allele carriers.N=309 healthy adults underwent structural and functional MRI during a frequently used emotional face-matching task. All subjects were genotyped for OXTR rs53576. Childhood maltreatment was assessed using the Childhood Trauma Questionnaire (CTQ) (13). Reward dependence (RD), a pro-social trait, was assessed using Tridimensional Personality Questionnaire (TPQ) scores (14). Gray matter volumes were investigated by means of voxel-based morphometry (VBM) across the entire brain. MR images were acquired on a 3 T-MR scanner (Gyroscan Intera 3T, Philips, Best, NL) with a 3D fast gradient echo sequence (TFE), TR/TE/FA 7.4 ms/3.4 ms/9°, inversion prepulse every 815 ms, acquired over a FoV of 256 mm (FH, frequency encoding) x 204 mm (AP, phase encoding) x 160 mm (RL, phase encoding, nominal slice selection direction) with cubic voxels of 1 mm edge length, reconstructed to 0.5 mm. The VBM8-toolbox was used for preprocessing the structural images with default parameters within a unified model (15) including DARTEL normalization. Group statistics were calculated using SPM8, CTQ scores were entered as covariate. In addition, to accomodate the skewed distribution of CTQ scores, a categorical model (history of maltreatment yes/no) was used. Amygdala responses to facial expressions were investigated using a paradigm consisting of 4 blocks of a face processing task - trios with faces of the Ekman & Friesen stimulus set (16) must be checked for matching faces - alternating with 5 blocks of a sensorimotor control task, frequently used for imaging genetics (9,17,18). fMRI data were acquired with a single shot EPI sequence, TR/TE/FA 2.1 s/30 ms/90°, 34 slices tilted 25° from the AC/PC line, no gap, matrix 64x64, cubic voxels with 3.6 mm edge length. After exclusions for technical reasons, N=264 datasets were available for fMRI analysis of bilateral amygdala, using a full-factorial model with genotype as between subjects factor and CTQ scores as covariate. While no significant main effects of genotype were detected, structural MRI data revealed a robust interaction of rs53576 genotype and CTQ-scores mapping specifically to the bilateral ventral striatum. GG homozygotes but not A-allele carriers showed strong gray matter reduction with increasing CTQ-scores (Fig. 1). The categorial model (maltreatment y/n) yielded almost identical results. An effect of gender on the GxE (gene x environment) effect was not observed. Concerning fMRI, a main effect of maltreatment mapping to the right amygdala was found, as well as a main effect of genotype: GG homozygotes showed increased amygdala response to angry and fearful faces, compared to A-allele carriers (Fig. 2). No GxE interaction was found. In turn, RD scores from the TPQ showed that lower ventral striatum gray matter volumes were associated with lower reward dependence.

In GG homozygotes, a negative correlation of gray matter volume in the ventral striatum and the degree of childhood maltreatment was detected, while A-allele carriers did not show this detrimental maltreatment effect. Second, amygdala responsiveness to emotional cues is stronger in GG homozygotes than in A-allele carriers.The higher receptiveness for social cues in GG homozygotes may result in a stronger profit from beneficial environments, but less resilience to detrimental effects of childhood maltreatment, while A-allele carriers are less affected by positive or negative aspects of their early environment.

In conclusion, the findings suggest that the G-allele constitutes a vulnerability factor for specifically altered limbic brain structure in individuals with adverse childhood experiences, complemented by increased limbic responsiveness to emotional interpersonal stimuli. While oxytocinergic signaling facilitates attachment and bonding in supportive social environments, this attunement for social cues may turn disadvantageous under early adverse conditions.