As a computational biologist, I was always excited by the possibility of analyzing multiple ‘omics datasets, both methodologically and for human health applications. From a purely computational standpoint, the ability to combine many data types to do better science was interesting in itself. But from the perspective of studying the microbiome, the capability of linking microbial metagenomic profiles - which describe the bugs in a community and their genetic potential - with microbial RNA (i.e. how microbial genes are transcribed compared to their abundances in DNA) was even more fascinating.
The opportunity presented itself when I joined the Huttenhower lab as a postdoc, almost two years ago now. The group focuses on functional analysis of microbial communities, both by developing new computational methods and by applying them to human population studies and microbial systems biology. Although it was not something I originally expected to work on, the group previously developed one of the first home stool sample collection kits that allowed easy, mail-in generation of multiple molecular data types from the gut microbiome. I’m glad I wasn’t personally involved in sample handling - but that earlier pilot study gave us a way to generate multiple, concurrent molecular profiles of gut microbes from human cohorts.
In the current study, we extended this to one of the largest population-scale investigations of human fecal metatranscriptome to date. Understanding person-to-person differences in the gut microbiome has been surprisingly challenging: everyone’s microbiomes have many more differences than do, say, their genomes. There’s almost no set of “core” microbes shared by everyone, although there do seem to be common pathways encoded metagenomically. Thus, in this study, we set out to determine whether there was also a “core” metatranscriptome - that is, pathways that are commonly transcribed by microbes in the gut.
A simple, initial approach to this was to identify the pathways transcribed in most samples and also prevalent in microbial DNA. We found that only 44% of DNA core pathways were actively transcribed in our cohort. Unlike the core metagenome’s complement of host-adapted microbial community features, the core metatranscriptome was enriched for housekeeping functions including nucleotide, carbohydrate, and amino acid metabolism, and they were often transcribed by many organisms per community.
Conversely, our data also revealed unique context- and species-specific aspects of the gut metatranscriptome. Many functions were encoded and transcribed by only a few species, in contrast to the metatranscriptomic core above. In addition, pathways with similar metagenomic contributions were not necessarily similarly transcribed, and transcribing organisms were not generally the most abundant (Fig. 1). This led us to the hypothesis that a functional ecological model drives microbial decision-making on what to transcribe and when. That is, within a particular community, one or a few bugs will “take over” transcription of a pathway in a particular niche, even if many organisms could be capable of carrying out the process. As a result, at any given time and environment, transcription is typically dominated by a few members of the community, even when those functions are universally encrypted in the DNA.
Overall, these results demonstrated that incorporating whole-community transcriptional profiling will provide an added benefit in many microbial community studies for human health. Methodologically, this further calls for a harmonious coexistence of DNA and RNA measurements - ‘happily ever after’ - in future large-scale analyses of the microbiome. These will require tying transcriptional activity to individual strains of microbes, and figuring out which stimuli (such as diet) trigger transcriptional regulatory responses.
I look forward to seeing the clinical and public health implications as well, particularly with respect to translation into microbiome based therapeutics and diagnostics, when this work is extended to even larger numbers of study participants and other disease areas in the future. Stay tuned for future research updates!
The corresponding published paper in Nature Microbiology is here: http://go.nature.com/2Di7sSM
Poster image created by Amoeba Sisters: http://www.amoebasisters.com/. Many thanks to colleagues Galeb Abu-Ali, Jason Lloyd-Price, and Kevin Bonham for offering constructive inputs on the first draft of this blog post.