Human microglial state dynamics in Alzheimer's disease progression.

Altered microglial states affect neuroinflammation, neurodegeneration, and disease but remain poorly understood. Here, we report 194,000 single-nucleus microglial transcriptomes and epigenomes across 443 human subjects and diverse Alzheimer's disease (AD) pathological phenotypes. We annotate 12 microglial transcriptional states, including AD-dysregulated homeostatic, inflammatory, and lipid-processing states. We identify 1,542 AD-differentially-expressed genes, including both microglia-state-specific and disease-stage-specific alterations. By integrating epigenomic, transcriptomic, and motif information, we infer upstream regulators of microglial cell states, gene-regulatory networks, enhancer-gene links, and transcription-factor-driven microglial state transitions. We demonstrate that ectopic expression of our predicted homeostatic-state activators induces homeostatic features in human iPSC-derived microglia-like cells, while inhibiting activators of inflammation can block inflammatory progression. Lastly, we pinpoint the expression of AD-risk genes in microglial states and differential expression of AD-risk genes and their regulators during AD progression. Overall, we provide insights underlying microglial states, including state-specific and AD-stage-specific microglial alterations at unprecedented resolution.

Epigenomic dissection of Alzheimer's disease pinpoints causal variants and reveals epigenome erosion.

Recent work has identified dozens of non-coding loci for Alzheimer's disease (AD) risk, but their mechanisms and AD transcriptional regulatory circuitry are poorly understood. Here, we profile epigenomic and transcriptomic landscapes of 850,000 nuclei from prefrontal cortexes of 92 individuals with and without AD to build a map of the brain regulome, including epigenomic profiles, transcriptional regulators, co-accessibility modules, and peak-to-gene links in a cell-type-specific manner. We develop methods for multimodal integration and detecting regulatory modules using peak-to-gene linking. We show AD risk loci are enriched in microglial enhancers and for specific TFs including SPI1, ELF2, and RUNX1. We detect 9,628 cell-type-specific ATAC-QTL loci, which we integrate alongside peak-to-gene links to prioritize AD variant regulatory circuits. We report differential accessibility of regulatory modules in late AD in glia and in early AD in neurons. Strikingly, late-stage AD brains show global epigenome dysregulation indicative of epigenome erosion and cell identity loss.

Regulatory genomic circuitry of human disease loci by integrative epigenomics.

Annotating the molecular basis of human disease remains an unsolved challenge, as 93% of disease loci are non-coding and gene-regulatory annotations are highly incomplete1-3. Here we present EpiMap, a compendium comprising 10,000 epigenomic maps across 800 samples, which we used to define chromatin states, high-resolution enhancers, enhancer modules, upstream regulators and downstream target genes. We used this resource to annotate 30,000 genetic loci that were associated with 540 traits4, predicting trait-relevant tissues, putative causal nucleotide variants in enriched tissue enhancers and candidate tissue-specific target genes for each. We partitioned multifactorial traits into tissue-specific contributing factors with distinct functional enrichments and disease comorbidity patterns, and revealed both single-factor monotropic and multifactor pleiotropic loci. Top-scoring loci frequently had multiple predicted driver variants, converging through multiple enhancers with a common target gene, multiple genes in common tissues, or multiple genes and multiple tissues, indicating extensive pleiotropy. Our results demonstrate the importance of dense, rich, high-resolution epigenomic annotations for the investigation of complex traits.

Sex-dependent placental methylation quantitative trait loci provide insight into the prenatal origins of childhood onset traits and conditions.

Molecular quantitative trait loci (QTLs) allow us to understand the biology captured in genome-wide association studies (GWASs). The placenta regulates fetal development and shows sex differences in DNA methylation. We therefore hypothesized that placental methylation QTL (mQTL) explain variation in genetic risk for childhood onset traits, and that effects differ by sex. We analyzed 411 term placentas from two studies and found 49,252 methylation (CpG) sites with mQTL and 2,489 CpG sites with sex-dependent mQTL. All mQTL were enriched in regions that typically affect gene expression in prenatal tissues. All mQTL were also enriched in GWAS results for growth- and immune-related traits, but male- and female-specific mQTL were more enriched than cross-sex mQTL. mQTL colocalized with trait loci at 777 CpG sites, with 216 (28%) specific to males or females. Overall, mQTL specific to male and female placenta capture otherwise overlooked variation in childhood traits.

Single-cell pair-wise relationships untangled by composite embedding model.

In multicellular organisms, cell identity and functions are primed and refined through interactions with other surrounding cells. Here, we propose a scalable machine learning method, termed SPRUCE, which is designed to systematically ascertain common cell-cell communication patterns embedded in single-cell RNA-seq data. We applied our approach to investigate tumor microenvironments consolidating multiple breast cancer datasets and found seven frequently observed interaction signatures and underlying gene-gene interaction networks. Our results implicate that a part of tumor heterogeneity, especially within the same subtype, is better understood by differential interaction patterns rather than the static expression of known marker genes.

Unraveling dynamically encoded latent transcriptomic patterns in pancreatic cancer cells by topic modeling.

Building a comprehensive topic model has become an important research tool in single-cell genomics. With a topic model, we can decompose and ascertain distinctive cell topics shared across multiple cells, and the gene programs implicated by each topic can later serve as a predictive model in translational studies. Here, we present a Bayesian topic model that can uncover short-term RNA velocity patterns from a plethora of spliced and unspliced single-cell RNA-sequencing (RNA-seq) counts. We showed that modeling both types of RNA counts can improve robustness in statistical estimation and can reveal new aspects of dynamic changes that can be missed in static analysis. We showcase that our modeling framework can be used to identify statistically significant dynamic gene programs in pancreatic cancer data. Our results discovered that seven dynamic gene programs (topics) are highly correlated with cancer prognosis and generally enrich immune cell types and pathways.

CoCoA-diff: counterfactual inference for single-cell gene expression analysis.

Finding a causal gene is a fundamental problem in genomic medicine. We present a causal inference framework, CoCoA-diff, that prioritizes disease genes by adjusting confounders without prior knowledge of control variables in single-cell RNA-seq data. We demonstrate that our method substantially improves statistical power in simulations and real-world data analysis of 70k brain cells collected for dissecting Alzheimer's disease. We identify 215 differentially regulated causal genes in various cell types, including highly relevant genes with a proper cell type context. Genes found in different types enrich distinctive pathways, implicating the importance of cell types in understanding multifaceted disease mechanisms.

Genetic drivers of m6A methylation in human brain, lung, heart and muscle.

The most prevalent post-transcriptional mRNA modification, N6-methyladenosine (m6A), plays diverse RNA-regulatory roles, but its genetic control in human tissues remains uncharted. Here we report 129 transcriptome-wide m6A profiles, covering 91 individuals and 4 tissues (brain, lung, muscle and heart) from GTEx/eGTEx. We integrate these with interindividual genetic and expression variation, revealing 8,843 tissue-specific and 469 tissue-shared m6A quantitative trait loci (QTLs), which are modestly enriched in, but mostly orthogonal to, expression QTLs. We integrate m6A QTLs with disease genetics, identifying 184 GWAS-colocalized m6A QTL, including brain m6A QTLs underlying neuroticism, depression, schizophrenia and anxiety; lung m6A QTLs underlying expiratory flow and asthma; and muscle/heart m6A QTLs underlying coronary artery disease. Last, we predict novel m6A regulators that show preferential binding in m6A QTLs, protein interactions with known m6A regulators and expression correlation with the m6A levels of their targets. Our results provide important insights and resources for understanding both cis and trans regulation of epitranscriptomic modifications, their interindividual variation and their roles in human disease.