We investigated the microbial and metabolic features of the faecal microbiota in both young and midlife subjects using in vitro batch fermentation model. Soluble starch, a representative of digestible carbohydrate, and inulin, a representative of non-digestible carbohydrate, were selected to study the metabolic responses. Both the feces and fermented broth samples were processed and subjected to 16S rRNA gene sequencing and SCFAs detection. We also determined gas production in the fermentation model as it is one of the most production metabolites beside SCFAs.
Difference in the structure of the microbiota community
Based on the results of the16S rRNA gene sequencing, the alpha-diversity indices, such as Shannon, Simpson and Chao1 index are similar between midlife and young in both faecal and fermented samples (Fig. 1A), suggesting the consistent diversity of the gut microbiota between the two age groups. This agrees with previous reports that the faecal microbiota structures of 20 to 60 years old are in a stable stage in this aspect.[35, 36] We also assessed overall community compositions by drawing bar charts of relative abundance at different taxonomic levels, of which at genus level was shown in supplementary (Supplemental Fig. 1). With the comparison of the relative abundance among top ten genera, Bifidobacterium in young group were significantly higher than those in midlife subjects (Fig. 1C), suggesting some differences started to show in the faecal microbiota between the two age groups. Furthermore, beta-diversity analysis was assessed to reveal differences between young and midlife group, of which the significance was examined by ANOSIM. Unfortunately, no significant difference was found between the two age groups. The ANOSIM results using Bray-Curtis were shown in supplementary (Supplemental Fig. 2). Similarly, PCA did not divided the faecal samples into two groups (Fig. 1B). Interestingly, the PCA scattered plots of the inulin broth samplescan be obviously clustered on age (Fig. 2B) despite of no difference between age groups in the starch broths (Fig. 2A), suggesting an age-related change in the microbial community structure of midlife.
Impact of environmental factors on microbiota
Canonical Correspondence Analysis (CCA) was performed to investigate the effects of age and carbon source on the variation in microbiota (Fig. 3). The scatter plot displayed the clustering on age and culture medium. The small percentage of the axis suggested that although age and carbon are not the only sources causing the variation in the system, they are the major ones with more contribution from carbon source than age. This is further supported by the observation that clustering plot shifts in the direction of both medium and age..
Difference in Taxonomic composition
Faecal samples
Linear discriminant analysis (LDA) effect size (LEfSe) was used to identify the age-related bacterial feature in the microbiota before (Fig. 1D) and after fermentation(Fig. 4). As for the faecal samples, LEfSe revealed 27 taxa overrepresent in young group but none in midlife group (Fig. 1D). The taxa involve 11 genera in 4 phyla, suggesting a wide range of changes in relative abundance between youth and midlife. Among the 27 taxa, the top 5 taxa were taxa within phylum Actinobacteria, including Actinobacteria (class), Actinobacteria (phylum), Bifidobacteriales (order), Bifidobacteriaceae (family) and Bifidobacterium (genus) in turn. Bifidobacteria dominates the infant gut microbiota and reduces with age.[37] Similar to the observation that Bifidobacterium disappears in middle-aged mice,[12] our data reveal that Bifidobacterium is the genus declined most in human midlife. Furthermore, 8 taxa, including Collinsella and Coriobacteriales (order) and Coriobacteriaceae (family), were observed to decline in midlife. This is also similar to the observation in mice that the relative abundance of both Bifidobacterium spp. and Coriobacteriaceae spp. decreases strongly and significantly in middle-aged mice compared to young subjects.[12]
Four genera, including Faecalibacterium, Butyricicoccus, Desulfovibrio and Parasutterella were abundant in young group but absent in midlife group. Faecalibacterium and Butyricicoccus are the main butyrate products found in the intestine.[38, 39] Thus, the relative abundance of butyrate-producing bacteria differs between young and midlife groups. Desulfovibrio is the dominant sulfate-reducing bacteria in the colon and reduces sulfate compounds to hydrogen sulfide (H2S)[40] which is one of the main microbiota-derived gases contributing to human colonic homeostasis.[41] Parasutterella was the one changed most in phylum Proteobacteria, a result also observed in aging mice.[12] Parasutterella is a core component of human gut microbiota and produces various metabolites including aromatic amino acid, bilirubin, purine, and bile acid derivatives.[42] Our results thus suggest the core composition in gut microbiota begins to reduce in midlife.
Upon community composition analysis of 16S rRNA gene sequencing results, Faecalibacterium, Bifidobacterium and Parasutterella were included in the top 10 genera with decreased relative abundance in midlife group (Fig. 1C), which is consistent with LEfSe, although only the reduction of Bifidobacterium was significant (p = 0.029). Furthermore, qPCR of Bifidobacterium also showed the significant reduction in feces of midlife group compared to young subjects (Fig. 5A). Bifidobacterium is thought to play pivotal roles in maintaining human health,[43, 44] and its numbers decline is one of the most marked changes in the elderly gut.[45] And Faecalibacterium also associates with the grip strength in elderly.[46] Hence, we conclude that microbial alterations associated with aging also occur in midlife, rather than the previous thought to begin in elderly.
Starch fermented samples
After starch fermentation (Fig. 4A), there were 11 taxa overrepresented in young group, predominated by Selenomonadales (order), Negativicutes (class), Veillonellaceae (family), Desulfovibrio (genus), Bifidobacterium (genus) and its belonging family and order, Comamonas (genus) and its belonging family, Catenibacterium (genus) and Bacillales (order). In contrast, only 6 taxa were overrepresented in midlife, including Proteobacteria (phylum), Peptoniphilus (genus), Lachnospiracea-incertae_sedis (genus) and Staphylococcus (genus). Veillonellaceae, belonging to class Negativicutes and further belonging to order Selenomonadales, is reported to play roles in host carbohydrates metabolism.[47] Veillonellaceae and its higher order taxa are the top taxa in starch fermentation indicates their roles seem to be more important for starch degradation compared to Bifidobacterium.
Inulin fermented samples
In term of the inulin fermentation (Fig. 4B), LEfSe revealed 8 taxa overrepresented in young group compared to midlife group. Five are taxa within phylum Actinobacteria, including Bifidobacteriumand the remaining 3 taxa are Catenibacterium (genus), Comamonas (genus) and its belonging family. These taxa are all present in LEfSe results of both feces and starch broth, suggesting they are the key bacteria in young microbiota for saccharolytic process. On the other side, LEfSe identified 14 taxa overrepresented after inulin fermentation in midlife group. The top 5 taxa are Proteobacteria (phylum), Escherichia-Shigella (genus), Enterobacteriaceae (family), Enterobacteriales (order), Gama-proteobacteria (class), and the remaining taxa are Peptostreptococcus (genus), Delftia (genus), Streptococcaceae (family),Fusicatenibacter (genus), Parasutterella (genus), Phascolarctobacterium (genus), Acidaminococcareae (family), Enterococcus (genus), Enterococcaceae (family). The shared taxa number within young group after fermentation is 8, contrast to it however, only 1 taxa, Proteobacteria (phylum) is common in midlife group, suggesting that the reduction of bifidobacteria and increase of Proteobacteria can be taxonomic characteristics for midlife microbiota to ferment carbohydrates.
Difference in metabolic response
SCFAs in both feces and fermented broths were measured. As shown in Fig. 5C, most individual SCFAs in the feces in midlife group were lower than those in young group, except for propionate, but without significant difference. After starch fermentation, the most individual SCFAs in midlife were lower than those in young group (Fig. 6C-H), which is similar to the case in feces. However, after inulin fermentation, the reduction of acetate in midlife (Fig. 6C) was significant (p = 0.0002) compared to young group despite of similar level of butyrate and propionate in the two age groups. A similar result has been observed in aging mice that butyrate and propionate do not differ between two age groups, but acetate is significantly lower in old mice.[12] It is worth pointing out that the total SCFAs in midlife broth of inulin declines pronouncedly (p = 0.052). In addition, inulin fermentation also caused significant gas production increase in midlife group (p = 0.008) (Fig. 6I), and inulin degradation decrease (p = 0.002) compared to young group (Fig. 6J), a result comparable to the decline of saccharolytic capability in aging mice.[11] Hence, we can conclude that the age-related metabolic responses occur only under inulin fermentation, consistent with the microbiota changes revealed by PCA (Fig. 2B) and CCA (Fig. 3). This is consistent with previous report of age-related functional characteristics in metagenomes that midlife with age 38-43 differentiates from young and elderly, and constitutes a good watershed in data clustering between age groups.[48]
Correlation between the metabolites and the genera
To understand the contribution of genera to the altered metabolites, Spearman’s correlation coefficient was used to analyze the association between metabolites concentration and genus abundance. As for the starch fermentation, 3 genera, including Bifidobacterium, Desulfovibrio and Staphylococcus, showed correlation with six metabolic parameters (Fig. 7A). As for inulin fermentation, 8 genera included Fusicatenibacter, Phascolarctobacterium, Peptostreptococcus, Delftia, Bifidobacterium, Parasutterella, Comamonas and Catenibacterium were involved and correlated with seven metabolic parameters (Fig. 7B). .
As shown in Fig. 7B, Bifidobacterium was the only genus shown significant correlation with both acetate (positive, red) and gas production (negative, bule) in the inulin fermentation. This is supported by the recent reports that Bifidobacterium is lack of hydrogenase genes in its genome[49] and acetate is its main metabolite.[50] In addition, Parasutterella and Peptostreptococcus, abundant in midlife group, also showed significant correlation with acetate and gas, respectively. For gas production, the relative low production in young group is consistent with its negative correlation with Bifidobacterium. Similarly, the relatively high gas production in midlife group is consistent with its positive correlation with Peptostreptococcus, the second abundant genus in this group. Therefore, we speculate that such regulations in the opposite direction seem to enhance the difference of gas production between the two age groups leading to significant difference. As for acetate production, both Bifidobacterium and Parasutterella showed positive correlation with acetate but with different p values, p < 0.01 and p < 0.05, respectively, suggesting Bifidobacterium has a stronger impact on acetate production than Parasutterella. This is also in agreement with the metabolite measurement that the young group, in which Bifidobacterium is the predominant genus, has higher acetate production than midlife group, in which Parasutterella is relatively abundant. Concern Parasutterella is the 5th abundant genus in midlife group by LEfSe, the dominant position of Parasutterella is much lower than that of Bifidobacterium in young group. Hence, we speculate that it is due to the lower influence on acetate production and lower dominance in its own group of Parasutterella compared to Bifidobacterium that midlife group showed significantly lower acetate production than young group in inulin fermentation. Thus, the absence of Bifidobacterium in midlife seemed contribute to both the significant increased gas production and the significant reduced acetate after inulin fermentation. In human intervention trial of resistant starch, some volunteers were found as “non-responders” with > 60 % of unfermented starch remaining in their stools while others with < 4%.[51] The difference in the initial composition of individual’s gut microbiota may be the cause.[52, 53] Therefore, the reduction of Bifidobacterium in midlife feces seems to be responsible for the significant drop in metabolic responses to inulin for midlife microbiota.
Accompanied with the declined Bifidobacterium, the significant reduced inulin degradation rate after in vitro fermentation suggests attenuated saccharolytic capacity of inulin in midlife group. Bifidobacteria is reported to own the capacity to ferment a variety of carbohydrate and fiber compounds.[54] The saccharolytic genes were found to decrease in elderly gut microbiota[21] and older age mice.[11]However, the age-related changes in microbial community structure and metabolites were only occurred in the fermentation of inulin but not starch in this study, suggesting that the microbial changes in midlife associated with aging are still in initial stage, leading to a partial functional decline.
In this study, acetate was significantly reduced in midlife after inulin fermentation compared to young group. SCFAs have a number of potential roles in modulating metabolic health[19]. Acetate is most productive acid among SCFAs and appears to stimulate leptin secretion in adipocytes, involving energy balance and appetite,[16] conditioning immune cell in response to protect against T2D[55] and regulating blood pressure.[56, 57] And the impaired acetate production might be associated with weight gain in midlife.[58] Considering that approximately 44% acetate in plasma is microbiota-derived,[59] the reduced production of acetate by microbiota might be associated with the physiological changes in midlife which leading obesity[7], T2D.[8] and cardiovascular disease.[10]
Interestingly, contrast to significantly reduced inulin degradation and acetate production, the reduction in number of Bifidobacterium is not significant after inulin fermentation. Concerning its diversity of inulin catabolism[60] and its niche- and strain-specific acetate production,[61] the loss of certain Bifidobacterium species or strains that produce large amounts of acetate might be the reason. This is supported by a recent study that changes in the composition of Bifidobacterium species occur with ageing.[37] Addition to the roles of acetate in the healthy problem of midlife, such as obesity and T2D, it is reported that bifidobacteria-produced acetate improves intestinal defense mediated by epithelial cells and thereby protects the host against lethal infections.[62] Therefore, further study is needed to clarify the species of Bifidobacterium reduced in midlife.