The investigation into function-species relationships undertaken here highlights some important aspects of microbiome studies and the possible inferences that can be made from such information. Although there are potential pitfalls with analysis of abundance of functions within a microbiome as has been done here such as insufficient sampling depth or incomplete sequencing of all DNA fragments, such approaches have revealed marked differences previously [5, 17]. It was found that the abundance of components of the peptides/nickel transport system differed between low and high BMI related samples, likely indicating a link between this system and obesity although such a correlation would require validation on other datasets. Taxonomic assignment of KO-associated reads showed that within the peptides/nickel transport system, there are multiple species associated with each KO, with dominance by one species being rare (Table 2). There are numerous possible reasons for this inconsistency of dominant species between KOs. As it is highly implausible that each protein is being created by different species and somehow incorporated separately into the transport systems, it is more likely LGT has resulted in operon or single gene transfers between organisms. This would result in conflicting phylogenetic relationships as observed here and makes determination of the true species involved in pathways difficult. This situation is likely due to the high degree of LGT known to occur in the human gut [18–20]. Although the presence of F. prausnitzii in all five KO sets may indicate that this species is one of the dominant organisms involved in this pathway, such extrapolation cannot be confirmed without knowing the exact history of LGT events within the microbiome, or with much deeper sequencing that allows for assembly of large genomic fragments as was performed in . Therefore further insight into detecting lateral gene transfer within the microbiome and determining the true species involved in each pathway is required before accurate relationships between species, functions and host properties such as disease be made with confidence.
Analysis of the peptides/nickel transport complex with F. prausnitzii revealed multiple operons associated with this function, each of which appeared to have been acquired through lateral gene transfer. Previous work on Fusobacterium nucleatum found an iron transport complex within the genome that resulted both from LGT of an entire operon and separate LGT events of single genes from multiple strains of other species resulting in two other operons of heterogeneous origins . Within F. prausnitzii it appears that a similar scenario has occurred within the peptides/nickel transporter with six operons types discovered. It was determined that each operon arose from separate LGT events through analysis of congruent gene trees within the operon (Additional file 4: Figure S3), which is a strong indicator of LGT [22, 23]. Five of the six operon types appear to be derived from the transfer of the whole operon into strains of F. prausnitzii, though the presence of the same operon type in some but not all strains suggests such transfers occurred prior to the divergence of certain strains. The remaining operon which was only found in a complete form within strain A2-165 appears to have been acquired from multiple sources, with the majority of the genes derived from Lachnospiraceae bacterium 3_1_57FAA_CT1 with the two ATP-binding related genes derived from other sources (Additional file 4: Figure S3). This may be due to a whole operon transfer followed by subsequent orthologous replacement and demonstrates that although the complexity hypothesis suggests such interactions between a new protein and the pre-existing complex would fail , heterogeneous integration can occur and may result in loss of fitness [25, 26], if this operon is active. Thus if multiple acquisitions did take place, this could point to a system of gradual gain of novel functions from multiple sources. However, functional assays (such as those performed in ) would be required to determine if this operon is transcribed and translated into a complex within this strain.
It may be that all five strains of F. prausnitzii acquired this transport system from independent sources within their environment (or across habitats from strains of closely related species) via gain-of-function LGT or already possessed the operon which was subsequently overwritten by multiple orthologous replacements, making the history of the lateral gene transfers difficult to trace. The relevance of nickel or short peptide transport within this species is difficult to interpret. Several enzymes such as ureases, hydrogenases, methane reductases and carbon monoxide dehydrogenases use nickel as a cofactor  though F. prausnitzii is not known to have urease activity or many hydrolases . However, a relationship between nickel concentration and butyrate production, a product of F. prausnitzii, has been postulated, and demonstrated in cattle . This could indicate that these strains are influencing the levels of butyrate within the surrounding environment. Concentrations of butyrate and butyrate-producing bacteria have been associated with lower carbohydrate intake  and also reduced obesity in mice . This suggests that a subset of the enzymatic functions associated with nickel , specifically links to butyrate production and may be connected to levels of obesity with the host, possibly through influence of butyrate production. Additionally, as this transport system can also be involved in more general transport of peptide from two to five amino acid residues in length it could be another unknown function being utilised by this species within the human digestive tract habitat. This module was characterised based upon the Opp complex in Salmonella typhimurium, which has been shown to be involved in modulating expression of surface-exposed proteins . These proteins may be involved in functions such as sporulation and virulence, both of which have been shown to be important in the human gut microbiome [19, 33]. Thus it is possible that this transporter is not involved in nickel regulation but actually modulating the cell surface responses to the digestive tract environment. As it has been shown that low levels of F. prausnitzii are associated with Crohn’s disease  and we have shown here that F. prausnitzii may also be associated with obesity, it is likely that LGT of systems such as peptides/nickel transport may contribute to host adaptation of this species, as has been observed with LGT in other species [35, 36], or play a role in determining the importance of the species within the microbiome. However, further experimental analysis would be required to confirm the link between this membrane transport system and host obesity and also determine is precise function.
Understanding the effect of habitat-directed LGT is a difficult problem. Microbiome data can be utilised to address this as has been shown here. We have found that although an overall signal for clustering of gut-associated organisms was not observed, this is not indicative of a lack of LGT. Each protein tree did not correlate exactly with a species tree as would be usually derived from single-gene studies based on 16S or other marker genes. Subsequent analysis revealed that some species that were clustered together in the protein trees were from taxonomically distant groups (Additional file 4: Figure S3). These species were usually found to be occupying similar environmental niches and were possibly associated with influencing the habitat, in this case the BMI of the host. Thus these findings signify that subsets of species may share genetic information within the environment and such LGT may impact how the habitat as a whole is shaped.