Additional contributors to this article: Jon Sanders, PHD, Anupriya Tripathi, Qiyun Zhu, Phd.
The American Gut Project1 is the country’s largest open source crowd-funded citizen science project in existence today. As of early 2017, the project has raised over $2 million and processed over 14,000 samples from over 10,000 individuals (and counting). Over the last two years, the American Gut Project has partnered with AGA and its Center for Gut Microbiome Research and Education on the Microbiome Active Learning Sessions at Digestive Disease Week®. In 2016, we performed 16S ribosomal RNA (rRNA) sequencing on stool samples collected from 54 AGA members, and learned that the gut bacteria of gastroenterologists are not particularly different from those of the average American Gut Project participant.2,3
The 16S rRNA gene is a ribosomal gene thought to exist in all bacteria and archaea. Many microbiome studies today target the V4 region of the 16S rRNA gene, as the primers that amplify this portion are least susceptible to bias and able to yield the greatest amount of information taxonomically from most sample types. The 16S method is also cost-effective since hundreds of samples can be sequenced at once. Most American Gut samples have been processed in this way, as the low cost makes personal microbiome sequencing attainable for many.
A limitation of 16S rRNA marker gene analysis is that it surveys only a small portion of each genome. In contrast, the shotgun metagenomics approach sequences all DNA present, yielding information about the entire genomes of all organisms in the sample, including the non-bacterial members of the microbiome such as viruses and fungi. Because most regions of the genome have higher variability than rRNA genes, shotgun data can also give greater resolution and distinguish between closely-related organisms. With sufficient sequencing depth, shotgun data can also yield direct information about the presence and abundance of particular functional gene pathways, or be assembled into draft genomes, yielding insights into the physiological capabilities of abundant organisms in the community. Due to the depth of sequencing required for shotgun analysis, it is more expensive than 16S rRNA sequencing; however, the Knight lab has been working on protocols to lower the cost of shotgun metagenomics sequencing and analysis. In 2017, the samples submitted by 53 AGA members were analyzed by both 16S and shotgun sequencing methods. Ten of these members were returning participants who also submitted a sample in 2016.
Figure 1. An unweighted UniFrac-based principal coordinates plot shows the AGA cohort samples (large blue dots; n = 53) in the context of all American Gut Project (AGP) stool samples (small red dots; n = 11,887).
As in 2016, we found that the 2017 cohort of AGA member participants were scattered across the greater American Gut Project “map” (Figure 1). 16S rRNA sequencing showed that the major phyla present in AGA member samples were Bacteroidetes and Firmicutes. Interestingly, three AGA participants had notable abundances (near 50 percent to over 70 percent) of Proteobacteria; in comparison, the average amount of Proteobacteria observed across the entire American Gut population is less than 10 percent. A number of diseases, including IBD, have been associated with an increase in Gammaproteobacteria. However, in the case of the three AGA participants, the main taxa were Alpha- and Betaproteobacteria. When we looked at the same samples by shotgun metagenomics sequencing, only two consistently yielded high levels of Proteobacteria.
Figure 2. Phylum level stacked bar charts reveal the relative abundances of Archaea, Bacteria, Fungi, and Viruses present in the AGA cohort. Taxonomy information and relative abundances were obtained using SHOGUN. The vast majority of detected taxa were bacterial, though a few individuals had archaea, fungi, or viruses present at a relative abundance of 1% or more.
Across the full 2017 AGA cohort, there is some consistency between the 16S and shotgun metagenomics approaches although some obvious differences are noted. The most obvious difference between the two approaches is the increased abundance of Bacteroidetes (and corresponding lower abundance of Firmicutes) overall in the shotgun metagenomics cohort. Additionally, while most of the taxa detected were bacteria, there were also archaea, fungi (eukaryotes) and viruses detected (Figure 2). A total of five archaeal genera, one parasitic (eukaryotic) genus, seven fungal (eukaryotic) genera and 25 viral genera were detected. In most cases, non-bacterial taxa were at low abundance, or not detected; in only six of the 53 individuals, non-bacterial taxa were detected at a relative abundance of 0.5 percent or more. These included the Archaea Methanobrevibacter, Methanosphaera, Methanomassiliicoccus and Methanobrevibacter, the eukaryotic taxon Candida, and the viral (phage) genera Nonagvirus and Pseudomonas_phage_O4 and Pseudomonas_phage_PA11.
Finally, assessing the top 20 pathways present across the dataset, we see a stable pattern in which most pathways are at similar relative abundances in all samples. However, in the samples in which Proteobacteria were present at relative abundances of over 70 percent, three pathways are noticeably higher: chorismate biosynthesis from 3-dehydroquinate, chorismate biosynthesis I and peptidoglycan biosynthesis I (Figure 3). The potential significance of the increased abundance of these three pathways is unclear, highlighting a new frontier of research as whole genome sequencing approaches become more common.
This year’s AGA participant report4 includes much more information than what is discussed here, including a case study of deep shotgun metagenomic sequencing performed on longitudinal stool samples collected from a single patient with IBD over three years. Importantly, while we now have the technology to understand what microbiota are in the gut, we do not yet have enough evidence to explain why they are there and how they impact health and disease. There is much more to learn about the relationship between humans and their microbiota. Thank you to the gastroenterologists who participated in the 2017 sequencing activity, and we look forward to seeing all of you at the next Microbiome Active Learning Session.
Dr. Hyde has no conflicts to disclose. Dr. Knight has no conflicts to disclose.
1. Preliminary Characterization of the American Gut Population. American Gut website.. February 2016. http://americangut.org.
2. Knight, R. The Gastroenterologist’s Microbiome: Here’s What We Found in the Guts of AGA Members. AGA Perspectives. 2016; vol.12, no.3:26-27. http://agaperspectives.gastro.org/gastroenterologists-microbiome-heres-found-guts-aga-members/#.WWZjkGgrJPY
3. American Gastroenterological Association and American Gut. Exploring IBD in the Context of the American Gut Project Using State of the Art Tools. Digestive Disease Week®: Active Learning Session on the Gut Microbiome. May 2016. http://www.gastro.org/about/initiatives/AGA-American_Gut_Handout_DDW_2016.pdf
4. American Gastroenterological Association and American Gut. Comparing 16S rRNA Marker Gene and Shotgun Metagenomics Datasets in the American Gut Project Using State of the Art Tools. Digestive Disease Week®: Microbiome Active Learning Session 200.May 2017. http://www.gastro.org/about/initiatives/AGA_2017_Microbiome_Active_Learning_Session_Handout_Final.pdf