Salmonella, E. coli, and Plasmids, Oh My! Solving the Plasmid Persistence Problem and the Virtues of Open Non-Equilibrium Models

Ed Yong has some very good coverage of a recent PNAS paper that describes how bacterial mini-chromosomes, known as plasmids and that often carry disease-related or antibiotic resistance genes, can jump from bacterium to bacterium at really high rates when there’s an infection:

Barbel Stecher from ETH Zürich showed that when mice are infected with Salmonella, the native E.coli in their guts start to bloom as well. Once mere bit-players, their numbers go up by more than 100 times. In some cases, E.coli accounted for more than 80 percent of the bacteria in the rodents’ guts.

Stecher found that several of the E.coli strains in this melange carried a plasmid that’s 99% identical to one normally borne by Salmonella. Known as P2, it includes several genes that allow bacteria to resist antibiotics. The E.coli probably picked up this plasmid from the Salmonella which had suddenly joined them. After all, the mice in the study had been reared in faultlessly clean and carefully monitored conditions. It’s unlikely that they had been exposed to any Salmonella before the experiment.

Blecher infected sterile mice with a strain of Salmonella that had P2, and one of the earlier E.coli strains that did not. She waited. Both species of bacteria bloomed and within four days, virtually every single E.coli had picked up the P2 plasmid from its neighbours. Bacteria pass genes to each other all the time, but in these infected guts, it happened with an efficiency that’s practically unheard of.

As Yong put it, “Infectious bacteria aren’t just making us ill – they’re also weaponizing our allies.” But I think this paper raises a broader issue: how are plasmids maintained? Plasmids can impose a cost on cells due to their replication machinery, and so most theory suggests that any beneficial genes (beneficial for the bacterium, not for humans or other hosts) should move to the bacterial chromosome, after which the plasmid-carrying bacteria should disappear since these bacteria no longer require the plasmid but still bear the burden. This has been referred to as the “plasmid persistence problem“. The authors note in the supplemental methods*:

In general, the maintenance of a plasmid poses a nonzero cost to the bacterial host. This fitness burden represents a selective pressure disfavoring the maintenance of the plasmid. The evolution and persistence of plasmids (despite this fitness burden) is therefore explained by two different—but not mutually exclusive— mechanisms, “infectious transmission” and plasmid-encoded beneficial genes enhancing the fitness of the host bacterium.

In theory, infectious transmission alone (i.e., without the need to encode beneficial genes) would be sufficient for explaining the maintenance of plasmids…. However, based on the available evidence it has been suggested that conjugative HGT rates of plasmids are generally too low to guarantee long term plasmid maintenance…. Our data …shows that high-rate conjugative plasmid transfer does indeed occur in nature, thus leading to nearly 100% transconjugant formation if plasmid-free recipients are encountered at sufficient frequency, e.g., in pathogen-inflicted bacterial blooms. Thus, occasional blooms represent an environmental feature favoring maintenance of conjugative plasmids.

Under certain conditions, beneficial genes can foster the long term maintenance of a plasmid. However, it should be noted that this plasmid-localization of beneficial genes is intrinsically instable: due to the fitness costs imposed by plasmid itself, beneficial genes should eventually move to the chromosome. Based on these theoretical considerations, it has been proposed that beneficial genes, i.e., those conferring a selective advantage only in specialized environments, do stabilize the maintenance of a given plasmid in a bacterial population (or superpopulation of different clones or species), if “sweeps” occur in the population …. These sweeps can be caused by the emergence of a new clone with superior growth properties, environmental changes favoring the growth of a particular clone or by new clones or species which newly enter the ecological niche …. If plasmid transfer occurs at an early stage of such a sweep, the growth advantage conferred by the plasmid-encoded beneficial gene favors the emergence of the transconjugant thus amplifying the beneficial plasmid. If sweeps occur often enough, the beneficial gene should foster the maintenance of the plasmid. However, so far there has been little data from naturally ecosystems to support these theoretical considerations… Our data suggest that pathogenesis-inflicted changes in the gut-ecosystem of an infected host might represent such a situation. Blooms of the incoming Salmonella spp. and the resident commensal E. coli species can be regarded as sweeps.

To translate into English, if bacterial populations are undergoing “sweeps”, where a strain rapidly replaces others, then plasmids will remain intact (and not have their genes integrate into the main bacterial chromosome) provided the plasmid can jump into the new strain (which the others demonstrate can happen). If the system were at equilibrium, then the plasmid would first lose its functions** and ultimately disappear. But if the plasmid can manage to ‘ride sweeps’, then it can escape to a new strain before it begins to disintegrate (of course, if the replaced strain with the plasmid is still ‘out there’, the plasmid might begin to disintegrate in that strain). Here’s a picture (from here):


If we viewed the problem as a closed system that reaches a steady-state equilibrium, we would be very misled. ‘Out there’ stuff gets complicated. We desperately need more experimental and population survey work to figure out what’s happening.

Tomorrow, I’ll talk about what this paper suggests about colicin plasmid evolution (don’t worry, I’ll translate into English or my approximation thereof).

*Have I ever mentioned how much I HATE supplemental method sections? This should have been front and center in the paper. Stupid page limits.

**While the plasmid persistence problem argued that genes would jump into the main bacterial chromosome, the alternative is that once these genes are no longer advantageous, they will simply be selected against and lost entirely.

Cited articles: Stecher B, Denzler R, Maier L, Bernet F, Sanders MJ, Pickard DJ, Barthel M, Westendorf AM, Krogfelt KA, Walker AW, Ackermann M, Dobrindt U, Thomson NR, Hardt WD. 2012 Jan. 9. doi: 10.1073/pnas.1113246109

Bergstrom CT, Lipsitch M, Levin BR. 2000. Natural Selection, Infectious Transfer and the Existence Conditions for Bacterial Plasmids. Genetics 155, 1505-1519.

This entry was posted in E. coli, Evolution, Plasmids, Salmonella. Bookmark the permalink.

3 Responses to Salmonella, E. coli, and Plasmids, Oh My! Solving the Plasmid Persistence Problem and the Virtues of Open Non-Equilibrium Models

  1. Geraldine says:

    Great discussion of a very interesting paper. There are some additional plasmid-centric considerations that deserve a mention though.

    One is the prevalence of “poison-antidote” style addiction systems, in which a plasmids carries two genes, one coding for a protein that is toxic to the cell, and the other coding for a protein that neutralizes the poison. Because the poison is individually more stable than the antidote, the cell has to keep a copy of the plasmid around to keep producing antidote. The regulation of relative expression and stability between the poison and the antidote is very finely tuned, between each other and also relative to the copy number of the plasmid, so the poison-antidote gene pair cannot simply be transferred to the chromosome to render the plasmid unnecessary.

    Then there are examples of pathogens like Bacillus anthracis, which causes anthrax. The main genetic components responsible for anthrax are located on two plasmids, pXO1 and pXO2. Anthrax strains from all around the world are highly clonal (they’re pretty much all the same, genetically) and have seemingly existed in that form for thousands of years. B. anthracis does not seem to have been subject to population sweeps as described in the paper, yet it has kept its two large (and, according to reigning assumptions, costly to maintain) plasmids around virtually unchanged, rather than incorporating their “helpful” (for anthrax!) genes into its chromosome and getting rid of the plasmids themselves. It is possible that this is due to particularities of its lifestyle, and that the picture is not as simple as previously thought — its close cousin Bacillus cereus has plasmids related to yet distinct from pXO1 and pXO2, that probably behave more like the “open non-equilibrium” described in the paper. Hopefully by studying the differences and similarities in the relationships between the plasmids and their hosts, we can figure out the range of plasmid persistence behaviors – and get a better grip on what that means for us as walking bacterial incubators!

    • Agreed. The partitioning system are key. I might follow up with a post about some of the shortcomings of the plasmid evolution theoretical literature. Or I might do some science. Not sure yet.

  2. Geraldine says:

    Science, huh? Is that important?

    (I would leave the plasmid evolution theoretical literature well enough alone — it doesn’t like to be taken out of its own bubble universe into the real world).

Comments are closed.