Bacteriocins are anti-bacterial molecules that typically kill closely related strains or species. These toxins have long been studied, often with the goal of finding new antimicrobials or food preservatives. However, their impact on shaping the composition of microbial communities remains somewhat unexplored. The Comstock lab studies antimicrobial molecules of the Bacteroidales, the most abundant Gram-negative order of bacteria in the human gut. The goal of these studies is to understand how these molecules influence the formation and stability of microbial communities and if they can be used to combat infectious diseases.
The lab had previously identified four Bacteroides strains of three different species that inhibited the growth of all Bacteroides and Parabacteroides strains tested contrary to all other Bacteroides secreted antibacterial molecules previously identified that specifically kill strains only of the same species. I spent a four-month long internship in the Comstock lab to find the molecule(s) that mediated this toxin activity. I performed a random transposon mutagenesis screen in one of these strains assaying for loss of killing activity. Excitingly, one transposon inserted next to a small gene annotated as a “GG cleavage motif type IIA bacteriocin”, which certainly looked like a smoking gun. Moreover, three other insertions mapped to a nearby gene encoding a putative ABC transporter with a GG cleavage peptidase domain likely involved in the secretion and maturation of a putative peptide toxin. Two adjacent genes encoded a protein of unknown function with five transmembrane domains and a putative thiol oxidoreductase. The exact same genetic region was present in one of the other three producer strains; whereas the other two producer strains contained a four gene region encoding proteins with 57% – 67% amino acid similarity. We confirmed that the small peptide genes identified encode the toxins, which we named bacteroidetocins, and that the three neighboring genes were also necessary for toxicity and were sufficient to confer the killing phenotype to Escherichia coli.
Regrettably, I had to return to France to start my PhD, were I knew I would have little time to devote to characterizing these toxins. Fortunately, the entire Comstock lab pitched in and, among other things, determined the potency and killing spectrum of these toxins. Interestingly, they found that the producing strain was sensitive to its own toxin and found no evidence for an immunity protein, but toxin production did not seem to impair the strain’s ability to colonize the gnotobiotic mouse gut. Moreover, they showed that the first two identified bacteroidetocin genes are wide-spread in human gut metagenomes and identified 17 new members of this family, two of which were also confirmed as anti-Bacteroidetes toxins.
From the first experiments in one strain of Bacteroides to the broader understanding obtained through phenotypic and bioinformatics analyses, we identified a new family of anti-bacterial toxin. These toxins are distinct from class IIa bacteriocins of Gram-positive bacteria in many ways and we proposed they be designated class IIf bacteriocins. The lab continues to study the mechanism of toxicity of these molecules and their potential role as therapeutics against diseases caused by pathogenic Bacteroidales species.