APOE4 genotype and aging impair injury-induced microglial behavior in brain slices, including toward Aβ, through P2RY12

Microglia are highly dynamic cells that play a critical role in tissue homeostasis through the surveillance of brain parenchyma and response to cues associated with damage. Aging and APOE4 genotype are the strongest risk factors for Alzheimer’s disease (AD), but how they affect microglial dynamics remains unclear. Using ex vivo confocal microscopy, we analyzed microglial dynamic behaviors in the entorhinal cortex (EC) and hippocampus CA1 of 6-, 12-, and 21-month-old mice APOE3 or APOE4 knock-in mice expressing GFP under the CX3CR1 promoter. To study microglia surveillance, we imaged microglia baseline motility for 20 min and measured the extension and retraction of processes. We found that APOE4 microglia exhibited significantly less brain surveillance (27%) compared to APOE3 microglia in 6-month-old mice; aging exacerbated this deficit. To measure microglia response to damage, we imaged process motility in response to ATP, an injury-associated signal, for 30 min. We found APOE4 microglia extended their processes significantly slower (0.9 µm/min, p < 0.005) than APOE3 microglia (1.1 μm/min) in 6-month-old animals. APOE-associated alterations in microglia motility were observed in 12- and 21-month-old animals, and this effect was exacerbated with aging in APOE4 microglia. We measured protein and mRNA levels of P2RY12, a core microglial receptor required for process movement in response to damage. We found that APOE4 microglia express significantly less P2RY12 receptors compared to APOE3 microglia despite no changes in P2RY12 transcripts. To examine if the effect of APOE4 on the microglial response to ATP also applied to amyloid β (Aβ), we infused locally Hi-Lyte Fluor 555-labeled Aβ in acute brain slices of 6-month-old mice and imaged microglia movement for 2 h. APOE4 microglia showed a significantly slower (p < 0.0001) process movement toward the Aβ, and less Aβ coverage at early time points after Aβ injection. To test whether P2RY12 is involved in process movement in response to Aβ, we treated acute brain slices with a P2RY12 antagonist before Aβ injection; microglial processes no longer migrated towards Aβ. These results provide mechanistic insights into the impact of APOE4 genotype and aging in dynamic microglial behaviors prior to gross Aβ pathology and could help explain how APOE4 brains are more susceptible to AD pathogenesis. Supplementary Information The online version contains supplementary material available at 10.1186/s13024-024-00714-y.


Introduction
Alzheimer's disease (AD) is a neurodegenerative disorder characterized by the accumulation of amyloid β (Aβ) plaques, neurofibrillary tangles formation, and cognitive decline [1].Aging is the strongest risk factor for AD, and the Apolipoprotein E (APOE) ε4 allele is the strongest genetic risk factor [2,3].Genetic studies have identified a large number of other genetic risk factors for AD, several of which are specifically expressed in microglia [4].Microglia have a role in Aβ clearance but can also cause chronic inflammation, facilitating or exacerbating AD pathology [5].Identifying alterations in microglial physiological functions associated with the maintenance of brain homeostasis prior to disease onset may guide toward therapeutic strategies for AD.
Microglia are the resident macrophages of the central nervous system (CNS) that play a pivotal role during neuronal development and maintenance of brain homeostasis [6].In a healthy brain, microglia express homeostatic membrane receptors that detect chemotactic cues, allowing them to monitor the extraneuronal environment and respond to pathological stimuli [6].Several mouse models of neurodegeneration identified a subset of disease-associated microglia (DAM) through single-cell and single-nuclei RNA sequencing [5,[7][8][9].DAM are characterized by the downregulation of key homeostatic genes associated with microglial surveillance and response to damage [10].Among the homeostatic microglial genes downregulated is the purinergic receptor P2RY12, which is a core marker of microglia in the healthy brain.P2RY12 is required for the movement of microglial processes in response to chemotactic cues associated with neuronal damage, such as ATP and laser ablation [11,12].These microglial transcriptional signatures are associated with chronic AD pathology, but it is necessary to determine how risk factors for AD, such as APOE4 and aging, affect microglial function during homeostasis and in response to subtle damages in healthy brains.
Microglia rely heavily on the motility of their processes to monitor CNS parenchyma to clear cellular debris and prune dysfunctional synapses.While fixed brain tissue has been informative in studying how APOE4 affects microglia morphology and interactions with Aβ during AD pathology [5,13], acute slices are essential in the study of microglial function as they preserve microglial phenotypes in near-physiological conditions [14][15][16].
Investigating APOE genotype and age-associated alterations at baseline and in response to injury in a physiological environment could reveal pathogenic mechanisms associated with AD risk factors.In the current study, we crossed APOE knock-in mice (KI) [17,18] with mice expressing GFP under the CX3CR1 promoter, which tags microglia.We used ex-vivo live imaging to measure spontaneous and responsive microglia motility to chemotactic cues associated with neuronal damages, including localized Aβ peptide.We further tested whether APOE4 was associated with the downregulation of key homeostatic microglial receptor P2RY12.These findings show that APOE genotype and aging alter microglia homeostasis and their response to injuries prior to the gross accumulation of AD pathology.

Mice
CX3CR1 GFP/GFP mice (JAX stock No. 005582) on a C57BL/6 J (JAX stock No. 000664) [19] were crossed with APOE3KI (JAX stock No. 029018) or APOE4KI (JAX stock No. 027894) [17] to obtain APOE3KI: CX3CR1 GFP/− and APOE4KI: CX3CR1 GFP/− (referred to in this study simply as APOE3 and APOE4 mice).All animals were housed with littermates and kept on a 12-h light/dark cycle with ad libitum access to chow and water.Male and female APOE3 and APOE4 mice were sacrificed at 6, 12, and 21 months of age (n = 4-5 mice /genotype/sex/age group).Sample sizes are included throughout the methods and in the figure legends.All studies were carried out following the Guide for the Care and Use of Laboratory Animals as adopted by the U.S. National Institute of Health and approved by Georgetown University Animal Care and Use Committee, approval protocol 2016-1160.
Total P2RY12 expression in CA1 was calculated as the fluorescent intensity across the entire image (arbitrary units) using Image J. To measure P2RY12 in microglial processes, we outlined the microglial soma, cropped out that portion of the images, and measured the remaining fluorescent intensity.The percent P2RY12 in processes was calculated as fluorescent intensity from processes over the total fluorescent intensity (processes plus soma).We analyzed 2 images from 2 brain slices for each animal (n = 4 animals/genotype).Experimenters were blinded to the group type during image analysis.

Confocal imaging
Confocal Z-and ZT-stacks were taken with a laser scanning microscope system (Thor Imaging System Division) equipped with 488/561/642 nm laser and Green/ Red/Far-red filters and mounted on an upright Elipse FN1 microscope (Nikon Instruments).284 × 284 x 20 μm (xyz) volumes of hippocampal slices containing the CA1 and cortical slices containing the entorhinal cortex (EC) were imaged through the 60 × water-dipping objective (CFI Fluor 60XW, NA = 1.0,WD = 2 mm, Nikon).Differential interference contrast images (on acute or fixed slices) were used to identify and confirm the region of interest as CA1 or EC.

Microglial density and morphology
For microglia density and morphology quantification, 2048 × 2048-pixel Z-stacks, 20 planes 0.5 μm apart, were taken from 30 μm fixed slices.We analyzed the maximal intensity projections across the z-axis.For microglia density, counts of GFP-positive cell bodies were calculated using Image J from single 284 × 284 x 20 μm fields containing CA1 or EC per hemisection.Microglial morphology was quantified as described [22].Briefly, fluorescent photomicrographs were transformed into gray-scale images and binarized by setting thresholds.Binarized microglia were then skeletonized, and the endpoint boxels (number of branches) and total branch length (total length of microglia) per individual microglia were measured.We analyzed 2 images from 2 brain slices for each animal (n = 4 animals/genotype/sex/age group).Experimenters were blinded to group type during image analysis.

Microglial process motility at baseline and in response to ATP
For baseline and responsive microglia motility experiments, 1024 × 1024 pixel ZT-stack images from 11 planes 1.5 μm apart were taken every 20 s.When necessary, time lapses were stabilized using the StackReg plugin [23] in Image J [24].Baseline motility was imaged for 20 min in both the CA1 and EC of naïve slices, or in these regions in the presence of an aCSF-containing pipette (0 mM ATP, the control condition for responsive motility experiments, tip resistance 3-5 m Ω).Time-lapse images were manually cropped, then thresholded and binarized in the region of interest (ROI) (adapted from [25]).Binary ROI from each timepoint t(x) were color-merged with a t(x + 1) ROI, generating images containing red pixels representing retracting structures, green pixels representing extending structures, and yellow pixels representing static structures.The numbers of each colored pixel were counted using the Color Counter plugin (Wayne Rasband in ImageJ.For each pair of frames (t(x) and t(x + 1), we calculated the motility index (number of green pixels + number of red pixels/number of yellow pixels).The motility index of each paired frame (e.g., t0 + t1; t1 + t2; t2 + t3) was averaged for the whole 20-min time-lapse.
For reactive microglia motility, a patch pipette containing 1, 3, or 10 mM ATP in aCSF was lowered into CA1 or EC 30 μm deep; the ATP was allowed to diffuse into the tissue without pressure in the pipette.The surrounding volume was imaged for 30 min.ATP concentration and imaging time were chosen based on previous studies [19].We quantified microglial process velocity by using the Manual Tracking plugin (Fabrice Cordelières) in ImageJ.Between 3 and 5 processes per microglia were manually tracked moving towards the ATP-containing pipette.If a process reached the pipet tip prior to 30 min, its velocity was only calculated until that time.To determine that process motility was due to microglia responding to ATP and not to the pipette piercing the tissue, a pipette containing 0 mM ATP (aCSF only) was lowered, and the surrounding volume was imaged.No directional motility was elicited to nearby microglia by this control.For all time-lapse analysis, we analyzed one time-lapse image from 1-2 brain slices for each animal (n = 4-6 animals/ genotype/sex/age group).Experimenters were blinded to group identity during image analysis.

Aβ-42 preparation and injection
HiLyte Fluor 555-labeled Aβ-42 (Cat# AS-60480-01, Anaspec) peptide was used in the experiments related to microglial processes responses to Aβ. HiLyte Fluor 555-labeled Aβ-42 (0.1 mg) was diluted in 1% NH 4 OH to obtain a 1 mg/ml peptide solution.The Aβ-42 solution was diluted in aCSF to 100 μM aliquots and stored at -20 °C for subsequent experiments.The same concentration of fluorescein amidites (5FAM) labeled scrambled Aβ-42 (Cat# AS-60892, Anaspec) was prepared and used as a negative control.The protocol for Aβ injection in acute brain slices was adapted from [26] and [27].Briefly, 5 μl of Aβ-42 was filled into a glass patch pipette (tip resistance 2-3 mΩ) connected with transparent tubing to a 3 ml syringe.The loaded glass patch pipette was carefully lowered into the EC and maneuvered into the center of the field, then the plunger of the syringe was moved slowly from the 3 ml to 1 ml position (in about 10 s).To avoid bleed-through, 2048 × 2048-pixel Z-stacks, 20 planes 0.5 μm apart, were acquired in separate channels every 15 min for 2 h.Time-lapse images were manually thresholded, binarized, and color-coded (green pixels = microglia, red pixels = Aβ-42, yellow pixels = microglia + Aβ-42) in Image J. The percent coverage was calculated as the area of yellow pixels over the area of the combination of yellow and red pixels.For all timelapse analysis, we analyzed one image from 1-2 brain slices for each animal (n = 4 animals/genotype).Experimenters were blinded for image analysis.

Statistical analyses
Data are expressed as mean ± SEM.For microglia morphology, cell density, and immunohistochemistry experiments, individual points represent each animal.For spontaneous and responsive microglia motility experiments, each individual point represents each cell.Datapoints were equally distributed across animals, and we did not detect datapoints for an individual animal skewing the results.Data were statistically analyzed by a twotailed t-test, a one-way ANOVA, or a two-way ANOVA, as indicated.p < 0.05 was considered statistically significant.When statistical significance was achieved, Sidak's multiple comparisons were used.Statistical analyses were done using GraphPad Prism 9.0 Software.

APOE4 genotype reduces microglia surveillance in the entorhinal cortex
Microglia continuously survey the cerebral microenvironment by extending and retracting their processes throughout the extracellular parenchyma [32,33].To investigate how microglial surveillance is affected by APOE genotype, we performed ex-vivo live imaging of baseline microglia motility for 20 min in acute brain slices from 6-month-old APOE3 and APOE4 mice using confocal microscopy (Fig. 2A).Microglia surveillance was quantified as a motility index.Microglia surveillance in EC was significantly lower (30%) in APOE4 mice (mean motility index = 0.19, p = 0.001, t = 3.49, df = 40, N = 4 animals, n = 19 cells) relative to APOE3 (mean motility index = 0.27, N = 5 animals, n = 23 cells) (Fig. 2B-C).This phenotype was not observed in the CA1 area (Fig. 2D).In addition, EC microglia had a 19% higher motility index than CA1 microglia (p = 0.032, t = 2.22, df = 35) in APOE3 brains; APOE4 microglia did not exhibit brain regiondependent differences in spontaneous motility (Figure S2E).These results suggest that APOE4 genotype promotes an alteration in spontaneous motility, resulting in the reduction of microglial surveillance.

APOE4 genotype reduces microglia responsiveness to ATP
ATP is a potent chemotactic cue of neuronal damages, and microglial processes migrate up ATP gradients [32].We investigated the effect of APOE genotype on responsive microglia motility by imaging the extension of processes in response to a patch pipette containing ATP for 30 min (Fig. 3A) and then calculating the velocity through manual tracking (Fig. 3B).Regardless of APOE genotype, microglial processes were observed responding to the ATP-filled pipet within one minute in both EC and CA1 processes.APOE3 microglia converged at the tip of the pipet by 30 min, while APOE4 microglial processes generally did not reach convergence (Fig. 3 C-D).
We observed similar results in the EC.There was a statistically significant 22% lower velocity of the APOE4 microglial processes in response to 1 mM ATP (p = 0.03, t = 2.66, df = 116, 1.66 ± 0.16 μm/min, n = 7 cells, N = 3 We observed that process velocity in both APOE3 and APOE4 microglia decreased in response to higher ATP concentration, perhaps due to a smaller focal point for microglia to extend the processes up the ATP gradient.Between the brain regions, there were no statistically significant differences in process velocity in response to ATP (e.g., at 3 mM ATP, Figure S2F).
We analyzed the effects of aging on spontaneous microglial motility using 12-month-old and 21-month-old mice.We found that spontaneous motility was lower at these ages than in the 6-month old mice above (Fig. 2).In the EC or the CA1 of older mice, spontaneous microglia motility was not altered by APOE genotype (Fig. 4D, and Figure S3C).However, in the EC, the aged, 21-month-old APOE4 microglia extended and retracted their processes significantly less (16%) relative to 12-month-old APOE4 microglia (p = 0.022, 12-month-old: 0.169 ± 0.006 vs 21-months-old: 0.142 ± 0.006) (Fig. 4D).

APOE4 microglia have less P2RY12, a key homeostatic gene associated with motility
To investigate a potential mechanism of the APOE4associated alteration in responsive microglia motility, we examined the homeostatic microglial ATP receptor P2RY12 [12].We immunostained coronal brain slices from 6-month-old APOE3 and APOE4 mice for P2RY12.P2RY12 labeling in the CA1 colocalized only with the cell membrane of microglia, mainly in the processes compared to the cell soma (Fig. 5A) (this co-localization further verifies the expression of GFP as limited to microglia).We found that P2RY12 fluorescent intensity overall in APOE3 microglia was 16% higher than in APOE4 microglia (p = 0.046, t = 2.51, df = 6, APOE3 = 37.3 ± 2.2 a.u., n = 4 vs APOE4 = 31.6± 0.5 a.u., n = 4) (Fig. 5B).Comparing P2RY12 labeling intensity between the microglial soma and processes, we found that nearly all P2RY12 fluorescent intensity was in the processes, but significantly less was in APOE4 processes compared to the APOE3 processes (p = 0.021, t = 3.1, DF = 6) (Fig. 5C).Through qPCR, we did not detect a difference between APOE genotypes in P2RY12 mRNA levels from cerebral cortices of six-month-old mice (Fig. 5D).Thus, we hypothesize that the APOE4 genotype affected the stability and subcellular distribution of P2RY12 rather than the gene transcription.

The response to Aβ-42 requires P2Y12 and is reduced in APOE4 microglia
We examined whether the slower motility of microglial responses toward ATP was recapitulated toward damage more relevant to AD, i.e., Aβ accumulations.We employed ex-vivo microglia live imaging as above, but introduced fluorescent synthetic Aβ-42 instead of ATP.The injected Aβ-42 was immediately visible at the injection site (Fig. 6A, time 0 min panels).We took Z-stack images every 15 min over the course of two hours (Fig. 6A).Processes from surrounding microglia responded to the Aβ-42 within 15 min, and, in general, processes from several microglia reached the Aβ-42 by two hours.The percent coverage was computed as the percent area of Aβ-42 covered by the microglial process over time.APOE3 microglia were able to reach Aβ-42 within the first 45 min and formed a lamellipodia structure that surrounded Aβ-42 for the remainder of the experiment (45-120 min).In contrast.APOE4 microglia demonstrated decreased capacity to cover Aβ-42.In the first 45 min, APOE4 microglia covered 34% Aβ-42, while APOE3 microglia covered 51% of the infusion (p = 0.025).Similar results were observed after 60 min (p = 0.009, APOE3: 48 ± 7.7% coverage vs APOE4: 37 ± 6.4% coverage) and 75 min (p = 0.049, APOE3: 60 ± 3.2% coverage vs APOE4: 46 ± 5% coverage) post-infusion.Two hours after infusion, there was no significant difference in Aβ-42 coverage between APOE genotypes (p = 0.237, APOE3: 70 ± 3.3% coverage vs APOE4: 57 ± 10% coverage).These results reflect a slower response to Aβ-42 by APOE4 microglia, suggesting an altered mechanism associated with initial responses to Aβ seeding.Internalization of Aβ peptide in microglial processes was observed in acute brain slices incubated for 6 h (Fig. 6C).
To test if P2RY12 plays a role in the observed microglial response to Aβ seeding, we incubated acute brain slices in 10 μM PSB0739, a selective P2RY12 inhibitor, for 30 min prior to Aβ injection.We found that microglial process motility towards the peptide deposit was completely disrupted by the inhibition of P2RY12 (n = 2 APOE3, n = 2 APOE4) (Fig. 6D).Thus, the ATP receptor P2RY12 is necessary for the movement of microglia processes in response to Aβ.

Discussion
Microglia play a pivotal role in maintaining brain homeostasis under physiological and pathological conditions [10,34].Large-scale transcriptomic and functional studies have identified microglia as a potential therapeutic target for neurodegenerative disorders [7,[35][36][37][38].Aging and APOE4 are the strongest risk factors for AD, but how they affect homeostatic microglia function remains unclear.In this study, we alterations in homeostatic and injury-induced microglial behavior associated with APOE4 in a preclinical mouse model lacking overt AD pathology.We found decreased spontaneous and responsive process motilities in APOE4 microglia, which are associated with the surveillance of brain parenchyma and the response to damage, respectively.These microglial phenotypes associated with APOE4 were exacerbated with aging.We also found that processes from APOE4 microglia surround newly injected Aβ-42 less than APOE3 microglia do, and the entire response is disrupted upon P2RY12 inhibition.This analysis was done in the cortex, given the early accumulation of Aβ deposits there.By examining microglial function under the physiological condition in the absence of AD pathology, we identified microglial behaviors associated with APOE4 and aging that could account for the predisposition to AD.
Microglial dynamics play a key role in many physiological functions [5,6,39].Investigating how risk factors such as APOE4 affect brain functions prior AD pathology may shed light on mechanisms that predispose the brain to neurodegeneration.The role of APOE genotype in AD pathology has been extensively explored through large-scale transcriptomic studies in mouse models, human brains, and iPSC-derived microglia [40][41][42][43][44]. Microglia from APOE4 carriers AD patients had downregulation in pathways associated with Aβ clearance and a proinflammatory transcriptomic profile [40].Work in iPSCs showed that APOE4 microglia had a proinflammatory transcription profile and reduced Aβ phagocytosis, suggesting that APOE4 impairs microglial function [42].In iPSC-derived microglia, APOE4 was also associated with dysregulation of lipid metabolism characterized by the accumulation of lipid droplets [43].APOE4 microglia isolated from targeted replacement mouse brains are proglycolytic due to increased expression of Hif1α and disrupted tricarboxylic acid cycle [41].Lipidome analysis of microglia isolated from mice expressing human APOE variants shows aberrant lipid metabolism in APOE4 microglia [45].Together, these studies suggest that APOE4 predisposes to AD pathogenesis through dysregulated immunometabolism.Dysregulation of lipid metabolism and neuroinflammation converge with transcriptional profiles associated with alteration of microglial function, such as chemotaxis and phagocytosis [46,47].With the large body of transcriptomic data available, it is imperative to validate the functional impact of APOE4 in studies that focus on microglial behaviors prior to AD pathology.A recent study in iPSC-derived microglia found that APOE4 genotype induces the accumulation of lipid droplets and impairs surveillance [48].In our study, we showed that APOE4 microglia have functional deficits exhibited by a decrease in brain surveillance and a slower response to chemotactic cues associated with acute damage.Our results complement data from iPSCs [48], as our model resembles more closely normal physiological conditions, including an environment with the basal levels of APOE released by astrocytes.We identified microglial phenotypes that are consistent with a proinflammatory transcriptomic profile.
We characterized microglial dynamics associated with APOE4 genotype in mice globally expressing human APOE.In recent work with human APOE3 or APOE4 expression only in microglia [49], dynamic functions were studied using in-vivo two-photon imaging.In line with our findings, microglia expressing APOE4 had impaired surveillance, and, after focal injury, microglial processes moved slower toward the lesion site compared to APOE3-expressing cells.These results suggest that the APOE4 genotype effects on microglial surveillance and responsiveness to damages are cell autonomous.It requires further investigation to determine whether microglial or astrocytic-derived APOE has a stronger impact on microglial homeostatic behaviors and the extent to which this impact is exacerbated by APOE isoform.Deletion of APOE4 in astrocytes in a mouse model of tauopathy showed upregulation of homeostatic microglial genes, i.e.P2RY12, downregulation of DAM genes, and decreased synaptic phagocytosis [50].These results suggest that astrocytic APOE influences the microglial molecular signature and function during AD pathology.Astrocytic and microglial APOE have different biochemical properties: microglial APOE is more glycosylated [51] while astrocytic APOE is more lipidated [52].In addition, APOE isoforms affect post-translational modification, i.e., APOE4 proteins are less lipidated [53,54] and glycosylated [55] than APOE3.It is possible that both the difference in biochemical properties of APOE isoforms as well as cell-specific APOE contributes to the alteration of microglial immunometabolism influencing homeostatic microglial dynamics.
Microglia exhibit different cellular phenotypes and dynamic behaviors during physiological and pathological conditions [34].These functional alterations are linked to different neurological disorders [56,57].Several bulk RNAseq and single-cell RNAseq studies have identified markers that constitute a signature of homeostatic microglia [10,58].Downregulation of these signature genes, including the ATP receptor P2RY12, is associated with altered microglial homeostasis and is consistent with the progression of AD [7,59,60].We demonstrated that APOE4 microglia expressed less P2RY12 than APOE3 microglia, particularly in their processes, results that are consistent with bulk and single-cell RNAseq from APOE4 mice and iPSC-derived microglia [41,44,48].P2RY12 expression is regarded as a marker of homeostatic or non-activated microglia; however, P2RY12 is necessary for microglial chemotaxis and process motility in response to neuronal damage [11,12,61,62].Consistent with this notion, we identified chemotactic defects associated with APOE4 genotype through live tissue imaging.We showed that APOE4 microglia have a slower response to ATP, a chemotactic cue of acute neuronal damage.These results suggest APOE4 genotype alters microglia homeostasis, affecting the expression of the P2RY12 receptor, which is necessary for injury-induced dynamic behavior.We proposed that the reduced homeostatic function of APOE4 microglia results in decreased surveillance of brain parenchyma and inefficient response to acute damages that, over time, lead to larger, chronic damages of neurodegeneration.
During normal aging, microglia undergo drastic changes in transcriptomics, microglial function, decreased phagocytosis, migration, and increased release of inflammatory cytokines associated with reduced homeostatic capacity [63].In the present study, our control group, APOE3 microglia, exhibited an age-associated decrease in surveillance of brain parenchyma and slower injury-induced response, consistent with aging studies of cortical and retinal microglia in wild-type animals [64,65].Bulk-and single-cell sequencing analysis shows that APOE4 and aging intractably downregulate transcriptional signatures associated with homeostatic microglia, including P2RY12, and upregulate pathways associated with inflammation [41,44].Consistent with this notion, we show that APOE4-associated decreased response to ATP is consistent in 12-and 21-month-old mice and that aging further exacerbates this effect.Although we did not detect an effect of APOE genotype on spontaneous motility in 12-and 21-month-olds, we showed that in APOE4 microglia, aging decreases the velocity movement of processes in response to ATP and spontaneous motility.These results suggest that the magnitude of ageassociated alteration on microglial behaviors will affect microglia differently depending on their APOE genotype.The decreased brain surveillance and response to acute damage in aged microglia might contribute to their loss of neuroprotective function.Microglia expressing APOE4 may lose their protective function at an earlier age, predisposing the brain to neuronal degeneration.
The entorhinal cortex is a primary site of AD pathogenesis, exhibiting early changes in the accumulation of Aβ plaques and neurofibrillary tangles [31].Several mechanisms influence microglia surveillance including P2RY13 [59] and neuronal network activity [66], and microglia manifest brain region-dependent heterogeneity in their phenotype [67].We observed that in 6-monthold APOE3 brains, EC microglia have a more ramified morphology and display higher surveillance compared to CA1 microglia but have no region-associated differences in response to ATP.However, with aging, the only regionassociated difference in microglial dynamics detected was in process velocity, with EC microglia responding to ATP more slowly than CA1 microglia.In APOE3 brains, the spontaneous motility of EC microglia decreased to levels observed in APOE4 microglia at 12-and 21 months old, suggesting that EC microglia are more susceptible to the effects of aging on microglial dynamics.These brain region-associated differences in microglia dynamics might be due to local cues coming from astrocytes and neurons affecting microglial gene expression.For example, EC neurons exhibit hyperexcitability due to increased spike firing rate [68][69][70].Astrocytes, besides secreting cholesterol and growth factors necessary for microglial viability and microglial function [15,71], have region-specific roles controlling brain function during health and disease [72,73].It is possible that epigenetic regulation of microglial functions makes EC more susceptible to the effects of aging and AD pathology.
In humans, the APOE4 allele is associated with the presence of amyloid at an early age [74] and increased amyloid burden [75,76].It remains unknown how microglia react during the early stages of amyloid seeding.In this study, we showed that APOE4 microglia extend their processes significantly slower after acute injection of Aβ compared to APOE3 microglia, and that P2RY12 inhibition prevents the process migration towards Aβ.In a recent in-vivo study, acute responses of microglia were studied in wild-type animals that received a cortical infusion of Aβ combined with either APOE3 protein (AβE3) or APOE4 protein (AβE4).In the brains that received AβE4 injection, microglia extended their processes to the infusion site significantly slower and had less coverage of the infusion [77].These results suggest that the presence of modifying APOE isoforms does not need to be expressed from microglia, but perhaps from other cells.It is possible that Aβ-APOE4 interaction and subsequent activation of APOE receptors such as TREM2 [78] are responsible for the deficient response to acute Aβ injection exhibited by APOE4 microglia.Several studies demonstrate that APOE protein binds to Aβ in an isoform-dependent manner, APOE4 > APOE3 > APOE2 [79,80].In addition, TREM2 is involved in microglia migration, phagocytosis, and expression of chemotactic receptors [7,81,82].Recent work in iPSC microglia shows that TREM2 mediates purinergic responses, expression of P2RY12, and process extension [20], suggesting that P2RY12 activity observed here may be influenced by TREM2 activation.Purinergic signaling through P2Y receptors plays a pivotal role in the crosstalk between astrocytes and microglia [83,84].Aβ peptide induces Ca + current-mediated release of ATP from astrocytes [85][86][87].While several microglial receptors are implicated in Aβ internalization (TREM2, SR-A1, CD33) and downstream activation of inflammatory responses (TLR, complement receptors, Fc receptors) [88], our data show that the ATP receptor P2YR12 is the initial mediator of microglial responsiveness to Aβ.It will be interesting to further explore whether Aβ causes the release of ATP by astrocytes or neurons that guide microglial processes, or the direct activation of TREM2 that causes hypersensitivity of P2RY12.
Long-term imaging in different mouse models of AD for at least 24 weeks shows an increase in Aβ plaque formation at early time points and a significant decline in the rate of plaque formation as Aβ load progresses over time [89][90][91].These findings strongly suggest that the initial seeding of Aβ is a critical time point in AD progression.The 5XFAD model of AD amyloid in the presence of APOE3 (E3FAD) or APOE4 (E4FAD) showed that APOE4 increases plaques number and size, as well as increased microglial reactivity [13,92,93].In a recent examination of microglial APOE in vivo [45], human APOE3 or APOE4 was conditionally expressed in microglia in APP/PS1 mice.Mice with APOE4 microglia had more Aβ plaque formation but less microglia surrounding the plaques compared to mice expressing APOE3 in microglia.Deletion of microglial APOE4 resulted in reduced Aβ pathology and increased microglial recruitment to plaques [45].However, these analyses could not address how APOE affects the acute responses to Aβ.In the present study, we injected Aβ-42 peptide in brain slices, allowing us to model the initial seeding of Aβ and study the acute microglial response.The slower process migration and decreased coverage of injected Aβ observed in APOE4 microglia suggest an altered mechanism associated with Aβ uptake compared to APOE3 microglia.In vitro studies have shown that APOE4 microglia exhibit a less efficient Aβ uptake [42] and have impaired digestion of phagocytized materials [94].It is imperative to study the interaction between microglia and Aβ peptides over time to investigate how APOE4-associated alteration in phagocytosis affects plaque formation in an in vivo system.Thus, understanding how APOE genotypes regulate microglial dynamic behaviors as well as their impact on homeostasis and responses to damage may shed light on their role in AD pathogenesis.

Limitations of the study
Our study has a few limitations.First, acute brain slices affect microglial activation, since they are highly sensitive to tissue disturbances and can adopt an activated phenotype over time [95].Acute brain slices have been an essential tool for studying dynamic microglial behaviors, as they provide an environment that preserves microglia's physiological states better than in vitro systems.Thus, the current work is not fully modeling an in vivo environment, and future analyses will be needed to test the in vivo effects of P2RY12 on microglial responses to amyloid across APOE genotypes and ages.Second, this work tests the acute effects of microglial responses to damage (including Aβ); more chronic effects would require similar experiments in mouse amyloid models.Third, our mouse model is heterozygous (CX3CR1 GFP/+ ) for CX3CR1; therefore, neuron-microglia interaction mediated by CX3CL1-CX3CR1 signaling might be somewhat impaired compared to wild-type animals, resulting in decreased cognitive and synaptic function [96].In addition, recent work [97] suggests that the loss of a single copy of CX3CR1 leads to alteration in microglial transcriptomics compared to wild-type animals.Although CX3CR1 is mainly expressed in microglia, expression can also be found in monocytes, subsets of NK, and dendritic cells [98].Therefore, the potential impact of peripheral deletion CX3CR1 on brain homeostasis warrants further investigation [99].

Conclusion
We determined that APOE4 microglia have decreased surveillance, slower process movement in response to ATP, and decreased coverage of Aβ peptide infusion, processes that are mediated in part by the ATP receptor P2RY12.Future studies with this model would provide more information on the effects of APOE genotype on microglia-neuron interaction and its impact on neuronal networks prior to AD pathology.These studies will help to identify the most important microglial behaviors that are altered by APOE4 and aging and could be used in therapeutic approaches to reduce the risk of AD.

Fig. 1
Fig. 1 APOE genotype does not affect microglia density and morphology.A Representative images of microglia in the entorhinal cortex (EC) and hippocampus (CA1) from APOE3 (E3) and APOE4 (E4) mice.Scale bar = 50 µm.B Quantification of microglia density.Bars represent the mean ± SEM cell density in analyzed fields, and symbols (circles, male; triangle, females) represent data points for each animal.2-3 brain slices per animal, 4 animals per genotype per sex.C Schematic of microglia morphology analysis.Scale bar 25µm.D Quantification of microglial endpoint/cell (left) and branch length/cell (right) of the EC and CA1 of E3 and E4 mice.Bars represent the mean ± SEM microglial endpoint/cell and branch length/cell in analyzed fields, and symbols (circles, male; triangle, females) represent data points for each animal.2-3 brain slices per animal, 4 animals per genotype per sex.*p < 0.05; unpaired two-tailed Student's t test

Fig. 2 APOE4
Fig. 2 APOE4 Microglia exhibit lower spontaneous motility in the entorhinal cortex.A Confocal images of microglia spontaneous motility time-lapse at different time points, and binary overlap.Scale bar 10 µm.B Binary overlaps of APOE3 and APOE4 microglia in the EC and CA1.Quantification of microglial baseline processes movement (motility index = green pixels + red pixels/yellow pixels) in the EC (C) and CA1 (D).Bars represent the average motility index of all analyzed microglia ± SEM, and symbols (circles, male; triangle, females) represent data points for each microglia (3-4 animals per genotype).***p < 0.001; unpaired two-tailed Student's t test

Fig. 3
Fig. 3 APOE4 microglia extend their processes in response to ATP slower than APOE3 microglia.A Representative confocal time-lapse showing microglial response (green) to 3 mM ATP in a patch pipette (red) in acute entorhinal slices.Scale bar 20 µm.B Illustration of the microglial processes manual tracking over the course of 30 min.Scale bar 20 µm.C-D APOE3 (E3) and APOE4 (E4) microglia responding to patch pipette containing ATP (shown as dashed triangle) in the CA1 (C) and EC (D).E-F Processes velocity in the CA1 (C) and EC (D) in response to 1 mM, 3 mM, and 10 mM ATP. Bar graphs represent the mean velocity ± SEM.Individual points (circles, male; triangle, females) represent the average process velocity (3 processes per cell) of each microglia (3-5 cells per animal).3-4 animals per genotype per sex (6 months old).Two-way ANOVA with Sidak's multiple comparison post-hoc analyses *p < 0.05, ****p < 0.0001

Fig. 4 Fig. 5 Fig. 6 APOE4
Fig.4APOE4-associated alterations in microglia behaviors are worsened with aging.A Quantification of microglia density in EC and CA1 of 12-and 21-month-old mice.Bars represent the mean ± SEM cell density in analyzed fields, symbols (circles, male; triangle females) represent data points for each animal.2-3 brain slices per animal, 3-4 animals per genotype per sex per age.Two-way ANOVA with Tukey's multiple comparison post-hoc analyses, *p = 0.014.B-C Quantification of microglial endpoint/cell (B) and branch length/cell (C) in the EC and CA1 of 12-and 21-month-old mice.Bars represent the mean ± SEM microglial endpoint/cell and branch length/cell in analyzed fields, and symbols (circles, male; triangle, females) represent data points for each animal.2-3 brain slices per animal, 3-4 animals per genotype per sex per age.D Bars represent the average motility index of all analyzed microglia ± SEM, and points represent data from each microglia 3-4 animals per genotype per sex per age.Two-way ANOVA with Tukey's multiple comparison post-hoc analyses *p = 0.022.E 12-and 21-months old APOE3 (E3) and APOE4 (E4) microglia responding to patch pipette containing ATP (shown as dashed triangle) in the CA1 and EC.Scale bar 20 µm.F Processes velocity in the CA1 (C) and EC (D) in response to 1 mM ATP. Bar graphs represent the mean velocity ± SEM.Individual points represent the average process velocity (3 processes per cell) of each microglia (3-5 cells per animal), 3-4 animals per genotype per sex (6 months old).Two-way ANOVA with Tukey's multiple comparison post-hoc analyses *p < 0.05, **p = 0.006.G Comparison of velocity of microglial processes in response to 1 mM ATP in 6-month (shown in Fig.3), 12-, and 21-month-old mice.Two-way ANOVA with Tukey's multiple comparison post-hoc analyses; data are compared to 6-month-old mice for either EC or CA1.****p = 0.0001