Robust and accurate estimation of cellular fraction from tissue omics data via ensemble deconvolution

Proteomic networks of gray and white matter reveal tissue-specific changes in human tauopathy

OBJECTIVE

To define tauopathy-associated changes in the human gray and white matter proteome.

METHOD

We applied tandem mass tagged labeling and mass spectrometry, consensus, and ratio weighted gene correlation network analysis (WGCNA) to gray and white matter sampled from postmortem human dorsolateral prefrontal cortex. The sampled tissues included control as well as Alzheimer's disease, corticobasal degeneration, progressive supranuclear palsy, frontotemporal degeneration with tau pathology, and chronic traumatic encephalopathy.

RESULTS

Only eight proteins were unique to gray matter while six were unique to white matter. Comparison of the gray and white matter proteome revealed an enrichment of microglial proteins in the white matter. Consensus WGCNA sorted over 6700 protein isoforms into 46 consensus modules across the gray and white matter proteomic networks. Consensus network modules demonstrated unique and shared disease-associated microglial and endothelial protein changes. Ratio WGCNA sorted over 6500 protein ratios (white:gray) into 33 modules. Modules associated with mitochondrial proteins and processes demonstrated higher white:gray ratios in diseased tissues relative to control, driven by mitochondrial protein downregulation in gray and upregulation in white.

INTERPRETATION

The dataset is a valuable resource for understanding proteomic changes in human tauopathy gray and white matter. The identification of unique and shared disease-associated changes across gray and white matter emphasizes the utility of examining both tissue types. Future studies of microglial, endothelial, and mitochondrial changes in white matter may provide novel insights into tauopathy-associated changes in human brain.

Proteomic changes in the human cerebrovasculature in Alzheimer's disease and related tauopathies linked to peripheral biomarkers in plasma and cerebrospinal fluid

INTRODUCTION

Cerebrovascular dysfunction is a pathological hallmark of Alzheimer's disease (AD). Nevertheless, detecting cerebrovascular changes within bulk tissues has limited our ability to characterize proteomic alterations from less abundant cell types.

METHODS

We conducted quantitative proteomics on bulk brain tissues and isolated cerebrovasculature from the same individuals, encompassing control (N = 28), progressive supranuclear palsy (PSP) (N = 18), and AD (N = 21) cases.

RESULTS

Protein co-expression network analysis identified unique cerebrovascular modules significantly correlated with amyloid plaques, cerebrovascular amyloid angiopathy (CAA), and/or tau pathology. The protein products within AD genetic risk loci were concentrated within cerebrovascular modules. The overlap between differentially abundant proteins in AD cerebrospinal fluid (CSF) and plasma with cerebrovascular network highlighted a significant increase of matrisome proteins, SMOC1 and SMOC2, in CSF, plasma, and brain.

DISCUSSION

These findings enhance our understanding of cerebrovascular deficits in AD, shedding light on potential biomarkers associated with CAA and vascular dysfunction in neurodegenerative diseases.

Native-state proteomics of Parvalbumin interneurons identifies unique molecular signatures and vulnerabilities to early Alzheimer's pathology

Dysfunction in fast-spiking parvalbumin interneurons (PV-INs) may represent an early pathophysiological perturbation in Alzheimer's Disease (AD). Defining early proteomic alterations in PV-INs can provide key biological and translationally-relevant insights. We used cell-type-specific in-vivo biotinylation of proteins (CIBOP) coupled with mass spectrometry to obtain native-state PV-IN proteomes. PV-IN proteomic signatures include high metabolic and translational activity, with over-representation of AD-risk and cognitive resilience-related proteins. In bulk proteomes, PV-IN proteins were associated with cognitive decline in humans, and with progressive neuropathology in humans and the 5xFAD mouse model of Aβ pathology. PV-IN CIBOP in early stages of Aβ pathology revealed signatures of increased mitochondria and metabolism, synaptic and cytoskeletal disruption and decreased mTOR signaling, not apparent in whole-brain proteomes. Furthermore, we demonstrated pre-synaptic defects in PV-to-excitatory neurotransmission, validating our proteomic findings. Overall, in this study we present native-state proteomes of PV-INs, revealing molecular insights into their unique roles in cognitive resiliency and AD pathogenesis.

Challenges and perspectives in computational deconvolution of genomics data

Deciphering cell-type heterogeneity is crucial for systematically understanding tissue homeostasis and its dysregulation in diseases. Computational deconvolution is an efficient approach for estimating cell-type abundances from a variety of omics data. Despite substantial methodological progress in computational deconvolution in recent years, challenges are still outstanding. Here we enlist four important challenges related to computational deconvolution: the quality of the reference data, generation of ground truth data, limitations of computational methodologies, and benchmarking design and implementation. Finally, we make recommendations on reference data generation, new directions of computational methodologies, and strategies to promote rigorous benchmarking.

Proteomic Changes in the Human Cerebrovasculature in Alzheimer's Disease and Related Tauopathies Linked to Peripheral Biomarkers in Plasma and Cerebrospinal Fluid

Dysfunction of the neurovascular unit stands as a significant pathological hallmark of Alzheimer's disease (AD) and age-related neurodegenerative diseases. Nevertheless, detecting vascular changes in the brain within bulk tissues has proven challenging, limiting our ability to characterize proteomic alterations from less abundant cell types. To address this challenge, we conducted quantitative proteomic analyses on both bulk brain tissues and cerebrovascular-enriched fractions from the same individuals, encompassing cognitively unimpaired control, progressive supranuclear palsy (PSP), and AD cases. Protein co-expression network analysis identified modules unique to the cerebrovascular fractions, specifically enriched with pericytes, endothelial cells, and smooth muscle cells. Many of these modules also exhibited significant correlations with amyloid plaques, cerebral amyloid angiopathy (CAA), and/or tau pathology in the brain. Notably, the protein products within AD genetic risk loci were found concentrated within modules unique to the vascular fractions, consistent with a role of cerebrovascular deficits in the etiology of AD. To prioritize peripheral AD biomarkers associated with vascular dysfunction, we assessed the overlap between differentially abundant proteins in AD cerebrospinal fluid (CSF) and plasma with a vascular-enriched network modules in the brain. This analysis highlighted matrisome proteins, SMOC1 and SMOC2, as being increased in CSF, plasma, and brain. Immunohistochemical analysis revealed SMOC1 deposition in both parenchymal plaques and CAA in the AD brain, whereas SMOC2 was predominantly localized to CAA. Collectively, these findings significantly enhance our understanding of the involvement of cerebrovascular abnormalities in AD, shedding light on potential biomarkers and molecular pathways associated with CAA and vascular dysfunction in neurodegenerative diseases.

scMD facilitates cell type deconvolution using single-cell DNA methylation references

The proliferation of single-cell RNA-sequencing data has led to the widespread use of cellular deconvolution, aiding the extraction of cell-type-specific information from extensive bulk data. However, those advances have been mostly limited to transcriptomic data. With recent developments in single-cell DNA methylation (scDNAm), there are emerging opportunities for deconvolving bulk DNAm data, particularly for solid tissues like brain that lack cell-type references. Due to technical limitations, current scDNAm sequences represent a small proportion of the whole genome for each single cell, and those detected regions differ across cells. This makes scDNAm data ultra-high dimensional and ultra-sparse. To deal with these challenges, we introduce scMD (single cell Methylation Deconvolution), a cellular deconvolution framework to reliably estimate cell type fractions from tissue-level DNAm data. To analyze large-scale complex scDNAm data, scMD employs a statistical approach to aggregate scDNAm data at the cell cluster level, identify cell-type marker DNAm sites, and create precise cell-type signature matrixes that surpass state-of-the-art sorted-cell or RNA-derived references. Through thorough benchmarking in several datasets, we demonstrate scMD's superior performance in estimating cellular fractions from bulk DNAm data. With scMD-estimated cellular fractions, we identify cell type fractions and cell type-specific differentially methylated cytosines associated with Alzheimer's disease.

scMD: cell type deconvolution using single-cell DNA methylation references

The proliferation of single-cell RNA sequencing data has led to the widespread use of cellular deconvolution, aiding the extraction of cell type-specific information from extensive bulk data. However, those advances have been mostly limited to transcriptomic data. With recent development in single-cell DNA methylation (scDNAm), new avenues have been opened for deconvolving bulk DNAm data, particularly for solid tissues like the brain that lack cell-type references. Due to technical limitations, current scDNAm sequences represent a small proportion of the whole genome for each single cell, and those detected regions differ across cells. This makes scDNAm data ultra-high dimensional and ultra-sparse. To deal with these challenges, we introduce scMD (single cell Methylation Deconvolution), a cellular deconvolution framework to reliably estimate cell type fractions from tissue-level DNAm data. To analyze large-scale complex scDNAm data, scMD employs a statistical approach to aggregate scDNAm data at the cell cluster level, identify cell-type marker DNAm sites, and create a precise cell-type signature matrix that surpasses state-of-the-art sorted-cell or RNA-derived references. Through thorough benchmarking in several datasets, we demonstrate scMD's superior performance in estimating cellular fractions from bulk DNAm data. With scMD-estimated cellular fractions, we identify cell type fractions and cell type-specific differentially methylated cytosines associated with Alzheimer's disease.

Native-state proteomics of Parvalbumin interneurons identifies novel molecular signatures and metabolic vulnerabilities to early Alzheimer's disease pathology

One of the earliest pathophysiological perturbations in Alzheimer's Disease (AD) may arise from dysfunction of fast-spiking parvalbumin (PV) interneurons (PV-INs). Defining early protein-level (proteomic) alterations in PV-INs can provide key biological and translationally relevant insights. Here, we use cell-type-specific in vivo biotinylation of proteins (CIBOP) coupled with mass spectrometry to obtain native-state proteomes of PV interneurons. PV-INs exhibited proteomic signatures of high metabolic, mitochondrial, and translational activity, with over-representation of causally linked AD genetic risk factors. Analyses of bulk brain proteomes indicated strong correlations between PV-IN proteins with cognitive decline in humans, and with progressive neuropathology in humans and mouse models of Aβ pathology. Furthermore, PV-IN-specific proteomes revealed unique signatures of increased mitochondrial and metabolic proteins, but decreased synaptic and mTOR signaling proteins in response to early Aβ pathology. PV-specific changes were not apparent in whole-brain proteomes. These findings showcase the first native state PV-IN proteomes in mammalian brain, revealing a molecular basis for their unique vulnerabilities in AD.

Robust differences in cortical cell type proportions across healthy human aging inferred through cross-dataset transcriptome analyses

Age-related declines in cognitive function are driven by cell type-specific changes in the brain. However, it remains challenging to study cellular differences associated with healthy aging as traditional approaches scale poorly to the sample sizes needed to capture aging and cellular heterogeneity. Here, we employed cellular deconvolution to estimate relative cell type proportions using frontal cortex bulk gene expression from individuals without psychiatric conditions or brain pathologies. Our analyses comprised 8 datasets and 6 cohorts (1142 subjects and 1429 samples) with ages of death spanning 15-90 years. We found aging associated with profound differences in cellular proportions, with the largest changes reflecting fewer somatostatin- and vasoactive intestinal peptide-expressing interneurons, more astrocytes and other non-neuronal cells, and a suggestive "U-shaped" quadratic relationship for microglia. Cell type associations with age were markedly robust across bulk-and single nucleus datasets. Altogether, we present a comprehensive account of proportional differences in cortical cell types associated with healthy aging.

Clustering ensemble in scRNA-seq data analysis: Methods, applications and challenges

With the rapid development of single-cell RNA-sequencing techniques, various computational methods and tools were proposed to analyze these high-throughput data, which led to an accelerated reveal of potential biological information. As one of the core steps of single-cell transcriptome data analysis, clustering plays a crucial role in identifying cell types and interpreting cellular heterogeneity. However, the results generated by different clustering methods showed distinguishing, and those unstable partitions can affect the accuracy of the analysis to a certain extent. To overcome this challenge and obtain more accurate results, currently clustering ensemble is frequently applied to cluster analysis of single-cell transcriptome datasets, and the results generated by all clustering ensembles are nearly more reliable than those from most of the single clustering partitions. In this review, we summarize applications and challenges of the clustering ensemble method in single-cell transcriptome data analysis, and provide constructive thoughts and references for researchers in this field.

Accurate estimation of rare cell type fractions from tissue omics data via hierarchical deconvolution

Bulk transcriptomics in tissue samples reflects the average expression levels across different cell types and is highly influenced by cellular fractions. As such, it is critical to estimate cellular fractions to both deconfound differential expression analyses and infer cell type-specific differential expression. Since experimentally counting cells is infeasible in most tissues and studies, in silico cellular deconvolution methods have been developed as an alternative. However, existing methods are designed for tissues consisting of clearly distinguishable cell types and have difficulties estimating highly correlated or rare cell types. To address this challenge, we propose Hierarchical Deconvolution (HiDecon) that uses single-cell RNA sequencing references and a hierarchical cell type tree, which models the similarities among cell types and cell differentiation relationships, to estimate cellular fractions in bulk data. By coordinating cell fractions across layers of the hierarchical tree, cellular fraction information is passed up and down the tree, which helps correct estimation biases by pooling information across related cell types. The flexible hierarchical tree structure also enables estimating rare cell fractions by splitting the tree to higher resolutions. Through simulations and real data applications with the ground truth of measured cellular fractions, we demonstrate that HiDecon significantly outperforms existing methods and accurately estimates cellular fractions.