Dopaminergic drug effects during reversal learning depend on anatomical connections between the orbitofrontal cortex and the amygdala

Abstract

Dopamine in the striatum is known to be important for reversal learning. However, the striatum does not act in isolation and reversal learning is also well-accepted to depend on the orbitofrontal cortex (OFC) and the amygdala. Here we assessed whether dopaminergic drug effects on human striatal BOLD signaling during reversal learning is associated with anatomical connectivity in an orbitofrontal-limbic-striatal network, as measured with diffusion tensor imaging (DTI). By using a fiber-based approach, we demonstrate that dopaminergic drug effects on striatal BOLD signal varied as a function of fractional anisotropy (FA) in a pathway connecting the OFC with the amygdala. Moreover, our experimental design allowed us to establish that these white-matter dependent drug effects were mediated via D2 receptors. Thus, white matter dependent effects of the D2 receptor agonist bromocriptine on striatal BOLD signal were abolished by co-administration with the D2 receptor antagonist sulpiride. These data provide fundamental insight into the mechanism of action of dopaminergic drug effects during reversal learning. In addition, they may have important clinical implications by suggesting that white matter integrity can help predict dopaminergic drug effects on brain function, ultimately contributing to individual tailoring of dopaminergic drug treatment strategies in psychiatry.

Keywords: OFC; amygdala; bromocriptine; diffusion tensor imaging; dopamine; reversal learning; striatum; sulpiride.

Figure 1

Fronto-striatal-limbic pathways. (A) Seed and waypoint masks that were used for probabilistic tractography, displayed on a MNI-template. (B) The four study-specific anatomical ROIs used for FA and MD data extraction. Colored masks represent the binarized group masks for pathways that were present in at least 75% of all subjects. Abbreviations: dPFC, dorsal prefrontal cortex; OFC, Orbitofrontal cortex.

Figure 2

Relationship between drug effects on BOLD and FA-values, revealed by ROI analyses. (A) Shown are effects of bromocriptine relative to sulpiride on striatal BOLD signal during unexpected outcomes in the reversal learning task [(unexpected—expected rewards) + (unexpected—expected punishments)] (x, y, z = −22, 18, 4, p_{fwe_striatum} = 0.03). These effects were reported previously in Van Der Schaaf et al. (2012) and depended on working memory capacity. (B) Linear relationship between FA values in the OFC-amygdala pathway and the effects of bromocriptine relative to sulpiride (left), bromocriptine relative to placebo (middle) and bromocriptine relative to combined administration of both drugs (right) on striatal BOLD. Only significant effects are shown. ^*p < 0.05 Abbreviations: Pla, placebo; Bro, bromocriptine; Sul, sulpiride; SB = combined administration of bromocriptine and sulpiride.

Figure 3

Relationship between drug effects on BOLD and FA-values, revealed by whole brain analyses. (A) White matter regions showing a linear relationship between FA-values and the effects of bromocriptine relative to sulpiride (left), bromocriptine relative to placebo (middle) and bromocriptine relative to combined administration of both drugs (right) on striatal BOLD. (B) Summed tractography maps of the individual pathways that originate from the FA-region displayed in Figure 3A (left). Probabilistic tractography from this region revealed an extensive network of pathways between the OFC, amygdala and striatum. Image is thresholded to present those tracks that were present in at least 25% of the subjects.