Abstract

   The gut microbiota of centenarians plays a vital role in promoting
   healthy longevity. We performed a cross-sectional study of 224 people
   from Jiaoling, China, which is globally recognised for the longevity of
   its residents. Compared with younger people, centenarians showed
   significantly increased alpha-diversity, enrichment of the beneficial
   bacteria Lactobacillus, Akkermansia, and Christensenella, and increased
   redox capacity in the gut microbiota. Serum metabolomics of
   centenarians showed significant enrichment of antioxidant metabolites,
   including L-ascorbic acid 2-sulphate and lipoic acid. Finally, we
   isolated and screened a strain of Lactobacillus plantarum 124 (LP124)
   with a good antioxidant effect on the gut microbiota of centenarians.
   Animal experiments further verified that mesaconic acid from LP124
   regulates the gut microbiota, is anti-inflammatory, relieves oxidative
   stress, maintains the intestinal barrier, and is the best-known
   anti-aging molecule. LP124 derived from the gut microbiota of
   centenarians and its metabolite mesaconic acid, have a significant
   positive effect on health and longevity.

   Subject terms: Metagenomics, Microbiome

Introduction

   Aging of the global population is becoming a serious problem. Evidence
   exists that human gut microbiota plays a key role in human health and
   chronic diseases^[68]1,[69]2. Therefore, the gut microbiota is an
   important factor in promoting human health^[70]3,[71]4. Evidence from
   an increasing number of studies suggests that gut microbiota disorders
   are key factors that lead to aging and chronic disease^[72]5–[73]7, and
   that the steady state of the gut microbiota is a healthy aging
   regulator^[74]8–[75]14. Therefore, the reconstruction of gut microbiota
   homoeostasis may be an important target for improving health and
   longevity.

   Centenarians provide a good natural model for studying longevity and
   aging. The composition of microbiota in centenarians plays an important
   role in healthy aging. Multiple cross-sectional studies have shown that
   as people age, their microbiota shows diverse changes compared with
   younger controls, with fewer beneficial microbes and more opportunistic
   disease-causing microbes^[76]15,[77]16. However, the core microbiota
   composition is reduced in centenarians and health-related bacteria
   (including Lactobacillus) are enriched^[78]17. In addition, locus
   cohort studies of people of different age groups and population
   locations have shown that novel bile acid biosynthesis pathways are
   abundant in the microbiota of centenarians and that the gut microbiota
   of centenarians has an exogenous material degradation
   function^[79]18,[80]19. Longevity-related gut microbiota (such as
   Christensenellaceae, Porphyromonadaceae and Rikenellaceae) and
   functions (such as xenobiotic degradation) play important roles in
   healthy longevity^[81]18–[82]23. Although these studies established an
   association between gut microbiota and longevity, they lacked further
   responses to these findings. Key beneficial or detrimental strains were
   not isolated, identified, preserved, and screened. Furthermore, the
   anti-aging function of key strains was not verified in vitro and in
   vivo. Finally, the key substance for the anti-aging effect of key
   strains was not determined and verified. Therefore, more detailed
   comparisons, larger sample sizes, longer trajectories, and further
   strain isolation screening and validation are needed to determine the
   function of the microbiome in centenarians and anti-aging measures to
   clarify the relationship between gut microbiota and health and
   longevity.

   Previously, we observed that the gut microbiota of long-lived
   individuals contained a high abundance of health-related
   Lactobacillus^[83]7. In this regard, we targeted the isolation of
   Lactobacillus in the gut microbiota of a long-lived township population
   and screened and verified Lactobacillus plantarum 124 (LP124) with good
   anti-aging function through in vitro screening and in vivo verification
   methods. Metabolomics analysis showed that the serum levels of
   mesoconic acid (MA), lysine (Lys), and butyrate (But) in mice
   administered with LP124 were significantly higher than those in the
   model group. To explore the potential anti-aging role of LP124, we
   studied the anti-inflammatory and anti-aging effects of MA, Lys, and
   But on the aging process.

   We performed a cross-sectional study involving 224 participants that
   included centenarians and other age groups (20–110 years). We examined
   the gut microbiota structure and function of all participants using
   metagenome analysis and characterised the serum differential
   metabolites of all participants using metabolome analysis. Based on our
   previous results, we further verified the anti-inflammatory and
   anti-aging effects of mesaconic acid, lysine, and butyrate on the
   metabolites of Lactobacillus plantarum 124 (LP124), a longevity-related
   strain, through animal experiments. This study aimed to explore that
   gut microbiota, especially gut probiotics and their metabolites, have a
   positive effect on health and longevity in long-lived people.

Results

Composition and functional characteristics of gut microbiota in long-lived
people of different ages

   According to age, the population in the long-lived areas was divided
   into middle-aged (Y40: 21–40 years and Y60: 41–60 years), older adults
   (Y80: 61–80 years and Y100: 81–99 years), and centenarian (Y120:
   100–110 years) groups. The top 35 relative abundance of the gut
   microbiota at the genus level were shown in Fig. [84]1A. Principal
   component analysis (PCA) identified two of the most variable classes
   across all populations, Bacteroides and Parabacteroides, which are
   major contributors to the human gut type (Fig. [85]1B). A constructed
   taxonomic phylogenetic tree showed that the genera level species
   abundance of the Y40, Y60, Y80, and Y100 groups differed significantly
   from that of the Y120 group (P < 0.05). Summarising the species
   selected by comparing Y40 vs. Y120, Y60 vs. Y120, Y80 vs. Y120, and
   Y100 vs. Y120 groups identified 281 distinct species (Fig. [86]1C).
   Diversity characteristics of the gut microbiota in long-lived people of
   different ages were examined. According to Chao1 index values, the gut
   microbiota diversity of the centenarian group (Y120) was significantly
   higher than that of the middle-aged groups (Y40 and Y60) (Fig.
   [87]2A–D). Principal co-ordinates analysis (PCoA) showed that the gut
   microbiota of the centenarian group (Y120) was different from the
   middle-aged (Y40, Y60) and the older adult (Y80, Y100) groups (Fig.
   [88]2E–H). A volcano map of the different bacteria was drawn according
   to the P-values of the DESeq2 analysis results. The beneficial bacteria
   Akkermansia, Lactobacillus, and Christensenella in the gut microbiota
   of the centenarian group (Y120) were significantly more prevalent than
   those in the middle-aged (Y40, Y60) and older adult (Y100) groups (Fig.
   [89]2I–L). The findings indicate that these beneficial bacteria play
   important roles in health and longevity.

Fig. 1. Distribution characteristics of gut microbiota in long-lived people
at different ages.

   [90]Fig. 1
   [91]Open in a new tab

   A Relative abundance heat map of the top 35 species at the genus level.
   B PCA map of the relative abundance level of the top 35 bacteria. PC,
   principal component. C Taxonomic phylogenetic tree for the difference
   analysis between the Y40, Y60, Y80, Y100, and Y120 groups. A total of
   257 species from 24 phyla were selected based on 281 species, with an
   average relative abundance >10^−6. The outer ring colour indicates that
   the species in each group had a significant increase (orange) or
   decrease (green) compared with the Y120 group. The inner ring shows the
   mean relative abundance of each species in all samples.

Fig. 2. Diversity characteristics of gut microbiota in long-lived people at
different ages.

   [92]Fig. 2
   [93]Open in a new tab

   A–D α-diversity analysis of Y40, Y60, Y80, Y100 and Y120 groups. E–H
   PCoA of the Y40, Y60, Y80, Y100, and Y120 groups. I–L Volcanic map of
   the Y40, Y60, Y80, Y100, and Y120 groups.

   To explore the role of the beneficial bacteria Akkermansia,
   Lactobacillus, and Christensenella in health and longevity, we screened
   281 distinct species with an abundance >10^−5. Spearman correlation
   analysis was performed for the Y40, Y60, Y80, Y100, and Y120 groups.
   The Akkermansia, Lactobacillus, and Christensenella genera correlated
   with absolute correlation coefficient values > 0.4 were presented at
   network and heat maps (Fig. [94]3). Akkermansia strongly correlated
   with Acholeplasma, Brevibacillus, Caloramator, and other genera in the
   Y120 group (Fig. [95]3A–F). Christensenella was strongly correlated
   with Massilimaliae, Provencibacterium, and Ruminiclostridium, in the
   Y40, Y60, Y80, Y100, and Y120 groups (Fig. [96]3A–E, G). Lactobacillus
   strongly correlated with Acholeplasma, Criibacterium, Desulfotomaculum,
   Lachnobacterium, and other genera in the Y120 group (Fig. [97]3A–E, H).
   Analysis of metabolic pathways revealed changes in gene abundance (Fig.
   [98]4). In the functional genes pathway with the largest difference in
   abundance, the abundance of related genes in the Y100 and Y120 groups
   was significantly higher than that in the Y40, Y60, and Y80 groups. The
   microbiota of centenarians showed a significantly increased redox
   capacity for methyltransferases, dehydrogenases, and oxidoreductases.

Fig. 3. Interaction characteristics of Akkermansia, Lactobacillus, and
Christensenella in the gut microbiota of long-lived people at different ages.

   [99]Fig. 3
   [100]Open in a new tab

   A–E Interactions between Akkermansia, Lactobacillus, and
   Christensenella in the Y40, Y60, Y80, Y100, and Y120 groups. The node
   size is proportional to the abundance of the genus, and the line
   segment width is proportional to the correlation strength. F
   Correlation analysis of Akkermansia in the Y40, Y60, Y80, Y100, and
   Y120 groups. G Correlation analysis of Christensenella abundance in the
   Y40, Y60, Y80, Y100, and Y120 groups. H Correlation analysis of
   Lactobacillus in the Y40, Y60, Y80, Y100, and Y120 groups. As
   Akkermansia, Lactobacillus, and Christensenella are mainly concerned,
   the species related to these three numbers (abundance >10^−5) in groups
   Y40, Y60, Y80, Y100, and Y120 are screened, and heat maps are drawn
   respectively. The mapping data are the values of correlation
   coefficients between each species and the three genera.

Fig. 4. Changes of longevity-related bacterial genes in the KEGG pathway
module and KO gene.

   [101]Fig. 4
   [102]Open in a new tab

   Four stages (Y40, Y60, Y80, and Y100) showed significantly increased or
   decreased bacterial genes abundance compared with the centenarian group
   (Y120). Significant changes (increase or decrease) are shown as
   follows: +++, increased P < 0.005; ++, increased P < 0.01; +, increased
   P < 0.05; −−−, reduced P < 0.005; −−, decreased P < 0.01; and −,
   decreased P < 0.05.

Serum metabolomics characteristics of longevity population at different ages

   We further characterised the serum metabolomics of the different age
   groups in the long-lived townships. Compared with the middle-aged (Y40,
   Y60) and older adult (Y80, Y100) groups, the relative expression levels
   of the L-ascorbic acid 2-sulphate, (R)-lipoic acid, L-arginine,
   lipoamide, fumaric acid, citraconic acid, glutamine, ascorbic acid, and
   taurolithocholic acid 3-sulphate metabolites were higher in the
   centenarian group (Y120) under the negative mode (Supplementary Fig.
   [103]1). The relative expression levels of the indole-3-butyric acid,
   glycochenodeoxycholic acid, methionine sulfoxide, dehydrocholic acid,
   glycocholic acid, arachidonoyl amide, and 5-hydroxyindole-3-acetic acid
   metabolites were higher in the Y120 group in the positive mode
   (Supplementary Fig. [104]2). The findings indicate that these
   metabolites play important roles in health and longevity.

   The analysis of differential serum metabolite characteristics between
   the middle-aged (Y40, Y60), older adult (Y80, Y100), and centenarian
   (Y120) groups is shown in Fig. [105]5. Glutamine, taurolithocholic acid
   3-sulphate, glycochenodeoxycholic acid, fumaric acid, indole-3-butyric
   acid, Dl-indole-3-lactic acid, and ascorbic acid was significantly
   upregulated in the centenarian group (Y120) compared to the other
   groups. Assessment of the characteristics of the serum metabolomic
   Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment pathways in
   the different age groups revealed significant differences between the
   middle-aged (Y40 and Y60), older adult (Y80 and Y100), and centenarian
   (Y120) groups (Supplementary Fig. [106]3). Significantly different KEGG
   enrichment pathways included ascorbate and aldarate metabolism, bile
   secretion, citric acid (TCA) cycle, ferroptosis, glutathione
   metabolism, hypoxia-inducible factor-1 (HIF-1) signalling pathway,
   vitamin digestion and absorption, and others.

Fig. 5. Characteristics of serum metabolites in the different age groups in
Jiaoling.

   [107]Fig. 5
   [108]Open in a new tab

   A–F Correlation diagram of different metabolite characteristics between
   the middle-aged (Y40) and centenarian (Y120) groups. G–L Correlation
   diagram of different metabolite characteristic analysis between the
   middle-aged group (Y60) and the centenarian group (Y120). M–R
   Correlation diagram for the analysis of differential metabolite
   characteristics between the older adult group (Y80) and the centenarian
   group (Y120). S–X Correlation graph for the analysis of differential
   metabolite characteristics between the older adult group (Y100) and the
   centenarian group (Y120). The metabolites labelled in the Fig. (A, G,
   M, S and D, J, P, V) setting a certain threshold of increase and
   decrease based on -log10 (p value) >3.0.

Correlation analysis of gut microbiota and serum metabolome in longevity
population at different ages

   We selected the top five differentially bacteria at the genus level and
   compared the top ten differentially expressed metabolites to analyse
   the correlation between the gut microbiota and metabolites. In the Y40
   vs. Y120 analysis in the negative mode (Supplementary Fig. [109]4A),
   Syntrophobotulus was significantly and positively correlated with
   glutamine levels. In the Y40 vs. Y120 analysis in the positive mode
   (Supplementary Fig. [110]4B), Alistipes and Desulfuromonas were
   significantly positively correlated with N3,N4-dimethyl-L-arginine.
   Marispirochaeta and Syntrophobotulus were significantly positively
   correlated with glycochenodeoxycholic acid. In the Y60 vs. Y120
   analysis in the negative mode (Supplementary Fig. [111]4C),
   Methylomicrobium was significantly positively correlated with gluconic
   acid. In the Y60 vs. Y120 analysis in the positive mode (Supplementary
   Fig. [112]4D), Methanopyrus was significantly positively correlated
   with coenzyme Q2. In the Y80 vs. Y120 analysis in the negative mode
   (Supplementary Fig. [113]4E), Bordetella, Dechloromonas, Eikenella,
   Holospora, and Leminorella were all associated with 3-indoxyl sulphate,
   whereas D-(-) -ribose, D-arabinose, gluconic acid, and glutamine were
   negatively correlated. In the Y80 vs. Y120 analysis in the positive
   mode (Supplementary Fig. [114]4F), Bordetella, Dechloromonas,
   Eikenella, Holospora, and Leminorella were negatively correlated with
   glycodeoxycholic acid, N3, N4-dimethyl-L-arginine, and
   proline-hydroxyproline. In the Y100 vs. Y120 analysis in the negative
   mode (Supplementary Fig. [115]4G), Cyclobacterium was significantly
   positively correlated with 1-methyladenosine, Slackia negatively
   correlated with glyoxylate levels, and Xanthomonas was positively
   correlated with D-arabinose and taurolithocholic acid 3-sulphate.
   Finally, in the Y100 vs. Y120 analysis in the positive mode
   (Supplementary Fig. [116]4H), Salegentibacter was significantly and
   positively correlated with indole-3-butyric acid, and Xanthomonas
   positively correlated with glycochenodeoxycholic acid content.

Effects of metabolites of L. plantarum 124 (LP124) in aging mice

   Previously, we targeted the isolation of Lactobacillus in the gut
   microbiota of a long-lived township population and screened and
   verified Lactobacillus plantarum 124 (LP124) with good anti-aging
   function through in vitro screening and in vivo verification methods.
   Metabolomics analysis showed that the serum levels of mesoconic acid
   (MA), lysine (Lys), and butyrate (But) in mice administered with LP124
   were significantly higher than those in the model group^[117]7. To
   explore the potential anti-aging role of LP124, we studied the
   anti-inflammatory effects of MA, Lys, and But on the aging process in
   mice.

   In terms of inflammatory factors (Fig. [118]6A), LP124, MA, Lys and But
   significantly decreased the levels of pro-inflammatory factors and
   significantly increased the levels of the anti-inflammatory factor. MA
   significantly reduced the levels of pro-inflammatory factors
   interleukin (IL)-1β, IL-6, IL-5, and tumour necrosis factor-alpha
   (TNF-α) in colon, liver, and kidney tissues, and significantly
   increased the levels of the anti-inflammatory factor IL-10 in these
   tissues. MA significantly decreased the levels of NOD-like receptor
   protein 3 (NLRP3), apoptosis-associated microprotein (ASC), and
   cysteine proteinase-1 (Casp-1) in colon, liver, and kidney tissues. In
   terms of oxidative stress (Fig. [119]6B), LP124, MA, and But
   significantly increased the levels of glutathione peroxidase (GSH-Px),
   superoxide dismutase (SOD), total antioxidant (T-AOC), and catalase
   (CAT), and significantly reduced the levels of reactive oxygen species
   (ROS), malondialdehyde (MDA), carbonyl protein, nuclear factor kappa B
   (NF-κB), P38 protein, and P65 protein in colon, liver and kidney
   tissues. In terms of the intestinal barrier (Fig. [120]6C), LP124, MA,
   and Lys significantly increased the levels of zonula occludens protein
   1 (ZO-1), tight junction protein (occludin), and claudin-1 in colon
   tissue. LP124, MA, and But significantly decreased the serum levels of
   d-lactic acid (D-LA), diamine oxidase (DAO), and endotoxin (ET).
   Hematoxylin and eosin staining revealed intact intestinal mucosa in the
   LP124, MA, Lys, and But groups; the epithelial cells were arranged
   neatly; no exfoliated, necrotic, or inflammatory cell infiltration was
   observed; there were more scattered goblet cells between columnar
   cells; and no obvious abnormalities were observed in the structure of
   the muscle and serosal layers. Periodic acid-Schiff (PAS) staining was
   positive (PAS++) for the LP124, MA, and Lys groups.
   Immunohistochemistry was moderately positive for mucin 2 (MUC2++) (Fig.
   [121]6D).

Fig. 6. Effects of LP124, MA, Lys, and But in aging mice.

   [122]Fig. 6
   [123]Open in a new tab

   Effects of LP124, MA, Lys, and But mesoconic acid, lysine, and butyrate
   on inflammatory factors (A), oxidative stress (B), intestinal barrier
   (C), haematoxylin and eosin (HE) staining, periodic acid-Schiff (PAS)
   staining, and immunohistochemistry of Mucin 2 (MUC2) in colonic tissue
   (D) in rapidly aging mice.

   The relative expression levels of mRNAs of the GSH-Px, SOD, thioredoxin
   reductase (TrxR), nuclear factor erythroid 2-related factor 2 (Nrf2),
   Kelch-like ECH-associated protein 1 (Keap1), B-cell leukaemia/lymphoma
   2 protein (Bcl2), Bcl-2 associated X-protein (Bax), NF-κB, NOD-, LRR-
   and pyrin domain-containing protein 3 (NLRP3), and Caspase-1 genes in
   colon tissues are shown in Supplementary Fig. [124]5A. LP124, MA, and
   Lys significantly increased the relative expression levels of GSH-Px,
   SOD, TrxR, and Nrf2 mRNA, and significantly reduced the relative
   expression levels of Bax and NLRP3 mRNA in colon tissues. LP124 and MA
   significantly reduced the relative mRNA expression levels of Keap1,
   NF-κB, and Caspase-1 genes in colon tissue. MA significantly increased
   the relative expression of Bcl2 mRNA in colon tissue. Finally, LP124
   and MA significantly increased the level of nicotinamide adenine
   dinucleotide (NAD+) in the colon, liver, and kidney tissues
   (Supplementary Fig. [125]5B).

Regulation of MA, Lys, and But on the gut microbiota of aging mice

   The α-diversity of mice gut microbiota was analysed. The results are
   provided in Supplementary Data [126]1. In terms of the abundance index,
   the observed species, Chao1, and Abundance Based Coverage Estimator
   (ACE) indices of the mouse gut microbiota were significantly increased
   by LP124, MA, Lys, and But group. In terms of the diversity index,
   LP124, MA, Lys, and But group significantly increased the Shannon index
   of the gut microbiota in mice. LP124, MA, Lys, and But group
   simultaneously increased the abundance of Parabacteroides,
   Rikenellaceae RC9 gut group, and Bacteroides, and simultaneously
   decreased the abundance of harmful bacteria Helicobacter pylori,
   Escherichia–Shigella, and Staphylococcus in the intestinal tract of
   mice. LP124 group increased the abundance of the beneficial bacteria
   Akkermansia, Ruminococcus, Desulfovibrio, Oscillibacter, Alistipes, and
   Colidextribacter in the intestines of mice. MA group increased the
   abundance of Ligilactobacillus, Erysipelatoclostridium, Dubosiella,
   Parasutterella, and Haemophilus in the intestinal tracts of the mice
   (Fig. [127]7A–F). Phylogenetic trees of the top 100 gut microbiota in
   mice at the genus level are shown in Supplementary Fig. [128]5C.

Fig. 7. Regulatory effects of LP124, MA, Lys, and But on the gut microbiota
of aging mice.

   [129]Fig. 7
   [130]Open in a new tab

   Composition characteristics of the gut microbiota in mice at the phylum
   (A, B), family (C, D), and genus (E, F) levels. G LEfSe analysis of
   different species among the mouse gut microbiota under the NC, MOD,
   LP124, MA, Lys, and But conditions. T-test analysis of different
   species among mouse gut microbiota groups under LP124 (H), MA (I), Lys,
   (J), and But (K) conditions.

   As shown in the histogram of the linear discriminant analysis (LDA)
   value distribution of the linear discriminant analysis effect size
   (LEfSe) analysis results of different species between NC, MOD, LP124,
   MA, Lys and But groups (Fig. [131]7G). NC group mainly contained the
   beneficial bacteria Odoribacter, Alistipes, and Alloprevotella. MOD
   group mainly harboured the harmful bacterium Helicobacter. LP124 group
   mainly contains the beneficial bacterium Akkermansia, and MA group
   mainly contains the butyricogenic bacteria Dubosiella. Lys group mainly
   contains Bifidobacterium; But group mainly contains Bacteroides
   acidifaciens and Bacteroides sartorii (Fig. [132]7G). In the
   statistical analysis, species with significant differences between
   groups at each classification level were identified using a t-test
   (p < 0.05). LP124, MA, Lys, and But group were significantly different
   from MOD group (Fig. [133]7H–K). The MA group contained Bacteroides,
   Dubosiella, Parabacteroides, and others (Fig. [134]7I).

Effects of MA, Lys, and But on serum metabolome of aging mice

   Next, we analysed the LP124 metabolites in the serum metabolome of
   aging mice. In the LP124 vs. MOD comparison group, 248 positive mode
   metabolites were significantly different, of which 164 were upregulated
   and 84 were downregulated. A total of 123 negative mode metabolites
   exist with significant differences, of which 75 were upregulated and 48
   were downregulated (Supplementary Fig. [135]6, Supplementary Data
   [136]2). The comparison revealed significantly upregulated metabolites,
   including taurochenodeoxycholic acid (sodium salt), taurocholic acid,
   tauroursodeoxycholic acid, and 3-hydroxybutyric acid (Supplementary
   Fig. [137]6A–F). A bubble diagram of KEGG enriched pathways for the
   LP124 vs. MOD comparison is shown in Supplementary Fig. [138]6G, H, and
   Supplementary Data [139]3 and [140]4. The results of the KEGG
   regulatory network analysis are provided in Supplementary Data [141]5
   and [142]6. In the MA vs. MOD comparison, 260 positive mode metabolites
   were significantly different, among which 155 were upregulated and 105
   were downregulated. A total of 131 negative mode metabolites were
   observed with significant differences, of which 61 were upregulated,
   and 70 were downregulated (Fig. [143]8, Supplementary Data [144]7). The
   MA vs. MOD comparison revealed significantly upregulated metabolites,
   including taurochenodeoxycholic acid (sodium salt), 3-hydroxybutyric
   acid, tauroursodeoxycholic acid, 7-ketolithocholic acid, and
   deoxycholic acid (Fig. [145]8A–F). A bubble diagram of KEGG enrichment
   data for MA vs. MOD is shown in Fig. [146]8G, H, and Supplementary Data
   [147]8 and [148]9. The KEGG regulatory network data for the MA vs. MOD
   comparison are provided in Supplementary Data [149]10 and [150]11.

Fig. 8. Effects of MA on serum metabolome of mice.

   [151]Fig. 8
   [152]Open in a new tab

   A, C Volcano and differential metabolite chords of mouse serum
   metabolites in MA mice in negative mode. B, D Volcano and differential
   metabolite chords of serum metabolites in MA mice in the positive mode.
   Correlation between differential metabolites in MA mouse serum in the
   negative (E) and positive (F) mode. Bubble diagram of KEGG enrichment
   of MA mouse serum metabolites in negative (G) and positive (H) mode.

   In the Lys vs. MOD comparison group, 250 positive mode metabolites were
   significantly different, of which 152 were upregulated and 98 were
   downregulated. A total of 124 negative mode metabolites were observed
   with significant differences, of which 68 were upregulated and 56 were
   downregulated (Supplementary Fig. [153]7, Supplementary Data [154]12).
   A bubble diagram of KEGG enrichment data for Lys vs. MOD is shown in
   Supplementary Fig. [155]7G, H, and Supplementary Data [156]13 and
   [157]14. The KEGG regulatory network data for the Lys vs. MOD
   comparison are provided in Supplementary Data [158]15 and [159]16. In
   the But vs. MOD comparison group, 210 positive mode metabolites were
   significantly different, of which 128 were upregulated and 82 were
   downregulated. A total of 108 negative mode metabolites were observed
   with significant differences, of which 51 were upregulated, and 57 were
   downregulated (Supplementary Fig. [160]8, Supplementary Data [161]17).
   A bubble diagram of KEGG enrichment data for the But vs. MOD is shown
   in Supplementary Fig. [162]8G, H, and Supplementary Data [163]18 and
   [164]19. The KEGG regulatory network data for the But vs. MOD
   comparison are provided in Supplementary Data [165]20 and [166]21.

Analysis of the correlation among gut microbiota, metabolites, and
physiological indicators of aging mice

   We selected the top five differentially abundant bacteria at the genus
   level and compared the findings with the top ten differential
   metabolites to analyse the correlation between the gut microbiota and
   metabolites. The results of LP124, MA, Lys, and But vs. MOD are shown
   in Supplementary Fig. [167]9A–H, Supplementary Data [168]22–[169]25. We
   selected the top 35 differentially prevalent bacteria genera and
   indicators, such as inflammatory factors, oxidative stress, and
   intestinal barrier, and analysed the correlation between the gut
   microbiota and physiological indicators (Supplementary Fig. [170]9I).
   Beneficial bacteria, including Odoribacter, Alloprevotella,
   Parasutterella, Rikenellaceae RC9 gut group, Dubosiella, Alistipes,
   Bifidobacterium, and Haemophilus, were significantly negatively
   correlated with pro-inflammatory factors IL-1β, IL-6, TNF-α, and NLRP3
   inflammasome; pro-oxidation indices MDA, carbonyl, NF-κB, ROS, and ET;
   and intestinal barrier indices ZO-1, Claudin-1, DLA, and DAO. Positive
   correlations were evident for the anti-inflammatory factor IL-10 and
   the antioxidant factors SOD, T-AOC, and CAT. These bacteria have active
   anti-inflammatory roles. By contrast, Lachnoclostridium was
   significantly positively correlated with pro-inflammatory factors,
   inflammasomes, and pro-oxidation and intestinal barrier indices. This
   genus was negatively correlated with anti-inflammatory and antioxidant
   factors.

Discussion

   The current cross-sectional analysis of a long-lived population cohort
   in Jiaoling, China, which is globally recognised for its long-lived
   residents, combined with metagenomics, metabolomics, and in vivo animal
   validation clarifies the gut microbiota characteristics and metabolic
   network of centenarians and verifies the key substance of anti-aging of
   the LP124 probiotic derived from centenarians (Fig. [171]9). Our
   results strongly suggest that the diversity of the gut microbiota in
   centenarians is significantly higher than that in young and middle-aged
   individuals, accompanied by the enrichment of beneficial symbiotic
   bacteria. These observations are similar to those of previous reports,
   which showed that, based on a variety of approaches, such as
   metabolomic capacity and gut microbiome diversity, centenarians have a
   gut microbiome associated with younger people, and their gut microbiome
   may be more similar to that of younger people^[172]24,[173]25. Because
   the gut microbiome plays a key role in host health and disease^[174]26,
   we speculate that this unique gut microbiome profile and beneficial
   symbionts and their metabolites may contribute to longevity.

Fig. 9. Anti-inflammation and anti-aging mechanisms of LP124 and mesaconic
acid metabolite of LP124.

   [175]Fig. 9
   [176]Open in a new tab

   Compared with young and middle-aged people, centenarians showed
   significantly increased α-diversity and Lactobacillus in the gut
   microbiota. Mesaconic acid from LP124 and centenarian metabolites have
   significant anti-inflammatory and anti-aging effects. Created using
   BioRender.com.

   One of the main findings of this study is the significantly greater gut
   microbial diversity in centenarians compared to young and middle-aged
   adults. Consistent with these findings, a follow-up study on
   centenarians reported a consistently increased evenness of the gut
   microbiota^[177]25. In addition, consistent with observations in
   previous studies on older or younger adults, long-lived people with
   higher microbiome diversity tend to have fewer microbiome changes
   during aging, suggesting that greater diversity or species uniformity
   in centenarians compared to other older adults may protect the gut
   microbiota from instability^[178]27,[179]28. The current findings show
   that, compared to young and middle-aged people, the gut microbiota of
   long-lived people was enriched with beneficial symbiotic bacteria, such
   as Lactobacillus and Akkermannia. Thus, preservation of beneficial
   symbiotic Lactobacillus may be a key feature associated with longevity.
   Several studies have reported that bacterial diversity decreases with
   age, and the colonisation of opportunistic pathogens (such as
   Enterobacteriaceae) increases. However, our study of the
   characteristics of the gut microbiota in centenarians during aging
   showed an increase in bacterial diversity and an increase in the
   abundance of the beneficial bacteria in the genus Lactobacillus.
   Several cross-sectional studies have shown that the composition of
   microbiota in centenarians plays an important role in healthy aging.
   For example, Biagi et al. reported that in the centenarian group, the
   core microbiota composition was reduced and health-related bacteria
   (including Akkermansia, Bifidobacterium, and Christensenella) were
   enriched^[180]17. These findings are consistent with the results of our
   study. The enrichment of these beneficial microorganisms likely
   reflects the anti-aging pattern of gut microbiota in centenarians.
   Given that these bacteria typically secrete short-chain fatty acids and
   possess anti-inflammatory and antioxidant properties, their unique
   pattern of enrichment of beneficial microbes during aging may be
   associated with longevity.

   Another major finding of this study from the metagenomic analysis
   results combined with the isolation, culture and screening in vitro was
   the identification of an LP124 strain with a good antioxidant effect.
   The basis of its anti-aging effect was explored through metagenomics,
   metabolomics, and experiments using rapidly aging mice. Finally, we
   confirmed that the MA metabolite of LP124 has potent,
   anti-inflammatory, and antioxidant effects. With the rapid development
   of second- and third-generation high-throughput sequencing
   technologies, most studies have been limited to metagenomic sequencing
   analysis of the gut microbiota. Most of these metagenomic sequencing
   results were not validated by further isolation of key strains for in
   vitro and in vivo experiments. For example, the gut microbiota
   structure of centenarians and young adults is reportedly highly
   similar^[181]25, with a higher abundance of potentially beneficial
   Bacteroides, such as Akkermansia, in long-lived people^[182]29. In
   addition to a higher abundance of specific bacteria such as
   Parabacteroides, most bacteria have anti-inflammatory effects in their
   host^[183]30. The latter authors hypothesised that a healthy gut
   microbiota in centenarians promotes healthy aging. They posited that an
   underlying reason is increased chronic low-grade inflammation in the
   older adult, which is associated with various chronic diseases^[184]31.
   This is consistent with the results of this study. Coupled with the
   beneficial effects of these bacteria^[185]32,[186]33 and the lower
   susceptibility of centenarians to chronic diseases^[187]34, these
   microbes might contribute to the health of centenarians^[188]25.
   However, these studies did not validate the results obtained through
   metagenomic analysis by further designing in vivo and in vitro
   experiments. The mechanisms underlying the enrichment of beneficial
   microorganisms in centenarians have not yet been fully elucidated.

   In the present study, LP124 characterised by a good antioxidant effect
   was isolated and screened from the gut microbiota of centenarians.
   Animal experiments further verified that LP124 exerts anti-inflammatory
   effects related to MA, Lys, and But, which are the key substance for
   anti-aging. The MA metabolite of LP124 displayed the best anti-aging
   effect. MA, itaconic acid, and citriconic acid are isomers. Itaconic
   acid is an anti-inflammatory metabolite via activation of Nrf2 through
   the alkylation of KEAP1^[189]35. Citraconic acid inhibits aconitate
   decarboxylase 1 (immune responsive gene 1) catalysis, reduces
   interferon response and oxidative stress, and regulates inflammation
   and cell metabolism^[190]36. MA is synthesised from itaconic acid and
   plays an immunomodulatory role in macrophages^[191]37. This is
   consistent with our findings that MA had a significant
   anti-inflammatory effect, alleviated oxidative stress, and protected
   the intestinal barrier. In the current study, MA significantly
   increased the richness and diversity of the gut microbiota in aging
   mice and increased the abundance of butyricogenic bacteria in the genus
   Dubosiella. The collective findings indicate that MA regulates the gut
   microbiota, relieves inflammation and oxidative stress, and maintains
   the intestinal barrier. Our results are consistent with those of Zhang
   et al., who used intestinal faecal 16S rDNA sequencing and targeted
   metabolomics sequencing technology^[192]38. An important protective
   effect of the specific gut microbiota Dubosiella and its butyric acid
   metabolite in sepsis-associated brain injury has been discovered. The
   findings established the important role of gut microbiota in
   sepsis-associated brain injury and provided a basic research basis for
   targeted gut microbiota therapy in clinical practice^[193]38. The
   results of Cai et al., based on faecal metabolomics of healthy
   centenarians from a Chinese longevous region showed that longevity was
   closely related to beneficial metabolites such as phospholipids, amino
   acids and short-chain fatty acids (SCFAs)^[194]39. These provide
   guidance for the integration of multiomics techniques to identify aging
   and anti-aging biomarkers^[195]40,[196]41.

   In conclusion, this study reports the composition and functional
   characteristics of gut microbiota in centenarians. Our results
   characterise a new metabolite strain with anti-inflammatory and
   anti-aging properties. The findings jointly highlight the positive role
   of the gut microbiota of long-lived people, especially the key
   intestinal symbiotic beneficial bacteria and their metabolites, which
   play an anti-aging role in health and longevity. The gut microbiota of
   centenarians is closely associated with longevity and is characterised
   by high species diversity and high levels of beneficial bacteria.
   Probiotics derived from the gut microbiota of centenarians, especially
   Lactobacillus plantarum 124 and its metabolite mesaconic acid, have a
   significant positive effect on health and longevity.

Methods

Ethics approval and study cohort

   The overall purpose of this study was to use metagenomic and
   metabolomic methods to characterise the composition and function of the
   gut microbiota and serum metabolomics in different age groups in a
   township renowned for the longevity of its residents. The key strains
   were isolated, screened in vitro, and used in the
   senescence-accelerated mice (SAMR1 and SAMP8) to further explore the
   role of the key strains and their metabolites in the gut microbiota in
   health and longevity. This study was approved by the Ethics Committee
   (Ethics number: 2022(9)) of the First Affiliated Hospital of Guangdong
   Pharmaceutical University.

   In total, 224 participants were randomly selected. All enroled
   participants signed an informed consent form before physical
   examination and biomaterial collection. Faecal samples were freshly
   collected from each subject and immediately frozen at −20 °C,
   transported to the laboratory on an ice pack, and stored in a −80 °C
   freezer until metagenomic analysis. Subsequently, serum samples were
   subjected to metabolomic analyses. For age-related group comparison
   analyses, we divided the participants from the Jiaoling cohort into
   middle-aged (Y40 and Y60, Y40: 21–40 years and Y60: 41–60 years), older
   adults (Y80 and Y100, Y80: 61–80 years and Y100: 81–99 years), and
   centenarian (Y120: 100–110 years) groups.

Metagenomic sequencing analysis

   DNA was extracted from faecal samples, as previously described^[197]42.
   In this study, the Illumina HiSeq sequencing platform was used to
   obtain 11,619.54 Mbp of raw data. After quality control, an average of
   11,604.59 Mbp was obtained. A mean of 219,259 open reading frames
   (ORFs) were obtained by gene prediction using MetaGeneMark software.
   The non-redundant gene set was compared with the MicroNR database using
   BLASTP, and species annotation was performed using the LCA algorithm.
   We used DIAMOND software^[198]43,[199]44 to perform common functional
   database annotations on non-redundant gene sets (e-value ≤ 10^−5). A
   total of 60.85% ORFs were compared to the KEGG database (Version
   2018-01-01, [200]http://www.kegg.jp/kegg/)^[201]45,[202]46, 59.72% were
   compared to the eggNOG database (Version 4.5,
   [203]http://eggnogdb.embl.de/#/app/home)^[204]47, and 3.09% were
   compared to the CAZy database (Version 20150704,
   [205]http://www.cazy.org/)^[206]48.

Analysis of composition characteristics of gut microbiota

   The use of DIAMOND software^[207]49 (v0.9.9.110,
   [208]https://github.com/bbuchfink/diamond/), Link Unigenes to the NCBI
   NR database (Version 2018-01-02, [209]https://www.ncbi.nlm.nih.gov/).
   For the comparison, the result of each sequence with
   e-value ≤ e-value*10^[210]50 was chosen. Since each sequence may have
   multiple comparison results, the LCA algorithm (applied to the
   systematic classification of MEGAN software^[211]51
   ([212]https://en.wikipedia.org/wiki/Lowest_common_ancestor) to
   determine the sequence of species annotation information. The abundance
   of a species in a sample is equal to the sum of the annotated gene
   abundances of the species^[213]52–[214]54. Starting from the abundance
   table at each classification level, Krona analysis^[215]55 was
   performed, showing the general situation of relative abundance and the
   clustering heat map of abundance. Dimension reduction analysis was
   performed using PCA^[216]56 (R ade4 package, Version 2.15.3) and
   NMDS^[217]57 (R vegan package, Version 2.15.3).

Animal experiments and determination of biochemical index

   The animal experiments were approved by the Institute of Microbiology,
   Guangdong Academy of Sciences (approval number: GT-IACUC202112281).
   SAMR1 and SAMP8 (SPF grade) mice obtained from Guangdong Jinzhihe
   Biotechnology Co., Ltd. (Guangdong, China) were housed under a 12-h
   light/dark cycle in gnotobiotic facilities. All mice were provided with
   sterile food and water ad libitum. Both the NC (SAMR1, 6 mice) and MOD
   (SAMP8, 6 mice) groups received 0.2 mL of normal saline by gavage
   daily. For the LP124 test group (SAMP8, 6 mice), 0.2 mL of LP124
   suspension containing 1.0 ± 0.05 × l0^9 colony forming units (CFU)/mL
   was administered^[218]7. For the MA test group (SAMP8, 6 mice), 0.2 mL
   of MA solution (200 mg/kg body weight/day) was administered. For the
   Lys test group (SAMP8, 6 mice), 0.2 mL of Lys solution (200 mg/kg body
   weight/day) was administered. For the But test group (SAMP8, 6 mice),
   0.2 mL of But solution (200 mg/kg body weight/day) was administered.
   All administrations were by gavage daily for 9 weeks. The animals used
   in this experiment were anesthetized. The anaesthetic drug was sodium
   pentobarbital, which was administered by intraperitoneal injection. The
   injection dose was 40–50 mg/kg (2%) intraperitoneally. The principle
   was that sodium pentobarbital directly acted on the central nervous
   system through the blood circulation, inhibiting the activity of
   neurons, thereby achieving the anaesthetic effect. Colon, liver, and
   kidney tissue homogenates were prepared for analysis for
   pro-inflammatory factors IL-1β, IL-6, and TNF-α, and anti-inflammatory
   factors IL-10, NLRP3, ASC, and Casp-1 were measured using ELISA kits
   (Beijing winter song Boye Biotechnology Co. Ltd., Beijing, China). The
   oxidative stress markers GSH-Px, SOD, TrxR, T-AOC, MDA, CAT, carbonyl
   protein, NF-κB, and ROS were determined by detection kits (Nanjing
   Jiancheng Bio Co., Nanjing, China). Intestinal barrier factors ET,
   ZO-1, occludin, claudin-1, D-LA, and DAO were measured using an ELISA
   kit and a related detection kit (Beijing Winter Song Boye Biotechnology
   Co. Ltd.).

Morphological observations and immunohistochemical analysis

   After the animals were euthanized, their colons were isolated. The
   tissues were fixed with 10% formalin, embedded in conventional
   paraffin, sectioned to a thickness of 4–5 μm, and observed using
   haematoxylin-eosin and PAS staining. MUC2 in the colon tissue was
   stained using immunohistochemistry. The intestinal morphology was
   observed and photographed using an inverted optical microscope.

Expression analysis of genes

   Total RNA was extracted from the mouse colon using an RNeasy Mini Kit
   (Huangshi Yanke Biotechnology Co., Ltd., Hubei, China) according to the
   manufacturer’s instructions. Quantitative RT-PCR was performed using
   SYBR Premix Ex Taq (Huangshi Yanke Biotechnology Co., Ltd., Hubei,
   China) on the 7500 Fast Real-Time PCR System (Applied Biosystems,
   Franklin Lakes, NJ, USA). The calculation of mRNA expression was
   performed by the 2^−ΔΔCT method using the geometric mean of the
   housekeeping genes GSH-Px, SOD, TrxR, Nrf2, Keap 1, Bax, Bcl 2, NF-κB,
   NLRP3, and Caspase 1. Gene and primer sequences are listed in
   Supplementary Data [219]26. Comparisons between groups were performed
   using a one-way analysis of variance with n = 6 per group. All results
   are expressed as mean ± SD of the biological replicates.

Mass spectrometry

   Serum samples were collected from people in the different age groups
   (Supplementary Data [220]27–[221]29), and the NC, MOD, LP124, MA, Lys,
   and But groups (n = 6 per group) (Supplementary Data [222]30, [223]31).
   Acetonitrile was used to extract metabolites by protein precipitation.
   Untargeted metabolomic profiling was performed using liquid
   chromatography coupled with a High-Field Qexactive mass spectrometer
   (Thermo Fisher Scientific, Waltham, MA, USA; HILIC/ESI+, C18/ESI−,
   85–1275 m/z, 120k resolution). Spectral features (m/z, retention time)
   corresponding to the identified and uncharacterized metabolites were
   integrated and aligned using apLCMS/xMSanalyzer software.
   MetaboAnalyst^[224]58 was used for statistical analysis, and Mummichog
   software^[225]59 was used for pathway enrichment analysis and verified
   by retention time, m/z, and tandem mass spectrometry using
   authenticated standards. Candidate molecules were identified following
   ion dissociation experiments on a Thermo Scientific Fusion mass
   spectrometer, and the spectral library was matched to the mzCloud,
   mzVault, and MassList libraries using Compound Discoverer 3.0.

Statistical analyses

   According to the different data, statistical analyses between groups
   were performed using Wilcoxon’s rank sum test, Student’s t test, or
   one-way ANOVA. Results are expressed as mean ± SD. Statistical analyses
   and data visualisation were performed using the R software (v2.15.3)
   with the WGCNA, stats, and ggplot2 software packages.

Supplementary information

   [226]Supplementary information^ (2.3MB, pdf)
   [227]Supplementary Data 1_31^ (5.2MB, xls)

Acknowledgements