Authors: Alyssa H. Mitchell, Tami D. Lieberman
Categories: Commentary, Biological Sciences
Source: Proceedings of the National Academy of Sciences of the United States of America
A major result of the human microbiome project was that individuals typically harbor remarkably different microbiomes at the species level, despite having microbes that, on the whole, perform similar functions (1). This individuality persists even within households (2, 3) and at multiple levels of genetic resolution—family members often have distinct strains of the same species (4, 5). The basis of this individuality has remained elusive and poses a challenge for the design of microbiome-based It remains difficult to predict whether a probiotic strain will successfully engraft in a given host.
While an attractive explanation for individuality is that colonization can only occur in early life, it is now clear that individuals acquire new species and strains throughout life (6, 7). However, whether these new colonizations can occur only at specific times or only by specific species remains unclear. The complexity of the gut microbiome, as well as the large differences between human microbiomes and animal model microbiomes, has historically made it challenging to tease apart the many potential factors that limit colonization—including exposure, metabolic or nonmetabolic (e.g., phage) competition with resident strains, and host features (including diet, genetics, and immune history).
Recently, empirical studies have indicated a role of propagule pressure—the combined effect of invading population size and frequency of exposure—in determining the outcome of probiotic success (8, 9). However, understanding the physiological basis of this dose dependence has been difficult, and no study has examined dose dependence across different colonizing strains and microbial communities. In PNAS, Goldman et al. (10), elegantly leverage stable in vitro microcosms derived from human gut microbiome samples, innovative experimental design, and mathematical modeling to investigate the extent and drivers of dose-dependent dynamics.
The controlled study of invasion in microbial communities that reflect the diversity found in human guts has remained Synthetic communities assembled in the laboratory have far fewer species stably coexisting than found in nature, experiments in mouse microbiomes do not allow separation of invasion and survival due to the coprophagic nature of their hosts, and experiments in humans are difficult to perform at scale. To obtain complex communities for controlled experimentation, Goldman et al. leveraged recent findings showing that serial batch culture of stool for 15 passages results in communities with high diversity and stable relative abundances of species (11). Using this process, the authors derived eight stable in vitro communities, each from the stool of a single human donor.
The authors then systematically cocultured pairs of these distinct, realistically complex communities across initial mixture ratios spanning six orders of magnitude (Fig. 1A). After approximately 40 generations (5 passages) of equilibration, they used 16S ribosomal RNA amplicon sequencing to determine the relative abundances of amplicon sequence variants (ASVs). While not every strain can be tracked with the ASV approach (because multiple strains from the same or different communities may share a sequence), this approach enables tracking of individual taxa in these complex communities.
Fig. 1. The high number of species in complex microbiomes decreases the availability of private resources and increases the likelihood of dose-dependent dynamics. (A) To systematically study the role of dose-dependent invasion into complex microbiome communities, Goldman et al. (10) mixed stable human stool-derived in vitro communities across ratios spanning six orders of magnitude. Similar follow-up experiments were performed with individual species. (B) 16S rRNA sequencing was performed to track relative abundances of individual sequence variants (ASVs, roughly corresponding to species) in these mixed communities. Many ASVs were not unique to a sample and thus could not be tracked across mixtures (not shown). The remaining ASVs could be categorized as noisy, dose-independent, or dose-dependent. Cartoon examples shown with the percent of all ASVs assigned to each category. (C) A two-species consumer–resource model revealed that competition for shared resources can drive dose dependence. This is further explored in experiments. (D) Gradually increasing complexity in both models and experiments, the authors demonstrated that increasing the number of species in a community directly enhances colonizer dose dependence by limiting available exclusive (unshared) resources.
While neutral ecology theory—which assumes no interaction between strains (12), metabolic or otherwise—predicts perfect coexistence and thus peak diversity at intermediate mixture ratios (1:1), all experimental mixtures showed lower-than-expected diversity. This consistent result indicates that the final community is not simply a passive mixture of the communities, thereby confirming the expectation that nonneutral interactions are common in gut microbial communities.
Investigation of individual ASVs demonstrated a variety of trends (Fig. 1B). Some ASVs always reached the same equilibrium frequency regardless of the initial starting ratio (“dose-independent”), while others showed nonmonotonic responses across mixture ratios (“noisy”) that underscore the complexity of microbial ecosystems. Intriguingly, across all community pairs tested, a substantial number of ASVs showed “dose-dependent” colonization patterns proportional to propagule pressure. Importantly, the authors found many cases where the same ASV showed distinct patterns across different community contexts.
The ubiquitous presence of dose-dependent colonizers across all community mixtures, often comprising a substantial portion of the equilibrated communities, highlighted the importance of understanding what drives dose-dependent behavior in complex microbial systems. To understand whether metabolism could explain these dynamics, the authors computationally explored a simple consumer resource model, wherein species compete for both shared and exclusive resources. This model revealed that dose dependence emerges from metabolic competition when niche overlap is high—that is when few resources are available exclusively to the invader (Fig. 1C). Notably, the model predicts that dose dependence is a transient phenomenon; all simulations eventually reach equilibrium relative abundances, except when species are exactly equivalent. However, as niche overlap increases, the effects of dose dependence are amplified and longer-lived.
Notably, the authors went beyond just presenting this potential explanation and performed experiments that strongly support that metabolic overlap is responsible for at least some of the dose dependence in the gut microbiome. The authors isolated individual strains from these communities and again performed pairwise mixing experiments across a range of mixture ratios. Consistent with expectation from modeling, these results revealed stronger dose-dependent dynamics between strains of the same taxonomic order, which are expected to consume similar resources, compared to strains from different orders.
To quantitatively assess shared versus exclusive resource usage between two strains, they developed a clever and simple assay using spent media; strains were determined to have more exclusive resources if they grew well in the spent media of other strains. The authors determined that the extent of shared resource dependence did not fully predict dose dependence. The key to predicting dose dependence was incorporation of the rates of resource consumption, estimated from another creative and straightforward assay—the depletion of peaks in untargeted metabolomics. Among strains with high niche overlap, those that consumed shared resources more slowly exhibited stronger dose dependence.
Building upon these mechanistic insights, Goldman et al. extended their investigation to more complex communities. The authors used computational modeling to demonstrate that adding an additional species can decrease the availability of exclusive resources and thereby further increase dose dependency (Fig. 1D). This theoretical prediction was experimentally validated using invasion experiments against single strains, communities of 3 to 4 strains, and the more complex donor-derived consortia, demonstrating increasing dose dependence. These results reveal an important ecological As microbial community complexity increases, the scarcity of species-exclusive resources naturally leads to stronger dose-dependent colonization dynamics.
The findings of Goldman et al. affirm an intuitive ecological principle—the more that a species is challenged by a resident community, the more opportunities it needs to establish itself. This work represents a significant advance because of its demonstration of dose dependence in realistically complex microcosms derived from real human gut microbiota, its clarification that metabolic competition for shared resources is sufficient to produce these dynamics, and theoretical framework extending from pairwise competition to complex communities. By demonstrating that dose dependence is driven by competition for shared resources and amplified in more diverse communities, this work provides a framework for predicting colonization dynamics based on metabolic relationships between incoming strains and resident communities.
Moreover, this study serves as a powerful example of how systematic experimental approaches capable of driving mechanistic understanding can be performed with complex communities. The various elegant experimental tricks used here—including controlled invasion of single species into stable consortia derived from human gut communities and untargeted mass spectrometry to compare resource consumption rates across strains and conditions—will inspire others to address mechanistic questions in diverse communities.
The dose-dependent results presented by Goldman et al. primarily apply to transient colonization, as all modeled communities eventually reach equilibrium. Yet, the ability to predict even transient colonization would have a large impact on the design of microbial therapies and basic microbiome science. While sustained engraftment is often prioritized, even transient probiotic colonization could restructure microbial communities with lasting effects. This is powerfully illustrated in the treatment of Clostridioides difficile infection, where the pathogen exploits disruptions in the commensal microbiome to establish itself and maintains an altered metabolic landscape that blocks commensal recovery. Even temporary interventions that break this cycle can allow commensals to regain their foothold, showing how transient effects can trigger lasting community shifts. During transient colonization, microbes may also alter community function, evolve to increase competitiveness in their new host, or modulate host immunity—all with potential long-term consequences.
Beyond these immediate therapeutic implications, the experimental systems developed and results presented here open new avenues for exploring other phenomena in community assembly that remain unexplained. One particular area worth investigating further is the “noisy” colonization patterns that suggest the presence of higher-order or more complex interactions (13). Another opportunity for future study would be the use of metagenomics to distinguish the dynamics of strains found in both initial communities and not resolvable at the ASV level. Continued mechanistic insights into colonization dynamics, building on the work of Goldman et al.’s framework, will not only advance rational therapeutic design but also deepen our understanding of why human microbiomes remain distinctly individualized despite shared environment and biology.
A.H.M. and T.D.L. wrote the paper.
The authors declare no competing interest.