Graphical abstract graphic file with name fx1.jpg [53]Open in a new tab Highlights * • Inpp5d haplodeficiency rescues microglial loss around Aβ plaques in Tyrobp KO mice * • Restored microglia affect plaque compaction, astrogliosis, and p-tau deposition * • TREM2/TYROBP and INPP5D exert opposing effects on a common PI3K signaling pathway __________________________________________________________________ Biological sciences; Neuroscience; Molecular neuroscience; Immunology; Immune system; Immunity; Immune response; Omics; Transcriptomics Introduction Alzheimer disease (AD) is pathologically characterized by the extracellular deposition of amyloid β (Aβ) fibrils in the brain parenchyma, the intraneuronal aggregation of tau proteins, and neuronal cell loss. The Aβ deposition drives the pathogenesis of AD, while tau aggregation is directly linked to neuronal dysfunction and ultimately cell death. Furthermore, recent whole-exome sequencing and genome-wide association studies have suggested that microglia also contribute to the pathogenesis of AD.[54]^1^,[55]^2^,[56]^3 A rare variant of triggering receptor expressed on myeloid cells 2 (TREM2) (p.R47H) is associated with the second-highest odds ratio for late-onset AD after APOE ε4.[57]^4^,[58]^5 TREM2 encodes a microglia-specific transmembrane receptor, which recognizes various extracellular ligands (e.g. lipids, Aβ, and lipoproteins) and transmits signals via its binding partner, TYRO protein tyrosine kinase-binding protein (TYROBP), to downstream targets, including Syk, phosphoinositide 3-kinases (PI3Ks), and mitogen-activated protein kinases (MAPKs).[59]^6^,[60]^7^,[61]^8^,[62]^9 AD-associated TREM2 variants disrupt TREM2-ligand interaction,[63]^10^,[64]^11^,[65]^12^,[66]^13^,[67]^14^,[68]^15 suggesting a loss-of-function mechanism underlying the AD pathogenesis. Loss or disease-associated mutations of TREM2 affect microglial clustering around amyloid plaques in mouse models of Aβ pathology as well as in human, and exacerbate neuritic dystrophy around plaques.[69]^11^,[70]^16^,[71]^17^,[72]^18^,[73]^19^,[74]^20 Importantly, dystrophic neurites are regarded as the place where tau is phosphorylated and aggregated,[75]^21 and Trem2 deficiency accelerates it only when Aβ pathology is present.[76]^22^,[77]^23 These suggest a crucial protective role of microglia and TREM2 in Aβ-driven spreading of tau pathology.[78]^24 Moreover, TREM2 sustains microglial proliferation/survival,[79]^11 maintains energy metabolism of microglia,[80]^25^,[81]^26 permits conversion of homeostatic microglia into disease-associated microglia (DAM)[82]^27 or neurodegenerative microglia,[83]^28 and regulates cerebral glucose uptake by microglia.[84]^29^,[85]^30^,[86]^31 These beneficial roles of TREM2 may offer a therapeutic opportunity in AD. Indeed, TREM2 agonistic antibodies enhance microglial responses to Aβ, promote microglial proliferation, and reduce Aβ burden and its neurotoxic effects[87]^32^,[88]^33^,[89]^34^,[90]^35^,[91]^36 and are starting to enter clinical trials to test their therapeutic potential. Considering the important roles of TREM2, it is reasonable to assume that other AD risk genes also converge on the same pathway. Among them, we focused on inositol polyphosphate-5-phosphatase D (INPP5D), a gene specifically expressed in microglia in the brain.[92]^2 INPP5D hydrolyzes phosphatidylinositol-(3,4,5)-trisphosphate [PI(3,4,5)P[3]] at its 5′-position and negatively regulates the function of this lipid second messenger.[93]^37 Since PI(3,4,5)P[3] activates many signaling molecules such as AKT serine/threonine kinase (AKT), INPP5D regulates a broad range of cellular events, including inflammation, cell proliferation, and cell motility.[94]^38 An AD risk variant of INPP5D (rs35349669) is associated with increased expression of INPP5D protein.[95]^39 In particular, this intronic polymorphism is thought to upregulate a specific truncated isoform of unknown function by promoting its internal transcription initiation.[96]^2 Intriguingly, INPP5D inhibits TREM2 signaling during osteoclastogenesis.[97]^40 However, the role of INPP5D in AD pathogenesis has not been investigated. In this study, we aimed to clarify the pathological role of INPP5D in mutant App knockin AD model mice (App^NL-G-F/NL-G-F; hereafter NLGF).[98]^41 Inpp5d-haplodeficient NLGF mice showed no effects on Aβ burden or plaque-associated changes of microglia, astrocytes, or neurites. In Tyrobp-deficient NLGF mice, however, Inpp5d reduction ameliorated the impaired microglial clustering around plaques, which was associated with plaque compaction, the reappearance of reactive astrocytes, and the reduction of phosphorylated tau in nearby neurites. Bulk RNA sequencing (RNA-seq) analysis of isolated microglia and immunohistochemical profiling showed that Inpp5d haplodeficiency had a minor rescue effect on DAM signature gene expression in Tyrobp-deficient microglia. Instead, an improvement in microglial adhesion to amyloid plaques or an increase in PI(3,4,5)P[3] levels was suggested. These data suggest that Inpp5d haplodeficiency ameliorates several defects in Tyrobp-deficient mouse microglia, thereby affecting p-tau deposition. Results Inpp5d haplodeficiency ameliorates the pathological changes in Tyrobp-deficient NLGF mice In NLGF mice, brain expression of INPP5D was restricted in parenchymal microglia and particularly high in plaque-associated microglia ([99]Figure 1A), similar to a previous report.[100]^42 To test the pathological role of INPP5D, we investigated the effect of Inpp5d haplodeficiency in NLGF mice, as homozygous but not heterozygous Inpp5d knockout mice develop lethal pulmonary inflammation.[101]^43 However, Inpp5d^+/−;NLGF mice did not show any detectable difference in Aβ deposition, microglial accumulation around amyloid plaques, astrocyte activation, or neurite dystrophy compared with NLGF mice ([102]Figures S1 and [103]S2). Figure 1. [104]Figure 1 [105]Open in a new tab Inpp5d haplodeficiency restored microglial clustering around amyloid plaques in Tyrobp-deficient NLGF mice (A) Immunohistochemistry for INPP5D (magenta), IBA1 (green), and Aβ[1-x](blue) in the brain cortical area of NLGF mouse. Scale bar: 50 μm. (B) Schematic showing inhibitory role of INPP5D on the TREM2/TYROBP-mediated PI(3,4,5)P[3] signaling pathway. (C) Flow cytometric quantification of INPP5D in isolated microglia from 5-month-old mice. Mean fluorescence intensity (MFI) of each genotype was shown after subtracting the value for Inpp5d^−/−;NLGF microglia (N = 2) as a background. (D) Immunohistochemistry for IBA1 (green) and Aβ[1-x] (magenta) in the hippocampal region of the indicated genotypes. Scale bar: 50 μm. (E) Quantification of the plaque coverage by IBA1-positive cells (%) in the hippocampus and cortex. (F) Frequency distribution of the plaque coverage by IBA1-positive cells, classified into dense (>40%), intermediate (20%–40%) and sparse (<20%). (G) Quantification of the mean distance of PU.1-positive myeloid cells from the center of the plaque in the hippocampus. (H) The number of plaque-associated myeloid cells per plaque (right) and the overall density of myeloid cells (left) in the hippocampus were assessed by PU.1 staining (N = 8–11). (I) Representative super-resolution images of Thioflavin S-labeled Aβ plaques of the indicated mice. Note that the brightness/contrast of images was differently adjusted for each image to highlight the morphological differences between groups. Scale bar: 10 μm. (J) The mean fluorescence intensity of thioflavin labeling on plaques was determined and shown by normalizing to the average of NLGF group as 100% (N = 9–10). (K and L) The size (K) and number (L) of plaques in the brain slices from the indicated mice were determined by anti-Aβ[1-x] antibody (82E1) staining for the cortex and hippocampus (Hippo) (N = 8–11). (M and N) Abundance of Aβ40 and Aβ42 in TBS-soluble (M) and FA-soluble (N) fractions was determined from the cortex and hippocampus of the indicated mice by ELISA. Data were represented by normalizing to NLGF mice as 100% (N = 3–11). Each point represents an individual mouse. Bars represent the mean ± SEM of all analyzed animals in each genotype. Nine-month-old male mice were analyzed except for C. One-way ANOVA was used except for F using two-way ANOVA, followed by Tukey’s multiple comparisons. ns: not significant (p > 0.05), ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. See also [106]Figures S1–S3. TREM2/TYROBP exerts its neuroprotective function possibly via their downstream molecules including PI3K.[107]^25^,[108]^44 Consistently, gene ontology (GO) term enrichment analysis of publicly available RNA-seq data[109]^20 showed enrichment for the term related to PI3K pathway as being downregulated by Trem2 deficiency ([110]Figure S3A). Considering that INPP5D inhibits PI3K functions and has an antagonistic role against TREM2 signaling,[111]^40 we hypothesized that INPP5D negatively regulates TREM2/TYROBP functions in the brain ([112]Figure 1B). To investigate their genetic interaction in vivo, we asked whether Inpp5d haplodeficiency could modulate phenotypes of Tyrobp^−/−;NLGF mice. Reduced expression of INPP5D was first confirmed in isolated microglia from Inpp5d^+/−;Tyrobp^−/−;NLGF mice by flow cytometry ([113]Figure 1C) as well as by immunoblotting of whole brain lysates ([114]Figures S3B and S3C). We then performed immunohistochemistry of the brains for Aβ and IBA1 (a myeloid cell marker) to analyze microglial association with plaques ([115]Figure 1D). At 9 months of age, Tyrobp^−/−;NLGF mice showed reduced microglial coverage of plaques versus NLGF mice, as assessed by the overlap of Aβ and IBA1 staining ([116]Figures 1E and [117]S3D). Specifically, the proportions of plaques showing high and intermediate (>20%) microglial coverage were decreased, whereas that of low (<20%) coverage was increased ([118]Figure 1F). Interestingly, additional haplodeficiency of Inpp5d almost normalized these defects ([119]Figures 1E, 1F, and [120]S3D). P2RY12-positive (brain resident) microglia were comparable between Inpp5d^+/−;Tyrobp^−/−;NLGF and NLGF mice ([121]Figure S3E), suggesting that the repopulated cells are of brain origin. Immunolabeling of PU.1, which visualizes microglial nuclei,[122]^45 revealed that the distance between individual plaques and surrounding microglia was affected in Tyrobp^−/−;NLGF mice, but it was also restored in Inpp5d^+/−;Tyrobp^−/−;NLGF mice ([123]Figures 1G and [124]S3F). For the number of plaque-associated microglia, Tyrobp^−/−;NLGF mice showed a significant reduction compared with NLGF mice, without affecting the overall density of microglia in brain sections ([125]Figure 1H). Interestingly, however, Inpp5d haplodeficiency did not restore the number of plaque-associated microglia in Tyrobp^−/−;NLGF mice. Taken together, these suggest that Inpp5d deficiency restores the microglial association with plaques, rather than promoting their proliferation/survival. Microglia form a physical barrier around Aβ deposits and promote the compaction of Aβ into inert, dense core plaques. Impaired microglia-plaque association in Trem2- or Tyrobp-deficient mice leads to plaque remodeling into a filamentous shape.[126]^17^,[127]^18^,[128]^20^,[129]^46 We therefore stained amyloid plaques with Thioflavin S, which only detects inert plaques, and examined their morphology using super-resolution microscopy ([130]Figure 1I). In contrast to the compact morphology of the plaques in NLGF mice, plaques in Tyrobp-deficient mice were often devoid of such a dense core, and instead were composed of reticular fibrils radiating from the vacant center. Notably, the plaques of Inpp5d^+/−;Tyrobp^−/−;NLGF mice demonstrated intermediate morphological characteristics between those of NLGF and Tyrobp^−/−;NLGF mice; the plaques had a dense central core and were surrounded by fibrillar Aβ deposits. Quantification of the plaque-associated thioflavin fluorescence showed a significant reduction of the staining in Tyrobp-deficient NLGF mice, and the reduction was corrected by additional Inpp5d haplodeficiency ([131]Figure 1J). Next, we asked whether Inpp5d haplodeficiency affects Aβ burden by anti-Aβ antibody (82E1) staining, which visualizes filamentous Aβ deposition. There was no detectable difference in the size or the number of plaques among NLGF, Tyrobp^−/−;NLGF, and Inpp5d^+/−;Tyrobp^−/−;NLGF mice ([132]Figures 1K, 1L, and [133]S3G). In addition, the amounts of Aβ40 and Aβ42 in the Tris-buffered saline-soluble and the formic acid-soluble fractions of brains were almost indifferent among all the genotypes ([134]Figures 1M and 1N). In summary, Inpp5d reduction restores plaque morphology in Tyrobp-deficient NLGF mice without affecting Aβ burden by facilitating the microglial clustering around plaques. Reactive microglia trigger astrocyte activation that is also implicated in the amyloid pathology in mice and humans.[135]^47^,[136]^48 Astrocytes in NLGF mice were activated around plaques, as indicated by glial fibrillary acidic protein (GFAP) positivity ([137]Figure 2A), which is absent in the cortex of wild-type mice.[138]^49^,[139]^50 The immunoreactivity was reduced in Tyrobp^−/−;NLGF mice ([140]Figure S4), similar to previous findings that Trem2 deficiency or microglial depletion reduces astrocyte activation.[141]^17^,[142]^20^,[143]^51^,[144]^52 Interestingly, Inpp5d haplodeficiency restored the GFAP expression in Tyrobp^−/−;NLGF mice ([145]Figures 2A and 2B). These results suggest that astrocyte activation depends on the physical association of microglia with amyloid plaques, which is restored by haplodeficiency of Inpp5d. Figure 2. [146]Figure 2 [147]Open in a new tab Inpp5d haplodeficiency restored astrocytic activation in Tyrobp-deficient NLGF mice (A) Representative images of GFAP staining around the somatosensory cortex. Scale bar: 200 μm. (B) Quantification of the GFAP-positive area (%) per unit area of the cortex (N = 4). Each point represents an individual mouse. Bars represent the mean ± SEM of all analyzed animals in each genotype. Nine-month-old male mice were analyzed. One-way ANOVA followed by Tukey’s multiple comparisons. ns: not significant (p > 0.05), ∗p < 0.05, ∗∗p < 0.01. See also [148]Figure S4. Apolipoprotein E (ApoE) is a unique factor that facilitates both plaque-associated microgliosis and plaque compaction.[149]^53 While ApoE is predominantly expressed by astrocytes, it is upregulated by plaque-associated microglia. In addition, recent reports in β-amyloidosis mice showed that Trem2 deficiency affects the abundance of plaque-associated ApoE.[150]^20^,[151]^54 We therefore asked whether Inpp5d deficiency alters ApoE deposition around plaques and whether such alteration correlates with observed changes in microgliosis, astrogliosis, and plaque morphology. In NLGF mice, ApoE immunoreactivity was prominent around amyloid plaques ([152]Figure 3A), but the labeling was unchanged in Tyrobp^−/−;NLGF mice unlike Trem2-deficient mice ([153]Figure 3B). In contrast, the ApoE deposition was significantly upregulated in Inpp5d^+/−;Tyrobp^−/−;NLGF mice, revealing an unexpected role of INPP5D in the ApoE regulation. However, no obvious change of plaque-associated ApoE in Tyrobp^−/−;NLGF mice suggests that ApoE is not a sole factor that can explain the phenotypes observed in our models. Figure 3. [154]Figure 3 [155]Open in a new tab Inpp5d haplodeficiency increased ApoE deposition around amyloid plaques in Tyrobp-deficient NLGF mice (A) Representative images of ApoE staining around the somatosensory cortex. Scale bar: 100 μm. (B) Quantification of the ApoE-positive area (%) per unit area of the cortex (N = 6–9). Each point represents an individual mouse. Bars represent the mean ± SEM of all analyzed animals in each genotype. Male mice at 11–12 months of age were analyzed. One-way ANOVA followed by Tukey’s multiple comparisons. ns: not significant (p > 0.05), ∗p < 0.05. Dystrophic neurite is an axonal swelling surrounding plaques, which reflects neuronal damage by Aβ. Because microglia restrain the dystrophic neurite formation by encapsulating plaques,[156]^18^,[157]^55 we asked the effect of Inpp5d haplodeficiency on this phenotype. We first examined beta-site APP cleaving enzyme 1 (BACE1), a known marker of dystrophic neurites.[158]^56 Tyrobp^−/−;NLGF mice had a larger BACE1-immunoreactive area compared with NLGF mice. However, additional Inpp5d haplodeficiency did not affect the BACE1 deposition ([159]Figure 4A). Considering that the molecular composition of dystrophic neurite is changed in disease progression,[160]^57 we also analyzed other markers, namely, LAMP1,[161]^55 ubiquitin,[162]^58 and phosphorylated tau (p-tau; pS202/pT205).[163]^17^,[164]^18 Similar to BACE1, the levels of these proteins around plaques were increased in Tyrobp^−/−;NLGF mice compared with NLGF mice ([165]Figures 4B–4D). Interestingly, p-tau immunoreactivity was specifically attenuated around plaques of Inpp5d^+/−;Tyrobp^−/−;NLGF mice, whereas that of neither LAMP1 nor ubiquitin was altered compared with Tyrobp-deficient mice ([166]Figure 4D). To ask why this effect is specific to p-tau, we compared the accumulation of p-tau and BACE1 in disease progression ([167]Figure 5A). As expected, overall levels of both markers increased concomitantly with amyloid deposition ([168]Figures 5B and 5C). However, the mean intensity of BACE1 per plaque area was already high as early as 4 months of age and remained constant among all ages, while that of p-tau showed a unique age-dependent increase ([169]Figure 5D). Furthermore, p-tau and BACE1 were found in juxtaposed but different dystrophic neurites ([170]Figure 5A), as previously described.[171]^22 These suggest that p-tau accumulates later than and in different dystrophic neurites from BACE1, and that Inpp5d deficiency likely affects the mechanism specific to p-tau deposition. Figure 4. [172]Figure 4 [173]Open in a new tab Inpp5d haplodeficiency attenuated p-tau accumulation in the dystrophic neurites of Tyrobp-deficient NLGF mice (A–C) Representative staining for BACE1 (A, N = 8–11), LAMP1 (B, N = 4), and ubiquitin (C, N = 4) in the cortex from 9-month-old mice of the indicated genotypes. Quantifications of the immunoreactive area (% of cortical area) are shown in right. Note that, for ubiquitin, the plaque-associated signal was specifically quantified after excluding smaller particles corresponding to nucleus-localized physiological ubiquitin. Scale bars: 100 μm. (D) Immunohistochemistry for p-tau (yellow) and Aβ (green) in a cortical region from 9-month-old mice of the indicated genotypes. The overlapping area of p-tau and Aβ signals was expressed as relative to the Aβ-positive area as 100% (N = 4). Scale bar: 50 μm. Each point represents an individual mouse. Bars represent the mean ± SEM of all analyzed animals in each genotype. One-way ANOVA followed by Tukey’s multiple comparisons. ns: not significant (p > 0.05), ∗p < 0.05, ∗∗∗∗p < 0.0001. Figure 5. [174]Figure 5 [175]Open in a new tab Differential accumulation of p-tau and BACE1 in dystrophic neurites (A–D) Triple labeling of p-tau (A, green), BACE1 (magenta), and Aβ (gray) in the brain sections from 4-, 6-, 9-, 12-, and 18-month-old (m.o.) female NLGF mice, around amygdala (A, Scale bar: 50 μm). The Aβ-positive area (% of brain section, B), and the overall fluorescent signals of BACE1 and p-tau per section (C) were elevated with aging. Of note, the mean intensity of BACE1 in plaques was relatively constant from as early as 4 months old, in contrast to p-tau being stronger with aging (D, N = 3–6). Each point represents an individual mouse. Inpp5d haplodeficiency showed a minor or partial rescue effect on DAM transition in Tyrobp-deficient NLGF mouse microglia Transcriptomic changes of microglia have been implicated in the progression of amyloid pathology. We performed RNA-seq on isolated microglia from NLGF, Inpp5d^+/−;NLGF, Tyrobp^−/−;NLGF, and Inpp5d^+/−;Tyrobp^−/−;NLGF mice at 9 months of age. Clustering analysis was performed using TCC-GUI[176]^59 ([177]Figure 6A). Whereas Trem2-dependent gene expression has been extensively analyzed using microglia isolated from β-amyloidosis mice,[178]^20^,[179]^27^,[180]^60 transcriptional changes in Tyrobp-deficient mice have only been assessed by bulk tissue RNA-seq.[181]^46^,[182]^61 Therefore, we first analyzed the expression profiles of microglia from Tyrobp^−/−;NLGF mice. Among the 1,823 differentially expressed genes (DEGs; > 2-fold change, adjusted p value: <0.05), 1,258 genes were upregulated and 565 genes were downregulated ([183]Figure 6B). The upregulated genes included S100a8 and S100a9 (associated with the inflammatory response), and Clec4e (C-type lectin). The downregulated genes included Ch25h (cholesterol metabolism), Mamdc2 (regulation of the MAPK signaling pathway), Spp1 (cytokine), Dkk2, and Fzd9 (WNT signaling), which were the same as those identified in microglia of Trem2-deficient β-amyloidosis mice.[184]^20^,[185]^60 Of note, Trem2-dependent stage 2 DAM signature genes[186]^27 were enriched in the downregulated genes (e.g., Clec7a, Cst7, Itgax, Lpl, and Spp1), supporting the essential role of TYROBP in conversion of homeostatic microglia into DAM. Furthermore, GO terms enriched in the downregulated genes included “cholesterol metabolic process” and “positive regulation of cell migration” ([187]Figure S5A), consistent with a previous study on Trem2-deficient microglia.[188]^62 Moreover, “positive regulation of phosphatidylinositol 3-kinase signaling” was also identified in our dataset ([189]Figures S3A and [190]S5A). Collectively, Tyrobp deficiency causes transcriptional changes of microglia in a similar manner to those observed in Trem2-deficient microglia. Figure 6. [191]Figure 6 [192]Open in a new tab RNA-seq analyses of Inpp5d- and/or Tyrobp-deficient NLGF mice (A) RNA-seq data of isolated microglia from 9-month-old mice were analyzed by TCC-GUI. Top 100 genes were shown in a heatmap, in which color indicates the relative expression level. Canberra distances were used to perform the clustering analysis. (B) Volcano plot of Tyrobp^−/−;NLGF versus NLGF. The negative log10 transformed p values are plotted against the log2-fold changes. The upregulated genes in Tyrobp^−/−;NLGF microglia were colored in red, while the downregulated genes were in blue. (C) Volcano plot of Inpp5d^+/−;Tyrobp^−/−;NLGF versus Tyrobp^−/−;NLGF. The negative log10 transformed p values are plotted against the log2-fold changes. The upregulated genes in Inpp5d^+/−;Tyrobp^−/−;NLGF microglia were colored in red, while the downregulated genes were in blue. (D) Validation of RNA-seq data by qRT-PCR. mRNA expressions of Tyrobp, Cst7, Lpl, Serinc3, and Tmem119 in isolated microglia were shown by normalizing to Gapdh. (E and F) Immunohistochemistry for IBA1 (E, cyan), CD11c (blue), and Tmem119 (yellow) in the cortical region. The overlapping area of IBA1/CD11c (F, left) or IBA1/Tmem119 (right) staining was shown as relative to the IBA1-positive area as 100%. Scale bar: 100 μm. (G) Top 5 genes of upregulated or downregulated in Inpp5d^+/−;Tyrobp^−/−;NLGF microglia compared with Tyrobp^−/−;NLGF microglia. Each point represents an individual mouse. Bars represent the mean ± SEM of all analyzed animals in each genotype. One-way ANOVA followed by Tukey’s multiple comparisons. ns: not significant (p > 0.05), ∗p < 0.05, ∗∗p < 0.01, ∗∗∗∗p < 0.0001. See also [193]Figure S5. We then investigated the effects of Inpp5d haplodeficiency on the transcriptional profile of microglia in Tyrobp^−/−;NLGF mice. Among the 870 DEGs (>2-fold change, adjusted p value: <0.05), 306 genes were upregulated, whereas 564 were downregulated. We observed that Tyrobp^−/−;NLGF mice have more P2RY12-positive “homeostatic” microglia compared with NLGF mice, and this defect was corrected by additional deletion of Inpp5d ([194]Figure S3E). We therefore asked whether Inpp5d reduction would restore DAM gene expression in Tyrobp-deficient mice. However, Inpp5d^+/−;Tyrobp^−/−;NLGF microglia showed almost no difference in DAM gene expression compared with Tyrobp^−/−;NLGF microglia ([195]Figure 6C). Exceptions included Lpl and Spp1, although their increases were subtle. Some of the results were validated by qRT-PCR ([196]Figure 6D). Immunohistochemical analyses of CD11c and Tmem119, a marker for DAM and homeostatic microglia, respectively, indicated that microglia from Tyrobp^−/−;NLGF and Inpp5d^+/−;Tyrobp^−/−;NLGF mice showed similar defects in the transition to the DAM phenotype ([197]Figures 6E and 6F). These data suggest that Inpp5d haplodeficiency in Tyrobp^−/−;NLGF mice has a minor or partial effect on the acquisition of the DAM phenotype by microglia. To obtain molecular insights into the beneficial role of Inpp5d haplodeficiency in Tyrobp^−/−;NLGF microglia, we compared DEGs between Tyrobp^−/−;NLGF and Inpp5d^+/−;Tyrobp^−/−;NLGF microglia. Trem2 levels were similar among all the genotypes ([198]Figure S5B), suggesting that the observed rescue effects were not explained by Trem2 upregulation. The top five upregulated genes were Wdfy1, Sox13, Fzd9, Mamdc2, and Pdcd1, whereas the top five downregulated genes were Neil3, Klrb1f, Slc25a31, Gm38248, and Xkr9 ([199]Figure 6G). Among them, Fzd9 and Mamdc2 were downregulated by Tyrobp deficiency in the microglia of NLGF mice, as well as in the microglia of Trem2-deficient PS2APP mice.[200]^20 Fzd9 encodes a receptor for WNT,[201]^63 which modulates the TREM2/TYROBP signaling pathway.[202]^44 Mamdc2, a gene upregulated in AD model mouse microglia, is involved in type-I interferon response.[203]^64 Pdcd1 (also known as PD-1), upregulated in plaque-associated microglia, regulates Aβ phagocytosis as well as inflammatory responses.[204]^65 Wdfy1 modulates Toll-like receptor signaling,[205]^66 which is also regulated by INPP5D.[206]^67 These molecules may therefore contribute to the beneficial effects of Inpp5d haplodeficiency through their signaling pathways. INPP5D negatively regulates the PI3K/AKT pathway in Tyrobp-deficient microglia To ask how INPP5D reduction exerts its effect independently of the transcriptional change induced by TREM2 and TYROBP, we investigated its role in PI(3,4,5)P[3] signaling by assessing the phosphorylation of AKT[207]^44 as indicative of PI(3,4,5)P[3] level. Phosphorylation level of AKT was reduced by Trem2 or Tyrobp knockdown (KD) ([208]Figures 7A and 7B), as well as Tyrobp deficiency in primary microglia ([209]Figures S6A and S6B), whereas it was increased by Inpp5d KD. More relevant to in vivo situation, additional KD of Inpp5d corrected p-AKT levels in Trem2 or Tyrobp KD cells ([210]Figures 7A and 7B), and the similar results were observed in ex vivo microglia isolated from Inpp5d^+/−;Tyrobp^−/−;NLGF mice, although it did not reach a statistical significance ([211]Figure S6C). These suggest that TREM2/TYROBP and INPP5D oppositely regulate PI(3,4,5)P[3] levels. Meanwhile, phosphorylation of phospholipase C-γ2 (PLCγ2), a protein activated downstream of PI(3,4,5)P[3],[212]^68^,[213]^69^,[214]^70^,[215]^71 was also reduced by Trem2 or Tyrobp KD, but was not affected by Inpp5d ([216]Figures 7A and 7B). This result can be explained by the fact that PLCγ2 activation requires not only PI(3,4,5)P[3] but also other upstream inputs (i.e., SYK activation). Phosphorylation levels of ERK1 and ERK2 were affected similarly to that of PLCγ2 ([217]Figures S6D and S6E), consistent with that MAPK is regulated independently of PI3K pathway. Considering that PI3K/AKT sits upstream of mTORC/autophagy, we also examined an autophagy marker LC3B ([218]Figures S6F and S6G). Trem2 or Tyrobp KD increased the level of LC3B in its lipidated form, suggesting that reduced PI3K activity unlocks autophagy.[219]^25 In contrast, Inpp5d KD reduced LC3B-II levels, but the effect was modest. The minor contribution of INPP5D to autophagy regulation is likely due to the existence of multiple PI(3,4,5)P[3] phosphatases in the cell. Figure 7. [220]Figure 7 [221]Open in a new tab Inpp5d reduction upregulated p-AKT levels and promoted cell adhesion in Tyrobp KD cells (A and B) Effects of Trem2, Tyrobp, or Inpp5d KD on the phosphorylation levels of AKT and PLCγ2 in primary microglia (A). The levels of p-AKT (B, left, N = 15) and p-PLCγ2 (right, N = 9) were shown as relative to the total levels of AKT and PLCγ2, respectively. (C) Schematic of PI(3,4,5)P[3] metabolisms and related enzymes. (D and E) Effects of simultaneous KD of Tyrobp and phosphoinositide phosphatases (D). The p-AKT level was shown as relative to the total AKT level (E, N = 3). (F and G) Phosphoproteomic analysis of primary microglia treated with siRNA against Trem2 and Inpp5d. Indicated are phosphoproteins whose expression levels (log2-transformed fold-change relative to the non-target siRNA control value) are oppositely regulated by Inpp5d and Trem2 (Inpp5d vs Trem2 > 1.5, Inpp5d vs non-target siRNA >1, Trem2 vs non-target siRNA <1). Their relative levels were color-coded in the heatmap (F). Asterisks indicate statistical significance (adjusted p value <0.05). Note that the list includes the proteins with sub-threshold p value (p > 0.05). The GO term enrichment analysis of the 94 proteins were shown in G. (H and I) Representative images of siRNA-transfected primary microglia on glass coverslips after brief exposure to lipopolysaccharides (H, scale bar: 50 μm). Quantification of the cell area was shown in I. (J) Transwell assay of siRNA-transfected primary microglia toward C5a. The number of migrated cells was shown by normalizing to the control siRNA-transfected, C5a-treated group as 100%. Each point represents an individual trial. Bars represent the mean ± SEM of all analyzed trials in each condition. One-way ANOVA was used except for B using repeated measures one-way ANOVA, followed by Tukey’s multiple comparisons. ns: not significant (p > 0.05), ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. See also [222]Figure S6. We further asked the role of other PI(3,4,5)P[3] phosphatases, namely, inositol polyphosphate phosphatase-like 1 (INPPL1; a closest homolog of INPP5D) and phosphatase and tensin homolog,[223]^38 as well as PI(3,4)P[2] phosphatases, inositol polyphosphate-4-phosphatase type I A and type II B, as they also contribute to AKT activation[224]^72 ([225]Figure 7C). Interestingly, the ability to revert the reduced p-AKT in Tyrobp KD cells was unique to Inpp5d and not seen by the other phosphatases KD ([226]Figures 7D and 7E). Simultaneous KD of Inppl1 with Inpp5d also did not modulate the effect of single Inpp5d KD, indicating that Inppl1 does not compensate for the lack of Inpp5d function. These data indicate a unique inhibitory role of INPP5D in the PI3K/AKT pathway. While our finding that INPP5D antagonizes PI(3,4,5)P[3] function and its inhibitory effect does not require TYROBP ([227]Figure 1), a previous study suggests that INPP5D also inhibits PI3K activation by binding to TYROBP, preventing its interaction with SYK.[228]^40 We thus examined the interaction between INPP5D and TYROBP. Their binding was unexpectedly not captured by co-immunoprecipitation of HEK293 expressing TREM2, TYROBP, and INPP5D ([229]Figure S6H). However, in situ proximity ligation assay[230]^73 of primary microglia indicated that INPP5D is in proximity (<30 nm) with TREM2 ([231]Figures S6I and S6[232]J) and TYROBP ([233]Figure S6K). A specific interaction between TREM2 and SYK was also observed, which was enhanced by Inpp5d KD ([234]Figure S6L), suggesting that INPP5D inhibits SYK activation process, probably by competitively binding to TREM2. These suggest that INPP5D regulates microglial function via multiple mechanisms that converge on PI(3,4,5)P[3] downregulation. Considering that an AD risk variant of INPP5D (rs35349669) reportedly increases INPP5D expression,[235]^39 we asked the effect of INPP5D overexpression in microglia. In murine MG6 cell line, phosphorylation of AKT was increased upon agonist antibody for TREM2 ([236]Figures S6M and S6N); however, its level was unchanged by overexpressing wild-type or phosphatase-dead mutant of INPP5D, suggesting that simple overexpression of INPP5D is not enough to affect PI(3,4,5)P[3] pathway. Importantly, a previous report suggests that INPP5D activation requires a conformational change to overcome an autoinhibitory intramolecular interaction.[237]^74 Of note, rs35349669 is located near internal transcription start site predicted to yield a truncated isoform of INPP5D[238]^75 and may affect the autoinhibition mechanism. It is therefore important to explore the consequence of this variant in future studies. INPP5D negatively regulates actin remodeling and cellular adhesion of microglia To further gain insight into the role of INPP5D in microglial regulation, we looked for proteins that are affected similarly to AKT. For this purpose, phosphoproteins of primary microglia treated with siRNA against Trem2 or Inpp5d were enriched and subjected to phosphoproteomic analysis. Considering an expected high off-target activity of siRNA, we only focused on phosphoproteins whose abundance is oppositely regulated by Trem2 and Inpp5d, rather than those that are affected by each of them (i.e., Inpp5d vs Trem2 > 1.5, Inpp5d vs non-target siRNA >1, Trem2 vs non-target siRNA <1 in log2-transformed fold-change). However, the phosphoproteins with significant change in abundance (adjusted p value <0.05) were limited, and pathway enrichment analysis was not possible. We therefore looked at proteins with sub-threshold p values ([239]Figure 7F). This criterion unfortunately did not lead to AKT identification possibly due to its low abundance but is nevertheless reliable since it found MAPK1/ERK2 and MAPK3/ERK1, for both of which we observed that the phosphorylation levels were increased by KD of Trem2 compared with Inpp5d ([240]Figures S6D and S6E). The GO term enrichment analysis of the resultant 94 proteins indicated “actin filament organization” as the most significant ([241]Figure 7G). In addition, the list included multiple AD risk factors (LILRB4, CD2AP, FRMD4A, ABI3, and BLNK), among which CD2AP and ABI3 are involved in the actin remodeling process. The finding of ABI3 is particularly interesting, since its deficiency in β-amyloidosis mice causes microglial dissociation from plaques.[242]^76 Given that TREM2 and INPP5D oppositely regulate protein phosphorylation related to actin remodeling process, we asked whether they are involved in actin-related functions of primary microglia. Because Inpp5d reduction affects microglial association with plaques in vivo, we first examined its role in cell adhesion. We noticed that Trem2 or Tyrobp KD caused a significant reduction of cell area on glass coverslips, particularly when briefly primed with lipopolysaccharide to mimic an in vivo inflammatory state ([243]Figure 7H), indicating impaired cell adhesion. Interestingly, simultaneous KD of Inpp5d restored the defects ([244]Figure 7I), suggesting that TREM2/TYROBP and INPP5D oppositely regulate cell adhesion. Next, we examined the effects on chemotaxis and cell proliferation/survival, since their deficits are implicated in the defective plaque association of Trem2-deficient microglia.[245]^11^,[246]^77^,[247]^78 Chemotaxis to C5a, a potent chemoattractant of microglia, was indeed affected by Tyrobp KD, but was not reverted by Inpp5d KD ([248]Figure 7J). Similarly, Inpp5d KD did not rescue the reduced viability of Trem2 or Tyrobp KD cells ([249]Figure S6O), consistent with in vivo data ([250]Figure 1H). Based on these observations, Inpp5d reduction specifically increases cell adhesion, which likely underlies the enhanced plaque association of microglia observed in vivo. Discussion In the present study, we investigated the pathological role of INPP5D, an AD risk gene expressed in microglia. Inpp5d haplodeficiency in NLGF mice did not affect Aβ deposition, gliosis, or neurite dystrophy. However, in Tyrobp-deficient NLGF mice showing TREM2 loss-of-function phenotypes, Inpp5d haplodeficiency restored microglial envelopment of plaques, altered plaque morphology, increased astrocyte activation, and reduced p-tau levels in dystrophic neurites. RNA-seq of isolated microglia and immunohistochemical analyses suggest that Inpp5d reduction has a minor or partial effect on the DAM transition of Tyrobp-deficient NLGF mouse microglia. Instead, KD experiments revealed that INPP5D has unique inhibitory roles in cell adhesion as well as in PI(3,4,5)P[3] signaling. Collectively, Inpp5d haplodeficiency alleviates the detrimental effects of Tyrobp deficiency in microglia, thereby affecting p-tau deposition around Aβ plaques. TREM2 is crucial for microglia to cluster around plaques and thereby protect neurons from amyloid toxicity. TREM2 regulates multiple aspects of microglial function, including cell proliferation/survival, energy metabolism, and transformation into DAM[251]^11^,[252]^25^,[253]^27; however, which factor contributes to the protective effect remains unclear. Ulland et al. reported that metabolic correction of Trem2-deficient mice by cyclocreatine restores the microglial envelopment of plaques and ameliorates neurite dystrophy.[254]^25 Similarly, we showed that INPP5D reduction in Tyrobp-deficient NLGF mice restores the microglial clustering and attenuates p-tau deposition, suggesting that physical interaction between microglia and plaques limits the p-tau deposition around plaques. On the other hand, the neuroprotective effects other than p-tau were not evident in our mice, suggesting that signaling pathways other than the PI3K pathway are required for this process. Inpp5d haplodeficiency in Tyrobp^−/−;NLGF mice did not significantly alter brain Aβ deposition, although it increased microglial association with Aβ plaques. While this may be due to the lack of a functional TREM2 receptor, two recent studies (published after the initial submission of our manuscript) have questioned the role of INPP5D in Aβ clearance using similar Inpp5d-deficient β-amyloidosis mice.[255]^79^,[256]^80 In fact, these studies report conflicting results regarding Aβ pathology, which may be due to differences in the mouse models (NLGF, APP/PS1, and 5xFAD) and detection methods used. However, it is interesting to note that homozygous deletion of Inpp5d in APP/PS1 mice does not reduce Aβ burden, and the increased microglial association with plaques in these mice does not appear to be sufficient to enhance Aβ clearance by microglia.[257]^79 While studies in Trem2 knockout mice suggest the importance of plaque-associated microglia in Aβ clearance,[258]^11^,[259]^51 pharmacological depletion of microglia in various β-amyloidosis mice did not increase Aβ deposition,[260]^52^,[261]^81 and thus the precise role of microglia in Aβ metabolism remains unclear. On the other hand, the loss of plaque-associated microglia in these studies was consistently associated with impaired plaque compaction, and plaques with different morphologies have different effects on surrounding neurons.[262]^17^,[263]^18^,[264]^20^,[265]^52 Since Inpp5d reduction alters plaque morphology to a compact shape in Tyrobp-deficient NLGF mice, the beneficial effect of Inpp5d deficiency may be due to the improvement of plaque compaction as well as the physical barrier around the plaques. Microglial envelopment of plaques suppresses periplaque neuronal dystrophy, which likely underlies the mechanism that TREM2 restrains the Aβ-mediated enhancement of p-tau accumulation.[266]^22^,[267]^23 Our findings that Inpp5d reduction in Tyrobp-deficient NLGF mice specifically mitigates p-tau deposition, and that p-tau accumulates later than and in different neurites from BACE1 suggest that INPP5D targets a putative-specific mechanism of p-tau deposition, which may be driven by cumulative damage in neurites, or by some neurotoxic factors that increase late in plaque accumulation. Among such candidates, Aβ oligomers disrupt microtubules in nearby axons,[268]^82 thereby contributing to cytosolic release of tau, and its phosphorylation and aggregation.[269]^21 Considering that microglia restrain the diffusion of Aβ oligomers around plaques,[270]^55 INPP5D may affect their diffusion by strengthening the contact with plaques. On the other hand, Inpp5d haplodeficiency in Tyrobp-deficient NLGF mice did not increase plaque-associated microglia or fully restore the plaque morphology, which may be further required to ameliorate the deposition of other markers. GFAP-positive reactive astrocytes were dramatically reduced in Tyrobp-deficient NLGF mice, reminiscent of the phenotype of Trem2 deficiency or CSF1R inhibition,[271]^20^,[272]^51^,[273]^52^,[274]^83 indicating an essential role of microglia in the induction of reactive astrocytes. Microglia contribute to neurotoxic astrocyte formation upon inflammatory stimuli.[275]^48 In addition, Aβ abundance in the brain is related to the GFAP elevation in preclinical AD.[276]^84^,[277]^85 These suggest that microglia and Aβ coordinately lead astrocyte activation. Here, we found that Inpp5d reduction restored astrocyte activation in Tyrobp-deficient NLGF mice, suggesting the importance of the physical interaction of microglia with plaques or PI(3,4,5)P[3] signaling in this process. Our RNA-seq results provided molecular insights into the phenotypes observed in this study. For the first time to our knowledge, we profiled the transcriptional changes of microglia in Tyrobp-deficient β-amyloidosis mice. The DEGs in Tyrobp-deficient microglia quite resemble those described in Trem2-deficient microglia, in agreement with assumption that TYROBP is essential for TREM2 function regulating the transition of microglia into the DAM state. Interestingly, a recent study on Inpp5d^flox/flox;Cx3cr1-CreER;APP/PS1 mice suggests that Inpp5d deficiency facilitates the transition to a DAM-like state in microglia,[278]^79 but Inpp5d reduction in Tyrobp-deficient mice shows little or partial effect on DAM gene expression, suggesting that multiple TYROBP-mediated signaling pathways, not just the PI3K pathway, are required for this process. Furthermore, Inpp5d reduction caused the transcriptional upregulation of Wdfy1, Sox13, Fzd9, Mamdc2, and Pdcd1, and the downregulation of Neil3, Klrb1f, Slc25a31, Gm38248, and Xkr9. Future studies should address their roles in the amyloid recognition/association process as well as the p-tau deposition. One primary phenotype oppositely regulated by TREM2/TYROBP and INPP5D was the cell adhesion activity, which correlated with the altered phosphorylation of proteins that are involved in actin assembly. The reduced cell adhesion of Trem2 KD microglia is likely linked to the reported defects in cytoskeletal organization,[279]^86 as well as the surface expression of β-integrins.[280]^87 Intriguingly, PI(3,4,5)P[3] regulates actin polymerization and integrin activation[281]^38; and β1-integrin is implicated in Aβ recognition by microglia.[282]^88 Furthermore, previous studies showed that microglia have a physical contact with plaques in a TREM2/TYROBP-dependent manner[283]^18^,[284]^55 and that microglia lacking the contact were not able to make plaques into compact shapes nor protect neurons,[285]^89 highlighting the importance of the physical contact. We therefore speculate that the enhanced adhesion by Inpp5d haplodeficiency would increase the physical contact with plaques and strengthen the neuroprotective function by microglia. TREM2/TYROBP and INPP5D exert opposing effects on the PI(3,4,5)P[3] downstream signaling. While INPP5D may exert this effect via the PI(3,4,5)P[3] dephosphorylating activity, our PLA results suggest a different inhibitory role in the TYROBP-SYK interaction, which is quite reminiscent of the previous finding that INPP5D inhibits SYK/PI3K activation by competitively binding to TYROBP.[286]^40 Given the central role of SYK in the TREM2-mediated microglial signaling pathway,[287]^90^,[288]^91 it is interesting to investigate whether INPP5D inhibition enhances SYK activity. On the other hand, we also found that INPP5D reduction upregulates PI(3,4,5)P[3] signaling in the absence of TYROBP, suggesting that INPP5D plays a role in the metabolism of a PI(3,4,5)P[3] pool that is presumably generated by other upstream receptors that activates PI3K. Interestingly, PI(3,4,5)P[3] phosphatases other than INPP5D (Pten and Inppl1) were not involved in p-AKT regulation in Tyrobp KD cells. This raises the interesting possibility that the PI(3,4,5)P[3] pool targeted by INPP5D has a specific role in the AKT activation. Further studies are required to elucidate the underlying mechanism and the different roles of the phosphatases in this pathway. A common variant of the INPP5D gene (rs35349669) is associated with an increased risk of late-onset AD.[289]^2 While its effect on INPP5D function remains unclear, rs35349669 was associated with higher p-tau levels in the cerebrospinal fluid (CSF) of patients with AD.[290]^92 Altered CSF tau was also reported for other INPP5D variants,[291]^93^,[292]^94 suggesting a link between INPP5D and p-tau deposition. Moreover, the potential role of INPP5D in the Aβ and tau crosstalk may provide a unique opportunity for AD therapy. INPP5D inhibitors may be beneficial to prevent tau deposition. Nonetheless, complete inhibition of INPP5D can cause an adverse effect particularly in peripheral tissues,[293]^43 and therefore it is necessary to develop a specific way to target it in microglia. For this purpose, it would be important to identify a microglia-specific regulator of INPP5D. Of note, INPP5D is upregulated near Aβ plaques[294]^42 in an unknown mechanism, which may involve druggable targets to regulate microglial expression of INPP5D. Such drugs would provide a novel microglia-targeting therapeutic/preventive approach for AD, specifically against Aβ-driven worsening of tau deposition. Limitations of the study Our findings on the beneficial effects of Inpp5d reduction were mainly obtained from a Tyrobp-deficient, TREM2 loss-of-function mouse model. To know the potential beneficial effect of INPP5D inhibition as an AD therapeutic, it would be helpful to analyze the single knockout effect of Inpp5d in AD model animals, preferably using multiple biological assays, including biomarker tests, electrophysiological analyses, and behavioral tests. STAR★Methods Key resources table REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies __________________________________________________________________ Rabbit polyclonal anti-Iba1 FUJIFILM Wako Cat# 019-19741, RRID: [295]AB_839504 Goat polyclonal anti-Iba1 FUJIFILM Wako Cat# 011-27991 Mouse monoclonal anti-Human Amyloid-beta (N) (82E1) Immuno-Biological Laboratories Cat# JP10323, RRID:[296]AB_163080 Rabbit monoclonalanti-β-Amyloid (D54D2) Cell Signaling Technology Cat# 8243, RRID: [297]AB_279764 Mouse monoclonal anti-β-Amyloid (6E10) Covance Cat# SIG-39320, RRID: [298]AB_662798 Rabbit monoclonalanti-PU.1 (9G7) Cell Signaling Technology Cat# 2258, RRID: [299]AB_218690 Rabbit monoclonal anti-BACE (D10E5) Cell Signaling Technology Cat# 5606, RRID: [300]AB_190390 Chicken polyclonal anti-GFAP BioLegend Cat# 829401, RRID: [301]AB_256492 Goat polyclonal anti-Apolipoprotein E Merck/Millipore Cat# AB947, RRID: [302]AB_2258475 Rabbit monoclonal anti-Phospho-Akt (Ser473) (D9E) Cell Signaling Technology Cat# 4060, RRID: [303]AB_231504 Rabbit monoclonal anti-Akt (pan) (C67E7) Cell Signaling Technology Cat# 4691, RRID: [304]AB_915783 Rabbit polyclonal anti-PLC-gamma-2 Cell Signaling Technology Cat# 3872, RRID: [305]AB_229958 Rabbit polyclonal anti-Phospho-PLC-gamma-2 (Tyr1217) Cell Signaling Technology Cat# 3871, RRID: [306]AB_229954 Mouse monoclonal anti-Phospho-Erk (E-4) Santa Cruz Biotechnology Cat# sc-7383, RRID: [307]AB_627545 Rabbit monoclonal anti-p44/42 MAPK (Erk1/2) (137F5) Cell Signaling Technology Cat# 4695, RRID: [308]AB_390779 Mouse monoclonal anti-LC3B (E5Q2K) Cell Signaling Technology Cat# 83506, RRID: [309]AB_2800018 Rabbit monoclonalanti-TREM2 (E7P8J) Cell Signaling Technology Cat# 76765, RRID: [310]AB_279988 Rabbit monoclonalanti-DAP12 (D7G1X) Cell Signaling Technology Cat# 12492, RRID: [311]AB_272112 Mouse monoclonal anti-SYK (SYK-01) BioLegend Cat# 626202RRID: [312]AB_2200268 Rabbit monoclonalanti-PTEN (138G6) Cell Signaling Technology Cat# 9559, RRID: [313]AB_390810 Mouse monoclonal anti-SHIP1 (P1C1) Santa Cruz Biotechnology Cat# sc-8425, RRID: [314]AB_628250 Rabbit monoclonal anti-Inpp4a (EP3425(2)) abcam Cat# ab109622, RRID: [315]AB_10887220 Rabbit monoclonal anti-Inppl1 ([316]EPR10955) abcam Cat# ab166916, RRID: [317]AB_268689 Mouse monoclonalanti-α-Tubulin (clone DM1A) Sigma-Aldrich Cat# T9026, RRID: [318]AB_477593 Rabbit polyclonal anti-Ubiquitin Dako Cat# Z0458, RRID: [319]AB_231552 Rabbit monoclonal anti-Phospho-tau ([320]EPR20390) abcam Cat# ab210703, RRID: [321]AB_2922760 Mouse monoclonal anti-Phospho-tau (Ser202, Thr205) (Clone AT8) Thermo Fisher Cat# MN1020, RRID: [322]AB_223647 Rabbit monoclonal anti-Tmem119 (28-3) abcam Cat# 209064, RRID: [323]AB_272808 Armenian Hamster monoclonal anti-mouse CD11c (Clone ID N418) BioLegend Cat# 117301, RRID: [324]AB_313770 Rabbit polyclonal anti-P2RY12 antibody Sigma-Aldrich Cat# HPA013796, RRID: [325]AB_1854884 Mouse monoclonal anti-DYKDDDDK tag (Clone 1E6) FUJIFILM Wako Cat# 012-22384, RRID: [326]AB_10659717 Rabbit polyclonal anti-DYKDDDDK tag Cell Signaling Technology Cat# 2368, RRID: [327]AB_2217020 Rabbit polyclonal anti-V5 tag MBL Cat# PM003, RRID: [328]AB_592941 Rat monoclonal anti-mouse/human CD11b FITC BioLegend Cat# 101205, RRID: [329]AB_312788 Mouse monoclonal anti-SHIP-1 (Clone P1C1-A5) PE BioLegend Cat# 656604, RRID: [330]AB_256286 Rat monoclonal anti-mouse CD45 Alexa Fluor® 647 BioLegend Cat# 103124, RRID: [331]AB_493533 Rat monoclonal anti-mouse CD16/32 BioLegend Cat# 101302, RRID: [332]AB_312801 Goat Anti-Rabbit Peroxidase Jackson ImmunoResearch Cat# 111-035-144, RRID: [333]AB_2307391 Goat Anti-Mouse Peroxidase Jackson ImmunoResearch Cat# 115-035-003, RRID: [334]AB_10015289 Horse Anti-Mouse IgG Antibody (H+L), Biotinylated Vector Laboratories Cat# BA-2000, RRID: [335]AB_2313581 Rabbit Anti-Goat IgG Antibody (H+L), Biotinylated Vector Laboratories Cat# BA-5000, RRID: [336]AB_2336126 Normal mouse IgG, whole molecule, purified FUJIFILM Wako Cat# 140-09511 Mouse TREM2 Antibody R&D Systems Cat# AF1729, RRID: [337]AB_354956 Normal Sheep IgG Control R&D Systems Cat# 5-001-A, RRID: [338]AB_10141430 __________________________________________________________________ Chemicals, peptides, and recombinant proteins __________________________________________________________________ 10×HBSS(−) without Phenol Red FUJIFILM Wako Cat# 082-09865 Percoll Plus GE Healthcare Cat# 17544502 HBSS(−) without Phenol Red FUJIFILM Wako Cat# 085-09355 Lipofectamine™ RNAiMAX Transfection Reagent Thermo Fisher Cat# 13778500 Thioflavin S, practical grade SIGMA Cat# T1892-25G Sudan Black B Merck Cat# 1.15928.0025 Propidium iodide Sigma-Aldrich Cat# P4170 Paraformaldehyde FUJIFILM Wako Cat# 162-16065 Albumin, from Bovine Serum, Fraction V pH7.0 FUJIFILM Wako Cat# 013-27054 Lipopolysaccharides from Escherichia coli O111:B4 Sigma-Aldrich Cat# L4391 Cholera Toxin Subunit B (Recombinant), Alexa Fluor 647 Conjugate Thermo Fisher Cat# [339]C34778 Recombinant Mouse Complement Component C5a Protein R&D Systems Cat# 2150-C5-025 ProLong Diamond Antifade Mountant with DAPI Thermo Fisher Cat# [340]P36962 Protein G Sepharose 4 Fast Flow Cytiva Cat# 17061801 __________________________________________________________________ Critical commercial assays __________________________________________________________________ Human β Amyloid(1-40) ELISA Kit Wako II FUJIFILM Wako Cat# 292-64601 Human/Rat β Amyloid(42) ELISA Kit Wako, High Sensitive FUJIFILM Wako Cat# 292-64501 Simoa NF-light Advantage Kit Quanterix Cat# 103186 Simoa NF-light Sample Diluent Reagent Quanterix Cat# 102252 ReverTra Ace® qPCR RT Master Mix with gDNA Remover TOYOBO CO., LTD Cat# FSQ-301 THUNDERBIRD® SYBR qPCR Mix TOYOBO CO., LTD Cat# QPS-201 RNeasy Plus Micro Kit QIAGEN Cat# 74034 TSA Plus Fluorescein 50-150 slides AKOYA Biosciences Cat# NEL741001KT TSA Plus TMR 50-150 slides AKOYA Biosciences Cat# NEL742001KT TSA Plus Cyanine 5 50-150 slides AKOYA Biosciences Cat# NEL745001KT TSA Blocking Reagent AKOYA Biosciences Cat# FP1020 Proteinase K TaKaRa Cat# 9034 BCA Protein Assay Kit TaKaRa Cat# T9300A Amicon Ultra 0.5 mL Merck/Millipore Cat# UFC505024 Cell culture insert, transparent PET membrane, 24-well, 8.0 μm pore size Corning Cat# 353097 alamarBlue® Bio-Rad Cat# BUF012A Duolink® In Situ Detection Reagents Green Sigma-Aldrich Cat# DUO92014-30RXN Duolink® In Situ PLA® Probe Anti-Mouse PLUS Sigma-Aldrich Cat# DUO92001, RRID: [341]AB_2810939 Duolink® In Situ PLA® Probe Anti-Rabbit MINUS Sigma-Aldrich Cat# DUO92005, RRID: [342]AB_2810942 Peroxidase Labeling Kit - NH2 Dojindo Cat# LK11 Titansphere Phos-TiO2 Kit GL Sciences Cat# 5010-21305 GL-Tip SDB GL Sciences Cat# 7820-11200 __________________________________________________________________ Deposited data __________________________________________________________________ Raw files of RNA-seq analysis This study NCBI-GEO:[343]GSE172499 LC-MS/MS analysis of phosphopeptides isolated from primary microglia This study PXID: PXD040217 (jPOST: JPST002047) __________________________________________________________________ Experimental models: Cell lines __________________________________________________________________ MG6 RIKEN BioResource Research Center RRID:CVCL_8732 1D4B (Frozen Cells) Developmental Studies Hybridoma Bank Cat# 1D4B-f, RRID: [344]AB_528127 __________________________________________________________________ Experimental models: Organisms/strains __________________________________________________________________ App(NL-G-F/NL-G-F) Saito et al.[345]^41 IMSR Cat# RBRC06344, RRID:IMSR_RBRC06344 Inpp5d(+/−) Nishio et al.[346]^95 N/A Tyrobp(−/−) Kaifu et al.[347]^96 N/A __________________________________________________________________ Oligonucleotides __________________________________________________________________ qRT-PCR primers, see [348]Table S1 N/A siRNAs, see [349]Table S1 N/A __________________________________________________________________ Software and algorithms __________________________________________________________________ GraphPad Prism 8 GraphPad Software RRID: SCR_002798 Imaris Oxford Instruments RRID: SCR_007370 Fiji (1.53c) ImageJ RRID: SCR_003070 RStudio (Version 1.3.1056) RStudio, PBC RRID: SCR_000432 R (Version 4.0.2) R Foundation for Statistical Computing RRID: SCR_001905 Proteome Discoverer (version 2.2.0.388) Thermo Fisher RRID: SCR_014477 [350]Open in a new tab Resource availability Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Taisuke Tomita ([351]taisuke@mol.f.u-tokyo.ac.jp). Materials availability Plasmids generated in this study will be available upon request. Experimental model and subject details Mice NLGF (IMSR Cat# RBRC06344, RRID:IMSR_RBRC06344),[352]^41 Inpp5d^+/−[353]^95 and Tyrobp^−/−[354]^96 mice were described previously. All animal experiments were performed according to the protocols approved by the Institutional Animal Care Committee of the Graduate School of Pharmaceutical Sciences at the University of Tokyo (protocol no. P27-6, P2-2). All mice were housed under standard conditions (12-hour light and 12-hour dark cycle, 22°C, 40 to 60% of humidity) with free access to food and water. We used male mice except for the experiments in [355]Figures 5 and [356]S6C, in which female mice were used. We did not mix male and female mice in individual analyses (except for primary cell preparation) to avoid possible sex-dependent variations. Primary microglia culture Primary microglia were prepared from 3-day-old C57BL/6J male and female mice. After removing meninges, brains were incubated in Hank’s balanced salt solution (HBSS) containing trypsin, deoxyribonuclease I, CaCl[2], MgSO[4] at 37°C for 20 min, gently triturating with a fine pipet tip every 5 min. Cells were added with DMEM supplemented with 10% FBS and penicillin/streptomycin (DMEM(+/+)) to quench the reaction and passed through a 100-μm strainer. Cells were spun at 300 g and room temperature for 10 min, resuspended in DMEM(+/+) and plated in T-75 flasks at the density of one brain per flask. At 4, 7 and 10 days in vitro (DIV), the medium was replaced with a fresh one. At 15 DIV, microglia were isolated by shaking flasks by hand for 1 min and centrifuged at 2,000 rpm and room temperature for 5 min. Cells were resuspended in medium supplemented with 10% L929-cell-conditioned medium (10% LCM) and cultured until use. The purity of isolated microglia was higher than 90% based on the flow cytometric analysis of the CD11b^+ cell population. Method details Brain sample collection and preparation Mice were deeply anesthetized and perfused transcardially with 15 mL ice-cold phosphate-buffered saline (PBS). The brains were quickly removed and dissected into two hemispheres. The left hemispheres (for biochemical analyses) were flash-frozen in liquid nitrogen and stored at −80°C until use. The right hemispheres (for immunohistochemistry) were fixed in 4% paraformaldehyde (PFA) for 24 h at 4°C, followed by three washes and a cryoprotection with 30% sucrose, 0.02% NaN[3] in PBS. The brains were quickly frozen on dry ice and cut into 30-μm-thick coronal sections by using a sliding microtome. Histochemistry on free-floating brain sections Free-floating brain sections were blocked and permeabilized with 10% donkey serum, 0.1% Triton X-100 and 0.02% NaN[3] in PBS for 1 h at room temperature. If needed, antigen retrieval at 95°C for 10 min was performed before blocking. After 3 washes with PBS, brain sections were incubated with primary antibodies diluted in PBS overnight at 4°C. After 3 washes with PBS, sections were treated with secondary antibodies diluted in PBS for 2 h at room temperature. After 3 washes with PBS, the sections were mounted with ProLong Diamond Antifade Mountant with DAPI. For the staining of PU.1, CD11c, and p-tau, TSA plus amplification kit (Perkin Elmer) was used in combination with subsequent conventional immunofluorescence staining for other antigens. To block endogenous peroxidase activity, brain sections were treated with 30% H[2]O[2] in PBS for 15 min at room temperature. After washing, sections were treated with TNB buffer (100 mM Tris-HCl (pH 7.5), 150 mM NaCl containing 0.5% blocking reagent from Perkin Elmer) for 1 h at room temperature, and then incubated with primary antibody in TNB buffer overnight at 4°C. The sections were incubated with appropriate peroxidase-conjugated secondary antibodies diluted in TNB buffer at 4°C overnight. For CD11c staining, goat anti-Armenian hamster antibody (Jackson Immunoresearch) was used after conjugation with peroxidase using a commercial kit (LK-11, Dojindo). After 3 washes with TNT buffer (0.1% Triton X-100, 100 mM Tris-HCl (pH 7.5), 150 mM NaCl), the sections were reacted with TMR-amplification reagent for 10 min at room temperature, washed with TNT. Thioflavin S staining was performed as below. To reduce autofluorescence, brain sections were treated with 0.1% Triton X-100 in PBS for 30 min and then incubated with 0.1% Sudan Black in 70% EtOH for 15 min. After 2 washes with 50% EtOH, sections were stained with 0.1% Thioflavin S in 20% EtOH for 15 min at room temperature, followed by 2 washes with 50% EtOH, 3 washes with 0.1% Triton X-100 in PBS, and final mounting with ProLong Diamond Antifade Mountant. If needed, this method was performed posterior to the immunofluorescence method as described above. Immunohistochemistry on paraffin-embedded sections For the longitudinal analysis of p-tau and BACE1 accumulation in [357]Figure 5, paraffin-embedded brain sections from female NLGF mice were used, and the signals were amplified using TSA plus amplification kits. Deparaffinized sections were autoclaved at 121°C for 5 min in 10 mM Citrate-HCl (pH 6.0). After inactivation of endogenous peroxidases using 0.3% H[2]O[2] in methanol for 30 min, the sections were blocked for 30 min in TNB buffer, and incubated with anti-phosphorylated tau (AT8) at 4°C overnight. On the second day, the sections were sequentially incubated with biotinylated anti-mouse and peroxidase-conjugated streptavidin, and then reacted with fluorescein-tyramide reagent for 10 min. After quenching the peroxidase activity using 0.3% H[2]O[2], 0.1% NaN[3] in PBS, the sections were blocked and incubated with anti-BACE1 (D10E5) at 4°C overnight. On the third day, the sections were incubated with peroxidase-conjugated anti-rabbit for 1 h, and then reacted with tetramethylrhodamine-tyramide reagent for 10 min. The sections were subjected to microwave processing to remove antibodies, followed by peroxidase inactivation using 0.3% H[2]O[2] in methanol. The sections were blocked and incubated with anti-Aβ (D54D2) at 4°C overnight. On the fourth day, the signal was amplified in the same way as BACE1, except for using Cy5-tyramide. Finally, the sections were mounted with ProLong Diamond Antifade Mountant with DAPI. For ApoE staining, paraffin-embedded brain sections from 11 to 12 m.o. male mice were used. Deparaffinized sections were subjected to antigen retrieval by microwave heating (500 W for 10 min and then 100–200 W for 10 min in 10 mM Citrate-HCl, pH 6.0) followed by proteinase K treatment (TaKaRa; Dilution: 1/200 (v/v)) at 37°C for 7 min. After inactivation of endogenous peroxidase and blocking, sections were incubated with Goat anti-ApoE antibody (AB947) at 4°C overnight. The sections were treated with HRP-conjugated anti-Goat antibody (BA5000) for 1 h and then visualized using fluorescein-tyramide reagent as above. Brain Aβ measurement by ELISA Frozen brain hemispheres were mechanically homogenized with 30 strokes of Potter-Elvehjem tissue grinder for cortex or with 20 strokes of BioMasher II for hippocampus in 10 volumes of ice-cold Tris-buffered saline (pH 7.6) with protease/phosphatase inhibitors (TBS). After ultracentrifugation at 312,000 g and 4°C for 20 min, the supernatant was collected as the TBS-soluble fraction. The pellet was similarly extracted with 10 volumes of ice-cold 2% (v/v) Triton X-100 in TBS and then with room temperature 2% (w/v) SDS in TBS, each followed by ultracentrifugation yielding the supernatants referred to as the TX-soluble and the SDS-soluble fraction, respectively. The final pellet was re-solubilized in 70% formic acid (FA), clarified by ultracentrifugation at 312,000 g and 4°C for 20 min. The supernatant was neutralized by adding with neutralization buffer (1 M Tris, 0.5 M Na[2]HPO[4], 0.05% NaN[3]) and referred to as the FA-soluble fraction. The abundance of Aβ40 or Aβ42 in each fraction was measured by ELISA according to the manufacturer’s instruction. Isolation of microglia from adult mouse brain Microglia were isolated from 9-month-old mouse brain by using fluorescence activated cell sorting (FACS). After perfusion with 15-mL HBSS, mice were quickly decapitated, and the brains were cut into two hemispheres. Only the left hemispheres were used for isolating cells, while the remaining hemispheres were kept for histochemical analyses. The left hemispheres were gently homogenized in HBSS using the Dounce homogenizer. Homogenates were centrifuged at 300 g and 4°C for 5 min, resuspended in 6 mL of 37% isotonic Percoll plus (#17544502, GE Healthcare), and underlaid with 6 mL of 70% isotonic Percoll solution. After centrifugation at 800g and room temperature for 30 min with no acceleration and no brake, the top layer (myelin debris) was discarded and 2 mL of the 37–70% interphase was collected. Cells were diluted with ice-cold HBSS and centrifuged at 300 g and 4°C for 5 min. Cells resuspended in FACS buffer (0.5% BSA, 2 mM EDTA, and 0.1% NaN[3] in Dulbecco’s PBS(−)) were reacted with Fc blocker (BioLegend, #101302, 1:100) for 15 min on ice and then with anti-CD11b-FITC (BioLegend, #101206, 1:300), anti-CD45-Alexa 647 (BioLegend, #103124, 1:300) and propidium iodide (PI for dead cells, Sigma-Aldrich, #P4170, final concentration 100 ng/mL) for 20 min on ice in the dark. Cells were added with 5 mL of FACS buffer and spun at 300 g and 4°C for 5 min. Cells were resuspended in 500 μL of FACS buffer and passed through a filter-top tube (BD Falcon). CD11b^+ CD45^+ PI^− cells were sorted using BD FACSAria III Cell Sorter (BD Biosciences). Sorted cells were spun at 5,000 g and 4°C for 10 min, flash-frozen in liquid nitrogen and stored at −80°C until RNA extraction. For quantitation of INPP5D by flow cytometry, isolated microglia were fixed with 4% PFA in PBS at room temperature for 15 min, washed with ice-cold PBS and permeabilized with ice-cold 90% methanol on ice for at least 20 min. After 2 washes with PBS, cells were treated with Fc blocker in FACS buffer on ice, and then added with anti-CD11b-FITC, anti-CD45-Alexa 647 and anti-SHIP-1-PE (INPP5D, BioLegend, #656604, 1:100) and incubated for 20 min on ice in the dark. INPP5D expression was measured in CD11b^+ CD45^+ cells. Sample preparation for immunoblotting Frozen brain hemispheres were mechanically homogenized with 30 strokes of Potter-Elvehjem tissue grinder in 10 volumes of ice-cold RIPA buffer with protease/phosphatase inhibitors. The homogenates were ultracentrifuged at 312,000 g and 4°C for 20 min. The supernatant was collected as a RIPA-soluble fraction and stored at −80°C until use. The pellet was re-solubilized in 70% formic acid, clarified by ultracentrifugation at 312,000 g and 4°C for 20 min, dried, solubilized in DMSO and used as a RIPA-insoluble, formic-acid-soluble fraction. For MG6 and primary microglial cells, cells rinsed with ice-cold PBS were lysed/sonicated in a RIPA buffer containing protease/phosphatase inhibitors. The protein concentration of the samples was measured by Bicinchoninic Acid Protein Assay (TaKaRa). 5× SDS-PAGE sample buffer and 2-mercaptoethanol to a final concentration of 1% (v/v) were added, and the samples were heated for 10 min at 100°C. Immunoblotting Proteins separated by Tris-Glycine or Tris-Tricine SDS polyacrylamide gel electrophoresis were transferred to PVDF membranes. The membranes were blocked for 1 h at room temperature with 5% skim milk or 3% BSA (for phospho-specific antibodies) in 0.1% (v/v) Tween 20 in TBS (TBST), followed by 3 washes with TBST. Membranes were incubated with appropriate dilutions of primary antibody in TBST with NaN[3] overnight at 4°C. After three washes with TBST, membranes were incubated with the appropriate peroxidase-conjugated secondary antibody (1:4,000) in TBST for 1 h at room temperature. After three washes with TBST, membranes were incubated with immunostar luminescence working solution. Signals were detected by ImageQuant LAS 4000 and processed using Image J/Fiji (1.53c). Transfection of primary microglia For transfection of small interfering RNA (siRNA; listed in [358]Table S1), primary microglia were plated in a low-cell-binding 12-well plate at the density of 1.5 × 10^5 cells/well with 10% LCM. After two days, cells were transfected with siRNA using Lipofectamine RNAiMAX according to the manufacturer’s instruction. After 24 h, the medium was replaced with a fresh one. Cells cultured for further 48 h were used for subsequent analyses. Phosphoproteomic analysis of primary microglia Primary microglia transfected with indicated siRNAs were lysed in SDS lysis buffer (4% SDS, 10 mM Tris pH 8.0), and proteins were precipitated using ice-cold acetone and resuspended in a buffer (10% 2-2-2-trifluorethanol, 100 mM ammonium bicarbonate). Proteins were reduced by tris(2-carboxyethyl)phosphine, S-alkylated with methyl methanethiosulfonate, and digested by LysC and trypsin overnight. Phosphopeptides were enriched using Titansphere Phos-TiO2 Kit and the eluates were desalted using GL-Tip SDB (GL Sciences) and dried under reduced pressure. The dried peptides were dissolved in a solvent (water: formic acid = 99.9: 0.1 by volume). Six microliter samples were subjected to EASY-nLC 1200 (Thermo Fisher Scientific) equipped with a 3-μm C18 nano HPLC capillary column (Nikkyo Technos, Cat#75-3-12), and separation was performed using mobile phase composed of solvent A (water/formic acid 100:0.1 (v:v)) and solvent B (water/acetonitrile/formic acid 5:95:0.1 (v:v:v)) at a constant flow rate of 300 nL/min, with the following conditions: an isocratic elution with 0% B for 5 min; a linear gradient of 10–40% B over the next 85 min; 40–95% B for 2 min; 95% B for 18 min. Mass spectrometry (MS) was performed on Q-Exactive Orbitrap mass spectrometer (Thermo Fisher Scientific) with the top 10 acquisition method: MS resolution 70,000, between 350 and 1,500 m/z, followed by MS/MS analysis (resolution 17,500) on the most intense 10 peaks. Raw MS data were processed using Proteome Discoverer (Thermo Fisher Scientific, version 2.2.0.388) with false-discovery rates < 0.01 at the levels of proteins and peptides. Enzyme specificity was set to trypsin, and the search included cysteine methylthiolation as a fixed modification, and N-acetylation of protein, oxidation of methionine, and phosphorylation of serine/threonine/tyrosine as variable modifications. Differential abundance significance was estimated from two biologically independent experiments (two technical replicates for each experiment) using background-based ANOVA with Benjamini-Hochberg correction to determine adjusted p-values. For the pathway analysis, we selected proteins whose levels (log2-fold change) were oppositely affected by siRNA against Inpp5d and Trem2 (Inpp5d vs Trem2 > 1.5, Inpp5d vs non-target siRNA > 1, Trem2 vs non-target siRNA < 1). Note that we included the proteins with sub-threshold p-values (p > 0.05) in the analysis. The gene-ontology term enrichment analysis was performed using DAVID ([359]https://david.ncifcrf.gov/home.jsp) and the terms significant in the biological process (p-value < 0.05) were shown. Adhesion assay of primary microglia Forty-eight hours after siRNA transfection, primary microglia were trypsinized, allowed to attach to glass coverslips overnight, and primed with 1 μg/mL lipopolysaccharide (O111:B4, Sigma-Aldrich, Cat#L4391) for 3 h. The cells were fixed and labeled with Alexa Fluor 647-conjugated cholera toxin subunit B (Invitrogen) and DAPI. For quantification, randomly selected 10–11 fields for each of three biological replicates were photographed using EVOS microscope (Thermo Fisher Scientific). The total area occupied by cells was determined for each image and divided by the number of DAPI-positive nuclei to obtain mean cell area. Chemotaxis assay of primary microglia Forty-eight hours after siRNA transfection, primary microglia were replated onto a Transwell insert (8.0-μm pore, 24-well, Corning) at the density of 5.0 × 10^4 cells/well with 10% LCM. Twenty ng/mL recombinant murine complement component C5a (R&D Systems) was added to the bottom well to stimulate downward migration. After 24 h, the cells were fixed with freshly prepared 4% PFA at RT for 15 min, and stained with 0.5% Crystal violet in 25% methanol at RT for 15 min. The cells on the upper surface were removed with cotton swabs. Images of migrated cells were acquired using EVOS microscope. The mean number of migrated cells were determined from five fields for each insert per experiment. alamarBlue assay of primary microglia Cell viability was determined using the alamarBlue assay kit (Bio-Rad). Briefly, primary microglia were plated in a 96-well plate at a density of 1.5 × 10^4 cells/well with 10% LCM and transfected with siRNA on the next day. Seventy-two hours after transfection, the medium was replaced by those containing 10% alamarBlue. After 3 h of culturing, fluorescence was measured by spectrophotometer. Values were shown by normalizing to the value of non-transfected cells. Co-immunoprecipitation experiment HEK293A cells were transfected with plasmids encoding FLAG-tagged human INPP5D, HA-tagged human TREM2 (an HA tag is inserted immediately after the signal sequence), and V5-tagged human TYROBP. Forty-eight hours after transfection, the cells were treated with or without 10 μM pervanadate for 30 min and then lysed with lysis buffer (50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% Triton X-100, protease/phosphatase inhibitors). The lysates were precleared, incubated with anti-FLAG (FUJIFILM Wako, clone 1E6) or irrelevant mouse IgG at the concentration of 2 μg/mL overnight at 4°C, and then added with Protein G Sepharose 4 Fast Flow (Cytiva) for 2 h. After washing with lysis buffer, immunoprecipitates were eluted from the beads by boiling in SDS-PAGE sample buffer containing 2-mercaptoethanol. In situ proximity ligation assay The in situ PLA was performed on fixed primary microglia with Duolink In Situ PLA probes and Detection Reagents Green (Sigma-Aldrich) according to the manufacturer’s instruction. Cells were permeabilized with 0.1% Triton X-100 in PBS for 5 min, incubated with blocking solution for 1 h at 37°C and then with pairs of primary antibodies (anti-TREM2 E7P8J, 1:250; anti-TYROBP D7G1X, 1:250; anti-INPP5D P1C1, 1:250; or anti-SYK SYK-01, 1:250) for 1 hr at RT. The cells were incubated with anti-mouse PLUS and anti-rabbit MINUS for 1 h at 37°C, followed by ligation reaction for 30 min and then by the rolling circle amplification for 100 min at 37°C. Finally, the cells were mounted with ProLong Diamond Antifade Mountant with DAPI. For quantification, randomly selected 9–13 fields for each of 2–4 biological replicates were photographed using EVOS microscope. The total number of PLA dots determined for each image was divided by the number of DAPI-positive nuclei to obtain mean number of dots per cell. The data were shown as relative to the averaged value of control group. INPP5D overexpression in MG6 cells and antibody-mediated stimulation of TREM2 Immortalized mouse microglial cell line MG6[360]^97^,[361]^98 was maintained in 10% LCM. Doxycycline-inducible lentiviral expression vectors for FLAG-tagged full-length INPP5D (wild-type and phosphatase-dead mutant D672A) were constructed by replacing Cas9 cassette from pCW-Cas9 (a gift from Eric Lander & David Sabatini; Addgene plasmid #50661; http://n2t.net/addgene:50661; RRID:Addgene_50661). Cells infected with the recombinant lentivirus were selected with 4 μg/mL puromycin, plated at a density of 5.0 × 10^4 cells/48-well in medium containing 5 μg/mL doxycycline and cultured for 48 h. The cells were then starved with HBSS(+) for 30 min, followed by stimulation with 15 μg/mL anti-TREM2 antibody (AF1729) or normal sheep IgG in HBSS(+) for 5 min, and harvested for immunoblotting. The antibodies were used after purification with Amicon Ultra device to remove sodium azide. RNA sequencing Total RNA was extracted from at least more than 12,000 isolated microglia for each sample using RNeasy Plus Micro Kit (QIAGEN, Cat#74034) according to the manufacturer’s instructions. RNA sequence library preparation, sequencing and mapping were performed by DNAFORM (Yokohama, Kanagawa, Japan). Qualities of RNA were assessed by Bioanalyzer (Agilent) to ensure that RNA integrity number is over 9.0. Double-stranded cDNA libraries were prepared using SMART-Seq v4 Ultra Low Input RNA Kit for Sequencing (TaKaRa) according to the manufacturer’s instruction and were sequenced using paired end reads (150 bp) on HiSeq. Obtained reads were mapped onto the mouse GRCm38.p6 (mm10) genome using STAR (version 2.7.3a). Total and mapped counts of reads for each sample were summarized in [362]Table S2. Reads on annotated genes were counted using featureCounts (version 1.6.4). The count data were analyzed using TCC-GUI[363]^59 to identify DEGs. qRT-PCR Complementary DNA (cDNA) was prepared from extracted RNA using ReverTra Ace qPCR RT Master Mix with gDNA Remover (FSQ-301, TOYOBO) according to the manufacturer’s instructions. cDNA was diluted in 1/25 with Milli-Q and used for the assay. Real-time reverse transcription was performed in duplicate using THUNDERBIRD SYBR qPCR Mix (QPS-201, TOYOBO) according to the manufacturer’s instructions in a LightCycler 480 Instrument II (Roche). Standard curves were generated for each gene by diluting one of the samples in 1/5, 1/25 and 1/125 with Milli-Q. Changes in gene expression were measured by comparative analysis of qRT-PCR by CT method which was normalized to Gapdh expression. The primers used in the assay were listed in [364]Table S1. Image acquisition and analysis Samples belonging to the same experiment were imaged during the same imaging session by using confocal laser scanning microscopy TCS-SP5 (Leica) or spinning disc microscopy Dragonfly 505 (Andor Technology). Super resolution images of Thioflavin S were acquired by LSM980 with Airyscan2 (Carl Zeiss Microscopy). If there is no additional description, five z-slices at the interval of 10 μm were obtained at 10× magnification for IBA1 (coverage analysis) and GFAP, or 20× magnification for PU.1, dystrophic neurite markers, Cd11c and Tmem119. For most cases, tiled images were stitched to reconstitute an entire image of the coronal section, the entire or subregion (hippocampus, cortex) of which were used for analyses. Except for p-tau analysis, images were maximally projected to the x-y plane by using ImageJ or Imaris (Oxford Instruments). ImageJ was also used for other image processing and analyses. If needed, we applied “mean filter” or “median filter” function to reduce random noise and “subtract background” function to reduce overall image background. Following these pre-processing, respective automated image processes were performed to images by using custom ImageJ macros. The same ImageJ macro was applied to each image in the same experiment. The methods for image processing were summarized as below. To quantify microglial coverage, the proportion of IBA1-positive area per individual Aβ plaque was calculated based on IBA1 and Aβ staining images that were preprocessed as follows. Aβ images were segmented by optimal threshold values determined by an automatic algorithm, and then by particle size. Considering that microglia approach Aβ plaques horizontally, the obtained binary images were dilated for 3 pixels to create Aβ plaque masks. IBA1 images were also segmented by optimal threshold values determined by an automatic algorithm to create IBA1 masks. Overlapping area between IBA1 and Aβ masks was normalized to plaque size as 100% for each plaque, and the average value of all plaques in each brain section was compared between groups. Otherwise, individual plaques in the hippocampal region were classified based on the IBA1-coverage into the following categories: dense (>40%), intermediate (20–40%) and sparse (<20%), and the proportion of each category was compared between groups. To quantify the immunoreactive area for GFAP, BACE1, LAMP1, and ubiquitin, the following preprocessing and analyses were performed on selected subregions in the cortex (regions of interest; ROIs) instead of the entire section. The original images were segmented by optimal threshold values determined by an automatic algorithm. For the BACE1 and LAMP1 analyses, smaller particles were excluded as non-specific signals. For the ubiquitin analysis, “watershed” function was applied to the threshold images to distinguish neuritic plaques from non-specific staining of nuclei, and then smaller particles were excluded. Based on the obtained images, the immunoreactive area was calculated and normalized to the ROI area, and the value was compared between groups. To quantify the number of microglia around Aβ plaques, PU.1-positive puncta were counted for each plaque based on PU.1 and Aβ staining images. Aβ masks were created as described above. PU.1 images were segmented by optimal threshold values determined by an automatic algorithm and were applied with the “watershed” function to distinguish adjacent PU.1-positive nuclei. The number of PU.1-positive puncta was determined for each plaque, and the average value of all plaques in each brain section was compared between groups. To quantify colocalization of p-tau and Aβ plaques, p-tau immunoreactive area was determined for each plaque based on pre-processed p-tau and Aβ images. We found that satisfactory p-tau labeling was only achieved within several micrometers beneath the surface of brain sections, possibly due to poor tissue penetration of the peroxidase-conjugated antibody used in the TSA method. Therefore, instead of using projected images, we selected an optimal staining image at a certain z-position for the analysis. Aβ plaque masks were created as described above. P-tau images were segmented by optimal threshold values determined by an automatic algorithm. From the threshold images, smaller particles (corresponding to non-specific neuronal staining) were excluded to obtain plaque-associated p-tau masks. Overlapping area between the two masks was quantified and normalized to the plaque size as 100% for each Aβ plaque, and the average value of all plaques in each observed region was compared between groups. To quantify the Thioflavin S staining of the plaques, images were segmented by optimal threshold values determined by an automatic algorithm. Smaller particles were excluded as non-specific signals. Mean intensities of individual plaques were determined, and the average value of all the plaques in each brain section was compared between groups. Quantification and statistical analysis Statistical analyses and graphical presentations were performed with GraphPad Prism 8/9 or the statistical language R. Data were analyzed by unpaired t-test with Welch’s correction for two group comparisons. For multiple group comparisons, ordinary one-way ANOVA followed by Tukey’s or Dunnett’s multiple comparison test was used, except for [365]Figures 7B, [366]S6E, S6G, and S6N (using repeated measures one-way ANOVA followed by Tukey’s test), [367]Figure 1F (using two-way ANOVA followed by Tukey’s test) and [368]Figure S4B (using two-way ANOVA followed by Sidak’s test). All error bars indicate the SE of mean. Sample sizes and any additional statistical details can be found in the figure legends. p-values lower than 0.05 were considered significant and represented by the following symbols: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Acknowledgments