Age-related shifts in gut microbiota contribute to cognitive decline in aged rats

Cognition changes in aged rats

The cognitive functions of young and aged rats were analyzed by operant-based delayed matching to position (DMTP) task (Figure 1A). The correct rate of lever-pressing at different delays is shown in Figure 1B. At the shortest delay, both young and aged rats performed well. As delays increased, the accuracy decreased in both young and aged rats, especially at 18 s and 24 s delay, the accuracy decreased significantly (main effect of delay: P < 0.05). Notably, compared with young rats, aged rats performed even worse at longer delays of 18 s and 24 s, and the accuracy was significantly lower than in the young rats (P < 0.05). When the delay was 24 s, the accuracy in aged rats was close to 50%, possibly a probability event (Figure 1B).

Using resting-state functional magnetic resonance (rs-fMRI) and regional homogeneity (ReHo) analytical method, we identified the brain regions showing differences in spontaneous blood oxygenation level dependent (BOLD) signal, representing neuronal activities in young and aged rats. Compared with young rats, aged rats had lower ReHo values in almost the entire brain; this is consistent with the structural and functional alterations in normal aging brain (Figure 1C).

In addition, there were significant changes of synaptic structures in medial prefrontal cortex (mPFC) and hippocampus in aged rats, including widened synaptic cleft, thinned post-synaptic density (PSD), and decreased curvature of synaptic interface (Figure 1D–1E). Aged rats had a decreased expression of synaptophysin in mPFC and hippocampus compared to young rats (P < 0.05, Figure 1F). Furthermore, the density of dendritic spines in mPFC and hippocampus decreased significantly in aged rats (Figure 1G–1J).

Immunohistochemistry (IHC) and western blotting analyses demonstrated that the expression of NR1, the essential subunit of N-methyl-D-aspartate (NMDA) receptor, decreased significantly in aged mPFC and hippocampus compared with young rats (Figure 2A–2C). In addition, the expression of brain derived neurotrophic factor (BDNF) in mPFC and hippocampus of aged rats decreased compared with young rats (Figure 2D–2F), while the expression of advanced glycation end products (AGEs) and receptor for AGEs (RAGE) increased (Figure 2G–2L).

Analysis of inflammatory cytokines in serum and brain tissues showed that the serum levels of IL-1β, TNF-α and IL-6 were increased in aged rats compared with young rats (Figure 3A–3C). Similarly, the expression of IL-1β, TNF-α and IL-6 was higher in mPFC and hippocampus of aged rats than in young rats (Figure 3D–3F).

We then examined activity of the antioxidant enzymes superoxide dismutase (SOD) and glutathione peroxidase (GSH-PX), as well as the expression of malondialdehyde (MDA). In serum, the SOD activity of aged rats declined, while the MDA expression increased; the activity of GSH-PX did not change compared to young rats (Figure 3G–3I). Similarly, both in mPFC and hippocampus of aged rats, the SOD activity declined, while the MDA level increased compared with young rats; the activity of GSH-PX did not change (Figure 3J–3L).

Gut microbiota varies in young and aged rats

Analysis of the composition of gut microbiota in young and aged rats showed that the diversity of microbiota was significantly lower in aged rats than in young rats (P < 0.05, Figure 4A, 4B). Nonmetric Multidimensional Scaling (NMDS) analysis showed that the microbial communities of aged rats were significantly different from young rats (P < 0.05, Figure 4C); there was no overlap between the microbiota of young and aged rats.

We then analyzed the composition of microbiota communities at various taxonomic levels. At the phylum level, Bacteroidetes and Firmicute were the main bacteria in the feces of rats (Figure 4D). The relative abundance of Bacteroidetes was lower in aged rats than in young rats, while the relative abundance of Firmicutes was higher (Figure 4E). The ratio of Firmicutes / Bacteroidetes (F/B) increased about 7.4 fold with aging, from 0.97 ± 0.07 to 7.22 ± 1.69 (P < 0.01, Figure 4F). In addition, the relative abundance of Actinobacteria, Spirochaetes and TM7 increased in aged rats compared to young rats (Figure 4E). There was no statistical difference in Proteobacteria between the young and aged rats (Figure 4E).

At the genus level, the microbiota composition in aged rats significantly differed from the young rats (Figure 4G). The relative abundance of Prevotella, Bacteroides, Sutterella and Parabacteroides decreased in aged rats, whereas Bulleidia, Collinsella, Lactobacillus, Treponema and Allobaculum increased in aged rats (Figure 4H).

Gut microbiota of young recipient rats is reshaped by fecal microbiota transplantation (FMT)

To analyze the effect of gut microbiota on cognitive decline in aged rats, we treated young rats with aged microbiota from aged rats by FMT. Within 2 months after transplantation, some characteristics of the microbiota in the young recipient rats (3 months) changed to characteristics similar for aged (24 months) rats, including diversity and dominant bacteria.

The diversity of gut microbiota in the FMT group was lower than in the non-fecal microbiota transplantation (NFMT) group, although the difference was not statistically significant (Figure 5A). NMDS analysis showed that the gut microbiota of the FMT rats was significantly different from the NFMT rats (P < 0.05, Figure 5B). UPGMA cluster analysis also showed that the microbiota in the control NFMT group was closer to the microbiota before transplantation, while the microbiota in the FMT group was significantly different from the NFMT group, and was closer to aged rats (Figure 5C).

Microbial community analysis also showed that FMT changed the gut microbiota of the young recipient rats. At the phylum level, the relative abundance of Bacteroidetes was lower in FMT rats than in NFMT rats, and the difference was significant (P < 0.01). In contrast, the abundance of Firmicutes was higher, although the difference was not statistically significant (Figure 5D–5E). Compared with NFMT group, the ratio of F/B in FMT group increased (P < 0.05, Figure 5F). The relative abundance of Actinobacteria increased in FMT group, although the difference was not statistically significant. The relative abundance of Spirochaetes did not change in FMT group (Figure 5E). At the genus level, FMT rats had lower levels of Prevotella, Bacteroide, Parabacteroides, and higher levels of Sutterella (Figure 5G–5H).

FMT of aged microbiota from aged rats impairs cognitive behavior in young rats

The cognitive behavior of NFMT and FMT rats was evaluated by DMTP task. The correct rate of lever-pressing under different delays is shown in Figure 6A. At the shortest delay, both NFMT and FMT rats performed well. However, at longer intervals (18 s and 24 s), the accuracy of lever-pressing in FMT rats was significantly lower than in the NFMT group (P < 0.05, Figure 6A), suggesting that the cognition of the young recipient rats was impaired by the FMT of aged microbiota.

FMT of aged microbiota decreases brain activity in young rats

Functional Magnetic Resonance Imaging (fMRI) is one of the best tools for studying functions of the brain. As with conventional fMRI, resting state fMRI (rs-fMRI) typically measures BOLD signals reflecting the brain hemodynamics. In contrast with conventional fMRI, rs-fMRI is a task-free technique that monitors the intrinsic and fundamental activities of the brain neural circuits [10]. ReHo is a method to assess the rs-fMRI data, which is used to measure the BOLD synchrony of a given voxel to those of its nearest surrounding voxels, thus measuring the consistency of neuronal activity in local brain regions in time series [11]. A higher ReHo value reflects a higher local BOLD synchrony, representing an increase in neuron activity in the particular brain area. In contrast, a lower ReHo indicates a decrease in neuron activity in the particular brain area. Our results showed that the FMT rats possessed a lower ReHo in mPFC and hippocampus compared with the NFMT rats, indicating that FMT decreased the functional activities in these two brain areas in the young recipient rats (Figure 6B).

FMT of aged microbiota changes synaptic structures and dendritic spines in young rats

To evaluate the alterations of neuronal structures induced by FMT, we analyzed the effect of FMT on synaptic structures, synaptophysin expression, and density of dendritic spines in mPFC and hippocampus. Compared with NFMT rats, the synaptic structures of FMT rats were significantly changed and similar to the structural characteristics of aged rats, including widened synaptic cleft (mPFC: 18.82 ± 2.30 vs 9.92 ± 1.97, P < 0.01; hippocampus: 15.87 ± 2.30 vs 8.99 ± 2.00, P < 0.01), decrease in the thickness of post-synaptic density (mPFC: 30.79 ± 4.49 vs 44.84 ± 7.58, P < 0.01; hippocampus: 26.41 ± 5.21 vs 47.76 ± 8.85, P < 0.01), as well as a decrease in the curvature of synaptic interface (mPFC: 1.08 ± 0.09 vs 1.17 ± 0.06, P < 0.05; hippocampus: 1.07 ± 0.04 vs 1.23 ± 0.09, P < 0.01, Figure 6C, 6D). In addition, the expression of synaptophysin in mPFC and hippocampus of FMT rats was decreased compared with NFMT rats (P < 0.05, Figure 6E). In mPFC and hippocampus, the dendritic spines of FMT rats decreased significantly compared with NFMT rats (P < 0.05, Figure 6F–6I).

FMT of aged microbiota changes expression of NR1, BDNF, and AGE/RAGE

NR1, an essential subunit of the NMDA receptor, and BDNF are important signaling proteins involved in neuronal proliferation, differentiation, synaptic plasticity, and survival. Immunoblotting and IHC analyses showed that the expression of NR1 and BDNF in mPFC and hippocampus of FMT rats was significantly decreased, and similar to that in aged rats, compared with NFMT rats (P < 0.05, Figure 7A–7F).

Advanced glycation end products (AGEs) and receptor for AGEs (RAGE) accumulate in neurons during normal aging, and have been used as markers of normal aging [12, 13]. Immunoblotting and IHC showed that the expression of AGE and RAGE in mPFC and hippocampus of FMT rats significantly increased compared with NFMT rats, having the characteristics of aged rats (P < 0.05, Figure 7G–7L).

FMT of aged microbiota increases inflammatory cytokine levels in young rats

Having shown that the levels of the inflammatory cytokines IL-1β, TNF-α and IL-6 increase in serum and brain of aged rats (Figure 3A–3F), we asked whether FMT of aged microbiota increases the inflammatory cytokine levels in young rats. Both serum and brain levels of IL-1β, TNF-α and IL-6 were significantly increased in FMT rats compared with NFMT rats (P < 0.05, Figure 8A–8F). These results suggest that inflammation might be one of the mechanisms by which gut microbiota contributes to the cognitive decline in normal aging.

FMT of aged microbiota increases oxidative stress in young rats

Since our results showed that the SOD activity in serum and brain declined, the MDA content increased, and the activity of GSH-PX did not change in aged rats (Figure 3G–3L), we tested whether the FMT of aged microbiota will cause similar changes in young rats. In serum, the SOD activity of FMT rats declined, and the MDA content increased compared with NFMT rats (P < 0.05, Figure 8G, 8I). The SOD activity also declined and the MDA content increased in mPFC and hippocampus of FMT rats compared with NFMT rats (P < 0.05, Figure 8J, 8L). The activity of GSH-PX did not change in serum and brain (Figure 8H, 8K). These results suggest that oxidative stress might represent another mechanism contributing to the cognitive decline in normal aging.

Rats

Young (~ 3 months) and aged (20 ~ 24 months) male SD rats were obtained from Chengdu Dashuo Laboratory Animal Co., Ltd (Chengdu, China). The rats were specific-pathogen-free (SPF) grade animals, and were fed in a barrier SPF environment under 12 h light / dark cycle. Two rats were housed in each cage. All rats were fed autoclaved chow and sterilized water ad libitum. All protocols were reviewed and approved by the Ethics Committee of Shanxi Medical University.

Delayed matching to position (DMTP) task

The tasks were conducted in well-controlled rat two-lever operant chambers (Med Associates, USA). The task consisted of a sample phase and a choice phase, separated by a variable delay. The rats were trained and tested as described previously [37]. During the initial sessions, there was 0-delay between the sample and choice phases. Once a rat reached >80% correct over 3 days, sets of delays were introduced (Set 1: 0, 1, 2, 3, 4, 5, 6 s; Set 2: 0, 1, 2, 4, 8, 12, 16 s; Set 3: 0, 2, 4, 8, 12, 18, 24 s). Upon establishing >80% correct over 3 days in Set 1, rats were progressed to Set 2. Each rat was tested for 5 consecutive sessions in Set 3.

Fecal sample collection and 16S rRNA gene sequencing

The fecal samples were collected directly from the anus of each rat. The total genomic DNA was extracted from each sample by the QIAamp DNA stool mini kit (Qiagen, Hilden, Germany). The extracted DNA was subjected to PCR to amplify the variable V3-V4 region of the 16S locus using the universal bacterial primers 338F/806R on the Illumina MiSeq platform, as described [38].

For bioinformatics analysis, overlapping paired-end reads form original DNA fragments were filtered to obtain clean reads by using QIIME (version 1.17). Then, to get tags, clean reads were paired using FLASH (version 1.2.11). The tags were clustered to Operational Taxonomic Unit (OTUs) by utilizing the software USEARCH (v7.0.1090). To describe the alpha diversity, the numbers of OTUs and Shannon diversity index were calculated using R package (v3.1.1). Nonmetric Multidimensional Scaling (NMDS) analysis based on weighted UniFrac distances was used to display the difference of OTU composition among samples using R package. UPGMA clustering analysis of weighed UniFrac distance was carried out Using QIIME and R package to evaluate the similarity among samples. Next, the OTU representative sequences were classified taxonomically using Ribosomal Database Project (RDP) Classifier v.2.2, using 0.8 confidence values as cutoff. Statistical analyses were conducted in R (v3.1.1).

Resting-state functional magnetic resonance imaging

All rats were scanned using 7.0T Bruker Scanner MRI system (Pharmascan, Bruker, Switzerland). T2-weighted anatomical images were acquired first, using a T2_TurboRARE sequence with the following parameters: repetition time (TR) = 5,300 ms, echo time (TE) = 33 ms, number of averages = 2, field of view = 35 mm × 35 mm, slice thickness = 0.55 mm, slices = 50, matrix = 256 × 256, flip angle = 90°. Functional scans were acquired using a T2Star_FID_EPI sequence with the following parameters: TR = 2000 ms, TE = 20 ms, number of averages = 1, field of view = 35 mm × 35 mm, slice thickness = 0.55 mm, slices = 50, matrix = 64 × 64, flip angle = 90°, 600 Repetitions. All original Bruker images were converted to DICOM format with the Paravision.

Preprocessing of all acquired resting-state data and the ReHo analysis were carried out using Statistical Parametric Mapping (SPM8, http://www.fil.ion.ucl.ac.uk/spm/software/spm8) and Data Processing Assistant for Resting-State fMRI (DPARSF, http://www.restfmri.net/forum/DPARSF). The dataset was spatially normalized to the standard rat brain atlas [39]. As described previously [23], ReHo analysis was performed using the REST software (http://restfmri.net/forum/REST). Individual ReHo maps were generated by calculating Kendall’s coefficient of concordance of the time series of a given voxel with those of its nearest 26 neighboring voxels. Then, all ReHo maps were smoothed with an isotropic Gaussian kernel of 2 mm (full-width half-maximum). The two sample t-test was performed to determine the difference between two groups. Clusters that were significantly different were selected by setting P < 0.005 with a Gaussian random fields (GRF) correction and clusters more than 50 voxels.

Fecal microbiota transplantation (FMT)

Before transplantation, the FMT rats received an antibiotic mixture consisting of ampicillin (180 mg/kg/d), vancomycin (72 mg/kg/d), metronidazole (90 mg/kg/d) and imipenem (90 mg/kg/d), twice daily (12 h interval) for three consecutive days [40]. Transplantation was conducted at day 4 (at least 24 h from the last antibiotic treatment). To prepare suspension of fecal bacteria for the transplantation, the fresh fecal contents of aged rats were surgically extracted, pooled, and diluted with PBS, as described [41]. The suspension was centrifuged at 800 rpm for 5 min. The supernatant was taken and given to the recipient rats by gavage immediately. The rats underwent FMT once a day for three days, then twice a week for two months.

Golgi staining

According to the instruction of FD Rapid Golgi Stain Kit (FD Neuro Technologies, Inc. USA), the tissues were cut into 100 μm-thick slices and mounted on gelatin-coated slides. After staining, the sections were examined under a light microscope. All images were analyzed by Image J software. The dendritic Spine density was expressed as spine number per 10 μm dendritic length.

Transmission electron microscopy

The synaptic ultrastructure features were observed under transmission electron microscope (JEM -100CX, JEOL, Japan), and gray type-I synapses were analyzed. The width of synaptic gap, the thickness of PSD, and the curvature of synaptic interface were quantitatively analyzed with Image J, as described [42].

Western blotting

Equal amounts of protein were separated by SDS-PAGE and transferred to a PVDF membrane. After blocking with 5% BSA, immunoblots were incubated with the following specific primary antibodies: NMDAR1 (1:1000, ab68144, Abcam), BDNF (1:1000, ab216443, Abcam), Synaptophysin (1:20000, ab32127, Abcam), AGEs (1:500, ab23722, Abcam), RAGE (1:500, ab3611, Abcam), IL-1β (1:500, ab9722, Abcam), TNF-α (1:1000, ab6671, Abcam), IL-6 (1:1000, ab9324, Abcam) and β-actin (1:400, BM0627, Boster). After incubation with HRP-labeled secondary antibodies, the bands were visualized using the ECL detection kit. The protein bands were quantitatively analyzed by using Image Lab software 5.2.

Immunohistochemistry

Paraffin-embedded tissues were cut into 4 μm-thick sections, and incubated with a primary antibody, followed by enzyme-labelled secondary antibody. The primary antibodies used were: NMDAR1 (1:100, ab68144, Abcam), BDNF (1:500, ab216443, Abcam), AGE (1:100, ab23722, Abcam), and RAGE (1:20, ab3611, Abcam). After incubation with a secondary antibody, the sections were visualized by using 3,3′-diaminobenzidine (DAB). Images were photographed and analyzed with Digital Pathology Solutions (Scanscope CS, Aperio, USA).

Enzyme-linked immunosorbent assay (ELISA)

Serum was collected by centrifugation at 1000 g for 10 min at 4°C. IL-1β, TNF-α, and IL-6 serum levels were analyzed by commercially available ELISA kits (SHANGHAI WESTANG BIO-TECH).

Analysis of oxidative stress level

Serum was collected as above, and the activities of SOD (ab65354, Abcam) and GSH-PX (ab102530, Abcam), and the level of MDA (ab118970, Abcam) were analyzed according to the manufacturer’s instructions.

The medial prefrontal cortex and hippocampus were homogenized, and the homogenates were centrifuged at 3000 rpm for 15 min. The supernatants were collected, and analyzed for the activities of SOD (A001; Nanjing Jiancheng Bioengineering Institute, Nanjing, China), GSH-PX (A005; Nanjing Jiancheng Bioengineering Institute, Nanjing, China), and the content of MDA (A003; Nanjing Jiancheng Bioengineering Institute, Nanjing, China).

Statistical analysis

Data are presented as means ± SEMs. Statistical analysis was performed by using SPSS 13.0. Two-factor repeated-measures ANOVA was used to analyze the performance of DMTP Task. To compare the data between two groups, t-test was used; one-way ANOVA was performed to analyze data from three or more groups. Difference was considered significant if P < 0.05.