Scutellarin Modulates the Microbiota-Gut-Brain Axis and Improves Cognitive Impairment in APP/PS1 Mice

Animals

Four-month-old male APPswe/PS1dE9 (APP/PS1) mice and C57BL/6J mice were purchased from Jackson Laboratory and bred in the animal room of the State Key Laboratory of Cognitive Neuroscience and Learning at Beijing Normal University. The APP/PS1 transgenic (Tg) mice express a chimeric mouse/human amyloid precursor protein (Mo/HuAPP695swe) and a mutant human presenilin 1 (PS1-dE9), both of which are controlled by independent mouse prion protein promoter elements [20]. After genotyping, all mice (37 males) were allocated to different cages and raised in a 12/12-h light-dark cycle with free access to food and water. The animal protocols were approved by the Animal Care and Use Committee of the State Key Laboratory of Cognitive Neuroscience and Learning at Beijing Normal University (Ref. No.: IACUC-BNU-NKLCNL-2013-10).

Drug preparation and treatment

The scutellarin (lot number: 20161003, purity ≥90%) used in this study was produced by Yunnan Bio Valley, China. According to our previous study, scutellarin was prepared at a low concentration (LS, 20 mg/kg) and a high concentration (HS, 100 mg/kg) [3]. The drugs were suspended in a 0.1% sodium carboxymethyl cellulose (CMC-Na, cat: 9004-32-4, Sigma-Aldrich) solution, which was used as a solvent control. In this study, the oral administration of the drugs occurred when the mice were 4 months old and continued until they developed cognitive deficits at 6 months of age [21]. Each mouse received the drug or control solution via oral gavage (0.3 ml/day) in order on the basis of the cage number every afternoon on weekdays.

After drug treatment, feces were collected from mice in different cages in each group for 16S rRNA sequencing and SCFA testing. Then, the mice were transferred to a behavior room and allowed to adapt to the laboratory conditions for 1 week under a 12/12-h light-dark cycle with free access to food and water. The mice were then subjected to behavioral tests, followed by sacrifice at seven months of age. The brain tissues were processed according to the different experimental requirements.

In addition, the inclusion/exclusion criteria for this study were related to the health of animals; animals showing signs of disease were excluded. The initial number of animals was set at 9 to 10 per group on the basis of previous research [3, 22]. After drug treatment, two mice in the WT group were excluded. Thus, the final numbers of animals used in the different groups were 7 in the WT group, 10 in the Tg group, 9 in the LS group, and 9 in the HS group.

Open field test

The open field test (OF) was carried out. Briefly, each mouse was placed in a 48-cm×48-cm OF arena and automatically recorded for 5 min by a suspended video tracking system connected to PC-based software (ANY-maze, Stoelting, USA). The total distance traveled in the arena and the time spent in the center (a 24-cm×24-cm imaginary square) were recorded. The arena was cleaned with paper towels and 70% ethanol between each trial.

Novel object recognition test

The novel object recognition test (NOR) was performed according to the methods in a previous study with minor modifications [23]. The NOR was conducted in the same apparatus as the OF. On the first day, two objects of similar size but different shapes and colors were placed opposite one another in the arena 8 cm from the side walls. The mice were placed in the center and allowed to explore the arena, including the two objects, for 10 min. After 24 h, one of the two objects was replaced with a third object (the novel object), which was of similar size but was an entirely different shape and color. The mice were placed in the arena and allowed to explore for 5 min, and their movements were recorded by a suspended video tracking system connected to PC-based software. The times the mice spent in the 2-cm areas around the novel object and the familiar object and the numbers of head entries into these areas were recorded. The discrimination index was calculated according to the formula discrimination index (%) = (time_novel –time_familiar)/(time_novel +time_familiar). The entry ratio was calculated according to the formula entry ratio (%) = (entries_novel –entries_familiar)/(entries_novel + entries_familiar).

Feces collection

Feces were collected in the morning after the day of last treatment. Before collecting the feces, antoclaved cages and tweezers were disinfected with 75% alcohol. Each mouse was placed in an empty cage, and the feces were collected immediately after excretion without being contaminated by urine. The feces were collected into sterile EP tubes pre-cooled on ice and then stored at –80°C until use. The tweezers were re-sterilized between collections from different mice with 75% alcohol.

16S rRNA sequencing

Total genomic DNA was extracted from fecal samples (n = 5 per group) with a kit (cat: DP328, TIANGEN BIOTECH, China), and the 16S rRNA genes were amplified using a specific primer (V4-V5:515F-907R) with a barcode. Polymerase chain reaction (PCR) was carried out in 30-μl reactions with 15μl Phusion High-Fidelity PCR Master Mix (cat: M0531L, New England Biolabs, USA), 0.2μM forward and reverse primers, and approximately 10 ng of template DNA. The thermal cycling program consisted of initial denaturation at 98°C for 1 min followed by 30 cycles of denaturation at 98°C for 10 s, annealing at 50°C for 30 s, and elongation at 72°C for 60 s. Finally, the samples were held at 72°C for 5 min. Sequencing libraries were generated using an NEB Next Ultra DNA Library Prep Kit for Illumina (cat: E7370L, New England Biolabs, USA) following the manufacturer’s recommendations, and index codes were added. After quality control, denoising, and chimaera removal, the samples were rarefied to an even sampling depth of 61,776 reads. The sequences were aligned using SILVA 128 (https://www.arb-silva.de). Sequence analysis was performed using the UPARSE-OTU and UPARSE-OTUref algorithms of the UPARSE software package (version 7.1). Sequences with ≥97% similarity were assigned to the same operational taxonomic unit (OTU).

To compute α diversity, we rarefied the OTU table and calculated three metrics using mothur software (version 1.30.1). The Chao1 index was used to estimate the species abundance, the observed species index was used to estimate the number of unique OTUs found in each sample, and the Shannon index was used to estimate diversity and evenness. The Quantitative Insights Into Microbial Ecology (QIIME 2) was used to calculate both weighted and unweighted UniFrac distances, which are phylogenetic measures of β diversity [24]. We used the unweighted UniFrac distances for principal coordinate analysis (PCoA) and then used the R package (version 3.1.3) to visualize the PCoA results. PCoA can be performed to obtain and visualize principal coordinates from complex multidimensional data. In the analysis, the factor responsible for the most variation is demonstrated by the first principal coordinate, the factor responsible for the second-most variation is demonstrated by the second principal coordinate, and so on.

Additionally, differential abundance analysis was performed using the linear discriminant analysis effect size (LEfSe) 1.0 [25] in order to quantify the biomarkers within different groups. This method is designed to analyze data for which the number of species is much higher than the number of samples and to provide biological class explanations in order to establish statistical significance, biological consistency, and effect-size estimates of predicted biomarkers.

Short-chain fatty acid testing

The concentrations of SCFAs associated with intestinal microbial metabolism in the fecal samples were quantified by liquid chromatography/mass spectrometry (LC/MS) (cat: QTRAP 6500 Plus, SCIEX, USA) at LipidALL Technologies (China). SCFAs were extracted from 100 mg of feces using a 1:1 ratio of water/acetonitrile (cat: A998, Thermo Fisher Scientific, USA), and appropriate internal standards were used before derivatization. The SCFAs were analyzed using a multiple reaction monitoring (MRM) scan on a SCIEX QTRAP 6500+ (USA) and an UltiMate^TM 3000 Standard Binary System (USA). Various SCFAs were isolated using Phenomenex C18 columns (cat: 00D-4462-AN, Phenomenex, USA) with mobile phase A (water:formic acid = 100:0.001), and mobile phase B (acetonitrile:formic acid = 100:0.001). Finally, the SCFAs were quantified with octanoic acid-1-4C (cat: 296457, Sigma-Aldrich, USA) as the internal standard. The SCFA concentrations are reported in units of μmol/g of feces.

Tissue sampling

Animals were anesthetized with 4% isoflurane to minimize pain, which did not affect the experimental accuracy, and transcardially perfused with a 0.9% saline solution to replace blood throughout the body. The left brain hemispheres from five mice were then immersed in 4% paraformaldehyde for 24 h, and then in 30% sucrose for 48 h to prepare for subsequent cryoprotection and histological analysis. These hemispheres were embedded in optimal cutting temperature (OCT) compound (cat: 4583, SAKURA, Japan) and then cut into coronal slices at a thickness of 30μm. The remaining left and right brain hemispheres were frozen at –80°C for protein analysis, RNA analysis, or ChIP. In each experiment, all the tissues from the different groups were ipsilateral.

The brains were thawed and homogenized in RIPA buffer (cat: C1053, Applygen Technologies, China) using a homogenizer (Bead Ruptor 12, OMNI International, USA) in three cycles of 20 s at a speed of 4.85 m/s separated by 10-s intervals on ice. Then, the homogenates were centrifuged for 15 min at 12,000×g at 4°C, and the supernatant was collected for western blots (WB) and dot blots. The RIPA buffer consisted of 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% NP-40 and 0.1% SDS. Before use, 200× protease inhibitors (cat: HY-K0010, MedChemExpress, China) were added to the RIPA buffer.

Western blots

This WB experiment was performed according to the methods described in our previous studies with minor modifications [3]. The homogenates were separated using an SDS-PAGE preparation kit (cat: C631100, Sangon Biotech, China) and the proteins were transferred onto nitrocellulose (NC) membranes (cat: 7064136, GVS, USA) and then blocked for 1 h. After three washes with Tween-TBS (cat: T1081, Solarbio, China), the membranes were incubated with primary antibodies overnight, and then with secondary anti-mouse IgG (cat: BE0102, EASYBIO, Beijing, China) and anti-rabbit IgG (cat: BE0101, EASYBIO, Beijing, China) for 1 h. The blots were finally visualized with an ECL Imaging System (Bio-Rad, cat: Gel Doc XR System, USA). The blot intensities were determined by calibration of the total amounts of pixels using the Fiji ImageJ image processing program (https://imagej.net/Fiji/Downloads). The antibodies used were as follows: ionized calcium-binding adaptor 1 (IBA1) (1:1000, cat: 019-19741, Wako, Japan), IBA1 (1:200, cat: MA5-27726, ThermoFisher Scientific), cAMP (1:200, cat: 12009-1-Ap, Proteintech, China), CREB (1:1000, #9197, Cell Signaling Technology, USA), phosphorylated-CREB (p-CREB, Ser133, 1:1000, #9198, Cell Signaling Technology, USA), phosphorylated PKA-R2 (p-PKA; Ser99, 1:2000, cat: ab32390, Abcam, UK), HDAC3 (1:500, cat: #3949, Cell Signaling Technology, USA), DM1A (1:1000, cat: T9026, Sigma-Aldrich), acetylated histone 3 (H3ace; 1:2000, cat: #06-599, Millipore, CA), and histone 3 (H3) (1:1000, cat: ab1791, Abcam, UK).

Dot blots

Homogenates were thawed and centrifuged for 2 min at 4,000×g at 4°C. The supernatants were then spotted onto an NC membrane, with 10μg of protein in each sample. After air drying, the membrane was blocked with 5% skim milk for 1 h and then incubated with a primary antibody (A11, 1:1000, cat: ABH0052, Invitrogen) and a secondary antibody. Finally, the membrane was developed with an ECL system, as described for WB analysis.

Immunohistochemistry and immunofluorescence

Immunohistochemistry (IHC) experiments were carried out according to the product instructions (cat: sp-9000, ZSGB-BIO, China). Floating sections were incubated with the following primary antibodies: IBA1 (1:200, cat: 019-19741, Wako, Japan) and A11 (1:100, cat: ABH0052, Thermo Fisher Scientific). For IHC, the sections from different groups were incubated for the same durations with diaminobenzidine tetrachloride (ZLI-9017, ZSGB-BIO, China). The images were imaged using a microscope (BX53, Olympus, Japan). Immunofluorescence (IF) experiments were carried out according to established methods [26]. The sections were incubated with the same primary antibodies as those used in the WB experiment, with the addition of an IBA1 antibody (1:200, cat: MA5-27726, Thermo Fisher Scientific). The sections were photographed using a laser confocal microscope (LSM 800, Zeiss, Germany).

Morphological examination of microglia

Morphometric analysis of microglia was performed in the cortices of IBA1-immunostained brain slices according to the methods in previous studies [27, 28]. Using the Fiji ImageJ package, skeleton analysis was carried out on binary (black and white) images. In order to obtain binary images of microglia from micrographs, a series of ImageJ plugins were progressively implemented. Then, the binary images were converted to skeletonized images, and skeleton analysis was conducted using the AnalyzeSkeleton (2D/3D) plugin, through which data about the number of endpoints and process length were collected. The cell body (soma) perimeter was measured in each IBA1-immunostained brain slice using the Analyze Particles command in the same software. Then, the cell body perimeters of microglia were measured as the cell body size according to the methods in a previous study [27].

Reverse transcriptase polymerase chain reaction (RT-PCR)

RNA was extracted from mouse hippocampal homogenates with TRIzol (Invitrogen) using a homogenizer (Bead Ruptor 12, OMNI International, USA). Reverse transcription was conducted using HiScript III RT SuperMix for quantitative polymerase chain reaction (qPCR) (+gDNA wiper) (cat: R323-01, Vazyme Biotech, China). Then, qPCR was conducted with a Thermal Cycler Dice^TM Real Time System (code TP800, Takara, Japan). The reaction system (40μl) included 0.5μM forward and reverse primers, 20μl SYBR Green PCR master mix (containing MgCl₂) and 1∼12μl cDNA. The primers (Invitrogen) were as follows: N-methyl-aspartate receptor subunit 1 (GluN1): sense 5′-TCTGGCCAGGAGGAGAGACAGAG-3′, anti-sense 5′-TGTCATTAGGCCCCGTACAGATCACC-3′; N-methyl-aspartate receptor subunit 2B (GluN2B): sense 5′-GAAGCTCTCTGGCTCACTGGC-3′, anti-sense 5′-TCATCACGGATTGGCGCTCCTC-3′; glutamate AMPA receptor subunit A1 (GluR1): sense 5′-AGAGGCTGGTGGTGGTTGACT-3′, anti-sense 5′-CGCCCTTTCTCGTTGAACTGC-3′; glutamate AMPA receptor subunit A2 (GluR2): sense 5′-AACGGCGTGTAATCCTTGAC-3′, anti-sense 5′-CTCCTGCATTTCCTCTCCTG-3′; interleukin-10 (IL-10): sense 5′-GGCGCTGTCATCGATTTCTC-3′, anti-sense 5′-CCTTGTAGACACCTTGGTCTTG-3′; IL-1β: sense 5′-GAAGAGCCCATCCTCTGTGACT-3′, anti-sense 5′-GTTGTTCATCTCGGAGCCTGTAG-3′; tumor necrosis factor-α (TNF-α): sense 5′-GACTAGCCAGGAGGGAGAACAG-3′, anti-sense 5′-CAGTGAGTGAAAGGGACAGAACCT-3′; β-actin: sense 5′-GAGACCTTCAACACCCCAGC-3′, anti-sense 5′-GAGAGCATAGCCCTCGTAGAT-3′.

Chromatin immunoprecipitation assay

The ChIP assay was carried out as previously described [26] with minor modifications. Samples were homogenized using a homogenizer (Bead Ruptor 12, OMNI International, USA) in three cycles of 20 s at a speed of 4.85 m/s separated by 10-s intervals on ice. Each sample was immunoprecipitated with 10μg primary antibody (H3ace). Nothing was added to one sample, which served as the input control. An anti-RNA polymerase II antibody was added to another sample as the positive control, and mouse IgG was added to a third sample as the negative control. The ChIP DNA (IL-1β) was normalized to the input DNA (β-actin). To quantify histone-associated DNA, qPCR was performed with the following primers: IL-1β: sense 5′-GAAGAGCCCATCCTCTGTGACT-3′, anti-sense 5′-GTTGTTCATCTCGGAGCCTGTAG-3′; β-actin: sense 5′-GAGACCTTCAACACCCCAGC-3′, anti-sense 5′-GAGAGCATAGCCCTCGTAGAT-3′.

Statistical analysis

The data were analyzed mainly using GraphPad Prism 6 software (GraphPad Software Inc., San Diego, CA, USA). The Spearman correlation analysis was performed using SPSS 19.0 (IBM SPSS Statistics, USA). The statistical methods and software used for the 16S rRNA sequencing results are described earlier in the Materials and Methods section. The specific statistical methods used in other experiments are stated in the figure legends. The data are presented as the mean±SEM, and differences with a p < 0.05 are accepted as statistically significant.

Scutellarin improves object recognition memory of APP/PS1 transgenic mice

Low (20 mg/kg, LS) and high (100 mg/kg, HS) concentrations of scutellarin were orally administered to 4-month-old APP/PS1 mice for 2 months. Once drug treatment was completed, the feces of the mice were collected. After one week of rest and acclimation, the mice were exposed to behavioral tests, including OF and NOR (Fig. 1a). During the 2-month drug treatment, there were no significant changes in body weight of the mice in the four groups (Fig. 1b).

OPEN IN VIEWER

Fig. 1

Scutellarin mitigates cognitive deficits and Aβ pathology of 7-month-old APP/PS1 mice. a) Experimental timeline. Four-month-old APP/PS1 mice were orally administered scutellarin for two months. Once drug treatment was completed, the feces of the mice were collected. After 1-week acclimation to the laboratory conditions, the mice were subjected to the OF and NOR, followed by sacrifice at seven month of age. Brain tissues were applied to different experimental assays. b) Changes in the body weights of mice from 16 weeks of age to 24 weeks of age during drug administration (two-way ANOVA followed by Bonferroni’s post hoc test, column factor, F (3, 124) = 1.078, p = 0.3611; ^*p < 0.05, n = 7∼10 animals per group). c) Total distance traveled in the arena during the OF, which indicated the activity level (one-way ANOVA followed by Bonferroni’s post hoc test, F (3, 31) = 3.230, p = 0.0357; ^*p < 0.05, n = 7∼10 animals per group). d) Time spent in the center during the OF, which provided a rough estimate of anxiety (one-way ANOVA followed by Bonferroni’s post hoc test, F (3, 31) = 1.652, p = 0.1977; ^*p < 0.05, n = 7∼10 animals per group). e) Total distance traveled in the arena during the NOR (one-way ANOVA followed by Bonferroni’s post hoc test, F (3, 31) = 1.790, p = 0.1696; ^*p < 0.05, n = 7∼10 animals per group). f) Time spent exploring the novel object during the NOR (one-way ANOVA followed by Bonferroni’s post hoc test, F (3, 31) = 3.428, p = 0.0290; ^*p < 0.05, n = 7∼10 of animals per group). g) Discrimination index during the NOR (one-way ANOVA followed by Bonferroni’s post hoc test, F (3, 31) = 6.189, p = 0.0020; ^*p < 0.05, ^**p < 0.01, n = 7∼10 animals per group). h) Number of times the head entered a 2-cm circle around the novel object during the NOR (one-way ANOVA followed by Bonferroni’s post hoc test, F (3, 31) = 6.960, p = 0.0010; ^*p < 0.05, ^***p < 0.001, n = 7∼10 animals per group). i) Ratio of head entries into the 2-cm circle around the objects during the NOR (one-way ANOVA followed by Bonferroni’s post hoc test, F (3, 31) = 4.164, p = 0.0137; ^*p < 0.05, n = 7∼10 animals per group). j-l) Scutellarin reduced Aβ oligomers levels in 7-month-old APP/PS1 mice. j-k) Dot blots of the A11 antibody targeted oligomeric forms of all proteins in the cortex of mouse brains. l) IHC staining of A11 antibody targeted oligomers in the cortex and hippocampi (cornu ammonis 3, CA3) of mouse brains. Scale bar = 50μm. m-p) The mRNA expression levels of several synapse-related proteins, including GluN1, GluN2B, GluR1, and GluR2 (one-way ANOVA followed by Tukey’s post hoc test, m: GluN1, F (3, 12) = 3.765, p = 0.0409; n: GluN2B, F (3, 12) = 2.333, p = 0.1257; o: GluR1, F (3, 12) = 0.8620, p = 0.4872; p: GluR2, F (3, 12) = 1.446, p = 0.2783; ^*p < 0.05, n = 4 animals per group).

By the OF, we evaluated the locomotor activity of the mice to rule out the influences of the drug treatments on locomotor activity in cognitive tests. Although the total distance traveled in the OF in the HS group was greater than that in the WT group, there were no significant differences between the Tg group and scutellarin-treated groups, suggesting similar activity levels (Fig. 1c). The time spent in the center was similar among the four groups, suggesting no differences in anxiety-like behaviors (Fig. 1d). In the NOR, no significant differences in total distance were detected among the four groups (Fig. 1e). However, the time taken to explore the new object in the HS group was significantly longer than that in the Tg group, while the discrimination indexes in the LS group and the HS group were significantly higher than that in the Tg group (Fig. 1f, g). In addition, the Tg group had impaired head entries compared to the WT control, and this was restored with HS treatment (Fig. 1h). The ratio of head entries was also restored by the LS and HS treatments (Fig. 1i). These data indicated that scutellarin significantly attenuated the impairment of object recognition memory of 7-month-old APP/PS1 mice.

Scutellarin reduces the levels of Aβ oligomers but does not affect the mRNA levels of synapse-related proteins of 7-month-old APP/PS1 mice

To unravel the mechanism underlying the ameliorative effects on memory, we examined Aβ pathology and synapse-related proteins in the brains of scutellarin-treated AD mice. As oligomers are the most toxic among various Aβ species [29], we detected their levels by using the A11 antibody [30, 31]. By IHC, we found that scutellarin reduced the accumulation of oligomers in the cortex and hippocampi of 7-month-old APP/PS1 mice (Fig. 1l). This result was further confirmed by dot blots (Fig. 1j, k).

To further elucidate the synaptic alterations affected by scutellarin, we measured the mRNA expression levels of synapse-associated proteins, including GluN1, GluN2B, GluR1, and GluR2. Through RT-PCR, we found no remarkable differences in the mRNA levels of these synapse-associated proteins in the WT, LS and HS groups compared with the Tg group (Fig. 1m-p).

Scutellarin improves the inflammatory state in the brains of APP/PS1 mice

Neuroinflammation is another crucial pathological characteristic of AD [32]; therefore, we tested the inflammatory state in the brains of AD mice after drug treatment. By IHC, we observed that scutellarin treatment significantly reduced IBA1 immunoreactivity, which reflects activated microglia, within the hippocampi and cortex of mouse models (Fig. 2a). Notably, the microglia in the scutellarin-treated mice had smaller cell bodies with fewer and shorter branches than those in the Tg mice (Fig. 2b–e). Moreover, WB analysis verified the role of scutellarin in reducing IBA1 expression in AD mice (Fig. 2f, g).

OPEN IN VIEWER

Fig. 2

Scutellarin ameliorates the inflammatory state in the brains of 7-month-old APP/PS1 mice. a) Immunohistochemical staining of IBA1 in the cortex and hippocampi of 7-month-old APP/PS1 mouse brain. The hippocampi includes the cornu ammonis 1 (CA1), cornu ammonis 3 (CA3), and dentate gyrus (DG) regions. Scale bar = 25μm. b) The cortical immunostaining images in the boxes were magnified and transformed into black-white skeletonized images to quantify microglia morphological characteristics. Scale bar = 10μm. c-e) Three parameters (the endpoint number, process length, and cell body size/perimeter) were used to characterize the morphology of activated microglia in the four groups (one-way ANOVA followed by Tukey’s post hoc test, c: F (3, 16) = 4.994, p = 0.0124; d: F (3, 16) = 14.13, p < 0.0001; e: F (3, 16) = 13.19, p = 0.0001; ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, ^****p < 0.0001, n = 5 animals per group). f) The IBA1 level in the cortices of 7-month-old APP/PS1 mice normalized to α-tubulin (DM1A) and measured by WB analysis. g) Statistical analysis of the relative intensity of IBA1 in WB analysis (one-way ANOVA followed by Tukey’s post hoc test, F (3, 8) = 5.472, p = 0.0243; ^*p < 0.05, n = 3 animals per group). h-j) The mRNA levels of the inflammatory cytokines IL-1β, TNF-α and IL-10 (one-way ANOVA followed by Tukey’s post hoc test, h: IL-1β, F (3, 12) = 6.067, p = 0.0094; i: TNF-α, F (3, 12) = 1.046, p = 0.4078; j: IL-10, F (3, 12) = 10.04, p = 0.0014; ^*p < 0.05, ^**p < 0.01, n = 4 animals per group).

To further confirm the inflammatory responses in the brain, the hippocampal levels of cytokines, including IL-1β, TNF-α, and IL-10, were detected by RT-PCR. Compared with the Tg group, the HS group had a significantly lower level of the pro-inflammatory cytokine IL-1β and a significantly higher level of the anti-inflammatory cytokine IL-10 in the hippocampi. However, the level of TNF-α was not altered by scutellarin treatment (Fig. 2h–j).

Scutellarin changes alpha diversity and beta diversity of the fecal microbiota in APP/PS1 mice

To uncover the effects of scutellarin on the gut microbiota of APP/PS1 mice, the fecal samples were subjected to 16S rRNA sequencing. Analysis of the α diversity of intestinal bacteria showed that the Shannon index of the HS group was significantly lower than that of the WT and Tg groups, while there were no significant differences in the number of OTUs and the Chao1 index among the four groups (Fig. 3a–c). The numbers of similar and different OTUs among the four groups are displayed in a Venn diagram (Fig. 3d). These data indicated that scutellarin decreased α diversity of intestinal bacteria in APP/PS1 mice.

OPEN IN VIEWER

Fig. 3

Scutellarin affects alpha and beta diversity of the fecal microbiota in APP/PS1 mice. a) Chao1 index, an estimator of species richness (one-way ANOVA followed by Tukey’s post hoc test, F (3, 16) = 2.664, p = 0.0831; ^*p < 0.05, n = 5 animals per group). b) Shannon index, an estimator of diversity and evenness (one-way ANOVA followed by Tukey’s post hoc test, F (3, 16) = 4.770, p = 0.0147; ^*p < 0.05, n = 5 animals per group). c) Number of OTUs (one-way ANOVA followed by Tukey’s post hoc test, F (3, 16) = 2.965, p = 0.0635; ^*p < 0.05, n = 5 animals per group). d) OTU Venn analysis performed with R packages; OTU samples with a similarity level of 97% were used for the analysis (n = 5 animals per group). e, f) PCoA based on unweighted UniFrac distances of the fecal microbiota from the four mouse groups (n = 5 animals per group). g) GraPhlAn plot of the LEfSe analysis results for microbiome profiles up to the family level from the 16S rRNA gene sequencing data. The clustering tree shows bacteria with significant differences (LDA score >2.0, p < 0.05). The different colors of the areas represent different groups. The red nodes in the branches represent the bacteria that play important roles in the red group, and so on. The yellow nodes represent the bacteria that do not play important roles in either group.

The β diversity of intestinal bacteria was further analyzed by PCoA. The results showed that the microbiota of the Tg group was significantly separated from the other three groups, reflecting the altered composition of the fecal microbiota after oral supplementation with scutellarin (Fig. 3e-f). Moreover, the LEfSe results of the microbiome profiles revealed key microbes, which may matter in the behavioral or molecular phenotypes of the four mouse groups induced by the different treatments. The results showed that Rikenellaceae, Peptococcaceae and Saccharimonadaceae mattered in the WT group; that Desulfovibrionales, Erysipelotrichaceae, Deltaproteobacteria, Burkholderiaceae, and Lachnospiraceae mattered in the Tg group; that Marinifilaceae, Enterobacteriaceae, and Gammaproteobacteria mattered in the LS group; and that Prevotellaceae, Bacteroidales, and Staphylococcaceae mattered in the HS group (Fig. 3g). The above data indicated that scutellarin affected α diversity and β diversity of intestinal bacteria in the APP/PS1 mice and the analysis screened out several potentially important families of microbiota.

Scutellarin alters the composition of the fecal microbiota in APP/PS1 mice at the phylum and genus levels

The specific taxonomic shifts were investigated in detail. At the phylum level, the murine fecal microbiota was dominated by Bacteroidetes and Firmicutes, the relative abundance of which differed among the four groups (Fig. 4a). The relative abundance of Bacteroidetes in the HS group was significantly higher than that in the Tg group. In contrast, the relative abundance of Firmicutes and Proteobacteria in the HS group was significantly lower than that in the Tg group (Fig. 4b–d). The mice in the WT group possessed a higher abundance of Tenericutes than the mice in the other three groups (Fig. 4e).

OPEN IN VIEWER

Fig. 4

Scutellarin alters the microbiota components of APP/PS1 mice up to the genus level. a) Microbial distribution at the phylum level. All phyla accounting for less than 1% of the total abundance were combined into the “Others” category. b-e) Relative abundances of selected phyla with significant differences among the four mouse groups. The relative abundances of bacteroidetes (b), firmicutes (c), proteobacteria (d), tenericutes (e) are shown (the nonparametric Kruskal–Wallis test was used to analyze the differences among the mouse groups, ^*p < 0.05, n = 5 animals per group). f, g) Relative abundances of selected genera with significant differences among the four mouse groups (the nonparametric Kruskal–Wallis test was used to analyze the differences among the mouse groups, ^*p < 0.05, ^**p < 0.01, n = 5 animals per group).

In accordance with the aforementioned results, at the genus level, Lachnospiraceae and Alloprevotella were the dominant microbial taxa with differences among the four groups (Fig. 4g). Compared with the Tg group, the significant increase in the abundance of Bacteroidetes phylum was attributed to a significant increase in Alloprevotella in the HS group. Similarly, the significant decrease in the abundance of Firmicutes and Proteobacteria was attributed to significant decreases in Lachnospiraceae and Parasutterella in the HS group, respectively. Significantly lower abundances of Parabacteroides, Desulfovibrionaceae, and Dubosiella were also found in the LS and HS groups compared to the Tg group. In addition, scutellarin administration significantly increased the abundance of Paraprevotella in APP/PS1 mice. The abundances of the following genera also increased, but without significant differences, such as Eubacterium coprostanoligenes, Candidatus Arthromitus, Mollicutes RF39, and Staphylococcaceae (Fig. 4f-g). In short, scutellarin significantly increased the abundances of 2 genera and significantly decreased the abundances of the 4 other genera in the APP/PS1 mice.

Scutellarin reduces the level of acetate in the feces of APP/PS1 mice

The levels of microbial-derived SCFAs in the feces can be used as indexes to evaluate gut microbiota activity. To assess the alterations in SCFAs, we extracted and quantified the acetate, propionate, butyrate, and isobutyrate by LC/MS. It was observed that the level of acetate in the LS and HS groups was significantly lower than that in the Tg group, whereas no significant differences were observed in the levels of the three other kinds of SCFAs among the four groups (Fig. 5a–d).

OPEN IN VIEWER

Fig. 5

Scutellarin changes the concentrations of short-chain fatty acids in the feces of APP/PS1 mice. a-d) Scutellarin reduced the concentration of acetate but had no effects on other SCFAs, such as butyrate, propionate and isobutyrate, in the feces of APP/PS1 mice (one-way ANOVA followed by Tukey’s post hoc test, a: acetate, F (3, 16) = 5.161, p = 0.0110; b: propionate, F (3, 16) = 0.4917, p = 0.6930; c: butyrate, F (3, 16) = 2.707, p = 0.0799; d: isobutyrate, F (3, 16) = 0.2259, p = 0.8770; ^*p < 0.05, n = 5 animals per group).

Correlations among cognitive behaviors, the gut microbiota and inflammatory cytokines

To investigate whether the altered intestinal bacteria were associated with cognitive impairment or neuroinflammation, Spearman correlation analysis was performed. The results showed that the mRNA level of IL-1β in the hippocampi was negatively associated with the time spent around the novel object, the discrimination index, the number of head entries into the 2-cm area around the novel object, and the entry ratio in the NOR, indicating a negative correlation between neuroinflammation and cognition. Additionally, significantly positive correlations of the relative abundances of Alloprevotella, Paraprevotella and Staphylococcus were revealed with the discrimination index in the NOR. In contrast, significantly negative correlations of Dubosiella, Parasutterella, and Parabacteroides were revealed with the discrimination index in the NOR. These results suggest that the fecal microbiota is associated with cognition. Moreover, a significantly positive correlation of IL-1β was revealed with the abundance of Dubosiella, while a negative correlation was revealed with the abundance of Alloprevotella. For IL-10, a significantly positive correlation was revealed with the abundance of Staphylococcus; significantly negative correlations were revealed with the abundance of Dubosiella, Desulfovibrionaceae, and Parasutterella (Fig. 6). Thus, it was concluded that among the altered microbiota, the increase in Alloprevotella and the decrease in Dubosiella played critical roles in mediating the therapeutic anti-inflammatory effects of scutellarin in the context of AD.

OPEN IN VIEWER

Fig. 6

Correlations among cognitive behaviors, the gut microbiota and inflammatory cytokines. The color of each square in the matrix codes for the level of correlation; red represents a negative correlation, and blue represents a positive correlation (Spearman correlation analysis). The mRNA level of IL-1β in the hippocampi was significantly negatively associated with the time spent around the novel object, the discrimination index, the number of head entries into the 2-cm area around the novel object and the entry ratio in the NOR. The discrimination index in the NOR was significantly positively associated with Alloprevotella, Paraprevotella, and Staphylococcus abundances and significantly negatively associated with Dubosiella, Parasutterella, and Parabacteroides abundances. The IL-1β level was significantly positively correlated with the abundance of Dubosiella and negatively correlated with the abundance of Alloprevotella. The IL-10 level was significantly positively associated with Staphylococcus abundance and significantly negatively associated with Dubosiella, Desulfovibrionaceae, and Parasutterella abundances.

Scutellarin reverses alterations in the cAMP-PKA-CREB-HDAC3 pathway downstream of SCFAs in the microglia of APP/PS1 mice

To test whether the cAMP-PKA-CREB-HDAC signaling downstream of SCFAs was affected in our models, we measured the protein levels of cAMP, p-PKA, p-CREB, total CREB and HDAC3 in the cortices of mice. In WB, we observed significant increases in cAMP, p-PKA, and p-CREB in the Tg group compared to the WT group, but only a trend of reductions in the HS group with respect to the Tg group. At the same time, the protein level of HDAC3 was significantly increased in the Tg group while that of HDAC3 was significantly decreased in the LS and HS groups in a dose-dependent manner (Fig. 7a, b). It was concluded that the change trends for cAMP, p-PKA, p-CREB, and HDAC3 were roughly consistent; that is, they increased in AD and decreased after drug administration, indicating the involvement of cAMP-PKA-CREB-HDAC3 signaling in our models.

OPEN IN VIEWER

Fig. 7

Scutellarin reverses the change in cAMP-PKA-CREB-HDAC3 signaling downstream of SCFAs in the microglia of 7-month-old APP/PS1 mice. a) WB analysis of cAMP-PKA-CREB-HDAC3 pathway proteins in the cortices of mouse brains. b) Statistical analysis of the relative intensity of cAMP-PKA-CREB-HDAC3 pathway proteins (one-way ANOVA followed by Tukey’s post hoc test, cAMP: F (3, 8) = 6.336, p = 0.0165; p-PKA: F (3, 8) = 7.365, p = 0.0109; p-CREB: F (3, 8) = 7.539, p = 0.0102; CREB: F (3, 8) = 0.3906, p = 0.7632; HDAC3: F (3, 8) = 26.25, p = 0.0002; ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, n = 3 animals per group). c-f) IF images of cAMP, p-PKA, p-CREB and HDAC3 in IBA1+ microglia of the cortex in brain slices. Scale bar = 20μm. g-j) Quantitative analysis of the double-positive cells, including cAMP+/IBA1+, p-PKA+/IBA1+, p-CREB+/IBA1+ and HDAC3+/IBA1+ cells, in the visual fields. The arrows indicate double-positive cells (one-way ANOVA followed by Tukey’s post hoc test, cAMP: F (3, 16) = 16.84, p < 0.0001; p-PKA: F (3, 16) = 16.32, p < 0.0001; p-CREB: F (3, 16) = 9.676, p = 0.0007; HDAC3: F (3, 16) = 13.16, p = 0.0001; ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, ^****p < 0.0001, n = 5 per group).

However, the role of the cAMP-PKA-CREB-HDAC3 signaling in mitigating neuroinflammation in our models is still unknown. Thus, we evaluated the alteration of this signaling in cortical microglia by performing IF staining. Co-localizations of the four proteins cAMP, p-PKA, p-CREB, and HDAC3 with IBA1 in the cortex were augmented in the Tg group but decreased in the scutellarin-treated groups. In addition, these proteins, especially p-CREB and HDAC3, co-localized not only with microglia, but also with other types of cells. Therefore, to some extent, the IF results supported changes in these four proteins revealed by WB (Fig. 7c–j). These results demonstrated that scutellarin reversed the alterations in cAMP-PKA-CREB-HDAC3 signaling in the microglia of APP/PS1 mice.

The level of H3ace associated with IL-1β promoter increases in AD, but it is reversed by scutellarin

HDACs are key epigenetic regulators that function by mediating dynamic histone acetylation at lysine residues. To reveal the changes of signaling downstream of HDAC3, we detected the levels of acetylated histones by WB. There were no significant differences in H3ace between the Tg group and the WT group, but the H3ace level was significantly higher in the HS group than that in the Tg group (Fig. 8a, b). The change in H3ace was roughly consistent with the alteration in HDAC3 after drug treatment, suggesting that scutellarin indeed inhibited HDAC3 and augmented histone acetylation in APP/PS1 mice. As histone acetylation regulates gene expression, we examined the relationship between the changed IL-1β and H3ace by performing a ChIP assay on hippocampal extracts. Surprisingly, compared with the WT group, the Tg group showed a significantly increase in the association between H3ace and the IL-1β promoter, but scutellarin treatment reduced this association in APP/PS1 mice (Fig. 8c). These data indicated that the decrease in H3ace associated with the promoter of the IL-1β led to the downregulation of IL-1β, strongly supporting the hypothesis that cAMP-PKA-CREB-HDAC3 signaling mediates the therapeutic anti-inflammatory mechanism of scutellarin in APP/PS1 mice.

OPEN IN VIEWER

Fig. 8

The level of H3ace associated with IL-1β promoter increases in AD, but it is reversed by scutellarin. a) The H3ace level in the cortices of APP/PS1 mice normalized to the histone 3 (H3) level, as measured by WB analysis. b) Statistical analysis of the relative intensity of H3ace (one-way ANOVA followed by Tukey’s post hoc test, F (3, 8) = 9.940, p = 0.0045; ^*p < 0.05, ^**p < 0.01, n = 3 animals per group). c) Amount of the H3ace associated with the IL-1β promoter in hippocampal extracts of 7-month-old APP/PS1 mouse, as measured by ChIP. Immunoprecipitated DNA (ChIP DNA) or genomic DNA (input) was measured by qPCR, and the ChIP DNA was normalized to the input DNA (one-way ANOVA followed by Tukey’s post hoc test, F (3, 8) = 7.651, p = 0.0098; ^*p < 0.05, ^**p < 0.01, n = 3 animals per group).