Dopaminergic suppression of brain deactivation responses during sequence learning

Abstract

Cognitive processing is associated with deactivation of the default mode network. The presence of dopaminoceptive neurons in proximity to the medial prefrontal node of this network suggests that this neurotransmitter may modulate deactivation in this region. We therefore used positron emission tomography to measure cerebral blood flow in 15 Parkinson's disease (PD) patients while they performed a motor sequence learning task and a simple movement task. Scanning was conducted before and during intravenous levodopa infusion; the pace and extent of movement was controlled across tasks and treatment conditions. In normal and unmedicated PD patients, learning-related deactivation was present in the ventromedial prefrontal cortex (p < 0.001). This response was absent in the treated condition. Treatment-mediated changes in deactivation correlated with baseline performance (p < 0.002) and with the val(158)met catechol-O-methyltransferase genotype. Our findings suggest that dopamine can influence prefrontal deactivation during learning, and that these changes are linked to baseline performance and genotype.

Figure 1.

Sequence learning performance: relationship of levodopa-mediated changes to baseline values. Correlation between baseline performance (RET_OFF) and the levodopa-mediated changes in learning (RET_ON − RET_OFF). A significant negative correlation (r = −0.58, p = 0.002) was observed between these variables. The solid line is the best linear fit; dashed lines represent the 95% confidence interval.

Figure 2.

Changes in learning-related rCBF with levodopa infusion. Treatment was associated with loss of learning-related deactivation responses (SEQ < CCW) in the ventromedial prefrontal cortex (top) and the posterior insula (bottom). In both regions (arrows), significant deactivation during learning was present in healthy volunteers (NL) and unmedicated patients (PD OFF). These responses were not present in the same patients when they performed the sequence learning task during levodopa infusion (PD ON). Left, Voxel-based display of significant task (SEQ, CCW) by condition (OFF, ON) interaction effects on rCBF (SPM slices depict significant clusters at a threshold of p < 0.001, uncorrected). Right, Post hoc analysis of rCBF data from VOIs corresponding to the clusters with significant interaction effects. Values for each task and condition are presented by box-and-whisker plots. ***p < 0.001, paired t test. Paired rCBF data across tasks for each subject are connected by lines.

Figure 3.

Effect of baseline performance on changes in learning-related neural responses with levodopa treatment. A, Bar graph of learning-related deactivation (SEQ < CCW) in the vmPFC for the patient subgroups with good and bad baseline performance (see Materials and Methods). In the unmedicated state (open bars), the good learners achieved a degree of deactivation that was comparable with healthy controls (shaded bar). In contrast, these responses were minimal in bad learners scanned at baseline. Neither of the two subgroups exhibited significant vmPFC deactivation in the treated state (filled bars). B, Bar graph of learning-related activation (SEQ > CCW) in the precuneus for the good and bad performance subgroups. In this region, levodopa enhanced activation in bad learners, but attenuated this response in good learners. Parallel treatment-mediated behavioral effects were observed in the concurrently recorded learning performance data for the two subgroups (see Results). The dashed lines represent significant interaction effects (2 × 2 RMANOVA). The solid lines represent significant pairwise differences (paired Student's t tests). *p < 0.05, **p < 0.01. Error bars indicate SEM.

Figure 4.

Effect of COMT genotype on changes in learning-related neural responses with levodopa treatment. A, Bar graph of learning-related deactivation (SEQ < CCW) in the vmPFC for patient subgroups with different COMT val¹⁵⁸met genotypes (see Materials and Methods). In this region, levodopa reduced the magnitude of deactivation in valine allele carriers, but enhanced this response in methionine homozygotes. B, Bar graph of learning-related activation (SEQ > CCW) in the cerebellum (lobule VI) for the two genotypic subgroups. In this region, levodopa reduced the magnitude of learning-related activation in valine allele carriers and enhanced in the methionine homozygotes. The baseline and treated conditions are represented by open and filled bars, respectively. The shaded bar represents reference values from healthy subjects. The dashed lines represent significant interaction effects (2 × 2 RMANOVA). The solid lines represent significant pairwise differences (paired Student's t tests). *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Error bars indicate SEM.

Figure 5.

Learning-related deactivation responses in the vmPFC displayed as an inverted-U function. Bands of low and high dopamine at the edges of the curve (shaded areas) represent zones of optimal function in which local deactivation responses (SEQ − CCW < 0) occur during task performance. The data were modeled using a simplex search method (Lagarias et al., 1998) to fit the measured learning-related responses to a Gaussian curve, f(v) = p₁ × exp(−(v)²/p₂) + p₃, where v is the hypothetical regional dopamine level and f(v) is the corresponding activation response. The model was based on two major assumptions: (1) for all subjects, the local dopamine level was assumed to be higher in ON relative to OFF (v_s,ON > v_s,OFF, where v_i,j is the dopamine level for subject i in condition j); and (2) because the valine allele is associated with greater activity of the enzyme in the brain (Chen et al., 2004), carriers of this allele were assumed to have relatively lower dopamine levels at baseline [mean (v_val/val,OFF) < mean (v_val/met,OFF) < mean (v_met/met,OFF)]. The model included two minor constraints: (1) the increment in dopamine level with treatment was of similar magnitude across genotypes [SD(v_i,ON − v_i,OFF)_{iϵ{all subjects}} is minimal]; and (2) the amplitude of f(v) was determined by the range of the measured regional data. This fitting procedure was implemented in Matlab 6.5 (MathWorks).