Thread by @conTAMInatedsci

Our paper is out in @cellhostmicrobe! Led by @Shijie_Zhao, myself, and @ejalm. If you liked it on @biorxivpreprint, get excited for new analyses.

Your gut microbiome changes within you, even during health. What role do mutations play?

sciencedirect.com/science/articl…

Its not clear what to expect:

Point mutations with strong adaptive effects have been seen in presence of antibiotics, in opportunistic infections (e.g. transition from soil to cystic fibrosis lung), and in lab experiments (e.g. lab strain to mice).

But what should we expect in healthy humans? Gut commensals have been living in mammalian guts for many millions of years. Perhaps, they have had time to evolve regulators and invertible promoters, to try out all the point mutations, and this makes adaptation by mutation rare.

Empiricial data is needed. Metagenomics alone doesn't work well, because we need to identify low-frequency mutations with high confidence. So we used a culture based approach.

We studied Bacteroides fragilis in 12 healthy stool donors (so healthy that their poop is medicine!)

First thing is first-- characterize the B. fragilis intraperson diversity at the 'strain level'. We use the word lineage instead of strain to get around semantics arguments (read paper for definition). We find that all 12 donors are colonized primarily by a single strain.

Within this single strain, there are lots of point-mutations (SNPs) and putative horizontal transfers (MEDs, mobile element differences). Here is an example phylogeny from one person, and some summary statistics.

There is nice consistency with a molecular clock, suggesting point mutations are coming from errors during replication. This clock-like behavior also allows us to estimate the age of the diversity in each person, which ranges between 1-12 years.

We present what may be the most direct observation to date of horizontal gene transfer happening within a microbiome. While it isn't a main focus of the paper, we think its pretty cool. We see a prophage hop into a B fragilis genome, and present new methods for HGT detection.

Ok but lets get back to the point: adaptation! The best methods for statistically distinguishing adaptation from neutrality are for SNPs. We find bright signals by means of parallel evolution (localization of mutations) and dN/dS (enrichment for amino-acid changes).

16 genes show a signal for within-person adaptation. They are concentrated in genes that control what the cell envelope of B fragilis looks like, and SusC/SusD genes, a family of importers of complex-polysaccharides (e.g. fibers)

Its not clear why these genes are under pressure to change. It could be dietary variation, but there are other possibilties. Its interesting that the same pathways (Sus and envelope) are driven by invertible promoters, suggesting both historical and ongoing pressure to vary them.

We next used available timeseries metagenomic data from 2 of the same subjects to further investigate these mutations. While the metagenomic sequencing can't be used to identify SNPs with high confidence, it can be used to track abundance of mutations we know are there.

This combination of data types shows us a beautiful picture of evolution-in-action. We see a pair of mutations (blue) rising from undetectable to 70% of the population in less than a year, implying a selective advantage of ~2% for these mutations combined.

We also see very interesting dynamics that we can't fully explain. The blue sublineage is clearly more fit (in this person) than its purple ancestor, but it doesn't outcompete the tan sublineage, from which it diverged ~10 years ago. See paper for in vitro competitions.

Ok, so mutations in core genes are providing strong benefits (~2%). If these mutations are so good, how come the bacteria are just getting them now? There may be many reasons.

One possibility is person/microbiome-specific selection, driven by the immune system, phage, or diet. The Sus genes appear to be under some sort of person-specific selection (X here means that gene is in the accessory genome and is not found in that lineage, dots are mutations).

Some selective pressure might be regional. Using available metagenomes, we investigated a SNP that emerged in 3/12 people (same amino acid!). We find that it emerges over and over again in Western, but not Chinese, guts. These gene is not homologous to anything well-annotated.

So: mutational adaptation is happening in your gut, even in health. We don't yet know if it impacts ecological dynamics or health.

It might be that each persons microbiome presents has a unique collection of selective forces, imposed by diet, immune system, other microbes, and more. Does adaptation to these differences environments contribute to strain stability? How will this affect microbiome engieering?

Lots of exciting questions left to ask!

Other authors involved in this work: @mathildpoyet @mgroussi @gibbological @k6logc Ramnik Xavier, @MITMicrobiome @MIT_CEE @MIT_IMES @MITdeptofBE

(This was my first attempt at paper tweet-storm. DM me with feedback on how to do it better next time.)

If you're interested in this topic, check out these recent papers from @nanditagarud @benjaminhgood @volatilebug @nathancrook @lucia_macareito
cell.com/cell-host-micr…
(lots of cool mice experiments)
journals.plos.org/plosbiology/ar…
(comparision between timepoints using metagenomics)

One more note: @Shijie_Zhao is responsible for the beautiful figures and many of the great insights in this paper. Keep an eye on him!!