Vestibular circuit stimulation for retuning locomotor dynamics in Parkinson's disease

Vestibular circuit stimulation for retuning locomotor dynamics in Parkinson's disease | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Vestibular circuit stimulation for retuning locomotor dynamics in Parkinson's disease Johannes Hartig, Maximilian Friedrich, Jérémy Signoret-Genest, and 9 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5851215/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Postural and locomotor dysfunction represent axial symptoms of Parkinson’s disease (PD), remaining poorly treated by medication and deep brain stimulation. Non-invasive neuromodulation of the vestibular system, centered on the vestibular nucleus complex (VNC), offers a novel therapeutic avenue. However, the underlying circuits remain ill-explored. In this study, we found that the VNC in mice feeds extensive Vglut2 -defined projections into striato-thalamo-subthalamic and caudal medulla motor hubs, but not the mesencephalic locomotor region. Optoactivation of excitatory VNC neurons below the threshold for vestibular symptoms promoted activity in these basal ganglia-brainstem axis targets. Unbiased analysis of pose dynamics revealed global enhancement of behavioural transitions and locomotion, confirmed by regular kinematic analyses. Therapeutically, it enabled resynchronization of naturalistic gait patterns and improved locomotor performance, but not capacity, in parkinsonian mice. Our data identify excitatory VNC circuit processes for therapeutic retuning of motor dysfunction in the context of PD. Health sciences/Diseases/Neurological disorders/Parkinson's disease Health sciences/Anatomy/Nervous system/Central nervous system/Brain Health sciences/Diseases/Neurological disorders/Neurodegeneration Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Parkinson's disease (PD) represents a significant and escalating public health challenge 1 , marked by dysfunction in dopaminergic circuits 2 . Increasingly recognized as a neurodegenerative spectrum disorder with a diverse phenotypical combination of motor and non-motor features 3 , the motor aspects, such as bradykinesia, rigidity, and tremor, typically respond well to dopaminergic medication and deep brain stimulation (DBS) targeting the subthalamic nucleus (STN) or globus pallidus internus. However, axial motor symptoms, encompassing dysfunctional gait (e.g., freezing of gait) and postural instability, exhibit poor responsiveness 4 – 7 and persist as a clinical concern. Occasionally classified as the PIGD subtype 5 (postural instability and gait difficulty), axial phenotypes remain both insufficiently understood and inadequately addressed by current therapeutic interventions 8 . Thus, there is an urgent need for innovative strategies to address treatment-resistant parkinsonian gait and posture disorders. Previous approaches towards improving axial PD symptoms included postulations about the involvement of basal ganglia-brainstem interactions 9 – 12 ( see Supplementary Fig. 1 for graphical summary). Specifically, neuromodulation strategies for PIGD have led to trials of DBS targeting brainstem mesencephalic locomotor region (MLR) nuclei – pedunculopontine and cuneiform – based on their implication in circuits for gait and posture 9 , 13 . However, those studies have yielded inconsistent results 14 , 15 , likely reflecting the hitherto poorly understood, complex nature of gait and posture (disorders) in PD 8 . Recognizing the advantages of non-invasive neuromodulation techniques in terms of ease-of-use and lower complication rates, there is a growing interest in exploring these approaches for different (patho-)physiological states 16 , 17 , including PD. However, a prerequisite for novel PIGD neuromodulation approaches is a better mechanistic understanding of the circuitry underlying predominantly axial phenotypes. Vestibular stimulation is a non-invasive method that utilizes, e.g. noisy (i.e. thresholded), galvanic (nGVS) or caloric means to activate (central) vestibular circuits via peripheral excitation of the vestibular organ 18 . This technique has shown promise in mitigating motor and non-motor symptoms of PD 19 – 22 , potentially via integrating vestibular information for movement into a vestibulo-thalamo-basal ganglia loop 23 , 24 . Notably, there are some data supporting a positive impact on axial postural deficits, in both PD 22 , 25 and atypical parkinsonism 26 , yet less so for gait in PD 27 . Further, nGVS has improved parkinsonian symptoms in rats 28 . Vestibular circuits likely exhibit glutamatergic predominance 29 – 31 , especially with regards to motor targets 32 , and while anatomical studies suggest possible circuit mechanisms involving vestibulo-thalamic inputs to basal ganglia loops 23 , 33 , 34 , the structural and functional 24 contributions of vestibular circuits for movement, specifically in PD 35 , remain insufficiently understood. Our study thus aimed to better understand the feed of the vestibular nucleus complex (VNC) into basal ganglia loops and contributions to motor dimensions, specifically in parkinsonian mice. First, we wanted to further virally trace and functionally map this association. Second, based on the hypothesis that thresholded human nGVS feeds prolocomotor information into basal ganglia loops, we aimed to check for therapeutic effects of a thresholded optical activation of the VNC. Results Titrated optoactivation of glutamatergic vestibular nucleus complex neurons yields a diverse posture and locomotor symptom continuum To functionally interrogate vestibular circuits governing postural and locomotor control, we used optogenetics by introducing the photosensitive optical actuator ChRmine 37 into the vestibular nucleus complex (VNC) of Vglut2 (vesicular glutamate transporter 2)- ires-cre mice ( Fig. 1a ). We rendered a fraction of these mice parkinsonian by overexpression of AAV-delivered human A53T-mutated α-synuclein in the substantia nigra pars compacta (SNc) and maintained some as control mice injected with empty vector (EV) virus. These are henceforth referred to as A53T and EV, respectively ( Fig. 1a ) 38 . To avoid confounding by group-differential opsin uptake, we estimated and compared transfected VNC neuron count by assessing mScarlet-positive cells, labelling the viral construct, in A53T vs. EV mice. The former showed no significant difference ( Fig. 1a , week 8; A53T: n=12, EV: n=10 mice). A sham optical control group was injected with a virus lacking an opsin ( Fig. 1b ). In optogenetic pilot experiments, we found that incremental laser power yielded differential posture and locomotor symptoms with VNC activation in both, A53T and EV mice. To comprehensively assess the effects of increasing levels of optical VNC actuation and generate a therapeutic stimulation regime, we devised two optogenetic approaches: the first consisted of a titration paradigm that allowed us to probe dose-response effects on posture and locomotor control by using incremental laser power thresholds ( Fig. 1c ). To capture these disruptive stimulation effects systematically, we used a Vestibular Symptom Scale (VSS) [see 31 for a similar scale] ( Fig. 1d ), based on the following findings from pilot experiments: frequently, the first optically-induced aberrant behaviour to appear was tilting of the head, most notably in the yaw and roll plane (= VSS 1 ). Additionally, some animals demonstrated head bobbing under low stimulation intensity. With increasing laser intensity, body turning, as well as crossed extensor-like 39 (both= VSS 2 ) and circling movements (= VSS 3 ) emerged. Finally, at high laser intensities, retropulsion or retro-locomotion (= VSS 4 ) appeared (see Supplementary Video 1, 2, 3 ). The titration experiments confirmed dose-dependent effects of VNC optoactivation on postural and locomotor control in both EV and parkinsonian A53T mice, but not sham mice. Specifically, we found that while VSS 2 symptoms (body tilting, crossed extensor-like responses) were observed in ~56% of animals at week 8 (human rater), more pronounced phenotypes (VSS 3 + 4) and VSS 1 (head bobbing and tilting) symptoms were observed less consistently. Since these were subjective rater-based observations, we next aimed to check whether a machine learning approach based on pose data would identify similar subtypes of vestibular symptoms. We made use of top-view open field (OF)-based pose estimation via DeepLabCut (DLC) 40,41 from systematic titration experiments (data from n=24 mice). These encompassed the whole spectrum of rater-observed induced vestibular symptoms, but also spontaneous behaviour, which we categorically annotated as ‘VSS-like’ and ‘naturalistic behaviour’, respectively. We then leveraged a semi-supervised machine learning pipeline (A-SoiD 42 ) to train an active learning classifier (week 4 titration data), which performed well in detecting both behavioural categories (average performance: ~90%) ( see Supplementary Figure 2a-c for details on workflow). We applied this active learning classifier to unseen titration data (week 8 titration data, n=22 videos, A53T: n=12, EV: n=10) and found that VSS symptoms were matched with those of a human rater ( Supplementary Figure 2d ). Building upon that, we used clustering and embedding algorithms (B-SoiD 36,42 ) inside A-SoiD to try and isolate distinct VSS-like behaviours (n=261 snippet videos, week 4 titrations) and found 5 significantly different clusters ( Fig. 1e ), while naturalistic behaviour yielded 10 distinct clusters ( Fig. 1f ). Based on clustering results, we in turn trained a novel, second active learning classifier on the resulting classes inside A-SoiD ( Supplementary Figure 2e ; average performance: ~90%), which revealed similar, however not identical subtypes as the VSS: Specifically, cluster 0 mapped to head bobbing (VSS 1) and cluster 4 to head & body tilting, circling and retrograde motion behaviours (VSS 2-4) (see Supplementary Videos 4-10 for classified subtypes). The major clusters in naturalistic behaviour mapped to idle exploration behaviour ( cluster 9 ) vs. locomotor exploration ( cluster 7 ), thus verifying overall sensible, yet highly stringent behavioural clustering and embedding. These results demonstrate that similar vestibular symptom subtypes are identified across subjective observation (VSS 1-4 scale) and automated analyses via pose estimation. We next asked whether the thresholds for vestibular symptoms dynamically changed or correlated with mScarlet-positive VNC cell count for any VSS class. We found that only for the high end of the symptom scale (VSS 4), laser power thresholds correlated with the number of mScarlet-positive cells ( Fig. 1g-j , all available mice from optogenetics cohort; A53T: n=10, EV: n=9). Nonetheless, all VSS thresholds were significantly lower at week 8 vs. baseline ( Fig. 1k , all available mice from optogenetics cohort; A53T: n=10, EV: n=9). Once we unraveled the strong symptomatic effects of VNC optoactivation on posture and locomotion, we blueprinted a second, therapeutic regime for optical neuromodulation: A Perithreshold Vestibular Optomodulation regime ( Fig. 1l ). With the latter, we aimed to mimic the perceptually thresholded electrical vestibular neuromodulation by nGVS 19,25,22,26 . We calibrated our optogenetic activation at the threshold of vestibular symptoms to appear and stimulated just below threshold, in order to achieve central activation of vestibular networks 24 , akin to GVS 43 . Yet, repeated stimulation may idiosyncratically change thresholds or responsiveness to stimulation during different (on-going) behaviours. We thus termed this ‘perithreshold stimulation’ instead of ‘subthreshold stimulation’. Taken together, our data show how titrated, optical activation of the VNC yields a diverse continuum of posture and locomotor symptoms in mice. These findings enabled us to develop a therapeutic optogenetic stimulation regime. Glutamatergic vestibular circuits connect to striatum, (sub)thalamus and caudal medulla, interlinking basal ganglia-brainstem motor hubs Having identified vestibular (sub-)phenotypes both by observation and from pose estimation data in titration experiments, we next focused on the underlying circuit elements and aimed to further unravel how VNC is embedded within pathways for movement, specifically those interlinking brainstem and basal ganglia 9 . For structural mapping of glutamatergic VNC connectivity, we took a myc-tagged AAV-delivered anterograde synaptophysin tracer injected into the VNC. To delineate overlaps with motor ensembles implicated in neuromodulation and motor rescue in PD rodents 45,46 , and due to evidence on a vestibulo-thalamo-striatal tract 23 , we complemented the anterograde tracing approach with retrograde tracer injections ( retroAAV-hSyn-DIO-mCherry ) into the STN (n=3) ( Fig. 2a ), or into the dorsolateral striatum (dlSTR; n=4). Additionally, we evaluated mScarlet projection patterns from A53T and EV animals. Focusing on subcortical regions related to basal ganglia-brainstem interactions associated with posture and locomotion 9 , we found that the VNC maintains an extensive glutamatergic connectome ( Fig. 2b,c ), projecting to important sensorimotor hubs in the subthalamus ( Fig. 2d,e , zona incerta), thalamus ( Fig. 2f-k , parafascicular-centromedian nucleus) and brainstem ( Fig. 2l , caudal medulla reticular formation). Synaptic punctae appeared in the parafasciular-centromedian (Pf-CM) nucleus complex and posterior oral (Po) nucleus ( Fig. 2f-i ), partially overlapping with thalamostriatal neurons projecting to the dlSTR, corroborating findings on a vestibulo-thalamo-striatal pathway in rodents 23 via the Pf-CM complex. Further, we found evidence of a strong vestibular pathway to the zona incerta (ZI) in the subthalamus yet observed no significant projection to the STN ( Fig. 2d,e ). Contrary to our initial hypotheses, we found no consistent evidence for a strong projection to the mesencephalic locomotor region (MLR), either ( Fig. 2m,n ), which we postulated to regulate gait adjustments. While some mScarlet fibres and isolated synapses reached the contralateral PPN, no consistent synaptic punctae were found in the MLR (either PPN or CnF). However, we found that the VNC directly targets the reticular formation in the caudal medulla, namely the gigantocellular (Gi) and sparsely the lateral paragigantocellular nucleus (LPGi) and anterior gigantocellular nucleus (GiA) ( Fig. 2l ), thus potentially leveraging ‘hyperdirect’ influence on reticulospinal effector pathways 47,12 . After we anatomically identified VNC targets within canonical motor circuitry, we aimed to verify functional relevance of VNC perithreshold stimulation in driving activity in those target areas. We used changes in c fos expression, an immediate early gene (IEG), as a surrogate for recent neuronal activity 48,49 , which we quantified using the deepflash2 pipeline ( Fig. 2p ). Therein, we focused on subcortical brain regions heavily targeted by the VNC ( Fig. 2d-o ), namely the ZI, the Pf, Gi and finally the VNC itself. Further, we aimed to confirm the effect of optical actuation vs. sham optogenetics by comparing A53T & EV with mCherry mice. Perithreshold vestibular optomodulation in an OF context resulted in increased bilateral cfos expression in ZI, Pf, Gi, and VNC in the A53T and EV group compared to mCherry-injected control animals, lacking a functional opsin ( Fig. 2q-t ; A53T: n=12, EV: n=9-10, mCherry: n=5-6). Overall, these structural (viral tracing) and functional (cFos) mapping data show that the VNC links both striato-thalamo-subthalamic and motor circuits in the reticular formation of the caudal medulla, consistent with the notion that the VNC serves as a basal ganglia-brainstem interface 9 in regulating axial motor functions. Human A53T-mutated α-synuclein overexpression in the substantia nigra leads to dopaminergic neurodegeneration and parkinsonian-like behaviour Based on structural and functional observations of VNC interactions with key motor brain areas, we next addressed potential kinematic and behavioural changes by thresholded actuation of Vglut2 -positive VNC neurons to probe putative therapeutic action in parkinsonian A53T mice. To that end, we first confirmed disease model effects in parkinsonian A53T mice via stereological, immunohistochemical and behavioural analyses. Previous studies used injection of AAV1/2-A53T (here: conc. 15x10 12 genomic particles/ml; 500 nanoliters (nl)) into the SN of mice to induce dopaminergic cell death via overexpression of α-synuclein (α-Syn) 38 , thereby establishing a model for parkinsonian α-synucleinopathy. We substantiated α-Syn deposition in the ipsilateral SN in A53T mice qualitatively by verifying immunoreactivity ( Fig. 3a,c ). Stereological analysis yielded significantly reduced tyrosine hydroxlase (TH)-positive cells in the SN in A53T vs. EV mice (week 8; A53T: n=12, EV: n=10) indicating dopaminergic cell death. For sound downstream pooling of behavioural readouts, we validated dopaminergic neurodegeneration at week 4 in an independent cohort of A53T mice (n=5) ( Fig. 3b ). Confirming these cells as actual neurons, we found a strong correlation of TH-positive cell count with Nissl-stained cells ( Fig. 3d ). Having checked model effects at both time points histologically, we confirmed and extended behavioural effects in the cylinder and amphetamine test, as well as OF kinematics 38,50 . As a surrogate of nigrostriatal dopaminergic dysfunction 51 , we measured total amphetamine-induced rotations (without optogenetic manipulation). These strongly discriminated A53T from EV mice (total rotations, week 8 off; AUROC=0.9375; A53T: n=12, EV: n=10), verifying effects on nigrostriatal dopaminergic output pathways in diseased mice ( Fig. 3e ). Our findings were supported by additional evidence from the cylinder test and top-view OF. In the latter, A53T mice demonstrated a bias in meander behaviour, defined as quotient of turn angle per distance moved, thus displaying changes in the steering of directionality ( Fig. 3f , week 8 off; A53T: n=12, EV: n=7). Mean meander significantly correlated with amphetamine-induced rotations ( Fig. 3g ), suggesting association of both effects with nigrostriatal dysfunction. In the cylinder test, we confirmed increased forepaw use ipsilateral to the injection site ( Fig. 3h ) in A53T vs. EV mice ( Fig. 3h , week 8 off; A53T: n=12 , EV: n=10) 38 . Beyond these previous results, we found that A53T mice displayed a strong, ipsilesional rotatory bias in the cylinder test setting ( Fig. 3i ), reflected by a skewed rotatory symmetry index vs. EV mice (week 8 off; A53T: n=12, EV: n=10). Of note, this (static) rotatory asymmetry did at most show a mild reduction with perithreshold actuation of the VNC. Lastly, the maximum velocity (V max ) on a treadmill correlated significantly with both dopaminergic neurodegeneration ( Fig. 3j ) and rotatory asymmetry in the cylinder test ( Fig. 3k ). Put together, these data confirm the development of a cellular and behavioural parkinsonian phenotype through virally induced α-synucleinopathy in the substantia nigra. Optoactivation of glutamatergic vestibular nucleus complex neurons allows retuning of locomotor performance but not capacity in parkinsonian mice With confirmation of a parkinsonian phenotype in A53T mice, we next took an in-depth look at changes on two hierarchical levels. Using top-view pose estimation, we first assessed higher-order behavioural patterns, followed by analysis of more fine-grained kinematic motor aberrations. We subsequently aimed to interrogate how both levels were altered by perithreshold optoactivation of glutamatergic VNC circuitry and whether therapeutic changes occur in parkinsonian mice. To this end, we extensively retrained the recurrent convolutional neural network (RCNN) ‘ superanimal_topviewmouse ’ from the DLC model zoo 40,41,53 , thereby achieving high tracking accuracy (mean average Euclidean error (MAE) on test data=2.04 pixels) ( Fig. 4a ) for top-view pose on our own data. Utilizing a large amount of diverse pose estimation data from the top-view OF (n=193 videos), we included data from sham-stimulated mCherry (n=6), EV (n=10) and parkinsonian A53T mice (n=14) from different time points and employed an unbiased, unsupervised machine learning approach via Keypoint-MoSeq 52 to identify distinct subsecond behavioural modules in pose dynamics, which were shown to be highly relevant to explaining mouse behaviour 54 . We found multiple principal behavioural ‘syllables’ (i.e. modules) (n=23, Fig. 4b ), that clearly mapped to naturally occurring behavioural motifs of mice, such as e.g. running (syllable 15; Supplementary Videos 11-33 & Supplementary Figure 3 ). Amongst this set of syllables, running was significantly different from most other syllables with regards to trajectory ( Fig. 4c ). Further, syllables 6 and 11 were both identified as running behaviours associated with prior head movements and rearing, respectively. We aggregated all three running-related syllables to define peak locomotor performance. We first aimed to understand transitions between behavioural modules regarding differences amongst groups and with perithreshold optoactivation. To get a quantitative assessment of this effect, we analyzed the transition matrices of A53T, EV and sham-stimulated mCherry mice, specifically probability differences. Via a matrix eigenvector space analysis, we found that optoactivation of VNC neurons leads to significant changes in both A53T and EV, but not in sham-stimulated mCherry mice ( Fig. 4d ). This underscores global changes in behavioural transitions induced by optoactivation of glutamatergic VNC neurons. In addition, A53T vs. EV group behavioural transition matrices in the absence of optical stimulation were significantly different (p=0.014), though not surviving correction for multiple comparisons. These results suggest that SNC α-synucleinopathy induces mild alterations in behavioural transition dynamics. We then created network plots to gain a visual representation of how the overall usage of (and transitions between) distinct behavioural syllables changes with perithreshold VNC actuation. The overall structure of up (green)- and downregulation (violet) shows that optoactivation in A53T and EV groups ( Fig. 4e-f ) leads to globally up-regulated (green) behavioural transitions vs. sham-stimulated mCherry mice ( Fig. 4g ), indicating enhanced pose shifting and enhanced modularity in pose with VNC activation. These behavioural changes seen with a comprehensive birds-eye view on a higher hierarchical level motivated us to investigate what thresholded actuation of glutamatergic VNC neurons would do to distinctly identified behavioural modules (i.e. syllables) themselves. Analysing syllable kinematics, we found that perithreshold optical actuation of Vglut2 + VNC neurons lead to an enhancement of peak locomotor performance, as reflected by increased mean velocity ( Fig. 4h ) and its variability ( Extended Data Fig. 2a ) in the running module in both A53T and EV vs. sham mCherry mice. However, a blunted response to VNC optoactivation, i.e. lower duration of optically-induced running, revealed impaired locomotor capacity in A53T mice ( Fig. 4i ). Underscoring this effect, we found that behavioural modules where running was associated with prior rearing (duration, syllable 11; Extended Data Fig. 2b ) or head movements (max. velocity; syllable 6; Extended Data Fig. 2c ) were sensitive to optoactivation in EV controls, but not in A53T mice. We next asked whether and how perithreshold and suprathreshold optoactivaction effects are related to each other, in part to cross verify the validity of the syllable analysis. In other words, are specific motor effects induced by higher stimulation intensities (such as turning, circling and body steering behaviours ipsilateral to stimulated/A53T or EV-injected hemisphere) part of a continuous behavioural pattern? Strikingly, we found that those behaviours bear a covert representation in behavioural modules under low-intensity stimulation conditions. For example, variability of angular velocity in ipsilateral (i.e. leftward) steering was enhanced by perithreshold optoactivation of glutamatergic VNC neurons (syllable 7; Extended Data Fig. 2e ). The same manipulation led to an increase in mean angular velocity during ipsilateral rotation in EV, but not A53T mice (syllable 19; Extended Data Fig. 2f ). Do our results stand irrespective of pre-defined behavioural patterns? To answer this question, we compared kinematics identified through unsupervised behavioural syllable analysis with general locomotion parameters across different behaviours. We investigated locomotor dynamics in the OF with keypoint tracking, identical to the basis of the syllable analysis ( Fig. 4j ). Overall locomotor capacity, as measured by distance moved, was reduced by habituation in EV controls and A53T groups over time. Although EV animals exhibited strong enhancement of locomotor capacity due to VNC optoactivation ( Fig. 4j ), this effect was strikingly missing in A53T animals. Sham-stimulated mCherry animals displayed a similar reduction in distance moved with sham stimulation, attributable to habituation effects. ( Fig. 4j ). Peak locomotor performance (as indicated by maximum velocity) was reduced in A53T mice vs. baseline. EV mice did not show this reduction vs. baseline. In contrast to the lack of effect on capacity, this SNC α-synucleinopathy-induced deficit was rescued by perithreshold optoactivation ( Fig. 4j ). This increased peak performance was observed in EV animals as well, however not in sham-stimulated mice, attributing the stimulation effect to specific activation of VNC glutamatergic neurons ( Fig. 4j ). In summary, pose estimation in subsecond behavioural modules reveals that thresholded actuation of glutamatergic VNC neurons leads to overall up-regulated transitions between behavioural modules, indicating increased dynamics in pose shifting. In addition, optoactivation enhances peak locomotor performance of running behaviours. Importantly, these effects were blunted by SNC α-synucleinopathy, reinforcing that vestibular excitatory modulation of the basal-ganglia loop is instrumental in regulating both peak performance and capacity of locomotor function. In parkinsonian mice, exogenous enhancement of VNC excitatory drive exerts a differential rescue effect on peak locomotor performance but not capacity. Optoactivation of glutamatergic vestibular nucleus complex neurons yields disease-specific retuning of parkinsonian gait dynamics We finally wondered, based on our findings of rather specific behavioural rescue effects due to VNC optical actuation, whether the same intervention would specifically affect gait, the motor pattern underlying both locomotor performance and capacity. To this end, we first assessed the maximum velocity on a treadmill and applied DLC-based analysis of gait kinematics equipped with a dual camera system ( Fig. 6a ). Whereas EV mice exhibited increased maximal velocity over time ( Fig. 6b ), A53T mice were unable to make the same adjustments. Interestingly, maximal velocity did not change with perithreshold VNC actuation, but closely correlated with both dopaminergic neurodegeneration and rotatory bias in the cylinder test (see Fig. 3j-k ). First, to cross-validate the specificity of perithreshold (vs. suprathreshold) optoactivation on locomotion, we opted for an assessment of gait failure by a three-level scoring system, which we termed fluidity of gait (FGI) index. We based the FGI on the number of consecutive strides (as conducted previously in disease models 55 ) and added gait interruptions as a complementing measure [one condition (worse counted) had to be met – FGI 1 = failure: consecutive strides ≤ 4, gait interruptions ≤ 3; FGI 2 = sufficient: consecutive strides = 5-7, gait interruptions ≤ 2; FGI 3 = fluid gait: consecutive strides ≥ 8, gait interruptions ≤ 1]. We found that perithreshold stimulation in both A53T and EV mice did not affect gait, whereas suprathreshold stimulation (mice stimulated at VSS 2-3 thresholds) caused markedly enhanced gait failure ( Fig. 6c,d ; Supplementary Video 34 ). Strikingly, while optical VNC actuation was ineffective in enhancing velocity, initial observations suggested that it resulted in a “smoothening” of gait. Hence, we specifically investigated the oscillatory pattern of different keypoint markers and matching keypoint pairs of hind- and forelimb. In detail, we analyzed the synchronicity, periodicity and stability of gait oscillations across speeds in both A53T (summary example overview via scalograms, Fig. 6e,f ) and EV mice ( Extended Data Fig. 3g-h ) with and without activation of glutamatergic VNC neurons. For a robust analysis of the oscillatory gait pattern of mice, we conducted an in-depth characterization across speed(s) (10, 15, 20, 25 cm/s) and conditions (baseline = presymptomatic, off, perithreshold) in estimated pose (n = 853 videos; A53T: n = 466, EV: n = 387). We found that A53T ( Fig. 6g ), but not EV mice ( Fig. 6h ), developed complex gait aberrations, reflected by a reduction of power in peak gait frequencies in ipsilesional/-stimulatory (left) distal limb markers, mainly driven by changes at 20cm/s ( Fig. 6k ). Vestibular optoactivation enabled us to retune these changes to baseline level. Using autocorrelation analysis, we further unraveled that simultaneously, the synchronicity of ipsilesional/-stimulatory distal limb markers across speeds decreases in A53T ( Fig. 6m,o,q ), but not EV mice ( Fig. 6n,p,r ). Importantly, we could optically retune this impairing effect of the parkinsonian model with a perithreshold activation of the glutamatergic VNC. Further, the relationship of peak amplitude and frequency across speeds (10-25 cm/s) collapsed in A53T mice and could be optically retuned to baseline level with VNC activation ( Fig. 6s,u ). EV animals did not show any changes due to either time or perithreshold stimulation ( Fig. 6t,v ). Whereas right-sided markers and vertical coordinates (y-axis) did not show any clear oscillatory changes, we found a trend for a change in the neck marker in A53T mice, which might indicate axial rigidity ( Extended Data Fig. 3a-f ). Taken together, these results demonstrate that a disease-specific retuning of parkinsonian oscillatory gait patterns is possible using thresholded vestibular neuromodulation by specific augmentation of excitatory VNC drive. Discussion This study in mice has two key findings: first, we identified structural and functional contributions of excitatory vestibular nucleus complex (VNC) cells to locomotor behaviour in general and to posture and gait dynamics in particular. Second, mimicking thresholded human vestibular neuromodulation, we optically retuned parkinsonian gait and locomotor deficits with activation of VNC glutamatergic neurons in mice. Incremental stimulation intensities yielded a diverse set of posturolocomotor alterations, blending into each other. This is consistent with phenotypes reported by previous human studies on effects of lesions, or targeted neuromodulation 56–59 and animal data 60–62,31 on manipulation of central vestibular pathways subserving head, body and eye movements in yaw, pitch and roll plane 63 . Indeed, structural mapping of glutamatergic VNC neurons exhibited extensive bilateral connectivity with key motor hubs, predominantly in the thalamus (centromedian-parafascicular (CM-Pf) nucleus complex), subthalamus (zona incerta), and brainstem (caudal medulla). Thresholded optical activation of VNC excitatory neurons led to enhanced activity of neuronal ensembles in these target regions, confirming functional relevance of a glutamatergic VNC circuit element linked to the quintessential basal ganglia-brainstem axis for movement 9 . These structural and functional circuit findings are in line with previous anatomical studies, which found a vestibulo-thalamo-striatal tract via the parafascicular nucleus in rats 23 and data from an optogenetic fMRI study in mice 24 , underscoring integration of thalamo-subthalamic with VNC circuitry. The strong overlap with thalamostriatal CM-Pf neurons projecting into the dlSTR hints at how the VNC likely conveys its strong influence onto motor systems: by the functionally segregated projections from the Pf to the striatum 64 . The Pf, and its interconnections to STR (and STN) specifically, have been implicated in motor rescue recently 45,46 . As a target of excitatory VNC inputs, this Pf function hints at putative pathways for locomotor rescue effects in our parkinsonian mice. Moreover, the Pf is known to cause ipsiversive head steering, and with prolonged stimulation, full body turns 65 , which corresponds to the pathway mediating the head tilting and turning effects caused by incremental VNC optoactivation. Specifically, the encoding of vector components in head velocity and directional steering by Pf 65 might underlie both thresholded VNC optoactivation effects on behavioural dynamics and point towards circuit dysfunction causing directionality bias in parkinsonian mice. VNC optoactivation favored locomotor performance involving running behaviours. Since we found that the VNC directly targets the reticular formation in the caudal medulla, it might leverage ‘hyperdirect’ locomotor influence via extensive connectivity to the Gi, sparsely also to LPGi, potentially explaining the distinct effects on behavioural motifs involving high-speed locomotion 47,12,66 . We corroborated these findings by showing that peak free, but not forced, locomotor performance (e.g. maximum velocity) is enhanced by thresholded optical vestibular neuromodulation. Further, we show that global changes in behavioural dynamics underlie specific changes in behavioural motifs, again enhancing locomotion. Yet, one of our primary hypotheses was that the vestibular system might interact with the mesencephalic locomotor region (MLR) in mediating effects on posture and locomotion, including under parkinsonian conditions 67,10,11,13 . Deep brain stimulation of the MLR in patients with Parkinson’s disease (PD) and a posture und gait-predominant phenotype has been explored, but results remain clinically inconsistent 15,14 . It is interesting to note that we did not find strong signs for glutamatergic connectivity of VNC neurons to neither pedunculopontine nor cuneiform nuclei; especially since electrical vestibular stimulation seems to enhance deficient pedunculopontine nucleus connectivity in PD patients 67 . This may, in part, relate to our cell type-specific approach focused on the optical activation of excitatory Vglut2 -VNC neurons. Given that GABAergic neurons seem to exert negligible global (motor) influence 31 , we postulate that the glutamatergic population subserves most functional influence on posture and locomotion. However, a primarily segregated (likely the second largest) vGlyT2 -defined glycinergic neuron population 30,32 remains to be functionally explored, which might exhibit MLR connectivity instead. Furthermore, a distinct glutamatergic population of vestibulo-cerebellar neurons, defined via vGLUT1 -expression, remains to be characterized in detail 32 . From a translational bedside-to-bench perspective, we sought to generate mechanistic insights into the central circuit-level effects of non-invasive vestibular stimulation, such as galvanic vestibular stimulation (GVS). The former shows promise in addressing both non-motor and motor symptoms in typical and atypical parkinsonian states, including hard-to-treat axial features 19,20,22,25,26 . We aimed to mimic a perceptually-thresholded stimulation with central vestibular network activation (like nGVS) by using optogenetics in EV and parkinsonian mice. Massive impairment of motor behaviour (e.g. gait) by suprathreshold optogenetic stimulation is in line with well-known dose-response relationships in human vestibular stimulation 18 . Our individually thresholded optoactivation of the VNC vs. sham stimulation shifted mice into a pro-locomotor state by enhancing behavioural transitions, peak locomotor performance and locomotor capacity. However, our data demonstrate specificity of these effects in a parkisonian context. Here, VNC optoactivation allowed for rescue of peak locomotor performance and improved gait patterns, but did not alter locomotor capacity. While the behavioural effects of thresholded VNC activation were studied in detail, the precise pathway-specific effects remain unidentified. Our tracing studies suggest the mediation of motor effects via vestibulo-(sub)thalamic or vestibulo-medullary projections, warranting further investigation into single-circuit effects combining e.g. projection specific in vivo electrophysiology and optogenetic manipulation. We chose the A53T model for its neuropathological profile, mimicking the progressive course of PD associated with α-synucleinopathy and the exploration of an early-stage PD level for potential disease modification. Exploring vestibular effects on severe hypo-/bradykinesia could be addressed in a more severe toxin-based mouse model for PD. Considering the broad impact of the vestibular system, including at a cortical level 24 , on navigation, affective and autonomic regulation, an integrated examination of both behavioural and autonomic readouts is imperative, especially from a translational perspective. Specifically, the complex effects of vestibular circuits on integrated behaviours (navigation, defense, etc.) require further investigation, not least because of its connectivity to a proposed global behavioural state regulator, the zona incerta (ZI) 68,69 . The strong influence of the VNC on sensorimotor circuitry is another hint towards the importance of brainstem-basal ganglia interactions for movement 9 . To sum up, through structural and functional mapping, as well as in vivo optogenetics, our data demonstrate a key regulatory role for excitatory vestibular neurons in axial movement, a function that can be exploited to retune and normalize motor symptoms in parkinsonian states. Declarations Funding This project was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) Project-ID 424778381-TRR 295 (A01 to J.V. and C.W.I., A06 to J.V. and C.W.I., B06 to P.T., S01 to R.P.) and by the Interdisciplinary Center for Clinical Research (IZKF) at the University of Würzburg (N-362). J.H. was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) Project-ID 424778381-TRR 295 (Gerok position). M.F. received funding by the Jung Stiftung für Wissenschaft and the Manfred and Ursula Mueller Stiftung. J.V. received funding from the European Union's Horizon 2020 research and innovation programme under the EJP RD COFUND-EJP N° 825575 (EurDyscover). C.W.I. received funding from the VERUM foundation. Contributions J.H. conceived ideas, acquired and analyzed data and wrote the manuscript. M.F. conceived ideas, acquired and analyzed data and contributed to the manuscript. J.S.G. analyzed data and contributed to the manuscript. S.T., N.S., D.D., R.P., S.K. and T.P. analyzed data and contributed to the manuscript. C.W.I., P.T. and J.V. contributed resources and edited the manuscript. Online Methods Animals and ethics approval All experiments were conducted in accordance with local and European animal welfare law, are ARRIVE-compliant and were approved by the local veterinary authorities and animal experimentation ethics committee (AZ 2-631 and 2-2030, Regierung von Unterfranken, Bavaria, Germany). Transgenic Vglut2-ires-cre ( STOCK Slc17a6 tm2(cre)Lowl /J , Jackson #016963, The Jackson Laboratory, USA) and C57Bl6/J mice (Charles River Laboratories Germany) were commercially acquired and bred at in-house breeding facilities (Institute of Clinical Neurobiology and Center for Experimental and Molecular Medicine, Wuerzburg, Germany). A total of n=42 mice were used (n=24 optogenetics, n=6 sham optogenetics, n=7 tracing cross-bred transgenic/wildtype, adult male Vglut2-ires-cre mice; n=5 adult male C57Bl/6J non-behavioural parkinsonian mice). Animals were housed at 14/10-hour light-dark cycles with ad libitum availability to rodent nutrition and water. No statistical methods were used to predetermine sample sizes. Mice were randomly assigned to experimental/control groups. Stereotaxic surgery procedures A stereotaxic apparatus (Kopf, David Kopf Instruments, USA or Neurostar, Neurostar Germany) was used for injecting viral tracers, optogenetic constructs, and AAVs containing A53T-mutated or EV genomic particles. The surgical procedures generally followed previously described protocols 70,71 . In detail, animals were maintained under isoflurane anesthesia (4% induction, 1-2% maintenance) throughout the procedures with systemic analgesic (Buprenorphine or Carprofen) administered ~30 minutes prior subcutaneously. Local anesthetic (Ropivacaine) was injected under the scalp, and a longitudinal incision was made after a two-minute wait. Veterinary eye ointment was applied to keep the eyes moisturized. A stereotaxic drill was used to create a small burr hole above the implantation and/or injection sites. We used the following stereotaxic coordinates, derived from the stereotaxic mouse brain atlas 72 : Vestibular Nucleus Complex (VNC): [AP: -6.00; ML: -1.20; DV: -4.20/30], Substantia Nigra Pars Compacta (SNC): [AP: -3.10, ML: -1.40, DV: -4.40, -4.35, -4.30], Dorsolateral Striatum (dlSTR): [AP: 0; ML: +2.20; DV: -3.00, -2.70, -2.50], Subthalamic Nucleus (STN): [AP: -2.00; ML: +1.50; DV: -4.90]. Optical fibers were placed ~200-300 µm above the injection site of the optogenetic viral construct. AAV1/2-A53T and AAV1/2-EV constructs were serially injected (DV: -4.40, -4.35, -4.30) to target the dorsoventral spread of dopaminergic SNC cells. retroAAV-mCherry was serially injected into the dlSTR (DV: -3.00, -2.70, -2.50). Post-injection, a ~6-minute wait time ensured adequate vector diffusion and minimized risk of viral spreading. The exposed skull was sutured for injection-only surgeries or covered with cyanoacrylate for optical fiber implantation to maximize stability. In postoperative care, animals were closely monitored, and analgesics (Carprofen or Meloxicam) were administered preemptively and repetitively. Mice received a minimum of 4 days of rest before starting any behavioural experiments and were continuously monitored day-to-day. Viral tracing For both anterograde and retrograde viral tracing studies, we utilized commercially available adeno-associated viruses (AAV) or AAV constructs that were gifted from other labs. Viruses were generated in an in-house viral preparation facility from plasmids according to standard protocols for viral cloning, clonal expansion, etc. For anterograde tracing of Vglut2 + glutamatergic VNC neurons, we utilized an myc-tagged anterograde synaptophysin viral tracer ( AAV2/5-CAG-Floxed-Syp10xMyc-rev.WPRE , UNC Vector Core). For concomitant retrograde tracing from the STR or STN, we used a retrograde tracer ( retroAAV-hSyn-DIO-mCherry , Addgene), driving the expression of mCherry. Viral tracers (both antero- and retrograde) were expressed for four weeks before euthanasia. Generation of the parkinsonian (A53T-aSyn) mouse model For the generation of a parkinsonian phenotype, we utilized the A53T-α-synuclein mouse model of PD 38 . We used a concentrated viral suspension (15x10 12 gp/ml) for both the AAV1/2-A53T and AAV1/2-EV construct (kindly provided by Dr. James Brotchie and Jonathan Koprich, Canada). The stereotaxic procedure and serial dorsoventral injections of the A53T and EV construct for generation of parkinsonian and EV mice, respectively, was conducted as described above. Verification of α-synuclein (aSyn) overexpression and dopaminergic neurodegeneration in the SNC was achieved by immunohistochemistry, which was followed by immunofluorescence microscopy and stereological estimation, respectively. Optogenetic experiments Mice used in optogenetic experiments with parkinsonian A53T vs. EV control, as well as the sham-stimulation mCherry group all had Vglut2-ires-cre background. For optical actuation in A53T and EV mice, we utilized ChRmine , a highly sensitive red-light activated cation-conducting channelrhodopsin (CCR) delivered via a plasmid inside an anterograde AAV construct ( AAV2/5-hSyn-DIO-ChRmine-mScarlet, Addgene). For sham actuation, the mCherry group was injected with an anterograde AAV construct, which delivered a plasmid lacking an optical actuator (construct ( AAV2/5-hSyn-DIO-mCherry, Addgene). Commercially acquired 4-4.5 mm optical fibres (Doric Lenses, Canada) were placed 0.2-0.3 mm above the target region. A numerical aperture of 0.53 was chosen for all used fibres. Laser power at the fibre tip was gauged before or after an optogenetic experiment with a photometer (ThorLabs Inc., New Jersey, USA). We utilized a laser driver (Doric Lenses, Canada) for generation of laser pulses and the accompanying Doric Neuroscience Studio software (Doric Lenses, Canada) for coding stimulation regimes. Considering data on tonic vestibular neuron firing rates and intrinsic properties of ChRmine 32,73,74 , we opted for 20 Hz photostimulation at 10 ms pulse width with 10 s pulse trains. Pause intervals were pseudorandomized to 30-40 s. In titration experiments (TV-OF), we stimulated at all 4 different VSS thresholds in an interspersed manner during a total recording period of 10 minutes per mouse. For experiments with perithreshold stimulation, we stimulated just below (i.e. 1 mA below threshold) the individual threshold for vestibular symptoms to appear (i.e. VSS 1 on the VSS) and in an interspersed manner during a total recording period of 10 minutes per mouse. On the treadmill, we stimulated on demand during each run. Behavioural testing and analysis Generalities and data pooling Mice for behavioural testing from the parkinsonian A53T and EV group with optogenetics were tested at baseline, a presymptomatic time point at week 1 after surgery. Repeat testing when A53T animals became symptomatic was conducted at week 4 and week 8. Since an independent cohort of A53T mice, euthanized at week 4 after injection of AAV1/2-A53T showed similar levels of dopaminergic cell loss, we pooled data from week 4 and 8 postoperatively for the A53T and EV group from optogenetic experiments. Data from week 4 off and week 8 off recordings were pooled as “off”. Data with perithreshold optoactivation of the VNC was pooled from week 4 and 8 as “perithreshold”. The mCherry sham group was tested at week 2 in off and with sham-optogenetics and with sham-optogenetics at week 3 after surgery, when viral expression plateaued 37 . Cylinder test The cylinder test was conducted as described previously 75,76 . We recorded mice in a transparent cylinder for 10 minutes after habituation on video. We then assessed both rotational asymmetry and asymmetry of forepaw use 38 as indeces by using the following formulae: Amphetamine-induced rotation test The amphetamine-induced rotation test was conducted as described previously 51 . At week 8, we injected mice intraperitoneally with 0,5 mg/kg body weight d-amphetamine and assessed for total rotations in the off vs. perithreshold condition with optogenetic actuation. Animals were pseudorandomized to start in the perithreshold vs. off condition, like in the cylinder test. A total 40 min (20 min-off; 20 min-perithreshold) were recorded. Ipsi-, contralateral and total rotations were counted by a human rater (M.F.). On this basis, statistical analyses were conducted. Top-view open field (TV-OF) Open field (OF) trials were conducted in a square, white open field box (39,5 x 39,5 x 39,5 cm). Individual trials lasted 10 minutes. We recorded videos at 30 frames per second (fps) with a proprietary behavioural analysis system (EthoVision, Noldus Inc., Wageningen, Netherlands). For kinematic feature analyses, we used a custom-trained deep learning model, trained within DeepLabCut (DLC) 40 on top-view videos of OF trials (see section Deep learning-based pose estimation). Afterwards, coordinate time-series were used for further analysis and feature extraction and analyzed via in-house built MATLAB (MATLAB R2022b, MathWorks, Natick, MA, USA) and Python (version 3.9. or 3.10) scripts. Treadmill gait analysis To assess gait thoroughly, we used a commercially available treadmill setup by Digigait (Mouse Specific Inc., Massachusetts, USA), which includes two (ventral and lateral) high-resolution cameras (Basler, 165 fps cameras, Basler AG, Ahrensburg, Germany). Animals were extensively habituated to the treadmill context before experiments. Both maximum velocity (V max , see next section) and recorded gait sessions (on video) at different treadmill belt velocities (10, 15, 20, 25 cm/s) were assessed at baseline, week 4 and week 8. First videos were recorded, then V max sessions conducted, to reduce systematic error due to exhaustion. Treadmill gait analysis - Maximum velocity (V max ) To assess gait-associated locomotor performance, maximum velocity (V max ) was used as a quantitative surrogate. Mice were tested to determine their V max at each time point, based on the hypothesis that increasing treadmill velocity requires greater physical effort to maintain stable gait. V max was tested with perithreshold stimulation and in the off condition at week 4 and week 8 (at baseline only off), to examine the effects of vestibular optostimulation. Animals were pseudorandomized to either perithreshold→off or off→perithreshold at weeks 4 and 8. V max analysis involved calculating the mean V max from 3 trials. V max was determined by progressively increasing treadmill speed until gait failure (hitting the bumper with the rear). Initial treadmill speed was set between 15 and 20 cm/s to ensure a start in stable gait, then increased progressively. V max was assessed regardless of treadmill belt direction (forwards or reverse). Treadmill gait analysis - Fluidity of gait index (FGI) In order to verify that suprathreshold vs. perithreshold stimulation impairs locomotion significantly, we devised a gait scoring system, based on previous findings of treadmill gait capacity with regards to consecutive strides in a murine disease model 55 . For this reason, we devised a trinomial scoring system, with which we aimed to capture overall gait stability per run: the fluidity of gait index (FGI) . We anchored the scoring system in a combined count of a) consecutive strides 55 and b) gait interruptions per treadmill run and animal as such that: one condition (worse counted) had to be met – FGI 1 = failure: consecutive strides ≤ 4, gait interruptions ≤ 3; FGI 2 = sufficient: consecutive strides = 5-7, gait interruptions ≤ 2; FGI 3 = fluid gait: consecutive strides ≥ 8, gait interruptions ≤ 1]. Evidently, any semiquantitative scoring system may be affected by interrater variability, hence we opted to validate the FGI via assessment of interrater agreement by Cohen’s kappa (k). Both behavioural experimenters (M.F. + J.H.) individually and blindly scored all treadmill runs according to the pre-devised FGI scoring system and interrater agreement was calculated. A Cohen’s kappa (k) of 0.619 (95% CI: 0.540-0.699) and weighted kappa of 0.671 (which is considered a more accurate measure when data are ordinally scaled) 77 ascertained the validity of the FGI scoring system and showed moderate two-rater inter-agreement in FGI allocation per individual trial. Deep learning-based pose estimation For body part keypoint tracking and pose estimation, we used DeepLabCut (DLC, version 2.3.1) 40,41 . Specifically, we trained n=2 different recurrent convolutional neural networks (RCNN) for a) the TV-OF and b) the combined lateral-ventral dual-view videos of treadmill gait. For the TV-OF RCNN, we utilized the pretrained RCNN ‘ superanimal_topviewmouse ’ 53 and labeled an additional n=255 number of frames taken from n=16 videos (then 95% was used for training). We used a ResNet-152-based neural network with default parameters for 7 training iterations. We validated with 1 shuffle, and found the test error was: 1.09 pixels, train: 2.04 pixels (image size was ≈ 640 by 624). We used a p-cutoff of 0.6. This network was then used to analyze videos from similar experimental settings. Coordinate time-series were further processed in the different custom-made python and MATLAB pipelines. For the RCNN used on combined lateral and ventral (i.e. bottom), dual-view videos (30 fps) of gait on the treadmill, we labeled n=1409 number of frames taken from n=118 videos (then 95% was used for training). We used a ResNet-50-based neural network with default parameters for 3 training iterations. We validated with 1 shuffle, and found the test error was: 1.66 pixels, train: 4.45 pixels (image size was ≈ 718 by 682). We used a p-cutoff of 0.6. This network was then used to analyze videos from similar experimental settings. Coordinate time-series were further processed in the different custom-made python and MATLAB pipelines. Deep learning-assisted gait analysis After training the RCNN via DLC, we analyzed an extensive dataset of treadmill videos (total n=387 videos from EV mice, total n=466 videos from A53T mice) from different speed settings 47 (10, 15, 20, 25 cm/s) from baseline, week 4 and week 8 testing. Table 1. Number of videos per condition (pooled from week 4 and week 8 evaluations): EV 10cm/s 15cm/s 20cm/s 25cm/s baseline 19 19 20 20 off 35 39 40 40 threshold 35 40 40 40 Parkinsonian (A53T) 10cm/s 15cm/s 20cm/s 25cm/s baseline 28 27 28 28 off 38 45 46 44 threshold 43 47 47 45 We then used the resultant coordinate time series for further analysis via a custom MATLAB script (MATLAB R2022b, MathWorks, Natick, MA, USA). Unfiltered time series of body part coordinates obtained from DeepLabCut were imported into MATLAB, where they were calibrated based on pre-recording snippets and aligned between the two cameras. Subsequently, frequency analysis was conducted using continuous wavelet transform via the Wavelet Toolbox 6.2 (MathWorks, Natick, MA, USA) after linearly interpolating missing values or values with a confidence score below 0.95. Data points falling outside the cone of influence, where wavelet analysis has low confidence due to limited data, were excluded from further analysis. Scalograms were visualized as montages, representing different speeds for a given mouse and condition. The magnitude of the wavelet coefficients over time was integrated for each frequency and normalized by the overall signal power, providing a power spectrum. Power spectrum, peak frequency and peak power were extracted per recording and averaged per mouse for pooled conditions (e.g. week 4 and week 8), and then across mice. Autocorrelograms were generated from normalized distances between pairs of markers. Power spectra and autocorrelograms were compared. Finally, an F-test was conducted to determine whether the relationships between peak frequency, peak amplitude, and speed differed significantly across experimental conditions. Linear models with and without interaction terms between these variables and the experimental condition were fit to the data. The F-test compared the goodness of fit between these models, with the p-value indicating whether the inclusion of interaction terms significantly improved the model's ability to explain the variance in the data. Unsupervised behavioural classification via Keypoint-MoSeq (KPMS) We trained a full KPMS model as described previously 52 after fitting an initial autoregressive hidden markov model (AR-HMM) with a kappa hyperparameter of 10000 (dimensionless) and resultant approximate median syllable duration of 12 at 25 frames per second (fps) ~ 300ms. Results of principal component analysis (PCA) on explained variance by the result of the AR-HMM showed that 6 principal components explained ≥ 90% of variance. The kappa (stickiness) hyperparameter was maintained at 10000 for fitting the full model (final median syllable duration also at 12 at 25 frames per second (fps) ~ 300ms). The model was applied on n=139 TV-OF videos from the parkinsonian A53T, EV (both at baseline, week 4, week 8, latter two with and without perithreshold stimulation) and sham mCherry mice (at week 2 (with and without (sham) stimulation) and week 3 with (sham) stimulation). Downstream statistical analysis and plotting was conducted as described previously 52,54 and is specifically described below: Syllable statistics All statistical analyses were performed using Python (version 3.9 or 3.10). To compare syllable statistics across multiple groups 52,54 , we performed Kruskal-Wallis tests, implemented manually with clustered permutation tests (10,000 permutations, significance threshold set at p < 0.05), including tie correction factors. For post-hoc analysis, Dunn’s z test was used for pairwise comparisons, with p-values adjusted using the Benjamini-Hochberg method to control the false discovery rate (FDR), and significance determined by an adjusted p-value threshold of 0.05. Syllable transition matrices and transition graphs Frequencies of syllable occurrences were calculated, excluding syllables with a frequency below 0.005. Transition matrices were generated to analyze transitions between syllables within each condition using custom Python scripts and functions, normalized by bigram counts, rows, and columns. Visualizations of transition matrices and syllable frequencies were created with Matplotlib and Seaborn. Transition graphs were generated using NetworkX and visualized with Matplotlib, with nodes representing syllables and edges representing transition probabilities. Node sizes were scaled based on syllable usage, and edges colored to represent transition probabilities. Differences in transition graphs between groups were analyzed by calculating differences in transition matrices and usage, visualized with NetworkX and Matplotlib. Eigenvalues and eigenvectors of transition matrices were computed to assess spectral properties, compared using permutation tests to generate a null distribution, and p-values were corrected using the Bonferroni method. Procrustes analysis was used to compare eigenvector spaces between groups. Node Scaling: Node sizes were determined by normalized usage differences, scaled by a factor (e.g., 3000) and a constant (e.g., 500) for visibility. The formula used was: Line Thickness Scaling: Edge thickness indicated transition probability differences, scaled by a factor (e.g., 1000) with colors indicating the direction of difference (green for positive, violet for negative). Semisupervised behavioural classification via A-SoiD To verify the Vestibular Symptom Scale (VSS) we devised, we made use of a semi-supervised machine learning approach within the A-SoiD pipeline, as described previously 42 by annotation of specific behaviours in an ethogram, leveraging the BORIS software 78 combined with pose estimation data. Specifically, we extensively trained an active learning classifier ( ‘main behaviour classifier’ ) on a diverse set of top-view pose estimation data using n = 31 snippet videos (~10 s length, i.e. single optostimulation pulse) of VSS symptomatology and n = 3 full titration videos (week 4, 10 min length) on identifying both vestibular symptoms and naturalistic behaviour. These videos were pose estimated within DLC with the model described previously in the section ‘Deep learning-based pose estimation’. Second, each video was loaded into BORIS and every single frame annotated in an ethogram with either one of two categories: 1) Naturalistic behaviour, 2) VSS-like behaviour. The ethograms were all exported as singular behaviour binary files at 0.04 s (25 Hertz = Hz). The training parameters were set as follows: DLC likelihood cutoff: 0.99, minimum bout duration: 0.05s, training fraction: 0.01, number of iterations: 100, confidence threshold for inclusion in training dataset: 0.95, video frame rate: 25 fps, ethogram rate: 0.04 s = 25 Hertz. The classifier achieved ~90% average performance ( see Supplementary Figure 2 for details ). We then predicted unseen titration data (week 8). Via a built-in unsupervised clustering/embedding algorithm (B-SoiD) 36 , we aimed to identify subtypes from pose, predicted by our active learning classifier. We clustered and embedded a diverse pose dataset of n = 261 snippet videos of vestibular symptomatology and naturalistic behaviour from week 4 titrations (hdbscan, min. percentage for cluster in both: 3%, normalization and relaxed embedding options deactivated). We fed these subclasses into a second, new active learning classifier ( ‘subtype behaviour classifier’ ) (~90% average performance), and re-predicted all full titration videos (10 min, week 4 and week 8 titration). The training parameters were set as follows: DLC likelihood cutoff: 0.99, minimum bout duration: 0.05s, training fraction: 0.01, number of iterations: 500, confidence threshold for inclusion in training dataset: 0.95, video frame rate: 25 fps, ethogram rate: 0.04 s = 25 Hz ( also see Supplementary Figure 2, Supplementary Videos 4-10 ). Brain processing and immunohistochemistry At the end of the experiments, animals were euthanized and transcardially perfused at 65 rpm at 90 minutes (for cFos expression) 49 after the last optogenetic experiment in the TV-OF with a mixture of 0.01mM phosphate-buffered saline (PBS) and 0.34% heparin sulfate for 3-5 minutes. Adequate perfusion was confirmed by observing hepatic blood clearing, adjusting needle placement or pump rate as needed. This was followed by 10-15 minutes of perfusion with 4% paraformaldehyde (PFA). Brains were then extracted, washed in 0.01mM PBS, and post-fixed in 4% PFA for 12 hours. Subsequently, brains were placed in a sterile-filtered 30% D-Saccharose solution for 48 hours until dehydration and then frozen in tissue gel matrix (O.C.T., TissueTek) at -20°C until cutting and staining. Brains were sectioned into 40µm slices using a cryotome (Leica PM1950, Leica) and placed in a cryoprotection solution (30% glycerol, 30% PBS, 40% ethanol). Before immunostaining, slices were washed in 0.01mM PBS and permeabilized with 0.1% Triton-X. For dopaminergic cell staining and α-synuclein overexpression verification, primary antibodies against tyrosine hydroxylase (chicken anti-tyrosine hydroxylase, 1:500, Abcam, ab76442) and α-synuclein (rabbit anti-α-synuclein, 1:30000, Sigma-Aldrich, Sigma #S3062) were used. Secondary antibodies were goat anti-chicken Alexa Fluor 488 (Invitrogen #A11039) and goat anti-rabbit Cy3 (Dianova #111-165-144). Slices were counterstained with DAPI (1:500,000, Sigma-Aldrich, Sigma #D8417) and mounted in embedding medium (AquaPolymount, Polysciences) for microscopy. For stereological estimation, primary antibody against tyrosine hydroxylase (rabbit anti-tyrosine hydroxylase, 1:1000, Abcam, ab112) were used with secondary antibody (goat anti-rabbit, 1:100, Vektor #BA-1000) followed by ABC (Thermo Scientific #32050) and DAB (Vektor #SK 4100). Slices were then mounted and counterstained with cresyl violet solution (1 gram cresyl violet (Merck#1.05235.0025) + 10ml acetic acid (100%; Sigma#33209-1L)) and subsequently put into an ascending concentration ethanol, (70-96-100%) and finally a xylol bath. Slices were mounted and embedded in Vitroclud medium (R. Langenbrinck GmbH, Emmendingen, Germany, Article-No. 04-0001). The mScarlet and mCherry fluorophore tagging the viruses was imaged natively. Exemplary images from Figure 1 and Figure 2o were counterstained with FluoroNissl (1:200, NeuroTrace™ 435/455 Blue Fluorescent, N21479, ThermoFischer). For viral tracing, primary antibodies against myc (goat-anti-Myc, 1:500, Abcam, ab9132) and red fluorescent protein (rabbit-anti-RFP, 1:500, Rockland/Biomol, #600-401-379) were used. Secondary antibodies were donkey anti-goat Cy5 (1:800, Jackson Laboratories #705-175-147) and donkey anti-rabbit Cy3 (1:800, Jackson Laboratories #711-165-152). Slices were counterstained with DAPI (1:500,000, Sigma-Aldrich, Sigma #D8417) and mounted in embedding medium (AquaPolymount, Polysciences) for microscopy. For the cFos staining 48,49 (the animals were transcardially perfused 90 minutes after final perithreshold optoactivation of the VNC in the TV-OF), we quenched brain slices for quantification of immunoreactive cells via DeepSlice2 with 100mM glycine at 7.4 pH and blocked and permeabilized with 0.3% Triton-X, 0.1% Tween 20 at 10% normal goat serum in 0.01mM PBS. Afterwards, slices were incubated in primary antibody (rabbit anti-cFos, 1:1000, Synaptic Systems, #226003) and tagged with secondary antibody (donkey anti-rabbit Cy5, 1:200, Jackson Laboratories, #711175152). Slices were counterstained with DAPI (1:500,000, Sigma-Aldrich, Sigma #D8417) and mounted in embedding medium (AquaPolymount, Polysciences) for microscopy. Microscopy and Stereology After immunofluorescence stainings, brain slices were imaged at a pseudoconfocal microscope (Axioimager M2, Zeiss) at different magnifications. Images for cFos analysis were taken on a region of interest (ROI) basis at 20x magnification of each hemisphere without a z-stack at 3 different anteroposterior levels. Viral tracing images were acquired in a similar fashion at the same pseudoconfocal microscope (Axioimager M2, Zeiss) at 10x and 20x magnification. Exemplary images in Figure 1 and Figure 2o were taken with a different pseudoconfocal imager (Leica Thunder DMi8 imager, Leica) at 10x magnification and z-stacked. Proprietary large volume computational clearing was used for background clearing. We used a maximum projection for these images. Brightness and contrast were adjusted for print visibility on all images. Direct immunohistochemistry images were further used for a stereological quantification of dopaminergic SN pars compacta (SNC) cells with an optical fractionator microscope (Olympus BX53, OM Digital Solutions), the StereoInvestigator 64-bit software (MBF Bioscience, Williston, USA) and a Prior ProScan III device (MBF Bioscience, Williston, USA) in an unbiased manner, as described previously 79 . For stereology, serial 40µm cuts of the SN were used and quantified (High magnification lens (used for quantification mode): 100x, Counting Frame X: 60.00 μm, Counting Frame Y: 60.00 μm, Grid Size X: 100.00 μm, Grid Size Y: 100.00 μm, Section Cut Thickness: 40.00 μm, GuardZone Type: Fixed Distance, GuardZone Height: 2.00 μm, Dissector Height: 16.00 μm). Both Nissl + (grandaverage neuron estimate) and TH + cells (dopaminergic neurons) were quantified and correlations established. cFos activity analysis and deep image segmentations We used an in-house built deep-learning model ensemble with U-net/encoder-decoder architecture for regional quantification of cFos immunoreactivity per ROI as described previously 44,80 with the deepflash2 pipeline. Three independent raters (J.H., T.P., D.D.) annotated a total of 27 immunohistochemistry images of cFos stainings (4 images from rats, 23 from mice). We used a test/train split of 8/19 (images) and trained an ensemble of 5 models. Ground truth was established and the U-net convolutional neural network (CNN) model ensemble compared to the 3 raters‘ performances. Considering the different dice scores (dc), we found that the model ensemble for the most part performed at least as well as the individual raters in terms of accuracy (dc to estimated ground truth≈0.8). Via the ensemble, segmentation went as follows: instance segmentations by Cellpose, semantic segmentations via deepflash2 . We imaged n=3 anteroposterior levels for all ROIs: anterior, middle, posterior for each mouse (A53T: n=12, EV: n=10, mCherry: n=6) at a pseudoconfocal microscope (Axioimager M2, Zeiss). Subsequently, we calculated the feature density in features per micrometer squared using a Python script (Python, Version 3.9). Finally, we pooled results from both hemispheres for further statistical analysis to get a sense of overall regional activation, considering the differential, yet overwhelmingly bilateral innervation of thalamo-subthalamic and caudal medulla targets. We analyzed a total of 526 images (A53T: n=216, EV: n=184, mCherry: n=127) from the groups above, by generating masks of ROIs by using atlas outlines of brain regions 72 . Transfected cell count estimation We estimated the amount of transfected Vglut2-positive neurons in the vestibular nucleus complex (VNC) manually by quantification inside Fiji/ImageJ (doi:10.1038/nmeth.2019). For each individual mouse, we took three images of the VNC along the anteroposterior axis, overlayed the atlas outline 72 and quantified the number of mScarlet-positive cells in the VNC. We formed an average across all 3 anteroposterior levels per animal. This average was taken as the transfected cell count estimate and analyzed subsequently. Statistical analysis, plotting and data reporting Statistical analyses and plotting of analyses were conducted using GraphPad Prism (Version 10.0). Further, we utilized custom-built MATLAB (R2022b, MathWorks, Natick, MA, USA) and Python (Python, Version 3.9 or 3.10) scripts for statistical analyses and plotting. Assessment for normality was conducted by using normality/lognormality testing (most reliable test according to sample size was used) and assessment of Q-Q plots. Individual statistical testing was conducted on a group level via analysis of variance (ANOVA, one- or two-way) or mixed-effects models whenever possible and relevant, to assess for factor and interaction effects. In case of non-gaussian distribution of data, corresponding standard non-parametric equivalents were used. The OF kinematics of sham mCherry mice were compared with a paired t-test. Interrater reliability of the FGI score was assessed by using the Cohen’s kappa online calculator by GraphPad, found on the GraphPad website (https://www.graphpad.com/quickcalcs/kappa1.cfm). For cFos quantification and VSS thresholds, we removed significant outliers with the ROUT method in GraphPad Prism (Version 10.0). Data plotting is described individually for each plot in the figure legends. Significance is annotated as *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Figures were created with Adobe Illustrator and BioRender. Source data and custom code is available from the corresponding authors upon reasonable request. References Lau, L. M. de & Breteler, M. M. Epidemiology of Parkinson’s disease. Lancet Neurol. 5 , (2006). Kalia, L. V. & Lang, A. E. Parkinson’s disease. The Lancet 386 , 896–912 (2015). Armstrong, M. J. & Okun, M. S. Time for a New Image of Parkinson Disease. JAMA Neurol. 77 , 1345 (2020). Beuter, A., Hernández, R., Rigal, R., Modolo, J. & Blanchet, P. J. Postural sway and effect of levodopa in early Parkinson’s disease. Can. J. Neurol. Sci. J. Can. Sci. Neurol. 35 , 65–68 (2008). Kim, S. D., Allen, N. E., Canning, C. G. & Fung, V. S. C. Postural Instability in Patients with Parkinson’s Disease. CNS Drugs 27 , (2013). Szlufik, S. et al. The Neuromodulatory Impact of Subthalamic Nucleus Deep Brain Stimulation on Gait and Postural Instability in Parkinson’s Disease Patients: A Prospective Case Controlled Study. Front. Neurol. 9 , (2018). Moreira, F., Gomes, I. R. & Januário, C. Freezing of gait and postural instability: the unpredictable response to levodopa in Parkinson’s disease. BMJ Case Rep. 12 , (2019). Kotagal, V. Is PIGD a legitimate motor subtype in Parkinson disease? Ann. Clin. Transl. Neurol. 3 , 473–477 (2016). Arber, S. & Costa, R. M. Networking brainstem and basal ganglia circuits for movement. Nat. Rev. Neurosci. 23 , 342–360 (2022). Fougère, M. et al. Optogenetic stimulation of glutamatergic neurons in the cuneiform nucleus controls locomotion in a mouse model of Parkinson’s disease. Proc. Natl. Acad. Sci. U. S. A. 118 , e2110934118 (2021). Masini, D. & Kiehn, O. Targeted activation of midbrain neurons restores locomotor function in mouse models of parkinsonism. Nat. Commun. 13 , 504 (2022). Caggiano, V. et al. Midbrain circuits that set locomotor speed and gait selection. Nature 553 , (2018). Ryczko, D. The Mesencephalic Locomotor Region: Multiple Cell Types, Multiple Behavioral Roles, and Multiple Implications for Disease. The Neuroscientist 10738584221139136 (2022) doi:10.1177/10738584221139136. Thevathasan, W. et al. Pedunculopontine nucleus deep brain stimulation in Parkinson’s disease: A clinical review. Mov. Disord. Off. J. Mov. Disord. Soc. 33 , 10–20 (2018). Wang, H., Gao, H., Jiao, T. & Luo, Z. A meta-analysis of the pedunculopontine nucleus deep-brain stimulation effects on Parkinson’s disease. NeuroReport 27 , 1336–1344 (2016). Violante, I. R. et al. Non-invasive temporal interference electrical stimulation of the human hippocampus. Nat. Neurosci. 26 , 1994–2004 (2023). Wessel, M. J. et al. Noninvasive theta-burst stimulation of the human striatum enhances striatal activity and motor skill learning. Nat. Neurosci. 26 , 2005–2016 (2023). Pires, A. P. B. de Á. et al. Galvanic vestibular stimulation and its applications: a systematic review. Braz. J. Otorhinolaryngol. 88 Suppl 3 , S202–S211 (2022). Khoshnam, M., Häner, D. M. C., Kuatsjah, E., Zhang, X. & Menon, C. Effects of galvanic vestibular stimulation on upper and lower extremities motor symptoms in parkinson’s disease. Front. Neurosci. 12 , (2018). Wilkinson, D. et al. Caloric vestibular stimulation for the management of motor and non-motor symptoms in Parkinson’s disease. Parkinsonism Relat. Disord. 65 , 261–266 (2019). Wilkinson, D. Caloric and galvanic vestibular stimulation for the treatment of Parkinson’s disease: rationale and prospects. Expert Rev. Med. Devices 18 , 649–655 (2021). Wuehr, M. et al. Effects of Low-Intensity Vestibular Noise Stimulation on Postural Instability in Patients with Parkinson’s Disease. J. Park. Dis. 12 , 1611–1618 (2022). Lai, H. et al. Morphological evidence for a vestibulo-thalamo-striatal pathway via the parafascicular nucleus in the rat. Brain Res. 872 , (2000). Leong, A. T. L. et al. Optogenetic fMRI interrogation of brain-wide central vestibular pathways. Proc. Natl. Acad. Sci. U. S. A. 116 , 10122–10129 (2019). Kataoka, H. et al. Effect of galvanic vestibular stimulation on axial symptoms in Parkinson’s disease. J. Cent. Nerv. Syst. Dis. 14 , 11795735221081599 (2022). Wuehr, M. et al. Low-intensity vestibular noise stimulation improves postural symptoms in progressive supranuclear palsy. J. Neurol. (2024) doi:10.1007/s00415-024-12419-9. Peto, D. et al. No evidence for effects of low-intensity vestibular noise stimulation on mild-to-moderate gait impairments in patients with Parkinson’s disease. J. Neurol. 271 , 5489 (2024). Samoudi, G., Nissbrandt, H., Dutia, M. B. & Bergquist, F. Noisy galvanic vestibular stimulation promotes GABA release in the substantia nigra and improves locomotion in Hemiparkinsonian rats. PLoS ONE 7 , (2012). Takazawa, T., Saito, Y., Tsuzuki, K. & Ozawa, S. Membrane and firing properties of glutamatergic and GABAergic neurons in the rat medial vestibular nucleus. J. Neurophysiol. 92 , 3106–3120 (2004). Bagnall, M. W., Stevens, R. J. & du Lac, S. Transgenic mouse lines subdivide medial vestibular nucleus neurons into discrete, neurochemically distinct populations. J. Neurosci. Off. J. Soc. Neurosci. 27 , 2318–2330 (2007). Montardy, Q. et al. Selective optogenetic stimulation of glutamatergic, but not GABAergic, vestibular nuclei neurons induces immediate and reversible postural imbalance in mice. Prog. Neurobiol. 204 , (2021). Kodama, T. et al. Neuronal Classification and Marker Gene Identification via Single-Cell Expression Profiling of Brainstem Vestibular Neurons Subserving Cerebellar Learning. J. Neurosci. 32 , 7819–7831 (2012). Stiles, L. & Smith, P. F. The vestibular–basal ganglia connection: Balancing motor control. Brain Res. 1597 , 180–188 (2015). Sabzevar, F. T., Vautrelle, N., Zheng, Y. & Smith, P. F. Vestibular modulation of the tail of the rat striatum. Sci. Rep. (2023). Smith, P. F. Vestibular Functions and Parkinson’s Disease. Front. Neurol. 9 , (2018). Hsu, A. I. & Yttri, E. A. B-SOiD, an open-source unsupervised algorithm for identification and fast prediction of behaviors. Nat. Commun. 12 , 5188 (2021). Kishi, K. E. et al. Structural basis for channel conduction in the pump-like channelrhodopsin ChRmine. Cell 185 , 672-689.e23 (2022). Ip, C. W. et al. AAV1/2-induced overexpression of A53T-α-synuclein in the substantia nigra results in degeneration of the nigrostriatal system with Lewy-like pathology and motor impairment: a new mouse model for Parkinson’s disease. Acta Neuropathol. Commun. 5 , 11 (2017). Laflamme, O. D. & Akay, T. Excitatory and inhibitory crossed reflex pathways in mice. J. Neurophysiol. 120 , 2897–2907 (2018). Mathis, A. et al. DeepLabCut: markerless pose estimation of user-defined body parts with deep learning. Nat. Neurosci. 21 , 1281–1289 (2018). Nath, T. et al. Using DeepLabCut for 3D markerless pose estimation across species and behaviors. Nat. Protoc. 14 , 2152–2176 (2019). Tillmann, J. F., Hsu, A. I., Schwarz, M. K. & Yttri, E. A. A-SOiD, an active-learning platform for expert-guided, data-efficient discovery of behavior. Nat. Methods 21 , 703–711 (2024). Becker-Bense, S. et al. Direct comparison of activation maps during galvanic vestibular stimulation: A hybrid H2[15 O] PET—BOLD MRI activation study. PLOS ONE 15 , e0233262 (2020). Griebel, M. et al. Deep learning-enabled segmentation of ambiguous bioimages with deepflash2. Nat. Commun. 14 , 1679 (2023). Watson, G. D. R. et al. Thalamic projections to the subthalamic nucleus contribute to movement initiation and rescue of parkinsonian symptoms. Sci. Adv. 7 , eabe9192 (2021). Zhang, Y. et al. Targeting thalamic circuits rescues motor and mood deficits in PD mice. Nature 607 , 321–329 (2022). Bellardita, C. & Kiehn, O. Phenotypic characterization of speed-associated gait changes in mice reveals modular organization of locomotor networks. Curr. Biol. CB 25 , 1426–1436 (2015). Bullitt, E. Expression of c-fos-like protein as a marker for neuronal activity following noxious stimulation in the rat. J. Comp. Neurol. 296 , 517–530 (1990). Herrera, D. G. & Robertson, H. A. Activation of c-fos in the brain. Prog. Neurobiol. 50 , 83–107 (1996). Paumier, K. L. et al. Behavioral Characterization of A53T Mice Reveals Early and Late Stage Deficits Related to Parkinson’s Disease. PLoS ONE 8 , e70274 (2013). Björklund, A. & Dunnett, S. B. The Amphetamine Induced Rotation Test: A Re-Assessment of Its Use as a Tool to Monitor Motor Impairment and Functional Recovery in Rodent Models of Parkinson’s Disease. J. Park. Dis. 9 , 17–29 (2019). Weinreb, C. et al. Keypoint-MoSeq: parsing behavior by linking point tracking to pose dynamics. Nat. Methods 21 , 1329–1339 (2024). Ye, S. et al. SuperAnimal pretrained pose estimation models for behavioral analysis. Nat. Commun. 15 , 5165 (2024). Wiltschko, A. B. et al. Mapping Sub-Second Structure in Mouse Behavior. Neuron 88 , 1121–1135 (2015). Vincelette, J. et al. Gait analysis in a murine model of collagen-induced arthritis. Arthritis Res. Ther. 9 , R123 (2007). Brandt, T. Vestibular disorders in (horizontal) yaw plane. in Vertigo: Its Multisensory Syndromes (ed. Brandt, T.) 215–218 (Springer, New York, NY, 2003). doi:10.1007/978-1-4757-3801-8_12. Brandt, T. & Dieterich, M. Vestibular syndromes in the roll plane: Topographic diagnosis from brainstem to cortex. Ann. Neurol. 36 , 337–347 (1994). Brandt, T. & Dietrich, M. Skew deviation with ocular torsion: a vestibular brainstem sign of topographic diagnostic value. Ann. Neurol. 33 , 528–534 (1993). Friedrich, M. U. et al. Current-dependent ocular tilt reaction in deep brain stimulation of the subthalamic nucleus: Evidence for an incerto-interstitial pathway? Eur. J. Neurol. 29 , 1545–1549 (2022). Lopez, C. & Blanke, O. The thalamocortical vestibular system in animals and humans. Brain Res. Rev. 67 , 119–146 (2011). Cullen, K. E. The vestibular system: multimodal integration and encoding of self-motion for motor control. Trends Neurosci. 35 , 185–196 (2012). Basaldella, E., Takeoka, A., Sigrist, M. & Arber, S. Multisensory Signaling Shapes Vestibulo-Motor Circuit Specificity. Cell 163 , 301–312 (2015). Brandt, T. & Dieterich, M. Central vestibular syndromes in roll, pitch, and yaw planes: Topographic diagnosis of brainstem disorders. Neuro-Ophthalmol. 15 , 291–303 (1995). Mandelbaum, G. et al. Distinct Cortical-Thalamic-Striatal Circuits through the Parafascicular Nucleus. Neuron 102 , 636-652.e7 (2019). Fallon, I. P. et al. The role of the parafascicular thalamic nucleus in action initiation and steering. Curr. Biol. 33 , 2941-2951.e4 (2023). Leiras, R., Cregg, J. M. & Kiehn, O. Brainstem Circuits for Locomotion. Annu. Rev. Neurosci. 45 , 63–85 (2022). Cai, J. et al. Galvanic Vestibular Stimulation (GVS) Augments Deficient Pedunculopontine Nucleus (PPN) Connectivity in Mild Parkinson’s Disease: fMRI Effects of Different Stimuli. Front. Neurosci. 12 , (2018). Wang, X., Chou, X., Zhang, L. I. & Tao, H. W. Zona Incerta: An Integrative Node for Global Behavioral Modulation. Trends Neurosci. 43 , 82–87 (2020). Mitrofanis, J. Some certainty for the “zone of uncertainty”? Exploring the function of the zona incerta. Neuroscience 130 , 1–15 (2005). Tovote, P. et al. Midbrain circuits for defensive behaviour. Nature 534 , 206–212 (2016). Letzkus, J. J. et al. A disinhibitory microcircuit for associative fear learning in the auditory cortex. Nature 480 , 331–335 (2011). Paxinos, G. & Franklin, K. Paxinos and Franklin’s the Mouse Brain in Stereotaxic Coordinates - 5th Edition . (2019). Av-Ron, E. & Vidal, P.-P. Intrinsic membrane properties and dynamics of medial vestibular neurons: a simulation. Biol. Cybern. 80 , 383–392 (1999). Johnston, A. R., MacLeod, N. K. & Dutia, M. B. Ionic conductances contributing to spike repolarization and after-potentials in rat medial vestibular nucleus neurones. J. Physiol. 481 , 61–77 (1994). Glajch, K. E., Fleming, S. M., Surmeier, D. J. & Osten, P. Sensorimotor assessment of the unilateral 6-hydroxydopamine mouse model of Parkinson’s disease. Behav. Brain Res. 230 , 309–316 (2012). Fleming, S. M., Ekhator, O. R. & Ghisays, V. Assessment of Sensorimotor Function in Mouse Models of Parkinson’s Disease. J. Vis. Exp. JoVE 50303 (2013) doi:10.3791/50303. Landis, J. R. & Koch, G. G. An application of hierarchical kappa-type statistics in the assessment of majority agreement among multiple observers. Biometrics 33 , 363–374 (1977). Friard, O. & Gamba, M. BORIS : a free, versatile open‐source event‐logging software for video/audio coding and live observations. Methods Ecol. Evol. 7 , 1325–1330 (2016). Baquet, Z. C., Williams, D., Brody, J. & Smeyne, R. J. A comparison of model-based (2D) and design-based (3D) stereological methods for estimating cell number in the substantia nigra pars compacta (SNpc) of the C57BL/6J mouse. Neuroscience 161 , 1082–1090 (2009). Segebarth, D. et al. On the objectivity, reliability, and validity of deep learning enabled bioimage analyses. eLife 9 , e59780 (2020). Additional Declarations There is NO Competing Interest. Supplementary Files SupplementaryVideo1Scale.mp4 Supplementary Video1: Titration framework SupplementaryVideo2Montage1388.avi Supplementary Video 2: Vestibular Symptom Scale (VSS) 1 SupplementaryVideo3Montage2438.avi Supplementary Video 3: Vestibular Symptom Scale (VSS) 2 SupplementaryVideo4VS0.mp4 Supplementary Video 4: Video for UMAP plot for Fig 1 e SupplementaryVideo5VS1.mp4 Supplementary Video 5: Video for UMAP plot for Fig 1 e SupplementaryVideo6VS2.mp4 Supplementary Video 6: Video for UMAP plot for Fig 1 e SupplementaryVideo7VS3.mp4 Supplementary Video 7: Video for UMAP plot for Fig 1 e SupplementaryVideo8VS4circling.mp4 Supplementary Video 8: Video for UMAP plot for Fig 1 e SupplementaryVideo9VS4circlingretrolocomotion.mp4 Supplementary Video 9: Video for UMAP plot for Fig 1 e SupplementaryVideo10VS4crossedextensorlike.mp4 Supplementary Video 10: Video for UMAP plot for Fig 1 e SupplementaryVideo11syllable0.mp4 Supplementary Video 11: Syllable video 0 for Fig 4 b SupplementaryVideo12syllable1.mp4 Supplementary Video 12: Syllable video 1 for Fig 4 b SupplementaryVideo13syllable2.mp4 Supplementary Video 13: Syllable video 2 for Fig 4 b SupplementaryVideo14syllable3.mp4 Supplementary Video 14: Syllable video 3 for Fig 4 b SupplementaryVideo15syllable4.mp4 Supplementary Video 15: Syllable video 4 for Fig 4 b SupplementaryVideo16syllable5.mp4 Supplementary Video 16: Syllable video 5 for Fig 4 b SupplementaryVideo17syllable6.mp4 Supplementary Video 17: Syllable video 6 for Fig 4 b SupplementaryVideo18syllable7.mp4 Supplementary Video 18: Syllable video 7 for Fig 4 b SupplementaryVideo19syllable8.mp4 Supplementary Video 19: Syllable video 8 for Fig 4 b SupplementaryVideo20syllable9.mp4 Supplementary Video 20: Syllable video 9 for Fig 4 b SupplementaryVideo21syllable10.mp4 Supplementary Video 21: Syllable video 10 for Fig 4 b SupplementaryVideo22syllable11.mp4 Supplementary Video 22: Syllable video 11 for Fig 4 b SupplementaryVideo23syllable12.mp4 Supplementary Video 23: Syllable video 12 for Fig 4 b SupplementaryVideo24syllable13.mp4 Supplementary Video 33: Syllable video 22 for Fig 4 b SupplementaryVideo25syllable14.mp4 Supplementary Video 25: Syllable video 14 for Fig 4 b SupplementaryVideo26syllable15.mp4 Supplementary Video 26: Syllable video 15 for Fig 4 b SupplementaryVideo27syllable16.mp4 Supplementary Video 24: Syllable video 13 for Fig 4 b SupplementaryVideo28syllable17.mp4 Supplementary Video 28: Syllable video 17 for Fig 4 b SupplementaryVideo29syllable18.mp4 Supplementary Video 27: Syllable video 16 for Fig 4 b SupplementaryVideo30syllable19.mp4 Supplementary Video 29: Syllable video 18 for Fig 4 b SupplementaryVideo31syllable20.mp4 Supplementary Video 31: Syllable video 20 for Fig 4 b SupplementaryVideo32syllable21.mp4 Supplementary Video 30: Syllable video 19 for Fig 4 b SupplementaryVideo33syllable22.mp4 Supplementary Video 32: Syllable video 21 for Fig 4 b SupplementaryVideo34SuprathresholdStimDual20RS25cms0degUP.mp4 Supplementary Video 34 SupplementaryVideo35KPMSalltrajectories.gif Supplementary Video 35: ExtendedDataandSupplementoryFigures.docx Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-5851215","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":407724815,"identity":"796074b0-6eed-4ac4-9df9-de73a27f08f9","order_by":0,"name":"Johannes Hartig","email":"","orcid":"https://orcid.org/0000-0002-3669-8856","institution":"University Hospital Wuerzburg, Department of Neurology, Wuerzburg, Germany","correspondingAuthor":false,"prefix":"","firstName":"Johannes","middleName":"","lastName":"Hartig","suffix":""},{"id":407724816,"identity":"14119096-1436-455a-a360-c0b1de9de1d5","order_by":1,"name":"Maximilian Friedrich","email":"","orcid":"https://orcid.org/0000-0002-3695-1268","institution":"University Hospital Wurzburg","correspondingAuthor":false,"prefix":"","firstName":"Maximilian","middleName":"","lastName":"Friedrich","suffix":""},{"id":407724817,"identity":"15d564cc-5e09-420c-87a3-0c87c6586dac","order_by":2,"name":"Jérémy Signoret-Genest","email":"","orcid":"https://orcid.org/0000-0001-6373-0324","institution":"Institute of Clinical Neurobiology, University Hospital Wuerzburg","correspondingAuthor":false,"prefix":"","firstName":"Jérémy","middleName":"","lastName":"Signoret-Genest","suffix":""},{"id":407724818,"identity":"a4f49ab4-a2c7-4d10-8678-0b8465b03654","order_by":3,"name":"Sawako Tabuchi","email":"","orcid":"","institution":"Department of Neurology, University Hospital Wuerzburg, Wuerzburg, Germany","correspondingAuthor":false,"prefix":"","firstName":"Sawako","middleName":"","lastName":"Tabuchi","suffix":""},{"id":407724819,"identity":"246c0b52-5c99-4bcd-979f-828ee921a3fb","order_by":4,"name":"Nina Schöneberg","email":"","orcid":"","institution":"Institute of Clinical Neurobiology, University Hospital Wuerzburg, Wuerzburg, Germany","correspondingAuthor":false,"prefix":"","firstName":"Nina","middleName":"","lastName":"Schöneberg","suffix":""},{"id":407724820,"identity":"1451c1a3-e176-4953-bdd5-60f3dfdfba33","order_by":5,"name":"Robert Peach","email":"","orcid":"https://orcid.org/0000-0002-8738-5825","institution":"Department of Neurology, University Hospital Wuerzburg, Wuerzburg, Germany","correspondingAuthor":false,"prefix":"","firstName":"Robert","middleName":"","lastName":"Peach","suffix":""},{"id":407724821,"identity":"cd7d4215-ab71-4a59-abb7-fbc1c12e574c","order_by":6,"name":"Dennis Doll","email":"","orcid":"https://orcid.org/0000-0002-3806-9324","institution":"Indoc Research Europe gGmbH","correspondingAuthor":false,"prefix":"","firstName":"Dennis","middleName":"","lastName":"Doll","suffix":""},{"id":407724822,"identity":"e7420450-dbb7-484a-83a3-67e9c4475b81","order_by":7,"name":"Tobias Petschner","email":"","orcid":"","institution":"Department of Neurology, University Hospital Wuerzburg, Wuerzburg, Germany","correspondingAuthor":false,"prefix":"","firstName":"Tobias","middleName":"","lastName":"Petschner","suffix":""},{"id":407724823,"identity":"07ce4f1e-e684-4233-9335-4529a8e039b9","order_by":8,"name":"Susanne Knorr","email":"","orcid":"","institution":"University Hospital of Wuerzburg, Department of Neurology","correspondingAuthor":false,"prefix":"","firstName":"Susanne","middleName":"","lastName":"Knorr","suffix":""},{"id":407724824,"identity":"46d1f2ec-ecfd-4e58-92ab-2985ae739832","order_by":9,"name":"Jens Volkmann","email":"","orcid":"https://orcid.org/0000-0002-9570-593X","institution":"University Hospital Wuerzburg, Department of Neurology, Wuerzburg, Germany","correspondingAuthor":false,"prefix":"","firstName":"Jens","middleName":"","lastName":"Volkmann","suffix":""},{"id":407724825,"identity":"d9bc1e4a-be67-49de-acf4-5ff74db2e0e0","order_by":10,"name":"Philip Tovote","email":"","orcid":"https://orcid.org/0000-0002-4598-722X","institution":"Institute of Clinical Neurobiology, University Hospital Wuerzburg","correspondingAuthor":false,"prefix":"","firstName":"Philip","middleName":"","lastName":"Tovote","suffix":""},{"id":407724814,"identity":"3e3ed80d-853a-4679-9653-d7758a123bf7","order_by":11,"name":"Chi Wang Ip","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAkklEQVRIiWNgGAWjYBACNhDxgSEBSCaQoIVxBklaQICZhyQtfNINjJ9t29ISG9iTDxDpMJkDzNK5bTmJDTzPiLSGTSKBjTm3rSK3QSLHgAQtlmAt+R9I0MLYlgOyhTgdYL9I9pxLq2/jeUakw+RnNzB++FGWbMzPnvyASGsk+CFeYCNSPUgL8UpHwSgYBaNgpAIA+AsjioO+xmYAAAAASUVORK5CYII=","orcid":"https://orcid.org/0000-0003-0484-385X","institution":"University Hospital Wuerzburg, Department of Neurology, Wuerzburg, Germany","correspondingAuthor":true,"prefix":"","firstName":"Chi","middleName":"Wang","lastName":"Ip","suffix":""}],"badges":[],"createdAt":"2025-01-17 18:20:53","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5851215/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5851215/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":75875678,"identity":"75380251-565c-4017-9f00-c05001993aac","added_by":"auto","created_at":"2025-02-10 07:52:25","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":352876,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGlutamatergic vestibular nucleus complex neurons regulate posturolocomotor subtypes. a.\u003c/strong\u003e Depiction of optogenetics group (n=24), segregated into parkinsonian A53T (n=14) and healthy EV (n=10) mice by additive viral vector injection. Vestibular nucleus complex (VNC)=dotted line, arrow=optical fibre tip, lines=optical fibre trajectory. Scale bar: 500 microns. Right top: Comparison of mScarlet-positive VNC cells in A53T vs. EV mice (Two-tailed unpaired t-test, p\u0026gt;0.05; A53T: n=12, EV: n=10). Data are mean ± s.e.m. Right bottom: Example image around an optical fibre tip. Scale bar: 100 microns. \u003cstrong\u003eb. \u003c/strong\u003eDepiction of sham optogenetics group (n=6). Scale bar: 500 microns. \u003cstrong\u003ec.\u003c/strong\u003e Titration framework for optical actuation of VNC at increasing thresholds yielding different vestibular symptoms (\u003cstrong\u003eSupplementary Video 1\u003c/strong\u003e). \u003cstrong\u003ed.\u003c/strong\u003e V\u003cem\u003eestibular Symptom Scale (VSS)\u003c/em\u003e according to the observed phenotype-power relationships with\u003cstrong\u003e \u003c/strong\u003eincreasing levels of VNC optoactivation (\u003cstrong\u003eSupplementary Video 2 and 3\u003c/strong\u003e). \u003cstrong\u003ee.\u003c/strong\u003e UMAP plot shows clustering and embedding of behaviourally classified pose estimation via B-SoiD \u003csup\u003e36\u003c/sup\u003e on n=261 snippet videos from week 4 titrations, which identified 5 distinct clusters in vestibular symptomatology (Kruskal-Wallis test, H-statistic: 3982.5, p\u0026lt;1×10\u003csup\u003e−300\u003c/sup\u003e, Dunn’s multiple comparisons all p\u0026lt;3.6149×10\u003csup\u003e−31\u003c/sup\u003e), which were used to predict on week 8 videos (\u003cstrong\u003eSupplementary Video 4-10\u003c/strong\u003e). \u003cstrong\u003ef. \u003c/strong\u003eUMAP plot shows ten significantly different clusters in naturalistic behaviour (Kruskal-Wallis test, H-statistic: 18404.1, p\u0026lt;1×10\u003csup\u003e−300\u003c/sup\u003e, Dunn’s multiple comparisons all p\u0026lt;2.3999×10\u003csup\u003e-4\u003c/sup\u003e). \u003cstrong\u003eg-j.\u003c/strong\u003e Assessment of relationship between laser power and mScarlet-positive VNC cell count (n=19 of total cohort, A53T: n=10, EV: n=9; VSS1: Spearman’s r=-0,4553, p=0.0501; VSS2: Spearman’s r=-0,2981, p=0.2151 VSS3: Spearman’s r=-0,2856, p=0.2359, VSS4: Spearman’s r=-0.4756, p=0,0396). Lines=linear intercepts, dotted lines=95% confidence intervals. \u003cstrong\u003ek.\u003c/strong\u003e VSS 1-4 thresholds longitudinally at baseline vs. week 8 (Friedman test with Dunn’s multiple comparisons; n=19 of total cohort; VSS1: p\u0026lt;0.05, VSS2: p\u0026gt;0.01, VSS3: p\u0026lt;0.01, VSS4: p\u0026lt;0.05. Lines are medians, boxes are interquartile ranges, whiskers are most extreme data points. \u003cstrong\u003el. \u003c/strong\u003ePerithreshold vestibular optostimulation paradigm as an ‘optogenetic mimic’ to human (noisy) galvanic vestibular stimulation (nGVS) for therapeutic neuromodulation by stimulation just below threshold for vestibular symptoms.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/e96de2ece4e69f6bde827668.png"},{"id":75875679,"identity":"feb16168-f089-4074-9ac5-e7d5db1c294a","added_by":"auto","created_at":"2025-02-10 07:52:25","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":387630,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGlutamatergic vestibular nucleus complex neurons structurally and functionally link with striato-thalamo-subthalamic and caudal medulla circuitry. a. \u003c/strong\u003eSketch denotes\u003cstrong\u003e \u003c/strong\u003etracing cohort (n=7, all SypMyc) and additive injection paradigms (retroAVV into STN: n=3, dlSTR: n=4). \u003cstrong\u003eb. \u003c/strong\u003eWiring diagram of glutamatergic VNC motor targets; for STN-injected animals, we found no consistent overlaps.\u003cstrong\u003e c. \u003c/strong\u003eApparent laterality of targets [human rating upon inspection] (3 sections/animal). +=sparse, ++=dense.\u003cstrong\u003e d. \u003c/strong\u003eSynaptic punctae in the zona incerta (ZI) (n=7). \u003cstrong\u003ee.\u003c/strong\u003e mScarlet fibres in ZI of A53T/EV animals (n=22). \u003cstrong\u003ef. \u003c/strong\u003eSynaptic punctae (n=7) in parafascicular (Pf) and posterior oral (Po) nucleus of thalamus. VNC synapses and thalamo-striatal neurons overlap (n=4) in Pf. \u003cstrong\u003eg. \u003c/strong\u003emScarlet fibres in Pf+Po (n=22). \u003cstrong\u003eh. \u003c/strong\u003eSynaptic punctae (n=7) and thalamo-striatal neurons overlap (n=4) in the centromedian (CM) nucleus. \u003cstrong\u003ei. \u003c/strong\u003emScarlet signal in CM (n=22). \u003cstrong\u003ej.\u003c/strong\u003e Some sparse thalamic projections in ventral nucleus group (n=7). \u003cstrong\u003ek. \u003c/strong\u003emScarlet projections in the former (n=22). \u003cstrong\u003el.\u003c/strong\u003e Medullary projections into (anterior) gigantocellular (Gi(A)), lateral paragigantocellular nuclei (LPGi) and contralateral VNC (commissural) (n=7). \u003cstrong\u003em.\u003c/strong\u003e Very few, if any, synaptic punctae in the MLR, specifically pedunculopontine nuclei (n=7). \u003cstrong\u003en.\u003c/strong\u003e Some mScarlet-positive fibres in the mesencephalic locomotor region (MLR, n=22). \u003cstrong\u003eo. \u003c/strong\u003eExample of thalamo-subthalamic innervation targets by VNC\u003cstrong\u003e. p. \u003c/strong\u003eQuantification of cFos-positive cells by \u003cem\u003edeepflash2\u003c/em\u003e \u003csup\u003e44\u003c/sup\u003e in the major projection targets and VNC. n=526 images (A53T: n=216, EV: n=184, mCherry: n=127) analyzed. Example microscopy image=around Pf in one animal (top), segmentations thereof by our model ensemble (bottom). \u003cstrong\u003eq.\u003c/strong\u003e Significantly higher cFos immunoreactive cell count in the bilateral VNC (H-statistic: 9.777, p=0.0075), gigantocellular nuclei (H-statistic: 10.75, p=0.0046, \u003cstrong\u003er.\u003c/strong\u003e), parafascicular nuclei (H-statistic: 6.095, p=0.0475, \u003cstrong\u003es.\u003c/strong\u003e) and zonae incertae (H-statistic: 8.011, p=0.0182, \u003cstrong\u003et.\u003c/strong\u003e) of ChRmine-A53T/EV vs. sham-stimulated mCherry mice (A53T: n=11; EV: n=9-10; mCherry: n=5-6 animals, Kruskal-Wallis tests with Dunn’s multiple comparisons correction; sum from both hemispheres). Lines are medians, boxes are interquartile ranges, whiskers are most extreme data points. Scale bars (a., d.-o.)=100 microns, scale bar of inset (f.)=20 microns, Scale bars (p.)=200 microns.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/ab85290555bc34264af94007.png"},{"id":75875680,"identity":"b554382d-21c4-4e92-a402-d17fe89500a6","added_by":"auto","created_at":"2025-02-10 07:52:25","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":293292,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eA53T-α-Synuclein overexpression induces a parkinsonian phenotype. a.\u003c/strong\u003e Example of substantia nigra of a A53T (top) and EV (bottom) animal at week 8. A53T mice show deposition of α-synuclein (aSyn\u003cem\u003e, \u003c/em\u003eyellow) in peri-nigral zone. Scale bars=500 microns. \u003cstrong\u003eb. \u003c/strong\u003eStereological estimation of tyrosine hydroxylase (TH)-positive neurons in the substantia nigra pars compacta (SNC) (one-way ANOVA, F=83,87, p\u0026lt;0.0001, Tukey’s multiple comparisons test; A53T-Week8: n=12, EV: n=10, non-behavioural cohort: A53T-Week4: n=5).\u003cem\u003e \u003c/em\u003eData shown as mean ± s.e.m. \u003cstrong\u003ec.\u003c/strong\u003e Example of immunoreactivity of TH, aSyn and DAPI in the SNC of an A53T (top) and EV (bottom) animal demonstrating deposition of aSyn in A53T. Scale bars=20 microns. \u003cstrong\u003ed.\u003c/strong\u003e TH\u003csup\u003e+\u003c/sup\u003e cells correlate with Nissl-stained cells [neuron count] (week 8; Spearman’s r=0.8453, p\u0026lt;0.0001; A53T: n=12, EV: n=10). \u003cstrong\u003ee.\u003c/strong\u003e D-amphetamine injection to verify nigrostriatal dopaminergic dysfunction. Simple logistic regression model discriminates A53T from EV mice (week 8 off, total rotations, A53T: n=12, EV: n=10; AUROC=0.9375, s.e.=0.05201, p=0.0005, Tjur’s R squared=0.6371). \u003cstrong\u003ef.\u003c/strong\u003e Kinematic analysis from Ethovision of top-view open field found a significant meander bias in A53T mice (week 8 off; two-tailed Mann-Whitney test, A53T: n=12, EV: n=7, p=0.0098). Dashed line is median, dotted lines are interquartile ranges. \u003cstrong\u003eg. \u003c/strong\u003eBias in meander significantly correlated with total off rotations from the amphetamine test (week 8; A53T: n=12, EV: n=7; Spearman’s r=0.6032, p=0.0063). \u003cstrong\u003eh. \u003c/strong\u003eAssessment of forepaw use in cylinder test showed canonically increased usage ipsilateral to injection in A53T \u003csup\u003e38\u003c/sup\u003e (A53T: n=12, EV: n=10; mixed-effects model with Tukey's multiple comparisons, time/stimulation\u003csup\u003ens\u003c/sup\u003e, group**, interaction\u003csup\u003ens\u003c/sup\u003e; week 8-off: adjusted p=0.0097). Lines are medians, boxes are interquartile ranges. Whiskers are most extreme data points. \u003cstrong\u003ei. \u003c/strong\u003eA53T mice show ipsilateral rotatory bias in cylinder test vs. EV (A53T: n=12, EV: n=10; mixed-effects model with Tukey’s multiple comparisons, time/stimulation****, group****, interaction****) at week 8 off (p\u0026lt;0.0001), perithreshold stimulation induced a mild reduction in asymmetry. Data shown as median ± 95% CI. Maximum velocity on a treadmill (see\u003cstrong\u003e Fig. 6\u003c/strong\u003e) correlates closely with both TH\u003csup\u003e+\u003c/sup\u003e cell count (\u003cstrong\u003ej.\u003c/strong\u003e, week 8-off; A53T: n=12, EV: n=10; Spearman’s r=0.6631, p=0.0008) and rotatory asymmetry (\u003cstrong\u003ek.\u003c/strong\u003e, cylinder test; week 8-off; A53T: n=12, EV: n=10; Spearman’s r=-0.6045, p=0.0029). Lines=linear intercepts, dotted lines=95% confidence intervals.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/892490f7db2f1fd90cc33f3f.png"},{"id":75875681,"identity":"919f9cb9-27f2-463f-8585-d6e256b1eb37","added_by":"auto","created_at":"2025-02-10 07:52:25","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":450338,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePerithreshold optoactivation of glutamatergic vestibular nucleus complex neurons governs prolocomotor changes in subsecond behavioural modules. a.\u003c/strong\u003e Behavioural dynamics via subsecond behavioural modules (termed syllables) by \u003cem\u003eKeypoint-MoSeq\u003c/em\u003e (KPMS) \u003csup\u003e52\u003c/sup\u003e from top-view open field (OF) pose derived within DeepLabCut (EV: n=10, A53T: n=14, mCherry: n=6).\u003cstrong\u003e b.\u003c/strong\u003e Our KPMS model identified distinct syllables, clearly mappable to different behaviours (\u003cstrong\u003eSupplementary Videos 11-33, 35\u003c/strong\u003e).\u003cstrong\u003e c. \u003c/strong\u003eSyllable dendrogram shows significant difference between syllables (distances between each syllable’s median trajectory), specifically running (15) and others. \u003cstrong\u003ed\u003c/strong\u003e. Heatmaps show transition (probability) differences comparing groups \u0026amp; conditions in matrices. Perithreshold VNC optoactivation leads to distinct, global changes in behavioural transition structure in both A53T and EV mice (Spectral procrustes analysis; both Bonferroni corr. p=0.00499), however not in sham-stimulated mCherry mice (Bonferroni corr. p=0.99999). Color bars indicate transition probabilities differences, grey=no change. X and Y-axis squares plot identical orders of syllables (0-16, 18, 19, 20, 21). Syllables 17+22 excluded due to low frequency (\u0026lt;0.005). \u003cstrong\u003ee-g. \u003c/strong\u003eNetwork plots demonstrate magnitude of usage (nodes) and transition (bars) differences between distinct syllables in A53T vs. EV vs. mCherry (sham stimulation) mice with perithreshold VNC actuation. Transitions between behavioural modules appear globally up-regulated (green) in A53T and EV vs. mCherry mice. Numbers=arbitrary scaling factors for visibility of changes (see \u003cstrong\u003eMethods\u003c/strong\u003e). Each node=one syllable in clockwise order (0-16, 18, 19, 20, 21). \u003cstrong\u003eh. \u003c/strong\u003ePerithreshold VNC optoactivation leads to kinematic changes within subsecond behavioural modules (all A53T: n=13, EV: n=10, mCherry: n=6 mice; clustered permutations Kruskal-Wallis testing with Dunn’s z-test for paired comparisons), specifically enhancing peak locomotor performance (running, mean velocity; Dunn’s z-test; A53T: p\u0026lt;0.01, EV: p\u0026lt;0.001) vs. sham (mCherry p\u0026gt;0.05).\u003cstrong\u003e i. \u003c/strong\u003eLocomotor capacity is strongly increased in EV, less so in A53T mice (running, duration; Dunn’s z-test; A53T: p\u0026lt;0.05, EV: p\u0026lt;0.0001). mCherry mice show no significant changes (p\u0026gt;0.05). Identical syllables excluded. Data shown as mean ± s.e.m. \u003cstrong\u003ej.\u003c/strong\u003e Regular OF kinematics confirm upregulation of locomotor capacity in EV (Tukey's multiple comparisons test, p\u0026lt;0.001), but not in A53T mice (p\u0026gt;0.05) through perithreshold VNC optoactivation (A53T: n=14, EV: n=10; mixed-effects model: time/stim****, groups\u003csup\u003ens\u003c/sup\u003e, interaction\u003csup\u003ens\u003c/sup\u003e). Both groups show reduced distance moved due to canonical OF habituation (p\u0026lt;0.001).\u003cstrong\u003e \u003c/strong\u003eSimilar reduction in locomotor capacity due to habituation with sham stimulation in mCherry mice (mCherry: n=6; paired t-test, t=4.456, p\u0026lt;0.01).\u003cstrong\u003e \u003c/strong\u003eMaximum locomotor velocity decreases in A53T mice vs. baseline (pre-symptomatic) (p\u0026lt;0.01) and can be increased with stimulation in both A53T (p\u0026lt;0.05) and EV mice (p\u0026lt;0.001) (A53T: n=14, EV: n=10; mixed-effects model: time/stim***, groups\u003csup\u003ens\u003c/sup\u003e, interaction***).\u003cstrong\u003e \u003c/strong\u003eLack of the latter effect in mCherry mice demonstrates specificity (mCherry: n=6; paired t-test, t=0.8052, p\u0026gt;0.05). Lines are medians, boxes are interquartile ranges, whiskers are most extreme data points.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/ec32922bb75c028c9e4b3305.png"},{"id":75876926,"identity":"047f2f15-ead1-4a53-a618-5afb9ab965f1","added_by":"auto","created_at":"2025-02-10 08:00:25","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":509876,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGlutamatergic vestibular nucleus complex neurons resynchronize gait oscillations in parkinsonian mice. a.\u003c/strong\u003e Sketch with treadmill setup. Exemplary predictions by DLC on dual-view videos. We analyzed gait videos (n=853; A53T: n=466, EV: n=387) for pose estimation at 4 different speeds (10, 15, 20, 25cm/s) \u003csup\u003e47,12\u003c/sup\u003e. \u003cstrong\u003eb.\u003c/strong\u003e Treadmill V\u003csub\u003emax\u003c/sub\u003e\u003csub\u003e\u003cstrong\u003e \u003c/strong\u003e\u003c/sub\u003eis significantly higher in EV vs. A53T in both off state and with perithreshold stimulation (A53T: n=12, EV: n=10; Two-way ANOVA: time/stim\u003csup\u003ens\u003c/sup\u003e, groups***, interaction*; Tukey’s multiple comparisons test, p\u0026lt;0.01). Heatmap shows group means. \u003cstrong\u003ec-d.\u003c/strong\u003e According to gait failure scoring system (fluidity of gait index, FGI, FGI1=failed, FGI2+3=no failure, see \u003cstrong\u003eMethods\u003c/strong\u003e), suprathreshold stimulation significantly impaired gait in both A53T (n=12; FGI1 vs. FGI2+3; Fisher’s exact test; p\u0026lt;0.0001) and \u003cstrong\u003ee.\u003c/strong\u003e EV (n=10; FGI1 vs. FGI2+3; Fisher’s exact test; p\u0026lt;0.0001) mice vs. the perithreshold paradigm. Data both show number of trials of all available animals per group and condition. \u003cstrong\u003ee.\u003c/strong\u003e Example scalograms (week 8) of keypoint-based gait analysis of the left ‘LateralFHandmid’ marker of a single A53T animal in off state and under perithreshold stimulation (\u003cstrong\u003ef.\u003c/strong\u003e) showing increased power in peak frequencies with the latter, especially at 20cm/s. Red lines indicate cone of influence for the wavelet analysis. \u003cstrong\u003eg-l. \u003c/strong\u003ePeriodograms demonstrate loss of frequency power in gait dynamics and retuning by perithreshold actuation in A53T across speeds (\u003cstrong\u003eg\u003c/strong\u003e), absent in EV mice (\u003cstrong\u003eh\u003c/strong\u003e), mainly driven by changes at 20 cm/s (\u003cstrong\u003ek\u003c/strong\u003e). Data shown as mean ± s.e.m. Plotted significances are Šídák's multiple comparisons tests of mixed-effects models. \u003cstrong\u003em-r.\u003c/strong\u003e Autocorrelation analysis of all matched limb markers demonstrate retuned gait synchronicity with perithreshold actuation of the VNC. Data shown as mean ± s.e.m. Plotted significance of Šídák's multiple comparisons test between peaks; between panels: time/stimulation term of mixed-effects models. \u003cstrong\u003es.\u003c/strong\u003e Peak frequency and amplitude for left distal limb marker in A53T mice shows a positive correlation at baseline, that is lost and subsequently retuned by perithreshold stimulation (F-test, F(2,138) = 4.15, p = 0.018) in across speeds analysis (10-25cm/s). \u003cstrong\u003et.\u003c/strong\u003e EV mice do not show these changes (F-test, F(2,113)=0.353, p = 0.703). \u003cstrong\u003eu.\u003c/strong\u003e Correlation of peak frequency and amplitude is not driven by isolated changes at 20cm/s (F-test, F(2,32)=0.478, p=0.624) in A53T mice. \u003cstrong\u003ev.\u003c/strong\u003e No changes in EV group for identical correlation at 20cm/s (F-test, F(2,24)=0.096, p = 0.909). Lines = linear intercepts.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/23b065f5708f56a8b4d7c052.png"},{"id":75935960,"identity":"9f1e0d33-2e1c-4a23-abde-8a65daabe243","added_by":"auto","created_at":"2025-02-10 17:05:53","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3630729,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/3e16cef9-6c07-4b26-83e9-d197dd35d957.pdf"},{"id":75875713,"identity":"b4eecfec-bb40-42bd-8dd4-50675dec12f6","added_by":"auto","created_at":"2025-02-10 07:52:26","extension":"mp4","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":55231625,"visible":true,"origin":"","legend":"Supplementary Video1: Titration framework","description":"","filename":"SupplementaryVideo1Scale.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/0f6421ebccc45d235898c132.mp4"},{"id":75875723,"identity":"b3f3ab04-a1f6-469c-af87-7c9900d16595","added_by":"auto","created_at":"2025-02-10 07:52:26","extension":"avi","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":76670840,"visible":true,"origin":"","legend":"Supplementary Video 2: Vestibular Symptom Scale (VSS) 1","description":"","filename":"SupplementaryVideo2Montage1388.avi","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/72da95b6cc08f48488bc6fe3.avi"},{"id":75875727,"identity":"fa509980-bb5d-42de-93a3-5d94fea4f96d","added_by":"auto","created_at":"2025-02-10 07:52:26","extension":"avi","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":77724970,"visible":true,"origin":"","legend":"Supplementary Video 3: Vestibular Symptom Scale (VSS) 2","description":"","filename":"SupplementaryVideo3Montage2438.avi","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/4023419c9bd4f88636a50d45.avi"},{"id":75876928,"identity":"3bb1eaab-30dd-44c9-a0e6-572999c10afe","added_by":"auto","created_at":"2025-02-10 08:00:25","extension":"mp4","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":2041871,"visible":true,"origin":"","legend":"Supplementary Video 4: Video for UMAP plot for Fig 1 e","description":"","filename":"SupplementaryVideo4VS0.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/11f2c267ff7d199dc334da94.mp4"},{"id":75880779,"identity":"fa3add65-fc6d-41ca-82de-ee13be0d7f82","added_by":"auto","created_at":"2025-02-10 08:24:25","extension":"mp4","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":2633142,"visible":true,"origin":"","legend":"Supplementary Video 5: Video for UMAP plot for Fig 1 e","description":"","filename":"SupplementaryVideo5VS1.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/c0a8f6c5353a01d0a99d978e.mp4"},{"id":75879762,"identity":"390b725e-7a5f-48ca-8fb6-13c65d096efc","added_by":"auto","created_at":"2025-02-10 08:16:25","extension":"mp4","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":2547328,"visible":true,"origin":"","legend":"Supplementary Video 6: Video for UMAP plot for Fig 1 e","description":"","filename":"SupplementaryVideo6VS2.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/2ef5f750cc5eb2440ffd0026.mp4"},{"id":75877864,"identity":"60953584-a7b4-4e8f-8809-dede123502a9","added_by":"auto","created_at":"2025-02-10 08:08:25","extension":"mp4","order_by":7,"title":"","display":"","copyAsset":false,"role":"supplement","size":2904487,"visible":true,"origin":"","legend":"Supplementary Video 7: Video for UMAP plot for Fig 1 e","description":"","filename":"SupplementaryVideo7VS3.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/d9ed7d3577f3a85c28871135.mp4"},{"id":75875689,"identity":"336aceee-af4c-4cfa-a070-a48aec5126f1","added_by":"auto","created_at":"2025-02-10 07:52:25","extension":"mp4","order_by":8,"title":"","display":"","copyAsset":false,"role":"supplement","size":3060380,"visible":true,"origin":"","legend":"Supplementary Video 8: Video for UMAP plot for Fig 1 e","description":"","filename":"SupplementaryVideo8VS4circling.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/e29028ae43faacf6ecec2a02.mp4"},{"id":75875686,"identity":"1754d0f2-e4c3-44aa-8379-ad46df11ba6d","added_by":"auto","created_at":"2025-02-10 07:52:25","extension":"mp4","order_by":9,"title":"","display":"","copyAsset":false,"role":"supplement","size":3870562,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 9: Video for UMAP plot for Fig 1 e\u003c/p\u003e","description":"","filename":"SupplementaryVideo9VS4circlingretrolocomotion.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/cf7a068bb08fbfaf03fe3a0f.mp4"},{"id":75877869,"identity":"d65889fd-d2a4-4f01-94a4-4d1969a5d3db","added_by":"auto","created_at":"2025-02-10 08:08:25","extension":"mp4","order_by":10,"title":"","display":"","copyAsset":false,"role":"supplement","size":3597630,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 10: Video for UMAP plot for Fig 1 e\u003c/p\u003e","description":"","filename":"SupplementaryVideo10VS4crossedextensorlike.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/89c91aaa7f23dd132c334fa6.mp4"},{"id":75876933,"identity":"2907d2db-db4e-413b-9703-9b0c1a935d9f","added_by":"auto","created_at":"2025-02-10 08:00:25","extension":"mp4","order_by":11,"title":"","display":"","copyAsset":false,"role":"supplement","size":3138800,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 11: Syllable video 0 for Fig 4 b\u003c/p\u003e","description":"","filename":"SupplementaryVideo11syllable0.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/85f211ed05f1cd9b4f45a66e.mp4"},{"id":75875703,"identity":"b279ff55-c405-4d2d-96d7-aa7faec6d736","added_by":"auto","created_at":"2025-02-10 07:52:25","extension":"mp4","order_by":12,"title":"","display":"","copyAsset":false,"role":"supplement","size":3092130,"visible":true,"origin":"","legend":"Supplementary Video 12: Syllable video 1 for Fig 4 b","description":"","filename":"SupplementaryVideo12syllable1.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/f55d0cb02b1bbb4c594c6b60.mp4"},{"id":75876941,"identity":"8d36829e-aa4a-4206-b24d-0b94c5c899f5","added_by":"auto","created_at":"2025-02-10 08:00:25","extension":"mp4","order_by":13,"title":"","display":"","copyAsset":false,"role":"supplement","size":2981274,"visible":true,"origin":"","legend":"Supplementary Video 13: Syllable video 2 for Fig 4 b","description":"","filename":"SupplementaryVideo13syllable2.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/7d119f75078735895455647e.mp4"},{"id":75875696,"identity":"bde3cac9-2ca3-45d0-8167-883f89a34bd1","added_by":"auto","created_at":"2025-02-10 07:52:25","extension":"mp4","order_by":14,"title":"","display":"","copyAsset":false,"role":"supplement","size":2912330,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 14: Syllable video 3 for Fig 4 b\u003c/p\u003e","description":"","filename":"SupplementaryVideo14syllable3.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/c52ee855c329fc46054d57b7.mp4"},{"id":75877883,"identity":"d05d006a-5029-418e-b275-82408af8e8a9","added_by":"auto","created_at":"2025-02-10 08:08:26","extension":"mp4","order_by":15,"title":"","display":"","copyAsset":false,"role":"supplement","size":3214922,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 15: Syllable video 4 for Fig 4 b\u003c/p\u003e","description":"","filename":"SupplementaryVideo15syllable4.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/cb272b4742b13390721483c1.mp4"},{"id":75875715,"identity":"6242e3f4-48a4-4351-b692-61ddeb24acf3","added_by":"auto","created_at":"2025-02-10 07:52:26","extension":"mp4","order_by":16,"title":"","display":"","copyAsset":false,"role":"supplement","size":2915659,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 16: Syllable video 5 for Fig 4 b\u003c/p\u003e","description":"","filename":"SupplementaryVideo16syllable5.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/6f44ece063d684bbb0744800.mp4"},{"id":75876935,"identity":"5d5246dc-a5d4-43aa-b806-9759297a6e59","added_by":"auto","created_at":"2025-02-10 08:00:25","extension":"mp4","order_by":17,"title":"","display":"","copyAsset":false,"role":"supplement","size":3096090,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 17: Syllable video 6 for Fig 4 b\u003c/p\u003e","description":"","filename":"SupplementaryVideo17syllable6.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/1f4dc42939547a62a869577d.mp4"},{"id":75877882,"identity":"414a5fed-e099-4339-b7a8-d3c3ee82e698","added_by":"auto","created_at":"2025-02-10 08:08:25","extension":"mp4","order_by":18,"title":"","display":"","copyAsset":false,"role":"supplement","size":3308924,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 18: Syllable video 7 for Fig 4 b\u003c/p\u003e","description":"","filename":"SupplementaryVideo18syllable7.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/c6a22118c8b4b86cb95c437b.mp4"},{"id":75875708,"identity":"99879bd5-7b76-4cb8-816f-bc7886ad318f","added_by":"auto","created_at":"2025-02-10 07:52:25","extension":"mp4","order_by":19,"title":"","display":"","copyAsset":false,"role":"supplement","size":2813958,"visible":true,"origin":"","legend":"Supplementary Video 19: Syllable video 8 for Fig 4 b","description":"","filename":"SupplementaryVideo19syllable8.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/1606ca4c9cb0842da79484a3.mp4"},{"id":75875704,"identity":"3e4e0eb1-e119-4aef-8967-9f695d121c2e","added_by":"auto","created_at":"2025-02-10 07:52:25","extension":"mp4","order_by":20,"title":"","display":"","copyAsset":false,"role":"supplement","size":3237422,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 20: Syllable video 9 for Fig 4 b\u003c/p\u003e","description":"","filename":"SupplementaryVideo20syllable9.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/7b16125d2fd4b71d99f2403e.mp4"},{"id":75876939,"identity":"f6cab2b8-6458-4119-b070-9e524210311d","added_by":"auto","created_at":"2025-02-10 08:00:25","extension":"mp4","order_by":21,"title":"","display":"","copyAsset":false,"role":"supplement","size":2455005,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 21: Syllable video 10 for Fig 4 b\u003c/p\u003e","description":"","filename":"SupplementaryVideo21syllable10.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/445a8b70cd7eb35211bc4086.mp4"},{"id":75875705,"identity":"59aa10ea-1b16-4447-bff1-c1500806531e","added_by":"auto","created_at":"2025-02-10 07:52:25","extension":"mp4","order_by":22,"title":"","display":"","copyAsset":false,"role":"supplement","size":3283164,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 22: Syllable video 11 for Fig 4 b\u003c/p\u003e","description":"","filename":"SupplementaryVideo22syllable11.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/983e74e461d1fefe6ae718a7.mp4"},{"id":75875721,"identity":"7b039e6a-d43c-4e43-a2ac-a89a7ede6745","added_by":"auto","created_at":"2025-02-10 07:52:26","extension":"mp4","order_by":23,"title":"","display":"","copyAsset":false,"role":"supplement","size":3260473,"visible":true,"origin":"","legend":"Supplementary Video 23: Syllable video 12 for Fig 4 b","description":"","filename":"SupplementaryVideo23syllable12.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/9ee1bf81b6c2ce67851474bf.mp4"},{"id":75875701,"identity":"ff9fb610-851f-411c-bd56-b23df6aa2762","added_by":"auto","created_at":"2025-02-10 07:52:25","extension":"mp4","order_by":24,"title":"","display":"","copyAsset":false,"role":"supplement","size":2988985,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 33: Syllable video 22 for Fig 4 b\u003c/p\u003e","description":"","filename":"SupplementaryVideo24syllable13.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/413942fd02addf050673fe84.mp4"},{"id":75875706,"identity":"e69c6f89-fa1f-4244-8488-5cb108c2675a","added_by":"auto","created_at":"2025-02-10 07:52:25","extension":"mp4","order_by":25,"title":"","display":"","copyAsset":false,"role":"supplement","size":3088191,"visible":true,"origin":"","legend":"Supplementary Video 25: Syllable video 14 for Fig 4 b","description":"","filename":"SupplementaryVideo25syllable14.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/f8b555badb3682528d14590a.mp4"},{"id":75875712,"identity":"705d56ca-4437-48f4-81f7-6df7ff021cdc","added_by":"auto","created_at":"2025-02-10 07:52:25","extension":"mp4","order_by":26,"title":"","display":"","copyAsset":false,"role":"supplement","size":2853714,"visible":true,"origin":"","legend":"Supplementary Video 26: Syllable video 15 for Fig 4 b","description":"","filename":"SupplementaryVideo26syllable15.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/c13ec5c7dc2434679eb6e0e7.mp4"},{"id":75875719,"identity":"bde9b852-c632-4967-a92f-7e0e0ec2fa69","added_by":"auto","created_at":"2025-02-10 07:52:26","extension":"mp4","order_by":27,"title":"","display":"","copyAsset":false,"role":"supplement","size":2623414,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 24: Syllable video 13 for Fig 4 b\u003c/p\u003e","description":"","filename":"SupplementaryVideo27syllable16.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/d5c43b47bbe0765b199bcf50.mp4"},{"id":75876946,"identity":"dd920114-ece8-4e3f-8a8d-74b4c6befecb","added_by":"auto","created_at":"2025-02-10 08:00:26","extension":"mp4","order_by":28,"title":"","display":"","copyAsset":false,"role":"supplement","size":3080268,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 28: Syllable video 17 for Fig 4 b\u003c/p\u003e","description":"","filename":"SupplementaryVideo28syllable17.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/c52a4076a2f9a095654a12b5.mp4"},{"id":75875709,"identity":"81c8b9ca-7c54-4349-af68-7743193e8e68","added_by":"auto","created_at":"2025-02-10 07:52:25","extension":"mp4","order_by":29,"title":"","display":"","copyAsset":false,"role":"supplement","size":2746966,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 27: Syllable video 16 for Fig 4 b\u003c/p\u003e","description":"","filename":"SupplementaryVideo29syllable18.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/8d720a6d327535b8682ec3cb.mp4"},{"id":75875718,"identity":"bd27c67f-f254-42a0-8574-5adc495c6e10","added_by":"auto","created_at":"2025-02-10 07:52:26","extension":"mp4","order_by":30,"title":"","display":"","copyAsset":false,"role":"supplement","size":3087255,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 29: Syllable video 18 for Fig 4 b\u003c/p\u003e","description":"","filename":"SupplementaryVideo30syllable19.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/557e5845a5166dd54a5796a5.mp4"},{"id":75877878,"identity":"7227d366-b673-4977-adc8-4b7fb20a7af5","added_by":"auto","created_at":"2025-02-10 08:08:25","extension":"mp4","order_by":31,"title":"","display":"","copyAsset":false,"role":"supplement","size":2831420,"visible":true,"origin":"","legend":"Supplementary Video 31: Syllable video 20 for Fig 4 b","description":"","filename":"SupplementaryVideo31syllable20.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/6d7e7f3d48c1d401f6dcbcff.mp4"},{"id":75876947,"identity":"75214b59-630b-4283-b301-391bc45e2ab2","added_by":"auto","created_at":"2025-02-10 08:00:26","extension":"mp4","order_by":32,"title":"","display":"","copyAsset":false,"role":"supplement","size":3144618,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 30: Syllable video 19 for Fig 4 b\u003c/p\u003e","description":"","filename":"SupplementaryVideo32syllable21.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/595a436dfb71d47434900dae.mp4"},{"id":75877888,"identity":"4bf7a9db-8524-4e8e-abf0-0f689d34e696","added_by":"auto","created_at":"2025-02-10 08:08:26","extension":"mp4","order_by":33,"title":"","display":"","copyAsset":false,"role":"supplement","size":3065255,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 32: Syllable video 21 for Fig 4 b\u003c/p\u003e","description":"","filename":"SupplementaryVideo33syllable22.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/603b4e28d89a5d7fa1b60ba5.mp4"},{"id":75876942,"identity":"df60e100-7713-4679-bd09-b700a312e45f","added_by":"auto","created_at":"2025-02-10 08:00:25","extension":"mp4","order_by":34,"title":"","display":"","copyAsset":false,"role":"supplement","size":3637202,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 34\u003c/p\u003e","description":"","filename":"SupplementaryVideo34SuprathresholdStimDual20RS25cms0degUP.mp4","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/e250acb8d84502e6b5b03246.mp4"},{"id":75876944,"identity":"613ad12b-0d04-4dbc-a9e1-038d0ff10fda","added_by":"auto","created_at":"2025-02-10 08:00:26","extension":"gif","order_by":35,"title":"","display":"","copyAsset":false,"role":"supplement","size":899960,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video 35:\u003c/p\u003e","description":"","filename":"SupplementaryVideo35KPMSalltrajectories.gif","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/a2b4ffcc4739e2c88e3ec414.gif"},{"id":75875724,"identity":"b63a2f8b-e2bb-4970-86bf-17a9b03f4af5","added_by":"auto","created_at":"2025-02-10 07:52:26","extension":"docx","order_by":36,"title":"","display":"","copyAsset":false,"role":"supplement","size":3449397,"visible":true,"origin":"","legend":"","description":"","filename":"ExtendedDataandSupplementoryFigures.docx","url":"https://assets-eu.researchsquare.com/files/rs-5851215/v1/fc8ebc62842f28f4b536fba6.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Vestibular circuit stimulation for retuning locomotor dynamics in Parkinson's disease","fulltext":[{"header":"Introduction","content":"\u003cp\u003eParkinson's disease (PD) represents a significant and escalating public health challenge \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e, marked by dysfunction in dopaminergic circuits \u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. Increasingly recognized as a neurodegenerative spectrum disorder with a diverse phenotypical combination of motor and non-motor features \u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e, the motor aspects, such as bradykinesia, rigidity, and tremor, typically respond well to dopaminergic medication and deep brain stimulation (DBS) targeting the subthalamic nucleus (STN) or globus pallidus internus. However, axial motor symptoms, encompassing dysfunctional gait (e.g., freezing of gait) and postural instability, exhibit poor responsiveness \u003csup\u003e\u003cspan additionalcitationids=\"CR5 CR6\" citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e and persist as a clinical concern. Occasionally classified as the PIGD subtype \u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e (postural instability and gait difficulty), axial phenotypes remain both insufficiently understood and inadequately addressed by current therapeutic interventions \u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. Thus, there is an urgent need for innovative strategies to address treatment-resistant parkinsonian gait and posture disorders. Previous approaches towards improving axial PD symptoms included postulations about the involvement of basal ganglia-brainstem interactions \u003csup\u003e\u003cspan additionalcitationids=\"CR10 CR11\" citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e (\u003cb\u003esee Supplementary Fig.\u0026nbsp;1\u003c/b\u003e for graphical summary). Specifically, neuromodulation strategies for PIGD have led to trials of DBS targeting brainstem mesencephalic locomotor region (MLR) nuclei \u0026ndash; pedunculopontine and cuneiform \u0026ndash; based on their implication in circuits for gait and posture \u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. However, those studies have yielded inconsistent results \u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e,\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e, likely reflecting the hitherto poorly understood, complex nature of gait and posture (disorders) in PD \u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. Recognizing the advantages of non-invasive neuromodulation techniques in terms of ease-of-use and lower complication rates, there is a growing interest in exploring these approaches for different (patho-)physiological states \u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e,\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e, including PD. However, a prerequisite for novel PIGD neuromodulation approaches is a better mechanistic understanding of the circuitry underlying predominantly axial phenotypes.\u003c/p\u003e \u003cp\u003eVestibular stimulation is a non-invasive method that utilizes, e.g. noisy (i.e. thresholded), galvanic (nGVS) or caloric means to activate (central) vestibular circuits via peripheral excitation of the vestibular organ \u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e. This technique has shown promise in mitigating motor and non-motor symptoms of PD \u003csup\u003e\u003cspan additionalcitationids=\"CR20 CR21\" citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e, potentially via integrating vestibular information for movement into a vestibulo-thalamo-basal ganglia loop \u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e,\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e. Notably, there are some data supporting a positive impact on axial postural deficits, in both PD \u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e,\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e and atypical parkinsonism \u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e, yet less so for gait in PD \u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e. Further, nGVS has improved parkinsonian symptoms in rats \u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e. Vestibular circuits likely exhibit glutamatergic predominance \u003csup\u003e\u003cspan additionalcitationids=\"CR30\" citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e, especially with regards to motor targets \u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e, and while anatomical studies suggest possible circuit mechanisms involving vestibulo-thalamic inputs to basal ganglia loops \u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e,\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e,\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e, the structural and functional \u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e contributions of vestibular circuits for movement, specifically in PD \u003csup\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e, remain insufficiently understood.\u003c/p\u003e \u003cp\u003eOur study thus aimed to better understand the feed of the vestibular nucleus complex (VNC) into basal ganglia loops and contributions to motor dimensions, specifically in parkinsonian mice. First, we wanted to further virally trace and functionally map this association. Second, based on the hypothesis that thresholded human nGVS feeds prolocomotor information into basal ganglia loops, we aimed to check for therapeutic effects of a thresholded optical activation of the VNC.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eTitrated optoactivation of glutamatergic vestibular nucleus complex neurons yields a diverse posture and locomotor symptom continuum\u003c/p\u003e\n\u003cp\u003eTo functionally interrogate vestibular circuits governing postural and locomotor control, we used optogenetics by introducing the photosensitive optical actuator \u003cem\u003eChRmine\u0026nbsp;\u003c/em\u003e\u003csup\u003e37\u003c/sup\u003e into the vestibular nucleus complex (VNC) of \u003cem\u003eVglut2\u0026nbsp;\u003c/em\u003e(vesicular glutamate transporter 2)-\u003cem\u003eires-cre\u003c/em\u003e mice (\u003cstrong\u003eFig. 1a\u003c/strong\u003e). We rendered a fraction of these mice parkinsonian by overexpression of AAV-delivered human A53T-mutated \u0026alpha;-synuclein in the substantia nigra pars compacta (SNc) and maintained some as control mice injected with empty vector (EV) virus. These are henceforth referred to as A53T and EV, respectively (\u003cstrong\u003eFig. 1a\u003c/strong\u003e) \u003csup\u003e38\u003c/sup\u003e. To avoid confounding by group-differential opsin uptake, we estimated and compared transfected VNC neuron count by assessing mScarlet-positive cells, labelling the viral construct, in A53T vs. EV mice. The former showed no significant difference (\u003cstrong\u003eFig. 1a\u003c/strong\u003e, week 8; A53T: n=12, EV: n=10 mice). A sham optical control group was injected with a virus lacking an opsin (\u003cstrong\u003eFig. 1b\u003c/strong\u003e).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn optogenetic pilot experiments, we found that incremental laser power yielded differential posture and locomotor symptoms with VNC activation in both, A53T and EV mice. To comprehensively assess the effects of increasing levels of optical VNC actuation and generate a therapeutic stimulation regime, we devised two optogenetic approaches: the first consisted of a titration paradigm that allowed us to probe dose-response effects on posture and locomotor control by using incremental laser power thresholds (\u003cstrong\u003eFig. 1c\u003c/strong\u003e). To capture these disruptive stimulation effects systematically, we used a \u003cem\u003eVestibular Symptom Scale (VSS)\u003c/em\u003e [see \u003csup\u003e31\u003c/sup\u003e for a similar scale] (\u003cstrong\u003eFig. 1d\u003c/strong\u003e), based on the following findings from pilot experiments: frequently, the first optically-induced aberrant behaviour to appear was tilting of the head, most notably in the yaw and roll plane (=\u003cem\u003eVSS 1\u003c/em\u003e). Additionally, some animals demonstrated head bobbing under low stimulation intensity. With increasing laser intensity, body turning, as well as crossed extensor-like \u003csup\u003e39\u003c/sup\u003e (both=\u003cem\u003eVSS 2\u003c/em\u003e) and circling movements (=\u003cem\u003eVSS 3\u003c/em\u003e) emerged. Finally, at high laser intensities, retropulsion or retro-locomotion (=\u003cem\u003eVSS 4\u003c/em\u003e) appeared (see\u003cstrong\u003e\u0026nbsp;Supplementary Video 1, 2, 3\u003c/strong\u003e). The titration experiments confirmed dose-dependent effects of VNC optoactivation on postural and locomotor control in both EV and parkinsonian A53T mice, but not sham mice.\u003c/p\u003e\n\u003cp\u003eSpecifically, we found that while VSS 2 symptoms (body tilting, crossed extensor-like responses) were observed in ~56% of animals at week 8 (human rater), more pronounced phenotypes (VSS 3 + 4) and VSS 1 (head bobbing and tilting) symptoms were observed less consistently. Since these were subjective rater-based observations, we next aimed to check whether a machine learning approach based on pose data would identify similar subtypes of vestibular symptoms. We made use of top-view open field (OF)-based pose estimation via DeepLabCut (DLC) \u003csup\u003e40,41\u003c/sup\u003e from systematic titration experiments (data from n=24 mice). These encompassed the whole spectrum of rater-observed induced vestibular symptoms, but also spontaneous behaviour, which we categorically annotated as \u0026lsquo;VSS-like\u0026rsquo; and \u0026lsquo;naturalistic behaviour\u0026rsquo;, respectively. We then leveraged a semi-supervised machine learning pipeline (A-SoiD \u003csup\u003e42\u003c/sup\u003e) to train an active learning classifier (week 4 titration data), which performed well in detecting both behavioural categories (average performance: ~90%) (\u003cstrong\u003esee Supplementary Figure 2a-c\u0026nbsp;\u003c/strong\u003efor details on workflow). We applied this active learning classifier to unseen titration data (week 8 titration data,\u0026nbsp;n=22 videos,\u0026nbsp;A53T: n=12, EV: n=10) and found that VSS symptoms were matched with those of a human rater (\u003cstrong\u003eSupplementary Figure 2d\u003c/strong\u003e). Building upon that, we used clustering and embedding algorithms (B-SoiD \u003csup\u003e36,42\u003c/sup\u003e) inside A-SoiD to try and isolate distinct VSS-like behaviours (n=261 snippet videos, week 4 titrations) and found 5 significantly different clusters (\u003cstrong\u003eFig. 1e\u003c/strong\u003e), while naturalistic behaviour yielded 10 distinct clusters (\u003cstrong\u003eFig. 1f\u003c/strong\u003e). Based on clustering results, we in turn trained a novel, second active learning classifier on the resulting classes inside A-SoiD (\u003cstrong\u003eSupplementary Figure 2e\u003c/strong\u003e; average performance: ~90%), which revealed similar, however not identical subtypes as the VSS: Specifically, \u003cem\u003ecluster 0\u003c/em\u003e mapped to head bobbing (VSS 1) and \u003cem\u003ecluster 4\u003c/em\u003e to head \u0026amp; body tilting, circling and retrograde motion behaviours (VSS 2-4) (see \u003cstrong\u003eSupplementary Videos 4-10\u0026nbsp;\u003c/strong\u003efor classified subtypes). The major clusters in naturalistic behaviour mapped to idle exploration behaviour (\u003cem\u003ecluster 9\u003c/em\u003e) vs. locomotor exploration (\u003cem\u003ecluster 7\u003c/em\u003e), thus verifying overall sensible, yet highly stringent behavioural clustering and embedding. These results demonstrate that similar vestibular symptom subtypes are identified across subjective observation (VSS 1-4 scale) and automated analyses via pose estimation.\u003c/p\u003e\n\u003cp\u003eWe next asked whether the thresholds for vestibular symptoms dynamically changed or correlated with mScarlet-positive VNC cell count for any VSS class. We found that only for the high end of the symptom scale (VSS 4), laser power thresholds correlated with the number of mScarlet-positive cells (\u003cstrong\u003eFig. 1g-j\u003c/strong\u003e, all available mice from optogenetics cohort; A53T: n=10, EV: n=9). Nonetheless, all VSS thresholds were significantly lower at week 8 vs. baseline (\u003cstrong\u003eFig. 1k\u003c/strong\u003e, all available mice from optogenetics cohort; A53T: n=10, EV: n=9).\u003c/p\u003e\n\u003cp\u003eOnce we unraveled the strong symptomatic effects of VNC optoactivation on posture and locomotion, we blueprinted a second, therapeutic regime for optical neuromodulation: A \u003cem\u003ePerithreshold Vestibular Optomodulation\u003c/em\u003e regime (\u003cstrong\u003eFig. 1l\u003c/strong\u003e). With the latter, we aimed to mimic the perceptually thresholded electrical vestibular neuromodulation by nGVS \u003csup\u003e19,25,22,26\u003c/sup\u003e. We calibrated our optogenetic activation at the threshold of vestibular symptoms to appear and stimulated just below threshold, in order to achieve central activation of vestibular networks \u003csup\u003e24\u003c/sup\u003e, akin to GVS \u003csup\u003e43\u003c/sup\u003e. Yet, repeated stimulation may idiosyncratically change thresholds or responsiveness to stimulation during different (on-going) behaviours. We thus termed this \u0026lsquo;perithreshold stimulation\u0026rsquo; instead of \u0026lsquo;subthreshold stimulation\u0026rsquo;.\u003c/p\u003e\n\u003cp\u003eTaken together, our data show how titrated, optical activation of the VNC yields a diverse continuum of posture and locomotor symptoms in mice. These findings enabled us to develop a therapeutic optogenetic stimulation regime.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;Glutamatergic vestibular circuits connect to striatum, (sub)thalamus and caudal medulla, interlinking basal ganglia-brainstem motor hubs\u003c/p\u003e\n\u003cp\u003eHaving identified vestibular (sub-)phenotypes both by observation and from pose estimation data in titration experiments, we next focused on the underlying circuit elements and aimed to further unravel how VNC is embedded within pathways for movement, specifically those interlinking brainstem and basal ganglia\u0026nbsp;\u003csup\u003e9\u003c/sup\u003e. For structural mapping of glutamatergic VNC connectivity, we took a myc-tagged AAV-delivered anterograde synaptophysin tracer injected into the VNC. To delineate overlaps with motor ensembles implicated in neuromodulation and motor rescue in PD rodents\u0026nbsp;\u003csup\u003e45,46\u003c/sup\u003e, and due to evidence on a vestibulo-thalamo-striatal tract\u0026nbsp;\u003csup\u003e23\u003c/sup\u003e, we complemented the anterograde tracing approach with retrograde tracer injections (\u003cem\u003eretroAAV-hSyn-DIO-mCherry\u003c/em\u003e) into the STN (n=3) (\u003cstrong\u003eFig. 2a\u003c/strong\u003e), or into the dorsolateral striatum (dlSTR; n=4). Additionally, we evaluated \u003cem\u003emScarlet\u003c/em\u003e projection patterns from A53T and EV animals.\u003c/p\u003e\n\u003cp\u003eFocusing on subcortical regions related to basal ganglia-brainstem interactions associated with posture and locomotion\u0026nbsp;\u003csup\u003e9\u003c/sup\u003e, we found that the VNC maintains an extensive glutamatergic connectome (\u003cstrong\u003eFig. 2b,c\u003c/strong\u003e), projecting to important sensorimotor hubs in the subthalamus (\u003cstrong\u003eFig. 2d,e\u003c/strong\u003e, zona incerta), thalamus (\u003cstrong\u003eFig. 2f-k\u003c/strong\u003e, parafascicular-centromedian nucleus) and brainstem (\u003cstrong\u003eFig. 2l\u003c/strong\u003e, caudal medulla reticular formation). Synaptic punctae appeared in the parafasciular-centromedian (Pf-CM) nucleus complex and posterior oral (Po) nucleus (\u003cstrong\u003eFig. 2f-i\u003c/strong\u003e), partially overlapping with thalamostriatal neurons projecting to the dlSTR, corroborating findings on a vestibulo-thalamo-striatal pathway in rodents\u0026nbsp;\u003csup\u003e23\u003c/sup\u003e via the Pf-CM complex. Further, we found evidence of a strong vestibular pathway to the zona incerta (ZI) in the subthalamus yet observed no significant projection to the STN (\u003cstrong\u003eFig. 2d,e\u003c/strong\u003e). Contrary to our initial hypotheses, we found no consistent evidence for a strong projection to the mesencephalic locomotor region (MLR), either (\u003cstrong\u003eFig. 2m,n\u003c/strong\u003e), which we postulated to regulate gait adjustments. While some mScarlet fibres and isolated synapses reached the contralateral PPN, no consistent synaptic punctae were found in the MLR (either PPN or CnF). However, we found that the VNC directly targets the reticular formation in the caudal medulla, namely the gigantocellular (Gi) and sparsely the lateral paragigantocellular nucleus (LPGi) and anterior gigantocellular nucleus (GiA) (\u003cstrong\u003eFig. 2l\u003c/strong\u003e), thus potentially leveraging \u0026lsquo;hyperdirect\u0026rsquo; influence on reticulospinal effector pathways\u0026nbsp;\u003csup\u003e47,12\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eAfter we anatomically identified VNC targets within canonical motor circuitry, we aimed to verify functional relevance of VNC perithreshold stimulation in driving activity in those target areas. We used changes in c\u003cem\u003efos\u003c/em\u003e expression, an immediate early gene (IEG), as a surrogate for recent neuronal activity\u0026nbsp;\u003csup\u003e48,49\u003c/sup\u003e, which we quantified using the \u003cem\u003edeepflash2\u003c/em\u003e pipeline (\u003cstrong\u003eFig. 2p\u003c/strong\u003e). Therein, we focused on subcortical brain regions heavily targeted by the VNC (\u003cstrong\u003eFig. 2d-o\u003c/strong\u003e), namely the ZI, the Pf, Gi and finally the VNC itself. Further, we aimed to confirm the effect of optical actuation vs. sham optogenetics by comparing A53T \u0026amp; EV with mCherry mice. Perithreshold vestibular optomodulation in an OF context resulted in increased bilateral \u003cem\u003ecfos\u003c/em\u003e expression in ZI, Pf, Gi, and VNC in the A53T and EV group compared to mCherry-injected control animals, lacking a functional opsin (\u003cstrong\u003eFig. 2q-t\u003c/strong\u003e; A53T: n=12, EV: n=9-10, mCherry: n=5-6).\u003c/p\u003e\n\u003cp\u003eOverall, these structural (viral tracing) and functional (cFos) mapping data show that the VNC links both striato-thalamo-subthalamic and motor circuits in the reticular formation of the caudal medulla, consistent with the notion that the VNC serves as a basal ganglia-brainstem interface\u0026nbsp;\u003csup\u003e9\u003c/sup\u003e in regulating axial motor functions.\u003c/p\u003e\n\u003cp\u003eHuman A53T-mutated \u0026alpha;-synuclein overexpression in the substantia nigra leads to dopaminergic neurodegeneration and parkinsonian-like behaviour\u003c/p\u003e\n\u003cp\u003eBased on structural and functional observations of VNC interactions with key motor brain areas, we next addressed potential kinematic and behavioural changes by thresholded actuation of \u003cem\u003eVglut2\u003c/em\u003e-positive VNC neurons to probe putative therapeutic action in parkinsonian A53T mice.\u003c/p\u003e\n\u003cp\u003eTo that end, we first confirmed disease model effects in parkinsonian A53T mice via stereological, immunohistochemical and behavioural analyses. Previous studies used injection of AAV1/2-A53T (here: conc. 15x10\u003csup\u003e12\u0026nbsp;\u003c/sup\u003egenomic particles/ml; 500 nanoliters (nl)) into the SN of mice to induce dopaminergic cell death via overexpression of \u0026alpha;-synuclein (\u0026alpha;-Syn)\u0026nbsp;\u003csup\u003e38\u003c/sup\u003e, thereby establishing a model for parkinsonian \u0026alpha;-synucleinopathy. We substantiated \u0026alpha;-Syn deposition in the ipsilateral SN in A53T mice qualitatively by verifying immunoreactivity (\u003cstrong\u003eFig. 3a,c\u003c/strong\u003e). Stereological analysis yielded significantly reduced tyrosine hydroxlase (TH)-positive cells in the SN in A53T vs. EV mice (week 8;\u0026nbsp;A53T: n=12, EV: n=10) indicating dopaminergic cell death. For sound downstream pooling of behavioural readouts, we validated dopaminergic neurodegeneration at week 4 in an independent cohort of A53T mice (n=5) (\u003cstrong\u003eFig. 3b\u003c/strong\u003e). \u0026nbsp;Confirming these cells as actual neurons, we found a strong correlation of TH-positive cell count with Nissl-stained cells (\u003cstrong\u003eFig. 3d\u003c/strong\u003e). Having checked model effects at both time points histologically, we confirmed and extended behavioural effects in the cylinder and amphetamine test, as well as OF kinematics\u0026nbsp;\u003csup\u003e38,50\u003c/sup\u003e. As a surrogate of nigrostriatal dopaminergic dysfunction\u0026nbsp;\u003csup\u003e51\u003c/sup\u003e, we measured total amphetamine-induced rotations (without optogenetic manipulation). These strongly discriminated A53T from EV mice (total rotations, week 8 off; AUROC=0.9375;\u0026nbsp;A53T: n=12, EV: n=10), verifying effects on nigrostriatal dopaminergic output pathways in diseased mice (\u003cstrong\u003eFig. 3e\u003c/strong\u003e). Our findings were supported by additional evidence from the cylinder test and top-view OF. In the latter, A53T mice demonstrated a bias in meander behaviour, defined as quotient of turn angle per distance moved, thus displaying changes in the steering of directionality (\u003cstrong\u003eFig. 3f\u003c/strong\u003e, week 8 off; A53T: n=12, EV: n=7). Mean meander significantly correlated with amphetamine-induced rotations (\u003cstrong\u003eFig. 3g\u003c/strong\u003e), suggesting association of both effects with nigrostriatal dysfunction. In the cylinder test, we confirmed increased forepaw use ipsilateral to the injection site (\u003cstrong\u003eFig. 3h\u003c/strong\u003e) in A53T vs. EV mice (\u003cstrong\u003eFig. 3h\u003c/strong\u003e, week 8 off; A53T: n=12 , EV: n=10)\u0026nbsp;\u003csup\u003e38\u003c/sup\u003e. Beyond these previous results, we found that A53T mice displayed a strong, ipsilesional rotatory bias in the cylinder test setting (\u003cstrong\u003eFig. 3i\u003c/strong\u003e),\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003ereflected by a skewed rotatory symmetry index vs. EV mice (week 8 off; A53T: n=12, EV: n=10). Of note, this (static) rotatory asymmetry did at most show a mild reduction with perithreshold actuation of the VNC. Lastly, the maximum velocity (V\u003csub\u003emax\u003c/sub\u003e) on a treadmill correlated significantly with both dopaminergic neurodegeneration (\u003cstrong\u003eFig. 3j\u003c/strong\u003e) and rotatory asymmetry in the cylinder test (\u003cstrong\u003eFig. 3k\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003ePut together, these data confirm the development of a cellular and behavioural parkinsonian phenotype through virally induced \u0026alpha;-synucleinopathy in the substantia nigra.\u003c/p\u003e\n\u003cp\u003eOptoactivation of glutamatergic vestibular nucleus complex neurons allows retuning of locomotor performance but not capacity in parkinsonian mice\u003c/p\u003e\n\u003cp\u003eWith confirmation of a parkinsonian phenotype in A53T mice, we next took an in-depth look at changes on two hierarchical levels. Using top-view pose estimation, we first assessed higher-order behavioural patterns, followed by analysis of more fine-grained kinematic motor aberrations. We subsequently aimed to interrogate how both levels were altered by perithreshold optoactivation of glutamatergic VNC circuitry and whether therapeutic changes occur in parkinsonian mice.\u003c/p\u003e\n\u003cp\u003eTo this end, we extensively retrained the recurrent convolutional neural network (RCNN) \u0026lsquo;\u003cem\u003esuperanimal_topviewmouse\u003c/em\u003e\u0026rsquo; from the DLC model zoo\u0026nbsp;\u003csup\u003e40,41,53\u003c/sup\u003e, thereby achieving high tracking accuracy (mean average Euclidean error (MAE) on test data=2.04 pixels) (\u003cstrong\u003eFig. 4a\u003c/strong\u003e) for top-view pose on our own data. Utilizing a large amount of diverse pose estimation data from the top-view OF (n=193 videos), we included data from sham-stimulated mCherry (n=6), EV (n=10) and parkinsonian A53T mice (n=14) from different time points and employed an unbiased, unsupervised machine learning approach via \u003cem\u003eKeypoint-MoSeq\u003c/em\u003e \u003csup\u003e52\u003c/sup\u003e to identify distinct subsecond behavioural modules in pose dynamics, which were shown to be highly relevant to explaining mouse behaviour\u0026nbsp;\u003csup\u003e54\u003c/sup\u003e. We found multiple principal behavioural \u0026lsquo;syllables\u0026rsquo; (i.e. modules) (n=23, \u003cstrong\u003eFig. 4b\u003c/strong\u003e), that clearly mapped to naturally occurring behavioural motifs of mice, such as e.g. running (syllable 15; \u003cstrong\u003eSupplementary Videos 11-33\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e\u0026amp; Supplementary Figure 3\u003c/strong\u003e). Amongst this set of syllables, running was significantly different from most other syllables with regards to trajectory (\u003cstrong\u003eFig. 4c\u003c/strong\u003e). Further, syllables 6 and 11 were both identified as running behaviours associated with prior head movements and rearing, respectively. We aggregated all three running-related syllables to define peak locomotor performance.\u003c/p\u003e\n\u003cp\u003eWe first aimed to understand transitions between behavioural modules regarding differences amongst groups and with perithreshold optoactivation. To get a quantitative assessment of this effect, we analyzed the transition matrices of A53T, EV and sham-stimulated mCherry mice, specifically probability differences. Via a matrix eigenvector space analysis, we found that optoactivation of VNC neurons leads to significant changes in both A53T and EV, but not in sham-stimulated mCherry mice (\u003cstrong\u003eFig. 4d\u003c/strong\u003e). This underscores global changes in behavioural transitions induced by optoactivation of glutamatergic VNC neurons. In addition, A53T vs. EV group behavioural transition matrices in the absence of optical stimulation were significantly different (p=0.014), though not surviving correction for multiple comparisons. These results suggest that SNC \u0026alpha;-synucleinopathy induces mild alterations in behavioural transition dynamics.\u003c/p\u003e\n\u003cp\u003eWe then created network plots to gain a visual representation of how the overall usage of (and transitions between) distinct behavioural syllables changes with perithreshold VNC actuation. The overall structure of up (green)- and downregulation (violet) shows that optoactivation in A53T and EV groups (\u003cstrong\u003eFig. 4e-f\u003c/strong\u003e)\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eleads to globally up-regulated (green) behavioural transitions vs. sham-stimulated mCherry mice (\u003cstrong\u003eFig. 4g\u003c/strong\u003e), indicating enhanced pose shifting and enhanced modularity in pose with VNC activation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThese behavioural changes seen with a comprehensive birds-eye view on a higher hierarchical level motivated us to investigate what thresholded actuation of glutamatergic VNC neurons would do to distinctly identified behavioural modules (i.e. syllables) themselves. Analysing syllable kinematics, we found that perithreshold optical actuation of \u003cem\u003eVglut2\u003csup\u003e+\u003c/sup\u003e\u003c/em\u003e VNC neurons lead to an enhancement of peak locomotor performance, as reflected by increased mean velocity (\u003cstrong\u003eFig. 4h\u003c/strong\u003e) and its variability (\u003cstrong\u003eExtended Data\u003c/strong\u003e \u003cstrong\u003eFig. 2a\u003c/strong\u003e) in the running module in both A53T and EV vs. sham mCherry mice. However, a blunted response to VNC optoactivation, i.e. lower duration of optically-induced running, revealed impaired locomotor capacity in A53T mice (\u003cstrong\u003eFig. 4i\u003c/strong\u003e). Underscoring this effect, we found that behavioural modules where running was associated with prior rearing (duration, syllable 11; \u003cstrong\u003eExtended Data\u003c/strong\u003e \u003cstrong\u003eFig. 2b\u003c/strong\u003e)\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eor head movements (max. velocity; syllable 6; \u003cstrong\u003eExtended Data\u003c/strong\u003e \u003cstrong\u003eFig. 2c\u003c/strong\u003e) were sensitive to optoactivation in EV controls, but not in A53T mice.\u003c/p\u003e\n\u003cp\u003eWe next asked whether and how perithreshold and suprathreshold optoactivaction effects are related to each other, in part to cross verify the validity of the syllable analysis. In other words, are specific motor effects induced by higher stimulation intensities (such as turning, circling and body steering behaviours ipsilateral to stimulated/A53T or EV-injected hemisphere) part of a continuous behavioural pattern? Strikingly, we found that those behaviours bear a covert representation in behavioural modules under low-intensity stimulation conditions. For example, variability of angular velocity in ipsilateral (i.e. leftward) steering was enhanced by perithreshold optoactivation of glutamatergic VNC neurons (syllable 7; \u003cstrong\u003eExtended Data\u003c/strong\u003e \u003cstrong\u003eFig. 2e\u003c/strong\u003e). The same manipulation led to an increase in mean angular velocity during ipsilateral rotation in EV, but not A53T mice (syllable 19; \u003cstrong\u003eExtended Data\u003c/strong\u003e \u003cstrong\u003eFig. 2f\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003eDo our results stand irrespective of pre-defined behavioural patterns? To answer this question, we compared kinematics identified through unsupervised behavioural syllable analysis with general locomotion parameters across different behaviours. We investigated locomotor dynamics in the OF with keypoint tracking, identical to the basis of the syllable analysis (\u003cstrong\u003eFig. 4j\u003c/strong\u003e). Overall locomotor capacity, as measured by distance moved, was reduced by habituation in EV controls and A53T groups over time. Although EV animals exhibited strong enhancement of locomotor capacity due to VNC optoactivation (\u003cstrong\u003eFig. 4j\u003c/strong\u003e), this effect was strikingly missing in A53T animals. Sham-stimulated mCherry animals displayed a similar reduction in distance moved with sham stimulation, attributable to habituation effects. (\u003cstrong\u003eFig. 4j\u003c/strong\u003e). Peak locomotor performance (as indicated by maximum velocity) was reduced in A53T mice vs. baseline. EV mice did not show this reduction vs. baseline. In contrast to the lack of effect on capacity, this SNC\u0026nbsp;\u0026alpha;-synucleinopathy-induced deficit was rescued by perithreshold optoactivation (\u003cstrong\u003eFig. 4j\u003c/strong\u003e). This increased peak performance was observed in EV animals as well, however not in sham-stimulated mice, attributing the stimulation effect to specific activation of VNC glutamatergic neurons (\u003cstrong\u003eFig. 4j\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003eIn summary, pose estimation in subsecond behavioural modules reveals that thresholded actuation of glutamatergic VNC neurons leads to overall up-regulated transitions between behavioural modules, indicating increased dynamics in pose shifting. In addition, optoactivation enhances peak locomotor performance of running behaviours. Importantly, these effects were blunted by SNC \u0026alpha;-synucleinopathy, reinforcing that vestibular excitatory modulation of the basal-ganglia loop is instrumental in regulating both peak performance and capacity of locomotor function. In parkinsonian mice, exogenous enhancement of VNC excitatory drive exerts a differential rescue effect on peak locomotor performance but not capacity.\u003c/p\u003e\n\u003cp\u003eOptoactivation of glutamatergic vestibular nucleus complex neurons yields disease-specific retuning of parkinsonian gait dynamics\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe finally wondered, based on our\u0026nbsp;findings of rather specific behavioural rescue effects due to VNC optical actuation, whether the same intervention would specifically affect gait, the motor pattern underlying both locomotor performance and capacity. To this end,\u0026nbsp;we first assessed the maximum velocity on a treadmill and applied DLC-based analysis of gait kinematics equipped with a dual camera system (\u003cstrong\u003eFig. 6a\u003c/strong\u003e). Whereas EV mice exhibited increased maximal velocity over time (\u003cstrong\u003eFig. 6b\u003c/strong\u003e), A53T mice were unable to make the same adjustments. Interestingly, maximal velocity did not change with perithreshold VNC actuation, but closely correlated with both dopaminergic neurodegeneration and rotatory bias in the cylinder test (see \u003cstrong\u003eFig. 3j-k\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003eFirst, to cross-validate the specificity of perithreshold (vs. suprathreshold) optoactivation on locomotion, we opted for an assessment of gait failure by a three-level scoring system, which we termed fluidity of gait (FGI) index. We based the FGI on the number of consecutive strides (as conducted previously in disease models\u0026nbsp;\u003csup\u003e55\u003c/sup\u003e) and added gait interruptions as a complementing measure [one condition (worse counted) had to be met \u0026ndash; FGI 1 = failure: consecutive strides \u0026le; 4, gait interruptions \u0026le; 3; FGI 2 = sufficient: consecutive strides = 5-7, gait interruptions \u0026le; 2; FGI 3 = fluid gait: consecutive strides \u0026ge; 8, gait interruptions \u0026le; 1]. We found that perithreshold stimulation in both A53T and EV mice did not affect gait, whereas suprathreshold stimulation (mice stimulated at VSS 2-3 thresholds) caused markedly enhanced gait failure (\u003cstrong\u003eFig. 6c,d\u003c/strong\u003e; \u003cstrong\u003eSupplementary Video 34\u003c/strong\u003e).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eStrikingly, while optical VNC actuation was ineffective in enhancing velocity, initial observations suggested that it resulted in a \u0026ldquo;smoothening\u0026rdquo; of gait. Hence, we specifically investigated the oscillatory pattern of different keypoint markers and matching keypoint pairs of hind- and forelimb. In detail, we analyzed the synchronicity, periodicity and stability of gait oscillations across speeds in both A53T (summary example overview via scalograms, \u003cstrong\u003eFig. 6e,f\u003c/strong\u003e) and EV mice (\u003cstrong\u003eExtended Data Fig. 3g-h\u003c/strong\u003e) with and without activation of glutamatergic VNC neurons.\u003c/p\u003e\n\u003cp\u003eFor a robust analysis of the oscillatory gait pattern of mice, we conducted an in-depth characterization across speed(s) (10, 15, 20, 25 cm/s) and conditions (baseline = presymptomatic, off, perithreshold) in estimated pose (n = 853 videos; A53T: n = 466, EV: n = 387). We found that A53T (\u003cstrong\u003eFig. 6g\u003c/strong\u003e), but not EV mice (\u003cstrong\u003eFig. 6h\u003c/strong\u003e), developed complex gait aberrations, reflected by a reduction of power in peak gait frequencies in ipsilesional/-stimulatory (left) distal limb markers, mainly driven by changes at 20cm/s (\u003cstrong\u003eFig. 6k\u003c/strong\u003e). Vestibular optoactivation enabled us to retune these changes to baseline level. Using autocorrelation analysis, we further unraveled that simultaneously, the synchronicity of ipsilesional/-stimulatory distal limb markers across speeds decreases in A53T (\u003cstrong\u003eFig. 6m,o,q\u003c/strong\u003e), but not EV mice (\u003cstrong\u003eFig. 6n,p,r\u003c/strong\u003e). Importantly, we could optically retune this impairing effect of the parkinsonian model with a perithreshold activation of the glutamatergic VNC. Further, the relationship of peak amplitude and frequency across speeds (10-25 cm/s) collapsed in A53T mice and could be optically retuned to baseline level with VNC activation (\u003cstrong\u003eFig. 6s,u\u003c/strong\u003e). EV animals did not show any changes due to either time or perithreshold stimulation (\u003cstrong\u003eFig. 6t,v\u003c/strong\u003e). Whereas right-sided markers and vertical coordinates (y-axis) did not show any clear oscillatory changes, we found a trend for a change in the neck marker in A53T mice, which might indicate axial rigidity (\u003cstrong\u003eExtended Data Fig. 3a-f\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003eTaken together, these results demonstrate that a disease-specific retuning of parkinsonian oscillatory gait patterns is possible using thresholded vestibular neuromodulation by specific augmentation of excitatory VNC drive.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study in mice has two key findings: first, we identified structural and functional contributions of excitatory vestibular nucleus complex (VNC) cells to locomotor behaviour in general and to posture and gait dynamics in particular. Second, mimicking thresholded human vestibular neuromodulation, we optically retuned parkinsonian gait and locomotor deficits with activation of VNC glutamatergic neurons in mice.\u003c/p\u003e\n\u003cp\u003eIncremental stimulation intensities yielded a diverse set of posturolocomotor alterations, blending into each other. This is consistent with phenotypes reported by previous human studies on effects of lesions, or targeted neuromodulation\u0026nbsp;\u003csup\u003e56–59\u003c/sup\u003e and animal data\u0026nbsp;\u003csup\u003e60–62,31\u003c/sup\u003e on manipulation of central vestibular pathways subserving head, body and eye movements in yaw, pitch and roll plane\u0026nbsp;\u003csup\u003e63\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eIndeed, structural mapping of glutamatergic VNC neurons exhibited extensive bilateral connectivity with key motor hubs, predominantly in the thalamus (centromedian-parafascicular (CM-Pf) nucleus complex), subthalamus (zona incerta), and brainstem (caudal medulla). Thresholded optical activation of VNC excitatory neurons led to enhanced activity of neuronal ensembles in these target regions, confirming functional relevance of a glutamatergic VNC circuit element linked to the quintessential basal ganglia-brainstem axis for movement\u0026nbsp;\u003csup\u003e9\u003c/sup\u003e. These structural and functional circuit findings are in line with previous anatomical studies, which found a vestibulo-thalamo-striatal tract via the parafascicular nucleus in rats\u0026nbsp;\u003csup\u003e23\u003c/sup\u003e and data from an optogenetic fMRI study in mice\u0026nbsp;\u003csup\u003e24\u003c/sup\u003e, underscoring integration of thalamo-subthalamic with VNC circuitry. The strong overlap with thalamostriatal CM-Pf neurons projecting into the dlSTR hints at how the VNC likely conveys its strong influence onto motor systems: by the functionally segregated projections from the Pf to the striatum\u0026nbsp;\u003csup\u003e64\u003c/sup\u003e. The Pf, and its interconnections to STR (and STN) specifically, have been implicated in motor rescue recently\u0026nbsp;\u003csup\u003e45,46\u003c/sup\u003e. As a target of excitatory VNC inputs, this Pf function hints at putative pathways for locomotor rescue effects in our parkinsonian mice. Moreover, the Pf is known to cause ipsiversive head steering, and with prolonged stimulation, full body turns\u0026nbsp;\u003csup\u003e65\u003c/sup\u003e, which corresponds to the pathway mediating the head tilting and turning effects caused by incremental VNC optoactivation. Specifically, the encoding of vector components in head velocity and directional steering by Pf\u0026nbsp;\u003csup\u003e65\u003c/sup\u003e might underlie both thresholded VNC optoactivation effects on behavioural dynamics and point towards circuit dysfunction causing directionality bias in parkinsonian mice.\u003c/p\u003e\n\u003cp\u003eVNC optoactivation favored locomotor performance involving running behaviours.\u0026nbsp;Since we found that the VNC directly targets the reticular formation in the caudal medulla, it might leverage ‘hyperdirect’ locomotor influence via extensive connectivity to the Gi, sparsely also to LPGi, potentially explaining the distinct effects on behavioural motifs involving high-speed locomotion\u0026nbsp;\u003csup\u003e47,12,66\u003c/sup\u003e. We corroborated these findings by showing that peak free, but not forced, locomotor performance (e.g. maximum velocity) is enhanced by thresholded optical vestibular neuromodulation. Further, we show that global changes in behavioural dynamics underlie specific changes in behavioural motifs, again enhancing locomotion.\u003c/p\u003e\n\u003cp\u003eYet, one of our primary hypotheses was that the vestibular system might interact with the mesencephalic locomotor region (MLR) in mediating effects on posture and locomotion, including under parkinsonian conditions\u0026nbsp;\u003csup\u003e67,10,11,13\u003c/sup\u003e. Deep brain stimulation of the MLR in patients with Parkinson’s disease (PD) and a posture und gait-predominant phenotype has been explored, but results remain clinically inconsistent\u0026nbsp;\u003csup\u003e15,14\u003c/sup\u003e. It is interesting to note that we did not find strong signs for glutamatergic connectivity of VNC neurons to neither pedunculopontine nor cuneiform nuclei; especially since electrical vestibular stimulation seems to enhance deficient pedunculopontine nucleus connectivity in PD patients\u0026nbsp;\u003csup\u003e67\u003c/sup\u003e. This may, in part, relate to our cell type-specific approach focused on the optical activation of excitatory \u003cem\u003eVglut2\u003c/em\u003e-VNC neurons. Given that GABAergic neurons seem to exert negligible global (motor) influence\u0026nbsp;\u003csup\u003e31\u003c/sup\u003e, we postulate that the glutamatergic population subserves most functional influence on posture and locomotion. However, a primarily segregated (likely the second largest) \u003cem\u003evGlyT2\u003c/em\u003e-defined glycinergic neuron population\u0026nbsp;\u003csup\u003e30,32\u003c/sup\u003e remains to be functionally explored, which might exhibit MLR connectivity instead. Furthermore, a distinct glutamatergic population of vestibulo-cerebellar neurons, defined via \u003cem\u003evGLUT1\u003c/em\u003e-expression, remains to be characterized in detail\u0026nbsp;\u003csup\u003e32\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eFrom a translational bedside-to-bench perspective, we sought to generate mechanistic insights into the central circuit-level effects of non-invasive vestibular stimulation, such as galvanic vestibular stimulation (GVS). The former shows promise in addressing both non-motor and motor symptoms in typical and atypical parkinsonian states, including hard-to-treat axial features\u0026nbsp;\u003csup\u003e19,20,22,25,26\u003c/sup\u003e. We aimed to mimic a perceptually-thresholded stimulation with central vestibular network activation (like nGVS) by using optogenetics in EV and parkinsonian mice. Massive impairment of motor behaviour (e.g. gait) by suprathreshold optogenetic stimulation is in line with well-known dose-response relationships in human vestibular stimulation\u003csup\u003e18\u003c/sup\u003e. Our individually thresholded optoactivation of the VNC vs. sham stimulation shifted mice into a pro-locomotor state by enhancing behavioural transitions, peak locomotor performance and locomotor capacity. However, our data demonstrate specificity of these effects in a parkisonian context. Here, VNC optoactivation allowed for rescue of peak locomotor performance and improved gait patterns, but did not alter locomotor capacity.\u003c/p\u003e\n\u003cp\u003eWhile the behavioural effects of thresholded VNC activation were studied in detail, the precise pathway-specific effects remain unidentified. Our tracing studies suggest the mediation of motor effects via vestibulo-(sub)thalamic or vestibulo-medullary projections, warranting further investigation into single-circuit effects combining e.g. projection specific \u003cem\u003ein vivo\u003c/em\u003e electrophysiology and optogenetic manipulation. We chose the A53T model for its neuropathological profile, mimicking the progressive course of PD associated with α-synucleinopathy and the exploration of an early-stage PD level for potential disease modification. Exploring vestibular effects on severe hypo-/bradykinesia could be addressed in a more severe toxin-based mouse model for PD.\u003c/p\u003e\n\u003cp\u003eConsidering the broad impact of the vestibular system, including at a cortical level\u0026nbsp;\u003csup\u003e24\u003c/sup\u003e, on navigation, affective and autonomic regulation, an integrated examination of both behavioural and autonomic readouts is imperative, especially from a translational perspective. Specifically, the complex effects of vestibular circuits on integrated behaviours (navigation, defense, etc.) require further investigation, not least because of its connectivity to a proposed global behavioural state regulator, the zona incerta (ZI)\u0026nbsp;\u003csup\u003e68,69\u003c/sup\u003e. The strong influence of the VNC on sensorimotor circuitry is another hint towards the importance of brainstem-basal ganglia interactions for movement \u003csup\u003e9\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eTo sum up, through structural and functional mapping, as well as \u003cem\u003ein vivo\u003c/em\u003e optogenetics, our data demonstrate a key regulatory role for excitatory vestibular neurons in axial movement, a function that can be exploited to retune and normalize motor symptoms in parkinsonian states.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis project was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) Project-ID 424778381-TRR 295 (A01 to J.V. and C.W.I., A06 to J.V. and C.W.I., B06 to P.T., S01 to R.P.) and by the Interdisciplinary Center for Clinical Research (IZKF) at the University of Würzburg (N-362). J.H. was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) Project-ID 424778381-TRR 295 (Gerok position). M.F. received funding by the Jung Stiftung für Wissenschaft and the Manfred and Ursula Mueller Stiftung. J.V. received funding from the European Union's Horizon 2020 research and innovation programme under the EJP RD COFUND-EJP N° 825575 (EurDyscover). C.W.I. received funding from the VERUM foundation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eContributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eJ.H. conceived ideas, acquired and analyzed data and wrote the manuscript. M.F. conceived ideas, acquired and analyzed data and contributed to the manuscript. J.S.G. analyzed data and contributed to the manuscript. S.T., N.S., D.D., R.P., S.K. and T.P. analyzed data and contributed to the manuscript. C.W.I., P.T. and J.V. contributed resources and edited the manuscript.\u003c/p\u003e"},{"header":"Online Methods","content":"\u003cp\u003e\u003cem\u003eAnimals and ethics approval\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eAll experiments were conducted in accordance with local and European animal welfare law, are ARRIVE-compliant and were approved by the local veterinary authorities and animal experimentation ethics committee (AZ 2-631 and 2-2030, Regierung von Unterfranken, Bavaria, Germany). Transgenic \u003cem\u003eVglut2-ires-cre\u0026nbsp;\u003c/em\u003e(\u003cstrong\u003eSTOCK Slc17a6\u003csup\u003etm2(cre)Lowl\u003c/sup\u003e/J\u003c/strong\u003e\u003cem\u003e,\u0026nbsp;\u003c/em\u003eJackson #016963, The Jackson Laboratory, USA) and C57Bl6/J mice (Charles River Laboratories Germany) were commercially acquired and bred at in-house breeding facilities (Institute of Clinical Neurobiology and Center for Experimental and Molecular Medicine, Wuerzburg, Germany). A total of n=42 mice were used (n=24 optogenetics, n=6 sham optogenetics, n=7 tracing cross-bred transgenic/wildtype, adult male \u003cem\u003eVglut2-ires-cre\u003c/em\u003e mice; n=5 adult male C57Bl/6J non-behavioural parkinsonian mice). Animals were housed at 14/10-hour light-dark cycles with ad libitum availability to rodent nutrition and water. No statistical methods were used to predetermine sample sizes. Mice were randomly assigned to experimental/control groups.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eStereotaxic surgery procedures\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eA stereotaxic apparatus (Kopf, David Kopf Instruments, USA or Neurostar, Neurostar Germany) was used for injecting viral tracers, optogenetic constructs, and AAVs containing A53T-mutated or EV genomic particles. The surgical procedures generally followed previously described protocols\u0026nbsp;\u003csup\u003e70,71\u003c/sup\u003e. In detail, animals were maintained under isoflurane anesthesia (4% induction, 1-2% maintenance) throughout the procedures with systemic analgesic (Buprenorphine or Carprofen) administered ~30 minutes prior subcutaneously. Local anesthetic (Ropivacaine) was injected under the scalp, and a longitudinal incision was made after a two-minute wait. Veterinary eye ointment was applied to keep the eyes moisturized. A stereotaxic drill was used to create a small burr hole above the implantation and/or injection sites. We used the following stereotaxic coordinates, derived from the stereotaxic mouse brain atlas\u0026nbsp;\u003csup\u003e72\u003c/sup\u003e: Vestibular Nucleus Complex (VNC): [AP: -6.00; ML: -1.20; DV: -4.20/30], Substantia Nigra Pars Compacta (SNC): [AP: -3.10, ML: -1.40, DV: -4.40, -4.35, -4.30], Dorsolateral Striatum (dlSTR): [AP: 0; ML: +2.20; DV: -3.00, -2.70, -2.50], Subthalamic Nucleus (STN): [AP: -2.00; ML: +1.50; DV: -4.90]. Optical fibers were placed ~200-300 \u0026micro;m above the injection site of the optogenetic viral construct. \u003cem\u003eAAV1/2-A53T\u003c/em\u003e and \u003cem\u003eAAV1/2-EV\u003c/em\u003e constructs were serially injected (DV: -4.40, -4.35, -4.30) to target the dorsoventral spread of dopaminergic SNC cells. \u003cem\u003eretroAAV-mCherry\u003c/em\u003e was serially injected into the dlSTR (DV: -3.00, -2.70, -2.50). Post-injection, a ~6-minute wait time ensured adequate vector diffusion and minimized risk of viral spreading. The exposed skull was sutured for injection-only surgeries or covered with cyanoacrylate for optical fiber implantation to maximize stability. In postoperative care, animals were closely monitored, and analgesics (Carprofen or Meloxicam) were administered preemptively and repetitively. Mice received a minimum of 4 days of rest before starting any behavioural experiments and were continuously monitored day-to-day.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eViral tracing\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eFor both anterograde and retrograde viral tracing studies, we utilized commercially available adeno-associated viruses (AAV) or AAV constructs that were gifted from other labs. Viruses were generated in an in-house viral preparation facility from plasmids according to standard protocols for viral cloning, clonal expansion, etc. For anterograde tracing of Vglut2\u003csup\u003e+\u003c/sup\u003e glutamatergic VNC neurons, we utilized an myc-tagged anterograde synaptophysin viral tracer (\u003cem\u003eAAV2/5-CAG-Floxed-Syp10xMyc-rev.WPRE\u003c/em\u003e, UNC Vector Core). For concomitant retrograde tracing from the STR or STN, we used a retrograde tracer (\u003cem\u003eretroAAV-hSyn-DIO-mCherry\u003c/em\u003e, Addgene), driving the expression of mCherry. Viral tracers (both antero- and retrograde) were expressed for four weeks before euthanasia.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eGeneration of the parkinsonian (A53T-aSyn) mouse model\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eFor the generation of a parkinsonian phenotype, we utilized the A53T-\u0026alpha;-synuclein mouse model of PD\u0026nbsp;\u003csup\u003e38\u003c/sup\u003e. We used a concentrated viral suspension (15x10\u003csup\u003e12\u003c/sup\u003e gp/ml) for both the AAV1/2-A53T and AAV1/2-EV construct (kindly provided by Dr. James Brotchie and Jonathan Koprich, Canada). The stereotaxic procedure and serial dorsoventral injections of the A53T and EV construct for generation of parkinsonian and EV mice, respectively, was conducted as described above. Verification of \u0026alpha;-synuclein (aSyn) overexpression and dopaminergic neurodegeneration in the SNC was achieved by immunohistochemistry, which was followed by immunofluorescence microscopy and stereological estimation, respectively.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eOptogenetic experiments\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eMice used in optogenetic experiments with parkinsonian A53T vs. EV control, as well as the sham-stimulation mCherry group all had \u003cem\u003eVglut2-ires-cre\u003c/em\u003e background. For optical actuation in A53T and EV mice, we utilized \u003cem\u003eChRmine\u003c/em\u003e, a highly sensitive red-light activated cation-conducting channelrhodopsin (CCR) delivered via a plasmid inside an anterograde AAV construct (\u003cem\u003eAAV2/5-hSyn-DIO-ChRmine-mScarlet,\u0026nbsp;\u003c/em\u003eAddgene). For sham actuation, the mCherry group was injected with an anterograde AAV construct, which delivered a plasmid lacking an optical actuator (construct (\u003cem\u003eAAV2/5-hSyn-DIO-mCherry,\u0026nbsp;\u003c/em\u003eAddgene). Commercially acquired 4-4.5 mm optical fibres (Doric Lenses, Canada) were placed 0.2-0.3 mm above the target region. A numerical aperture of 0.53 was chosen for all used fibres. Laser power at the fibre tip was gauged before or after an optogenetic experiment with a photometer (ThorLabs Inc., New Jersey, USA). We utilized a laser driver (Doric Lenses, Canada) for generation of laser pulses and the accompanying Doric Neuroscience Studio software (Doric Lenses, Canada) for coding stimulation regimes. Considering data on tonic vestibular neuron firing rates and intrinsic properties of ChRmine\u0026nbsp;\u003csup\u003e32,73,74\u003c/sup\u003e, we opted for 20 Hz photostimulation at 10 ms pulse width with 10 s pulse trains. Pause intervals were pseudorandomized to 30-40 s.\u003c/p\u003e\n\u003cp\u003eIn titration experiments (TV-OF), we stimulated at all 4 different VSS thresholds in an interspersed manner during a total recording period of 10 minutes per mouse.\u003c/p\u003e\n\u003cp\u003eFor experiments with perithreshold stimulation, we stimulated just below (i.e. 1 mA below threshold) the individual threshold for vestibular symptoms to appear (i.e. VSS 1 on the VSS) and in an interspersed manner during a total recording period of 10 minutes per mouse. On the treadmill, we stimulated on demand during each run.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eBehavioural testing and analysis\u003c/em\u003e\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eGeneralities and data pooling\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eMice for behavioural testing from the parkinsonian A53T and EV group with optogenetics were tested at baseline, a presymptomatic time point at week 1 after surgery. Repeat testing when A53T animals became symptomatic was conducted at week 4 and week 8. Since an independent cohort of A53T mice, euthanized at week 4 after injection of AAV1/2-A53T showed similar levels of dopaminergic cell loss, we pooled data from week 4 and 8 postoperatively for the A53T and EV group from optogenetic experiments. Data from week 4 off and week 8 off recordings were pooled as \u0026ldquo;off\u0026rdquo;. Data with perithreshold optoactivation of the VNC was pooled from week 4 and 8 as \u0026ldquo;perithreshold\u0026rdquo;. The mCherry sham group was tested at week 2 in off and with sham-optogenetics and with sham-optogenetics at week 3 after surgery, when viral expression plateaued\u0026nbsp;\u003csup\u003e37\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eCylinder test\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eThe cylinder test was conducted as described previously \u003csup\u003e75,76\u003c/sup\u003e. We recorded mice in a transparent cylinder for 10 minutes after habituation on video. We then assessed both rotational asymmetry and asymmetry of forepaw use \u003csup\u003e38\u003c/sup\u003e as indeces by using the following formulae:\u003c/p\u003e\n\u003cp\u003e\u003cimg src=\"data:image/png;base64,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\"\u003e\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eAmphetamine-induced rotation test\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eThe amphetamine-induced rotation test was conducted as described previously\u0026nbsp;\u003csup\u003e51\u003c/sup\u003e. At week 8, we injected mice intraperitoneally with 0,5 mg/kg body weight d-amphetamine and assessed for total rotations in the off vs. perithreshold condition with optogenetic actuation. Animals were pseudorandomized to start in the perithreshold vs. off condition, like in the cylinder test. A total 40 min (20 min-off; 20 min-perithreshold) were recorded. Ipsi-, contralateral and total rotations were counted by a human rater (M.F.). On this basis, statistical analyses were conducted.\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eTop-view open field (TV-OF)\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eOpen field (OF) trials were conducted in a square, white open field box (39,5 x 39,5 x 39,5 cm). Individual trials lasted 10 minutes. We recorded videos at 30 frames per second (fps) with a proprietary behavioural analysis system (EthoVision, Noldus Inc., Wageningen, Netherlands). For kinematic feature analyses, we used a custom-trained deep learning model, trained within DeepLabCut (DLC)\u0026nbsp;\u003csup\u003e40\u003c/sup\u003e on top-view videos of OF trials (see section Deep learning-based pose estimation). Afterwards, coordinate time-series were used for further analysis and feature extraction and analyzed via in-house built MATLAB (MATLAB R2022b, MathWorks, Natick, MA, USA) and Python (version 3.9. or 3.10) scripts.\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eTreadmill gait analysis\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eTo assess gait thoroughly, we used a commercially available treadmill setup by Digigait (Mouse Specific Inc., Massachusetts, USA), which includes two (ventral and lateral) high-resolution cameras (Basler, 165 fps cameras, Basler AG, Ahrensburg, Germany). Animals were extensively habituated to the treadmill context before experiments. Both maximum velocity (V\u003csub\u003emax\u003c/sub\u003e, see next section) and recorded gait sessions (on video) at different treadmill belt velocities (10, 15, 20, 25 cm/s) were assessed at baseline, week 4 and week 8. First videos were recorded, then V\u003csub\u003emax\u003c/sub\u003e sessions conducted, to reduce systematic error due to exhaustion.\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eTreadmill gait analysis - Maximum velocity (V\u003csub\u003emax\u003c/sub\u003e)\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eTo assess gait-associated locomotor performance, maximum velocity (V\u003csub\u003emax\u003c/sub\u003e) was used as a quantitative surrogate. Mice were tested to determine their V\u003csub\u003emax\u003c/sub\u003e at each time point, based on the hypothesis that increasing treadmill velocity requires greater physical effort to maintain stable gait. V\u003csub\u003emax\u0026nbsp;\u003c/sub\u003ewas tested with perithreshold stimulation and in the off condition at week 4 and week 8 (at baseline only off), to examine the effects of vestibular optostimulation. Animals were pseudorandomized to either perithreshold\u0026rarr;off or off\u0026rarr;perithreshold at weeks 4 and 8. V\u003csub\u003emax\u003c/sub\u003e analysis involved calculating the mean V\u003csub\u003emax\u003c/sub\u003e from 3 trials. V\u003csub\u003emax\u003c/sub\u003e was determined by progressively increasing treadmill speed until gait failure (hitting the bumper with the rear). Initial treadmill speed was set between 15 and 20 cm/s to ensure a start in stable gait, then increased progressively. V\u003csub\u003emax\u003c/sub\u003e was assessed regardless of treadmill belt direction (forwards or reverse).\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eTreadmill gait analysis - Fluidity of gait index (FGI)\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eIn order to verify that suprathreshold vs. perithreshold stimulation impairs locomotion significantly, we devised a gait scoring system, based on previous findings of treadmill gait capacity with regards to consecutive strides in a murine disease model\u0026nbsp;\u003csup\u003e55\u003c/sup\u003e. For this reason, we devised a trinomial scoring system, with which we aimed to capture overall gait stability per run: \u003cem\u003ethe fluidity of gait index (FGI)\u003c/em\u003e. We anchored the scoring system in a combined count of a) consecutive strides\u0026nbsp;\u003csup\u003e55\u003c/sup\u003e and b) gait interruptions per treadmill run and animal as such that:\u0026nbsp;one condition (worse counted) had to be met \u0026ndash; FGI 1 = failure: consecutive strides \u0026le; 4, gait interruptions \u0026le; 3; FGI 2 = sufficient: consecutive strides = 5-7, gait interruptions \u0026le; 2; FGI 3 = fluid gait: consecutive strides \u0026ge; 8, gait interruptions \u0026le; 1].\u0026nbsp;Evidently, any semiquantitative scoring system may be affected by interrater variability, hence we opted to validate the FGI via assessment of interrater agreement by Cohen\u0026rsquo;s kappa (k). Both behavioural experimenters (M.F. + J.H.) individually and blindly scored all treadmill runs according to the pre-devised FGI scoring system and interrater agreement was calculated. A Cohen\u0026rsquo;s kappa (k) of 0.619 (95% CI: 0.540-0.699) and weighted kappa of 0.671 (which is considered a more accurate measure when data are ordinally scaled)\u0026nbsp;\u003csup\u003e77\u003c/sup\u003e ascertained the validity of the FGI scoring system and showed moderate two-rater inter-agreement in FGI allocation per individual trial.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u003cstrong\u003eDeep learning-based pose estimation\u003c/strong\u003e\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eFor\u0026nbsp;body part keypoint tracking and pose estimation, we used DeepLabCut (DLC, version 2.3.1)\u0026nbsp;\u003csup\u003e40,41\u003c/sup\u003e. Specifically, we trained n=2 different recurrent convolutional neural networks (RCNN) for a) the TV-OF and b) the combined lateral-ventral dual-view videos of treadmill gait.\u003c/p\u003e\n\u003cp\u003eFor the TV-OF RCNN, we utilized the pretrained RCNN\u0026nbsp;\u0026lsquo;\u003cem\u003esuperanimal_topviewmouse\u003c/em\u003e\u0026rsquo;\u0026nbsp;\u003csup\u003e53\u003c/sup\u003e and labeled an additional n=255 number of frames taken from n=16 videos (then 95% was used for training). We used a ResNet-152-based neural network with default parameters for 7 training iterations. We validated with 1 shuffle, and found the test error was: 1.09 pixels, train: 2.04 pixels (image size was \u0026asymp; 640 by 624). We used a p-cutoff of 0.6. This network was then used to analyze videos from similar experimental settings. Coordinate time-series were further processed in the different custom-made python and MATLAB pipelines.\u003c/p\u003e\n\u003cp\u003eFor the RCNN used on combined lateral and ventral (i.e. bottom), dual-view videos (30 fps) of gait on the treadmill, we labeled n=1409 number of frames taken from n=118 videos (then 95% was used for training). We used a ResNet-50-based neural network with default parameters for 3 training iterations. We validated with 1 shuffle, and found the test error was: 1.66 pixels, train: 4.45 pixels (image size was \u0026asymp; 718 by 682). We used a p-cutoff of 0.6. This network was then used to analyze videos from similar experimental settings. Coordinate time-series were further processed in the different custom-made python and MATLAB pipelines.\u003c/p\u003e\n\u003ch3\u003eDeep learning-assisted gait analysis\u003c/h3\u003e\n\u003cp\u003eAfter training the RCNN via DLC, we analyzed an extensive dataset of treadmill videos (total n=387 videos from EV mice, total n=466 videos from A53T mice) from different speed settings\u0026nbsp;\u003csup\u003e47\u003c/sup\u003e (10, 15, 20, 25 cm/s) from baseline, week 4 and week 8 testing.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 1. Number of videos per condition (pooled from week 4 and week 8 evaluations):\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eEV\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15.5629%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e10cm/s\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20.1987%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e15cm/s\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20.1987%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e20cm/s\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19.0397%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e25cm/s\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003ebaseline\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15.5629%;\"\u003e\n \u003cp\u003e19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20.1987%;\"\u003e\n \u003cp\u003e19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20.1987%;\"\u003e\n \u003cp\u003e20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19.0397%;\"\u003e\n \u003cp\u003e20\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003eoff\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15.5629%;\"\u003e\n \u003cp\u003e35\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20.1987%;\"\u003e\n \u003cp\u003e39\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20.1987%;\"\u003e\n \u003cp\u003e40\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19.0397%;\"\u003e\n \u003cp\u003e40\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003ethreshold\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15.5629%;\"\u003e\n \u003cp\u003e35\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20.1987%;\"\u003e\n \u003cp\u003e40\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20.1987%;\"\u003e\n \u003cp\u003e40\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19.0397%;\"\u003e\n \u003cp\u003e40\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eParkinsonian (A53T)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15.5629%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e10cm/s\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20.1987%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e15cm/s\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20.1987%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e20cm/s\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19.0397%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e25cm/s\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003ebaseline\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15.5629%;\"\u003e\n \u003cp\u003e28\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20.1987%;\"\u003e\n \u003cp\u003e27\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20.1987%;\"\u003e\n \u003cp\u003e28\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19.0397%;\"\u003e\n \u003cp\u003e28\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003eoff\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15.5629%;\"\u003e\n \u003cp\u003e38\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20.1987%;\"\u003e\n \u003cp\u003e45\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20.1987%;\"\u003e\n \u003cp\u003e46\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19.0397%;\"\u003e\n \u003cp\u003e44\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003ethreshold\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15.5629%;\"\u003e\n \u003cp\u003e43\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20.1987%;\"\u003e\n \u003cp\u003e47\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 20.1987%;\"\u003e\n \u003cp\u003e47\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 19.0397%;\"\u003e\n \u003cp\u003e45\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eWe then used the resultant coordinate time series for further analysis via a custom MATLAB script (MATLAB R2022b, MathWorks, Natick, MA, USA). Unfiltered time series of body part coordinates obtained from DeepLabCut were imported into MATLAB, where they were calibrated based on pre-recording snippets and aligned between the two cameras. Subsequently, frequency analysis was conducted using continuous wavelet transform via the Wavelet Toolbox 6.2 (MathWorks, Natick, MA, USA) after linearly interpolating missing values or values with a confidence score below 0.95. Data points falling outside the cone of influence, where wavelet analysis has low confidence due to limited data, were excluded from further analysis. Scalograms were visualized as montages, representing different speeds for a given mouse and condition. The magnitude of the wavelet coefficients over time was integrated for each frequency and normalized by the overall signal power, providing a power spectrum. Power spectrum, peak frequency and peak power were extracted per recording and averaged per mouse for pooled conditions (e.g. week 4 and week 8), and then across mice. Autocorrelograms were generated from normalized distances between pairs of markers. Power spectra and autocorrelograms were compared. Finally, an F-test was conducted to determine whether the relationships between peak frequency, peak amplitude, and speed differed significantly across experimental conditions. Linear models with and without interaction terms between these variables and the experimental condition were fit to the data. The F-test compared the goodness of fit between these models, with the p-value indicating whether the inclusion of interaction terms significantly improved the model\u0026apos;s ability to explain the variance in the data.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eUnsupervised behavioural classification via Keypoint-MoSeq (KPMS)\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eWe trained a full KPMS model as described previously\u0026nbsp;\u003csup\u003e52\u003c/sup\u003e after fitting an initial autoregressive hidden markov model (AR-HMM) with a kappa hyperparameter of 10000 (dimensionless) and resultant approximate median syllable duration of 12 at 25 frames per second (fps) ~ 300ms. Results of principal component analysis (PCA) on explained variance by the result of the AR-HMM\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eshowed that 6 principal components explained \u0026ge; 90% of variance. The kappa (stickiness) hyperparameter was maintained at 10000 for fitting the full model (final median syllable duration also at 12 at 25 frames per second (fps) ~ 300ms). The model was applied on n=139 TV-OF videos from the parkinsonian A53T, EV (both at baseline, week 4, week 8, latter two with and without perithreshold stimulation) and sham mCherry mice (at week 2 (with and without (sham) stimulation) and week 3 with (sham) stimulation). Downstream statistical analysis and plotting was conducted as described previously\u0026nbsp;\u003csup\u003e52,54\u003c/sup\u003e and is specifically described below:\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSyllable statistics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll statistical analyses were performed using Python (version 3.9 or 3.10). To compare syllable statistics across multiple groups \u003csup\u003e52,54\u003c/sup\u003e, we performed Kruskal-Wallis tests, implemented manually with clustered permutation tests (10,000 permutations, significance threshold set at p \u0026lt; 0.05), including tie correction factors. For post-hoc analysis, Dunn\u0026rsquo;s z test was used for pairwise comparisons, with p-values adjusted using the Benjamini-Hochberg method to control the false discovery rate (FDR), and significance determined by an adjusted p-value threshold of 0.05.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSyllable transition matrices and transition graphs\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFrequencies of syllable occurrences were calculated, excluding syllables with a frequency below 0.005. Transition matrices were generated to analyze transitions between syllables within each condition using custom Python scripts and functions, normalized by bigram counts, rows, and columns. Visualizations of transition matrices and syllable frequencies were created with Matplotlib and Seaborn. Transition graphs were generated using NetworkX and visualized with Matplotlib, with nodes representing syllables and edges representing transition probabilities. Node sizes were scaled based on syllable usage, and edges colored to represent transition probabilities. Differences in transition graphs between groups were analyzed by calculating differences in transition matrices and usage, visualized with NetworkX and Matplotlib. Eigenvalues and eigenvectors of transition matrices were computed to assess spectral properties, compared using permutation tests to generate a null distribution, and p-values were corrected using the Bonferroni method. Procrustes analysis was used to compare eigenvector spaces between groups.\u003c/p\u003e\n\u003cp\u003eNode Scaling: Node sizes were determined by normalized usage differences, scaled by a factor (e.g., 3000) and a constant (e.g., 500) for visibility. The formula used was:\u003c/p\u003e\n\u003cp\u003e\u003cimg src=\"data:image/png;base64,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\"\u003e\u003c/p\u003e\n\u003cp\u003eLine Thickness Scaling: Edge thickness indicated transition probability differences, scaled by a factor (e.g., 1000) with colors indicating the direction of difference (green for positive, violet for negative).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eSemisupervised\u003c/em\u003e\u003cem\u003e\u0026nbsp;behavioural classification via A-SoiD\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eTo verify the Vestibular Symptom Scale (VSS) we devised, we made use of a semi-supervised machine learning approach within the A-SoiD pipeline, as described previously\u0026nbsp;\u003csup\u003e42\u003c/sup\u003e by annotation of specific behaviours in an ethogram, leveraging the BORIS software\u0026nbsp;\u003csup\u003e78\u003c/sup\u003e combined with pose estimation data. Specifically, we extensively trained an active learning classifier (\u003cem\u003e\u0026lsquo;main behaviour classifier\u0026rsquo;\u003c/em\u003e) on a diverse set of top-view pose estimation data using n = 31 snippet videos (~10 s length, i.e. single optostimulation pulse) of VSS symptomatology and n = 3 full titration videos (week 4, 10 min length) on identifying both vestibular symptoms and naturalistic behaviour. These videos were pose estimated within DLC with the model described previously in the section \u0026lsquo;Deep learning-based pose estimation\u0026rsquo;. Second, each video was loaded into BORIS and every single frame annotated in an ethogram with either one of two categories: 1) Naturalistic behaviour, 2) VSS-like behaviour. The ethograms were all exported as singular behaviour binary files at 0.04 s (25 Hertz = Hz). The training parameters were set as follows: DLC likelihood cutoff: 0.99, minimum bout duration: 0.05s, training fraction: 0.01, number of iterations: 100, confidence threshold for inclusion in training dataset: 0.95, video frame rate: 25 fps, ethogram rate: 0.04 s = 25 Hertz. The classifier achieved ~90% average performance (\u003cstrong\u003esee Supplementary Figure 2 for details\u003c/strong\u003e). We then predicted unseen titration data (week 8).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eVia a built-in unsupervised clustering/embedding algorithm (B-SoiD)\u0026nbsp;\u003csup\u003e36\u003c/sup\u003e, we aimed to identify subtypes from pose, predicted by our active learning classifier. We clustered and embedded a diverse pose dataset of n = 261 snippet videos of vestibular symptomatology and naturalistic behaviour from week 4 titrations (hdbscan, min. percentage for cluster in both: 3%, normalization and relaxed embedding options deactivated). We fed these subclasses into a second, new active learning classifier (\u003cem\u003e\u0026lsquo;subtype behaviour classifier\u0026rsquo;\u003c/em\u003e) (~90% average performance), and re-predicted all full titration videos (10 min, week 4 and week 8 titration). The training parameters were set as follows: DLC likelihood cutoff: 0.99, minimum bout duration: 0.05s, training fraction: 0.01, number of iterations: 500, confidence threshold for inclusion in training dataset: 0.95, video frame rate: 25 fps, ethogram rate: 0.04 s = 25 Hz (\u003cstrong\u003ealso see Supplementary Figure 2, Supplementary Videos 4-10\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eBrain processing and immunohistochemistry\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eAt the end of the experiments, animals were euthanized and transcardially perfused at 65 rpm at 90 minutes (for cFos expression)\u0026nbsp;\u003csup\u003e49\u003c/sup\u003e after the last optogenetic experiment in the TV-OF with a mixture of 0.01mM phosphate-buffered saline (PBS) and 0.34% heparin sulfate for 3-5 minutes. Adequate perfusion was confirmed by observing hepatic blood clearing, adjusting needle placement or pump rate as needed. This was followed by 10-15 minutes of perfusion with 4% paraformaldehyde (PFA). Brains were then extracted, washed in 0.01mM PBS, and post-fixed in 4% PFA for 12 hours. Subsequently, brains were placed in a sterile-filtered 30% D-Saccharose solution for 48 hours until dehydration and then frozen in tissue gel matrix (O.C.T., TissueTek) at -20\u0026deg;C until cutting and staining. Brains were sectioned into 40\u0026micro;m slices using a cryotome (Leica PM1950, Leica) and placed in a cryoprotection solution (30% glycerol, 30% PBS, 40% ethanol). Before immunostaining, slices were washed in 0.01mM PBS and permeabilized with 0.1% Triton-X.\u003c/p\u003e\n\u003cp\u003eFor dopaminergic cell staining and \u0026alpha;-synuclein overexpression verification, primary antibodies against tyrosine hydroxylase (chicken anti-tyrosine hydroxylase, 1:500, Abcam, ab76442) and \u0026alpha;-synuclein (rabbit anti-\u0026alpha;-synuclein, 1:30000, Sigma-Aldrich, Sigma #S3062) were used. Secondary antibodies were goat anti-chicken Alexa Fluor 488 (Invitrogen #A11039) and goat anti-rabbit Cy3 (Dianova #111-165-144). Slices were counterstained with DAPI (1:500,000, Sigma-Aldrich, Sigma #D8417) and mounted in embedding medium (AquaPolymount, Polysciences) for microscopy.\u003c/p\u003e\n\u003cp\u003eFor stereological estimation, primary antibody against tyrosine hydroxylase (rabbit anti-tyrosine hydroxylase, 1:1000, Abcam, ab112) were used with secondary antibody (goat anti-rabbit, 1:100, Vektor #BA-1000) followed by ABC (Thermo Scientific #32050) and DAB (Vektor #SK 4100). Slices were then mounted and counterstained with cresyl violet solution (1 gram cresyl violet (Merck#1.05235.0025) + 10ml acetic acid (100%; Sigma#33209-1L)) and subsequently put into an ascending concentration ethanol, (70-96-100%) and finally a xylol bath. Slices were mounted and embedded in Vitroclud medium (R. Langenbrinck GmbH, Emmendingen, Germany, Article-No. 04-0001).\u003c/p\u003e\n\u003cp\u003eThe mScarlet and mCherry fluorophore tagging the viruses was imaged natively. Exemplary images from Figure 1 and Figure 2o were counterstained with FluoroNissl (1:200, NeuroTrace\u0026trade; 435/455 Blue Fluorescent, N21479, ThermoFischer).\u003c/p\u003e\n\u003cp\u003eFor viral tracing, primary antibodies against myc (goat-anti-Myc, 1:500, Abcam, ab9132) and red fluorescent protein (rabbit-anti-RFP, 1:500, Rockland/Biomol, #600-401-379) were used. Secondary antibodies were donkey anti-goat Cy5 (1:800, Jackson Laboratories #705-175-147) and donkey anti-rabbit Cy3 (1:800, Jackson Laboratories #711-165-152). Slices were counterstained with DAPI (1:500,000, Sigma-Aldrich, Sigma #D8417) and mounted in embedding medium (AquaPolymount, Polysciences) for microscopy.\u003c/p\u003e\n\u003cp\u003eFor the cFos staining\u0026nbsp;\u003csup\u003e48,49\u003c/sup\u003e (the animals were transcardially perfused 90 minutes after final perithreshold optoactivation of the VNC in the TV-OF), we quenched brain slices for quantification of immunoreactive cells via \u003cem\u003eDeepSlice2\u0026nbsp;\u003c/em\u003ewith 100mM glycine at 7.4 pH and blocked and permeabilized with 0.3% Triton-X, 0.1% Tween 20 at 10% normal goat serum in 0.01mM PBS. Afterwards, slices were incubated in primary antibody (rabbit anti-cFos, 1:1000, Synaptic Systems, #226003) and tagged with secondary antibody (donkey anti-rabbit Cy5, 1:200, Jackson Laboratories, #711175152). Slices were counterstained with DAPI (1:500,000, Sigma-Aldrich, Sigma #D8417) and mounted in embedding medium (AquaPolymount, Polysciences) for microscopy.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eMicroscopy and Stereology\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eAfter immunofluorescence stainings, brain slices were imaged at a pseudoconfocal microscope (Axioimager M2, Zeiss) at different magnifications. Images for cFos analysis were taken on a region of interest (ROI) basis at 20x magnification of each hemisphere without a z-stack at 3 different anteroposterior levels. Viral tracing images were acquired in a similar fashion at the same pseudoconfocal microscope (Axioimager M2, Zeiss) at 10x and 20x magnification.\u003c/p\u003e\n\u003cp\u003eExemplary images in Figure 1 and Figure 2o were taken with a different pseudoconfocal imager (Leica Thunder DMi8 imager, Leica) at 10x magnification and z-stacked. Proprietary large volume computational clearing was used for background clearing. We used a maximum projection for these images. Brightness and contrast were adjusted for print visibility on all images.\u003c/p\u003e\n\u003cp\u003eDirect immunohistochemistry images were further used for a stereological quantification of dopaminergic SN pars compacta (SNC) cells with an optical fractionator microscope (Olympus BX53, OM Digital Solutions), the StereoInvestigator \u0026nbsp;64-bit software (MBF Bioscience, Williston, USA) and a Prior ProScan III device (MBF Bioscience, Williston, USA) in an unbiased manner, as described previously\u0026nbsp;\u003csup\u003e79\u003c/sup\u003e. For stereology, serial 40\u0026micro;m cuts of the SN were used and quantified (High magnification lens (used for quantification mode): 100x, Counting Frame X: 60.00 \u0026mu;m, Counting Frame Y: 60.00 \u0026mu;m, Grid Size X: 100.00 \u0026mu;m, Grid Size Y: 100.00 \u0026mu;m, Section Cut Thickness: 40.00 \u0026mu;m, GuardZone Type: Fixed Distance, GuardZone Height: 2.00 \u0026mu;m, Dissector Height: 16.00 \u0026mu;m). Both Nissl\u003csup\u003e+\u0026nbsp;\u003c/sup\u003e(grandaverage neuron estimate) and TH\u003csup\u003e+\u003c/sup\u003e cells (dopaminergic neurons) were quantified and correlations established.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003ecFos activity analysis and deep image segmentations\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eWe used an in-house built deep-learning model ensemble with U-net/encoder-decoder architecture for regional quantification of cFos immunoreactivity per ROI as described previously \u003csup\u003e44,80\u003c/sup\u003e with the \u003cem\u003edeepflash2\u003c/em\u003e pipeline. Three independent raters (J.H., T.P., D.D.) annotated a total of 27 immunohistochemistry images of cFos stainings (4 images from rats, 23 from mice). We used a test/train split of 8/19 (images) and trained an ensemble of 5 models. Ground truth was established and the U-net convolutional neural network (CNN) model ensemble compared to the 3 raters\u0026lsquo; performances. Considering the different dice scores (dc), we found that the model ensemble for the most part performed at least as well as the individual raters in terms of accuracy (dc to estimated ground truth\u0026asymp;0.8). Via the ensemble, segmentation went as follows: instance segmentations by Cellpose, semantic segmentations via \u003cem\u003edeepflash2\u003c/em\u003e. We imaged n=3 anteroposterior levels for all ROIs: anterior, middle, posterior for each mouse (A53T: n=12, EV: n=10, mCherry: n=6) at a pseudoconfocal microscope (Axioimager M2, Zeiss). Subsequently, we calculated the feature density in features per micrometer squared using a Python script (Python, Version 3.9). Finally, we pooled results from both hemispheres for further statistical analysis to get a sense of overall regional activation, considering the differential, yet overwhelmingly bilateral innervation of thalamo-subthalamic and caudal medulla targets. We analyzed a total of 526 images (A53T: n=216, EV: n=184, mCherry: n=127) from the groups above, by generating masks of ROIs by using atlas outlines of brain regions \u003csup\u003e72\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eTransfected cell count estimation\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eWe estimated the amount of transfected Vglut2-positive neurons in the vestibular nucleus complex (VNC) manually by quantification inside Fiji/ImageJ (doi:10.1038/nmeth.2019). For each individual mouse, we took three images of the VNC along the anteroposterior axis, overlayed the atlas outline \u003csup\u003e72\u003c/sup\u003e and quantified the number of mScarlet-positive cells in the VNC. We formed an average across all 3 anteroposterior levels per animal. This average was taken as the transfected cell count estimate and analyzed subsequently.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eStatistical analysis, plotting and data reporting\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eStatistical analyses and plotting of analyses were conducted using GraphPad Prism (Version 10.0). Further, we utilized custom-built MATLAB (R2022b, MathWorks, Natick, MA, USA) and Python (Python, Version 3.9 or 3.10) scripts for statistical analyses and plotting. Assessment for normality was conducted by using normality/lognormality testing (most reliable test according to sample size was used) and assessment of Q-Q plots. Individual statistical testing was conducted on a group level via analysis of variance (ANOVA, one- or two-way) or mixed-effects models whenever possible and relevant, to assess for factor and interaction effects. In case of non-gaussian distribution of data, corresponding standard non-parametric equivalents were used. The OF kinematics of sham mCherry mice were compared with a paired t-test. Interrater reliability of the FGI score was assessed by using the Cohen\u0026rsquo;s kappa online calculator by GraphPad, found on the GraphPad website (https://www.graphpad.com/quickcalcs/kappa1.cfm). For cFos quantification and VSS thresholds, we removed significant outliers with the ROUT method in GraphPad Prism (Version 10.0). Data plotting is described individually for each plot in the figure legends. Significance is annotated as *p\u0026lt;0.05, **p\u0026lt;0.01, ***p\u0026lt;0.001, ****p\u0026lt;0.0001. Figures were created with Adobe Illustrator and BioRender. Source data and custom code is available from the corresponding authors upon reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eLau, L. M. de \u0026amp; Breteler, M. M. Epidemiology of Parkinson\u0026rsquo;s disease. \u003cem\u003eLancet Neurol.\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, (2006).\u003c/li\u003e\n \u003cli\u003eKalia, L. V. \u0026amp; Lang, A. E. Parkinson\u0026rsquo;s disease. \u003cem\u003eThe Lancet\u003c/em\u003e \u003cstrong\u003e386\u003c/strong\u003e, 896\u0026ndash;912 (2015).\u003c/li\u003e\n \u003cli\u003eArmstrong, M. J. \u0026amp; Okun, M. S. Time for a New Image of Parkinson Disease. \u003cem\u003eJAMA Neurol.\u003c/em\u003e \u003cstrong\u003e77\u003c/strong\u003e, 1345 (2020).\u003c/li\u003e\n \u003cli\u003eBeuter, A., Hern\u0026aacute;ndez, R., Rigal, R., Modolo, J. \u0026amp; Blanchet, P. J. Postural sway and effect of levodopa in early Parkinson\u0026rsquo;s disease. \u003cem\u003eCan. J. Neurol. Sci. J. Can. Sci. Neurol.\u003c/em\u003e \u003cstrong\u003e35\u003c/strong\u003e, 65\u0026ndash;68 (2008).\u003c/li\u003e\n \u003cli\u003eKim, S. D., Allen, N. E., Canning, C. G. \u0026amp; Fung, V. S. C. Postural Instability in Patients with Parkinson\u0026rsquo;s Disease. \u003cem\u003eCNS Drugs\u003c/em\u003e \u003cstrong\u003e27\u003c/strong\u003e, (2013).\u003c/li\u003e\n \u003cli\u003eSzlufik, S. \u003cem\u003eet al.\u003c/em\u003e The Neuromodulatory Impact of Subthalamic Nucleus Deep Brain Stimulation on Gait and Postural Instability in Parkinson\u0026rsquo;s Disease Patients: A Prospective Case Controlled Study. \u003cem\u003eFront. Neurol.\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, (2018).\u003c/li\u003e\n \u003cli\u003eMoreira, F., Gomes, I. R. \u0026amp; Janu\u0026aacute;rio, C. Freezing of gait and postural instability: the unpredictable response to levodopa in Parkinson\u0026rsquo;s disease. \u003cem\u003eBMJ Case Rep.\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, (2019).\u003c/li\u003e\n \u003cli\u003eKotagal, V. Is PIGD a legitimate motor subtype in Parkinson disease? \u003cem\u003eAnn. Clin. Transl. Neurol.\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, 473\u0026ndash;477 (2016).\u003c/li\u003e\n \u003cli\u003eArber, S. \u0026amp; Costa, R. M. Networking brainstem and basal ganglia circuits for movement. \u003cem\u003eNat. Rev. Neurosci.\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e, 342\u0026ndash;360 (2022).\u003c/li\u003e\n \u003cli\u003eFoug\u0026egrave;re, M. \u003cem\u003eet al.\u003c/em\u003e Optogenetic stimulation of glutamatergic neurons in the cuneiform nucleus controls locomotion in a mouse model of Parkinson\u0026rsquo;s disease. \u003cem\u003eProc. Natl. Acad. Sci. U. S. A.\u003c/em\u003e \u003cstrong\u003e118\u003c/strong\u003e, e2110934118 (2021).\u003c/li\u003e\n \u003cli\u003eMasini, D. \u0026amp; Kiehn, O. Targeted activation of midbrain neurons restores locomotor function in mouse models of parkinsonism. \u003cem\u003eNat. Commun.\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 504 (2022).\u003c/li\u003e\n \u003cli\u003eCaggiano, V. \u003cem\u003eet al.\u003c/em\u003e Midbrain circuits that set locomotor speed and gait selection. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e553\u003c/strong\u003e, (2018).\u003c/li\u003e\n \u003cli\u003eRyczko, D. The Mesencephalic Locomotor Region: Multiple Cell Types, Multiple Behavioral Roles, and Multiple Implications for Disease. \u003cem\u003eThe Neuroscientist\u003c/em\u003e 10738584221139136 (2022) doi:10.1177/10738584221139136.\u003c/li\u003e\n \u003cli\u003eThevathasan, W. \u003cem\u003eet al.\u003c/em\u003e Pedunculopontine nucleus deep brain stimulation in Parkinson\u0026rsquo;s disease: A clinical review. \u003cem\u003eMov. Disord. Off. J. Mov. Disord. Soc.\u003c/em\u003e \u003cstrong\u003e33\u003c/strong\u003e, 10\u0026ndash;20 (2018).\u003c/li\u003e\n \u003cli\u003eWang, H., Gao, H., Jiao, T. \u0026amp; Luo, Z. A meta-analysis of the pedunculopontine nucleus deep-brain stimulation effects on Parkinson\u0026rsquo;s disease. \u003cem\u003eNeuroReport\u003c/em\u003e \u003cstrong\u003e27\u003c/strong\u003e, 1336\u0026ndash;1344 (2016).\u003c/li\u003e\n \u003cli\u003eViolante, I. R. \u003cem\u003eet al.\u003c/em\u003e Non-invasive temporal interference electrical stimulation of the human hippocampus. \u003cem\u003eNat. Neurosci.\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e, 1994\u0026ndash;2004 (2023).\u003c/li\u003e\n \u003cli\u003eWessel, M. J. \u003cem\u003eet al.\u003c/em\u003e Noninvasive theta-burst stimulation of the human striatum enhances striatal activity and motor skill learning. \u003cem\u003eNat. Neurosci.\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e, 2005\u0026ndash;2016 (2023).\u003c/li\u003e\n \u003cli\u003ePires, A. P. B. de \u0026Aacute;. \u003cem\u003eet al.\u003c/em\u003e Galvanic vestibular stimulation and its applications: a systematic review. \u003cem\u003eBraz. J. Otorhinolaryngol.\u003c/em\u003e \u003cstrong\u003e88 Suppl 3\u003c/strong\u003e, S202\u0026ndash;S211 (2022).\u003c/li\u003e\n \u003cli\u003eKhoshnam, M., H\u0026auml;ner, D. M. C., Kuatsjah, E., Zhang, X. \u0026amp; Menon, C. Effects of galvanic vestibular stimulation on upper and lower extremities motor symptoms in parkinson\u0026rsquo;s disease. \u003cem\u003eFront. Neurosci.\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, (2018).\u003c/li\u003e\n \u003cli\u003eWilkinson, D. \u003cem\u003eet al.\u003c/em\u003e Caloric vestibular stimulation for the management of motor and non-motor symptoms in Parkinson\u0026rsquo;s disease. \u003cem\u003eParkinsonism Relat. Disord.\u003c/em\u003e \u003cstrong\u003e65\u003c/strong\u003e, 261\u0026ndash;266 (2019).\u003c/li\u003e\n \u003cli\u003eWilkinson, D. Caloric and galvanic vestibular stimulation for the treatment of Parkinson\u0026rsquo;s disease: rationale and prospects. \u003cem\u003eExpert Rev. Med. Devices\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 649\u0026ndash;655 (2021).\u003c/li\u003e\n \u003cli\u003eWuehr, M. \u003cem\u003eet al.\u003c/em\u003e Effects of Low-Intensity Vestibular Noise Stimulation on Postural Instability in Patients with Parkinson\u0026rsquo;s Disease. \u003cem\u003eJ. Park. Dis.\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 1611\u0026ndash;1618 (2022).\u003c/li\u003e\n \u003cli\u003eLai, H. \u003cem\u003eet al.\u003c/em\u003e Morphological evidence for a vestibulo-thalamo-striatal pathway via the parafascicular nucleus in the rat. \u003cem\u003eBrain Res.\u003c/em\u003e \u003cstrong\u003e872\u003c/strong\u003e, (2000).\u003c/li\u003e\n \u003cli\u003eLeong, A. T. L. \u003cem\u003eet al.\u003c/em\u003e Optogenetic fMRI interrogation of brain-wide central vestibular pathways. \u003cem\u003eProc. Natl. Acad. Sci. U. S. A.\u003c/em\u003e \u003cstrong\u003e116\u003c/strong\u003e, 10122\u0026ndash;10129 (2019).\u003c/li\u003e\n \u003cli\u003eKataoka, H. \u003cem\u003eet al.\u003c/em\u003e Effect of galvanic vestibular stimulation on axial symptoms in Parkinson\u0026rsquo;s disease. \u003cem\u003eJ. Cent. Nerv. Syst. Dis.\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 11795735221081599 (2022).\u003c/li\u003e\n \u003cli\u003eWuehr, M. \u003cem\u003eet al.\u003c/em\u003e Low-intensity vestibular noise stimulation improves postural symptoms in progressive supranuclear palsy. \u003cem\u003eJ. Neurol.\u003c/em\u003e (2024) doi:10.1007/s00415-024-12419-9.\u003c/li\u003e\n \u003cli\u003ePeto, D. \u003cem\u003eet al.\u003c/em\u003e No evidence for effects of low-intensity vestibular noise stimulation on mild-to-moderate gait impairments in patients with Parkinson\u0026rsquo;s disease. \u003cem\u003eJ. Neurol.\u003c/em\u003e \u003cstrong\u003e271\u003c/strong\u003e, 5489 (2024).\u003c/li\u003e\n \u003cli\u003eSamoudi, G., Nissbrandt, H., Dutia, M. B. \u0026amp; Bergquist, F. Noisy galvanic vestibular stimulation promotes GABA release in the substantia nigra and improves locomotion in Hemiparkinsonian rats. \u003cem\u003ePLoS ONE\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, (2012).\u003c/li\u003e\n \u003cli\u003eTakazawa, T., Saito, Y., Tsuzuki, K. \u0026amp; Ozawa, S. Membrane and firing properties of glutamatergic and GABAergic neurons in the rat medial vestibular nucleus. \u003cem\u003eJ. Neurophysiol.\u003c/em\u003e \u003cstrong\u003e92\u003c/strong\u003e, 3106\u0026ndash;3120 (2004).\u003c/li\u003e\n \u003cli\u003eBagnall, M. W., Stevens, R. J. \u0026amp; du Lac, S. Transgenic mouse lines subdivide medial vestibular nucleus neurons into discrete, neurochemically distinct populations. \u003cem\u003eJ. Neurosci. Off. J. Soc. Neurosci.\u003c/em\u003e \u003cstrong\u003e27\u003c/strong\u003e, 2318\u0026ndash;2330 (2007).\u003c/li\u003e\n \u003cli\u003eMontardy, Q. \u003cem\u003eet al.\u003c/em\u003e Selective optogenetic stimulation of glutamatergic, but not GABAergic, vestibular nuclei neurons induces immediate and reversible postural imbalance in mice. \u003cem\u003eProg. Neurobiol.\u003c/em\u003e \u003cstrong\u003e204\u003c/strong\u003e, (2021).\u003c/li\u003e\n \u003cli\u003eKodama, T. \u003cem\u003eet al.\u003c/em\u003e Neuronal Classification and Marker Gene Identification via Single-Cell Expression Profiling of Brainstem Vestibular Neurons Subserving Cerebellar Learning. \u003cem\u003eJ. Neurosci.\u003c/em\u003e \u003cstrong\u003e32\u003c/strong\u003e, 7819\u0026ndash;7831 (2012).\u003c/li\u003e\n \u003cli\u003eStiles, L. \u0026amp; Smith, P. F. The vestibular\u0026ndash;basal ganglia connection: Balancing motor control. \u003cem\u003eBrain Res.\u003c/em\u003e \u003cstrong\u003e1597\u003c/strong\u003e, 180\u0026ndash;188 (2015).\u003c/li\u003e\n \u003cli\u003eSabzevar, F. T., Vautrelle, N., Zheng, Y. \u0026amp; Smith, P. F. Vestibular modulation of the tail of the rat striatum. \u003cem\u003eSci. Rep.\u003c/em\u003e (2023).\u003c/li\u003e\n \u003cli\u003eSmith, P. F. Vestibular Functions and Parkinson\u0026rsquo;s Disease. \u003cem\u003eFront. Neurol.\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, (2018).\u003c/li\u003e\n \u003cli\u003eHsu, A. I. \u0026amp; Yttri, E. A. B-SOiD, an open-source unsupervised algorithm for identification and fast prediction of behaviors. \u003cem\u003eNat. Commun.\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 5188 (2021).\u003c/li\u003e\n \u003cli\u003eKishi, K. E. \u003cem\u003eet al.\u003c/em\u003e Structural basis for channel conduction in the pump-like channelrhodopsin ChRmine. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e185\u003c/strong\u003e, 672-689.e23 (2022).\u003c/li\u003e\n \u003cli\u003eIp, C. W. \u003cem\u003eet al.\u003c/em\u003e AAV1/2-induced overexpression of A53T-\u0026alpha;-synuclein in the substantia nigra results in degeneration of the nigrostriatal system with Lewy-like pathology and motor impairment: a new mouse model for Parkinson\u0026rsquo;s disease. \u003cem\u003eActa Neuropathol. Commun.\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 11 (2017).\u003c/li\u003e\n \u003cli\u003eLaflamme, O. D. \u0026amp; Akay, T. Excitatory and inhibitory crossed reflex pathways in mice. \u003cem\u003eJ. Neurophysiol.\u003c/em\u003e \u003cstrong\u003e120\u003c/strong\u003e, 2897\u0026ndash;2907 (2018).\u003c/li\u003e\n \u003cli\u003eMathis, A. \u003cem\u003eet al.\u003c/em\u003e DeepLabCut: markerless pose estimation of user-defined body parts with deep learning. \u003cem\u003eNat. Neurosci.\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 1281\u0026ndash;1289 (2018).\u003c/li\u003e\n \u003cli\u003eNath, T. \u003cem\u003eet al.\u003c/em\u003e Using DeepLabCut for 3D markerless pose estimation across species and behaviors. \u003cem\u003eNat. Protoc.\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 2152\u0026ndash;2176 (2019).\u003c/li\u003e\n \u003cli\u003eTillmann, J. F., Hsu, A. I., Schwarz, M. K. \u0026amp; Yttri, E. A. A-SOiD, an active-learning platform for expert-guided, data-efficient discovery of behavior. \u003cem\u003eNat. Methods\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 703\u0026ndash;711 (2024).\u003c/li\u003e\n \u003cli\u003eBecker-Bense, S. \u003cem\u003eet al.\u003c/em\u003e Direct comparison of activation maps during galvanic vestibular stimulation: A hybrid H2[15 O] PET\u0026mdash;BOLD MRI activation study. \u003cem\u003ePLOS ONE\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, e0233262 (2020).\u003c/li\u003e\n \u003cli\u003eGriebel, M. \u003cem\u003eet al.\u003c/em\u003e Deep learning-enabled segmentation of ambiguous bioimages with deepflash2. \u003cem\u003eNat. Commun.\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 1679 (2023).\u003c/li\u003e\n \u003cli\u003eWatson, G. D. R. \u003cem\u003eet al.\u003c/em\u003e Thalamic projections to the subthalamic nucleus contribute to movement initiation and rescue of parkinsonian symptoms. \u003cem\u003eSci. Adv.\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, eabe9192 (2021).\u003c/li\u003e\n \u003cli\u003eZhang, Y. \u003cem\u003eet al.\u003c/em\u003e Targeting thalamic circuits rescues motor and mood deficits in PD mice. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e607\u003c/strong\u003e, 321\u0026ndash;329 (2022).\u003c/li\u003e\n \u003cli\u003eBellardita, C. \u0026amp; Kiehn, O. Phenotypic characterization of speed-associated gait changes in mice reveals modular organization of locomotor networks. \u003cem\u003eCurr. Biol. CB\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, 1426\u0026ndash;1436 (2015).\u003c/li\u003e\n \u003cli\u003eBullitt, E. Expression of c-fos-like protein as a marker for neuronal activity following noxious stimulation in the rat. \u003cem\u003eJ. Comp. Neurol.\u003c/em\u003e \u003cstrong\u003e296\u003c/strong\u003e, 517\u0026ndash;530 (1990).\u003c/li\u003e\n \u003cli\u003eHerrera, D. G. \u0026amp; Robertson, H. A. Activation of c-fos in the brain. \u003cem\u003eProg. Neurobiol.\u003c/em\u003e \u003cstrong\u003e50\u003c/strong\u003e, 83\u0026ndash;107 (1996).\u003c/li\u003e\n \u003cli\u003ePaumier, K. L. \u003cem\u003eet al.\u003c/em\u003e Behavioral Characterization of A53T Mice Reveals Early and Late Stage Deficits Related to Parkinson\u0026rsquo;s Disease. \u003cem\u003ePLoS ONE\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, e70274 (2013).\u003c/li\u003e\n \u003cli\u003eBj\u0026ouml;rklund, A. \u0026amp; Dunnett, S. B. The Amphetamine Induced Rotation Test: A Re-Assessment of Its Use as a Tool to Monitor Motor Impairment and Functional Recovery in Rodent Models of Parkinson\u0026rsquo;s Disease. \u003cem\u003eJ. Park. Dis.\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 17\u0026ndash;29 (2019).\u003c/li\u003e\n \u003cli\u003eWeinreb, C. \u003cem\u003eet al.\u003c/em\u003e Keypoint-MoSeq: parsing behavior by linking point tracking to pose dynamics. \u003cem\u003eNat. Methods\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 1329\u0026ndash;1339 (2024).\u003c/li\u003e\n \u003cli\u003eYe, S. \u003cem\u003eet al.\u003c/em\u003e SuperAnimal pretrained pose estimation models for behavioral analysis. \u003cem\u003eNat. Commun.\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 5165 (2024).\u003c/li\u003e\n \u003cli\u003eWiltschko, A. B. \u003cem\u003eet al.\u003c/em\u003e Mapping Sub-Second Structure in Mouse Behavior. \u003cem\u003eNeuron\u003c/em\u003e \u003cstrong\u003e88\u003c/strong\u003e, 1121\u0026ndash;1135 (2015).\u003c/li\u003e\n \u003cli\u003eVincelette, J. \u003cem\u003eet al.\u003c/em\u003e Gait analysis in a murine model of collagen-induced arthritis. \u003cem\u003eArthritis Res. Ther.\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, R123 (2007).\u003c/li\u003e\n \u003cli\u003eBrandt, T. Vestibular disorders in (horizontal) yaw plane. in \u003cem\u003eVertigo: Its Multisensory Syndromes\u003c/em\u003e (ed. Brandt, T.) 215\u0026ndash;218 (Springer, New York, NY, 2003). doi:10.1007/978-1-4757-3801-8_12.\u003c/li\u003e\n \u003cli\u003eBrandt, T. \u0026amp; Dieterich, M. Vestibular syndromes in the roll plane: Topographic diagnosis from brainstem to cortex. \u003cem\u003eAnn. Neurol.\u003c/em\u003e \u003cstrong\u003e36\u003c/strong\u003e, 337\u0026ndash;347 (1994).\u003c/li\u003e\n \u003cli\u003eBrandt, T. \u0026amp; Dietrich, M. Skew deviation with ocular torsion: a vestibular brainstem sign of topographic diagnostic value. \u003cem\u003eAnn. Neurol.\u003c/em\u003e \u003cstrong\u003e33\u003c/strong\u003e, 528\u0026ndash;534 (1993).\u003c/li\u003e\n \u003cli\u003eFriedrich, M. U. \u003cem\u003eet al.\u003c/em\u003e Current-dependent ocular tilt reaction in deep brain stimulation of the subthalamic nucleus: Evidence for an incerto-interstitial pathway? \u003cem\u003eEur. J. Neurol.\u003c/em\u003e \u003cstrong\u003e29\u003c/strong\u003e, 1545\u0026ndash;1549 (2022).\u003c/li\u003e\n \u003cli\u003eLopez, C. \u0026amp; Blanke, O. The thalamocortical vestibular system in animals and humans. \u003cem\u003eBrain Res. Rev.\u003c/em\u003e \u003cstrong\u003e67\u003c/strong\u003e, 119\u0026ndash;146 (2011).\u003c/li\u003e\n \u003cli\u003eCullen, K. E. The vestibular system: multimodal integration and encoding of self-motion for motor control. \u003cem\u003eTrends Neurosci.\u003c/em\u003e \u003cstrong\u003e35\u003c/strong\u003e, 185\u0026ndash;196 (2012).\u003c/li\u003e\n \u003cli\u003eBasaldella, E., Takeoka, A., Sigrist, M. \u0026amp; Arber, S. Multisensory Signaling Shapes Vestibulo-Motor Circuit Specificity. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e163\u003c/strong\u003e, 301\u0026ndash;312 (2015).\u003c/li\u003e\n \u003cli\u003eBrandt, T. \u0026amp; Dieterich, M. Central vestibular syndromes in roll, pitch, and yaw planes: Topographic diagnosis of brainstem disorders. \u003cem\u003eNeuro-Ophthalmol.\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 291\u0026ndash;303 (1995).\u003c/li\u003e\n \u003cli\u003eMandelbaum, G. \u003cem\u003eet al.\u003c/em\u003e Distinct Cortical-Thalamic-Striatal Circuits through the Parafascicular Nucleus. \u003cem\u003eNeuron\u003c/em\u003e \u003cstrong\u003e102\u003c/strong\u003e, 636-652.e7 (2019).\u003c/li\u003e\n \u003cli\u003eFallon, I. P. \u003cem\u003eet al.\u003c/em\u003e The role of the parafascicular thalamic nucleus in action initiation and steering. \u003cem\u003eCurr. Biol.\u003c/em\u003e \u003cstrong\u003e33\u003c/strong\u003e, 2941-2951.e4 (2023).\u003c/li\u003e\n \u003cli\u003eLeiras, R., Cregg, J. M. \u0026amp; Kiehn, O. Brainstem Circuits for Locomotion. \u003cem\u003eAnnu. Rev. Neurosci.\u003c/em\u003e \u003cstrong\u003e45\u003c/strong\u003e, 63\u0026ndash;85 (2022).\u003c/li\u003e\n \u003cli\u003eCai, J. \u003cem\u003eet al.\u003c/em\u003e Galvanic Vestibular Stimulation (GVS) Augments Deficient Pedunculopontine Nucleus (PPN) Connectivity in Mild Parkinson\u0026rsquo;s Disease: fMRI Effects of Different Stimuli. \u003cem\u003eFront. Neurosci.\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, (2018).\u003c/li\u003e\n \u003cli\u003eWang, X., Chou, X., Zhang, L. I. \u0026amp; Tao, H. W. Zona Incerta: An Integrative Node for Global Behavioral Modulation. \u003cem\u003eTrends Neurosci.\u003c/em\u003e \u003cstrong\u003e43\u003c/strong\u003e, 82\u0026ndash;87 (2020).\u003c/li\u003e\n \u003cli\u003eMitrofanis, J. Some certainty for the \u0026ldquo;zone of uncertainty\u0026rdquo;? Exploring the function of the zona incerta. \u003cem\u003eNeuroscience\u003c/em\u003e \u003cstrong\u003e130\u003c/strong\u003e, 1\u0026ndash;15 (2005).\u003c/li\u003e\n \u003cli\u003eTovote, P. \u003cem\u003eet al.\u003c/em\u003e Midbrain circuits for defensive behaviour. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e534\u003c/strong\u003e, 206\u0026ndash;212 (2016).\u003c/li\u003e\n \u003cli\u003eLetzkus, J. J. \u003cem\u003eet al.\u003c/em\u003e A disinhibitory microcircuit for associative fear learning in the auditory cortex. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e480\u003c/strong\u003e, 331\u0026ndash;335 (2011).\u003c/li\u003e\n \u003cli\u003ePaxinos, G. \u0026amp; Franklin, K. \u003cem\u003ePaxinos and Franklin\u0026rsquo;s the Mouse Brain in Stereotaxic Coordinates - 5th Edition\u003c/em\u003e. (2019).\u003c/li\u003e\n \u003cli\u003eAv-Ron, E. \u0026amp; Vidal, P.-P. Intrinsic membrane properties and dynamics of medial vestibular neurons: a simulation. \u003cem\u003eBiol. Cybern.\u003c/em\u003e \u003cstrong\u003e80\u003c/strong\u003e, 383\u0026ndash;392 (1999).\u003c/li\u003e\n \u003cli\u003eJohnston, A. R., MacLeod, N. K. \u0026amp; Dutia, M. B. Ionic conductances contributing to spike repolarization and after-potentials in rat medial vestibular nucleus neurones. \u003cem\u003eJ. Physiol.\u003c/em\u003e \u003cstrong\u003e481\u003c/strong\u003e, 61\u0026ndash;77 (1994).\u003c/li\u003e\n \u003cli\u003eGlajch, K. E., Fleming, S. M., Surmeier, D. J. \u0026amp; Osten, P. Sensorimotor assessment of the unilateral 6-hydroxydopamine mouse model of Parkinson\u0026rsquo;s disease. \u003cem\u003eBehav. Brain Res.\u003c/em\u003e \u003cstrong\u003e230\u003c/strong\u003e, 309\u0026ndash;316 (2012).\u003c/li\u003e\n \u003cli\u003eFleming, S. M., Ekhator, O. R. \u0026amp; Ghisays, V. Assessment of Sensorimotor Function in Mouse Models of Parkinson\u0026rsquo;s Disease. \u003cem\u003eJ. Vis. Exp. JoVE\u003c/em\u003e 50303 (2013) doi:10.3791/50303.\u003c/li\u003e\n \u003cli\u003eLandis, J. R. \u0026amp; Koch, G. G. An application of hierarchical kappa-type statistics in the assessment of majority agreement among multiple observers. \u003cem\u003eBiometrics\u003c/em\u003e \u003cstrong\u003e33\u003c/strong\u003e, 363\u0026ndash;374 (1977).\u003c/li\u003e\n \u003cli\u003eFriard, O. \u0026amp; Gamba, M. BORIS : a free, versatile open‐source event‐logging software for video/audio coding and live observations. \u003cem\u003eMethods Ecol. Evol.\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, 1325\u0026ndash;1330 (2016).\u003c/li\u003e\n \u003cli\u003eBaquet, Z. C., Williams, D., Brody, J. \u0026amp; Smeyne, R. J. A comparison of model-based (2D) and design-based (3D) stereological methods for estimating cell number in the substantia nigra pars compacta (SNpc) of the C57BL/6J mouse. \u003cem\u003eNeuroscience\u003c/em\u003e \u003cstrong\u003e161\u003c/strong\u003e, 1082\u0026ndash;1090 (2009).\u003c/li\u003e\n \u003cli\u003eSegebarth, D. \u003cem\u003eet al.\u003c/em\u003e On the objectivity, reliability, and validity of deep learning enabled bioimage analyses. \u003cem\u003eeLife\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, e59780 (2020).\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-5851215/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5851215/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003ePostural and locomotor dysfunction represent axial symptoms of Parkinson’s disease (PD), remaining poorly treated by medication and deep brain stimulation. Non-invasive neuromodulation of the vestibular system, centered on the vestibular nucleus complex (VNC), offers a novel therapeutic avenue. However, the underlying circuits remain ill-explored. In this study, we found that the VNC in mice feeds extensive \u003cem\u003eVglut2\u003c/em\u003e-defined projections into striato-thalamo-subthalamic and caudal medulla motor hubs, but not the mesencephalic locomotor region. Optoactivation of excitatory VNC neurons below the threshold for vestibular symptoms promoted activity in these basal ganglia-brainstem axis targets. Unbiased analysis of pose dynamics revealed global enhancement of behavioural transitions and locomotion, confirmed by regular kinematic analyses. Therapeutically, it enabled resynchronization of naturalistic gait patterns and improved locomotor performance, but not capacity, in parkinsonian mice. Our data identify excitatory VNC circuit processes for therapeutic retuning of motor dysfunction in the context of PD.\u003c/p\u003e","manuscriptTitle":"Vestibular circuit stimulation for retuning locomotor dynamics in Parkinson's disease","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-02-10 07:52:20","doi":"10.21203/rs.3.rs-5851215/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-communications","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"NCOMMS","sideBox":"Learn more about [Nature Communications](http://www.nature.com/ncomms/)","snPcode":"","submissionUrl":"https://mts-ncomms.nature.com/","title":"Nature Communications","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Communications","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"5268de3a-dd7f-44d4-a941-11d8d9d0a678","owner":[],"postedDate":"February 10th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":43471105,"name":"Health sciences/Diseases/Neurological disorders/Parkinson's disease"},{"id":43471106,"name":"Health sciences/Anatomy/Nervous system/Central nervous system/Brain"},{"id":43471107,"name":"Health sciences/Diseases/Neurological disorders/Neurodegeneration"}],"tags":[],"updatedAt":"2025-02-10T07:52:20+00:00","versionOfRecord":[],"versionCreatedAt":"2025-02-10 07:52:20","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-5851215","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5851215","identity":"rs-5851215","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}