Spontaneous recovery of gut microbiota following antibiotic treatment in aged mice
Previous studies have shown that antibiotic treatment can disrupt the gut microbiota in young hosts, leading to a substantial decrease in microbial load [33, 34] and microbiota diversity [11, 35, 36]. Afterwards, there is a long-term restorative process, during which the gut microbial ecosystem is temporarily imbalanced and significantly distinct from the original state [11, 35-37]. However, little is known about the recoverability of gut microbiota in aged hosts.
To study the disruption of gut microbiota by antibiotics and the ensuing spontaneous recovery in aged hosts, we treated aged (20-month-old) female C57BL/6J mice with oral gavage of a broad-spectrum antibiotic cocktail (ampicillin, vancomycin, neomycin, and metronidazole) at a high dose (ABH group) or low dose (ABL group) or with water (Ctrl group). Fecal samples were collected before treatment and then regularly collected until 56 days after treatment (Figure 1A). We found that antibiotic treatment led to a substantial but transient decrease in bacterial load (the total copy number of the 16S rRNA gene) in feces (Figure 1B). In contrast, antibiotic treatment had a significant and long-term (8 weeks after treatment) impact on the number of observed species (Figure 1C) and the Shannon diversity index (Figure 1D) of gut microbiota.
The alpha diversity of gut microbiota in aged mice did not recover to the baseline level, suggesting the elimination of community members during the antibiotic treatment [36]. These observations are consistent with previous studies in young mice [11] and in an independent experiment that we carried out in 2-month-old mice (Figure S1).
Since heterochronic FMT is usually the only option for elderly patients in clinical situations, it is essential to investigate whether FMT donors with significantly different microbiota are equally successful in facilitating microbiota restoration in recipients. To verify whether donor-recipient compatibility affects the outcome, two parallel groups of aged mice were given a high-dose antibiotic cocktail treatment and then received FMT from age-matched 20-month-old donor mice (FMT-M), or aged-unmatched 2-month-old donor mice (FMT-UM). FMT was performed by oral gavage for three consecutive days after antibiotics treatment (Figure 2A and Methods). We found that the two groups of donor mice (2 months old vs. 20 months old) had very distinct microbiota (Figure S3A and S3B). Differential abundance analysis using LEfSe (Linear discriminant analysis Effect Size) revealed that Prevotellaceae and Bilophila were enriched in the baseline microbiota of young mice, whereas Bacteroides and Desulfovibrio were enriched in aged mice (Figure S3C).
The change in the number of observed species (Figure 2B) and Shannon index (Figure 2C) showed that FMT substantially accelerated the recovery of gut microbiota diversity in comparison with spontaneous recovery. The gut microbiota diversity of aged mice that received FMT was fully restored by Day 21 after antibiotic treatment (Figure 2B and 2C). This is consistent with a previous study, which showed that autologous FMT improved the post-antibiotic reconstitution of the gut microbiome in young hosts (mice and human volunteers) [11].
On Day 21, the microbiota composition of mice that received FMT was more similar to the baseline of young or aged mice (average microbiota composition of donors on Day 0), in comparison to mice in the SR group (Figure 2D). In addition, the gut microbiota of mice in the FMT-M group was closer to the baseline microbiota of aged donors, while there was larger variation in the gut microbiota of mice in the FMT-UM group, with one sub-group having a microbiota close to the baseline of young donors (Figure 2E). Kinetics variation of microbiota composition recovery after antibiotic treatment has been observed in previous studies in conventional and humanized young mice [36]. However, our results revealed a novel finding that aged mice in the FMT-M group had more consistent compositional kinetics than those in the FMT-UM group. The individual differences in gut microbial composition on Day 21 between mice in both the FMT-M and FMT-UM groups were not fully attributable to cage effects (Figure S4).
We demonstrated that age-unmatched donors were as effective as age-matched donors in restoring the diversity of the microbiota, with both groups showing a significant increase in diversity when compared with the SR group. Moreover, the composition of newly established microbiota was donor dependent. We followed five recipient mice in each FMT group until Day 56 (Figure S5). For each mouse, we calculated the Bray-Curtis dissimilarity between its gut microbiota at different time points and the original baseline gut microbiota (i.e., Day 0). We found that the gut microbiota of mice in the FMT-M group was closer to the baseline of aged mice than the other two groups throughout the entire experiment (Figure 2F and Figure S5A).
Long-term effects of FMT on the gut metagenome and colon gene expression
Previous research has shown that that functional restoration of bacteriomes and viromes revealed by metagenomic sequencing could be an indicator of successful FMT [41]. To evaluate the long-term effects of FMT in aged mice with different donors, we performed shotgun metagenomic sequencing of Day-56 fecal samples from each treatment group (SR, FMT-M, and FMT-UM) together with the baseline of 20-month-old mice (matched donors, MD) and 2-month-old mice (unmatched donors, UMD). We found that 35 out of 219 functional pathways were differentially expressed between the SR group and MD (Figure 3A). We then examined the levels of these pathways in the FMT-M and FMT-UM groups after 56 days of recovery. We found that most of the 35 pathways that did not spontaneously recover to baseline levels were restored in the FMT-M group. In contrast, FMT-UM was not as effective in facilitating recovery of microbiota function. When all 219 functional pathways were taken into consideration, endpoint samples of both the FMT-M and FMT-UM groups were closer to the baseline samples compared with samples from the SR group on Day 56 (Figure S6).
The shift in the transcriptome of the host following antibiotics treatment can be dramatic [11, 38, 42], and the response is determined by the combined effects of microbiota depletion and the remaining antibiotic-resistant microbes as well as the direct effects of antibiotics on host tissues [42]. It was shown that autologous FMT in young mice helped restore the gene expression profile across GI tracts to some extent [11]. However, the effects of FMT donor on the transcriptional profiles of aged hosts remain unknown. We performed RNA sequencing (RNA-seq) of upper colon samples of aged mice before treatment (Baseline), Day 1 after antibiotics (ABH-Day1), and at the end point (Day 56) for the three treatment groups (SR, FMT-M, and FMT-UM). We first examined the restoration of differentially expressed genes (DEGs) between ABH-Day1 and the Baseline on Day 56 in the SR, FMT-M, and FMT-UM groups (Figure 3B). More genes that were affected by antibiotics treatment were reverted to the Baseline configuration in the FMT groups than in the SR group, suggesting that FMT facilitated the restoration of the host transcriptome, which is consistent with previous reports in young mice and human donors [11]. We then examined the DEGs between each group and the Baseline and the overlap among them (Figure 3C and Figure S7). Interestingly, a large fraction of DEGs were not shared between the different treatment groups (SR, FMT-M, and FMT-UM), which demonstrated that the transcriptional landscapes on Day 56 were highly dependent on FMT and the choice of donor. We also found that the FMT from age-matched donors did not outperform that from unmatched donors in the reversion of the transcriptional landscape (the number of total DEGs compared with the Baseline was 174 in FMT-M and 98 in FMT-UM).
The resilience of transplanted gut microbiota against colon inflammation
Whether the transplanted microbiota is resilient against perturbations to gut homeostasis reflects its long-term stability. We utilized a self-limiting colon inflammation model induced by administration of 3% dextran sulfate sodium (DSS) in the drinking water of the mice for 7 days, which is widely used to induce acute colitis in rodents [43, 44]. Previous studies had shown that DSS-induced-colitis caused changes in the gut microbiota composition, which were spontaneously restored to the original state after the end of DSS treatment [45-47].
In the sections above we showed that the microbiota diversity returned to baseline levels on Day 21 in the FMT-M and FMT-UM groups (Figure 2B and 2C). Also, microbial compositions largely stabilized after Day 21 (Figure S5). Hence, we chose Day 26 as the starting point for DSS treatment. From Day 26 to Day 31 after antibiotic treatment, aged recipients of FMT-M or FMT-UM were given 3% DSS treatment for seven consecutive days and then switched back to normal drinking water for 42 days (Figure 4A). Aged recipient mice of FMT-M or FMT-UM both showed typical symptoms of acute colitis as shown by weight loss (Figure S8A) and diarrhea and blood in feces (Figure S8B). These disease symptoms slowly resolved after the mice were switched back to normal drinking water, which is the same as what was reported in young and non-transplanted mice [48-50].
PCoA based on Bray-Curtis dissimilarity was used to compare the changes in microbiota composition in FMT recipients with or without DSS treatment (Figure 4B and 4C). For mice in the FMT-M-DSS group, the microbiota restoration process was only briefly disrupted by the induced inflammation; the microbiota composition after Day 53 was similar to that of FMT-M group (without DSS treatment) (Figure 4B). At the end point of the experiments, the microbiota composition of FMT-M mice and that of FMT-M-DSS mice showed no significant difference in the distance to that of their aged donors (Figure 4D).
In contrast, for mice in the FMT-UM-DSS group, the microbiota composition after DSS treatment gradually shifted toward the original baseline of aged mice (Figure 4C). At the end point, compared with the FMT-UM mice, the microbiota composition of FMT-UM-DSS mice was significantly further from the baseline of its young donors (Figure 4D). These observations suggested that under the conditions of colon inflammation, the transplanted gut microbiota from age-unmatched donors may not be as stable as that from age-matched donors.