GPCR19 regulates P2X7R-mediated NLRP3 inammasomal activation of microglia by Amyloid β in Alzheimer’s diseases

Amyloid β (Aβ) and/or ATP activates NLRP3 inammasome (N3I) by P2 × 7R ion channel of microglia, which is crucial in neuroinammation shown in Alzheimer’s disease (AD). Due to polymorphisms, subtypes, and ubiquitous expression of P2 × 7R, inhibition of P2 × 7R has not been effective for AD. We rst report that GPCR19 is a prerequisite for P2 × 7R-mediated N3I activation and Taurodeoxycholate (TDCA), a GPCR19 ligand, inhibited the priming phase of N3I activation, suppressed P2 × 7R expression and P2 × 7R-mediated Ca ++ mobilization, and N3I oligomerization which is essential for production of IL-1β/IL-18. Further, TDCA increased expression of scavenger receptor (SR) A, enhanced phagocytosis of Aβ, and decreased Aβ plaque numbers in the brain of 5x Familial Alzheimer’s disease (5xFAD) mice. TDCA also reduced microgliosis, prevented neuronal loss, and improved memory function of 5xFAD mice. The pleiotropic roles of GPCR19 in P2 × 7-mediated N3I activation suggest that targeting GPCR19 might resolve neuroinammation in AD patients.


Introduction
Despite Alzheimer's disease (AD) being the most common cause of dementia, no effective treatments are currently available 1 . Cholinergic, tau, and amyloid hypotheses have been suggested to explain the pathophysiology of AD 2 . Currently, the treatment of choices for AD patients are mostly based on cholinergic neurotransmission, which could not su ciently mitigate the progression of AD 3 . Although the glial cells constitute brain cells more than neurons, neurons received more attention than glial cells for a long time, possibly due to prominent neurological symptoms of AD patients 4 . Within the last few years, however, clinical trials have moved to reduce neuroin ammation incurred by reactive microglia 5 .
Neuroin ammation led by Aβ-activated microglia induces neuronal apoptosis in the hippocampus and cortex of AD patients 6 . In the brains of AD patients, pro-in ammatory mediators, such as reactive oxygen species (ROS), reactive nitrogen species (RNS), IL-1β, IL-6, and tumor necrosis factor (TNF)-α, are frequently increased 7 . The insoluble aggregates of Aβ and hyperphosphorylated tau, which make neuro brillary tangles, are possibly main initiators of neuroin ammation in these patients 8 . These damage-associated molecular patterns (DAMPs) interact with pattern recognition receptors (PRR) on membranes of brain cells or in the cytosol to initiate pro-in ammatory pathways 9 . Sustained neuronal apoptosis may unleash more DAMPs in the brain, which further amplify sterile in ammation in the brain 10 . Considering that neuroin ammation plays crucial roles in the cognitive and memory de cits by neuronal loss, controlling neuroin ammation may provide promising therapeutic strategies 11 .
The in ammasome plays central roles in the pathogenesis of many in ammatory disorders, including AD 12 . The NACHT, LRR, and PYD domains-containing protein 3 (NLRP3) polymorphisms are closely related with AD incidence 13 . Among the several in ammasomes, the most crucial contributor in AD pathologies is the NLRP3 14 . Aβ was e ciently cleared and cognition was improved in the AD mouse model by inhibiting activation of the NLRP3 in ammasome (N3I) 15 , suggesting that N3I is crucial in in ammatory neurodegeneration of AD 16 .
Unfortunately, however, strategies targeting components of N3I have not been successful in clinical trials for AD until now 18 . This may be due to redundant pro-in ammatory pathways activated by Aβ. For example, several PRRs recognize cytoplasmic tau and extracellular Aβ. Furthermore, the in ammasome is activated by canonical and non-canonical pathways, consisting of pro-caspase-1/4/5/11, gasdermin D, ASC, NLR proteins (such as NLRP1, NLRP3, NLRC4, NLRP6, or NLRP12), absent in melanoma 2 (AIM2), IFN-inducible protein 16 (IF116), and pyrin 19,20 , although their roles in Aβ-mediated in ammasomal activation have not been elucidated in detail yet. The signi cance of redundancy in developing in ammasomal inhibitors was well demonstrated in studies using the NLRP3-speci c inhibitor, MCC950 21 . MCC950 showed promising e cacy in preclinical settings, but not in clinical settings 22 . The redundancy of in ammasomal activation pathways implies evolutionary signi cance of the in ammasome responding to diverse environmental or endogenous threats to maintain tissue homeostasis 23 . For these reasons, a molecule at a higher level of the in ammasomal signaling cascade requires regulation to overcome redundancy in in ammasomal activation. A more plausible approach may involve targeting P2 × 7R, which is one of the top regulators of the signaling cascade necessary for N3I activation 24 .
Brain cells, including microglia, express purinergic receptors, and both ionotropic P2X and metabotropic P2Y receptors are crucial in AD pathogenesis 25 . P2 × 7R is the ion-channel primarily studied in terms of N3I activation by Aβ 26 . Upon binding with ATP or Aβ, P2 × 7R renders cell membranes permeable to K + and Ca ++ , which activate the in ammasome 27 . N3I of P2 × 7R −/− microglia was not activated in response to Aβ 28 , suggesting the essential role of P2 × 7R in Aβ-mediated neuroin ammation. P2 × 7R promotes the assembly of N3I, secretion of IL1β/18, and pyroptosis 29 . Intriguingly, the P2 × 7R is overexpressed in glial cells from AD patients and Aβ injection into the hippocampus increases P2 × 7R expression 30 . Taken together, the P2 × 7R plays a key role in chronic neuroin ammation and neurodegeneration in AD.
Apyrase blocks activation of N3I by Aβ 31 , suggesting that ATP-P2 × 7R interaction is crucial in N3I activation in response to Aβ 32 . High levels of ATP are passively released from necrotic cells and act as a pro-in ammatory DAMP, binding to the P2 × 7R and activating the N3I 33 . P2 × 7R activation creates membrane pores, through which ATP can leak further 34 .
Activation of P2 × 7R is also crucial in impairing phagocytosis of Aβ 35 . In AD patients, the phagocytic ability of microglia was insu cient to clear Aβ 5 . P2 × 7R −/− microglia phagocytosed Aβ more e ciently that wild type 36 . These ndings clearly show both pro-in ammatory roles and anti-phagocytic functions of P2 × 7R in glial cells 37 .
Thus, many P2 × 7R inhibitors have been developed to control in ammasomal activation without success until now 38 . These inhibitors effectively decreased in ammatory responses in AD mice 39 . However, many of these did not meet the clinical needs 40 . The human P2 × 7R is highly polymorphic and there are several isotypes 41 . Ten human P2 × 7R gene splice variants (P2 × 7RA-J) might produce a complex combination of P2 × 7R with various haplotypes that cause a broad spectrum of responsiveness to P2 × 7R inhibitors in clinical settings.
For these reasons, a non-polymorphic physiological regulator for P2 × 7R might be an alternative target to control in ammasome activation. We found that GPCR19 regulates P2 × 7R in N3I activation. TDCA, a GPCR19 agonist, inhibited the priming phase of N3I activation of microglia by activating adenylate cyclase, as reported earlier in other types of cells, and inhibited the activation phase of N3I by inhibiting P2 × 7R function.

Results
GPCR19 is a prerequisite for P2X7R-mediated Ca ++ mobilization in microglia.
We sorted primary microglia using a magnetic column from the brains of C57BL/6 (B6), P2X7 -/-, and GPCR19 -/mice. The purity was above 95% (Supplementary Figure 1a). Upon treatment with ATP or BzATP, Ca ++ mobilization signi cantly decreased in P2X7R -/or in GPCR19 -/microglia compared with microglia from WT B6 (Figure 1a Treatment of BV2 cells with Aβ ± ATP signi cantly downregulated expression of GPCR19 on the cell surface, while this was recovered by TDCA treatment. On the contrary, the surface expression of P2X7R was upregulated by Aβ ± ATP, which was inhibited by TDCA treatment (Figure 1g, 1h; Supplementary  Figure 2a). Although treatment with Aβ ± ATP also downregulated cytoplasmic GPCR19 expression in BV2 cells and upregulated cytoplasmic P2X7R expression, TDCA treatment for 1h did not normalize cytoplasmic expression levels of these two molecules (Supplementary Figure 2b~d). However, TDCA treatment for 24h signi cantly upregulated cytoplasmic GPCR19 expression which was suppressed by Aβ ( Supplementary Figure 2e, 2f). GPCR19 expression in the frontal cortex of mice in the group of 5xFAD-TDCA (1 mg/kg, i.p., q.d. for 10 weeks) was signi cantly higher than that of mice in the group of 5xFAD-PBS (Figure 1i, 1k; Supplementary Figure 2g). The expression levels of P2X7R in the cortex of mice in the group of 5xFAD-TDCA (1 mg/kg, i.p., q.d. for 10 weeks) were signi cantly lower than that of mice in the group of 5xFAD-PBS (Figure 1j, 1k; Supplementary Figure 2h). The expression levels of GPCR19 in the frontal cortex were signi cantly lower in six-or nine-month-old 5xFAD mice compared to that in threemonth-old 5xFAD mice or B6 mice (Figure 1l, m). The expression levels of GPCR19 in the frontal cortex of six to nine-month-old 5xFAD mice did not differ signi cantly from that of three-month-old GPCR19 -/mice.
Conversely, expression levels of P2X7R in the frontal cortex were signi cantly higher in three-to ninemonth-old 5xFAD mice than in three-month-old B6 mice (Figure 1n, 1o). The expression level of P2X7R in the frontal cortex of nine-month-old 5xFAD mice was signi cantly higher than that of three-month-old 5xFAD mice.
In primary microglial cells, Aβ signi cantly increased expression of NLRP3 and ASC (Figure 2b, 2c). Treatment with Aβ + ATP further increased expression of NLRP3 and ASC. Without TDCA, NLRP3 colocalized with DAPI + nucleus (white arrows), although the level of cytoplasmic NLRP3 also increased by treatment with Aβ ± ATP. TDCA treatment signi cantly suppressed expression of these two molecules, as well as their colocalization (yellow dots, Figure 2b, 2d; Supplementary Figure 3a). Interestingly, levels of nuclear NLRP3 (white arrows) were signi cantly downregulated by TDCA treatment. In the BV2 cell line, Aβ + ATP increased NLRP3 and ASC expression, but not by Aβ alone. As observed in primary microglia of mice, TDCA also suppressed expression and colocalization of NLRP3 and ASC in BV2 cells To elucidate how TDCA inhibits transcription and expression of N3I components, we analyzed the GPCR19-cAMP-PKA-NF-kB axis after BV2 cells were treated with Aβ ± TDCA 40 ( Figure 2f~2l). TDCA increased cAMP production in BV2 cells irrespective of Aβ treatment (Figure 2f). Adenylyl cyclase inhibitor (KH7) blocked TDCA-mediated cAMP production in BV2 cells (Figure 2f). BV2 cells increased production of IL-1β, IL-18, and TNF-α upon treatment with Aβ ± ATP, while TDCA suppressed production of these cytokines (Figure 2g~2i). BV2 cells produced ROS in response to Aβ (Supplementary Figure 3g), and this was inhibited by TDCA treatment (Figure 2j).
In vivo, TDCA (1 mg/kg, i.p., q.d.) administration for 10 weeks decreased expression of NLRP3 and ASC in the frontal cortex of 5xFAD mice when compared with those of PBS-treated 5xFAD mice (Figure 2k Proteogenomic analysis of brain tissues of 5xFAD mice treated with TDCA The proteomes of brain tissues from 5xFAD mice were analyzed post-administration of TDCA (1 mg/kg, i.p., q.d.) for 10 weeks (Figure 3a, 3b). In total, 3,259 unique proteins were identi ed at a protein threshold of a 1.0% false discovery rate. Among these proteins, 460 proteins showing peptide spectral counts in more than two assays from triplicate assays, with a fold change of more than 2 between PBS-and TDCAtreated groups, are depicted on the heat map, plotted with the Perseus software platform (http://www.perseus-framework.org). Proteomic analysis indicated that 56 proteins exhibited more than 2-fold changes in the 95% con dence interval (Figure 3a), demonstrating two distinct proteome clusters that were upregulated in TDCA-treated groups and downregulated in PBS-treated groups, or vice versa.
Functions of these proteins were further analyzed based on QIAGEN's IPA database (Figure 3b). Notably, a canonical pathway, 'Regulation of eIF4 and p70S6K', was enriched by TDCA treatment, suggesting that TDCA plays critical roles in translational regulation followed by calcium signaling, which could exert allosteric regulatory effects on many enzymes and proteins. Based on proteomic analysis, we further analyzed transcript levels of several pro-in ammatory cytokines in both the hippocampus and cortex of 5xFAD mice. The TDCA treatment for 10 weeks downregulated IL-1β, TNF-α, IL-33, 1L-12, CCL-11, and CCL-5 transcripts. On the contrary, increased transcripts of IFN-γ, IL-10, CCL-17, GPCR19, CD47, FPR-2, CD36, SRB1, and SRA in the brain of 5xFAD mice (Figure 3c).

TDCA increases expression of scavenger receptor (SR) A and phagocytosis of Aβ
We further investigated the effects of TDCA on SRA, since the expression of SRA was increased by TDCA treatment in the proteomic analysis of 5xFAD mice brains. In vitro, treatment of BV2 cells with Aβ signi cantly downregulated SRA expression and was normalized by TDCA (Figure 3d; Supplementary  Figure 4a, 4b). SRA transcripts were signi cantly downregulated in BV2 cells upon treatment with Aβ for 24h and were dose-dependently increased by TDCA treatment (Supplementary Figure 4c). In vivo, CD11b int CD45 int primary microglia isolated from the brains of 5xFAD mice treated with TDCA for 10 weeks showed signi cantly higher SRA levels when compared with that of microglia from PBS-treated 5xFAD mice (Figure 3e). Since P2X7 is known to inhibit phagocytosis 35,37 and SRA is known to increase phagocytosis, we hypothesized that TDCA-induced suppression of P2X7 expression and TDCA-induced SRA expression might contribute to increased phagocytosis of Aβ by microglia. As expected, primary microglia from B6 mice showed decreased expression of P2X7R and increased phagocytosis of uorescent Aβ oligomers (fAβ) upon treatment with TDCA ( Figure 3f; Supplementary Figure 5a).
Interestingly, P2X7R high microglia (white arrows) did not phagocytose fAβ even after treatment with TDCA ( Figure 3f). After being phagocytosed, fAβ were co-localized with LAMP-2 + phagosomes in BV2 cells and TDCA decreases the number of microglia and increases the number of MDSCs in the brains of 5xFAD mice Brain cells were analyzed using FACS after 5xFAD mice received TDCA treatment for 10 weeks (1 mg/kg, i.p., q.d., Figure 5). DAPIsinglet cells were gated and expression of CD45 and CD11b were determined  Figure 8b). The number of CD45 int CD11b int microglia in 5xFAD mice treated with PBS was signi cantly higher than that in age-matched B6 mice ( Figure 5c). After TDCA treatment for 10 weeks, the number of microglia was signi cantly lower than that observed in brains of 5xFAD mice treated with PBS ( Figure 5c). Interestingly, the number of CD45 int CD11b hi MDSCs was signi cantly lower in PBS-treated 5xFAD mice than age-matched WT B6 mice and was higher in the TDCA-treated group when compared with the PBS-treated group of 5xFAD mice ( Supplementary Figure 9). The number of Ly6C int Ly6G + PMN-MDSCs in the spleen of 5xFAD mice was signi cantly increased by TDCA treatment (Figure 6b). The number of Ly6G -Ly6C hi M-MDSCs in the spleen of 5xFAD mice treated with TDCA was signi cantly lower than that of PBS-treated 5xFAD mice ( Figure 6c).
TDCA decreases apoptosis of neurons in 5xFAD mice brain The NeuN + cells in the brains of 5xFAD mice were stained after treatment with TDCA for 10 weeks (1 mg/kg, i.p., q.d.). The frontal cortex and hippocampus (CA1, CA3, and DG) were observed using confocal microscopy ( Figure 7a). NeuN + cells in the frontal cortex were signi cantly higher in 5xFAD mice treated with TDCA when compared to 5xFAD mice treated with PBS ( Figure  Spatial learning and memory were also assessed by the Y-maze test ( Figure 8e) and Novel object recognition test (NOR) (Figure 8f, 8g). The % alternation of the mice in the 5xFAD-PBS group was signi cantly lower than that of mice in the B6-PBS group and 5xFAD-TDCA group ( Figure 8e). However, the total number of arm entry did not differ signi cantly between the groups of mice (Supplementary Figure 11f). In the NOR test, exploration time to a new object compared to an old object was signi cantly higher in B6-PBS and 5xFAD-TDCA mice (Figure 8f). 5xFAD-PBS mice did not exhibit a difference in exploration time between old and new objects ( Figure 8f). The discrimination index of the mice in the 5xFAD-TDCA group was signi cantly higher than that of mice in the 5xFAD-PBS group (Figure 8g).
After oral administration of TDCA (5 or 10 mg/kg, p.o., q.d.) for 14 weeks, spatial learning and memory of mice were also tested (Supplementary Figure 12a). No obvious changes in body weight were observed (Supplementary Figure 12b). Escape latency of mice in the 5xFAD-TDCA (10 mg/kg, p.o., q.d.) group decreased (Figure 8h), as much as observed in mice treated with TDCA i.p.. In the probe test at 5 d, 5xFAD-TDCA (10 mg/kg) mice exhibited increased numbers of crossings over the target quadrant ( Figure  8i). 5xFAD-TDCA (10 mg/kg) mice remained in the target quadrant for longer periods of time and spent less time in the opposite quadrant when compared with 5xFAD-vehicle mice ( Figure 8j). The mean speed and total travel distance of 5xFAD mice were similar between groups (Supplementary Figure 12c, 12d). Spatial learning and memory were also assessed by the Y-maze test after feeding TDCA orally ( Figure   8k). The % alternation of the mice in the 5xFAD-vehicle group was signi cantly lower than that of mice in the 5xFAD-TDCA (10 mg/kg) group ( Figure 8k).

Discussion
Senile Aβ plaques in the human brain incur microgliosis, which is responsible for neuroin ammatory cascades, causing memory and cognitive impairment, eventually progressing to dementia 6 . Reactive microgliosis accompanied by neuronal damages aggravates AD 5 . Aβ-mediated activation of N3I exacerbates the pathogenesis of AD by inducing neuroin ammaiton 14 . However, the intervention of N3I activation enough to suppress neuroin ammation of AD is still not successful. Therefore, many resources have been put forward to halt the progression of Aβ-mediated N3I activation, which might delay neuronal loss due to the neuroin ammatory cascade. In this study, we found that administration of a GPCR19 agonist, TDCA (1 mg/kg, i.p., q.d. or 10 mg/kg, p.o., q.d.), for 10 or 14 weeks, signi cantly improved learning and memory of 5xFAD mice (Fig. 8).
The possible mode of action of TDCA could be explained in four ways. First, TDCA could suppress the priming phase of N3I activation (transcription of NLRP3, ASC, and Pro-caspase-1) (Fig. 2a) and the activation phase of N3I (production of mature IL-1β and IL-18 by NLRP3-ASC oligomerization) ( Fig. 2b-2 h) in response to Aβ and ATP. Second, TDCA could inhibit production of crucial proin ammatory mediators, such as ROS and TNF-α, of microglia independent of N3I activation (Fig. 2i, 2j). Third, TDCA could augment clearance of Aβ by enhancing phagocytosis and by suppressing P2 × 7 expression ( Fig. 2f, 2j). Lastly, TDCA could increase the number of regulatory MDSCs in the brain that might further suppress neuroin ammation by Aβ (Fig. 5a, 5d). Overall, TDCA reduces neuroin ammation and prevents neuronal apoptosis, which delays impairment of spatial learning and memory ( Fig. 7, 8).
TDCA regulates N3I activation by altering functions and expression of P2 × 7R after binding with GPCR19 ( Fig. 1). The P2 × 7R on microglia is crucial in Ca ++ mobilization that initiates N3I activation 42 . Many glial functions are mediated by Ca ++30 , such as production of cytokines and chemokines 43 . ATP released from damaged neurons could activate P2 × 7R in AD.
Upon stimulation with P2 × 7R agonists (ATP or BzATP), GPCR19 −/− microglia could not mobilize cytosolic Ca ++ as much as WT microglia (Fig. 1a, 1b However, none have been reported for the role of GPCR19 in this process until now. Several ndings in this study suggest the bi-phasic role of GPCR19 in regulating P2 × 7R. In immediate phase in response to pro-in ammatory cues such as Aβ and/or ATP (50-150 sec), TDCA may regulate function of P2 × 7R by allosteric modulation of GPCR19 that interferes GPCR19-P2 × 7R interaction. GPCR19 and the P2 × 7R are co-localized on the cell membrane of resting microglia ( Fig. 1) and Ca ++ is mobilized within 100 sec in response to ATP. A GPCR19 agonist, TDCA, inhibited Ca ++ mobilization when microglia were stimulated with Aβ + P2 × 7R agonist (ATP or BzATP) within 100 sec, further supporting the idea that GPCR19 is a prerequisite for the activation of P2 × 7R and TDCA interferes interaction of these two molecules ( Fig. 1d-1f). The nding suggests that binding of TDCA with GPCR19 might cis-regulate opening the pore of the P2 × 7R. It is supposed that binding of TDCA with GPCR19 might alter tertiary structure of GPCR19 necessary for opening the pore of the P2 × 7R in response to its ligands. In delayed phase in response to Aβ and/or ATP (1 h), TDCA suppresses P2 × 7R expression and increases expression of GPCR19 on the cell membrane (Fig. 1g, 1 h), suggesting plausible trans-regulation of P2 × 7R expression by the GPCR19mediated signaling cascade. In summary, the TDCA-GPCR19 complex might transmit signals necessary for inhibiting P2 × 7R expression on the membrane within an hour (delayed response) and might alter structures of GPCR19 necessary for Ca ++ current incurred by P2 × 7R activation within a couple of seconds (immediate response) ( Supplementary Fig. 13).
In the brains of nine-month-old 5xFAD mice, the expression of GPCR19 was signi cantly lower and the expression of P2 × 7R was signi cantly higher than levels observed in the brains of three-month-old 5xFAD mice or B6 mice (Fig. 1m, 1o). The expression levels of GPCR19 and P2 × 7R on microglia were reciprocally regulated after stimulation with Aβ ± ATP in vitro, which was reverted by TDCA treatment (Fig. 1g). Taken together, these ndings suggest that neuroin ammation by Aβ might be responsible for downregulation of anti-in ammatory GPCR19 and upregulation of proin ammatory P2 × 7R upon aging.
The relative expression levels of GPCR19 and P2 × 7R on microglia might be crucial biomarkers indicating the severity of neuroin ammation of AD patients.
P2 × 7R inhibits phagocytosis of Aβ in various pro-in ammatory microenvironments in AD 35 . TDCA enhanced phagocytosis of Aβ and coincided with downregulation of P2 × 7R expression in the microglia (Fig. 3f). Furthermore, TDCA treatment increased expression of SRA, which might further enhance phagocytosis of Aβ by microglia (Fig. 3g). Downregulation of P2 × 7R, which inhibits phagocytosis, and upregulation of SRA, which enhances phagocytosis, might explain the possible roles of TDCA in clearing Aβ plaques in the brain of 5xFAD mice in vivo (Fig. 4a).
Numbers of PMN-MDSCs signi cantly increased in the brain ( features in the AD brain, in addition to direct suppression of N3I pro-in ammatory pathways by TDCA. Proteomic analysis supports these ideas. Whole brain lysates exhibited global editing of the brain proteome that leads to an anti-in ammatory microenvironment in the AD brain (Fig. 3a). For example, upregulation of clathrin-mediated endocytosis, FXR pathway, mTOR signaling, and acute phase response signaling, in addition to downregulation of NO signaling and PI3K/AKT signaling were observed after treatment with TDCA (Fig. 3b).
Various bile acids are GPCR19 agonists and could reduce in ammation in the brain 47 . However, many studies were carried out with bile acids in concentrations that are unobtainable in pharmacological or pathological conditions in vivo 48 . Thus, the exact role of bile acids in modulating brain in ammation in AD could not be concluded. In this study, we showed that TDCA, one of the bile acids interacting with GPCR19 49 , could suppress brain in ammation in 5xFAD mice by inhibiting the P2 × 7R-N3I axis. Further, oral administration of TDCA also improved spatial learning and memory, as observed with i.  BV2 cells were seeded on 25 mm cover glasses in a 6 well plate. After starvation, cells were treated with Aβ (2 µM) with or without TDCA (400 ng/ml) for 24 h. Cells with glass coverslips were transferred and loaded with 2 µM Fluo-4/AM for 30 min at 37°C in a physiological external solution mentioned above. After loading, cells were transferred to an open perfusion chamber and uorescence was measured at 494/506 nm as previously described.
BV2 cell Ca ++ sensing was measured using a BD calcium assay kit (BD bioscience, San Diego, CA, USA) according to the manufacturer's protocol. In a 6 well plate, 1.5 x 10 5 cells/2 ml media were treated with Aβ with or without TDCA for 24 h after serum starvation. Cells were harvested in a FACS tube from the cell culture plate, washed with complete RPMI media, loaded with loading dye, and incubated for 1 h at 37°C in a 5% CO 2 atmosphere. Cells were acquired for 1 min for basal signaling, then incubated with BzATP To stain in ammasomal components, BV2 or primary microglia cells isolated from the brains of one~two-day-old B6 mice were seeded on 12 mm microscope cover glasses in 24 well plates, and were treated with Aβ (2 µM

Animals
The 5xFAD mice co-overexpress high levels of APP with three FAD mutations (Swedish (K670N/M671L), Florida (I716V), and London (V717I)), and high levels of presenilins 1 (PSEN1) with two FAD mutations (M146L, L286V), which are speci cally overexpressed in the brain, regulated by neural-speci c Thy1 promoter 51 . 5xFAD mice were maintained by breeding male 5xFAD mice with female B6 mice. The SJL F1 hybrid was produced by an SJL male and a B6 female. PCR was performed for genotyping of the mice. 5xFAD or SJL mice were kindly provided by Professor Mook-Jung, In-hee or Professor Sung, Jung-Joon, respectively, of Seoul National University. All animal experiments were approved by the institutional animal care and use committees (IACUC) of Seoul National University (SNU-170517-25) and performed in accordance with animal ethics regulations. Mice were maintained in speci c pathogen-free conditions at the animal facility of Wide River Institute of Immunology.
Drug administration 5xFAD male and female mice (8 to 10 weeks old) were injected intraperitoneally (i.p.) with either 1 mg/kg of TDCA or PBS ve times/week for ten weeks. Age and gender matched non-transgenic littermates were used as a control group. Behavioral tests were performed after treatment and then mice were sacri ced for further experimentation. In the case of oral (p.o.) treatment, 5 mg/kg or 10 mg/kg TDCA was mixed with Monoolein (1-Oleoyal-rac-glucerol) and Oleic acid (TDCA: Monoolein: Oleic acid = 1: 2: 1) in PBS.
TDCA (5 or 10 mg/kg) or PBS p.o. were administered using gastric gavage ve/week for 14 weeks. Behavioral tests were performed during the last two~three weeks of treatment.

Morris water maze test
The maze was composed of a circular pool (1.5 m in diameter, 80 cm in height) with spatial cues at three different locations. Before testing, the pool was lled with opaque water adjusted to 20 ± 1°C. On the rst day, mice were allowed to freely swim in the water for 60 seconds to nd the escape platform located in one quadrant of the pool. When they failed to nd the platform, the mice were guided to the platform.
Once on the platform, mice were allowed to remain there for 30 sec. From the next day for four consecutive days, the same procedure was repeated from three different starting points to train the mouse, and the time to reach the platform was recorded every day. After the four-day training period, the probe test was performed in the same manner, but without the platform. Each mouse was allowed to swim from one starting point for 60 seconds, which was recorded using a video camera. The video was analyzed for the movement of mice in the water using tracking software (SMART3.0, Panlab Harvard Apparatus, Barcelona, Spain) to count the number of crossings, the time on the platform, and to measure the time spent at each quadrant of the pool.

Y-maze test
Mouse functional behavior tests were performed on TDCA-or PBS-treated 5xFAD mice and WT (B6) control mice. The Y-maze test was assessed over the course of four days. On the rst two days, individual mice were habituated to the task room and experimenter for 5 min. On the third day after task room and experimenter habituation, mice were allowed to habituate to the Y-maze for less than 1 min. On the last day, mice entered the middle of the Y-maze and were allowed to move freely within the maze for 8 min.
Each mouse movement was recorded using a video camera. The video was analyzed for all mouse entries regarding limbs that pass through each half arm of the maze. Total arm entries and percentage of alteration were counted for each mouse and compared between groups of mice.

Novel object recognition test (NORT)
NORT was assessed over the course of four days. The apparatus consisted of a white acrylic box (350 mm x 450 mm x 250 mm). The basement of the box was divided into 6 equal rectangles. On the rst two days, each mouse was habituated to the box for 10 min. On the third day, two similar cylindrical objects were xed in the box and mice were allowed to explore the objects freely for 10 min. On the last day, one cylindrical object was replaced by a similar size different object and the mouse explored this for 5 min. Exploration of two cylindrical objects on the third day, and exploration of the novel object and cylindrical object on day four was recorded for each mouse using a video camera. Total time and frequency of novel and old object exploration were counted for each mouse using the video footage. The percent discrimination index of the novel object was calculated from exploration time of novel and old object.
Flow cytometry analysis of brain and spleen cells Cells from the isolated brain and spleen were prepared from TDCA-or PBS-administered mice after ten Primary microglia isolated from one~two-day-old B6 mice were cultured and treated with Aβ (2 µM) and TDCA (400 ng/ml) for 48 h. Cells were harvested and incubated with CD86-PE-cy7 (Clone: GL1, BD Pharmingen) and CD206-PE (Clone: C068C2, Biolegend) for 30 min at 4°C. Samples were washed in DPBS, suspended in a FACS buffer containing DAPI (0.3 µg/ml), and immediately acquired using ow cytometry. Further analysis was performed using FlowJo.
Immunohistochemistry TDCA-or PBS-treated 5xFAD or B6 mice were anesthetized and perfused with ice-cold PBS. Brains were harvested and maintained in 10% neutral buffer formalin (Sigma-Aldrich) for 24 h at 4°C and then embedded in para n (Lecia, Illinois, USA). The para n embedded brains were cut (3 μΜ) using microtome (Thermo Fisher Scienti c), then depara nized in xylene, and rehydrated in a graded ethanol series (100 %, 90%, 80%, and 70%). Antigen unmasking was performed by heating the brain sections in citrate-based buffer (pH 6, Vector laboratories, Burlingam, CA, USA). The sections were incubated in 0.3% Triton X-100 for 30 min at room temperature for intracellular staining, and then blocked with blocking solution of 10% normal goat serum (Thermo Fisher Scienti c) and 1% BSA in PBS for 1 h. Tissue samples were incubated overnight at 4°C with the following primary antibodies: GPCR19 (Novus Biologicals), NLRP3 (Abcam), ASC (Santa Cruz Biotechnology), NeuN (Clone A60, Merk Millipore, Temecula, CA, USA), Iba-1 (Wako, Osaka, Japan), and GFAP (Thermo Fisher Scienti c). Slices were subsequently incubated for 1 h at room temperature with Alexa 488, 532, or 546-conjugated IgG secondary antibodies, as appropriate, then counterstained with DAPI for 10 min The Aβ core plaque was labeled by treating tissue with 1% Thio avin-S (Sigma-Aldrich) in PBS for 10 min at room temperature after being depara nized in xylene and rehydrated in a graded ethanol series (100 %, 90%, 80%, and 70%). The tissue slides were then washed thrice with 70% ethanol following DW three times. Fluorescence imaging was performed using Confocal Microscope A1 (Nikon). NIS-Elements.AR.Ink (version 4.2, Nikon) was used to measure mean uorescence intensity (MFI) of ROI.

Global protein pro ling
Anesthetized Mice were perfused with cold PBS and the whole brain was extracted, followed by snapfreezing using liquid nitrogen. The frontal cortex and hippocampus region of each mouse was collected

Statistical analysis
The data are expressed as the mean ± SEM and were analyzed using a two-sided, unpaired Student's t test. The mean value between groups were compared using GraphPad Prism 5.0 software (GraphPad Software, La Jolla, CA, USA), unless otherwise indicated. P values < 0.05 were considered statistically signi cant.  were analyzed. Expression levels of GPCR19 (i) and P2X7R (j) in the frontal cortex of 5xFAD mice (n = 6/group) were analyzed using confocal microscopy after treatment with 1 mg/kg TDCA i.p for ten weeks.
k, MFI ± SEM of 2~3 ROI (× 200) from panel i and j were analyzed. m, GPCR19 expression and o, P2X7R expression levels in the frontal cortex of B6, 5xFAD, GPCR19-/-, and P2X7R-/-mice were analyzed using WB. The expression levels of GPCR19 (l) and P2X7R (n) in B6 mice were considered as 100%. The individual samples are shown with the mean ± SEM. *P < 0.05 using the Student's unpaired t-test. P2X7R (Red) on the surface of BV2 cells were analyzed using confocal microscopy after treatment with Aβ (2 µM), ATP (1 mM), and/or TDCA (400 ng/ml) for 1 h. h, MFI ± SEM of 4~5 ROI (× 600) from panel g were analyzed. Expression levels of GPCR19 (i) and P2X7R (j) in the frontal cortex of 5xFAD mice (n = 6/group) were analyzed using confocal microscopy after treatment with 1 mg/kg TDCA i.p for ten weeks.
Average intensity re ecting intracellular Ca++ mobilization (shown in left panels) and delta intensity (max-min) after stimulation was analyzed (right panel). c, Co-localization of GPCR19 and P2X7 on membranes of primary microglia are shown using confocal microscopy. d, e, The representative plots (left panels) of three independent experiments (right panels) showing Ca++ mobilization of BV2 cells in response to Aβ (2 µM) and TDCA (400 ng/ml) in presence of ATP (d) or BzATP (e) are depicted. f, Ca++ mobilization of BV2 cells treated with Aβ (2 µM) and TDCA (200~800 ng/ml) in response to BzATP (300 µM) was measured using ow-cytometry. The representative FACS plots (left panels) of three independent experiments (right panels) are depicted. g, The expression levels of GPCR19 (green) and P2X7R (Red) on the surface of BV2 cells were analyzed using confocal microscopy after treatment with Aβ (2 µM), ATP (1 mM), and/or TDCA (400 ng/ml) for 1 h. h, MFI ± SEM of 4~5 ROI (× 600) from panel g were analyzed. Expression levels of GPCR19 (i) and P2X7R (j) in the frontal cortex of 5xFAD mice (n = 6/group) were analyzed using confocal microscopy after treatment with 1 mg/kg TDCA i.p for ten weeks.        The number of Aβ plaques and microglia in 5xFAD mouse brain The 5xFAD mice (n = 9/group) were treated with TDCA (1 mg/kg, i.p) for ten weeks. a, Para n sections of brains were stained with Thio avin-  The number of Aβ plaques and microglia in 5xFAD mouse brain The 5xFAD mice (n = 9/group) were treated with TDCA (1 mg/kg, i.p) for ten weeks. a, Para n sections of brains were stained with Thio avin-  The number of Aβ plaques and microglia in 5xFAD mouse brain The 5xFAD mice (n = 9/group) were treated with TDCA (1 mg/kg, i.p) for ten weeks. a, Para n sections of brains were stained with Thio avin-S to show Aβ plaques (green dots indicated with white arrows) in the left panel. The number of plaques and total area of plaques were quantitated using Image J after selecting a random eld from each brain sample at 100 × magni cation. b, Frozen sections of 5xFAD mice brain were stained with Iba-     TDCA prevents neuronal apoptosis in 5xFAD mouse brain. Confocal microscopy showing NeuN+ cells in frontal cortex, CA, and DG of 5xFAD mice (n = 8/group) treated with 1 mg/kg TDCA i.p., q.d. for ten weeks.    The red arrows and red circles indicate release points and positions of the platform, respectively. Black dots indicate the end points for individual mice. e, Alteration percentages of mice (n = 9~12) in the Y maze test were calculated by; 100 × # spontaneous alteration / # of total arm entry. f, Total exploration time for each object of individual mice in the NOR test are depicted. g, Discrimination index (%) = 100 × time spent to explore novel object / exploration time for both novel and old object for each mouse (n = 10). Memory of 5xFAD and B6 mice (n = 8/group) were tested using the MWM test and the Y maze test after feeding the mice with TDCA (5 or 10 mg/kg) for 14 weeks (h ~ k). h, Escape latency l, Number of platform crosses and j, time in target and opposite quadrant were measured using the MWM test. k, Alteration percentages in the Y maze test were calculated. The individual samples are shown with the mean ± SEM. *P < 0.05 using the Student's unpaired t-test. The red arrows and red circles indicate release points and positions of the platform, respectively. Black dots indicate the end points for individual mice. e, Alteration percentages of mice (n = 9~12) in the Y maze test were calculated by; 100 × # spontaneous alteration / # of total arm entry. f, Total exploration time for each object of individual mice in the NOR test are depicted. g, Discrimination index (%) = 100 × time spent to explore novel object / exploration time for both novel and old object for each mouse (n = 10). Memory of 5xFAD and B6 mice (n = 8/group) were tested using the MWM test and the Y maze test after feeding the mice with TDCA (5 or 10 mg/kg) for 14 weeks (h ~ k). h, Escape latency l, Number of platform crosses and j, time in target and opposite quadrant were measured using the MWM test. k, Alteration percentages in the Y maze test were calculated. The individual samples are shown with the mean ± SEM. *P < 0.05 using the Student's unpaired t-test.