What is already known
The human gut comprises hundreds of microbial species which create a complex and interdependent metabolic network. More than half of the metabolites consumed by these gut microbes result from microbial metabolism, where the waste generated by one species serves as nutrients for others. Despite this interdependence, accurately measuring and thoroughly understanding these cross-feeding interactions still pose significant challenges.
What this research adds
A recent study has introduced a Metabolite Exchange Score (MES) to measure the interactions between gut microorganisms. By employing metabolic models of prokaryotic metagenome-assembled genomes from more than 1600 individuals, MES enables identification and ranking of metabolic interactions influenced by the absence of cross-feeding partners. Specifically, this method identified a lack of species capable of consuming hydrogen sulfide as the primary distinctive microbiome characteristic of Crohn’s disease.
Conclusions
The conceptual framework presented in the study will help prioritizing in-depth analyses, experiments, and clinical targets. Focusing on restoring microbial cross-feeding interactions emerges as a promising mechanism-informed strategy to reconstruct a healthy gut ecosystem.
The interdependence among species can make microorganisms susceptible to local extinction if a partner is lost, unless alternative species are present to fill that ecological niche. In such a scenario, having functionally redundant species capable of producing or consuming the same nutrients proves advantageous for the host. Although it is known that robust human gut microbiomes are characterized by high functional redundancy, the health implications of redundancy in metabolic interactions remain largely unexplored.
Re-establishing the diversity of cross-feeding microbial partners presents a logical yet largely unexplored approach to combat a broad spectrum of diseases associated with an imbalanced gut microbiome.
Foster and colleagues, recently published a study on Nature communications journal, where they present a scoring system for metabolite exchange, designed to pinpoint the microbial cross-feeding interactions most affected in diseases.
By applying a conceptual framework to an integrated dataset of 1661 publicly available stool metagenomes, spanning 15 countries and 11 disease phenotypes, the authors identified both established and novel associations between the microbiome and diseases. Notable findings include a connection between colorectal cancer and microbial ethanol metabolism, a link between rheumatoid arthritis and microbially-derived ribosyl nicotinamide, and associations between Crohn’s disease and specific bacteria metabolizing hydrogen sulfide.
This scoring system facilitates the quantification and identification of context-dependent disruptions of microbial interactions, offering potential targets for microbiome-based medicines.
Quantification of potential cross-feeding interactions
To understand the connection between cross-feeding interactions and disease, the Metabolite Exchange Score (MES) was designed. MES is calculated by multiplying the diversity of taxa predicted to consume and taxa predicted to produce a specific metabolite, then normalizing it by the total number of involved taxa.
The potential production, consumption, and exchange of metabolites by each microbiome member, for which metagenome-assembled genomes (MAGs) can be reconstructed, is estimated through metabolic modeling. Metabolites with high MESs are likely crucial components in the microbial food chain; whereas metabolites with an MES of zero are either not produced or not consumed by any community member.
By comparing MESs for each metabolite across healthy and diseased microbiomes, one can rank and pinpoint the metabolites most affected by the loss of cross-feeding partners. Once prioritized with MESs, it becomes possible to integrate taxa abundances and their estimated metabolic fluxes to identify a consortium of species acting as the primary producers or consumers of the targeted metabolites. This approach has been proposed as a strategy for hypothesis generation, to guide new discoveries, targeted experiments, and clinical trials.
Key metabolic interactions within gut microorganisms in health and disease: meta-analysis of 1661 microbiomes
In order to understand the correlation between cross-feeding interactions and various diseases, 1661 well-characterized and deeply sequenced gut metagenome samples were analyzed. This dataset encompassed 871 individuals without reported diseases and 790 individuals with various diseases, drawn from 33 published studies across 15 countries and representing 11 disease phenotypes.
40 bacterial and 1 archaeal species were exclusively found in diseased individuals, whereas healthy individuals harbored 59 bacterial and 1 archaeal species not observed in any diseased individual. To infer metabolic exchanges between microbes, community-scale metabolic models were built for each individual based on species-level abundances, and MES was calculated.
Initially, the authors’ focus was on identifying the metabolic exchanges with the highest diversity of cross-feeding partners in healthy microbiomes. This involved analyzing the MESs of each metabolite across the entire healthy group. Notably, metabolites exhibited considerable variation in MESs among individuals. Those with the highest mean MES included nucleobases like uracil and thymine, essential nutrients such as phosphate and iron, and sugars like glucose and galactose.
To pinpoint metabolites most impacted by the loss of cross-feeding partners during disease, MESs between the healthy group and 11 disease phenotypes were compared and a significant loss of cross-feeding partners was observed for specific metabolites in all disease groups, except for schizophrenia.
Metabolites crucial for human health, such as vitamin B1 (thiamin) and precursors of short-chain fatty acids (e.g., malate, glucose, galactose), which had high MESs in healthy individuals, were significantly affected in multiple disease phenotypes. Thiamin exhibited the highest difference in MESs between healthy and diseased microbiomes in cirrhosis and ankylosing spondylitis, ranking second in Inflammatory Bowel Disease (IBD), suggesting a potential microbial mediation of thiamin deficiency in cirrhosis and IBD.
Similarly, a connection between microbially-derived ribosyl nicotinamide and rheumatoid arthritis was suggested for the first time. The results also confirmed previously reported microbially-mediated disease-metabolite associations, such as ethanol in colorectal cancer and hydrogen sulfide in IBD, emphasizing the potential of this innovative approach to identify plausible relationships.
Interestingly, 22 metabolites, including hydrogen sulfide (H2S) and D-galactose, were identified to have significantly higher MESs in type 2 diabetes (T2D)-associated microbiomes compared to healthy microbiomes, in agreement with a previous study that quantified microbial metabolic exchanges in the gut linked with glucose intolerance and T2D.
Species diversity exhibits unique associations with producers and consumers of exchanged metabolites
The diversity of microbial species in the gut community is commonly regarded as an indicator of health. Microbiomes linked to five diseases exhibited a significant reduction in alpha diversity. However, microbiomes from individuals with T2D showed a significantly higher alpha diversity compared to the healthy group. Diseases characterized by low species diversity, such as Inflammatory Bowel Disease (IBD), exhibited the most pronounced differences in MESs, aligning with the expectation that the number of microbial species engaged in metabolite exchange naturally correlates with the overall number of species in the community.
To better understand the relationship between diversity and metabolite exchange, the authors investigated whether the number of species influences the exchange of substances in a community. They compared the impact of species diversity on both the organisms producing and consuming these substances. The authors wanted to test if the increase in the number of species affects both producers and consumers equally.
The findings indicate that in most instances, either the producers or consumers are more influenced by species diversity. This means that as the variety of species increases, the number of producers or consumers may not increase at the same rate. For example, when looking at metabolites with the highest impact (highest MESs), only producers and consumers of glycerol showed no significant difference in response to species richness.
The restoration of the microbial food web as a potential therapeutic approach for Crohn’s disease
To explore how the application of MES and the modeling framework can guide the identification of potential therapeutic targets, the subtype of IBD, namely Crohn’s disease (CD), was investigated. A single case-control study, featuring the largest number of samples from both healthy and diseased individuals within the quality-controlled dataset, was selected.
Hydrogen sulfide (H2S) – a gas previously linked to CD and IBD symptoms – was found as the metabolite most affected by the loss of cross-feeding microbial partners. While previous studies have focused on H2S production by the gut microbiome, but not the consumption of this gas, the authors suggest that bacteria can consume H2S and incorporate it into sulfur-containing amino acids such as cysteine.
The microbiome of healthy individuals was found to harbor more species with the potential to produce H2S, as well as to consume H2S, compared to CD-associated microbiomes. Notably, the diversity of potential H2S consumers was more impacted in CD patients than the diversity of H2S producers, resulting in a significantly higher H2S producer-to-consumer ratio in individuals affected by CD.
Similar trends were observed when investigating the flux of H2S among microorganisms. In the disease state, the total estimated capacity of the microbiome to consume H2S was reduced by 74%, while total production remained unaffected, leading to a higher H2S production-to-consumption ratio in CD. However, the excess of H2S was not significantly different between healthy and diseased subjects. Therefore, H2S consumers are more affected than H2S producers in CD.
To identify the key species linked to H2S imbalance in CD, a comparison of each species’ contribution to total H2S production or consumption was conducted in both healthy and CD cohorts. For each species, the H2S flux was estimated, and species exhibiting the most substantial increase in H2S production in CD patients included members of the Clostridia, Bacteroidia, and Bacilli. Enterocloster clostridioformis (Clostridia) and Enterococcus_B faecium (Bacilli) were exclusive to the CD cohort.
Many species demonstrated the ability to both produce and consume H2S, with their role contingent on their community context. Phocaeicola dorei (Bacteroidia) exhibited the highest difference in predicted H2S production between healthy and CD individuals, despite being common in both cohorts.
Members of the Clostridia, including Roseburia intestinalis, Blautia_A obeum, and two Faecalibacterium species, were the H2S consumers with the most substantial reduction in H2S consumption in CD microbiomes.
Next, the authors compared the results from their metabolic modeling approach with traditional compositional microbiome analyses and identified species contributing most to the differences between healthy and CD-associated microbiomes. Some of these species overlapped with the metabolic modeling approach used in this study, including the H2S consumers Roseburia intestinalis, Escherichia coli, and Anaerostipes hadrus, as well as the H2S producer Clostridium_Q symbiosum.
Overall, the present study introduced a novel conceptual framework based on MES revealing a notable reduction in potential cross-feeding interactions within the microbiomes linked to 10 diseases and identified promising therapeutic targets in Crohn’s disease.
The analytical framework presented here, successfully detected known and new associations between the microbiome and diseases. This offers a cost-efficent and mechanistically grounded strategy to prioritize experiments and guide clinical trials.
In conclusions, the authors anticipate that metabolic models informed by metagenomic data, along with an evaluation of microbial cross-feeding interactions, will overcome a major obstacle in the advancement of microbiome therapies, by prioritizing which species or metabolites to target. By focusing on the restoration of crucial elements of gut ecology, it might be possible to introduce more impactful and long-lasting modifications in the human gut microbiome.
