Background
Although symptoms of cognitive impairment and memory decline in Alzheimer’s disease (AD) ultimately manifests through neuronal dysfunction, AD is also obligately associated with neuroinflammation where glial cell types show chronic activation in AD brain [
1,
23]. Recent identification of genetic variants linked to enhanced risk in sporadic AD onset through genome-wide association studies (GWAS) implicate numerous risk genes which feature enriched expression in microglia, including TREM2 and CD33 [
9,
22,
25,
34,
50]. As an immune receptor almost exclusively expressed in microglia and myeloid cell types, mutational variants in TREM2 such as the rare R47H mutation has been shown to significantly increase risk of AD onset [
22,
34], where TREM2 R47H has been shown to alter Aβ plaque morphology, microglial energy metabolism, and impaired binding to TREM2 ligands such as APOE and Aβ oligomers [
3,
60,
65‐
67]. While a role for TREM2 in microglia has been described for microglial response to Aβ, whether TREM2 can similarly alter progression of tau pathology during AD onset is unclear. Recent studies suggest
Trem2 deletion (
Trem2 KO) can aggravate tau hyperphosphorylation and enhance morphological activation phenotypes in mouse models of tauopathy [
7]. Further,
Trem2 KO or TREM2 R47H promote pathological tau seeding in an APP/PS1 AD mouse model [
42];
Trem2 deletion also aggravated tau spreading/pathology in a P301L tau/PS2 APP mouse model [
41] as well as in 5xFAD brain injected with tau aggregates from human AD brain [
21], together suggesting that in addition to Aβ [
69], TREM2 dysfunction can also promote tau pathology. At this point, however, it is unclear how TREM2 can mediate tau dispersion and whether these mechanisms can impact tau-dependent impairment of cognitive function.
Given that TREM2 is expressed exclusively in microglia in brain, it seems likely that modulation of TREM2 may influence certain aspects of microglia behavior or function to consequently aggravate downstream tau pathology. Although microglia do not express tau, a potential role for microglia in altering tau pathology and spreading has been previously implicated in mouse models of AD [
15]. Deletion of the murine fractalkine receptor CX3CR1 was previously shown to enhance microglial activation and reduce Aβ pathology in APP/PS1 mouse models [
40], and was associated with enhanced tau pathology and memory [
10]. Adoptive cell transfer of CX3CR1 knockout (KO) microglia could likewise induce tau pathology in hTau mouse brain [
45], indicating that microglial activation could potentially aggravate tau pathogenesis. In agreement with this, recent evidence indicates that assembly of the NLRP3 inflammasome (comprising NLRP3, ASC and caspase-1) during microglia activation is required for tau hyperphosphorylation and aggregation [
32]. Intriguingly, inflammasome activation is also induced through microglial deletion of an essential mediator of autophagosome biogenesis, Atg7, resulting in aggravated tau pathology and tau spreading in PS19 mouse brain [
64]. Although enhancement of tau pathology is observed with NLRP3, CX3CR1 or microglial Atg7 deletion [
10,
32,
45,
64], depletion of microglia in Tg4510 (P301L tau) mice at 12 months of age using the PLX3397 CSF1R inhibitor showed little effect on tau burden [
8], despite upregulation of disease-associated microglia (DAM) gene signatures [
8,
38,
39]. This suggests that microglia activation at early stages of tau pathology may be required to aggravate tau spreading and pathological accumulation.
Although growing evidence indicates that microglia can alter tau pathology, exact cellular mechanisms underlying microglia-dependent tau dispersion remain somewhat unclear. Previous studies demonstrate microglia can propagate pathological tau through exosomal extrusion/transmission [
2]. Furthermore, pathological tau has been observed in microglia in PS19 (P301S tau) mouse brain [
2], and tau seeds have been observed in microglia isolated from human AD brain, as well as in transgenic Tg4510 (P301L tau) mouse brain [
27]. Microglia have also been shown to convert, package and release microvesicles containing Aβ species with enhanced toxicity [
35]. Despite growing evidence that microglia exosomes can transduce proteotoxic protein species such as Aβ and tau, it remains unknown whether and how genetic components related to AD risk genes such as TREM2 and APOE can influence the dispersion and pathology of tau through exosome-associated mechanisms.
Herein, we observed that Trem2 deletion enhanced transfer of human P301L tau from the medial entorhinal cortex (MEC) to the hippocampal dentate gyrus (DG) region in mouse brain, which correlated with increased tau pathology, reduced synaptic transmission and impaired spatial and fear memory. Using a microfluidic 3-chamber assay system, we show that Trem2 KO microglia can transfer tau between isolated neuronal layers and demonstrate enhanced trafficking of internalized tau into pre-exosomal trafficking compartments, whereas treatment with exosome inhibitors suppressed extracellular tau extrusion in both WT and Trem2 KO microglia. Purified exosomes from Trem2 KO microglia exposed to tau oligomers also feature elevated levels of tau and show enhanced tau-seeding competency in vitro and induced enhanced pathological tau phosphorylation in WT mouse brain compared to exosomes from WT microglia. Together, these results implicate Trem2 as a suppressor of pathological tau dispersion and demonstrate that Trem2 deletion can aggravate pathological tau spreading through microglial extracellular exosomes.
Discussion
Early studies characterizing the progression of human neurofibrillary pathology during AD progression noted distinctive spreading of tangles originating from the entorhinal region to isocortical regions at later stages of disease [
11]. Interestingly, tau pathology appears to spread progressively along neuroanatomically interconnected regions within the brain [
24,
56,
61], potentially through the dispersion of tau aggregates between synaptically adjoined neurons to promote seeded aggregation in unaffected downstream cells [
16,
17,
19,
43]. Various mechanisms have been proposed to describe how intracellular tau aggregates are transferred between cells primarily through studies in vitro; potential means of intracellular tau transfer include cell to cell spreading through interconnected membrane nanotubes [
59], vesicle-free mechanisms through tau secretion [
36,
49], or unconventional pore-mediated tau extrusion [
12]. Despite these observations, whether tau is exclusively transmitted between neurons and how transmission occurs remains somewhat unclear.
Interestingly, recent evidence indicates that microglia play a fundamental role in mediating tau spreading through tau extrusion and dispersion in extracellular exosomes [
2], which may involve stimulation of exosome release through the activation of purinergic receptors such as P2RX7 [
52]. Microglia feature a distinctive expression profile in AD mouse brain, and modulate a specific subset of “disease-associated” or “neurodegenerative” microglial (DAM, MGnD) transcripts [
38,
39]. Tetraspanin components such as Cd9 previously shown to mediate exosomal biogenesis [
13] are upregulated in microglia from 5xFAD mouse brain [
38]; however, how age-related AD stress potentially affects microglia exosomal biogenesis remains yet unclear. Given that genomic studies have identified enrichment of microglia gene variants related to AD risk such as TREM2 [
9,
22,
25,
34,
50], it seems likely that pathological events such as exosome dysregulation may be affected by altered function of TREM2 and other microglial AD risk variants.
Although our results revealed no difference in internalization of tau oligomers, we observed enhanced trafficking of tau to the late endosome and compartments containing the ESCRT-I component Tsg101 and the MVB-enriched tetraspanin, CD63, suggesting that
Trem2 deletion can enhance enrichment of certain cargo such as tau to exosome-destined compartments. Interestingly,
Trem2 deletion has been recently shown to alter exosomal content in macrophages, where increased levels of the micro RNA miR-106b-5p in macrophage-derived,
Trem2 KO exosomes induced mitochondrial impairment in hepatocytes [
28]. Additionally, iPSC-derived microglia from heterozygous carriers of the AD-associated R47H allele also featured alterations in exosomal content, and exosomes derived from TREM2 R47H microglia showed reduced protection from H
2O
2 toxicity in SH-SY5Y cells [
44]. Thus, growing evidence indicates that TREM2 dysfunction can alter exosomal content and potential downstream toxicity. At this point however, how
Trem2 deletion or dysfunction can influence exosome biogenesis or trafficking remains unclear. Interestingly, our proteomic analysis indicates that
Trem2 deletion can potentially alter proteins associated with exosomes, intracellular exosome biogenesis, and tau-binding with tau and LPS/ATP stimulation, suggesting that
Trem2 deletion may alter proteomic profiles linked to tau trafficking and extrusion pathways.
Although genetic perturbations associated with microglia activation such as deletion of CX3CR1, NRPL3 or Atg7 have been previously shown to aggravate tau pathology as discussed above [
10,
32,
45,
64], the connection between microglia activation and tau appears to be complex. With respect to microglia activation,
Trem2 has been shown to be essential in the transition of microglia from homeostatic to late-stage DAM states in neurodegenerative mouse brain [
38,
39]. Microglia are thought to transition between homeostatic and an early DAM state (stage 1 DAM) independent of
Trem2 which involves the downregulation of homeostatic genes such as
Cx3cr1 and the purinergic receptor
P2ry12 [
38].
Trem2-dependent progression from early to late-stage DAM (stage 2 DAM) involves upregulation of various transitional markers (Lpl, Cst7, Cd9, Axl) which are associated with upregulation of various pathways linked to lysosomal, phagocytic and lipid metabolic function. Since
Trem2 deletion is anticipated to suppress microglia conversion from homeostatic to disease-associated states, one possibility is that cellular changes uncoupled to DAM or MGnD transcriptional signatures could account for enhanced tau spreading in
Trem2 KO microglia. In support of this, our proteomics analysis implicates alteration of exosome and tau-associated proteins in cultured
Trem2 KO microglia with tau or LPS/ATP stimulation that may enhance exosome-mediated tau spreading with
Trem2 deletion.
In agreement with our results indicating that
Trem2 deletion can aggravate tau dispersion, previous studies have shown that tau pathology and/or spreading is enhanced in most tau mouse model systems where
Trem2 is downregulated or deleted. For example, shRNA-mediated
Trem2 downregulation [
33],
Trem2 haploinsufficiency [
18,
55],
Trem2 deletion [
7] in mouse tauopathy models, or
Trem2 deletion in 5xFAD mouse brain injected with tau derived from human AD brain (“AD-tau”) [
21] can aggravate tau pathology which together, suggests that
Trem2 normally restrains pathological effects associated with tau. Interestingly, some differences with respect to
Trem2 haploinsufficiency (
Trem2 +/−) and homozygous deletion (
Trem2 −/−) have been observed with respect to tau pathology in mice; while
Trem2 haploinsufficiency enhances pathological ptau (AT8) [
18,
55] and fibrillary tau forms [
55], homozygous
Trem2 deletion potentially attenuates ptau and fibrillary tau in PS19 brain [
55]. Although this indicates that homozygous
Trem2 deletion may have reduced pathogenicity compared to
Trem2 haploinsufficiency, we note that since tau pathology was quantified at late stages (8–9 months) in PS19/Trem2
+/− and PS19/Trem2
−/− mice [
55], analysis at an advanced pathological age may mask potential effects on tau spreading which would occur during early stages of pathological tau accumulation. In support of this notion, depletion of microglia showed little or no effect on tau plaque burden in aged 12 month-old Tg4510 mouse brain [
8], further suggesting that microglia alterations at early stages impact late stage pathology in mouse tauopathy models.
Additionally, pathological context may also manifest differences observed between
Trem2 haploinsufficiency and homozygous deletion; recent evidence indicates that homozygous
Trem2 deletion can affect tau pathology in a P301L tau, PS2APP (PS2 N141I; APP KM670/671NL) combined mouse line, with little effect in
Trem2 WT or haploinsufficient (het) animals [
41]. Given that
Trem2 has been shown to bind directly to Aβ and alter microglia response [
66,
67], it is tempting to speculate that
Trem2 haploinsufficiency can still limit tau pathogenesis in response to Aβ proteotoxicity. Aside from differences in tau pathogenesis with homozygous and heterozygous deletion, downregulation of
Trem2 function appears to have deleterious consequences in exacerbating tau pathology and spreading. As our results here only provide evidence that homozygous
Trem2 deletion can enhance tau spreading, future work may determine how
Trem2 levels can differentially affect global tau pathology (phosphorylation and conformation) and tau spreading, and its relationship to Aβ.
Although our results here and from other published studies implicate an active role for microglia activation in aggravating tauopathy, the question remains why microglia depletion in some instances, for example in 5xFAD mouse brain with AD-tau injection, can also aggravate tau pathology [
21], while other studies that show reduction in tau pathology in PS19 mouse brain with microglia depletion [
2]. One possibility may be that microglia-independent mechanisms of tau transduction may predominate differently in various combined Aβ/tau or tau-only mouse models, which may involve pathological effects between cell types. For example, direct transfer of tau between neurons has been demonstrated, and appears to be enhanced by neuronal stimulation [
63], and recent evidence also indicates that tau transfer can also occur between neurons and astrocytes [
46]. This suggests that in addition to contributions from microglia, tau transduction mechanisms may involve other cell types which may introduce additional effects in the presence of amyloid proteotoxicity.
Methods
Mouse models
Trem2 knockout (KO) (Stock no. 027197) and Wild-type (WT) C57BL/6 J mouse strains were purchased from The Jackson Laboratory (JAX). Both KO and WT animals were maintained as homozygous lines. Postnatal day 2–3 WT and Trem2 KO mice were used for dissection and generation of primary microglial cultures. Since hormonal effects are negligible at this stage, mixed primary microglial cultures were taken from both males and females. Animal procedures did not include additional drug or treatment regimen other than that described. Mouse lines were housed with littermates with free access to food and water under a 12 hour light/day cycle. All animal procedures, including husbandry were performed under the guidelines of the Institutional Animal Care and Use Committee at Sanford Burnham Prebys Medical Discovery Institute.
Stereotaxic AAV injection
Control and recombinant AAV-P301L tau (AAV9) vectors have been described previously; AAV-GFP and AAV-P301L tau plasmid vectors were generously provided by Dr. Tsuneya Ikezu [
2]. 1 × 10
9 AAV particles were stereotaxically/bilaterally injected into the MEC (layer II/III) using the following coordinates: anteroposterior, − 4.75 mm; lateral, + 2.9; dorsoventral, − 4.6. Mice were then subjected to behavioral analysis, or histological staining 14 (behavior only) or 35 (histology and behavior) days following stereotaxic injection.
Immunohistological staining and analysis
Mice stereotaxically injected with AAV particles were sacrificed for immunohistological analysis at 35 days post-injection. Mice were anesthetized with 4% isoflurane and intracardially perfused with PBS. Brain tissues were harvested and fixed in 4% paraformaldehyde at 4C for 24 h. Tissues were washed in PBS and cryoprotected in PBS containing 30% sucrose. Tissues were embedded in OCT containing 30% sucrose (at 1:1 v/v) and free-floating coronal brain cryostat sections (25 mm) were collected. Data were collected and analyzed in a double-blind fashion.
For detection of Iba1, ptau, and tau, brain slices were stained using antibodies as follows: Goat anti-Iba1 (1:400, ab5076, Abcam), AT8 (1:500, MN1020 ThermoFisher), AT180 (1:500, MN1040 ThermoFisher), T13 (1:500, #835201 Biolegend), PSD95 (1:200, #3450 s Cell Signaling), Cd68 (1:250, MCA341R BioRad), P2y12 (1:200, #69766 Cell Signaling Technology), Cd9 (1:100, 20,597-I-AP Proteintech), Tmem119 (1:150, #90840 Cell Signaling), Clec7a (1:100, Invivogen), GFP (1:250, NB600–308 Novus). Alexa Fluor 488, 568 or 647 secondary antibodies (1:400, ThermoFisher) were used, DAPI counterstains were applied to the sections, and image z stacks were acquired from multiple sections (up to 9) from each animal using a Zeiss LSM 710 laser-scanning confocal microscope.
Staining intensity of human tau in cultured neurons was quantified using Imaris image analysis software (Bitplane, Oxford Instruments), and phospho-tau epitopes quantified in histological sections were scored manually by identifying AT8 or AT180-positive cells in imaged MEC or DG regions. Quantification of PSD95 puncta in the stratum moleculare within the hippocampus was also performed using Imaris by defining a region of interest (ROI) defined immediately above of the granular cell layer (GCL). The number of PSD95 puncta in each ROI was measured using Imaris, and calculated using a set intensity threshold, expressed as number of PSD95 puncta/mm2. Percentage of P2y12, Cd68, Tmem119, Clec7a, CD9 positive microglia and P2y12/CD68, Tmem119/Clec7a, CD9/Iba1 double positive microglia was quantified using Imaris.
Behavioral analysis
Barnes maze test was performed as described previously [
4]. Briefly, WT or
Trem2 KO mice stereotaxically injected with AAV-tau or AAV control (AAV9-synapsin GFP) were habituated on day 1, where mice were placed in the center of the maze underneath a clear 3500-ml glass beaker for 30 s. Mice were slowly guided to the target hole leading to the escape cage by moving the glass beaker over the target hole within a span of 10–15 s. Mice were then given 3 min to enter the target hole, and gently forced to enter in the event no entry was apparent. Training was initiated the next day. Mice were placed inside an opaque cardboard cylinder, 10″ tall and 7″ in diameter, in the center of the Barnes maze for 15 s. The cylinder was then removed, and mice were allowed to explore the maze for 2 mins. Upon entering the escape cage, the mouse was allowed to remain in the escape cage for 1 min; otherwise, the mouse was gently guided to the escape hole using a glass beaker and allowed to freely enter the escape cage. In case the mouse did not enter the escape cage within 3 min, it was gently nudged with the beaker to enter. Five trials were performed during training, with 3 trials on day 1, and 2 trials on day 2. Forty-eight hours after the last training session, a probe test to analyze the behavioral characteristics of the mice seeking the target area was monitored using a digital camera controlled by the ANY-maze video tracking system (Stoelting Co.). Subsequent analyses of the probe test parameters were processed using ANY-maze software, where statistical analyses and significance values were calculated using GraphPad Prism (Dotmatics, Boston, MA).
Contextual fear conditioning behavior tests were performed using the Freeze Detector System (San Diego Instruments, CA). Twenty-four hours before training initiated, mice were placed in the conditioning chamber and allowed to freely explore the chamber for 5 mins. On the first day of training, mice were placed in the conditioning chamber and allowed to freely explore for 120 s, where a 0.4 mA electrical foot-shock was subsequently applied to the mice for 2 s. After 60 s, another 0.4 mA electric shock was given to the mouse for 2 s. Following shocks, mice were left in the chamber for an additional 120 s. Twenty-four hours after training, each mouse was monitored in the same chamber for 5 min. Freezing time was automatically recorded and analyzed by the Freeze Detector System.
Electrophysiology
Following behavior analysis, ex vivo hippocampal slices were prepared from WT and
Trem2 KO mice stereotaxically injected with either AAV-control or AAV-tau using methods described previously [
31]. Briefly, mice were decapitated under deep terminal anesthesia, and brains were surgically removed in ice-cold, sucrose-based artificial cerebrospinal fluid (aCSF) (190 mM sucrose, 25 mM D-glucose; 25 mM NaHCO
3, 3 mM KCl, 1.25 mM NaH
2PO
4, 5 mM MgSO
4, 10 mM NaCl, and 0.5 mM CaCl
2) saturated with carbogen (95% O
2/5% CO
2) at pH 7.4. A vibrating-blade microtome (Leica VT1000S) was used to cut 400-μm-thick coronal slices containing both cortex and hippocampus. Slices were transferred to a holding chamber containing a warmed (32 °C) aCSF formulation for recording (125 mM NaCl, 25 mM NaHCO
3, 3.0 mM KCl, 1.25 mM NaH
2PO
4, 2.0 mM CaCl
2, 1.0 mM MgSO
4, and 10 mM D-glucose) saturated with carbogen (95% O
2/5% CO
2) at pH 7.4. Slices were left to recover at room temperature in oxygenated aCSF for at least 30 min before recording. Population spike amplitude was measured to determine synaptic transmission within the excitatory perforant pathway in acute hippocampal slices. Concentric bipolar stimulating electrodes were positioned in the middle molecular layer of the DG while recording electrodes were positioned in the dentate granule cell body layer. Stimuli (0.1 ms in duration) were applied at 0.05 Hz in increments of 20 μA from 0 to 200 μA, at each time-point, five recordings of evoked responses were averaged.
Primary microglial and neuronal culture
Primary microglial cultures were prepared as described previously [
3,
68]. Briefly, brains were removed from WT or
Trem2 KO mice at postnatal day 2–3. After removal of the meninges, brains were treated with a Papain Dissociation System (Worthington Biochemical Corporation) according to manufacturer’s specifications. Mixed glial cells were plated in flasks coated with poly-D-lysine and grown in DMEM containing 10% FBS (VWR Life Science Seradigm). Twenty-five nanograms per milliliter GM-CSF (R&D Systems) was added into the cultures after 5 days and removed before harvesting. Microglial cells were harvested twice by shaking (200 rpm, 60 min) 10–14 days after plating and subjected to various treatments within 24 h of harvest.
Neurons were dissected as described previously [
58]. Primary neurons were dissected from embryonic days 17–18 (E17-E18) embryos from pregnant female C57BL/6 mice, and hippocampal and cortical neurons were isolated by microdissection from the cerebral cortex and hippocampus using a stereomicroscope. Tissue was dispersed by trypsin and DNase 1 digestion for 30 minutes at 37 °C, followed by trituration in DMEM+ penicillin /streptomycin + HEPES. Neurons were maintained separately on poly-D-lysine coated coverslips or seeded onto poly-D-lysine coated layers in microfluidic chamber systems. Neurons were cultured in Neurobasal Medium Plus supplemented with B27 Plus, glutamine, and penicillin/streptomycin, where half of the media was replaced every 2–3 days.
3-chamber interneuronal tau dispersion assay
A three-layer microfluidic chamber (TCND1000, Xona Microfluidics) was designed to reconstitute intraneuronal tau dispersion through cells within an intermediary layer, with microgrooves adjoining the three chambers. Two side-reservoirs connected the poly-D-lysine coated culture chambers to facilitate neuronal adhesion. Primary cortical neurons from wildtype C57BL/6 J E18 embryos were dissected as previously described [
58] (see “Primary microglial and neuronal culture”) and 2 × 10
5 cells were plated in chamber layers 1 and 3 in 50% neurobasal with B27 medium and 50% DMEM/F12 with FBS medium, and allowed to adhere overnight. The following day, medium was changed to complete neurobasal with B27 and maintained at 37 °C in 5% CO
2. Neurons in layer 1 were transduced with 1 × 10
10 VP/ml AAV-tau particles at DIV2. At DIV7, primary WT or
Trem2 KO microglia were detached from mixed glial cultures and 1 × 10
4 cells were seeded into chamber layer 2 in 50% Neurobasal with B27 containing/50% DMEM with 10% FBS, 25 ng/ml GM-CSF (a control without microglia in layer 2 was also included); and neurons/microglia were cultured for another 7 days. At DIV14, chambers were fixed and stained to visualize tubulin (#5568, Cell Signaling Technology), Iba1 (ab5076, Abcam) and human tau (T13, Biolegend) by fluorescence microscopy.
Intensity of human tau (T13) staining and neuronal density in layers 1 and 3 were calculated from intensity measurements using Imaris (Bitplane) in imaged fields, and neuronal number were quantified by DAPI measurements in layers 1 and 3 (normalized per cm2). Iba1-positive microglia were also quantified in chamber layer 2, and overlap in human tau with Iba-1 positive microglia were quantified, and normalized to Iba1/tau-positive microglia in WT layer 2 (set to 1.0).
Microglia tau uptake
Binding/uptake of tau oligomers in cultured WT and Trem2 KO microglia was measured using purified recombinant 2N4R human tau 1–441 (500 μg/ml, #AS-55556-50, AnaSpec) oligomerized in 30 μM heparin (#07980, StemCell Technologies Inc.) for 24 h. Tau oligomers were subsequently conjugated to Alex555 (“tau-555”) using an Alexa Fluor 555 Microscale Protein Labeling Kit (#A30007, ThermoFisher) according to the manufacturer’s instructions. Binding/uptake of tau-Alexa Fluor 555 was assayed in microglia cultures seeded at 50,000 cells in 24-well plates, and tau-555 binding/uptake was measured in real time at a final concentration of 10 μg/ml where a fixed area in each well was serially imaged every 15 min using a Nikon N-SIM microscope. Serial confocal images were acquired for 10 h, and tau-555 intensity/area was quantified. Fluorescence intensity was normalized to microglia cell number using automated IMARIS imaging software (Bitplane); fluorescence thresholds for tau-555 were set to a value of 10, and individual cells were identified by differential interference contrast (DIC) imaging. Fluorescence measurements normalized to cell number from 9 total confocal images from 3 independent microglia batches/experiments (three independent wells per batch) at time points ranging from 15 to 180 min (for all experiments). The phagocytic index (PI) for varying time points was then calculated using the following formula: PI = It / I15, where It represents averaged fluorescence intensity at various time points and I15 represents fluorescence intensity at 15 min following addition of tau-555. The resulting value represents fold change of tau-555over the 15 min time point for each microglia culture. Ratios were then normalized to the PI value in WT microglia under control conditions at 15 min (where WT microglia under control conditions/15 min are set to 1.0) for each experiment.
Flow cytometry and TUNEL analysis
Flow cytometry for tau oligomer uptake
Tau oligomers were conjugated to an Alexa-488 fluor label using an Alexa Fluor™ 488 Protein Labeling Kit (A10235, ThermoFisher); WT and Trem2 KO microglia were incubated with tau-Alexa488 (tau488) for 6 hours. Microglia were washed in 1xPBS, harvested and fixed with 4% PFA. Fluorescent signals were detected and quantified by flow cytometry (Novocyte, ACEA Bioscience). WT microglia (“cell only”) without tau uptake was used as an unlabeled control.
TUNEL analysis
Effects of LPS and ATP treatment on cell death in WT and Trem2 KO microglia were quantified by TUNEL labeling and flow cytometry analysis. 1 × 106 WT and Trem2 KO microglia left untreated or treated with 1 μg/ml LPS (#L3024, Sigma-Aldrich) for 3 h, and 5 mM ATP, or WT microglia exposed to 10 μM camptothecin for 3 h were processed for TUNEL staining using the Apo-Direct TUNEL Assay Kit (APT110, Millipore Sigma) according to the manufacturer’s recommendations, and subjected to flow cytometry and FlowJo analysis. TUNEL staining in WT and Trem2 KO microglia under varying treatments was performed using the Click-IT Plus TUNEL assay system (C10617, ThermoFisher), and stained cells were imaged by confocal microscopy.
Induction and purification of microglia exosomes
Exosomes were purified from primary microglia cultures as described previously [
2,
54]. Briefly, 1 × 10
7 WT and
Trem2 KO microglia cultured under untreated conditions or incubated with 2.5 μg/ml tau oligomers for 24 h; cells were then washed in 1xPBS and incubated in DMEM with 10% exosome-depleted FBS (A2720803, ThermoFisher) and treated with 1 μg/ml LPS (#L3024, Sigma-Aldrich) for 3 h, and 5 mM ATP for 15 mins. Conditioned media (5 ml) was collected and centrifuged at 2000×g, and supernatant subsequently centrifuged at 10,000×g for 30 min at 4 °C to remove cell debris. Supernatants were then diluted to 10 ml in 1x PBS and centrifuged at 100,000×g for 90 mins. To precipitate exosomes, pellets were then resuspended in 10 ml 1xPBS and re-centrifuged at 100,000×g to remove contaminants and non-exosomal debris. Resulting pellets were then resuspended in 50 μl 1xPBS for electron microscopy (EM), ELISA, or RIPA buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1% NP-40, 0.1% SDS, 0.5% DOC) for immunoblot analysis.
Electron microscopy
Formvar-carbon-coated copper grids (100 mesh, Electron Microscopy Sciences, Hatfield, PA) were placed on 20 μl drops of each sample solution displayed on a Parafilm sheet. After allowing material to adhere to the grids for 10 minutes, grids were washed 3 times by rinsing through 200 μl drops of milli-Q water before being left for 1 min on 2% (wt/vol) uranyl acetate (Ladd Research Industries, Williston, VT). Excess solution was removed with Whatman 3MM blotting paper, and grids were left to dry for a few minutes before viewing. Grids were examined using a JEOL JEM-1400Plus transmission electron microscope operating at 80 kV. Images were recorded using a Gatan OneView 4 K digital camera.
Immunoblot and ELISA analysis
Exosome preparations or lysates from primary microglia were generated/resuspended in RIPA buffer in the presence of protease and phosphatase inhibitors (#78430, ThermoFisher). Proteins were separated by SDS-PAGE using 4–20% gradient gels, transferred onto nitrocellulose, and blocked in 5% non-fat milk in 1xPBS. Blots were then probed with primary antibodies overnight in 5% BSA/1xPBS, washed in 1xPBS with 0.1% tween-20, and probed with HRP-conjugated secondary antibodies. Blots were then incubated with ECL and immunoblot signals were acquired using a Chemidoc imaging system (BioRad). ELISA measurements for exosomal human tau were performed using ELISA kits for tau (KHB0041, ThermoFisher) according to specifications supplied by the manufacturer. Primary antibodies used for immunoblot analysis include Alix (1:1000, 2171 Cell Signaling), T13 (htau) and actin (1:5000, A5441 Sigma).
Sample preparation for proteomics analysis and LC-MS/MS
Primary Trem2 KO or WT microglia were left untreated, or treated with tau oligomers (2.5 μg/ml for 24 h), or 1 μg/ml LPS for 3 h, and 5 mM ATP for 15 mins (run 2), and cell pellets were lysed in UAB buffer (8 M urea, 50 mM ammonium bicarbonate (ABC) and Benzonase 24 U/100 ml). Protein concentration was determined using BCA assays (ThermoFisher) according to the manufacturer’s instructions. Proteins were then reduced by the addition of 5 mM tris(2-carboxyethyl) phosphine (TCEP) at 30 °C for 60 min, followed by alkylation of cysteines with 15 mM iodoacetamide (IAA) for 30 minutes in the dark at room temperature. Urea concentration was reduced to 1 M by adding 50 mM ammonium bicarbonate. Samples were digested overnight with Lys-C/trypsin (Promega) at room temperature with constant agitation at a 1:25 enzyme:protein ratio. Following digestion, samples were acidified using 0.1% FA and desalted using AssayMap C18 cartridges mounted on an Agilent AssayMap BRAVO liquid handling system. Cartridges were sequentially conditioned with 100% acetonitrile (ACN) and 0.1% FA; samples were then loaded, washed with 0.1% FA, and eluted with 60% ACN, 0.1% FA. Peptide concentration was determined using a NanoDrop spectrophotometer (Thermo Fisher).
Samples were subjected to mass spectrometry analysis using an EASY nanoLC system (ThermoFisher). Buffer A consisted of H2O/0.1% FA; Buffer B consisted of 80% ACN/0.1% FA. Samples were separated over a 90-min gradient of increasing Buffer B on analytical C18 Aurora column (75 μm × 250 mm, 1.6 μm particles; IonOpticks) at a flow rate of 300 nL/min. The mass spectrometer was operated in positive data-dependent acquisition mode, and the Thermo FAIMS Pro device was set to standard resolution with the temperature of FAIMS inner and outer electrodes set to 100 °C. A three-experiment method was set up where each experiment utilized a different FAIMS Pro compensation voltage: - 50, − 70, and − 80 Volts, and each of the three experiments had a 1 second cycle time. A high resolution MS1 scan in the Orbitrap (m/z range 350 to 1500, 60 k resolution at m/z 200, AGC 4e5 with maximum injection time of 50 ms, RF lens 30%) was collected in top speed mode with 1-second cycles for the survey and the MS/MS scans. For MS2 spectra, ions with charge state between + 2 and + 7 were isolated with the quadrupole mass filter using a 0.7 m/z isolation window, fragmented with higher-energy collisional dissociation (HCD) with normalized collision energy of 30% and the resulting fragments were detected in the ion trap as rapid scan mode with AGC of 5e4 and maximum injection time of 35 ms. The dynamic exclusion was set to 20 sec with a 10 ppm mass tolerance around the precursor.
Proteomic data analysis
Raw files were searched with SpectroMine software (Biognosys, version 2.7.210226.47784) using the BGS default settings. The search criteria were set as follows: full tryptic specificity was, 2 missed cleavages were allowed, carbamidomethylation (C) was set as fixed modification and oxidation (M) as a variable modification. The false identification rate was set to 1%. Spectra were searched against the curated Uniprot
mus musculus database including common contaminants from the GPM cRAP sequences. Data was further processed using the MSstats package (version 4.2) in R [
14]. We avoided use of imputation of missing values prior to statistical test using MSstats; instead we calculated a pseudo Log
2 fold-change (L2FC), adj.pvalue and pvalue of proteins completely missing in one condition after failing to perform the statistical test. The imputed (pseudo) L2FC was calculated as the sum of intensities of the protein (i.e., sum of feature intensities of a given protein within a given sample) across all replicates of the same group that it was detected, divided by 3.3. On the other hand, the imputed pvalue and adj.pvalue was calculated by dividing 0.05 or 0.1, respectively, by the number of replicates that a given protein was confidently identified multiplied by the number of features quantified. Therefore, the imputed L2FC gives an estimate of the protein abundance in condition that it is detected, while the imputed pvalue or adj.pvalue reports the confidence of the imputation in the sense of consistency of protein detection in the group that it is detected.
Proteins with an adjusted
p value< 0.05 (adjp< 0.05) were selected as significantly differentially expressed proteins (DEPs). GO analysis was performed by entering DEPs into the GO DAVID input interface, and KEGG, Biological Process (BP), Cellular Component (CC) and Molecular Function (MF) categories were retrieved for each query [
29,
30]. Principal Component Analysis (PCA) was carried out in R version 4.1.2 with PCATools package (version 2.6.0) using log2 protein intensity for all proteins summarized by
dataProcess function from MSstats (version 4.2). Contaminant proteins (non-murine, “
Bos taurus” proteins) or proteins with gene names that failed to map from gene ID’s were manually removed from the datasets prior to analysis. To calculate z-score values within each replicate, row (protein-wise) z-scores were computed in R version 4.0.2 using the scale function by subtracting mean intensity of each protein from the corresponding intensities of the biological replicates, and dividing the resulting values by the standard deviation of the intensities.
Tau internalization and quantification
WT and Trem2 KO microglia were seeded at a density of 0.2 million cells on coverslips on 24-well plates, and treated with 2.5 μg/ml tau oligomers for 0.5, 2, 4 and 24 h. Cells were then fixed and stained with antibodies to visualize colocalization of recombinant human tau (T13) with Rab5, Rab7, LAMP1, CD63 and Tsg101. Images were acquired by confocal microscopy and overlapping signals from tau and intracellular markers were quantified; manual identification, demarcation and quantification of regions of overlap were performed using Imaris, and normalized to the total area of tau staining in fluorescence images.
In vitro exosomal tau seeding assay
Tau seeding was performed using a FRET biosensor HEK293T cell line stably expressing tau-RD P301S-CFP and P301S-YFP (tau-RD cells) as described previously [
26]. To assay tau seeding capacity of exosomes purified from cultured microglia, exosomes were purified from WT or
Trem2 KO microglia pre-incubated with tau oligomers and induced with LPS/ATP and resuspended in 50 μl PBS. Tau-RD cells were then exposed to 20 μl purified exosomes and transduced with Lipofectamine 2000 (#11668019, ThermoFisher) for 48 hours, and FRET activity from tau aggregation was visualized by excitation at 405 nm and consequent fluorescence at 525/50 nm was imaged using a Zeiss LSM 710 microscope. DIC images were concurrently acquired in FRET images and used to normalize FRET activity according to cell number. FRET area was calculated from positive cell signals in tau-RD cells through defined FRET-positive regions in Imaris, and FRET-positive area (μm
2) was normalized over cell number; cells without FRET signals were included in the analyses.
Stereotaxic exosome injection and analysis in vivo
Exosomes were prepared from 3 × 10
7 tau-loaded and untreated (control), LPS/ATP-treated WT and
Trem2 KO microglia as described under “
Induction and purification of microglia exosomes”. Exosome pellets were resuspended in 20 μl sterile 1xPBS and 2 μl of the exosome preparations purified from WT and
Trem2 KO microglia were stereotaxically injected into the DG region (coordinates: AP, − 2.0 mm; ML, ± 1.3 mm; DV, 2.1 mm) in 6 month C57BL/6 WT mice. Three weeks post-injection, mice were perfused with 4% PFA and stained with AT8 (MN1020, Thermo Fisher), β3-Tubulin (5568, Cell Signaling Technology) and DAPI to examine tau pathology. AT8-positive cells were scored in 2.5 × 10
5 μm
2 imaging areas in independently-injected animals.
Statistical analysis
All statistical analyses were performed using R scripts (proteomics) or Graphpad Prism as indicated.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.