A protective PLCγ2 gene variant may reduce Alzheimer's-related problems by boosting helpful immune cell activity in the brain

In this study, the APPswe/PS1dE9 transgenic mouse line (later simply APP/PS1) [19] was crossbred with the PLCγ2-P522R knock-in (KI) mouse line [7] to investigate the effects of the PLCγ2-P522R variant on AD-related β-amyloid and associated neuronal pathologies, microglial activation, and behavioral changes. Female and male mice at the ages of 7 and 13 months were subjected to TSPO-PET imaging with the [18F]FEPPA ligand to monitor microglial activation [7]. The mice were sacrificed at the age of 13 months, and half of the brain was cut into coronal sections and used for immunohistochemical analyses, whereas the other half was dissected into regional blocks and utilized for biochemical assays. The entorhinal cortex and hippocampus were subjected to immunohistochemical analyses, while the temporo-occipital cortex and hippocampus of the same mice were subjected to biochemical analyses. The whole brains of a subset of 13-month-old mice were used for CD11b+ microglia isolation and the isolated CD11b+ cells were used for RNA sequencing studies. Behavioral testing was conducted on 12- to 13-month-old mice. All immunohistochemical, biochemical, and behavioral analyses were conducted in female mice.

CD11b+ microglia acutely isolated from 13-month-old PLCγ2-P522R KI and wild-type (WT) mice and cultured primary microglia isolated from 0–3-day-old pups under different conditions to induce cellular stress were subjected to global transcriptomic and proteomic analyses as well as functional studies. Transcriptomic analyses were conducted in blood monocyte-derived microglia (MDMi) isolated from PLCγ2-P522R variant carriers and matched controls to study how well the findings translated to human cells. Induced pluripotent stem cell-derived microglia (iMGL) homozygous for the PLCγ2-P522R variant with either an APOE3/3 or APOE4/4 genetic background and isogenic control lines were used to functionally assess the key findings.

The APP/PS1 line in the C57BL/6J background was used as a model for AD [19]. This mouse line coexpresses a single transgene, a chimeric mouse/human APP695 harboring the Swedish K670M/N671L mutation (Mo/Hu APPswe), and a human PS1 with exon 9 deletion (PS1dE9) under the mouse prion protein promoter. This line was crossbred with a PLCγ2-P522R KI mouse line generated using CRISPR/Cas9-assisted gene editing on a C57Bl/6J background, as we reported previously [7]. WT (C57BL/6J) mice were used as controls in in vitro experiments. The mice were kept in an environment with a constant temperature of 22 ± 1 °C and 50–60% humidity. The lights were on from 07:00–19:00, and food and water were available ad libitum. All mouse studies were carried out in accordance with the guidelines of the European Community Council Directives 86/609/EEC and approved by the Animal Experiment Board of Finland/Regional State Administrative Agency and Project Authorization Board (ESAVI 21203–2019, EKS-004–2019).

Genomic DNA was purified from ear biopsies by incubating the samples with 50 mM NaOH for 1 h at 95 °C and neutralizing them with 1 M Tris, pH 8.0. Genotyping of the PLCγ2-P522R locus was carried out with SNP genotyping. Ten nanograms of DNA were mixed with TaqMan Genotyping master mix (Applied Biosystems) and a TaqMan SNP genotyping assay (Applied Biosystems) targeted to trinucleotide variation, leading to the substitution of proline 522 with arginine [7]. The reaction mixture was amplified for 10 s at 95 °C, 15 s at 95 °C, and 60 s at 60 °C for 40 cycles and measured with a LightCycler® 480 II (Roche).

To confirm the expression of the APP and PSEN1 genes, loci harboring the human PSEN1 and mouse prion protein promoters were amplified using the following primers: 5'-CCTCTTTGTGACTATGTGGACTGATGTCGG-3', 5'-GTGGATACCCCCTCCCCCAGCCTAGACC-3', and 5'-CAGGTGGTGGAGCAAGATG-3'. Touchdown PCR was performed using the following program: 3 min at 94 °C; 30 s at 94 °C, 1 min at 62–58 °C, and 2 min at 72 °C, for two cycles at each temperature; 30 s at 94 °C, 1 min at 57 °C, and 2 min at 72 °C for 25 cycles; and 5 min at 72 °C. The 1300 bp (PSEN1) and 750 bp (internal control) bands were detected with agarose gel electrophoresis.

Starting at the age of 12 months, the mice were subjected to a behavioral test battery. The mice were weighed before testing began.

Spontaneous explorative activity was assessed with an automated activity monitor (TruScan, Coulbourn Instruments, Whitehall, PA, USA) based on infrared photobeam detection. The system consisted of an observation cage with white plastic walls (26 cm × 26 cm × 39 cm) and two frames of photo detectors for monitoring horizontal and vertical activity. The test cage was cleaned with 70% ethanol before each mouse to remove odor traces. The following parameters were measured during a 10-min session: ambulatory distance (gross horizontal locomotion), rearing time, and time in the area center vs. periphery.

A light‒dark box with an electric grid floor in the dark chamber was used to test spontaneous fear towards the illuminated chamber and for foot-shock reinforced passive avoidance. On Day 1, the mouse was first placed in the illuminated chamber with free access to the dark chamber through a narrow opening, and the time spent in the light and dark chambers was recorded for 5 min. Thereafter, the mouse was placed again in the illuminated chamber for 30 s with the entrance to the dark chamber closed. The slide door separating the chambers was then opened. As soon as the mouse entered the dark side (cutoff 3 min), the sliding door was closed, and a mild foot shock was delivered (2 × 2 s at 0.30 mA). The mouse was then returned to its home cage. On Day 3 (48 h later), the mouse was again placed in the illuminated chamber with the sliding door open, and the time to enter the dark chamber was recorded as an index for fear memory. Finally, the mouse was confined to the dark chamber, and the pain threshold to electric foot shock was determined by gradually increasing the current on the grid floor until the mouse reacted.

Motor coordination and balance were tested using an automated rotarod device (Uno Bacile, Italy). Each mouse was first familiarized with the rotating rod twice for 2 min by allowing it to rotate at a constant speed of 4 rpm. On a separate day, the mouse was tested twice, with the speed of the rod increasing stepwise from 4 to 40 rpm. The cutoff time was 6 min. The trial with longer latency to fall was recorded.

The pain threshold was tested using a hotplate with surface temperatures of 47 °C and 51 °C. The time to lick a paw or jump was recorded until a cutoff time of 60 s. Each mouse was tested three times at both temperatures (lower temperature first), with 1 h between the test sessions. The average reaction time at each temperature was recorded.

Spatial learning and memory were assessed using the Morris water maze (MWM) test. The test was conducted in a white circular wading pool (diameter of 120 cm), and a transparent platform (14 cm × 14 cm) was submerged underneath the water. The pool was open to landmarks in the room. The water temperature was 20 ± 0.5 °C. The testing was preceded by two training days with a guiding alley to the platform. During the acquisition phase (Days 1–5), the location of the hidden platform was kept constant, and the start position varied between four different locations at the pool edge. Each mouse was placed in the water with its nose pointing toward the pool wall. If the mouse failed to find the escape platform within 60 s, it was placed on the platform for 10 s by the experimenter. The acquisition phase consisted of five daily trials with a 10 min intertrial interval. On Day 5, the search bias was tested in a 60-s probe trial (the 5th trial) without the platform. The mouse was video tracked, and a video analysis program (EthoVision, Noldus, The Netherlands) was used to calculate the escape latency, swimming speed, path length and time at the pool periphery (10 cm from the wall) and in the platform zone (diameter of 30 cm).

[18F] FEPPA-PET imaging was conducted on female and male mice at the ages of 7 and 13 months. [18F]FEPPA is a specific positron emission tomography (PET) ligand for the mitochondrial translocator protein 18 kDa (TSPO) that is highly expressed in activated microglia [20]. The mice were anesthetized with isoflurane (1.5% with N2/O2 70%/30% through a nose cone) and placed on a heated animal holder (Équipement Vétérinaire Minerve, Esternay, France) on the scanner bed in a prone position and secured with tape to prevent movement during scanning. For the 7-month timepoint, the mice were imaged using a dedicated PET scanner (Inveon DPET, Siemens Healthcare) and immediately afterward with a CT scanner (Flex SPECT/CT, Gamma Medica, Inc.) for anatomical reference images using the same animal holder. For the 13-month timepoint, imaging was performed with a preclinical PET/MRI scanner (MR Solutions Ltd., Surrey, UK). The change in scanners was due to unexpected technical problems with the dedicated PET scanner. Dynamic imaging for 70 min was started at the time of activity. 18F-FEPPA (8.0 ± 2.0 MBq) was injected as a slow bolus over a period of 30 s through the tail vein. PET data were gathered in list-mode form and corrected for deadtime, randomness, scatter and attenuation. The data were reconstructed using the scanner manufacturers' software with a 2D- (7 mo) or 3D- (13 mo) ordered-subsets expectation maximization (OSEM) algorithm. Regions of interest (ROIs) were drawn for the whole brain (excluding the cerebellum and olfactory bulb), frontal cortex, pons, hippocampus and cerebellum. The ROIs were drawn using Carimas 2.10 software (Turku PET Centre, Finland). The final uptake values (percent of injected dose per ml of tissue) were the average values from 20 to 70 min from the ligand injection.

At the age of 13 months, terminal anesthesia was administered as intraperitoneal injection of 60 mg/ml pentobarbital, after which mice were transcardially perfused with ice-cold phosphate-buffered saline (DPBS, #17-512F, BioNordica) for 5 min to rinse blood from the brain. The brain was quickly removed and placed on ice. One brain hemisphere was immersed in 4% paraformaldehyde (PFA) in phosphate buffer (0.05 M, pH 7.6) for 24 h at 4 °C, followed by incubation in 30% sucrose in DPBS overnight. The fixed tissue was stored in antifreeze (0.5 M sucrose, 37.5% (v/v) ethylene glycol, 0.03 M phosphate buffer (pH 7.6), 0.004 M sodium azide) at − 20 °C until it was cut into 35 µm-thick coronal sections using a freezing sliding microtome.

The temporo-occipital cortex and hippocampus were dissected from the other hemisphere and snap-frozen in liquid nitrogen. The tissue pieces were mechanically homogenized in 300 µl of ice-cold DPBS. Fifty microliters of the lysate was mixed with 500 µl of TRIzol reagent (#T9424, Sigma‒Aldrich) and used for RNA extraction (below). The remaining lysate was supplemented with 1:100 HALT™ EDTA-free protease (#87785, Thermo Scientific) and HALT^™ phosphatase (#78420, Thermo Scientific) inhibitor cocktails. Proteins were extracted by incubating the lysate on ice for 30 min with Tissue Protein Extraction Reagent (T-PER, #78510, Thermo Scientific), and the supernatant was collected after centrifugation for 10 min at 10 000 × g at 4 °C. The fraction of the PBS lysate was ultracentrifuged at 100 000 × g for 2 h at 4 °C, and the supernatant containing the soluble fraction was collected in clean tubes. The pellets containing the insoluble fraction were solubilized in 5 M guanidine-HCl by vortexing for 3 h at room temperature. The samples were subsequently centrifuged at 10 000 × g for 10 s. The protein concentrations were measured using a BCA protein assay kit (#23225, Pierce).

Total protein lysates (15–20 μg) were supplemented with NuPAGE LDS sample buffer (#NP0007, Invitrogen) containing β-mercaptoethanol and separated via SDS‒PAGE via NuPAGE 4–12% Bis–Tris Midi protein gels (#WG1402BOX, Invitrogen). Proteins were subsequently transferred to polyvinylidene difluoride (PVDF) membranes (#IB24001, Invitrogen) using the iBlot 2 Dry Blotting System (Invitrogen). Unspecific antibody binding was blocked by incubating the blots in 5% nonfat milk or 5% bovine serum albumin (BSA, #A9647, Sigma‒Aldrich) in 1 × Tris-buffered saline with 0.1% Tween 20 (TBST) for 1 h at room temperature. Proteins were detected from the blots using the following primary antibodies diluted in 1 × TBST and incubated overnight at 4 °C: mouse anti-β-amyloid 1:1000 (6E10, #SIG-39320, BioLegend), rabbit anti-sAPPβ 1:100 (#18957, IBL America), mouse anti-APP N-terminal 1:1000 (22C11, #MAB348, Merck), rabbit anti-APP C-terminal 1:1000 (#A8717, Merck), and mouse anti-GAPDH 1:10 000 (#ab8226, Abcam). The blots were subsequently incubated with sheep anti-mouse 1:5000 (#NA931V, Cytiva) or donkey anti-rabbit 1:5000 (#NA934V, Cytiva) horseradish peroxidase (HRP)-conjugated secondary antibodies in 1 × TBST for 1 h at room temperature. Enhanced chemiluminescence Prime (#RPN2232, Cytiva) and Select (#RPN2235, Cytiva) Western blotting detection reagents were used to detect the protein bands. SeeBlue Plus2 Prestained Standard (#LC5925, Invitrogen) was used to estimate the size of the detected bands. The blots were imaged with a ChemiDoc MP system (Bio-Rad), and the images were quantified using the ImageLab (Bio-Rad) software. Results are shown as percentage of A+/P^wt/wt group.

Aβ40 and Aβ42 levels in the soluble and insoluble lysates were determined with Amyloid β (40) Human/Rat (#294–64701, FUJIFILM, Wako) and High-Sensitivity Amyloid β (42) Human/Rat (#292–64501, FUJIFILM, Wako) ELISA Kits following the manufacturers’ instructions. Aβ levels were normalized to the total protein concentration within each sample.

Soluble and insoluble APOE were measured in lysates using an Apolipoprotein E (APOE) Mouse ELISA Kit (#ELK2007, Gentaur) according to the manufacturer’s instructions. APOE levels were normalized to the total protein concentration within the same sample.

For sTREM2 measurement from the respective lysates, a 96-well Costar Assay Plate (9018, Corning) was coated with 500 ng/ml human/mouse anti-TREM2 antibody (#MAB17291-100, R&D Systems) in 0.5 M carbonate-bicarbonate buffer (pH 9.6, #C3041, Sigma‒Aldrich) and incubated at 4 °C overnight. The wells were subsequently washed 3 times with 0.05% Tween 20 (#93773-250G, Sigma‒Aldrich) in DPBS and blocked with 1% Block Ace (#BUF029, Bio-Rad) in DPBS for 4 h at room temperature. The wells were washed once before adding Recombinant Mouse TREM2 Protein in concentrations 12.21–781.25 pg/ml (#50,149-M08H, Sino Biological, Inc.), and the samples were diluted in assay buffer (1% BSA (#A9647, Sigma‒Aldrich) and 0.05% Tween 20 in DPBS). The standards and samples were incubated at 4 °C overnight. The wells were washed five times following incubation with 66.67 pg/ml mouse TREM2 biotinylated antibody (#BAF1729, R&D Systems) for 1 h at room temperature, after which the wells were washed 3 times. Streptavidin poly-HRP40 (#20102011, Fitzgerald) diluted 1:3000 in assay buffer was added, and the mixture was incubated for 1 h at room temperature in the dark, after which the wells were washed 5 times. 3,3′,5,5′-Tetramethylbenzidine Liquid Substrate (TMB, Super Slow, #T5569, Sigma‒Aldrich) was added, and the mixture was incubated for 15 min at room temperature. The reaction was stopped with 1 M H₃PO_4, and the absorbances were measured with a plate reader (Tecan Infinite, 450 nm).

The levels of inflammatory cytokines and transforming growth factor beta (TGF-β) 1–3 in the cortical and hippocampal lysates were determined with an MSD mouse V-plex Proinflammatory Panel 1 (#K15048, Mesoscale) and a U-plex TGFβ Combo Panel (#K15242K, Mesoscale), respectively, according to the manufacturer’s instructions. The results were normalized to the total protein concentration in each sample.

Three to four brain sections between the bregma − 3.1 mm and − 3.5 mm coronal planes according to Paxinos and Franklin’s Mouse Brain in Stereotaxic Coordinates (2001) were selected from each mouse for every staining. Before staining, the sections were incubated in sodium phosphate buffer overnight at room temperature. The next day, the sections were pretreated with 0.05 M citrate solution (pH 6.0) for 30 min at 80 °C and cooled in sodium phosphate buffer for 20 min. To visualize β-amyloid plaques, the sections were incubated with 0.1 mM X-34 (#SML1954, Sigma‒Aldrich) for 1 h at room temperature in a solution containing 60% DPBS and 40% EtOH. The sections were then rinsed three times with 60% DPBS + 40% EtOH followed by washing with TBST three times for 5 min. Nonspecific antibody binding was blocked with 3% BSA in TBST, after which the sections were incubated with the following primary antibodies at 4 °C overnight: a rabbit anti-β-amyloid antibody (1:1000, 6E10, #SIG-39320, BioLegend) to detect diffuse β-amyloid, a mouse anti-APP N-terminal antibody (1:2000, 22C11, #MAB348, Merck) to visualize dystrophic neurites, a rabbit anti-IBA1 antibody (1:1000 #019–19741, FUJIFILM Wako) to detect microglia, a rabbit anti-GFAP antibody (1:500, #Z0334, Dako) to detect astrocytes, and a goat anti-APOE antibody (1:1000, #178479, Merck). After being washed three times for 5 min each with TBST at room temperature, the sections were incubated with Alexa Fluor^™ 568 donkey anti-mouse (#A-10037, Thermo Scientific), Alexa Fluor^™ 488 goat anti-rabbit (#A-11008, Invitrogen), Alexa Fluor™ 488 donkey anti-rabbit (#A-21206, Thermo Scientific), Alexa Fluor™ 568 goat anti-rabbit (#A-11036, Invitrogen), and Alexa Fluor™ 568 donkey anti-goat (#A-11057, Thermo Scientific) secondary antibodies for 1 h at room temperature. The sections were again washed three times for 5 min and then mounted with Vectashiled HardSet Antifade Mounting Medium (#H-1400, Vector Laboratories) on gelatin-coated slides. Control sections without primary antibody treatment were processed simultaneously.

For X-34-positive β-amyloid plaque coverage analysis, the slices were imaged with a Leica Thunder Imager 3D Tissue slide scanner equipped with a K5-14400955 camera and an HC PL FLUOTAR 20x (NA 0.5) DRY objective (Leica Microsystems). Z-stack images were taken with 19 layers with a total distance of 27 μm. The whole entorhinal cortex and hippocampus were scanned, stitched, and subjected to automated Thunder image processing computational clearance. All the stained samples, which included more than one fluorophore, were imaged with a Zeiss Axio Observer inverted microscope with a Zeiss LSM 800 confocal module (Carl Zeiss Microimaging GmbH) with a 20 × (NA 0.5) objective. A Plan-Apochromat 63 × (NA 1.4) objective was used for imaging 6E10 and IBA1 for Imaris-based colocalization analysis. Three 1024 × 1024-pixel fields from both brain areas were taken via the Z-stack imaging method 15–19 layers with a total distance of 22.5–27 μm. All laser, light, and detector settings were kept constant for all immunostained samples. Images were taken by investigators who were blinded to the mouse groups.

Quantitative image analysis was carried out with Fiji (ImageJ, version 1.53c) software. Maximum intensity projections were generated for each channel, and the rolling ball algorithm (radius 50) and Gaussian blurring (sigma 2) were applied to subtract the background and remove noise, respectively. The following automatic thresholding methods were used to create a mask for each stain: Otsu for X-34, Moments for 22C11 and IBA1, and Li for GFAP and APOE. For Leica Thunder Imager 3D tissue slide scanner images, the entorhinal cortex and hippocampal ROIs were manually drawn for each image. The X-34-positive β-amyloid plaque area and the number of plaques in the entorhinal cortex and hippocampus were subsequently normalized to the area of the respective ROI. For the analysis of IBA1, GFAP, 22C11, and APOE within and surrounding the β-amyloid plaque area, the outlines of the β-amyloid plaques were first determined from X-34 staining, and then the IBA1+, GFAP+, 22C11+, and APOE+ areas were analyzed within 0–40 µm, 0–40 µm, 0–14 µm, and 0–10 µm from the plaque outline, respectively (Figure S1A-D). The smallest area covering most of the IBA1, GFAP, 22C11 and APOE signals surrounding β-amyloid plaques was used to choose the width of the analyzed area. The image preprocessing and thresholding methods were kept constant between the samples within each staining. No signal was detected in the negative control samples. The ZEN 2012 software Blue edition (Carl Zeiss Microimaging GmbH) was used for visualization purposes.

Imaris (10.1.1) software was used for quantitative image analysis of the colocalization of 6E10 and IBA1. A surface for 6E10 and IBA1 was created using the surface tool. A Gaussian filter was applied via a smoothing function, and the surface detail value was set to 0.198 µm. The absolute intensity thresholding method was used for surface creation, with a set threshold of 3.8 for 6E10 and 12 for IBA1. Surfaces below 600 voxels in the IBA1 channel and 70.7 voxels in the 6E10 channel were filtered out from the analysis to exclude small particles. A mask was created from the 6E10 channel using the created IBA1 surface. The voxel intensity outside the IBA1 surface was set to 0 for this mask. A new surface was created from the mask using the same settings as those used to create the previous 6E10 surfaces. The new surface represents the overlapping signal of 6E10 IBA1. All image analysis parameters were kept the same for all the samples. The data are presented as the 6E10 volume inside the IBA1 surface as a percentage of the total 6E10 volume. All image analyses were performed by an investigator blinded to the genetic background of the mice.

At 13 months of age, the mice were anesthetized with 60 mg/ml pentobarbital and briefly perfused with ice-cold DPBS to eliminate blood cells. The mice were then decapitated, and the brains were placed in ice-cold Hank's balanced salt solution without Ca²⁺ or Mg²⁺ (HBSS, #14,175,129, Gibco). The brains were mechanically dissociated into small pieces and further dissociated into single-cell suspensions via enzymatic digestion. Cell isolation was performed according to the instructions of the Adult Brain Dissociation Kit (#130-107-677, Miltenyi), and CD11b+ microglia were separated via CD11b MicroBeads, human and mouse (#130-093-634, Miltenyi) kits instructions. Magnetic separation was performed via a QuadroMACS Separator (#130-090-976, Miltenyi) with LS Columns (#130-042-401, Miltenyi). The cell pellet was washed twice with DPBS and immediately used for RNA extraction (below) or stored as a dry pellet at − 80 °C for proteomic analysis (below).

Cognitively healthy PLCγ2-P522R monoallelic (GC, APOE3/3, age 68 ± 6.7, n = 7) carriers and matched controls (CC, APOE3/3, age 70 ± 5.5, n = 7) were recruited from 2022 to 2024 through the Auria Biobank on the basis of preexisting genome data returned to the biobank from FinnGen [21]. Blood samples for monocyte and DNA isolation were collected from the study subjects following written informed consent from each participant.

To extract peripheral blood mononuclear cells (PBMCs), 60–100 ml of peripheral venous blood was collected from the participants and processed within 24 h of sample collection. PBMCs were extracted by gradient centrifugation using Ficoll-Paque PLUS (#17–1440-02, Cytiva) in SepMate-50 tubes (#85450, Stemcell Technologies). Positive selection of monocytes from PBMCs was performed with human CD14 MicroBeads (#130-050-201, Miltenyi Biotec) and magnet-activated cell sorting.

Monocytes were plated at a density of 1 million cells per well in 12-well plates. Monocyte differentiation into MDMis was performed according to a previously published protocol [22] with modifications described in Martiskainen et al. [23]. After MDMi differentiation for 12 days in vitro, the conditioned medium was replaced with fresh MDMi culture medium or with culture medium containing myelin (25 µg/ml) or LPS (200 ng/ml, O26:B6, L5543, Sigma Aldrich), and the cells were cultured for 24 h prior to sample collection. For RNA extraction, the cells were washed three times with ice-cold PBS and stored in TRIzol reagent (#T9424 Sigma‒Aldrich). The RNA extraction protocol is described below.

The mouse temporo-occipital cortex was homogenized in TRIzol reagent (#T9424, Sigma‒Aldrich), after which chloroform was added (one-fifth of the volume of TRIzol reagent), and the tubes were shaken vigorously. The mixtures were incubated at room temperature for 5 min, and phase separation was performed by centrifugation at 12 000 × g for 20 min at 4 °C. The RNA was precipitated from the aqueous phase with 2-propanol by mixing and incubating at room temperature for 30 min. The samples were subsequently centrifuged at 12 000 × g for 25 min at 4 °C, after which the pellets were washed twice with 75% ethanol. The RNA pellets were air-dried and dissolved in RNAse-free H₂O.

Isolated mouse CD11b+ microglia were resuspended in DPBS and immediately mixed with lysis buffer included in a High Pure RNA Isolation Kit (#11828665001, Roche). RNA was extracted following the kit instructions and sent for sequencing to Novogene Europe (below).

MDMi cells were washed once with 1 × DPBS, and the cells were collected with TRIzol reagent, where one-fifth of the TRIzol reagent volume of chloroform was added. The samples were mixed and incubated at room temperature for 2 min, followed by centrifugation at 12 000 × g for 15 min at 4 °C. The aqueous phase was collected, and 2-propanol was added at room temperature. The samples were vortexed and incubated at room temperature for 10 min. The RNA was washed twice with 75% ethanol and centrifuged at 13 000 × g for 5 min at 4 °C. The supernatant was removed, and the remaining ethanol was allowed to evaporate at room temperature for approximately 10 min. The RNA pellet was resuspended in RNase-free H₂O and sent for sequencing to Novogene Europe (below).

Library preparation and RNA sequencing were conducted by Novogene (Cambridge, UK) and executed with an Illumina high-throughput sequencing platform. In brief, mRNA enrichment was performed with oligo(dT) bead pulldown, from which the pulldown material was subjected to fragmentation, followed by reverse transcription, second strand synthesis, A-tailing, and sequencing adaptor ligation. The final amplified and size-selected library comprised 250–300 bp of inserted cDNA and 150 bp paired-end sequences. Sequencing yielded reads for 20.2–31.34, 29.4–49.9, and 19.3–31.8 million library fragments per CD11b+ mouse microglia, mouse cortex, and human MDMi cells, respectively.

The 150-nucleotide pair-end RNA-seq reads were quality controlled using FastQC (version 0.11.7) (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/↗). The reads were then trimmed with Trimmomatic (version 0.39) [24] to remove Illumina sequencing adapters and poor-quality read ends via the following essential settings: ILLUMINACLIP:2:30:10:2:true, SLIDINGWINDOW:4:10, LEADING:3, TRAILING:3, and MINLEN:50. The trimmed reads were aligned to the Gencode mouse transcriptome version M25 (for genome version GRCm38) or the human transcriptome version 38 (for genome version GRCh38) using STAR (version 2.7.9a) [25] with essential nondefault settings: –seedSearchStartLmax 12, –alignSJoverhangMin 15, –outFilterMultimapNmax 100, –outFilterMismatchNmax 33, –outFilterMatchNminOverLread 0, –outFilterScoreMinOverLread 0.3, and –outFilterType BySJout. The unstranded, uniquely mapped, genewise counts for primary alignments produced by STAR were collected in R (version 4.2.2 or 4.3.2) using Rsubread::featureCounts (version 2.12.0, 2.12.3 or 2.16.1)[26], ranging from 16.2 to 19.7, 25.1 to 43.1, or 14.5 to 25.1 million per CD11b+ mouse microglia, mouse cortex, and human MDMi cells, respectively. Differentially expressed genes (DEGs) between experimental groups were identified in R (version 4.3.2) using DESeq2 (version 1.38.3) [27] by employing the Wald statistic and lfcShrink for FC shrinkage (type = “apeglm”) [28]. For the human MDMi samples, comparisons between PLCγ2-P522R variant carriers (GC) and matched controls (CC) were performed while adjusting for the sample delivery batch. Pathway enrichment analysis was performed on the gene lists ranked by the pairwise DEG test log2FCs in R (version 4.2.3) using clusterProfiler::GSEA (version 4.6.2) [29] with Molecular Signatures Database gene sets (MSigDB, version 2022.1 for mouse CD11b+ microglia, and version 2023.1 for the mouse cortex and human MDMi) [30].

Sample preparation for proteomic analyses of acutely isolated CD11b+ microglia from PLCγ2-P522R mice was carried out as described previously by Martiskainen et al. [23]. A total of 350 ng of peptide was separated on an in-house packed C18 analytical column (15 cm × 75 µm ID, ReproSil-Pur 120 C18-AQ, 1.9 µm, Dr. Maisch GmbH) using a 120 min long binary gradient of water and acetonitrile (B) containing 0.1% formic acid at a flow rate of 300 nL/min and a column temperature of 50 °C for CD11b+ microglia. A Data Independent Acquisition Parallel Accumulation–Serial Fragmentation (DIA-PASEF) method with a cycle time of 1.9 s was used for spectrum acquisition. Briefly, ion accumulation and separation via trapped ion mobility spectrometry (TIMS) was set to a ramp time of 100 ms. One scan cycle included one TIMS full MS scan followed by 17 DIA PASEF scans with two 26 m/z windows covering the m/z range from 350–1200 m/z.

The raw data were analyzed using the software DIA-NN version 1.8 [31] for protein label-free quantification (LFQ). One protein per gene canonical fasta database of Mus musculus (download date: January 12th 2023, 21957 entries) from UniProt and a fasta database with 125 potential human contaminations from MaxQuant [32]. Trypsin was defined as a protease. DIA-NN was used to generate a spectral library on the basis of the individual datasets and to perform label-free quantification. Two missed cleavages were allowed, and peptide charge states were set to 2–4. Carbamidomethylation of cysteine was defined as a static modification. Acetylation of the protein N-term as well as oxidation of methionine were set as variable modifications. The false discovery rate for both peptides and proteins was adjusted to less than 1%. Data normalization was disabled.

The identification of differentially expressed proteins between genotypes, nonnormalized intensities as the starting point, and subsequent pathway enrichment were performed for the corresponding RNA-seq data, with the exception of MSigDB version 2023.1 [30]. To facilitate direct comparisons between proteomic and RNA-seq data, the respective gene symbols were used instead of the UniProt protein IDs.

Homozygous PLCγ2-P522R KI and WT mouse pups (P0-P3) were decapitated, and the brains were removed and placed in 6-cm petri dishes in ice-cold HBSS (#24020091, Thermo Fisher). Meninges were removed under a microscope, and the brains were washed three times with ice-cold HBSS. For tissue lysis, 0.25% trypsin (#15090046, Gibco) was added, and the brains were incubated for 10 min at 37 °C in a cell culture incubator. The reaction was stopped by adding culture medium containing DMEM with 4.5 g/L glucose without L-glutamine (#ECB7501L, BioNordica), 10% (v/v) heat-inactivated fetal bovine serum (FBS, #10500–064, Gibco), 2 mM L-glutamine (#BE17-605E, Lonza Bioscience), and 1% (v/v) penicillin‒streptomycin (#DE17-602E, Lonza Bioscience). Next, 400 µl of 5 mg/ml deoxyribonuclease I (#D4527-40KU, Sigma) was added, and the cells were separated by pipetting carefully. The cells were then centrifuged for 10 min at 800 × g at room temperature and resuspended in culture medium. The cells were cultured in T-75 cell culture flasks coated with 0.02% poly-L-lysine hydrobromide (PLL, #P6407, Sigma) until the microglia were shaken one week later and then every 3‒4 days.

Myelin was isolated from the brains of adult male C57BL/6J mice by a decreasing sucrose gradient according to a protocol described previously [33, 34]. Briefly, the brain tissue was homogenized in a solution containing 10 mM HEPES, 5 mM EDTA, 0.32 M sucrose and protease inhibitor. The homogenate was layered on top of 0.85 M sucrose solution prepared in HEPES-EDTA buffer and centrifuged at 75 000 × g for 30 min with low acceleration and deceleration using a Beckman Ultracentrifuge with a Ti 50.2 rotor. The white myelin layer was collected from the interface, suspended in sterile water, and centrifuged at 75 000 × g for 15 min. The pellet was purified twice by hypo-osmotic shock using ice-cold water and centrifuged at 12 000 × g for 10 min. After that, the sucrose gradient and hypo-osmotic shocks were repeated to purify the myelin fraction. Finally, the myelin pellet was resuspended in HEPES-EDTA buffer and stored at − 80 °C. The concentration of myelin was determined using a BCA protein assay kit (#23225, Pierce).

The extracted myelin was labeled with a pHrodo^™ iFL Red Microscale Protein Labeling Kit (Thermo Fisher Scientific, P36014↗) by adding 25 µl of dye per 1 mg of myelin and incubating for 60 min at room temperature. The myelin was then pelleted by centrifuging for 10 min at 15 000 × g and washed twice with 1 × DPBS to remove the excess dye. pHrodo-labeled myelin was resuspended in 1 × DPBS at a concentration of 1 mg/ml and stored as aliquots at − 80 °C.

To analyze lipid accumulation in mouse primary microglia, 50 000 cells were plated onto a 48-well plate on top of glass cover slips precoated with 0.02% PLL in microglial culture medium (above) with reduced (5%) FBS. The next day, the cells were treated with 5 µg/ml LPS (#L5543; Sigma‒Aldrich) for 24 h or with 30 µg/ml myelin for 48 h. Control wells were left untreated. The cells were then washed three times with warm DPBS and stained with 2 µM BODIPY 493/503 (#D3922, Invitrogen) for 15 min in a cell culture incubator. The cells were rinsed several times with DPBS, fixed with 4% PFA for 20 min at room temperature and mounted on microscope glasses with a Vectashield Hard Set containing phalloidin (#H-1699, Vector Laboratories).

Images of primary microglia stained with BODIPY 493/503 to visualize neutral lipid droplets (LDs) and phalloidin to visualize the actin cytoskeleton (cell border) were acquired with a Zeiss Axio Observer inverted microscope with a Zeiss LSM 700 confocal module using a 20 × (NA 0.5) objective. Five to ten 1024 × 1024-pixel fields per cover slip were obtained. All the laser and detector settings were kept constant for all the samples. Imaging was performed by an investigator blinded to the genetic background and treatment of the cells.

LDs were analyzed via ImageJ (version 1.53c). LDs were segmented via the Intermodes automatic threshold method. Outliers with a radius of 3 and threshold of 50 were removed from the mask, after which particles larger than 100 pixels were included in the particle analysis. The number of LD-positive cells was manually counted. Analysis was performed by an investigator blinded to the genetic background and treatment of the cells. The image preprocessing and thresholding methods were kept constant throughout the analysis. The ZEN 2012 software Blue edition (Carl Zeiss Microimaging GmbH) was used for visualization purposes. The percentage of LD-positive cells among all the cells and the average LD size (µm²) per imaged field are shown. The numbers of analyzed images per group were as followed: WT untreated = 11, LPS = 22, myelin = 42 and KI PLCγ2-P522R untreated = 6, LPS = 12, myelin = 35.

The original iPSC line used in this study, carrying APOE4/4 and control PLCG2 genotype CC/APOE4/4 (UKBi011-A, https://hpscreg.eu/cell-line/UKBi011-A↗), was described previously [35]. The iPSC line CC/APOE3/3 (UKBi011-A3, https://hpscreg.eu/cell-line/UKBi011-A-3↗) was generated from the original line using CRISPR/Cas9 as previously described [36]. For the generation of homozygous PLCγ2-P522R lines on the APOE3/3 (GG/APOE3/3) and APOE4/4 (GG/APOE4/4) backgrounds, the gRNA CCCCAAAATGTAGTTCTGTA was used. The sequence of the ssODN used as an HDR donor was GTGAGACAGAAGGACCTGTCTAGTGATGCTGGGGTTTGGTCCAAGGCTTTCAGAAACCCCTCCTCTCTTTGCGGCCCAGGATATACGCCCGACGGAGCTACATTTTGGGGAGAAATGGTTCCACAAG.

Human iPSC lines were cultured in Essential 8 (E8) medium (#A1517001, Gibco) on 3.5 cm dishes coated with Matrigel (#356231, Corning) (#83.3900.500, Sarstedt) at 37 °C in 5% CO₂ and split with 0.5 mM EDTA every 4–5 days. For iMGL differentiation from iPSCs, a protocol described previously was used [37–39]. Briefly, iPSC colonies were detached using ReLeSR reagent (#5872, STEMCELL Technologies) and plated at a density of 3‒6 colonies per cm² on Matrigel-coated 6-well plates (#356231, Corning) in E8 medium supplemented with 5 µM ROCK inhibitor (#Y-27632, Merck). On the following day, differentiation into hematopoietic stem cells was initiated via a commercial STEMdiff Hematopoietic kit (#05310, STEMCELL Technologies). After 11‒13 days of differentiation, the floating hematopoietic progenitors were collected and plated at a density of 7000‒8000 cells per cm² on new Matrigel-coated 6-well plates in microglial differentiation medium consisting of DMEM/F12 (#21331020, Gibco), 2 × insulin-transferrin-selenite (#41400045, Invitrogen), 2 × B27 (#17504044, Invitrogen), 0.5 × N2, 1 × GlutaMAX (#35050‒038, Invitrogen), 1 × nonessential amino acids (#11140050, Invitrogen), 400 μM monothioglycerol (#M1753, Sigma), 5 μg/mL human insulin (#19278, Sigma), 100 ng/mL human IL-34 (#200‒34, PeproTech), 50 ng/mL human TGF-β1 (#100‒21, PeproTech), and 25 ng/mL human M-CSF (#300‒25, PeproTech). The cells were grown for 27 days, and fresh medium was added every other day. During the last 4 days of culture, microglial maturation was promoted by adding 100 ng/mL human CD200 (#77002, Biolegend) and 100 ng/mL human CX3CL1 (#300–31, PeproTech) to the cells.

To perform the cell Mito Stress test in the iMGLs, 40 000 cells per well were seeded in 200 μl of maturation medium one week before the experiment. Every other day, half of the medium was replaced with fresh medium. On the day of the experiment, Seahorse XF assay medium was prepared by adding 2 mM Glutamax (#35050–038, Invitrogen) to Seahorse DMEM. First, the cells were rinsed with 180 μl of Seahorse XF medium following the addition of Seahorse XF medium to the cells to a final volume of 180 μl. The cells were incubated in a non-CO₂ incubator for 1 h at 37 °C before being run on an XFe96 Analyzer. At the beginning of the assay, the XFe96 Analyzer added 10 mM glucose and 1 mM sodium pyruvate (#11360,039, Gibco) to the cells. Next, modulators of the electron transport chain were injected at a 1 μM concentration into the cells in the following order: oligomycin (#11342, Cayman Chemical), carbonyl cyanide-4 (trifluoromethoxy) phenylhydrazone (FCCP) (#15218, Cayman Chemical) and a mixture of rotenone (#13995, Cayman Chemical) and antimycin A (#A8674, Sigma‒Aldrich). During the assay, the oxygen consumption rate (OCR) was directly measured by the XFe96 Analyzer. The results were normalized to the cell confluence determined by an IncuCyte S3 (Sartorius) before the beginning of the assay. The results are shown from three independent Seahorse XF96 cell culture plates, each with 3‒4 technical replicates per line.

GraphPad Prism 10.1.2 software was used for all the statistical analyses and for the visualization of the data. The Shapiro‒Wilk test was used to test whether the data had a normal distribution. Accordingly, the statistical significance was tested via an unpaired samples t test or the Mann‒Whitney test. In the case of multiple dependent variables, two-way ANOVA together with Šídák's or Tukey’s multiple comparisons test was used. Correlation analyses were performed via Pearson correlation coefficients. P values less than 0.05 were considered statistically significant. All the data are presented as the means ± SEMs.

PLCG2 is highly expressed in microglia [3]. We have previously shown that the protective PLCγ2-P522R (rs72824905, p.P522R, Fig. 1A) variant increases microglia activation in WT mice with no β-amyloid pathology [7]. We now analyzed whether the PLCγ2-P522R variant promotes microglial activation in APP/PS1 (hereafter referred to as A+/P^wt/wt) mice using PET imaging with the TSPO ligand [18F]FEPPA, which is highly expressed in activated microglia [20]. To increase statistical power, data from female and male mice were pooled for analysis. Given the baseline difference between the sexes, the data were normalized to the average A+/P^wt/wt value within each sex. There was no difference in [18F]FEPPA ligand uptake between the genotypes in 7-month-old mice (Fig. 1B). However, in 13-month-old animals, increased [18F]FEPPA ligand uptake was observed in the whole brain (p = 0.032) of APP/PS1 x PLCγ2-P522R KI (hereafter referred to as A+/P^ki/ki) mice compared with A+/P^wt/wt mice. When the brain regions were examined separately, [18F]FEPPA ligand uptake was significantly increased in the pons (p = 0.013) and hippocampus (p = 0.012) (Fig. 1C). These findings suggest that microglial activation in the brain tissue of APP/PS1 mice carrying the PLCγ2-P522R variant moderately increases at 13 months of age.

Given that microglial activation was observed in the brain tissue of 13-month-old mice but not in those of younger animals, the following characterizations of brain pathology were carried out in female mice at this time point. In general, the APP/PS1 female mice used in this study presented more pronounced β-amyloid-related pathology and greater learning and memory impairments than their male counterparts did [40]. Therefore, we concentrated our subsequent assessments on female A+/P^ki/ki and A+/P^wt/wt mice. The PLCγ2-P522R variant was previously shown to reduce β-amyloid pathology in 5XFAD mice [2, 17]. To study whether the PLCγ2-P522R variant affects brain β-amyloid pathology in APP/PS1 mice, we stained β-amyloid plaques using the fluorescent amyloid stain X-34 [41]. Image analysis revealed that β-amyloid plaque number (normalized to the size of the analyzed area, p = 0.005) and β-amyloid plaque coverage (total plaque area % of the size of the analyzed area, p = 0.006) were both significantly lower in the entorhinal cortex of the A+/P^ki/ki mice than in that of the A+/P^wt/wt mice (Fig. 2A). However, a similar reduction was not detected in the hippocampus of the same mice (Supplementary Figures, Figure S2A). The average size of individual X-34+ β-amyloid plaques did not differ between the genotypes in either the entorhinal cortex (Fig. 2A) or the hippocampus (Supplementary Figures, Figure S2A).

The soluble and insoluble levels of Aβ40 and Aβ42 in the ELISA did not significantly differ in the temporo-occipital cortex between A+/P^ki/ki and A+/P^wt/wt mice (Fig. 2B). Similarly, there were no significant changes in the levels of soluble and insoluble Aβ40 and Aβ42 in the hippocampus of A+/P^ki/ki mice compared with those in the hippocampus of A+/P^wt/wt mice (Supplementary Figures, Figure S2B). Importantly, there were no differences in the levels of full-length APP (APPtot), APP C-terminal fragments (C83 and C99, Fig. 2C and Supplementary Figures, S2C, left panel), or soluble APPα and -β (sAPPα and -β, Fig. 2C and Supplementary Figures, S2C, right panel) in the temporo-occipital cortex or hippocampus according to Western blot analyses of these mice, suggesting that the histochemically determined decrease in the β-amyloid pathology of A+/P^ki/ki mice was not due to alterations in APP stability or processing.

To assess whether PLCγ2-P522R affects the microglial response to β-amyloid deposition, we quantified the IBA1+ area within and surrounding X-34-positive β-amyloid plaques in the APP/PS1 mouse brain (Supplementary Figures, Figure S1A). The IBA1+ area was increased within the 0–40 µm radius from the β-amyloid plaque outline in the entorhinal cortex (p = 0.03, Fig. 3A) and hippocampus (p = 0.037, Supplementary Figures, Figure S3A) of A+/P^ki/ki mice compared with A+/P^wt/wt mice. No difference was found in the IBA1+ signal overlapping the β-amyloid plaque area itself, either in the entorhinal cortex or hippocampus. The total IBA1+ area in the entorhinal cortex was marginally increased, whereas in the hippocampus, the total IBA1+ area was decreased (p = 0.043). Together, these data suggest that the PLCγ2-P522R variant increases the area of microglia around β-amyloid plaques and consequently, may resist β-amyloid deposition more efficiently.

Since activated microglia secrete chemokines and cytokines that promote astrocyte activation [42], we stained astrocytes with a glial fibrillary acidic protein (GFAP) antibody to study potential changes in astrocytes in the entorhinal cortex and hippocampus of APP/PS1 mice [42]. We found no differences in the total GFAP+ area or in the GFAP+ area overlapping or surrounding (within the 0–40 µm radius) X-34-positive β-amyloid plaques (Supplementary Figures, Figure S1B) in either the entorhinal cortex (Fig. 3B) or the hippocampus (Supplementary Figures, Figure S3B) of A+/P^ki/ki mice compared with A+/P^wt/wt mice, suggesting that PLCγ2-P522R variant-driven microglial activation does not promote astrocyte activation in the brain tissue of 13-month-old APP/PS1 mice.

Microglia clustered around β-amyloid plaques have been suggested to pack diffuse β-amyloid into a more compacted and less toxic form [43]. Hence, to investigate β-amyloid plaque morphology, we used X-34 to detect compact plaques and the anti-β-amyloid antibody 6E10 to detect diffuse plaques. Despite the evident improvement in the microglial response toward β-amyloid plaques, we found no differences in the total 6E10+ area or in the percentage of X-34+ or 6E10+ area of total β-amyloid pathology (X-34 + 6E10) either in the entorhinal cortex (Fig. 4A) or in the hippocampus (Supplementary Figures, Figure S4A) between the mice with different genotypes. Furthermore, no differences in the colocalization of 6E10 within the IBA1+ area were detected between the mice with different genotypes (Fig. 4B and Supplementary Figures, S4B), suggesting that microglial uptake of diffuse β-amyloid nor the ratio of diffuse and compact β-amyloid in the brain tissue are affected in 13-month-old APP/PS1 mice carrying the PLCγ2-P522R variant.

To assess the formation of dystrophic neurites around β-amyloid plaques in the brain tissue of APP/PS1 mice carrying the PLCγ2-P522R variant, brain sections were stained with the N-terminal anti-APP antibody 22C11. It is well documented that IHC staining with 22C11 shows the locations of swollen neuronal processes linked to abnormal protein accumulation, including abnormal N-terminal APP and phosphorylated Tau (p-Tau) accumulation, in different in vivo models of AD [44–47]. Since the 22C11 signal was strongly dependent on the Aβ plaque size (Pearson correlation coefficient (r) = 0.81, p < 0.0001) (Fig. 4C), we normalized the 22C11+ area around each plaque to the size of the analyzed plaque. The 22C11+ area overlapping the plaque area in the entorhinal cortex was significantly smaller (p = 0.033) in A+/P^ki/ki mice than in A+/P^wt/wt mice, and a similar but nonsignificant trend was observed within the 0–14 µm radius from the plaque outline (Supplementary Figures, Figure S1C and 4C). In the hippocampus, the 22C11+ area was significantly decreased within the 0–14 µm radius from the plaque outline (p = 0.010) (Supplementary Figures, Figure S4C). In addition to the N-terminal APP, dystrophic neurites contain various other proteins, including p-Tau [44–47]. Thus, the levels of p-Tau relative to total Tau (ratio of p-Tau/total Tau) were assessed to gain insight into the extent of p-Tau pathology and its association with dystrophic neurites in protein lysates from the temporo-occipital cortex and hippocampus of 13-month-old female A+/P^ki/ki and A+/P^wt/wt mice (Supplementary Figures, Figure S5A and S5B). Consequently, Western blot analysis did not reveal significant differences in the ratio of p-Tau/total Tau between A+/P^ki/ki and A+/P^wt/wt mice in the tissue lysates studied. Taken together, these findings suggest that the PLCγ2-P522R variant decreases the formation of dystrophic neurites around β-amyloid plaques, as shown by N-terminal APP staining, whereas a similar decrease in the p-Tau/total Tau ratio in the tissue lysates was not detected.

The well-established ability of microglia to compact β-amyloid deposits and protect surrounding neurons is considered beneficial. However, microglial activation also promotes neuroinflammation, which is harmful to neuronal health over time. Therefore, we next analyzed the levels of several inflammatory markers in the temporo-occipital cortex and hippocampal lysates of 13-month-old mice via an MSD mouse inflammatory panel. In general, A+/P^ki/ki mice presented lower levels of several inflammatory markers in both the temporo-occipital cortex and hippocampus than did A+/P^wt/wt mice (Fig. 5). The levels of IL-12-p70 (p70 refers to an active heterodimer, p = 0.049) and IL-6 (p = 0.003) in A+/P^ki/ki mice were significantly lower in the cortex, while those of IL-1β (p = 0.021) and IL-8 (p = 0.026) were lower in the hippocampus. These findings suggest that despite increased microglial activation, the PLCγ2-P522R variant does not increase chronic β-amyloid-associated neuroinflammation in the APP/PS1 mouse brain upon aging.

β-Amyloid plaque-associated APOE has been suggested to increase plaque compaction and support microglial responses toward β-amyloid deposition [48, 49]. To address whether APOE plays a role in the PLCγ2-P522R variant-mediated microglial response to Aβ, we analyzed the APOE+ area within and surrounding plaques. We found no differences in the APOE+ area overlapping or within a 10 µm radius from the plaque outline (Supplementary Figures, Figure S1D) between A+/P^ki/ki and A+/P^wt/wt mice in the entorhinal cortex (Fig. 6A) or in the hippocampus (Supplementary Figures, Figure S6A). In line with these findings, soluble and insoluble APOE levels in the temporo-occipital cortex and hippocampus did not differ between the mice with different genotypes (Supplementary Figures, Figure S6B).

The upregulation of Apoe is part of the disease-associated microglia (DAM) signature [5]. DAM activation is regulated by Trem2 expression and is needed for restricting β-amyloid pathology in AD [5]. Hence, to study whether the protective variant exacerbates the RNA signature related to DAM activation in the brain tissue of 13-month-old APP/PS1 mice, RNA sequencing was carried out in the bulk temporo-occipital cortex and in CD11b+ microglia acutely isolated from the whole brain. The expression of both Apoe and Trem2 in cortical tissue and CD11b+ microglia did not differ between the A+/P^ki/ki and A+/P^wt/wt mice, suggesting that the DAM signature is not upregulated in APP/PS1 mice carrying the PLCγ2-P522R variant (Fig. 6B and C). In fact, no DEGs or enriched biological pathways were detected in the temporo-occipital cortex (Supplementary Figures, Figure S6D) or isolated CD11b+ microglia (Supplementary Figures, Figure S6E) when A+/P^ki/ki were compared with A+/P^wt/wt mice.

At the protein level, shedding of soluble TREM2 (sTREM2) terminates the signaling of full-length TREM2 and therefore influences microglial activation [18]. To assess whether PLCγ2-P522R modulates TREM2 shedding, we measured TREM2 levels in the PBS soluble fraction of temporo-occipital and hippocampal lysates. There were no differences between mice with different genotypes in the levels of sTREM2 in these brain areas (Supplementary Figures, Figure S6C), suggesting that TREM2 shedding is not promoted in the mice carrying the PLCγ2-P522R variant.

We next wanted to investigate whether the expression of genes related to cell motility and phagocytosis was altered in CD11b+ microglia isolated from female A+/P^ki/ki mice to understand the increased microglial area around β-amyloid plaques and the reduced pathology of β-amyloid plaques. More specifically, we examined the expression of fibronectin and integrin classes α and β genes, as described by Obst et al., which reported decreased cell adhesion and motility in human iPSC-derived macrophages upon PLCγ2 deficiency [50]. We also assessed other well-known cell motility and phagocytosis genes via gene set enrichment analysis (GSEA) core enrichment gene sets. Although several targets related to cell motility and phagocytosis were slightly upregulated in A+/P^ki/ki microglia compared with A+/P^wt/wt microglia, these results were not statistically significant (Supplementary Figures, Figure S7A and S7B).

Given that β-amyloid deposition and associated neuronal pathologies are mitigated in the brain tissue of APP/PS1 mice carrying PLCγ2-P522R, we next wanted to investigate whether this protective variant influences the behavioral alterations observed in APP/PS1 mice. First, the body weight showed an APP/PS1 x PLCγ2-P522R genotype interaction (p = 0.04), such that the APP/PS1 wild-type (-) A-/P^ki/ki mice had a lower weight than the A-/P^wt/wt littermates did (p = 0.024), but the same was not detected among the APP/PS1 hemizygous (+) mice (Fig. 7A). Given that the PLCγ2-P522R variant did not influence the ambulatory distance traveled in the spontaneous activity test (Fig. 7B), the observed weight reduction was not related to increased activity. However, the time spent in the center of the light–dark box was significantly shorter in the P^ki/ki than in the P^wt/wt mice, regardless of the APP/PS1 genotype (p = 0.011, Fig. 7C), suggesting that the PLCγ2-P522R variant increases the avoidance of open places. In the light‒dark box, the mice were given a free choice between compartment the illuminated and dark chamber. In this test, ANOVA revealed a main effect of both the PLCγ2-P522R (p = 0.03) and APP/PS1 (p = 0.04) genotypes and their interaction (p = 0.004). The P^ki/ki mice showed increased avoidance of the illuminated chamber, which was further accentuated in mice with the APP/PS1+ genotype (Fig. 7D). In the passive avoidance test, which was performed one day after the mice had received a foot shock when they entered the dark chamber, ANOVA revealed an even stronger main effect of both the PLCγ2-P522R (p < 0.0001) and APP/PS1 (p = 0.02) genotypes and their interaction (p < 0.0001). On the day after the foot shock, the P^ki/ki mice rushed into the dark chamber, whereas the P^wt/wt mice were hesitant to enter (Fig. 7E). This tendency was further accentuated by the APP/PS1+ genotype. Given that P^ki/ki and P^wt/wt mice did not differ in their motor abilities on the rotating rod (Fig. 7F) and that they had similar pain thresholds (Fig. 7G), this outcome was not explained by decreased motor ability or impaired ability to sense pain. In fact, the P^ki/ki mice had a stronger behavioral response (crab-like walking backward) to foot shock than did the P^wt/wt mice (Fig. 7H). Finally, the PLCγ2-P522R variant did not influence spatial learning or memory in the Morris water maze in either APP/PS1- or APP/PS1+ mice (Fig. 7I), whereas APP/PS1+ mice exhibited strongly impaired learning ability in terms of escape latency (p < 0.001). This finding suggests that rushing into the dark chamber behavior is not a result of memory dysfunction related to the foot shock in the P^ki/ki mice.

To elucidate the molecular pathways associated with the PLCγ2-P522R variant, global RNA sequencing and proteomics analyses were performed in CD11b+ microglia. As the PLCγ2-P522R variant did not promote transcriptomic changes on an APP/PS1 background, the analyses were also carried out in microglia acutely isolated from 13-month-old male and female PLCγ2-P522R KI and WT mice. RNA sequencing revealed dramatically different RNA signatures between male PLCγ2-P522R KI and WT microglia, including 3719 DEGs (FDR 0.05) (Supplementary Figures, Figure S8A). Among these genes, 1302 were downregulated, and 2417 were upregulated. Gene enrichment analysis revealed several enriched biological pathways in the upregulated and downregulated DEGs between PLCγ2-P522R KI microglia and WT microglia (Fig. 8A). Among these pathways, calcium signaling, fatty acid (FA) metabolism, cholesterol homeostasis, and oxidative phosphorylation were prominently enriched in the upregulated DEGs in PLCγ2-P522R KI microglia, whereas pathways related to the inflammatory response and the interferon-α (IFN-α) response were enriched in the downregulated DEGs. Closer examination revealed that the core DEGs within the calcium signaling pathway included Itpr1, Itpka, Camk4, Camk2a and Camk2b, whereas the calcium-sensitive MEF2 transcription factor family members Mef2a, Mef2c, and Mef2d were downregulated in PLCγ2-P522R KI microglia (Fig. 8B; Supplementary Figures, Figure S8C). The upregulated genes within the FA metabolism pathway included, e.g., genes linked to LD lipolysis and FA synthesis (Mgll, Fasn, Fads2, Fabp7, Fabp5, Fabp3, Fig. 8B; Supplementary Figures, Figure S8C). Interestingly, several genes encoding PI3K subunits, Pik3cd, Pik3cg, Pik3r5, Pik3ap1, Pik3r1, and Pik3r4, were downregulated (Supplementary Figures, Figure S8C). Furthermore, several mitochondrial complex I (Ndufa1, Ndufa2, Ndufc2, Ndufb8, Ndufs6, Ndufa4, Ndufa8, Ndufa5, Ndufa3, and Ndufb6), complex II (Sdha), complex III (Uqcrh, Uqcrq, and Uqcr10), and complex IV (Cox5a, Cox7a2, Cox7b, Cox6a1, Cox4i1, Cox6c, Cox5b, Cox6b1, and Cox7c) genes as well as genes encoding other mitochondrial enzymes involved in FA oxidation (FAO, e.g., Eci1) were upregulated. Finally, the downregulated DEGs in PLCγ2-P522R KI microglia related to the inflammatory response, TLR signaling and the IFN-α response included Tlr9, Tlr3, P2rx4, P2rx7, and Casp8 (Fig. 8B). Conversely, similar RNA-seq analysis of CD11b+ microglia from female mice did not yield as prominent results as microglia from male mice did (Supplementary Figures, Figure S8E). Thus, we decided to focus on assessing microglia isolated from males via proteomic analyses instead of microglia isolated from females to increase the likelihood of detecting significant changes owing to the PLCγ2-P522R variant.

Proteomic analysis revealed 34 differentially expressed proteins (DEPs, FDR 0.05), 14 of which were downregulated and 20 of which were upregulated in male PLCγ2-P522R KI microglia compared with WT microglia (Fig. 8C; Supplementary Figures, Figure S8B). DEPs and biological pathways were only mildly associated with changes detected in gene expression. In contrast to gene expression, core enriched proteins related to oxidative phosphorylation were downregulated in the proteomics dataset (Fig. 8C). In contrast, proteins related to FA oxidation were mildly upregulated in PLCγ2-P522R KI microglia, which was similar to what was observed at the gene expression level (Supplementary Figures, Figure S8D). Together, the results of the RNA and proteomic analyses suggest that pathways related to calcium signaling, the inflammatory/IFN response, mitochondrial oxidative metabolism (FAO), and lipid/FA metabolism are modulated in acutely isolated microglia from male mice carrying the PLCγ2-P522R variant.

The downregulation of genes involved in LD dynamics, including FABPs, has been shown to be associated with lipid accumulation and impaired mitochondrial function in TREM2 KO and PLCγ2 KO iMGLs [6, 8]. As several genes associated with LD accumulation (Pi3k subunits) and LD lipolysis (e.g., Fabp) were differentially expressed in CD11b+ PLCγ2-P522R KI mouse microglia (Supplementary Figures, Figure S8C), we next addressed whether the PLCγ2-P522R variant influences LD accumulation under LPS- and myelin-induced stress conditions. Both LPS and myelin have previously been shown to cause lipid dyshomeostasis and promote LD accumulation in similar models [10, 51]. Compared with no treatment, LPS and even more myelin treatment increased the number of LD-positive microglia, whereas genotype had no effect on the percentage of LD-accumulating microglia (Fig. 9). However, the size of the LDs was clearly smaller in PLCγ2-P522R KI microglia than in WT microglia (p < 0.0001) upon myelin treatment. A similar but nonsignificant trend was also detected in LPS-treated microglia. These findings suggest that PLCγ2-P522R KI reduces the buildup of LDs in microglia under conditions of cellular stress.

To assess whether PLCγ2-P522R-driven mechanisms are translated into human microglial models, we conducted RNA sequencing of human MDMi cells derived from PLCγ2-P522R variant carriers (CG) and matched control individuals (CC) under untreated conditions and after myelin and LPS treatments. Compared with those in CC MDMi cells, only 11 significantly altered DEGs were identified in the untreated group, and 10 DEGs were identified in the myelin-treated group in the CG (Fig. 10A-B). LPS had the most robust effect on gene expression, resulting in 236 DEGs in CG cells compared with CC cells (Fig. 10C). Under LPS-treated conditions, gene enrichment analysis revealed the enrichment of oxidative phosphorylation and MYC targets (V1 and V2) in the upregulated DEGs, whereas pathways related to the inflammatory response, TNF-α signaling, the IFN-γ response, and KRAS signaling were most clearly enriched pathways in the downregulated DEGs in the MDMis of CG carriers (Fig. 10D). The core enriched genes in the oxidative phosphorylation pathway included ACADVL, an enzyme regulating FA oxidation; the mitochondrial complex I gene; NDUFB7; mitochondrial ATP synthase; ATP5MC1; and FXN, a gene associated with mitochondrial iron transport and respiration (Fig. 10E). Upregulated genes linked to the MYC target pathway and involved in RNA regulation included FARSA, MRTO4, and WDR74. Several IFN (IRF9, IRF4, and STAT4), TNF (TNFAIP2, TNFAIP3, and TNFAIP6) and IL (IL1-α and IL-7R) family members, immune sensors (CD14, TLR8, TLR2, and NLRP3), and related transcription factors (NFKB1 and NFAT5) were downregulated and associated with the enriched pathways. To examine whether a reduction in inflammatory cytokine expression can be detected at the protein level, IL-6 and TNF-α levels were measured in the conditioned medium of LPS-treated MDMi cells. However, no significant changes were found in the levels of either of the cytokines (Supplementary Figures, Figure S9A). Together with findings in mouse CD11b+ microglia, these data confirm a central role for PLCγ2-P522R in the regulation of genes associated with mitochondrial function and FA oxidation as well as inflammatory/IFN responses.

To further investigate mitochondrial function in human microglia, we employed PLCγ2-P522R homozygous (GG) iMGLs and isogenic control lines (CCs) with APOE3/3 or APOE4/4 backgrounds (Supplementary Figures, Figure S9B-C). In general, the mitochondrial OCR significantly differed between the genotypes (p < 0.0001). Specifically, maximal respiration was greater in PLCγ2-P522R GG/APOE3/3 iMGLs than in all the other groups (CC/APOE3/3 p = 0.001, CC/APOE4/4 p < 0.0001, CG/APOE4 p < 0.0001) (Fig. 10F). In addition, the spare respiratory capacity was significantly greater in GG/APOE3/3 cells than in CC/APOE4/4 (p = 0.009) and GG/APOE4/4 (p = 0.027) iMGLs. However, PLCγ2-P522R did not increase mitochondrial respiration in the APOE4/4 background. Taken together, both the mouse and human data in this study point toward promoted lipid/FA metabolism and improved mitochondrial function in microglia from individuals carrying PLCγ2-P522R variant.

The animals were raised and handled at the Laboratory Animal Center of the University of Eastern Finland, Kuopio, Finland. All mouse studies were carried out in accordance with the guidelines of the European Community Council Directives 86/609/EEC and approved by the Regional State Administrative Agency and Project Authorization Board (ESAVI/7315/2024, EKS-004–2019). All study protocols concerning human samples were approved by the Medical Research Ethics Committee of Wellbeing Services County of North Savo (formerly the Medical Research Ethics Committee of North Savo Hospital District, 833/2023, 123/2016). Participant recontacting was approved by the Scientific Steering Committee of Auria Biobank (BB_2020-0062).

All the authors have approved the final version of the manuscript and provided their consent for publication.

CH collaborates with Denali Therapeutics and is a member of the advisory boards of AviadoBio and Cure Ventures.

Dataset generated and/or analyzed in the current study are included within the article and its additional files. The RNA sequencing datasets of mouse cortical tissue and isolated CD11b+ microglia are available in the Gene Expression Omnibus (NCBI) repository, GSE277167↗, https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc↗ = GSE277167↗. The study participant consent does not allow opening the sequencing data of the carrier-derived MDMi cells which was generated and analyzed during the current study, but they are available from the corresponding authors (MT or MH) on a reasonable request.