Ming Hu

Although cerebellum is neuropathologically affected in Alzheimer's disease (AD) and AD-related dementia (ADRD), the cell type-specific genomic and epigenomic changes remain poorly understood. Here, we present single-nucleus multiome (RNA-seq and ATAC-seq) analysis of 103,861 nuclei isolated from both the cerebellum and frontal cortex of diseased individuals and cognitively healthy controls. We identify 431,834 significant peak-to-gene linkages across cell subtypes, revealing cis-regulatory elements linked to cell type-specific and disease-associated transcriptomic and regulatory network changes. Differential trajectory analysis pinpoints disease-associated genes in cerebellar granule cells, such as STMN1 and CLU. Integrating epigenomic, genetic, and three-dimensional genome data, we pinpoint two likely causal variants linked to SEZ6L2 in Purkinje cells and KANSL1 in granule cells. Through CRISPR interference in patient-derived cerebellar organoids, we show that silencing the variant-containing regulatory elements suppresses SEZ6L2 and KANSL1 expression, which perturbs neurodegeneration-associated gene programs. This cell type-specific regulatory landscape provides genomic and epigenomic insights into the pathological role of the cerebellum in AD/ADRD.

Hi-C technologies are widely used to study genome-wide chromosome spatial organization. Among various Hi-C data analyses, detecting long-range chromatin interactions (i.e., 3D peak calling) is particularly critical due to its direct relevance to gene regulation. However, most existing peak callers fail to account for spatial dependencies in high-resolution (e.g., ≤10 kb) Hi-C data, often resulting in suboptimal accuracy. To address this limitation, we introduce LAWS-HiC, a novel computational method based on the Locally Adaptive Weighting and Screening (LAWS) approach. LAWS-HiC adjusts p-values from a standard Hi-C peak caller by incorporating local spatial dependence within topologically associating domains (TADs) in a data-driven manner. Benchmarking in two deeply sequenced Hi-C datasets from human lymphoblastoid cell line GM12878 and mouse embryonic stem cells (mESCs), LAWS-HiC consistently improves the area under the precision–recall curve (PRAUC) across varying sequencing depths, with larger gains at lower sequencing depths. At the standard Benjamini–Hochberg false discovery rate (BH-FDR) ≤ 0.05, biological feature overlap analysis confirms that LAWS-HiC calls were more strongly enriched for regulatory annotations relative to non-calls than those from alternative methods. LAWS-HiC is freely available as an R package on GitHub.

Background/Objectives: The purpose of this study was to evaluate associations between quantitative ellipsoid zone (EZ) integrity metrics derived from spectral-domain OCT (SD-OCT) and visual outcomes, including clinically meaningful visual acuity (VA) loss, in dry age-related macular degeneration (AMD). Methods: This single-center retrospective image-analysis cohort included 351 eyes from 351 patients with intermediate AMD or atrophic changes. Macular SD-OCT scans were analyzed using a certified-reader-validated machine learning-based segmentation platform to delineate the EZ, retinal pigment epithelium (RPE), and Bruch’s membrane (BM). EZ metrics included EZ-RPE thickness (i.e., outer segment thickness) and EZ attenuation, defined as partial attenuation (EZ-RPE thickness of ≤20 µm, abnormal outer segment thinning)) or total attenuation/loss (0 µm). Drusen volume was quantified from the RPE-BM compartment. VA was converted to ETDRS letters. Associations between baseline OCT metrics and baseline VA, Year 2 VA, and ≥15-letter VA loss were assessed using nonparametric testing, linear regression, and ANCOVA. Results: Greater EZ-RPE thickness and lower EZ attenuation were associated with better baseline VA (all p < 0.001), and EZ metrics remained significant in cross-sectional regression models. Among 188 eyes with Year 2 follow-up, 38 (20%) experienced ≥15-letter VA loss. These eyes had significantly lower baseline EZ-RPE thickness (i.e., abnormal outer segment thinning) and greater EZ attenuation than eyes without such loss (all p < 0.001). In ANCOVA models, EZ metrics remained associated with Year 2 VA (all p < 0.001), whereas drusen volume was not. Conclusions: Quantitative EZ integrity metrics were associated with cross-sectional and longitudinal visual outcomes in dry AMD. EZ-RPE thickness and EZ attenuation may serve as clinically relevant OCT biomarkers reflecting photoreceptor integrity and risk of meaningful VA loss.

Background: Diffuse middle glioma (DMG or DIPG) is a fatal pediatric brain tumor. Although chimeric antigen receptor (CAR) T-cell therapy shows promise, clinical outcomes remain inconsistent due to premature exhaustion, underscoring a critical need to improve CAR-T persistence. A major barrier to CAR-T efficacy is antigen-independent tonic signaling, yet the extent to which tonic signaling shapes CAR-T durability and clinical outcomes, particularly in DMG, remains incompletely defined. Methods: Using a clinically investigated B7-H3 MGA271-based CAR as a reference platform, we generated alternative B7-H3 CARs incorporating either a human codon-optimized 376.96 (B7H3.BC) or Hu8H9 scFv antigen-binding domain to systematically assess scFv-dependent effects on tonic signaling and therapeutic efficacy. CAR-T cells were evaluated using integrated in vitro and in vivo functional assays, alongside multi-omics profiling and computational modeling. We further derived a tonic signaling-associated gene signature and evaluated its predictive performance across independent clinical datasets. Results: B7H3.BC CAR-T cells exhibit markedly restrained tonic signaling compared with MGA271- and Hu8H9-based counterparts, accompanied by superior antitumor activity and enhanced persistence across patient-derived DMG cells. Integrated multi-omics and single-cell profiling further identified a tonic signaling-associated gene signature that outperforms conventional T-cell exhaustion signatures in predicting therapeutic efficacy across multiple clinical trials, including DMG and other tumors. Conclusions: Our findings establish that scFv-dependent modulation of tonic signaling critically governs CAR-T persistence and antitumor efficacy in DMG. By linking CAR design to transcriptional and epigenetic programs, our study provides a principle-based and predictive framework to inform rational CAR engineering and improve therapeutic outcomes.

Although genetic variants are among the strongest risk factors for Alzheimer’s disease (AD), most associated variants reside in non-coding regions with unclear mechanisms. Here, we address that gap by fine-tuning the Akita DNA foundation model on high-resolution Hi-C data generated from neuronal progenitor cells (NPCs) and microglia to produce iNeuroAkita , a cell-context-aware three-dimensional (3D) genome model. iNeuroAkita shows improved correlation with experimental Hi-C contact maps and more accurate contact predictions at key AD loci (e.g. APOE and PSEN2) , and captures CTCF-dependent chromatin loops. Using in silico saturation mutagenesis (ISM) across 11,856 AD-associated single-nucleotide variants (SNVs) and 120 structural variants (SVs), we quantified impacts on chromatin spatial organization and identified two SNVs predicted to strongly disrupt 3D chromatin architecture: rs636317 at the MS4A locus that is predicted to disrupt local chromatic interactions and microglia eQTL rs7599054 that creates gained interactions connecting its surrounding region to CCNT2 and TMEM163 , consistent with enhancer–promoter activation in epigenomic and single-cell transcriptomic data. We also identified AD-associated deletions at the KANSL1 and ARMS2 that rewire local regulatory contacts and associate with cell-type-specific expression changes, demonstrating iNeuroAkita’s utility for systematic characterization of noncoding AD risk variants underlying 3D genome regulation.

Centromeres are essential chromosome components yet remain poorly understood due to their highly repetitive sequence architecture. Using fully-phased telomere-to-telomere diploid assemblies from a three-generation pedigree integrated with long-read epigenomes from matched peripheral blood mononuclear cells, induced pluripotent stem cells, and neural progenitor cells, we generate allele-resolved single basepair resolution maps of centromere genetic and epigenetic dynamics across inheritance, reprogramming, and differentiation. We show that centromeric dip regions (CDRs), which define the functional core of centromeres, are positionally stable across generations and cell-fate transitions. In contrast, CDR epigenetic architecture is highly dynamic. Reprogramming markedly attenuates CDR hypomethylation, which is partially restored during differentiation in parallel with global hypomethylation of active alpha-satellite arrays and coordinated changes in nucleosome organization and protein occupancy. Centromeric remodeling is insulated from X-chromosome status, including Xa, Xi, and erosion. Finally, de novo mutations arising during reprogramming are enriched in centromeric regions but depleted within functional centromeric cores.

The dynamic three-dimensional (3D) organization of the human genome (the 4D nucleome) is linked to genome function. Here we describe efforts by the 4D Nucleome Project ¹ to map and analyse the 4D nucleome in widely used H1 human embryonic stem cells and immortalized fibroblasts (HFFc6). We produced and integrated diverse genomic datasets of the 4D nucleome, each contributing unique observations, which enabled us to assemble extensive catalogues of more than 140,000 looping interactions per cell type, to generate detailed classifications and annotations of chromosomal domain types and their subnuclear positions, and to obtain single-cell 3D models of the nuclear environment of all genes including their long-range interactions with distal elements. Through extensive benchmarking, we describe the unique strengths of different genomic assays for studying the 4D nucleome, providing guidelines for future studies. Three-dimensional models of population-based and individual cell-to-cell variation in genome structure showed connections between chromosome folding, nuclear organization, chromatin looping, gene transcription and DNA replication. Finally, we demonstrate the use of computational methods to predict genome folding from DNA sequence, which will facilitate the discovery of potential effects of genetic variants, including variants associated with disease, on genome structure and function.

Background/Objectives: Hydroxychloroquine (HCQ) retinopathy can be underrecognized early, as structural changes in OCT may precede symptoms and are often subtle. Early detection is crucial to prevent irreversible damage. This study evaluated longitudinal OCT changes preceding overt HCQ toxicity using ellipsoid zone (EZ) mapping. Methods: Patients on long-term HCQ underwent two macular cube scans at least one year apart using Cirrus HD-OCT. Scans were analyzed with an EZ-mapping platform and manually validated. Patients with baseline OCT signs of toxicity or co-existing macular disease were excluded based on masked expert review. Results: Three hundred and seventy-three eyes of 373 patients were included. The mean age was 57.0 ± 12.6 years, the mean HCQ dose was 379.4 ± 59.4 mg, the treatment duration was 5.6 ± 3.7 years, and the OCT interval was 3.1 ± 0.9 years. Outer retinal metrics remained stable across the cohort. The mean en face EZ attenuation increased from 3.3% to 3.9% (p = 0.24). Thirty-four eyes (9.1%) experienced an absolute increase of ≥4% (~1.5 mm²) in EZ attenuation. This increase was significantly associated with age at HCQ initiation (p < 0.001), age at the time of the first and second OCT (p < 0.001), and baseline visual acuity (p = 0.01), and demonstrated changes in other outer retinal metrics (p < 0.01). Only 3/34 eyes (8.9%) were diagnosed by the managing clinician with HCQ toxicity at the time of the second OCT. However, 26 of these eyes (76.5%) had signs of HCQ toxicity by expert review, suggesting the overall greater sensitivity of these quantitative outer retinal metrics for detecting toxicity compared with clinician review. Conclusions: Longitudinal OCT assessment revealed overall stability in outer retinal metrics in eyes on HCQ, but a subset showed increased EZ attenuation, which correlated with age at the time of HCQ initiation, baseline visual acuity, and expert OCT review. These changes may help identify at-risk eyes and eyes with early toxicity and warrant further validation as potential screening biomarkers.

The biological basis of resilience to cognitive decline and Alzheimer’s disease (AD) remains unclear, hindering the development of effective interventions for brain health. Here, we present the largest human brain single-nucleus RNA-seq atlas to date (~14 million nuclei across 31 brain regions from 2,239 deeply phenotyped postmortem samples, including cognitively normal (CN), AD (clinico-pathologically confirmed), cognitively resilient (CR), and primary age-related tauopathy (PART)), using a scalable single-cell foundation model. This resource identifies >50 neuronal and glial cell types and reveals mitochondrial bioenergetics in somatostatin (SST) neurons linked to cognitive resilience despite amyloid-beta/tau pathology, which we validated immunohistochemically. Age-related downregulation of protective genes (i.e., PLCG2) and upregulated pro-inflammatory signals (TGF-β, IL18) in microglia drove AD progression, with female-specific markers (USP18, IFIT3) linked to microglial state transition (i.e., disease-associated microglia). Glutamate-mediated T cell-microglia crosstalk and proliferative microglia mediated complement signaling are associated with AD pathogenesis. Integrating deep clinical and genomic data under the single-cell foundation model, this comprehensive brain cell atlas deciphers brain-immune interactions underlying brain resilience and offers an open-access platform to accelerate therapies for neurodegeneration.