Abstract
The subthalamic nucleus (STN) is a common target for deep brain stimulation (DBS) treatments of Parkinsonian motor symptoms. According to the dominant model, the STN output can suppress movement by enhancing inhibitory basal ganglia (BG) output via the indirect pathway, and disrupting STN output using DBS can restore movement in Parkinson's patients. But the mechanisms underlying STN DBS remain poorly understood, as previous studies usually relied on electrical stimulation, which cannot selectively target STN output neurons. Here, we selectively stimulated STN projection neurons using optogenetics and quantified behavior in male and female mice using 3D motion capture. STN stimulation resulted in movements with short latencies (10–15 ms). A single pulse of light was sufficient to generate movement, and there was a highly linear relationship between stimulation frequency and kinematic measures. Unilateral stimulation caused movement in the ipsiversive direction (toward the side of stimulation) and quantitatively determined head yaw and head roll, while stimulation of either STN raises the head (pitch). Bilateral stimulation does not cause turning but raised the head twice as high as unilateral stimulation of either STN. Optogenetic stimulation increased the firing rate of STN neurons in a frequency-dependent manner, and the increased firing is responsible for stimulation-induced movements. Finally, stimulation of the STN's projection to the brainstem mesencephalic locomotor region was sufficient to reproduce the behavioral effects of STN stimulation. These results question the common assumption that the STN suppresses movement, and instead suggest that STN output can precisely specify action parameters via direct projections to the brainstem.
SIGNIFICANCE STATEMENT Our results question the common assumption that the subthalamic nucleus (STN) suppresses movement, and instead suggest that STN output can precisely specify action parameters via direct projections to the brainstem.
Introduction
The subthalamic nucleus (STN) is a key diencephalic nucleus that is strongly and reciprocally connected with the basal ganglia (BG) and mesencephalic locomotor region (MLR) of the brainstem (Parent and Hazrati, 1995). It is sometimes considered to be a part of the indirect pathway in the BG. According to classic models, the direct and indirect pathways play opposing roles, with the former responsible for movement initiation and the latter for movement inhibition (Albin et al., 1989; Kravitz et al., 2010). The direct pathway exerts its influence through inhibition of BG output nuclei such as internal globus pallidus (GPi) and substantia nigra pars reticulata (SNr), whereas the indirect pathway can disinhibit the GPi and SNr, in part via the STN (Smith and Bolam, 1989; Alexander and Crutcher, 1990; DeLong, 1990). Given the STN's glutamatergic projections to the GPi and SNr, its activation is thought to oppose the effects of the direct pathway, resulting in movement cancelation (Nambu et al., 2002; Aron and Poldrack, 2006; Fife et al., 2017). Consequently, STN is often used as the site of deep brain stimulation (DBS), a major treatment option for Parkinson's disease (PD), as STN stimulation at high frequency is thought to reduce hyperactive indirect pathway activity caused by DA depletion (DeLong, 2001).
However, evidence for the STN's role in canceling action is mixed. Although human fMRI work using a stop-signal task suggested that the STN is more activated during successful stop compared with go trials (Aron and Poldrack, 2006), electrophysiological results in rats using a similar task showed that STN neurons increase firing after stop cues regardless of whether stopping was successful (Schmidt et al., 2013). There have also been conflicting observations on the effects of STN stimulation on movement. Bilateral STN photo-stimulation was reported to reduce horizontal movement while inhibition increased horizontal movement, but the opposite pattern was observed in grooming behavior (Guillaumin et al., 2021). Moreover, in another study, STN stimulation failed to cause a significant effect on locomotion in the open field altogether (Heston et al., 2020). Bilateral lesions of the rat STN have also been shown to affect stopping accuracy in a stop-signal task, but stop signal reaction time was unaffected, and the stopping deficit did not depend on the stop-signal delay, suggesting that lesions may actually have increased impulsive behavior rather than simply caused problems stopping behavior (Eagle et al., 2008). Indeed, impulsive responding has been found with bilateral rat STN lesions in several tasks (Baunez et al., 1995; Baunez and Robbins, 1997; Uslaner and Robinson, 2006). Fife et al. (2017) reported that photo-stimulation of the mouse STN cancels ongoing lick bouts or induces a pause when bouts are not canceled, but videos suggest that stimulation actually caused head and body movements that interfered with licking. Although there was some data indicating that STN activation increases head movement and turning behavior, the role of the STN in movement has not been examined closely (Heston et al., 2020; Guillaumin et al., 2021).
A major limitation of previous work is the lack of precise quantification of the movements during STN manipulations. In the present study, we determined the relationship between photo-stimulation parameters and movement measured with high temporal and spatial resolution. Specifically, 3D motion capture measuring movement at 200 frames per second was used to quantify the subtle changes in head and body angle caused by optogenetic stimulation. To selectively target STN projection neurons, we used a vGlut2-Cre driver line combined with a Cre-dependent viral vector with channelrhodopsin (ChR2). We found that selective STN activation generates movement of the head and torso with very short latency, and there was a highly linear relationship between stimulation frequency and movement kinematics. This effect was largely reproduced by stimulating the STN's projection to the MLR.
Materials and Methods
Experimental design
Subjects
Data were collected from 21 mice (13 male; 8 female), aged three to eight months old. Vglut2-ires-Cre (Slc17a6tm2(cre)Lowl, The Jackson Laboratory) and wild-type mice were used as the ChR2 and control groups, respectively. Vglut2-ires-Cre mice express Cre-recombinase under control of Vglut2 receptor regulatory elements, which nearly all STN neurons express and surrounding areas lack, so these mice enabled selective expression of Cre-dependent ChR2 in the STN. Wild-type mice were used as the control group because they do not express Cre and thus would not express Cre-dependent ChR2 anywhere. Seven ChR2 and four control mice were used in the main STN stimulation experiments, two ChR2 mice were used in the optrode stimulation experiment, and four ChR2 and four control mice were used in the STN-MLR projection stimulation experiment. All mice were housed in groups of two to five and given free access to food and water. They were maintained on a 12/12 h light/dark cycle, and experiments were conducted during the light phase. All experimental procedures were approved by the Duke University Institutional Animal Care and Use Committee.
Surgery
Mice were anesthetized with 3% isoflurane in oxygen flowing at a rate of 0.6 l/min. They were then secured in a stereotactic frame (David Kopf Instruments) and injected with Meloxicam (2 mg/kg) before incision; 150–400 nl of a viral vector containing a Cre-dependent ChR2 (pAAV5.EF1a.DIO.hChR2(E123T/T159C).eYFP, Addgene) was then injected into the STN bilaterally, or unilaterally into the left STN for the optrode experiments (Anterior/Posterior (AP): −2.0, Medial/Lateral (ML): ±1.6, Dorsal/Ventral (DV): −4.5 mm relative to bregma), at 1 nl/s with a Nanoject III microinjector (Drummond Scientific). Next, for the STN and STN-MLR projection stimulation experiments, respectively, custom-made optic fibers (0.22 NA, MM, 105-µm core, ∼80% transmittance) were implanted bilaterally either 4.35 mm ventral to bregma in the injection tract or in the MLR (AP: −4.4, ML: ±1.5, DV: −3.2 mm relative to bregma). For the optrode stimulation experiments, a 16-channel drivable electrode array (Innovative Neurophysiology) with a custom-made optic fiber attached at a 16° angle to the end of the electrode cannula was implanted just above the left STN (AP: −2.0, ML: −1.7, DV: −4.0 mm relative to bregma). All implants were anchored to skull screws using dental acrylic and custom made headbars (H.E. Parmer) were then affixed to the acrylic (Fig. 1A). Mice were given four to six weeks to recover from surgery before testing.
Placement of viral injections, optic fibers, and electrodes. A, Schematic of surgery strategy for bilateral STN stimulation experiments, along with representative histology and optic fiber tip locations. B, Schematic of surgery strategy for left hemisphere STN optrode stimulation experiment, along with representative histology and electrophysiological recording locations. C, Schematic of surgery strategy for STN-MLR projection stimulation experiments, along with optic fiber tip locations.
Open field stimulation and 3D motion capture
Mice were placed on a one foot tall, wall-less, 8 × 8-inch open field platform, set on a table centrally located in the experiment room; 6.4-mm diameter infrared markers (B&L Engineering) were attached just above and to either side of the head, at the base of the optic fibers and on the headbars, with another added to the tail base. Six Osprey motion tracking cameras (Motion Analysis) positioned around the table captured marker positions at 200 Hz, and the Cortex Motion Analysis computer program saved them in 3D cartesian coordinates (Fig. 2A; Bartholomew et al., 2016).
Quantifying movement kinematics using 3D motion capture. A, Schematic of the experimental setup. B, Body angle calculation. A right triangle was made in the X-Y plane with the tail marker and center of the left and right head markers (head center) as vertices. The angle at the tail marker was shifted into the proper polar quadrant based on the mouse's orientation and was used as body angle. C, To calculate yaw, a triangle was made in the X-Y plane with the tail marker, right head marker, and head center as vertices. The angle at head center minus 90 was used as yaw. D, To compute roll, a triangle was made in the X-Z plane with the left head marker, head center, and a float point 10 mm below head center as vertices. The angle at head center minus 90 was used as roll. E, To calculate pitch, a triangle was made in 3D space with the top head marker, head center, and a float point 10 mm in front of head center as vertices. The angle at head center minus 90 was used as pitch. F, Illustration of the key measures: body angle, head yaw, roll, and pitch.
MATLAB (MathWorks) was used to make a NI USB-6211 (National Instruments) operate a 473-nm diode-pumped solid state laser (Shanghai Laser & Optics Century Co, Ltd.) at specific frequencies and pulse widths, detailed below. Laser light was relayed through a 100-µm core optic fiber (Precision Fiber Products, 0.22 NA, MM), to a FRJ_1x2i_FC-2FC_0.22 commutator (Doric Lenses), into another 100-µm core optic fiber (Precision Fiber Products, 0.22 NA, MM), which was connected to the mice's optic fiber implants. A Cerebus Neural Signal Processor (Blackrock Microsystems) was used to record the laser control signal from the NI device and the frame signal from the Osprey cameras so that movement could be aligned to laser pulses. During the optrode stimulation experiments, the Cerebus system was also used to record electrophysiology data.
For the main STN stimulation experiments, the laser power was adjusted on a mouse-by-mouse basis so that 5.5–9.5 mW of power was transmitted to the STN, given the tested 80% transmittance of the optic fiber implants. For the STN-MLR projection stimulation experiments, the power range used was 4–9.5 mW, and for the STN optrode stimulation experiments, the power was set at 7 mW.
All frequency stimulation data from a given hemisphere, or combination of hemispheres, was collected in a separate session. Half second, 5-, 10-, 20-, and 40-Hz laser trains were delivered pseudo-randomly, so that each occurred 15 times, with a 20- to 30-s intertrain interval; 10-ms pulse widths were used in all frequency stimulation experiments, except the optrode ones, where a 20-ms pulse width was used. All single pulse stimulation data from a given hemisphere, or combination of hemispheres, was also collected in a separate session. Five-, 10-, 20-, and 40-ms single laser pulses were delivered pseudo-randomly, so that each occurred 15 times, with a 15- to 20-s interpulse interval. All data from a given mouse was collected in a single day, with the exception of the STN optrode mice. In this experimental group, the electrode locations were changed between sessions and frequency stimulation data were collected repeatedly so that different STN neurons could be recorded at different depths.
In vivo electrophysiology
Implanted drivable electrodes were single-drive movable micro-bundles of tungsten electrodes (1 × 16; 23 μm in diameter) placed within a guide cannula (Innovative Neurophysiology). Electrophysiological data were recorded via CerePlex µ headstages that communicated with the Cerebus data acquisition system used to collect the behavioral and optogenetic stimulation timestamps (Blackrock Microsystems). A 250-Hz to 5-kHz bandpass filter was applied to the electrophysiological data during online sorting in Cerebus' Central user-interface program. Acquired data were re-sorted offline using Offline Sorter V4 (Plexon). All waveforms and raster plots of spiking activity were generated using NeuroExplorer 5 (Nex Technologies) and MATLAB.
Histology
Mice were perfused with 0.1 m PBS followed by 4% paraformaldehyde. Brains were then stored in 4% paraformaldehyde with 30% sucrose for 72 h before being transferred to 30% sucrose for another 24 h. Brains were then sliced into 60-µm coronal sections with a Leica CM1850 cryostat. Afterwards, sections were mounted and immediately coverslipped with Fluoromount-G with DAPI medium (catalog #17984-24, Electron Microscopy Sciences). To confirm viral expression and implant placement, slices were examined with an Axio Zoom V16 microscope (Zeiss; Fig. 1B,C).
Quantification and statistical analyses
Behavioral analysis
Calculation of body angle, head yaw, roll, and pitch was done in MATLAB for each motion sample. The MATLAB function “unwrap” was used to remove discontinuities caused by the body angle changing from 0° to 360°, or vice versa (Fig. 2B). Velocity was calculated for each variable by finding the difference in kinematic angle across adjoining samples and dividing by the inverse frame rate.
Movements caused by frequency stimulations were calculated for each combination of mouse, kinematic variable, hemisphere, and frequency. Frequency driven movements were measured by subtracting the kinematic angle at stimulation onset from the kinematic angle achieved at the end of the stimulation train, 500 ms later. Movements caused by single laser pulses were calculated for each combination of mouse, kinematic variable, hemisphere, and pulse width. Single pulse driven movements were measured by subtracting the kinematic angle at stimulation onset from the kinematic angle achieved 100 ms later. Latency to move following unilateral STN stimulation onset was determined for each combination of mouse, kinematic variable, and single STN by manual inspection of velocity changes caused by 40-ms single pulse stimulation. The time of the first significant change from prestimulation baseline in the effect direction was chosen on trials in which such a change occurred before stimulation offset, and the results were averaged across trials. Velocity 60 and 120 ms after single laser pulse onset was calculated for each combination of mouse, kinematic variable, STN, and pulse width by finding the velocity 60 and 120 ms after each laser pulse onset and averaging across each time point.
Optotagging analysis
For the optrode stimulation experiments, neurons were considered ChR2-positive (tagged) if the first light pulse of each laser train evoked significant spiking within 10 ms and caused a 200% or greater increase in firing rate from baseline. To identify significant evoked spike rates above the baseline firing rate of STN neurons, we applied a random jitter (±20 ms) to the spike train of each trial over 1000 iterations. We then determined a 95% confidence interval of the resulting distribution of firing rates expected by chance within each 1-ms time bin. This preserved the firing rates of the neurons but allowed us to detect above-chance levels of spike synchrony across trials resulting from laser stimulation. A neuron demonstrated significant spiking if a minimum of one time bin within 10 ms of stimulation onset had observed firing rates above the 95% confidence interval. To determine whether stimulation produced a significant increase in firing rate from baseline, the mean number of spikes in the 20 ms before and after stimulation onset was compared.
Statistical analysis
All statistical tests (linear regressions, mixed-effects models, t tests) were performed in GraphPad Prism (GraphPad Software).
Results
Laser frequency manipulation
The STN of ChR2 and control mice were stimulated unilaterally and bilaterally for 500 ms at 5, 10, 20, and 40 Hz, and stimulation-induced changes in body angle, head yaw, head roll, and head pitch were quantified. For unilateral stimulation, regression analyses revealed strong linear relationships between stimulation frequency and changes in all kinematic variables. Ipsiversive changes occurred in body angle, head yaw, and head roll, while stimulation of either STN increased head pitch (Table 1; Fig. 3, left). Significant interactions in two-way mixed-effects models elucidated the relationship between stimulation hemisphere and experimental group in the 40-Hz condition. For body angle, head yaw, and head roll, this confirmed the ipsiversive direction of the movement, while for head pitch, it demonstrated that left hemisphere stimulation increased head pitch more than right. For all kinematic variables, this confirmed the existence of the movement versus controls (Table 2; Fig. 3, middle left).
Linear regression results for STN ChR2+ laser frequency manipulation
Interaction between hemisphere and experimental group caused by 40-Hz laser stimulation
Photo-stimulation of STN ChR2+ neurons causes movement (frequency manipulation). A, left, Linear relationship between unilateral stimulation frequency and ipsiversive turning. Middle left, 40-Hz unilateral stimulation causes significant ipsiversive turning versus controls. Middle right, No relationship between stimulation frequency and turning with bilateral stimulation. Right, 20-Hz bilateral stimulation does not cause turning. B, Left, Linear relationship between unilateral stimulation frequency and ipsiversive head yaw. Middle left, 40-Hz unilateral stimulation causes significant ipsiversive head yaw across hemispheres. Middle right, No relationship between stimulation frequency and head yaw with bilateral stimulation. Right, 20-Hz bilateral stimulation does not cause head yaw. C, Left, Linear relationship between unilateral stimulation frequency and ipsiversive head roll. Middle left, 40-Hz unilateral stimulation causes significant ipsiversive head roll versus controls. Middle right, No relationship between stimulation frequency and head roll with bilateral stimulation. Right, 20-Hz bilateral stimulation does not cause head roll. D, Left, Linear relationship between unilateral stimulation frequency and upwards head pitch. Middle left, 40-Hz unilateral stimulation causes significant upwards head pitch versus controls. Left hemisphere stimulation increases head pitch more than right. Middle right, Nonsignificant linear relationship between stimulation frequency and upwards head pitch with bilateral stimulation. Right, 20-Hz bilateral stimulation causes a significant increase in head pitch versus controls. n = 7 ChR2 and 4 control mice. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Error bars indicate SEM.
For bilateral stimulation, regression analyses did not reveal a relationship between stimulation frequency and changes in body angle, head yaw, and head roll, however, there was a nonsignificant, but highly linear relationship between stimulation frequency and increased head pitch, up to 20 Hz (Table 1; Fig. 3, middle right); 40-Hz data were not included in any bilateral analysis because the stimulation caused mice to jump out of camera view, consistent with STN stimulation increasing the vertical orientation of the head. Consistent with the regression analyses, 20-Hz stimulation did not cause a significant change in body angle, head yaw, or head roll, but it did cause a significant increase in head pitch (two-tailed t test: t(9) = 0.5691, p = 0.5832; t(9) = 0.5928, p = 0.5679; t(9) = 1.809, p = 0.1039; t(9) = 3.690, p = 0.0050; Fig. 3, right). The increase in head pitch caused by bilateral stimulation was greater than the increase caused by unilateral stimulation of either hemisphere alone at every tested frequency up to 20Hz, and roughly two times greater in the 20-Hz condition. When taken together with the lack of a horizontal movement produced by bilateral stimulation, this indicates that bilateral STN stimulation produces a movement consistent with the sum of the unilateral stimulation effects.
ChR2-induced firing underlies the movement
The discovery that STN stimulation creates movement was unexpected, so we repeated the frequency stimulation experiment in a new group of mice with optrodes implanted in the left STN to verify that our manipulation was causing neurons to fire. As expected, a positive linear relationship was observed between stimulation frequency and firing rate in ChR2-positive STN neurons (linear regression: r2 = 0.9719, p = 0.0142; Fig. 4A).
Left STN ChR2 stimulation drives firing in a frequency dependent manner, and this firing underlies the stimulation-induced movement. A, Linear relationship between stimulation frequency and firing rate in tagged units. B, Mean opto-evoked spike latency measured from the onset of the first pulse in each laser train. C, D, Representative unit. Left, Spontaneous and evoked action potentials have the same shape. Right, Stimulation drives firing at all frequencies. E–H, Linear relationship between stimulation frequency and changes in all kinematic variables, except yaw. n = 24 units (17 tagged) recorded from 2 mice. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Error bars indicate SEM.
To link the stimulation-induced STN firing directly to movement, we also quantified the changes in kinematic variables caused by stimulation. Once again, significant linear relationships emerged between stimulation frequency and ipsiversive body turning, ipsiversive head roll, and upwards head pitch (linear regression: r2 = 0.9947, p = 0.0026; r2 = 0.9968, p = 0.0016; r2 = 0.9713, p = 0.0144). Although no relationship emerged between stimulation frequency and ipsiversive head yaw (linear regression: r2 = 0.3923, p = 0.3736; Fig. 4E–H). Altogether, this demonstrates that ChR2 stimulation drives STN firing in a frequency dependent manner, and this firing underlies the stimulation-induced movement.
Laser pulse width manipulation
The STN of the original mice were also stimulated unilaterally and bilaterally with single laser pulses of 5-, 10-, 20-, and 40-ms width, and stimulation-induced changes in body angle, head yaw, head roll, and head pitch 100 ms after stimulation onset were quantified. For unilateral stimulation, regression analyses revealed strong linear relationships between pulse width and changes in all kinematic variables. As before, ipsiversive changes occurred in body angle, head yaw, and head roll, while stimulation of either STN increased head pitch (Table 3; Fig. 5, left). Significant results from two-way mixed-effects models elucidated the relationship between stimulation hemisphere and experimental group in the 40-ms condition. Interactions obtained from the analyses of body angle, head yaw, and head roll confirmed the ipsiversive directionality of the movement, while this combined with the main effect of experimental group in the analysis of head pitch confirmed the existence of the movement versus controls (main effect head pitch: F(1,9) = 20.15, p = 0.0015; Table 4; Fig. 5, middle left).
Linear regression results for STN ChR2+ laser pulse width manipulation
Interaction between hemisphere and experimental group caused by 40-ms single laser pulse stimulation
Photo-stimulation of STN ChR2+ neurons causes movement (pulse width manipulation). A, Left, Linear relationship between unilateral stimulation pulse width and ipsiversive turning. Middle left, 40-ms unilateral stimulation causes significant ipsiversive turning versus controls. Middle right, Linear relationship between pulse width and leftwards turning with bilateral stimulation. Right, 40-ms bilateral stimulation does not cause significant turning versus controls. B, Left, Linear relationship between pulse width and ipsiversive head yaw with right STN stimulation. Middle left, 40-ms unilateral stimulation causes significant ipsiversive head yaw across hemispheres, and versus controls in the right hemisphere. Middle right, No relationship between pulse width and head yaw with bilateral stimulation. Right, 40-ms bilateral stimulation does not cause head yaw. C, Left, Linear relationship between unilateral stimulation pulse width and ipsiversive head roll. Middle left, 40-ms unilateral stimulation causes significant ipsiversive head roll across hemispheres, and versus controls in the left hemisphere. Middle right, Linear relationship between pulse width and leftwards head roll with bilateral stimulation. Right, 40-ms bilateral stimulation does not cause significant head roll versus controls. D, Left, Linear relationship between pulse width and upwards head pitch with unilateral stimulation. Middle left, 40-ms unilateral stimulation causes significant upwards head pitch versus controls. Middle right, Linear relationship between pulse width and upwards head pitch with bilateral stimulation. Right, 40-ms bilateral stimulation causes a significant increase in head pitch versus controls. n = 7 ChR2 and 4 control mice. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Error bars indicate SEM.
For bilateral stimulation, regression analyses revealed linear relationships between pulse width and changes in body angle, head roll, and head pitch. Body angle and head roll went to the left, the direction of the strongest unilateral effect, with a slope roughly half as steep as was observed from left hemisphere unilateral stimulation. Head pitch, on the other hand, increased with a slope roughly twice as steep as was observed from unilateral stimulation of either hemisphere alone (Table 3; Fig. 5, middle right). When the 40-ms stimulation-induced movement was tested for significance versus controls, only the increase in head pitch was significant (two-tailed t test: t(9) = 1.266, p = 0.2372; t(9) = 0.9324, p = 0.3755; t(9) = 1.063, p = 0.3157; t(9) = 3.790, p = 0.0043; Fig. 5, right). This is again consistent with the interpretation that bilateral STN stimulation produces a movement consistent with the sum of the unilateral stimulation effects, although it is clear a leftwards bias exists in our sample.
Movement onset latency
We also examined the mean latency to movement onset for each combination of kinematic variable and unilateral STN. Movements caused by 40-ms unilateral laser pulses were chosen for analysis because they reliably cause changes in all kinematic variables. Body angle changes began 12.26 and 15.78 ms after left and right STN stimulation onset (two-tailed t test: t(6) = 3.561, p = 0.0119). Head yaw changes began 11.68 and 11.75 ms after left and right STN stimulation onset (two-tailed t test: t(6) = 0.03,963, p = 0.9697). Head roll changes began 13.42 and 14.55 ms after left and right STN stimulation onset (two-tailed t test: t(6) = 0.8264, p = 0.4402). Head pitch changes began 12.84 and 13.16 ms after left and right STN stimulation onset (two-tailed t test: t(6) = 0.1506, p = 0.8852; Fig. 6). The extremely fast movement onset latency suggests the effect may be through the STN's projection to the brainstem MLR.
Mean latency to move from the onset of a single STN-targeted 40-ms laser pulse. n = 7 ChR2 mice. *p < 0.05. Error bars indicate SEM.
Single pulse movement velocity, effect and rebound
We next examined mean body and head angular velocities 60 and 120 ms after unilateral single pulse stimulation onset because velocity peaked in the effect direction after 60 ms and then rebounded to the opposite direction 60 ms later in head angle measures. For head yaw and roll, this meant a contraversive rebound movement followed the initial stimulation-driven ipsiversive movement, and for head pitch, a downwards movement followed the initial upwards movement. No rebound occurred in measure of body angle, although the stimulation-driven increase in ipsiversive turning velocity was quite clear.
Specifically, at the 60-ms time point, linear relationships were observed between stimulation pulse width and peak effect velocity in all kinematic measures, except left hemisphere head yaw. At the 120-ms time point, linear relationships were observed between stimulation pulse width and peak rebound velocity in measures of left and right hemisphere head roll, as well as left hemisphere head pitch (Table 5). In the 40-ms stimulation condition, the effect and rebound movements were evident in measures of velocity at the 60- and 120-ms time points, respectively, for all measures of head angle except left hemisphere head yaw (Table 6; Fig. 7).
Linear regression results for analysis of single pulse manipulation driven changes in velocity
Interaction between time point and experimental group on angular velocity caused by 40-ms single pulse laser stimulation illustrates presence of effect and rebound movement
Single pulse photo-stimulation of STN. A, Mean turning velocity. Left, Linear relationship between pulse width and ipsiversive turning velocity 60 ms after left STN stimulation. No relationship 120 ms poststimulation. Middle left, Significant interaction (see Table 5) between time point and experimental group on turning velocity following left STN 40-ms pulse stimulation. Significant ipsiversive turning velocity at 60-ms time point. Middle right, Linear relationship between pulse width and ipsiversive turning velocity 60 and 120 ms after right STN stimulation. Right, Significant interaction between time point and experimental group on turning velocity following right STN 40-ms pulse stimulation. Significant ipsiversive turning velocity only at the 60-ms time point. B, Mean head yaw velocity. Left, No relationship between pulse width and yaw velocity 60 or 120 ms after left STN stimulation. Middle left, No effect of time point or experimental group on yaw velocity following left STN 40-ms pulse stimulation. Middle right, Linear relationship between pulse width and ipsiversive yaw velocity 60 ms after right STN stimulation. Right, Significant interaction between time point and experimental group on yaw velocity following right STN 40-ms pulse stimulation. Stimulation caused significant ipsiversive and contraversive yaw velocity at the 60- and 120-ms time points, respectively. C, Mean head roll velocity. Left, Linear relationships between pulse width and roll velocity 60 and 120 ms after left STN stimulation, respectively. Middle left, Significant interaction between time point and experimental group on roll velocity following left STN 40-ms pulse stimulation. Stimulation caused significant ipsiversive and contraversive roll velocity at the 60- and 120-ms time points, respectively. Middle right, Linear relationships between pulse width and ipsiversive and contraversive roll velocity 60 and 120 ms after right STN stimulation, respectively. Right, Significant interaction between time point and experimental group on roll velocity following right STN 40-ms pulse stimulation. Stimulation caused significant ipsiversive and contraversive roll velocity at the 60- and 120-ms time points, respectively. D, Mean head pitch velocity. Left, Linear relationships between pulse width and upwards and downwards pitch velocity 60 and 120 ms after left STN stimulation, respectively. Middle left, Significant interaction between time point and experimental group on pitch velocity following left STN 40-ms pulse stimulation. Stimulation caused significant upwards and downwards pitch velocity at the 60- and 120-ms time points, respectively. Middle right, Linear relationship between pulse width and upwards pitch velocity 60 ms after right STN stimulation. Right, Significant interaction between time point and experimental group on pitch velocity following right STN 40-ms stimulation. Stimulation caused significant upwards and downwards pitch velocity at the 60- and 120-ms time points, respectively. n = 7 ChR2 and 4 control mice. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Error bars indicate SEM.
The head rebound movement is notable because it indicates that stimulation induced changes in head yaw, roll, and pitch were quickly corrected, and thus perceived by the mice as involuntary. Consistent with the rebound being a correction of undesired movement, roll and pitch rebound velocities scaled linearly with stimulation pulse width, and thus effect magnitude. The perceived involuntary nature of the movement is also consistent with the possibility that it is driven by the STN's projection to the MLR.
Rebound in 10-Hz stimulation trains
The head rebound is also caused by individual laser pulses in unilateral 10-Hz stimulation trains. To demonstrate this, we illustrate mean pulse driven changes in velocity and angle for all kinematic measures, across time. The rebound is clearly evident in the velocity and angular change plots for roll and pitch following stimulation of either STN, as well as for yaw following stimulation of the left STN. For yaw and roll, velocity first peaks in the direction ipsilateral to the stimulated hemisphere, then peaks again in the contralateral direction. This can be seen as an initial ipsiversive movement followed by a contraversive correction in the angular change plots. For pitch, velocity peaks first in the upwards direction, then peaks again in the downwards direction. This too can be seen in the angular change plots (Fig. 8).
Individual laser pulses delivered to STN ChR2+ neurons as part of 10-Hz photo-stimulation trains cause head rebound movement. Left, Mean velocity change caused by left STN stimulation. Middle left, Mean velocity change caused by right STN stimulation. Middle right, Mean angular change caused by left STN stimulation. Right, Mean angular change caused by right STN stimulation. A, Body angle. B, Head yaw. C, Head roll. D, Head pitch. n = 7 ChR2 and 4 control mice. Error bars indicate SEM.
STN-MLR terminal stimulation
To test whether stimulation of the STN's projection to the MLR was sufficient to produce the movement caused by whole STN stimulation, we injected ChR2 into the STN and implanted optic fibers bilaterally in the MLR. ChR2 and control mice were stimulated unilaterally and bilaterally for 500 ms at 5, 10, 20, and 40 Hz, and stimulation-induced changes in body angle, head yaw, head roll, and head pitch were quantified, just as was done for whole STN stimulation.
For unilateral stimulation, regression analyses revealed strong linear relationships between stimulation frequency and changes in all kinematic variables, except head yaw. Ipsiversive changes occurred in body angle, head yaw, and head roll, while stimulation of either STN increased head pitch (Table 7; Fig. 9, left). Significant results from two-way mixed-effects models elucidated the relationship between stimulation hemisphere and experimental group in the 40-Hz condition. Interactions obtained from the analyses of body angle, head yaw, and head roll confirmed the ipsiversive directionality of the movement, while this combined with the main effect of experimental group in the analysis of head pitch confirmed the existence of the movement versus controls (main effect head pitch: F(1,12) = 81.17, p < 0.0001; Table 8; Fig. 9, middle left).
Linear regression results for STN ChR2+ terminal stimulation in the MLR
Interaction between hemisphere and experimental group caused by 40-Hz STN-MLR terminal stimulation
Photo-stimulation of STN ChR2+ terminals in the MLR reproduces the effect of STN cell body stimulation. A, Left, Linear relationship between unilateral stimulation frequency and ipsiversive turning (right hemisphere trending). Middle left, 40-Hz unilateral stimulation causes significant ipsiversive turning versus controls. Middle right, No relationship between stimulation frequency and turning with bilateral stimulation. Right, 40-Hz bilateral stimulation does not cause turning. B, Left, No clear linear relationship between unilateral stimulation frequency and ipsiversive head yaw. Middle left, 40-Hz unilateral stimulation causes significant ipsiversive head yaw across hemispheres, and versus controls in the left hemisphere. Middle right, No relationship between stimulation frequency and head yaw with bilateral stimulation. Right, 40-Hz bilateral stimulation does not cause head yaw. C, Left, Linear relationship between unilateral stimulation frequency and ipsiversive head roll. Middle left, 40-Hz unilateral stimulation causes significant ipsiversive head roll versus controls. Middle right, No relationship between stimulation frequency and head roll with bilateral stimulation. Right, 40-Hz bilateral stimulation does not cause head roll. D, Left, Linear relationship between unilateral stimulation frequency and upwards head pitch. Middle left, 40-Hz bilateral stimulation significantly increases head pitch versus controls. Middle right, Linear relationship between stimulation frequency and upwards head pitch with bilateral stimulation. Right, 40-Hz bilateral stimulation significantly increases head pitch versus controls. n = 4 ChR2 and 4 control mice. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Error bars indicate SEM.
For bilateral stimulation, regression analyses did not reveal a relationship between stimulation frequency and changes in body angle, head yaw, and head roll, however, there was a clear linear relationship between stimulation frequency and increased head pitch (Table 7; Fig. 9, middle right). Consistent with the regression analyses, 40-Hz stimulation did not cause a significant change in body angle, head yaw, or head roll, but it did cause a significant increase in head pitch (two-tailed t test: t(6) = 0.9383, p = 0.3843; t(6) = 1.093, p = 0.3162; t(6) = 0.0199, p = 0.9848; t(6) = 7.256, p = 0.0003; Fig. 9, right).
This pattern of results fits that of whole STN stimulation and demonstrates that stimulation of the STN's projection to the MLR is largely sufficient to reproduce the movement caused by stimulating STN cell bodies. However, there are important differences in the magnitude of the kinematic changes caused by unilateral stimulation of the STN and its MLR projection. While the magnitude of the head pitch increase was similar across stimulation targets, the magnitude of the ipsiversive changes in body angle, head yaw, and head roll caused by STN stimulation were two-to-three times greater than the ones caused by stimulating the STN's projection to the MLR. This leaves open the possibility that the STN's projection to the SNr may also play a role in generating ipsiversive movements.
STN-MLR terminal stimulation causes rebound movement
To test whether stimulation of STN projections to the MLR also caused a rebound movement, we once again examined movements caused by individual laser pulses in unilateral 10-Hz stimulation trains. As was the case for STN cell body stimulation, the rebound occurred in roll and pitch following stimulation of either hemisphere, as well as for yaw following stimulation of the left hemisphere (Fig. 10).
Individual laser pulses delivered to STN ChR2+ terminals in the MLR as part of 10-Hz photo-stimulation trains also cause head rebound movement. Left, Mean angular change caused by left hemisphere STN-MLR stimulation. Right, Mean angular change caused by right hemisphere STN-MLR stimulation. (A) Body angle. (B) Head yaw. (C) Head roll. (D) Head pitch. n = 4 ChR2 and 4 control mice. Error bars indicate SEM.
Discussion
Selective activation of glutamatergic (vGlut2+) STN projection neurons produces systematic and highly predictable movements. Unilateral stimulation produces body turning and head yaw and roll movements in the ipsiversive direction, toward the side of stimulation, while stimulation of either hemisphere causes changes in pitch, raising the head. Bilateral stimulation doubles the increase in head pitch caused by unilateral stimulation, but it does not cause body turning or changes in head yaw and roll. Bilateral stimulation generates a movement equal to the sum of the movements caused by unilateral stimulation. All these kinematic changes are a linear function of stimulation parameters with short latencies. Notably, stimulation-induced movements did not differ as a function of optic fiber placement location (Figs. 3 and 5). However, as the mouse STN is very small, it is possible that photo-stimulation activated most of the STN regardless of where the fibers were placed.
We also verified that our stimulation parameters generate movement by driving STN output. In ChR2+ neurons, there is a strong linear relationship between stimulation frequency and firing rate, and between stimulation-induced firing rate and the co-recorded frequency dependent changes in movement kinematics (Fig. 4).
The short latency from stimulation onset to movement initiation is also notable. On average, movement began around 10–15 ms after stimulation onset (Fig. 6). The short latency and high precision of stimulation-induced movement suggests that STN output can rapidly influence effectors to generate movement. This is likely enabled by the STN's projection to the MLR. When this projection was stimulated, it was able to reproduce the basic changes in body angle, head yaw, roll, and pitch caused by STN cell body stimulation (Fig. 9). It is possible that the STN-MLR terminal stimulation effect could be caused by antidromic activation of STN cell bodies and thus the activation of another STN projection, but this is unlikely given the latency of movement onset. The shortest path from the STN to a primary motor neuron is through the MLR.
Head angle changes caused by single light pulse stimulation of the STN are often quickly corrected by the animal if the pulse is not immediately followed by another. There was often a “rebound” movement in the opposite direction after 50–60 ms, bringing the head back toward the baseline position (Fig. 7). When stimulation frequency increases from 10 to 20 Hz, the gap between pulses shortens, so there is no time to recover between pulses. The quick correction of movement could be an indirect result of activating the STN's projection to the MLR. With descending projections to the lower brainstem and spinal cord, the MLR is perfectly situated to cause rapid deviations in posture (Garcia-Rill et al., 1986). By injecting an artificial signal into the MLR via stimulation of the STN, we could have induced an involuntary deviation in posture that acted as a disturbance. Under natural conditions, movements caused by the STN may not be reflexively corrected, because they would be aligned with top-down command signals.
While our results suggest that the STN's projection to the MLR is responsible for the stimulation-induced movements, the role of the STN's projection to the SNr should not be ruled out. Unilateral stimulation of the STN's MLR projection was not able to produce half as much ipsiversive change in body angle, head yaw, and head roll as STN cell body stimulation, but it did fully increase head pitch. This discrepancy could be because of the placement of the optic fibers within the relatively large MLR, but it could also indicate that the STN's SNr projection contributes to ipsiversive movement generation. Although the STN (or the indirect pathway) has long been thought to suppress movement, previous work has established that indirect pathway activation can produce ipsiversive turning (Kravitz et al., 2010). Via either direct excitation of the SNr by the STN, or disinhibition through the GPe, indirect pathway output is expected to increase SNr output. Our observation is therefore consistent with ipsiversive turning observed after indirect pathway stimulation. Further, many SNr neurons show high correlations with postural disturbances and head position, suggesting that SNr output may influence downstream circuits to specify head position (Barter et al., 2014, 2015; Yin, 2017).
Recently, Ji et al. (2023) demonstrated that unilateral optogenetic stimulation of the STN's projections to both the GPi and SNr reduced distance traveled. But they did not employ any other behavioral measures, so it is impossible to know whether normal locomotion was replaced by ipsiversive turning or grooming, as has been observed in other STN stimulation studies (Guillaumin et al., 2021; Parolari et al., 2021; Ji et al., 2023). STN projections to the SNr could still contribute to the generation of ipsiversive movements.
Previous STN stimulation experiments in the open field differ from our experiments in important ways. Previous work compared differences in gross behavioral variables such as locomotion, turning, or grooming across long stimulation and nonstimulation epochs, whereas the present study used short-duration stimulations and discovered rapid, parametric changes in head and body angle. Each previous experiment revealed decreased locomotion during stimulation, but increased grooming was also found, and Guillaumin et al. (2021) observed occasional jumping, too. Additionally, each experiment that measured turning during unilateral stimulation detected more ipsiversive turns (Guillaumin et al., 2021; Parolari et al., 2021; Ji et al., 2023). These results were interpreted through the lens of canonical basal ganglia models by focusing on the reduced locomotion. However, the decrease in locomotion during stimulation could be the result of increased grooming, and the occasional jumping could be the consequence of intense stimulation driving a sustained increase in head pitch.
Parolari et al. (2021) concluded that ipsiversive turning was because of motor deficits in the contralateral limbs. Our results do not support this interpretation. We found that stimulation rapidly generated ipsiversive turning. In a stationary mouse, such turning cannot be explained by stimulation-induced motor deficits in the contralateral limbs.
In rats, GPe neurons (arkypallidal) are selectively activated on successful stop trials just before cancellation of movement-related activity in the striatum (Schmidt et al., 2013; Mallet et al., 2016; Schmidt and Berke, 2017). It was concluded that these neurons project directly to the striatum to stop ongoing behavior. Higher STN activity was also associated with longer reaction times, suggesting that the STN could pause behavior. However, the only behavioral measures used were beam breaks that detect nose pokes. With such measures it is impossible to know whether movements were suppressed. Alternatively, their results could be explained by STN's role in causing ipsiversive movement and head deviation.
Our results also have implications for understanding STN DBS (Vitek, 2008). According to the standard model, dopamine depletion causes hypoactivity of direct pathway neurons and hyperactivity of indirect pathway neurons, resulting in excessive inhibition of prokinetic thalamic and brainstem targets (Albin et al., 1989; Alexander and Crutcher, 1990; Chevalier and Deniau, 1990; DeLong, 1990). It is commonly believed that STN DBS suppresses STN output, thereby making it easier for the direct pathway to disinhibit the thalamus and brainstem and restore normal movement (Chevalier and Deniau, 1990; Benazzouz et al., 1995; Benabid et al., 2001; Beurrier et al., 2001; Wichmann et al., 2018). However, it remains unclear whether STN DBS suppresses STN output, or disrupts abnormal bursting, synchronization, and oscillations in the basal ganglia (Hashimoto et al., 2003; Degos et al., 2005; Meissner et al., 2005; Shi et al., 2006; Hammond et al., 2007; Galvan and Wichmann, 2008; Hahn et al., 2008; McConnell et al., 2012; Zhuang et al., 2018; Yu et al., 2020). Recent work found that selective activation of glutamatergic projections from the parafascicular nucleus of the thalamus to the STN restores normal movement in a mouse model of PD (Watson et al., 2021). These results also support the hypothesis that STN output can directly generate movements. Moreover, in dopamine-depleted mice with complete akinesia, activation of the Pf-STN pathway at 25 Hz is sufficient to rescue movements. This frequency is comparable to the frequencies used here, but much lower than that used in traditional STN DBS (∼130 Hz). Since traditional DBS using electrical stimulation is not specific to cell bodies or a specific population of projection neurons, it could activate a combination of afferent terminals from multiple areas as well as fibers of passage. Our results suggest that the clinical efficacy of DBS could be because of the activation, rather than suppression, of STN neurons. Although recent work has demonstrated that inhibition of the STN's projections to either the GPi or SNr is sufficient to rescue movement in a mouse model of PD, so the potential benefit of STN activation would be limited to the MLR projection (Ji et al., 2023).
In short, using precise targeting of STN output neurons, we revealed a highly linear relationship between stimulation parameters and kinematics, suggesting that the STN can quantitatively scale movements, especially turning in the ipsiversive direction. These results question the prevailing model that the STN functions to suppress movement, and instead suggests that STN output plays an active role in scaling movement parameters.
Footnotes
This work was supported by National Institutes of Health Grants NS094754 and MH112883 (to H.H.Y.). We thank Konstantin Bakhurin for help with data analysis, Guozhong Yu and Marina Roshchina for help with surgery and histology, and Francesco Paolo Ulloa Severino for help with Adobe Illustrator.
The authors declare no competing financial interests.
- Correspondence should be addressed to Henry H. Yin at hy43{at}duke.edu
