FMT of human feces on C57BL/6 mice has variable effects on muscle strength, manifested as strong, moderate, weak changes
To identify the gut microbes responsible for regulating muscle strength, a mixture of human gut microbiome was transplanted into C57BL/6 mice whose gut microbiome had been depleted by administering a combination of three broad-spectrum antibiotics and antifungal drugs (Fig. 1a). To assess the muscle strength, we used two muscle strength tests: Rotarod and Wire suspension. The Rotarod and Wire suspension is commonly used to measure the motor co-ordination and balance of the mice.17 First, we compared changes in Rotarod record in the same mice before and after FMT to evaluate precisely the influence of the gut microbiome on muscle strength while excluding genetic factors. The replacement of their original gut microbiome with FMT resulted in varying effects on muscle strength (Fig. 1b-c). Over the three-month experimental period, the changes in Rotarod record could be categorized into three groups: the strengthened group (Rota_SMS; n = 7; Δ latency to fall = 18.6 ± 6.2 sec), the moderate group (Rota_MMS; n = 7; Δ latency to fall = − 0.3 ± 1.4 sec), and the weakened group (Rota_WMS; n = 7; Δ latency to fall = − 42.6 ± 5.6 sec) (Fig. 1b and c). Second, we compared changes in Wire suspension record in the same mice before and after FMT (Fig. 1d-e). Over the three-month experimental period, the changes in Wire suspension record could be categorized into three groups: the strengthened group (Wire_SMS; n = 8; Δ hanging time = 24.5 ± 4.7 sec), the moderate group (Wire_MMS; n = 8; Δ hanging time = − 7.2 ± 2.6 sec), and the weakened group (Wire_WMS; n = 8; Δ hanging time = − 28.8 ± 2.9 sec) (Fig. 1b-c).
We checked the body weight, blood glucose and blood lipids of mice in the two experimental groups before and 3 months after FMT (Supplementary Fig. 1). Overall, 3 months after FMT, body weight and blood glucose increased, reflecting the natural aging process of the mice in the experiment (Supplementary Fig. 1a, b, g and h). However, it is worth noting that in both experiments, high density lipoprotein (HDL) cholesterol decreased except in the group where the muscle strength increased (Supplementary Fig. 1e and k).
FMT altered the composition of the gut microbiome of each mouse
The phenotypic changes of individual mice after FMT prompted us to investigate the changes of the gut microbiome isolated from fecal samples. The gut microbiome before and after FMT were analysed by 16S rRNA amplicon sequencing. The V3-V4 sites of the 16S rRNA genes of the isolated genomic DNAs of the gut microbiome after filtering sequencing errors, represented as operational taxonomic units (OTUs), were used to identify individual microbes by matching the 16S rRNA sequences with the SILVA reference (region V3-V4) database (https://www.arb-silva.de/) (Supplementary Table S1). All of the identified 16S rRNA sequences were classified to the species level.
As shown in Fig. 2a, Supplementary Fig. S2a, and Supplementary Table S2, in the Rotarod experimental group, the species richness measured by Fisher and ACE dramatically improved, indicating that the number of microbial species constituting the gut microbiome increased with FMT. Despite the increase in microbial species in the gut microbiome, the species evenness measured by Simpson, Evenness, InvSimpson, and Shannon indices indicated that the patterns of species distribution did not change significantly (Fig. 2a, Supplementary Fig. S2a, Supplementary Table S2).
In accordance with the species richness, phylogenetic classification showed that the microbial compositions dramatically altered and diversified after FMT. The gut microbiome was mainly consisted of Bacteroidetes (57.120%) and Firmicutes (39.869%) before FMT (Fig. 2b). However, FMT increased the relative abundance of Verrucomicrobia (9.029%) and decreased the abundance of Bacteroidetes (50.352%) and Firmicutes (36.649%) (Fig. 2b). The microbial distribution at lower classification levels are shown in Supplementary Fig. S2b-d. The analyses of β-diversities (Fig. 2c and Supplementary Fig. S2e-g), phylogenetic trees (Fig. 2d), and co-occurrence network (Fig. 2e and Supplementary Table S3) further confirmed the distinct nature of the gut microbiome before and after FMT in the Rotarod experimental group.
Similar to the Rotarod experimental group, in the Wire suspension experimental group, the species richness measured by Fisher and ACE dramatically improved, and the species evenness measured by Simpson, Evenness, InvSimpson and Shannon indices were not changed significantly (Fig. 2f, Supplementary Fig. S3a and Supplementary Table S2). The gut microbiome before FMT, which was mainly consisted of Bacteroidetes (54.994%) and Firmicutes (41.866%), showed an increase in the abundance of Verrucomicrobia (9.060%) and a decrease in the abundance of Bacteroidetes (49.785%) and Firmicutes (37.415%) after FMT (Fig. 2g). The microbial distribution at lower classification levels are shown in Supplementary Fig. S3b-d. The analyses of β-diversities (Fig. 2h and Supplementary Fig. S3e-g), phylogenetic trees (Fig. 2i), and co-occurrence network (Fig. 2j and Supplementary Table S3) further confirmed the distinct nature of the gut microbiome before and after FMT in the Wire suspension experimental group. Thus, it was established that FMT effectively replaces the gut microbiota in both experimental groups.
The fecal microbiome cluster analysis confirms the association between gut microbiome clustering and muscle strength, but it is not sufficient
We confirmed that both the muscle strength of mice and the composition of the gut microbiome were altered through FMT. Specifically, we analyzed the possible association between the gut microbiome and muscle strength using fecal samples. In the Rotarod experimental group, the species richness measured by Fisher and ACE indices dramatically improved in all groups. When comparing between groups, the richness indices of the Rota_SMS group were decreased compared to the other groups (Fig. 3a, Supplementary Fig. S5a and Supplementary Table S2). Despite the difference in richness indices of the gut microbiome between groups, the species evenness measured by Simpson, Evenness, Inverse Simpson, and Shannon indices indicated that the patterns of species distribution did not differ significantly in all groups (Fig. 3a and Supplementary Table S2). Phylogenetic classification showed that the microbial compositions dramatically altered and diversified after FMT (Fig. 3b). The gut microbiome was mainly consisted of Bacteroidetes (58.434% in the Rota_SMS group, 50.874% in the Rota_MMS group, and 62.053% in the Rota_WMS group) and Firmicutes (38.199% in the Rota_SMS group, 45.477% in the Rota_MMS group, and 35.930% in the Rota_WMS group) before FMT (Fig. 3b and Supplementary Table S4). However, FMT caused the proliferation of Verrucomicrobia (7.221% in the Rota_SMS group, 13.833% in the Rota_MMS group, and 6.031% in the Rota_WMS group), and thus reduction of the abundances of Bacteroidetes (52.640% in the Rota_SMS group, 44.480% in the Rota_MMS group, and 53.937% in the Rota_WMS group) and Firmicutes (35.574% in the Rota_SMS group, 38.509% in the Rota_MMS group, and 35.864% in the Rota_WMS group) (Fig. 3b and Supplementary Table S4). The microbial distribution at lower classification levels is shown in Supplementary Fig. S5b and Supplementary Tables S5–S9. Although alpha (α) diversity indices showed slight evidence of an association between the gut microbiome and muscle strength (Fig. 3a, Supplementary Fig. S5a, and Supplementary Table S2), phylogenetic trees (Supplementary Fig. S5c), co-occurrence network analysis (Supplementary Fig. S5d and Supplementary Table S3), and the beta (β) diversity analyses (Fig. 3c-d and Supplementary Fig. S6-7) did not show an association.
In the Wire suspension experimental group, the species richness measured by Fisher and ACE significantly improved in all groups after FMT, while the species evenness measured by Simpson, Evenness, InvSimpson, and Shannon indices did not differ significantly between groups (Fig. 3e, Supplementary Fig. S8a, and Supplementary Table S2). Additionally, there was no consistent pattern observed according to changes in Wire suspension records. However, phylogenetic classification showed that the microbial compositions dramatically altered and diversified after FMT (Fig. 3f). The gut microbiome was mainly consisted of Bacteroidetes (53.026% in the Wire_SMS group, 58.645% in the Wire_MMS group, and 53.312% in the Wire_WMS group) and Firmicutes (44.074% in the Wire_SMS group, 37.880% in the Wire_MMS group, and 43.644% in the Wire_WMS group) before FMT (Fig. 3f and Supplementary Table S10). However, FMT caused the proliferation of Verrucomicrobia (8.463% in the Wire_SMS group, 10.368% in the Wire_MMS group, and 8.348% in the Wire_WMS group), and thus reduction of the abundances of Bacteroidetes (47.909% in the Wire_SMS group, 50.059% in the Wire_MMS group, and 51.388% in the Wire_WMS group) and Firmicutes (40.619% in the Wire_SMS group, 35.485% in the Wire_MMS group, and 36.142% in the Wire_WMS group) (Fig. 3f and Supplementary Table S10). The microbial distribution at lower classification levels are shown in Supplementary Fig. S8b and Supplementary Tables S11–S15. Neither α-diversity indices (Fig. 3e, Supplementary Fig. S8a and Supplementary Table S2) nor phylogenetic trees (Supplementary Fig. S8c), co-occurrence network analysis (Supplementary Fig. S8d and Supplementary Table S3), nor β-diversity analyses (Fig. 3g-h and Supplementary Fig. S9-10) showed an association between gut microbiome and muscle strength.
The gastrointestinal microbiome analysis elucidated the association of the gut microbiome with muscle strength
We further analysed a possible association between the gut microbiome and muscle strength by using GI contents. In the Rotarod experimental group, the species richness and distribution pattern of the gut microbiome isolated from the GI contents (GI gut microbiome) were very different from those isolated from feces (fecal gut microbiome) (Fig. 4a, Supplementary Fig. S11a and Supplementary Table S2). The composition of the gut microbes constituting the gut microbiome was not only different from that of the fecal gut microbiome, but there was also a clear group difference (Fig. 4b, Supplementary Fig. S11b–d, and Supplementary Tables S4–S9). The α-diversities (Fig. 4a, Supplementary Fig. S11a, and Supplementary Table S2), β-diversities analyses (Fig. 4c-d, and Supplementary Fig. S12), phylogenetic trees (Fig. 4e), and co-occurrence network analysis (Fig. 4f and Supplementary Table S3) also confirmed that the gut microbiome of the three groups differed from each other.
Similar to the Rotarod experimental group, the α-diversity indices (Fig. 5a, Supplementary Fig. S13a, and Supplementary Table S2), the composition of the gut microbiome (Fig. 5b, Supplementary Fig. S13b–d, and Supplementary Tables S10–S15), the β-diversity analyses (Fig. 5c-d, and Supplementary Fig. S14), phylogenetic trees (Fig. 5e), and co-occurrence network analysis (Fig. 5f and Supplementary Table S3) isolated from the GI contents (GI gut microbiome) in the Wire suspension experimental group were very different from those isolated from feces (fecal gut microbiome). Moreover, it was also confirmed that the gut microbiome of the three groups isolated from the GI contents differed from each other.
We believe that through the analysis of the gastrointestinal microbiome, we have visualized the association between the gut microbiome and muscle strength. Our results demonstrate that the gastrointestinal microbiome is more diverse and accurately represents the real gut microbiome compared to the fecal gut microbiome. This finding is in perfect agreement with recent research indicating that the fecal gut microbiome does not fully represent the entire gut microbiome.18
The gut microbiomes that regulate muscle strength were identified through differential analysis
After identifying the association of the GI gut microbiome with muscle strength, we aimed to identify the individual gut microbial species associated with muscle strength. The associative relationship at the species level was analyzed by utilizing the DESeq2 method, which exhibited differential abundance of gut microbes as a fold change in the gut microbiome of each group. In the Rotarod experimental group, a total of 9 bacterial species were found to have significantly different abundance (adjusted P-value of 0.05 or lower) between the Rota_SMS and Rota_WMS groups (Fig. 6a-b). Among the 9 bacterial species, 7 were enriched in the Rota_SMS group, while 2 were enriched in the Rota_WMS group (Fig. 6a-b). In the Wire suspension experimental group, a total of 9 bacterial species were found to have significantly different abundance (adjusted P-value of 0.05 or lower) between the Wire_SMS and Wire_WMS groups (Fig. 6c-d).
Among the 9 bacterial species, 4 were enriched in the Wire_SMS group, while 5 were enriched in the Wire_WMS group (Fig. 6c-d). Among the gut microbial species selected through the two experimental methods, the three species that had a common effect on muscle strength improvement were Lactobacillus johnsonii, Limosilactobacillus reuteri, and Turicibacter sanguinis. All three gut microbial species were abundantly present in the gut microbiome of the mice groups, affecting the muscle strength of their hosts, and were also linearly correlated with the increase of muscle strength in an abundance-dependent manner.
Lactobacillus johnsonii and Limosilactobacillus reuteri increased the mouse muscle strength
We then performed two commonly used motor behavioral tests to demonstrate the significant improved muscle strength effect of L. johnsonii and L. reuteri (Fig. 7a). The Rotarod was conducted at one-month intervals while administering L. johnsonii and L. reuteri for 3 months. In quantitative measurements, there was a difference in the time to fall between the control group (CTRL) and the bacterial administration group, with the time to fall in the bacterial administration group slightly increasing (Fig. 7b). In the Wire suspension, the initial hanging time was set to 5 min, but it was changed to 10 min because mice in all experimental groups stayed suspended for more than 5 min (Fig. 7c). The hanging time increased after one month of oral administration of L. reuteri, but after that, the hanging time tended to decrease compared to the CTRL (Fig. 7c). The body weights of mice were monitored monthly (Fig. 7d). Overall, the bacteria administration group’s body weight decreased by 3.83 ~ 21.84% (Fig. 7d). Among them, the L. johnsonii (LJ) and mixture of L. johnsonii and L. reuteri (LJ + LR) groups showed high weight loss rates of up to 21.11 ± 3.46, 21.84 ± 2.56%, respectively (Fig. 7d). The amount of muscle weight according to LJ, LR group were not significant compared to the CTRL (Fig. 7e). But in LJ + LR group, the muscle weight increased by 157.47 ± 41.48%. In addition, the muscle weight of the LJ + LR group was greater than that of the LJ group (Fig. 7e).
Follistatin (FST) is a secreted glycoprotein that regulates the activity of the TGF-β family and strongly inhibits myostatin.19,20 Genetic deletion of FST is associated with muscle hypotrophy, while transgenic overexpression of FST in mouse muscle results in increased muscle mass.19,20 The expression of FST in LJ group increased by 99.20 ± 3.07% compared to the CTRL group, but the other groups were not significant (Fig. 7f). IGF1 is a potent growth factor affecting muscle growth during development and acting both systemically as a typical hormone produced by the liver under the CTRL of growth hormone and locally as a paracrine/autocrine factor produced by skeletal muscle.21 IGF-I is known to be a major positive regulator of skeletal muscle mass. Indeed, IGF-I overexpression specifically in skeletal muscle increases skeletal muscle mass.22 The expression of IGF-1 in LJ + LR group increased by 164.00 ± 56.03% compared to the CTRL group, but the other groups were not significant (Fig. 7f).
The cross-section area of soleus (Sol), gastrocnemius (GA) and extensor digitorum longus (EDL) muscle fiber was increased in all test groups compared to CTRL (Fig. 7g-h). According to the MT staining, collagen accumulation was observed in muscle fibers in test groups compared to CTRL (Fig. 7h).
The LJ + LR group showed a decrease in Triglyceride (TG), total cholesterol (TCHO) and low-density lipoprotein-cholesterol (LDL-CHO) at T3 compared to CTRL (Supplementary Fig. S15a-d). It suggests that administration of L. johnsonii and L. reuteri helps to maintain vascular health in mice. The interleukin (IL)-6 and IL-1β, the cytokines than play important role in inflammatory and immune responses, showed that IL-6 content increase only in LJ groups, and IL-1β did not differ compared to CTRL (Supplementary Fig. S14e and g). The IL-8 content, a cytokine than occurs primarily in the inflammatory response, was increased in the LJ group and LJ + LR group at T3 (Supplementary Fig. S14f). C-reactive protein (CRP) is a crucial blood marker of inflammation that is produced in the liver and primarily regulated by the proinflammatory cytokine IL-6.23 Overall, the CRP content increased in the test groups, and in particular, the LJ + LR group increased the CRP content in the T1, T2 and T3 (Supplementary Fig. S14h).