Targeting the lateral mammillary nucleus rescues sleep disturbances in Parkinson’s disease

Targeting the lateral mammillary nucleus rescues sleep disturbances 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 Targeting the lateral mammillary nucleus rescues sleep disturbances in Parkinson’s disease Qizhou Ke, Pengxiang Xu, Wenjie Meng, Bo Yang, Yu Wang, Dongsheng Sun, and 6 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9084382/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Sleep disturbances are highly prevalent, early‑onset non‑motor symptoms in Parkinson’s disease (PD) that severely compromise quality of life. However, these neural substrates for sleep disturbance of PD remain poorly understood. Here, we reveal that the lateral part of the medial mammillary nucleus (ML) within the mammillary body (MB) is a critical hub for sleep-wake control and a key substrate of PD-related sleep dysfunction. Radiomic analysis of multi-modal MRI data from PD patients reveal that structural heterogeneity within the MB robustly predicts the presence of significant sleep disturbances, establishing it as a novel imaging biomarker. In mice, we demonstrate that ML neurons exhibit heterogeneous, state-specific activity patterns across sleep-wake cycles. Chemogenetic manipulation establishes that ML neuronal activity is both sufficient to promote NREM and REM sleep, and necessary for normal sleep architecture. Electrode recordings demonstrate that broad suppression of ML neuronal firing alongside a paradoxical hyperactivity of REM-active subpopulations. Strikingly, chemogenetic activation of ML neurons rescues these sleep disturbances, increasing total, NREM, and REM sleep without altering motor function. Therefore, our findings reveal a pathological neural mechanism underlying sleep dysfunction in PD and highlight the ML as a promising therapeutic target for PD-related sleep disorders. Health sciences/Neurology Biological sciences/Neuroscience the lateral mammillary nucleus Parkinson’s disease sleep disturbances rescue REM sleep NREM sleep Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction PD non-motor symptoms, particularly sleep disorders are increasingly recognized as core features that significantly impair quality of life and often precede motor onset by years 1 , 2 . Among these, rapid eye movement (REM) sleep behavior disorder (RBD), insomnia, non-REM (NREM) sleep fragmentation and excessive daytime sleepiness are particularly prevalent, affecting up most PD patients 3 , 4 . Current evidence suggests that sleep disturbances in PD arise from multifactorial etiologies, including primary neurodegeneration of sleep-regulating neural circuits, dopaminergic dysregulation, circadian rhythm dysfunction, and secondary effects of motor symptoms or medications 5 . However, the precise neural substrates underlying sleep dysfunction in PD remain incompletely understood, limiting the development of targeted therapeutic interventions. The mammillary body (MB) is a small deep brain structure located in the posterior hypothalamus, which is a key component of the limbic system and has long been recognized for its role in memory 6 , 7 , brain oscillation 8 , 9 , and spatial navigation 10 . The region is composed of the medial mammillary nucleus (MMn) and lateral mammillary nucleus (LM), and the MMn can be further subdivided into medial (MM) and lateral (ML) parts 11 , 12 . The ML receives inputs from the ventral tegmental nucleus of Gudden (VTg) and the medial prefrontal cortex (mPFC) which are the critical sleep-wake centers 13 . However, how the ML mediates sleep-wake cycles remains largely unknown. Magnetic resonance imaging (MRI) has demonstrated that the MB volumes show a marked volume reduction in obstructive sleep apnea (OSA) patients 14 . Therefore, like in the OSA, we hypothesized that ML neuronal activities are organized in PD sleep disorders to regulate the sleep-wake circles. Here, by applying radiomic analysis based on MRI data, we demonstrated MB structural features associated with sleep disturbances in PD patients. We further identified that ML neurons were recruited in sleep-wake regulation via a combination of c-Fos staining, in vivo single-unit recording, and DREADD-mediated chemical genetics. Also, we demonstrated that chemogenetic activation of ML neurons could rescue sleep disturbances in PD mice. Together, our data reveal that therapeutic potential of targeting ML neurons represents a promising strategy to alleviate sleep disturbances associated with PD, which address critical gaps in our understanding of the neural mechanisms underlying one of the most disabling non-motor symptoms of PD. Results Decoding Sleep Disorders in Parkinson's Disease through Radiomic Analysis of the MB. To investigate the correlation between the MB and sleep disorders in patients with Parkinson's disease, we performed radiomics analysis based on MRI data (Fig. 1 a). From T1-FLAIR and T2-PROPELLER sequences after standardized preprocessing, including intensity normalization, resampling, and manual region-of-interest (ROI) delineation, we selected features spanned first-order statistics, texture matrices (GLDM, GLSZM, GLRLM), and NGTDM, capturing heterogeneity, structural complexity, and intensity distribution within the MB (Fig. 1 b). We selected optimal regularization parameter (λ) determined via 10-fold cross-validation, with the minimum mean squared error (MSE) observed at λ ≈ 0.1 (Fig. 1 c). As λ increased, the coefficients of less informative features gradually shrank to zero, indicating effective feature selection (Fig. 1 d). Also, we found that ten features, including InterquartileRange.1, Imc.1, MCC.1, MCC.2, Kurtosis.3, ZoneEntropy.3, ZoneEntropy.5, SmallAreaLowGrayLevelEmphasis.7, Strength.8, and LargeDenpedenceHighGrayLevelEmphasis.7 were retained in the final model, and Pearson correlation coefficients among the selected features revealed moderate to strong correlations between several pairs (Fig. 1 e). Notably, InterquartileRange.10 and MCC.9 exhibited the highest absolute coefficients (Fig. 1 f). Next, we extracted and selected ten radiomic features using Lasso regression, which demonstrated robust predictive power for classifying PD patients with significant sleep disorders (area under the curve, AUC = 0.78) (Fig. 1 g). Crucially, our analysis revealed that integrating information from both T1-FLAIR and T2-PROPELLER sequences significantly enhanced classification accuracy (85%) compared to using either sequence alone (60% and 75%, respectively) (Fig. 1 h). This improvement highlights the complementary value of multi-modal imaging in capturing distinct aspects of tissue pathology within small, deep brain structures. Together, these results indicate that structural alterations in the MB, as revealed by multi-modal radiomic features, are closely associated with sleep disturbances in Parkinson's disease. This identifies the MB as a novel imaging biomarker and potential therapeutic target for addressing sleep pathology in PD patients. ML neurons in sleep regulations. The MB is composed of medial mammillary nucleus (MMn), lateral mammillary nucleus (LM), where the MMn can be further divided into the medial part (MM) and the lateral part (ML). To identify which part of the MB involved in sleep-active neurons, we measured c-Fos activity in a recovery sleep after deprivations (RS) paradigm. Sleep deprivation was induced using the fondling method, which maintains wakefulness via gentle handling without inducing stress (Fig. 2 a). Then, we designed three experimental paradigms: control mice (no sleep manipulation), sleep-deprived mice (6 h of sleep deprivation, SD) and mice undergoing recovery sleep (2 h of recovery sleep following 6 h of sleep deprivation, RS) (Fig. 2 b). To assess neural activity, we examined c-Fos-positive (c-Fos+) cells in the MB, and found that the number of c-Fos+ cells in the ML was significantly increased in RS mice compared to both control and SD groups (Fig. 2 c). Next, to explore how ML neurons are engaged in sleep-wake state, we employed in vivo single-unit and EEG/EMG recordings simultaneously (Figure S1 a). We confirmed the electrode locations in all recorded mice (Figure S1 b and Figure S2a). We classified neurons based on their activity profiles across wakefulness, NREM sleep, and REM sleep using a state-preference index. As shown in Figure S1 c, a scatter plot of 231 recorded neurons plotted the Wake-NREM activity ratio (W-NR)/(W + NR) against the REM-NREM activity ratio (R-NR)/(R + NR). Five distinct neuronal subtypes were identified: REM-active (pink), REM/Wake-active (salmon), Wake-active (green), NREM-active (orange), and neurons with no state preference (gray). The distribution revealed clustered patterns, reflecting specialized activity preferences for each sleep-wake state. We further illustrated the proportion of each subtype: REM-active neurons comprised 11.69%, REM/Wake-active 16.02%, Wake-active 10.39%, NREM-active 2.16%, and 59.74% exhibited no clear state-related activity (Figure S1 d). Collectively, these data demonstrate that the ML contains diverse neuronal populations with distinct sleep-wake state-related activity profiles, uncovering the heterogeneity of neural circuits involved in sleep-wake regulation. Overall, our results reveal that distinct sleep-wake state-related neural activity patterns in the ML and identify multiple neuronal subtypes differentially engaged in wakefulness, NREM, and REM sleep. ML neurons are sufficient to promote REM and NREM sleep. The observation that ML neurons were recruited during sleep-wake rhythm suggests these neurons might be functionally important. To test this, we unilaterally injected AAV carrying excitatory Gq-coupled DREADDs (rAAV-hSyn-hM3D(Gq)-EGFP) into the ML of C57BL/6J mice (Fig. 3 a and Figure S2b). Following a two-week recovery period after viral injection, we implanted EEG/EMG electrodes, and treated animals with CNO (1 mg/kg) and saline (NS) administration while recording sleep behavior (Fig. 3 b). Strikingly, during the light phase (9:00–12:00 AM), chemogenetic activation of ML neurons significantly increased total sleep duration (Fig. 3 c), primarily driven by an increase in NREM sleep (Fig. 3 d). REM sleep also showed a significant increase (Fig. 3 e), while wakefulness duration was markedly reduced (Fig. 3 f). These results indicate that ML neuron activation promotes sleep across both NREM and REM states during the light phase. Meanwhile, during the dark phase (19:30 − 22:30 PM), which corresponds to the rodent’s natural active period, ML activation still induced a significant increase in total sleep (Fig. 3 g) and NREM sleep (Fig. 3 h), with no significant change in REM sleep (Fig. 3 i), but a reduction in wakefulness (Fig. 3 j). Notably, the magnitude of sleep promotion was greater during the light phase, suggesting circadian modulation of ML-mediated sleep control. In summary, these data indicate that chemogenetic activation of ML neurons potently promotes sleep, particularly NREM and REM, across both light and dark phases, with concomitant suppression of wakefulness-demonstrating a critical role for the ML in regulating sleep-wake balance in mice. ML neurons are Required for REM and NREM sleep. To further study the necessity of ML neurons in sleep-wake regulation, we employed chemogenetic silencing using rAAV-hSyn-hM4D(Gi)-EGFP to selectively inhibit the activity of ML neurons. Viral vectors were bilaterally injected into the ML, with successful transduction confirmed by EGFP expression localized to the ML (Fig. 4 a and Figure S2b). After two weeks of recovery, we implanted EEG/EMG electrodes, and monitored sleep behavior during both light and dark phases via administration of CNO (1 mg/kg) and saline (NS) (Fig. 4 b). We found that chemogenetic inhibition of ML neurons significantly reduced total sleep duration (Fig. 4 c), NREM sleep (Fig. 4 d) and REM sleep (Fig. 4 e) during the light phase. Conversely, wakefulness duration was markedly increased in the same phase (Fig. 4 f). Interestingly, during the dark phase, chemogenetic inhibition of ML neurons led to a significant reduction in total sleep (Fig. 4 g) and NREM sleep (Fig. 4 h), along with a reduction in REM sleep (Fig. 4 i). However, Wakefulness was significantly prolonged under CNO treatment (Fig. 4 j). Therefore, these data indicate that ML neurons are required for both NREM and REM states in the light and dark phases. ML Neurons Exhibit Alter Firing Dynamics in PD Mice with Impaired REM Sleep . To investigate neural mechanisms underlying sleep disturbances in PD, we generated a PD mouse model by unilateral injection of rAAV-CMV-SYN-A53T into the substantia nigra pars compacta (SNc) of C57BL/6 mice (Figure S3a). This approach induces progressive α-synuclein aggregation and dopaminergic neurodegeneration, recapitulating key pathological features of PD. To test the success of the PD model establishment, we conducted behavioral tests and histopathological analysis. Firstly, we performed the pole assay (Figure S3b). As expected, PD mice exhibited significantly prolonged crawling duration in, indicating impaired motor coordination and bradykinesia (Figure S3c). Next, we performed the cylinder test (Figure S3d). Notably, PD mice displayed marked asymmetry in forelimb use that left claw utilization was significantly reduced (Figure S3e), while right claw usage increased (Figure S3f), confirming contralateral motor impairments due to unilateral SNc degeneration. Then, we performed histological analysis to confirm robust dopaminergic pathology. Indeed, immunofluorescence staining revealed a dramatic loss of tyrosine hydroxylase-positive (TH+) neurons in the SNc of PD mice (Figure S3g, h). Also, TH+ fiber density in the caudate-putamen (CPU) was also significantly reduced (Figure S3i, g), demonstrating extensive striatal denervation consistent with PD pathophysiology. Altogether, these data indicate that the PD models are successfully established. How changes of firing dynamics of ML neurons in the PD mice is largely unknown. To test it, we performed in vivo single-unit recordings on ML neurons in PD mice during sleep-wake state (Fig. 5 a). We confirmed the electrode locations in all recorded mice (Fig. 5 b and Figure S2a). We recorded the activity of ML neurons (n = 252 units, 4 mice) in PD mice and compared the neuronal activity in different brain-state phases to the controls. Interestingly, we found that there were distinct functional clusters, including REM-active, REM/wake-active, wake-active, NREM-active, NREM-REM-active, and no state-dependent cycle (Fig. 5 c, d). Notably, in PD mice, the proportion of REM-active ML neurons was significantly increased compared to controls (Fig. 5 e), suggesting a pathological reorganization of ML neuronal activity toward REM-specific firing patterns. However, we found that the average firing rates of ML neurons in PD mice significantly reduced during the overall brain-state, wake and NREM phase. While the average firing rates during the REM phase were higher in PD mice than controls, indicating a selective hyperactivity in a subset of REM-active neurons that may reflect compensatory or pathological circuit reorganization (Fig. 5 f). Overall, these findings suggest that while ML neuronal activity is broadly suppressed in PD, REM-associated neurons exhibit enhanced firing, potentially contributing to REM sleep disturbances commonly observed in Parkinson’s disease. Chemogenetic Activation of ML Neurons Promotes Sleep in PD Mice. To investigate the role of ML neurons in sleep regulation within a PD model, we employed a chemogenetic approach using the excitatory DREADD receptor hM3Dq (Fig. 6 a). Using a similar viral injection strategy, we generated a PD mouse model by unilateral injection of rAAV-CMV-SYN-A53T into the SNc of C57BL/6 mice (Fig. 6 a and Figure S2b). To characterize the sleep-wake architecture in PD mice, we performed EEG/EMG recordings in freely moving PD mice and control littermates (Figure S4a). We found that PD mice exhibited a significant increase in wakefulness percentage compared to controls, indicating heightened arousal or impaired sleep consolidation (Figure S4b). Concurrently, the proportion of time spent in NREM sleep was significantly reduced, suggesting disrupted slow-wave sleep homeostasis (Figure S4c). Most strikingly, REM sleep was profoundly suppressed in PD mice (Figure S4d), consistent with REM sleep disturbances commonly observed in human PD patients. Furthermore, PD mice displayed a dramatic increase in the number of sleep-wake transitions (Figure S4e), reflecting severe sleep fragmentation and instability. These data indicate that PD mice exhibit increased wakefulness, reduced NREM and REM sleep, and profound sleep fragmentation, revealing a comprehensive disruption of sleep-wake homeostasis that parallels core clinical features of PD. It is unknown how ML neurons regulate the sleep patterns in the PD mice. To test it, we performed EEG/EMG recordings during both light (9:00–12:00 AM) and dark (19:30 − 22:30 PM) phases following intraperitoneal injection of either CNO or NS (Fig. 6 a). We found that CNO treatment significantly increased total sleep duration compared to NS control (Fig. 6 b), driven primarily by an increase in NREM and REM sleep duration during the light phase (Fig. 6 c, d). Consistently, wakefulness duration was significantly reduced after CNO administration (Fig. 6 e), indicating that ML neuron-activation promotes sleep and suppresses arousal. Alao, in the dark phase, similar effects were observed: total sleep duration was significantly increased (Fig. 6 f), NREM and REM sleep duration trended upward (Fig. 6 g, h), while wakefulness duration was significantly decreased (Fig. 6 i), confirming that ML activation enhances sleep propensity across circadian phases. Notably, chemogenetic activation of ML neurons did not rescue motor asymmetry or enhance forelimb use in PD mice (Figure S5). Altogether, these findings demonstrate that chemogenetic activation of ML neurons increases sleep duration and reduces wakefulness in PD mice, without altering motor function, identifying the ML as a key regulator of sleep architecture and a potential therapeutic target for sleep disturbances in PD. Discussion Our study shows that the ML as the lateral part of the MB is a novel and critical hub for sleep-wake regulation. Furthermore, we identify the dysfunction of ML neuronal dynamics as a substrate for sleep disturbances in PD and demonstrate the therapeutic potential of targeting this nucleus. Altogether, these data advance the understanding of sleep circuit pathophysiology in PD and propose a new deep brain target for intervention. PD is a progressive neurodegenerative disorder characterized by motor symptoms such as bradykinesia, rigidity, and resting tremor 15 . Patients with PD also have non-motor symptoms, specifically sleep disturbances, which have emerged as equally devastating manifestations and received limited attention 1 . However, the precise neural substrates underlying sleep dysfunction in PD remain unknown. Our radiomic analysis of multi-modal MRI data from PD patients reveals that structural heterogeneity within the MB is closely associated with the presence and severity of sleep disorders, with the combined use of T1-FLAIR and T2-PROPELLER sequences yielding a high classification accuracy for identifying PD patients with significant sleep disorders (Fig. 1 ). This aligns with growing evidence that non-motor symptoms in PD are rooted in early pathology of limbic and hypothalamic circuits beyond the SNc 2 , 16 . The MB, a key component of this network, has long been implicated in memory and spatial navigation 6 – 10 , but its role in sleep physiology and dysfunction in PD has remained understudied. Our identification of ten optimal radiomic features that robustly predict PD sleep disturbances fills this gap, positioning the MB as a novel imaging biomarker for a highly disabling non-motor symptom of PD (Fig. 1 e, f). Notably, previous MRI studies have demonstrated volume reductions in the MB in patients with OSA 14 , a condition characterized by recurrent sleep fragmentation and daytime sleepiness that overlap with PD sleep disturbances. This convergence suggests that the MB structural integrity may be a shared vulnerability factor across sleep disorders with distinct etiologies. Our radiomic approach captured subtle structural heterogeneity, provides a more sensitive tool for detecting MB pathology in PD, and offers potential clinical utility for early identification of patients at risk for severe sleep disturbances 17 . The MB is composed of medial mammillary nucleus (MMn), lateral mammillary nucleus (LM), where the MMn can be further divided into the medial part (MM) and the lateral part (ML). However, it is unclear that which part of the MB is involved in the regulation of sleep-wake cycle. Overall, we found that the activity of ML neurons significantly increased in RS mice (Fig. 2 ). The data indicate that ML neurons are recruited during sleep rebound, which reflects the homeostatic regulation of sleep 18 . Specifically, we identified diverse neuronal subtypes in the ML with distinct sleep-wake state-related activity profiles, including REM-active, REM/Wake-active, Wake-active, NREM-active, and non-state-preferring neurons (Figure S1 ), which establishes the ML as a functional mosaic within the sleep-regulatory network. This heterogeneity is consistent with its complex afferent inputs from key sleep-wake centers like the VTg and the mPFC 13 , and efferent projections to the anterior thalamus and reticular formation 19 , 20 . Our findings extend this principle to the ML, showing that distinct subtypes of ML neurons are differentially engaged in wakefulness, NREM, and REM sleep. Therefore, this functional specialization suggests the ML is not a monolithic structure but a circuit node where distinct neuronal subpopulations may integrate limbic and homeostatic signals to gate specific sleep states 21 , 22 . Moreover, we demonstrate a potent and bidirectional causal role for ML neurons in sleep-wake control via chemogenetic approaches. Activation of ML neurons significantly increased total sleep duration, primarily driven by increases in NREM and REM sleep, while suppressing wakefulness—effects observed across both light and dark phases (Fig. 3 ). Notably, the magnitude of sleep promotion was greater during the light phase, suggesting circadian modulation of ML function. Conversely, inhibition of ML neurons induced insomnia-like phenotypes, reducing total sleep, NREM, and REM sleep (Fig. 4 ). Overall, our findings confirm that the activity of ML neurons is required for the maintenance of normal sleep architecture, particularly for NREM and REM sleep. The consistent effects of ML manipulation across light and dark phases suggest that ML neurons play a tonic role in sleep regulation, rather than being restricted to a specific circadian window. This is in contrast to some sleep-regulating neurons, such as SCN neurons, which exhibit strong circadian rhythmicity in their activity 23 . Therefore, these data shown that the tonic role of ML neurons may reflect their function as a key node in integrating homeostatic sleep pressure with circadian cues to maintain sleep-wake balance. However, the precise neurotransmitter phenotype of sleep-modulating ML neurons remains unknown, warranting cell-type-specific manipulations. Further studies are needed to resolve this issue. Importantly, consistent with clinical observations in PD patients 2 , 24 , we found that PD mice exhibited severe sleep disturbances, including increased wakefulness, reduced NREM and REM sleep, and profound sleep fragmentation (Figure S4). Specially, we observed a profound disruption of ML neuronal firing patterns. Also, the broad reduction in average firing rates during wakefulness and NREM sleep aligns with general neuronal dysfunction in neurodegenerative disease 25 . Nevertheless, the paradoxical increase in the proportion and firing rate of REM-active neurons is a striking finding. Together, this dual pattern of ML neuronal dysfunction that there is broad hypoactivity alongside REM-active neuron hyperactivity may contribute to PD-related sleep disturbances. Moreover, this pathological reorganization may represent a compensatory response to the severe REM sleep suppression observed in PD mice and in patients 26 . Alternatively, this hyperactivity could be a pathological consequence of α-synuclein aggregation or dopaminergic degeneration, further disrupting REM sleep regulation 27 , 28 . For example, α-synuclein pathology is known to spread through neural circuits via prion-like mechanisms 29 , and it is possible that α-synuclein accumulation in the ML disrupts the balance of neuronal activity, favoring REM-active subtypes. Future work should be investigated. The most translational finding of our study is that chemogenetic activation of ML neurons in PD mice effectively rescued key sleep abnormalities, increasing total, NREM, and REM sleep duration, while reducing wakefulness and sleep fragmentation, which were observed during both light and dark phases (Fig. 6 ). Importantly, this sleep-promoting effect was not accompanied by improvements in motor function (Figure S5), indicating that ML neuron activation specifically targets sleep circuits without modulating motor pathways. This specificity is clinically relevant, as current PD therapies primarily target motor symptoms and often exacerbate sleep disturbances 30 . Notably, the ability of ML activation to rescue both NREM and REM sleep disturbances in PD mice, while REM sleep disturbances, such as RBD, are among the most disabling and treatment-refractory non-motor symptoms of PD 31 . RBD, which is characterized by the loss of REM sleep atonia and the acting out of dreams, is associated with more rapid disease progression 32 . Our data show that ML activation increases REM sleep in PD mice suggests that targeting ML neurons could provide a novel therapeutic strategy for RBD and other REM-related sleep disturbances in PD. This is supported by our observation that ML neurons contain a large population of REM-active subtypes, whose hyperactivity in PD may reflect a failed compensatory attempt to restore REM sleep—an attempt that could be augmented by targeted activation of ML neurons. Future work should explore mechanisms that may underlie the sleep-rescuing effect of ML activation in PD mice. In summary, we identify the ML as a critical regulator of sleep-wake states and demonstrate its pathological engagement in PD-related sleep disturbances. By bridging clinical imaging with causal mechanistic studies in animals, we propose that the structural and functional integrity of the ML is essential for normal sleep architecture and is disrupted in PD. Moreover, our findings that modulating ML activity can rectify sleep deficits without impacting motor function, highlighting the potential of targeting ML neurons for the treatment of PD non-motor symptoms. Thus, these data fill critical gaps in our understanding of the neural mechanisms underlying PD sleep disturbances and provide a foundation for the development of targeted therapies for this disabling feature of PD. Methods Animals. All animals experimental procedures were approved by the Hubei Provincial Animal Care and Use Committee and complied with the experimental guidelines of the Animal Experimentation Ethics Committee of Hubei University of Medicine (Approval code: 2025-16). C57BL/ 6 mice (12–16 weeks old) were purchased from Shulaibao (Wuhan) Biotechnology Co., Ltd., and were group-housed and bred the laboratory animal centre of Hubei University of Medicine under a constant temperature (20 ± 2℃), humidity (50%–60%), illumination intensity (15–20 lux) and 12-hr light/dark cycle (7:00 am to 19:00 pm). Virus. rAAV-SYN(BrainVTA)-SNCA(A53T)-WPRE-bGHpA, rAAV-hSyn-hM3D(Gq) -EGFP-WPRE-hGH polyA, and rAAV-hSyn-hM4D(Gi) -EGFP-WPRE-hGH polyA were purchased from BrainVTA (Wuhan) Co., Ltd. All viruses were stored in aliquots at -80℃. Radiomics analysis. Ethical approval for this study was obtained from Ethics Committee of Shiyan Taihe hospital, Hubei University of Medicine (Approval code: 2025KS26). This study was performed in accordance with the Declaration of Helsinki , and was waived the requirement for written informed consent. 150 PD accompanied by sleep disorders and 100 controls were recruited in the study, who had undergone 3.0-T MRI (GE Healthcare, Discovery 750w, Wisconsin, USA) imaging including T1WI-Flair, T2-FLAIR, and T2WI-PROPELLER sequences. Radiomics analysis was performed as previously described 33 – 35 . The inclusion criteria for PD were as follows: (1) Patients met the diagnostic criteria for idiopathic Parkinson’s disease. (2) Patients had not undergone deep brain stimulation (DBS) surgery and were required to discontinue dopaminergic medications for one week prior to participation. (3) Patients had no history of central nervous system disorders—including intracerebral hemorrhage, cerebral infarction, Alzheimer’s disease, amyotrophic lateral sclerosis, or Huntington’s disease—or other neurological or psychiatric conditions. (4) Patients were aged between 55 and 75 years inclusive. The exclusion criteria for PD were as follows: (1) Presence of image artifacts. (2) Incomplete clinical or imaging data. (3) History of neurological or psychiatric disorders. (4) Prior exposure to antipsychotic medications or other psychoactive drugs. The criteria for controls (HC) were as follows: (1) Age-matched to the experimental group, with no restriction on sex. (2) No history of neurological disorders, no large-area cerebral infarction, and no evidence of brain atrophy. MRI image preprocessing. To minimize variability arising from differences in imaging acquisition parameters, MRI image preprocessing was performed using the MR Radiomics Platform (MRP). All images were resampled to an isotropic voxel size of 0.50 × 0.50 × 3.00 mm³. Rigid registration of T1-FLAIR and T2-FLAIR sequences to the T2-PROPELLER sequence was conducted using a six-degree-of-freedom rigid-body transformation with a mutual information similarity metric. Intensity normalization was applied to standardize MR signal intensities across all subjects into a consistent range for each imaging modality. Region-of-Interest (ROI) delineation. The mammillary bodies were manually segmented slice-by-slice on axial views of three MRI sequences using 3D Slicer software. ROIs were carefully delineated by trained raters blinded to clinical information. Feature extraction. Radiomic features were extracted from both original and wavelet-filtered images for each sequence using the open-source PyRadiomics platform, yielding a total of 851 features per sequence. These features comprised14 three-dimensional shape-based features, 18 first-order intensity statistics, 24 gray-level co-occurrence matrix (GLCM) features, 14 gray-level dependence matrix (GLDM) features, 16 gray-level run-length matrix (GLRLM) features, 16 gray-level size zone matrix (GLSZM) features, and 5 neighboring gray-tone difference matrix (NGTDM) features. Given five sequences per subject, a total of 5,255 radiomic features were initially extracted. Detailed definitions of these features are available on the PyRadiomics documentation website. Feature selection. Feature distributions were assessed for normality using the Kolmogorov–Smirnov (K–S) test, and homogeneity of variance was evaluated using Levene’s test. Depending on the distribution, either Student’s t -test or non-parametric alternatives were used to identify features significantly differentiating high- versus low-expression groups. To reduce dimensionality and select the most informative features, least absolute shrinkage and selection operator (LASSO) regression with 10-fold cross-validation was applied exclusively on the training set. The optimal regularization parameter (lambda) was determined based on the minimum cross-validated error criterion. This pipeline was repeated to investigate the impact of multi-sequence MRI on feature performance. Feature fusion. Fusion was implemented at both the feature level and the decision level. At the feature level, the most common approach—concatenation—was employed: features extracted from multiple sequences were combined into a single high-dimensional feature matrix, which was then subjected to the feature selection pipeline. Alternatively, features from individual sequences were first selected separately and then fused. A canonical correlation analysis (CCA)-based fusion method was also explored, which identifies linear transformations of two input feature sets to maximize their inter-set correlation. At the decision level, ensemble strategies—including majority voting, weighted averaging, and stacking with a meta-classifier—were used to integrate predictions from multiple models and enhance predictive stability. Model evaluation. The dataset was randomly stratified into independent training (70%) and testing (30%) sets. Baseline clinical and ROI characteristics were compared between the two sets to ensure no statistically significant imbalances. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), along with 95% confidence intervals (CIs), accuracy, sensitivity, and specificity. The classifier yielding the highest AUC for each task was selected as optimal. Statistical significance of AUC improvements was assessed using DeLong’s test. All statistical analyses were performed using Python (v3.8) and R (v4.1.3), with a two-sided P < 0.05 considered statistically significant. Stereotaxic injections. Mice were fixed in a stereotaxic injection frame (RWD Life Science Co., Ltd., China) after anesthetized (0.1 ml/10 g) by 1% pentobarbital sodium. 0.20 µl of rAAV vectors were delivered with a glass micropipette connected KDS-120 Pressure Micro Injector (Harvard Bioscience, Inc., USA). Coordinates used for injection were as follows: ML, − 2.92 mm from bregma, 0.55 mm lateral from midline, and 5.25 mm vertical from the cortical surface; SNc, − 3.08 mm from bregma, 0.7 5 mm lateral from midline, and 4.7 mm vertical from the cortical surface. Immunohistochemistry. After anaesthetization, mice were perfused with phosphate buffered saline (PBS), followed by 4% paraformaldehyde (PFA, w/v in PBS). Brains were dissected and postfixed in 4% PFA at 4℃ overnight, and then were cut into 40 µm thick coronal sections by a vibratome (Leica VM 1000). Brain sections were incubated in blocking solution (PBS containing 0.1% Triton X-100 and 10% normal goat serum) with anti-c-Fos (1:12,000, rabbit, Synaptic System, 226003) for 40 h or anti-TH (1:500, rabbit, Millipore, ab6211) overnight at 4℃. Then, sections were washed three times with PBS, incubated with secondary antibodies (goat anti-rabbit, Alexa Fluor 488 conjugate 1:500, Thermo Fisher Scientific, A-11008) for 2 h at room temperature, mounted and imaged. For examining the virus expression and electrode locations, brain slices were washed three times with PBS and incubated for 30 min with DAPI at room temperature. Images were captured with Leica SP8 MP confocal microscope. Chemogenetic manipulation. C57BL/6J mice expressing hM3Dq or hM4Di in the ML were injected intraperitoneally 30 min prior to recording at 8:30/19:00 with saline (NS, day 1) and clozapine-N-oxide (CNO, 1 mg/kg, Enzo Life Sciences Inc., Exeter, UK, day 2). Recordings were carried out for 3 h during 9:00–12:00 and 19:30 − 22:30, respectively. Sleep Deprivation and Sleep Recovery Experiments. In the sleep deprivation and recovery experiments, mice were randomly assigned to three groups: control group (Ctrl), sleep-deprived group (SD), and sleep recovery group (RS). Each group was individually housed in a testing cage, which was placed inside an acoustic/light-isolated enclosure to minimize external disturbances. Prior to the sleep deprivation procedure, all mice were acclimatized to the arena for 48 h under a standard 12-h light/dark cycle (lights on from 7:00 a.m. to 7:00 p.m.). SD began 7:00 a.m. and was maintained for 6 h by gently stroking the mice’s ears and tail with a soft brush whenever signs of behavioral sleep were observed. Care was taken throughout the procedure to avoid startling the animals-no loud noises or mechanical vibrations were introduced during the intervention. Immediately after the 6 h SD, SD and Ctrl mice were deeply anesthetized and transcardially perfused for brain collection. RS mice were allowed 2 h of undisturbed sleep following deprivation before undergoing perfusion and brain extraction. Pole test. The pole test was performed as previously described 36 . A custom-made wooden pole (diameter: 18 cm; height: 50 cm) was used, with a small wooden ball wrapped in gauze (to provide grip) fixed at the top. During testing, each mouse was placed head-up at the top of the pole. The time from movement onset to reaching the bottom of the pole (crawling duration) were recorded. Each mouse was tested five times with a 1-min interval for each trial. The average values across the five trials were used for analysis. Cylinder test. The behavioral setup was performed as previously described 37 . The test was administrated at the same period (14:00–18:00 pm) and performed in a glass cylinder (20 cm × 40 cm). Prior to test, mice were habituated to the testing room for at least 30 min to minimize stress-induced behavioral artifacts. Each mouse was allowed to freely explore the cylinder for 10 min and recorded the number of times it reared up and touched the cylinder wall. The number of times included scored for the left, right, or both paws that touched the wall. In vivo single-unit recordings and data analysis. In vivo single-unit recordings and data analyses were performed as described in previous studies 38 . Briefly, a 16-channel electrode array constructed from 25.4-µm formvar-insulated nichrome wire (cat no. 761500, A-M Systems, USA) was affixed to a screw-driven microdrive, and the impedance of the electrode tips was measured to be 700–800 kΩ. Mice were implanted with the 16-channel electrodes targeting the ML. Neuronal recordings were performed using NeuroLego System (Jiangsu Brain Medical Technology Co.ltd), band-pass filtered at 300–6,000 Hz, sampled at 30 kHz. During the recordings, an amplitude threshold of 50 µV was applied to exclude background noise. Electrodes were advanced incrementally in steps of 62.5 µm each recording session. Recordings data were collected at the depth where electrode tips were located in the ML. All data analyses were performed using MATLAB 2014b (The Mathworks, Inc., Natick, Massachusetts, USA). Manual spike sorting was performed on the basis of the three principal components of spike waveforms and waveform energy features using MATLAB toolbox (MClust 4.4). Isolation distance (> 20) and L-ratio (< 0.1) were calculated to identify neurons, and only units with inter-spike intervals shorter than 2 ms accounting for < 1% of total spikes were included analyzed. Additionally, cross-correlation comparisons were performed to exclude duplicate units in a session. For the simultaneous multi-channel and EEG/EMG recordings, data were collected for 1 h. The neuronal activity was aligned to wakefulness, non-rapid eye movement sleep (NREM) and rapid eye movement sleep (REM) states as classified by EEG spectral features and EMG signals. To quantify the relative firing rates of individual neurons in different brain states, we computed and plotted REM-NREM modulation ((R REM - R NREM ) / (R REM + R NREM )) versus Wake-NREM modulation ((R Wake - R NREM ) / (R Wake + R NREM )), where R represented the mean firing rate in each state. Neurons exhibiting modulation values falling outside the light-blue shaded region, denoting to a greater than two-fold change in firing rate (|modulation| > 0.33), were classified as showing “state-dependent” activity 39 , 40 . EEG/EMG recordings and analysis. Electroencephalographic (EEG)/electromyographic (EMG) recordings and data analyses were performed as described in previous studies 38 , 40 . In brief, mice were implanted with two stainless steel screw electrodes served as EEG electrodes (recording location: AP = 1.75 mm, ML = 0.4 mm and reference location: cerebellum) together with two EMG wire electrodes, all of which were pre-soldered to a four-pin connector. The EMG electrodes were inserted into the bilateral neck muscles. After 5 d recovery, mice were connected to recording headstages via flexible recording cables and adapted for at least 2 d prior to data collection. Signals were recorded using a Microelectrode AC Amplifier Model 1800 (A-M Systems, USA), filtered (0.1–500 Hz for EEG and 10–500 Hz for EMG recording), and digitized at 250 Hz by Intracept Chart software. A notch filter was applied at 50 Hz by the amplifier. Vigilance states (wakefulness, NREM sleep, REM sleep) were classified for consecutive 4 s epochs using a custom-written MATLAB algorithm, based on EEG/EMG waveforms and power spectra. Wakefulness was identified by desynchronized, low-amplitude EEG and high EMG activity; NREM sleep was characterized by synchronized, high-amplitude, low-frequency EEG activity (0.5–4 Hz) and low EMG activity relative to wakefulness; REM sleep was defined as desynchronized EEG with high power at theta frequencies (6–9 Hz) and low EMG activity, reflecting muscle atonia. Statistical analysis. All statistical analysis was performed using MATLAB 2014b and GraphPad Prism 10. Experimental conditions were randomly allocated, and behavioral data were analyzed using a double-blind approach. Statistical analyses comprised the Wilcoxon rank-sum test, two-tailed paired and unpaired t-tests, one-way ANOVA test and chi-squared test, with P values derived from each respective test. The significance was denoted as: * P < 0.05; ** P < 0.01; *** P < 0.001; **** P < 0.0001. See the figure legends. Declarations Competing interests The authors declare no competing interests. Author Contribution Q. K., P. X. and W. M. performed most of the experiments. Q. K. and P. X. performed *in vivo* single-unit and EEG/EMG recordings. W. C., P. X. and B. Y. performed Radiomics analysis. Q. K., P. X. and W. M. performed behavioral manipulation experiments and evaluations. Q. K., Y. W., D. S., W. D., X. Y. and Y. X. performed immunohistochemistry. H. Y., C. K. and W. C. designed and supervised the study. H. Y., Q. K. and C. K. wrote the manuscript. All authors read and approved the final manuscript. Acknowledgement This study was supported by National Natural Science Foundation of China (Original Exploration Program, 32250018 and National Natural Science Foundation of China，82501490), Hubei Provincial Natural Science Foundation (2024AFB720 and JCZRYB202500348), the Scientific and Technological Project of Shiyan City of Hubei Province (25Y002), and Cultivating Project for Young Scholar at Hubei University of Medicine (2022QDJZR010). We thank the Biomedical Research Institute of Hubei University of Medicine for providing an experimental platform. Data Availability All data supporting the findings of this study are provided within the paper and its supplementary information. All additional information will be made available upon reasonable request to the authors. References Iranzo, A. et al. Neurodegenerative disease status and post-mortem pathology in idiopathic rapid-eye-movement sleep behaviour disorder: an observational cohort study. Lancet Neurol. 12, 443–453 (2013). Chaudhuri, K. R., et al , & National Institute for Clinical Excellence. Non-motor symptoms of Parkinson’s disease: diagnosis and management. Lancet Neurol. 5, 235–245 (2006). Iranzo, A., et al . M. Sleep and sleep disorders in people with Parkinson’s disease. Lancet Neurol. 23, 925–937 (2024). Li, J. et al. Sleep disorders in Parkinson’s disease: evaluating pharmacological treatments through systematic review and meta-analysis. BMC Neurol. 25, 504 (2025). Lima, M. M. S. Sleep disturbances in Parkinson’s disease: the contribution of dopamine in REM sleep regulation. Sleep Med. Rev. 17, 367–375 (2013). Vann, S. D. & Nelson, A. J. D. The mammillary bodies and memory: more than a hippocampal relay. Prog. Brain Res. 219, 163–185 (2015). Roy, D. S. et al. Distinct Neural Circuits for the Formation and Retrieval of Episodic Memories. Cell 170, 1000–1012.e19 (2017). Jiang, J., et al . Mammillary body regulates state-dependent fear by alternating cortical oscillations. Sci. Rep. 8, 13471 (2018). Ruan, M., et al . Bi-Directional Theta Modulation between the Septo-Hippocampal System and the Mammillary Area in Free-Moving Rats. Front. Neural Circuits 11, 62 (2017). Dillingham, C. M. & Vann, S. D. Why Isn’t the Head Direction System Necessary for Direction? Lessons From the Lateral Mammillary Nuclei. Front. Neural Circuits 13, 60 (2019). Hou, Y. et al. Topographical organization of mammillary neurogenesis and efferent projections in the mouse brain. Cell Rep. 34, 108712 (2021). Żakowski, W. & Zawistowski, P. Neurochemistry of the mammillary body. Brain Struct. Funct. 228, 1379–1398 (2023). Dillingham, C. M., et al . How do mammillary body inputs contribute to anterior thalamic function? Neurosci. Biobehav. Rev. 54, 108–119 (2015). Kumar, R. et al. Reduced mammillary body volume in patients with obstructive sleep apnea. Neurosci. Lett. 438, 330–334 (2008). Zhang, Y. et al. Targeting thalamic circuits rescues motor and mood deficits in PD mice. Nature 607, 321–329 (2022). Poewe, W. Non-motor symptoms in Parkinson’s disease. Eur. J. Neurol. 15 Suppl 1, 14–20 (2008). Roberts, A. G. et al. Technical Feasibility of Quantitative Susceptibility Mapping Radiomics for Predicting Deep Brain Stimulation Outcomes in Parkinson Disease. Neurosurgery https://doi.org/10.1227/neu.0000000000003721 (2025). Lu, J., et al . A putative flip-flop switch for control of REM sleep. Nature 441, 589–594 (2006). Jankowski, M. M. et al. The anterior thalamus provides a subcortical circuit supporting memory and spatial navigation. Front. Syst. Neurosci. 7, 45 (2013). Aggleton, J. P. et al. Hippocampal-anterior thalamic pathways for memory: uncovering a network of direct and indirect actions. Eur. J. Neurosci. 31, 2292–2307 (2010). Weber, F. & Dan, Y. Circuit-based interrogation of sleep control. Nature 538, 51–59 (2016). Scammell, T. E., et al . Neural Circuitry of Wakefulness and Sleep. Neuron 93, 747–765 (2017). Tu, H.-Q. et al. Rhythmic cilia changes support SCN neuron coherence in circadian clock. Science 380, 972–979 (2023). Postuma, R. B. et al. MDS clinical diagnostic criteria for Parkinson’s disease. Mov. Disord. Off. J. Mov. Disord. Soc. 30, 1591–1601 (2015). Harris, S. S. et al. Alzheimer’s disease patient-derived high-molecular-weight tau impairs bursting in hippocampal neurons. Cell 188, 3775–3788.e21 (2025). Boeve, B. F. REM sleep behavior disorder: Updated review of the core features, the REM sleep behavior disorder-neurodegenerative disease association, evolving concepts, controversies, and future directions. Ann. N. Y. Acad. Sci. 1184, 15–54 (2010). W, S., S, Z. & Dj, S. Striatal synaptic adaptations in Parkinson’s disease. Neurobiol. Dis. 167, (2022). Marino, G. et al. Intensive exercise ameliorates motor and cognitive symptoms in experimental Parkinson’s disease restoring striatal synaptic plasticity. Sci. Adv. 9, eadh1403 (2023). Surmeier, D. J., et al . Selective neuronal vulnerability in Parkinson disease. Nat. Rev. Neurosci. 18, 101–113 (2017). Hunt, J. et al. Sleep and circadian rhythms in Parkinson’s disease and preclinical models. Mol. Neurodegener. 17, 2 (2022). During, E.H., et al . Symptomatic treatment of REM sleep behavior disorder (RBD): A consensus from the international RBD study group - Treatment and trials working group. Sleep Med. 132, 106554 (2025). D'Amelio, M., et al . Long-term survival of Parkinson’s disease: a population-based study. J Neurol. 253, 33–37 (2006). Kang, J. J. et al. Combining quantitative susceptibility mapping to radiomics in diagnosing Parkinson’s disease and assessing cognitive impairment. Eur. Radiol. 32, 6992–7003 (2022). Hu, X. et al. Multivariate radiomics models based on 18F-FDG hybrid PET/MRI for distinguishing between Parkinson’s disease and multiple system atrophy. Eur. J. Nucl. Med. Mol. Imaging 48, 3469–3481 (2021). Guan, X.-J. et al. A multiple-tissue-specific magnetic resonance imaging model for diagnosing Parkinson’s disease: a brain radiomics study. Neural Regen. Res. 17, 2743–2749 (2022). Jiang, P.-E. et al. Behavioral Assessments of Spontaneous Locomotion in a Murine MPTP-induced Parkinson’s Disease Model. JoVE . 143, 58653 (2019). Magno, L. A. V., Collodetti, M., Tenza-Ferrer, H. & Romano-Silva, M. A. Cylinder Test to Assess Sensory-motor Function in a Mouse Model of Parkinson’s Disease. Bio-Protoc. 9, e3337 (2019). Yu, H. et al. Periaqueductal gray neurons encode the sequential motor program in hunting behavior of mice. Nat. Commun. 12, 6523 (2021). Hua, R. et al. Calretinin Neurons in the Midline Thalamus Modulate Starvation-Induced Arousal. Curr. Biol. CB 28, 3948–3959 (2018). Xu, M. et al. Basal forebrain circuit for sleep-wake control. Nat. Neurosci. 18, 1641–1647 (2015). Additional Declarations No competing interests reported. Supplementary Files Supplementaryinformation.docx Cite Share Download PDF Status: Posted 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-9084382","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":630603585,"identity":"3e8ec9b8-5d58-48f1-a628-166d672aa806","order_by":0,"name":"Qizhou Ke","email":"","orcid":"","institution":"Hubei University of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Qizhou","middleName":"","lastName":"Ke","suffix":""},{"id":630603588,"identity":"a0b9bb3e-a8b2-4d04-a337-6cc1e026a6d7","order_by":1,"name":"Pengxiang Xu","email":"","orcid":"","institution":"Sinopharm Dongfeng General Hospital","correspondingAuthor":false,"prefix":"","firstName":"Pengxiang","middleName":"","lastName":"Xu","suffix":""},{"id":630603590,"identity":"d6037c8e-4b1a-49d6-84c2-cbaf4c97a890","order_by":2,"name":"Wenjie Meng","email":"","orcid":"","institution":"Hubei University of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Wenjie","middleName":"","lastName":"Meng","suffix":""},{"id":630603592,"identity":"640ad44c-7d12-4ae2-8905-ecad633da72d","order_by":3,"name":"Bo Yang","email":"","orcid":"","institution":"Hubei University of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Bo","middleName":"","lastName":"Yang","suffix":""},{"id":630603594,"identity":"a498b14e-2610-474d-b287-1bd66bbf2678","order_by":4,"name":"Yu Wang","email":"","orcid":"","institution":"Hubei University of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Yu","middleName":"","lastName":"Wang","suffix":""},{"id":630603596,"identity":"b97017aa-0f43-4e80-877c-4255c054fcd0","order_by":5,"name":"Dongsheng Sun","email":"","orcid":"","institution":"Hubei University of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Dongsheng","middleName":"","lastName":"Sun","suffix":""},{"id":630603598,"identity":"a0746abb-642c-4270-9b51-fd3bdce76394","order_by":6,"name":"Wanqi Ding","email":"","orcid":"","institution":"Hubei University of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Wanqi","middleName":"","lastName":"Ding","suffix":""},{"id":630603612,"identity":"7fd00981-ef93-4d04-8179-368a2b11f14e","order_by":7,"name":"Xiuqin Yu","email":"","orcid":"","institution":"Hubei University of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Xiuqin","middleName":"","lastName":"Yu","suffix":""},{"id":630603614,"identity":"d241fb34-11bb-4603-87ff-23ca8843db91","order_by":8,"name":"Yangqiao Xiao","email":"","orcid":"","institution":"Hubei University of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Yangqiao","middleName":"","lastName":"Xiao","suffix":""},{"id":630603620,"identity":"3edae458-d617-4e61-9e9a-c83b6ccd560c","order_by":9,"name":"Wen Chen","email":"","orcid":"","institution":"Hubei University of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Wen","middleName":"","lastName":"Chen","suffix":""},{"id":630603622,"identity":"70cc6007-9213-4fc9-84dc-2103c196f774","order_by":10,"name":"Changbin Ke","email":"","orcid":"","institution":"Hubei University of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Changbin","middleName":"","lastName":"Ke","suffix":""},{"id":630603627,"identity":"10e4c7df-5232-4a1a-94ad-975595810c19","order_by":11,"name":"Hong Yu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAwUlEQVRIiWNgGAWjYFACHhBhk2AA5rARryWNdC2HSdBicCP34OOCX+fzzMXOGDB8KDvMwD+7gZCWvGTjmX23iy1n5xgwzjh3mEHizgFCWnLMpHl7biduuJ1jwMzbdpjBQCKBKC3nIFr+Eq2F58cBiBZGYrRInnljbMzbkJy4c3ZawcGec+k8EjcIaOE7nmP4mOePXeJ26eSND36UWcvxzyCgReEAkGBsg3BAbB786oFAvgFE/iGobhSMglEwCkYyAAC2xkbSXpFooAAAAABJRU5ErkJggg==","orcid":"","institution":"Hubei University of Medicine","correspondingAuthor":true,"prefix":"","firstName":"Hong","middleName":"","lastName":"Yu","suffix":""}],"badges":[],"createdAt":"2026-03-10 13:08:42","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9084382/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9084382/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108406371,"identity":"fa50ff3a-b4b5-4485-8e14-ea141f77107a","added_by":"auto","created_at":"2026-05-04 09:42:13","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":263025,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eRadiomics analysis of sleep disorders in Parkinson's disease.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003ea\u003c/strong\u003e) ROIs segmented on T2WI images. Two red dots represent the MB. (\u003cstrong\u003eb\u003c/strong\u003e) Schematic representation of multi-slice MRI data (left) being processed into a set of extracted radiomic features (right), including first-order statistics, Gray Level Co-occurrence Matrix (GLCM), Gray Level Dependence Matrix (GLDM), Gray Level Size Zone Matrix (GLSZM), Gray Level Run Length Matrix (GLRLM), and Neighboring Gray Tone Difference Matrix (NGTDM). (\u003cstrong\u003ec\u003c/strong\u003e) MSE plot for model selection across different values of the regularization parameter λ. The dotted line represents the optimal λ value that minimizes prediction error. (\u003cstrong\u003ed\u003c/strong\u003e) The LASSO coefficient profiles of 10 features with non-zero coefficients. The dotted line represents the optimal λ selected based on cross-validation. (\u003cstrong\u003ee\u003c/strong\u003e) Heatmap of the correlation matrix between selected radiomic features and clinical outcomes or biomarkers. (\u003cstrong\u003ef\u003c/strong\u003e) Coefficient values of the top 10 most influential radiomic features in the predictive model. (\u003cstrong\u003eg\u003c/strong\u003e)\u003cstrong\u003e \u003c/strong\u003eROC curve depicting the diagnostic performance of the radiomic model. AUC = 0.78. (\u003cstrong\u003eh\u003c/strong\u003e)\u003cstrong\u003e \u003c/strong\u003eClassification accuracy across different MRI sequence. T1-Flair sequence: 61%, T2-PROPELLER sequence: 76%, T2-Flair sequence: 73%, T1-Flair+T2-Propeller sequence: 88%, T1-Flair+T2-Flairsequence: 73%, T2-Propeller+T2-Flairsequence: 82%, T1-Flair+T2-Propeller+T2-Flair sequence: 88%。\u003c/p\u003e","description":"","filename":"Picture1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-9084382/v1/20b6d69111142e31356bc294.jpg"},{"id":108406372,"identity":"8234ed9f-0bac-445e-8fee-3fe08927f18a","added_by":"auto","created_at":"2026-05-04 09:42:13","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":143280,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eML neurons are active during wake–sleep transitions.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003ea\u003c/strong\u003e) Scheme of sleep deprivation. (\u003cstrong\u003eb\u003c/strong\u003e) Experimental schematic. Animals were divided into three groups: control (2-h sweeping before perfusion), SD (6-h sleep deprivation), RS (2-h recovery sleep after deprivation). (\u003cstrong\u003ec\u003c/strong\u003e)\u003cstrong\u003e \u003c/strong\u003eLeft,\u003cstrong\u003e \u003c/strong\u003eimages of c-Fos immunostaining (green) in the ML of three groups: control, SD and RS. Scale bars, 150 μm. Right, quantitation of c-Fos+ cells in the ML in each group (n = 3 mice for control, n = 4 for SD, n = 4 for RS，one-way ANOVA, Tukey's multiple comparisons test, ****\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.0001 for control vs RS, ****\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.0001 for SD vs RS, \u003cem\u003eP =\u003c/em\u003e 0.999 for control vs SD). Data are presented as the mean±SEM.\u003c/p\u003e","description":"","filename":"Picture2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-9084382/v1/c485236c4bc87a6ca6117702.jpg"},{"id":108406373,"identity":"ce243bcc-135f-44e4-abbc-64c50f10cd09","added_by":"auto","created_at":"2026-05-04 09:42:13","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":201813,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eChemogenetic activation of ML neurons promotes REM and NREM sleep.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003ea\u003c/strong\u003e) Left, scheme for chemogenetic activation of ML neurons. Right, histology of injection site in the ML. Scale bar, 150 μm. (\u003cstrong\u003eb\u003c/strong\u003e) Experimental protocol of the chemogenetic activation. (\u003cstrong\u003ec-f\u003c/strong\u003e) Changes in all sleep duration (\u003cstrong\u003ec\u003c/strong\u003e), NREM duration (\u003cstrong\u003ed\u003c/strong\u003e), REM duration (\u003cstrong\u003ee\u003c/strong\u003e) and wakefulness duration (\u003cstrong\u003ef\u003c/strong\u003e) during the light phase after intraperitoneal injection of NS or CNO (1 mg kg\u003csup\u003e-1\u003c/sup\u003e). n = 6 mice, two-tailed paired t test, \u003cem\u003eP \u003c/em\u003e= 0.0007, \u003cem\u003eP \u003c/em\u003e= 0.0021, \u003cem\u003eP \u003c/em\u003e= 0.0334, \u003cem\u003eP \u003c/em\u003e= 0.0007. (\u003cstrong\u003eg-j\u003c/strong\u003e) Changes in all sleep duration (\u003cstrong\u003eg\u003c/strong\u003e), NREM duration (\u003cstrong\u003eh\u003c/strong\u003e), REM duration (\u003cstrong\u003ei\u003c/strong\u003e) and wakefulness duration (\u003cstrong\u003ej\u003c/strong\u003e) during the dark phase after intraperitoneal injection of NS or CNO (1 mg kg\u003csup\u003e-1\u003c/sup\u003e). n = 6 mice, two-tailed paired t test, \u003cem\u003eP \u003c/em\u003e= 0.0155, \u003cem\u003eP \u003c/em\u003e= 0.0107, \u003cem\u003eP \u003c/em\u003e= 0.0198, \u003cem\u003eP \u003c/em\u003e= 0.0179. Data are presented as the mean±SEM.\u003c/p\u003e","description":"","filename":"Picture3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-9084382/v1/7b34ec61b1ceab25eec237d4.jpg"},{"id":108406375,"identity":"98392db6-9514-40ed-bd30-2162f493da72","added_by":"auto","created_at":"2026-05-04 09:42:13","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":190548,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eChemogenetic inhibition of ML neurons supresses REM and NREM sleep.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003ea\u003c/strong\u003e) Left, scheme for chemogenetic inhibition of ML neurons. Right, histology of injection site in the ML. Scale bar, 150 μm. (\u003cstrong\u003eb\u003c/strong\u003e) Experimental protocol of the chemogenetic inhibition. (\u003cstrong\u003ec-f\u003c/strong\u003e) Changes in all sleep duration (\u003cstrong\u003ec\u003c/strong\u003e), NREM duration (\u003cstrong\u003ed\u003c/strong\u003e), REM duration (\u003cstrong\u003ee\u003c/strong\u003e) and wakefulness duration (\u003cstrong\u003ef\u003c/strong\u003e) during the light phase after intraperitoneal injection of NS or CNO (1 mg kg\u003csup\u003e-1\u003c/sup\u003e). n = 6 mice, two-tailed paired t test, \u003cem\u003eP \u003c/em\u003e= 0.0063, \u003cem\u003eP \u003c/em\u003e= 0.0238, \u003cem\u003eP \u003c/em\u003e= 0.0003, \u003cem\u003eP \u003c/em\u003e= 0.0056. (\u003cstrong\u003eg-j\u003c/strong\u003e) Changes in all sleep duration (\u003cstrong\u003eg\u003c/strong\u003e), NREM duration (\u003cstrong\u003eh\u003c/strong\u003e), REM duration (\u003cstrong\u003ei\u003c/strong\u003e) and wakefulness duration (\u003cstrong\u003ej\u003c/strong\u003e) during the dark phase after intraperitoneal injection of NS or CNO (1 mg kg\u003csup\u003e-1\u003c/sup\u003e). n = 6 mice, two-tailed paired t test,\u0026nbsp; \u003cem\u003eP \u003c/em\u003e= 0.0061, \u003cem\u003eP \u003c/em\u003e= 0.0157, \u003cem\u003eP \u003c/em\u003e= 0.0154, \u003cem\u003eP \u003c/em\u003e= 0.0078. Data are presented as the mean±SEM.\u003c/p\u003e","description":"","filename":"Picture4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-9084382/v1/fc083b4bc0416386ecbd3fdc.jpg"},{"id":108493433,"identity":"8c595e37-6dcc-4b79-a381-5504b3d33234","added_by":"auto","created_at":"2026-05-05 10:00:21","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":221723,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePD pathology alters ML neuronal dynamics across sleep-wake cycles.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003ea\u003c/strong\u003e) Scheme of simultaneous electrode and polysomnography recordings in PD mice.\u003cstrong\u003e \u003c/strong\u003e(\u003cstrong\u003eb\u003c/strong\u003e) Histology of electrode trace in the ML. Scale bar, 150 μm. (\u003cstrong\u003ec\u003c/strong\u003e) Scatter plot of ML neuron activity profiles across sleep-wake states in PD mice.\u003cstrong\u003e \u003c/strong\u003eEach dot represents one neuron (n = 252), plotted by the normalized activity ratio of wakefulness vs. NREM sleep [(W-NR)/(W+NR)] on the x-axis and REM sleep vs. NREM sleep [(R-NR)/(R+NR)] on the y-axis. Neurons are color-coded based on their state-specific activity patterns: REM active (pink), REM/wake active (light pink), wake active (green), NREM active (orange), NREM-REM active (beige), and no cycle (gray).\u003cstrong\u003e \u003c/strong\u003e(\u003cstrong\u003ed\u003c/strong\u003e)\u003cstrong\u003e \u003c/strong\u003ePie chart showing the proportion of ML neurons exhibiting distinct activity profiles in PD mice (REM active: 22.54%, REM/wake active: 14.75%, wake active: 6.15%, NREM active: 4.92%, NREM-REM active: 1.64%, no cycle: 50%). (\u003cstrong\u003ee\u003c/strong\u003e)\u003cstrong\u003e \u003c/strong\u003eComparing the percentage of state-specific activity patterns between Ctrl and PD mice. Ctrl.: n = 231 neurons, PD: n = 252 neurons, two-sided chi-square test, \u003cem\u003eP \u003c/em\u003e= 0.0028, \u003cem\u003eP \u003c/em\u003e= 0.59, \u003cem\u003eP \u003c/em\u003e= 0.079, \u003cem\u003eP \u003c/em\u003e= 0.116, \u003cem\u003eP \u003c/em\u003e= 0.207. (\u003cstrong\u003ef\u003c/strong\u003e)\u003cstrong\u003e \u003c/strong\u003eMean firing rates of ML neurons during different behavioral states in Ctrl. (gray) and PD (purple) mice. Wilcoxon rank-sum test, ****\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.0001. Data are presented as the mean±SEM.\u003c/p\u003e","description":"","filename":"Picture5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-9084382/v1/28a60041f924f10aecabf39f.jpg"},{"id":108493190,"identity":"b2b46a7c-81cd-411c-9aed-99311a125479","added_by":"auto","created_at":"2026-05-05 09:59:35","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":218298,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eChemogenetic activation of ML neurons rescues sleep disturbancesin PD.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003ea\u003c/strong\u003e) Experimental protocol of the chemogenetic activation in PD mice. (\u003cstrong\u003eb-e\u003c/strong\u003e) Changes in all sleep duration (\u003cstrong\u003eb\u003c/strong\u003e), NREM duration (\u003cstrong\u003ec\u003c/strong\u003e), REM duration (\u003cstrong\u003ed\u003c/strong\u003e) and wakefulness duration (\u003cstrong\u003ee\u003c/strong\u003e) during the light phase after chemogenetic activation ML neurons in PD mice. n = 7 mice, two-tailed paired t test, \u003cem\u003eP \u003c/em\u003e= 0.0069, \u003cem\u003eP \u003c/em\u003e= 0.0078, \u003cem\u003eP \u003c/em\u003e= 0.0043, \u003cem\u003eP \u003c/em\u003e= 0.015. (\u003cstrong\u003ef-i\u003c/strong\u003e) Changes in all sleep duration (\u003cstrong\u003ef\u003c/strong\u003e), NREM duration (\u003cstrong\u003eg\u003c/strong\u003e), REM duration (\u003cstrong\u003eh\u003c/strong\u003e) and wakefulness duration (\u003cstrong\u003ei\u003c/strong\u003e) during the dark phase after chemogenetic activation ML neurons in PD mice. n = 7 mice, two-tailed paired t test, \u003cem\u003eP \u003c/em\u003e= 0.0044, \u003cem\u003eP \u003c/em\u003e= 0.0282, \u003cem\u003eP \u003c/em\u003e= 0.0039, \u003cem\u003eP \u003c/em\u003e= 0.0136. Data are presented as the mean±SEM.\u003c/p\u003e","description":"","filename":"Picture6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-9084382/v1/7faa30ff59aca6690fa9e044.jpg"},{"id":109203681,"identity":"fc783c36-c3b4-4306-acd2-eec4872cd773","added_by":"auto","created_at":"2026-05-13 14:42:24","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1625613,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9084382/v1/7b5b2ef2-f90c-4378-96b0-16949c0a748b.pdf"},{"id":108406370,"identity":"440d0377-b72f-4471-88b9-fb1d92f5bcaa","added_by":"auto","created_at":"2026-05-04 09:42:13","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":4532941,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaryinformation.docx","url":"https://assets-eu.researchsquare.com/files/rs-9084382/v1/14e381d50cec14dcd1d355ff.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Targeting the lateral mammillary nucleus rescues sleep disturbances in Parkinson’s disease","fulltext":[{"header":"Introduction","content":"\u003cp\u003ePD non-motor symptoms, particularly sleep disorders are increasingly recognized as core features that significantly impair quality of life and often precede motor onset by years\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. Among these, rapid eye movement (REM) sleep behavior disorder (RBD), insomnia, non-REM (NREM) sleep fragmentation and excessive daytime sleepiness are particularly prevalent, affecting up most PD patients\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e,\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e. Current evidence suggests that sleep disturbances in PD arise from multifactorial etiologies, including primary neurodegeneration of sleep-regulating neural circuits, dopaminergic dysregulation, circadian rhythm dysfunction, and secondary effects of motor symptoms or medications\u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e. However, the precise neural substrates underlying sleep dysfunction in PD remain incompletely understood, limiting the development of targeted therapeutic interventions.\u003c/p\u003e \u003cp\u003eThe mammillary body (MB) is a small deep brain structure located in the posterior hypothalamus, which is a key component of the limbic system and has long been recognized for its role in memory\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e,\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e, brain oscillation\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e, and spatial navigation\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e. The region is composed of the medial mammillary nucleus (MMn) and lateral mammillary nucleus (LM), and the MMn can be further subdivided into medial (MM) and lateral (ML) parts\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e,\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e. The ML receives inputs from the ventral tegmental nucleus of Gudden (VTg) and the medial prefrontal cortex (mPFC) which are the critical sleep-wake centers\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. However, how the ML mediates sleep-wake cycles remains largely unknown. Magnetic resonance imaging (MRI) has demonstrated that the MB volumes show a marked volume reduction in obstructive sleep apnea (OSA) patients\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. Therefore, like in the OSA, we hypothesized that ML neuronal activities are organized in PD sleep disorders to regulate the sleep-wake circles.\u003c/p\u003e \u003cp\u003eHere, by applying radiomic analysis based on MRI data, we demonstrated MB structural features associated with sleep disturbances in PD patients. We further identified that ML neurons were recruited in sleep-wake regulation via a combination of c-Fos staining, \u003cem\u003ein vivo\u003c/em\u003e single-unit recording, and DREADD-mediated chemical genetics. Also, we demonstrated that chemogenetic activation of ML neurons could rescue sleep disturbances in PD mice. Together, our data reveal that therapeutic potential of targeting ML neurons represents a promising strategy to alleviate sleep disturbances associated with PD, which address critical gaps in our understanding of the neural mechanisms underlying one of the most disabling non-motor symptoms of PD.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e \u003cb\u003eDecoding Sleep Disorders in Parkinson's Disease through Radiomic Analysis of the MB.\u003c/b\u003e To investigate the correlation between the MB and sleep disorders in patients with Parkinson's disease, we performed radiomics analysis based on MRI data (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea). From T1-FLAIR and T2-PROPELLER sequences after standardized preprocessing, including intensity normalization, resampling, and manual region-of-interest (ROI) delineation, we selected features spanned first-order statistics, texture matrices (GLDM, GLSZM, GLRLM), and NGTDM, capturing heterogeneity, structural complexity, and intensity distribution within the MB (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb). We selected optimal regularization parameter (λ) determined via 10-fold cross-validation, with the minimum mean squared error (MSE) observed at λ\u0026thinsp;\u0026asymp;\u0026thinsp;0.1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec). As λ increased, the coefficients of less informative features gradually shrank to zero, indicating effective feature selection (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ed). Also, we found that ten features, including InterquartileRange.1, Imc.1, MCC.1, MCC.2, Kurtosis.3, ZoneEntropy.3, ZoneEntropy.5, SmallAreaLowGrayLevelEmphasis.7, Strength.8, and LargeDenpedenceHighGrayLevelEmphasis.7 were retained in the final model, and Pearson correlation coefficients among the selected features revealed moderate to strong correlations between several pairs (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ee). Notably, InterquartileRange.10 and MCC.9 exhibited the highest absolute coefficients (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ef).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eNext, we extracted and selected ten radiomic features using Lasso regression, which demonstrated robust predictive power for classifying PD patients with significant sleep disorders (area under the curve, AUC\u0026thinsp;=\u0026thinsp;0.78) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eg). Crucially, our analysis revealed that integrating information from both T1-FLAIR and T2-PROPELLER sequences significantly enhanced classification accuracy (85%) compared to using either sequence alone (60% and 75%, respectively) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eh). This improvement highlights the complementary value of multi-modal imaging in capturing distinct aspects of tissue pathology within small, deep brain structures. Together, these results indicate that structural alterations in the MB, as revealed by multi-modal radiomic features, are closely associated with sleep disturbances in Parkinson's disease. This identifies the MB as a novel imaging biomarker and potential therapeutic target for addressing sleep pathology in PD patients.\u003c/p\u003e \u003cp\u003e \u003cb\u003eML neurons in sleep regulations.\u003c/b\u003e The MB is composed of medial mammillary nucleus (MMn), lateral mammillary nucleus (LM), where the MMn can be further divided into the medial part (MM) and the lateral part (ML). To identify which part of the MB involved in sleep-active neurons, we measured c-Fos activity in a recovery sleep after deprivations (RS) paradigm. Sleep deprivation was induced using the fondling method, which maintains wakefulness via gentle handling without inducing stress (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea). Then, we designed three experimental paradigms: control mice (no sleep manipulation), sleep-deprived mice (6 h of sleep deprivation, SD) and mice undergoing recovery sleep (2 h of recovery sleep following 6 h of sleep deprivation, RS) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb). To assess neural activity, we examined c-Fos-positive (c-Fos+) cells in the MB, and found that the number of c-Fos+ cells in the ML was significantly increased in RS mice compared to both control and SD groups (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eNext, to explore how ML neurons are engaged in sleep-wake state, we employed \u003cem\u003ein vivo\u003c/em\u003e single-unit and EEG/EMG recordings simultaneously (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003ea). We confirmed the electrode locations in all recorded mice (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eb and Figure S2a). We classified neurons based on their activity profiles across wakefulness, NREM sleep, and REM sleep using a state-preference index. As shown in Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003ec, a scatter plot of 231 recorded neurons plotted the Wake-NREM activity ratio (W-NR)/(W\u0026thinsp;+\u0026thinsp;NR) against the REM-NREM activity ratio (R-NR)/(R\u0026thinsp;+\u0026thinsp;NR). Five distinct neuronal subtypes were identified: REM-active (pink), REM/Wake-active (salmon), Wake-active (green), NREM-active (orange), and neurons with no state preference (gray). The distribution revealed clustered patterns, reflecting specialized activity preferences for each sleep-wake state. We further illustrated the proportion of each subtype: REM-active neurons comprised 11.69%, REM/Wake-active 16.02%, Wake-active 10.39%, NREM-active 2.16%, and 59.74% exhibited no clear state-related activity (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003ed). Collectively, these data demonstrate that the ML contains diverse neuronal populations with distinct sleep-wake state-related activity profiles, uncovering the heterogeneity of neural circuits involved in sleep-wake regulation. Overall, our results reveal that distinct sleep-wake state-related neural activity patterns in the ML and identify multiple neuronal subtypes differentially engaged in wakefulness, NREM, and REM sleep.\u003c/p\u003e \u003cp\u003e \u003cb\u003eML neurons are sufficient to promote REM and NREM sleep.\u003c/b\u003e The observation that ML neurons were recruited during sleep-wake rhythm suggests these neurons might be functionally important. To test this, we unilaterally injected AAV carrying excitatory Gq-coupled DREADDs (rAAV-hSyn-hM3D(Gq)-EGFP) into the ML of C57BL/6J mice (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea and Figure S2b). Following a two-week recovery period after viral injection, we implanted EEG/EMG electrodes, and treated animals with CNO (1 mg/kg) and saline (NS) administration while recording sleep behavior (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb). Strikingly, during the light phase (9:00\u0026ndash;12:00 AM), chemogenetic activation of ML neurons significantly increased total sleep duration (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec), primarily driven by an increase in NREM sleep (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ed). REM sleep also showed a significant increase (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ee), while wakefulness duration was markedly reduced (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ef). These results indicate that ML neuron activation promotes sleep across both NREM and REM states during the light phase. Meanwhile, during the dark phase (19:30\u0026thinsp;\u0026minus;\u0026thinsp;22:30 PM), which corresponds to the rodent\u0026rsquo;s natural active period, ML activation still induced a significant increase in total sleep (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eg) and NREM sleep (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eh), with no significant change in REM sleep (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ei), but a reduction in wakefulness (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ej). Notably, the magnitude of sleep promotion was greater during the light phase, suggesting circadian modulation of ML-mediated sleep control. In summary, these data indicate that chemogenetic activation of ML neurons potently promotes sleep, particularly NREM and REM, across both light and dark phases, with concomitant suppression of wakefulness-demonstrating a critical role for the ML in regulating sleep-wake balance in mice.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eML neurons are Required for REM and NREM sleep.\u003c/b\u003e To further study the necessity of ML neurons in sleep-wake regulation, we employed chemogenetic silencing using rAAV-hSyn-hM4D(Gi)-EGFP to selectively inhibit the activity of ML neurons. Viral vectors were bilaterally injected into the ML, with successful transduction confirmed by EGFP expression localized to the ML (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea and Figure S2b). After two weeks of recovery, we implanted EEG/EMG electrodes, and monitored sleep behavior during both light and dark phases via administration of CNO (1 mg/kg) and saline (NS) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb). We found that chemogenetic inhibition of ML neurons significantly reduced total sleep duration (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ec), NREM sleep (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ed) and REM sleep (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ee) during the light phase. Conversely, wakefulness duration was markedly increased in the same phase (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ef). Interestingly, during the dark phase, chemogenetic inhibition of ML neurons led to a significant reduction in total sleep (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eg) and NREM sleep (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eh), along with a reduction in REM sleep (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ei). However, Wakefulness was significantly prolonged under CNO treatment (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ej). Therefore, these data indicate that ML neurons are required for both NREM and REM states in the light and dark phases.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eML Neurons Exhibit Alter Firing Dynamics in PD Mice with Impaired REM Sleep\u003c/b\u003e. To investigate neural mechanisms underlying sleep disturbances in PD, we generated a PD mouse model by unilateral injection of rAAV-CMV-SYN-A53T into the substantia nigra pars compacta (SNc) of C57BL/6 mice (Figure S3a). This approach induces progressive α-synuclein aggregation and dopaminergic neurodegeneration, recapitulating key pathological features of PD. To test the success of the PD model establishment, we conducted behavioral tests and histopathological analysis. Firstly, we performed the pole assay (Figure S3b). As expected, PD mice exhibited significantly prolonged crawling duration in, indicating impaired motor coordination and bradykinesia (Figure S3c). Next, we performed the cylinder test (Figure S3d). Notably, PD mice displayed marked asymmetry in forelimb use that left claw utilization was significantly reduced (Figure S3e), while right claw usage increased (Figure S3f), confirming contralateral motor impairments due to unilateral SNc degeneration. Then, we performed histological analysis to confirm robust dopaminergic pathology. Indeed, immunofluorescence staining revealed a dramatic loss of tyrosine hydroxylase-positive (TH+) neurons in the SNc of PD mice (Figure S3g, h). Also, TH+ fiber density in the caudate-putamen (CPU) was also significantly reduced (Figure S3i, g), demonstrating extensive striatal denervation consistent with PD pathophysiology. Altogether, these data indicate that the PD models are successfully established.\u003c/p\u003e \u003cp\u003eHow changes of firing dynamics of ML neurons in the PD mice is largely unknown. To test it, we performed \u003cem\u003ein vivo\u003c/em\u003e single-unit recordings on ML neurons in PD mice during sleep-wake state (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea). We confirmed the electrode locations in all recorded mice (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eb and Figure S2a). We recorded the activity of ML neurons (n\u0026thinsp;=\u0026thinsp;252 units, 4 mice) in PD mice and compared the neuronal activity in different brain-state phases to the controls. Interestingly, we found that there were distinct functional clusters, including REM-active, REM/wake-active, wake-active, NREM-active, NREM-REM-active, and no state-dependent cycle (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ec, d). Notably, in PD mice, the proportion of REM-active ML neurons was significantly increased compared to controls (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ee), suggesting a pathological reorganization of ML neuronal activity toward REM-specific firing patterns. However, we found that the average firing rates of ML neurons in PD mice significantly reduced during the overall brain-state, wake and NREM phase. While the average firing rates during the REM phase were higher in PD mice than controls, indicating a selective hyperactivity in a subset of REM-active neurons that may reflect compensatory or pathological circuit reorganization (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ef). Overall, these findings suggest that while ML neuronal activity is broadly suppressed in PD, REM-associated neurons exhibit enhanced firing, potentially contributing to REM sleep disturbances commonly observed in Parkinson\u0026rsquo;s disease.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eChemogenetic Activation of ML Neurons Promotes Sleep in PD Mice.\u003c/b\u003e To investigate the role of ML neurons in sleep regulation within a PD model, we employed a chemogenetic approach using the excitatory DREADD receptor hM3Dq (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ea). Using a similar viral injection strategy, we generated a PD mouse model by unilateral injection of rAAV-CMV-SYN-A53T into the SNc of C57BL/6 mice (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ea and Figure S2b). To characterize the sleep-wake architecture in PD mice, we performed EEG/EMG recordings in freely moving PD mice and control littermates (Figure S4a). We found that PD mice exhibited a significant increase in wakefulness percentage compared to controls, indicating heightened arousal or impaired sleep consolidation (Figure S4b). Concurrently, the proportion of time spent in NREM sleep was significantly reduced, suggesting disrupted slow-wave sleep homeostasis (Figure S4c). Most strikingly, REM sleep was profoundly suppressed in PD mice (Figure S4d), consistent with REM sleep disturbances commonly observed in human PD patients. Furthermore, PD mice displayed a dramatic increase in the number of sleep-wake transitions (Figure S4e), reflecting severe sleep fragmentation and instability. These data indicate that PD mice exhibit increased wakefulness, reduced NREM and REM sleep, and profound sleep fragmentation, revealing a comprehensive disruption of sleep-wake homeostasis that parallels core clinical features of PD.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eIt is unknown how ML neurons regulate the sleep patterns in the PD mice. To test it, we performed EEG/EMG recordings during both light (9:00\u0026ndash;12:00 AM) and dark (19:30\u0026thinsp;\u0026minus;\u0026thinsp;22:30 PM) phases following intraperitoneal injection of either CNO or NS (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ea). We found that CNO treatment significantly increased total sleep duration compared to NS control (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eb), driven primarily by an increase in NREM and REM sleep duration during the light phase (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ec, d). Consistently, wakefulness duration was significantly reduced after CNO administration (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ee), indicating that ML neuron-activation promotes sleep and suppresses arousal. Alao, in the dark phase, similar effects were observed: total sleep duration was significantly increased (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ef), NREM and REM sleep duration trended upward (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eg, h), while wakefulness duration was significantly decreased (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ei), confirming that ML activation enhances sleep propensity across circadian phases. Notably, chemogenetic activation of ML neurons did not rescue motor asymmetry or enhance forelimb use in PD mice (Figure S5). Altogether, these findings demonstrate that chemogenetic activation of ML neurons increases sleep duration and reduces wakefulness in PD mice, without altering motor function, identifying the ML as a key regulator of sleep architecture and a potential therapeutic target for sleep disturbances in PD.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eOur study shows that the ML as the lateral part of the MB is a novel and critical hub for sleep-wake regulation. Furthermore, we identify the dysfunction of ML neuronal dynamics as a substrate for sleep disturbances in PD and demonstrate the therapeutic potential of targeting this nucleus. Altogether, these data advance the understanding of sleep circuit pathophysiology in PD and propose a new deep brain target for intervention.\u003c/p\u003e \u003cp\u003ePD is a progressive neurodegenerative disorder characterized by motor symptoms such as bradykinesia, rigidity, and resting tremor\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e. Patients with PD also have non-motor symptoms, specifically sleep disturbances, which have emerged as equally devastating manifestations and received limited attention\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. However, the precise neural substrates underlying sleep dysfunction in PD remain unknown. Our radiomic analysis of multi-modal MRI data from PD patients reveals that structural heterogeneity within the MB is closely associated with the presence and severity of sleep disorders, with the combined use of T1-FLAIR and T2-PROPELLER sequences yielding a high classification accuracy for identifying PD patients with significant sleep disorders (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). This aligns with growing evidence that non-motor symptoms in PD are rooted in early pathology of limbic and hypothalamic circuits beyond the SNc\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e,\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e. The MB, a key component of this network, has long been implicated in memory and spatial navigation\u003csup\u003e\u003cspan additionalcitationids=\"CR7 CR8 CR9\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e, but its role in sleep physiology and dysfunction in PD has remained understudied. Our identification of ten optimal radiomic features that robustly predict PD sleep disturbances fills this gap, positioning the MB as a novel imaging biomarker for a highly disabling non-motor symptom of PD (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ee, f). Notably, previous MRI studies have demonstrated volume reductions in the MB in patients with OSA\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e, a condition characterized by recurrent sleep fragmentation and daytime sleepiness that overlap with PD sleep disturbances. This convergence suggests that the MB structural integrity may be a shared vulnerability factor across sleep disorders with distinct etiologies. Our radiomic approach captured subtle structural heterogeneity, provides a more sensitive tool for detecting MB pathology in PD, and offers potential clinical utility for early identification of patients at risk for severe sleep disturbances\u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eThe MB is composed of medial mammillary nucleus (MMn), lateral mammillary nucleus (LM), where the MMn can be further divided into the medial part (MM) and the lateral part (ML). However, it is unclear that which part of the MB is involved in the regulation of sleep-wake cycle. Overall, we found that the activity of ML neurons significantly increased in RS mice (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). The data indicate that ML neurons are recruited during sleep rebound, which reflects the homeostatic regulation of sleep\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e. Specifically, we identified diverse neuronal subtypes in the ML with distinct sleep-wake state-related activity profiles, including REM-active, REM/Wake-active, Wake-active, NREM-active, and non-state-preferring neurons (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e), which establishes the ML as a functional mosaic within the sleep-regulatory network. This heterogeneity is consistent with its complex afferent inputs from key sleep-wake centers like the VTg and the mPFC\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e, and efferent projections to the anterior thalamus and reticular formation\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e,\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e. Our findings extend this principle to the ML, showing that distinct subtypes of ML neurons are differentially engaged in wakefulness, NREM, and REM sleep. Therefore, this functional specialization suggests the ML is not a monolithic structure but a circuit node where distinct neuronal subpopulations may integrate limbic and homeostatic signals to gate specific sleep states\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e,\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eMoreover, we demonstrate a potent and bidirectional causal role for ML neurons in sleep-wake control via chemogenetic approaches. Activation of ML neurons significantly increased total sleep duration, primarily driven by increases in NREM and REM sleep, while suppressing wakefulness\u0026mdash;effects observed across both light and dark phases (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Notably, the magnitude of sleep promotion was greater during the light phase, suggesting circadian modulation of ML function. Conversely, inhibition of ML neurons induced insomnia-like phenotypes, reducing total sleep, NREM, and REM sleep (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). Overall, our findings confirm that the activity of ML neurons is required for the maintenance of normal sleep architecture, particularly for NREM and REM sleep. The consistent effects of ML manipulation across light and dark phases suggest that ML neurons play a tonic role in sleep regulation, rather than being restricted to a specific circadian window. This is in contrast to some sleep-regulating neurons, such as SCN neurons, which exhibit strong circadian rhythmicity in their activity\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e. Therefore, these data shown that the tonic role of ML neurons may reflect their function as a key node in integrating homeostatic sleep pressure with circadian cues to maintain sleep-wake balance. However, the precise neurotransmitter phenotype of sleep-modulating ML neurons remains unknown, warranting cell-type-specific manipulations. Further studies are needed to resolve this issue.\u003c/p\u003e \u003cp\u003eImportantly, consistent with clinical observations in PD patients\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e,\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e, we found that PD mice exhibited severe sleep disturbances, including increased wakefulness, reduced NREM and REM sleep, and profound sleep fragmentation (Figure S4). Specially, we observed a profound disruption of ML neuronal firing patterns. Also, the broad reduction in average firing rates during wakefulness and NREM sleep aligns with general neuronal dysfunction in neurodegenerative disease\u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e. Nevertheless, the paradoxical increase in the proportion and firing rate of REM-active neurons is a striking finding. Together, this dual pattern of ML neuronal dysfunction that there is broad hypoactivity alongside REM-active neuron hyperactivity may contribute to PD-related sleep disturbances. Moreover, this pathological reorganization may represent a compensatory response to the severe REM sleep suppression observed in PD mice and in patients\u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e. Alternatively, this hyperactivity could be a pathological consequence of α-synuclein aggregation or dopaminergic degeneration, further disrupting REM sleep regulation\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e,\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e. For example, α-synuclein pathology is known to spread through neural circuits via prion-like mechanisms\u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e, and it is possible that α-synuclein accumulation in the ML disrupts the balance of neuronal activity, favoring REM-active subtypes. Future work should be investigated.\u003c/p\u003e \u003cp\u003eThe most translational finding of our study is that chemogenetic activation of ML neurons in PD mice effectively rescued key sleep abnormalities, increasing total, NREM, and REM sleep duration, while reducing wakefulness and sleep fragmentation, which were observed during both light and dark phases (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). Importantly, this sleep-promoting effect was not accompanied by improvements in motor function (Figure S5), indicating that ML neuron activation specifically targets sleep circuits without modulating motor pathways. This specificity is clinically relevant, as current PD therapies primarily target motor symptoms and often exacerbate sleep disturbances\u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e. Notably, the ability of ML activation to rescue both NREM and REM sleep disturbances in PD mice, while REM sleep disturbances, such as RBD, are among the most disabling and treatment-refractory non-motor symptoms of PD\u003csup\u003e31\u003c/sup\u003e. RBD, which is characterized by the loss of REM sleep atonia and the acting out of dreams, is associated with more rapid disease progression\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e. Our data show that ML activation increases REM sleep in PD mice suggests that targeting ML neurons could provide a novel therapeutic strategy for RBD and other REM-related sleep disturbances in PD. This is supported by our observation that ML neurons contain a large population of REM-active subtypes, whose hyperactivity in PD may reflect a failed compensatory attempt to restore REM sleep\u0026mdash;an attempt that could be augmented by targeted activation of ML neurons. Future work should explore mechanisms that may underlie the sleep-rescuing effect of ML activation in PD mice.\u003c/p\u003e \u003cp\u003eIn summary, we identify the ML as a critical regulator of sleep-wake states and demonstrate its pathological engagement in PD-related sleep disturbances. By bridging clinical imaging with causal mechanistic studies in animals, we propose that the structural and functional integrity of the ML is essential for normal sleep architecture and is disrupted in PD. Moreover, our findings that modulating ML activity can rectify sleep deficits without impacting motor function, highlighting the potential of targeting ML neurons for the treatment of PD non-motor symptoms. Thus, these data fill critical gaps in our understanding of the neural mechanisms underlying PD sleep disturbances and provide a foundation for the development of targeted therapies for this disabling feature of PD.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e \u003cb\u003eAnimals.\u003c/b\u003e \u003c/p\u003e \u003cp\u003e All animals experimental procedures were approved by the Hubei Provincial Animal Care and Use Committee and complied with the experimental guidelines of the Animal Experimentation Ethics Committee of Hubei University of Medicine (Approval code: 2025-16). C57BL/ 6 mice (12\u0026ndash;16 weeks old) were purchased from Shulaibao (Wuhan) Biotechnology Co., Ltd., and were group-housed and bred the laboratory animal centre of Hubei University of Medicine under a constant temperature (20\u0026thinsp;\u0026plusmn;\u0026thinsp;2℃), humidity (50%\u0026ndash;60%), illumination intensity (15\u0026ndash;20 lux) and 12-hr light/dark cycle (7:00 am to 19:00 pm).\u003c/p\u003e \u003cp\u003e \u003cb\u003eVirus.\u003c/b\u003e \u003c/p\u003e \u003cp\u003erAAV-SYN(BrainVTA)-SNCA(A53T)-WPRE-bGHpA, rAAV-hSyn-hM3D(Gq) -EGFP-WPRE-hGH polyA, and rAAV-hSyn-hM4D(Gi) -EGFP-WPRE-hGH polyA were purchased from BrainVTA (Wuhan) Co., Ltd. All viruses were stored in aliquots at -80℃.\u003c/p\u003e \u003cp\u003e \u003cb\u003eRadiomics analysis.\u003c/b\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eEthical approval\u003c/strong\u003e \u003cp\u003e for this study was obtained from Ethics Committee of Shiyan Taihe hospital, Hubei University of Medicine (Approval code: 2025KS26). This study was performed in accordance with the \u003cem\u003eDeclaration of Helsinki\u003c/em\u003e, and was waived the requirement for written informed consent. 150 PD accompanied by sleep disorders and 100 controls were recruited in the study, who had undergone 3.0-T MRI (GE Healthcare, Discovery 750w, Wisconsin, USA) imaging including T1WI-Flair, T2-FLAIR, and T2WI-PROPELLER sequences. Radiomics analysis was performed as previously described\u003csup\u003e\u003cspan additionalcitationids=\"CR34\" citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003c/p\u003e \u003cp\u003eThe inclusion criteria for PD were as follows: (1) Patients met the diagnostic criteria for idiopathic Parkinson\u0026rsquo;s disease. (2) Patients had not undergone deep brain stimulation (DBS) surgery and were required to discontinue dopaminergic medications for one week prior to participation. (3) Patients had no history of central nervous system disorders\u0026mdash;including intracerebral hemorrhage, cerebral infarction, Alzheimer\u0026rsquo;s disease, amyotrophic lateral sclerosis, or Huntington\u0026rsquo;s disease\u0026mdash;or other neurological or psychiatric conditions. (4) Patients were aged between 55 and 75 years inclusive.\u003c/p\u003e \u003cp\u003eThe exclusion criteria for PD were as follows: (1) Presence of image artifacts. (2) Incomplete clinical or imaging data. (3) History of neurological or psychiatric disorders. (4) Prior exposure to antipsychotic medications or other psychoactive drugs.\u003c/p\u003e \u003cp\u003eThe criteria for controls (HC) were as follows: (1) Age-matched to the experimental group, with no restriction on sex. (2) No history of neurological disorders, no large-area cerebral infarction, and no evidence of brain atrophy.\u003c/p\u003e \u003cp\u003eMRI image preprocessing. To minimize variability arising from differences in imaging acquisition parameters, MRI image preprocessing was performed using the MR Radiomics Platform (MRP). All images were resampled to an isotropic voxel size of 0.50 \u0026times; 0.50 \u0026times; 3.00 mm\u0026sup3;. Rigid registration of T1-FLAIR and T2-FLAIR sequences to the T2-PROPELLER sequence was conducted using a six-degree-of-freedom rigid-body transformation with a mutual information similarity metric. Intensity normalization was applied to standardize MR signal intensities across all subjects into a consistent range for each imaging modality.\u003c/p\u003e \u003cp\u003eRegion-of-Interest (ROI) delineation. The mammillary bodies were manually segmented slice-by-slice on axial views of three MRI sequences using 3D Slicer software. ROIs were carefully delineated by trained raters blinded to clinical information.\u003c/p\u003e \u003cp\u003eFeature extraction. Radiomic features were extracted from both original and wavelet-filtered images for each sequence using the open-source PyRadiomics platform, yielding a total of 851 features per sequence. These features comprised14 three-dimensional shape-based features, 18 first-order intensity statistics, 24 gray-level co-occurrence matrix (GLCM) features, 14 gray-level dependence matrix (GLDM) features, 16 gray-level run-length matrix (GLRLM) features, 16 gray-level size zone matrix (GLSZM) features, and 5 neighboring gray-tone difference matrix (NGTDM) features. Given five sequences per subject, a total of 5,255 radiomic features were initially extracted. Detailed definitions of these features are available on the PyRadiomics documentation website.\u003c/p\u003e \u003cp\u003eFeature selection. Feature distributions were assessed for normality using the Kolmogorov\u0026ndash;Smirnov (K\u0026ndash;S) test, and homogeneity of variance was evaluated using Levene\u0026rsquo;s test. Depending on the distribution, either Student\u0026rsquo;s \u003cem\u003et\u003c/em\u003e-test or non-parametric alternatives were used to identify features significantly differentiating high- versus low-expression groups. To reduce dimensionality and select the most informative features, least absolute shrinkage and selection operator (LASSO) regression with 10-fold cross-validation was applied exclusively on the training set. The optimal regularization parameter (lambda) was determined based on the minimum cross-validated error criterion. This pipeline was repeated to investigate the impact of multi-sequence MRI on feature performance.\u003c/p\u003e \u003cp\u003eFeature fusion. Fusion was implemented at both the feature level and the decision level. At the feature level, the most common approach\u0026mdash;concatenation\u0026mdash;was employed: features extracted from multiple sequences were combined into a single high-dimensional feature matrix, which was then subjected to the feature selection pipeline. Alternatively, features from individual sequences were first selected separately and then fused. A canonical correlation analysis (CCA)-based fusion method was also explored, which identifies linear transformations of two input feature sets to maximize their inter-set correlation. At the decision level, ensemble strategies\u0026mdash;including majority voting, weighted averaging, and stacking with a meta-classifier\u0026mdash;were used to integrate predictions from multiple models and enhance predictive stability.\u003c/p\u003e \u003cp\u003eModel evaluation. The dataset was randomly stratified into independent training (70%) and testing (30%) sets. Baseline clinical and ROI characteristics were compared between the two sets to ensure no statistically significant imbalances. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), along with 95% confidence intervals (CIs), accuracy, sensitivity, and specificity. The classifier yielding the highest AUC for each task was selected as optimal. Statistical significance of AUC improvements was assessed using DeLong\u0026rsquo;s test. All statistical analyses were performed using Python (v3.8) and R (v4.1.3), with a two-sided \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 considered statistically significant.\u003c/p\u003e \u003cp\u003e \u003cb\u003eStereotaxic injections.\u003c/b\u003e \u003c/p\u003e \u003cp\u003eMice were fixed in a stereotaxic injection frame (RWD Life Science Co., Ltd., China) after anesthetized (0.1 ml/10 g) by 1% pentobarbital sodium. 0.20 \u0026micro;l of rAAV vectors were delivered with a glass micropipette connected KDS-120 Pressure Micro Injector (Harvard Bioscience, Inc., USA). Coordinates used for injection were as follows: ML, \u0026minus;\u0026thinsp;2.92 mm from bregma, 0.55 mm lateral from midline, and 5.25 mm vertical from the cortical surface; SNc, \u0026minus;\u0026thinsp;3.08 mm from bregma, 0.7 5 mm lateral from midline, and 4.7 mm vertical from the cortical surface.\u003c/p\u003e \u003cp\u003e \u003cb\u003eImmunohistochemistry.\u003c/b\u003e \u003c/p\u003e \u003cp\u003eAfter anaesthetization, mice were perfused with phosphate buffered saline (PBS), followed by 4% paraformaldehyde (PFA, w/v in PBS). Brains were dissected and postfixed in 4% PFA at 4℃ overnight, and then were cut into 40 \u0026micro;m thick coronal sections by a vibratome (Leica VM 1000). Brain sections were incubated in blocking solution (PBS containing 0.1% Triton X-100 and 10% normal goat serum) with anti-c-Fos (1:12,000, rabbit, Synaptic System, 226003) for 40 h or anti-TH (1:500, rabbit, Millipore, ab6211) overnight at 4℃. Then, sections were washed three times with PBS, incubated with secondary antibodies (goat anti-rabbit, Alexa Fluor 488 conjugate 1:500, Thermo Fisher Scientific, A-11008) for 2 h at room temperature, mounted and imaged. For examining the virus expression and electrode locations, brain slices were washed three times with PBS and incubated for 30 min with DAPI at room temperature. Images were captured with Leica SP8 MP confocal microscope.\u003c/p\u003e \u003cp\u003e \u003cb\u003eChemogenetic manipulation.\u003c/b\u003e \u003c/p\u003e \u003cp\u003eC57BL/6J mice expressing hM3Dq or hM4Di in the ML were injected intraperitoneally 30 min prior to recording at 8:30/19:00 with saline (NS, day 1) and clozapine-N-oxide (CNO, 1 mg/kg, Enzo Life Sciences Inc., Exeter, UK, day 2). Recordings were carried out for 3 h during 9:00\u0026ndash;12:00 and 19:30\u0026thinsp;\u0026minus;\u0026thinsp;22:30, respectively.\u003c/p\u003e \u003cp\u003e \u003cb\u003eSleep Deprivation and Sleep Recovery Experiments.\u003c/b\u003e \u003c/p\u003e \u003cp\u003eIn the sleep deprivation and recovery experiments, mice were randomly assigned to three groups: control group (Ctrl), sleep-deprived group (SD), and sleep recovery group (RS). Each group was individually housed in a testing cage, which was placed inside an acoustic/light-isolated enclosure to minimize external disturbances. Prior to the sleep deprivation procedure, all mice were acclimatized to the arena for 48 h under a standard 12-h light/dark cycle (lights on from 7:00 a.m. to 7:00 p.m.). SD began 7:00 a.m. and was maintained for 6 h by gently stroking the mice\u0026rsquo;s ears and tail with a soft brush whenever signs of behavioral sleep were observed. Care was taken throughout the procedure to avoid startling the animals-no loud noises or mechanical vibrations were introduced during the intervention. Immediately after the 6 h SD, SD and Ctrl mice were deeply anesthetized and transcardially perfused for brain collection. RS mice were allowed 2 h of undisturbed sleep following deprivation before undergoing perfusion and brain extraction.\u003c/p\u003e \u003cp\u003e \u003cb\u003ePole test.\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe pole test was performed as previously described\u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e. A custom-made wooden pole (diameter: 18 cm; height: 50 cm) was used, with a small wooden ball wrapped in gauze (to provide grip) fixed at the top. During testing, each mouse was placed head-up at the top of the pole. The time from movement onset to reaching the bottom of the pole (crawling duration) were recorded. Each mouse was tested five times with a 1-min interval for each trial. The average values across the five trials were used for analysis.\u003c/p\u003e \u003cp\u003e \u003cb\u003eCylinder test.\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe behavioral setup was performed as previously described\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e. The test was administrated at the same period (14:00\u0026ndash;18:00 pm) and performed in a glass cylinder (20 cm \u0026times; 40 cm). Prior to test, mice were habituated to the testing room for at least 30 min to minimize stress-induced behavioral artifacts. Each mouse was allowed to freely explore the cylinder for 10 min and recorded the number of times it reared up and touched the cylinder wall. The number of times included scored for the left, right, or both paws that touched the wall.\u003c/p\u003e \u003cp\u003e \u003cb\u003eIn vivo\u003c/b\u003e \u003cb\u003esingle-unit recordings and data analysis.\u003c/b\u003e\u003c/p\u003e \u003cp\u003e \u003cem\u003eIn vivo\u003c/em\u003e single-unit recordings and data analyses were performed as described in previous studies\u003csup\u003e\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e. Briefly, a 16-channel electrode array constructed from 25.4-\u0026micro;m formvar-insulated nichrome wire (cat no. 761500, A-M Systems, USA) was affixed to a screw-driven microdrive, and the impedance of the electrode tips was measured to be 700\u0026ndash;800 kΩ. Mice were implanted with the 16-channel electrodes targeting the ML. Neuronal recordings were performed using NeuroLego System (Jiangsu Brain Medical Technology Co.ltd), band-pass filtered at 300\u0026ndash;6,000 Hz, sampled at 30 kHz. During the recordings, an amplitude threshold of 50 \u0026micro;V was applied to exclude background noise. Electrodes were advanced incrementally in steps of 62.5 \u0026micro;m each recording session. Recordings data were collected at the depth where electrode tips were located in the ML.\u003c/p\u003e \u003cp\u003eAll data analyses were performed using MATLAB 2014b (The Mathworks, Inc., Natick, Massachusetts, USA). Manual spike sorting was performed on the basis of the three principal components of spike waveforms and waveform energy features using MATLAB toolbox (MClust 4.4). Isolation distance (\u0026gt;\u0026thinsp;20) and L-ratio (\u0026lt;\u0026thinsp;0.1) were calculated to identify neurons, and only units with inter-spike intervals shorter than 2 ms accounting for \u0026lt;\u0026thinsp;1% of total spikes were included analyzed. Additionally, cross-correlation comparisons were performed to exclude duplicate units in a session.\u003c/p\u003e \u003cp\u003eFor the simultaneous multi-channel and EEG/EMG recordings, data were collected for 1 h. The neuronal activity was aligned to wakefulness, non-rapid eye movement sleep (NREM) and rapid eye movement sleep (REM) states as classified by EEG spectral features and EMG signals. To quantify the relative firing rates of individual neurons in different brain states, we computed and plotted REM-NREM modulation ((R\u003csub\u003eREM\u003c/sub\u003e - R\u003csub\u003eNREM\u003c/sub\u003e) / (R\u003csub\u003eREM\u003c/sub\u003e + R\u003csub\u003eNREM\u003c/sub\u003e)) versus Wake-NREM modulation ((R\u003csub\u003eWake\u003c/sub\u003e - R\u003csub\u003eNREM\u003c/sub\u003e) / (R\u003csub\u003eWake\u003c/sub\u003e + R\u003csub\u003eNREM\u003c/sub\u003e)), where R represented the mean firing rate in each state. Neurons exhibiting modulation values falling outside the light-blue shaded region, denoting to a greater than two-fold change in firing rate (|modulation| \u0026gt; 0.33), were classified as showing \u0026ldquo;state-dependent\u0026rdquo; activity\u003csup\u003e\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e,\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003e \u003cb\u003eEEG/EMG recordings and analysis.\u003c/b\u003e \u003c/p\u003e \u003cp\u003eElectroencephalographic (EEG)/electromyographic (EMG) recordings and data analyses were performed as described in previous studies\u003csup\u003e\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e,\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e. In brief, mice were implanted with two stainless steel screw electrodes served as EEG electrodes (recording location: AP\u0026thinsp;=\u0026thinsp;1.75 mm, ML\u0026thinsp;=\u0026thinsp;0.4 mm and reference location: cerebellum) together with two EMG wire electrodes, all of which were pre-soldered to a four-pin connector. The EMG electrodes were inserted into the bilateral neck muscles. After 5 d recovery, mice were connected to recording headstages via flexible recording cables and adapted for at least 2 d prior to data collection. Signals were recorded using a Microelectrode AC Amplifier Model 1800 (A-M Systems, USA), filtered (0.1\u0026ndash;500 Hz for EEG and 10\u0026ndash;500 Hz for EMG recording), and digitized at 250 Hz by Intracept Chart software. A notch filter was applied at 50 Hz by the amplifier.\u003c/p\u003e \u003cp\u003eVigilance states (wakefulness, NREM sleep, REM sleep) were classified for consecutive 4 s epochs using a custom-written MATLAB algorithm, based on EEG/EMG waveforms and power spectra. Wakefulness was identified by desynchronized, low-amplitude EEG and high EMG activity; NREM sleep was characterized by synchronized, high-amplitude, low-frequency EEG activity (0.5\u0026ndash;4 Hz) and low EMG activity relative to wakefulness; REM sleep was defined as desynchronized EEG with high power at theta frequencies (6\u0026ndash;9 Hz) and low EMG activity, reflecting muscle atonia.\u003c/p\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis.\u003c/h2\u003e \u003cp\u003eAll statistical analysis was performed using MATLAB 2014b and GraphPad Prism 10. Experimental conditions were randomly allocated, and behavioral data were analyzed using a double-blind approach. Statistical analyses comprised the Wilcoxon rank-sum test, two-tailed paired and unpaired t-tests, one-way ANOVA test and chi-squared test, with P values derived from each respective test. The significance was denoted as: *\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05; **\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.01; ***\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001; ****\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0001. See the figure legends.\u003c/p\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e \u003ch2\u003eCompeting interests\u003c/h2\u003e \u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eQ. K., P. X. and W. M. performed most of the experiments. Q. K. and P. X. performed *in vivo* single-unit and EEG/EMG recordings. W. C., P. X. and B. Y. performed Radiomics analysis. Q. K., P. X. and W. M. performed behavioral manipulation experiments and evaluations. Q. K., Y. W., D. S., W. D., X. Y. and Y. X. performed immunohistochemistry. H. Y., C. K. and W. C. designed and supervised the study. H. Y., Q. K. and C. K. wrote the manuscript. All authors read and approved the final manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eThis study was supported by National Natural Science Foundation of China (Original Exploration Program, 32250018 and National Natural Science Foundation of China，82501490), Hubei Provincial Natural Science Foundation (2024AFB720 and JCZRYB202500348), the Scientific and Technological Project of Shiyan City of Hubei Province (25Y002), and Cultivating Project for Young Scholar at Hubei University of Medicine (2022QDJZR010). We thank the Biomedical Research Institute of Hubei University of Medicine for providing an experimental platform.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eAll data supporting the findings of this study are provided within the paper and its supplementary information. All additional information will be made available upon reasonable request to the authors.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eIranzo, A. \u003cem\u003eet al.\u003c/em\u003e Neurodegenerative disease status and post-mortem pathology in idiopathic rapid-eye-movement sleep behaviour disorder: an observational cohort study. \u003cem\u003eLancet Neurol.\u003c/em\u003e 12, 443\u0026ndash;453 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChaudhuri, K. R., \u003cem\u003eet al\u003c/em\u003e, \u0026amp; National Institute for Clinical Excellence. Non-motor symptoms of Parkinson\u0026rsquo;s disease: diagnosis and management. \u003cem\u003eLancet Neurol.\u003c/em\u003e 5, 235\u0026ndash;245 (2006).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIranzo, A., \u003cem\u003eet al\u003c/em\u003e. M. Sleep and sleep disorders in people with Parkinson\u0026rsquo;s disease. \u003cem\u003eLancet Neurol.\u003c/em\u003e 23, 925\u0026ndash;937 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi, J. \u003cem\u003eet al.\u003c/em\u003e Sleep disorders in Parkinson\u0026rsquo;s disease: evaluating pharmacological treatments through systematic review and meta-analysis. \u003cem\u003eBMC Neurol.\u003c/em\u003e 25, 504 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLima, M. M. S. Sleep disturbances in Parkinson\u0026rsquo;s disease: the contribution of dopamine in REM sleep regulation. \u003cem\u003eSleep Med. Rev.\u003c/em\u003e 17, 367\u0026ndash;375 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVann, S. D. \u0026amp; Nelson, A. J. D. The mammillary bodies and memory: more than a hippocampal relay. \u003cem\u003eProg. Brain Res.\u003c/em\u003e 219, 163\u0026ndash;185 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRoy, D. S. \u003cem\u003eet al.\u003c/em\u003e Distinct Neural Circuits for the Formation and Retrieval of Episodic Memories. \u003cem\u003eCell\u003c/em\u003e 170, 1000\u0026ndash;1012.e19 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJiang, J., \u003cem\u003eet al\u003c/em\u003e. Mammillary body regulates state-dependent fear by alternating cortical oscillations. \u003cem\u003eSci. Rep.\u003c/em\u003e 8, 13471 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRuan, M., \u003cem\u003eet al\u003c/em\u003e. Bi-Directional Theta Modulation between the Septo-Hippocampal System and the Mammillary Area in Free-Moving Rats. \u003cem\u003eFront. Neural Circuits\u003c/em\u003e 11, 62 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDillingham, C. M. \u0026amp; Vann, S. D. Why Isn\u0026rsquo;t the Head Direction System Necessary for Direction? Lessons From the Lateral Mammillary Nuclei. \u003cem\u003eFront. Neural Circuits\u003c/em\u003e 13, 60 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHou, Y. \u003cem\u003eet al.\u003c/em\u003e Topographical organization of mammillary neurogenesis and efferent projections in the mouse brain. \u003cem\u003eCell Rep.\u003c/em\u003e 34, 108712 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eŻakowski, W. \u0026amp; Zawistowski, P. Neurochemistry of the mammillary body. Brain Struct. \u003cem\u003eFunct.\u003c/em\u003e 228, 1379\u0026ndash;1398 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDillingham, C. M., \u003cem\u003eet al\u003c/em\u003e. How do mammillary body inputs contribute to anterior thalamic function? \u003cem\u003eNeurosci. Biobehav. Rev.\u003c/em\u003e 54, 108\u0026ndash;119 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKumar, R. \u003cem\u003eet al.\u003c/em\u003e Reduced mammillary body volume in patients with obstructive sleep apnea. \u003cem\u003eNeurosci. Lett.\u003c/em\u003e 438, 330\u0026ndash;334 (2008).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang, Y. \u003cem\u003eet al.\u003c/em\u003e Targeting thalamic circuits rescues motor and mood deficits in PD mice. \u003cem\u003eNature\u003c/em\u003e 607, 321\u0026ndash;329 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePoewe, W. Non-motor symptoms in Parkinson\u0026rsquo;s disease. \u003cem\u003eEur. J. Neurol.\u003c/em\u003e 15 Suppl 1, 14\u0026ndash;20 (2008).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRoberts, A. G. \u003cem\u003eet al.\u003c/em\u003e Technical Feasibility of Quantitative Susceptibility Mapping Radiomics for Predicting Deep Brain Stimulation Outcomes in Parkinson Disease. \u003cem\u003eNeurosurgery\u003c/em\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1227/neu.0000000000003721\u003c/span\u003e\u003cspan address=\"10.1227/neu.0000000000003721\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLu, J., \u003cem\u003eet al\u003c/em\u003e. A putative flip-flop switch for control of REM sleep. \u003cem\u003eNature\u003c/em\u003e 441, 589\u0026ndash;594 (2006).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJankowski, M. M. \u003cem\u003eet al.\u003c/em\u003e The anterior thalamus provides a subcortical circuit supporting memory and spatial navigation. \u003cem\u003eFront. Syst. Neurosci.\u003c/em\u003e 7, 45 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAggleton, J. P. \u003cem\u003eet al.\u003c/em\u003e Hippocampal-anterior thalamic pathways for memory: uncovering a network of direct and indirect actions. \u003cem\u003eEur. J. Neurosci.\u003c/em\u003e 31, 2292\u0026ndash;2307 (2010).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWeber, F. \u0026amp; Dan, Y. Circuit-based interrogation of sleep control. \u003cem\u003eNature\u003c/em\u003e 538, 51\u0026ndash;59 (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eScammell, T. E., \u003cem\u003eet al\u003c/em\u003e. Neural Circuitry of Wakefulness and Sleep. \u003cem\u003eNeuron\u003c/em\u003e 93, 747\u0026ndash;765 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTu, H.-Q. \u003cem\u003eet al.\u003c/em\u003e Rhythmic cilia changes support SCN neuron coherence in circadian clock. \u003cem\u003eScience\u003c/em\u003e 380, 972\u0026ndash;979 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePostuma, R. B. \u003cem\u003eet al.\u003c/em\u003e MDS clinical diagnostic criteria for Parkinson\u0026rsquo;s disease. \u003cem\u003eMov. Disord. Off. J. Mov. Disord. Soc.\u003c/em\u003e 30, 1591\u0026ndash;1601 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHarris, S. S. \u003cem\u003eet al.\u003c/em\u003e Alzheimer\u0026rsquo;s disease patient-derived high-molecular-weight tau impairs bursting in hippocampal neurons. \u003cem\u003eCell\u003c/em\u003e 188, 3775\u0026ndash;3788.e21 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBoeve, B. F. REM sleep behavior disorder: Updated review of the core features, the REM sleep behavior disorder-neurodegenerative disease association, evolving concepts, controversies, and future directions. \u003cem\u003eAnn. N. Y. Acad. Sci.\u003c/em\u003e 1184, 15\u0026ndash;54 (2010).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eW, S., S, Z. \u0026amp; Dj, S. Striatal synaptic adaptations in Parkinson\u0026rsquo;s disease. \u003cem\u003eNeurobiol. Dis.\u003c/em\u003e 167, (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMarino, G. \u003cem\u003eet al.\u003c/em\u003e Intensive exercise ameliorates motor and cognitive symptoms in experimental Parkinson\u0026rsquo;s disease restoring striatal synaptic plasticity. \u003cem\u003eSci. Adv.\u003c/em\u003e 9, eadh1403 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSurmeier, D. J., \u003cem\u003eet al\u003c/em\u003e. Selective neuronal vulnerability in Parkinson disease. \u003cem\u003eNat. Rev. Neurosci.\u003c/em\u003e 18, 101\u0026ndash;113 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHunt, J. \u003cem\u003eet al.\u003c/em\u003e Sleep and circadian rhythms in Parkinson\u0026rsquo;s disease and preclinical models. \u003cem\u003eMol. Neurodegener.\u003c/em\u003e 17, 2 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDuring, E.H., \u003cem\u003eet al\u003c/em\u003e. Symptomatic treatment of REM sleep behavior disorder (RBD): A consensus from the international RBD study group - Treatment and trials working group. \u003cem\u003eSleep Med.\u003c/em\u003e 132, 106554 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eD'Amelio, M., \u003cem\u003eet al\u003c/em\u003e. Long-term survival of Parkinson\u0026rsquo;s disease: a population-based study. \u003cem\u003eJ Neurol.\u003c/em\u003e 253, 33\u0026ndash;37 (2006).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKang, J. J. \u003cem\u003eet al.\u003c/em\u003e Combining quantitative susceptibility mapping to radiomics in diagnosing Parkinson\u0026rsquo;s disease and assessing cognitive impairment. \u003cem\u003eEur. Radiol.\u003c/em\u003e 32, 6992\u0026ndash;7003 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHu, X. \u003cem\u003eet al.\u003c/em\u003e Multivariate radiomics models based on 18F-FDG hybrid PET/MRI for distinguishing between Parkinson\u0026rsquo;s disease and multiple system atrophy. \u003cem\u003eEur. J. Nucl. Med. Mol. Imaging\u003c/em\u003e 48, 3469\u0026ndash;3481 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuan, X.-J. \u003cem\u003eet al.\u003c/em\u003e A multiple-tissue-specific magnetic resonance imaging model for diagnosing Parkinson\u0026rsquo;s disease: a brain radiomics study. \u003cem\u003eNeural Regen. Res.\u003c/em\u003e 17, 2743\u0026ndash;2749 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJiang, P.-E. \u003cem\u003eet al.\u003c/em\u003e Behavioral Assessments of Spontaneous Locomotion in a Murine MPTP-induced Parkinson\u0026rsquo;s Disease Model. \u003cem\u003eJoVE\u003c/em\u003e. 143, 58653 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMagno, L. A. V., Collodetti, M., Tenza-Ferrer, H. \u0026amp; Romano-Silva, M. A. Cylinder Test to Assess Sensory-motor Function in a Mouse Model of Parkinson\u0026rsquo;s Disease. \u003cem\u003eBio-Protoc.\u003c/em\u003e 9, e3337 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYu, H. \u003cem\u003eet al.\u003c/em\u003e Periaqueductal gray neurons encode the sequential motor program in hunting behavior of mice. \u003cem\u003eNat. Commun.\u003c/em\u003e 12, 6523 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHua, R. \u003cem\u003eet al.\u003c/em\u003e Calretinin Neurons in the Midline Thalamus Modulate Starvation-Induced Arousal. \u003cem\u003eCurr. Biol. CB\u003c/em\u003e 28, 3948\u0026ndash;3959 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu, M. \u003cem\u003eet al.\u003c/em\u003e Basal forebrain circuit for sleep-wake control. \u003cem\u003eNat. Neurosci.\u003c/em\u003e 18, 1641\u0026ndash;1647 (2015).\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e "}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"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":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"the lateral mammillary nucleus, Parkinson’s disease, sleep disturbances, rescue, REM sleep, NREM sleep","lastPublishedDoi":"10.21203/rs.3.rs-9084382/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9084382/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eSleep disturbances are highly prevalent, early‑onset non‑motor symptoms in Parkinson\u0026rsquo;s disease (PD) that severely compromise quality of life. However, these neural substrates for sleep disturbance of PD remain poorly understood. Here, we reveal that the lateral part of the medial mammillary nucleus (ML) within the mammillary body (MB) is a critical hub for sleep-wake control and a key substrate of PD-related sleep dysfunction. Radiomic analysis of multi-modal MRI data from PD patients reveal that structural heterogeneity within the MB robustly predicts the presence of significant sleep disturbances, establishing it as a novel imaging biomarker. In mice, we demonstrate that ML neurons exhibit heterogeneous, state-specific activity patterns across sleep-wake cycles. Chemogenetic manipulation establishes that ML neuronal activity is both sufficient to promote NREM and REM sleep, and necessary for normal sleep architecture. Electrode recordings demonstrate that broad suppression of ML neuronal firing alongside a paradoxical hyperactivity of REM-active subpopulations. Strikingly, chemogenetic activation of ML neurons rescues these sleep disturbances, increasing total, NREM, and REM sleep without altering motor function. Therefore, our findings reveal a pathological neural mechanism underlying sleep dysfunction in PD and highlight the ML as a promising therapeutic target for PD-related sleep disorders.\u003c/p\u003e","manuscriptTitle":"Targeting the lateral mammillary nucleus rescues sleep disturbances in Parkinson’s disease","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-04 09:42:04","doi":"10.21203/rs.3.rs-9084382/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"3c2cc83e-9d9f-487c-9a1b-cb25cee6bccd","owner":[],"postedDate":"May 4th, 2026","published":true,"recentEditorialEvents":[{"type":"decision","content":"Rejected","date":"2026-04-30T13:27:47+00:00","index":"","fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":67118181,"name":"Health sciences/Neurology"},{"id":67118183,"name":"Biological sciences/Neuroscience"}],"tags":[],"updatedAt":"2026-05-05T09:58:07+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-04 09:42:04","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9084382","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9084382","identity":"rs-9084382","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}