This week’s BEACON Researchers at Work post is by MSU postdoc Bjørn Østman.
We would like to know how soil bacteria evolve. They are important for humans and other living things, as they are involved in chemical processes that are both beneficial and harmful to us. They emit and absorb greenhouse gases: carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O), and they are part of the nitrogen cycle, which is important for agriculture. Which you care about even if you don’t eat your veggies.
There is a reason why soil bacteria are not model organisms in biology. They are hard to study because they are difficult to grow in the laboratory. Also, there are many different species of soil bacteria, so even if we did mange to grow a couple in the lab, we would not likely learn much about the overall function of the bacterial communities in soil. So what to do?
You take a handful of soil and you sequence that – after removing earthworms. Simple. It’s called metagenomics.
That gives you a bunch of sequences that you at first don’t know what to do with, but after thinking about it for a while, you realize that you can find specific genes among these metagenomes. But you also realize that they are from many different species of bacteria – bacteria that you don’t know what are. Some of them can be inferred to be from known species, but there are likely millions of different species of bacteria in soil, so no doubt many of them are utterly foreign to us. We can, in other words, not know if one metagenomic sequence is from the same organism or even species as another.
Kellogg Biological Station has been running a Long-Term Ecological Research program for over twenty years where different agricultural treatments have been managed for research purposes. From KBS we have obtained metagenomic sequences from agricultural soils (AG) used for growing wheat, soy, and corn that are fertilized with ammonium nitrate, as well as from unfertilized deciduous forest soils (DF), where trees have not been cut down in recorded history.
Our question is if and how the soil bacteria evolve differently in response to transforming DF to AG. What happens to the bacterial communities when we cut down the forest, grow one crop a year, add fertilizer, and reduce the amount of oxygen by increasing the level of moisture?
In the soils, nitrate (fertilizer) is chemically reduced in the denitrification pathway:
Notice that one byproduct is nitrous oxide, the third-most potent greenhouse gas. Each of these reactions are catalyzed by a different gene. We look at nitrite reductase (nirK), which reduces nitrite (NO2–) to nitric oxide (NO). The first indicator that this is an important gene is that an estimated 1/9 of bacteria in DF have this gene, while 1/3 of bacteria in AG have it. This increase in nirK abundance suggests that bacteria with nirK have higher fitness in AG than in DF. That nirK should affect fitness makes sense because under anaerobic conditions (no oxygen), some bacteria can use nitrate to make energy. They basically breathe nitrate. It’s not as efficient as using oxygen, but clearly much better than doing nothing at all in times of limited O2.
In order to estimate the selection pressure that the change of environment exerts on the bacteria, we turn to the formalism of dN/dS. This is the ratio between the rates of non-synonymous and synonymous substitutions. dN is the rate at which nucleotide substitutions change an amino acid in the protein-coding sequence of a gene (e.g., CTG → CCG, which code for leucine and proline). Similarly, dS is the rate of nucleotide substitutions that changes a codon, but not the amino acid (e.g., CTG → CTA, which both code for leucine).
Assuming that synonymous nucleotide substitutions don’t change the efficacy of the protein (a fair assumption, but not always a certainty), we can infer that if dN=dS, then it doesn’t matter what the sequence of amino acids is, and that there is no selection acting on the gene (or amino acid site). If the environment dictates that the protein must consist of a certain sequence of amino acids, then selection will favor those that are of that particular sequence, and purge those that deviate from it. This is purifying selection, and results in dN being smaller than dS. On the other hand, if the protein is not optimal, some amino acid changes will be favorable to the organism, and we might observe that dN is larger than dS. Thus, measuring dN/dS is informative about the selection pressure exerted on the gene in question.
Given a phylogenetic tree based on the nirK sequences, we can compute dN/dS for each amino acid site in the gene. Using some 30 unique sequences from both AG and DF, we can compare dN/dS between the two.
Two phylogenetic trees from AG (red nodes) and DF (green nodes). The sequences at every internal node is estimated based on a substitution model and on the known sequences at the tips. dN/dS is then computed between all pairs of sequences separated by a branch. The two trees have 30 sequences that are chosen at random among the full set of sequences from KBS.
The result is that for both the bacteria from the agricultural soils and the forest soils the majority of amino acid sites have a dN/dS much smaller than one. Most sites are thus under purifying selection. However, DF does contain a fair number of sites with a dN/dS close to one, while there are nearly none of those in AG. In other words, not only are bacteria in AG more likely to have a copy of nirK, those copies are also more stringently optimized in AG than they are in DF. It thus seems to be of higher importance for the bacteria to possess a well-functioning nirK gene when their environment is farmland.
And this is bad news for us, because it means that traditional farming practices increases the amount of N2O in the atmosphere, thereby exacerbating global climate change. The good news is that the selection pressure in AG does not cause the bacteria that thrived in DF to disappear completely, so it is possible that the bacterial communities in farmland left to its own devices will eventually change their composition back into what they were in the past.
For more information about Bjørn’s work, please contact him at ostman at msu dot edu. Bjørn blogs at Pleiotropy.
