TREM2 fuels the anabolic adaptation required for microglial resilience in Alzheimer’s disease

TREM2 fuels the anabolic adaptation required for microglial resilience in Alzheimer’s disease | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article TREM2 fuels the anabolic adaptation required for microglial resilience in Alzheimer’s disease Jie Gao, Da Lin, Jeffrey Atkinson, Min Chen, Sohan Jayasekara, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8896508/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract The microglial response to Alzheimer’s disease (AD) pathology has canonically been defined by the transcriptional transition to a Disease-Associated Microglia (DAM) state. However, the specific metabolic programs required to fuel the high-demand functions of this reactive state, such as Aβ encapsulation and clearance, remain obscure. Here, we identify a TREM2-dependent anabolic adaptation as the critical driver of microglial resilience to Aβ pathology. By integrating proteomics, transcriptomics, and in vivo metabolic labeling in the AppNL-G-F mouse model, we demonstrate that plaque-associated microglia undergo a synchronized metabolic shift, coupling enhanced protein synthesis with local mitochondrial biogenesis to support the bioenergetic demands of phagocytosis. We show that TREM2 signaling acts as the essential "metabolic licensor" for this process, driving anabolic remodeling directly at the site of phagocytic activity. In the absence of TREM2, this adaptive response collapses: microglia fail to renew their metabolic machinery, resulting in a state of bioenergetic exhaustion characterized by the accumulation of depolarized mitochondria. Strikingly, we discover that these metabolically compromised cells utilize exophergenesis – the extrusion of large, cargo-filled vesicles – as a compensatory mechanism to purge undigested synaptic and amyloid debris during proteostatic failure. Furthermore, we find that these extruded exophers contain hyperphosphorylated Tau, identifying a potential non-cell-autonomous mechanism for pathology seeding. Single-cell analysis confirms that this anabolic capacity is functionally distinct from the canonical DAM transcriptional signature. Our findings redefine TREM2 not merely as a pathogen sensor, but as a metabolic regulator that safeguards microglial viability and prevents neurotoxic spreading under chronic proteotoxic stress. Biological sciences/Neuroscience/Diseases of the nervous system/Alzheimer's disease Biological sciences/Neuroscience/Glial biology/Microglia Biological sciences/Neuroscience/Molecular neuroscience Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Introduction In Alzheimer’s disease (AD), microglia serve as a primary line of defense, tasked with the containment and clearance of neurotoxic Aβ aggregates 1 . Genetic studies have solidly placed microglial function at the center of AD risk, suggesting the necessity of a robust innate immune response to delay pathogenesis 2 3 . However, this protective capacity is not static. As pathology progresses, microglia undergo profound phenotypic changes- initially mounting a containment response, but frequently transitioning toward dysregulated, senescent, or exhausted states that fail to limit neurodegeneration 4 5 . Understanding the mechanisms that allow microglia to maintain functional resilience in the face of chronic proteotoxic stress is a critical priority for therapeutic intervention. Recent advances in single-cell transcriptomics have mapped the trajectory of this response, defining a conserved 'Disease-Associated Microglia' (DAM) or 'Microglial Neurodegenerative Phenotype' (MGnD) signature 6 7 . This transcriptional program, driven largely by the triggering receptor expressed on myeloid cells 2 (TREM2), involves the downregulation of homeostatic checkpoints (e.g., P2ry12, Cx3cr1) and the induction of lipid-sensing and phagocytic pathways (e.g., Apoe, Lpl, Clec7a). The vital importance of TREM2 signaling is underscored by the fact that TREM2 loss-of-function variants (e.g., R47H) triple the risk of AD, and are associated with impaired microglial clustering around plaques,leading to diffuse, neurotoxic amyloid pathology 8 9 10 11 . However, the acquisition of a transcriptional signature is merely a blueprint; executing the functions of plaque-associated microglia, such as proliferation, migration, and continuous phagocytosis, imposes a formidable bioenergetic and biosynthetic demand on the cell. Phagocytes must constantly synthesize new membranes, hydrolytic enzymes, and organelles to replace those consumed during the degradation of Aβ. How microglia acquire the metabolic resources to sustain this high-demand state in the nutrient-deprived or toxic milieu of the AD brain remains poorly understood. While immunometabolism is well-characterized in peripheral macrophages, where activation is coupled to distinct metabolic switches (e.g., glycolysis vs. oxidative phosphorylation) 12 13 , the metabolic dependencies of plaque-associated microglia remain largely unexplored 14 . Specifically, it is unknown whether TREM2 signaling simply instructs the cell what to do, or if it also initiates the anabolic drive required to sustain these functions over time. Here, we combine integrated proteomics, transcriptomics, and in vivo metabolic labeling to define the bioenergetic requirements of microglial resilience. We demonstrate that Aβ pathology triggers a TREM2-dependent "anabolic adaptation", a coordinated surge in protein synthesis and mitochondrial biogenesis at the site of phagocytosis. We show that this anabolic drive is distinct from the canonical DAM transcriptional signature and is essential for maintaining microglial proteostasis. In the absence of TREM2, microglia fail to mount this anabolic response, leading to a state of metabolic exhaustion characterized by mitochondrial stagnation, stalled phagocytic flux, and the extrusion of undigested cargo via neurotoxic exophers. Our findings redefine TREM2 as a metabolic regulator that couples sensing to cellular anabolism, ensuring microglial survival and preventing the secondary seeding of pathology during chronic neurodegeneration. Results Microglia mount a TREM2-dependent anabolic adaptation to Aβ pathology . To gain mechanistic insight into microglial adaptation to amyloid pathology, we performed integrated proteomic and transcriptomic profiling of primary microglia isolated from App NL-G-F and App NL-G-F ; Trem2 KO mice at 3 and 9 months of age, corresponding to early and intermediate stages of amyloid deposition, respectively (Fig. 1A). In App NL-G-F microglia, proteomic profiling identified over 7,000 proteins, of which 2,621 were differentially abundant between 3 and 9 months, indicating extensive proteome remodeling as pathology progressed. Ingenuity Pathway Analysis (IPA) revealed a robust induction of protein synthesis pathways, including Eukaryotic Translation Initiation, Elongation, and Termination, and EIF2 Signaling (Fig. 1B, red arrow). Gene Set Enrichment Analysis (GSEA) of both Cellular Component (CC) and Biological Process (BP) terms confirmed the increased abundance of translation machinery, including cytosolic and mitochondrial ribosomal subunits (Fig. 1C-D, red arrows; Extended Fig. 1A). In parallel, proteins associated with degradative processes, such as ‘autophagosome membrane’, ‘lysosomal membrane’, and ‘phagocytic vesicles’, were also enriched ((Fig. 1C-D, orange arrows). These findings indicate that microglia respond to Aβ deposition by engaging a coordinated anabolic-catabolic program that supports continuous proteome renewal. Transcriptomic profiling corroborated this adaptive response, though to a lesser extent, revealing increased expression of genes associated with protein translation (Fig. 1E, red arrow), ribosomal biogenesis, mitochondrial respiratory chain complexes (Fig. 1F, Extended Fig. 1B, red arrow), and lysosomal/CLEAR pathways (Fig. 1E-F, orange arrows). Together, these data suggest that Aβ pathology induces a broad metabolic remodeling in microglia that couples elevated protein synthesis with enhanced degradative capacity. To determine whether this metabolic remodeling is TREM2-dependent, we compared the profiles of App NL-G-F and App NL-G-F ; Trem2 KO microglia. The impact of TREM2 was tightly linked to disease stage. At 3 months, when Aβ burden is minimal, TREM2 deficiency produced no significant change in the microglial proteome (Extended Fig. 2A). By 9 months, however, loss of TREM2 profoundly disrupted the proteome, resulting in 1,480 significantly altered proteins (Fig. 2A). Strikingly, proteomic and transcriptomic changes showed poor correlation (Fig. 2A-B), underscoring the importance of proteome-level analysis to reveal functional remodeling under Aβ stress. A dominant effect of TREM2 deficiency was collapse of the anabolic response. IPA revealed marked downregulation of translation-related pathways, including SRP-dependent protein targeting, translation elongation, and translation termination (Fig. 2C, blue arrow). GSEA confirmed broad reductions in cytosolic and mitochondrial ribosomal components, as well as ribosome biogenesis (Fig. 2D; Extended Fig. 2B, blue arrows), demonstrating that TREM2 is indispensable for inducing anabolic pathways during Aβ pathology. Transcriptomic profiling reinforced these findings, showing significantly reduced expression of genes involved in protein translation, ribosomal assembly, and mitochondrial respiratory complexes (Fig. 2E-F; Extended Fig. 2C). Consistent with defective proteome renewal, Trem2 KO microglia accumulated synaptic and myelin proteins (Fig. 2C-D, red arrows), materials normally degraded following phagocytosis. Because microglia routinely clear damaged synapses and myelin debris, this accumulation indicates impaired proteostasis when anabolic support is lost. Directional integration of proteomic and transcriptomic datasets highlighted that that while App NL-G-F microglia coordinate protein translation, transport, and degradation, Trem2 KO microglia fail to mount this remodeling response (Extended Fig. 2D-E). Together, these findings suggest that microglia adapt to Aβ pathology by activating a TREM2-dependent anabolic program that drives proteome remodeling and maintains proteostasis. TREM2 drives nascent protein synthesis and phagocytic processing in plaque-associated microglia . To validate the anabolic adaptation revealed by our omics analysis in vivo , we employed Fluorescent Non-Canonical Amino Acid Tagging (FUNCAT) to label newly synthesized proteins (Fig. 3A). Sixteen hours after intraperitoneal injection of azidohomoalanine (AHA), a methionine analog incorporated into nascent proteins, we analyzed CX3CR1 + ; CD45 + microglia by flow cytometry (Extended Fig. 3A). At 3 months of age, FUNCAT intensity was comparable between App NL-G-F and App NL-G-F ; Trem2 KO microglia. By 9 months, however, microglia from App NL-G-F mice exhibited an approximately twofold increase in protein synthesis, whereas TREM2 deficiency completely abolished this induction (Fig. 3B-C). Histograms revealed a distinct “FUNCAT-high” shoulder in App NL-G-F microglia (Fig. 3B, red arrow), indicating the emergence of a metabolically active subpopulation. Because CD45 high microglia are known to accumulate around plaques, we examined their relationship to protein synthesis. While the CD45 high subpopulation expanded with the progression of Aβ pathology (Extended Fig. 3B-C), this expansion was largely absent in Trem2 KO mice. CD45 levels correlated positively with FUNCAT intensity (Fig. 3D), suggesting plaque-associated microglia as the primary population undergoing upregulated nascent protein synthesis. In situ FUNCAT labeling further confirmed a robust induction of protein synthesis specifically within the plaque-associated microglia of App NL-G-F mice, an induction that was largely abolished in App NL-G-F ; Trem2 KO mice (Fig. 3E-F). Previous studies have linked TREM2 signaling to the activation of mTORC1, a key regulator of ribosome biogenesis and anabolic metabolism. Consistent with this mechanism, phosphorylated ribosomal protein S6 (p-RPS6), a canonical downstream marker of mTORC1 activity, was markedly increased in plaque-associated microglia of App NL-G-F mice but not in those of App NL-G-F ; Trem2 KO mice (Fig. 3G-H). Together, these findings suggest that TREM2-dependent activation of mTORC1 drives nascent protein synthesis and sustains anabolic remodeling in plaque-associated microglia. Our proteomic data further suggested that loss of TREM2 compromises microglial proteostasis and phagocytic processing. To test this directly, we examined the intracellular accumulation of phagocytic cargo. Confocal imaging of the postsynaptic marker PSD95, together with Iba1 and Aβ, revealed a substantial buildup of PSD95 puncta within Trem2 KO microglia near plaques (Fig. 4B-C). These inclusions were surrounded by CD68+ lysosomes (Extended Fig. 4A), indicating stalled degradation of engulfed synaptic material. The presynaptic marker vGluT1 similarly accumulated in Trem2 KO microglia (Extended Fig. 4B, C). This defective clearance extended beyond synaptic debris; staining for Myelin Basic Protein (MBP) showed a striking increase in intracellular myelin accumulation in Trem2 KO microglia at plaque sites (Fig. 4D-E). The simultaneous buildup of synaptic and myelin proteins, despite the presence of lysosomal markers, indicates a general failure of phagocytic degradation when TREM2-dependent anabolic support is absent. These results suggest that TREM2-dependent anabolic remodeling is essential for maintaining effective phagocytic processing in plaque-associated microglia. TREM2-dependent anabolic adaptation enables microglial resilience to Aβ-induced phagocytic stress. Because our earlier findings suggested that TREM2-dependent anabolic remodeling supports microglial proteostasis, we next sought to determine how this program shapes transcriptional states under increasing phagocytic stress. To this end, we profiled the transcriptomes of Aβ-engaged microglia using Methoxy-X04 (X04), a fluorescent probe that selectively labels fibrillar Aβ (Fig. 5A). Primary microglia from App NL-G-F and App NL-G-F ; Trem2 KO mice were sorted into X04⁺ (Aβ-phagocytic) and X04⁻ (non-phagocytic) populations. To capture the full spectrum of Aβ load, X04⁺microglia were further stratified into low-, medium-, and high-intensity groups (Fig. 5B). At early stages of pathology (5-6 months), the proportion of X04⁺microglia was comparable between genotypes, indicating that TREM2 is not required for the initial engulfment of Aβ fibrils. By the mid-stage (9-10 months), however, TREM2 deficiency markedly reduced the total X04⁺population. This reduction was driven largely by the specific loss of the highly phagocytic X04 high subset (Fig. 5B-C). Because X04 intensity reflects cumulative Aβ uptake and processing, these findings suggest that TREM2 is essential for sustaining phagocytic function and survival under chronic Aβ burden, rather than for the initiation of phagocytosis. Transcriptomic profiling of X04⁺ and X04⁻ microglia from 9-month-old wild-type, App NL-G-F , and App NL-G-F ; Trem2 KO mice revealed clear genotype-specific clustering by principal-component analysis (PCA) (Fig. 5D). Within each genotype, samples also aligned along the X04-intensity gradient, indicating a shift of transcriptional states proportional to Aβ-induced stress (Fig. 5D; Extended Fig. 5A-B). Differential expression analysis between X04⁺ and X04⁻ microglia in App NL-G-F mice revealed a strong induction of mitochondrial pathways, including Oxidative Phosphorylation, Electron Transport Chain, and Mitochondrial Translation (Fig. 5E, red arrows), as well as protein translational programs, such as SRP-dependent protein targeting, Translation Initiation, and Translation Termination (Fig. 5E, orange arrows). These results indicate that Aβ phagocytosis elicits a coordinated anabolic program that couples protein synthesis with mitochondrial biogenesis to meet the energetic and biosynthetic demands of phagocytic activity. This adaptive transcriptional program was completely abolished in TREM2-deficient microglia (Extended Fig. 5C). Analysis across the X04 subpopulations (X04⁻, low, medium, high) revealed distinct kinetics: anabolic gene induction peaked in X04 low microglia and declined progressively as phagocytic burden increased (Fig. 5F). In contrast, pathways associated with the disease-associated microglia (DAM) signature, such as Neutrophil Degranulation and MHC II Antigen Presentation, increased linearly with X04 intensity. Notably, classical inflammatory pathways (TNF, IFN-γ) were suppressed or unchanged in X04⁺microglia, demonstrating that anabolic activation represents a metabolically adaptive, non-inflammatory response. Loss of TREM2 largely abolished this anabolic response in X04 low and X04 medium microglia (Fig. 5G). As Aβ load increased, Trem2 KO microglia displayed reduced oxidative phosphorylation and mitochondrial translation, accompanied by increased signatures of mitochondrial dysfunction, a compensatory upregulation of glycolysis, and heightened pro-inflammatory signaling (Fig. 5G). Thus, across the X04 gradient, TREM2 loss had minimal effects on non-phagocytic microglia but was indispensable for inducing anabolic adaptation and mitochondrial biogenesis in phagocytic populations (Fig. 5H). Together, these results demonstrate that microglial adaptation to Aβ-induced phagocytic stress depends on a TREM2-driven anabolic program that integrates protein synthesis and mitochondrial biogenesis. In the absence of TREM2, this adaptive network collapses, leading to metabolic failure and the loss of microglial resilience under chronic Aβ stress. TREM2 deficiency impairs mitochondrial renewal, driving proteostatic failure and the extrusion of neurotoxic exophers To determine whether the anabolic defect caused by TREM2 deficiency translates to a functional breakdown of bioenergetics in vivo , we assessed mitochondrial dynamics using high-dimensional flow cytometry. We reasoned that if TREM2-deficient microglia fail to mount the anabolic program required for mitochondrial biogenesis, they would be unable to maintain a healthy organelle pool under the phagocytic stress of amyloid pathology. Freshly isolated microglia were stained with MitoTracker Green (measuring total mitochondrial mass) and MitoTracker Red CMXRos (measuring membrane potential). In App NL-G-F mice, the CD45 high (plaque-associated) population exhibited a synchronized elevation of both mitochondrial membrane potential and lysosomal mass, suggesting that resilient microglia boost their bioenergetic capacity to meet the high metabolic demands of Aβ clearance (Extended Fig. 6A, B). However, Trem2 KO microglia displayed a profound bioenergetic uncoupling: while total mitochondrial mass was significantly increased compared to controls, these organelles exhibited significantly reduced membrane potential. Consequently, the Mitochondrial Health Index (calculated as the ratio of respiratory-active mitochondria to total mass) was markedly lower in Trem2 KO microglia (Fig. 6A), indicating a severe accumulation of metabolically incompetent organelles. Notably, lysosomal mass (LysoTracker) remained unchanged between genotypes (Fig. 6B), suggesting that the accumulation of depolarized mitochondria in Trem2 KO microglia unlikely stems from lysosomal deficiency. We hypothesized that this defect arose from a failure in mitochondrial renewal, particularly in the distal processes where phagocytosis actively occurs. To definitively test for impaired biogenesis, we combined metabolic labeling with a proximity ligation assay (PLA). Mice were injected with the amino acid analog AHA (L-azidohomoalanine) to tag nascent proteins, followed by click-mediated biotinylation and detection of newly synthesized TOM20 using an anti-biotin/anti-TOM20 PLA antibody pair. In App NL-G-F mice, plaque-associated microglia displayed a significant increase in nascent mitochondrial signal (TOM20-PLA + ) specifically within IBA1 + regions near plaques (Fig. 6C, D), indicating robust biogenesis in response to amyloid deposition. In contrast, Trem2 KO microglia failed to increase (TOM20-PLA + ) signals, demonstrating that TREM2 is required to boost mitochondrial biogenesis. Furthermore, co-staining of FUNCAT and TOM20 revealed colocalization of nascent protein signal (FUNCAT) with mitochondria (TOM20) within microglial processes (Fig. 6E, Extended Fig. 6C). This spatial pattern correlated with the induction of TREM2 expression in the processes of plaque-associated microglia (Extended Fig. 6D), suggesting that TREM2 drives local protein synthesis and mitochondrial biogenesis directly at the site of phagocytic activity. In contrast, Trem2 KO microglia around the plaques failed to recruit mitochondria to their processes (Fig. 6E). Despite the reduction in new biogenesis, quantification of confocal images revealed that Trem2 KO microglia around the plaques exhibited increased total mitochondrial mass (% of TOM20+ voxel in Iba1+ area) (Fig. 6F, G), corroborating the flow cytometry data. Upon further image analysis of brain sections from App NL-G-F ; Trem2 KO mice, we observed a striking phenomenon: the presence of numerous large (2-8 µm), spherical Iba1 + /DAPI - vesicles within the plaque niche that were filled with mitochondria (TOM20+) (Fig. 6F). This morphology - large, membrane-bound vesicles packed with mitochondria - bore a strong resemblance to exophers, a specialized extrusion mechanism previously reported in neurons and cardiomyocytes to eject damaged mitochondria and protein aggregates under proteostatic stress. Staining for the outer mitochondrial membrane protein VDAC1 confirmed that these vesicles were densely packed with mitochondria (Extended Fig. 6E, F). Crucially, co-staining with the autophagy receptor p62/SQSTM1 revealed that the VDAC1 in these exophers displayed high levels of p62 (Extended Fig. 6G), indicating that these vesicles are selectively enriched with damaged mitochondria that failed to undergo mitophagy. Prompted by this resemblance to exophaer, we performed high-resolution 3D reconstruction from multiple confocal image datasets. We found that these vesicles mostly emerged directly from microglial processes, appearing as segmental enlargements connected by thin membrane nanotubes (Fig. 7A, D; Extended Fig. 7A-D). To rigorously rule out that these structures are derived from apoptotic fragmentation, we stained for apoptotic marker cleaved caspase-3 (CC3). We validated our CC3 antibody using postnatal day 7 (P7) mouse brains, which undergo significant programmed cell death. While P7 neurons showed clear CC3 signal (Extended Fig. 7E), these exopher-like vehicles and adjacent microglia in App NL-G-F ; Trem2 KO brains were consistently CC3-negative (Extended Fig. 7F), confirming they represent a genuine non-apoptotic extrusion event. In addition to damaged mitochondria (VDAC1 + ;p62 high ), the exophers were packed with Aβ, undigested synaptic (PSD95 + ), and myelin (MBP + ) debris (Fig. 7B, E), consistent with their function as emergency "trash bags" for material the cell could not degrade. Quantitative analysis revealed a striking disparity in the frequency of these extrusion events. Within a standardized plaque-associated volume (10 5 um 3 ), we detected an average of 7-8 PSD95 + or MBP + exophers in AKTK mice, compared to less than 0.5 in AK controls (Fig. 7C, F). When normalized to the number of plaque-associated microglia, we found that 60-70% of Trem2 KO microglia in the plaque niche were associated with exopher extrusion, compared to less than 2% of WT microglia. This suggests that exophergenesis is not a rare anomaly, but a dominant phenotype of metabolic failure in the absence of TREM2. Importantly, a subset of these exophers contained hyperphosphorylated Tau (AT8 + ) (Fig. 7G, H). This finding suggests that exopher extrusion may effectively package pathological Tau into concentrated, extracellular seeds, facilitating the conversion and spreading of Tau pathology by releasing aggregation-prone material back into the parenchyma. Supporting this notion, we observed a marked increase in AT8 + area around Aβ plaques in the cortex of AKTK brains compared to controls, spatially correlating with regions of active exopher extrusion (Fig. 7I, J). Together, these data demonstrate that TREM2-dependent mitochondrial renewal is essential for microglial proteostasis and resilience against Aβ, and its failure leads to the extrusion of neurotoxic exophers that may actively drive Tau pathology. The Anabolic Adaptation signature defines phagocytic competence better than the canonical DAM signature . To place anabolic adaptation within the broader landscape of microglial activation, we performed single-cell RNA sequencing on X04⁺ (phagocytic) and X04⁻ microglia (non-phagocytic) microglia isolated from 9-10 month-old App NL-G-F mice. Clustering analysis identified the expected spectrum of transcriptional state (Sala Frigerio et al., 2019; Mancuso et al., 2024), including Homeostatic (HM), Activated Response (ARM), and Transition Response (TRM) microglia (Extended Fig. 8A-B). However, projecting the canonical Disease-Associated Microglia (DAM) signature onto this map revealed a critical discrepancy: DAM genes were expressed broadly across both X04⁺and X04⁻populations (Extended Fig. 8C) 15 . This suggests that the acquisition of the transcriptional DAM signature is insufficient to distinguish microglia that are functionally engaged in phagocytosis from those that are merely responsive to pathological environment. To resolve the specific molecular determinants of phagocytic competence, we interrogated the spatial heterogeneity within these clusters. UMAP visualization revealed that X04⁺and X04⁻ microglia occupied distinct regions within the nominally defined HM and TRM clusters (Fig. 8A). Increasing the clustering resolution resolved this heterogeneity, splitting HM into subclusters 0 (Non-phagocytic enriched) and 2 (Phagocytic enriched), and TRM into subclusters 6 (Non-phagocytic) and 1 (Phagocytic) (Fig. 8B-C). We then explored which transcriptional programs effectively discriminated these functional states. "DAM Module Scores" were similar between phagocytic (Cluster 1, 2) and non-phagocytic (Cluster 6, 0) pairs (Extended Fig. 8D), confirming that the DAM signature correlates poorly with phagocytic states. In contrast, differential expression analysis between Cluster 1, 2 and Cluster 6, 0 revealed that the phagocytic state was defined by a robust, synchronized upregulation of genes involved in ribosome biogenesis, protein synthesis, and mitochondrial oxidative phosphorylation (Fig. 8D, Extended Fig. 8E). GSEA confirmed the enrichment of these pathways in the cellular components (Fig. 8E; Extended Fig. 8F), which mirrored the proteomic "Anabolic Adaptation" signature identified earlier. This program was featured by the specific induction of rate-limiting translational initiators ( Eif3f ) and structural ribosomal subunits ( Rps5 , Rpl23 ), coupled with respiratory chain subunits spanning the input ( Ndufa1 ), catalytic core ( Cox4i1 ), and ATP-generating output ( Atp5e ) of the electron transport chain (Fig. 8F). Crucially, this anabolic induction appeared distinct from inflammatory activation. We observed that neuroinflammatory signaling pathways were reduced in the phagocytic clusters (1 and 2) compared to their non-phagocytic counterparts (6 and 0) (Fig. 8D, Extended Fig. 8E), supporting the view that anabolic adaptation represents a functional metabolic engagement rather than a generic inflammatory state. Finally, we spatially validated the relationship between the DAM signature and anabolic activities in situ . We co-stained brain sections from 9-month-old App NL-G-F mice for Iba1, the canonical DAM marker CD74, and the nascent protein label FUNCAT. While most plaque-associated (Iba1 high ) microglia exhibited robust anabolic activity (high FUNCAT signal), CD74 expression was only induced in a subset of these microglia. In addition, CD74 intensity showed no correlation with FUNCAT signal within Iba1 + microglia (Pearson's R = -0.12, P=0.187; Extended Fig. 8G, H). Collectively, these data demonstrate that the anabolic adaptation signature better define the phagocytic microglial state, distinguishing the functional capacity to engage with stress from the broader transcriptional response to Aβ stress. Discussion Microglial activation in Alzheimer’s disease has canonically been defined by the transition to a Disease-Associated Microglia (DAM) state. Because this transcriptional signature includes the upregulation of phagocytic receptors (e.g., Clec7a, Axl, MerTK), the field has largely operated under the assumption that the DAM phenotype is synonymous with enhanced phagocytic function. Our findings challenge this assumption. We demonstrate that the DAM signature and phagocytic competence are not inextricably linked. Instead, we identify a distinct TREM2-dependent anabolic adaptation, characterized by enhanced protein synthesis and mitochondrial biogenesis, as the true driver of functional resilience. Our data suggest a fundamental reinterpretation of microglial states: the DAM signature likely represents a cellular stress response to pathology, whereas the anabolic program represents the functional capacity to resolve it. While TREM2 deficiency is known to impair the full acquisition of the DAM signature, we show that the functional impotence of these cells is not merely due to a lack of ability to phagocytosis. Rather, it stems from a bioenergetic collapse: without TREM2, microglia fail to mount the 'metabolic engine' required to sustain activity. This distinction explains why purely transcriptional definitions of microglial states often fail to predict functional outcomes, as the stress response (DAM) and the capacity to act (Anabolism) are molecularly distinct programs. A central finding of our study is that the containment of Aβ pathology (catabolism) strictly requires an accompanying anabolic surge. In peripheral macrophage biology, metabolic states are often dichotomized: pro-inflammatory (M1-like) activation is typically driven by aerobic glycolysis (the Warburg effect), while anti-inflammatory (M2-like) states rely on oxidative phosphorylation (OXPHOS) 16 13 . Our data challenge this binary framework in the context of AD. We reveal that plaque-associated microglia adopt a unique "hyper-metabolic" phenotype that simultaneously engages both biomass synthesis (anabolism) and mitochondrial respiration. This state mirrors the "effector expansion" observed in T-cell biology, where rapid proliferation and cytokine production demand a massive, mTOR-driven upregulation of translational capacity and mitochondrial mass 17 18 . For microglia, the functional burden is proteostatic: the continuous internalization of Aβ fibrils and cellular debris consumes lysosomal enzymes, membranes, and ATP at a rate that homeostatic synthesis cannot match. We propose that the extensive proteome remodeling observed in App NL-G-F microglia, specifically the upregulation of ribosomal and mitochondrial biogenesis, represents a compensatory maneuver to maintain "proteostatic bandwidth." The cell must physically construct new degradative machinery to replace what is expended. Consequently, the accumulation of synaptic and myelin debris we observed in Trem2 KO microglia is likely not a primary defect in recognition or uptake, but a failure of renewal. Without the TREM2-driven anabolic engine to replenish mitochondria, the cells simply exhaust the machinery required to digest their cargo. Our results identify a failure of mitochondrial renewal as the primary driver of microglial dysfunction in the absence of TREM2. While seminal studies have characterized the TREM2-deficient state as a general "metabolic collapse" or loss of energetic fitness 14 , our single-cell metabolic profiling provides a more granular mechanism: the decoupling of organelle mass from function. We observe that Trem2 KO microglia exhibit a paradoxical "High Mass / Low Potential" phenotype. We propose that this stems fundamentally from an inability to renew the mitochondrial network, forcing the cell into a state of metabolic stagnation. Under chronic phagocytic stress, microglia must maintain a high rate of organelle turnover, constantly synthesizing new mitochondria to replace those damaged by ROS or consumed during lysosomal fusion. This mirrors the metabolic reprogramming seen in activated T cells, where mitochondrial biogenesis is a prerequisite for sustained effector function 17 . Our in vivo AHA-PLA data confirm that TREM2 is the driver of this essential biogenic surge. In the absence of this anabolic signal, the cycle of renewal arrests. We posit that without the capacity to synthesize replacements, Trem2 KO microglia are forced to retain aging, damaged mitochondria to maintain basal viability. Because the cell cannot "afford" to degrade its existing power sources (mitophagy) if it cannot replace them (biogenesis), these "damaged" mitochondria accumulate, leading to the observed increase in total mass. However, because they are depolarized and structurally compromised, they offer diminishing bioenergetic returns and likely contribute to oxidative stress 19 . Thus, the metabolic defect in Trem2 KO microglia is not merely a failure to clear waste (catabolism), but a failure to build the machinery (anabolism) that allows clearance to happen. Perhaps the most striking finding of our study is the identification of exopher-like structures emerging from metabolically compromised microglia. Originally characterized in C. elegans neurons and recently described in mammalian cardiomyocytes, exophers represent a conserved, primordial mechanism for ejecting aggregation-prone proteins and damaged organelles when intracellular degradation pathways are overwhelmed 20 21 . To our knowledge, our study provides one of the first lines of evidence for this phenomenon in CNS microglia in situ. We propose that exophergenesis acts as a 'proteostatic emergency valve' for TREM2-deficient microglia. Our data suggest that when the anabolic support required for lysosomal digestion fails, the cell is left with a fatal accumulation of undigested myelin, amyloid, and 'damaged' mitochondria. Unable to degrade this toxic burden internally, the cell physically ejects it to preserve its own viability. This aligns with findings in other phagocytic contexts where cells jettison indigestible cargo to avoid programmed cell death 22 . However, this survival maneuver likely comes at a high cost to the neural environment. By sequestering proteopathic seeds but failing to digest them, these cells effectively "re-package" aggregates into a mobile, bioactive form. We propose that the "metabolic uncoupling" observed in Trem2 variants transforms microglia from guardians into "Trojan horses," absorbing toxic seeds only to regurgitate them into the parenchyma. This mechanism offers a novel, active explanation for the 'diffuse plaque' and accelerated tau pathology consistently observed in TREM2-deficient AD patients 23 24 . This exopher-mediated expulsion may represent a novel, non-neuronal mechanism for the prion-like spreading of Aβ and Tau pathology, particularly in the context of aging or genetic risk where microglial metabolism is waned. Determining whether these microglial exophers contribute to the trans-synaptic seeding of tau pathology or the physical disruption of neural circuits in human AD tissue represents a critical frontier for future investigation. Our single-cell RNA seq analysis on phagocytic (X04+) and non-phagocytic (X04-) also challenges the prevailing dogma that the DAM transcriptional signature is synonymous with functional microglial competence. Since the initial characterization of the DAM/MGnD state 6 7 , the upregulation of sensing receptors (e.g., Clec7a, Axl, MerTK) has been widely interpreted as a proxy for enhanced phagocytic activity. Our data reveal a critical distinction between these two programs. We observed that the DAM module score correlates linearly with phagocytic load (X04 intensity) even in cells that are metabolically failing. This suggests that the DAM signature functions primarily as a cellular stress response, a transcriptional "cry for help" triggered by proteotoxic burden, rather than a guarantee of functional execution. In the absence of TREM2-driven anabolism, microglia can mount this transcriptional sensing response (demand) but fail to back it up with the biosynthetic machinery (supply) required to act. Finally, our findings offer critical insights into the optimization of current anti-amyloid immunotherapies. The recent clinical success of monoclonal antibodies against Aβ (e.g., lecanemab, donanemab) has validated amyloid clearance as a primary therapeutic target 25 26 . However, the efficacy of these therapies relies heavily on the underlying functional capacity of microglia to engage in Fc-receptor-mediated phagocytosis, a process we now identify as imposing a formidable bioenergetic cost on the cell. Our data suggest a potential vulnerability in this approach: pharmacologically driving microglia toward a hyper-phagocytic phenotype without ensuring adequate anabolic support may be counterproductive. If the 'demand' for clearance (induced by antibody tagging) exceeds the cell's 'supply' of metabolic machinery (anabolic capacity), this metabolic mismatch could accelerate exhaustion and drive the extrusion of neurotoxic exophers. We propose that future therapeutic regimens should adopt a combinatorial approach. Interventions that specifically bolster the anabolic axis, for example, via TREM2 agonists, mitochondrial co-factors, or amino acid supplementation, which could provide the necessary fuel to sustain the immunotherapeutic response. With this combinatorial approach, we may be able to extend the therapeutic window during which microglia can effectively clear Aβ without succumbing to proteostatic collapse. Methods Animals App NL-G-F mice mice were kindly provided by Dr. Takaomi Saido (RIKEN Center for Brain Science) 27 . Trem2 knockout mice (Trem2 KO ) were obtained from The Jackson Laboratory (Strain #027197). To generate the double-mutant cohort, App NL-GF mice were crossed with Trem2 KO mice to produce generate App NL-G-F ; Trem2 KO homozygous for both manipulations. Mice were housed in a pathogen-free barrier facility with a standard 12-hour light/dark cycle and ad libitum access to food and water. Housing density was maintained at maximum 5 mice per cage. Both male and female mice were used for all experiments. All animal procedures were conducted in strict accordance with the National Institutes of Health (NIH) guidelines and approved by The Ohio State University Institutional Animal Care and Use Committee (IACUC). In Vivo Metabolic Labeling (FUNCAT) For labeling of nascent proteins, mice were injected intraperitoneally (i.p.) with 50 mg/kg Azidohomoalanine (AHA; Click Chemistry Tools) dissolved in sterile PBS. Animals were sacrificed 16 hours post-injection. For Flow Cytometry: Dissociated cells were reacted with Alexa Fluor 647-Alkyne (Thermo Fisher Scientific) using the Click-iT Cell Reaction Buffer Kit (Thermo Fisher Scientific, Cat #C10269) according to the manufacturer’s instructions. For Proximity Ligation Assay (PLA): Free-floating brain sections were reacted with Biotin-Alkyne (Thermo Fisher Scientific) using the Click-iT Cell Reaction Buffer Kit prior to antibody incubation. Microglial Isolation and Sorting Adult microglia were isolated using magnetic-activated cell sorting (MACS) as described previously 28 . Briefly, mice were anesthetized and perfused transcardially with ice-cold PBS to remove circulating leukocytes. Brains were dissected, chilled on ice, and dissociated using the Neural Tissue Dissociation Kit (P) (Miltenyi Biotec, #130-107-677) and the gentleMACS Tissue Dissociator. Cell suspensions were filtered through a 70 µm cell strainer and centrifuged at 300 × g for 10 minutes. Myelin was depleted using Myelin Removal Beads II (Miltenyi Biotec, #130-096-733) via magnetic separation. The resulting single-cell suspension was used immediately for flow cytometry or further enriched for transcriptomic analysis using CD11b MicroBeads (Miltenyi Biotec, #130-049-601) and LS columns. Evaluation of Phagocytic Microglia To evaluate amyloid load, Methoxy-X04 (X04; Tocris Bioscience, #4920) was prepared by dissolving the compound in DMSO, followed by dilution in a 1:1 mixture of propylene glycol and PBS to obtain a stable yellowish-green emulsion. The solution was prepared freshly and injected i.p. at 10 mg/kg 16 hours prior to tissue harvest. For flow cytometric analysis, dissociated cells were incubated with Fixable Viability Dye eFluor 780 (1:1000, eBioscience) and anti-CD16/32 (Fc Block, 1:200, clone 2.4G2) to exclude dead cells and prevent non-specific binding. Cells were subsequently stained with fluorophore-conjugated antibodies against CD11b (1:500, clone M1/70, Invitrogen), CD45 (1:100, clone 30-F11, Invitrogen), and CX3CR1 (1:50, R&D Systems). Live microglia (CD45 int ; CD11b + ; CX3CR1 + ) were gated as X04 + or X04 - based on fluorescence in the DAPI excitation channel (405 nm). Wild-type animals injected with Methoxy-X04 served as biological negative controls to define gating thresholds. In Situ Proximity Ligation Assay (PLA) : To visualize nascent mitochondrial proteins in situ, free-floating brain sections from AHA-injected mice were first subjected to the click reaction with Biotin-Alkyne as described above. Sections were extensively washed and incubated overnight at 4 °C with rabbit anti-TOM20 (1:500, 11802-1-AP, Proteintech) and mouse anti-Biotin (1:500, 1D4-C5, BioLegend). PLA probes (Duolink In Situ, Sigma-Aldrich) were applied, and ligation and amplification steps were performed according to the manufacturer's protocol. Bulk RNA-seq and Bioinformatics Analysis Library Preparation and Sequencing Total RNA from primary microglia was extracted using Quick-RNA miniprep (R1055, Zymo Research). RNA quality was evaluated by TapeStation using high Sensitivity RNA ScreenTape (5067-5579, Agilent). RNA samples with RNA integrity numbers greater than 8 were used for cDNA library construction. RNA seq libraries were prepared using SMART Seq® mRNA LP Kit (Takara Bio) following the manufacturer’s instructions. The qualities of the cDNA library were assessed using TapeStation using High Sensitivity D5000 ScreenTape (5067-5592, Agilent). cDNA library samples were then pooled and sequenced with the HiSeq 4000 System (Illumina) by AZENTA life sciences. Raw Reads Preprocessing and Sequence Alignment Demultiplexed FASTQ files of bulk RNA sequencing data were aligned to the mouse genome (Mus_musculus.GRCm39) using STAR (version 2.7.10a) 29 30 31 . Adapters were trimmed using Flexbar (version 3.5.0.) 32 . Reads mapped to genomic features were counted using featureCounts (version 2.0.3) 33 . The count matrix was imported in R (version 4.3.3) for analysis. Differential Expression and Principal Component Analysis (PCA) Differential gene expression analysis was performed using the DESeq2 package (version 1.42.1) 34 in R. Raw count data were filtered to remove genes with low library representation (total counts < 10). For basic comparisons, a single-factor generalized linear model was used. In experiments involving Batch/Sex/Time, a multi-factor model was implemented to control for confounding variables. DESeq2’s internal median-of-ratios method was used for normalization. To control the false discovery rate (FDR), p-values were adjusted using the Benjamini-Hochberg procedure. Genes were considered significantly differentially expressed if they reached an adjusted p-value < 0.05. Differential expression results were visualized using volcano plots generated by the EnhancedVolcano package (version 1.20.0) 35 , with significance defined as an adjusted p-value < 0.05. PCA was performed on the transformed counts extracted from DESeq2. If a multi-factor design was used for DEA to measure the effect of the genotypes controlling for batch differences, the PCA was plotted with batch variation removed by using the removeBatchEffect() function from limma (version 3.54.2) 36 . Functional Enrichment and Pathway Analysis IPA Core Analysis Core Canonical Pathway Analysis was performed by QIAGEN’s Ingenuity® (IPA®, QIAGEN Redwood City, www.qiagen.com/ingenuity ). Complete lists of DEGs and DAPs, along with their log2 fold change expression values and FDR were inputted into IPA for identifying canonical pathways, biological functions, and upstream regulators using a cutoff of FDR < 0.05. The p-value of overlap, calculated using the right-tailed Fischer’s Exact Test with a statistical threshold of 0.05, is used to indicate the probability of association of molecules from test dataset with the canonical pathway by random chance alone. A positive or negative regulation z-score value indicates that a function is predicted to be activated or inhibited. No activity prediction by IPA results in ineligible z-score which is represented by grey bars. IPA Comparison Analysis To identify biological pathways modulated across the different XO4 subgroups of microglia within specific genotype, differentially expressed genes (DEGs) from the Low, Medium, and High groups from each genotype--each compared against a common baseline control of XO4- dataset--were first processed through individual Core Analyses in IPA. Subsequently, the results of contrast within each genotype were integrated using the IPA Comparison Analysis platform to juxtapose the canonical pathway profiles across all groups of the specific genotype. Findings were visualized as a heatmap where color intensity represents the activation z-score (grey indicates an unpredictable directional trend). To denote statistical confidence, pathways failing to reach the significance threshold (e.g., Fisher’s exact test p-value > 0.05) were marked with an asterisk (*). Pathways were further grouped and annotated by their broader functional categories to identify overarching biological themes. Gene Set Enrichment Analysis (GSEA) Using the R package clusterProfiler (version 4.11.0.2 37 , genes or proteins were ranked by values of log2 Fold Change and −log10(p-value) ∗ sign(log2FoldChange) respectively to form the ranking metrics for transcriptomic and proteomic datasets. Enrichment was conducted against the Gene Ontology (GO) database, specifically targeting Biological Process (BP) and Cellular Component (CC) categories. Results were considered statistically significant at an adjusted p-value < 0.05. Results were visualized using dot plots, where selected GO terms were segregated by their direction of regulation (activated vs. suppressed) based on the Normalized Enrichment Score (NES). The color scale represents the Benjamini-Hochberg adjusted p-value (q-value) to indicate statistical significance, while the size of each dot corresponds to the gene count (the number of genes from the dataset coregulated within a specific GO term). Gene-concept Network To reveal the molecule-level information associated with the significant pathways of interest, we constructed gene-concept networks using the cnetplot() function within the clusterProfiler package in R. Quantitative Proteomic Analysis: Brain samples were lysed using 5% SDS and 50mM TEAB buffer, sonicated, cleaned, and quantified via BCA assay. Proteomic analysis was conducted by BGI Genomics Co., Ltd. Briefly, samples were then prepared using the STrap Midi MS sample prep device (Protifi), with each sample containing 2000μg of protein. This preparation involved reduction with dithiothreitol (DTT), alkylation with iodoacetamide (IAM), quenching of the IAM reaction with DTT, and overnight digestion with Trypsin/Lys-C within the STrap device. Peptides were eluted from the STrap, with 60μg per sample dried via SpeedVac and reconstituted in 50% acetonitrile for tandem mass tag (TMT) labeling in 50mM TEAB (pH 8.5). After TMT labeling, samples were pooled, acidified with 1% formic acid, and analyzed for label check on a nano LC-MS/MS system. Following successful label checks, samples were dried, reconstituted in 2% formic acid, desalted using EVOLUTE® EXPRESS ABN (Biotage), and fractionated via offline HPLC into 12 fractions. These fractions were analyzed by LC-MS/MS after being reconstituted with mobile phase A; approximately 5% of each was injected using the TMT method. TMT quantification and identity discovery were performed using Proteome Discoverer 2.5 (Thermo Fisher). False discovery rate (FDR) was calculated based on The Benjamini-Hochberg Procedure. FDR <= 0.05 was considered to be significant. Histological analysis : Brains were sectioned on a cryostat at 40-mm thickness. For immunofluorescence staining, free-floating sections were blocked with PBS containing 10% normal goat serum (NGS) at room temperature for 30 minutes, incubated with primary antibody in blocking solution (PBS with 1% NGS) at 4°C for 24-48 hours, and then incubated with secondary antibody at room temperature for 2 hours. Sections were mounted on slides with ProLong Diamond (Life Technologies). Images were captured on a ZEISS Axio Observer and/or the Nikon AXR point scanning confocal microscope. 2D Image quantification was performed using ImageJ software. Auto Threshold methods “Otsu” or “Triangle” were used to define the region of interest (ROI). Statistical analyses were conducted using a two-tailed unpaired t-test or one-way ANOVA. Primary antibodies used in this study are: Human Amyloid β (N) (82E1, IBL Co., LTD.), Iba1 (019-19741, or 011-27991 from Wako Co), Phos-RPS6 (4858S, Cell Signaling), , PSD95 sdAb (N3702-AF568-L, Synaptic Systems), VGLUT1 sdAb (N1602-At488-L, Synaptic Systems), TOM20 (11802-1-AP, Proteintech), VDAC1 (CL488-10866, Proteintech), MBP (78896, Cell Signaling), Cleaved Caspase-3 (Asp175) (9661, Cell Signaling), CD74 (151002, Biolegend). All secondary antibodies were purchased from ThermoFisher or Jackson Immunoresearch. Image Acquisition and Statistical Analysis 2D Epifluorescence Microscopy and Analysis Two-dimensional epifluorescence imaging was performed using a ZEISS Axio Observer. This modality was employed for analyses where axial depth was not a primary variable. Image quantification was conducted using ImageJ software (NIH) 38 , focusing on aggregate metrics per field of view. To ensure unbiased measurement, regions of interest (ROIs) were defined using automated thresholding algorithms ('Otsu' or 'Triangle', depending on signal-to-noise characteristics). Statistical comparisons between experimental groups were performed using either a two-tailed unpaired t-test or one-way ANOVA, as appropriate, based on the number of groups and data distribution. 3D Confocal Imaging and Hierarchical Statistical Modeling High-resolution three-dimensional image stacks were acquired using the Nikon AXR point scanning confocal microscope to quantify the morphological and biochemical properties of Iba1-positive (Iba1+) regions of interest (ROIs). Due to the high density and clustering of microglia in the App NL-G-F mice, individual ROIs were defined as distinct morphological units rather than individual cells. All 3D reconstructions and quantitative analyses were performed using Imaris (v11.0; Oxford Instruments, Zurich, Switzerland), a multidimensional image analysis software. Iba1+ surfaces were generated using a machine learning-based thresholding algorithm to ensure objective and consistent segmentation of complex microglial morphologies. To quantify the volumetric colocalization of target markers within Iba1+ ROIs, marker surfaces were generated via manually defined thresholds, allowing for the correction of staining-specific background noise. Volumes were converted to voxel counts prior to calculating the overlapping voxel ratio and performing model fitting. To assess protein expression levels, the mean fluorescence intensity of the specific target marker channel was extracted directly from the reconstructed Iba1+ surface. To assess the impact of AD pathology, Iba1+ ROIs were spatially categorized based on their interaction with amyloid pathology. An ROI was classified as Plaque-Associated Microglia (PAM) if any portion of the Iba1+ surface, including processes or the soma, was in direct contact with a plaque cluster. All other ROIs were designated as Non-Plaque-Associated Microglia (NPAM). To account for the hierarchical structure of the 3D data (multiple ROIs nested within individual imaging fields), we employed a Generalized Linear Mixed Model (GLMM) framework, allowing us to treat the animal or image stack as a random effect, thereby controlling for intra-subject variation and ensuring a more accurate estimation of the fixed effects of the genotypes, Plaque Status (PAM vs. NPAM), and their interaction term. All statistical analyses were conducted in R using the glmmTMB 39 and emmeans 40 packages. The choice of statistical model was tailored to the distribution and mathematical constraints of each quantified metric. For the volumetric occupancy of target markers within Iba1+ ROIs, a beta-binomial GLMM (logit link) was selected to account for the bounded nature of proportional data and to adjust for overdispersion derived from biological variability. Continuous measurements, such as mean intensities, were modeled using a Gamma distribution (log link). Model fit was rigorously assessed using the DHARMa package 41 , utilizing a simulation-based approach to verify distribution assumptions, dispersion, and residual patterns. Estimated Marginal Means (EMMs) and 95% confidence intervals (CIs) were back-transformed from the link scale to the response scale for reporting. Pairwise comparisons were performed as planned contrasts between genotypes and plaque conditions. Exopher Density Quantification and Non-Parametric Analysis Following 3D confocal acquisition as described above, we performed a targeted quantification of microglial exophers. Exopher density was quantified via a dual-blind system: Aβ plaque areas were manually delineated by an analyst blinded to the channels of exopher markers (PSD95, MBP, and AT8), and exophers were subsequently counted within these regions by a second independent analyst. Exopher densities were calculated as the number of exophers per 10 5 μm 3 of plaque volume to ensure human-readable scaling. Due to the distinct biological nature of the groups—where the control (AK) genotype exhibited "structural zeros" (near-total absence of symptoms) and the treatment (AKTK) genotype showed a robust, high-variance phenotype—standard count-based Generalized Linear Mixed Models (GLMMs) failed to reach mathematical convergence. To address the complete separation and non-normal distribution of the data, a non-parametric Wilcoxon rank-sum test with exact permutation and rank transformation using the R package coin (version 1.4.3) 42 was employed to account for the high frequency of tied zeros in the control group. Exopher density distributions were visualized in violin plots, with jittered points indicating individual plaque ROIs. Medians and interquartile ranges (IQRs) are indicated by dots and bars, respectively. Statistical significance is reported as p values based on the exact permutation test. Single-cell Seq and Data Processing Library Construction Single cell RNA-seq libraries were generated using the 10x Genomics Chromium NEXT GEM Single Cell 3’ Reagent Kit. Briefly, primary microglia isolated from adult mice were loaded onto chromium chips with a capture target of 10,000 cells per sample. Libraries were prepared following the provided protocol and sequenced on an Illumina NovaSeq with a targeted sequencing depth of 50,000-100,000 reads per cell. FASTQ files from sequencing were then used as inputs to the 10X Genomics Cell Ranger pipeline. Read Processing, Quality Control and Filtering Gene expression matrices were generated with the Cell Ranger Pipeline (v7.0.0; 10x Genomics) and aligned to the Mouse (mm10) reference transcriptome. The resulting digital gene expression matrix was filtered, normalized, and clustered using R version 4.2.0 and Seurat version 4.1.1 43 . Genes that are expressed in less than 10 cells, and cells with greater than 5% of reads mapped to mitochondrial genes, or with less than 1500 features and 3000 UMIs were removed. Initial Normalization and Doublet Removal We performed an initial normalization of post-QC dataset in Seurat to stabilize variance and did not regress out variation associated with percent.mito or percent.rb due to the metabolic relevance mitochondrial and ribosomal gene expression features, despite observed differences between these fractions. DoubletFinder version 2.0.3 44 was used to identify false-negative Demuxlet classifications caused by doublets formed from cells with identical SNP profiles, and an average of 10% of cells per sample were confidently predicted as doublets and removed. In total, 26,096 cells were identified as putative singlets and retained for downstream analysis. SCTransform Normalization, Integration, and Clustering The gene expression matrix was normalized and scaled using the Seurat function SCTransform which also identifies the most variable genes, of which the top 3,000 were used for dimensionality reduction. Four samples were integrated to correct for any potential library batch effect by using the Seurat functions FindIntegrationAnchors and IntegrateData based on reciprocal PCA with n = 5 neighbors (k.anchor). Integrated matrix was used for downstream analysis. Cells were clustered using the Louvain algorithm based on the first 20 principal components with a resolution of 0.3. The Uniform Manifold Approximation and Projection (UMAP) 45 was used for non-linear reduction and two-dimensional data visualization. Cluster Annotation, Subsetting, and Reclustering Cell-type annotations were assigned to each cluster based on two levels of evidence. First, the Seurat function FindAllMarkers was used to identify cluster marker genes based on one-versus-all Wilcoxon rank sum differential expression tests for each cluster. Second, cell-type identities were predicted by comparing transcriptomic profiles to a curated panel of marker genes derived from a previously published single-cell RNA sequencing dataset of the App NL-G-F mouse brain 46 . Based on this classification, we retained 25,776 cells identified as putative microglia for downstream analyses, while non-microglial cell types were excluded. For reclustering putative microglia, we applied SCTransform normalization, recomputed PCA and used the top 20 PCs for dimensionality reduction by UMAP, followed by unbiased clustering using the Seurat function FindNeighbors with the resolution of granularity set to 0.2. This led to the identification of 7 clusters each representing a microglial state defined by unique or transitory profiles. Differential Gene Expression Analysis Differentially expressed genes of specific cell states were found by applying the Seurat function FindAllMarkers for overall DE and FindMarkers for side-by-side comparisons. Genes with adjusted p values (using a Bonferroni correction) < 0.05 were considered significantly differentially expressed. Canonical Pathways Analysis by IPA was used to test for gene sets enriched in DE genes. Gene Module Scoring and Signature Analysis To compare the transcriptomic profile of our clusters with previously described microglial states, we calculated module scores using the AddModuleScore function in Seurat. Signatures were defined based on published marker lists for homeostatic microglia (HM) markers (Tmem119, P2ry12, Cx3cr1), activated response microglia (ARM) markers (Apoe, Cst7, Itgax, Lpl, Spp1, Gpnmb, Dkk2, Cd74, H2-Aa, H2-Ab1), transiting response microglia (TRM) markers (Apoe, Cst7, Itgax, Cd74, H2-Aa, H2-Ab1), interferon response microglia (IRM) markers (Ifit2, Ifit3, Ifitm3, Oasl2, and Irf7), cycling and proliferating microglia (CPM) markers (Top2a, Mcm2, Tubb5, Mki67, Cdk1) 47 , a Ribosomal Microglia signature (Tpt1, Rps3a, Rpl13, Rps23) 48 , and Disease-associated Mciroglia (DAM) markers (Cd9, Apoe, Trem2, Tyrobp, Cd63, Lgals3, Axl, Spp1, Cstb, Ctsd, Lpl, Itgax, B2m, Cst7, Gpnmb, Igf1, Irf8, Fth1, Lyz2, Ccl3, Ccl6, Timp2) in App NL-G-F mice as curated in 15 . Assessment of Signature-level Enrichment across Cell States To assess the distribution of gene expression across the identified transcriptomic states in our dataset (resolution 0.3), we generated stacked violin plots using the Seurat package. This visualization displays the log-normalized expression levels of representative marker genes, which were grouped along the y-axis into modules corresponding to curated microglial signatures and canonical pathways identified via Ingenuity Pathway Analysis (IPA). Individual cell clusters, denoted by their index numbers, are arrayed along the x-axis. Multi-omics Integration and Comparative Analysis To elucidate the regulatory landscape across different molecular layers, we performed a comparative and integrative analysis of the transcriptomic (RNA-seq) and quantitative proteomic profiles from primary microglia isolated from 9-month-old and 2-month-old App NL-G-F and APP NL-G-F ;Trem2 KO mice. Direct Profile Comparison We first assessed the direct correspondence between identified transcripts and proteins. Initial comparison was conducted by mapping identified transcripts to their corresponding proteins. A Venn analysis was utilized to determine the overlap between the two datasets, identifying molecules consistently regulated at both levels as well as those uniquely detected within a single "omic" layer. The resulting intersections were visualized as a Venn diagram using the VennDiagram package (version 1.7.3) 49 in R. Integrative Pathway Analysis To identify biological processes consistently active across both molecular layers, we employed ActivePathways, an integrative method that uses directionality and significance estimates of molecules to identify significantly enriched pathways by combining evidence from multiple omic sources 50 . Integration metrics . P-values and log2 fold-change (log2FC) values derived from the 9-month-old versus 2-month-old comparisons were processed using the ActivePathways framework. Statistical evidence from the transcriptomic and proteomic datasets was integrated via Data-driven P-value Merging (DPM), with Brown’s method utilized as a robust reference for determining combined significance. To identify biologically convergent profiles, a weighted constraint vector [mRNA=1,protein=1] was applied to prioritize genes exhibiting direct, concordant associations between the two molecular layers. Conversely, genes with conflicting directional signals or those failing to meet the integration criteria were penalized, ensuring the final pathway enrichment was driven by consistent cross-omic evidence. The relationship between the two molecular layers was visualized using a concordance scatter plot. Integrative pathway enrichment analysis. Functional enrichment was performed using the ActivePathways R package to identify biological processes significantly represented across the integrated datasets. The analysis utilized a gene list ranked by merged P-values derived from directional data integration. To determine optimal enrichment of Gene Ontology (GO) terms, a ranked hypergeometric test was applied. The gene set collection (m5.go.v2023.2.Mm.symbols.gmt) was filtered to include only pathways containing between 10 and 500 annotated genes in order to minimize biases from excessively specific or overly generic terms. Significant GO terms were defined using a Holm Family-Wise Error Rate (FWER) < 0.05. To facilitate biological interpretation, unique and significant GO terms were visualized via bar charts with the x-axis representing the −log10(adjusted P-values) and the y-axis listing the specific GO terms which were further grouped by their broader functional categories. Statistical Analysis Statistical analyses were performed using GraphPad Prism 9 software (v9.5.1; GraphPad Software, San Diego, CA, USA). Data are presented as mean ± standard error of the mean (SEM). Pairwise comparisons were analyzed using two-tailed unpaired t-tests. Multiple comparisons were analyzed using one-way or two-way analysis of variance (ANOVA) followed by Tukey’s or Sidak’s post-hoc tests. A P value < 0.05 was considered statistically significant. References Hansen, D. V., Hanson, J. E. & Sheng, M. Microglia in Alzheimer’s disease. J Cell Biol 217 , 459–472 (2018). Lambert, J.-C. et al. Meta-analysis of 74,046 individuals identifies 11 new susceptibility loci for Alzheimer’s disease. Nature genetics 45 , 1452–1458 (2013). Bellenguez, C. et al. New insights into the genetic etiology of Alzheimer’s disease and related dementias. Nat Genet (2022). Streit, W. J., Braak, H., Xue, Q.-S. & Bechmann, I. Dystrophic (senescent) rather than activated microglial cells are associated with tau pathology and likely precede neurodegeneration in Alzheimer’s disease. Acta neuropathologica 118 , 475–485 (2009). Navarro, V. et al. Microglia in Alzheimer’s disease: activated, dysfunctional or degenerative. Frontiers in aging neuroscience 10 , 140 (2018). Keren-Shaul, H. et al. A Unique Microglia Type Associated with Restricting Development of Alzheimer’s Disease. Cell 169 , 1276–1290.e17 (2017). Krasemann, S. et al. The TREM2-APOE Pathway Drives the Transcriptional Phenotype of Dysfunctional Microglia in Neurodegenerative Diseases. Immunity 47 , 566–581.e9 (2017). Guerreiro, R. et al. TREM2 variants in Alzheimer’s disease. New England Journal of Medicine 368 , 117–127 (2013). Jonsson, T. et al. Variant of TREM2 associated with the risk of Alzheimer’s disease. New England Journal of Medicine 368 , 107–116 (2013). Wang, Y. et al. TREM2 Lipid Sensing Sustains the Microglial Response in an Alzheimer’s Disease Model. Cell 160 , 1061–1071 (2015). Yuan, P. et al. TREM2 haplodeficiency in mice and humans impairs the microglia barrier function leading to decreased amyloid compaction and severe axonal dystrophy. Neuron 90 , 724–739 (2016). O’Neill, L. A. J., Kishton, R. J. & Rathmell, J. A guide to immunometabolism for immunologists. Nature Reviews Immunology 16 , 553–565 (2016). Pearce, E. L. & Pearce, E. J. Metabolic pathways in immune cell activation and quiescence. Immunity 38 , 633–643 (2013). Ulland, T. K. et al. TREM2 Maintains Microglial Metabolic Fitness in Alzheimer’s Disease. Cell 170 , 649–663.e13 (2017). Chen, Y. & Colonna, M. Microglia in Alzheimer’s disease at single-cell level. Are there common patterns in humans and mice? J Exp Med 218 , e20202717 (2021). O’Neill, L. A. J., Kishton, R. J. & Rathmell, J. A guide to immunometabolism for immunologists. Nature Reviews Immunology 16 , 553–565 (2016). Ron-Harel, N. et al. Mitochondrial biogenesis and proteome remodeling promote one-carbon metabolism for T cell activation. Cell metabolism 24 , 104–117 (2016). Chi, H. Regulation and function of mTOR signalling in T cell fate decisions. Nature reviews immunology 12 , 325–338 (2012). Palikaras, K., Lionaki, E. & Tavernarakis, N. Coordination of mitophagy and mitochondrial biogenesis during ageing in C. elegans. Nature 521 , 525–528 (2015). Melentijevic, I. et al. C. elegans neurons jettison protein aggregates and mitochondria under neurotoxic stress. Nature 542 , 367–371 (2017). Nicolás-Ávila, J. A. et al. A network of macrophages supports mitochondrial homeostasis in the heart. Cell 183 , 94–109. e23 (2020). Alvarez, M. & Casadevall, A. Phagosome extrusion and host-cell survival after Cryptococcus neoformans phagocytosis by macrophages. Current Biology 16 , 2161–2165 (2006). McQuade, A. et al. Gene expression and functional deficits underlie TREM2-knockout microglia responses in human models of Alzheimer’s disease. Nature communications 11 , 1–17 (2020). Meilandt, W. J. et al. Trem2 Deletion Reduces Late-Stage Amyloid Plaque Accumulation, Elevates the Aβ42:Aβ40 Ratio, and Exacerbates Axonal Dystrophy and Dendritic Spine Loss in the PS2APP Alzheimer’s Mouse Model. J Neurosci 40 , 1956–1974 (2020). van Dyck, C. H. et al. Lecanemab in Early Alzheimer’s Disease. N Engl J Med 388 , 9–21 (2023). Sevigny, J. et al. The antibody aducanumab reduces Aβ plaques in Alzheimer’s disease. Nature 537 , 50–56 (2016). Saito, T. et al. Single App knock-in mouse models of Alzheimer’s disease. Nat Neurosci 17 , 661–663 (2014). Lin, D. et al. Arachidonic Acid Mobilization and Peroxidation Promote Microglial Dysfunction in Aβ Pathology. J Neurosci 44 , e0202242024 (2024). Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29 , 15–21 (2013). Dobin, A. & Gingeras, T. R. Mapping RNA-seq Reads with STAR. Curr Protoc Bioinformatics 51 , 11.14.1–19 (2015). Dobin, A. & Gingeras, T. R. Optimizing RNA-Seq Mapping with STAR. Methods Mol Biol 1415 , 245–262 (2016). Roehr, J. T., Dieterich, C. & Reinert, K. Flexbar 3.0–SIMD and multicore parallelization. Bioinformatics 33 , 2941–2942 (2017). Liao, Y., Smyth, G. K. & Shi, W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics (2014). Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 15 , 550 (2014). Blighe, K., Rana, S. & Lewis, M. EnhancedVolcano: Publication-ready volcano plots with enhanced colouring and labeling. R package version 1.24. 0. 2024. (2024). Ritchie, M. E. et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic acids research 43 , e47–e47 (2015). Wu, T. et al. clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. Innovation (Camb) 2 , 100141 (2021). Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Nat Methods 9 , 676–682 (2012). Brooks, M. E. et al. glmmTMB Balances Speed and Flexibility Among Packages for Zero-inflated Generalized Linear Mixed Modeling. The R Journal 9 , 378–400 (2017). Russell, V. L. et al. emmeans: Estimated Marginal Means, aka Least-Squares Means. (2025). Florian, H. DHARMa: Residual Diagnostics for Hierarchical (Multi-Level / Mixed) Regression Models. (2025). Hothorn, T., Hornik, K., van de Wiel, M. A. & Zeileis, A. Implementing a Class of Permutation Tests: The coin Package. Journal of Statistical Software 28 , 1–23 (2008). Hao, Y. et al. Integrated analysis of multimodal single-cell data. Cell 184 , 3573–3587.e29 (2021). McGinnis, C. S., Murrow, L. M. & Gartner, Z. J. DoubletFinder: Doublet Detection in Single-Cell RNA Sequencing Data Using Artificial Nearest Neighbors. Cell Syst 8 , 329–337.e4 (2019). McInnes, L., Healy, J. & Melville, J. Umap: Uniform manifold approximation and projection for dimension reduction. arXiv preprint arXiv:1802.03426 (2018). Frigerio, C. S. et al. The major risk factors for Alzheimer’s disease: age, sex, and genes modulate the microglia response to Aβ plaques. Cell reports 27 , 1293–1306. e6 (2019). Sala Frigerio, C. et al. The Major Risk Factors for Alzheimer’s Disease: Age, Sex, and Genes Modulate the Microglia Response to Aβ Plaques. Cell Rep 27 , 1293–1306.e6 (2019). Mancuso, R. et al. Xenografted human microglia display diverse transcriptomic states in response to Alzheimer’s disease-related amyloid-β pathology. Nat Neurosci 27 , 886–900 (2024). Chen, H. & Boutros, P. C. VennDiagram: a package for the generation of highly-customizable Venn and Euler diagrams in R. BMC Bioinformatics 12 , 35 (2011). Slobodyanyuk, M. et al. Directional integration and pathway enrichment analysis for multi-omics data. Nat Commun 15 , 5690 (2024). Additional Declarations There is NO Competing Interest. Supplementary Files ExtendedFigureLegend.docx Extended Figure legend ExtendedFiguresv3.pdf Extended figures Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8896508","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":599215299,"identity":"57eccfe6-d29e-4e0f-92cd-5e13f5d8cd84","order_by":0,"name":"Jie Gao","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAxUlEQVRIiWNgGAWjYBACxmYILQcXaCBWizHxWmAgEaaSsBbmduYDjF9qbNL7+c+YbvjBYCO74QBBh7ElMMscS8udOSPH7GYPQ5oxEVp4DJgl2A7nbrjBY3abgeFwIhFa+D8wS/w7nG5w/gxIy39itPAwMH5sO5xgcCAHpOUAMVrYDA4z9qUZzpyRVnazxyDZeCYhLYb9hx8+/PHNRp6f//C2Gz8q7GT7CGppAHqZB841IKAcBORBjvtBhMJRMApGwSgYwQAAQVVDCSTbk/0AAAAASUVORK5CYII=","orcid":"https://orcid.org/0000-0001-5448-6606","institution":"Ohio State University","correspondingAuthor":true,"prefix":"","firstName":"Jie","middleName":"","lastName":"Gao","suffix":""},{"id":599215300,"identity":"f26359d8-c44d-4572-826d-a2db10058408","order_by":1,"name":"Da Lin","email":"","orcid":"","institution":"Ohio State University","correspondingAuthor":false,"prefix":"","firstName":"Da","middleName":"","lastName":"Lin","suffix":""},{"id":599215301,"identity":"fbe08e94-a3de-49eb-8a94-1ed54a680a56","order_by":2,"name":"Jeffrey Atkinson","email":"","orcid":"https://orcid.org/0000-0002-0250-2210","institution":"Ohio State University","correspondingAuthor":false,"prefix":"","firstName":"Jeffrey","middleName":"","lastName":"Atkinson","suffix":""},{"id":599215302,"identity":"bc08d333-67f2-49fd-92a2-d0b7e491775d","order_by":3,"name":"Min Chen","email":"","orcid":"","institution":"The Ohio State University","correspondingAuthor":false,"prefix":"","firstName":"Min","middleName":"","lastName":"Chen","suffix":""},{"id":599215303,"identity":"8cb73da5-6a28-40ff-9cd3-92281e8b9193","order_by":4,"name":"Sohan Jayasekara","email":"","orcid":"","institution":"Ohio State University","correspondingAuthor":false,"prefix":"","firstName":"Sohan","middleName":"","lastName":"Jayasekara","suffix":""},{"id":599215304,"identity":"a10ffeb2-b89e-477e-a986-51463e46e2d4","order_by":5,"name":"Benjamin Segal","email":"","orcid":"","institution":"Ohio State University","correspondingAuthor":false,"prefix":"","firstName":"Benjamin","middleName":"","lastName":"Segal","suffix":""}],"badges":[],"createdAt":"2026-02-16 22:30:13","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8896508/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8896508/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":105567383,"identity":"16710eea-f75e-4201-93b3-2187dbf88ae6","added_by":"auto","created_at":"2026-03-27 12:59:15","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":388998,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMicroglia mount a proteome-wide anabolic adaptation to Aβ pathology. (A)\u003c/strong\u003e Experimental schematic. Primary microglia were isolated from 3- and 9-month-old \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e and\u0026nbsp;\u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e;Trem2\u003c/em\u003e\u003csup\u003e\u003cem\u003eKO\u003c/em\u003e\u003c/sup\u003e mice for integrated proteomic and transcriptomic profiling. \u003cstrong\u003e(B)\u003c/strong\u003e Ingenuity Pathway Analysis (IPA) of differentially abundant proteins (DAP) in 9-month-old \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003csup\u003e \u003c/sup\u003evs. 3-month-old \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003csup\u003e \u003c/sup\u003emicroglia, highlighting the enrichment of Protein synthesis pathways. \u003cstrong\u003e(C–D)\u003c/strong\u003e Gene Set Enrichment Analysis (GSEA) dot plots for Cellular Component (C) and Biological Process (D) terms. Red arrows\u0026nbsp;indicate the concurrent enrichment of \"Ribosome\" (anabolic) signature and orange arrows\u0026nbsp;indicate\u0026nbsp; \"Lysosomal membrane\" (catabolic) signatures enriched in 9-month-old vs. 3-month-old \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003csup\u003e \u003c/sup\u003emicroglia. \u003cstrong\u003e(E) \u003c/strong\u003eIPA and \u003cstrong\u003e(F) \u003c/strong\u003eGSEA of differentially expressed gene (DEG) in 9-month-old \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e, highlighting genes involved in Protein Translation (red arrows) and catabolic process (yellow arrows).\u003c/p\u003e","description":"","filename":"Figuresv31.png","url":"https://assets-eu.researchsquare.com/files/rs-8896508/v1/a2a3927fd9eba68f6514d197.png"},{"id":105567100,"identity":"bf624f7f-7e73-49ae-94c4-88c251a01d4d","added_by":"auto","created_at":"2026-03-27 12:58:19","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":269941,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTREM2 deficiency causes a collapse of the anabolic response. (A)\u003c/strong\u003e Venn diagram of\u0026nbsp;unique and shared genes/proteins between transcriptome and proteome. \u003cstrong\u003e(B)\u003c/strong\u003e Volcano plot comparing the proteome of 9-month-old \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e and\u0026nbsp;\u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e;Trem2\u003c/em\u003e\u003csup\u003e\u003cem\u003eKO\u003c/em\u003e\u003c/sup\u003e microglia. Triangle dots indicate uniquely altered proteins in the \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e;Trem2\u003c/em\u003e\u003csup\u003e\u003cem\u003eKO\u003c/em\u003e\u003c/sup\u003e microglia. \u003cstrong\u003e(C)\u003c/strong\u003e IPA of top-ranked pathways in \u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e;\u003cem\u003eTrem2\u003c/em\u003e\u003csup\u003eKO\u003c/sup\u003e microglia, showing specific suppression of protein translation pathways (blue arrows) and the increased levels of synaptic proteins (red arrows). \u003cstrong\u003e(D)\u003c/strong\u003e GSEA dot plots confirming the broad reduction of cytosolic and mitochondrial ribosomal components in \u003cem\u003eTrem2\u003c/em\u003e\u003csup\u003e\u003cem\u003eKO\u003c/em\u003e\u003c/sup\u003e microglia (blue arrows). \u003cstrong\u003e(E) \u003c/strong\u003eIPA and \u003cstrong\u003e(F) \u003c/strong\u003eGSEA of differentially expressed gene (DEG) in 9-month-old \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e and \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e;Trem2\u003c/em\u003e\u003csup\u003e\u003cem\u003eKO\u003c/em\u003e\u003c/sup\u003e microglia, highlighting suppressed genes involved in protein translation and ribosome (blue arrows).\u003c/p\u003e","description":"","filename":"Figuresv32.png","url":"https://assets-eu.researchsquare.com/files/rs-8896508/v1/54f7b3b7dc07c902f64552b2.png"},{"id":105567166,"identity":"be782f6a-1b8d-452a-bce2-361a30d10e7a","added_by":"auto","created_at":"2026-03-27 12:58:31","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":2264949,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTREM2 drives nascent protein synthesis in plaque-associated microglia. (A)\u003c/strong\u003e Schematic of \u003cem\u003ein vivo\u003c/em\u003e FUNCAT labeling. Mice were injected with Azidohomoalanine (AHA, 50 mg/kg) 16 hours prior to microglia isolation and analysis. \u003cstrong\u003e(B)\u003c/strong\u003e Representative flow cytometry histograms of FUNCAT intensity in CX3CR1\u003csup\u003e+\u003c/sup\u003e CD45\u003csup\u003e+\u003c/sup\u003e microglia. Red arrow indicates the \"FUNCAT-high\" shoulder present in \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003csup\u003e \u003c/sup\u003emicroglia but absent in \u003cem\u003eTrem2\u003c/em\u003e\u003csup\u003e\u003cem\u003eKO\u003c/em\u003e\u003c/sup\u003e mice. \u003cstrong\u003e(C)\u003c/strong\u003e Quantification of mean fluorescence intensity (MFI) for FUNCAT signal. \u003cstrong\u003e(D)\u003c/strong\u003e Comparison of FUNCAT intensity between CD45\u003csup\u003ehigh\u003c/sup\u003e and CD45\u003csup\u003elow\u003c/sup\u003e microglial population. \u003cstrong\u003e(E)\u003c/strong\u003e Representative confocal images of cortical plaques stained for Iba1 (green), Aβ (82E1, grey), and nascent proteins (FUNCAT, red). Scale bar: 50 µm. \u003cstrong\u003e(F)\u003c/strong\u003e Quantification of FUNCAT intensity in plaque-associated microglia. \u003cstrong\u003e(G)\u003c/strong\u003e Representative confocal images of phosphorylated Ribosomal Protein S6 (p-S6, Ser235/236) in plaque-associated microglia. \u003cstrong\u003e(H)\u003c/strong\u003e Quantification of p-S6 intensity per microglial cell. n = 6-8 mice per group. ***\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.001 by two-tailed unpaired t-test.\u003c/p\u003e","description":"","filename":"Figuresv33.png","url":"https://assets-eu.researchsquare.com/files/rs-8896508/v1/31aac82ebfc9194ff3816370.png"},{"id":105558559,"identity":"813d22f7-2a7e-4afd-914d-4febcd1f6b88","added_by":"auto","created_at":"2026-03-27 11:38:04","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":1933524,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eLoss of TREM2 leads to proteostatic failure and the intracellular accumulation of undigested cargo.\u003c/strong\u003e (A) Gene-concept network (cnet) plots displaying the top-ranked differentially abundant proteins associated with the Gene Ontology (GO) terms ‘Synapse’, ‘Myelin Sheath’, and ‘Lysosome’. Note that these cargo proteins are specifically enriched in \u003cem\u003eTrem2\u003c/em\u003e\u003csup\u003e\u003cem\u003eKO\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e \u003c/em\u003emicroglia compared to \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F \u003c/em\u003e\u003c/sup\u003econtrols at 9 months of age, indicating a failure to degrade internalized material. \u003cstrong\u003e(B, D)\u003c/strong\u003e Representative confocal images of plaque-associated microglia in 9-month-old mice co-stained for PSD95 (postsynaptic marker, B) or Myelin Basic Protein (MBP, D), together with Iba1 and Aβ. White arrows highlight the pronounced intracellular accumulation of PSD95 and MBP puncta within the soma of \u003cem\u003eTrem2\u003c/em\u003e\u003csup\u003e\u003cem\u003eKO\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e \u003c/em\u003emicroglia. Scale bar: 50 μm. \u003cstrong\u003e(C, E)\u003c/strong\u003e Quantification of cargo accumulation relative to plaque proximity. Violin plots display the voxel overlap percentage of PSD95 (C) or MBP (E) within Iba1+ ROIs, classified as Non-Plaque-Associated (NPAM) or Plaque-Associated (PAM). Colored points represent individual ROIs; solid black circles and error bars represent Estimated Marginal Means (EMMs) ± 95% CIs derived from a beta-binomial Generalized Linear Mixed Model (GLMM). Lines connect paired regions within biological replicates (n = 5-6 mice per group). Pairwise comparisons were performed using the Delta method. Significance levels: *P \u0026lt; 0.05, **P \u0026lt; 0.01, ***P \u0026lt; 0.001, ****P \u0026lt; 0.0001. exact p-values are shown for non-significant trends.\u003c/p\u003e","description":"","filename":"Figuresv34.png","url":"https://assets-eu.researchsquare.com/files/rs-8896508/v1/b4fbf13b0628b99b75a2d36f.png"},{"id":105558556,"identity":"8d9b5551-7401-408b-955c-bc99b34aca41","added_by":"auto","created_at":"2026-03-27 11:38:04","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":400387,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTREM2 is indispensable for initiating anabolic adaptation at the onset of Aβ phagocytosis.\u003c/strong\u003e \u003cstrong\u003e(A)\u003c/strong\u003e Schematic of the gating strategy used to sort Methoxy-X04 (X04) positive (X04\u003csup\u003e+\u003c/sup\u003e) and negative (X04\u003csup\u003e-\u003c/sup\u003e) microglia for transcriptional profiling. \u003cstrong\u003e(B)\u003c/strong\u003e Representative flow cytometry plots illustrating the gating of Aβ-phagocytic microglia. Upper panels\u003cstrong\u003e: \u003c/strong\u003eComparison of Wild-Type (WT) and \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003csup\u003e \u003c/sup\u003emice confirms the specificity of the X04 signal. Lower panels: Stratification of the X04\u003csup\u003e+\u003c/sup\u003e population into Low, Medium, and High fluorescence intensity bins, representing increasing phagocytic load, in \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003csup\u003e \u003c/sup\u003eand \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e;Trem2\u003c/em\u003e\u003csup\u003e\u003cem\u003eKO\u003c/em\u003e\u003c/sup\u003e\u003csup\u003e \u003c/sup\u003emice. \u003cstrong\u003e(C)\u003c/strong\u003e Temporal analysis of Aβ-phagocytic microglia. Upper panels: Expansion of the X04\u003csup\u003e+\u003c/sup\u003e population in \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003csup\u003e\u0026nbsp; \u003c/sup\u003emice from 5–6 months to 9–10 months of age. Lower panels: Comparison of X04 intensity distributions in 9-month-old \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003csup\u003e \u003c/sup\u003eand \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e;Trem2\u003c/em\u003e\u003csup\u003e\u003cem\u003eKO\u003c/em\u003e\u003c/sup\u003e mice, showing the marked reduction of X04\u003csup\u003ehigh\u003c/sup\u003e population in TREM2\u003csup\u003eKO\u003c/sup\u003e microglia. \u003cstrong\u003e(D)\u003c/strong\u003e Principal Component Analysis (PCA) of bulk RNA-seq samples (\u003cem\u003en\u003c/em\u003e = 46). Dots represent individual biological replicates across genotypes: WT (\u003cem\u003en\u003c/em\u003e = 5), \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e(\u003cem\u003en\u003c/em\u003e = 23), and \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e;Trem2\u003c/em\u003e\u003csup\u003e\u003cem\u003eKO\u003c/em\u003e\u003c/sup\u003e\u003csup\u003e \u003c/sup\u003e(\u003cem\u003en\u003c/em\u003e = 18). Colors denote X04 classification (X04\u003csup\u003e-\u003c/sup\u003e vs. X04\u003csup\u003e+\u003c/sup\u003e subsets: Low, Medium, High). Note that X04 levels drive separation along PC2, defining distinct plaque-phagocytic states. \u003cstrong\u003e(E)\u003c/strong\u003e IPA of top-ranked pathways enriched in X04\u003csup\u003e+\u003c/sup\u003e vs. X04\u003csup\u003e-\u003c/sup\u003e \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e microglia. Arrows highlight the robust induction of Mitochondrial Biogenesis/Oxidative Phosphorylation (red) and Protein Targeting/Synthesis (orange) upon Aβ engagement. \u003cstrong\u003e(F, G)\u003c/strong\u003e IPA Comparison Analysis illustrating transcriptional changes of microglia across the phagocytic gradient (X04\u003csup\u003ehigh\u003c/sup\u003e, X04\u003csup\u003emed\u003c/sup\u003e, X04\u003csup\u003elow\u003c/sup\u003e) relative to X04\u003csup\u003e-\u003c/sup\u003e controls in \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003csup\u003e \u003c/sup\u003emice (F) and at \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e;Trem2\u003c/em\u003e\u003csup\u003e\u003cem\u003eKO\u003c/em\u003e\u003c/sup\u003e\u003csup\u003e \u003c/sup\u003emice (G) at 9 months of age. \u003cstrong\u003e(H)\u003c/strong\u003e\u0026nbsp;IPA Comparison of\u0026nbsp;\u0026nbsp;\u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003csup\u003e\u0026nbsp; \u003c/sup\u003evs \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e;Trem2\u003c/em\u003e\u003csup\u003e\u003cem\u003eKO\u003c/em\u003e\u003c/sup\u003e\u0026nbsp;stratified by Methoxy-X04 intensity at 9 months of age. The activation z-scores for the selected significantly enriched canonical pathways (Fisher’s exact test; adjusted \u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05) are presented as a heatmap. Color scale indicates predicted pathway directionality: red corresponds to positive z-scores (predicted activation), blue represents negative z-scores (predicted inhibition), and grey denotes pathways for which IPA was unable to predict directionality. Pathways lacking statistical significance are marked with an asterisk (*). Rows of pathways are grouped into four categories according to their biological relevance.\u003c/p\u003e","description":"","filename":"Figuresv35.png","url":"https://assets-eu.researchsquare.com/files/rs-8896508/v1/1d2aeca603dfff084301dc4c.png"},{"id":105567099,"identity":"55b1a200-23ef-4cf7-ae7c-6571cd4ba803","added_by":"auto","created_at":"2026-03-27 12:58:18","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":2610657,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTREM2 drives local mitochondrial biogenesis and prevents metabolic stagnation in plaque-associated microglia.\u003c/strong\u003e \u003cstrong\u003e(A)\u003c/strong\u003e Flow cytometry analysis of mitochondrial fitness in primary microglia from 9-month-old \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003csup\u003e \u003c/sup\u003eand \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e;Trem2\u003c/em\u003e\u003csup\u003e\u003cem\u003eKO\u003c/em\u003e\u003c/sup\u003e mice. Quantification shows Mitochondrial Mass (MitoTracker Green), Membrane Potential (MitoTracker Red CMXRos), and the Mitochondrial Health Index (defined as the ratio of Membrane Potential to Mass). Note that Trem2\u003csup\u003eKO\u003c/sup\u003e microglia exhibit a \"High Mass / Low Potential\" phenotype, indicative of the accumulation of metabolically incompetent organelles.\u003cstrong\u003e (B) \u003c/strong\u003eFlow cytometry analysis of lysosomal mass (LysoTracker Deep Red). The lack of significant difference between genotypes suggests that the mitochondrial accumulation in Trem2\u003csup\u003eKO\u003c/sup\u003e microglia is not driven by a general lysosomal deficiency. \u003cstrong\u003e(C) \u003c/strong\u003eVisualization of nascent mitochondrial biogenesis in situ using the AHA-TOM20 Proximity Ligation Assay (PLA). Representative confocal images show newly synthesized TOM20 protein (TOM20-PLA puncta, red) within Iba1+ microglia (green) in the plaque niche. Note the robust biogenic signal in \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003csup\u003e\u0026nbsp; \u003c/sup\u003emicroglia compared to the blunted response in Trem2\u003csup\u003eKO\u003c/sup\u003e microglia. \u003cstrong\u003e(D) \u003c/strong\u003eQuantification of nascent TOM20-PLA puncta volume per microglia (30-40 plaques from 5-6 mice were quantified per group). \u003cstrong\u003e(E) \u003c/strong\u003eRepresentative confocal images of plaque-associated microglia co-stained for TOM20 (mitochondria), Iba1, and FUNCAT (nascent protein synthesis). Maximum intensity projection (MIP) of 4 μm z-stack. White boxes highlight the colocalization of nascent proteins and mitochondria within the distal processes of App\u003csup\u003eNL-G-F\u003c/sup\u003e microglia. Scale bar: 20 μm. \u003cstrong\u003e(F)\u003c/strong\u003e Discovery of microglial exophers. Representative confocal images (MIP, 5 μm) of plaque-associated microglia co-stained for TOM20, Iba1, and Aβ. White arrows indicate large (\u0026gt;2 μm), detached exopher-like structures filled with mitochondrial content emerging from Trem2\u003csup\u003eKO\u003c/sup\u003e microglia. Scale bar: 20 μm. \u003cstrong\u003e(G)\u003c/strong\u003e Quantification of mitochondrial mass accumulation relative to plaque proximity. Violin plots display the distribution of TOM20 voxel overlap within Iba1+ ROIs, classified as Non-Plaque-Associated (NPAM) or Plaque-Associated (PAM). Colored points represent individual ROIs; solid circles and error bars represent Estimated Marginal Means (EMMs) ± 95% CIs derived from a beta-binomial Generalized Linear Mixed Model (GLMM). Lines connect paired regions within biological replicates (n = 5-6 mice). Pairwise comparisons were performed using the Delta method. ***P \u0026lt; 0.001, ****P \u0026lt; 0.0001.\u003c/p\u003e","description":"","filename":"Figuresv36.png","url":"https://assets-eu.researchsquare.com/files/rs-8896508/v1/8e34df7bf9f6dd1b6c878345.png"},{"id":105558564,"identity":"f88b86b7-d828-49a9-85b1-c6a403c92f22","added_by":"auto","created_at":"2026-03-27 11:38:05","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":3116166,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMetabolic collapse triggers the extrusion of neurotoxic exophers containing synaptic and tauopathic cargo.\u0026nbsp; (A)\u003c/strong\u003e 3D surface rendering (Imaris) of a plaque-associated microglial cell in \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e;Trem2\u003c/em\u003e\u003csup\u003e\u003cem\u003eKO\u003c/em\u003e\u003c/sup\u003e mice, co-stained for DAPI, PSD95 (green), Iba1 (magenta), and Aβ (cyan). Note the extrusion of a large, cargo-filled exopher. \u003cstrong\u003e(B)\u003c/strong\u003e High-magnification (63X) single optical slice (0.17 μm) confirming the enclosure of PSD95 puncta and Aβ aggregates within the Iba1+ membrane of an exopher. \u003cstrong\u003e(C)\u003c/strong\u003e Quantification of PSD95+ exopher density in the plaque niche (10\u003csup\u003e5\u003c/sup\u003e μm\u003csup\u003e3\u003c/sup\u003e ROI). Each point represents an individual plaque ROI. Medians and interquartile ranges (IQRs) are indicated by dots and bars, respectively. Statistical significance was determined using a non-parametric Wilcoxon rank-sum test with exact permutation (R package coin, v1.4.3). \u003cstrong\u003e(D)\u003c/strong\u003e 3D surface rendering of an exopher containing Myelin Basic Protein (MBP) debris. \u003cstrong\u003e(E)\u003c/strong\u003e Single optical slice (0.17 μm) confirmation of MBP cargo within the exopher lumen. (F) Quantification of MBP+ exopher density between genotypes. Statistics as in (C). \u003cstrong\u003e(G)\u003c/strong\u003e Discovery of Tau-seeding exophers. Representative confocal image (MIP, 5 μm) of plaque-associated microglia co-stained for hyperphosphorylated Tau (AT8, red), Iba1 (green), and Aβ (blue). White box highlights an exopher filled with pathological Tau emerging from a Trem2\u003csup\u003eKO\u003c/sup\u003emicroglia. Scale bar: 20 μm. \u003cstrong\u003e(H)\u003c/strong\u003e Quantification of AT8+ exopher density. Statistics as in (C). \u003cstrong\u003e(I)\u003c/strong\u003e Representative low-magnification images of the entorhinal cortex (40 μm sections) stained for AT8, Iba1, and Aβ. Note the marked increase in AT8 burden surrounding plaques in \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e;Trem2\u003c/em\u003e\u003csup\u003e\u003cem\u003eKO\u003c/em\u003e\u003c/sup\u003e mice compared to controls. \u003cstrong\u003e(J)\u003c/strong\u003e Quantification of total AT8+ area in the entorhinal cortex. Data represent mean ± SEM (n = 8-9 mice per group). Unpaired t-test.\u003c/p\u003e","description":"","filename":"Figuresv37.png","url":"https://assets-eu.researchsquare.com/files/rs-8896508/v1/48bf0e4a7152e2569e9ba897.png"},{"id":105558561,"identity":"a4d7ed91-9f36-4a8c-bae2-36f41d874720","added_by":"auto","created_at":"2026-03-27 11:38:04","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":1170349,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThe Anabolic Adaptation signature defines phagocytic competence more accurately than the canonical DAM signature. (A) \u003c/strong\u003eUMAP projection of single-cell RNA-seq profiles from sorted X04+ (phagocytic) and X04- (non-phagocytic) microglia isolated from 9-10 month-old \u003cem\u003eApp\u003c/em\u003e\u003csup\u003e\u003cem\u003eNL-G-F\u003c/em\u003e\u003c/sup\u003e mice. Note that phagocytic status drives cells into distinct topological niches even within the canonically defined Homeostatic (HM) and Transition (TRM) global clusters. \u003cstrong\u003e(B)\u003c/strong\u003e High-resolution sub-clustering of the HM and TRM populations. The analysis resolves discrete subpopulations, splitting HM into clusters 0 and 2, and TRM into clusters 1, 3, and 6, that segregate based on phagocytic competence. \u003cstrong\u003e(C)\u003c/strong\u003e Quantification of cluster composition, showing the relative enrichment of specific sub-states within the X04+ versus X04- fractions. \u003cstrong\u003e(D)\u003c/strong\u003e Ingenuity Pathway Analysis (IPA) and (E) Gene Set Enrichment Analysis (GSEA) comparing the phagocytic-enriched sub-cluster (Cluster 2) versus the non-phagocytic sub-cluster (Cluster 0). The analysis reveals a robust enrichment of Ribosome and Oxidative Phosphorylation pathways specifically in the phagocytic sub-population. \u003cstrong\u003e(F)\u003c/strong\u003e Violin plots displaying the expression levels of the Homeostatic signature (Tmem119, P2ry12, Cx3cr1), DAM signature (Trem2, Cst7, Itgax, Cd74, H2-Aa), and the Anabolic Adaptation signature (Atp5e, Cox4i1, Ndufa1, Eif3f, Rps5, Rpl23) across identified microglial subpopulations. Note that while the DAM signature is comparable between phagocytic and non-phagocytic states, the Anabolic Adaptation signature is strictly upregulated in the phagocytic clusters (1 and 2).\u003c/p\u003e","description":"","filename":"Figuresv38.png","url":"https://assets-eu.researchsquare.com/files/rs-8896508/v1/44b1247eeab879432140f058.png"},{"id":105570318,"identity":"105629ea-d983-4ae2-b773-3d7469615bec","added_by":"auto","created_at":"2026-03-27 13:16:06","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":13620231,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8896508/v1/21b50801-1120-4cf6-a4e6-ce71cf43678d.pdf"},{"id":105558554,"identity":"d1d1f09f-710f-4d1f-973c-a102e8d824bd","added_by":"auto","created_at":"2026-03-27 11:38:04","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":21719,"visible":true,"origin":"","legend":"Extended Figure legend","description":"","filename":"ExtendedFigureLegend.docx","url":"https://assets-eu.researchsquare.com/files/rs-8896508/v1/bd25ed057f7377fff3f3bc5b.docx"},{"id":105558562,"identity":"3021da66-5156-4f55-bd75-dea80bc6ecac","added_by":"auto","created_at":"2026-03-27 11:38:04","extension":"pdf","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":12611798,"visible":true,"origin":"","legend":"Extended figures","description":"","filename":"ExtendedFiguresv3.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8896508/v1/3062a38be21afcb0bb71ba53.pdf"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"TREM2 fuels the anabolic adaptation required for microglial resilience in Alzheimer’s disease","fulltext":[{"header":"Introduction","content":"\u003cp\u003eIn Alzheimer\u0026rsquo;s disease (AD), microglia serve as a primary line of defense, tasked with the containment and clearance of neurotoxic A\u0026beta; aggregates \u003csup\u003e1\u003c/sup\u003e. Genetic studies have solidly placed microglial function at the center of AD risk, suggesting the necessity of a robust innate immune response to delay pathogenesis \u003csup\u003e2\u003c/sup\u003e \u003csup\u003e3\u003c/sup\u003e. However, this protective capacity is not static. As pathology progresses, microglia undergo profound phenotypic changes- initially mounting a containment response, but frequently transitioning toward dysregulated, senescent, or exhausted states that fail to limit neurodegeneration \u003csup\u003e4\u003c/sup\u003e \u003csup\u003e5\u003c/sup\u003e. Understanding the mechanisms that allow microglia to maintain functional resilience in the face of chronic proteotoxic stress is a critical priority for therapeutic intervention.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eRecent advances in single-cell transcriptomics have mapped the trajectory of this response, defining a conserved \u0026apos;Disease-Associated Microglia\u0026apos; (DAM) or \u0026apos;Microglial Neurodegenerative Phenotype\u0026apos; (MGnD) signature \u003csup\u003e6\u003c/sup\u003e \u003csup\u003e7\u003c/sup\u003e. This transcriptional program, driven largely by the triggering receptor expressed on myeloid cells 2 (TREM2), involves the downregulation of homeostatic checkpoints (e.g., P2ry12, Cx3cr1) and the induction of lipid-sensing and phagocytic pathways (e.g., Apoe, Lpl, Clec7a). The vital importance of TREM2 signaling is underscored by the fact that TREM2 loss-of-function variants (e.g., R47H) triple the risk of AD, and are associated with impaired microglial clustering around plaques,leading to diffuse, neurotoxic amyloid pathology \u003csup\u003e8\u003c/sup\u003e \u003csup\u003e9\u003c/sup\u003e \u003csup\u003e10\u003c/sup\u003e \u003csup\u003e11\u003c/sup\u003e. However, the acquisition of a transcriptional signature is merely a blueprint; executing the functions of plaque-associated microglia, such as proliferation, migration, and continuous phagocytosis, imposes a formidable bioenergetic and biosynthetic demand on the cell. Phagocytes must constantly synthesize new membranes, hydrolytic enzymes, and organelles to replace those consumed during the degradation of A\u0026beta;. How microglia acquire the metabolic resources to sustain this high-demand state in the nutrient-deprived or toxic milieu of the AD brain remains poorly understood. While immunometabolism is well-characterized in peripheral macrophages, where activation is coupled to distinct metabolic switches (e.g., glycolysis vs. oxidative phosphorylation) \u003csup\u003e12\u003c/sup\u003e \u003csup\u003e13\u003c/sup\u003e, the metabolic dependencies of plaque-associated microglia remain largely unexplored \u003csup\u003e14\u003c/sup\u003e. Specifically, it is unknown whether TREM2 signaling simply instructs the cell what to do, or if it also initiates the anabolic drive required to sustain these functions over time.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eHere, we combine integrated proteomics, transcriptomics, and \u003cem\u003ein vivo\u003c/em\u003e metabolic labeling to define the bioenergetic requirements of microglial resilience. We demonstrate that A\u0026beta; pathology triggers a TREM2-dependent \u0026quot;anabolic adaptation\u0026quot;, a coordinated surge in protein synthesis and mitochondrial biogenesis at the site of phagocytosis. We show that this anabolic drive is distinct from the canonical DAM transcriptional signature and is essential for maintaining microglial proteostasis. In the absence of TREM2, microglia fail to mount this anabolic response, leading to a state of metabolic exhaustion characterized by mitochondrial stagnation, stalled phagocytic flux, and the extrusion of undigested cargo via neurotoxic exophers. Our findings redefine TREM2 as a metabolic regulator that couples sensing to cellular anabolism, ensuring microglial survival and preventing the secondary seeding of pathology during chronic neurodegeneration.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u003cu\u003eMicroglia mount a TREM2-dependent anabolic adaptation to A\u0026beta; pathology\u003c/u\u003e\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e.\u003c/strong\u003e To gain mechanistic insight into microglial adaptation to amyloid pathology, we performed integrated proteomic and transcriptomic profiling of primary microglia isolated from \u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e and \u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e;\u003cem\u003eTrem2\u003c/em\u003e\u003csup\u003eKO\u003c/sup\u003e mice at 3 and 9 months of age, corresponding to early and intermediate stages of amyloid deposition, respectively (Fig. 1A).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn \u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u0026nbsp;\u003c/sup\u003emicroglia, proteomic profiling identified over 7,000 proteins, of which 2,621 were differentially abundant between 3 and 9 months, indicating extensive proteome remodeling as pathology progressed. Ingenuity Pathway Analysis (IPA) revealed a robust induction of protein synthesis pathways, including \u003cem\u003eEukaryotic Translation Initiation, Elongation, and Termination, and EIF2 Signaling\u003c/em\u003e (Fig. 1B, red arrow). Gene Set Enrichment Analysis (GSEA) of both Cellular Component (CC) and Biological Process (BP) terms confirmed the increased abundance of translation machinery, including cytosolic and mitochondrial ribosomal subunits (Fig. 1C-D, red arrows; Extended Fig. 1A). In parallel, proteins associated with degradative processes, such as \u0026lsquo;autophagosome membrane\u0026rsquo;, \u0026lsquo;lysosomal membrane\u0026rsquo;, and \u0026lsquo;phagocytic vesicles\u0026rsquo;, were also enriched ((Fig. 1C-D, orange arrows). These findings indicate that microglia respond to A\u0026beta; deposition by engaging a coordinated anabolic-catabolic program that supports continuous proteome renewal. Transcriptomic profiling corroborated this adaptive response, though to a lesser extent, revealing increased expression of genes associated with protein translation (Fig. 1E, red arrow), ribosomal biogenesis, mitochondrial respiratory chain complexes (Fig. 1F, Extended Fig. 1B, red arrow), and lysosomal/CLEAR pathways (Fig. 1E-F, orange arrows). Together, these data suggest that A\u0026beta; pathology induces a broad metabolic remodeling in microglia that couples elevated protein synthesis with enhanced degradative capacity.\u003c/p\u003e\n\u003cp\u003eTo determine whether this metabolic remodeling is TREM2-dependent, we compared the profiles of \u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e and \u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e;\u003cem\u003eTrem2\u003c/em\u003e\u003csup\u003eKO\u003c/sup\u003e microglia. The impact of TREM2 was tightly linked to disease stage. At 3 months, when A\u0026beta; burden is minimal, TREM2 deficiency produced no significant change in the microglial proteome (Extended Fig. 2A). By 9 months, however, loss of TREM2 profoundly disrupted the proteome, resulting in 1,480 significantly altered proteins (Fig. 2A). Strikingly, proteomic and transcriptomic changes showed poor correlation (Fig. 2A-B), underscoring the importance of proteome-level analysis to reveal functional remodeling under A\u0026beta; stress.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eA dominant effect of TREM2 deficiency was collapse of the anabolic response. IPA revealed marked downregulation of translation-related pathways, including \u003cem\u003eSRP-dependent protein targeting, translation elongation,\u0026nbsp;\u003c/em\u003eand\u003cem\u003e\u0026nbsp;translation termination\u003c/em\u003e (Fig. 2C, blue arrow). GSEA confirmed broad reductions in cytosolic and mitochondrial ribosomal components, as well as ribosome biogenesis (Fig. 2D; Extended Fig. 2B, blue arrows), demonstrating that TREM2 is indispensable for inducing anabolic pathways during A\u0026beta; pathology. Transcriptomic profiling reinforced these findings, showing significantly reduced expression of genes involved in protein translation, ribosomal assembly, and mitochondrial respiratory complexes (Fig. 2E-F; Extended Fig. 2C). Consistent with defective proteome renewal, Trem2\u003csup\u003eKO\u003c/sup\u003e microglia accumulated synaptic and myelin proteins (Fig. 2C-D, red arrows), materials normally degraded following phagocytosis. Because microglia routinely clear damaged synapses and myelin debris, this accumulation indicates impaired proteostasis when anabolic support is lost. Directional integration of proteomic and transcriptomic datasets highlighted that\u0026nbsp;that while\u003cem\u003e\u0026nbsp;App\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e microglia coordinate protein translation, transport, and degradation, Trem2\u003csup\u003eKO\u003c/sup\u003e microglia fail to mount this remodeling response (Extended Fig. 2D-E). Together, these findings suggest that microglia adapt to A\u0026beta; pathology by activating a TREM2-dependent anabolic program that drives proteome remodeling and maintains proteostasis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u003cu\u003eTREM2 drives nascent protein synthesis and phagocytic processing in plaque-associated microglia\u003c/u\u003e\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e.\u0026nbsp;\u003c/strong\u003eTo validate the anabolic adaptation revealed by our omics analysis \u003cem\u003ein vivo\u003c/em\u003e, we employed Fluorescent Non-Canonical Amino Acid Tagging (FUNCAT) to label newly synthesized proteins (Fig. 3A). Sixteen hours after intraperitoneal injection of azidohomoalanine (AHA), a methionine analog incorporated into nascent proteins, we analyzed CX3CR1\u003csup\u003e+\u003c/sup\u003e; CD45\u003csup\u003e+\u003c/sup\u003e\u0026nbsp; microglia by flow cytometry (Extended Fig. 3A). At 3 months of age, FUNCAT intensity was comparable between\u0026nbsp;\u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e and \u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e;\u003cem\u003eTrem2\u003c/em\u003e\u003csup\u003eKO\u003c/sup\u003e microglia. By 9 months, however, microglia from\u0026nbsp;\u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e mice exhibited an approximately twofold increase in protein synthesis, whereas TREM2 deficiency completely abolished this induction (Fig. 3B-C). Histograms revealed a distinct \u0026ldquo;FUNCAT-high\u0026rdquo; shoulder in\u0026nbsp;\u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e microglia (Fig. 3B, red arrow), indicating the emergence of a metabolically active subpopulation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eBecause CD45\u003csup\u003ehigh\u003c/sup\u003e microglia are known to accumulate around plaques, we examined their relationship to protein synthesis. While the CD45\u003csup\u003ehigh\u003c/sup\u003e subpopulation expanded with the progression of A\u0026beta; pathology (Extended Fig. 3B-C), this expansion was largely absent in\u0026nbsp;\u003cem\u003eTrem2\u003c/em\u003e\u003csup\u003eKO\u003c/sup\u003e mice. CD45 levels correlated positively with FUNCAT intensity (Fig. 3D), suggesting plaque-associated microglia as the primary population undergoing upregulated nascent protein synthesis. In situ FUNCAT labeling further confirmed a robust induction of protein synthesis specifically within the plaque-associated microglia of\u0026nbsp;\u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e mice, an induction that was largely abolished in\u0026nbsp;\u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e;\u003cem\u003eTrem2\u003c/em\u003e\u003csup\u003eKO\u003c/sup\u003e mice\u0026nbsp;(Fig. 3E-F). Previous studies have linked TREM2 signaling to the activation of mTORC1, a key regulator of ribosome biogenesis and anabolic metabolism. Consistent with this mechanism, phosphorylated ribosomal protein S6 (p-RPS6), a canonical downstream marker of mTORC1 activity, was markedly increased in plaque-associated microglia of\u0026nbsp;\u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e mice but not in those of\u0026nbsp;\u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e;\u003cem\u003eTrem2\u003c/em\u003e\u003csup\u003eKO\u003c/sup\u003e mice (Fig. 3G-H). Together, these findings suggest that TREM2-dependent activation of mTORC1 drives nascent protein synthesis and sustains anabolic remodeling in plaque-associated microglia.\u003c/p\u003e\n\u003cp\u003eOur proteomic data further suggested that loss of TREM2 compromises microglial proteostasis and phagocytic processing. To test this directly, we examined the intracellular accumulation of phagocytic cargo. Confocal imaging of the postsynaptic marker PSD95, together with Iba1 and A\u0026beta;, revealed a substantial buildup of PSD95 puncta within Trem2\u003csup\u003eKO\u003c/sup\u003e microglia near plaques (Fig. 4B-C). These inclusions were surrounded by CD68+ lysosomes (Extended Fig. 4A), indicating stalled degradation of engulfed synaptic material. The presynaptic marker vGluT1 similarly accumulated in Trem2\u003csup\u003eKO\u0026nbsp;\u003c/sup\u003emicroglia (Extended Fig. 4B, C). This defective clearance extended beyond synaptic debris; staining for Myelin Basic Protein (MBP) showed a striking increase in intracellular myelin accumulation in Trem2\u003csup\u003eKO\u003c/sup\u003e microglia at plaque sites (Fig. 4D-E). The simultaneous buildup of synaptic and myelin proteins, despite the presence of lysosomal markers, indicates a general failure of phagocytic degradation when TREM2-dependent anabolic support is absent. These results suggest that TREM2-dependent anabolic remodeling is essential for maintaining effective phagocytic processing in plaque-associated microglia.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u003cu\u003eTREM2-dependent anabolic adaptation enables microglial resilience to A\u0026beta;-induced phagocytic stress.\u003c/u\u003e\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eBecause our earlier findings suggested that TREM2-dependent anabolic remodeling supports microglial proteostasis, we next sought to determine how this program shapes transcriptional states under increasing phagocytic stress. To this end, we profiled the transcriptomes of A\u0026beta;-engaged microglia using Methoxy-X04 (X04), a fluorescent probe that selectively labels fibrillar A\u0026beta; (Fig. 5A). Primary microglia from\u0026nbsp;\u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e and\u0026nbsp;\u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e;\u003cem\u003eTrem2\u003c/em\u003e\u003csup\u003eKO\u003c/sup\u003e mice were sorted into X04⁺\u0026nbsp;(A\u0026beta;-phagocytic) and X04⁻\u0026nbsp;(non-phagocytic) populations. To capture the full spectrum of A\u0026beta; load, X04⁺microglia were further stratified into low-, medium-, and high-intensity groups (Fig. 5B).\u0026nbsp;At early stages of pathology (5-6 months), the proportion of X04⁺microglia\u0026nbsp;was comparable between genotypes, indicating that TREM2 is not required for the initial engulfment of A\u0026beta; fibrils. By the mid-stage (9-10 months), however, TREM2 deficiency markedly reduced the total X04⁺population. This reduction was driven largely by the specific loss of the highly phagocytic X04\u003csup\u003ehigh\u003c/sup\u003e\u003csup\u003e\u0026nbsp;\u003c/sup\u003esubset (Fig. 5B-C). Because X04 intensity reflects cumulative A\u0026beta; uptake and processing, these findings suggest that TREM2 is essential for \u003cem\u003esustaining\u003c/em\u003e phagocytic function and survival under chronic A\u0026beta; burden, rather than for the initiation of phagocytosis.\u003c/p\u003e\n\u003cp\u003eTranscriptomic profiling of X04⁺\u0026nbsp;and X04⁻\u0026nbsp;microglia from 9-month-old wild-type,\u0026nbsp;\u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e, and\u0026nbsp;\u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e;\u003cem\u003eTrem2\u003c/em\u003e\u003csup\u003eKO\u003c/sup\u003e mice revealed clear genotype-specific clustering by principal-component analysis (PCA) (Fig. 5D). Within each genotype, samples also aligned along the X04-intensity gradient, indicating a shift of transcriptional states proportional to A\u0026beta;-induced stress (Fig. 5D; Extended Fig. 5A-B). Differential expression analysis between X04⁺\u0026nbsp;and X04⁻\u0026nbsp;microglia in\u0026nbsp;\u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e mice revealed a strong induction of mitochondrial pathways, including \u003cem\u003eOxidative Phosphorylation, Electron Transport Chain,\u0026nbsp;\u003c/em\u003eand\u003cem\u003e\u0026nbsp;Mitochondrial Translation\u003c/em\u003e (Fig. 5E, red arrows), as well as protein translational programs, such as \u003cem\u003eSRP-dependent protein targeting, Translation Initiation, and Translation Termination\u003c/em\u003e (Fig. 5E, orange arrows). These results indicate that A\u0026beta; phagocytosis elicits a coordinated anabolic program that couples protein synthesis with mitochondrial biogenesis to meet the energetic and biosynthetic demands of phagocytic activity. This adaptive transcriptional program was completely abolished in TREM2-deficient microglia (Extended Fig. 5C).\u0026nbsp;Analysis across the X04 subpopulations (X04⁻, low, medium, high) revealed distinct kinetics: anabolic gene induction peaked in X04\u003csup\u003elow\u003c/sup\u003e microglia and declined progressively as phagocytic burden increased (Fig. 5F). In contrast, pathways associated with the disease-associated microglia (DAM) signature, such as Neutrophil Degranulation and MHC II Antigen Presentation, increased linearly with X04 intensity. Notably, classical inflammatory pathways (TNF, IFN-\u0026gamma;) were suppressed or unchanged in X04⁺microglia, demonstrating that anabolic activation represents a metabolically adaptive, non-inflammatory response. Loss of TREM2 largely abolished this anabolic response in X04\u003csup\u003elow\u003c/sup\u003e and X04\u003csup\u003emedium\u003c/sup\u003e microglia (Fig. 5G). As A\u0026beta; load increased, Trem2\u003csup\u003eKO\u003c/sup\u003e microglia displayed reduced oxidative phosphorylation and mitochondrial translation, accompanied by increased signatures of mitochondrial dysfunction, a compensatory upregulation of glycolysis, and heightened pro-inflammatory signaling (Fig. 5G). Thus, across the X04 gradient, TREM2 loss had minimal effects on non-phagocytic microglia but was indispensable for inducing anabolic adaptation and mitochondrial biogenesis in phagocytic populations (Fig. 5H).\u003c/p\u003e\n\u003cp\u003eTogether, these results demonstrate that microglial adaptation to A\u0026beta;-induced phagocytic stress depends on a TREM2-driven anabolic program that integrates protein synthesis and mitochondrial biogenesis. In the absence of TREM2, this adaptive network collapses, leading to metabolic failure and the loss of microglial resilience under chronic A\u0026beta; stress.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u003cu\u003eTREM2 deficiency impairs mitochondrial renewal, driving proteostatic failure and the extrusion of neurotoxic exophers\u003c/u\u003e\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo determine whether the anabolic defect caused by TREM2 deficiency translates to a functional breakdown of bioenergetics \u003cem\u003ein vivo\u003c/em\u003e, we assessed mitochondrial dynamics using high-dimensional flow cytometry. We reasoned that if TREM2-deficient microglia fail to mount the anabolic program required for mitochondrial biogenesis, they would be unable to maintain a healthy organelle pool under the phagocytic stress of amyloid pathology. Freshly isolated microglia were stained with MitoTracker Green (measuring total mitochondrial mass) and MitoTracker Red CMXRos (measuring membrane potential). In \u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u0026nbsp;\u003c/sup\u003emice, the CD45\u003csup\u003ehigh\u003c/sup\u003e (plaque-associated) population exhibited a synchronized elevation of both mitochondrial membrane potential and lysosomal mass, suggesting that resilient microglia boost their bioenergetic capacity to meet the high metabolic demands of A\u0026beta; clearance (Extended Fig. 6A, B). However, Trem2\u003csup\u003eKO\u0026nbsp;\u003c/sup\u003emicroglia displayed a profound bioenergetic uncoupling: while total mitochondrial mass was significantly increased compared to controls, these organelles exhibited significantly reduced membrane potential. Consequently, the Mitochondrial Health Index (calculated as the ratio of respiratory-active mitochondria to total mass) was markedly lower in Trem2\u003csup\u003eKO\u0026nbsp;\u003c/sup\u003emicroglia (Fig. 6A), indicating a severe accumulation of metabolically incompetent organelles. Notably, lysosomal mass (LysoTracker) remained unchanged between genotypes (Fig. 6B), suggesting that the accumulation of depolarized mitochondria in Trem2\u003csup\u003eKO\u003c/sup\u003e microglia unlikely stems from lysosomal deficiency.\u003c/p\u003e\n\u003cp\u003eWe hypothesized that this defect arose from a failure in mitochondrial renewal, particularly in the distal processes where phagocytosis actively occurs. To definitively test for impaired biogenesis, we combined metabolic labeling with a proximity ligation assay (PLA). Mice were injected with the amino acid analog AHA (L-azidohomoalanine) to tag nascent proteins, followed by click-mediated biotinylation and detection of newly synthesized TOM20 using an anti-biotin/anti-TOM20 PLA antibody pair. In \u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u0026nbsp;\u003c/sup\u003emice, plaque-associated microglia displayed a significant increase in nascent mitochondrial signal (TOM20-PLA\u003csup\u003e+\u003c/sup\u003e) specifically within IBA1\u003csup\u003e+\u003c/sup\u003e regions near plaques (Fig. 6C, D), indicating robust biogenesis in response to amyloid deposition. In contrast, Trem2\u003csup\u003eKO\u003c/sup\u003e microglia failed to increase (TOM20-PLA\u003csup\u003e+\u003c/sup\u003e) signals, demonstrating that TREM2 is required to boost mitochondrial biogenesis. Furthermore, co-staining of FUNCAT and TOM20 revealed colocalization of nascent protein signal (FUNCAT) with mitochondria (TOM20) within microglial processes (Fig. 6E, Extended Fig. 6C). \u0026nbsp;This spatial pattern correlated with the induction of TREM2 expression in the processes of plaque-associated microglia (Extended Fig. 6D), suggesting that TREM2 drives local protein synthesis and mitochondrial biogenesis directly at the site of phagocytic activity. In contrast, Trem2\u003csup\u003eKO\u003c/sup\u003e microglia around the plaques failed to recruit mitochondria to their processes (Fig. 6E).\u003c/p\u003e\n\u003cp\u003eDespite the reduction in \u003cem\u003enew\u003c/em\u003e biogenesis, quantification of confocal images revealed that Trem2\u003csup\u003eKO\u003c/sup\u003e microglia around the plaques exhibited increased \u003cem\u003etotal\u003c/em\u003e mitochondrial mass\u0026nbsp;(% of TOM20+ voxel in Iba1+ area)\u0026nbsp;(Fig. 6F, G), corroborating the flow cytometry data. Upon further image analysis of brain sections from \u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e;\u003cem\u003eTrem2\u003c/em\u003e\u003csup\u003eKO\u003c/sup\u003e mice,\u0026nbsp;we observed a striking phenomenon: the presence of numerous large (2-8 \u0026micro;m), spherical Iba1\u003csup\u003e+\u003c/sup\u003e/DAPI\u003csup\u003e-\u003c/sup\u003e vesicles within the plaque niche that were filled with mitochondria (TOM20+) (Fig. 6F). This morphology - large, membrane-bound vesicles packed with mitochondria - bore a strong resemblance to exophers, a specialized extrusion mechanism previously reported in neurons and cardiomyocytes to eject damaged mitochondria and protein aggregates under proteostatic stress. Staining for the outer mitochondrial membrane protein VDAC1 confirmed that these vesicles were densely packed with mitochondria (Extended Fig. 6E, F). Crucially, co-staining with the autophagy receptor p62/SQSTM1 revealed that the VDAC1 in these exophers displayed high levels of p62 (Extended Fig. 6G), indicating that these vesicles are selectively enriched with damaged mitochondria that failed to undergo mitophagy.\u003c/p\u003e\n\u003cp\u003ePrompted by this resemblance to exophaer, we performed high-resolution 3D reconstruction from multiple confocal image datasets. We found that these vesicles mostly emerged directly from microglial processes, appearing as segmental enlargements connected by thin membrane nanotubes (Fig. 7A, D; Extended Fig. 7A-D). To rigorously rule out that these structures are derived from apoptotic fragmentation, we stained for apoptotic marker cleaved caspase-3 (CC3). We validated our CC3 antibody using postnatal day 7 (P7) mouse brains, which undergo significant programmed cell death. While P7 neurons showed clear CC3 signal (Extended Fig. 7E), these exopher-like vehicles and adjacent microglia in \u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e;\u003cem\u003eTrem2\u003c/em\u003e\u003csup\u003eKO\u003c/sup\u003e brains were consistently CC3-negative (Extended Fig. 7F), confirming they represent a genuine non-apoptotic extrusion event. In addition to damaged mitochondria (VDAC1\u003csup\u003e+\u003c/sup\u003e;p62\u003csup\u003ehigh\u003c/sup\u003e), the exophers were packed with A\u0026beta;, undigested synaptic (PSD95\u003csup\u003e+\u003c/sup\u003e), and myelin (MBP\u003csup\u003e+\u003c/sup\u003e) debris (Fig. 7B, E), consistent with their function as emergency \u0026quot;trash bags\u0026quot; for material the cell could not degrade. Quantitative analysis revealed a striking disparity in the frequency of these extrusion events. Within a standardized plaque-associated volume (10\u003csup\u003e5\u003c/sup\u003e um\u003csup\u003e3\u003c/sup\u003e), we detected an average of 7-8 PSD95\u003csup\u003e+\u003c/sup\u003e or MBP\u003csup\u003e+\u003c/sup\u003e exophers in AKTK mice, compared to less than 0.5 in AK controls (Fig. 7C, F). When normalized to the number of plaque-associated microglia, we found that 60-70% of Trem2\u003csup\u003eKO\u003c/sup\u003e microglia in the plaque niche were associated with exopher extrusion, compared to less than 2% of WT microglia. This suggests that exophergenesis is not a rare anomaly, but a dominant phenotype of metabolic failure in the absence of TREM2. \u0026nbsp;Importantly, a subset of these exophers contained hyperphosphorylated Tau (AT8\u003csup\u003e+\u003c/sup\u003e) (Fig. 7G, H). This finding suggests that exopher extrusion may effectively package pathological Tau into concentrated, extracellular seeds, facilitating the conversion and spreading of Tau pathology by releasing aggregation-prone material back into the parenchyma. Supporting this notion, we observed a marked increase in AT8\u003csup\u003e+\u003c/sup\u003e area around A\u0026beta; plaques in the cortex of AKTK brains compared to controls, spatially correlating with regions of active exopher extrusion (Fig. 7I, J). Together, these data demonstrate that TREM2-dependent mitochondrial renewal is essential for microglial proteostasis and resilience against A\u0026beta;, and its failure leads to the extrusion of neurotoxic exophers that may actively drive Tau pathology.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u003cu\u003eThe Anabolic Adaptation signature defines phagocytic competence better than the canonical DAM signature\u003c/u\u003e\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e.\u0026nbsp;\u003c/strong\u003eTo place anabolic adaptation within the broader landscape of microglial activation, we performed single-cell RNA sequencing on X04⁺\u0026nbsp;(phagocytic) and X04⁻\u0026nbsp;microglia\u0026nbsp;(non-phagocytic) microglia isolated from 9-10 month-old \u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e mice. Clustering analysis identified the expected spectrum of transcriptional state (Sala Frigerio et al., 2019; Mancuso et al., 2024), including Homeostatic (HM), Activated Response (ARM), and Transition Response (TRM) microglia (Extended Fig. 8A-B). However, projecting the canonical Disease-Associated Microglia (DAM) signature onto this map revealed a critical discrepancy: DAM genes were expressed broadly across both X04⁺and X04⁻populations (Extended Fig. 8C) \u003csup\u003e15\u003c/sup\u003e. This suggests that the acquisition of the transcriptional DAM signature is insufficient to distinguish microglia that are functionally engaged in phagocytosis from those that are merely responsive to pathological environment.\u003c/p\u003e\n\u003cp\u003eTo resolve the specific molecular determinants of phagocytic competence, we interrogated the spatial heterogeneity within these clusters. UMAP visualization revealed that X04⁺and X04⁻\u0026nbsp;microglia occupied distinct regions\u0026nbsp;within the nominally defined HM and TRM clusters (Fig. 8A). Increasing the clustering resolution resolved this heterogeneity, splitting HM into subclusters 0 (Non-phagocytic enriched) and 2 (Phagocytic enriched), and TRM into subclusters 6 (Non-phagocytic) and 1 (Phagocytic) (Fig. 8B-C). We then explored which transcriptional programs effectively discriminated these functional states. \u0026quot;DAM Module Scores\u0026quot; were similar between phagocytic (Cluster 1, 2) and non-phagocytic (Cluster 6, 0) pairs (Extended Fig. 8D), confirming that the DAM signature correlates poorly with phagocytic states. In contrast, differential expression analysis between Cluster 1, 2 and Cluster 6, 0 revealed that the phagocytic state was defined by a robust, synchronized upregulation of genes involved in ribosome biogenesis, protein synthesis, and mitochondrial oxidative phosphorylation\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e(Fig. 8D, Extended Fig. 8E). GSEA confirmed the enrichment of these pathways in the cellular components (Fig. 8E; Extended Fig. 8F), which mirrored the proteomic \u0026quot;Anabolic Adaptation\u0026quot; signature identified earlier. This program was featured by the specific induction of rate-limiting translational initiators (\u003cem\u003eEif3f\u003c/em\u003e) and structural ribosomal subunits (\u003cem\u003eRps5\u003c/em\u003e, \u003cem\u003eRpl23\u003c/em\u003e), coupled with respiratory chain subunits spanning the input (\u003cem\u003eNdufa1\u003c/em\u003e), catalytic core (\u003cem\u003eCox4i1\u003c/em\u003e), and ATP-generating output (\u003cem\u003eAtp5e\u003c/em\u003e) of the electron transport chain (Fig. 8F). Crucially, this anabolic induction appeared distinct from inflammatory activation. We observed that neuroinflammatory signaling pathways were reduced in the phagocytic clusters (1 and 2) compared to their non-phagocytic counterparts (6 and 0) (Fig. 8D, Extended Fig. 8E), supporting the view that anabolic adaptation represents a functional metabolic engagement rather than a generic inflammatory state.\u003c/p\u003e\n\u003cp\u003eFinally, we spatially validated the relationship between the DAM signature and anabolic activities \u003cem\u003ein situ\u003c/em\u003e. We co-stained brain sections from 9-month-old \u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u0026nbsp;\u003c/sup\u003emice for Iba1, the canonical DAM marker CD74, and the nascent protein label FUNCAT. While most plaque-associated (Iba1\u003csup\u003ehigh\u003c/sup\u003e) microglia exhibited robust anabolic activity (high FUNCAT signal), CD74 expression was only induced in a subset of these microglia. In addition, CD74 intensity showed no correlation with FUNCAT signal within Iba1\u003csup\u003e+\u0026nbsp;\u003c/sup\u003emicroglia (Pearson\u0026apos;s R = -0.12, P=0.187; Extended Fig. 8G, H). Collectively, these data demonstrate that the anabolic adaptation signature better define the phagocytic microglial state, distinguishing the functional capacity to engage with stress from the broader transcriptional response to A\u0026beta; stress.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eMicroglial activation in Alzheimer\u0026rsquo;s disease has canonically been defined by the transition to a Disease-Associated Microglia (DAM) state. Because this transcriptional signature includes the upregulation of phagocytic receptors (e.g., Clec7a, Axl, MerTK), the field has largely operated under the assumption that the DAM phenotype is synonymous with enhanced phagocytic function. Our findings challenge this assumption. We demonstrate that the DAM signature and phagocytic competence are not inextricably linked. Instead, we identify a distinct TREM2-dependent anabolic adaptation, characterized by enhanced protein synthesis and mitochondrial biogenesis, as the true driver of functional resilience. Our data suggest a fundamental reinterpretation of microglial states: the DAM signature likely represents a cellular stress response to pathology, whereas the anabolic program represents the functional capacity to resolve it. While TREM2 deficiency is known to impair the full acquisition of the DAM signature, we show that the functional impotence of these cells is not merely due to a lack of ability to phagocytosis. Rather, it stems from a bioenergetic collapse: without TREM2, microglia fail to mount the \u0026apos;metabolic engine\u0026apos; required to sustain activity. This distinction explains why purely transcriptional definitions of microglial states often fail to predict functional outcomes, as the stress response (DAM) and the capacity to act (Anabolism) are molecularly distinct programs.\u003c/p\u003e\n\u003cp\u003eA central finding of our study is that the containment of A\u0026beta; pathology (catabolism) strictly requires an accompanying anabolic surge. In peripheral macrophage biology, metabolic states are often dichotomized: pro-inflammatory (M1-like) activation is typically driven by aerobic glycolysis (the Warburg effect), while anti-inflammatory (M2-like) states rely on oxidative phosphorylation (OXPHOS)\u0026nbsp;\u003csup\u003e16\u003c/sup\u003e \u003csup\u003e13\u003c/sup\u003e. Our data challenge this binary framework in the context of AD. We reveal that plaque-associated microglia adopt a unique \u0026quot;hyper-metabolic\u0026quot; phenotype that simultaneously engages both biomass synthesis (anabolism) and mitochondrial respiration. This state mirrors the \u0026quot;effector expansion\u0026quot; observed in T-cell biology, where rapid proliferation and cytokine production demand a massive, mTOR-driven upregulation of translational capacity and mitochondrial mass\u0026nbsp;\u003csup\u003e17\u003c/sup\u003e \u003csup\u003e18\u003c/sup\u003e. For microglia, the functional burden is proteostatic: the continuous internalization of A\u0026beta; fibrils and cellular debris consumes lysosomal enzymes, membranes, and ATP at a rate that homeostatic synthesis cannot match. We propose that the extensive proteome remodeling observed in \u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e microglia, specifically the upregulation of ribosomal and mitochondrial biogenesis, represents a compensatory maneuver to maintain \u0026quot;proteostatic bandwidth.\u0026quot; The cell must physically construct new degradative machinery to replace what is expended. Consequently, the accumulation of synaptic and myelin debris we observed in Trem2\u003csup\u003eKO\u003c/sup\u003e microglia is likely not a primary defect in recognition or uptake, but a failure of renewal. Without the TREM2-driven anabolic engine to replenish mitochondria, the cells simply exhaust the machinery required to digest their cargo.\u003c/p\u003e\n\u003cp\u003eOur results identify a failure of mitochondrial renewal as the primary driver of microglial dysfunction in the absence of TREM2. While seminal studies have characterized the TREM2-deficient state as a general \u0026quot;metabolic collapse\u0026quot; or loss of energetic fitness\u0026nbsp;\u003csup\u003e14\u003c/sup\u003e, our single-cell metabolic profiling provides a more granular mechanism: the decoupling of organelle mass from function. We observe that Trem2\u003csup\u003eKO\u003c/sup\u003e microglia exhibit a paradoxical \u0026quot;High Mass / Low Potential\u0026quot; phenotype. We propose that this stems fundamentally from an inability to renew the mitochondrial network, forcing the cell into a state of metabolic stagnation. Under chronic phagocytic stress, microglia must maintain a high rate of organelle turnover, constantly synthesizing new mitochondria to replace those damaged by ROS or consumed during lysosomal fusion. This mirrors the metabolic reprogramming seen in activated T cells, where mitochondrial biogenesis is a prerequisite for sustained effector function\u0026nbsp;\u003csup\u003e17\u003c/sup\u003e. Our in vivo AHA-PLA data confirm that TREM2 is the driver of this essential biogenic surge. In the absence of this anabolic signal, the cycle of renewal arrests. We posit that without the capacity to synthesize replacements, Trem2\u003csup\u003eKO\u003c/sup\u003e microglia are forced to retain aging, damaged mitochondria to maintain basal viability. Because the cell cannot \u0026quot;afford\u0026quot; to degrade its existing power sources (mitophagy) if it cannot replace them (biogenesis), these \u0026quot;damaged\u0026quot; mitochondria accumulate, leading to the observed increase in total mass. However, because they are depolarized and structurally compromised, they offer diminishing bioenergetic returns and likely contribute to oxidative stress \u003csup\u003e19\u003c/sup\u003e. Thus, the metabolic defect in Trem2\u003csup\u003eKO\u003c/sup\u003e microglia is not merely a failure to clear waste (catabolism), but a failure to build the machinery (anabolism) that allows clearance to happen.\u003c/p\u003e\n\u003cp\u003ePerhaps the most striking finding of our study is the identification of exopher-like structures emerging from metabolically compromised microglia. Originally characterized in C. elegans neurons and recently described in mammalian cardiomyocytes, exophers represent a conserved, primordial mechanism for ejecting aggregation-prone proteins and damaged organelles when intracellular degradation pathways are overwhelmed\u0026nbsp;\u003csup\u003e20\u003c/sup\u003e \u003csup\u003e21\u003c/sup\u003e. To our knowledge, our study provides one of the first lines of evidence for this phenomenon in CNS microglia in situ. We propose that exophergenesis acts as a \u0026apos;proteostatic emergency valve\u0026apos; for TREM2-deficient microglia. Our data suggest that when the anabolic support required for lysosomal digestion fails, the cell is left with a fatal accumulation of undigested myelin, amyloid, and \u0026apos;damaged\u0026apos; mitochondria. Unable to degrade this toxic burden internally, the cell physically ejects it to preserve its own viability. This aligns with findings in other phagocytic contexts where cells jettison indigestible cargo to avoid programmed cell death\u0026nbsp;\u003csup\u003e22\u003c/sup\u003e. However, this survival maneuver likely comes at a high cost to the neural environment. By sequestering proteopathic seeds but failing to digest them, these cells effectively \u0026quot;re-package\u0026quot; aggregates into a mobile, bioactive form. We propose that the \u0026quot;metabolic uncoupling\u0026quot; observed in Trem2 variants transforms microglia from guardians into \u0026quot;Trojan horses,\u0026quot; absorbing toxic seeds only to regurgitate them into the parenchyma. This mechanism offers a novel, active explanation for the \u0026apos;diffuse plaque\u0026apos; and accelerated tau pathology consistently observed in TREM2-deficient AD patients\u0026nbsp;\u003csup\u003e23\u003c/sup\u003e \u003csup\u003e24\u003c/sup\u003e. This exopher-mediated expulsion may represent a novel, non-neuronal mechanism for the prion-like spreading of A\u0026beta; and Tau pathology, particularly in the context of aging or genetic risk where microglial metabolism is waned. Determining whether these microglial exophers contribute to the trans-synaptic seeding of tau pathology or the physical disruption of neural circuits in human AD tissue represents a critical frontier for future investigation.\u003c/p\u003e\n\u003cp\u003eOur single-cell RNA seq analysis on phagocytic (X04+) and non-phagocytic (X04-) also challenges the prevailing dogma that the DAM transcriptional signature is synonymous with functional microglial competence. Since the initial characterization of the DAM/MGnD state\u0026nbsp;\u003csup\u003e6\u003c/sup\u003e \u003csup\u003e7\u003c/sup\u003e, the upregulation of sensing receptors (e.g., Clec7a, Axl, MerTK) has been widely interpreted as a proxy for enhanced phagocytic activity. Our data reveal a critical distinction between these two programs. We observed that the DAM module score correlates linearly with phagocytic load (X04 intensity) even in cells that are metabolically failing. This suggests that the DAM signature functions primarily as a cellular stress response, a transcriptional \u0026quot;cry for help\u0026quot; triggered by proteotoxic burden, rather than a guarantee of functional execution. In the absence of TREM2-driven anabolism, microglia can mount this transcriptional sensing response (demand) but fail to back it up with the biosynthetic machinery (supply) required to act.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFinally, our findings offer critical insights into the optimization of current anti-amyloid immunotherapies. The recent clinical success of monoclonal antibodies against A\u0026beta; (e.g., lecanemab, donanemab) has validated amyloid clearance as a primary therapeutic target\u0026nbsp;\u003csup\u003e25\u003c/sup\u003e \u003csup\u003e26\u003c/sup\u003e. However, the efficacy of these therapies relies heavily on the underlying functional capacity of microglia to engage in Fc-receptor-mediated phagocytosis, a process we now identify as imposing a formidable bioenergetic cost on the cell. Our data suggest a potential vulnerability in this approach: pharmacologically driving microglia toward a hyper-phagocytic phenotype without ensuring adequate anabolic support may be counterproductive. If the \u0026apos;demand\u0026apos; for clearance (induced by antibody tagging) exceeds the cell\u0026apos;s \u0026apos;supply\u0026apos; of metabolic machinery (anabolic capacity), this metabolic mismatch could accelerate exhaustion and drive the extrusion of neurotoxic exophers. We propose that future therapeutic regimens should adopt a combinatorial approach. Interventions that specifically bolster the anabolic axis, for example, via TREM2 agonists, mitochondrial co-factors, or amino acid supplementation, which could provide the necessary fuel to sustain the immunotherapeutic response. With this combinatorial approach, we may be able to extend the therapeutic window during which microglia can effectively clear A\u0026beta; without succumbing to proteostatic collapse.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cstrong\u003eAnimals\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eApp\u003csup\u003eNL-G-F\u003c/sup\u003e\u003c/em\u003e mice mice were kindly provided by Dr. Takaomi Saido (RIKEN Center for Brain Science)\u0026nbsp;\u003csup\u003e27\u003c/sup\u003e. Trem2 knockout mice (Trem2\u003csup\u003eKO\u003c/sup\u003e) were obtained from The Jackson Laboratory (Strain #027197). To generate the double-mutant cohort, \u003cem\u003eApp\u003csup\u003eNL-GF\u003c/sup\u003e\u003c/em\u003e mice were crossed with Trem2\u003csup\u003eKO\u003c/sup\u003e mice to produce generate \u003cem\u003eApp\u003csup\u003eNL-G-F\u003c/sup\u003e; Trem2\u003csup\u003eKO\u003c/sup\u003e\u003c/em\u003e homozygous for both manipulations. Mice were housed in a pathogen-free barrier facility with a standard 12-hour light/dark cycle and ad libitum access to food and water. Housing density was maintained at maximum 5 mice per cage. Both male and female mice were used for all experiments. All animal procedures were conducted in strict accordance with the National Institutes of Health (NIH) guidelines and approved by The Ohio State University Institutional Animal Care and Use Committee (IACUC).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIn Vivo Metabolic Labeling (FUNCAT)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;For labeling of nascent proteins, mice were injected intraperitoneally (i.p.) with 50 mg/kg Azidohomoalanine (AHA; Click Chemistry Tools) dissolved in sterile PBS. Animals were sacrificed 16 hours post-injection. For Flow Cytometry: Dissociated cells were reacted with Alexa Fluor 647-Alkyne (Thermo Fisher Scientific) using the Click-iT Cell Reaction Buffer Kit (Thermo Fisher Scientific, Cat #C10269) according to the manufacturer\u0026rsquo;s instructions. For Proximity Ligation Assay (PLA): Free-floating brain sections were reacted with Biotin-Alkyne (Thermo Fisher Scientific) using the Click-iT Cell Reaction Buffer Kit prior to antibody incubation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMicroglial Isolation and Sorting\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAdult microglia were isolated using magnetic-activated cell sorting (MACS) as described previously\u0026nbsp;\u003csup\u003e28\u003c/sup\u003e. Briefly, mice were anesthetized and perfused transcardially with ice-cold PBS to remove circulating leukocytes. Brains were dissected, chilled on ice, and dissociated using the Neural Tissue Dissociation Kit (P) (Miltenyi Biotec, #130-107-677) and the gentleMACS Tissue Dissociator. Cell suspensions were filtered through a 70 \u0026micro;m cell strainer and centrifuged at 300 \u0026times; g for 10 minutes. Myelin was depleted using Myelin Removal Beads II (Miltenyi Biotec, #130-096-733) via magnetic separation. The resulting single-cell suspension was used immediately for flow cytometry or further enriched for transcriptomic analysis using CD11b MicroBeads (Miltenyi Biotec, #130-049-601) and LS columns.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEvaluation of Phagocytic Microglia\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo evaluate amyloid load, Methoxy-X04 (X04; Tocris Bioscience, #4920) was prepared by dissolving the compound in DMSO, followed by dilution in a 1:1 mixture of propylene glycol and PBS to obtain a stable yellowish-green emulsion. The solution was prepared freshly and injected i.p. at 10 mg/kg 16 hours prior to tissue harvest. For flow cytometric analysis, dissociated cells were incubated with Fixable Viability Dye eFluor 780 (1:1000, eBioscience) and anti-CD16/32 (Fc Block, 1:200, clone 2.4G2) to exclude dead cells and prevent non-specific binding. Cells were subsequently stained with fluorophore-conjugated antibodies against CD11b (1:500, clone M1/70, Invitrogen), CD45 (1:100, clone 30-F11, Invitrogen), and CX3CR1 (1:50, R\u0026amp;D Systems). Live microglia (CD45\u003csup\u003eint\u003c/sup\u003e; CD11b\u003csup\u003e+\u003c/sup\u003e; CX3CR1\u003csup\u003e+\u003c/sup\u003e) were gated as X04\u003csup\u003e+\u003c/sup\u003e or X04\u003csup\u003e-\u0026nbsp;\u003c/sup\u003ebased on fluorescence in the DAPI excitation channel (405 nm). Wild-type animals injected with Methoxy-X04 served as biological negative controls to define gating thresholds.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIn Situ Proximity Ligation Assay (PLA)\u003c/strong\u003e: To visualize nascent mitochondrial proteins in situ, free-floating brain sections from AHA-injected mice were first subjected to the click reaction with Biotin-Alkyne as described above. Sections were extensively washed and incubated overnight at 4 \u0026deg;C with rabbit anti-TOM20 (1:500, 11802-1-AP, Proteintech) and mouse anti-Biotin (1:500, 1D4-C5, BioLegend). PLA probes (Duolink In Situ, Sigma-Aldrich) were applied, and ligation and amplification steps were performed according to the manufacturer\u0026apos;s protocol.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBulk RNA-seq and Bioinformatics Analysis\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cu\u003eLibrary Preparation and Sequencing\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eTotal RNA from primary microglia was extracted using Quick-RNA miniprep (R1055, Zymo Research). RNA quality was evaluated by TapeStation using high Sensitivity RNA ScreenTape (5067-5579, Agilent). RNA samples with RNA integrity numbers greater than 8 were used for cDNA library construction. RNA seq libraries were prepared using SMART Seq\u0026reg; mRNA LP Kit (Takara Bio) following the manufacturer\u0026rsquo;s instructions. The qualities of the cDNA library were assessed using TapeStation using High Sensitivity D5000 ScreenTape (5067-5592, Agilent). cDNA library samples were then pooled and sequenced with the HiSeq 4000 System (Illumina) by AZENTA life sciences.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cu\u003eRaw Reads Preprocessing and Sequence Alignment\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eDemultiplexed FASTQ files of bulk RNA sequencing data were aligned to the mouse genome (Mus_musculus.GRCm39) using STAR (version 2.7.10a)\u0026nbsp;\u003csup\u003e29\u003c/sup\u003e \u003csup\u003e30\u003c/sup\u003e \u003csup\u003e31\u003c/sup\u003e. Adapters were trimmed using Flexbar (version 3.5.0.) \u003csup\u003e32\u003c/sup\u003e. Reads mapped to genomic features were counted using featureCounts (version 2.0.3)\u0026nbsp;\u003csup\u003e33\u003c/sup\u003e. The count matrix was imported in R (version 4.3.3) for analysis.\u003c/p\u003e\n\u003cp\u003e\u003cu\u003eDifferential Expression and Principal Component Analysis (PCA)\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eDifferential gene expression analysis was performed using the DESeq2 package (version 1.42.1) \u003csup\u003e34\u003c/sup\u003e in R. Raw count data were filtered to remove genes with low library representation (total counts \u0026lt; 10). For basic comparisons, a single-factor generalized linear model was used. In experiments involving Batch/Sex/Time, a multi-factor model was implemented to control for confounding variables. DESeq2\u0026rsquo;s internal median-of-ratios method was used for normalization. To control the false discovery rate (FDR), p-values were adjusted using the Benjamini-Hochberg procedure. Genes were considered significantly differentially expressed if they reached an adjusted p-value \u0026lt; 0.05. Differential expression results were visualized using volcano plots generated by the EnhancedVolcano package (version 1.20.0)\u0026nbsp;\u003csup\u003e35\u003c/sup\u003e, with significance defined as an adjusted p-value \u0026lt; 0.05.\u003c/p\u003e\n\u003cp\u003ePCA was performed on the transformed counts extracted from DESeq2.\u0026nbsp;If a multi-factor design was used for DEA to measure the effect of the genotypes controlling for batch differences, the PCA was plotted with batch variation removed by using the\u0026nbsp;removeBatchEffect()\u0026nbsp;function from\u0026nbsp;limma\u003cem\u003e\u0026nbsp;\u003c/em\u003e(version 3.54.2)\u0026nbsp;\u003csup\u003e36\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunctional Enrichment and Pathway Analysis\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cu\u003eIPA Core Analysis\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eCore Canonical Pathway Analysis was performed by QIAGEN\u0026rsquo;s Ingenuity\u0026reg; (IPA\u0026reg;, QIAGEN Redwood City,\u0026nbsp;\u003ca href=\"http://www.qiagen.com/ingenuity\"\u003ewww.qiagen.com/ingenuity\u003c/a\u003e). Complete lists of DEGs and DAPs, along with their log2 fold change expression values and FDR were inputted into IPA for identifying canonical pathways, biological functions, and upstream regulators using a cutoff of FDR \u0026lt; 0.05. The p-value of overlap, calculated using the right-tailed Fischer\u0026rsquo;s Exact Test with a statistical threshold of 0.05, is used to indicate the probability of association of molecules from test dataset with the canonical pathway by random chance alone. A positive or negative regulation z-score value indicates that a function is predicted to be activated or inhibited. No activity prediction by IPA results in ineligible z-score which is represented by grey bars.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cu\u003eIPA Comparison Analysis\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eTo identify biological pathways modulated across the different XO4 subgroups of microglia within specific genotype, differentially expressed genes (DEGs) from the Low, Medium, and High groups from each genotype--each compared against a common baseline control of XO4- dataset--were first processed through individual Core Analyses in IPA. Subsequently, the results of contrast within each genotype were integrated using the IPA Comparison Analysis platform to juxtapose the canonical pathway profiles across all groups of the specific genotype. Findings were visualized as a heatmap where color intensity represents the activation z-score (grey indicates an unpredictable directional trend). To denote statistical confidence, pathways failing to reach the significance threshold (e.g., Fisher\u0026rsquo;s exact test p-value \u0026gt; 0.05) were marked with an asterisk (*). Pathways were further grouped and annotated by their broader functional categories to identify overarching biological themes.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cu\u003eGene Set Enrichment Analysis (GSEA)\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eUsing the R package clusterProfiler (version 4.11.0.2\u0026nbsp;\u003csup\u003e37\u003c/sup\u003e, genes or proteins were ranked by values of log2 Fold Change and \u0026minus;log10(p-value)\u0026nbsp;\u0026lowast;\u0026nbsp;sign(log2FoldChange) respectively to form the ranking metrics for transcriptomic and proteomic datasets. Enrichment was conducted against the Gene Ontology (GO) database, specifically targeting Biological Process (BP) and Cellular Component (CC) categories. Results were considered statistically significant at an adjusted p-value \u0026lt; 0.05. Results were visualized using dot plots, where selected GO terms were segregated by their direction of regulation (activated vs. suppressed) based on the Normalized Enrichment Score (NES). The color scale represents the Benjamini-Hochberg adjusted p-value (q-value) to indicate statistical significance, while the size of each dot corresponds to the gene count (the number of genes from the dataset coregulated within a specific GO term).\u003c/p\u003e\n\u003cp\u003e\u003cu\u003eGene-concept Network\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eTo reveal the molecule-level information associated with the significant pathways of interest, we constructed\u0026nbsp;gene-concept networks using the\u0026nbsp;cnetplot()\u0026nbsp;function within the\u0026nbsp;clusterProfiler\u0026nbsp;package in R.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eQuantitative Proteomic Analysis:\u003c/strong\u003e Brain samples were lysed using 5% SDS and 50mM TEAB buffer, sonicated, cleaned, and quantified via BCA assay. Proteomic analysis was conducted by BGI Genomics Co., Ltd. Briefly, samples were then prepared using the STrap Midi MS sample prep device (Protifi), with each sample containing 2000\u0026mu;g of protein. This preparation involved reduction with dithiothreitol (DTT), alkylation with iodoacetamide (IAM), quenching of the IAM reaction with DTT, and overnight digestion with Trypsin/Lys-C within the STrap device. Peptides were eluted from the STrap, with 60\u0026mu;g per sample dried via SpeedVac and reconstituted in 50% acetonitrile for tandem mass tag (TMT) labeling in 50mM TEAB (pH 8.5). After TMT labeling, samples were pooled, acidified with 1% formic acid, and analyzed for label check on a nano LC-MS/MS system. Following successful label checks, samples were dried, reconstituted in 2% formic acid, desalted using EVOLUTE\u0026reg; EXPRESS ABN (Biotage), and fractionated via offline HPLC into 12 fractions. These fractions were analyzed by LC-MS/MS after being reconstituted with mobile phase A; approximately 5% of each was injected using the TMT method. TMT quantification and identity discovery were performed using Proteome Discoverer 2.5 (Thermo Fisher). False discovery rate (FDR) was calculated based on The Benjamini-Hochberg Procedure. FDR \u0026lt;= 0.05 was considered to be significant.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eHistological analysis\u003c/strong\u003e: \u0026nbsp;Brains were sectioned on a cryostat at 40-mm thickness. For immunofluorescence staining, free-floating sections were blocked with PBS containing 10% normal goat serum (NGS) at room temperature for 30 minutes, incubated with primary antibody in blocking solution (PBS with 1% NGS) at 4\u0026deg;C for 24-48 hours, and then incubated with secondary antibody at room temperature for 2 hours. Sections were mounted on slides with ProLong Diamond (Life Technologies). Images were captured on a ZEISS Axio Observer and/or the Nikon AXR point scanning confocal microscope. 2D Image quantification was performed using ImageJ software. Auto Threshold methods \u0026ldquo;Otsu\u0026rdquo; or \u0026ldquo;Triangle\u0026rdquo; were used to define the region of interest (ROI). Statistical analyses were conducted using a two-tailed unpaired t-test or one-way ANOVA.\u003c/p\u003e\n\u003cp\u003ePrimary antibodies used in this study are: Human Amyloid \u0026beta; (N) (82E1, IBL Co., LTD.), Iba1 (019-19741, or 011-27991 from Wako Co), Phos-RPS6 (4858S, Cell Signaling), , PSD95 sdAb (N3702-AF568-L, Synaptic Systems), VGLUT1 sdAb (N1602-At488-L, Synaptic Systems), TOM20 (11802-1-AP, Proteintech), VDAC1 (CL488-10866, Proteintech), MBP (78896, Cell Signaling), Cleaved Caspase-3 (Asp175) (9661, Cell Signaling), CD74 (151002, Biolegend). All secondary antibodies were purchased from ThermoFisher or Jackson Immunoresearch.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eImage Acquisition and Statistical Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cu\u003e2D Epifluorescence Microscopy and Analysis\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eTwo-dimensional epifluorescence imaging was performed using a\u0026nbsp;ZEISS Axio Observer. This modality was employed for analyses where axial depth was not a primary variable. Image quantification was conducted using\u0026nbsp;ImageJ software (NIH)\u0026nbsp;\u003csup\u003e38\u003c/sup\u003e, focusing on\u0026nbsp;aggregate metrics per field of view. To ensure unbiased measurement, regions of interest (ROIs) were defined using automated thresholding algorithms (\u0026apos;Otsu\u0026apos; or \u0026apos;Triangle\u0026apos;, depending on signal-to-noise characteristics). Statistical comparisons between experimental groups were performed using either a\u0026nbsp;two-tailed unpaired t-test\u0026nbsp;or\u0026nbsp;one-way ANOVA, as appropriate, based on the number of groups and data distribution.\u003c/p\u003e\n\u003cp\u003e\u003cu\u003e3D Confocal Imaging and Hierarchical Statistical Modeling\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eHigh-resolution three-dimensional image stacks were acquired using\u0026nbsp;the Nikon AXR point scanning confocal microscope\u0026nbsp;to quantify the morphological and biochemical properties of\u0026nbsp;Iba1-positive (Iba1+) regions of interest (ROIs). Due to the high density and clustering of microglia in the App\u003csup\u003eNL-G-F\u003c/sup\u003e mice, individual ROIs were defined as distinct morphological units rather than individual cells.\u003c/p\u003e\n\u003cp\u003eAll 3D reconstructions and quantitative analyses were performed using Imaris (v11.0; Oxford Instruments, Zurich, Switzerland), a multidimensional image analysis software. Iba1+ surfaces were generated using a machine learning-based thresholding algorithm to ensure objective and consistent segmentation of complex microglial morphologies. To quantify the volumetric colocalization of target markers within Iba1+ ROIs, marker surfaces were generated via manually defined thresholds, allowing for the correction of staining-specific background noise. Volumes were converted to voxel counts prior to calculating the overlapping voxel ratio and performing model fitting. To assess protein expression levels, the mean fluorescence intensity of the specific target marker channel was extracted directly from the reconstructed Iba1+ surface.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo assess the impact of AD pathology, Iba1+\u0026nbsp;ROIs were spatially categorized based on their interaction with amyloid pathology. An ROI was classified as\u0026nbsp;Plaque-Associated Microglia (PAM)\u0026nbsp;if any portion of the\u0026nbsp;Iba1+\u0026nbsp;surface, including processes or the soma, was in direct contact with a plaque cluster. All other ROIs were designated as\u0026nbsp;Non-Plaque-Associated Microglia (NPAM).\u003c/p\u003e\n\u003cp\u003eTo account for the hierarchical structure of the 3D data (multiple ROIs nested within individual imaging fields), we employed a Generalized Linear Mixed Model (GLMM) framework, allowing us to treat the animal or image stack as a random effect, thereby controlling for intra-subject variation and ensuring a more accurate estimation of the fixed effects of the genotypes, Plaque Status (PAM vs. NPAM), and their interaction term.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAll statistical analyses were conducted in\u0026nbsp;R\u0026nbsp;using the\u0026nbsp;glmmTMB\u0026nbsp;\u003csup\u003e39\u003c/sup\u003e and emmeans\u0026nbsp;\u003csup\u003e40\u003c/sup\u003e packages.\u0026nbsp;The choice of statistical model was tailored to the distribution and mathematical constraints of each quantified metric.\u0026nbsp;For the\u0026nbsp;volumetric occupancy\u0026nbsp;of target markers within\u0026nbsp;Iba1+\u0026nbsp;ROIs, a\u0026nbsp;beta-binomial GLMM (logit link)\u0026nbsp;was selected to account for the\u0026nbsp;bounded nature of proportional data\u0026nbsp;and to adjust for\u0026nbsp;overdispersion derived from biological variability. Continuous measurements, such as mean intensities, were modeled using a\u0026nbsp;Gamma distribution (log link).\u003c/p\u003e\n\u003cp\u003eModel fit was rigorously assessed using the\u0026nbsp;DHARMa package\u0026nbsp;\u003csup\u003e41\u003c/sup\u003e, utilizing a simulation-based approach to verify distribution assumptions, dispersion, and residual patterns. Estimated Marginal Means (EMMs) and 95% confidence intervals (CIs) were back-transformed from the link scale to the response scale for reporting. Pairwise comparisons were performed as planned contrasts between genotypes and plaque conditions.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cu\u003eExopher Density Quantification and Non-Parametric Analysis\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eFollowing 3D confocal acquisition as described above, we performed a targeted quantification of microglial\u0026nbsp;exophers.\u0026nbsp;Exopher density was quantified via a dual-blind system:\u0026nbsp;A\u0026beta;\u0026nbsp;plaque areas were manually delineated by an analyst blinded to the channels of exopher markers (PSD95, MBP, and AT8), and\u0026nbsp;exophers were subsequently counted within these regions by a second independent analyst.\u003c/p\u003e\n\u003cp\u003eExopher densities were calculated as the number of exophers per\u0026nbsp;10\u003csup\u003e5\u003c/sup\u003e \u0026mu;m\u003csup\u003e3\u003c/sup\u003e of plaque volume to ensure human-readable scaling. Due to the distinct biological nature of the groups\u0026mdash;where the control (AK) genotype exhibited \u0026quot;structural zeros\u0026quot; (near-total absence of symptoms) and the treatment (AKTK) genotype showed a robust, high-variance phenotype\u0026mdash;standard count-based Generalized Linear Mixed Models (GLMMs) failed to reach mathematical convergence. To address the complete separation and non-normal distribution of the data, a non-parametric Wilcoxon rank-sum test with exact permutation and rank transformation using the R package coin (version 1.4.3)\u0026nbsp;\u003csup\u003e42\u003c/sup\u003e was employed to account for the high frequency of tied zeros in the control group. Exopher density distributions were visualized in violin plots, with jittered points indicating individual plaque ROIs.\u0026nbsp;Medians and interquartile ranges (IQRs) are indicated by dots and bars, respectively. Statistical significance is reported as p values based on the exact permutation test.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSingle-cell Seq and Data Processing\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cu\u003eLibrary Construction\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eSingle cell RNA-seq libraries were generated using the 10x Genomics Chromium NEXT GEM Single Cell 3\u0026rsquo; Reagent Kit. Briefly, primary microglia isolated from adult mice were loaded onto chromium chips with a capture target of 10,000 cells per sample. Libraries were prepared following the provided protocol and sequenced on an Illumina NovaSeq with a targeted sequencing depth of 50,000-100,000 reads per cell. FASTQ files from sequencing were then used as inputs to the 10X Genomics Cell Ranger pipeline.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cu\u003eRead Processing, Quality Control and Filtering\u0026nbsp;\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eGene expression matrices were generated with the Cell Ranger Pipeline (v7.0.0;\u0026nbsp;10x Genomics)\u0026nbsp;and aligned to the Mouse (mm10) reference transcriptome. The resulting digital gene expression matrix was filtered, normalized, and clustered using R version 4.2.0 and Seurat version 4.1.1\u0026nbsp;\u003csup\u003e43\u003c/sup\u003e. Genes that are expressed in less than 10 cells, and cells with greater than 5% of reads mapped to mitochondrial genes, or with less than 1500 features and 3000 UMIs were removed.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cu\u003eInitial Normalization and Doublet Removal\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eWe performed an initial normalization of post-QC dataset in Seurat to stabilize variance and did not regress out variation associated with percent.mito or percent.rb due to the metabolic relevance mitochondrial and ribosomal gene expression features, despite observed differences between these fractions. DoubletFinder version 2.0.3\u0026nbsp;\u003csup\u003e44\u003c/sup\u003e was used to identify false-negative Demuxlet classifications caused by doublets formed from cells with identical SNP profiles, and an average of 10% of cells per sample were confidently predicted as doublets and removed. In total, 26,096 cells were identified as putative singlets and retained for downstream analysis.\u003c/p\u003e\n\u003cp\u003e\u003cu\u003eSCTransform Normalization, Integration, and Clustering\u0026nbsp;\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eThe gene expression matrix was normalized and scaled using the Seurat function SCTransform which also identifies the most variable genes, of which the top 3,000 were used for dimensionality reduction. Four samples were integrated to correct for any potential library batch effect by using the Seurat functions FindIntegrationAnchors and IntegrateData based on reciprocal PCA with n = 5 neighbors (k.anchor). Integrated matrix was used for downstream analysis. Cells were clustered using the Louvain algorithm based on the first 20 principal components with a resolution of 0.3. The Uniform Manifold Approximation and Projection (UMAP)\u0026nbsp;\u003csup\u003e45\u003c/sup\u003e was used for non-linear reduction and two-dimensional data visualization.\u003c/p\u003e\n\u003cp\u003e\u003cu\u003eCluster Annotation, Subsetting, and Reclustering\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eCell-type annotations were assigned to each cluster based on two levels of evidence. First, the Seurat function FindAllMarkers was used to identify cluster marker genes based on one-versus-all Wilcoxon rank sum differential expression tests for each cluster. Second, cell-type identities were predicted by comparing transcriptomic profiles to a curated panel of marker genes derived from a previously published single-cell RNA sequencing dataset of the \u003cem\u003eApp\u003csup\u003eNL-G-F\u003c/sup\u003e\u003c/em\u003e mouse brain\u0026nbsp;\u003csup\u003e46\u003c/sup\u003e. Based on this classification, we retained 25,776 cells identified as putative microglia for downstream analyses, while non-microglial cell types were excluded. For reclustering putative microglia, we applied SCTransform normalization, recomputed PCA and used the top 20 PCs for dimensionality reduction by UMAP, followed by unbiased clustering using the Seurat function FindNeighbors with the resolution of granularity set to 0.2. This led to the identification of 7 clusters each representing a microglial state defined by unique or transitory profiles.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cu\u003eDifferential Gene Expression Analysis\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eDifferentially expressed genes of specific cell states were found by applying the Seurat function FindAllMarkers for overall DE and FindMarkers for side-by-side comparisons. Genes with adjusted p values (using a Bonferroni correction) \u0026lt; 0.05 were considered significantly differentially expressed. Canonical Pathways Analysis by IPA was used to test for gene sets enriched in DE genes.\u003c/p\u003e\n\u003cp\u003e\u003cu\u003eGene Module Scoring and Signature Analysis\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eTo compare the transcriptomic profile of our clusters with previously described microglial states, we calculated module scores using the AddModuleScore function in Seurat. Signatures were defined based on published marker lists for homeostatic microglia (HM) markers (Tmem119, P2ry12, Cx3cr1), activated response microglia (ARM) markers (Apoe, Cst7, Itgax, Lpl, Spp1, Gpnmb, Dkk2, Cd74, H2-Aa, H2-Ab1), transiting response microglia (TRM) markers (Apoe, Cst7, Itgax, Cd74, H2-Aa, H2-Ab1), interferon response microglia (IRM) markers (Ifit2, Ifit3, Ifitm3, Oasl2, and Irf7), cycling and proliferating microglia (CPM) markers (Top2a, Mcm2, Tubb5, Mki67, Cdk1)\u0026nbsp;\u003csup\u003e47\u003c/sup\u003e, a Ribosomal Microglia signature (Tpt1, Rps3a, Rpl13, Rps23)\u0026nbsp;\u003csup\u003e48\u003c/sup\u003e, and Disease-associated Mciroglia (DAM) markers (Cd9, Apoe, Trem2, Tyrobp, Cd63, Lgals3, Axl, Spp1, Cstb, Ctsd, Lpl, Itgax, B2m, Cst7, Gpnmb, Igf1, Irf8, Fth1, Lyz2, Ccl3, Ccl6, Timp2) in \u003cem\u003eApp\u003c/em\u003e\u003csup\u003eNL-G-F\u003c/sup\u003e mice as curated in\u0026nbsp;\u003csup\u003e15\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cu\u003eAssessment of Signature-level Enrichment across Cell States\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eTo assess the distribution of gene expression across the identified transcriptomic states in our dataset (resolution 0.3), we generated stacked violin plots using the Seurat package. This visualization displays the log-normalized expression levels of representative marker genes, which were grouped along the y-axis into modules corresponding to curated microglial signatures and canonical pathways identified via Ingenuity Pathway Analysis (IPA). Individual cell clusters, denoted by their index numbers, are arrayed along the x-axis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMulti-omics Integration and Comparative Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo elucidate the regulatory landscape across different molecular layers, we performed a comparative and integrative analysis of the transcriptomic (RNA-seq) and quantitative proteomic profiles from primary microglia isolated from 9-month-old and 2-month-old \u003cem\u003eApp\u003csup\u003eNL-G-F\u003c/sup\u003e\u003c/em\u003e\u003csup\u003e\u0026nbsp;\u0026nbsp;\u003c/sup\u003eand \u003cem\u003eAPP\u003csup\u003eNL-G-F\u003c/sup\u003e;Trem2\u003csup\u003eKO\u003c/sup\u003e\u003c/em\u003e mice.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cu\u003eDirect Profile Comparison\u0026nbsp;\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eWe first assessed the direct correspondence between identified transcripts and proteins. Initial comparison was conducted by mapping identified transcripts to their corresponding proteins. A\u0026nbsp;Venn analysis\u0026nbsp;was utilized to determine the overlap between the two datasets, identifying molecules consistently regulated at both levels as well as those uniquely detected within a single \u0026quot;omic\u0026quot; layer. The resulting intersections were visualized as a Venn diagram using the\u0026nbsp;VennDiagram package (version 1.7.3)\u0026nbsp;\u003csup\u003e49\u003c/sup\u003e in R.\u003c/p\u003e\n\u003cp\u003e\u003cu\u003eIntegrative Pathway Analysis\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eTo identify biological processes consistently active across both molecular layers, we employed\u0026nbsp;ActivePathways, an integrative method that uses directionality and significance estimates of molecules to identify significantly enriched pathways by combining evidence from multiple omic sources\u0026nbsp;\u003csup\u003e50\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eIntegration metrics\u003c/em\u003e. P-values and log2 fold-change (log2FC) values derived from the 9-month-old versus 2-month-old comparisons were processed using the ActivePathways framework. Statistical evidence from the transcriptomic and proteomic datasets was integrated via Data-driven P-value Merging (DPM), with Brown\u0026rsquo;s method utilized as a robust reference for determining combined significance. To identify biologically convergent profiles, a weighted constraint vector [mRNA=1,protein=1] was applied to prioritize genes exhibiting direct, concordant associations between the two molecular layers. Conversely, genes with conflicting directional signals or those failing to meet the integration criteria were penalized, ensuring the final pathway enrichment was driven by consistent cross-omic evidence. The relationship between the two molecular layers was visualized using a concordance scatter plot.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eIntegrative pathway enrichment analysis.\u003c/em\u003e Functional enrichment was performed using the ActivePathways R package to identify biological processes significantly represented across the integrated datasets. The analysis utilized a gene list ranked by merged P-values derived from directional data integration. To determine optimal enrichment of Gene Ontology (GO) terms, a ranked hypergeometric test was applied. The gene set collection (m5.go.v2023.2.Mm.symbols.gmt) was filtered to include only pathways containing between 10 and 500 annotated genes in order to minimize biases from excessively specific or overly generic terms. Significant GO terms were defined using a Holm Family-Wise Error Rate (FWER) \u0026lt; 0.05. To facilitate biological interpretation, unique and significant GO terms were visualized via bar charts with the x-axis representing the \u0026minus;log10(adjusted P-values) and the y-axis listing the specific GO terms which were further grouped by their broader functional categories.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistical Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eStatistical analyses were performed using GraphPad Prism 9 software (v9.5.1; GraphPad Software, San Diego, CA, USA). Data are presented as mean \u0026plusmn; standard error of the mean (SEM). Pairwise comparisons were analyzed using two-tailed unpaired t-tests. Multiple comparisons were analyzed using one-way or two-way analysis of variance (ANOVA) followed by Tukey\u0026rsquo;s or Sidak\u0026rsquo;s post-hoc tests. A P value \u0026lt; 0.05 was considered statistically significant.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eHansen, D. V., Hanson, J. E. \u0026amp; Sheng, M. Microglia in Alzheimer\u0026rsquo;s disease. \u003cem\u003eJ Cell Biol\u003c/em\u003e \u003cstrong\u003e217\u003c/strong\u003e, 459\u0026ndash;472 (2018).\u003c/li\u003e\n\u003cli\u003eLambert, J.-C. \u003cem\u003eet al.\u003c/em\u003e Meta-analysis of 74,046 individuals identifies 11 new susceptibility loci for Alzheimer\u0026rsquo;s disease. \u003cem\u003eNature genetics\u003c/em\u003e \u003cstrong\u003e45\u003c/strong\u003e, 1452\u0026ndash;1458 (2013).\u003c/li\u003e\n\u003cli\u003eBellenguez, C. \u003cem\u003eet al.\u003c/em\u003e New insights into the genetic etiology of Alzheimer\u0026rsquo;s disease and related dementias. \u003cem\u003eNat Genet\u003c/em\u003e (2022).\u003c/li\u003e\n\u003cli\u003eStreit, W. J., Braak, H., Xue, Q.-S. \u0026amp; Bechmann, I. Dystrophic (senescent) rather than activated microglial cells are associated with tau pathology and likely precede neurodegeneration in Alzheimer\u0026rsquo;s disease. \u003cem\u003eActa neuropathologica\u003c/em\u003e \u003cstrong\u003e118\u003c/strong\u003e, 475\u0026ndash;485 (2009).\u003c/li\u003e\n\u003cli\u003eNavarro, V. \u003cem\u003eet al.\u003c/em\u003e Microglia in Alzheimer\u0026rsquo;s disease: activated, dysfunctional or degenerative. \u003cem\u003eFrontiers in aging neuroscience\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 140 (2018).\u003c/li\u003e\n\u003cli\u003eKeren-Shaul, H. \u003cem\u003eet al.\u003c/em\u003e A Unique Microglia Type Associated with Restricting Development of Alzheimer\u0026rsquo;s Disease. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e169\u003c/strong\u003e, 1276\u0026ndash;1290.e17 (2017).\u003c/li\u003e\n\u003cli\u003eKrasemann, S. \u003cem\u003eet al.\u003c/em\u003e The TREM2-APOE Pathway Drives the Transcriptional Phenotype of Dysfunctional Microglia in Neurodegenerative Diseases. \u003cem\u003eImmunity\u003c/em\u003e \u003cstrong\u003e47\u003c/strong\u003e, 566\u0026ndash;581.e9 (2017).\u003c/li\u003e\n\u003cli\u003eGuerreiro, R. \u003cem\u003eet al.\u003c/em\u003e TREM2 variants in Alzheimer\u0026rsquo;s disease. \u003cem\u003eNew England Journal of Medicine\u003c/em\u003e \u003cstrong\u003e368\u003c/strong\u003e, 117\u0026ndash;127 (2013).\u003c/li\u003e\n\u003cli\u003eJonsson, T. \u003cem\u003eet al.\u003c/em\u003e Variant of TREM2 associated with the risk of Alzheimer\u0026rsquo;s disease. \u003cem\u003eNew England Journal of Medicine\u003c/em\u003e \u003cstrong\u003e368\u003c/strong\u003e, 107\u0026ndash;116 (2013).\u003c/li\u003e\n\u003cli\u003eWang, Y. \u003cem\u003eet al.\u003c/em\u003e TREM2 Lipid Sensing Sustains the Microglial Response in an Alzheimer\u0026rsquo;s Disease Model. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e160\u003c/strong\u003e, 1061\u0026ndash;1071 (2015).\u003c/li\u003e\n\u003cli\u003eYuan, P. \u003cem\u003eet al.\u003c/em\u003e TREM2 haplodeficiency in mice and humans impairs the microglia barrier function leading to decreased amyloid compaction and severe axonal dystrophy. \u003cem\u003eNeuron\u003c/em\u003e \u003cstrong\u003e90\u003c/strong\u003e, 724\u0026ndash;739 (2016).\u003c/li\u003e\n\u003cli\u003eO\u0026rsquo;Neill, L. A. J., Kishton, R. J. \u0026amp; Rathmell, J. A guide to immunometabolism for immunologists. \u003cem\u003eNature Reviews Immunology\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 553\u0026ndash;565 (2016).\u003c/li\u003e\n\u003cli\u003ePearce, E. L. \u0026amp; Pearce, E. J. Metabolic pathways in immune cell activation and quiescence. \u003cem\u003eImmunity\u003c/em\u003e \u003cstrong\u003e38\u003c/strong\u003e, 633\u0026ndash;643 (2013).\u003c/li\u003e\n\u003cli\u003eUlland, T. K. \u003cem\u003eet al.\u003c/em\u003e TREM2 Maintains Microglial Metabolic Fitness in Alzheimer\u0026rsquo;s Disease. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e170\u003c/strong\u003e, 649\u0026ndash;663.e13 (2017).\u003c/li\u003e\n\u003cli\u003eChen, Y. \u0026amp; Colonna, M. Microglia in Alzheimer\u0026rsquo;s disease at single-cell level. Are there common patterns in humans and mice? \u003cem\u003eJ Exp Med\u003c/em\u003e \u003cstrong\u003e218\u003c/strong\u003e, e20202717 (2021).\u003c/li\u003e\n\u003cli\u003eO\u0026rsquo;Neill, L. A. J., Kishton, R. J. \u0026amp; Rathmell, J. A guide to immunometabolism for immunologists. \u003cem\u003eNature Reviews Immunology\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 553\u0026ndash;565 (2016).\u003c/li\u003e\n\u003cli\u003eRon-Harel, N. \u003cem\u003eet al.\u003c/em\u003e Mitochondrial biogenesis and proteome remodeling promote one-carbon metabolism for T cell activation. \u003cem\u003eCell metabolism\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, 104\u0026ndash;117 (2016).\u003c/li\u003e\n\u003cli\u003eChi, H. Regulation and function of mTOR signalling in T cell fate decisions. \u003cem\u003eNature reviews immunology\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 325\u0026ndash;338 (2012).\u003c/li\u003e\n\u003cli\u003ePalikaras, K., Lionaki, E. \u0026amp; Tavernarakis, N. Coordination of mitophagy and mitochondrial biogenesis during ageing in C. elegans. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e521\u003c/strong\u003e, 525\u0026ndash;528 (2015).\u003c/li\u003e\n\u003cli\u003eMelentijevic, I. \u003cem\u003eet al.\u003c/em\u003e C. elegans neurons jettison protein aggregates and mitochondria under neurotoxic stress. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e542\u003c/strong\u003e, 367\u0026ndash;371 (2017).\u003c/li\u003e\n\u003cli\u003eNicol\u0026aacute;s-\u0026Aacute;vila, J. A. \u003cem\u003eet al.\u003c/em\u003e A network of macrophages supports mitochondrial homeostasis in the heart. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e183\u003c/strong\u003e, 94\u0026ndash;109. e23 (2020).\u003c/li\u003e\n\u003cli\u003eAlvarez, M. \u0026amp; Casadevall, A. Phagosome extrusion and host-cell survival after Cryptococcus neoformans phagocytosis by macrophages. \u003cem\u003eCurrent Biology\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 2161\u0026ndash;2165 (2006).\u003c/li\u003e\n\u003cli\u003eMcQuade, A. \u003cem\u003eet al.\u003c/em\u003e Gene expression and functional deficits underlie TREM2-knockout microglia responses in human models of Alzheimer\u0026rsquo;s disease. \u003cem\u003eNature communications\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 1\u0026ndash;17 (2020).\u003c/li\u003e\n\u003cli\u003eMeilandt, W. J. \u003cem\u003eet al.\u003c/em\u003e Trem2 Deletion Reduces Late-Stage Amyloid Plaque Accumulation, Elevates the A\u0026beta;42:A\u0026beta;40 Ratio, and Exacerbates Axonal Dystrophy and Dendritic Spine Loss in the PS2APP Alzheimer\u0026rsquo;s Mouse Model. \u003cem\u003eJ Neurosci\u003c/em\u003e \u003cstrong\u003e40\u003c/strong\u003e, 1956\u0026ndash;1974 (2020).\u003c/li\u003e\n\u003cli\u003evan Dyck, C. H. \u003cem\u003eet al.\u003c/em\u003e Lecanemab in Early Alzheimer\u0026rsquo;s Disease. \u003cem\u003eN Engl J Med\u003c/em\u003e \u003cstrong\u003e388\u003c/strong\u003e, 9\u0026ndash;21 (2023).\u003c/li\u003e\n\u003cli\u003eSevigny, J. \u003cem\u003eet al.\u003c/em\u003e The antibody aducanumab reduces A\u0026beta; plaques in Alzheimer\u0026rsquo;s disease. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e537\u003c/strong\u003e, 50\u0026ndash;56 (2016).\u003c/li\u003e\n\u003cli\u003eSaito, T. \u003cem\u003eet al.\u003c/em\u003e Single App knock-in mouse models of Alzheimer\u0026rsquo;s disease. \u003cem\u003eNat Neurosci\u003c/em\u003e \u003cstrong\u003e17\u003c/strong\u003e, 661\u0026ndash;663 (2014).\u003c/li\u003e\n\u003cli\u003eLin, D. \u003cem\u003eet al.\u003c/em\u003e Arachidonic Acid Mobilization and Peroxidation Promote Microglial Dysfunction in A\u0026beta; Pathology. \u003cem\u003eJ Neurosci\u003c/em\u003e \u003cstrong\u003e44\u003c/strong\u003e, e0202242024 (2024).\u003c/li\u003e\n\u003cli\u003eDobin, A. \u003cem\u003eet al.\u003c/em\u003e STAR: ultrafast universal RNA-seq aligner. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e29\u003c/strong\u003e, 15\u0026ndash;21 (2013).\u003c/li\u003e\n\u003cli\u003eDobin, A. \u0026amp; Gingeras, T. R. Mapping RNA-seq Reads with STAR. \u003cem\u003eCurr Protoc Bioinformatics\u003c/em\u003e \u003cstrong\u003e51\u003c/strong\u003e, 11.14.1\u0026ndash;19 (2015).\u003c/li\u003e\n\u003cli\u003eDobin, A. \u0026amp; Gingeras, T. R. Optimizing RNA-Seq Mapping with STAR. \u003cem\u003eMethods Mol Biol\u003c/em\u003e \u003cstrong\u003e1415\u003c/strong\u003e, 245\u0026ndash;262 (2016).\u003c/li\u003e\n\u003cli\u003eRoehr, J. T., Dieterich, C. \u0026amp; Reinert, K. Flexbar 3.0\u0026ndash;SIMD and multicore parallelization. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e33\u003c/strong\u003e, 2941\u0026ndash;2942 (2017).\u003c/li\u003e\n\u003cli\u003eLiao, Y., Smyth, G. K. \u0026amp; Shi, W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. \u003cem\u003eBioinformatics\u003c/em\u003e (2014).\u003c/li\u003e\n\u003cli\u003eLove, M. I., Huber, W. \u0026amp; Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. \u003cem\u003eGenome Biol\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 550 (2014).\u003c/li\u003e\n\u003cli\u003eBlighe, K., Rana, S. \u0026amp; Lewis, M. EnhancedVolcano: Publication-ready volcano plots with enhanced colouring and labeling. R package version 1.24. 0. 2024. (2024).\u003c/li\u003e\n\u003cli\u003eRitchie, M. E. \u003cem\u003eet al.\u003c/em\u003e limma powers differential expression analyses for RNA-sequencing and microarray studies. \u003cem\u003eNucleic acids research\u003c/em\u003e \u003cstrong\u003e43\u003c/strong\u003e, e47\u0026ndash;e47 (2015).\u003c/li\u003e\n\u003cli\u003eWu, T. \u003cem\u003eet al.\u003c/em\u003e clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. \u003cem\u003eInnovation (Camb)\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e, 100141 (2021).\u003c/li\u003e\n\u003cli\u003eSchindelin, J. \u003cem\u003eet al.\u003c/em\u003e Fiji: an open-source platform for biological-image analysis. \u003cem\u003eNat Methods\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 676\u0026ndash;682 (2012).\u003c/li\u003e\n\u003cli\u003eBrooks, M. E. \u003cem\u003eet al.\u003c/em\u003e glmmTMB Balances Speed and Flexibility Among Packages for Zero-inflated Generalized Linear Mixed Modeling. \u003cem\u003eThe R Journal\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 378\u0026ndash;400 (2017).\u003c/li\u003e\n\u003cli\u003eRussell, V. L. \u003cem\u003eet al.\u003c/em\u003e emmeans: Estimated Marginal Means, aka Least-Squares Means. (2025).\u003c/li\u003e\n\u003cli\u003eFlorian, H. DHARMa: Residual Diagnostics for Hierarchical (Multi-Level / Mixed) Regression Models. (2025).\u003c/li\u003e\n\u003cli\u003eHothorn, T., Hornik, K., van de Wiel, M. A. \u0026amp; Zeileis, A. Implementing a Class of Permutation Tests: The coin Package. \u003cem\u003eJournal of Statistical Software\u003c/em\u003e \u003cstrong\u003e28\u003c/strong\u003e, 1\u0026ndash;23 (2008).\u003c/li\u003e\n\u003cli\u003eHao, Y. \u003cem\u003eet al.\u003c/em\u003e Integrated analysis of multimodal single-cell data. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e184\u003c/strong\u003e, 3573\u0026ndash;3587.e29 (2021).\u003c/li\u003e\n\u003cli\u003eMcGinnis, C. S., Murrow, L. M. \u0026amp; Gartner, Z. J. DoubletFinder: Doublet Detection in Single-Cell RNA Sequencing Data Using Artificial Nearest Neighbors. \u003cem\u003eCell Syst\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 329\u0026ndash;337.e4 (2019).\u003c/li\u003e\n\u003cli\u003eMcInnes, L., Healy, J. \u0026amp; Melville, J. Umap: Uniform manifold approximation and projection for dimension reduction. \u003cem\u003earXiv preprint arXiv:1802.03426\u003c/em\u003e (2018).\u003c/li\u003e\n\u003cli\u003eFrigerio, C. S. \u003cem\u003eet al.\u003c/em\u003e The major risk factors for Alzheimer\u0026rsquo;s disease: age, sex, and genes modulate the microglia response to A\u0026beta; plaques. \u003cem\u003eCell reports\u003c/em\u003e \u003cstrong\u003e27\u003c/strong\u003e, 1293\u0026ndash;1306. e6 (2019).\u003c/li\u003e\n\u003cli\u003eSala Frigerio, C. \u003cem\u003eet al.\u003c/em\u003e The Major Risk Factors for Alzheimer\u0026rsquo;s Disease: Age, Sex, and Genes Modulate the Microglia Response to A\u0026beta; Plaques. \u003cem\u003eCell Rep\u003c/em\u003e \u003cstrong\u003e27\u003c/strong\u003e, 1293\u0026ndash;1306.e6 (2019).\u003c/li\u003e\n\u003cli\u003eMancuso, R. \u003cem\u003eet al.\u003c/em\u003e Xenografted human microglia display diverse transcriptomic states in response to Alzheimer\u0026rsquo;s disease-related amyloid-\u0026beta; pathology. \u003cem\u003eNat Neurosci\u003c/em\u003e \u003cstrong\u003e27\u003c/strong\u003e, 886\u0026ndash;900 (2024).\u003c/li\u003e\n\u003cli\u003eChen, H. \u0026amp; Boutros, P. C. VennDiagram: a package for the generation of highly-customizable Venn and Euler diagrams in R. \u003cem\u003eBMC Bioinformatics\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 35 (2011).\u003c/li\u003e\n\u003cli\u003eSlobodyanyuk, M. \u003cem\u003eet al.\u003c/em\u003e Directional integration and pathway enrichment analysis for multi-omics data. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 5690 (2024).\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8896508/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8896508/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"The microglial response to Alzheimer’s disease (AD) pathology has canonically been defined by the transcriptional transition to a Disease-Associated Microglia (DAM) state. However, the specific metabolic programs required to fuel the high-demand functions of this reactive state, such as Aβ encapsulation and clearance, remain obscure. Here, we identify a TREM2-dependent anabolic adaptation as the critical driver of microglial resilience to Aβ pathology. By integrating proteomics, transcriptomics, and in vivo metabolic labeling in the AppNL-G-F mouse model, we demonstrate that plaque-associated microglia undergo a synchronized metabolic shift, coupling enhanced protein synthesis with local mitochondrial biogenesis to support the bioenergetic demands of phagocytosis. We show that TREM2 signaling acts as the essential \"metabolic licensor\" for this process, driving anabolic remodeling directly at the site of phagocytic activity. In the absence of TREM2, this adaptive response collapses: microglia fail to renew their metabolic machinery, resulting in a state of bioenergetic exhaustion characterized by the accumulation of depolarized mitochondria. Strikingly, we discover that these metabolically compromised cells utilize exophergenesis – the extrusion of large, cargo-filled vesicles – as a compensatory mechanism to purge undigested synaptic and amyloid debris during proteostatic failure. Furthermore, we find that these extruded exophers contain hyperphosphorylated Tau, identifying a potential non-cell-autonomous mechanism for pathology seeding. Single-cell analysis confirms that this anabolic capacity is functionally distinct from the canonical DAM transcriptional signature. Our findings redefine TREM2 not merely as a pathogen sensor, but as a metabolic regulator that safeguards microglial viability and prevents neurotoxic spreading under chronic proteotoxic stress.","manuscriptTitle":"TREM2 fuels the anabolic adaptation required for microglial resilience in Alzheimer’s disease","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-27 11:37:59","doi":"10.21203/rs.3.rs-8896508/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-communications","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"NCOMMS","sideBox":"Learn more about [Nature Communications](http://www.nature.com/ncomms/)","snPcode":"","submissionUrl":"https://mts-ncomms.nature.com/","title":"Nature Communications","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Communications","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"99820cf0-af44-4411-b8c5-0918a4bb65f1","owner":[],"postedDate":"March 27th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":63761504,"name":"Biological sciences/Neuroscience/Diseases of the nervous system/Alzheimer's disease"},{"id":63761505,"name":"Biological sciences/Neuroscience/Glial biology/Microglia"},{"id":63761506,"name":"Biological sciences/Neuroscience/Molecular neuroscience"}],"tags":[],"updatedAt":"2026-03-27T11:37:59+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-27 11:37:59","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8896508","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8896508","identity":"rs-8896508","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}