APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Jan. 2005, p. 354–362 Vol. 71, No. 1


Interactions between Oral Bacteria: Inhibition of Streptococcus mutans
Bacteriocin Production by Streptococcus gordonii

Bing-Yan Wang and Howard K. Kuramitsu

First, let me get this off my chest: I'm very excited about this paper, even before I read it. I wish, I really really wish, they had used colicins instead of gram + bacteriocins, but still ... this is a project I've been considering a long time, whether I can demonstrate that bacteriocins are modulated by quorum sensing.

Clearly, I had missed:
Kleerebezem, M., and L. E. Quadri. 2001. Peptide pheromone-dependent
regulation of antimicrobial peptide production in Gram-positive bacteria: a
case of multicellular behavior. Peptides 22:1579–1596.

So, on with the paper...

The Big Picture:
Bacteriocins, specifically colicins, are a topic I talk and write about often, so I'm providing an 'attenuated' version of the topic here. These are antibiotics, but different from the normal antibiotics we use in the pharacopiea.
1. They are produced by bacteria or eubacteria, not fungi or streptomycetes. While yeast and paramecium both have things that may be ecologically analagous (killing factors) the bacteriocins proper are not produced by them.
2. They are proteins. The classical antibiotics are small organic molecules. Many other antagonistic agents are inorganic molecules (peroxide, acid, etc).
3. They are narrow spectrum. While acidifying the environment may kill lots of things, and penicillin targets a broad range of bacteria, the bacteriocins target a small group.
4. This group is related to the organism producing the colicin, and thus is a relatively direct competitor.
5. Many of the bacteriocins are born on plasmids. This used to be a significant trait but is no longer, because we keep discovering them in the chromosome. Of the plasmids, there are large and small plasmids, so there is no particular rhyme or reason to that trait.
6. The bacteriocins require immunity proteins to keep from killing other cells that carry the bacteriocin genes.
7. The bacteriocins require some sort of bacteriocin release protein, otherwise known as BRP or the lysis protein. This kills the cell when the bacteriocin is released.
8. Bacteriocins are generally regulated by the SOS DNA damage response pathway, among others.
9. There are several kinds of bacteriocins with respect to structure. The classic colicins have three domains: one for target acquisition (the receptor binding domain), one for molecular judo (the translocation domain), and one for the coup de grace (the killing domain). Other bacteriocins may be too small for all this structure (like the microcins).
10. Naming a bacteriocin means taking the target species/genus name and roughly adding -cin to it.

So, with that background out of the way... we have suicidal cells that produce toxins that kills closely related cells, expect those that happen to share the toxin genes. Pretty complicated. How does the bacteria know when to release it? Suicide is pretty drastic if there is not threat around! Also, why would one single bacteria want to commit suicide, if there aren't relatives around to use the newly reserved resources?
So, it would make sense that possibly the bacteriocins are under some sort of quorum sensing control. After all, the current known controls for colicins (SOS) are pretty rudimentary. They only turn on in pretty much certain cell death circumstances. And yet, this doesn't square with experience. Colicins are produced by about 1% of the population all the time in a lab culture, and it goes up in stationary phase, to maybe 3%... but which 1%/3%? And how and why? Random? Lethal mutants? Additional control pathways?

This paper sets out to demonstrate that a particular bacteriocin of Stept. mutans (an important resident of the dental flora) is regulated by quorum sensing. They use the classic methods, both on agar and in broth, to estimate bacteriocin production. They also use sequencing and mutagenesis to demonstrate the involvement of the particular quorum sensing pathway.
The use of these mutants allowed them to figure out which genes were necessary for stimulating the system and which for detecting the quorum sensing molecule. It also allowed them to separate production of the molecule from production of the bacteriocin, so they could add back the chemical in a controlled fashion. They proved their point beyond a shadow of a doubt. Bacteriocin production is positively regulated by quorum sensing in Strept. mutans.

Interestingly, some of the bacteria involved secrete something into broth that is heat liable and destroys the quorum sensing molecule, not the bacteriocin. This is an interesting observation for any number of reasons. It isn't something included in that many models of competition, but it makes sense - if reliable communication is an advantage, disrupting it removes that advantage. In fact, by making the communication unreliable, it changes the entire course of selective pressures, possibly adding a need for back-up communication. Another way to make it unreliable would be to generate false signals... clearly, there is a significant lack of understanding of the difficult environment that bacteria navigate with respect to their intercellular communication.
They also found that the communication was very robust. Large amounts of the molecule were produced but only small amounts were needed to stimulate full bacteriocin production (or, maximally stimulated, at least). This indicates that the protease must be very effective. Attempts to knock out one protease did not eliminate the full proteolysis, presumably because there are others working as well. The system gets more and more complex as you look longer and harder.

As an aside, effectively, there was also some methods development as they studied the impact of different media on the bacteriocin production. What this indicates (to me) is less of in vivo importance and more about the difficulty in designing good experimental systems for in vitro bacteriocin studies, given the huge impact of relatively minor changes to the media.

There are also issues because biofilm formation is regulated by quorum sensing, and so things may be different in broth or on agar, or on other surfaces. This was briefly explored, but quickly becomes a whole other set of studies. The findings of this paper were more or less insensitive to biofilms, but then, the populations weren't mixed either, nor was it an oral model system.

This paper was extremely provacative and interesting, adding new factors to any decent model of bacteriocin production, regulation, and ecological function. One wonders whether the proteases eating the quorum sensing molecules would be important in limiting bacteriocin production. One option is that the cells producing the quorum sensing molecules are so close together and the production level is so high relative to the saturation levels for detection that the degredation doesn't matter. What it might do is insulate neighboring populations from each other's signals, making the whole system much MORE robust but not effectively limiting bacteriocin production (especially given that sensitive cells may not approach the producing colonies due a halo of toxin).
Some theories in the past suggested that bacteriocins were limited in effectiveness because of other exogenous proteases, in the intestines or mouth, for example, but this was largely a response to flawed in vivo studies that couldn't detect bacteriocin effectiveness, and so that hypothesis has been set aside.

All told, this paper is an important contribution to the field of bacteriocin research. It is thorough, within its scope, and seems to be very useful. I hope to see it built upon in the near future.