Dysregulated glial genes in Alzheimer's disease are essential for homeostatic plasticity: Evidence from integrative epigenetic and single cell analyses

Abstract Synaptic homeostatic plasticity is a foundational regulatory mechanism that maintains the stability of synaptic and neural functions within the nervous system. Impairment of homeostatic regulation has been linked to synapse destabilization during the progression of Alzheimer's disease (AD). Recent epigenetic and transcriptomic characterizations of the nervous system have revealed intricate molecular details about the aging brain and the pathogenesis of neurodegenerative diseases. Yet, how abnormal epigenetic and transcriptomic alterations in different cell types in AD affect synaptic homeostatic plasticity remains to be elucidated. Various glial cell types play critical roles in modulating synaptic functions both during the aging process and in the context of AD. Here, we investigated the impact of glial dysregulation of histone acetylation and transcriptome in AD on synaptic homeostatic plasticity, using computational analysis combined with electrophysiological methods in Drosophila. By integrating snRNA‐seq and H3K9ac ChIP‐seq data from the same AD patient cohort, we pinpointed cell type‐specific signature genes that were transcriptionally altered by histone acetylation. We subsequently investigated the role of these glial genes in regulating presynaptic homeostatic potentiation in Drosophila. Remarkably, nine glial‐specific genes, which were identified through our computational method as targets of H3K9ac and transcriptional dysregulation, were found to be crucial for the regulation of synaptic homeostatic plasticity in Drosophila. Our genetic evidence connects abnormal glial transcriptomic changes in AD with the impairment of homeostatic plasticity in the nervous system. In summary, our integrative computational and genetic studies highlight specific glial genes as potential key players in the homeostatic imbalance observed in AD.


| INTRODUC TI ON
Alzheimer's disease (AD) is a prominent neurodegenerative disorder and the leading cause of dementia in older adults (Knopman et al., 2021).Dysfunctional synaptic activity is closely linked with AD pathology progression (Selkoe, 2002).Homeostatic plasticity plays a critical role in preserving synaptic physiology, neural activity, and overall brain function (Davis, 2006;Turrigiano, 2008).Disruptions in this mechanism can lead to neural network instability and are implicated in conditions such as epilepsy and neurodegenerative disorders (Orr et al., 2020;Styr & Slutsky, 2018;Wondolowski & Dickman, 2013).Presynaptic homeostatic potentiation (PHP) is a fundamental form of homeostatic regulation that is evolutionarily conserved from Drosophila to humans (Cull-Candy et al., 1980;Davis & Muller, 2015;Plomp et al., 1992).PHP ensures stability in postsynaptic excitation by counteracting perturbations in glutamate receptors through compensatory upregulation in presynaptic neurotransmitter release (Dickman & Davis, 2009;Frank et al., 2006).
Recent studies have linked shifts in the epigenetic landscape to potential causes of AD, deepening our understanding of the disease's molecular etiology (Klein et al., 2019;Lord & Cruchaga, 2014;Morabito et al., 2021;Nativio et al., 2018).Notably, single-cell transcriptomic investigations in both AD mouse models and patient cohorts have illuminated critical roles of glia in AD pathogenesis (Grubman et al., 2019;Habib et al., 2020;Keren-Shaul et al., 2017;Lau et al., 2020;Zhou et al., 2020).While neuronal deficits in AD have been the subject of extensive research, there is growing evidence underscoring the important function glial cells have in modulating synaptic transmission and plasticity, both under normal conditions and in the diseased brain (Chung et al., 2015;Hong et al., 2016;Keren-Shaul et al., 2017;Stogsdill & Eroglu, 2017).Our previous work has shown that glial epigenetic signaling, as well as interactions between glia and neurons, are fundamental for inducing and sustaining synaptic homeostatic plasticity in Drosophila (Wang et al., 2014;Wang et al., 2020).Yet, the influence of abnormal epigenetic and transcriptomic shifts in glial cells on synaptic homeostatic plasticity within the context of AD remains largely unexplored.
In this study, we investigated how the aberrant histone acetylation and glial transcriptomic shifts associated with AD affect synaptic homeostasis.By employing computational methods, we integrated data from previously published ChIP-seq and snRNA-seq datasets (Klein et al., 2019;Mathys et al., 2019).Through this, we identified cell type-specific differentially expressed genes (DEGs) in post-mortem AD brains, which were consistently regulated by histone acetylation.Following this, we examined the function of these glial-specific DEGs on homeostatic plasticity through both genetic and electrophysiological methods in Drosophila models.Notably, all nine glial-specific DEGs we analyzed were indispensable for PHP.This emphasizes the pivotal role glia play in homeostatic regulation and the preservation of synaptic stability.

| CELL T YPE-S PECIFI C DYS REG UL ATI ON OF H3K9AC-TARG E T G ENE S IN AD
To understand the molecular interplay between aberrant histone acetylation, transcriptomic changes in AD, and homeostatic plasticity, we began by pinpointing genes affected transcriptionally by abnormal histone acetylation in AD.We were particularly drawn to the dysregulation at the histone H3 lysine nine site (H3K9ac) for several reasons.First, prior studies underscored an anomalous shift in H3K9ac-dependent epigenetic modulation within AD brains, but the direct consequences of this dysregulation on synaptic functions are yet to be elucidated (Klein et al., 2019;Santana et al., 2022;Tasaki et al., 2019).Second, our previous research showed that the Spt-Ada-Gcn5 acetyltransferase (SAGA) complex-driven histone acetylation at H3K9ac in glia is crucial for the glial-neuron interactions necessary for synaptic homeostatic plasticity (Wang et al., 2020).Third, in a subsequent study exploring glial epigenetic regulation in Drosophila AD models, we observed a pronounced reduction in H3K9ac levels within glial nuclei in transgenic models of Amyloid β42 (Aβ42).
Additionally, overexpression of human Aβ42 in neurons (but not tau) led to PHP deficits (Yin et al., 2023).This indicates that glial cells may detect changes in Aβ levels, prompting abnormal H3K9ac-mediated signaling that impairs PHP.Given these insights, our study focused on investigating H3K9ac-driven transcriptomic changes across various cell types in AD, with a spotlight on glia.
We performed an integrative analysis by combining a previously published H3K9ac ChIP-seq dataset (Klein et al., 2019) with a snRNA-seq dataset (Mathys et al., 2019).Both datasets originated from the prefrontal cortex of post-mortem brains and were part of the same cohort studies on aging and dementia, specifically the ROSMAP project (Figure 1a).Recent studies underline that the accumulation of Aβ not only leads to glial activation but also markedly compromises glial functionality (Huffels et al., 2023;Sierksma et al., 2020;Zeng et al., 2023).In our previous work, we observed that Aβ42 overexpression caused reduction in glial H3K9ac and led to PHP impairment in Drosophila AD models (Yin et al., 2023).Therefore, we centered our analysis on the H3K9ac peaks associated with Aβ.First, we analyzed these H3K9ac peaks in the ChIP-seq dataset, using the criteria delineated in the original publication (FDR < 0.05).A subsequent Gene Ontology (GO) assessment of the 329 genes, which were annotated based on chromatin states from the original study, pinpointed pathways that were specifically related to Aβ (Figure 1b; Figure S1a).It is worth noting that the genes associated with Aβ play crucial roles in modulating signaling processes related to axon and plasma membrane activities (Figure 1b; Figure S1a).
In our subsequent analysis, we aimed to identify DEGs targeted by H3K9ac that exhibit dysregulation across various cell types in the AD brain.Drawing from the published snRNA-seq study, we found a total of 4982 AD-associated DEGs (Figure 1a).By overlapping these DEGs with the 329 Aβ-associated H3K9ac-target genes sourced from the ChIP-seq dataset, we sought to determine cell type-specific transcriptomic alterations.We found that 100 H3K9ac-targeted cell type-specific DEGs (representing 87 distinct genes) displayed dysregulation, either being up-or downregulated in AD (Figure 1a,c).
Notably, the most prominent dysregulations were evident in excitatory neurons, inhibitory neurons, and oligodendrocytes (Figure 1c).
A detailed breakdown revealed that the majority of DEGs in excitatory neurons (77%), inhibitory neurons (100%), and microglia (67%) were downregulated.In contrast, 70% of DEGs specific to oligodendrocytes showed upregulation (Figure 1d).The opposing transcriptional changes observed in neurons and oligodendrocytes suggest

| TR ANSCRIP TOMIC CHANG E S CORREL ATE WITH A Β-LINK ED HK9AC ALTER ATI ON S IN AD
To assess the impact of Aβ pathology-related H3K9ac modifications on cell-specific gene transcription in AD, we probed the correlation between coefficient Aβ (from ChIP-seq) and cell type-specific transcriptomic shifts (from snRNA-seq).Coefficient Aβ denotes the extent to which H3K9ac peak reads in the ChIP-seq dataset are influenced by the severity of Aβ pathology.Specifically, a positive coefficient Aβ indicates that H3K9ac peak reads vary in the same direction with the overall amyloid load, and an elevated absolute value suggests that a given amyloid burden induces a more pronounced change in H3K9ac peak reads (Klein et al., 2019).We conducted a correlation analysis between coefficient Aβ and the transcriptional log fold change for each of the identified H3K9actargeted DEGs (a total of 100 DEGs).A significant positive correlation was found between coefficient Aβ and transcriptional fold change across all cell types (Pearson correlation = 0.37, p = 1.44e-4, Figure 1e).When we examined this correlation within specific cell types, we found significant correlations in both excitatory and inhibitory neurons (Figure 1g).Although statistical significance was not achieved for astrocytes, oligodendrocyte progenitor cells (OPCs), and microglia, possibly due to the limited number of DEGs, a Pearson correlation of 0.36 was observed for oligodendrocytes (p = 0.09; Figure 1f,g).
F I G U R E 2 Glial DEGs that are direct targets of H3K9ac control PHP.(a) Schematic to show criteria used to select genes for the electrophysiology-based genetic screen in Drosophila (upper left panel).Genes that are H3K9ac targets, cell type-specific DEGs, acquire coefficient Aβ and transcriptional log 2 fold change with the "same sign," and have identified Drosophila homologues are selected for the electrophysiological characterization.If a gene is a DEG in multiple cell types, it is assigned to the cell type with the largest absolute value of log 2 fold change.The Spearman correlation between the expression of the top 20 highest expressed human "synaptic" genes in various cell types and the expression of their homologues in Drosophila perineurial glia (lower left panel).Human gene names, corresponding homologues in Drosophila, and GO terms associated with human glial DEGs that are selected for electrophysiological screen are shown.From this illustration, it is evident that, depending upon the respective "signs" of the coefficient Aβ and log fold change, an incremental increase in Aβ levels can result in either the up-or downregulation of gene expression.
Remarkably, across all examined cell types, 69 out of the 100 DEGs (or 69%) displayed "same sign" in their transcriptional log fold change and coefficient Aβ values (Figure 1e,h,i; Figure S1b, with detailed data in Tables S1 and S2).Notably, previously identified AD risk factors, including SPP1, have been discovered as genes with "same sign."To determine if this correlation was statistically significant, we conducted a two-sided binomial test, comparing the observed percentage of genes with the "same sign" against 50% (the likelihood if signs were assigned randomly).This analysis revealed that 66% of DEGs in excitatory neurons (p = 0.03) and a striking 78% in oligodendrocytes (p = 0.01) displayed this "same sign" trait (Figure 1f,g).Given that coefficient Aβ essentially represents the impact of Aβ pathology on H3K9ac, this significant overrepresentation of genes with the "same sign" underscores the positive co-regulation between H3K9ac dysregulation and the transcriptional abnormalities in AD.

| G LIAL DEG S IDENTIFIED IN AD ARE E SS ENTIAL FOR PHP IN DROSOPHIL A
To explore the potential association between impaired homeostatic plasticity and the concurrent anomalies in H3K9ac and transcriptomes within AD, we set forth a series of selection criteria for a genetic screen in Drosophila (Figure 2a).We utilized the following sequential criteria: 1.The gene is a direct target of H3K9ac associated with Aβ.
2. The gene is a dysregulated cell type-specific DEG in AD.Applying the first 4 criteria, we pinpointed 69 DEGs that fulfill these conditions (Figures 1e,i and 2a; Tables S1 and S2).Moreover, our data analysis indicated the majority of DEGs to be downregulated in AD (Figure 1d).Building on this, we selected all the downregulated glial-specific DEGs (8 DEGs) that had direct Drosophila homologues (six meet all the criteria) and added three upregulated glial DEGs to our list for an in-depth electrophysiological examination (Figures 1i   and 2a).
At the NMJ, three distinct peripheral glial cells-wrapping glia, subperineurial glia, and perineurial glia-interact with both the presynaptic motoneurons and postsynaptic muscles (Wang et al., 2020;Yildirim et al., 2019).This unique setup aids in decoding the intricacies of glia-neuron communications essential for PHP.Our earlier work unveiled the crucial role of perineurial glia's secretory function in PHP.There is a transcriptional upregulation of the extracellular matrix molecule, Multiplexin, in these glial cells during PHP, which is triggered by chronic inhibition of postsynaptic glutamate receptor activity, and is mediated by the SAGA complex-dependent H3K9ac (Wang et al., 2020).These results emphasize glial signaling molecules' central role in modulating neuronal activities during PHP.Consequently, we sought to systematically examine the regulatory roles of glial-specific DEGs (from computational analysis) in PHP, through genetic and electrophysiological examinations at the fly NMJ.
While the DEGs were identified from distinct glial cell types (astrocytes, microglia, oligodendrocytes, and OPCs) in the human brain, we first sought to determine if the perineurial glia in Drosophila exhibit activities similar to those of mammalian glial cells in terms of regulating synaptic functions.To this end, we conducted a computational analysis to assess whether genes linked with synaptic functions in various human glial cell types are also "enriched" and show correlative expression in Drosophila perineurial glia.We analyzed a scRNA-seq dataset from Drosophila brains (Davie et al., 2018) and a snRNA-seq dataset from human brains (non-pathological, [Mathys et al., 2019]) and identified genes that possess "synaptic" functions as defined by GO terms with the keyword "synaptic."We discovered that for the top 20 (we observed consistent results, as detailed below, when analyzing either the top 30 or 40 genes) most highly expressed synaptic-related human genes in each cell type (excitatory neurons, inhibitory neurons, astrocytes, microglia, oligodendrocytes, and OPCs), all the corresponding Drosophila homologues are expressed in perineurial glia.
Next, we conducted a gene Expression Weighted Cell-type Enrichment ("enrichment," Skene & Grant, 2016) analysis to determine whether the Drosophila homologues of the top 20 "synaptic" genes expressed in human glial cell types are represented in perineurial glia.The "enrichment" refers to the likelihood of the top 20 Drosophila homologues within a specific human cell type displaying higher expression levels in perineurial glia than what would be expected by chance.We found that, in perineurial glia, the Drosophila homologues of top 20 human "synaptic" genes exhibited higher expressions than 99.8% of randomly sampled expressed genes in astrocytes, 95.7% in microglia, and 87.7% in oligodendrocytes.These findings suggest that perineurial glia play roles in regulating synaptic properties, which further underscores the potential synaptic functional parallels between these cell types.
Furthermore, for the top 20 most highly expressed human "synaptic" genes, we found that correlations between human expression levels (in various cell types) and Drosophila homologue expression levels in perineurial glia are positive for oligodendrocytes (Spearman correlation = 0.37, p = 0.11) and astrocytes (0.30, p = 0.20), while negative for excitatory neurons (−0.14, p = 0.54) and mild for inhibitory neurons (0.13, p = 0.13).Even though the p-values did not reach the conventional 0.05 significance level, these findings nonetheless suggest a directional relationship between the expression levels of "synaptic" genes in Drosophila perineurial glia and those in human astrocytes and oligodendrocytes (Figure 2a, bottom panel).
Additionally, there exists a positive correlation between the expression levels of "synaptic" genes in oligodendrocytes (Spearman correlation = 0.40, p = 0.08) and astrocytes (0.27, p = 0.25), and the percentage of perineurial glia expressing these genes (Figure S2a).Therefore, even though Drosophila perineurial glia do not share identical cell identities with mammalian glia in the central nervous system, it is plausible that the mechanisms for regulating synaptic functions are conserved.
Next, we examined the GO terms and found that the glial DEGs included in the genetic screen are generally involved in regulating the functions associated with axon, dendritic spine, synapse, myelination, chromatin, extracellular space, and apoptotic process (Figure 2a).Notably, these GO terms align with those associated with Aβ in the H3K9ac ChIP-seq dataset (Figure 1b).To further understand the functions of these DEGs in regulating PHP, we investigated the rapid induction of PHP (Frank et al., 2006).We assessed PHP in loss-of-function genetic mutants of Septin5, CG42709, h, dpn, E2f1, NimC1, M6, Mkp3, and Cep97.In order to examine whether the identified DEGs function in glia for PHP, we knocked down h, dpn, NimC1, M6, and grk specifically in perineurial glia using RNAi strategies.Taking into account the expression of human E2F1 is upregulated in oligodendrocytes in AD, we also explored the impact of overexpressing E2f1 in glial cells on PHP in Drosophila.Moreover, as an important control, we assessed whether motoneuron-specific knockdown of dpn, NimC1, and M6 affect PHP (Figure 2b-e).
Bath application of philanthotoxin (PhTX) significantly reduced the mEPSP amplitude in all genotypes (Figure 2b-e).We calculated quantal content (QC) as the ratio of EPSP amplitude to mEPSP amplitude, using it as an estimate for the number of vesicles released from presynaptic terminals.In the wild-type, a decrease in mEPSP amplitude is associated with an increase in presynaptic release, as inferred from QC values.This relationship indicates a negative correlation between mEPSP amplitude and QC, underscoring a compensatory mechanism that facilitates neurotransmitter release in the presence of PhTX (Figure 2b,c).Quantitatively, a significant increase in presynaptic release, as measured by QC, counterbalances the decrease in mEPSP amplitude, ensuring that postsynaptic excitation remains at the baseline value observed in the wild-type (see Figure 2d,e).Conversely, there is no change in QC and the average EPSP amplitude is significantly decreased in all mutant alleles of Septin5, CG42709, h, dpn, E2f1, NimC1, M6, Mkp3, and Cep97, as well as perineurial glial-specific knockdown of h, dpn, NimC1, M6, and grk (Figure 2b-e).Interestingly, both loss-of-function and gain-of-function (glial overexpression) of E2f1 block PHP, highlighting a dosage-sensitive activity of E2f1 in synaptic homeostasis (Figure 2b-e).Notably, the negative correlation between QC and mEPSP amplitude is absent in all genetic, perineurial glial-specific RNAi knockdown, and overexpression mutants that were tested, suggesting that PhTX-induced homeostatic compensation is impaired when the expression of these glial DEGs is perturbed (Figure 2b,c).Taken together, we demonstrated that H3K9ac-target genes that are dysregulated in glia in AD are critical for homeostatic plasticity in Drosophila.
While human HES4 (Drosophila dpn) and TNFRSF21 (Drosophila NimC1) showed the most significant transcriptional changes in oligodendrocytes (Figure 1i), they were identified as neuronal DEGs in AD as well.Thus, we decided to further explore whether they function in neurons or glia for PHP.As a control, we also examined the neuronal role of M6 (human GPM6A, a DEG exclusively in oligodendrocytes) in PHP.We discovered that when we knockdown dpn and NimC1 in motoneurons, PHP is still expressed, albeit to a lesser degree than in the wild-type control (Figure 2c-g).Interestingly, PHP is completely normal when M6 is knocked down specifically in motoneurons (Figure 2c-g).There is a significant increase in neurotransmitter release when these three genes were knockdown specifically in motoneurons (Figure 2f).Moreover, we analyzed the percentage EPSP amplitude and QC in the presence of PhTX compared to the basal conditions, when dpn, NimC1, and M6 were knocked down specifically in perineurial glia or motoneurons (Figure 2g).We found that there was a more pronounced reduction in EPSP amplitude and a smaller increase in QC in the presence of PhTX, when these genes were specifically knocked in perineurial glia than in neurons (Figure 2g).These findings suggest that dpn, NimC1, and M6 primarily operate in glia for synaptic homeostatic plasticity.The impairment of PHP is consistent with the observation of greater transcriptional dysregulation of dpn, NimC1, and M6 in human glial cell types than in neurons (Figure 2g).
Finally, we examined whether the % EPSP amplitude and QC in PHP (normalized to baseline values in the absence of PhTX) is quantitatively correlated with the magnitude of transcriptional fold change and coefficient Aβ in AD.We found that the % EPSP amplitude and QC in mutants in the presence of PhTX exhibited negative correlations with the magnitude of transcriptional log fold change in AD (absolute value, Spearman correlation = −0.66,p = 0.02 for EPSP amplitude, −0.37, p = 0.11 for QC, Figure 2h).The % QC in the presence of PhTX also showed a negative correlation with the absolute value of coefficient Aβ (Spearman correlation = −0.25),while QC at baseline in the absence of PhTX was not correlated with coefficient Aβ (Spearman correlation = 0.05, Figure 2h).Given that limited number of genes (nine genes) were used in the analysis, only the correlation between the % EPSP amplitude and the absolute value of transcriptional log fold change is statistically significant (p = 0.02).
Nonetheless, these findings underscore that the extent of PHP impairment (and not baseline synaptic transmission deficits) correlates with the extent of H3K9ac dysregulation and transcriptomic shifts in AD.
In conclusion, by integrating H3K9ac ChIP-seq with snRNAseq datasets from AD patients, we identified genes targeted by H3K9ac that exhibit transcriptional dysregulation across various cell types in AD.Leveraging Drosophila models, we investigated nine glial-specific DEGs and found that each of them is necessary to PHP.Tissue-specific RNAi knockdown studies suggest these genes primarily function in perineurial glia to regulate PHP.Notably, while conventional gene discovery approaches typically yield a hit rate of 3%-4% in screens partially biased towards studying synaptic molecule functions (Dickman & Davis, 2009;Muller et al., 2011), and 1.4% in unbiased screens (Hauswirth et al., 2018) to identify genes essential for PHP, our study found that 100% of the genes we tested are required in regulating PHP.This underscores the robustness of computational analysis-driven gene discovery in understanding the molecular mechanisms of synaptic function.
Our results suggest an important role of PHP in maintaining synaptic stability in AD, and we have identified nine genes connecting AD-related epigenetic and transcriptomic shifts to synaptic instability.The genes we have identified serve as a valuable resource for further investigations into the mechanisms of synaptic destabilization in neurodegenerative diseases.A notable finding was the inverse correlation between the % EPSP amplitude when PhTX is present (normalized to the baseline value) and the magnitude of transcriptional log fold changes in AD (Figure 2h).As the homeostatic compensation deteriorates, the EPSP amplitude diminishes in the presence of PhTX.This indicates that more pronounced transcriptional changes in AD are linked to greater disruptions in PHP.This discovery presents, for the first time, a quantitative association between the level of PHP impairment and the scale of transcriptional abnormalities.Notably, while this correlation does not directly clarify the cell type-specific roles of these genes in PHP, it emphasizes the broader relationship between PHP dysfunction and transcriptional dysregulation.We propose that the functions of these glial genes are evolutionarily conserved: their importance to PHP is directly related to their vulnerability to transcriptional dysregulation in AD.

F
H3K9ac-target Genes Are Differentially Regulated in Various Cell Types in AD.(a) Schematic to show an integrative analysis comparing H3K9ac-target genes (ChIP-seq) and single cell transcriptomic changes (snRNA-seq) in human AD patients versus controls.Criteria for gene selection and the total number of overlapping unique genes are shown.(b) GO terms identified from genes (annotated by chromatin states as in the published study) that are associated with Aβ in the H3K9ac ChIP-seq dataset.(c, d) The total number (c) and percentage of downregulated genes (d) that are both H3K9ac targets and cell type-specific DEGs in excitatory neuron (Ex), inhibitory neuron (In), astrocyte (Ast), oligodendrocyte (Oli), oligodendrocyte progenitor cell (OPC), and microglia (Mic).(e, f) Correlations between coefficient Aβ and log 2 fold change of DEGs in all cell types (e) and oligodendrocytes (f).DEGs that have coefficient Aβ and log 2 fold change with the "same sign" are shown in red.(g, h) The correlation between coefficient Aβ and log 2 fold change of DEGs (g) and percentage of DEGs that have coefficient Aβ and log 2 fold change with the "same sign" (h) in different cell types.*p < 0.05; Pearson correlation for (g) and two-sided binomial test for (h).(i) Transcriptional log 2 fold change (snRNA-seq) and coefficient Aβ (ChIP-seq) for genes that are dysregulated (up and downregulated) and have the "same sign."Cell type-specific DEGs are highlighted with yellow boxes.When a gene is a DEG in multiple cell types, it is assigned to the cell type with the largest absolute value of log 2 fold change.Glial DEGs that have homologues in Drosophila and are selected for PHP characterization are labeled with asterisks.that the downregulation in gene expression within neurons could be a consequence of disease pathology, while the upregulation in gene expression in oligodendrocytes could potentially indicate a compensatory regulatory response linked to AD.These findings underscore the cell type-specific changes of H3K9ac-targeted DEGs in the context of AD.

3.
Both the coefficient Aβ and the gene's transcriptional log fold change display the "same sign" in AD, suggesting H3K9ac and transcriptomic changes are coherently modulated.4. If a gene is a DEG across multiple cell types, it is assigned to the cell type where it has the highest magnitude of transcriptional log fold change (absolute value).5.There is an identified Drosophila homologue for the gene, signifying evolutionary conservation from Drosophila to humans.
and designed the project; Tingting Wang and Stefano Vicini supervised the project; Yimei Cai, Tao Cui, and Paxton Paganelli collected data and performed analyses; Pengqi Yin performed data analyses; and Tingting Wang interpreted the data and wrote the manuscript.