Parkinsonism originates in a discrete secondary and dystonia in a primary motor cortical‐basal ganglia subcircuit

Although manifesting contrasting phenotypes, Parkinson's disease and dystonia, the two most common movement disorders, can originate from similar pathophysiology. Previously, we demonstrated that lesioning (silencing) of a discrete dorsal region in the globus pallidus (rodent equivalent to globus pallidus externa) in rats and produced parkinsonism, while lesioning a nearby ventral hotspot‐induced dystonia. Presently, we injected fluorescent‐tagged multi‐synaptic tracers into these pallidal hotspots (n = 36 Long Evans rats) and permitted 4 days for the viruses to travel along restricted connecting pathways and reach the motor cortex before sacrificing the animals. Viral injections in the Parkinson's hotspot fluorescent labeled a circumscribed region in the secondary motor cortex, while injections in the dystonia hotspot labeled within the primary motor cortex. Custom probability mapping and N200 staining affirmed the segregation of the cortical territories for Parkinsonism and dystonia to the secondary and primary motor cortices. Intracortical microstimulation localized territories specifically to their respective rostral and caudal microexcitable zones. Parkinsonian features are thus explained by pathological signaling within a secondary motor subcircuit normally responsible for initiation and scaling of movement, while dystonia is explained by abnormal (and excessive) basal ganglia signaling directed at primary motor corticospinal transmission.


| INTRODUC TI ON
Parkinson's disease (PD) is characterized by paucity, diminutive, and slow movements.In contrast, dystonia is heralded by excessive and sustained, uncontrolled involuntary movements with cocontractions of antagonistic muscles.These dissimilar conditions can originate from similar pathophysiology and commonly co-occur.
A deficiency in DA causes disinhibition of GABAergic DA D2 receptor inhibitory striatal neurons, which is expected to silence globus pallidus externa (GPe) neurons (Alexander, 2004;Tritsch & Sabatini, 2012).To more directly elucidate the behavioral and downstream neuronal effects of silencing GPe neurons, we previously injected the neurotoxin ibotenate into the posterolateral motor territory of GP (rodent equivalent of GPe) in rats (Kumbhare et al., 2017).
Larger motor territory ibotenate lesions induced both parkinsonism and dystonia, while more restricted lesions separately induced these two clinical features (Kumbhare et al., 2017).A circumscribed dorsal hotspot lesion produced pure characteristic parkinsonian features, including severe immobility and contralateral forelimb flexion posturing.In distinction, a focal ventral hotspot lesion induced isolated, prominent contralateral and truncal dystonic extension posturing, with characteristic EMG dystonic co-contraction activity.In ibotenate-lesioned dystonic rats, the downstream abnormal neuronal discharge rates and patterned activity in the entopeduncular nucleus (EP; rodent equivalent of the GPi internus (GPi)) were indistinguishable from those recorded in jaundiced, kernicterus dystonic rats.The EP neuronal activity in parkinsonian lesioned rats was typical of that in humans with PD, with high discharge rates, prominent burst and 4-6 Hz oscillatory activity (Francois et al., 2004).The projections from GP to EP and GPe to GPi innervate topographically equivalent discrete regions of these nuclei (Lee, 2020).Therefore, to further address the translatability of these findings to humans, we scrutinized our blinded postoperative selection of effective GPi deep brain stimulation (DBS) contacts (4 metal contacts, 1.5 mm in length, with .5 mm separation) in PD and dystonia patients, typically placed with the lowest contact at the bottom and the third contact near the top of GPi.As predicted from with our rodent studies, the effective DBS contact for treating PD was located in the dorsal posterolateral motor territory of GPi and for dystonia was located nearby, more ventrally (Kumbhare et al., 2017).
The BG are organized in largely segregated motor, associative, and limbic reentrant BG thalamocortical (BGTC) circuits (Alexander et al., 1986;Francois et al., 2004).Furthermore, our previous findings suggested that disturbances in distinct BGTC motor subcircuits can give rise to contrasting movement conditions.Based on the locations of GPi labeling from multi-synaptic retrograde viral tracer injections in different motor cortical territories in primates (Akkal et al., 2007), we hypothesized that the dorsal Parkinson's GPe-GPi projections contribute to a BG-supplementary motor area (SMA) subcircuit and the ventral dystonia projections to a BG-primary motor subcircuit.These proposed cortical territorial contributions to PD and to dystonia are supported by a majority of (though not all) functional imaging studies (Eidelberg et al., 1995;Haslinger et al., 2010;Playford et al., 1992;Pujol et al., 2000;Simonyan & Ludlow, 2010).
We reasoned that higher order motor signaling disturbances in SMA could account for the poor initiation and abnormally scaled features of parkinsonism, while excessive, poorly regulated activation of primary motor cortex projections to the spinal cord could explain dystonic motor features.To test our hypothesis, we injected fluorescent-tagged multi-synaptic tracers into the Parkinson's and dystonia GP hotspots and defined the resultant cortical labeling.

Note:
The gray-shaded rows represents rat brains that were used for the GP-cortical correlation.Bold values represents the primary labeled site.
*Parkinsonian rat shown in Figure 1b1.^Dystonic rat shown in Figure 1b2.
+ In animals with prolonged incubations beyond 4 days, the cortical fluorescent labeling consistently advanced extensively beyond the primary cortical projection sites.All experiments were conducted in strict accordance with the ARRIVE guidelines.To increase scientific rigor, rats were randomly chosen for the experiments, irrespective of the strain, sex, age, and weight.A total of 22 females and 14 males were used, age range: 8-48 weeks old and weights: 200-720 g at the time of stereotactic injection of the tracers.See Table 1 for details of the rats and stereotactic injections.All data were reviewed by multiple team members to ensure its validity and to minimize operator biases.

| Viral vector
All tracing experiments utilized the transsynaptic vesicular stomatitis virus (VSV) native glycoprotein (G) VSV-Venus [VSV-G VSV-Venus] Replication Competent vector (7.56E + 10 TU/mL), obtained from the Salk Institute Viral Vector Core and aliquots (5 μL) were stored at −80°C.Center of Disease Control guidelines for viral handling were strictly followed.

| Surgery and target localization
The stereotactic injection surgeries were carried out under 2%-4% isoflurane anesthesia (with 1 L/min oxygen) using sterile techniques.
Adequate depth of sedation was assured by regularly assessing for responses to toe pinch.Ophthalmic ointment was applied to protect the eyes during surgery.The rat's body temperature was monitored and maintained via rectal temperature probe and a feedbackcontrolled heating pad.Respiration and heart rate were monitored at regular intervals.The fur on the rat's head was clipped.Then, the position of the rat's head was secured with a bite bar, non-rupturing ear bars, and the stereotactic device (KOPF).After disinfection with betadine, an incision was made on the top of the head and the overlying tissue was scraped to expose the skull.To ensure proper targeting, the pitch of the rat's head was measured using a dial indicator and adjusted to make the head level.A 2-5 mm burr hole was made with a drill bit in the skull centered above the targeted GP (L 3.3 mm, AP −1.6 mm referenced to bregma).Prior to viral tracer injection, the location of the GP dystonia or parkinsonian hotspot target was confirmed via stereotaxic microelectrode recording (MER), typically over 2-3 penetrations, utilizing ultrafine 100 μm heptodes (Thomas Recording, GmbH) mounted in a 7-heptode capacity Eckhorn manipulator (Thomas Recording) attached to the KOPF stereotaxic system.Refer to our prior publication (Kumbhare et al., 2015) for additional details of the mapping procedures.

| Tracer injection
Prior to initiating MER, a glass micropipette (80 μm I.D., Thomas Recording) was secured in one of the heptode positions in the Eckhorn manipulator and connected via Tygon flexible tubing (.25 mm I.D.) to a Hamilton syringe mounted in a syringe pump.To assure the administration of the viral load, 2-4 μL of Dulbecco's phosphate buffered saline (dPBS) was drawn first into the micropipette with a small air bubble (~2 mm).dPBS was used, rather than water or regular PBS, to minimize the risk of hydrocephalus.Also, some of the utilized tracer agents are prepared in Dulbecco's PBS.
Next, fluorescent tagged VSV tracer solution (.2-1 μL) was drawn into the micropipette.After confirming the precise location of the intended dorsal versus ventral GP target using MER, the tracer was unilaterally focally injected in the right GP at a rate of 100-200 nL/ min while monitoring the movement of the air bubble.To minimize leakage of tracer on withdrawal of the micropipette, the system was left in place for an additional 10 min prior to retracting.
After completion of a tracer injection, the burr hole was sealed with bone wax and the incision was sutured closed.Bupivacaine was injected into the skin around the incision and buprenorphine (.25-1.6 mg/kg, i.p.) was administered prior to discontinuing the isoflurane.Subsequently, rats were placed in a clean cage and allowed 24 h for recovery.

| Behavioral assessment
Successful discrete targeting of injections into either the dorsal or ventral GP hotspots was affirmed by induction of selective motor conditions, parkinsonism or dystonia, respectively, upon delayed local virus-induced neuronal degeneration.The health status and motor behavior of rats were assessed at least twice daily following injections.Health status monitoring included weights, pain and stress indicators, including vocalization, posture, grooming, porphyrin staining, and suture condition.Motor behavior assessments included recording of the level and direction of any turning tendency.Parkinsonian and dystonic symptoms were quantified based on previously reported rating scales [appendix B of (Kumbhare F I G U R E 1 Induction of pure parkinsonism or dystonia affirms the successful injection of VSV in the respective GP motor territory hotspot.Box and whiskers plots of motor scores (0 = nl to 4 = marked disability) for rats injected with VSV in the dorsal and ventral GP are shown.Outliers are plotted individually using the "+" marker symbol.Previously reported (Kumbhare et al., 2017) behavioral results from dorsal and ventral GP ibotenate lesions are shown for comparison.(a) Beginning typically 4 days after VSV injections in the dorsal GP hotspot (n = 5), the rats began to display (a1) paucity of spontaneous generalized movement, (a2) minimal response to audio and tactile stimuli, and (a3) prominent contralateral forelimb (and hindlimb) hypokinesia.(b) In contrast, rats with ventral GP hotspot injections (n = 5) showed essentially normal occurrences of spontaneous and stimulation-induced movement, while displaying prominent (b1) contralateral dystonic limb extension, (b2) neck and truncal twisting, and (b3) dystonic posturing related falls.The central red marks in each box indicate the median, and the bottom and top edges of the box indicate the 25th and 75th percentiles, respectively.The outliers are plotted individually using the "+" marker symbol.Statistically significant (p < .05)comparisons are indicated as "*".  of 0-12).For features presenting midway in severity between one category and the next higher one, an additional .5 was added.

| Histology and imaging
After 1-7 days of viral incubation, the rats were sacrificed with an overdose of pentobarbital and transcardiac perfusion via the left ventricle, with saline (200 mL), followed by 4% formaldehyde solution (300 mL).The brains were extracted and initially placed in 4% formaldehyde overnight for post-fixation.The brains then were placed in 30% sucrose in PBS solution until the brains sank (~2 days) for cryopreservation.The cerebrum was vertically bisected along the interhemispheric fissure.The injected side of the cerebral hemispheres was then molded in Tissue-Tek OCT (Sakura).Using a Leica CM1850 cryostat, the brains were then sliced into sagittal sections (50 μm), typically saving every third section from approximately lat..5 to 4.0 mm for a total of 24 sections and floated in PBS.The sections were next mounted on slides (Fisherbrand Superfrost Plus) and stained with fluorescent nuclear stain 4′,6-diamidino-2-phenylindole (DAPI).
In the earliest experiments, prior to the acquisition of a microscopy capable of directly imaging yellow fluorescent proteins (YFP), fluorescence was enhanced using standard immuno-fluorescent techniques.Briefly, after mounting on slides, tissue were placed in acetone for antigen retrieval, blocked with normal goat serum, then treated with rabbit/anti-GFP primary antibody.After incubation, rinsing, and additional blocking, the tissue were treated with goat anti-rabbit secondary antibody conjugated with Alexa 488 fluorescent dye for visualization with standard GFP filter sets.
In most cases, the brain sections were imaged with a Keyence BZ-X800 digital microscope.The slides were scanned with 4× objective lens (Nikon PlanFluor DL 4× .13/16.50 mm PhL) in two fluorescence channels: EYFP, DAPI (CHROMA 49003-UF1-ET-EYFP, 49000-UF1-ET-DAPI), and in phase contrast visible light.For each channel, each slide was scanned as a 231-image mosaic composed of 960 × 720 pixel 8-bit monochrome images; 21 columns × 11 rows with approximately 15% overlap.The Keyence image stitching feature was then employed to produce a single large 14,405 × 5768pixel image for each channel (Figure S1c).Also, to investigate the extent and timing of delayed VSVinduced neuronal damage in the injected hotspot areas, we additionally stained sections from select VSV injected rats with silver stain.For silver staining methodological details, refer to our prior publication (Kumbhare et al., 2017).The fluorescent cells that were readily identifiable as neurons based on their morphology were considered.

| N200 labeling
N200 histology was performed using a standard primary/secondary antibody immunofluorescence process for floating tissue.Briefly, formalin fixed brains were immersed in 30% sucrose/PBS solution until the tissue sank.50 μm thick tissue sections were sliced in the sagittal plane on a cryostat and placed in PBS in individual wells of a 24-well plate.Tissue sections were first immersed in 3% hydrogen peroxide for 1 h.Sections were rinsed three times with PBS/Triton X-100 solution and then blocked in a solution of PBS/Triton X-100/ normal donkey serum for 1 h.Each section was then treated with 250 μL of primary antibody solution consisting of PBS, Triton X-100, donkey serum, and rabbit anti-neurofilament 200 (Sigma) diluted 1:400.The tissue was incubated overnight at 4°C, and then rinsed three times and blocked again for 1 h.Secondary antibody solution, 250 μL/ well, similar to the primary solution but with Alexa Fluor 568 donkey anti-rabbit (Invitrogen) diluted 1:500 was then applied and incubated at room temperature for 2 h.The tissue was again rinsed three times before mounting on slides.After drying for 10-30 min, the slides were cover-slipped with Prolong Glass antifade mounting media (Invitrogen).The mounting media were allowed to cure overnight before microscope imaging.

| Registration and reconstruction
Single-channel Keyence images of entire sagittal sections were imported into GIMP image software [version 2.10.12] as individual layers and overlaid.This allowed for each channel to be selectively switched on and off for visual analysis of the fluorescent labeling within the context of the whole brain slice (Figure S1).Registration of the sagittal slices was then performed by cropping and rotation of the individual brain slice images before aligning them with respect to each other as separate layers in GIMP.After alignment, slice images were exported in a format to be imported into ImageJ for further analysis.This facilitated 3D reconstruction of each brain hemisphere.Figure S1 shows representative images from each step.The cortical projected neuronal cell body labeling was then digitally registered for each rat using custom MATLAB script employing image processing toolbox [MATLAB 2019A], which also converts imageJ coordinates into atlas coordinates.The location and the spread of injections in GP was similarly registered.The labeling of neuropil and axonal fiber were subjectively removed from the analysis.Only clearly labeled cortical cell bodies were considered.

| Correlating cortical labeling (cell body) with the injection
Tetrachoric correlation was used to assess for significant correlations between the dorsal (parkinsonian) and ventral (dystonia) GP injections and the cortical labeling.For the dorsal GP injected cases, for example, the digitized projection coordinates in M1 and M2 were assigned into categories "dorsal GP-M1" and "dorsal GP-M2," respectively.Similar assignments were performed for ventral GP projections in M1 and M2.The sum of all the labeled coordinates for each of the four paired combinations were calculated.The percentage of each category with respect to total M1 and M2 coordinates was then arranged in a correlation matrix and the tetrachoric correlation was calculated based on following formula:

| Projection probability modeling
A novel 3D spatial algorithm was developed to define statistically significant cortical regions anatomically connected to the two GP hotspots related to parkinsonism and dystonia.This where DV i is the voxel score for ith voxel in rat number n, Ds is the dorsal spread score estimated in point 1 above, d i is the distance between the center of voxel V i and the nearest VSV labeled voxel.
4. Arrays of DV and ventral hotspot (VV) were generated for each of the voxels from the cortical ROI for each rat.For each rat, three scores were separately assigned to each voxel that lies within the volume of lesion spread.Therefore, each voxel was assigned 17 different scores, equal to the number of lesioned animals. 5. Next, for individual cortical labeled voxels, medians and standard deviations were calculated across dorsal lesioned and across ventral lesioned rats.Only voxels that satisfied the following three criteria were accepted as projections from the corresponding GPe hotspot: (i) median score > threshold-1 (Th1, defined below), (ii) standard deviation (SD) < Th2, and (iii) number of non-zero scores > Th3.To define stringent and robust parkinsonian and dystonia cortical hotspots, the thresholds were set as follows: Th1 = 6, Th2 = 1, and Th3 = 3.For Th1, a value of 6 (i.e., >60% of maximum Ds or Vs scores) was chosen to represent high correlations between the cortical labeling and the corresponding GP hotspot.Th2 was selected to assure that the Ds or Vs correlations are consistent across animals.This was performed by excluding voxels with SD >1 (i.e., SD > 10% of the maximum value of Ds or VS).Th3 was selected to ensure that the voxels are consistently labeled across rats (appearing in at least 3 rats).To additionally define a more inclusive representation of the cortical hotspots, we also performed relaxed hotspot analyses.For these analyses, less strict criteria were set by lowering the Th1 and Th2 requirements to Th1 = 5 (i.e., at least 50% of max.Ds or Vs), Th2 = 2 (i.e., SD < 2), while maintaining Th3 = 3.Although the relaxed criteria almost certainly included regions that do not in fact contribute to the defined motor subcircuits, this was included to define the full potential extent of the cortical involvement.For both the strict and relaxed criteria, voxels satisfying the set criteria were then plotted in a final 3D realization representing cortical probability maps (2D realization are presented in Figures 3 and 4, left column).Voxels with score 1 are drawn in the 3D realization and the values were compared.

| Hotspot analysis
The final GP and cortical hotspots defined from the above steps were tested for spatial autocorrelations and significant associations.Global Moran's I function (GMI) (Jackson et al., 2010;Schmal et al., 2017), modified for 3D data (Kumbhare et al., 2017), was used on the filtered voxelated data to assess spatial autocorrelation.GMI classifies the overall spatial distribution as clustered (GMI ~ +1), dispersed (GMI ~ −1), or random (GMI ≈ 0).Next, Getis-Ord GI* statistics (Getis & Ord, 2010), modified for 3D data (Kumbhare et al., 2017), was calculated providing a statistical hotspot analysis using the local pattern of spatial association to identify local spatial clusters with high or low efficacy values in the ROI.For statistically significant positive z-values (GI*; z-score >1.96 (p < .05)), the larger the z-values, the more intense the clustering of hotspots.The probability distribution maps generated in above section (Figures 3 and 4, middle column) were subsequently filtered to include only significant z-scores (GI*) and to generate final maps revealing the statistically significant hotspots.

| Independent hotspot test
To ensure that the two hotspots are spatially significantly different from each other, the distribution of the two hotspot coordinates (distance from bregma) were compared using independent sample test.

| Intracortical microstimulation
The rats were placed under anesthesia using i.p. ketamine (100 mg/ whiskers, shoulder, jaw, tail) were recorded.The rats were perfused immediately after the experiment and the brain was processed to confirm the anatomy.

| Experimental design and statistical analysis
Comparisons between different multiple groups were performed using Kruskal-Wallis test followed by Tukey's honestly significant difference (HSD) multiple comparison test.A probability value of <.05 was considered statistically significant for comparisons between groups.Spatial statistics is discussed in hotspot analysis of "Connectivity modeling" section above.All statistical analyses were conducted using standard MATLAB (MathWorks) functions.

| Select VSV-induced motor features affirm accurate delivery of viral constructs to the targeted motor subcircuit
After several days, VSV begins to be lethal to infected cells (Beier et al., 2011;Kretzschmar et al., 1997).Thus, after a delay, we expected to observe behavioral changes coincident with VSV-induced cytotoxicity (silencing) of GP hotspot neurons at the injection site.Among rats receiving a larger volume of viral construct (.75-.9 μL), moderate to severe parkinsonism (rats #8, 9, 23) or dystonia (rats #11, 14) de-

| Focal GP hotspot injections label distinct cortical motor territories
Post-mortem sections were imaged using a Keyence digital microscope and aligned to Paxinos & Watson rat brain atlas sections (Paxinos & Watson, 2006) (Figure S1b).The coordinates of the cortical projections were then digitally registered (Figure S1c).Of 22 4-DPI rats, 17 (n = 10 predominant dorsal, n = 6 ventral, n = 1 dorsal and ventral GP hotspot injections) were used for the labeling reconstructions (Table 1).The additional five 4-DPI rats were excluded due to missed target, poor perfusion, processing issues or excessive leakage.Among the 17 reconstructed injections, all injections leaked to varying extents into the nearby unintended hotspot with one injection evenly encompassing both hotspots.The injections also all showed varying degrees of leakage into the traversed striatum, reticular thalamus, and substantia innominate.Most injections produced dense cortical neuronal labeling, with three injections (rat # 20, 27, 32) producing more modest labeling due to limited transfection of GP neurons.The dorsal (parkinsonian) GP injections were observed to project predominately to secondary motor cortex (M2; Figure 2b1a,b1b).In contrast, ventral (dystonia) injections predominately projected to primary motor cortex (M1; Figure 2b2a,b2b).

| Probability mapping of the parkinsonian and dystonia cortical territories
The 17 reconstructed injections were next scrutinized in detail and subjected to statistical verification of their GP-cortical relationship.For consistency, the projection modeling studies, with the exception of one older rat, were restricted to younger rats (8-34 weeks).For convenience, available rats, ages 8-48 weeks, For each voxel, median scores were generated across 17 animals.
Cortical voxels were then defined as being related or not related to the dorsal or ventral hotspots based on the following criteria: i. a high median score (≥threshold Th1, defined below), ii.low variability (standard deviation ≤Th2), and ii.repeat labeling (≥Th3).Using stringent criteria, thresholds were set as follows: Th1 = 6, Th2 = 1, Th3 = 3 rats to accommodate 70% active (non-zero) data points in the region of interest.The distribution of these probability maps related to dorsal and ventral GP hotspots were then compared.

TA B L E 2
Tetrachoric matrix percentage paired distributions of GP hotspot injections and M1-M2 projection labeling.

| N200 staining and ICMS testing affirm that the parkinsonian subcircuit is confined to the microexcitable M2 motor territory
Additionally, to better define the borders of M2 and thereby assure that the dorsal GP hotspot subcircuit does not encroach into the cingulate cortex rostrally or into M1 posteriorly, we stained representative brains (n = 3) with N200 (Ueta et al., 2014;Voelker et al., 2004) (see Methods for details).The N200 staining affirmed that the cortical fluorescent labeling from the dorsal GP hotspot injections was confined to M2 (Figure 5a).Although the existence of an SMA equivalent in rodents remains controversial, the motor responses evoked by electrical stimulation in the microexcitable zone in M2 (Deffeyes et al., 2015) and the direct physiological influences of the M2 microexcitable area on M1 (Deffeyes et al., 2015) closely resemble that of SMA.Thus, to better define the M2 parkinsonian territory and its potential relation to the human SMA, as well as to better define the M1 dystonia cortical hotspot, we also performed intracortical microstimulation (ICMS) in additional rats (n = 6).These studies indeed revealed the parkinsonian cortical territory lie chiefly within the M2 microexcitable zone.More specifically, it was found to be localized to the (proximal and distal) forelimb area, encroaching on the more lateral jaw area (Figure 5b).Surprisingly, as for other groups, hindlimb responsive neurons were not identified in M2 with ICMS (Deffeyes et al., 2015).The dystonia cortical hotspot encompassed the M1 microexcitable forelimb area, as well as potentially the hindlimb area.

| DISCUSS ION
We focally injected muti-synaptic tracers in GP in rats and delineated the associated cortical territories of two of the most comment disorders, parkinsonism and dystonia.Consistent with our prior observations of classical parkinsonian motor features from ibotenate lesions in the dorsal GP hotspot and typical dystonic features from lesions in a ventral GP hotspot, rats here displayed these same features beginning around 4 days after VSV injection into these hotspots.
Cortical labeling reconstructions here revealed that parkinsonism originates along a circumscribed dorsal BG-secondary motor cortex subcircuit and dystonia along a ventral BG-primary motor cortex subcircuit (Figure 6b).These distributions were consistent with those from tracer (Akkal et al., 2007;Hoover & Strick, 1993) and electrophysiology (Yoshida et al., 1993) investigations in primates.Hoover and Strick (1993), for example, injected herpes simplex retrograde multi-synaptic tracers into distinct arm regions of the frontal cortex and traced their separate labeling in the posterolateral motor portion of GPi.Comparable to our findings in rodents, the investigators traced the GPi-SMA subcircuit dorsal and adjacent to the more ventral GPi-primary motor subcircuit.
In conflict with the long-standing classical BGTC model, which attempts to explain hypokinetic and hyperkinetic movement disorders by opposing up-and down-signaling within the same motor circuit (Albin et al., 1989;Alexander et al., 1986), our studies bring to light the differential contributions of distinct BGTC motor subcircuits.While Francois and colleagues (Francois et al., 2004) showed that dyskinesia, attention deficits, and stereotypy are inducible along separate motor, associative, and limbic circuits, respectively, our studies revealed that different movement disoriginate along distinct motor subcircuits.Our previous mapping of the effective DBS contact for treating PD to the dorsal posterolateral motor territory of GPi and that for dystonia more ventrally (Kumbhare et al., 2017) further supports the translational relevance of our rodent study findings to primates, and more specifically humans.
Kaneko and colleagues (Kaneko, 2013;Koshimizu et al., 2013;Kuramoto et al., 2015) used GABAergic and glutamatergic staining to distinguish pallidal-versus cerebellar-receiving thalamic motor neurons and completed electron microscopic single neuronal thalamocortical projection reconstructions in rats.Although not an objective of the author's investigations, the figures reveal that pallidothalamic projections to M2 and M1, respectively, maintain their dorsal-ventral relationship within the thalamus.Specifically, neurons in the pallidalreceiving dorsal, rostral portion of the ventral anterior-ventral lateral (VA-VL) thalamic complex are seen to project predominately to M2, while neurons in the pallidal-receiving ventral, rostral VA-VL complex project predominately to M1.Also, pallidal-receiving neurons in the rostral portion of the medial ventral (MV) thalamus are seen to project predominately to M2, while those in the caudal VM project predominately to M1 (Kuramoto et al., 2015).Although motor information processing through the basal ganglia is largely delegated to distinct BGTC motor subcircuits, there, nevertheless, appears to be some crosstalk at the level of the basal ganglia and thalamus.Furthermore, using neural tracing and immunostaining, Karube et al. (2019)   The GP-cortical labeling correlations here show that a circumscribed territory in M1 is integral to a ventral BG-M1 dystonia subcircuit.
further localized the parkinsonian territory chiefly to the M2 forelimb area.Although an SMA equivalent has not, to our knowledge, been previously defined in rodents, additional lines of evidence support that the M2 parkinsonian territory represents the rodent equivalent of SMA.Both anatomical tracer (Hoover & Strick, 1993) and cortical stimulation (Yoshida et al., 1993) investigations in primates have demonstrated that the BGTC SMA forearm subcircuit traverses the posterolateral GPi at mid-depth between the dorsally located primary motor and the ventral premotor forearm subcircuits.This localization and that of our immediate upstream hotspot in GP, in which we previously induced parkinsonism in rats via ibotenate lesions (Kumbhare et al., 2017) and induced here via virus-induced neuronal toxicity in rodents, correspond very well.
Also, as we have reported (Kumbhare et al., 2017), the position of the hotspot for inducing parkinsonism in rodents corresponds well with the circumscribed DBS target in GPi for reversing parkinsonism in PD patients.Also, the SMA, in difference from other premotor areas, provides substantial direct projections to the F I G U R E 5 The parkinsonian and dystonia subcircuits are localized to the M2 and M1 cortical microexcitable zones, respectively.(a) To address whether the parkinsonian cortical hotspot might encroach on the cingulate cortex rostromedially, we first stained select brains (n = 3) with N200.N200 densely labels layer 3-5 neurons in cingulate cortex and M1, while more sparsely labeling these neurons in M2 (Voelker et al., 2004).As illustrated, N200 staining revealed the parkinsonian cortical hotspot (green outlined circle) to be confined specifically to M2.The white arrows indicate rostrally, the margins of cingulate cortex area 1 (cg1) and M2 and caudally, the margins of M2 and M1.(b) Next, we used microstimulation to define the topographical relations of the parkinsonian and dystonia cortical hotspots to the M2 (rostral) and M1 (caudal) microexcitable zones, which denote the two predominant direct cortical projections to the spinal cord (Deffeyes et al., 2015).The parkinsonian (green) cortical hotspot and the statistically relaxed territory (dashed green) can be seen to be primarily confined to the rostral forelimb area (RFA, blue oval) of M2, encroaching on the lateral jaw (light green) area.The dystonia cortical hotspot was chiefly confined to the caudal forelimb area (CFA, purple oval) of M1, encroaching caudally on the hindlimb (orange oval) area.
In difference, not only do lateral premotor area neurons require substantially larger stimulation currents to induce movements but also these area neurons chiefly signal distal, and not proximal, forelimb movements.
The parkinsonian and dystonia-related GP sites targeted here with viral tracers are relatively small, circumscribed regions that are vertically separated by only ~.3 mm.Thus, it was important to incorporate methodology that would reliably distinguish between cortical labeling originating from the two targeted GP sites.
Because we were unable to find adequate prior methodology to account for unavoidable tracer leakage and imprecisions in targeting, we developed our own probability modeling algorithm.To complement the dependability of our algorithms, we also varied the trajectorial approach in different experiments.This served to avoid regular leaking into the same regions and thereby, enhanced our ability to delineate contributions of unintended tracer injections into transversed sites and from extension beyond the intended injection sites.We are confident that the algorithm output accurately delineated cortical voxels that correspond to the corresponding subcircuit of the targeted dorsal versus ventral GP hotspot.Furthermore, by defining high z-scores for the clustered cortical projections, we are confident that the Getis-Ord clustering hotspot analysis effectively masked any cortical labeling originating from leakage or spread.Although the derived final cortical hotspots in M1 and M2 were largely non-overlapping, our analyses did suggest that a small percentage of projections might project more diffusely across regions of premotor and primary motor cortex.While we suggest that this represents a degree of true overlap in the projections, our methodology cannot fully correct for inadvertent overlap of the injections into the nearby unintended hotspot.Furthermore, our methodology cannot completely exclude a component of cortico-cortical tracer spread of the viral tracer.Additional studies utilizing electrophysiological and neuromodulation approaches will be important to further substantiate the extent of segregation of the motor subcircuitry.
We propose that in PD, loss of DA input to STR, chiefly along the SMA-BG reentrant subcircuit, ultimately produces pathological SMA-corticospinal and SMA-primary motor-cortical spinal signaling.Because the SMA contributes to the executive control of movement (Russo et al., 2019;Tanji & Shima, 1994), pathological SMA executive motor signaling could account for the classical poor  et al., 1995;Grafton, 2004;Kübel et al., 2018;Playford et al., 1992), while investigations in dystonic subjects have, in contrast, suggested principal disturbances in primary motor cortex, though also variably implicating prefrontal motor regions (Ceballos-Baumann, , Marsden, & Brooks, 1995;Ceballos-Baumann, Passingham, Warner, et al., 1995;Norris et al., 2020).In further support of the translational implications of our findings to humans, Magno et al. (2019) reported that optogenetic stimulation of ICMS defined M2 pyramidal neurons ameliorated parkinsonian features in DA depleted mice.

Passingham
All experiments were approved and monitored by the Institutional Animal Care and Use Committee of the Hunter Holmes McGuire Veterans Affairs Medical Center and performed in accordance with regulatory guidelines.Long Evans wild-type rats were used for this study.Animals were initially obtained from Charles River, MA, United States and maintained as an in-house breeding colony in the McGuire Research Institute's animal facility.The rats were housed on 12-h light/12-h dark cycle with food and water ad libitum.Animals were housed in groups of two or three per cage before procedures and were single-housed post-surgery.SignificancePreviously, we showed that two common movement conditions, Parkinson's disease (PD) and dystonia, could be separately induced in rats by silencing neurons in distinct, adjoining hotspot territories in the basal ganglia, deep in the brain.Presently, we injected viral tracers into the two identified deep brain hotspots.By tracing the spread of the viruses to the surface of the brain, the PD subcircuit was shown to involve a distinct pre-motor cortical territory, while dystonia involves the principal motor cortex.These findings will push the field to explain other movement disorders by comparable up-or down-signaling in specific motor subcircuits.TA B L E 1 Details for individual rats injected with fluorescent labeled rVSV.
viral dilution.The grey-shaded rows represents rat brains that were used for the Gp-Cortical correlation.Bold Values represents the primary labeled site.
et al., 2017)].Briefly, parkinsonian features were scored from 0 to 4 (normal to extreme) for (1) generalized spontaneous and (2) stimulation-induced movement and for (3) contralateral hypokinesiaflexion posturing.Dystonic features were scored for (1) contralateral fore-and hindlimb extension, (2) midline posturing, and (3) dystoniarelated falls (producing total parkinsonian and dystonia scores each was accomplished by correlating the location and spread of viral infection in GP injection to the distribution of the neuronal projection labeling in the cortex.1.First, the distribution of VSV spread in GP and beyond was established based on visual inspection of fluorescent labeling and/or silver staining of damaged neurons.Scores (from 0 = none to 10 = complete labeling) were then assigned to following: (1) dorsal GP hotspot (Ds) and (2) ventral GP hotspot (Vs).2. The cortical region of interest (ROI) was defined to encompass M1, M2, cingulate cortex, and surrounding regions (L 0-5 mm, AP +5 to −5 mm, and D 0-4 mm).The ROI was divided into a 50 × 100 × 40-point grid, totaling 200,000 voxels, each of .1 × .1 × .1 mm 3 .3.From the generated 3D cortical labeling plots (Methods, Registration and Reconstruction (FigureS1)), all voxels which are fluorescently labeled or have labeling in their immediate surrounding voxels are scored (0-10) each for correlations with Ds and Vs.For example, for Ds, the cortical voxel scores are calculated as follows: kg) and xylazine (16.67 mg/kg) and secured to the stereotaxic device with ear bars.A burr hole was made in the skull exposing the M1 and M2 regions.Ketamine (30 mg/kg) was subsequently injected every 70 min for the duration of the experiment.A stainless steel EEGhead screw electrode (Invivo1) was secured in the skull contralateral to the burr hole for electrical stimulation return.Next, a tungsten monopolar electrode was inserted through the exposed dura in precise locations within M1 or M2 at a depth of 1.6-2.4mm from the surface.Using an AM systems stimulus generator and isolator, biphasic pulse trains (pulse width 200-500 μs, inter-pulse frequency 100-500 Hz, train length = 5-20 pulses, train frequency = .5-5Hz, amplitude = up to 500 μA) were delivered and corresponding contralateral body part movements (front paw, hind limb, trunk, neck, when sufficient survival times were permitted.The classical movement disorder features were preceded on Day 2-3 by the onset of a turning tendency ipsilateral to the injection side.The movement disorder features (Video S1) developed rapidly on Day 4 (rat #8 and 11) and were indistinguishable from that induced previously with ibotenate(Kumbhare et al., 2017) and closely resembled the human conditions.Dorsal hotspot rVSV injections induced parkinsonism characterized by a paucity of spontaneous generalized movement, with a highly suppressed response to audio and tactile stimuli.The contralateral forelimb was held in a flexed posture and infrequently utilized.Ventral hotspot injections induced prominent neck and truncal twisting, and contralateral forelimb and hindlimb extensions.These animals showed normal spontaneous activity, though exhibited regular falls due to the dystonic posturing.See Figure1a,b and Video S1 for scoring of parkinsonian and dystonia features.For additional details, refer to our previously published videos(Kumbhare et al., 2017) for comparable illustrative behavioral features.Kruskal-Wallis test revealed that there were statistically significant differences between rats from the four categories (dorsal (n = 4) and ventral (n = 5) ibotenate lesions and dorsal (n = 5) and ventral (n = 5) VSV injections) for their Parkinsonian scores: (1) generalized spontaneous movement, F(3,15) = 149.1,p < .05;(2) generalized induced movement, F(3,15) = 159.6,p < .05;(3) forelimb movement, F(3,15) = 156.3,p < .05;as well as dystonia scores: (1) Hindlimb extension, F(3,15) = 156.4,p < .05;(2) Trunk posturing, F(3,15) = 144.3,p < .05;and (3) falling, F(3,15) = 148.5,p < .05.Tukey's HSD test for multiple comparisons found that rats with VSV injected in dorsal GP have significantly different scores (all p < .05).The results for ibotenate lesioned rats were reported previously.The onset of the parkinsonian or dystonic features coincided with the timing (on Day 4) of delayed local viral-induced GP neuronal degeneration, as seen on silver-stained sections (not shown).The reproducibility of the isolated clinical syndromes confirmed both the overall accuracy and discreteness of the rVSV injections.The largely successful, discrete targeting of the intended GP hotspots was additionally supported by chiefly isolated fluorescent transsynaptic cortical labeling (Figure 2).Subsequently, to limit the severity of the motor features and to better restrict the local spread of the viral labeling, the viral dosage was limited to .2-.4 μL and/or diluted (1/5th or 1/10th dilution using Dulbecco's PBS) in the last 13 animals.At these lower viral dosages, the animals self-maintained their body weights and when permitted to survive for 4 days (rats # 26-36) consistently displayed similar but milder parkinsonian or dystonic motor features.At the lower dosages, the consistent induction of mild behavioral features was sufficient to affirm the accuracy of each injection.Although the lower dosages permitted us to keep animals for extended periods, we quickly realized that maintaining animals beyond 4 days led to confounding further transmission of the virus beyond the primary targeted cortical regions to connecting motor and somatosensory cortical areas.As such, we restricted the subsequent reconstruction of the primary cortical projections to 4-day post-injection (DPI) rats (perfused on day 4).Fortuitously, on day 4, the virus consistently reaches the cortex from GP at approximately the same time that it induces prominent degeneration of GP neurons and the resultant induction of the behavioral phenotype.The 4-DPI timing of transsynaptic cortical labeling was consistent with reports of VSV transit of one synapse per day(Beier et al., 2011).

F
I G U R E 2 Focal VSV multi-synaptic tracer injections in the parkinsonian and dystonia GP hotspots label distinct secondary and primary cortical motor territories.(a) Illustrated are dorsal (green) and ventral (magenta) hotspots in GP for induction of parkinsonism and dystonia, respectively, via VSV injections in rats.(b) Representative sagittal images are shown to illustrate florescent cortical labeling resulting from dorsal (b1) and ventral (b2) injections in GP.Irrespective of the surrounding spread, stereotactic injections of VSV tracer into the previously defined dorsal parkinsonian GP hotspot produced florescent labeling chiefly confined to a circumscribed region of the secondary motor cortex (M2) (b1a), while ventral GP dystonia hotspot injections predominately labeled a region in the primary motor cortex (M1) (b2a).High power magnifications demonstrate dense neuronal labeling of (b1b) M2 and (b2b) M1 neurons.The yellow scale bars in b1a and b2a equal 1 mm and in b1b and b2b equal 100 μm.were utilized for the shorter incubation studies.Although not thoroughly studied, we were not aware of, nor did we appreciate, any age-related differences in viral transmission.First, tetrachoric correlation was used to grossly support the anatomical relations between the cortical labeling and the two principal injection sites (step 1).Upon supporting the relations, a custom probability algorithm was implemented to delineate cortical voxels which have strong and consistent associations with either of the two GP hotspots (step 2).Next, after filtering out poorly correlated voxels, spatial autocorrelation (global Moran I) and Getis-Ord GI hotspot analysis were used to establish whether the GP-cortical projections are both significantly clustered (step 3).Finally, independent t-test on the filtered cortical voxels was used to confirm that the dorsal versus ventral GP hotspots project to two non-overlapping cortical regions (step 4).Details of the results for each step are as follows:1.Tetrachoric matrix supports anatomical relations between the cortical labeling and the two principal injection sites Upon deriving the impression that the cortical labeling from the dorsal GP hotspot injections was projecting predominately to M2 and that from ventral GP injections to M1, a tetrachoric correlation was calculated to objectively assess the strength of the correlations between the binary injection locations (dorsal and ventral GP) and the primary projection locations (M2 and M1).Table2 indicatesthe tetrachoric matrix with percentage of M1-M2 digitized projections paired with each GP injection sites.The correlation coefficient of the matrix was −.818, indicating a strong relation between the individual GP injection sites and the specific M1 versus M2 cortical labeling.The negative value indicates a correlation of the ventral GP site with M1 and the dorsal site with M2.2.Probability mapping to delineate cortical voxels with strong and consistent associations with either of the two GP hotspots Next, towards defining the separate cortical parkinsonian and dystonia territories, we derived a novel 3D spatial custom GPcortical correlation algorithm (methods above).The distribution of VSV within and surrounding GP was established based on visual inspection of fluorescent labeling and/or silver staining (FD NeuroSilver kit II) of damaged neurons and scored (from 0 = none to 10 = complete labeling) for the dorsal and ventral GP hotspots.The cortical region of interest (ROI) was then defined to encompass M1 and M2 and surrounding regions and segmented into a .1 × .1 × .1 mm 3 voxel grid.From 3D cortical labeling plots, the algorithm generated individual voxel scores (0-10) for extent of labeling, accounting for immediate and/or surrounding voxel labeling.

3.
Spatial autocorrelation and hotspot analysis establish that both GP-cortical projections are significantly clustered Global Moran's I (GMI) function values for the M1 (GMI = +.453) and M2 (GMI = +.585)territories derived from the probability mapping indicate significant spatial autocorrelations, both p < .05(i.e., less than a 5% probability that the distinct clusters resulted randomly).Getis-Ord GI* statistics (Getis & Ord, 2010) (z-scores) modified for 3D data were calculated providing a statistical hotspot analysis using the local pattern of spatial association to identify local spatial clusters with high median scores (Figures 3 and 4, middle columns).For statistically significant positive zvalues (GI*; z-score >1.96 (p < .05)), the larger the z-values, the more intense is the clustering of the cortical labeling.The Getis-Ord distribution maps reveal the progressive inwards confidence for the defined parkinsonian M2 and dystonia M1 central hotspot regions.Subsequently, the probability distribution maps generated in step 2 (Figures 3 and 4, left column) were filtered to include only significant z-scores (GI*).The resultant final distribution maps define the statistically significant cortical projects (Figures 3 and 4, right column).The final maps define two distinct cortical territories, one contributing to a dorsal BG-rostral cortical (M2) parkinsonian subcircuit (center of mass (COM) for the stringent protocol: L2.07, A3.08 D3; Figure 3 right column and Figure 6a) and one to a ventral BG-caudal cortical (M1) dystonia subcircuit (COM: L2.12, A-.14, D1.90; Figure 4, right column, and Figure 6a).4.Independent t-test on the filtered cortical voxels confirms that the dorsal and ventral GP hotspots project to two nonoverlapping cortical regions Finally, to verify that the two hotspots are spatially significantly different from each other, the distributions (distance from bregma) of the two hotspot coordinates were compared using independent sample test.The test rejected the null hypothesis that the two hotspot coordinate distributions are statistically at the same location (p < .05).

F
Probability modeling and hotspot analysis for parkinsonian projections.To account for viral leakage and extension of the GP injections beyond the targeted hotspots, a custom spatial algorithm was employed with the results shown here for the dorsal parkinsonian GP injections (n = 17).Illustrated here is the defined cortical region of fluorescent neuronal labeling that correlated with neuronal labeling or cellular damage in the dorsal GP hotspot.(a) The cortical labeling defined by the algorithm is shown for four sagittal planes.(b) Getis-Ord GI* statistics maps show the distribution of the z score values (color bar).(c) Final masked output based on Getis scores, superimposed on the cortical labeling on demarcated M1 (yellow) and M2 (green) cortical representations (defined by the Paxinos and Watson 6th edition atlas).The final labeling shows a circumscribed territory in M2 is integral to a dorsal BG-M2 parkinsonian subcircuit.

F
Probability modeling and hotspot analysis for dystonia projections.To account for viral leakage and extension of the GP injections beyond the targeted hotspots, a custom spatial algorithm was employed with the results shown here for the ventral parkinsonian GP injections.Illustrated here is the defined cortical region of fluorescent neuronal labeling that correlated with neuronal labeling or cellular damage in the ventral GP hotspot.(a) The cortical labeling defined by the algorithm is shown for four sagittal planes.(b) Getis-Ord GI* statistic maps show the distribution of the z score values (color bar).(c) Final masked output based on Getis scores, superimposed on the cortical labeling on demarcated M1 (yellow) and M2 (green) cortical representations (defined by the Paxinos and Watson 6th edition atlas).
initiation and abnormally scaled movement features in PD.In contrast, per our modeling, pathophysiological BG signaling along the M1 subcircuit produces excessive and poorly regulated thalamocortical activation of M1, leading to the characteristic excessive F I G U R E 6 Cortical hotspots and segregated subcircuitry.(a) The defined parkinsonian (green) and dystonia (magenta) cortical territories are shown together in 3D to illustrate their positions and separation from each other.(b) Illustrated is our new working model of segregated pathological BGTC motor subcircuits, specifically for parkinsonism and dystonia, two of the most common movement disorders.As discussed in the main text, the confinement of the parkinsonian cortical hotspot to the rostral microexcitable zone supports that the parkinsonian subcircuit is likely to represent the SMA subcircuit.The M2 (purported SMA) parkinsonian cortical territory provides high level (executive) motor-related signaling to the spinal cord and to the M1 (dystonia) territory (not illustrated), with the latter, providing direct corticospinal programming of the motor action.Disturbances in motor signaling isolated to these two subcircuits can readily explain the characteristic clinical features of Parkinson's disease and dystonia.dystonic co-contractions with spread of muscle activations to unintended joints.Consistent with our present findings, functional cortical imaging studies in humans with PD have demonstrated abnormal cortical activation principally in the SMA (Eidelberg