NEW YORK(GenomeWeb) – In a study of human fecal samples, a team led by researchers at the University of Helsinki was able to distinguish between obese and non-obese subjects by their metaproteomic profiles.
Detailed in a paper published last week in Proteomics, the effort identified distinct differences in the metaproteomes of obese and non-obese subjects and found that this data provided distinguishing information beyond that offered by the subjects' microbiome profiles alone.
Notably, said Carolin Kolmeder, a University of Helsinki researcher and first author on the paper, she and her colleagues found that while obese subjects had less bacteria from the phylum Bacteroidetes – which have in the past been linked to weight control – these bacteria expressed proteins at a higher rate in obese versus non-obese subjects.
This, she noted, suggests that looking only at the numbers of Bacteroidetes present in obese subjects might underestimate their actual functional contribution in such people.
More generally, Kolmeder told GenomeWeb, the finding indicates the potential value metaproteomics can bring to microbiota studies.
"If you only look at the compositional side, you still don't know what is happening from a functional point of view, what pathways are expressed," she said. "If you have a certain set of microbes, it doesn't always mean that in each person they are [behaving] the same. Then, also, many times you don't even know what the microbes can do, so it is worth looking at the protein level."
"For instance, there may be no difference in the composition [of the samples' microbiomes], but there may be a difference in the [protein] expression," she added.
In the study, the researchers looked at samples from 29 subjects ranging from "lean to morbidly obese," characterizing them using LC-MS/MS analysis on a Thermo Fisher Scientific Orbitrap Elite and adding this information to data they had previously collected on the sample's microbiomes.
To increase their number of identifications, Kolmeder and her colleagues used two different algorithms, OMSSA and X!Tandem to make peptide matches. They followed this with EggNOG database searching to obtain functional annotations for the identified bacterial and human proteins.
The researchers identified 25 groups of bacterial proteins that differed considerably between obese and non-obese subjects, the majority of which were linked to carbohydrate metabolism. Obese subjects had significantly higher levels of proteins involved in starch and pectin metabolism, indicating higher consumption of these substances and suggesting the role of eating habits in the differences in the subjects' microbial activity.
As a technique, metaproteomics — the study of proteomes in their naturally occurring environments — is still very much in its infancy and, Kolmeder noted, significantly more complicated than a typical proteomics experiment.
"When you are looking at one species and you have the genome, it is kind of straightforward," she said. "Of course, you may miss some splice variants, but most of the proteins you expect are covered in the database."
This is often not the case in metaproteomics where researchers are typically looking at samples whose compositions are not so well nailed down.
For instance, Kolmeder said, "at least 50 percent of the species you find in the gut haven't been sequenced.
Additionally, as French Institute of Environmental Biology and Biotechnology (CEA) researcher Jean Armengaud told GenomeWeb last month, many of the reference databases used in metaproteomic work are either inaccurate or incomplete, further adding to the challenge of such studies. Armengaud and his colleagues are currently working on a mass spec-based proteogenomic approach for assessing the quality of these genome assemblies.
In the Proteomics study, the researchers used their own in-house Human Intestinal Metaproteome database which they constructed from a variety of sources including the human genome, other metagenomes, and bacterial and plant genomes from organisms expected to be present in the fecal samples.
Additionally, the large sizes of the databases necessarily used in metaproteomic studies limits the usefulness of the target-decoy false discovery rate approach commonly used for assessing the quality of peptide matches in traditional proteomics experiments.
To address this issue it might be necessary to use de novo sequencing and reduce database sizes, Kolmeder said, adding that she thinks such database challenges are "the main bottleneck" in metaproteomics at the moment.