A Sneak Peak into What is Possible: Postbiotic Profiling of Next Generation Microbes - Fitbiomics

A Sneak Peak into What is Possible: Postbiotic Profiling of Next Generation Microbes

I am a recent graduate from the University of Vermont majoring in Biomedical Engineering with minors in Math and Computer Science. I am also a varsity division I soccer play with a strong passion for athletics and the human body. I recently started working for FitBiomics full time and below I will summarize the fascinating projects I worked on while I was an intern.

Probiotic bacteria are known to have health benefits, but most probiotics are derived from food. FitBiomics creates probiotics derived from athletes’ microbiomes. Last year, FitBiomics launched Nella, a probiotic with three strains of Lactobacillus that were isolated from some of the most elite athletes in the world. We know that Nella has some exciting benefits because survey data was collected during a beta test of Nella and it was found that 45.1% of participants experienced improved sleep quality, 38.5% experienced shorter recovery time, 38.5% experienced shoer recovery time, and 34.6% experienced better and more regular bowel movements.

We were not satisfied with just knowing that there are benefits, though. We wanted to understand why these effects might be happening, so that we can keep improving our probiotics and making them even more effective. One of the techniques we use to do this is metabolomics. Metabolomics is a way to identify and measure all the molecules that are in the mixture. When we grow our bacterial strains in a lab, metabolomics tells us what types of compounds the bacteria are producing, which helps us figure out how these probiotics are affecting our customers’ health and performance.

In these experiments, we grew our probiotic bacteria alone and in cultures together. It is important to grow them together because they are delivered together in the capsule. When different bacterial species are grown together, they tend to interact. These new and synergistic interactions can produce new molecules, and we needed a method to try and understand what types of interactions might be happening when we grew different bacterial strains together.

The four FitBiomics strains we analyzed as part of these experiments were Lactobacillus acidophilus, Lactobacillus plantarum, Lactobacillus rhamnosus and Veillonella atypica. Veillonella atypica is an exciting strain that may be important for improving endurance. We are working on getting it approved for use in humans soon! The co-cultures were different combinations of these four strains. We also looked at cultures of commercial Lactobacillus strains of these species, so we can start to understand what the differences are between FitBiomics’ strains and commercials probiotics. The experiments are important for understanding FitBiomics’ probiotic strain compositions and to determine if there are any synergistic interactions. Furthermore, our metabolomics analysis can be used as a basis for a future machine learning model that we hope will be able to predict the beneficial effects of different probiotics.

First, we did hierarchical clustering (seen in Figure 1) because we wanted to understand whether the metabolite profile could predict the genetic relationship (phylogeny) between FitBiomics’ Lactobacillus and commercial Lactobacillus strains. Each FitBiomics’ Lactobacillus strain correlated most closely with the commercial counterpart of the same species. This suggests that even though FitBiomics’ strains are extracted from elite athletes instead of the common source of soil, baby poop, etc., the molecules they produce (at least in these growth conditions) are very similar to commercial probiotic strains that are already known to positively impact users. 

Figure 1. The hierarchical clustering performed on each of the 7 single cultures revealing the similarities across FitBiomics’ and commercial strains. FitBiomics Strains: 0 (Lactobacillus acidophilus), 1 (Lactobacillus plantarum), 2 (Lactobacillus rhamnosus), and 3 (Veillonella atypica). Commercial strains: 4 (Lactobacillus plantarum), 5 (Lactobacillus rhamnosus), and 2 (Lactobacillus acidophilus). The x-axis represents the 7 different single culture clusters and y-axis represents the distance between clusters.

Next, we wanted to further our understanding of the relationships between each strain and their interactions within co-cultures, so we performed a technique called Principal Component Analysis (PCA), which lets us visualize the relationships of a very large complex dataset on a two-dimensional plot (seen in Figure 2). It was hypothesized from the hierarchical clustering that co-cultures involving the Veillonella atypica strain would be visually separate from other cultures because as a different genus, it would be producing a very different set of metabolites from the Lactobacillus strains.

Based on the plot that was formed, our hypothesis was correct in that the five co-cultures that included Veillonella atypica were all distinctly separate in grouping from the rest of the cultures. Potentially more interesting were the results that showed that some synergistic interactions might be occurring. We found that the Lactobacillus co-cultures clustered near but slightly away from Lactobacillus single cultures suggesting that the metabolites the co-cultures produce are similar but not identical to the single cultures. If the strains were not interacting at all, the PCA plot would place them in with the single cultures. This is very exciting, so now we want to determine what new molecules are being produced by the single cultures!

Figure 2. The PCA plot indicating Veillonella atypica had the greatest differences among all the analyzed strains. The data is made up of single and co-culture strains. The red data points are single culture Lactobacillus strains, the orange are co-cultures consisting of only Lactobacillus cultures, and the grey is any single or co-culture that has Veillonella aytpica. 

This analysis is very important for FitBiomics because it helps FitBiomics understand the beneficial role of the microbiome in the world of health and performance. Also, this work gives us a foundation as we begin to build a machine learning model that will help us understand microbiome-host interactions and identify novel mechanisms of actions by which the microbiome impacts our health and fitness.


WRITTEN BY:  Alex West


Alex is a recent college undergraduate from the University of Vermont. She graduated with a degree in Biomedical Engineering specializing in Systems and Network Biology. She was a collegiate varsity soccer player who loves to combine her love for data and human health to have a positive impact on the biotech and health industries. In her free time she can be found reading or building legos.
Back to blog