Given the increased attention to the vital role the gut microbiota plays in human health, in vitro systems for controlled experimental investigation have been extensively developed and implemented. Most notable among them are bioreactors for propagating and maintaining microbial communities derived directly from human fecal samples. However, despite their versatility and functionalities, one common limitation is that these systems are typically dependent on additional specialized expensive lab equipment and setup, such as anaerobic chambers or multi-channel pumps 15 . Despite the commercialization of several models, most systems’ applications are limited within the lab of creation and the accessibility world-wide is usually restrained 33 . As such, one of our primary foci for MiCoMo design was to ensure the system is low-cost and can be easily established by most labs. The whole system costs ~$1,500 CAD and can be assembled by personnel with limited engineering experience with ease (A list of components and price can be found in Supplementary Information). The small working volume of reactors (30 ml) and the compact design also reduce the system footprint, allowing the whole MiCoMo to fit on a typical lab bench or within a biosafety cabinet. Compared to other small scale-systems, MiCoMo operates independently of anaerobic chambers, which are usually expensive and cumbersome to setup and maintain. Notably, MiCoMo is also equipped with pH control, which also makes it suitable for mimicking physiological conditions leading to a pH shift in the GI tract. Together, these features allow for easy replication across laboratories, as well as multiplexing capabilities by establishing multiple sets of MiCoMo systems in parallel within relatively small spaces, if desired.
Our validation experiments demonstrated that MiCoMo can maintain anoxic conditions at specific pH levels, leading to suitable growth conditions for two strict anaerobes and allowing for investigators to adjust the pH according to their own experimental needs. At a more fundamental level, the triplicate reactors of MiCoMo can be easily reconfigured to connect to each other in series instead of in parallel. Individually equipped with pH control system, these reactors could, when connected in series, mimic the human GI tract from stomach to colon by adjusting the pH setpoint and inoculating with different samples. This setup would enable the investigation of how the gut microbiota responds to perturbations along the GI tract.
When analyzing complex microbial community stability and structure, we were most interested in whether our system achieved a performance comparable to the currently available in vitro systems and animal models. Our observed Shannon index from fecal samples is comparable to previously published values that typically range from 4-6.5 for healthy individuals34. The decrease in alpha diversity and in observed ASVs for microbial communities grown in MiCoMo likely reflects a selection process by the specific growth conditions used (media, retention time, etc.) as well as the initial composition of the fecal inoculum. Notably, due to lack of incubation time (feed cycles were immediately started after inoculation), some slow-growing bacteria might have been washed off during the initial transition period before being able to adapt to the new ex vivo conditions. Such selection processes were commonly observed in other in vitro system as well15,35. Importantly, despite the decrease, MiCoMo-grown communities demonstrated an alpha diversity similar to that observed in previously reported in vitro systems after stabilization 15,34, indicating that MiCoMO was able to support growth of complex and diverse communities from a variety of fecal samples.
A big challenge for assessing stability of microbial communities in in vitro systems lies in the lack of a clear consensus for defining community stability and distinguishing natural variations within communities from major community shifts. Here, by adopting previously published analyzes and diversity metrics, we are able to directly compare our system to a previously validated in vitro system, such as the MBRA15. Notably, Auchtung et al. not only reported stability metrics of their in vitro system, but also analyzed and compared these metrics to those observed in mouse models36. It was reported that the six weaned mice with stable microbial communities analyzed by Auchtung et al. demonstrate a day-to-day variation in Bray-Curtis similarity (daily similarity) of 0.79 ± 0.06, and a between-replicate similarity of 0.71 ± 0.05. Meanwhile, the MBRA system had a daily similarity of 0.74 ± 0.05 and a between-replicate similarity of 0.54 ± 0.07 to 0.61 ± 0.08 during stable operations, depending on the volunteer. The MiCoMo system, with a daily similarity of 0.81 ± 0.07 during stable operation (all volunteers included; Day 5 – Day 14 for individual B and Day 3 – Day 14 for all other individuals) and a between-replicate similarity of 0.72 \(\pm\) 0.13, thus exhibited similar performance (no significance difference between MiCoMo and mice, unpaired t-test with p > 0.5 for both categories). This demonstrates that MiCoMo is able to support stable microbial community growth, with variations comparable with an in vivo mouse model and a previously reported in vitro systems, from various fecal inocula, after a timeframe of 3–5 days.
When analyzing the principal component analysis plots, we observed that the communities developed from the pooled sample (Individuals A and C) interestingly clustered closely and almost exclusively with one of its source donors, individual C, by Jaccard distance for all three replicates; whereas this was not the case for the Bray-Curtis distance. The difference between replicate reactors from the same pooled fecal sample emphasizes the need for technical replicates. Further, this distinction hints at the importance of using the number of individual human donor samples as the statistical inference unit when conducting large-scale perturbation analysis, as suggested by Walter et al.8, as opposed to only using the number of technical replicates (replicate mice or bioreactors with same inoculum).
Looking at the taxonomy of MiCoMo-grown microbial communities, except for a selected few known members of the gut microbiota, we limited the taxonomic assignment to the genus level, as there is extensive literature discussing the limitations of 16S rRNA sequencing with selected variable regions to reach species-level identification37,38.
We first report an overall decrease in the relative abundance of Firmicutes, likely due to their extreme intolerance to oxygen (loss of cultivability after < 2 min of oxygen exposure for some species has been reported39), in addition to possible nutrient preferences. Other validated in vitro systems have reported similar observations, with either a decrease in abundance or a complete loss of members of this phylum15,35. In addition, the expansion of facultative bacterial species belonging to the Proteobacteria phylum has been observed in various in vitro fermentation systems15,35, likely due to their high resilience to oxygen exposure and short doubling time40–42. Interestingly, we did not observe this phenomenon for most volunteers in MiCoMo after the first few days of inoculation. Rather, we observed an expansion of several known members of the gut microbiota, such as Bacteroides uniformis and B. thetaiotaomicron, both species being strict gut anaerobes29,43,44. These observations indicate that some of the underlying microbial interactions known to take place in the gut could also be occurring in MiCoMo, such as limiting the expansion of Proteobacteria. Importantly, MiCoMo does not seem to select for the most adaptable and aero-tolerant species, although these may establish their niche early on during the stabilization period.
In this paper, we demonstrate that MiCoMo is able to support stable and distinct microbial communities from different volunteers, using a previously validated culture medium, as a first proof-of-functionality of MiCoMo. However, the strength of the MiCoMo system lies in its versatility: with user-customizable pH setpoint, gas sparging and feed schedules, one can easily adjust the MiCoMo environment to better accommodate individual-specific gut conditions. For instance, the pH setpoint can be decreased along with a gas sparging with increased interval in order to mimic the gut environment of IBD patients with reduced pH and increased oxygen concentration45,46. In order to better support a mucosal microbial communities, mucin could also be supplemented into the system, as previously done in the SHIME system47.