Research ArticleNEUROSCIENCE

Sodium rutin ameliorates Alzheimer’s disease–like pathology by enhancing microglial amyloid-β clearance

See allHide authors and affiliations

Science Advances  27 Feb 2019:
Vol. 5, no. 2, eaau6328
DOI: 10.1126/sciadv.aau6328

Abstract

The accumulation of aggregated amyloid-β (Aβ) in the brain is the first critical step in the pathogenesis of Alzheimer’s disease (AD), which also includes synaptic impairment, neuroinflammation, neuronal loss, and eventual cognitive defects. Emerging evidence suggests that impairment of Aβ phagocytosis and clearance is a common phenotype in late-onset AD. Rutin (quercetin-3-rutinoside) has long been investigated as a natural flavonoid with different biological functions in some pathological circumstances. Sodium rutin (NaR), could promote Aβ clearance by increasing microglial by increasing the expression levels of phagocytosis-related receptors in microglia. Moreover, NaR promotes a metabolic switch from anaerobic glycolysis to mitochondrial OXPHOS (oxidative phosphorylation), which could provide microglia with sufficient energy (ATP) for Aβ clearance. Thus, NaR administration could attenuate neuroinflammation and enhance mitochondrial OXPHOS and microglia-mediated Aβ clearance, ameliorating synaptic plasticity impairment and eventually reversing spatial learning and memory deficits. Our findings suggest that NaR is a potential therapeutic agent for AD.

INTRODUCTION

Alzheimer’s disease (AD) is the most common form of dementia in the elderly. It is estimated that AD will affect more than 100 million people worldwide by 2050, which will cause a huge burden for families and societies (1, 2). Mounting evidence suggests that accumulation and aggregation of amyloid-β (Aβ) is a key initiating event in the pathogenesis of AD (3). It is believed that imbalance of Aβ production and clearance is the main reason for Aβ accumulation and aggregation in the brain, which triggers many subsequent pathological events such as synaptic impairment and degeneration, oxidative stress, neuroinflammation, and neuronal loss (46). Accumulating evidence also indicates that impairment of Aβ clearance is a common prelude to late-onset AD (6). In addition, recent studies show that a mutation in the triggering receptor expressed on myeloid cells 2 (Trem2) impairs microglial Aβ phagocytosis and clearance and accelerates AD pathology (7, 8). Therefore, great efforts have been made to discover effective drugs that could promote Aβ clearance. A number of drugs targeting Aβ clearance (such as Aβ antibodies) or production (such as Aβ secretase inhibitors) have proceeded to clinical trials, but most of them failed to improve the cognitive functions in patients, which are mainly attributed to the signal targets, side effects, or the neuroinflammatory responses of these drugs (9, 10). By contrast, many traditional medicines and natural products have shown great progress toward AD prevention and treatment (11, 12). These studies also shed new light on the discovery and development of useful agents from established traditional medicines and safe natural products to halt the progression of neurodegenerative disease.

Rutin (quercetin-3-rutinoside) has long been investigated as a natural flavonoid with different biological functions under pathological circumstances (13). Among various pharmacological properties, rutin exhibits antioxidative and anti-inflammatory activities in diabetes, obesity, and AD (14, 15). Widely existing in nature (including plants, fruits, and vegetables), rutin is consumed in the daily diet with no side effects (15). However, despite various biological activities, poor solubility in aqueous media (approximately 0.125 g/liter at room temperature) has largely limited its usage owing to its poor bioavailability (16). Recently, various strategies have been developed to improve the solubility and bioavailability of compounds that are poorly soluble in water (17, 18), among which salt formation is the most common and most effective method to increase the solubility and dissolution rates of drugs (16, 19). In the present study, we processed rutin into sodium salt [hereafter called sodium rutin (NaR)], which is highly water soluble and bioavailable. Then, we evaluated the effects of NaR on two mouse models of AD, APP (amyloid precursor protein)/PS1 (presenilin 1) and 5XFAD mice, and we found that NaR treatment could substantially rescue learning and memory defects in both AD models. In addition, our results showed that NaR treatment attenuated synaptic impairment, neuroinflammation, and Aβ burden in AD brains. Furthermore, we observed that the microglial cells surrounding Aβ plaques are significantly increased upon NaR treatment in the AD mouse model. We found that NaR increased the expression levels of phagocytic receptors and, specifically, NaR treatment increased the expression and recycling of TREM2. The phagocytic dysfunction of Trem2 deletion or other receptors’ individual knockdown microglia could eventually be rescued by NaR treatment.

Microglial phagocytosis needs a large amount of energy, and Trem2 deficiency leads to a significant decrease of adenosine 5´-triphosphate (ATP) production (20, 21). In addition, metabolic dysfunction has been shown to be involved in the etiopathogenesis of sporadic AD (22). However, it is largely unknown how the microglial metabolic state is affected by AD or how AD development is influenced by microglial metabolism. Actually, a metabolic switch from mitochondrial oxidative phosphorylation (OXPHOS) to anaerobic glycolysis has been described in the activated microglia in AD and in M1 macrophages activated by bacteria, lipopolysaccharide (LPS), or interferon-γ (21, 23, 24). In the present study, we observed that NaR enhances mitochondrial OXPHOS to provide sufficient energy (ATP) to microglia for Aβ clearance. Moreover, the enhancement of microglial anaerobic glycolysis and the suppression of mitochondrial OXPHOS by LPS stimulation could be rescued by NaR treatment, suggesting that NaR could regulate metabolic dysfunction in microglia under pathological conditions.

Together, our results demonstrate that NaR increases microglia-mediated Aβ clearance and alleviates amyloid pathology, with an implication of a promising drug candidate for AD treatment.

RESULTS

Salt formation of rutin improves its water solubility and bioavailability

Previous studies have shown the various biological effects of rutin (the structure shown in Fig. 1A), but poor water solubility greatly limited its usage and biological activity. In the present study, we processed rutin into NaR using alkaline aqueous solution (Fig. 1B), which greatly improved its water solubility (by at least 200-fold). Improvement of water solubility caused a marked increase of blood absorption in mice after oral administration of NaR in comparison to regular rutin (fig. S1A). NaR signal was detected in the brain homogenates (fig. S1B), indicating that NaR could penetrate the blood-brain barrier (BBB). To compare the biological effects of NaR and rutin, three dosages (0.05, 0.25, and 1 mg/ml) of each drug were orally administered in 5XFAD mice (schematic drug treatment shown in fig. S2A) and assessed by the attenuation of Aβ burden. Compared with rutin, NaR significantly reduced Aβ deposits even at a low concentration (0.05 and 0.25 mg/ml) (fig. S2, B and C). Although rutin could ameliorate Aβ burden at a dose of 1 mg/ml, the effect was much lower than that of an equal amount of NaR (fig. S2, B and C). This phenomenon might be explained by the lower absorption and BBB penetration of rutin in mice, whereas NaR overcame these problems.

Fig. 1 Preparation and characterization of NaR.

(A) Molecular structure of rutin. (B) Preparation procedures of NaR; the final concentration of NaR stock solution was 10 mg/ml. (C) HPLC analysis of NaR (0.1 mg/ml in water; injection volume was 20 μl) and rutin [0.1 mg/ml in dimethyl sulfoxide (DMSO); injection volume was 20 μl]. mAU, milli–absorbance units. (D) TLC assay of NaR and rutin. NaR(S): Solid NaR was dissolved with double-distilled water (ddH2O) at 0.5 mg/ml and 20 μl was injected. NaR(L): NaR stock solution was diluted with ddH2O at 0.5 mg/ml and 20 μl was injected. Rutin was dissolved with DMSO at 0.5 mg/ml and 20 μl was injected. (E) 13C and 1H shift changes of NaR compared with rutin and significant heteronuclear multiple bond coherence correlations of NaR.

To further identify the physiochemical properties of NaR, high-performance liquid chromatography (HPLC), thin-layer chromatography (TLC), and nuclear magnetic resonance (NMR) analyses were performed. According to the result of the HPLC assay, we found that the retention time of NaR was ahead of the rutin standard (Fig. 1C) under the same mobile phase conditions, indicating higher polarity of NaR. Moreover, the TLC assay also confirmed a higher polarity of NaR (the lower spots shown in Fig. 1D). Meanwhile, the rutin spot was also visualized in the sample of liquid and resolved solid NaR (the top spots shown in Fig. 1D), which led us to speculate about the existence of a chemical transformation between NaR and rutin under the acidic mobile phase. This assumption was confirmed by the NMR assay, which showed only a set of 1H and 13C signals in the sample of NaR. On the basis of the 1H and 13C chemical shift changes of NaR in contrast with rutin, the replacement of Na+ was speculated to occur at C7-OH, which was in agreement with the marked deshielded effects that happened at C-7, C-6, and C-8 and at shielded H-6 and H-8 (Fig. 1E). Together, we provide a method to produce a sodium salt of rutin (NaR), which maintains the integrated structure of rutin but markedly improves its water solubility and bioavailability.

NaR ameliorates learning and memory deficits and rescues synaptic impairment in mouse models of AD

To further investigate the biological effects of NaR, two mouse models of AD, APP/PS1 and 5XFAD mice, were used. NaR was orally administered to 6- to 13-month-old APP/PS1 mice (Fig. 2A) and to 4- to 8-month-old 5XFAD mice (fig. S3A). First, we examined the spatial learning and memory of mice through a Morris water maze. Compared with wild-type (WT) mice, AD mice showed significant spatial learning and memory deficits as reflected by longer escape latency time during the training trails, a lower crossing number on the platform, and less time spent in the target quadrant in the probe trail (Fig. 2, B to F, and fig. S3, B to E). However, NaR treatment significantly reversed these defects in AD mice (Fig. 2, B to F, and fig. S3, B to E). Alleviation of these phenotypes upon NaR treatment was not affected by the swimming speed of mice (Fig. 2E). Learning and memory deficits are usually caused by the impairment of synaptic plasticity, which is accompanied by the progression of AD (25). According to Golgi staining results, we found that NaR significantly reduced synaptic loss in APP/PS1 mice (Fig. 2, G and H). In addition, impairment of long-term potentiation (LTP) induced by theta-burst stimulation in 5XFAD was also rescued by NaR treatment (fig. S3, F and G). Together, these results suggest that NaR treatment could reverse synaptic dysfunction and ameliorate spatial learning and memory deficits in AD mice.

Fig. 2 NaR ameliorates learning and memory deficits and reduces synaptic loss in APP/PS1 mice.

(A) Timeline of NaR treatment in APP/PS1 mice. (B) Escape latency to the platform during the training trails in a Morris water maze. (C) Target (platform) entries in the probe test. (D) Time spent in target quadrant in the probe test. (E) Mean swimming speed of mice. (F) Representative track images of mice in the probe test. (G) Representative images of dendritic spines in CA1 hippocampal neurons. Scale bar, 5 μm. (H) Quantification of different types of spine density, including thin, stubby, mushroom, and total (n = 22 to 31 from three mice per group). Data are means ± SEM. *P < 0.05 and **P < 0.01, two-way (B) or one-way (C to E and H) analysis of variance (ANOVA), followed by Tukey’s multiple comparisons test. N.S., not significant.

NaR alleviates Aβ burden without altering APP processing

We then asked whether NaR exerts beneficial effects on alleviation of Aβ pathology, one of the most important hallmarks of AD. To examine the amyloid burden in APP/PS1 mice, the brain sections were immunostained with an anti-Aβ antibody, and the amount of Aβ plaques was quantified. Compared with APP/PS1 control mice, the mice treated with NaR showed a remarkable decrease of Aβ deposition in brains (Fig. 3, A and B). To further analyze which Aβ fractions were affected by NaR, the soluble and insoluble Aβ forms were extracted from the prefrontal cortex (PFC), followed by Western blot analysis. We found that there was no significant difference in the soluble Aβ fraction between control and NaR-treated APP/PS1 mice, while the insoluble Aβ fraction level was markedly reduced by NaR treatment (Fig. 3, C and D). In addition, the levels of Aβ1-40 and Aβ1-42 were also markedly decreased in SDS and formic acid (FA) fractions, but no significant change was observed in TBS fraction upon NaR treatment (fig. S3H). These results indicate that NaR might only reduce Aβ deposition but not Aβ production. To test this hypothesis, we further examined a series of key factors that were involved in Aβ production. Compared with APP/PS1 control mice, there was no significant change in the expression levels of APP [APP full length (APPfl)], soluble APPα (sAPPα), APP-CTFs (C-terminal fragments) (CTFα and CTFβ), and APP processing secretases, including Beta-site-APP Cleaving Enzyme (BACE) and γ-secretase complex, nicastrin, presenilin enhancer 2 (PEN2), and PS2, between control and NaR-treated APP/PS1 mice (Fig. 3, E and F). Together, these findings demonstrate that NaR treatment alleviates Aβ burden without altering APP expression and processing in APP/PS1 mice.

Fig. 3 NaR reduces Aβ deposition but does not alter APP processing.

(A) Representative images of Aβ (6E10) staining in the PFC and hippocampal DG region. (B) Quantification of Aβ plaques in the PFC (n = 13 to 14 slices from three mice per group) and hippocampal DG region (n = 14 to 18 slices from three mice per group). ND, not determined. (C) The amount of soluble and insoluble Aβ fractions extracted from PFC was examined by Western blot analysis. (D) Quantification of the lanes’ intensity with ImageJ software (n = 6 mice per group). (E and F) Expression levels of APPfl, sAPPα, CTFα, CTFβ, BACE, nicastrin, PEN2, and PS2 in PFC were examined by Western blot and quantified with ImageJ software (n = 6 mice per group). Data are means ± SEM. **P < 0.01 and ***P < 0.001, one-way ANOVA, followed by Tukey’s multiple comparisons test.

NaR decreases neuroinflammation in AD mice

Chronic neuroinflammation accompanied by abnormal activation of astrocytes and microglia has usually been observed in patients with AD and AD mouse models (26, 27). On the basis of the properties of rutin, we next investigated whether NaR treatment can decrease neuroinflammation in AD mice. Brain sections were immunostained with glial fibrillary acidic protein (GFAP) antibody, a specific marker of astrocyte, followed by quantification. As expected, higher astrocytic activation was observed in APP/PS1 mice compared with WT mice, especially in the PFC and hippocampal dentate gyrus (DG) region (fig. S4, A and B). In contrast, astrocytic activation was markedly suppressed by NaR treatment in APP/PS1 mice (fig. S4, A and B). Further Western blot analysis confirmed that the expression levels of GFAP were significantly decreased in APP/PS1 mice treated with NaR (fig. S4C). At the same time, the brain sections were immunostained with ionized calcium binding adaptor molecule 1 (Iba1) antibody, a specific marker of microglia in the brain, to examine the extent of microgliosis. Overactive microglia were observed in APP/PS1 mice, which were significantly alleviated by NaR treatment (fig. S4, D and E). Additional Western blot analysis showed that Iba1 protein level was reduced upon NaR treatment in APP/PS1 mice compared with APP/PS1 control mice (fig. S4F). In addition, the proinflammatory cytokines, including interleukin-1β (IL-1β), IL-6, and tumor nescrosis factor–α (TNF-α), were significantly reduced (fig. S4G), while the anti-inflammatory cytokines, including IL-4 and IL-10 were up-regulated upon NaR treatment in AD mice (fig. S4H). These data demonstrate that NaR treatment significantly decreases chronic neuroinflammation in AD mice.

NaR enhances microglial recruitment around the plaques and promotes Aβ phagocytosis and clearance

Aβ accumulation and aggregation are attributed to the imbalance of Aβ production and clearance (6). Given the significant reduction of Aβ deposits without unchanged Aβ production in AD mice upon NaR treatment, we next sought to determine whether NaR accelerated Aβ clearance. In the brain, extracellular Aβ is mainly phagocytosed and cleared by microglia. We then performed immunochemistry analysis using anti-Iba1 and anti-Aβ antibodies to colabel microglia and Aβ plaques. Confocal images showed that microglia were recruited adjacent to the plaques in APP/PS1 mice (Fig. 4A). NaR treatment markedly enhanced the recruitment of microglia surrounding Aβ plaques (Fig. 4, A and B). Quantification of the microglia around Aβ plaques also demonstrated the increased clustering of microglia upon NaR treatment in APP/PS1 mice (Fig. 4C). Furthermore, we examined whether NaR influenced microglial proliferation. Surprisingly, we found that NaR significantly reduced microglial proliferation (5-bromo-2´-deoxyuridine and Iba1 double-positive cells) in APP/PS1 mice, including plaque-associated and non–plaque-associated microglia (fig. S5A). However, NaR treatment significantly promoted the microglial migration (fig. S5B).

Fig. 4 NaR enhances microglial phagocytosis of Aβ in vivo and in vitro.

(A) Representative images of Aβ (6E10) plaques and microglia (Iba1) costaining in the cortex from APP/PS1 and NaR-treated APP/PS1 mice. (B) Quantification of microglial cells within 20 μm of the plaque surface (n = 107 to 144 plaques per group). (C) Plaques were divided into small, medium, and large according to their size, and the number of microglia per plaque was quantified. (D) Representative images of Aβ (6E10) plaques, microglia (Iba1), and phagosome (CD68) costaining in the cortex from APP/PS1 and NaR-treated APP/PS1 mice. (E and F) Quantification of the percentage of phagosome area and internalized Aβ using ImageJ Pro Plus software (n = 10 per group). (G) Confocal analysis of microglial phagocytosis of FITC-Aβ42 in the presence or absence of NaR after uptake for 4 hours in cultured primary microglia; CytoD served as a negative control. (H) Quantification of internalized FITC-Aβ42 using ImageJ software (n = 48 to 60 per group). a.u., arbitrary units. (I) FITC-Aβ42 uptake index in the presence or absence of NaR after uptake for the indicated time in cultured primary microglia (n = 6 per group). Data are means ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001, Student’s t test.

To elucidate whether the enhanced microglia recruitment around plaques could facilitate Aβ engulfment and clearance, CD68, Aβ, and Iba1 antibodies were used for coimmunostaining (28, 29), and we quantified the CD68+ microglial phagosomes and internalized Aβ. Compared to a lower but diffuse distribution of CD68 phagosomes with lower internalized Aβ in APP/PS1 control mice, NaR treatment markedly increased CD68 expression and microglial Aβ engulfment (Fig. 4, D to F). Together, these findings suggest that NaR treatment enhances microglial Aβ phagocytosis.

To further demonstrate that NaR could enhance microglial Aβ phagocytosis, microglia were isolated from mouse brain for primary culture, followed by FITC (fluorescein isothiocyanate)–Aβ uptake assays. Consistently, NaR treatment significantly increased the capacity and efficiency of FITC-Aβ uptake by microglia (Fig. 4, G to I and fig. S6, A and B). As a negative control, cytochalasin D (CytoD), an inhibitor of actin filaments, almost abolished the phagocytic effects of microglia (Fig. 4, G and H). Together, these results demonstrate that NaR treatment enhances microglial recruitment adjacent to the plaques and improves microglial phagocytic capacity, thus promoting Aβ clearance in AD mice.

NaR increases the expression and recycling of microglial phagocytic receptors

To explore how NaR enhanced microglial phagocytic capacity, we extracted total RNA from PFC and performed quantitative real-time polymerase chain reaction (qRT-PCR) to examine the mRNA expression levels of putative microglial phagocytic receptors, including Trem2, complement receptor 3 (CR3), G protein–coupled receptor 34 (GPR34), Mer receptor tyrosine kinase (MerTK), and pyrimidinergic receptor P2Y6 (P2Y6). We found that there was almost no significant difference in the mRNA levels of these receptors if normalized by β-actin in control and NaR-treated APP/PS1 mice (fig. S6C). We noticed that NaR treatment decreased the microgliosis in APP/PS1 mice (fig. S4, D to F), and there was a significant decrease of Iba1 mRNA level in NaR-treated APP/PS1 mice compared with APP/PS1 control mice (fig. S6D). Since the phagocytic receptors are mainly expressed in microglia, we normalized the expression levels of these receptors to Iba1 and found that all of the receptor genes markedly increased in the NaR treatment group (fig. S6E). To further validate these results, we isolated CD45lowCD11b+ cells as microglia from control or NaR-treated APP/PS1 mice by using fluorescence-activated cell sorting (FACS) (Fig. 5A) and measured the expression levels of phagocytic receptors through qPCR and Western blot analysis. The microglial phagocytic receptors were significantly up-regulated in both mRNA and protein levels upon NaR treatment (Fig. 5B and fig. S6F). Accordingly, NaR treatment significantly increased TREM2 immunofluorescent intensity (Fig. 5, C and D) and the recycling of TREM2 in microglial cells (Fig. 5, E and F), which might functionally facilitate microglial phagocytosis (30). Together, NaR treatment increased the expression and recycling of phagocytic receptors, which might be functionally involved in the NaR-mediated enhancement of Aβ phagocytosis and clearance.

Fig. 5 NaR treatment increases the expression and recycling of phagocytic receptors.

(A) Representative FACS plots and gating strategy to isolate microglia from the brain of adult mice. PE, phycoerythrin. SSC-A, Side-scattered-area; FSC-A, Forward-scattered-area. (B) Relative mRNA levels of phagocytic receptors in sorted microglia from APP/PS1 mice treated with or without NaR (n = 4, 15-month-old mice were treated with or without NaR starting from 13 months of age). (C) Representative images of Aβ (6E10) plaques, microglia (Iba1), and Trem2 costaining in the cortex from APP/PS1 and NaR-treated APP/PS1 mice. (D) Quantification of Trem2 expression level in microglia (APP/PS1, n = 55; APP/PS1 + NaR, n = 63). (E and F) Trem2 recycling assay of microglial BV2 cells in the presence or absence of NaR (n = 13 to 14 per group from three independent experiments). (G) FITC-Aβ42 uptake index after uptake for the indicated time in WT and Trem2 KO cultured primary microglia treated with or without NaR (n = 4 per group). (H) Representative images of thioflavin S staining in the PFC and hippocampal DG region. (I) Quantification of the numbers and areas of Aβ plaques (n = 9 to 12 brain slices from three mice per group). Data are means ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001, Student’s t test (B, D, and F) or two-way (G) or one-way (I) ANOVA, followed by Tukey’s multiple comparisons test.

To further investigate whether these phagocytic receptors are required for NaR-mediated enhancement of Aβ phagocytosis, we examined the microglial phagocytic capacity of primary microglia from WT or Trem2 knockout (KO) mice. As expected, NaR treatment remarkably enhanced Aβ uptake in WT microglia, while Trem2 deletion almost abolished the effect of NaR treatment at an early uptake time point (3 hours) (Fig. 5G). However, longer treatment (24 hours) of NaR could still enhance the Aβ uptake capability of Trem2 KO microglia (Fig. 5G). Similarly, the individual knockdown of Trem2, CR3, and MerTK impaired the NaR-increased microglia phagocytic ability at an early time point (4 hours), but long-term (24 hours) NaR treatment still significantly increased microglial phagocytosis in vitro (fig. S6, G and H). The Aβ pathology in Trem2-deficient AD mice was ameliorated by NaR treatment (Fig. 5, H and I). Together, these data suggest that microglial phagocytic receptors are dispensable for the NaR-mediated enhancement of Aβ phagocytosis and clearance.

NaR enhances energetic metabolism by promoting mitochondrial OXPHOS in microglia

Recently, it has been shown that Trem2 maintains microglial metabolic fit that is critical for Aβ clearance (21), and we found that NaR could ameliorate Aβ pathology in Trem2-deficient AD mice. These findings led us to ask whether NaR regulates energetic metabolism of microglia. Since ATP is important for microglial phagocytosis and Aβ clearance (20), we first observed that the ATP production significantly increased in primary cultured microglia under Aβ treatment (Fig. 6A). The ATP level of microglia is further elevated upon NaR treatment, even under the condition without Aβ stress (Fig. 6A), indicating that NaR could enhance energetic metabolism in microglia. It is well known that glycolysis and tricarboxylic acid (TCA) cycle coupling with OXPHOS are the two major pathways for ATP production to supply energy for cellular function (Fig. 6B). To investigate how NaR enhances microglial energetic metabolism, Seahorse extracellular flux assay was performed in microglia. Extracellular acidification rate (ECAR) reflects the glycolytic flux, and oxygen consumption rate (OCR) reflects mitochondrial oxidative respiration. We found that ECAR was not affected (Fig. 6, C and D), but OCR was significantly increased by NaR treatment (Fig. 6, E and F), suggesting that the increased microglial energetic metabolism by NaR treatment was due to the enhancement of mitochondrial OXPHOS but not glycolysis. In addition, we observed that the synthesized biotin-rutin was mainly localized to mitochondria (fig. S7, A and B), indicating that NaR might directly function on mitochondria.

Fig. 6 NaR enhances energetic metabolism by promoting mitochondrial OXPHOS in microglia.

(A) Total ATP production of primary microglia under different conditions as indicated. Aβ and/or NaR was added to the cells for 24 hours before measurement (n = 6 from three independent experiments). (B) Schematic diagram of the glycolytic pathway and the mitochondrial TCA cycle, two major pathways of ATP production. (C and D) ECAR measurements of primary cultured microglia in the presence or absence of NaR treatment for 24 hours. Oligomycin (Oligo), an inhibitor of ATP synthase; 2-deoxyglucose (2-DG), a glucose analog. (E and F) OCR measurements of primary cultured microglia in the presence or absence of NaR treatment for 24 hours. p-Trifluoromethoxy carbonyl cyanide phenylhydrazone (FCCP), the reversible inhibitor of OXPHOS; Rote/AA, the mitochondrial complex I and complex III inhibitor. (G) Total ATP production of primary microglia under different conditions as indicated. Galactose (10 μM)–replaced glucose medium was used to incubate cells for 24 hours, and Rote/AA (0.5 μM) was added to the cells (glucose medium) 2 hours before measurement (n = 5). (H) FITC-Aβ42 uptake index after uptake for 4 hours under different cultured conditions as indicated. Galactose-replaced glucose medium was preincubated for 24 hours, and Rote/AA was preadded to the cells (glucose medium) 2 hours before adding FITC-Aβ42 (n = 5). (I and J) The values of ECAR and basal OCR of activated primary microglia induced by LPS (1 μg/ml) or IL-4 (1 μg/ml) with or without NaR treatment. Data are means ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001, Student’s t test (D and F) or one-way ANOVA, followed by Tukey’s multiple comparisons test (A, C, and G to J).

To reveal the relationship between mitochondrial OXPHOS and microglia Aβ phagocytosis, rotenone/antimycin A (Rote/AA) or galactose was used to inhibit or promote OXPHOS (Fig. 6G) (31), respectively. The results showed that Rote/AA markedly reduced microglial Aβ phagocytosis, and galactose pretreatment significantly enhanced microglial Aβ uptake (Fig. 6H), suggesting that mitochondrial OXPHOS is critical for microglial phagocytosis of Aβ. In addition, we found that NaR treatment rescued the energetic metabolism deficit in Trem2-deficient microglia (fig. S8A) by enhancing mitochondrial OXPHOS (fig. S8, B and C), which may explain that long-term NaR treatment could rescue phagocytic dysfunction in Trem2-deficient microglia. To our surprise, NaR failed to alter mitochondrial OXPHOS in neurons (fig. S9, A to D) and astrocytes (fig. S9, E to H), suggesting that NaR might specifically regulate mitochondrial OXPHOS in microglia.

To determine how NaR affects the microglial metabolism under pathological conditions, LPS or IL-4 was used to induce microglial polarization. Upon LPS stimulation, microglia displayed increased glycolysis with reduced mitochondrial OXPHOS (Fig. 6, I and J), which is similar to the activated microglia in AD (23). LPS-induced glycolysis was significantly suppressed by NaR treatment (Fig. 6I), but LPS-induced inhibition of mitochondrial OXPHOS was significantly rescued by NaR treatment (Fig. 6J). In contrast to LPS, IL-4 enhanced microglial OXPHOS but not glycolysis, and NaR treatment failed to further increase OXPHOS in IL-4–treated microglia (Fig. 6, I and J). Together, these data demonstrate that NaR enhances the energetic metabolism by improving mitochondrial OXPHOS. Therefore, we argue that NaR regulates a metabolic switch from anaerobic glycolysis to mitochondrial OXPHOS under pathological conditions, thus providing sufficient energy (ATP) through an efficient metabolic way for microglial Aβ clearance (fig. S10A).

DISCUSSION

In past decades, great efforts have been made to investigate the pathogenesis of AD, and amyloid hypothesis is supported by most of the laboratories and clinics worldwide (32). A number of drugs targeting Aβ clearance or production have been investigated to reduce Aβ level in animal models or patients with AD, but so far, none of them showed significant benefits on its clinical endpoints (32). However, failures do not mean the end of clinical trials or the fallacy of Aβ hypothesis. On the contrary, lessons from failures, such as time points, side effects, and the neuroinflammatory responses of these drugs, should be learned and more effective and safe drugs or strategies are urgently needed (33, 34). In the present study, we report a small molecule, NaR (a sodium salt of rutin), that could significantly alleviate learning and memory deficits in APP/PS1 and 5XFAD mice (Fig. 2 and fig. S3), indicating a promising drug candidate for AD treatment.

Rutin is a powerful phenolic antioxidant that has various pharmacological effects under several pathological circumstances. In a recent study, Xu et al. (14) reported that rutin ameliorates spatial memory defect in AD mice by reducing Aβ level and attenuating oxidative stress and neuroinflammation. However, this study has not shown the mechanism by which Aβ oligomer levels were reduced by rutin treatment. The poor water solubility and low bioavailability greatly limit the usage of rutin and other natural flavonoids (16). Therefore, strategies to improve the solubility and bioavailability of flavonoids could extend its application for clinical use, especially for central nervous system diseases. In our current study, rutin was converted to NaR, which gives it a high water solubility and bio-absorption in vivo (Fig. 1B and fig. S1A). NaR could penetrate the BBB in mice (fig. S1B), and NaR treatment markedly reduced Aβ deposits in the brain of AD mice (Fig. 3, A and B).

The accumulation and deposition of Aβ in the brain have been hypothesized to drive the pathogenic cascades of AD (3). Microglia, the resident immune cells of the brain, are the first responders to Aβ accumulation and phagocytosis (28). However, the phagocytic capacity of microglia decreases with age, resulting in an impaired Aβ clearance during AD progression (35). In particular, it has been reported that the impairment of Aβ clearance is a common prelude to late-onset AD (6). Here, we show that NaR treatment enhances microglial recruitment around Aβ plaques with increased CD68+ phagosome expression in AD mice (Fig. 4, A to F) and augments microglial Aβ uptake both in vivo and in vitro (Fig. 4), indicating that NaR treatment restores microglial phagocytic capacity and promotes Aβ clearance.

It is well known that phagocytosis controls brain homeostasis by removing the unwanted cellular debris and misfolded proteins (36). The impairment in different phagocytic receptors’ expression or trafficking would lead to neurodegeneration (30). Accordingly, up-regulation of these receptors would be beneficial for phagocytosis of aggregated protein such as Aβ and delays neuropathology (37, 38). In the present study, we found that NaR specifically enhances microglial Aβ phagocytosis by increasing the expression of microglial phagocytic receptors (Fig. 5B and fig. S6, C to F). In addition, we showed that NaR accelerates Trem2 recycling in the cultured microglial cells (Fig. 5, E and F). Microglia phagocytic dysfunction caused by Trem2 KO or individual knockdown of phagocytic receptors could be eventually rescued by NaR treatment (Fig. 5G and fig. S6 H), suggesting that microglial phagocytic receptors are dispensable for the NaR-mediated enhancement of Aβ phagocytosis and clearance. Temporal/synergistic interplays among these different receptors in the process of Aβ phagocytosis might exist, which needs to be further investigated.

Microglial phagocytosis requires dynamic reorganization of the cytoskeleton, for which a large amount of energy is demanded (20, 39). To quickly increase ATP production for Aβ phagocytosis, a metabolic switch from mitochondrial OXPHOS to anaerobic glycolysis in microglia, especially the cells surrounding Aβ plaques, was found in AD pathology (23). However, only two ATP molecules are obtained from each glucose molecule through anaerobic glycolysis (Fig. 6B). Metabolic switch to anaerobic glycolysis is always accompanied by the generation of metabolites such as lactate, which is toxic to cells (39). Therefore, microglial metabolic reprogramming might be a potential avenue for AD treatment. In our study, we observe that proinflammatory LPS stimulation increases microglial anaerobic glycolysis and inhibits mitochondrial OXPHOS, which could be rescued by NaR treatment (Fig. 6, I and J). We also observe that NaR treatment significantly decreased microglial M1-type inflammatory cytokines (including IL-1β, IL-6, and TNF-α) (fig. S4G) and increased the expression of microglial M2-type factors (including IL-4 and IL-10) in AD mice (fig. S4H). Therefore, we speculate that NaR treatment converts metabolic program from anaerobic glycolysis to mitochondrial OXPHOS and increases microglial M2 polarization, which explains that NaR treatment could reduce neuroinflammation and promote Aβ clearance in AD brain.

In summary, we provide a method to generate a salt formation of rutin (NaR) with high water solubility, bioavailability, and BBB penetration. NaR treatment significantly increases the expression and recycling of microglial phagocytosis-related receptors. NaR treatment promotes the metabolic switch from anaerobic glycolysis to mitochondrial OXPHOS, providing sufficient energy (ATP) through an efficient metabolic way (fig. S10A). NaR treatment promotes microglial recruitment to the plaques and enhances Aβ phagocytosis (fig. S10B), thus alleviating Aβ burden and the pathological repertoires, including neuroinflamamtion, synaptic loss, and plasticity impairment. As a result, NaR treatment reverses learning and memory deficits in AD mice, suggesting that NaR might be a promising drug candidate for AD treatment.

MATERIALS AND METHODS

Preparation and identification of NaR

Rutin (169050, J&K Scientific, Beijing, China) standard was of 98% purity. For NaR preparation, the rutin was dissolved in 0.1 M NaOH water solution to a final concentration of 10 mg/ml. The pH of NaR stock solution was adjusted to approximately 9.0 with 1 M HCl (low pH value would lead to rutin separating out). To identify NaR physicochemical properties, NaR was first purified using an MCI GEL CHP20P to remove the remaining NaOH in stock solution. The NaR attached on the MCI GEL was washed with double-distilled water (ddH2O) and eluted with methanol. After evaporation, dark brown solid NaR was obtained and used for further analysis.

HPLC analysis was performed with an Ultimate XB-C18 column (4.6 × 250 mm, 5 μm, Welch, Shanghai, China). The column temperature was set at 30°C, and the injection volume was 20 μl (0.1 mg/ml). The mobile phase was acetonitrile and 0.3% phosphoric acid (20:80, v/v), and the flow rate was 1 ml/min. Detection was performed at 254 nm.

TLC analysis was carried out on silica gel plates using ethyl acetate/methanol/FA/water (7:3:0.5:0.5, v/v) as mobile phase. The spots were visualized under an ultraviolet lamp (254 nm) with the help of sulfuric acid–vanillin.

For chemical structure identification of NaR, the 13C-NMR spectra were measured in CD3OD and recorded at 125 MHz. The 1H-NMR spectra were measured in CD3OD and recorded at 400/600 MHz. Chemical shifts were given in δ value.

13C-NMR (CD3OD, 125 MHz) of rutin: δ 179.4 (C-4), 166.0 (C-7), 163.0 (C-5), 159.3 (C-9), 158.5 (C-2), 149.8 (C-4′), 145.8 (C-3′), 135.6 (C-3), 123.6 (C-1′), 123.1 (C-6′), 117.7 (C-2′), 116.1 (C-5′), 105.6 (C-10), 104.7 (C-1″), 102.4 (C-1′′′), 100.0 (C-6), 94.9 (C-8), 78.2 (C-3″), 77.2 (C-5″), 75.7 (C-2″), 73.9 (C-4′′′), 72.2 (C-3′′′), 72.1 (C-2′′′), 71.4 (C-4″), 69.7 (C-5′′′), 68.6 (C-6″), and 17.9 (C-6′′′).

13C-NMR (CD3OD, 125 MHz) of NaR: δ 178.5 (C-4), 172.7 (C-7, invisible in the 13C broadband decoupling spectrum, assigned based on heteronuclear multiple bond coherence cross-peak), 162.5 (C-5), 159.0 (C-2), 158.5 (C-9), 150.6 (C-4′), 146.1 (C-3′), 135.4 (C-3), 123.5 (C-6′), 122.8 (C-1′), 117.3 (C-2′), 116.1 (C-5′), 105.6 (C-1″), 103.5 (C-10), 102.5 (C-6), 102.3 (C-1′′′), 96.5 (C-8), 78.3 (C-3″), 77.2 (C-5″), 75.7 (C-2″), 74.0 (C-4′′′), 72.2 (C-3′′′), 72.1 (C-2′′′), 71.4 (C-4″), 69.7 (C-5′′′), 68.7 (C-6″), and 17.9 (C-6′′′).

1H-NMR (CD3OD, 400 MHz) of rutin: δ 7.66 (1H, d, J = 1.9 Hz, H-2′), 7.62 (1H, d, J = 8.5, 1.9 Hz, H-6′), 6.86 (1H, d, J = 8.5 Hz, H-5′), 6.38 (1H, J = d, 1.9 Hz, H-8), 6.19 (1H, d, J = 1.9 Hz, H-6), 5.10 (1H, d, J = 7.5 Hz, H-1′′), 4.51 (1H, br s, H-1′′′), 3.82 to 3.24 (10H, m), and 1.12 (3H, d, J = 6.2 Hz, H-6′′′).

1H-NMR (CD3OD, 600 MHz) of NaR: δ 7.66 (1H, d, J = 1.9 Hz, H-2′), 7.63 (1H, d, J = 8.5, 1.9 Hz, H-6′), 6.84 (1H, d, J = 8.5 Hz, H-5′), 6.23 (1H, br s, H-8), 6.08 (1H, br s, H-6), 4.97 (1H, d, J = 7.8 Hz, H-1′′), 4.52 (1H, br s, H-1′′′), 3.82 to 3.24 (10H, m), and 1.15 (3H, d, J = 6.2 Hz, H-6′′′).

Animals

The APP/PS1 (APPswe/PSEN1dE9) double-transgenic mice (40) were obtained from the Model Animal Research Center of Nanjing University (Nanjing, China). The 5XFAD mice overexpressing the K670 N/M671 L (Swedish), I716V (Florida), and V717I (London) mutations in human APP (695), as well as M146 L and L286 V mutations in human PS1 (41), were provided by C. Zhang (Peking University, Beijing, China). Trem2 conventional KO mice were generated by deleting 7 base pairs (AAGCGGA) of exon2 from 384 to 390 via the CRISPR-Cas9 system. Genotypes were confirmed by PCR analysis of tail biopsy specimens. Mice of mixed genotypes were housed four to five per cage with a 12-hour light/12-hour dark cycle and food and water ad libitum. Male mice (C56BL/6J background) were used within all experiments to avoid the influence of gender. All experimental animal procedures were approved by the Institutional Animal Care and Use Committees of the Beijing Institute of Basic Medical Sciences.

Drug treatment

For in vivo NaR treatment, 6-month-old APP/PS1 mice or 4-month-old 5XFAD mice were fed with NaR-containing water, and the treatment continued until mice were euthanized (APP/PS1 mice were 13 months and 5XFAD mice were 8 months, except those specific explanation). NaR was administered in drinking water at a final concentration of 0.25 mg/ml. The desired intake of NaR was approximately 18 to 25 mg/kg of body weight per day, which was similar with the troxerutin used in humans of about 18 mg/kg of body weight per day. For in vitro treatment, NaR stock solution was filtered with a 0.22-μm membrane and diluted in sterile phosphate-buffered saline (PBS) to the designed concentration.

Morris water maze

The effect of NaR on spatial learning and memory performance of mice was tested by the Morris water maze. Briefly, the mice were allowed to habituate the water maze (110 cm in diameter) 1 day before the experiment. The maze filled with opacified water was drained every day, and the temperature of the water was maintained at 19° to 22°C. During the training period, the mice were allowed to freely swim for 60 s to find the platform (10 cm in diameter), which was fixed 1 cm beneath the water surface. Mice that failed to find the platform were guided to it and allowed to stay for 30 s. The mice were trained twice per day, and the data presented are the average of the two trials. Twenty-four hours after the last training trail, the platform was removed and the mice were tested for memory retention in a probe trial. The swimming activity of each mouse was monitored using a video camera mounted overhead and was automatically recorded via ANY-maze behavioral tracking software (Stoelting, Wood Dale, IL, USA).

Electrophysiology

Electrophysiology was performed as described previously (42). Briefly, brains from 6-month-old mice (mice were treated with or without NaR from 4 months old) were rapidly removed and placed in ice-cold oxygenated dissection buffer (213 mM sucrose, 10 mM glucose, 3 mM KCl, 1.0 mM NaH2PO4, 0.5 mM CaCl2, 10 mM MgCl2, and 25 mM NaHCO3). Brain slices were cut into 350 μm using a vibratome (Leica VT1200S, Leica, Solms, Germany) and then transferred to the incubation chamber containing artificial cerebrospinal fluid (ACSF; 10 mM glucose,125 mM NaCl, 5 mM KCl, 1.2 mM NaH2PO4, 2.6 mM CaCl2, 1.3 mM MgCl2, and 26 mM NaHCO3). The slices were incubated at 30°C for 20 min and at room temperature for 40 min before transferring to the recording chamber. The ACSF was perfused at 1 ml/min. The slices were visualized with a ×40 water immersion lens, differential interference contrast optics, and a charge-coupled device camera. Patch pipettes were pulled from borosilicate glass capillary tubes using a PC-10 pipette puller. For recording, stimulation pluses were delivered using a concentric bipolar electrode, and the excitatory postsynaptic potentials were recorded by a glass pipette (4 to 7 MΩ) filled with ACSF. The LTP was induced by a theta-burst stimulation protocol after recording a 20-min baseline.

Golgi staining

Golgi staining was performed using the manufacturer’s protocols (FD Rapid GolgiStain Kit, FD NeuroTechnologies, Columbia, MD, USA), and the density of spines was quantified using NeuronStudio software as described previously (42).

Immunohistochemistry

The coronal sections (40 μm) of mouse brains were cut using a cryostat (Leica CM3050S) and stored at −20°C in cryoprotective storage solution [300 g of sucrose, 10 g of polyvinylpyrrolidone, 500 ml of phosphate buffer (0.1 M), 300 ml of ethylene glycol, and up to 1000 ml of ddH2O] until use. Upon use, the sections were washed three times with PBS and incubated with 3% H2O2 for 15 min to inhibit endogenous peroxidases. After three washes with PBS, the sections were blocked for 1.5 hours with blocking buffer (0.3% Triton X-100 + 5% goat serum + 2% bovine serum albumin in PBS) at room temperature, followed by primary antibodies overnight at 4°C. Primary antibodies used include mouse monoclonal anti-6E10 (SIG-39320, Covance, Princeton, NJ, USA), rabbit polyclonal anti-Iba1 (019-19741, Wako, Richmond, VA, USA), mouse monoclonal anti-GFAP (MAB360, Millipore, Darmstadt, Germany), rat monoclonal anti-CD68 (ab53444, Abcam, Cambridge, MA, USA), rat monoclonal anti-Trem2 (#237916, R&D Systems, Minneapolis, MN, USA), and rabbit polyclonal anti-Tom20 (Sc-11415, Santa Cruz Biotechnology, Santa Cruz, CA, USA).

For immunohistochemical staining, the signal was visualized with biotinylated immunoglobulin G (IgG), followed by streptavidin-conjugated horseradish peroxidase (VECTASTAIN ABC Kit, Zhongshan Jinqiao, Beijing, China), and finally reacted with hydrogen peroxide (DAB Kit, Zhongshan Jinqiao). Images were captured using a Leica SCN400 scanner. The GFAP- and Iba1-positive cells and the number of Aβ plaques in the PFC and hippocampal DG region were counted blindly using ImageJ Pro Plus software.

For immunofluorescent costaining, the sections were labeled with fluorescent secondary antibodies as follows: Alexa Fluor 488 AffiniPure donkey anti-rat IgG (712-545-150, Jackson ImmunoResearch, West Grove, PA, USA), TRITC AffiniPure goat anti-rabbit IgG (111-025-003, Jackson ImmunoResearch), Cy5-conjugated donkey anti-mouse IgG (715-175-150, Jackson ImmunoResearch), Alexa Fluor 594 AffiniPure goat anti-mouse IgG (115-585-003, Jackson ImmunoResearch), and FITC AffiniPure goat anti-rabbit IgG (111–095-003, Jackson ImmunoResearch). The images were acquired using a Nikon confocal microscope (Nikon, Melville, NY, USA).

Thioflavin S staining

Thioflavin S staining was used to label the Aβ plaques. Briefly, brain sections were stained with 0.002% TS (catalog no. T1892-25G, Sigma-Aldrich) in the dark for 8 min in 50% ethanol followed by two washes with 50% ethanol and three washes with PBS. Afterward, sections were mounted for imaging or processed for further immunofluorescent costaining with other antibodies.

Confocal analyses

Images were acquired as a Z-series stack using a Nikon confocal microscope. For Aβ plaque–associated microglia quantification, Iba1-positive cells within a 20-μm range of the plaques were manually counted. Plaques were grouped into small (<250 μm2), medium (250 to 600 μm2), and large (>600 μm2) sizes. At least 100 plaques per group and over 30 plaques of each size were analyzed. Microglial CD68+ phagosome area was calculated as the Iba1 and CD68 double-positive area divided by the Iba1-positive area. For quantification of the Aβ internalization ratio, the area of Aβ-internalized microglia (Aβ+Iba1+) was normalized to the Aβ-positive area. The calculating performance was conducted using ImageJ Pro Plus software. All the images were captured and analyzed blindly using coded slides.

Primary microglia culture and transfection

Postnatal 0 to 3 day WT or Trem2−/− mice were used to prepare primary microglia. The procedure has been previously described (43). After 14 days of culture, primary microglia were collected and used for experiments. For small interfering RNA (siRNA) transfection, primary microglia were seeded overnight and then transfected with 50 nM siRNA using Lipofectamine RNAiMAX (Invitrogen, Waltham, MA, USA). Aβ phagocytosis assays were performed at 48 to 72 hours after transfection. siRNA sequences used are listed as follows: (i) Trem2 #1: uucuccugagcaaguuucuug, Trem2 #2: caucacucugaagaaccucca; (ii) CR3 #1: aucuuuccugcuaauucugag, CR3 #2: ccugcuaauucugaggucaca; (iii) GPR34 #1: auaaccaccaagcaaaguauu, GPR34 2#: cucauggaaugcacuuuauaa; (iv) MerTK #1: guuuaaucacaccauuggaca, MerTK #2: agauguugugauugacagaaa.

Phagocytosis assays

The phagocytosis of aggregated Aβ1-42 was analyzed similarly to a previously described method (44). Briefly, FITC-Aβ1-42 (AS-60479, AnaSpec, Fremont, CA, USA) was aggregated for 24 hours at 37°C with agitation. Primary microglia were plated at 2 × 104 cells per well in poly-l-ornithine–coated black-walled 96-well plates (Thermo Fisher Scientific, Waltham, MA, USA) and cultured overnight. After NaR (20 μg/ml) treatment for 24 hours, Aβ was added to a final concentration of 1 μg/ml and incubated for an indicated time at 37°C. As a negative control, 10 mM CytoD (PHZ1063, Life Technologies, Frederick, MD, USA) was added 30 min before the addition of Aβ. Before measurement, Aβ-containing cultured medium was removed, and the cells were washed twice with prewarmed Dulbecco’s modified Eagle’s medium (DMEM)/F12 to remove extracellular Aβ. Last, fluorescence was measured at 485-nm excitation/538-nm emission using a Spark Control Magellan reader (Spark 10M, Tecan, Zurich, Switzerland).

In addition, microglial Aβ phagocytosis was verified by a confocal microscope and Western blot. Briefly, 1 × 105 cells were plated in poly-l-ornithine–coated 24-well plates. After Aβ uptake, the cells were processed for immunofluorescence and Western blot. For confocal measurement, anti-Iba1 antibody was used to label the cell shape, and 4′,6-diamidino-2-phenylindole (DAPI) was used to stain the nuclei. Images were acquired, and the intracellular Aβ was quantified using the Fiji software “Analyze particles” plugin. The values are represented as the average intensity of intracellular FITC-Aβ1-42 signal per cell and normalized to the vehicle group.

For Western blotting analysis, the cells were lysed with 1× loading buffer and represented for intracellular Aβ. Supernatants were collected and processed for Aβ extraction. Briefly, the supernatants and equal volumes of methanol were well mixed and then a 1:4 volume (supernatants/methanol) of trichloromethane was added, followed by mixing upside down. After centrifugation (14,000g, 4°C, 15 min), there would be three layers, and the interlayer protein was collected and left air-dried. The protein was lysed with 1× loading buffer and processed for Western blot. Supernatant Aβ indicated the remaining Aβ that was not uptaken by microglia.

Soluble and insoluble Aβ extraction

Brain tissue samples were stored at −80°C until use. Upon use, each sample was homogenized with 300 μl of lysis buffer [50 mM tris (pH 7.4), 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 1% sodium deoxycholate, and 0.1% SDS] containing protease inhibitors on ice, followed by centrifugation at 14,000g for 15 min. The supernatants were collected, and the protein concentration was measured using bicinchoninic acid assays and adjusted to the same final concentration and then processed for Western blotting to measure soluble Aβ. The pellet was resuspended with 300 μl of lysis buffer, followed by centrifugation at 14,000g for 15 min; this step was repeated three times to clean up the soluble Aβ. The final pellet was resuspended with 200 μl of lysis buffer and sonicated until the lysis buffer was clear. After denaturation, the lysates were processed for Western blotting to measure insoluble Aβ.

Enzyme-linked immunosorbent assay

Frozen PFC was homogenized in TBS [25 mM tris-HCl (pH 7.4), 1 mM EDTA, and 1 mM EGTA] containing protease inhibitor cocktail (B14001, Biotool) followed by centrifugation at 100,000g for 30 min at 4°C. The supernatant was collected as TBS fraction, and the pellet was solubilized in 2% SDS and 25 mM tris-HCl (pH 7.4) followed by centrifugation at 100,000g for 30 min at 4°C. The SDS-insoluble pellet was extracted with 70% FA in water. The concentrations of Aβ1-40 (DAB140B, R&D Systems, Minneapolis, MN, USA) and Aβ1-42 (DAB142, R&D Systems, Minneapolis, MN, USA) were measured using enzyme-linked immunosorbent assay (ELISA) kits according to the manufacturer’s instructions. In addition, the levels of IL-1β, IL-6, and TNF-α (BioLegend) in TBS fraction were also quantitatively measured by ELISA. The data were normalized to total protein.

Western blotting

Western blotting analysis was performed as described previously (45). The following antibodies were used: mouse monoclonal anti-6E10 (recognizes Aβ and sAPPα; SIG-39320, Covance), rabbit polyclonal anti-Iba1 (019-19741, Wako), rabbit polyclonal anti-GFAP (Z0334, Dako), rabbit polyclonal anti-APP (recognizes APPfl, CTFα, and CTFβ; #2452, Cell Signaling Technology, Cambridge, MA, USA), rabbit monoclonal anti-nicastrin (#5887, Cell Signaling Technology), rabbit monoclonal anti-PEN2 (#5887, Cell Signaling Technology), rabbit monoclonal anti-PS2 (#5887, Cell Signaling Technology), rabbit monoclonal anti-BACE (#5606, Cell Signaling Technology), sheep polyclonal anti-Trem2 (AF1729, R&D Systems, Minneapolis, MN, USA), mouse polyclonal anti-GPR34 (ab169455, Abcam, Cambridge, MA, USA), mouse monoclonal anti-MerTK (#9178S, Cell Signaling Technology), goat polyclonal anti-P2Y6 (Sc-15217, Santa Cruz Biotechnology, Santa Cruz, CA, USA), and mouse monoclonal anti–β-actin (66009-1-lg, Proteintech, Wuhan, China).

Isolation of microglia from adult mice

For adult mice microglia isolation, brains were quickly removed and washed with prechilled PBS. Then, the brains were cut into small pieces (2 to 3 mm3) and immersed in homogenization buffer [1× PBS containing 2% fetal bovine serum (FBS) and collagenase (1 mg/ml)], followed by incubation for 15 to 20 min at 37°C. A 10-ml syringe plug was used to grind the tissue, and samples were filtered through a 70-μm strainer. The homogenates were then pelleted at 600g for 6 min, resuspended in 2 ml of 37% isotonic Percoll, underlain with 3 ml of 70% isotonic Percoll, covered by 2 ml of 30% isotonic Percoll and 3 ml of 1× PBS, and centrifuged at 2000g for 20 min at 4°C with minimal acceleration and no deceleration. Following Percoll gradient centrifugation, cells at the 37 to 70% interphase were collected and washed with 1× PBS, followed by centrifugation at 600g for 6 min. The pellet was resuspended with staining buffer (1× PBS containing 2% FBS) and incubated with anti-CD11b Alexa Fluor 488 (clone M1/70, 101217, BioLegend) and anti-CD45 phycoerythrin/Cy5 (clone 30-F11, 103110, BioLegend) for 30 min on ice in the dark. Microglia were then purified with a BD Influx and defined as CD11b+ CD45Low single cells and processed for RNA/protein extraction.

Quantitative real-time polymerase chain reaction

RNA extraction, complementary DNA production, and SYBR Green–based qPCR were performed as described previously (42). The primer sequences used are listed in table S1. The mRNA expression levels were normalized to the β-actin or Iba1 as indicated.

Transwell migration assay

For microglial migration evaluation, Transwell assays were performed using 8-μm-pore-diameter inserts (MCEP24H48, Millipore, Darmstadt, Germany). Briefly, 2 × 105 primary microglial cells were plated in the upper chamber with 200 μl of serum-free medium, and the chamber was then placed within the bottom wells (24-well plate) containing 600 μl of normal medium supplemented with or without NaR (20 g/ml) or 5 μM Aβ as indicated, followed by incubation for 24 hours. The nonmigrating cells on the upper membrane were removed with a cotton swab, and the cells on the lower surface of the membrane were fixed with 4% paraformaldehyde for 15 to 30 min. After staining with 0.2% crystal violet (30 min at room temperature), images were randomly captured, and the number of migrating cells were manually quantified.

Trem2 recycling assay

For Trem2 recycling assay, 2 × 105 BV2 microglial cells were seeded on poly-l-ornithine–coated glass coverslips in 24-well plates and cultured overnight. The cultured medium was replaced by fresh medium containing NaR (20 μg/ml) or vehicle (PBS) and cultured for 24 hours, followed by Trem2 recycling assay as described previously (43). Briefly, antibodies against TREM2 (#237916, R&D Systems, Minneapolis, MN, USA) were added to the cultured cells in DMEM/F12 containing 1% FBS at 37°C for 1 hour. Cells were then washed three times with prechilled DMEM/F12 at pH 2.0 and then cultured in DMEM/F12 containing 10% FBS at 37°C for 1 hour. Then, the secondary antibodies coupled with Alexa Fluor 488 were added to the cells in DMEM/F12 containing 1% FBS and incubated at 37°C for another 1 hour, followed by acid washed with prechilled DMEM/F12 at pH 2.0 and PBS at pH 7.4. After fixing and staining with DAPI, cells on the coverslips were mounted for imaging. The values of recycled Trem2 were presented as the fluorescence intensity per cell using the Fiji software “Analyze particles” plugin and normalized to the vehicle group.

Measurement of ATP levels

For the primary microglial, neuronal, and astrocytic ATP measurement, cultured medium was removed and 100 μl of lysis buffer containing luciferase reagents (Promega, Madison, WI, USA) was added and incubated for 10 min; then, the luminescent signal was measured in a microplate reader, and ATP levels were normalized to the protein concentration.

Seahorse extracellular flux assay

The ECAR and OCR of primary microglia, neurons, and astrocytes were determined using a Seahorse XFe 96 Extracellular Flux Analyzer (Seahorse Bioscience). The ECAR and OCR were measured using the Seahorse XF Glycolysis Stress Test Kit (103020-100, Agilent Technologies) and the Seahorse XF Cell Mito Stress Test Kit (103015-100, Agilent Technologies), and the experimental procedures were performed according to the manufacturer’s instructions. Briefly, primary microglia or astrocytes were seeded into a Seahorse XF 96-well culture microplate (2 × 104 cells per well) overnight (primary cortical neurons were seeded at 1 × 104 cells per well and cultured for 6 days before use), treated with drugs for 24 hours, and then used for ECAR or OCR measurement. Upon measurement, cultured cells were washed twice and maintained in XF assay medium. After baseline measurements, glucose, oligomycin, and 2-deoxyglucose solution or oligomycin, p-trifluoromethoxy carbonyl cyanide phenylhydrazone (1 μM for neurons and astrocytes and 2 μM for microglia), and Rote/AA solution was sequentially injected into the wells of a utility plate at the indicated time points for ECAR or OCR analysis. Data were analyzed using Seahorse XF 96 Wave software, and the results were normalized to cell number and presented as mpH/min for ECAR or pmol/min for OCR.

Statistical analysis

All data were expressed as means ± SEM. All statistical analyses were performed using GraphPad Prism version 6.0 software. The significance of differences was assessed by unpaired Student’s t test or one-way or two-way analysis of variance (ANOVA) followed by Tukey’s multiple comparisons test as indicated. The significant threshold was set at P < 0.05.

SUPPLEMENTARY MATERIALS

Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/5/2/eaau6328/DC1

Fig. S1. Salt formation of rutin benefits absorption and crosses the BBB in mice.

Fig. S2. The effect of rutin or NaR on Aβ burden in 5XFAD mice.

Fig. S3. NaR ameliorates learning and memory deficits, rescues synaptic impairment, and reduces Aβ burden in 5XFAD mice.

Fig. S4. NaR treatment reduces neuroinflammation in AD mice.

Fig. S5. NaR reduces proliferation but increases migration of microglia.

Fig. S6. NaR increases the expression of phagocytic receptors and enhances microglial Aβ phagocytosis.

Fig. S7. Subcellular localization of biotin-rutin.

Fig. S8. NaR rescues energetic metabolism deficit in Trem2 KO microglia.

Fig. S9. NaR fails to alter mitochondrial OXPHOS in primary cultured neurons and astrocytes.

Fig. S10. Proposed working model of NaR.

Table S1. List of qPCR primers.

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.

REFERENCES AND NOTES

Acknowledgments: We thank C. Zhang (Peking University, Beijing, China) for providing us 5XFAD mice and Y. Tian (Institute of Biophysics, Chinese Academy of Sciences, Beijing, China) for providing us Trem2−/− mice. Funding: This work was supported by the National Nature Science Foundation of China (grant no. 81630026 to Z.Y. and grant no. 81400987 to J.C.), the Beijing Nature Science Foundation (7161009 to Z.Y.), and the Beijing Innovation Program (16CXZ028 to Z.Y.). Author contributions: R.-Y.P., J.M., and Z.Y. conceived and designed the study and drafted the manuscript. X.-X.K. performed the electrophysiological recording and analysis. X.-F.W. and C.-H.T. identified the chemical structure of NaR. Q.L. synthesized the biotin-rutin. All authors analyzed the data and critically revised the manuscript. Competing interests: Z.Y., J.C., and R.-Y.P. are inventors on two patents related to this work (application nos. 201810227319.6 and 201810227298.8, filed by the Chinese Patent Office on 20 March 2018). The authors declare no other competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.
View Abstract

Navigate This Article