Seminars in Cell and Developmental Biology xxx (xxxx) xxx Contents lists available at ScienceDirect Seminars in Cell and Developmental Biology journal homepage: www.elsevier.com/locate/semcdb Revealing cell vulnerability in Alzheimer’s disease by single-cell transcriptomics Carlos A. Saura a, b, *, Angel Deprada a, b, Maria Dolores Capilla-López a, b, Arnaldo Parra-Damas a, b a Institut de Neurociències, Department de Bioquímica i Biologia Molecular, Universitat Autònoma de Barcelona, Barcelona 08193, Spain b Centro de Investigación Biomédica en Red Enfermedades Neurodegenerativas (CIBERNED), Spain A R T I C L E I N F O A B S T R A C T Keywords: Alzheimer’s disease (AD) is a neurodegenerative disorder that by affecting specific brain cell types and regions Alzheimer’s disease cause severe pathological and functional changes in memory neural circuits. A comprehensive knowledge of the Brain pathogenic mechanisms underlying AD requires a deeper understanding of the cell-specific pathological re­ Dementia sponses through integrative molecular analyses. Recent application of high-throughput single-cell tran­ Inflammation Microglia scriptomics to postmortem tissue has proved powerful to unravel cell susceptibility and biological networks Neurodegeneration responding to amyloid and tau pathologies. Here, we review single-cell transcriptomic studies successfully Pathology applied to decipher cell-specific gene expression programs and pathways in the brain of AD patients. Tran­ RNA-Seq scriptional information reveals both specific and common gene signatures affecting the major cerebral cell types, Single-cell transcriptomics including astrocytes, endothelial cells, microglia, neurons, and oligodendrocytes. Cell type-specific tran­ Systems biology scriptomes associated with AD pathology and clinical symptoms are related to common biological networks affecting, among others pathways, synaptic function, inflammation, proteostasis, cell death, oxidative stress, and myelination. The general picture that emerges from systems-level single-cell transcriptomics is a spatiotemporal pattern of cell diversity and biological pathways, and novel cell subpopulations affected in AD brain. We argue that broader implementation of cell transcriptomics in larger AD human cohorts using standardized protocols is fundamental for reliable assessment of temporal and regional cell-type gene profiling. The possibility of applying this methodology for personalized medicine in clinics is still challenging but opens new roads for future diagnosis and treatment in dementia. 1. Introduction (i.e., hippocampus, entorhinal cortex and amygdala) and then spreads to limbic and association cortices [4]. Tau pathology correlates tightly Neurodegenerative diseases are incurable brain disorders that affect with cognitive decline and propagates spatially in parallel with micro­ millions of people, causing enormous medical, societal and economic glial activation, whereas Aβ potentiates the effects of microglial acti­ impacts. Alzheimer’s disease (AD) is the most common cause of de­ vation on tau spreading in the neocortex [5]. Despite major research mentia affecting currently 57 million cases worldwide [1]. These efforts to decipher the genetic and pathological causes of AD, our un­ neurological diseases are characterized by abnormal conformation, ag­ derstanding of the molecular mechanisms underlying this regional and gregation and accumulation of pathological proteins in particular cellular vulnerability is still poor, which has limited the development of degenerating neuronal and glial populations through still unclear effective therapies to ameliorate or reverse progression of dementia. mechanisms. The presence of amyloid plaques containing amyloid-β One of the major scientific challenges to understand neurodegener­ (Aβ) peptides, neurofibrillary tangles (NFTs) formed by ative diseases is the complexity of the human brain that contains a vast paired-helicament filaments (PHFs) of phosphorylated diversity of cells, including billions of neurons of distinct (sub)types. The microtubule-associated protein tau (MAPT), inflammatory responses, presence of neurons and non-neuronal cells displays even further di­ and synapse and neuron loss are characteristic of AD patients’ brain [2, versity of molecular, morphological and functional properties of brain 3]. Among these hallmarks, amyloid plaques are widely distributed, network connectivity. Precisely, understanding the complexity of whereas tau pathology appears first in the medial temporal lobe regions physiological changes occurring in the dysfunctional degenerative brain * Corresponding author at: Institut de Neurociències, Department de Bioquímica i Biologia Molecular, Universitat Autònoma de Barcelona, Barcelona 08193, Spain. E-mail address: carlos.saura@uab.cat (C.A. Saura). https://doi.org/10.1016/j.semcdb.2022.05.007 Received 25 February 2022; Received in revised form 12 May 2022; Accepted 12 May 2022 1084-9521/© 2022 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by- nc-nd/4.0/). Please cite this article as: Carlos A. Saura, Seminars in Cell and Developmental Biology, https://doi.org/10.1016/j.semcdb.2022.05.007 C.A. Saura et al. Seminars in Cell and Developmental Biology xxx (xxxx) xxx requires advanced systems-level analyses of biological, pathological, studies indicated temporal and regional vulnerability of synaptic and neuroimaging and clinical data integrated through cross-modal activity-dependent genes during AD pathology and clinical symptoms computational modeling, as recently applied to define a mechanistic ([20], for a review), but they did not clarify the contribution of indi­ framework of the mammalian brain [6]. Moreover, comprehensive an­ vidual cell transcriptomes to regional vulnerability in AD brain. alyses accounting for cellular heterogeneity, novel cell subtypes and More recently, bulk transcriptomics and co-expression network responses to pathological changes are required for a deep understanding analysis showed AD-associated neuronal and glial gene modules (e.g., of the molecular mechanisms occurring in brain diseases. In this sce­ cell survival/death, immune response) [21–25], and highlighted nario, bulk transcriptomics, single-cell transcriptomics and genomics oxidative phosphorylation as a cluster that predicts AD [26]. Gene coupled with gene network analysis represent ideal tools for investi­ co-expression analysis by combining publicly available AD tran­ gating cell-specific molecular mechanisms of disease (Fig. 1). It is scriptomic datasets has allowed the identification of cell-specific gene conceivable that these powerful technologies will unravel pathogenic signatures associated with pathological and cognitive status [23,27,28]. mechanisms underlying cell vulnerability while providing novel mo­ For instance, large scale gene expression and network analyses across lecular targets for drug discovery in neurodegenerative diseases. In this nineteen AD brain regions identified selective regional vulnerability and review, we focus on single-cell transcriptomic studies that have shifted abnormal gene modules in astrocytes, neurons and oligodendrocytes our current understanding of the cellular diversity and pathways [29]. Additionally, a transcriptome-wide association study (TWAS) affected in AD. We finally discuss current limitations and future chal­ combining genomic and transcriptomic data identified twenty-four lenges of applying cell transcriptomics to understand the biological hippocampal genes (APOE, DMPK, ERRC2, EXOC3L2, KAT8, PCDHA4, complexity of AD brain. PTPN9, QPCTL, SNRPD2, TOMM40, etc.) clustered in five functional modules related to AD cellular processes, including amyloid processing, 2. Transcriptome changes in Alzheimer’s disease brain tau phosphorylation, neuronal apoptosis, neurogenesis, chaperone-mediated autophagy, transcription and telomerase regula­ Brain transcriptomics is useful to decipher gene networks and tran­ tion [30]. Combining genome-wide association studies (GWAS), tran­ scriptional mechanisms affected during biological and pathological scriptome and proteome analyses has further revealed a set of genes aging. Pioneering microarray-based transcriptomic analyses uncovered related to translational repression by miRNAs in the lateral amygdala of multiple differentially expressed genes (DEGs) in vulnerable AD brain the Religious Order Study/Rush Memory and Aging Project (ROSMAP) regions affecting, among others, energy metabolism, transcriptional donors with anxiety traits phenotypes [31]. Bulk brain transcriptomics regulation, inflammation and synaptic function [7–10]. In CA1 hippo­ has been useful to decipher gene biological networks in cells of a ho­ campus, altered genes related to metabolic, synaptic, immune response mogenous material, but it does not allow the identification of vulnerable and myelination pathways, correlate with clinical and NFT scores, cell subpopulations and to elucidate transcriptional changes at single suggesting that these pathways are tightly associated with disease pro­ cell level, particularly in less abundant cell types. Alternatively, gression [10,11]. Particularly, genes related to metabolism, endocytosis, advanced molecular profiling of individual cells (i.e., cell tran­ and synaptic function are upregulated in mild cognitive impairment, scriptomics) using single-cell (sc; intact cells) and single-nucleus (sn; and then decline as the disease progresses [12–15], which is relevant only nuclei) RNA-seq in fresh and frozen postmortem tissue has allowed considering that deregulation of synaptic genes/proteins can lead to unprecedented molecular characterization of cell types and tissue re­ synapse dysfunction and cognitive decline in AD and mouse models [16, sponses to physiological and pathological changes (Fig. 1). 17]. Recently, bulk RNA sequencing (RNA-seq) and digital pathology using machine learning algorithms identified five genes (e.g., SER­ 3. Cell-specific vulnerability in Alzheimer’s disease PINA5, RYBP, SLC38A2, FEM1B, and PYDC1) associated with neuro­ pathological changes and hippocampal vulnerability in AD [18]. In the Single-cell transcriptomics has revealed substantial cellular hetero­ prefrontal cortex, elevated synaptic function and plasticity genes are geneity and vulnerability, and cell-type specific but also coordinated detected at early pre-symptomatic stages, and their expression decrease gene expression responses in AD brain. Subcluster analysis of individual coinciding with the appearance of pathological traits and clinical cell types in the mammalian brain using snRNA-seq shows subcellular impairment [19]. However, cognitive decline correlates tightly with heterogeneity, comprising multiple subpopulations of neurons, changes in synaptic transmission/plasticity and myelin genes in the including excitatory and inhibitory neurons, microglia, astrocytes, oli­ entorhinal cortex and hippocampus but not in neocortical regions [15]. godendrocytes, oligodendrocyte precursor cells (OPCs) and endothelial Therefore, despite multiple gene changes ongoing in the degenerating cells, affected in AD brain [32–36] (Fig. 2). These cellular sub­ brain, some gene modules are conserved among brain regions. These populations are not static or quiescent as changes in cell subtypes occur Fig. 1. Experimental design of cell transcriptomic analysis of human AD brain. Schematic illustration of the experimental transcriptomic workflow, including the design, sample processing and integrative bioinformatic analysis of bulk, single-cell (sc) and single-nucleus (sn) RNA sequencing (RNA-seq) of control and AD brains. Darker red shades in the cartoon brains indicate increase of AD neuropathology. GWAS: genome-wide association studies; TF: transcription factor. 2 C.A. Saura et al. Seminars in Cell and Developmental Biology xxx (xxxx) xxx Fig. 2. Changes in gene expression in vulnerable cell types in AD brain. Diagram showing the average number of differentially expressed genes (DEGs) per cell type, region, and AD pathological stage (Braak III-IV/V-VI) in human brain. The average of DEGs for each cell type and brain region and/or pathological stage is pro­ portional to the circle size. Affected cell types are depicted in different colors and letter codes as indicated at the bottom right of the image. EC: entorhinal cortex; PFC: Prefrontal cortex. in vulnerable AD brain regions during the disease process, including cortical neurons from AD subjects, which affect coding exons and are decreased neurons and increased number of microglia, astrocytes and predicted to induce further transcriptional alterations [39]. These re­ endothelial cells [23]. Accordingly, reduced excitatory neuron subtypes sults are important because imply that transcriptomic analyses are and increased microglia are detected first in the entorhinal cortex and useful to follow region-dependent cell-specific transcriptional vulnera­ later in the superior frontal gyrus [33,36]. In AD prefrontal cortex, bility in AD. microglial and astrocytic subpopulations increase and excitatory neu­ The regional vulnerability of cell subpopulations is accompanied by rons tend to decrease, whereas oligodendrocytes suffer little changes cell-type transcriptional changes affecting distinct but also overlapping during disease progression [22,23,34,37]. On the contrary, inhibitory biological networks (Table 1; Fig. 3). This implies that specific and neurons subpopulations are not globally affected in the entorhinal cor­ common biological pathways are regulated simultaneously in distinct tex, superior frontal gyrus and prefrontal cortex [36]. Specific cell-type cell types in the degenerating brain. Some overlapping gene clusters are subpopulations change depending on pathology, cognitive status and simultaneously increased in a variety of AD cell types, particularly those sex, which suggests a strong association of specific cell subpopulations associated with protein folding, autophagy, apoptosis, cell stress, im­ and AD pathological stage [32]. Surprisingly, when cell subtypes from mune response and mRNA regulation [32,33,35,40]. AD brain contains the same origin are analyzed as unique populations, the proportion of enrichment of cell death-related genes in astrocytes, oligodendrocytes, cell types is not apparently affected in AD prefrontal cortex, except for OPCs, and endothelial cells, and upregulation of glial differentiation, an increase of endothelial cells [35]. This discrepancy in cell type myelination, inflammatory responses and mitochondrial/cellular respi­ abundance could be explained, among others, by methodological dif­ ration/metabolic genes in neurons, astrocytes and/or oligodendrocytes ferences in: sample dissociation and preparation; ages of the cohorts [33,35,41]. Alternatively, synapse function and learning/cognition (mean age, [32]: 86.7 in AD and 87.1 in controls; [35]: 74.6 in AD and genes are downregulated in most cell types, including astrocytes, 85.4 in controls); and cell-type annotation, which is based exclusively on endothelial cells, microglia, neurons, oligodendrocytes and OPCs [33, cell markers and transcriptomes rather than morphological and func­ 35]. Other pathology-related biological pathways are shared mainly tional features. In conclusion, temporal changes in cell-type composition between two cell types in the prefrontal cortex, including those related involving selective regional vulnerability of excitatory neurons, and to synaptic signaling in neurons and astrocytes, myelination/axonal appearance of microgliosis and astrocytosis occur in AD brain. integrity/immune system in endothelial cells and oligodendrocytes, and The differential composition of cell subclusters and gene expression immune response genes in microglia and endothelial cells [32,35] profiles indicate disease-associated segregation of transcriptional (Fig. 3). Together, cell transcriptomics studies indicate that: 1) cell-type changes across cell types. The global trend of gene changes seems cell transcriptional changes affecting shared biological pathways occur specific as indicated by the fact that a vast majority of DEGs in the coordinately in response to AD pathology; and 2) cell-specific dis­ prefrontal cortex of the ROSMAP cohort are downregulated in excitatory ease-associated gene signatures implicate multiple cell types in AD and inhibitory neurons, and upregulated in astrocytes, microglia and pathophysiology. In conclusion, coordinated protective and compensa­ oligodendrocytes [32]. In cerebral vasculature of AD patients, 61–78% tory responses against protein aggregation and cell damage occur in of deregulated genes are downregulated in endothelial cells, fibroblasts distinct cell types of AD brain. and astrocytes [38]. Similarly, excitatory neurons but not inhibitory neurons show the most striking gene differences in AD entorhinal cortex, 4. Cell transcriptome changes in Alzheimer’s disease but the most coordinated gene expression changes occur in astrocytes, endothelial cells and microglia [33,36]. The fact that some gene changes Cell transcriptomics show that most DEGs affect a single cell type are more pronounced in entorhinal and temporal cortices compared whereas very few genes are simultaneously affected in several cell types with frontal cortex is consistent with the temporal pathological pro­ in AD brain. Despite this minimal overlap of DEGs among cell subtypes, gression from the medial temporal lobe to frontal cortex in AD. Inter­ numerous biological processes are affected across cell types (Table 1, estingly, a recent single-cell genomics study found age-dependent Fig. 3). accumulation of somatic mutations in hippocampal and prefrontal 3 C.A. Saura et al. Seminars in Cell and Developmental Biology xxx (xxxx) xxx Table 1 Biological pathways differentially affected in cell types of the AD human brain revealed by single-cell transcriptomics. Specific gene network modules upregulated or downregulated in human brain at distinct Braak pathological stages. The information is collected only from published studies using single nucleus/cell transcriptomics and shorted by alphabetical name. Biological pathways are named according to the specific description in the original studies, The original manuscripts from which the biological pathway information has been extracted are indicated with numbers using superscripts as follows: 1: [33]; 2: [35]; 3: [36]; 4: [32]; 5: [34]; 6: [40]; 7: [38]; 8: [37]. The number of individuals of each study are: [33]: 8 AD and 8 control 1; [35]:12 AD and 9 control2; [36]: 3 AD and 3 control3; [32]: 24 AD and 24 control4; [34]: 11 AD and 7 control5; [40]:6 AD and 6 control6; [38]: 9 AD and 8 control7; [37]: 11 AD and 11 control8. The data in reference [38] were obtained from both frontal cortex and hippocampal samples. The following abbreviations are used: EC: entorhinal cortex. PFC: prefrontal cortex. OPCs: Oligodendrocytes precursor cells. Cell type Region Biological pathway Upregulated Downregulated Braak III-IV Braak V-VI Braak III-IV Braak V-VI Astrocytes EC Chaperone mediated Actin filament based process1 Glutamate transporters6 Cognition1 protein assembly6 Cellular response to stress1 Glutamate-glutamine uptake and Inflammatory processes6 Glial cell differentiation and metabolism3 Mitochondrial oxidation myelination1 Neuronal system3 phosphorylation6 Negative regulation of cell death1 Reuptake of GABA3 Nrf2 activation pathways6 Regulation of immune response and Synapse organization and Response to metal ions6 response to cytokine1 transmission1,3 Response to topologically incorrect protein1 PFC Extracellular matrix Action potential5 Lipid and oxidative metabolism8 BMP signaling pathway5 molecules8 Anterograde trans-synaptic Glutamate neurotransmission2,4,5 signaling5 Negative regulation of glial cell Chaperone mediated protein differentiation5 folding2,4 Neurogenesis2 Presynaptic membrane assembly5 Post synapse organization and Protein localization to ER4 synaptic signaling2 Response to stress and mechanical Regulation of membrane potential4 stimulous2,4 Regulation of PI3K signaling5 Translation initiation4 Endothelial cells PFC Angiogenesis1,2 Cation transport1 Cell motility2 Cognition1 Cellular responses to stress1 Metal ion transport1 Cytokine secretion and immune response1 IFN α/β signaling7* IL-1 regulation of ECM7* IL-2 signaling7* Immune signaling7* Myeloid leukocyte activation2 Negative regulation of cell death1 Response to topologically incorrect protein1 Ribosomal processes1 Neurons (general) EC Cellular response to stress1 Synapse organization and cognition1 Glial cell development1 Regulation of cell death1 Regulation of cytokine production1 Response to topologically incorrect protein1 Excitatory PFC Glycosaminoglycan metabolism3 Amyloid fibril formation5 neurons Learning, memory, and cognition2,4 Axon development and guidance2,5 Myelination, axonal outgrowth, CREB phopphorylation3 and regeneration4 Ion homeostasis3 Negative regulation of inclusion Mitochondrial respiratory chain body assembly5 assembly4 Negative regulation of protein Organelle transport along phosphorylation4 microtubule5 Negative regulation of protein Pre- and pot- synaptic proteins2,3 ubiquitination5 Protein folding4 Neuron cell-cell adhesion5 Ribonucleoprotein complex Neuron death4 biogenesis4 Protein phosphorylation2 Wnt/calcium signaling pathway5 Regulation of IGF transport and uptake3 Regulation of protein catabolic process4 Synaptic transmission and signaling4,5 Inhibitory PFC Cellular protein complex Cell-cell adhesion2,4,5 neurons disassembly4 Membrane repolarization5 Cellular respiration4 Protein dephosphorylation5 ERBB2 signaling pathway5 Synaptic membrane adhesion, Learning, memory, and cognition2 organization and signaling2,4,5 Neuronal death4,5 (continued on next page) 4 C.A. Saura et al. Seminars in Cell and Developmental Biology xxx (xxxx) xxx Table 1 (continued ) Cell type Region Biological pathway Upregulated Downregulated Braak III-IV Braak V-VI Braak III-IV Braak V-VI Phosphorylation2 Positive regulation of glucose transport5 Regulation of mRNA processing5 Response to metal ion4 Microglia EC Carbohydrate metabolic Cellular response to stress2 Behavior and cognition2 processes6 Response to topologically incorrect Cell-cell adhesion2 IL-1 related pathway6 protein2 G-protein coupled receptor MAPK cascade6 Ribosomal processes and signaling2 Microglia pathogen translation initiation2 Homeostatic processes2 pahgocytosis6 Regulation of cytokine production2 Response to unfolded Response to lipid2 protein6 Selective autophagy6 PFC ERK1 and ERK2 cascade8 Cellular respiration related Cellular response to ionizing IFN-γ mediated signaling pathway5 NFκβ signaling pathway8 processes2 radiation8 Immune system process2 Regulation of glial cell Cytoplasmic translation4 Defense response to other Membrane organization2 migration8 Inflammatory response4 organism8 Negative regulation of immune Regulation of lipid Lipid metabolic process4 Metal-ion homeostasis8 effector process5 metabolic process8 Pathways associated with Negative regulation of protein Phospholipid transport5 Response to IL-68 β-amyloid clearance4 phosphorylation8 Regulation of GTPase activity4 Regulation of apoptotic process5 Response to IFN-γ8 Regulation of Phosphatidylinositol Vesicle targeting through ER to cis- 3-kinase signaling4 Golgi5 Response to cytokine2 T cell activation4 Oligodendrocytes EC Glial cell proliferation and Synapse organization and cognition1 myelination1 Mitochondrion1 Negative regulation of cell death1 Regulation of cytokine production1 Response to topologically incorrect protein1 PFC Lipid accumulation8 Amyloid-β metabolic process5 Differentiation of precursor cells Cellular response to catecholamine Osmotic imablance8 Axon guidance, outgrowth, and into mature myelin-forming cells8 stimulus4 Oxidative stress8 regeneration4,5 Myelination8 Dicarboxylic acid transport4 pH and electrolites Cell-cell adhesion via plasma Myelination2 control8 membrane5 Negative regulation of protein Cellular respiration processes2 polymerization5 Chaperone mediated protein Nervous system development5 folding5 Neuron projection morphogenesis5 Lipid biosynthetic process5 Neuronal and axonal ensheathment2 Protein stabilization5 Postsynaptic membrane assembly5 Regulation of cellular response to Regulation of cholesterol heat5 biosynthetic process5 Regulation of inclusion body Regulation of GTPase activity4 assembly5 Regulation of release of sequestered Regulation of synaptic assembly5 calcium5 OPCs EC Cellular response to stress1 Synapse organization cognition1 Negative regulation of cell death1 Response to topologically incorrect protein1 PFC Neuron development and neuron Granulocyte activation4 projection morphogenesis4,5 Myelination5 Ionotropic glutamate receptor Positive regulation of cell signaling pathway4 proliferation5 Regulation of membrane potential4 Protein folding4 Synapse organization4 4.1. Neuronal transcriptomes in Alzheimer’s disease be due to technical differences, divergence of ages between groups (control: 87 years vs AD: 71.5 years) and/or that specific neuronal Single-cell transcriptomic analyses demonstrate differential subtypes could be more susceptible to AD pathology. For instance, three neuronal susceptibility in response to AD pathology depending on the excitatory neuron subpopulations expressing the RAR-related Orphan pathological Braak stage, cognitive status and sex [32,33,36]. Excitatory Receptor B (RORB) are more vulnerable to tau pathology and cell death neuronal subpopulations are reduced but inhibitory neurons are un­ in entorhinal cortex but not in the superior frontal gyrus at early path­ changed in vulnerable AD brain regions, such as entorhinal and pre­ ological stages [36]. It seems that region-specific vulnerability is linked frontal cortices, and superior frontal gyrus [32,36]. These results agree with strong transcriptional changes in neurons in response to AD with recent evidence indicating major susceptibility of excitatory neu­ pathology. rons in AD pointing to the use of glutamatergic system modulators as a Neurons show, together with astrocytes, the most dramatic gene promising therapeutic strategy [42]. Other studies have failed to find signatures changes compared with other cell types (Table 1, Fig. 2). In differences in the total number of neurons [35]. Such discrepancy could AD prefrontal cortex, a high percentage of deregulated genes in 5 C.A. Saura et al. Seminars in Cell and Developmental Biology xxx (xxxx) xxx Fig. 3. Cell specific and common biological pathways and deregulated genes affected in AD. Alluvial plot depicting the link of each cell type with the most significant biological pathways affected in AD according to Table 1. For simplicity, altered biological networks reported in AD -irrespective of stage or region- were grouped into seven main terms. Cell types are depicted in different colors, and cell type-specific genes and those shared by several cell types are indicated in the right. The original studies from which the gene information was extracted are indicated in brackets. excitatory (75%) and inhibitory (95%) neurons are downregulated [32], 4.2. Astrocyte transcriptomes in Alzheimer’s disease whereas several gene modules from excitatory and inhibitory neurons correlate significantly with AD diagnosis [34]. AD excitatory and Cerebral inflammation, a common pathological feature of aging and inhibitory neurons exhibit differential expression of genes involved in neurodegenerative diseases, is characterized by the presence of in­ proteostasis, including protein folding, autophagy, apoptosis and stress flammatory molecules, reactive astrocytes and activated microglial cells response pathways [32–34] (Table 1, Fig. 3). Interestingly, genes [49]. Astrocytes maintain brain homeostasis and function by supporting involved in axon outgrowth, regeneration and myelination are the most metabolic and trophic factors to neurons. They also play active roles in deregulated genes in AD prefrontal cortex excitatory neurons, but not pathology and cognitive impairment in AD, for instance by engulfing inhibitory neurons [32]. This is surprising because although myelin dystrophic neurites and synapses, and causing hippocampal dysfunction ensheathes mainly excitatory axons in neocortical regions it also covers [50–52]. Astrocytes adopt a diversity of pathological and cellular inhibitory axons, specially of parvalbumin interneurons [43]. In addi­ changes depending on pathological and clinical AD stages [32,33]. tion, excitatory neurons show downregulated genes related to amyloid snRNA-seq analysis of AD prefrontal cortex shows astrocyte sub­ fibril formation, Wnt/calcium signaling pathway, and axon guidance populations with elevated levels of stress response (CRYAB, GFAP, and transport, and increased expression of synaptic trans­ LINGO1) and synaptic assembly/signaling genes, and downregulation of mission/memory, protein ubiquitination/phosphorylation, and cell glial differentiation and glutamate transport/metabolism genes death/adhesion [34]. The similar transcriptomes of affected excitatory (SLC1A2, GLUL) [32,35]. Similar analysis in astrocytes from AD ento­ neurons in distinct vulnerable brain regions suggest the possibility for rhinal and somatosensory cortices show enrichment of genes associated common regional mechanisms of excitatory vulnerability in response to with proteostasis (CRYAB, HSPB1, HSPH1, HSP90AA1), mitochondrial AD pathology. In inhibitory neurons, the ERBB2 signaling, glucose oxidation, inflammatory, glial cell differentiation and myelination transport, mRNA processing regulation and neuron death pathways are pathways, and decreased homeostatic genes related to glutamate upregulated, whereas genes involved in synapse organization and transport/metabolism and synapse remodeling (SLC1A3, SLC1A2, signaling, cell adhesion, protein dephosphorylation and membrane CSF1R) [33,36,40] (Table 1, Fig. 3). These astrocytic gene changes are repolarization are downregulated [32,34]. associated with AD pathological and clinical markers suggesting that In agreement with previous bulk transcriptomic analyses [12,13,15, astrocytes respond to pathological and network changes by modulating 19,44], expression of synaptic genes are deregulated in excitatory and their transcriptomes. It is currently unclear which specific astrocytic inhibitory neurons in AD prefrontal and entorhinal cortices [33,35,36] subtypes or transcriptomic responses contribute to neurodegeneration. (Table 1). Neuronal subpopulations enriched for axon, ion and vesicle It is intriguing that APOE, the major genetic risk factor associated transport, synapse and neurotransmitter signaling genes are decreased with late onset AD, is downregulated in AD astrocytes, oligodendrocytes in AD [33,35]. Particularly, excitatory and inhibitory neurons of AD and OPCs but upregulated in microglia [32,33]. A balance of APOE entorhinal cortex show decreased expression of genes regulating excit­ levels or isoforms seems critical for astrocyte function as suggested by atory synaptic transmission (SNAP25, RIMS1), and ion transport and recent studies showing that astrocytic APOE4 promotes neuronal acti­ memory/cognition (CCK, SST, RELN, VIP, KCNIP4), respectively [33]. vation, disrupts clearance of Aβ and cerebrovascular integrity, while Changes of synaptic genes are relevant considering the possible role of selective inactivation of APOE3 and APOE4 in astrocytes reduces plaque excitatory/inhibitory imbalance in the cognitive decline in AD [45]. deposition and microglial reactivity[53–56]. Moreover, differential in­ Indeed, excitatory/inhibitory disbalance is associated with altered crease of calcium signaling due to APOE4 vs APOE3 expression may lead excitatory and inhibitory postsynaptic proteins as well as degener­ to changes of activity-regulated transcriptional alterations in astrocytes ation/dysfunction of specific inhibitory neurons [46–48]. Despite the [57,58]. Other genes simultaneously increased in astrocytes and neu­ distinct vulnerability of neuron subtypes in AD brain, it remains unclear rons or oligodendrocytes include: the Bridging Integrator 1 (BIN1), the the specific neuronal subtypes and individual or sets of excitatory and second largest genetic risk factor for sporadic AD associated with tau inhibitory neuronal genes that contribute to excitatory/inhibitory and amyloid pathologies; LINGO1, a negative regulator of myelination; transmission imbalance, memory circuit dysfunction and neuro­ and NEAT1, a regulatory long non-coding RNA [33,34]. Interestingly, degeneration in AD. the AD risk genes CLU and IQCK, found in an astrocytic gene module enriched in metal ion homeostasis and proteostasis, are upregulated in reactive astrocytes and associate with tau and/or amyloid pathologies 6 C.A. Saura et al. Seminars in Cell and Developmental Biology xxx (xxxx) xxx [32,40]. These results indicate that astrocytes acquire a reactive related to ERK and NFκB signaling, homeostasis (CX3CR1, P2RY12, immunological phenotype characteristic of astrocytosis in AD brain. TMEM119) and DAM (APOE, CD68, CH13L1, SORL1, TREM2), whereas Accordingly, a subcluster of astrocytes expressing high levels of GFAP in the metal-ion homeostasis gene cluster is the top downregulated AD entorhinal cortex show reduced levels of transcripts associated with pathway in microglia [37] (Table 1). Among these genetic factors, APOE glutamate/GABA homeostasis and synaptic adhesion/maintenance, is consistently upregulated across studies in microglia of AD brain [32, suggesting that these are reactive astrocytes with compromised ho­ 33,40]. This result is relevant considering that immune responses and Aβ meostasis [36]. Whether astrocyte dysfunction is due homeostatic gene clearance are affected in human APOE4-expressing iPSCs-derived changes is unclear, but a recent bulk RNA-seq meta-analysis of microglia [55]. Future investigations should discern the functional astrocyte-specific gene sets identified altered astrocytic genes associated and pathological consequences of altered APOE isoforms in microglia with endolysosomes and mitochondrial dysfunction in AD [27]. Future activity, phagocytosis and dysfunction. studies should discern the relevance of these astrocytic subpopulations and gene changes in the pathogenesis and clinical progression of AD. 4.4. Oligodendrocyte transcriptomes in Alzheimer’s disease 4.3. Microglia transcriptomes in Alzheimer’s disease Oligodendrocytes are the unique source of axonal myelin that allows electrical transmission and metabolic coupling in the adult brain. Gross Microglia are immune cells of the central nervous system that dysregulation of genes profiles and hubs in oligodendrocyte sub­ respond to pathogens and degenerating cells and contribute directly to populations suggests a pathogenic involvement of oligodendrocytes and synapse and cell loss in AD [5,59]. Disrupted microglial transcriptional OPCs in this disease (Table 1, Fig. 2). In AD prefrontal cortex, oligo­ signatures are associated with Aβ/tau pathologies and phagocytosis in dendrocyte subtypes show downregulation of ribosomal subunits and AD brain [25,40], which suggests that microglia may adapt to cerebral myelination genes (MAG, CNP, PLP1), and upregulation of protein changes in aging and neurodegeneration through deep transcriptomic folding, translation and cell death genes (CADM2, CRYAB, QDPR, changes [60]. The increase of microglial cells in AD brain should be NLGN1) [32,34]. Transcriptional changes in specific oligodendrocyte considered when analyzing microglia transcriptomes [22,23,36]. Ac­ subpopulations are also observed in AD entorhinal cortex [33,36]. These cording to transcriptome profiles, and depending on the region and results reinforce the idea that oligodendrocytes and myelination-related physiological state, the human brain contains from 4 to 14 microglia pathways may be affected across vulnerable brain regions leading to subpopulations, some containing disease-associated microglia (DAM) age-dependent demyelination and memory loss in mild cognitive signatures [32,35,36,61]. Particularly, three microglia subpopulations, impairment, AD and vascular dementia [66,67]. one of them exhibiting reduced expression of homeostatic, synaptic Biological pathways affecting oligodendrocyte proliferation, differ­ pruning and cytokine genes, contribute significantly to microglial entiation and myelination are consistently detected across studies. OPCs transcriptomic changes in AD prefrontal cortex [35]. An independent show increased expression of glutamatergic receptor signaling, synaptic single-cell study of prefrontal cortex revealed genes changes in several and neural development genes, and downregulation of protein folding, microglial clusters enriched in genes related to Aβ and PHFs [61]. mitochondrial complex, cell proliferation and myelination genes [32, However, microglia cluster 7, the most enriched for DAM, MHC antigen 34] (Fig. 3). AD oligodendrocytes show enrichment of protein folding, presentation and endosomal/vacuolar genes, is the only cluster whose metabolic, oxidative phosphorylation/stress and cell death genes, and genes are downregulated in relation to pathological and clinical diag­ dysregulation of myelination genes, involving upregulation (BIN1, nosis of dementia [61]. CNTN2) or downregulation (CTNNA2, OPALIN, SLIT2) [33–35, 37] This raises the possibility that subpopulations of DAM participate in (Fig. 3). Interestingly, cell subcluster analysis reveals increased number inflammatory responses and synaptic pruning imbalance in AD. of oligodendrocytes expressing remyelinating genes and reduced mature Accordingly, microglia exhibit reduced homeostatic genes and age- oligodendrocytes enriched in myelination genes [33–35]. Whether these independent increase in inflammatory gene signatures in AD brain transcriptional changes result from oligodendrocyte dysfunction and/or [32, 33, 62–64] (Table 1, Fig. 3). In AD entorhinal cortex, microglia a shift from OPC differentiation to oligodendrocytes in response to pa­ show enrichment in autophagy, phagocytosis, inflammation and pro­ thology is still unclear. Nonetheless, increased oligodendrocytes sub­ teostasis pathways, and downregulation of homeostatic (CX3CR1, types and reduced OPCs at late pathological stages strongly supports a P2RY12, P2RY13), cell adhesion (CD86, CD83), lipid response (LPAR6) role of differentiation of OPCs into mature oligodendrocytes with and G-protein receptors (GPR183, LPAR6) genes [33,40]. Loss of basal increasing pathology [22]. In fact, a gene set enriched in oligodendro­ microglia functions, indicated by reduced homeostatic microglial genes, cyte differentiation and myelination pathways correlates positively with could represent a microglial response to AD pathology [62]. It is amyloid and NFT pathologies in prefrontal cortex [32], whereas oligo­ important to consider that technical differences, including age/­ dendrocyte gene markers (MAG, MBP, MOBP, CLDN11, CNP) are postmortem delay, sample preparation, sequencing approaches, cell reduced at Braak III-IV stages in AD precuneus, a parietal cortex region annotation, and downstream data analysis pipelines (see below), may vulnerable to early amyloid deposition [62]. Of relevance, a spatial directly affect the results, which can explain some discrepancies of gene transcriptomics study showed that amyloid plaques alter expression of profiles among these studies. For example, prominent artifactual gene oligodendrocytes genes (e.g., CRYAB, QDPR) resulting in gene changes signatures reported in microglia are caused by enzymatic dissociation of in inflammation, oxidative stress and lysosome pathways [68]. These human and mouse brains [65], whereas in some cases snRNA-seq results likely reflect oligodendrocyte compensatory responses to neuron methods fail to detect changes in microglial activation and DAM genes and/or myelin loss occurring in AD. Together, these transcriptomic [64]. analyses further support a role of oligodendrocytes on neuronal demy­ The identification of microglial-specific AD risk genes supports not elination in AD. Whether changes in oligodendrocyte maturation or only a causative role of microglia in AD but also provides a strong link myelination genes mediate the effects of Aβ and/or tau on axonal between genetic risk factors and microglial responses in AD [49]. Gene demyelination and degeneration requires further investigation. association studies have identified several genetic risk factors (e.g., APOE, TREM2, MEF2C, PICALM, HLA-DRB, HLA-DRB5) in microglial 4.5. Endothelial cells transcriptomes in Alzheimer’s disease gene modules of AD prefrontal cortex [32]. Expression of microglial genes linked with AD (APOE, MS4A6A, PILRA) or other neurodegener­ Endothelial cells, fibroblasts and pericytes, together with neurons ative diseases (LRRK2, SNCA, GPNMB, GRN) are positively correlated and glial cells, are part of the cerebrovasculature of the blood-brain with Aβ and pTau in AD entorhinal cortex [40]. Notably, microglia from barrier (BBB) that controls entry of pathogens and delivery of nutri­ AD patients harboring TREM2 variants show upregulation of genes ents and oxygen to the brain. Dynamic magnetic resonance imaging 7 C.A. Saura et al. Seminars in Cell and Developmental Biology xxx (xxxx) xxx shows that BBB damage and permeability in the hippocampus is exac­ contribute to apparent changes of transcriptome profiles, as suggested erbated in people with mild-cognitive impairment likely contributing to by a recent study showing that altered microglia and astrocyte gene early memory impairment during aging [69,70]. Single-cell tran­ modules arise from cell-type composition changes [23]. scriptomics revealed selective vascular cell vulnerability, comprising Single-cell transcriptomics has satisfactorily demonstrated multiple reduction of endothelial cells, fibroblasts, pericytes and arterial smooth transcriptomic and molecular changes occurring in distinct cell pop­ muscle cells, in hippocampal and frontal cortex vasculature of AD brain ulations of vulnerable anatomical regions of AD brain (Fig. 2). However, [38], which likely contributes to the neurovascular damage in dementia it is noteworthy that the number of overlapping DEGs in specific cell [71]. It is interesting that the abundance of specific endothelial cells subtypes is relatively low across studies, which may be caused by dif­ subpopulations increases in AD brain at early and late pathological ferences in the experimental and data analysis methods. Due to diffi­ stages [35,36]. Particularly, three endothelial cell subpopulations -e1, culties for isolating undamaged cells from fixed or frozen post-mortem e3, and e4- that vastly contribute to upregulation (89%) of endothelial brain, a step required for optimal gene profiling, most studies employ transcriptomic changes double in prefrontal cortex of AD brain [35]. The snRNA-seq that does not represent exactly the whole cellular tran­ biological relevance of the selective vulnerability of specific vascular scriptional state [77]. As recently reported in microglia [64], differential cell subpopulations is still unclear, but it may represent the cellular basis analysis by snRNA-seq or scRNA-seq can directly affect the transcrip­ or a response to brain vasculature damage during neurodegeneration. tional cell profiles. Other important constrains of snRNA-seq studies are The most representative gene networks affected in endothelial cells the limited number of sequenced cell types and analysis of nuclear are related to angiogenesis, cell motility/adhesion, stress cell death, transcripts that are not necessary a direct measure of translational inflammation, ion transport, cellular respiration and rRNA processing profiling. Other technical issues, particularly related to sample prepa­ (Table 1, Fig. 3). Endothelial cells in AD prefrontal cortex adopt ration, can severely impact on gene profiling [77]. Fortunately, arti­ angiogenic and immune response features characterized by upregula­ factual gene signatures in glial cells due to enzymatic dissociation can be tion of angiogenic factors and receptors (EGR, FLT1, VWF), and antigen overcome using alternative protocols [65]. Differences and limitations presentation (B2M, HLA-E) genes [35]. Similarly, endothelial cells in AD of the experimental designs and protocols, particularly regarding small entorhinal cortex overexpress genes related to cytokine secretion and sample sizes, sample preparation, restriction to specific brain areas and immune responses (HLA-E, MEF2C, NFKB1A), and ribosomal processes pathological stages, may lead to discrepancies in the absolute tran­ and translation initiation (RPS19, RPS28) [33]. Genes related to scriptional profiles (i.e., number of DEGs). Despite this, most studies immunological response (interleukin-1, interleukin-2 and IFN signaling) have revealed consistent cell-specific gene expression alterations, at are upregulated in endothelial cells in hippocampal and cortical least within similar brain regions. Intriguingly, single-cell analyses vasculature of AD patients [38]. Considering the established relation­ performed in entorhinal cortex, an early pathological affected region, ship between cerebral inflammation and angiogenesis in dementia [71], highlighted deregulation of synaptic-related genes in neurons and glial it is plausible that inflammatory factors may contribute to changes in cells [33,36], whereas changes in glial-related pathways with minor or activation and/or proliferation of endothelial cells in AD and other unreferenced disturbances in synaptic pathways are detected in the neurological disorders [72]. Indeed, activation of the innate immune prefrontal cortex [32,34,37], consistent with previous network-based system in endothelial and glial cells, and disruption of neurovascular microarray analysis [24]. This is surprising, given the roles that glial proteins, characterize Huntington’s disease brains [73]. More recently, cells play on regulating neuronal and synaptic activities in health and deregulated genes related to vasoconstriction and blood flow in mural disease, and it may suggest that synaptic dysfunction may not be cells and fibroblasts, and selective vulnerability of endothelial cells, properly captured at the transcriptional level at distal cortical areas pericytes and arterial smooth muscle cells were detected in AD cerebral affected during late pathological stages. vasculature [38]. These results provide potential molecular mechanisms Single-cell sequencing applied to postmortem human brain has for cerebrovascular damage and cerebral blood flow dysfunction in AD multiple applications but still important caveats, such as low number of [74]. Considering that cerebrovascular disruption is a common feature brain specimens, properly represented cell types and detected DEGs, and of neurodegenerative diseases, it may be relevant to integrate tran­ lack of spatial resolution. Indeed, a limitation of currently available scriptomic information from several neurodegenerative diseases to single-cell technologies, reflected in the reviewed literature, is the dif­ discern whether common biological pathways and genes in endothelial ficulty to account for human biological variability, which in the case of a cells are affected in distinct brain pathologies. complex age-dependent brain disease such as AD requires large cohort studies. The quality and validity of single-cell transcriptomics data 5. Conclusions and future challenges depend on considering biological variability, and methods that ignore this are biased and result in false discoveries [78]. Compared to general Single-cell transcriptomics has uncovered diverse cellular heteroge­ scRNA-Seq, another important limitation of snRNA-seq from postmor­ neity and dynamics of the aging and degenerating brains, particularly of tem tissue is that some cell subtypes are under- or over-represented due postmortem AD samples. The identification of disease-related gene to the dissociation process or artificial filtering, which negatively affects networks and pathways in multiple cell types has further unraveled gene profiling [77]. Optimization of cell isolation methods from post­ unique and shared biological responses to pathology in the degenerating mortem brain will undoubtedly increase the number and quality of brain, shifting the neurocentric view of brain diseases. Gene regulatory sequenced cell types. Standardization of cell clustering annotation network analyses of scRNA-seq data have allowed the identification of methods (computational, manually curated, machine learning) can potential -not validated yet- transcription factors mediating transcrip­ facilitate the classification of cell types for raw datasets comparison, tional disturbances in specific cell subtypes of AD brain [33,34,36,40]. which will reduce discrepancies in cell classification between studies For instance, the transcription factor RORB was identified in vulnerable [77]. Finally, a major weakness of most of the reported scRNA-seq excitatory neurons of AD preferentially affected by NFTs [36], but its studies is the lack of tissue spatial information within an anatomical role in neurodegeneration and pathology is still unknown. Although region. Incorporating spatial cell transcriptomics is critical for dissecting transcriptome dysregulation in AD brain differs from normal aging [22], the complexity of cellular interactions in the context of local patholog­ the fact that several cellular pathways are shared between aging and AD ical changes occurring in the degenerating brain [77]. Overall, this re­ suggests that some transcriptomic changes could be due to the aging veals the need for reliable studies using rigorous experimental designs process [75]. This may explain why gene expression profiles and sub­ and validation protocols directed at assessing brain regions affected at populations of microglia do not significantly differ in cortical regions in early AD stages when diagnosis and therapeutic interventions are criti­ non-demented elderly people and dementia patients [25,76]. It is also cally needed. possible that changes in the relative abundance of cell subtypes may Thus, considering the marked protocol differences among the 8 C.A. Saura et al. Seminars in Cell and Developmental Biology xxx (xxxx) xxx reviewed studies, which directly affect data analysis and interpretation, [13] S.D. Ginsberg, M.J. Alldred, S. Che, Gene expression levels assessed by CA1 pyramidal neuron and regional hippocampal dissections in Alzheimer’s disease, future studies should consider benchmarking, standardization and Neurobiol. Dis. 45 (1) (2012) 99–107. quality control of experimental and computational pipelines [79]. We [14] S.D. Ginsberg, M.J. Alldred, S.E. Counts, A.M. Cataldo, R.L. Neve, Y. Jiang, J. Wuu, think that integrating multiple single-cell omics modalities (genomics, M.V. Chao, E.J. Mufson, R.A. Nixon, S. Che, Microarray analysis of hippocampal transcriptomics and proteomics) in larger human cohort studies is crit­ CA1 neurons implicates early endosomal dysfunction during Alzheimer’s disease progression, Biol. Psychiatry 68 (10) (2010) 885–893. ical to reveal disease-associated alterations, especially at early AD [15] N.C. Berchtold, M.N. Sabbagh, T.G. Beach, R.C. Kim, D.H. Cribbs, C.W. Cotman, stages. Finally, future studies should include spatial gene profiling in Brain gene expression patterns differentiate mild cognitive impairment from early vulnerable brain regions at different pathological stages to identify normal aged and Alzheimer’s disease, Neurobiol. Aging 35 (9) (2014) 1961–1972. [16] A. Parra-Damas, J. Valero, C. Meng, J. España, E. Martin, I. Ferrer, J. Rodriguez- relevant cell-specific mechanisms of therapeutic and diagnostic value. It Alvarez, C.A. Saura, Crtc1 activates a transcriptional program deregulated at early is expected that a deeper understanding of the cell-specific mechanisms Alzheimer´s disease-related stages, J. Neurosci. 34 (17) (2014) 5776–5787. mediating neurodegeneration will facilitate the development of novel [17] K. Hatanpaa, K.R. Isaacs, T. Shirao, D.R. Brady, S.I. Rapoport, Loss of proteins regulating synaptic plasticity in normal aging of the human brain and in Alzheimer therapeutic strategies effective to combat this devastating disorder. disease, J. Neuropathol. Exp. Neurol. 58 (6) (1999) 637–643. [18] A.M. Crist, K.M. Hinkle, X. Wang, C.M. Moloney, B.J. Matchett, S.A. Labuzan, Funding sources I. Frankenhauser, N.O. Azu, A.M. Liesinger, E.R. Lesser, D.J. Serie, Z.S. Quicksall, T.A. Patel, T.P. Carnwath, M. DeTure, X. Tang, R.C. Petersen, R. Duara, N.R. Graff- Radford, M. Allen, M.M. Carrasquillo, H. Li, O.A. Ross, N. Ertekin-Taner, D. This work was supported by grants from Agencia Estatal de Inves­ W. Dickson, Y.W. Asmann, R.E. Carter, M.E. Murray, Transcriptomic analysis to tigación/Ministerio de Ciencia e Innovación of Spain with FEDER funds identify genes associated with selective hippocampal vulnerability in Alzheimer’s disease, Nat. Commun. 12 (1) (2021) 2311. (PID2019-106615RB-I00 and PDC2021-121350-I00 to CAS), Instituto [19] K. Bossers, K.T. Wirz, G.F. Meerhoff, A.H. Essing, J.W. van Dongen, P. Houba, C. de Salud Carlos III (CIBERNED CB06/05/0042) and Generalitat de G. Kruse, J. Verhaagen, D.F. Swaab, Concerted changes in transcripts in the Catalunya (2017 SGR749). APD is supported by a Juan de la Cierva- prefrontal cortex precede neuropathology in Alzheimer’s disease, Brain 133 (Pt 12) Incorporación postdoctoral fellowship (IJC2019-042468-I). AD and (2010) 3699–3723. [20] C.A. Saura, A. Parra-Damas, L. Enriquez-Barreto, Gene expression parallels MDCL are supported by predoctoral fellowships from Formación Pro­ synaptic excitability and plasticity changes in Alzheimer’s disease, Front. Cell fesorado Universitario (FPU 18/02486) and Formación Personal Inves­ Neurosci. 9 (2015) 318. tigador (BES-2017-082072) programs of Ministerio de Universidades [21] D.L. Brokaw, I.S. Piras, D. Mastroeni, D.J. Weisenberger, J. Nolz, E. Delvaux, G. E. Serrano, T.G. Beach, M.J. Huentelman, P.D. Coleman, Cell death and survival and Ministerio de Ciencia e Innovación, respectively. pathways in Alzheimer’s disease: an integrative hypothesis testing approach utilizing -omic data sets, Neurobiol. Aging 95 (2020) 15–25. Declaration of interests [22] S. Morabito, E. Miyoshi, N. Michael, V. Swarup, Integrative genomics approach identifies conserved transcriptomic networks in Alzheimer’s disease, Hum. Mol. Genet. 29 (17) (2020) 2899–2919. The authors declare no financial competing conflicts of interest in [23] T.S. Johnson, S. Xiang, T. Dong, Z. Huang, M. Cheng, T. Wang, K. Yang, D. Ni, relation to this work. K. Huang, J. Zhang, Combinatorial analyses reveal cellular composition changes have different impacts on transcriptomic changes of cell type specific genes in Alzheimer’s Disease, Sci. Rep. 11 (1) (2021) 353. References [24] B. Zhang, C. Gaiteri, L.G. Bodea, Z. Wang, J. McElwee, A.A. Podtelezhnikov, C. Zhang, T. Xie, L. Tran, R. Dobrin, E. Fluder, B. Clurman, S. Melquist, [1] G.B.D.D.F. Collaborators, Estimation of the global prevalence of dementia in 2019 M. Narayanan, C. Suver, H. Shah, M. Mahajan, T. Gillis, J. Mysore, M. and forecasted prevalence in 2050: an analysis for the Global Burden of Disease E. MacDonald, J.R. Lamb, D.A. Bennett, C. Molony, D.J. Stone, V. Gudnason, A. Study 2019, Lancet Public Health (2022). J. Myers, E.E. Schadt, H. Neumann, J. Zhu, V. Emilsson, Integrated systems [2] M. Goedert, M.G. Spillantini, A century of Alzheimer’s disease, Science 314 (5800) approach identifies genetic nodes and networks in late-onset Alzheimer’s disease, (2006) 777–781. Cell 153 (3) (2013) 707–720. [3] T.L. Spires-Jones, B.T. Hyman, The intersection of amyloid-β and tau at synapses in [25] A. Patir, B. Shih, B.W. McColl, T.C. Freeman, A core transcriptional signature of Alzheimer’s disease, Neuron 82 (4) (2014) 756–771. human microglia: Derivation and utility in describing region-dependent alterations [4] D. Mrdjen, E.J. Fox, S.A. Bukhari, K.S. Montine, S.C. Bendall, T.J. Montine, The associated with Alzheimer’s disease, Glia 67 (7) (2019) 1240–1253. basis of cellular and regional vulnerability in Alzheimer’s disease, Acta [26] J. Cheng, H.P. Liu, W.Y. Lin, F.J. Tsai, Machine learning compensates fold-change Neuropathol. 138 (5) (2019) 729–749. method and highlights oxidative phosphorylation in the brain transcriptome of [5] T.A. Pascoal, A.L. Benedet, N.J. Ashton, M.S. Kang, J. Therriault, M. Chamoun, Alzheimer’s disease, Sci. Rep. 11 (1) (2021) 13704. M. Savard, F.Z. Lussier, C. Tissot, T.K. Karikari, J. Ottoy, S. Mathotaarachchi, [27] E. Galea, L.D. Weinstock, R. Larramona-Arcas, A.F. Pybus, L. Gimenez-Llort, J. Stevenson, G. Massarweh, M. Scholl, M.J. de Leon, J.P. Soucy, P. Edison, C. Escartin, L.B. Wood, Multi-transcriptomic analysis points to early organelle K. Blennow, H. Zetterberg, S. Gauthier, P. Rosa-Neto, Microglial activation and tau dysfunction in human astrocytes in Alzheimer’s disease, Neurobiol. Dis. 166 propagate jointly across Braak stages, Nat. Med. 27 (9) (2021) 1592–1599. (2022), 105655. [6] B.I.C.C. Network, A multimodal cell census and atlas of the mammalian primary [28] S. Mostafavi, C. Gaiteri, S.E. Sullivan, C.C. White, S. Tasaki, J. Xu, M. Taga, H. motor cortex, Nature 598 (7879) (2021) 86–102. U. Klein, E. Patrick, V. Komashko, C. McCabe, R. Smith, E.M. Bradshaw, D.E. Root, [7] N.C. Berchtold, P.D. Coleman, D.H. Cribbs, J. Rogers, D.L. Gillen, C.W. Cotman, A. Regev, L. Yu, L.B. Chibnik, J.A. Schneider, T.L. Young-Pearse, D.A. Bennett, P. Synaptic genes are extensively downregulated across multiple brain regions in L. De Jager, A molecular network of the aging human brain provides insights into normal human aging and Alzheimer’s disease, Neurobiol. Aging 34 (6) (2013) the pathology and cognitive decline of Alzheimer’s disease, Nat. Neurosci. 21 (6) 1653–1661. (2018) 811–819. [8] W.S. Liang, T. Dunckley, T.G. Beach, A. Grover, D. Mastroeni, K. Ramsey, R. [29] M. Wang, P. Roussos, A. McKenzie, X. Zhou, Y. Kajiwara, K.J. Brennand, G.C. De J. Caselli, W.A. Kukull, D. McKeel, J.C. Morris, C.M. Hulette, D. Schmechel, E. Luca, J.F. Crary, P. Casaccia, J.D. Buxbaum, M. Ehrlich, S. Gandy, A. Goate, M. Reiman, J. Rogers, D.A. Stephan, Neuronal gene expression in non-demented P. Katsel, E. Schadt, V. Haroutunian, B. Zhang, Integrative network analysis of individuals with intermediate Alzheimer’s disease neuropathology, Neurobiol. nineteen brain regions identifies molecular signatures and networks underlying Aging 31 (4) (2010) 549–566. selective regional vulnerability to Alzheimer’s disease, Genome Med. 8 (1) (2016) [9] W.S. Liang, E.M. Reiman, J. Valla, T. Dunckley, T.G. Beach, A. Grover, T. 104. L. Niedzielko, L.E. Schneider, D. Mastroeni, R. Caselli, W. Kukull, J.C. Morris, C. [30] N. Liu, J. Xu, H. Liu, S. Zhang, M. Li, Y. Zhou, W. Qin, M.J. Li, C. Yu, I. Alzheimer’s M. Hulette, D. Schmechel, J. Rogers, D.A. Stephan, Alzheimer’s disease is disease Neuroimaging, Hippocampal transcriptome-wide association study and associated with reduced expression of energy metabolism genes in posterior neurobiological pathway analysis for Alzheimer’s disease, PLoS Genet. 17 (2) cingulate neurons, Proc. Natl. Acad. Sci. USA 105 (11) (2008) 4441–4446. (2021), e1009363. [10] E.M. Blalock, J.W. Geddes, K.C. Chen, N.M. Porter, W.R. Markesbery, P. [31] M. Jurkiewicz, D. Moser, A. Koller, L. Yu, E.I. Chen, D.A. Bennett, T. Canli, W. Landfield, Incipient Alzheimer’s disease: microarray correlation analyses reveal Integration of postmortem amygdala expression profiling, GWAS, and functional major transcriptional and tumor suppressor responses, Proc. Natl. Acad. Sci. USA cell culture assays: neuroticism-associated synaptic vesicle glycoprotein 2A (SV2A) 101 (7) (2004) 2173–2178. gene is regulated by miR-133a and miR-218, Transl. Psychiatry 10 (1) (2020) 297. [11] J.A. Miller, M.C. Oldham, D.H. Geschwind, A systems level analysis of [32] H. Mathys, J. Davila-Velderrain, Z. Peng, F. Gao, S. Mohammadi, J.Z. Young, transcriptional changes in Alzheimer’s disease and normal aging, J. Neurosci. 28 M. Menon, L. He, F. Abdurrob, X. Jiang, A.J. Martorell, R.M. Ransohoff, B. (6) (2008) 1410–1420. P. Hafler, D.A. Bennett, M. Kellis, L.H. Tsai, Single-cell transcriptomic analysis of [12] S.E. Counts, M.J. Alldred, S. Che, S.D. Ginsberg, E.J. Mufson, Synaptic gene Alzheimer’s disease, Nature 570 (7761) (2019) 332–337. dysregulation within hippocampal CA1 pyramidal neurons in mild cognitive [33] A. Grubman, G. Chew, J.F. Ouyang, G. Sun, X.Y. Choo, C. McLean, R.K. Simmons, impairment, Neuropharmacology 79 (2014) 172–179. S. Buckberry, D.B. Vargas-Landin, D. Poppe, J. Pflueger, R. Lister, O.J.L. Rackham, E. Petretto, J.M. Polo, A single-cell atlas of entorhinal cortex from individuals with 9 C.A. Saura et al. Seminars in Cell and Developmental Biology xxx (xxxx) xxx Alzheimer’s disease reveals cell-type-specific gene expression regulation, Nat. [56] R.J. Jackson, J.C. Meltzer, H. Nguyen, C. Commins, R.E. Bennett, E. Hudry, B. Neurosci. 22 (12) (2019) 2087–2097. T. Hyman, APOE4 derived from astrocytes leads to blood-brain barrier impairment, [34] S. Morabito, E. Miyoshi, N. Michael, S. Shahin, A.C. Martini, E. Head, J. Silva, Brain (2021). K. Leavy, M. Perez-Rosendahl, V. Swarup, Single-nucleus chromatin accessibility [57] R. Larramona-Arcas, C. Gonzalez-Arias, G. Perea, A. Gutierrez, J. Vitorica, and transcriptomic characterization of Alzheimer’s disease, Nat. Genet. 53 (8) T. Garcia-Barrera, J.L. Gomez-Ariza, R. Pascua-Maestro, M.D. Ganfornina, E. Kara, (2021) 1143–1155. E. Hudry, M. Martinez-Vicente, M. Vila, E. Galea, R. Masgrau, Sex-dependent [35] S.F. Lau, H. Cao, A.K.Y. Fu, N.Y. Ip, Single-nucleus transcriptome analysis reveals calcium hyperactivity due to lysosomal-related dysfunction in astrocytes from dysregulation of angiogenic endothelial cells and neuroprotective glia in APOE4 versus APOE3 gene targeted replacement mice, Mol. Neurodegener. 15 (1) Alzheimer’s disease, Proc. Natl. Acad. Sci. USA 117 (41) (2020) 25800–25809. (2020) 35. [36] K. Leng, E. Li, R. Eser, A. Piergies, R. Sit, M. Tan, N. Neff, S.H. Li, R.D. Rodriguez, C. [58] P. Carriba, L. Pardo, A. Parra-Damas, M.P. Lichtenstein, C.A. Saura, A. Pujol, K. Suemoto, R.E.P. Leite, A.J. Ehrenberg, C.A. Pasqualucci, W.W. Seeley, S. Spina, R. Masgrau, E. Galea, ATP and noradrenaline activate CREB in astrocytes via H. Heinsen, L.T. Grinberg, M. Kampmann, Molecular characterization of selectively noncanonical Ca(2+) and cyclic AMP independent pathways, Glia 60 (9) (2012) vulnerable neurons in Alzheimer’s disease, Nat. Neurosci. 24 (2) (2021) 276–287. 1330–1344. [37] Y. Zhou, W.M. Song, P.S. Andhey, A. Swain, T. Levy, K.R. Miller, P.L. Poliani, [59] M. Puigdellivol, S. Milde, A. Vilalta, T.O.J. Cockram, D.H. Allendorf, J.Y. Lee, J. M. Cominelli, S. Grover, S. Gilfillan, M. Cella, T.K. Ulland, K. Zaitsev, A. Miyashita, M. Dundee, K. Pampuscenko, V. Borutaite, H.N. Nuthall, J.H. Brelstaff, M. T. Ikeuchi, M. Sainouchi, A. Kakita, D.A. Bennett, J.A. Schneider, M.R. Nichols, S. G. Spillantini, G.C. Brown, The microglial P2Y6 receptor mediates neuronal loss A. Beausoleil, J.D. Ulrich, D.M. Holtzman, M.N. Artyomov, M. Colonna, Human and memory deficits in neurodegeneration, Cell Rep. 37 (13) (2021), 110148. and mouse single-nucleus transcriptomics reveal TREM2-dependent and TREM2- [60] D. Boche, M.N. Gordon, Diversity of transcriptomic microglial phenotypes in aging independent cellular responses in Alzheimer’s disease, Nat. Med. 26 (1) (2020) and Alzheimer’s disease, Alzheimers Dement 18 (2) (2021) 360–376. 131–142. [61] M. Olah, V. Menon, N. Habib, M.F. Taga, Y. Ma, C.J. Yung, M. Cimpean, [38] A.C. Yang, R.T. Vest, F. Kern, D.P. Lee, M. Agam, C.A. Maat, P.M. Losada, M. A. Khairallah, G. Coronas-Samano, R. Sankowski, D. Grun, A.A. Kroshilina, B. Chen, N. Schaum, N. Khoury, A. Toland, K. Calcuttawala, H. Shin, R. Palovics, D. Dionne, R.A. Sarkis, G.R. Cosgrove, J. Helgager, J.A. Golden, P.B. Pennell, A. Shin, E.Y. Wang, J. Luo, D. Gate, W.J. Schulz-Schaeffer, P. Chu, J. M. Prinz, J.P.G. Vonsattel, A.F. Teich, J.A. Schneider, D.A. Bennett, A. Regev, A. Siegenthaler, M.W. McNerney, A. Keller, T. Wyss-Coray, A human brain vascular W. Elyaman, E.M. Bradshaw, P.L. De Jager, Single cell RNA sequencing of human atlas reveals diverse mediators of Alzheimer’s risk, Nature (2022). microglia uncovers a subset associated with Alzheimer’s disease, Nat. Commun. 11 [39] M.B. Miller, A.Y. Huang, J. Kim, Z. Zhou, S.L. Kirkham, E.A. Maury, J.S. Ziegenfuss, (1) (2020) 6129. H.C. Reed, J.E. Neil, L. Rento, S.C. Ryu, C.C. Ma, L.J. Luquette, H.M. Ames, D. [62] A. Sobue, O. Komine, Y. Hara, F. Endo, H. Mizoguchi, S. Watanabe, S. Murayama, H. Oakley, M.P. Frosch, B.T. Hyman, M.A. Lodato, E.A. Lee, C.A. Walsh, Somatic T. Saito, T.C. Saido, N. Sahara, M. Higuchi, T. Ogi, K. Yamanaka, Microglial gene genomic changes in single Alzheimer’s disease neurons, Nature 604 (2022) signature reveals loss of homeostatic microglia associated with neurodegeneration 714–722. of Alzheimer’s disease, Acta Neuropathol. Commun. 9 (1) (2021) 1. [40] A.M. Smith, K. Davey, S. Tsartsalis, C. Khozoie, N. Fancy, S.S. Tang, E. Liaptsi, [63] K. Srinivasan, B.A. Friedman, A. Etxeberria, M.A. Huntley, M.P. van der Brug, M. Weinert, A. McGarry, R.C.J. Muirhead, S. Gentleman, D.R. Owen, P. O. Foreman, J.S. Paw, Z. Modrusan, T.G. Beach, G.E. Serrano, D.V. Hansen, M. Matthews, Diverse human astrocyte and microglial transcriptional responses to Alzheimer’s patient microglia exhibit enhanced aging and unique transcriptional Alzheimer’s pathology, Acta Neuropathol. 143 (1) (2022) 75–91. activation, Cell Rep. 31 (13) (2020), 107843. [41] X.L. Wang, L. Li, Cell type-specific potential pathogenic genes and functional [64] N. Thrupp, C. Sala Frigerio, L. Wolfs, N.G. Skene, N. Fattorelli, S. Poovathingal, pathways in Alzheimer’s Disease, BMC Neurol. 21 (1) (2021) 381. Y. Fourne, P.M. Matthews, T. Theys, R. Mancuso, B. de Strooper, M. Fiers, Single- [42] V.N. Bukke, M. Archana, R. Villani, A.D. Romano, A. Wawrzyniak, K. Balawender, nucleus RNA-Seq is not suitable for detection of microglial activation genes in S. Orkisz, S. Beggiato, G. Serviddio, T. Cassano, The dual role of glutamatergic humans, Cell Rep. 32 (13) (2020), 108189. neurotransmission in Alzheimer’s disease: from pathophysiology to [65] S.E. Marsh, A.J. Walker, T. Kamath, L. Dissing-Olesen, T.R. Hammond, T.Y. de pharmacotherapy, Int J. Mol. Sci. 21 (20) (2020). Soysa, A.M.H. Young, S. Murphy, A. Abdulraouf, N. Nadaf, C. Dufort, A.C. Walker, [43] K.D. Micheva, D. Wolman, B.D. Mensh, E. Pax, J. Buchanan, S.J. Smith, D.D. Bock, L.E. Lucca, V. Kozareva, C. Vanderburg, S. Hong, H. Bulstrode, P.J. Hutchinson, D. A large fraction of neocortical myelin ensheathes axons of local inhibitory neurons, J. Gaffney, D.A. Hafler, R.J.M. Franklin, E.Z. Macosko, B. Stevens, Dissection of Elife 5 (2016). artifactual and confounding glial signatures by single-cell sequencing of mouse and [44] S.D. Ginsberg, S.E. Hemby, V.M. Lee, J.H. Eberwine, J.Q. Trojanowski, Expression human brain, Nat. Neurosci. 25 (3) (2022) 306–316. profile of transcripts in Alzheimer’s disease tangle-bearing CA1 neurons, Ann. [66] M. Bouhrara, D.A. Reiter, C.M. Bergeron, L.M. Zukley, L. Ferrucci, S.M. Resnick, R. Neurol. 48 (1) (2000) 77–87. G. Spencer, Evidence of demyelination in mild cognitive impairment and dementia [45] B. Styr, I. Slutsky, Imbalance between firing homeostasis and synaptic plasticity using a direct and specific magnetic resonance imaging measure of myelin content, drives early-phase Alzheimer’s disease, Nat. Neurosci. 21 (4) (2018) 463–473. Alzheimers Dement 14 (8) (2018) 998–1004. [46] D.R. Brady, E.J. Mufson, Parvalbumin-immunoreactive neurons in the [67] S.E. Nasrabady, B. Rizvi, J.E. Goldman, A.M. Brickman, White matter changes in hippocampal formation of Alzheimer’s diseased brain, Neuroscience 80 (4) (1997) Alzheimer’s disease: a focus on myelin and oligodendrocytes, Acta Neuropathol. 1113–1125. Commun. 6 (1) (2018) 22. [47] E. Sanchez-Mejias, C. Nunez-Diaz, R. Sanchez-Varo, A. Gomez-Arboledas, J. [68] W.T. Chen, A. Lu, K. Craessaerts, B. Pavie, C. Sala Frigerio, N. Corthout, X. Qian, A. Garcia-Leon, J.J. Fernandez-Valenzuela, M. Mejias-Ortega, L. Trujillo-Estrada, J. Lalakova, M. Kuhnemund, I. Voytyuk, L. Wolfs, R. Mancuso, E. Salta, S. Balusu, D. Baglietto-Vargas, I. Moreno-Gonzalez, J.C. Davila, J. Vitorica, A. Gutierrez, A. Snellinx, S. Munck, A. Jurek, J. Fernandez Navarro, T.C. Saido, I. Huitinga, Distinct disease-sensitive GABAergic neurons in the perirhinal cortex of J. Lundeberg, M. Fiers, B. De Strooper, Spatial transcriptomics and in situ Alzheimer’s mice and patients, Brain Pathol. 30 (2) (2020) 345–363. sequencing to study Alzheimer’s disease, Cell 182 (4) (2020) 976–991 e19. [48] J.C. Lauterborn, P. Scaduto, C.D. Cox, A. Schulmann, G. Lynch, C.M. Gall, C. [69] A. Montagne, S.R. Barnes, M.D. Sweeney, M.R. Halliday, A.P. Sagare, Z. Zhao, A. D. Keene, A. Limon, Increased excitatory to inhibitory synaptic ratio in parietal W. Toga, R.E. Jacobs, C.Y. Liu, L. Amezcua, M.G. Harrington, H.C. Chui, M. Law, B. cortex samples from individuals with Alzheimer’s disease, Nat. Commun. 12 (1) V. Zlokovic, Blood-brain barrier breakdown in the aging human hippocampus, (2021) 2603. Neuron 85 (2) (2015) 296–302. [49] F. Leng, P. Edison, Neuroinflammation and microglial activation in Alzheimer [70] B. Hussain, C. Fang, J. Chang, Blood-Brain Barrier breakdown: an emerging disease: where do we go from here? Nat. Rev. Neurol. 17 (3) (2021) 157–172. biomarker of cognitive impairment in normal aging and dementia, Front. Neurosci. [50] A.M. Arranz, B. De Strooper, The role of astroglia in Alzheimer’s disease: 15 (2021), 688090. pathophysiology and clinical implications, Lancet Neurol. 18 (4) (2019) 406–414. [71] T.V. Procter, A. Williams, A. Montagne, Interplay between brain Pericytes and [51] A. Gomez-Arboledas, J.C. Davila, E. Sanchez-Mejias, V. Navarro, C. Nunez-Diaz, endothelial cells in dementia, Am. J. Pathol. 191 (11) (2021) 1917–1931. R. Sanchez-Varo, M.V. Sanchez-Mico, L. Trujillo-Estrada, J.J. Fernandez- [72] A.H. Vagnucci Jr., W.W. Li, Alzheimer’s disease and angiogenesis, Lancet 361 Valenzuela, M. Vizuete, J.X. Comella, E. Galea, J. Vitorica, A. Gutierrez, Phagocytic (9357) (2003) 605–608. clearance of presynaptic dystrophies by reactive astrocytes in Alzheimer’s disease, [73] F.J. Garcia, N. Sun, H. Lee, B. Godlewski, K. Galani, B. Zhou, J. Mantero, D. Glia 66 (3) (2018) 637–653. A. Bennett, M. Sahin, M. Kellis, M. Heiman, Single-cell dissection of the human [52] K. Richetin, P. Steullet, M. Pachoud, R. Perbet, E. Parietti, M. Maheswaran, brain vasculature, Nature (2022). S. Eddarkaoui, S. Begard, C. Pythoud, M. Rey, R. Caillierez, Q.D.K. Halliez S., [74] K. Kisler, A.R. Nelson, A. Montagne, B.V. Zlokovic, Cerebral blood flow regulation P. Bezzi, L. Buee, G. Leuba, M. Colin, N. Toni, N. Deglon, Tau accumulation in and neurovascular dysfunction in Alzheimer disease, Nat. Rev. Neurosci. 18 (7) astrocytes of the dentate gyrus induces neuronal dysfunction and memory deficits (2017) 419–434. in Alzheimer’s disease, Nat. Neurosci. 23 (12) (2020) 1567–1579. [75] S. Peng, L. Zeng, J.V. Haure-Mirande, M. Wang, D.M. Huffman, V. Haroutunian, M. [53] S.C. Konings, L. Torres-Garcia, I. Martinsson, G.K. Gouras, Astrocytic and neuronal E. Ehrlich, B. Zhang, Z. Tu, Transcriptomic changes highly similar to Alzheimer’s Apolipoprotein E isoforms differentially affect neuronal excitability, Front disease are observed in a subpopulation of individuals during normal brain aging, Neurosci. 15 (2021), 734001. Front. Aging Neurosci. 13 (2021), 711524. [54] T.E. Mahan, C. Wang, X. Bao, A. Choudhury, J.D. Ulrich, D.M. Holtzman, Selective [76] A.M. Alsema, Q. Jiang, L. Kracht, E. Gerrits, M.L. Dubbelaar, A. Miedema, reduction of astrocyte apoE3 and apoE4 strongly reduces Abeta accumulation and N. Brouwer, E.M. Hol, J. Middeldorp, R. van Dijk, M. Woodbury, A. Wachter, S. Xi, plaque-related pathology in a mouse model of amyloidosis, Mol. Neurodegener. 17 T. Moller, K.P. Biber, S.M. Kooistra, E. Boddeke, B.J.L. Eggen, Profiling microglia (1) (2022) 13. from Alzheimer’s disease donors and non-demented elderly in acute human [55] Y.T. Lin, J. Seo, F. Gao, H.M. Feldman, H.L. Wen, J. Penney, H.P. Cam, E. Gjoneska, postmortem cortical tissue, Front. Mol. Neurosci. 13 (2020) 134. W.K. Raja, J. Cheng, R. Rueda, O. Kritskiy, F. Abdurrob, Z. Peng, B. Milo, C.J. Yu, S. Elmsaouri, D. Dey, T. Ko, B.A. Yankner, L.H. Tsai, APOE4 causes widespread molecular and cellular alterations associated with Alzheimer’s disease phenotypes in human iPSC-derived brain cell types, Neuron 98 (6) (2018) 1141–1154 e7. 10 C.A. Saura et al. Seminars in Cell and Developmental Biology xxx (xxxx) xxx [77] A. Kulkarni, A.G. Anderson, D.P. Merullo, G. Konopka, Beyond bulk: a review of [79] E. Mereu, A. Lafzi, C. Moutinho, C. Ziegenhain, D.J. McCarthy, A. Alvarez-Varela, single cell transcriptomics methodologies and applications, Curr. Opin. Biotechnol. E. Batlle, Sagar, D. Grun, J.K. Lau, S.C. Boutet, C. Sanada, A. Ooi, R.C. Jones, 58 (2019) 129–136. K. Kaihara, C. Brampton, Y. Talaga, Y. Sasagawa, K. Tanaka, T. Hayashi, [78] J.W. Squair, M. Gautier, C. Kathe, M.A. Anderson, N.D. James, T.H. Hutson, C. Braeuning, C. Fischer, S. Sauer, T. Trefzer, C. Conrad, X. Adiconis, L.T. Nguyen, R. Hudelle, T. Qaiser, K.J.E. Matson, Q. Barraud, A.J. Levine, G. La Manno, M. A. Regev, J.Z. Levin, S. Parekh, A. Janjic, L.E. Wange, J.W. Bagnoli, W. Enard, A. Skinnider, G. Courtine, Confronting false discoveries in single-cell differential M. Gut, R. Sandberg, I. Nikaido, I. Gut, O. Stegle, H. Heyn, Benchmarking single- expression, Nat. Commun. 12 (1) (2021) 5692. cell RNA-sequencing protocols for cell atlas projects, Nat. Biotechnol. 38 (6) (2020) 747–755. 11