Getting the timing right for artificial microbiome selection

“Understanding microbial community dynamics to improve optimal microbiome selection” – that’s the title of my first paper (which I’m very excited to say has just been published here!), so I thought I’d post a post an explanation on what we show in it. Our aim in this paper was to use artificial selection– similar to the method used for getting different breeds of dogs or crop plants with desired traits – on whole microbial communities (microbiomes) to see if they could be selected to become better at degrading something. This was with the overall aim of using it on plastics, but first of all we used chitin, the most abundant natural polymer in the oceans, and the material that makes up the shells of organisms such as crabs and shrimps. This had the advantage that chitin degradation is already well studied, so we know that lots of microbes can degrade it, and we know the enzymesand pathways that they use for this. There are also already tests for these chitin-degrading enzymes (chitinases) that are very quick and easy to do.

Picture taken at Plymouth Aquarium some time ago – the shell is made of chitin.

To carry out artificial selection of microbial communities, I incubate 30 communities with chitin as their only source of carbon (see figure below), and then at the end of the incubation period I test their chitinase activity, select the three with the highest activity, and use these to start my next generation of 30 communities. The aim is that the activity will get better and better over time. The only problem was that, when I first did this, the chitinase activity was actually getting worse over time. I was originally incubating my communities for nine days between generations, or selection events, because this was the best length of time to incubate in a preliminary experiment that I did. When I tested this again, I found out that the optimal incubation time had been reduced to four days, so I carried out a second, parallel experiment where I incubated the communities for four days, instead. This worked for a few generations – the chitinase activity was getting higher) – but then it dropped off again. I tested the optimal incubation time again, and it had been reduced to two days now!

Method for artificial selection of chitin-degrading communities. Picture taken from my Biological Sciences Review magazine article.

To find out why the chitinase activity kept dropping off like this, and the optimal incubation time kept being reduced, we decided to sequence the microbial communities (you can read my previous blog on microbial community sequencing here if you don’t know what this is) both across the whole experiment, and within an incubation period (so across four days). We wanted to find out if there were community members that were particularly associated with chitin degradation, and if growth rates and dynamics of these were driving the chitin-degrading community. We sequenced the 16S and 18S rRNA genes, so these are fairly universal markers for the prokaryotes (bacteria and archaea) and the eukaryotes (includes everything from single-celled protists to multicellular organisms like people), respectively. What this gave us was the abundance of thousands of different microbes, for us to try to look for patterns in. We found that the abundance of two bacteria in particular was correlated with chitinase activity; they peaked in abundance on day two (of four), when chitinase activity also peaked, and then they were overtaken by “cheaters” and grazers (see figure below, and you can see the graph showing the abundance in Figure 3 of my paper).

Microbial community succession.

This idea fits really well with what we already know about microbial communities; they go through distinct stages of succession. The first organisms that we see (on the first day, in this case) are just good at colonising, and sticking to new particles, but they can’t necessarily degrade it. Next we see a selection phase, for organisms that can actually degrade the particle (day two for us, when chitinase activity was highest), and on this day we saw two bacteria that can degrade chitin reach almost 50% abundance of our whole prokaryotic community (with thousands of other species!). Then, on days three and four, we see the succession phase. Our prokaryotic community was overtaken by cheaters – bacteria that can’t degrade chitin, but use the sub-products from others that do – and grazers. About 95% of our eukaryotic community on these days was made up of one species, a grazer. This species is described as a bactivorous marine flagellate, so basically it can swim, and it swims around eating bacteria. What was also interesting over the entire experiment was that we showed that almost all of the bacteria that were able to degrade chitin in our community came from one taxonomic class (for people, our taxonomic class is Mammalia, if this puts it into perspective), the Gammaproteobacteria. At the beginning, when we first collected the community from the sea, the Gammaproteobacteria made up about 5% of the community – within the normal ranges in the ocean – but by the end, and on the days with highest chitinase activity, they made up about 75% of the community.

Cartoon and microscope pictures of the grazer, Cafeteria sp.

All of this information about the community taken together really explains why we aren’t seeing chitin degradation if the time isn’t right – if we don’t select our community at the right time, then we don’t see an increase in chitinase activity because the community isn’t actually good at degrading chitin anymore. To try to prove this, I did a second experiment where I measured the chitinase activity daily, and selected my next community on the day when chitinase activity was highest. I found that doing this for 50 days gave me almost 100x higher chitinase activity than I’d seen in the first experiment (carried out over almost 200 days)! We’re hoping that, going forward, this technique might be able to be applied to the degradation of other polymers, like plastics, but it does need a good and quick enzymatic activity test for a relevant enzyme, which I don’t think has been developed yet. Hopefully someone will find this at some point, and then we’ll be able to apply it to make microbial communities better at degrading plastics!