Dai Zhang’s research while affiliated with Chinese Academy of Medical Sciences & Peking Union Medical College and other places

Neurodynamic models that simulate how micro-level alterations propagate upward to impact macroscopic neural circuits and overall brain function may offer valuable insights into the pathological mechanisms of schizophrenia (SCZ). In this study, we integrated a neurodynamic model with the classical Contrastive Variational Autoencoder (CVAE) to extract and evaluate macro-scale SCZ-specific features, including subject-level, region-level parameters, and time-varying states. Firstly, we demonstrated the robust fitting of the model within our multi-site dataset. Subsequently, by employing representational similarity analysis and a deep learning classifier, we confirmed the specificity and disorder-related information capturing ability of SCZ-specific features. Moreover, analysis of the attractor characteristics of the neurodynamic system revealed significant differences in attractor space patterns between SCZ-specific states and shared states. Finally, we utilized Partial Least Squares (PLS) regression to examine the multivariate mapping relationship between SCZ-specific features and symptoms, identifying two sets of correlated modes implicating unique molecular mechanisms: one mode corresponding to negative and general symptoms, and another mode corresponding to positive symptoms. Our results provide valuable insights into disorder-specific neurodynamic features and states associated with SCZ, laying the foundation for understanding the intricate pathophysiology of this disorder.

Schizophrenia is a severe psychiatric disorder with high heritability, characterized by positive and negative symptoms as well as cognitive abnormalities. Dysfunction in glutamate synapse is strongly implicated in the pathophysiology of schizophrenia. However, the precise role of the perturbed glutamatergic system in contributing to the cognitive abnormalities of schizophrenia at the synaptic level remains largely unknown. Although our previous work found that Opcml promotes spine maturation and Opcml -deficient mice exhibit schizophrenia-related cognitive impairments, the synaptic mechanism remains unclear. By using whole-cell patch clamp recording, we found that decreased neuronal excitability and alterations in intrinsic membrane properties of CA1 PNs in Opcml -deficient mice. Furthermore, Opcml deficiency leads to impaired glutamatergic transmission in hippocampus, which is closely related to postsynaptic AMPA/NMDA receptors dysfunction, resulting in the disturbances of E/I balance. Additionally, we found that the aripiprazole which we used to ameliorate abnormal cognitive behaviors also rescued the impaired glutamatergic transmission in Opcml -deficient mice. These findings will help to understand the synaptic mechanism in schizophrenia pathogenesis, providing insights into schizophrenia therapeutics with glutamatergic disruption.

Background Observational studies have shown a controversial relationship between dietary fat intake and Alzheimer's disease, and the causal effects are unclear. Aims To assess the causal effects of total fat, saturated fat and polyunsaturated fat (PUF) intakes on the risk of Alzheimer's disease. Method A two-sample Mendelian randomisation analysis was performed using genome-wide association study summary statistics on different types of fat intake from UK Biobank ( n = 51 413) and on late-onset Alzheimer's disease (LOAD; 4282 cases, n = 307 112) and all forms of Alzheimer's disease (6281 cases, n = 309 154) from the FinnGen consortium. In addition, a multivariable Mendelian randomisation (MVMR) analysis was conducted to estimate the effects independent of carbohydrate and protein intakes. Results Genetically predicted per standard deviation increase in the total fat and saturated fat intakes were associated with 44 and 38% higher risks of LOAD (total fat: odds ratio = 1.44, 95% CI 1.03–2.02; saturated fat: odds ratio = 1.38, 95% CI 1.002–1.90; P = 0.049). The associations remained significant in the MVMR analysis (total fat: odds ratio = 3.31, 95% CI 1.74–6.29; saturated fat: odds ratio = 2.04, 95% CI 1.16–3.59). Total fat and saturated fat intakes were associated with a higher risk of all forms of Alzheimer's disease in the MVMR analysis (total fat: odds ratio = 2.09, 95% CI 1.22–3.57; saturated fat: odds ratio = 1.60, 95% CI 1.01–2.52). The PUF intake was not associated with LOAD or all forms of Alzheimer's disease. Conclusions This study indicated that total dietary fat intake, especially saturated fat, contributed to the risk of Alzheimer's disease, and the effects were independent of other nutrients. These findings informed prevention strategies and management for Alzheimer's disease directly towards reducing dietary saturated fat intake.

Background Air pollution, a reversible environmental factor, was significantly associated with the cognitive domains that are impaired in major depressive disorder (MDD), notably processing speed. Limited evidence explores the interactive effect of air pollution and the genetic risk of depression on cognition. This cross-sectional study aims to extend the research by specifically examining how this interaction influences depression-related cognitive impairment and resting-state brain function. Methods Eligible participants were 497 healthy adult volunteers (48.7% males, mean age 24.5) living in Beijing for at least 1 year and exposed to relatively high air pollution from the local community controlling for socioeconomic and genomic. Six months’ ambient air pollution exposures were assessed based on residential addresses using monthly averages of fine particulate matter with a diameter of less than or equal to 2.5 μm (PM2.5). A cross-sectional analysis was conducted using functional magnetic resonance imaging (fMRI) and cognitive performance assessments. The polygenic risk score (PRS) of MDD was used to estimate genetic susceptibility. Results Using a general linear model and partial least square regression, we observed a negative association between resting-state local connectivity in precuneus and PRS-by-PM2.5 interactive effect (PFWE = 0.028), indicating that PM2.5 exposure reduced the spontaneous activity in precuneus in individuals at high genetic risk for MDD. DNA methylation and gene expression of the SLC30A3 gene, responsible for maintaining zinc-glutamate homeostasis, was suggestively associated with this local connectivity. For the global functional connectivity, the polygenic risk for MDD augmented the neural impact of PM2.5 exposure, especially in the frontal-parietal and frontal-limbic regions of the default mode network (PFDR < 0.05). In those genetically predisposed to MDD, increased PM2.5 exposure positively correlated with resting-state functional connectivity between the left angular gyrus and left cuneus gyrus. This connectivity was negatively associated with processing speed. Conclusions Our cross-sectional study suggests that air pollution may be associated with an increased likelihood of cognitive impairment in individuals genetically predisposed to depression, potentially through alterations in the resting-state function of the occipitoparietal and default mode network.

Objective Emotional dysfunctions are prevalent across various psychiatric disorders, leading to diverse emotional problems. Disrupted emotional episodic memory is a prominent deficit and may underlie various affective symptoms in clinical phenotypes. However, diagnosis-specific of neurodiverse disruptions remain elusive. Methods We used task-based functional magnetic resonance imaging (fMRI) and a normative modelling framework to establish a reference for functional activation during emotional episodic memory, drawing from a large dataset of healthy individuals (n = 409). Individualized deviations from this reference were evaluated using a clinical dataset of 328 participants, which included 168 healthy controls and patients with major depressive disorder (MDD, n = 56), bipolar disorder (BD, n = 31), and schizophrenia (SZ, n = 73). Regional deviations were mapped to four large-scale emotional regulation networks and used to predict affective symptoms across different mental disorders. Results We constructed a verifiable normative model of functional activation during emotional episodic memory to parse clinical heterogeneity. Diagnosis-specific regional deviations were enriched in the non-overlapping large-scale emotional regulation networks: MDD showed enrichment in emotion regulation network related to emotion perception and generation, BD in cognitive appraisal and emotional reactivity, and SZ in working memory and response inhibition. Individualized deviations significantly predicted affective symptom in distinct disorder, and specific emotional regulation network showed maximum feature weight. Conclusions These findings have potential implications for the understanding of dissociable neuropathological patterns of affective symptoms and improving individualized clinical diagnosis and treatment in psychiatric disorders.

Autism Spectrum Disorders (ASDs) are reported as a group of neurodevelopmental disorders. The structural changes of brain regions including the hippocampus were widely reported in autistic patients and mouse models with dysfunction of ASD risk genes, but the underlying mechanisms are not fully understood. Here, we report that deletion of Trio , a high-susceptibility gene of ASDs, causes a postnatal dentate gyrus (DG) hypoplasia with a zigzagged suprapyramidal blade, and the Trio -deficient mice display autism-like behaviors. The impaired morphogenesis of DG is mainly caused by disturbing the postnatal distribution of postmitotic granule cells (GCs), which further results in a migration deficit of neural progenitors. Furthermore, we reveal that Trio plays different roles in various excitatory neural cells by spatial transcriptomic sequencing, especially the role of regulating the migration of postmitotic GCs. In summary, our findings provide evidence of cellular mechanisms that Trio is involved in postnatal DG morphogenesis.

To elucidate the brain‐wide information interactions that vary and contribute to individual differences in schizophrenia (SCZ), an information‐resolved method is employed to construct individual synergistic and redundant interaction matrices based on regional pairwise BOLD time‐series from 538 SCZ and 540 normal controls (NC). This analysis reveals a stable pattern of regionally‐specific synergy dysfunction in SCZ. Furthermore, a hierarchical Bayesian model is applied to deconstruct the patterns of whole‐brain synergy dysfunction into three latent factors that explain symptom heterogeneity in SCZ. Factor 1 exhibits a significant positive correlation with Positive and Negative Syndrome Scale (PANSS) positive scores, while factor 3 demonstrates significant negative correlations with PANSS negative and general scores. By integrating the neuroimaging data with normative gene expression information, this study identifies that each of these three factors corresponded to a subset of the SCZ risk gene set. Finally, by combining data from NeuroSynth and open molecular imaging sources, along with a spatially heterogeneous mean‐field model, this study delineates three SCZ synergy factors corresponding to distinct symptom profiles and implicating unique cognitive, neurodynamic, and neurobiological mechanisms.

Background Response to antipsychotic drugs (APD) varies greatly among individuals and is affected by genetic factors. This study aims to demonstrate genome-wide associations between copy number variants (CNVs) and response to APD in patients with schizophrenia. Methods A total of 3030 patients of Han Chinese ethnicity randomly received APD (aripiprazole, olanzapine, quetiapine, risperidone, ziprasidone, haloperidol and perphenazine) treatment for six weeks. This study is a secondary data analysis. Percentage change on the Positive and Negative Syndrome Scale (PANSS) reduction was used to assess APD efficacy, and more than 50% change was considered as APD response. Associations between CNV burden, gene set, CNV loci and CNV break-point and APD efficacy were analysed. Findings Higher CNV losses burden decreased the odds of 6-week APD response (OR = 0.66 [0.44, 0.98]). CNV losses in synaptic pathway involved in neurotransmitters were associated with 2-week PANSS reduction rate. CNV involved in sialylation (1p31.1 losses) and cellular metabolism (19q13.32 gains) associated with 6-week PANSS reduction rate at genome-wide significant level. Additional 36 CNVs associated with PANSS factors improvement. The OR of protective CNVs for 6-week APD response was 3.10 (95% CI: 1.33–7.19) and risk CNVs was 8.47 (95% CI: 1.92–37.43). CNV interacted with genetic risk score on APD efficacy (Beta = −1.53, SE = 0.66, P = 0.021). The area under curve to differ 6-week APD response attained 80.45% (95% CI: 78.07%–82.82%). Interpretation Copy number variants contributed to poor APD efficacy and synaptic pathway involved in neurotransmitter was highlighted. Funding 10.13039/501100001809National Natural Science Foundation of China, 10.13039/501100012166National Key R&D Program of China, 10.13039/501100002858China Postdoctoral Science Foundation.

Background and Hypothesis Schizophrenia (SZ) is a prevalent mental disorder that imposes significant health burdens. Diagnostic accuracy remains challenging due to clinical subjectivity. To address this issue, we explore magnetic resonance imaging (MRI) as a tool to enhance SZ diagnosis and provide objective references and biomarkers. Using deep learning with graph convolution, we represent MRI data as graphs, aligning with brain structure, and improving feature extraction, and classification. Integration of multiple modalities is expected to enhance classification. Study Design Our study enrolled 683 SZ patients and 606 healthy controls from 7 hospitals, collecting structural MRI and functional MRI data. Both data types were represented as graphs, processed by 2 graph attention networks, and fused for classification. Grad-CAM with graph convolution ensured interpretability, and partial least squares analyzed gene expression in brain regions. Study Results Our method excelled in the classification task, achieving 83.32% accuracy, 83.41% sensitivity, and 83.20% specificity in 10-fold cross-validation, surpassing traditional methods. And our multimodal approach outperformed unimodal methods. Grad-CAM identified potential brain biomarkers consistent with gene analysis and prior research. Conclusions Our study demonstrates the effectiveness of deep learning with graph attention networks, surpassing previous SZ diagnostic methods. Multimodal MRI’s superiority over unimodal MRI confirms our initial hypothesis. Identifying potential brain biomarkers alongside gene biomarkers holds promise for advancing objective SZ diagnosis and research in SZ.