Integrated Multi-omics Approach Reveals the Interaction of Brain-gut in D-galactose-induced Aging Model Mice

WANG Jing, HAN Le, KANG Jia-Chao, MENG Jie, CHEN Dong-Mei,WU Ping-Min, TIAN Yi-Hong, DUAN Yong-Qiang

(1)School of Public Health, Gansu University of Chinese Medicine, Lanzhou 730000, China;2)Clinical Laboratory Center, Gansu Provincial Maternity and Child-care Hospital, Lanzhou 730000, China;3)Departments of Laboratory Diagnosis, Daqing Oilfield General Hospital, Daqing 163000, China;4)Teaching Experiment Training Center, Gansu University of Chinese Medicine, Lanzhou 730000, China;5)School of Chinese Medicine, Ningxia Medical University, YinChuan 750004, China)

Abstract Multiple organs are physiologically and pathologically interconnected during aging, and the brain plays a central role in this process. There is a direct two-way communication between the brain and the gut called “brain-gut interaction”, which is of great significance for the study of aging, and the molecular mechanism remains to be further studied. The aim of this study is to explore the mechanism of aging in the context of brain-gut interaction. The results of general physical signs of mice showed that the amount of exercise decreased, body weight and food intake decreased significantly in aged mice (P<0.001, P<0.05). The thymus index of aged mice was significantly lower than that of normal mice (P<0.05), and the thymic pathological results showed that the thymic cortex of aging mice was thinner, the boundary between medulla and cortex was blurred, and the cells were loosely arranged. Metabolomics analysis revealed 317 differential metabolites in feces and 100 differential metabolites in hippocampus. The results of microbiome showed that Bacteroidetes and Firmicutes were the dominant phyla of gut microbiota. Bacteroidetes showed an upward trend and Firmicutes showed a downward trend after aging. KEGG pathway results showed that 26 metabolic pathways were related to the study of aging, among which galactose metabolism, ABC transporter and purine metabolism were of great significance for the brain-gut interaction. The results of Spearman correlation analysis of the three groups showed that the types of metabolites involved were mainly lipids and lipid-like molecules and organic acids and derivatives, and the gut microbiota involved were mainly Bacteroidetes and Firmicutes. In conclusion, the present study demonstrated that the synergistic changes between brain and gut in aging mice were related to the mechanism of aging, which provided new insights into the mechanism of aging process.

Key words metabolomics; 16S rDNA; D-galactose; aging; interaction of brain-gut

Aging is accompanied by a decline in functional reserves of multiple organs[1]. Brain aging plays a central role in disease pathogenesis and incidence[2]. The interaction between the brain and digestive tract during aging is varied and complex[3]. During aging, the contractility of intestinal smooth muscle changes, as do regulatory interactions between the nervous and gastrointestinal systems[4].For example, ghrelin is a gastrointestinal hormone that is also linked to protection of neuronal viability, mood regulation, and promotion of neural plasticity. These effects are associated with homeostasis and have been implicated in various diseases. D-galactose-induced aging model is a commonly used animal model for aging research[5], many studies showed that long-term exposure to D-galactose causes aging characteristics that are similar to those of natural aging[6].

Recent studies have found that the gut is closely related to brain development and function[7,8]. Direct bidirectional communication between the brain and gut is known as the brain-gut axis and is of great significance in the study of aging[9]. Although metabolomics and 16S ribosomal deoxyribonucleic acid (rDNA) sequencing are often used to analyze metabolites and gut microbiota, single omics technology has systematic limitations. For example, Herteletal[10]proposed the use of metabolomics for biological age prediction, which can be used to predict patient prognosis and may have applications in other fields of medicine. Additionally, Yuetal[11]used microbiome to find that healthy adults aged 20-50 years have differences in gut microbiota, and suggested that effective interventions in gut microbiota may help to delay aging and prevent diseases. However, the changes in metabolites or gut microbiota alone are not able to explain all aging-related changes across the lifespan[12].

The joint application of multiple omics technologies is therefore imperative. The use of multi-level strategies and methods can simplify complex data and ensure the accuracy and consistency of results[13-15]. Previous brain-gut axis and aging research has mainly focused on omics analysis of brain and gut microbiota metabolites. However, because of the large number, variety, and complex structure of metabolites and gut microbiota, these studies have not been comprehensive[16,17]. We therefore used a combined omics approach to explore potential mechanisms of aging by examining changes in metabolites and gut microbiota.

Therefore, the present study used a combined omics approach to explore potential mechanisms of aging by examining changes in metabolites and gut microbiota in a D-galactose-induced aging mouse model. Applying bioinformatics methods, the correlations between the changes of fecal metabolites and gut microbiota, hippocampal metabolites and gut microbiota, as well as fecal metabolites and hippocampal metabolites, were characterized in the context of brain-gut interactions.

1 Materials and Methods

1.1 Animals and sample collection

In this study, 20 male 8-week-old specific pathogen-free grade C57BL/6 mice (Scientific Experimental Animal Center of Gansu University of Chinese Medicine, Lanzhou, Gansu, China) were used (Gansu Experimental Animal Quality Certificate License, SCXK (Gan) 2020-0001, No. 62001000000652).The average weight of the mice was 18-22 g. All mice were divided into 4 cages and they were raised under light at 22-25 ℃ with water and food provided freely (Gansu Experimental Animal Facility Use License, SYXK (Gan) 2020-0009, No. 00000628). All experimental procedures and protocols were in accordance with the guidelines of the Institutional Animal Ethics Committee of Gansu University of Chinese Medicine (No. 2020-309).

A total of 20 mice were randomly divided into the normal group (N group,n=10) and aging group (A group,n=10). Mice in the aging group were intraperitoneally injected with 10 g/L D-galactose per mouse per day (Injection volume, 0.02 mL/g). Mice in the normal group were given the same amount of normal saline. Continuous modelling was performed for 42 days. During modeling, the activity degree and exercise amount of mice in each group was observed. Body weight was measured every two days and food intake every three days. D-galactose was purchased from Sigma (St. Louis, Missouri, United States, Lot number: 10131055), and 10 g/L was prepared with normal saline before use. The feces samples were collected after modeling, and the mice were anesthetized by intraperitoneal injection with 3% sodium pentobarbital at a dose of 30 mg/Kg. After anesthetized, the mice were perfused for fixation, and tissues were collected. When the mice could not drink, eat and move freely, the humanized endpoints were considered to be reached, and then the mice were euthanized by cervical dislocation. And then the thymus and hippocampus samples from mice were collected. The carcasses of mice were delivered to the Laboratory Animal Center of Gansu University of Chinese Medicine for unified and harmless disposal.

1.2 Thymus index and Hematoxylin and eosin (H&E) staining

Thymus tissues were washed with saline, drained using filter paper, and weighed using an electronic balance. The thymus index was then calculated for each mouse as follows: Thymus index (%) = thymus weight (g) / body weight (g)×100%. To observe the pathological and morphological changes in thymus tissue in each group of mice, the thymus tissue of each group was placed in 4 % paraformaldehyde solution. Thymus tissues were then fixed at room temperature, embedded in paraffin, and cut into 3 μm slices. After dewaxing, washing, and performing H&E staining, the slices were observed under a microscope.

1.3 Fecal and hippocampus untargeted metabolomics

ACQUITY UPLC I-Class system (Waters Corporation, Milford, USA) coupled with VION IMS QTOF Mass spectrometer (Waters Corporation, Milford, USA) was used to analyze the metabolic profiling. An ACQUITY UPLC BEH C18 column (1.7 μm, 2.1 mm×100 mm) were employed. The flow rate was 0.4 mL/min, and column temperature was 45 ℃. All the samples were kept at 4 ℃ during the analysis. The injection volume was 1 μL.

Metabolomics data processing software Progenesis QI v2.3 was used to process the study data. Principle Component Analysis (PCA) was used to observe the overall distribution among the samples and the stability of the whole analysis process. Partial Least-Squares-Discriminant Analysis (PLS-DA) and Orthogonal Partial Least-Squares-Discriminant Analysis (OPLS-DA) were utilized to distinguish the metabolites that differ between groups. Variable Importance of Projection (VIP) values obtained from the OPLS-DA model were used to rank the overall contribution of each variable to group discrimination. A two-tailed Student’sT-test was further used to verify whether the metabolites of difference between groups were significant. Differential metabolites were selected with VIP values larger than 1.0 andP-values less than 0.05. In order to more intuitively display the relationship between samples and the expression differences of metabolites among different samples, hierarchical clustering was conducted to analyze the top 50 differential metabolites selected according to VIP values in feces and in hippocampus.

1.4 16S rDNA analysis

The genomic DNA of gut microbiota was extracted by DNA extraction kit, and then the purity and concentration of DNA were detected by agarose gel electrophoresis. The 16S rDNA gene of the microbes were amplified with the following primers: forward F 5′-TACGGRAGGCAGCAG-3′ and reverse R 5′-AGGGTATCTAATCCT-3′. The library sequencing was conducted by OE biotech Co., Ltd. (Shanghai, China). The original image data files obtained by high-throughput sequencing were analyzed by Base Calling, and the original two-ended sequences were generated by Illumina MiSeq sequencing, which was called raw data. The results were stored in FASTQ file format, which contained sequence information of sequencing reads and their corresponding sequencing quality information. Cutadapt software was used to cut out primer sequences for raw data sequences. Then qualified two-ended raw data quality filtering, noise reduction, mosaic and chimera and other quality control analysis, to get the representative sequence and the valid tags for later Operational Taxonomic Unit (OTU) partitioning. QIIME software package was used to select representative sequences of each OTU, and all representative sequences were compared and annotated with the database. Species alignment was annotated using BLAST software. Alpha diversity was used to evaluate the differences in bacterial diversity between the two groups. Linear discriminant analysis (LDA) coupled with effect size measurements (LEfSe) analysis revealed the composition of two or more groups of different species in the biome. These results were used to roughly compare the degree of sample dispersion within the group and the difference of the index between different groups. Total 16S rDNA sequencing data were uploaded to SRA database at the NCBI (https://www.ncbi.nlm.nih.gov/Traces/study/, accession no. PRJNA901653). Please note that the PRJNA901653 database is freely available and was released on December 6, 2022.

1.5 Data mining

Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis was performed for differential metabolites in feces and hippocampus of mice to investigate the biological functions.P-value <0.05 was used as the threshold for screening, and the results of the first 20 pathways were selected to display.

1.6 Association analysis

Univariate correlation analysis was used to integrate omics data. In this study, Spearman correlation analysis was applied to assess the possible relationships between gut microbiota and metabolites. The following three correlation analyses were performed by using Spearman analysis: the correlation between differential metabolites in feces and the top 30 (genus level) gut microbiota, the correlation analysis between differential metabolites in hippocampus and the top 30 (genus level) gut microbiota, and the correlation analysis between differential metabolites from hippocampus and feces. Bioinformatic analysis was performed using the OECloud tools at https://cloud.oebiotech.cn. The results of the correlation analysis were presented in the form of heatmap and network graph. After Spearman correlation analysis, an Excel table was formed containing key information (Correlation pairs, Correlation, andP-value). Then pick the key correlation pairs to form Tables 1-3, withP-value <0.01 and ∣Correlation∣ >0.5 as the criterion.

1.7 Statistical analysis

Data were analyzed using SPSS 26.0 statistical software, and expressed as the means ± standard deviation (SD). Additionally, Student’s unpairedT-test was used for comparison between the two groups.Pvalue <0.05 was considered to indicate a statistically significant difference.

2 Results

2.1 The decreases of body weight and food intake in D-galactose-induced aging mice

The exercise volume of mice in each group was observed, and it was found that compared with the normal group, the exercise ability was weakened and the exercise volume was reduced of mice in the aging group. Moreover, initial body weight did not significantly differ between the normal and aging groups. However, body weight (P<0.001) and daily food (P<0.05) intake were significantly lower in the aging group than normal group after the modelling (Fig.1).

Fig.1 Body weight and food intake between the normal group and the aging group (A) Body weight before and after the modelling. (B) Daily food intake between the normal group and the aging group induced by D-galactose. Data were shown as means ± SD (n=10). P values were calculated by Student’s unpaired T-test.*P <0.05 vs. normal group,*** P <0.001 vs. normal group

2.2 The pathological changes of thymic in D-galactose-induced aging mice

The thymus index was significantly lower in the aging group than the normal group (P< 0.05)(Fig.2A). In the normal group mice, the thymus cortex was relatively thick and intensely stained. In addition, lymphocytes in the cortex were arranged in an orderly fashion and thymic epithelial cells were scattered throughout the medulla. The cells grew vigorously, the capsule was intact, and overall cell distribution and arrangement were orderly. Thymus tissue in the aging group had a slightly thinner cortex and lighter staining, furthermore, the medulla-cortex boundary was blurred and cells were arranged more loosely (Fig.2B).

Fig.2 The pathological changes of thymus tissue between the normal group and the aging group (A) Thymus index between the normal group and the aging group induced by D-galactose. Data were shown as means ± SD (n=10). P values were calculated by Student’s unpaired T-test.*P <0.05 vs. normal group. (B) The pathological changes of thymus tissue between the normal group and the aging group induced by D-galactose (H&E ×10). Thymus tissues were stained with H&E for microscope examination of thymic pathological changes. Medulla (▲), cortex (★)

2.3 Characteristic changes of fecal and hippocampal metabolite in D-galactose-induced aging mice

The PLS-DA model diagrams of fecal and hippocampal metabolites revealed that the samples from the normal and aging groups could be clearly separated, indicating a significant change in metabolic profile with age (Fig.3A, B). Metabolomics analysis showed 317 differential metabolites in feces and 100 in the hippocampus. As shown in Fig.3C, in the feces of the mice, red indicated the metabolites with high expression, and green indicated the metabolites with low expression, so the expression of these metabolites in the aging group and the normal group were opposite. Similarly, as shown in Fig.3D, in the hippocampus of mice, metabolites that were highly expressed of the normal group were poorly expressed in the aging group, so the expression of these metabolites in the aging group and the normal group were also opposite.

Fig.3 Metabolomics analysis of fecal and hippocampal metabolites between the normal group and the aging group (A) The PLS-DA model diagram of fecal metabolites between the normal group and the aging group induced by D-galactose (n=10). (B) The PLS-DA model diagram of hippocampal metabolites between the normal group and the aging group induced by D-galactose (n=10). (C) Hierarchical clustering result of the top 50 differential fecal metabolites between the normal group and the aging group induced by D-galactose (n=10). The differential metabolites were selected on the basis of the combination of a statistically significant threshold of VIP value and P-value, in which VIP value> 1.0, and P value<0.05. (D) Hierarchical clustering result of the top 50 differential hippocampal metabolites between the normal group and the aging group induced by D-galactose (n=10). The differential metabolites were selected on the basis of the combination of a statistically significant threshold of VIP value and P-value, in which VIP value> 1.0, and P value<0.05

2.4 Characteristic changes of microbiome profiles of D-galactose-induced aging mice

A flower map of the OTU distribution showed the number of OTUs in each sample and the total number of OTUs in all samples (Fig.4A).BacteroidetesandFirmicuteswere the dominant phyla of gut microbiota. Both Chao1 richness (P<0.05) (Fig.4B) and the Shannon index (measures of alpha diversity) (P<0.01) (Fig.4C) were significantly increased in aging group compared with normal group. LEfSe analysis revealed the composition of two or more groups of different species in the biome, which showed that the structure and composition of gut microbiota changed significantly with age (Fig.4D). A histogram showing the top 15 species in abundance order at the taxonomic levels of phylum, class, order, family, and genus from different groups was presented in Fig.5. At the phylum level,BacteroidetesandFirmicuteswere the dominant bacteria of mice (Fig.5A), in which the levels ofBacteroidetes andFirmicuteswere up- and downregulated, respectively, after aging. The same was true for the other four taxonomic levels, where the types of gut microbiota were not significantly different between the two groups of mice, but the number of the first two dominant species was upregulated or downregulated after aging, respectively.

Fig.4 Microbiome analysis of the gut microbiota between normal group and the aging group (A) Flower map of OTUs distribution of gut microbiota between the normal group and the aging group induced by D-galactose (n=10). (B) Chao1 richness between the normal group and the aging group induced by D-galactose (n=10). The values of Chao 1 richness were the maximum, minimum, upper quartile, lower quartile, and median of the data from the normal group and the aging group induced by D-galactose. P values were calculated by Student’s unpaired T-test.* P <0.05 vs. normal group. (C) Shannon index between the normal group and the aging group induced by D-galactose (n=10). The values of Shannon index were the maximum, minimum, upper quartile, lower quartile, and median of the data from the normal group and the aging group induced by D-galactose. P values were calculated by Student’s unpaired T-test.** P <0.01 vs. normal group. (D) The LDA scores (log 10) of dominant gut microbiota between the normal group and the aging group induced by D-galactose (n=10)

Fig.5 The histogram of species distribution at the phylum (A), class (B), order (C), family (D) and genus (E) levels between normal group and the aging group induced by D-galactose Different colors represent different bacteria at phylum, class, order, family, and genus levels

2.5 The pathway analysis of differential fecal metabolites and differential hippocampal metabolites

The KEGG pathway enrichment results of the fecal and hippocampal differential metabolites are shown in Fig.6. In this figure, the significant of signal pathway was represented by the height of the bar, when the top of the bar was higher than the blue line, the signal pathway represented by it was significant. The differential metabolites between the feces and hippocampus were involved in 177 metabolic pathways. In the fecal differential metabolites, six metabolic pathways including Galactose metabolism and ATP-binding cassette (ABC) transporters were significantly enriched in the aging group compared with the normal group. In the hippocampal differential metabolites, twenty metabolic pathways including Purine metabolism and Retrograde endocannabinoid signaling were significantly enriched in the aging group compared with the normal group.

Fig.6 KEGG database was used to analyze the results of the top 20 pathways of differential metabolites in feces and hippocampus (A) Histogram of pathway enrichment analysis of differential metabolites in feces. (B) Histogram of pathway enrichment analysis of differential metabolites in hippocampus. KEGG (https://www.kegg.jp/) database was used for pathway analysis. The P-value in the metabolic pathway represented the significance of enrichment. The red line indicated a P-value of 0.01, and the blue line indicated a P-value of 0.05

2.6 Correlation analysis between differential fecal metabolites and gut microbiota

The top 50 differential fecal metabolites associated with gut microbiota at the genus level are shown in Fig.7. Among these, 70 pairs (4.7%) were significant, 28 (40.0%) pairs were positively correlated and 42 (60.0%) were negatively correlated. The first five key pairs are shown in Table 1, and Spearman correlation analysis was used. The metabolite types that correlated with gut microbiota were mainly lipids and lipid-like molecules (32.4%), Organic acids and derivatives (17.6%), and Organic oxygen compounds (11.8%). The gut microbiota that correlated with metabolite types were mainlyOdoribacter(15.7%),Lachnoclostridium(14.3%), andRikenellaceae_RC9_gut_group(11.4%). According to coefficient, “Alanyl-Proline-Lachnoclostridium” and “Phenylalanylproline-Lachnoclostridium” were found as the key correlated pairs in this study (Table 1).

Table 1 The top5 pairs of correlation analysis between fecal metabolites and gut microbiota

2.7 Correlation analysis between differential hippocampal metabolites and gut microbiota

The top 50 differential hippocampal metabolites associated with gut microbiota at the genus level are shown in Fig.8. Among these, 133 pairs (8.9%) were significant,56 pairs (42.1%) were positively correlated and 77 (57.9%) were negatively correlated. The first five key pairs are shown in Table 2. Spearman correlation analysis was used for statistical analysis in Table 2, and Spearman correlation analysis was used. The metabolite types that correlated with gut microbiota were mainly Lipids and lipid-like molecules (48.8%), Nucleosides, nucleotides, and analogs (14.6%), and Organic acids and derivatives (14.6%). The gut microbiota that correlated with metabolite types were mainlyParasutterella(25.6%),Lachnospiraceae_NK4A136_group(12.0%),Bacteroides(9.0%), andLachnospiraceae_UCG-001 (9.0%). According to Spearman coefficient,“PC(20∶4(5Z,8Z,11Z,14Z)/18∶0)-Rikenellaceae_RC9_gut_group”, “LysoPC (16∶0)-Parasutterella” and “Adenosine-Parasutterella” were found as key correlated pairs in this study (Table 2).

Fig.8 Spearman analysis was used to analyze the correlation between differential hippocampal metabolites and gut microbiota Red and blue indicated positive and negative correlations respectively. Spearman correlation analysis was used for the data, and bioinformatic analysis was performed using the OECloud tools at https://cloud.oebiotech.cn.* P<0.05,** P<0.01,*** P<0.001

2.8 Correlation analysis between fecal metabolites and hippocampal metabolites

Differential fecal metabolites associated with differential hippocampal metabolites are shown in Fig.9. Among these, 799 pairs (32.0%) pairs were significant, 390 pairs (48.8%) were positively correlated, and 409 (51.2%) were negatively correlated. The first five key pairs are shown in Table 3, and Spearman correlation analysis was used. The metabolite types that were correlated were mainly lipids and lipid-like molecules (30.5%), Organic acids and derivatives (17.9%), and Organic oxygen compounds (12.6%). According to Spearman coefficient, “7-oxolithocholic acid-glycolic acid” and “Allose-PI (16∶2(9Z,12Z)/22∶3(10Z,13Z,16Z))” were found as key correlated pairs in this study (Table 3).

Table 3 The top5 pairs of correlation analysis between differential fecal metabolites and differential hippocampal metabolites

3 Discussion

The studies of the brain-gut axis have elucidated the interactions between the gut and brain. The body weight, food intake, thymus index, and pathological examination findings in our study indicated successful establishment of the aging model[18,19], as general signs, aging-related indicators, and tissues degraded with age.

In this study, a total of 26 metabolic pathways were found to be relevant to the study of aging, among which Galactose metabolism, ABC transporters and Purine metabolism were of great significance for the brain-gut axis. Galactose metabolism modifies glycolipids and glycoproteins also plays an important role in the generation of intracellular energy[20]. Galactose is associated with cognitive development and can be affected by hyperlactatemia[21]. In addition, Wilms Eetal[22]found that oligomeric galactose can increase fecal bifidobacteria in older adults. Thus, galactose may regulate the relative abundance of related bacteria by affecting intestinal carbohydrate digestion and utilization. ABC transporters are widely expressed throughout the brain, they provide the brain with nutritional requirements and play a key role in central nervous system diseases such as Alzheimer’s disease[23]. Ou L,etal.[24]reported that ABC transporters participated in horizontal gene transfer, which is involved in bifidobacteria evolution, therefore, they likely play a role in generating diversity among gut microbiota. Lennicke C,etal.[25]demonstrated that excessive dietary sugar promoted the development of metabolic disorders that induce purine metabolism and therefore affected life expectancy. Furthermore, Lactobacillus in the gut regulate intestinal purine metabolism in the host[26], which suggests that Lactobacillus regulation may interfere with aging processes.

In this study, the fecal metabolites Alanyl-Proline and Phenylalanylproline were significantly negatively correlated with gut microbiota “Lachnoclostridium”. It was reported thatLachnoclostridiumaffects the metabolism of the body, by influencing brain physiology and behavior[21]. Alanyl-Proline and Phenylalanylproline belong to Organic acids and derivatives. Some metabolites produced by probiotics, such as organic acids protect the gut’s epithelial barrier[27]. Unfortunately, the high content in organic acids could also cause undesirable intestinal side effects in some cases[28]. Irritable bowel syndrome (IBS) is disorder that relates to brain-gut interactions, studies have shown that IBS patients may have abnormal gut microbiota as well as increased organic acids[29]. Thus, some microbiota and organic acids exhibit potential causal relationships, and increased organic acid levels have been associated with enteric diseases.

Hippocampal metabolites “PC(20∶4(5Z,8Z,11Z,14Z)/18∶0) and gut microbiota “Rikenellaceae_RC9_gut_group” was found as positively correlated pairs in this study. The metabolites PC(20∶4(5Z,8Z,11Z,14Z)/18∶0) belonged to lipids and lipid-like molecules, which were reported in inflammation, pain, and central nervous system diseases[30]. Besides, it was reported that gut microbiotaRikenellaceae_RC9_gut_groupwas high related to the development of Parkinson’s disease by affecting amino acid metabolism[31]. The finding in study coincided with these reports, indicating thatRikenellaceae_RC9_gut_groupmay be associated with brain aging by interaction with hippocampal metabolites. In this study, hippocampal metabolites “LysoPC(16∶0)” was found to be correlated with gut microbiota “Parasutterella”. Li JW,etal.[32]Reported thatParasutterellaregulated the hypothalamic-pituitary-adrenal axis as well as the immune response and gut barrier. Hippocampal metabolites Adenosine was involved in brain diseases, such as epilepsy, neurodegenerative disorders, psychiatric conditions, and cerebrovascular ischemia[33]. These correlations indicated that there is a synergistic change between brain and gut, which is involved in the mechanism of aging.

Fecal metabolite “7-oxolithocholic acid” belongs to lipids and lipid-like molecules, and it was found to be negatively correlated with hippocampal metabolites “glycolic acid”, which belongs to organic acids and derivatives. It was reported that glycolic acid was effective in reversing the signs of aging and photodamage, and glycolic acid also participated in the induction of antioxidant activity and ceramide biosynthesis[34, 35]. Thus, the correlation of glycolic acid with fecal metabolite in this study is a relative mechanism of glycolic acid in inducing aging. Allose is a fecal metabolite that belongs to organic oxygen compounds, which is reported to have therapeutic potential against brain ischema-reperfusion injury by attenuating blood-brain barrier disruption[36]. This study found that allose is negatively correlated with PI (16∶2(9Z,12Z)/22∶3(10Z,13Z,16Z)), a lipid molecule in hippocampal. These findings showed that the correlated metabolites present in the brain and gut are relevant to the study of aging.

In conclusion, this study used muti-omics approach to profilethe metabolites changes of fecal and hippocampal, as well as the changes in gut microbiota during D-galacose induced aging of C57BL/6 mice. Integrating omics data and bioinformatics methods, the correlations between changes of fecal metabolites and gut microbiota, fecal metabolites and hippocampal metabolites, as well as hippocampal metabolites and gut microbiota, were characterized in the context of aging process. According to three groups of correlation analysis, the metabolite types involved were mainly lipids and lipid-like molecules, organic acids and derivatives, and the gut microbiota involved were mainlyBacteroidetesandFirmicutes. This indicated that the above substances and the association networks they participate were of great significance for the study of aging and aging-related mechanisms. The results of this study suggested that muti-omics approach could be used to assist the identification of intrinsic interaction underlying the brain-gut axis. Examining correlations between the brain and gut provided a multi-tiered approach for revealing the mechanism of aging. At the same time, based on the data of this study, we hope to develop drugs to intervene the association pair in the future to verify whether there is an interaction between the association pair, which will open a new view for drug development and further in-depth exploration.

Acknowledgements

The authors thank Pro.Yong Huang of Gansu University of Chinese Medicine for his help in histopathological analysis. We also thank Liwen Bianji(Edanz)(https://www.liwenbianji.cn)for editing the language of a draft of this manuscript.