- Split View
-
Views
-
Cite
Cite
Po See Chen, Asif Jamil, Lin-Cho Liu, Shyh-Yuh Wei, Huai-Hsuan Tseng, Michael A Nitsche, Min-Fang Kuo, Nonlinear Effects of Dopamine D1 Receptor Activation on Visuomotor Coordination Task Performance, Cerebral Cortex, Volume 30, Issue 10, October 2020, Pages 5346–5355, https://doi.org/10.1093/cercor/bhaa116
- Share Icon Share
Abstract
Dopamine plays an important role in the modulation of neuroplasticity, which serves as the physiological basis of cognition. The physiological effects of dopamine depend on receptor subtypes, and the D1 receptor is critically involved in learning and memory formation. Evidence from both animal and human studies shows a dose-dependent impact of D1 activity on performance. However, the direct association between physiology and behavior in humans remains unclear. In this study, four groups of healthy participants were recruited, and each group received placebo or medication inducing a low, medium, or high amount of D1 activation via the combination of levodopa and a D2 antagonist. After medication, fMRI was conducted during a visuomotor learning task. The behavioral results revealed an inverted U-shaped effect of D1 activation on task performance, where medium-dose D1 activation led to superior learning effects, as compared to placebo as well as low- and high-dose groups. A respective dose-dependent D1 modulation was also observed for cortical activity revealed by fMRI. Further analysis demonstrated a positive correlation between task performance and cortical activation at the left primary motor cortex. Our results indicate a nonlinear curve of D1 modulation on motor learning in humans and the respective physiological correlates in corresponding brain areas.
Introduction
Dopamine is one of the most important neuromodulators associated with diverse brain functions at both physiological and behavioral levels. Clinically, brain diseases associated with dopaminergic dysfunctions, such as Parkinson’s disease and schizophrenia, are paralleled by profound cognitive deficits and deficient plasticity (Brown and Marsden 1990; Morgante et al. 2006; Hasan et al. 2013). Dopaminergic modulation of cognitive functions reveals a complex picture, as it can lead to both facilitation and impairment of task performance, which has been suggested to depend on receptor subtypes, amount of dopaminergic activation, type of plasticity, baseline level of performance, and types or stages of the cognitive task under study (Seamans and Yang 2004; Kuo et al. 2008b; Cools and D’Esposito 2011).
Central to the exploration of the effect of dopamine on cognitive functions so far was the prefrontal cortex (PFC), due to its vast dopaminergic input from the ventral tegmental area and substantia nigra (Williams and Goldman-Rakic 1998; Robbins 2000), primarily with respect to executive functions. Here, opposing effects of dopamine have been suggested, depending on modulation of D1- and D2-class receptors. Computational models discern between a D1-dominated state that favors robust online maintenance of information and a D2-dominated state that is beneficial for flexible and fast switching between representational states (Seamans and Yang 2004; Durstewitz and Seamans 2008). In other words, a D1-dominant state would be beneficial for tasks requiring cognitive stability by focusing on high activity states and increasing the threshold for noise to prevent destabilization of activity patterns, whereas a D2-dominated state would allow for cognitive flexibility in tasks requiring updating of information by transition from different low activity states (Frank 2005). For prefrontal cognitive functions, this proposed model is in accordance with a deleterious effect of D1 but not D2 receptor antagonism on working memory performance (Seamans et al. 1998), which requires holding information robustly in respective memory buffers. In contrast, D2 receptors have been shown to be involved in intentional flexibility, which was enhanced by the application of the D2 receptor agonist bromocriptine (Stelzel et al. 2013). Beyond its relevance for the explanation of dopaminergic effects on executive functions, this model is also applicable for learning and memory. For elementary long-term memory formation and consolidation, which requires stabilization of newly encoded representations, D1 receptor activity has been shown to be of critical importance in animal models (O’Carroll et al. 2006; Bethus et al. 2010; Daba Feyissa et al. 2019; Roughley and Killcross 2019). In contrast, D2 receptor activity enhances reversal learning, which requires flexibility of information processing (Linden et al. 2018; Horst et al. 2019). At the neurophysiological level, it has been shown that both D1 and D2 receptors are important for the induction of synaptic plasticity in the prefrontal cortex (Kolomiets et al. 2009; Bai et al. 2014). Nevertheless, the respective physiological correlates for the abovementioned results also demonstrated a glutamatergic long-term potentiation (LTP)-enhancing effect of D1 receptors, which has been demonstrated in animal models, as well as in humans (Otmakhova and Lisman 1996, 1998; Nitsche et al. 2009; Fresnoza et al. 2014a), via enhancement of the strength of synaptic connections. In difference, D2 receptors are assumed to have a minor impact on LTP (Hagena and Manahan-Vaughan 2016), or a LTP-blocking effect (Price et al. 2014), and might be more closely involved in the generation of glutamatergic LTD (Monte-Silva et al. 2009; Fresnoza et al. 2014b; Banks et al. 2015), which is in good accordance with a flexibility-enhancing effect of activation of this receptor. A certain amount of D2 receptor activity seems however to be required also for LTP induction (Abe et al. 2009; Monte-Silva et al. 2011).
Beyond the general contribution of dopamine, as well as its receptor subtypes to cognitive functions and brain physiology, the amount of activation plays an important role for its effects, as shown by the well-known inverted U-shaped dose-response curve of dopamine demonstrated in electrophysiological and cognitive studies in animals (Williams and Goldman-Rakic 1995; Seamans and Yang 2004). Such an inverted U-shaped, dose-response effect was also observed in humans, especially for prefrontal cognitive functions, such as working memory (for review, see Cools and D’Esposito 2011). Recently, similar effects of dopaminergic activation, including D1- and D2-like receptor subtypes, have been demonstrated for neurophysiological measures, such as plasticity, in humans, by aid of noninvasive brain stimulation techniques (Nitsche et al. 2006; Monte-Silva et al. 2009, 2010; Thirugnanasambandam et al. 2011; Fresnoza et al. 2014a).
With respect to the impact of dopaminergic effects on learning and memory formation in humans, a relevant improvement of receptive processes by global dopaminergic enhancement has been demonstrated in different domains (Knecht et al. 2004; Flöel et al. 2005, 2008; Rösser et al. 2008). The contribution of dopaminergic subtypes on performance, as well as a respective dosage-dependency, which can be expected from the results of animal, and human plasticity studies, is however so far unclear. Furthermore, it remains unclear to which degree the model of dopaminergic mechanisms in working memory is applicable also for other cognitive functions, especially learning, and long-term memory processes. A previous study has demonstrated a D1-dominant modulation of long-term potentiation (LTP)-like neuroplasticity in the human motor cortex (Fresnoza et al. 2014a), and LTP is of critical importance for motor learning (Rioult-Pedotti et al. 2000). In the present study, we thus aimed to clarify the contribution of D1-like receptor activation to the dopaminergic effects on memory formation, including its dosage-dependency, using the motor cortex as model system. We further explored the impact of D1 receptor activation on initial, as well as advanced learning states, because a predominant impact of D1 receptor activation on memory consolidation has been proposed (Yamasaki and Takeuchi 2017).
To this aim, we compared the effect of D1-like receptor activation on performance during the learning and over-learned stage of a visuomotor tracking task. We applied three dosages (25, 100, 200 mg) of L-DOPA combined with sulpiride, a D2-like receptor antagonist, or placebo medication, to test task performance and monitor brain activation via functional magnetic resonance tomography during the task. We hypothesized that D1 activation should lead to general enhancement of task performance according to its LTP-enhancing effect and that this effect might be larger during the overlearned stage, based on the impact of the intervention on consolidation. Moreover, we expected a dose-dependent, inverted U-shaped modulation of performance, with superior effects of the medium dosage of the intervention, based on respective nonlinear results of D1 activation on LTP-like motor cortex plasticity in humans (Fresnoza et al. 2014a).
Materials and Methods
Participants
Eighty right-handed, healthy volunteers participated in the experiment. They were randomly assigned to four groups for different drug conditions, which were sex- and age-matched (Table 1). Participants had no history of acute or chronic medical, neurological, or psychiatric diseases, and metal or electric implants in the body. All participants were nonsmokers and had normal or corrected visual acuity. Pregnancy was ruled out by a pregnancy test. The study was approved by the Ethics Committee of the National Cheng Kung University Hospital and conformed to the Declaration of Helsinki. Subjects signed a written informed consent prior to participation.
. | Placebo . | 25 mg . | 100 mg . | 200 mg . | P value (F value) . | |
---|---|---|---|---|---|---|
Sex (F/M) . | 12/7 . | 10/5 . | 6/10 . | 9/13 . | . | |
Age (years ± SD) | 23.43 ± 3.52 | 22.23 ± 2.84 | 24.91 ± 4.29 | 23.17 ± 3.60 | 0.225 (1.490) | |
Education (years ± SD) | 16.22 ± 1.22 | 15.56 ± 1.46 | 16.31 ± 1.78 | 16.67 ± 2.48 | 0.365 (1.077) | |
Body weight (kg ± SD) | 57.56 ± 11.45 | 61.14 ± 12.72 | 63.09 ± 12.71 | 59.14 ± 7.58 | 0.483 (0.828) | |
Body height (cm ± SD) | 163.59 ± 9.18 | 163.41 ± 8.18 | 166.56 ± 7.85 | 167.18 ± 6.18 | 0.335 (1.151) | |
Interval (days ± SD) | 10.63 ± 4.84 | 8.94 ± 3.86 | 10.75 ± 3.11 | 11.41 ± 4.69 | 0.387 (1.024) | |
Baseline CPT (d’ ± SD) | Session 1 | 4.67 ± 0.16 | 4.73 ± 0.31 | 4.60 ± 0.35 | 4.70 ± 0.26 | 0.574 (0.669) |
Session 2 | 4.56 ± 0.43 | 4.66 ± 0.25 | 4.67 ± 0.35 | 4.68 ± 0.29 | 0.674 (0.514) |
. | Placebo . | 25 mg . | 100 mg . | 200 mg . | P value (F value) . | |
---|---|---|---|---|---|---|
Sex (F/M) . | 12/7 . | 10/5 . | 6/10 . | 9/13 . | . | |
Age (years ± SD) | 23.43 ± 3.52 | 22.23 ± 2.84 | 24.91 ± 4.29 | 23.17 ± 3.60 | 0.225 (1.490) | |
Education (years ± SD) | 16.22 ± 1.22 | 15.56 ± 1.46 | 16.31 ± 1.78 | 16.67 ± 2.48 | 0.365 (1.077) | |
Body weight (kg ± SD) | 57.56 ± 11.45 | 61.14 ± 12.72 | 63.09 ± 12.71 | 59.14 ± 7.58 | 0.483 (0.828) | |
Body height (cm ± SD) | 163.59 ± 9.18 | 163.41 ± 8.18 | 166.56 ± 7.85 | 167.18 ± 6.18 | 0.335 (1.151) | |
Interval (days ± SD) | 10.63 ± 4.84 | 8.94 ± 3.86 | 10.75 ± 3.11 | 11.41 ± 4.69 | 0.387 (1.024) | |
Baseline CPT (d’ ± SD) | Session 1 | 4.67 ± 0.16 | 4.73 ± 0.31 | 4.60 ± 0.35 | 4.70 ± 0.26 | 0.574 (0.669) |
Session 2 | 4.56 ± 0.43 | 4.66 ± 0.25 | 4.67 ± 0.35 | 4.68 ± 0.29 | 0.674 (0.514) |
Shown are demographic data (sex, age, education), baseline attention level (CPT), as well as the average interval between two sessions.
. | Placebo . | 25 mg . | 100 mg . | 200 mg . | P value (F value) . | |
---|---|---|---|---|---|---|
Sex (F/M) . | 12/7 . | 10/5 . | 6/10 . | 9/13 . | . | |
Age (years ± SD) | 23.43 ± 3.52 | 22.23 ± 2.84 | 24.91 ± 4.29 | 23.17 ± 3.60 | 0.225 (1.490) | |
Education (years ± SD) | 16.22 ± 1.22 | 15.56 ± 1.46 | 16.31 ± 1.78 | 16.67 ± 2.48 | 0.365 (1.077) | |
Body weight (kg ± SD) | 57.56 ± 11.45 | 61.14 ± 12.72 | 63.09 ± 12.71 | 59.14 ± 7.58 | 0.483 (0.828) | |
Body height (cm ± SD) | 163.59 ± 9.18 | 163.41 ± 8.18 | 166.56 ± 7.85 | 167.18 ± 6.18 | 0.335 (1.151) | |
Interval (days ± SD) | 10.63 ± 4.84 | 8.94 ± 3.86 | 10.75 ± 3.11 | 11.41 ± 4.69 | 0.387 (1.024) | |
Baseline CPT (d’ ± SD) | Session 1 | 4.67 ± 0.16 | 4.73 ± 0.31 | 4.60 ± 0.35 | 4.70 ± 0.26 | 0.574 (0.669) |
Session 2 | 4.56 ± 0.43 | 4.66 ± 0.25 | 4.67 ± 0.35 | 4.68 ± 0.29 | 0.674 (0.514) |
. | Placebo . | 25 mg . | 100 mg . | 200 mg . | P value (F value) . | |
---|---|---|---|---|---|---|
Sex (F/M) . | 12/7 . | 10/5 . | 6/10 . | 9/13 . | . | |
Age (years ± SD) | 23.43 ± 3.52 | 22.23 ± 2.84 | 24.91 ± 4.29 | 23.17 ± 3.60 | 0.225 (1.490) | |
Education (years ± SD) | 16.22 ± 1.22 | 15.56 ± 1.46 | 16.31 ± 1.78 | 16.67 ± 2.48 | 0.365 (1.077) | |
Body weight (kg ± SD) | 57.56 ± 11.45 | 61.14 ± 12.72 | 63.09 ± 12.71 | 59.14 ± 7.58 | 0.483 (0.828) | |
Body height (cm ± SD) | 163.59 ± 9.18 | 163.41 ± 8.18 | 166.56 ± 7.85 | 167.18 ± 6.18 | 0.335 (1.151) | |
Interval (days ± SD) | 10.63 ± 4.84 | 8.94 ± 3.86 | 10.75 ± 3.11 | 11.41 ± 4.69 | 0.387 (1.024) | |
Baseline CPT (d’ ± SD) | Session 1 | 4.67 ± 0.16 | 4.73 ± 0.31 | 4.60 ± 0.35 | 4.70 ± 0.26 | 0.574 (0.669) |
Session 2 | 4.56 ± 0.43 | 4.66 ± 0.25 | 4.67 ± 0.35 | 4.68 ± 0.29 | 0.674 (0.514) |
Shown are demographic data (sex, age, education), baseline attention level (CPT), as well as the average interval between two sessions.
Pharmacological Intervention
One hour before the start of the experiment, the participants received either placebo medication or low (25 mg), medium (100 mg), or high (200 mg) dosages of L-DOPA combined with 200 mg sulpiride. This combination of substances was chosen to achieve a relative selective activation of D1-like receptors via unspecific activity enhancement of dopamine receptors via L-DOPA and block of D2 receptors by sulpiride. The doses were chosen based on prior studies, which showed a nonlinear, dose-dependent effect of D1 activation on neuroplasticity (Nitsche et al. 2009; Fresnoza et al. 2014a). One hour after substance intake, at the time of visuomotor task performance, the drugs have reached peak plasma concentrations and have prominent effects on CNS physiology (Maltby et al. 2005; Kuo et al. 2008a; Fresnoza et al. 2014a). As a preventive measure for possible systemic side effects of L-DOPA, such as nausea and vomiting, subjects received 20 mg of the peripheral acting dopaminergic antagonist domperidone three times per day, for 2 days prior to the experiment, and also 2 h before medication intake.
Visuomotor Task
The visuomotor coordination task was conducted via a MRI-compatible joystick system (Thomas RECORDING GmbH, Gießen, Germany). During the task, subjects were lying inside the MRI scanner with the joystick fixed at the waist level with an elastic bandage, and instructed to move the cursor, which appeared as a yellow dot (diameter 9 mm), by the joystick, to track the target circle (diameter 12 mm) as accurate as possible. Participants viewed the display through a mirror that was placed away from the head coil about 10–13 cm, and the size of the mirror was 13.5 × 9 cm. Each trial started after the participant fixed the cursor in the screen center. Then the target circle began to move in one of four directions (left to right, right to left, top to bottom, or bottom to top) with the speed of 35 mm/s. Participants were instructed to follow the target circle with the cursor from when it had reached the center and from then on to keep the cursor within the target region until the circle stopped at the periphery. The total tracking distance was 140 mm, and the tolerance radius was set at 9 mm, which was invisible for the participants but turned the cursor color to red when it was out of range. Errors were defined as the target leaving the tolerance range of the cursor. The task session contained three blocks, each including 40 trials. The tracking time of each trial was around 4 s, and the total duration of each block was slightly variable as the exact interval between each trial within the block was determined by the participants.
fMRI
fMRI was conducted in a 3 T scanner (GE Medical System, LLC, Waukesha, USA) using a standard eight-channel phased array head coil. Initially, anatomic images based on a T1-weighted MRI sequence at 0.875 × 0.875 × 1 mm3 isotropic resolution were recorded (repetition time [TR] = 7.652 ms, inversion time: 450 ms, echo time [TE] = 2.928 ms, flip angle: 12°). For BOLD fMRI, a multislice T2-sensitive gradient-echo echoplanar imaging (EPI) sequence (TR = 1800 ms, TE = 30 ms, and flip angle 90°) at 3.75 × 3.75 mm2 resolution was used. Thirty-four consecutive sections at 3.6 mm thickness angulated in an axial-to-coronal orientation, covering the whole brain, were acquired. Four hundred contiguous EPI volumes were acquired for each block as one fMRI dataset.
Experimental Procedure
The study was conducted in single-blinded, randomized, and placebo-controlled design. Participants were first screened to exclude contraindications, and demographic data, such as body height, weight, years of education, and exercise level, were obtained. They also performed the continuous performance test (CPT) for a baseline measure of attention (Davies and Parasuraman 1982; Hsieh et al. 2005). One hour after medication, they were positioned supine in the scanner with headphones for noise protection. The task was performed in three blocks during the scan, with about 1-min break between each block. A second session with the same medication and task conditions was repeated after more than 1 week interval.
Data Analysis and Statistics
Behavioral Data
Chi-square tests and one-way analyses of variance (ANOVA) were performed to ensure between-group homogeneity of gender, age, body weight and height, years of education, as well as baseline CPT performance. For the visuomotor learning task, performance was defined as success rate (percentage of correct trial performances) in each block. A repeated measure ANOVA was conducted with “performance” as dependent variable, “session” and “block” as within-subject factors, and drug conditions (“dose”) as between-subject factor. In case of significance (P < 0.05), exploratory post hoc Student’s t-tests were performed (two-tailed, P < 0.05). In order to exclude a possible influence of different gender distribution in the four groups, additional analyses were also performed with “gender” as covariate.
fMRI Data
SPM12 in MATLAB was applied, and functional data sets were preprocessed with slice timing correction, motion correction, co-registration, segmentation, normalization, and smoothing. Head movement was calculated as root mean square of six directions (x, y, z translation, as well as pitch, roll, and yaw estimates) for each block, and a repeated-measures ANOVA was implemented to exclude a possible impact of head motion on the imaging results.
Since the start of each trial was determined by the time that a participant would need to place the cursor at the correct position in the screen center, the duration of each block was therefore varied between sessions and subjects. Due to the temporally variable nature of the task-related movement dynamics, we used independent component analysis (ICA) for a data-driven extraction of time course and components of interest from the fMRI response (Penney and Koles 2006; Calhoun et al. 2009, 2013). Among the several documented approaches for conducting ICA on fMRI data, we followed the approach suggested by Calhoun et al. (2001) and Esposito et al. (2005), where single-subject ICA was performed for each single run of the task, before output was combined and analyzed at the group-level (Calhoun et al. 2001; Esposito et al. 2005). ICA was implemented using the open-source GIFT toolbox (http://icatb.sourceforge.net/). Briefly, we decomposed the volume set of each block into the 20 most relevant (% variance explained) spatiotemporal independent components (ICs) which captured the most variance in the data and performed “back-reconstruction.” We identified the relevant ICs corresponding to task-related visuomotor activity based on the activity time series and spatial region of neural activation. The resulting components for each of the three blocks within a session were averaged and entered into a second level analysis using a general linear model (GLM) approach, in order to contrast different conditions including “drug dosage,” “run,” and “session.” To compare between different dosages, results from the second-level analysis were entered into a third-level mixed effects analysis modeling of ANOVA, separately for each session. Four explanatory variables were defined corresponding to the four drug conditions, and pairwise contrasts were then defined to determine specific changes of each drug dosage, compared to the placebo group. The probability Z-maps were thresholded with clusters determined by Z = 2.3 and a significance threshold of P = 0.05 with a cluster-based family-wise error correction for multiple comparisons, for all analyses described above. Additionally, nonparametric analysis with randomize/FSL was also applied for the control of possible false positives (Winkler et al. 2014; Eklund et al. 2016). Moreover, “gender” was specified as a nuisance regressor in the statistical model in order to exclude the possible influence of gender distribution in the four groups.
Finally, in order to determine the correlation between individual behavioral performance and brain activation, a multiple regression approach was used. The mean behavioral accuracy score of session performance of each subject was entered as a regressor in the GLM, and an additional F-test was conducted to model the interaction of this covariate with the drug conditions. In case of significant activation, follow-up comparisons were conducted between drug conditions by extracting the parameter estimates (beta value) using the MarsBaR toolbox (insert URL of the MarsBar toolbox).
Results
Data of 11 subjects were excluded from analysis due to vomit, personal reasons, and poor quality of imaging data, resulting in 72 participants included for the behavioral analysis, and 69 subjects for fMRI analysis (the latter are given in brackets) (n: placebo = 19 (17), 25 mg = 15, 100 mg = 16, 200 mg = 22 (21)). The demographic data, such as gender, age, body weight, and height, as well as CPT scores and interval between the two sessions, did not differ significantly between the four groups (Table 1). ANOVA results of head motion in fMRI data revealed no significant difference between subject groups (F = 1.848, P = 0.148) (Supplementary Table 1).
Behavioral Data
The results of the ANOVA (Table 2) revealed significant main effects of “dose” (df = 3, F = 3.149, P = 0.031), “session” (df = 1, F = 31.780, P < 0.001), “block” (df = 2, F = 25.073, P < 0.001), as well as a significant interaction of “session” and “block” (df = 2, F = 9.715, P < 0.001). Task performance was improved over blocks, and the slope of the learning curve decreased by sessions (Fig. 1).
. | df . | F . | P . |
---|---|---|---|
Dose | 3 | 3.149 | 0.031* |
Session | 1 | 31.780 | <0.001* |
Session × dose | 3 | 0.827 | 0.484 |
Block | 2 | 25.073 | <0.001* |
Block × dose | 6 | 1.350 | 0.240 |
Session × block | 2 | 9.715 | <0.001* |
Session × block × dose | 6 | 1.403 | 0.218 |
. | df . | F . | P . |
---|---|---|---|
Dose | 3 | 3.149 | 0.031* |
Session | 1 | 31.780 | <0.001* |
Session × dose | 3 | 0.827 | 0.484 |
Block | 2 | 25.073 | <0.001* |
Block × dose | 6 | 1.350 | 0.240 |
Session × block | 2 | 9.715 | <0.001* |
Session × block × dose | 6 | 1.403 | 0.218 |
*P < 0.05.
. | df . | F . | P . |
---|---|---|---|
Dose | 3 | 3.149 | 0.031* |
Session | 1 | 31.780 | <0.001* |
Session × dose | 3 | 0.827 | 0.484 |
Block | 2 | 25.073 | <0.001* |
Block × dose | 6 | 1.350 | 0.240 |
Session × block | 2 | 9.715 | <0.001* |
Session × block × dose | 6 | 1.403 | 0.218 |
. | df . | F . | P . |
---|---|---|---|
Dose | 3 | 3.149 | 0.031* |
Session | 1 | 31.780 | <0.001* |
Session × dose | 3 | 0.827 | 0.484 |
Block | 2 | 25.073 | <0.001* |
Block × dose | 6 | 1.350 | 0.240 |
Session × block | 2 | 9.715 | <0.001* |
Session × block × dose | 6 | 1.403 | 0.218 |
*P < 0.05.
. | Region . | Extent (mm3) . | t-value . | x . | y . | z . |
---|---|---|---|---|---|---|
100 mg > plc | R postcentral gyrus | 381 | 7.133 | 30 | −38 | 54 |
R angular gyrus | 381 | 5.766 | 30 | −58 | 48 | |
200 mg > plc | L midcingulate cortex | 627 | 6.640 | −4 | −20 | 52 |
R posterior-medial frontal | 627 | 5.719 | 2 | −4 | 64 | |
L inferior parietal lobule | 306 | 6.042 | −28 | −60 | 58 | |
L postcentral pyrus | 306 | 5.887 | −32 | −36 | 54 |
. | Region . | Extent (mm3) . | t-value . | x . | y . | z . |
---|---|---|---|---|---|---|
100 mg > plc | R postcentral gyrus | 381 | 7.133 | 30 | −38 | 54 |
R angular gyrus | 381 | 5.766 | 30 | −58 | 48 | |
200 mg > plc | L midcingulate cortex | 627 | 6.640 | −4 | −20 | 52 |
R posterior-medial frontal | 627 | 5.719 | 2 | −4 | 64 | |
L inferior parietal lobule | 306 | 6.042 | −28 | −60 | 58 | |
L postcentral pyrus | 306 | 5.887 | −32 | −36 | 54 |
. | Region . | Extent (mm3) . | t-value . | x . | y . | z . |
---|---|---|---|---|---|---|
100 mg > plc | R postcentral gyrus | 381 | 7.133 | 30 | −38 | 54 |
R angular gyrus | 381 | 5.766 | 30 | −58 | 48 | |
200 mg > plc | L midcingulate cortex | 627 | 6.640 | −4 | −20 | 52 |
R posterior-medial frontal | 627 | 5.719 | 2 | −4 | 64 | |
L inferior parietal lobule | 306 | 6.042 | −28 | −60 | 58 | |
L postcentral pyrus | 306 | 5.887 | −32 | −36 | 54 |
. | Region . | Extent (mm3) . | t-value . | x . | y . | z . |
---|---|---|---|---|---|---|
100 mg > plc | R postcentral gyrus | 381 | 7.133 | 30 | −38 | 54 |
R angular gyrus | 381 | 5.766 | 30 | −58 | 48 | |
200 mg > plc | L midcingulate cortex | 627 | 6.640 | −4 | −20 | 52 |
R posterior-medial frontal | 627 | 5.719 | 2 | −4 | 64 | |
L inferior parietal lobule | 306 | 6.042 | −28 | −60 | 58 | |
L postcentral pyrus | 306 | 5.887 | −32 | −36 | 54 |
Post hoc Student’s t-tests revealed an inverted U-shaped dose-dependent effect of medication on performance. While performance in the 100 mg L-DOPA condition was improved in relation to the placebo condition with the exception of block 1, performance under 25 mg L-DOPA was indistinguishable from placebo, and 200 mg L-DOPA increased performance relative to placebo only in block 2. Moreover, 100 mg L-DOPA improved performance relative to the 25 mg condition in blocks 2, 3, 4, and 6 and relative to 200 mg L-DOPA in block 4, while trendwise performance differences in the other blocks were identically directed.
The overall performance, calculated as grand average across all six blocks, in the 100 mg L-DOPA condition was significantly better than performance in the placebo and low-dose groups (P = 0.012 and 0.028, respectively), while low- and high-dose conditions did not differ significantly from the placebo group and between each other (Fig. 3).
Interestingly, the overall performance significantly differed when considering “gender” as covariate (F = 9.860, df = 1, P = 0.003). Male subjects showed higher overall accuracy (44.32% ± 13.67) as compared to females (31.13% ± 17.90). However, there was no significant interaction of “gender” × “session” (F = 0.011, df = 1, P = 0.916), “gender” × “block” (F = 1.654, df = 2, P = 0.195), as well as “gender” × “session” × “block” (F = 0.112, df = 2, P = 0.894), indicating no impact of gender on the visuomotor learning process.
fMRI
Dosage Effect
Comparisons between each real medication dosage with the placebo group showed dose-dependent effects of D1 receptor activation on brain activity associated with visuomotor learning (Fig. 2). As summarized in Table 3, enhanced activation in the left sensorimotor area corresponding to the task performing hand was observed in all real medication conditions as contrasted to the placebo group. It was further revealed that medium D1 activation resulted in a more enhanced increase of activity of the right motor and parietal areas (MNI [30, −38, 54], t = 7.133, extent = 381; [30, −58, 48], t = 5.766, extent = 381) (Fig. 2). Further parametric analyses with gender as covariate also showed similar results, including increasing task-related sensorimotor activation, and slightly enhanced right motor-parietal activity (Supplementary Fig. 1 and Supplementary Table 2).
The unthresholded statistical maps are available through the NeuroVault repository under https://neurovault.org/collections/7198/.
Correlation Between Performance and Brain Activity
An explorative multiple regression analysis showed that task performance was associated with activation of the left primary motor area (MNI [x, y, z] = [−50, −6, 30]; height threshold T = 3.11; P < 0.001 [uncorrected]). The estimated parameters (i.e., coefficients) of this region were extracted for each drug condition (data shown as mean [SD]; placebo: 0.12 [0.07]; 25 mg: 0.16 [0.07]; 100 mg: 0.21 [0.06]; 200 mg: 0.19 [0.06]) (Fig. 3).
Discussion
The results of the present study show a dosage-dependent effect of D1 activity on different stages of motor learning. Healthy humans were acquired to learn and perform a visuomotor task under different levels of D1 activation. The outcomes reveal a dose-dependent modulation of performance. Participants exposed to medium-dose but not low- and high-dose, D1 activation showed better task performance when compared to the placebo group. Functional imaging data also demonstrated differences in the pattern of cerebral activation in the three dosage groups, as contrasted to placebo. While increasing task-related sensorimotor and occipital activation of the left hemisphere was observed with increasing D1-activating dosages, medium-dose D1 activation resulted in most prominent activity of the right sensorimotor network. Further multiple regression analysis revealed a positive correlation of task performance specifically with activation of the left M1.
Modulation of Learning Processes via D1 Activation
In humans, an impact of global dopaminergic enhancement on learning processes has been shown in the motor as well as the language domain (Knecht et al. 2004; Flöel et al. 2005, 2008; Rösser et al. 2008). The results of the present study are in general accordance with the results of those previous studies. Similar as in respective animal experiments, the results of the present study furthermore suggest that for this learning–improving effect, D1 receptor activity seems to be critical. This fits well with previous results of the impact of D1 receptor activation on LTP-like plasticity of the human motor cortex induced by noninvasive brain stimulation, where the intervention, which had a clear improving effect on motor learning in the present study, enhanced plasticity (Nitsche et al. 2009; Fresnoza et al. 2014a). It is also in accordance with results from animal models, where D1 receptor activation has been shown to be critically involved in learning processes and memory consolidation (Molina-Luna et al. 2009; Rossato et al. 2009; Rypma et al. 2015). More specifically, in the motor system, the D1 receptor is involved in the induction of synaptic plasticity in rats and thereby associated with motor skill learning (Molina-Luna et al. 2009; Rioult-Pedotti et al. 2015). Furthermore, D1 receptor binding potential has been shown to be predictive for recognition memory in humans (Rypma et al. 2015).
Beyond the general improving effect of D1 receptor activation on motor memory formation, the results of the present study show also an inverted U-shaped nonlinear effect. Medium dosage of D1 receptor activation improved performance, while low- and high-dosage D1 receptor activity enhancements had no effects. Respective effects have so far not been explored in humans; they are however well-known in animal models of cognitive performance. For working memory, medium D1 receptor activation results in optimal performance, whereas too little or excessive D1 activation impairs it (Goldman-Rakic et al. 2000; Seamans and Yang 2004). Such an inverted U-shaped effect of D1 receptor activation was also observed for corresponding cortical activities (Williams and Goldman-Rakic 1995; Vijayraghavan et al. 2007). Furthermore, similar results have been obtained in animal models of long-term memory formation. Medium dosage of a D1 agonist facilitated long-term memory in rats and mice (Castellano et al. 1991; Bach et al. 1999; Floresco and Phillips 2001; Hotte et al. 2006; de Lima et al. 2011), and titrating a D1 receptor antagonist revealed a similar dose dependency for modulation of memory consolidation (Bethus et al. 2010). Moreover, a better long-term memory retention was demonstrated in mice under a medium-dosed D1 agonist. In further accordance, LTP-like plasticity of the human motor cortex was preserved under medium D1 receptor activation but not via high- or low-dose interventions (Fresnoza et al. 2014a).
Finally, the results of the present study show an involvement of D1 receptor activation in the initial stage of motor learning but also at a later stage, which involves consolidation. This pattern of results is in accordance with respective results of animal models, where D1 receptor activation has been shown to be relevant for both stages of memory formation (Karunakaran et al. 2016; Daba Feyissa et al. 2019; Papp et al. 2019; Roughley and Killcross 2019). This might be relevant, because it suggests that the terminal gain of performance could be enhanced by D1 receptor activity.
D1 Modulation of Task-Related Brain Activity
A drug-specific effect on cortical activation during the learning process was revealed when conditions with real medication were contrasted to the placebo group, especially in task-associated cortical networks, which include visual, parietal, and motor areas (Kruse et al. 2002). When contrasted to the placebo group, high-dosage D1 activation resulted in broader activation of visuomotor networks, as compared to medium-dosage activation. Interestingly, medium-dose D1 activation resulted in also higher activation of the sensorimotor-parietal region ipsilateral to the performing hand, as compared to the low- and high-dose groups, suggesting a performance-associated bi-hemispheric activation during task learning. It has been proposed that the accuracy of directional movement is determined by the fine-tuning of distinguished inhibitory circuits in motor networks, to which the ipsilateral motor area contributes (Mahan and Georgopoulos 2013). Moreover, the right frontal-parietal network has also been shown to be critical for direction selectivity in reaching movements independently from the performing hand (Fabbri et al. 2010). In accordance, the behavioral outcome under medium-dose D1 activation might be associated with enhanced spatial and attentional orienting operated by the right parietal area (Colby and Goldberg 1999). Respective mechanisms with regard to D1 modulations are however speculative at present and remain to be further investigated and confirmed.
For the correlation of cortical activation with task performance, a multiple regression analysis was performed. The results show a positive correlation for the left motor cortex. As for task accuracy, the extracted beta value of the corresponding brain area showed a dose-dependent, inverted U-shaped pattern (Fig. 3). This finding indicates a critical association of primary motor cortex activity for visuomotor task performance modulated by D1 activation. On the one hand, this result fits well with a performance-improving effect of activity enhancement of the left primary motor cortex via noninvasive brain stimulation obtained for this task in previous studies (Antal et al. 2004). On the other hand, it is also in accordance with the impact of different levels of D1 activation on motor cortex plasticity induced by noninvasive brain stimulation. Here, both low- and high-dosage D1 activation reduced LTD-like plasticity, whereas medium D1 receptor activation preserved it (Fresnoza et al. 2014a). Taken together, it can be concluded that D1 receptor activation enhances motor learning and memory performance in humans via an LTP-enhancing impact on the primary motor cortex and that this effect requires fine tuning of respective D1 receptor activation.
Some limitations of the present study should be taken into account. The study was conducted in a single-blind design. However, interaction between participants and the experimenters were highly formalized; thus, an impact of this limitation on the results is improbable. Furthermore, serum levels of drug concentrations were not evaluated. Instead, standard dosages of the substance were applied. This might have enhanced variability of the results. However, since age and body weight did not differ between groups, a systematic impact of these factors on the results can be ruled out. Regarding the pharmacological approach with a combination of substances in order to achieve D1 activation, an impact of other receptor subtypes on the results cannot be completely excluded, as L-DOPA activates all dopaminergic receptor subtypes, while sulpiride only blocks D2 receptors. In addition, the age range of recruited subjects was relatively homogeneous, including young adults. Thus, transferability of the results to the elderly, or patients with dopamine activity alterations, cannot be taken for granted. Furthermore, individual differences in basal dopamine activity have been shown to be critical for cognitive performance as well as response to dopaminergic modulation (Granon et al. 2000; Phillips et al. 2004), which might contribute to variability of the impact of different dosages of D1 activation at the level of the individual. It should also be noted that baseline performance without medication was not evaluated due to the experimental design, which required a 1-h interval after drug intake to reach peak plasma concentration of medication. Therefore, a baseline measurement in the scanner before medication would not have been feasible. However, attentional level was assessed with the CPT before each session started and did not differ between groups. Moreover, since performance in block 1 did not differ significantly between groups, we assume that baseline performance would have been comparable.
Finally, the analysis of covariance revealed a significant main effect of gender on overall task performance. Male participants showed higher accuracy regardless of block and session, but the learning curve did not differ in both genders. Such gender-dependent difference in performance has been observed in similar visuomotor coordination task (Harwell et al. 2018). It is important to note that this did not interfere statistically with the task learning curve in the present study, and imaging data also demonstrated the specific dose-dependent activation effect after regressing out a possible gender influence.
Taken together, the present study shows that the dopaminergic D1 receptor has a beneficial but nonlinear effect on motor learning in humans. This inverted U-type modulation was reflected by respective activation of the left motor cortex, a main component of task-related neural networks. Based on these results, targeting specific receptors with the optimal dosage during specific cognitive training or rehabilitation might have potential for future clinical applications.
Funding
National Science Council of Taiwan (NSC 102-2314-B-006-017-MY2 to P.-S.C.); German Research Foundation (DFG NI683/6-1 to M.A.N.).
Notes
We thank the “Mind Research and Imaging Center” at the National Cheng Kung University for consultation and instrument availability. The “Mind Research and Imaging Center” is supported by the Ministry of Science and Technology, Taiwan. Conflicts of Interest: MAN is on the advisory board of NeuroElectrics and NeuroDevice.
References
Author notes
Michael A. Nitsche and Min-Fang Kuo have contributed equally to this work.