Are processed foods unintentionally engineering our microbiomes?

Cover art by Krista Armbruster, Ph.D.

Like Comment
Read the paper

Xanthan gum (XG) is a common food additive introduced into the human diet around 50 years ago. In most foods, XG is used at 0.1-0.5% (w/w) but in gluten-free baked goods XG can be consumed in up to gram quantities per serving! The rise of celiac disease and popularity of gluten-free diets will likely increase consumption of xanthan gum. Although there was some evidence that XG could be broken down by human gut microbiomes, it was unknown which organisms were involved and how the breakdown process occurred. We wanted to find answers to these questions and ultimately determine how consumption of XG is influencing human microbiomes and possibly human health.

We started by screening several dozen fecal samples from healthy human adults by exposing them to XG in liquid culture and only found one culture that could degrade XG. This was very lucky since without this initial culture showing growth on XG, we may have shelved this project to work on other research leads. By removing selective antibiotics from our culture media and carrying out a second screen, we eventually found numerous samples that degraded XG. Interestingly, while some cultures had a Bacteroides species that grew to high abundance, the only common microbe across all the degrading cultures was a Ruminococcaceae from currently uncultured genus 13 (R. UCG13). After trying and failing to isolate individual bacteria that could grow on XG, we decided to take a different approach to understanding how these cultures were breaking down the polysaccharide. By performing metagenomics and transcriptomics on one of these cultures growing on XG, we identified an expressed locus with all the expected enzymes for degrading XG. In several pathways used by soil microbes, XG is degraded with a lyase that removes the mannose from the side chain of the polysaccharide, then subsequently depolymerized into smaller oligosaccharides. The locus we identified from R. UCG13 had a lyase with some homology to established XG lyases, so we hypothesized that this lyase would work analogously to those in known XG pathways.

After trying numerous conditions to purify and test the recombinant protein, I simply could not get the lyase to break down XG. In the face of what felt like failure at the time, I began to consider other ways that XG could be broken down and the possibility that we had identified the wrong locus or even the wrong organism. Given these doubts, I decided to take a step back and try a different approach, using the fact that the cultures I was working with were making all the enzymes necessary to degrade and consume XG. First, I took a culture growing on XG and found that supernatant from this culture could degrade XG overnight. With this result in hand, I worked on using activity-guided fractionation and proteomics to narrow down which enzymes were depolymerizing XG. Eventually, I identified a glycoside hydrolase from family 5 (GH5) in the locus we had identified in the R. UCG13 metagenome assembled genome (MAG). At the same time, I developed a method to look at the oligosaccharides produced by these reactions; instead of lyase-treated oligosaccharides, the oligosaccharides were generated from direct hydrolysis of XG. Purifying a recombinant form of the R. UCG13 GH5 confirmed its ability to hydrolyze XG directly, without pre-treatment of the substrate with a lyase. The lyase that I had so much trouble with ‘failing’ at the beginning of the project, ended up working on smaller XG oligosaccharides but didn’t have activity on the full polymeric XG. This is a great example of why it’s important to consider multiple hypotheses and be open to what your results are telling you.

Finding the GH5 that initiates XG depolymerization really opened the door for this project to flourish. Although there were still ups and downs, we were able to characterize the rest of the enzymes in the R. UCG13 MAG that break down XG. We also discovered another organism, B. intestinalis, that could consume fragments of XG released by R. UCG13 but couldn’t grow on polymeric XG by itself. We were able to characterize most of the enzymes B. intestinalis uses to break down XG oligosaccharides, but once again couldn’t detect activity of the lyase from our identified locus. Since the other enzymes in the locus worked and were all upregulated when the B. intestinalis was growing on the oligosaccharides, we assumed that the recombinant lyase was not working because of a design issue or a missing cofactor. After we submitted our paper, reviewers astutely pointed out that our expression data alone was not strong proof that the suspected lyase was active on XG oligosaccharides. Once again, we went back to the actual organisms and were able to show that cultures of B. intestinalis growing on XG oligosaccharides produce one or more enzymes that catalyze the lyase reaction on XG oligosaccharides. We’re still working on identifying which enzyme(s) from B. intestinalis perform the lyase activity on XG oligosaccharides, but my preliminary data suggest that it may truly be a different enzyme found outside of the locus we identified. The reviewers for our initial submission helped us overcome our own biased expectations of how our system was working, which will ultimately result in a more accurate model of how this microbe consumes XG oligosaccharides. Although most of us have an expected hypothesis for the experiments we set up, it’s important to be open to alternative explanations and critiques that make our science stronger.

By screening healthy human adults, we had already found that R. UCG13 was present in around half of individuals, but we wanted to look at XG consuming microbes at a broader scale. My talented co-authors were able to search for R. UCG13 and B. intestinalis in hundreds of publicly available datasets from all different parts of the world. Remarkably, R. UCG13 was common in industrialized societies but completely absent in hunter-gatherer or pre-industrialized societies. We even found a sample from a 1-year-old infant with R. UCG13, suggesting that in some cases this microbe could influence assembly of the microbiome during childhood. The B. intestinalis was less prevalent, but similarly absent in pre-industrialized societies. We also found that feeding XG to humanized mice led to an increase in R. UCG13 and B. intestinalis. Together, these results suggest that consuming XG could be causing an increase in R. UCG13 and B. intestinalis in industrialized microbiomes.

Eating more fiber generally has positive effects on human health, but outcomes vary depending on a person’s initial microbiome. At the same time, microbiomes in industrialized societies are distinct from those in pre-industrialized societies and may be involved in the rise of chronic diseases. XG is both a fiber and a novel food ingredient that could be influencing the ecological trajectory of microbiomes in populations that consume it. While XG is generally well-tolerated and safe for consumption, our work provides a mechanistic foundation for understanding how the microbiome responds to this widespread food additive and how its introduction to human diets may be unintentionally engineering our microbiomes.

Read the full article here: https://doi.org/10.1038/s41564-022-01093-0

Matthew P. Ostrowski

Postdoctoral Fellow, University of Michigan