Team:Dalhousie Halifax NS/Proof
Metagenomic Library Proof Of Concept
Would it work?
Using gut microbes as a "mine" for enzymes using a metagenomic library approach is a fairly novel idea. This concept does raise some questions.
Is it even possible?
How different are microbiomes to begin with, will different animals have different microbiomes?
What enzymes are we likely to find in what animals?
Are the enzymes found in each animal predictable?
These questions guided our proof of concept experiments and allowed us to gather some evidence for our approach. This page will answer the questions raised above and will act to provide some of evidence to the feasibility of our approach.
How did we find evidence?
The Integrated Microbiome Resource at Dalhousie does 16S and metagenomic DNA sequencing and is a bioinformatic hub at the University. With their support we sequenced the 16S rRNA genes found in the environmental DNA extracted from feces of 21 mammals at the Shubenacadie Wildlife Park. With this information we were able to address a few of the questions that were mentioned above. We then chose the porcupine, beaver, arctic wolf and coyote samples to sequence in replicate, and then applied a bioinformatic tool called PICRUSt (Phylogenetic Investigation of Communities by Reconstruction of Unobserved States) to obtain information of microbial gene content found in these fecal samples. These animals were chosen because the beaver microbiome is similar to the porcupine microbiome according to our 16S data, and the arctic wolf and coyote microbiomes are similar to each other but different from the beaver and porcupine microbiomes. Our goal is to use microbial diversity and gene content in fecal samples as an approximation of the microbiomes of the mammals at the wildlife park. With this approximation we can determine what enzymes we are likely to find in the microbiomes of particular animals.
Results
These are the results of our sequencing and bioinformatic analysis separated by section and tool:
Initial 16S Sequencing
The initial 16S sequencing provided us with an overall picture of species diversity and similarity, and allowed us to create beta-diversity plots that compare the similarities of the microbiomes that were sequenced. A in figure 1 shows a bar graph that was constructed from the 16S data:
This bar graph shows us two things. The first is that the microbiomes of these different animals are very diverse. There is significant differences between each sample and their is tons of diversity. This is important because we would want a diverse representation of genetic content to start with when building a metagenomic library. The second thing is that you can notice similarities between the samples. The porcupine and beaver samples share a microbe that is of the family S24-7 and the order Bacteroidales, which is denoted by a light blue colour. You can also notice that the red deer, moose and elk share a large brown bar that matches up to the order Bacteriodales. This is by far an exhaustive list of similarites, but these two function to illustrate these similarities. To further show similarities, we built a beta-diversity plot (Figure 1. B) that allows us to see how similar or how different each sample is. This is particularly interesting because they tend to group by family of the animal they came from. So if indeed their microbiomes contain a useful and diverse group of enzymes, then we could choose multiple animals from a similar family to build a metagenomic library.
Figure 1.1. Bar Graph and Beta-Diversity Plot for Microbiome Survey Data of 21 Mammals from the Shubenacadie Wildlife Park. Bar Graph Legend here
Figure 1.2. Beta-Diversity Plot of Animal Microbiomes shows relationship between microbiomes.
From this initial 16S sequencing, it was noticed that the Arctic Wolf and Coyote were similar, and the beaver was similar to the porcupine. The Canidae and the Rodentia however were different from one another. We then decided to re-seqeuence new samples from the coyote, arctic wolf, beaver and porcupine in order to apply these to PICRUSt and see what gene content these samples have. This is found below
PICRUSt
Using PICRUSt allows us to predict the gene content found in 16S samples. It does this by matching 16S sequences to a metagenomic library of a known sequencing organisms or a metagenome whose gene content has been inferred based on the ancestors of that metagenome's gene content. What we found is that in many cases, the gene content of the beaver and porcupine was similar, and it was similar in enzymes that we would be particularly interested. That can be seen in the bar graphs for the cellulobiose phosphorylase enzyme and the endoglucanase enzyme. Cellulobiose Phosphorylase is an important enzyme for the degradation of cellulobiose and it's maintenance inside the cell. Endoglucanase breaks down the cellulose crystal, by cleaving internal bonds, into cellulobiose. These would be important enzymes in the creation of an E. coli cell that could degrade cellulose.
This same PICRUSt analysis was done on enzymes that we might expect to find in the Arctic Wolf and Coyote microbiome more often than the beaver and porcupine. The arctic wolf and the coyote microbiomes would see starch more often than the porcupines and the beavers. For this reason we would expect to see higher starch phosphorylase enzyme in the arctic wolf and coyote microbiomes. That is indeed what we see. The second enzyme that we found at a higher frequency in the arctic wolf and coyote microbiomes was the transketolase enzyme shown below. We do not have a good explanation as to why we see this, but it is interesting none-the less that we see it at a higher frequency in the arctic wolf and coyote microbiomes. This shows that enzyme content does vary by microbiomes, and it actually varies similarly between different animals. For example, we see higher cellulose degrading enzymes in the beaver and porcupine, but we see lower starch degrading enzymes. The opposite happens in the arctic wolf and coyote microbiomes.
Conclusions
All of the above information leads us to believe that our metagenomic library idea would work. Not only would the gene content in the porcupine microbiome match our predictions, we also found that we could use other animals that are similar to the porcupine to get cellulose degrading enzymes. We also found that other animals might contain useful enzymes in their microbiomes, and those are also predictable. We are confident that a metagenomic library build from fecal sample DNA of an animals whose microbiome we expect to preform a particular function would indeed lead us to useful enzymes.
