The Effects of LW-AFC on Intestinal Microbiome in Senescence-Accelerated Mouse Prone 8 Strain,a Mouse Model of Alzheimer’s Disease

Preparation of LW-AFC

LW-AFC was prepared from LiuweiDihuang decoction and includes polysaccharide fraction (LWB-B), glycosides fraction (LWD-b), and oligosaccharide fraction (CA-30). LiuweiDihuang decoction was prepared as previously described in studies by Yang et al. [43], Zhang et al. [44, 45] and Cheng et al. [46, 47]. After the prepared LiuweiDihuang decoction was filtered by 6-layer gauze, the extraction solutions were centrifuged (2500 rpm/min, 25 min). The supernatant was concentrated into quintessence. The quintessence was extracted in ethanol to produce the supernatant (LWD), and the sedimentation left in deionized water was concentrated into dried polysaccharide fraction (LWB-B). LWD was concentrated and removed ethanol, then dissolved in deionized water and eluted in turn by deionized water (6 column volume) and 30% ethanol (4 column volume) on macroporous adsorptive resins. On the one hand, the 30% ethanol elution of LWD was cryodesiccated into glycosides fraction (LWD-b). On the other hand, the water elution of LWD was concentrated and eluted in turn by 5% ethanol (6 column volume) and 30% ethanol (2 column volume) on active carbon absorption column (GH-15, Φ150×1500 mm, diameter height ratio was 1/9). Then the 30% ethanol elution was concentrated and removed ethanol, cryodesiccated into the oligosaccharide fraction (CA-30). Finally, LW-AFC was composed of 20.3% polysaccharide fraction (LWB-B), 15.1% glycosides fraction (LWD-b), and 64.6% oligosaccharide fraction (CA-30) in the dry weight.

Animals and drug administration

Original SAMR1 and SAMP8 mice were kindly provided by Dr. T. Takeda at Kyoto University, Japan. The mice were maintained in the Beijing Institute of Pharmacology and Toxicology under standard housing conditions (room temperature 22±1°C and humidity of 55±5%) with a 12-h light/12-h dark cycle, and were allowed free access to water and food. The 6-month-old male SAMP8 and SAMR1 mice were separated into 3 groups at random, each group had 5 mice. LW-treated SAMP8 mice were orally administered LW-AFC at the doses of 1.6 g/kg/day, for 5.5 months. SAMP8 mice as negative controls and SAMR1 mice as positive controls, were given with equal volume of distilled water respectively. The weight of mice was evaluated every 3 days. After administrating LW-AFC for 90 consecutive days, animals were used to conduct Morris-water maze test, novel object recognition test and shuttle-box test one by one. Following the behavioral experiments, the stool was collected for metagenomic analysis. The animal treatment, husbandry and experimental protocols in this study were approved by the Institute of Animal Care and Use Committee (IACUC) of the National Beijing Center for Drug Safety Evaluation and Research (NBCDSER).

Morris water maze test

The procedure of Morris water maze test was according to Vorhees and Williams [48] and Hu et al. [49]. This behavioral task included hidden-platform training (spatial learning) and probe trial (spatial memory) session. In the hidden-platform training session, the mouse was allowed 4 daily trials in the presence of the platform, for 5 subsequent days. In this session, mice were devoted into the pool facing the wall in one of the four quadrants. When the mouse located the platform, it was allowed to stay on the platform for 10 s, and if the mouse did not locate within 60 s, it was placed on the platform for 10 s to familiarize. In the probe trial session, the platform was removed, and the mouse was allowed to swim to search it for 60 s. During the whole Morris water maze test, the escape latency (the time taken to find the hidden platform) in hidden-platform training session, the escape latency (the first time that the mice crossed the removed platform) and number of times that the mice crossed the removed platform in probe trial session were recorded and analyzed.

Novel object recognition test

The procedure of novel object recognition test was according to Bevins and Besheer [50] and Xu et al. [51]. The apparatus was placed in an illuminated soundproof room. The procedure includes three phases: habituation, training, and testing. For two consecutive days, the animal was allowed to freely explore the vacant chamber to familiarize with the testing environment for 20 min per day. On the third day, placed two identical objects (sample object A and B) in the box. Then, each mouse allowed to explore the objects for 16 min and then returned to its home cage. The training-to-testing interval was 1 h. To test the object recognition memory, after 1 h training-to-testing interval, the mouse was placed back into a similar chamber where one of the two identical objects was switched to a new one (novel object C); the test session was 4 min. The object exploration time (the length of time when mouse directing its nose to the object within 2 cm, or pawing or sniffing the object) of object A, B, and C was recorded. The discrimination index was calculated to assess the short and long term object recognition memory of mice.

Shuttle-box test

The procedure of shuttle-box was according to Cheng et al. [52] and Yang et al. [43]. The shuttle box apparatus (Med Associates Inc., East Fairfield, VT, USA), consisting of two 50 cm×15 cm×40 cm compartments connected by an automated gate. Each tasting session began with acclimatization to the chambers for 2 min followed by 30 trials, and inter-trial interval was 30 s. Tone (60 dB) and light (8 W) were used as the conditioned stimulus (CS), and the CS was presented for 10 s. An electrical foot shock (0.2 mA), used as the unconditioned stimulus, was given for 5 s, subsequently. If the mouse escaped to the other chamber in the period of CS, no electrical foot shock presented and an active avoidance was recorded. If the mouse did not move to the opposite compartment during the CS, the shock was given until the mouse crossed to the other side or the time of shock had passed. If the mouse moved while the foot shock was being presented, a passive avoidance was recorded. The shuttle-box procedure was performed for 5 days. The computer scored the number of active avoidances. At the 6th day, all mice were submitted to another session of shuttle-box to test the level of the learning and memory ability. This test was also consisted of 30 trials, and was preceded by a permitted ambulation (2 min) before the first CS, but no shock. The number of active avoidances was recorded during the test.

16S rRNA amplicon sequencing of gut microbiota

All fecal specimens were freely defecated by mice and immediately collected. 180–220 mg of fresh stool was collected from each mouse. Total genomic DNA (gDNA) of every mouse’s fresh stool was isolated using QIAamp DNA Stool Mini Kit (50) (51504, Qiagen) according to the manufacturer’s protocol. The integrity of gDNA was confirmed using an agarose gel and the final gDNA yield was quantified using the ND-2000c(CCC) (Thermo, USA). Universal primers 356F (5^′CCTACGGGNGGCWGCAG3^′) and 803R (5^′GACTACHVGGGTATCTAATCC3^′) were employed to amplify V3-V4 regions of bacterial 16S rRNA gene from gDNA. A two-step PCR based library construction protocol was described as previously [53]. The libraries were sequenced on Illumina Miseq platform (Illumina, San Diego, CA, USA) to generate 2×300-bp pair-end sequencing reads using the standard protocol under standard conditions.

Taxonomic analysis of 16S sequencing data

The procedure for analyzing 16S sequencing data was described in our previous study [53] but with some modifications. Due to the low quality of reverse reads, only forward reads rather than pair-end reads were used as input of QIIME [54]. Thus, FLASH [55], a software previously used to merge pair-end reads, was not employed either. The whole procedure, including data quality control, chimera removal, operational taxonomic unit (OTU) clustering, and taxonomic assignment, was performed in QIIME.

Microbial diversity analysis

Rarefaction curves, measured by observed OTU, Shannon index, and Chao1 index, were generated by using the scripts “multiple_rarefactions.py”, “alpha_diversity.py” and “make_rarefaction_plots.py” in QIIME with a step size of 1000 reads. In order to eliminate potential bias introduced by differential sequencing depths of all these samples, alpha diversity indices (observed OTU, Goods coverage, Chao1 index, ACE index, and Shannon index) and beta diversity indices (unweighted UniFrac distance and weighted UniFrac distance) were calculated by randomly subsampling the total reads of each sample to a uniform depth for 600 times as previously described [53]. Principal coordinate analyses (PCoA) based on the subsampled unweighted UniFrac distances and subsampled weighted UniFrac distances were also performed in QIIME.

Statistical analysis

All data were expressed as mean±S.D. GraphPad Prism 6.0 (GraphPad Software, Inc., La Jolla, CA, USA) was used to plot and analyze data. Data between two groups were compared by Student’s t-test. Inter-group differential taxonomic ranks (from phylum to genus) were detected by using Student’s test in R (v3.1.2). Spearman correlation coefficients (R, v3.1.2) were employed to measure the correlation between the differentially abundant genera and the mice behavioral metadata. p < 0.05 was taken as statistically significant.

The treatment of LW-AFC improved cognitive impairments of SAMP8 mice

Morris water maze test was employed to measure the ability of spatial learning and memory of mice. Results showed that in the learning task, SAMP8 mice showed longer escape latency than SAMR1 mice on the final day, while the treatment of LW-AFC made this latency short in SAMP8 mice (Fig. 1A). In the probe trial, the number of crossing platform was decrease in SAMP8 mice compared to SAMR1 mice, while the administration of LW-AFC increased this number in SAMP8 mice (Fig. 1C). This data indicated that LW-AFC administration ameliorated the impairment of spatial learning and memory in SAMP8 mice.

The shuttle-box test was performed to investigate the active avoidance of mice. The successful avoidance times of SAMP8 mice were significantly fewer than SAMR1 mice since the 2nd day of the training phase, and the successful avoidance times increased significantly by LW-AFC administration since the 3th day of training phase (Fig. 2A). In the testing phase, the successful avoidance times of SAMP8 mice decreased comparing with SAMR1 mice, while increased after LW-AFC administration (Fig. 2B). This indicated that LW-AFC had protective effect on active avoidance deficits in SAMP8 mice.

The novel object recognition test was used to investigate the object recognition memory of mice. Results showed that the discrimination index (the percentage of time spent with the novel object with respect to the total exploration time) in SAMP8 mice was significantly decreased in the object recognition memory test (Fig. 3), while the treatment of LW-AFC significantly reversed this memory (Fig. 3A). This indicated that deterioration in the object recognition memory of SAMP8 mice were improved by LW-AFC treatment.

The treatment of LW-AFC restored the imbalance of gut microbiota in SAMP8 mice

We collected 15 stool samples and performed 16S rRNA amplicon sequencing (16S sequencing) on gDNA extracted from 15 samples. We obtained an average of 160,684 raw reads per sample. After quality control and operational taxonomic unit (OTU) clustering, a total of 13,727 OTUs were detected among all samples (range: 601–5,930). The comparison of alpha diversity indices based on randomly subsampling the reads of each sample revealed that the SAMP8 mice had reduced alpha diversities when compared to SAMR1 mice (observed OTU, p = 0.011; Goods coverage, p = 0.014; Chao1 index, p = 0.02; ACE index, p = 0.022) (Fig. 4A–C). No significant differences were observed between SAMP8 mice and SAMP8 mice administrated with LW-AFC. At the phylum level, Firmicutes, Bacteroidetes, Proteobacteria, Actinobacteria, Deferribacteres, and TM7 were dominant bacteria taxa, accounting for 96.04–99.81% of the total abundance of each sample (Fig. 4D). Among these phyla, Firmicutes, Bacteroidetes, and Proteobacteria were detected in all samples. The phylum TM7, usually identified as an environmental taxon [56], was detected in 11 samples with ≥0.1% abundance. At the genus level, a total of 62 genera were identified with ≥0.1% abundance in at least two samples. There were 36 of these genera were detected in at least 7 samples (Fig. 4E). Among all these stool samples, Lactobacillus, Bacteroides, Prevotella, Ruminococcus, and Flexispira were highly abundant.

Principal coordinate analysis (PCoA) based on unweighted UniFrac distances revealed that the gut microbiota of SAMP8 mice administrated with LW-AFC group and SAMR1 mice group were clustered together, while the gut microbiota of SAMP8 group were distinct from the other two groups (Fig. 5A). PCoA analysis based on weighted UniFrac distances also demonstrated similar trends (Fig. 5B). According to the PCoA results, the intestinal microbiota of the treatment of LW-AFC were more similar to those of the SAMR1 mice group than the SAMP8 mice group, which indicated that the treatment of LW-AFC might have the ability to restore, at least in part, the gut microbial communities of SAMP8 mice to those of the SAMPR1 mice.

We used Student’s t-test to identify the specific bacterial phylotypes that were different between SAMR1 and SAMP8 mice, and those altered by LW-AFC treatment. Compared with SAMR1 mice, there were 8 significantly increased and 12 decreased OTUs in SAMP8 mice (Fig. 5C, D). The treatment of LW-AFC altered 22 (16 increased and 6 decreased) OTUs in SAMP8 mice (Fig. 5C, D). Among the 22 OTUs, 15 OTUs could be reversed by LW-AFC treatment (Fig. 5C, D). These results suggested that LW-AFC could be the effective treatment for modulating the gut microbiota in SAMP8 mice.

Detailed analysis of the 15 OTUs reversed by LW-AFC indicated that at the genus level, Adlercreutzia, Anaerotruncus and Ruminococcus were decreased, while Prevotella, Streptococcus, Veillonella, and Bilophila were increased in SAMP8 mice comparing with SAMR1 mice, while these all were restored by LW-AFC treatment (Fig. 5C, D). Notably, in comparison with SAMP8 mice, LW-AFC treatment enhanced Coprococcus which revealed no difference between SAMP8 and SAMR1 mice. It was also notable that Candidatus Arthromitus, Lachnospiraceae [Ruminococcus], and Clostridium were differential between SAMP8 and SAMR1 mice, while they were not altered by LW-AFC (Fig. 5C, D). Collectively, these results showed that LW-AFC modulated the gut microbiota of the SAMP8 mice, resulting in a microbial composition similar to that of SAMR1 mice.

LW-AFC impacted on bacterial taxa correlated with the abilities of learning and memory in SAMP8 mice

In order to investigate whether the ability of learning and memory in central nerve system is related to gut microbiota or not, we further calculated the correlation between the differentially abundant OTUs and the cognitive abilities of all mice using Spearman correlation analyses. Results showed that there were 12 OTUs correlated with the number of crossing platform in the probe trial of Morris water maze test, including 4 negatively correlative genus, such as Streptococcus and Veillonella, 8 positively correlative genus, such as Anaerotruncus (Fig. 6A, D). There were 12 OTUs correlated with the successful avoidance times in testing phase of shuttle-box test, including 5 negatively correlative genus, such as Streptococcus, Veillonella, and Candidatus Arthromitus, and 7 positively correlative genus, such as Adlercreutzia, Lachnospiraceae [Ruminococcus], and Ruminococcaceae Ruminococcus (Fig. 6B, D). There were 14 OTUs correlated with the discrimination index in object recognition memory test, including four negatively correlative genus, such as Veillonella and Bilophila, and 10 positively correlative genus, such as Adlercreutzia, Lachnospiraceae [Ruminococcus], and Ruminococcaceae Ruminococcus (Fig. 6C, D). There were 7 OTUs significantly correlated with all the three types of cognitive abilities, at the order level, including Bacteroidales, Clostridiales, Desulfovibrionales, and CW040 (Fig. 6D). These data suggested that intestinal microbiota might be linked to the impairment of learning and memory in SAMP8, a mouse model of AD.

At the family level, Desulfovibrionaceae, Veillonellaceae, and an unclassified family had significantly negative correlation with spatial learning and memory ability, active avoidance response, and object recognition memory capability (Fig. 6D). At the order level, CW040, Clostridiales, and two unclassified orders had positive correlation with all these cognitive abilities (Fig. 6D). Furthermore, the abundance of all these 7 OTUs correlating with cognition were differential between SAMP8 and SAMR1 mice, and LW-AFC treatment restored the gut microbiota of SAMP8 mice to a composition similar to those of SAMR1 mice (Fig. 6D). These results indicated that the effects of LW-AFC improving cognitive impairments of SAMP8 mice might be through modulating intestinal microbiome.