BNST neurocircuitry in humans

Abstract

Anxiety and addiction disorders are two of the most common mental disorders in the United States, and are typically chronic, disabling, and comorbid. Emerging evidence suggests the bed nucleus of the stria terminalis (BNST) mediates both anxiety and addiction through connections with other brain regions, including the amygdala and nucleus accumbens. Although BNST structural connections have been identified in rodents and a limited number of structural connections have been verified in non-human primates, BNST connections have yet to be described in humans. Neuroimaging is a powerful tool for identifying structural and functional circuits in vivo. In this study, we examined BNST structural and functional connectivity in a large sample of humans. The BNST showed structural and functional connections with multiple subcortical regions, including limbic, thalamic, and basal ganglia structures, confirming structural findings in rodents. We describe two novel connections in the human brain that have not been previously reported in rodents or non-human primates, including a structural connection with the temporal pole, and a functional connection with the paracingulate gyrus. The findings of this study provide a map of the BNST's structural and functional connectivity across the brain in healthy humans. In large part, the BNST neurocircuitry in humans is similar to the findings from rodents and non-human primates; however, several connections are unique to humans. Future explorations of BNST neurocircuitry in anxiety and addiction disorders have the potential to reveal novel mechanisms underlying these disabling psychiatric illnesses.

Keywords: Addiction; Anxiety; Connectivity; DTI; Resting state; fMRI.

Fig. 1. The human bed nucleus of the stria terminalis (BNST)

The human brain is shown as an illustration (Gray, 1918) with the BNST highlighted in yellow. For reference, a similar slice is shown for fixed tissue (Mai et al., 2008). The traced BNST mask used for this study is outlined in yellow (right) on a 7T gradient spin echo (GRASE) magnetic resonance image.

Fig. 2

The traced BNST mask is shown in yellow on a template brain (top row) and on four individual participant brains (rows 2 through 5). The anatomical landmarks used to confirm correct placement within known boundaries are shown in green (internal capsule), orange (fornix), and blue (anterior commissure).

Fig. 3

The stria terminalis mask (blue) and BNST mask (red) are shown on a 3-dimensional template brain (left). The group DTI streamlines map, masked by the stria terminalis mask, is shown in green on a 3-dimensional template brain (middle). Each voxel of the stria terminalis mask overlapped with the group streamlines map, demonstrating that tractography streamlines were coursing through the entire stria terminalis. For reference, a histological section of the stria terminalis is shown (right).

Fig. 4

Structural and functional connectivity of the bed nucleus of the stria terminalis (BNST) in the human brain. Voxel-wise connectivity maps shown on a template brain illustrate spatial patterns of structural (A) and functional connectivity (B) within regions identified as having significant likelihood of connectivity with the BNST. Color bars indicate strength of connectivity with lighter colors representing stronger connectivity. Using DTI (n = 82) and rs-fMRI (n = 99) to assess structural and functional BNST connectivity, respectively, a varied pattern of BNST spatial connectivity is demonstrated within each larger target region. Bar graphs show mean connectivity values by region for structural (C) and functional (D) analyses, with error bars representing the standard error of the mean. The vertical dotted line indicates the threshold of significant likelihood of connectivity as determined using bootstrap methods (10,000 permutations, > 95% confidence interval of the mean). Target regions were determined to have a significant likelihood of connectivity with the BNST if 1) the mean connectivity value passed the threshold for significance (structural connectivity > .077, functional connectivity > .117), and 2) at least 50% of individual connectivity values passed the threshold for significance (DTI, n > 40; rs-fMRI, n > 49). Out of 108 regions tested, 17 regions passed both threshold criteria for significant structural connectivity (C) and 11 regions passed both threshold criteria for significant functional connectivity (D). Note: *Amygdala = positive control region in the structural connectivity analysis; **Medial prefrontal cortex = negative control region in the structural connectivity analysis

Fig. 5. Exploratory voxel-wise functional connectivity findings

The results of this voxel-wise analysis were consistent with the results from the target region analysis, identifying large clusters within each of the significant target regions. The voxel-wise analysis also revealed significant functional connectivity within regions which were not identified using the target region approach, including limbic regions (amygdala, subcallosal cortex, parahippocampal gyrus, anterior insula), prefrontal cortex regions (anterior cingulate cortex, medial frontal gyrus, superior frontal gyrus, inferior frontal gyrus, and middle frontal gyrus), posterior cingulate cortex, calcarine fissure, precuneus, and precentral gyrus.

Fig. 6. Differential structural and functional connectivity of amygdala subnuclei

An illustration of the three amygdala subnuclei is shown on a template brain in coronal, sagittal, and axial slices (A). Bar graphs show strength of structural (B, left) and functional (B, right) bed nucleus of the stria terminalis (BNST) connectivity by amygdala subnuclei, with error bars representing the standard error of the mean. The centromedial amygdala had greater structural connectivity with the BNST compared to both the basolateral and superficial subnuclei (p’s < .001) (B, left). BNST functional connectivity was greater for the left and right superficial subnucleus, compared with the left and right laterobasal subnucleus (left: p < .001, right: p = .002) and the left centromedial subnucleus (p = .026) (B, right).