Sean GibbonsMay 1, 2023

Is there a rational way to predict stable engraftmet of a non-indigenous bacterial taxon in the context of an individual's gut microbiota?

In our latest preprint, @avcarrology tackles this fundamental question. @ISBNitinBaliga @thaasophobia @isbsci

https://www.biorxiv.org/content/10.1101/2023.04.28.538771v1

🧵

Show threadSean GibbonsMay 1, 2023
Prior work has looked at metagenomic features predictive of probiotic engraftment (https://pubmed.ncbi.nlm.nih.gov/27693307/), ML approaches to predicting FMT donor strain engraftment (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9499858/), or orthogonal niches that enable engraftment in the gut (https://pubmed.ncbi.nlm.nih.gov/30110640/).
Stable Engraftment of Bifidobacterium longum AH1206 in the Human Gut Depends on Individualized Features of the Resident Microbiome - PubMed

Live bacteria (such as probiotics) have long been used to modulate gut microbiota and human physiology, but their colonization is mostly transient. Conceptual understanding of the ecological principles as they apply to exogenously introduced microbes in gut ecosystems is lacking. We find that, when …

PubMed
Show threadSean GibbonsMay 1, 2023

In this preprint, we use our MICOM platform to build microbial community-scale metabolic models (MCMMs) to predict personalized engraftment potential across >14,000 people.

We apply this approach to an important pathobiont known to colonize up to 30-40% of adults: C. difficile.

Show threadSean GibbonsMay 1, 2023
Basically, we 'inject' a non-indigenous taxon into a MCMM at a specified propagule pressure (e.g., 10% relative abundance), and then use its predicted growth rate as a proxy for engraftment/colonization probability within that individual (i.e., an engraftment risk index).
Show threadSean GibbonsMay 1, 2023

As an initial proof-of-principle for the accuracy of our MICOM/MCMM model, we leveraged time series data from a known asymptomatic C. difficile colonization event (following a bout of Salmonella food poisoning; https://www.nature.com/articles/s41564-020-0668-2).

The results were pretty good.

Diarrhoeal events can trigger long-term Clostridium difficile colonization with recurrent blooms - Nature Microbiology

Gastrointestinal disturbances, such as non-antibiotic-induced diarrhoea, promote susceptibility to colonization by Clostridium difficile, providing additional insights into antibiotic-independent triggers for this infection.

Nature
Show threadSean GibbonsMay 1, 2023
Next, we wanted to assess this engraftment risk index following an FMT treatment. We pulled down pre- and post-FMT sequencing data from https://pubmed.ncbi.nlm.nih.gov/25825673/ and we saw that risk was high pre-FMT and lower post-FMT. However, many healthy individuals were susceptible to C. diff.
Dynamic changes in short- and long-term bacterial composition following fecal microbiota transplantation for recurrent Clostridium difficile infection - PubMed

Dynamic behavior is an intrinsic property of normal fecal microbiota and should be accounted for in comparing microbial communities among normal individuals and those with disease states. This also suggests that more frequent sample analyses are needed in order to properly assess success of FMT proc …

PubMed
Show threadSean GibbonsMay 1, 2023
10^-6 was the lower limit of our solver accuracy, so any growth rate smaller than this number was effectively zero (i.e., see the grey shading in the prior plot).
Show threadSean GibbonsMay 1, 2023

Our metabolic model provides detailed mechanistic insights into what factors govern C. diff growth.

We looked at the top metabolites that C. diff consumed across models (i.e., its niche), & found that C. diff clustered into three growth niches: high-, moderate-, and no-growth.

Show threadSean GibbonsMay 1, 2023
The same exact genome-scale metabolic model of C. diff was used across these simulations, so these growth clusters show the plasticity of C. diff's metabolic niche across communities.
Show threadSean GibbonsMay 1, 2023
We see many metabolites known to promote C. diff growth from the literature, like succinate, ornithine, trehalose, and Stickland fermentation amino acids. We see other emergent metabolites important for growth as well, like fructose and cysteine/methionine.
Show threadSean GibbonsMay 1, 2023
MCMMs provide us with insight into potential competitors with C. diff, who import metabolites that it uses, like in the case of Phocaeicola and succinate (below). Even within the same taxon, these competitive interactions can vary a lot across individuals.
Show threadSean GibbonsMay 1, 2023
And MCMMs reveal taxa that cross-feed C. diff, like several of the butyrate-producing Clostridia, which appear to export succinate. Similarly, these cooperative interactions can vary a lot from person-to-person, even for the same bugs, highlighting the need for a systems approach
Show threadSean GibbonsMay 1, 2023
When we looked across 4 independent data sets, encompassing >14,000 individuals, we see that the three C. diff growth niches (import flux clusters) are highly consistent across populations.
Show threadSean GibbonsMay 1, 2023
If we embed these niches in the context of other commensal gut bacteria (grey clusters) in these models, we see that they are quite distinct from one another, relative to cross-taxon niche variation.
Show threadSean GibbonsMay 1, 2023
Previously, C. diff recurrence risk was observed to be higher in low-diversity microbiomes (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5684761/). We saw this in our MCMMs too, & we found that engraftment risk also increased at the top-end of the diversity spectrum. Risk was lowest at intermediate diversity.
Identifying predictive features of Clostridium difficile infection recurrence before, during, and after primary antibiotic treatment

Colonization by the pathogen Clostridium difficile often occurs in the background of a disrupted microbial community. Identifying specific organisms conferring resistance to invasion by C. difficile is desirable because diagnostic and therapeutic strategies ...

PubMed Central (PMC)
Show threadSean GibbonsMay 1, 2023
Finally, as a proof-of-concept, we re-constructed a 6-member probiotic cocktail (i.e., a subset of VE303, with available metabolic models @VedantaBio; https://jamanetwork.com/journals/jama/fullarticle/2803996), which was able to protect most MCMMs from C. diff invasion in silico.
Bacterial Consortium Therapy for Prevention of Recurrent C difficile Infection

This clinical trial compares the efficacy of 2 different doses of VE303, a defined bacterial consortium of 8 strains of commensal Clostridia, and placebo in preventing recurrence of Clostridioides difficile infection among adults at high risk.

Show threadSean GibbonsMay 1, 2023
However, there were clear responders and non-responders. The in silico efficacy of this probiotic cocktail was associated with the average growth rate of the probiotic taxa and/or with the average proximity of the probiotic taxa to the realized metabolic niche of C. diff.
Show threadSean GibbonsMay 1, 2023
And again, these models are very transparent, so we were able to look at how probiotics competed with C. diff for specific metabolites. We were also able to calculate the engraftment probability (growth rates) of the probiotoic strains themselves within each model.
Show threadSean GibbonsMay 1, 2023
We believe this work represents an advance in our ability to rationally predict personalized pathobiont colonization risk and assess efficacy of therapeutic probiotic interventions. Unlike more black-box machine learning approaches, MCMMs do not rely upon training data.
Show threadSean GibbonsMay 1, 2023
MCMMs provide transparent, mechanistically-grounded predictions that should be fairly robust across human populations. As our genome-scale metabolic model databases improve, these MCMMs will only grow more powerful.
Show threadSean GibbonsMay 1, 2023
This preprint is our first proof-of-concept for how MCMMs can be leveraged to predict engraftment potential. There is no reason why this same approach couldn't be applied to any other pathobiont or probiotic taxon in the human gut, or to more complex interventions, like FMTs.
Show threadSean Gibbons
Big thanks to the researchers who made their data available for this analysis: https://pubmed.ncbi.nlm.nih.gov/25825673/; https://journals.asm.org/doi/10.1128/mSystems.00031-18; https://pubmed.ncbi.nlm.nih.gov/24336217/; and https://pubmed.ncbi.nlm.nih.gov/36109637/
Dynamic changes in short- and long-term bacterial composition following fecal microbiota transplantation for recurrent Clostridium difficile infection - PubMed

Dynamic behavior is an intrinsic property of normal fecal microbiota and should be accounted for in comparing microbial communities among normal individuals and those with disease states. This also suggests that more frequent sample analyses are needed in order to properly assess success of FMT proc …

PubMed
May 1, 2023 at 6:37pmWeb
Show threadSean GibbonsMay 1, 2023

Amazing work from @avcarrology, with @thaasophobia and @ISBNitinBaliga at @isbsci.

💩💪🦠🤩

Show threadSean GibbonsMay 1, 2023
And finally, thanks to the funders who made this work possible: @wrfseattle and @NIDDKgov
Show threadSean GibbonsMay 1, 2023
If you want to hear me talk about this work, in addition to other recent work from our group, check out my @MicroSocEquity talk from earlier in the year: https://video.maine.edu/media/%22Personalized+nutrition+and+the+human+gut+microbiome%22%2C+by+Dr.+Sean+Gibbons/1_qpcnw6md
"Personalized nutrition and the human gut microbiome", by Dr. Sean Gibbons

Microbes and Social Equity speaker series 2023This series explores the way that microbes connect public policy, social disparities, and human health, as well as the ongoing research, education, policy, and innovation in this field.I would like to thank the UMaine Institute of Medicine for their support for this series, and acknowledge the work of our MSE members helping to organize this: Emily Wissel, Katherine Daiy, Kieran O’Doherty, Hannah Holland-Moritz, Mallory Choudoir, and Mustafa Saifuddin. I would also like to recognize that the University of Maine is located on Marsh Island in the homeland of the Penobscot Nation.This presentation is "Personalized nutrition and the human gut microbiome", by Dr. Sean Gibbons.Dr. Sean Gibbons is an Associate Professor at the Institute for Systems Biology, a non-profit research consortium. His lab develops computational and experimental tools for exploring and manipulating host-microbe systems. Added by Sue: The work from Sean's group and collaborators has been reshaping the way that host microbial researchers approach their work, by revealing trends through large metanalyses and novel perspectives on using data. Their most recent work has evaluated host-microbial interactions, metabolites, and health. For the last three years, Sean's lab has hosted the ISB Virtual Microbiome Series, which is freely available and attracts several thousand participants. The series includes a two day workshop that teaches data analysis skills, and a day-long symposium featuring discussions of current discoveries and conceptualizes the future of microbiome research. Finally, Sean and his research group have been making science a more welcoming and inclusive place.

University of Maine System