Data were collected from two Japanese macaque monkeys [Macaca fuscata: Monk-1 (female, 5.0 kg, 5 years old) and Monk-2 (male, 6.8 kg, 6 years old)] provided through the National BioResource Project of the Japan Ministry of Education, Culture, Sports, Science and Technology (MEXT). All animal experimental procedures were approved by the Committees for Animal Experiments at the National Institute of Natural Sciences and at the Graduate School of Medicine in Kyoto University and were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. The animals were fed regularly with diet pellets. Monk-1 was under water-restriction for the behavioral task, while Monk-2 had free access to water. The monkeys were prepared using standard surgical techniques described elsewhere (Sawada et al., 2015). Briefly, a head-holder, a chamber located over the VSt to gain easy access for muscimol injections, and an electrocorticographic (ECoG) electrode array comprised of 12 channel (4 × 3 grid) electrodes over the forelimb area of the primary motor cortex (M1), the primary somatosensory cortex (S1), and premotor cortex (PM) on the left hemisphere, were implanted under isoflurane anesthesia and aseptic conditions.

The manipulandum, which can measure wrist joint torque and joint angle in two dimensions (x, flexion-extension; y, ulnar-radial direction), was used for the present study. The monkeys controlled the one-dimensional (flexion-extension) position of a blue cursor on a video monitor with auxotonic (spring-loaded) contraction (Monk-1) or with isometric contraction of the wrist joint (Monk-2) and acquired targets displayed on the screen (Figure 1A). The cursor was adjusted to the rest position when the wrist joint torque was neutral. When the cursor stayed on the rest position for 0.8 s, one of the red targets appeared as an instruction cue. After an additional stay in the rest position for 0.45–0.55 s (Monk-1) or 0.7–1.2 s (Monk-2) during the instruction period, the target color changed from red to yellow, indicating a go-cue. Then, the monkeys were required to move and maintain the cursor on the target for 0.8 s (Monk-1) or 0.7 s (Monk-2) within 10 s after the go cue. After the acquisition of the target, the monkeys were required to back the cursor to the rest position and maintain this position for 0.15–0.2 s to receive a liquid (Monk-1) or apple sauce reward (Monk-2). Monk-1 and Monk-2 performed a five-graded and a two-graded torque-tracking task, respectively. The Moderate target accepted a wide range of torque output by auxotonic torque production (Monk-1) or isometric torque production (Monk-2) about the wrist joint. The Precise target required precise control of torque output. Since motor control of auxotonic contraction is assumed to be easier compared with isometric contraction, an additional three targets were imposed for Monk-1, as follows. A More precise target requiring more precise control, a High target requiring high torque output, and a High-Precise target requiring precise control of high torque output. Each of the targets was presented in randomized order with a preset probability. Monk-1 and Monk-2 performed four and nine sessions in the control conditions and five and five sessions in the VSt inactivation sessions, respectively (Supplementary Table 1). Monk-2 also performed one saline injection session.

To investigate the contribution of the VSt to motor control, the VSt in the left hemisphere, contralateral to the task-engaged right forelimb, was inactivated pharmacologically. Muscimol, a gamma aminobutyric acid (GABA)_A receptor agonist [5 μg/μl, dissolved in 0.1 M phosphate buffer (PB, pH 7.4)], was pressure injected at a rate of 0.4 μl/min (2–8 sites, 0.5–2 μl/site, 2–4 μl/session) to focally inactivate the contralateral VSt (Figures 2A,B). Considering that a 1.5 μl injection of muscimol inactivates a 3–4 mm in diameter spherical volume of brain tissue (Murata et al., 2015), which is consistent with those of other previous studies (Martin, 1991; Arikan et al., 2002), injection sites and muscimol volumes (2–8 sites, 0.5–2 μl/site, 2–4 μl/session) were determined to infiltrate the entire VSt with muscimol, assuming that 1 μl of muscimol would diffuse within a 2 mm radius. A 10 μl Hamilton syringe (Hamilton Company, Reno, Nevada, USA) was held on a syringe pump fixed on the stereotaxic manipulator. The grid-guided needle was inserted stereotaxically into the VSt. Injection sites were estimated by MRI images. As a vehicle condition, saline was injected in Monk-2. The behavioral session began approximately 10 min after the infusion was complete.

During the experiments, task parameters such as target positions, timing of trial events, wrist torque (flexion-extension direction), and ECoG data were recorded simultaneously using a Cerebus multichannel data acquisition system (Blackrock Microsystems, Salt Lake City, UT, USA) at a sampling rate of 2 kHz.

To compare behavioral outcomes between control (no intervention and/or vehicle) and the VSt inactivation sessions, we analyzed the initial 30-min data from each session.

To confirm the muscimol injection sites, histological analysis was conducted. Detailed procedures are described elsewhere (Sawada et al., 2015). Briefly, at the end of the experiment, electrocoagulation for the identification of muscimol injection sites in the VSt was made using a rectangular constant current at 40 μA for 20 s through the stimulating electrode under deep anesthesia with an overdose of sodium pentobarbital (50 mg/kg, i.v). Then, the monkeys were euthanized by transcardial perfusion with 0.1 M phosphate-buffered saline (pH 7.4), followed by 4% paraformaldehyde in 0.1 M phosphate-buffer (PB, pH 7.4). The perfused brain was postfixed in the same fresh fixative overnight at 4°C, and saturated with 10% sucrose in 0.1 M PB, followed by saturation in 20% and 30% sucrose solutions. The perfused brain was cut serially into 50 μm coronal sections, and every fifth section was mounted onto a gelatin-coated glass slide and Nissl-stained with 0.1% cresyl violet. Muscimol injection sites were estimated based on the location of coagulation in coronal sections of the brains.

The behavioral task procedure is shown in Figure 1A. The correct rates for targets requiring precise torque control were significantly lower than other targets in both monkeys (Figure 1B) [(Monk-1:one-way ANOVA, F_{4, 15} = 17.94, P < 0.001; Post hoc multiple comparison: Moderate vs. More-Precise, P < 0.05; Moderate vs. High-Precise, P < 0.001; Precise vs. More-Precise, P < 0.05; Precise vs. High-Precise, P < 0.001; High vs. High-Precise, P < 0.001), (Monk-2: paired t-test, t₍₈₎ = 6.45, P < 0.001)]. These results indicate that the precise control of torque output is demanding in both monkeys.

To investigate whether the VSt has an impact on motor control, we inactivated the contralateral VSt to the task-engaged hand pharmacologically and compared performance and cortical activity between the control and the VSt inactivation sessions (Figures 2A,B).

Figure 2C shows a typical example of task-related modulations of the wrist torque (1st line panels: control, black; VSt inactivation, blue; data obtained from Monk-1) and oscillatory activity obtained from the M1 and PM sites (enlarged red and cyan circles in Figure 2A, respectively) in correct trials. Compared with control trials, torque outputs, except for the High-Precise target, were significantly decreased during the VSt inactivation trials (2nd line panels in Figure 2C). Torque outputs in the High-Precise target showed a similar trend, but the differences were not significant. In control trials, representative M1 and PM sites revealed increased activity in high-γ, γ, and θ bands around the go cue for all targets compared with baseline (rest period) activity (3rd and 6th line panels in Figure 2C). On the other hand, such modulation in high-γ and γ bands during VSt inactivation was decreased (4th and 7th line panels in Figure 2C). Statistical analysis showed a significant reduction of modulation in high-γ and γ bands during VSt inactivation compared to the control trials in both M1 and PM among targets (5th and 8th panels in Figure 2C). In contrast, significantly increased modulation in lower frequency bands, α and θ bands during the VSt inactivation trials, was observed in the M1 site only for Moderate, Precise, and More-Precise targets. A similar effect of VSt inactivation on both torque output and cortical activities was observed for the Moderate target in Monk-2. Torque outputs for the Precise target also showed a decreasing trend as well, but significant differences were not found in cortical activity (Supplementary Figure 1).

Torque output traces in a typical example show a decreasing trend during VSt inactivation trials (1st and 2nd line panels in Figure 2C). We compared peak torque values (Figure 3A) of all trials for each target between control and inactivation sessions. The result in both monkeys showed significantly decreased peak torque for almost all targets (Figure 3B, Wilcoxon-signed rank test with Bonferroni correction; P < 0.001). Although peak torque values in correct trials for the Precise target in Monk-2 did not show significant change, they tended to be decreased compared with the control trials. To succeed in the trials for the Precise target by isometric contraction, the precise torque control, a minimal fluctuation of torque production was required. Therefore, a significant decrease in both the peak torque and high-frequency component in movement-related activity (Supplementary Figure 1) would not be observed in correct trials. Instead, the number of failed trials where the monkey could not maintain torque for the required time increased for the Precise target (Control, 7.3%; VSt inactivation, 11.8%), but not for the Moderate target (Control, 2.1%; VSt inactivation, 1.9%). These results suggest that the VSt inactivation interferes in maintaining torque output requiring precise control by isometric contraction.

In addition to the peak torque, VSt inactivation might affect the initiation of motor response, which is considered an indicator of motivational state (Hollerman et al., 1998; Hassani et al., 2001; Adcock et al., 2006; Mir et al., 2011). Thus, we analyzed the RTs (Figure 3A). VSt inactivation did not affect the RTs in both monkeys (Figure 3C, Wilcoxon-signed rank test with Bonferroni correction).

The correct rates for each target also showed no significant difference between the control and VSt inactivation sessions in both monkeys (Figure 3D, Wilcoxon-signed rank test with Bonferroni correction). This result seems consistent with previous findings showing that a VSt lesion or depletion of dopamine in the VSt did not affect instrumental responses requiring movements (Taylor and Robbins, 1986; Parkinson et al., 1999; Corbit et al., 2001; Cardinal and Cheung, 2005; Corbit and Balleine, 2011). Furthermore, omission trials were not increased by VSt inactivation [(Monk-1: Control, 1.38 ± 1.38% per session; VSt inactivation, 0.059 ± 0.059%), (Monk-2: Control, 4.26 ± 0.93%; VSt inactivation, 0%); average and standard error of the mean are shown]. These results combined with the results in RTs suggest that VSt inactivation does not induce a decrease in motivation to perform the task.

Since the movement-related power change in representative sites in the M1 and PM showed a common reduction of high-frequency modulation among targets (Figure 2C), we combined the data for individual targets into a single set of data and compared the population data in the time window in which the movement-related power change was observed (−0.5 to 0.5 s and 0–1 s from the go cue in Monk-1 and Monk-2, respectively). The results showed similar changes in the PM and M1. VSt inactivation decreased movement-related power changes at high-γ, γ, and β bands (Figure 4, Wilcoxon-signed rank test, P < 0.01; P-values are shown in Supplementary Table 2). In contrast, the θ band in both the PM and M1 increased power changes during VSt inactivation in the two monkeys. While the S1 also showed similar changes in β, α, and θ bands, as well as the PM and M1, the difference in high-γ and γ bands between the VSt inactivation and control trials was contradictory in the two monkeys.

VSt inactivation may affect cortical activity in general during the rest phase of the task. Therefore, frequency power during the rest periods was analyzed. Frequency power at the β band in VSt inactivation trials was significantly decreased during the rest periods compared with control trials (Supplementary Figure 2; Wilcoxon-signed rank test, P < 0.001; P-values are shown in Supplementary Table 3), which was common in both monkeys except for Monk-2’s PM. In addition, high-γ power in the S1 increased in the VSt inactivation trials (Wilcoxon-signed rank test: Monk-1, P < 0.001; Monk-2, P < 0.01) while γ band power in the S1 decreased in both monkeys (Wilcoxon-signed rank test, P < 0.001). On the other hand, other frequency bands were contradictory in the two monkeys, although significant changes were instigated by VSt inactivation. This result indicates that SMC activities in the rest phase were also affected by VSt inactivation.