Targeting the posterior thalamic hub: hemispheric-specific encoding of pain-depression comorbidity after hemorrhagic stroke Authors: Yujia Jin, Yating Lv, Xinye Wu, Chengjia Liu, Wanjun Guo, Yihui Cui, Yiyan Dong, Lusha Tong Published: 2026 ABSTRACT AbstractBackgroundPost-stroke pain (PSP) and post-stroke depression (PSD) frequently co-occur after thalamic hemorrhage and may mutually exacerbate one another, with persistent pain contributing to affective dysfunction and depression altering sensory processing. Because current symptom-specific treatments often provide limited benefit, identifying shared neural mechanisms underlying PSP–PSD comorbidity may offer a more clinically relevant framework for developing targeted circuit-based therapies.MethodsTwelve patients with isolated anterior or posterior thalamic hemorrhage underwent parallel multimodal assessments, including standardized pain, mood, anxiety, and cognitive evaluations together with structural MRI, diffusion tensor imaging, and resting-state functional MRI to characterize lesion topography and thalamocortical connectivity. Guided by the clinical findings, we subsequently established nucleus-specific murine models of thalamic hemorrhage using stereotactic collagenase injection. Mice underwent sequential behavioral phenotyping, fiber photometry recording, and viral tracing analyses to investigate the circuit mechanisms underlying pain–depression comorbidity.ResultsIn patients, posterior thalamic hemorrhage was the exclusive locus for PSP-PSD comorbidity and exhibited severe disruption of thalamocortical functional connectivity, with right-sided lesions having the most profound effect. In mice, focal hemorrhage confined to the posterior thalamic nucleus (PO), but not adjacent parafascicular or ventral posterolateral nuclei, recapitulated both mechanical hyperalgesia and delayed depression-like behaviors. In the murine model, we uncovered a striking hemispheric specialization: left PO lesions induced anhedonia, while right PO lesions led to behavioral despair. PO neurons became hyperexcitable to pain after hemorrhage, a state amplified with PSD onset. Viral tracing revealed lateralized PO projections to somatosensory cortices.ConclusionsThe posterior thalamic nucleus represents a lateralized thalamocortical substrate for post-stroke pain–depression comorbidity after hemorrhagic stroke. These findings support the potential value of lesion-topography-based prognostic stratification and circuit-guided neuromodulation strategies targeting both sensory and affective symptoms. A major limitation of this study is the relatively small but carefully phenotyped clinical cohort, which warrants validation in larger longitudinal studies.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12916-026-05014-4.Keywords: Posterior thalamic nucleus, Pain-depression comorbidity, Thalamocortical circuitry, Hemispheric lateralization, Thalamic hemorrhage, Functional connectivity Background Intracerebral hemorrhage (ICH) is characterized by parenchymal bleeding and the accumulation of neurotoxic blood-derived products [1–3]. As a devastating stroke subtype affecting approximately two million individuals worldwide each year, ICH is associated with high mortality and substantial long-term disability among survivors [2]. Neuropsychiatric sequelae are particularly burdensome following thalamic hemorrhage. Clinical studies have shown that post-stroke pain (PSP) affects up to 80% of thalamic stroke survivors [4–8], while post-stroke depression (PSD) prevails in 40% of ICH population [9–11]. Importantly, PSP and PSD frequently co-occur and may bidirectionally exacerbate one another, with persistent pain contributing to affective dysfunction and depression altering sensory processing [9, 12–17]. However, current symptom-specific treatments often provide limited benefit, partly because the shared neural mechanisms underlying PSP–PSD comorbidity remain poorly understood [8, 9, 18–20]. The thalamus is composed of multiple anatomically and functionally distinct nuclei that participate in diverse sensory, cognitive, and affective processes [21–24]. Although relatively small in size, the posterior thalamus contains several specialized subnuclei with distinct corticothalamic and thalamocortical connectivity patterns that contribute to different modes of sensory and affective information processing [14, 25–27]. This anatomical heterogeneity suggests that the thalamus functions as an integrative thalamocortical network rather than a uniform sensory relay structure. Functionally, the thalamus plays an important role in integrating sensory and affective processes, including nociceptive signaling and emotional regulation [8, 28–36]. Clinical neuroimaging studies have shown that disruptions in thalamocortical connectivity are associated with both central neuropathic pain and depressive symptoms [28, 29, 37, 38]. Emerging evidence further suggests hemispheric specialization within thalamic networks, with left-sided lesions more commonly associated with verbal and cognitive dysfunction and right-sided lesions linked to visuospatial and affective disturbances [39]. However, several unresolved questions remain, including which thalamic subnuclei contribute to post-stroke pain–depression comorbidity, the neural mechanisms underlying their co-occurrence, and the extent to which hemispheric specialization influences affective–nociceptive integration following intracerebral hemorrhage. To address these questions, we recruited 12 survivors with isolated thalamic intracerebral hemorrhage (6 anterior and 6 posterior) from our prospective ICH cohort. Participants underwent standardized clinical assessments together with multimodal neuroimaging, including resting-state functional MRI (rs-fMRI) and diffusion tensor imaging, to compare anterior and posterior thalamic hemorrhage and evaluate their associations with PSP–PSD comorbidity and thalamocortical connectivity alterations. Guided by the clinical findings, we subsequently established a targeted murine model of thalamic hemorrhage using stereotactic collagenase injection to generate nucleus-restricted lesions. We focused on three posterior thalamic nuclei with distinct but complementary roles in nociceptive and affective processing: the ventral posterolateral nucleus (VPL), a classical first-order somatosensory relay involved in sensory–discriminative aspects of pain [28, 30, 37]; the parafascicular nucleus (PF), an intralaminar nucleus implicated in pain-related arousal and affective–motivational regulation [14, 27, 40, 41]; and the posterior thalamic nucleus (PO), a higher-order integrative nucleus with extensive thalamocortical connections that has been increasingly linked to pathological pain amplification and affective modulation [14, 42, 43]. Based on this framework, we hypothesized that: [1] a specific thalamic nucleus serves as a shared substrate for PSP–PSD comorbidity; [2] localized thalamic hemorrhage in mice could reproduce the temporal progression and hemispheric lateralization of pain and affective symptoms observed in patients; [3] pathological thalamic hyperexcitability and thalamocortical circuit dysregulation contribute to the development of PSP–PSD comorbidity. Methods Human study Thalamic hemorrhage cohort We retrospectively reviewed data from a previously established ICH cohort at the Department of Neurology, Second Affiliated Hospital, Zhejiang University School of Medicine [44]. The cohort included consecutive patients with primary ICH confirmed by computed tomography (CT) or MRI. Patients with secondary causes of ICH, including trauma, vascular malformations, or brain tumors, were excluded. For the present study, additional inclusion criteria were: [1] age ≥ 18 years; and [2] radiologically confirmed isolated thalamic hemorrhage restricted to either the anterior or posterior thalamic subdivisions. Exclusion criteria included: pre-stroke psychiatric disorders (including depression or anxiety), chronic pain conditions (including rheumatoid arthritis, cancer-related pain, or gout), dementia or neurodegenerative diseases, previous non-lacunar stroke, age > 80 years, and contraindications to MRI. Eligible participants were recalled for parallel multimodal clinical and neuroimaging assessments: [1] multimodal 3T MRI including high-resolution structural MRI (sMRI), susceptibility-weighted imaging (SWI), diffusion tensor imaging (DTI), and rs-fMRI to delineate hemorrhage topography and thalamocortical connectivity; and [2] standardized clinical phenotyping using validated instruments for pain, depression, anxiety, and global cognition. Following rigorous screening, 12 patients met all criteria: 6 with isolated anterior thalamic intracerebral hemorrhage (atICH) and 6 with posterior thalamic hemorrhage (ptICH), with hemorrhage topography confirmed by two independent neuroradiologists (Y.J. and C.L.) using established anatomic landmarks. Healthy control cohort We retrospectively reviewed data from a previously established imaging cohort of healthy community-dwelling older adults (> 60 years) without a history of stroke [45]. Participants were recruited through community-based sampling and underwent standardized neuroimaging assessments. Exclusion criteria included a history of brain trauma, neurological disorders (including neurodegenerative diseases or epilepsy), psychiatric disorders (including major depression or anxiety disorders), or systemic diseases with potential cerebral involvement. Common age-related vascular comorbidities, including hypertension, diabetes mellitus, and hyperlipidemia, were permitted. All healthy control participants underwent the same MRI acquisition protocol as the thalamic hemorrhage cohort, including sMRI, SWI, DTI, and rs-fMRI. Pain, emotion and cognition evaluation Pain and neuropsychiatric outcomes were assessed using standardized questionnaires and structured interviews designed to capture participants’ symptoms since stroke onset up to the time of assessment (i.e., post-stroke course), rather than only transient states at the return-visit time point. Pain symptomatology was quantified using three validated instruments: Visual Analogue Scale (VAS) [46], a unidimensional 100-mm line wherein patients mark pain intensity between “no pain” (0 mm) and “worst imaginable pain” (100 mm), with severity calculated as the millimeter distance from the origin; Numeric Rating Scale (NRS) [47], an 11-point Likert-scale adaptation of the VAS (0 = no pain; 10 = maximum pain) requiring patients to verbally rate current pain intensity; and Douleur Neuropathique 4 Questions (DN4) [48], a clinician-administered 10-item tool discriminating neuropathic from nociceptive pain through symptom descriptors (7 items) and sensory examination signs (3 items). Each DN4 item is scored dichotomously (1 = present; 0 = absent), with total scores ≥ 4/10 indicating neuropathic pain. Concurrently, affective domains were assessed using: [1] the Hamilton Anxiety Scale (HAMA), evaluating somatic (e.g., muscular, sensory) and psychic (e.g., tension, fears) anxiety subdomains; [2] depression metrics via the self-reported Patient Health Questionnaire-9 (PHQ-9) and clinician-rated 24-item Hamilton Depression Scale (HAMD-24), the latter capturing seven symptom dimensions: somatization, weight change, cognitive impairment, diurnal variation, psychomotor retardation, sleep disturbance, and hopelessness. Global cognition was measured by the Mini-Mental State Examination (MMSE), a 30-point screening tool assessing five domains: orientation (temporal/spatial), registration (immediate recall), attention/calculation, recall (delayed memory), and language abilities (naming, repetition, comprehension). All instruments were administered by trained neurologists blinded to hemorrhage topography. Multimodal MRI data acquisition The MRI data were acquired on a 3-Tesla scanner (Siemens MAGNETOM Vida) at Department of Neurology of Second Affiliated Hospital, Zhejiang University School of Medicine. During the scanning, all participants were instructed to stay awake, relax with their eyes closed and hold still. Please see detailed scanning parameters for MRI data: rs-fMRI, sMRI, DTI, SWI in Supplementary Methods. Imaging processing Hemorrhagic lesion map Using ITK-SNAP software (https://www.itksnap.org), one neurologist (Y.J. with 5 years of experience), blinded to clinical information of the patients, manually traced hemorrhagic lesion mask based on SWI image for each patient. The lesion masks were further normalized to the MNI space via deformation fields derived from tissue segmentation of individual sMRI images. Preprocessing of rs-fMRI data The rs-fMRI data were preprocessed using the Data Processing & Analysis for Brain Imaging (DPABI, http://rfmri.org/dpabi) based on SPM12 (http://www.fil.ion.ucl.ac.uk/spm) and included following steps [49]: [1] removal of the first 20 volumes; [2] correction for inter-volume head motion based on rigid-body transformation; [3] spatial normalization to the MNI space by deformation fields derived from tissue segmentation of sMRI images; [4] spatial smoothing with an isotropic Gaussian kernel of 6 mm full width at half maximum (FWHM); [5] removal of linear trends; [6] regressing out nuisance signals including 24-parameter head motion profiles [50], white matter signals, cerebrospinal fluid signals, and other artifacts using ICA-based denoising approach. The global signals were not regressed out due to its controversy for rs-fMRI data [51]; [7] band-pass filtering (0.01–0.08 Hz). Notably, no participants were excluded based on the head movement criterion of maximal displacement > 3 mm, rotation > 3°. Functional connectivity analyses As proposed in previous study [52], the thalamus in each hemisphere was subdivided into anterior thalamus (aTHA) and posterior thalamus (pTHA). Four regions of interest (ROIs) (aTHA.L, aTHA.R, pTHA.L, pTHA.R) were then selected as seeds for functional connectivity (FC) analyses. Voxel-wise FC analyses: for each participant in both stroke patients with ptICH and healthy controls (HCs), the Pearson correlation analyses were performed between the averaged time course of each pTHA ROI and the time course of every brain voxel. The resultant correlation coefficients were then converted to z-value using Fisher’s r-to-z transformation. ROI-wise FC analyses: additionally, to measure the specific connectivity changes of pTHA.R with sensorimotor areas, the Pearson correlation coefficients were calculated between averaged time course of pTHA.R and each ROI in primary somatosensory cortex and primary motor cortex based on the Brodmann Area (BA) atlas (8 ROIs, BA 1–4 in each hemisphere) in ptICH and HCs groups. This resulted in a total of 8 paired-wise FC for each participant. The resultant correlation coefficients were further transformed to z-value. Similarly, for each participant in both stroke patients with atICH and HCs, the Pearson correlation coefficients were calculated between the averaged time course of each aTHA ROI and the time course of every brain voxel, and further transformed to z-value. Diffusion image processing The motion and geometrical distortion of individual DTI images were first corrected using the EDDY tool as implemented in the FMRIB Software Library (FSL) (http://www.fmrib.ox.ac.uk/fsl). Then, the deterministic tractography was performed in the native diffusion space using TrackVis software (https://www.trackvis.org). Fiber Assignment by Continuous Tracking (FACT) algorithm was used to reconstruct white matter fiber pathways (streamline tractography) in the brain, with angle threshold of 45° and fractional anisotropy (FA) threshold set at 0.2. For each participant, the whole-brain streamline tractography was transformed to the MNI space using the trackmath tool from the Tract Querier software package (https://github.com/demianw/tract_querier) via deformation fields derived from normalization of sMRI images using ANTs (https://github.com/ANTsX/ANTs). Subsequently, each of four ROIs in thalamus (aTHA.L, aTHA.R, pTHA.L, pTHA.R) was selected to isolate those fibers passing through the ROI. The union of all tractographies of each ROI in each group was generated using the function ‘tract_merge’ in the trackmath tool. The group tractography of each ROI was displayed after removing short tracts (streamline length < 100 mm) using ‘tract remove_short tracts’ function in the trackmath tool. Due to the limited sample size, tractography results were used for descriptive visualization of structural connectivity patterns rather than quantitative statistical comparison between groups. Murine study Reagents Rabbit anti-NeuN was from Millipore. Goat anti-rabbit (Alexa Fluor 647) was from Thermo Fisher Scientific. AAV2/2-hSyn-mCherry-WPRE-pA was purchased from Shanghai Taitool Bioscience Co. AAV2/5-hSyn-EGFP was from Sunbio Medical Biotech Co., Ltd. AAV2/1-CaMKIIa1.3-GCaMP6s-WRPE was from OBiO Technology (Shanghai) Corp, Ltd. DAPI was purchased from Beyotime. Collagenase IV was purchased from Sigma (C5138). Animals 7 ~ 10-week aged male C57BL/6J mice (SLAC Laboratory Animal Co., Shanghai) were used if not specified. Male mice, aged 8–12 weeks, were used for behavior test. Two to five mice were group-housed at 23 ± 1 °C on a 12 h light–dark cycle (light on from 8a.m. to 8p.m.) with unrestricted access to food and water. Each cage had a constant position throughout the experiment. Animals were randomly assigned to different experimental groups. Sample sizes were estimated based on our previous experience from similar studies and were sufficient for statistical analyses (Supplementary Table 1) [25, 53, 54]. Only male mice were used in this study. This choice was made to minimize biological variability related to the estrous cycle, which is known to influence nociceptive sensitivity and affective behaviors [55]. Using male mice allowed for more stable baseline behavioral and neurophysiological measurements and facilitated detection of nucleus-specific effects in the context of hemorrhagic stroke. Stroke model Solid powder of Collagenase IV was dissolved in phosphate-buffered saline (PBS) and diluted to 1 U/ml. Before stereotaxic surgeries, mice were deeply anesthetized with pentobarbital NEMBUTAL (100 mg/kg, Sigma-Aldrich, intraperitoneal injection (i.p.)). Collagenase was initially sealed in a glass pipette which was fixed on a stereotactic apparatus and then were injected into the following brain regions: extensive thalamic hemorrhage (AP -1.61 mm from bregma, ML ± 1.50 mm, DV -3.05 mm from the dura); parafascicular thalamic nucleus (PF) (AP -2.19 mm from bregma, ML ± 0.56 mm, DV -3.25 mm from the dura); PO (AP -2.20 mm from bregma, ML ± 1.33 mm, DV -3.25 mm from the dura); VPL (AP -1.90 mm from bregma, ML ± 1.13 mm, DV -3.50 mm from the dura). Lesions were made in either the left or right hemisphere, as indicated by the ML ± X mm coordinates, and were not intended for bilateral injections. Collagenase was injected with 25nL per site inducing extensive thalamic hemorrhage, 5 nL per site into VPL, and 6 nL per site into PO and PF at a rate of 5 nL/min. The glass pipette was withdrawn 10 min after ending injection. After the surgeries, mice were transferred to a heat pad for anesthesia recovery. An anterior thalamic hemorrhage model was also established (AP − 0.85 mm from bregma, ML ± 0.75 mm, DV − 3.15 mm from the dura), with 7 nL collagenase injected. Behavioral testing Behavioral testing was performed on adult male mice by investigators blinded to the experimental groups. Mice were assessed for neuropsychiatric-like phenotypes following stroke using a battery of established tests. Sensory allodynia was evaluated using the von Frey test [56, 57]. Depression-like behavior was assessed using the forced swim test (FST) and sucrose preference test (SPT). Anxiety-like behavior was measured using the elevated plus maze (EPMT), light-dark box (LDBT), and open field tests (OFT). Motivational and compulsive behaviors were evaluated with the novelty-suppressed feeding test (NSFT), and social interaction (SIT) was assessed in a dedicated social test. Detailed descriptions of all behavioral procedures are provided in the Supplementary Methods. Virus injection Before stereotaxic surgeries, mice were deeply anesthetized with pentobarbital NEMBUTAL (100 mg/kg, Sigma-Aldrich, intraperitoneal injection (i.p.)). Viruses were initially sealed in a glass pipette which was fixed on a stereotactic apparatus and then were injected into the PO (AP -2.20 mm from bregma, ML ± 1.33 mm, DV -3.25 mm from the dura). Viruses were injected with 80–100 nL per site at a rate of 100–150 nL/min into the brain regions. The glass pipette was withdrawn 10 min after ending injection. After the surgeries, mice were transferred to a heat pad for anesthesia recovery. The injection sites were verified by post immunohistochemistry. For fiber photometry recordings, optic fibers (200 mm diameter, 0.37 numerical aperture, Inper Ltd., China) were implanted 200 mm above the viral injection site, and cemented onto the skull using dental cement. Fiber photometry von Frey For fiber photometry, mice were habituated for 2 h in small chambers. After the baseline recording, von Frey filament (gram = 0.07) was used to poke the contralateral hindpaw from below to elicit instantaneous mechanical allodynia. Each mouse was tested with at least 5 pokes, with a 15-second interval between each poke. The inter-stimulus interval was maintained until the real-time calcium signal returned to baseline levels before proceeding with the next trial. Tail suspension (TS) For fiber photometry, after the baseline recording in the homecage, the tails of mice were held by experimenters and mice were suspended for 3–5 s every time with a 15–30 s interval. Restraint (Res) For fiber photometry, after the baseline recording in the homecage, mice were subjected to physical restraint applied by experimenter’s left hand for 3–5 s every time with a 15–30 s interval. Footshock (FS) Mice were individually placed into a footshock chamber with an electrifiable metal grid floor. For fiber photometry, mice were habituated for 1 min in the chamber and then subjected to 5 shocks (intensity: 0.8 mA; duration: 1 s; interval: random) in ~ 10 min and then transferred to the homecage. Recording and data analysis GCaMP6s fluorescence signal was collected at the frequency of 40 Hz through the optic fiber with a fiber photometry system (Inper Ltd., China). Ca2+-dependent fluorescence signal was acquired under the stimulation of a 470-nm LED (20 ~ 30 µW at the fiber tip), while the Ca2+-independent isosbestic signal was acquired under the stimulation of a 410-nm LED (20 ~ 30 µW at the fiber tip). During recording, a video camera (Daheng Imaging) was placed directly above the tested mouse to track its behaviors. Besides, a screen recorder (EV Capture, Hunan Yiwei) was used to simultaneously capture the mouse behaviors and fluorescence signals, aiming to align the time points of behavioral events and corresponding calcium activity. Before each recording session, mice were habituated to the new environment for at least 1 h. After 2-min baseline recording in the homecage, each mouse separately went through mechanical pain induced by von Frey (gram = 0.07) or TS, Res, FS. Mouse behaviors were manually annotated via BORIS software [58]. To calculate the Z score of fluorescence signals, raw data was exported by processing software (Inper Ltd., China) and analyzed with custom-made MATLAB codes and also codes from Thinker Tech Nanjing Biotech Co., Ltd. 410-nm signal was linearly scaled with the least-squares regression to minimize the difference with 470-nm signal. The scaled 410-nm signal was then subtracted from the 470-nm signal to correct the movement artifacts and photo-bleaching effect. After these corrections, Z score was used to optimize the comparison between hemorrhagic stroke and non-stroke neurons in PO. Z score was presented by heatmaps or average plots with a shaded area to indicate the SEM. Fiber photometry signal processing and bleaching correction: The fluorescence signals from the 470 nm (calcium-dependent) and 410 nm (isosbestic) channels were recorded simultaneously. A robust linear fit was applied to the 410 nm signal, and the fitted signal was subtracted from the 470 nm signal to correct for photobleaching and slow baseline drift: Z-score normalization: Following the bleaching correction, the calcium-dependent fluorescence signal was normalized using the Z-score method to account for any remaining fluctuations. The Z-score was calculated as follows: where F(t) is the fluorescence signal at time t, and µ(baseline) and σ(baseline) are the mean and standard deviation of the baseline signal, respectively. Histological verifications and immunohistochemistry To ensure the accuracy and consistency of collagenase-induced lesions, histological verification was performed. Mice were perfused and brains were sectioned coronally (50 μm). Sections were subjected to NeuN immunostaining to visualize neuronal loss and delineate lesion boundaries. Lesion location was confirmed by matching the lesion core and peri-lesional tissue damage with the corresponding stereotaxic coordinates in the Paxinos and Franklin mouse brain atlas. Because collagenase induces hemorrhage through vascular degradation rather than focal mechanical injury, lesion spread is not sharply confined to anatomical borders and may extend into peri-nuclear regions. Stereotaxic targeting is technically more challenging and minor lesion spread may occur even with accurate injection placement. Therefore, we allowed limited peri-nuclear extension if: [1] the lesion core was centered within the intended nucleus [2], the majority of tissue disruption was located within that nucleus, and [3] lesion spread did not substantially involve functionally distinct adjacent nuclei known to mediate pain-affective processing. For photometry, to verify viral expression and fiber placement, brains were collected after completion of experiments and sectioned coronally (50 μm). Native fluorescence of GCaMP6s (EGFP) was used to confirm expression within the target nucleus (PO). Optic fiber tracks were identified histologically, and animals were included only when fiber tips were located directly above the viral expression zone within PO. Animals with misplaced fibers, off-target viral expression, weak or absent fluorescence, or excessive spread beyond the intended thalamic region were excluded. Raw fluorescence signals were visually inspected for baseline stability and signal-to-noise ratio. Recordings with unstable baselines, excessive bleaching, or motion artifacts that could not be corrected by isosbestic regression were excluded. Mice were anesthetized with 1% pentobarbital NEMBUTAL (100 mg/kg) and perfused with 20-mL PBS and 20-mL 4% PFA (paraformaldehyde) in PBS (pH 7.4). The brains were then extracted, post-fixed in 4% PFA at 4 °C for 24 h, and dehydrated in 30% sucrose in PBS at 4 °C for 48 h. Brains were cryo-sectioned into coronal slices at 50-µm thickness via a cryostat microtome (Thermo Scientific NX50) and stored in PBS at 4 °C before immunostaining. Free-floating slices were then transferred into the blocking buffer (5% normal goat serum and 0.4% Triton X-100 in PBS) for 1 h at room temperature. For NeuN immunostaining, slices were then incubated within the primary antibody (rabbit anti-NeuN, Millipore, 1:500) dissolved in antibody solution (0.3% Triton X-100, 1% bovine serum albumin in PBS) for 48 h at 4 °C. Slices were then rinsed with PBS for 3 times (10 min each) and incubated for 2 h at room temperature with the secondary antibody (goat anti-rabbit 647, Thermofisher Scientific, 1:1000). After incubation with the secondary antibody, brain slices were washed with PBS for 3 times (10 min each) and mounted. Fluorescent images were acquired by an Olympus VS120 microscope. Statistical analyses Human studies Patient characteristics are presented as mean ± standard deviation (SD) for normally distributed data, and median (interquartile range, IQR) for skewed data. Between-group (atICH vs. ptICH) comparisons were performed using Mann–Whitney U tests for non-normally distributed continuous variables, independent t-tests for normally distributed variables, and Fisher’s exact test for categorical variables. For each ROI, voxel-wise FC differences between groups were assessed using two-sample t-tests. The resulting statistical parametric maps (T-maps) were thresholded at a voxel-level significance of p < 0.05, combined with a cluster extent threshold of > 20 contiguous voxels. Between-group differences in the 8 pre-defined pair-wise FC values were also evaluated using two-sample t-tests, with significance set at p < 0.05. Within the ptICH patient group, Spearman’s rank-order correlation analysis was employed to examine the associations between each pairwise connectivity (posterior thalamus right, pTHA.R, to sensorimotor areas) and clinical measures of pain, anxiety, depression, and cognitive function. Additionally, a series of analyses were further performed to validate main findings of voxel-wise FC: [1] Head motion correction alone may be insufficient to eliminate physiological noise in resting-state fMRI data. Thus, we re-analyzed between-group differences in voxel-wise FC after regressing out artifacts using ICA-based denoising approach [2]. The T-maps of between-group difference in voxel-wise FC were further corrected for multiple comparisons using Gaussian Random Field theory (GRF) approach (voxel p < 0.05, cluster p < 0.05, GRF corrected) [3]. Hemorrhage occurred in different hemispheres across six ptICH patients (2 in left side, 3 in right side and 1 bilateral). To assess the possible effects of lesion laterality, voxel-wise FC differences were re-analyzed between right ptICH patients and healthy controls. Murine studies Required sample sizes were estimated based on effect sizes observed in prior, analogous experiments. Mice were randomly assigned to experimental groups for all procedures. Data analysis was conducted using GraphPad Prism software (Version 10). Datasets were excluded from analysis if histological verification confirmed that viral vector or collagenase injection sites lay outside the target brain regions. Prior to inferential testing, the normality of data distribution was evaluated using the D’Agostino & Pearson omnibus normality test; for sample sizes smaller than 8, the Shapiro-Wilk test was applied. Homogeneity of variances was assessed using the Brown–Forsythe test. Statistical significance was defined as p < 0.05. For data satisfying assumptions of normality and equal variance, between-group comparisons were made using two-tailed unpaired Student’s t-tests, or paired t-tests where appropriate. Non-normally distributed data were analyzed using non-parametric tests (Mann-Whitney U-test for unpaired data). All murine data are expressed as mean ± standard error of the mean (S.E.M.). Results Posterior thalamic hemorrhage underlies pain-depression comorbidity with distinct functional connectivity profiles Our study cohort comprised 12 patients with isolated thalamic hemorrhage, stratified into two groups based on hemorrhage topography: 6 atICH and 6 ptICH (Fig. 1A). Demographic and clinical characteristics are summarized in Supplementary Table 2. Lesion mapping confirmed the anatomical segregation of hemorrhages of atICH and ptICH (Fig. 1B, D). The time from stroke onset to the study visit differed between groups (median ~ 15 months in ptICH vs. ~ 40 months in atICH, Supplementary Table 2). Standardized clinical assessments revealed a clear divergence in neurological sequelae: patients with atICH exhibited minimal post-stroke pain, anxiety, or depressive symptoms, whereas those with ptICH presented prominent pain, with higher levels of anxiety and depressive scale scores (Supplementary Table 2). These findings implicate the posterior thalamus as a critical hub for PSP–PSD comorbidity. Neuroimaging and functional connectivity profiles in patients with thalamic hemorrhagic stroke. (A) Schematic overview of patient recruitment, multimodal MRI acquisition, and standardized clinical assessment. (B) Representative lesion overlap map of anterior thalamic hemorrhage (atICH). Hemorrhagic lesions are depicted in yellow. (C ) Whole-brain functional connectivity (FC) differences between atICH patients and healthy controls after regressing out artifacts using ICA-based denoising approach. Seed-based analysis using left (top) and right (bottom) anterior thalamus (aTHA) as regions of interest (voxel-wise 𝑝<0.05, cluster size > 20). Blue indicates decreased FC in patients, red indicates increased FC in patients. (D) Lesion overlap map of posterior thalamic hemorrhage (ptICH). Hemorrhagic lesions are shown in blue. (E) Whole-brain FC differences between ptICH patients and healthy controls after regressing out artifacts using ICA-based denoising approach (voxel-wise 𝑝<0.05 cluster size > 20). Seeds placed in left (top) and right (bottom) posterior thalamus (pTHA). Blue: decreased FC in patients; red: increased FC in patients. (F) Seed regions in left (top) and right (top) aTHA, and left (bottom) and right (bottom) pTHA. (G) Bar graph showing significantly reduced FC between the right posterior thalamus (pTHA.R) and bilateral sensorimotor cortices (Brodmann areas 1–4) in ptICH patients compared to healthy controls. *𝑝<0.05, #𝑝<0.1; two-sample t-test FC analysis revealed distinct patterns of thalamocortical network disruption depending on anterior and posterior location. Both atICH and ptICH thalamic hemorrhage groups exhibited significant alterations in functional connectivity compared to healthy controls (voxel-wise p < 0.05, cluster size > 20). Notably, the laterality of the seed region itself significantly influenced FC profiles: when using left and right posterior thalamic (pTHA) seeds in ptICH patients, each produced divergent cortical connectivity patterns (Fig. 1C, E, F; Supplementary Tables 3–6). The most extensive FC disruptions were observed in ptICH patients specifically when using the right posterior thalamus (pTHA.R) as the seed region. Moreover, validation analyses indicated that the between-group differences in voxel-wise FC were not affected by lesion laterality. That is, differences of right pTHA-seeded FC and left pTHA-seeded FC remained largely unchanged when compared between right ptICH patients and healthy controls (Supplementary Fig. 1). However, decreased FC between right posterior thalamus and bilateral sensorimotor cortex was only observed in ptICH patients when correcting for multiple comparisons (voxel p < 0.05, cluster p < 0.05, cluster size > 283 voxels, GRF corrected) (Supplementary Fig. 2). Seed-based pairwise analysis further demonstrated significantly reduced FC between pTHA.R and key sensorimotor regions—including right BA1, bilateral BA3, and bilateral BA4—relative to healthy controls (Fig. 1G). These findings highlight: first, atICH and ptICH engage distinct neural networks; second, there exists notable hemispheric lateralization within the thalamus, with lesions of the right posterior thalamus conferring the most severe functional connectivity disruptions. Furthermore, we observed trends linking clinical scores with FC strength within the ptICH group. Pain intensity (especially VAS) showed a positive correlation with FC between pTHA.R and sensorimotor regions (Supplementary Fig. 3, Supplementary Table 7), while anxiety and depression scores (HAMA, PHQ-9 and HAMD) exhibited negative correlations with these connectivity measures (Supplementary Fig. 4, Supplementary Tables 8 and 9). Cognitive function (assessed by MMSE) was not affected (Supplementary Fig. 5, Supplementary Table 10). Although these correlations did not reach significance—likely due to limited sample size (n = 6)—they suggest a potential relationship between clinical severity and thalamocortical network integrity. Supporting these functional findings, DTI tractography indicated broader and more extensive structural connectivity between the posterior thalamus and the cortex compared to anterior thalamic regions (Supplementary Fig. 6). These results demonstrate that [1] the posterior thalamus is a convergent substrate for PSP-PSD comorbidity [2], thalamic hemorrhages produce location- and lateralization-dependent disruptions in thalamocortical circuitry, and [3] right posterior thalamic hemorrhage manifests the most profound dysconnectivity, particularly with ipsilateral somatosensory and bilateral sensorimotor cortices. To establish causality and investigate the underlying subnuclear and circuit mechanisms, we proceeded to a parallel series of experiments in a murine model of thalamic hemorrhage. Establishing murine models of subregion-specific thalamic hemorrhagic stroke In order to create a model of PSP and PSD, we first replicated a murine model of extensive thalamic hemorrhage by stereotactic injection of collagenase (collagenase IV, 1 U/mL, 25nL; Fig. 2A-C). Histological analysis (NeuN staining) revealed lesion encroachment into a number of subregions (Fig. 2B, C; Supplementary Fig. 7A), among which PF, PO and VPL nuclei—regions implicated in nociception were given special attention [14, 28]. We then induced hemorrhagic stroke with different collagenase dosage in separate brain regions (PF, PO and VPL) to observe the consequences of hemorrhagic stroke in different regions affected (Fig. 2D-L, Supplementary Fig. 7B-D). Mapping of thalamic hemorrhagic stroke. c57 mice were sacrificed one day after collagenase injection. (A) Schematics of collagenase-induced extensive thalamic hemorrhage. (B) Representative images of extensive thalamic hemorrhage (middle) and affected neurons (right) by infusion of collagenase IV. Scale bar, 200 μm. Red dashed line indicates hemorrhage area. (C) Schematics of extensive thalamic hemorrhage induced by collagenase. Blue patterns represent the hemorrhage area. (D, G and J) Schematics of collagenase-induced hemorrhagic stroke in right PF (D), PO (G), and VPL (J). (E, H and K) Representative images of hemorrhagic stroke (middle) and affected neurons (right) in PF (E), PO (H), and VPL (K). Scale bar, 200 μm. Red dashed line indicates hemorrhage area. (F, I and L) Schematics of hemorrhagic stroke in PF (F), PO (I) and VPL (L). Blue patterns represent the hemorrhage area. PF, parafascicular nucleus; PO, posterior thalamic nucleus. VPL, ventral posterolateral nucleus Therefore, we established a total of 4 thalamic hemorrhagic stroke models: Extensive thalamic hemorrhage: 25 nL collagenase disrupted a number of subnuclei including PF, PO, and VPL (Fig. 2A-C, Supplementary Fig. 7A). Extensive thalamic hemorrhage: 25 nL collagenase disrupted a number of subnuclei including PF, PO, and VPL (Fig. 2A-C, Supplementary Fig. 7A). Focal lesions: 5–6 nL collagenase restricted bleeding to single nuclei (PF, PO and VPL; Fig. 2D-L, Supplementary Fig. 7B-D). Focal lesions: 5–6 nL collagenase restricted bleeding to single nuclei (PF, PO and VPL; Fig. 2D-L, Supplementary Fig. 7B-D). To determine the bleeding range for modeling, mice were subjected to NeuN immunostaining (Fig. 2 and Supplementary Fig. 7). In this study, at the beginning, all stroke models were induced in the right hemisphere, as previous studies utilizing hemorrhagic stroke models have typically targeted the right side. This is consistent with the established protocols in the literature [8, 30]. In Figs. 3 and 4, all lesions were localized to the right hemisphere. In Fig. 5, we further examined left-sided PO hemorrhage to perform a comparison between the left and right hemispheres, providing a stepwise progression of the experiment. PO governs thalamic hemorrhage-induced post-stroke pain. (A-C) Mechanical pain thresholds (assessed by von Frey filament testing) in the acute phase (7–9 days post-stroke) following extensive thalamic hemorrhage (A), focal PO hemorrhage (B), or focal PF hemorrhage (C). (D-F) Chronic-phase mechanical hypersensitivity (23 days post-stroke) in mice with extensive thalamic hemorrhage (D), PO hemorrhage (E) or PF hemorrhage (F). (G) Longitudinal trajectory of mechanical pain thresholds following extensive thalamic hemorrhage, demonstrating early onset (day 3) and sustained hyperalgesia through chronic stages (day 23). Data are mean ± SEM; n = 10–19 mice per group. Mann-Whitney test for (A-E and G); unpaired t test for (F). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 vs. sham; n.s. = not significant. PF, parafascicular nucleus; PO, posterior thalamic nucleus. See also Supplementary Table 1 PO orchestrates delayed post-stroke depression and selective anxiety comorbidity. (A) Schematics of thalamic hemorrhagic stroke induced by collagenase injection and evaluation of motor function and behavioral tests. Collagenase-induced thalamic hemorrhage (day 0), motor assessment (day 6), and behavioral phenotyping (days 7–34). (B) Quantification of motor deficit score of mice at day 6 post-stroke, confirming equivalent and normal mobility across hemorrhage models and sham controls. (C-G) Quantification of immobile duration in the FST in mice with acute (day 7, C) or chronic (day 25, D) extensive thalamic hemorrhage; direct comparison of acute vs. chronic effects (E). Quantification of immobile duration in the FST at 25th day after PO (F) and PF (G) hemorrhage. (H-J) Quantification of total distance and time in the center zone in OFT in mice with extensive thalamic hemorrhage (H), PO (I) and PF (J) hemorrhage. (K-M) Quantification of time in the open arm in EPMT in mice with extensive thalamic hemorrhage (K), PO (L) and PF (M) hemorrhage. (N-P) Quantification of time in light in LDBT in mice with extensive thalamic hemorrhage (N), PO (O) and PF (P) hemorrhage. (Q-S) Quantification of social interaction ratio in SIT in mice with extensive thalamic hemorrhage (U), PO (V) and PF (W) hemorrhage. (T-V) Quantification of percentages of sucrose preference in SPT in mice with extensive thalamic hemorrhage (T), PO (U) and PF (V) hemorrhage. (W-X) Quantification of latency to feeding in NSF in mice with extensive thalamic hemorrhage (W) and PO (X) hemorrhage. Data are mean ± SEM; n = 4–18 mice per group. unpaired t test for (C–D, F-L, S-T, W-X); Paired t test for (E); Mann-Whitney test for (H-I, M-R, U-V); ns., not significant; *P < 0.05, **P < 0.01, ***P < 0.001. PF, parafascicular nucleus; PO, posterior thalamic nucleus; FST, forced swim test; SPT, sucrose preference test; EPMT, elevated plus maze test; NSFT, novelty-suppressed feeding test; OFT, open field test; LDBT, light-dark box test; SIT, social interaction test Hemispheric lateralization of PO. (A) Schematics of von Frey and behavioral test timeline. (B-C) Mechanical pain thresholds (von Frey test) in mice with left or right PO hemorrhage during acute (7–9 days, B) and chronic (23 days, C) phases. (D) Motor function test confirming no post-stroke impairment. (E-K) Behavioral tests: quantification of time in the center zone in OFT (E), open arm in EPMT (F), light side in LDBT (G); quantification of social interaction ratio in SIT (H); quantification of immobile duration in FST (I); quantification of percentages of sucrose preference in SPT (J); quantification of latency to feeding in NSFT (K). Data are mean ± SEM; n = 5–15 mice/group. Unpaired t test and Mann-Whitney test were used (Supplementary Table 1). *P < 0.05, **P < 0.01, ****P < 0.0001; n.s., not significant. PO, posterior thalamic nucleus; FST, forced swim test; SPT, sucrose preference test; EPMT, elevated plus maze test; NSFT, novelty-suppressed feeding test; OFT, open field test; LDBT, light-dark box test; SIT, social interaction test. In addition, an anterior thalamic hemorrhage model was established using stereotactic collagenase injection (AP − 0.85 mm from bregma, ML ± 0.75 mm, DV − 3.15 mm from the dura; 7 nL), serving as an anatomical control for region-specific effects. PO hemorrhage governs thalamic hemorrhage-induced PSP To monitor the PSP in mice, von Frey was used to examine the mechanical pain threshold of the contralateral (to stroke side) hindpaws in the acute (7–9 days) and chronic (23 days) stages after hemorrhagic stroke (Fig. 3A-F). Contralateral hindpaw withdrawal thresholds were significantly reduced in both acute (7–9 days post-stroke; Fig. 3A) and chronic (23 days; Fig. 3D) phases after extensive thalamic hemorrhage. Strikingly, focal hemorrhage confined to the PO recapitulated this phenotype with comparable severity (Fig. 3B and E), whereas lesions in the adjacent PF—anatomically analogous but functionally distinct—failed to evoke mechanical hyperalgesia (Fig. 3C and F). Notably, hemorrhage confined to the VPL, a canonical somatosensory relay, also showed no effect (Supplementary Fig. 8A), underscoring PO’s unique role in PSP pathogenesis. As a supplement, we also made a model of anterior thalamic hemorrhage and found that it did not lead to mechanical hyperalgesia (Supplementary Fig. 8B-E). Note that, we measured the temporal profiling of the threshold of PSP after hemorrhagic stroke. We found that stroke-driven mechanical hyperalgesia remained stable through acute (day 3 post-stroke) and chronic stages (day 23; Fig. 3G), indicating persistent nociceptive network dysregulation. PO hemorrhage drives delayed maladaptive behaviors after PSP Stroke is a crippling disease leading to movement disability, we therefore reserved sufficient recovery time for mice after hemorrhagic stroke. Motor-function test was performed for each mouse at the 6th day after stroke to confirm they have regained equivalent level of mobility as the sham group and had no motor impairment (Fig. 4A, B; Supplementary Fig. 9A, Supplementary Table 11) [59, 60]. We tested maladaptive behavioral phenotypes either on day 7 or after day 20, when the mice exhibited an early phase or a late phase of decreased pain threshold, respectively. We found an absence of acute-phase depression-like behaviors (revealed with a non-significant difference of FST immobility) (Fig. 4C) but a pronounced despair-like behavior (increased FST immobility) in mice with extensive thalamic hemorrhage by day 25 (Fig. 4D-E). These data suggest a delayed onset of PSD after PSP in different post hemorrhagic stroke stages. Consistent with subregion specific PSP phenotypes, focal hemorrhage confined to the PO recapitulated this delayed depression-like behavior with comparable severity (Fig. 4F), whereas lesions in the adjacent PF (Fig. 4G), VPL (Supplementary Fig. 9B) failed to evoke this phenotype, suggesting PO as the critical locus for both PSP and delayed PSD. Clinical evidence showed stroke frequently cause significant psychiatric sequelae including emotional and social impairments [61–63]. To comprehensively determine whether mice with thalamic hemorrhagic stroke recapitulate other clinical affective conditions, such as anxiety [62, 64], anhedonia and impaired social interaction [61], we performed a set of behavioral assays, including OFT, EPMT, LDBT, SIT, SPT and NSFT, from 20 days after stroke (Fig. 4A). Extensive thalamic hemorrhage and PO, PF hemorrhage spared most affective domains (no deficits in OFT locomotion, center time, light-dark preference, social interaction, or sucrose preference; Fig. 4H-J and N-V), except in the EPMT and NSFT PO hemorrhage induced reduced open-arm time (Fig. 4K-M) and longer latency to feed (Fig. 4W-X). VPL hemorrhage did not cause psychiatric sequelae (Supplementary Fig. 9). Characteristic lateralization of PO in post-stroke pain and affective comorbidity Although the two hemispheres of human brain look almost perfectly symmetrical, investigations on motor, sensory, emotional and cognitive systems show a deeply lateralized organization [65, 66]. To test whether hemispheric specialization exists in PSP and PSD after PO hemorrhagic stroke, we injected collagenase into left and right PO respectively to induce hemorrhagic stroke in left and right hemisphere (Fig. 5A). von Frey tests were performed in the acute (7–9 days after stroke) and chronic (23 days after stroke) stages (Fig. 5A). Unilateral collagenase-induced hemorrhage in left or right PO elicited contralateral mechanical hyperalgesia, with right PO lesions exacerbating acute-phase hyperalgesia compared to left PO counterparts (Fig. 5B and C). Motor-function tests confirmed mice had no motor deficits before behavioral tests across groups (Fig. 5D) [59, 60]. From the 20th day onward, OFT, EPMT, LDBT, SIT, FST, SPT and NSFT were performed in turn to examine the anxiety-like and depression-like behaviors (Fig. 5E-K). Chronic behavioral profiling revealed hemispheric lateralization of PO. Quantitative analysis demonstrated that mice with left PO hemorrhage had reduced sucrose preference in the SPT (Fig. 5J) compared with sham-treated mice, yet spared anxiety or social deficits in OFT, EPMT, LDBT, SIT, and NSFT or behavioral despair measured by FST (Fig. 5E -I, K). Right PO hemorrhage induced behavioral despair (prolonged FST immobility; Fig. 5I) and anxiety-like avoidance (reduced EPMT open-arm time, Fig. 5F; longer latency to feed, Fig. 5K), while preserving hedonic drive (SPT; Fig. 5J) and social interaction (Fig. 5H), and did not affect anxiety deficits in OFT and LDBT (Fig. 5E and G). These results highlight the hemispheric lateralization of PO. Although PO hemorrhage in both hemispheres could induce PSP and PSD, their affective phenotypes diverged. Left PO hemorrhage leads to behavioral despair measured by FST, and right PO hemorrhage leads to anhedonia-like phenotypes measured by SPT. Besides, right PO leads to anxiety-like phenotypes measured by the NSFT and EPMT respectively, positioning PO as a lateralized hub where lesion laterality dictates symptom dimensionality. Hemorrhagic stroke primes PO neurons for pain hyperexcitability amplified by comorbid depression To dissect the neural basis of thalamic hemorrhage-induced PSP and PSD, we next set out to test whether and how the PO neurons were activated during the acute and chronic stages after stroke. We thus performed fiber photometry recordings to directly visualize whether these PO neurons respond to PSP and PSD. We virally expressed the fluorescent Ca2+ sensor GCaMP6s in left and right PO of c57 mice. 3 weeks after virus injection, we injected collagenase into PO and implanted an optic fiber above PO to collect the emitted photons from neuronal soma (Fig. 6A-B). After 3 days of recovery from stroke, optical fiber photometer recordings were performed to investigate whether PO neurons are sensitized to subthreshold stimuli, using von Frey filament (gram = 0.07) to elicit prompt allodynia on the contralateral hindpaw (Fig. 6A). The populational calcium response of PO was recorded in the acute and chronic stages. Histological verifications of optic fibers and viral expression profiles in fiber photometry were showed in Supplementary Fig. 10. Hemorrhage-induced hyperactivity in PO. (A) Schematic of virus injection and secondary surgical (collagenase injection and optic fiber implantation) paradigm for fiber photometry. (B) Immunohistochemical (NeuN, blue) verifications of optic fiber location (black) and GCaMP6s (green) expression profile in left PO (right) and right PO (left). (C-K) Acute-phase (day 4) calcium dynamics (averaged Z score of calcium signals, quantification of Z score peaks, and heatmaps of Z score) in left PO (C-F) and right PO (G-K) neurons responding to mechanical pain elicited by von Frey filament (gram = 0.07). Black arrow indicates the onset of pain. (L-O) Chronic-phase (day 26) averaged Z score of calcium signals (L), quantification of Z score peaks (M), and heatmaps of Z score (N-O) in left PO neurons responding to mechanical pain elicited by von Frey filament (gram = 0.07). Black arrow indicates the onset of pain. (P) Quantification of Z score peaks in left PO neurons responding to mechanical pain elicited by von Frey filament (gram = 0.07) at the 4th and 26th day after stroke. (Q-T) Chronic-phase (day 26) averaged Z score of calcium signals (Q), quantification of Z score peaks (R), and heatmaps of Z score (S-T) in right PO neurons responding to mechanical pain elicited by von Frey filament (gram = 0.07). Black arrow indicates the onset of pain. (U) Quantification of Z score peaks in right PO neurons responding to mechanical pain elicited by von Frey filament (gram = 0.07) at the 4th and 26th day after stroke. Data are mean ± SEM; n = 5–6 mice/group. Unpaired t test and Mann-Whitney test were used (Supplementary Table 1). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; n.s., not significant. PO, posterior thalamic nucleus In the acute stage (the 4th day) after stroke, when stroke mice had PSP but no PSD, Ca2+ signals were increased following 0.07-g von Frey filament stimuli in both sham-treated and stroke group (Fig. 6C-K). Strikingly, right PO hemorrhage showed consistent increased Ca2+ signals and much bigger calcium peak amplitude compared with sham-treated mice (Fig. 6G-K), while left PO did not show this phenomenon (Fig. 6C-F), aligning with the result that right PO hemorrhage exacerbated hyperalgesia compared with left PO in the acute phase (lateralized mechanical hyperalgesia, Fig. 5B). By chronic stages (day 26), when PSD emerged, both left and right PO hemorrhage cohorts exhibited amplified calcium transients versus sham controls (Fig. 6L–U). Moreover, mice with both left and right PO hemorrhage exhibited an overwhelmingly increased Ca2 + activity on the 26th day when PSD exists compared with acute stage (Fig. 6P and U). These data suggest that hemorrhagic stroke activates PO neurons toward mechanical pain, and the excitability of PO neurons is further enhanced under depressive comorbidity. We then investigated whether hemorrhagic stroke affects PO’s response to other stressors. TS, Res and FS were performed in the acute stage (6 days after stroke, Fig. 7A-X). Hemorrhagic PO further demonstrated stressor-specific sensitization: FS (Fig. 7Q-X), but not TS (Fig. 7A-H) or Res (Fig. 7I-P), evoked right PO hyperactivity acutely, without affecting left PO. This aligned with the result that right PO hemorrhage exacerbated hyperalgesia compared with left PO (lateralized mechanical hyperalgesia, Fig. 5B), and the consistent increased Ca2+ signals and much bigger calcium peak amplitude in von Frey test in the acute phase (Fig. 6C-K). Collectively, these results confirmed the hemorrhagic stroke-induced hyperactivity toward pain in the PO. PO activity responding to stressors after hemorrhagic stroke. (A-H) Acute-phase (day 6) calcium dynamics (averaged Z score of calcium signals, quantification of Z score peaks, AUC, and heatmaps of Z score) in left PO (A-D) and right PO (E-H) neurons responding to mechanical pain elicited by TS. Black arrow indicates the onset of TS. (I-P) Acute-phase (day 6) calcium dynamics (averaged Z score of calcium signals, quantification of Z score peaks, AUC, and heatmaps of Z score) in left PO (I-L) and right PO (M-P) neurons responding to mechanical pain elicited by Res. Black arrow indicates the onset of Res. (Q-X) Acute-phase (day 6) calcium dynamics (averaged Z score of calcium signals, quantification of Z score peaks, AUC, and heatmaps of Z score) in left PO (Q-T) and right PO (U-X) neurons responding to mechanical pain elicited by Res. Black arrow indicates the onset of Res. Data are mean ± SEM; n = 5–6 mice/group. Unpaired t test and Mann-Whitney test were used (Supplementary Table 1). *P < 0.05, **P < 0.01, ****P < 0.0001; n.s., not significant. PO, posterior thalamic nucleus; TS = tail suspension; RS = restraint; FS = footshock; AUC, area under curve PO exhibits lateralized thalamocortical projections linking sensory and affective circuits The most distinctive of feature of the thalamus is its interconnection with the cerebral cortex [67]. To resolve the structural basis of PO’s hemispheric specialization, we next dissected the connectivity of PO-cortex circuits. AAV2/5-hSyn-EGFP and AAV2/2-hSyn-mCherry-WPRE-pA (anterograde tracing) were infused into left and right PO of c57 mice respectively (Fig. 8A-B, Supplementary Fig. 11). Three weeks later, we observed mCherry+ fibers in many regions, including bilateral anterodorsal thalamic nucleus (AD), and ipsilateral ventral pallidum (VP), primary visual cortex (V1), primary (S1) and secondary (S2) somatosensory cortex (Fig. 8C-D and G), which is well known to be relevant to tactile sensation and pain [14]. Neurons positive for green fluorescent protein (EGFP+) were observed in ipsilateral primary motor cortex (M1), S1 and S2, granular insular cortex (GI), and VP (Fig. 8E-F and H). Projections of the PO neurons. (A) Schematic of PO injection of AAV2/5-hSyn-EGFP (left PO) and AAV2/2-hSyn-mCherry-WPRE-pA (right PO) in c57 mice. (B) PO: representative image of mCherry (left) and EGFP (right) labeling neurons with PO injection of AAV2/2-hSyn-mCherry-WPRE-pA and AAV2/5-hSyn-EGFP. Scale bars, 500 μm. (C) AAV2/2-hSyn-mCherry-WPRE-pA-mediated identification of outputs of right PO neurons. (D) Representative images showing output neurons in the ipsilateral S2, S1ULp and VP, and bilateral AD. Scale bars, 500 μm. (E) AAV2/5-hSyn-EGFP-mediated identification of outputs of left PO neurons. (F) Representative images showing output neurons in the ipsilateral M1, S1J, S1DZO, S1ULp, GI, S2 and VP. Scale bars, 500 μm. (G-H) A series of coronal sections, ipsilateral to the site of injection, from a representative mouse showing the outputs of right (G) and left (H) PO neurons. Scale bar, 1 mm. PO, posterior thalamic nucleus; S2, secondary somatosensory cortex; S1ULp, primary somatosensory cortex, upper lip region; VP, ventral pallidum; AD, anterodorsal thalamic nucleus; V1, primary visual cortex; M1, primary motor cortex; S1J, primary somatosensory cortex, jaw region; S1DZO, primary somatosensory cortex, oral dysgranular zone; GI, granular insular cortex Notably, although both left and right PO projected predominantly to ipsilateral cortices, they exhibited striking laterality. Left PO innervated M1 and broad S1 subdomains (S1J, S1DZO, S1ULp; Fig. 8E-F). Right PO showed restricted S1ULp targeting and unique visual cortex (V1) connectivity (Fig. 8C-D). All projections converged on cortical layers III–V—key sites for corticothalamic integration (Fig. 8D and F). The segregated connectivity indicates anatomically basis for hemispheric lateralization of PO in pain and affective comorbidity and provides probable targets for neuromodulation such as transcranial magnetic stimulation (TMS), which could selectively alleviate left or right PO-driven PSP/PSD while sparing other networks. Discussion This study investigated the neural basis of PSP–PSD comorbidity following thalamic hemorrhagic stroke. In a clinically phenotyped cohort of patients with isolated thalamic intracerebral hemorrhage, multimodal neuroimaging showed that posterior thalamic lesions were preferentially associated with the co-occurrence of PSP and PSD. Posterior thalamic hemorrhage was also associated with altered thalamocortical functional connectivity, particularly within sensorimotor networks, suggesting that lesion topography within the thalamus may contribute to distinct neuropsychiatric outcomes after stroke. Together, these clinical findings support the possibility that the posterior thalamus represents a shared thalamocortical substrate for pain–depression comorbidity following hemorrhagic stroke. To further investigate subnuclear specificity, we established nucleus-targeted murine models of thalamic hemorrhage and demonstrated that focal injury confined to the PO was sufficient to induce persistent mechanical hypersensitivity and delayed depression-like behaviors. Moreover, unilateral PO lesions revealed hemispheric dissociation of affective phenotypes, with left PO hemorrhage preferentially linked to anhedonia-like deficits and right PO hemorrhage to behavioral despair. Fiber photometry analyses further suggested that hemorrhagic injury alters PO neuronal activity during nociceptive processing, with heightened responses becoming more pronounced during the chronic stage when affective abnormalities emerged. Together, these findings support the possibility that PO-associated thalamocortical circuit dysfunction contributes to both sensory and affective disturbances following thalamic hemorrhage. We further propose that the observed anhedonia-like behaviors may develop progressively over time as a consequence of persistent pain states rather than as an immediate affective response to hemorrhagic injury itself. The persistence of mechanical hypersensitivity during the chronic stage supports the presence of sustained nociceptive dysfunction that may contribute to emotional dysregulation. Although spontaneous pain was not assessed in the present study, future studies using approaches such as the Mouse Grimace Scale may help clarify the contribution of ongoing pain to the development of affective symptoms. The evolving focus on post-stroke rehabilitation now prioritizes multidimensional recovery, extending beyond motor or speech deficits to address the interplay of neuropsychiatric sequelae [10, 62, 68–72]. For thalamic hemorrhagic stroke survivors, the dual strike of concurrent PSP and PSD—compounded by their bidirectional exacerbation—represents a critical unmet need, since routine therapies offering weak help [5, 9, 26, 28, 73–76]. In the past, preclinical models relied on extensive thalamic lesions (affecting a group of brain regions) to study these comorbidities [8, 30, 77–79], which is unable to distinguish the specific deficiency of single nuclei—a gap that has hindered precise understanding of thalamic functions and targeted interventions. By employing stereotaxic collagenase injections to induce thalamic hemorrhagic lesions in different subregions, we establish a nuclei-specific hemorrhagic model that isolates PSP-PSD pathophysiology to a defined nucleus. This approach not only confirms PO’s dominance in thalamic pain-depression comorbidity but also represents typical clinical observation: the lateralization of affective symptoms and the temporal divergence of pain-depression trajectories. Crucially, our identification of PO→S1/S2 projections provides an anatomically cortical target for feasible treatment, e.g. TMS or acupuncture. These advances position PO-centric models as a practical tool for deconstructing circuit-specific mechanisms and testing pre-clinical neuromodulation therapies. The thalamus serves as a pivotal relay for nociceptive signaling to the cortex [80–83], yet the functional specialization of its individual subnuclei under pathological conditions remains incompletely understood. Clinical observations have previously indicated that thalamic stroke lesions produce heterogeneous deficits in nociceptive and affective processing depending on their anatomical location—whether lateral, medial, posterior, or anterior [37, 61]. In particular, lesions associated with thalamic PSP most commonly involve lateral and posterior thalamic nuclei [37]. Furthermore, patients with posterior thalamic damage frequently exhibit significant emotional disturbances compared to those with anterior lesions [61], underscoring the particular importance of the posterior thalamus in mood regulation. Our findings are consistent with these clinical reports and reinforce the critical role of the posterior thalamus in the integration of pain and affective processing. While the VPL has often been implicated in post-stroke sensory deficits [28, 37], our experimental results challenge its contribution to the comorbidity of pain and depression. In mice, focal hemorrhage restricted to the VPL failed to elicit either PSP or PSD phenotypes, whereas adjacent lesions of the PO robustly promoted both conditions. This discrepancy may arise from the frequent use of extensive thalamic hemorrhage models in earlier studies, in which bleeding encroaching upon the PO—given its anatomical proximity to VPL—could have led to the misattribution of nociceptive and affective symptoms to VPL [8, 30, 77–79]. Contrary to the classical view that stroke universally suppresses neuronal activity, our data demonstrate that hemorrhagic injury induces a state of pathological hyperexcitability within the thalamic lesion area. Fiber photometry recordings revealed that neurons in the posterior nucleus (PO) exhibited heightened responses to subthreshold mechanical stimuli as early as four days post-stroke, with a further significant increase in activity upon the emergence of post-stroke depression (PSD) by day 26. This hyperexcitability was not limited to nociceptive stimuli but also extended to other stressors, indicating a broad sensitization of PO networks. We propose that extravasated blood components—a hallmark of hemorrhagic stroke—generate a neuroinflammatory environment that primes PO neurons for hyperexcitability. This view is further supported by our clinical observation of a positive correlation between pain intensity, as measured by VAS scores, and functional connectivity strength between the posterior thalamus and sensorimotor cortex. Unlike ischemic stroke, which is primarily characterized by hypoxic necrosis due to vascular occlusion [84–88], hemorrhagic stroke results in the accumulation of hemoglobin-derived products, iron-mediated toxicity, persistent glial activation, and chronic hemosiderosis [89–95]. These processes are driven by a cascade of pro-inflammatory cytokines, including IL-1β and IL-6 [1, 96, 97], which contribute to sustained neuronal dysregulation. This excitatory phenotype represents a fundamental distinction from the thalamocortical hypoactivity typically observed in ischemic stroke models [84, 85]. Our findings thus redefine thalamic hemorrhagic stroke as an “irritative” lesion, pathologically enhancing—rather than diminishing—neural activity in surviving circuits. This study offers the subnucleus-resolution evidence of such a mechanism in thalamic networks. Furthermore, the temporal progression of neuronal hyperactivity, which peaks alongside PSD onset, supports a feedforward model whereby early nociceptive sensitization predisposes neural circuits to subsequent affective dysfunction. Another intriguing dimension of our findings concerns the robust functional laterality observed within the thalamus. Since Broca’s seminal work on speech lateralization, hemispheric specialization has been well-established in domains such as language and visual processing [98–102]. However, the mechanisms underlying hemispheric asymmetries in post-stroke pain and depression have remained largely unexplored. Our clinical data reveal that both anterior and posterior thalamic hemorrhages produce lateralized disruptions in whole-brain functional connectivity, with right-sided lesions exerting more extensive and pronounced effects on network integrity compared to left-sided lesions. This clinical observation is further substantiated by our experimental findings in mice, which challenge the conventional reliance on symmetric or right-centric models of thalamic stroke [8, 14, 30, 79, 103]. By employing precise unilateral PO lesions, we demonstrate that left- and right-sided hemorrhages drive dissociable affective phenotypes: left PO injury preferentially induces anhedonia and deficits in novelty-suppressed feeding, whereas right PO injury leads to behavioral despair and open-arm avoidance in the elevated plus maze. These behavioral distinctions are paralleled by asymmetries in functional neuronal activity, as evidenced by fiber photometry, which revealed earlier and more intense calcium hyperexcitability in the right PO following peripheral mechanical stimulation. Although both hemispheres ultimately engage in nociceptive-affective processing, viral tracing confirmed that thalamocortical projection patterns differ substantially between left and right PO, suggesting divergent circuit-level integration. The observed lateralization may arise from inherent differences in intra- and interhemispheric connectivity [104]. Notably, while bilateral PO neurons were activated by allodynia after hemorrhagic stroke, right PO demonstrated an earlier and more pronounced calcium hyperactivity relative to left-sided PO, mirroring its acute-phase hypersensitivity in von Frey assays. The delayed engagement of left PO neurons implies possible compensatory recruitment of alternative networks, a mechanism warranting further investigation. Several limitations should be acknowledged. First, the sample size of our clinical cohort is relatively small, reflecting the inherent difficulty of recruiting survivors with isolated anterior or posterior thalamic hemorrhage who could be successfully recalled for comprehensive post-discharge neuroimaging and neuropsychiatric assessment. To minimize anatomical and clinical heterogeneity, we deliberately restricted inclusion to patients with well-demarcated thalamic lesions and excluded those with confounding neurological or psychiatric comorbidities. While this approach limits statistical power and generalizability, it enabled a lesion-specific and mechanistically focused analysis, which was further complemented by causal testing in nucleus-targeted murine models. Second, another limitation of the clinical component is the imbalance in the time interval from stroke onset to the study visit between the anterior and posterior thalamic ICH groups, which could influence cross-sectional imaging measures and symptom reporting. Although symptom assessments were anchored to the post-stroke course (since onset), future longitudinal studies with matched follow-up intervals and repeated imaging will be important to more precisely define temporal dynamics and nucleus-specific contributions. Third, only male mice were included to reduce variability related to the estrous cycle; therefore, the potential influence of sex-specific mechanisms on post-stroke pain–depression comorbidity warrants further investigation in future studies. Forth, a further limitation is that the murine experiments did not include MRI/fMRI, which would have provided a direct cross-species imaging bridge. Although this was not feasible due to the absence of dedicated small-animal MRI infrastructure, we mitigated this limitation by combining histological lesion verification with complementary circuit-level assays (fiber photometry and viral tracing) to assess functional excitability and thalamocortical connectivity. Future studies integrating mouse MRI/fMRI will be important to directly compare imaging biomarkers across species. Finally, our mechanistic investigation in mice focused on anterograde tracing from the PO. Retrograde tracing and functional manipulations were not performed. This decision was made in consideration of the potential differences in structural and functional connectivity between murine models and the human brain. Instead, we prioritized the translational relevance of our findings by directly correlating clinical symptom severity with thalamocortical functional connectivity strength in our patient cohort, thereby providing a more direct link to human pathophysiology and patient experience. Clinical and translational implications PSP and PSD represent major unmet clinical challenges in hemorrhagic stroke survivors, particularly following thalamic injury, where treatment responses remain poor and symptom trajectories are difficult to predict. Current management is largely symptomatic and non-specific, partly because the neural substrates linking sensory and affective disturbances remain insufficiently defined. Our findings provide a circuit-level framework with direct clinical relevance in three aspects. First, we identify the posterior thalamus—specifically the PO nucleus—as a convergent anatomical substrate for PSP–PSD comorbidity. This suggests that lesion topography within the thalamus may serve as a prognostic marker for identifying patients at high risk for combined pain and affective complications after hemorrhagic stroke. Second, we demonstrate hemispheric specialization of PO involvement, with differential behavioral phenotypes linked to lesion laterality. This lateralization may help explain the heterogeneity of post-stroke affective symptoms and could inform symptom-specific clinical stratification. Third, by showing that hemorrhagic injury induces a persistent hyperexcitable state in PO neurons, we provide a mechanistic rationale for targeting thalamocortical circuits using neuromodulation strategies such as TMS or non-invasive cortical stimulation, guided by lesion location. From a translational perspective, the nucleus-targeted hemorrhage model developed here enables subnucleus-level investigation of PSP–PSD mechanisms, bridging clinical lesion mapping and circuit neuroscience. This approach supports the development of precision neuromodulation strategies based on lesion-defined circuit vulnerability rather than symptom-based trial-and-error treatment. However, several limitations temper immediate clinical translation. Future studies should pursue longitudinal clinical imaging to validate posterior thalamic involvement as a prognostic biomarker, integrate structural and functional connectivity measures for patient stratification, and test circuit-targeted neuromodulation guided by lesion topography. Experimental work should further examine whether selective modulation of PO or its cortical projections can reverse both nociceptive and affective phenotypes. Together, these efforts may enable mechanism-based, location-informed therapeutic strategies for post-stroke neuropsychiatric comorbidity. Conclusions In conclusion, our study identifies a specific brain area called the posterior thalamic nucleus as the key hub responsible for the debilitating combination of chronic pain and depression that often follows thalamic hemorrhage. We discovered that the left and right sides of this brain region control different emotional symptoms—loss of pleasure versus behavioral despair—and mapped their distinct brain connections. These findings provide a scientific basis for developing targeted treatments like precision brain stimulation, opening new possibilities for personalized therapy based on a patient’s specific symptom profile and lesion location. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Acknowledgements AcknowledgementsWe thank Yangyang Lu, Min Li, Zijian Lv for guidance on stereotaxic surgeries, Qifeng Xie, Liang Yang, Shulan Xie for guidance on fiber photometry and advice on AAVs; Yuxiao Chen for helping in photometry analysis; Jing Yang and Bingjie Tang for guidance on von Frey tests; Yiqin Liu for guidance on behavioral tests; Wenxi Qiu for help in NeuN immunostaining; Zhiwei Zheng for advice on tracing virus; Zhenzhong Xu for technical supports. We are grateful to the Core Facilities of Zhejiang University Institute of Neuroscience for technical assistance. AI-assisted language editing statementThe authors used ChatGPT (OpenAI) to assist with language polishing and grammar refinement of the manuscript. All scientific content, interpretation, and final manuscript preparation were reviewed and approved by the authors. Abbreviations intracerebral hemorrhage post-stroke pain post-stroke depression posterior thalamic nucleus ventral posterolateral nucleus parafascicular thalamic nucleus forced swim test sucrose preference test elevated plus maze test novelty-suppressed feeding test tail suspension restraint footshock open field test light-dark box test social interaction test secondary somatosensory cortex primary somatosensory cortex, upper lip region ventral pallidum anterodorsal thalamic nucleus primary visual cortex primary motor cortex primary somatosensory cortex, jaw region primary somatosensory cortex, oral dysgranular zone granular insular cortex transcranial magnetic stimulation Numeric Rating Scale Douleur Neuropathique 4 Questions, VAS, Visual Analogue Scale Patient Health Questionnaire-9 Hamilton Depression Scale Hamilton Anxiety Scale Mini-Mental State Examination susceptibility-weighted imaging diffusion tensor imaging Author contributions YJ and YL contributed equally to this work. YJ and YL conceptualized the study, conducted investigations, performed formal analysis, curated data, created visualizations, and wrote the original draft. YJ also acquired funding. YW conducted investigations, curated data, and performed formal analysis. CL provided resources, curated data, and performed validation. WG provided resources and contributed to methodology. YC acquired funding, provided supervision, and reviewed and edited the manuscript. YD acquired funding, provided supervision, administered the project, and reviewed and edited the manuscript. LT conceptualized the study, acquired funding, provided supervision, administered the project, provided resources, and reviewed and edited the manuscript. All authors have read and approved the final version of the manuscript. Funding This work was supported by grants from the “Pioneer” and “Leading Goose” R&D Program of Zhejiang Province (2024C03006 to L.T.), Zhejiang Natural Science Foundation (LMS25H090003 to L.T.), Scientific Research Fund of Zhejiang University (XY2025074 to Y.J.), National Natural Science Foundation of China (32400830 to Y.D.), Leading Innovation and Entrepreneurship Team of Zhejiang Province (2023R01005 to L.T.), the Leading Innovation and Entrepreneurship Team of Hangzhou City (TD2024003 to L.T.), and Yunnan Fundamental Research Projects (202601CJ070009 to Y. C.). Data availability The Study Protocol and Statistical Analysis Plan supporting this study are publicly available as additional files. Data requests will be considered from qualified researchers who provide a methodologically sound proposal, subject to approval by the institutional ethics committee and the execution of a data use agreement. Requests should be directed to the corresponding author (email: 2310040@zju.edu.cn). Declarations Ethics approval and consent to participate The human study was conducted in accordance with the Declaration of Helsinki and received ethical approval from the Institutional Review Board of the Second Affiliated Hospital of Zhejiang University (Approval No. 2018083). Written informed consent was obtained from all participants prior to their inclusion in the study. All animal experiments were performed in compliance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Laboratory Animal Welfare and Ethics Committee of Zhejiang University (Approval No. ZJU20240210). Consent for publication All participants in this study provided written informed consent for the publication of their de-identified data. The consent forms specifically included provisions for the use of their anonymized clinical and imaging data in scientific publications. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Yihui Cui, Email: cuiyihui@mail.kiz.ac.cn. Yiyan Dong, Email: dongyiyan@mail.kiz.ac.cn. Lusha Tong, Email: 2310040@zju.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Data Availability Statement The Study Protocol and Statistical Analysis Plan supporting this study are publicly available as additional files. Data requests will be considered from qualified researchers who provide a methodologically sound proposal, subject to approval by the institutional ethics committee and the execution of a data use agreement. Requests should be directed to the corresponding author (email: 2310040@zju.edu.cn).