I just read a recently published paper titled “Extended-spectrum β-lactamase-producing Enterobacteriaceae in healthy companion animals living in nursing homes and in the community” in American Journal of Infection Control. Unfortunately it is not open access and not easy to find, but it was interesting. Interesting to me mostly from the standpoint of pets being a sentinel of the household. Most of the studies examining ESBL-producing E. coli to date have focused on clinical isolates, so we are ignoring the major reservoir of these bacteria which is the healthy host. Since there appears to be more household sharing of microbes between pet-owner than even between spouses, studying pets is extremely relevant from a human health standpoint. Here, the authors attempted to find ESBL-producing E. coli from dogs and cats coming in for routine vaccinations or at nursing homes using rectal swabs. They looked at a total of 376 samples and found only 9 ESBL-producing isolates, a rate of 2.5%.
I applaud the authors for undertaking this study, as these types of projects are needed to get at the true occurrence of ESBL-producers in healthy animals and humans. I do have some issues with this type of study, though. The prevalence reported here is likely vastly underestimating the true prevalence of ESBL-producers that are out there. When you look at a single isolate from a rectal swab, you are ignoring the entire E. coli population present in an animal and therefore are probably missing a lot of other potential ESBL-producers that are out there. I think what is really needed is a more comprehensive study possibly using pre-enrichment to find ESBL-producers and better characterize the genetic elements encoding for them, the genetic background of the E. coli carrying them, etc. Beyond E. coli, the plasmids carrying many of these ESBL-encoding genes are present in other Proteobacteria that may be in the animal gut but also may be in the environment. What we see in single isolates of E. coli from rectal swabs is likely the tip of the iceberg related to what is actually out there. Using companion animals to measure the occurrence of such traits in bacterial populations is an excellent way to determine what might be floating around in households.
We have struggled a lot with the best measures for plasmid cost and conjugative capacity. The challenges lie in 1) relevance, and 2) choosing an in vitro method. I often waffle (to my students’ disliking) about the best way to go about experiments measuring these parameters. If you look in the literature, there is some fantastic work that addresses in vitro measures of plasmid fitness cost (just google scholar “plasmid fitness cost”). In its most primitive form, you can measure the fitness cost of a plasmid by competing plasmid-free versus plasmid-containing bacteria in broth cultures over the course of several passages, measuring the proportions of each cell population on a daily basis (see here). Some great work by Eva Top’s group has come up with “sexier” and likely more accurate means to measure fitness cost of plasmids using multiple parameters and modeling approaches (see here). In our experience, any of these approaches result in a great deal of variation even when controlling for everything we can think of. I think this is because of the nature of the plasmids we are studying (IncA/C) which imposes a very small fitness cost on its hosts. Contrast this to previous work where fitness costs are much larger and easier to measure accurately. A second confounding component – I believe – is that rapid adaptation to IncA/C plasmids occurs when introduced into a host, but there are many factors about this we don’t understand (conditional dependency, cell density impact, temperature impact, role of host genetic background, plasmid-encoded transcriptional regulators, etc.).
We are finding that host background plays a huge role in the persistence and cost of carrying IncA/C plasmids. It has been shown that IncA/C plasmids will be lost over time in some in vitro approaches (see here), however we are finding that in wild type recipients these plasmids are rarely lost in vitro and impose virtually no fitness cost. My opinion is that doing these experiments in a strain like E. coli DH10B does not come even close to reality, and should be avoided.
As for conjugation, filter matings have become the method of choice for measuring conjugative ability of a plasmid. We have tried about every type of mating experiment we can think of, including time point filter matings, liquid conjugations, solid plate matings, etc. for IncA/C plasmids. Measuring conjugation with these plasmids seems to be more reliable and straightforward than measuring fitness cost, however we again see a variety of conjugative frequencies depending on donor and recipient. This is not at all unusual for a broad-host-range plasmid (see here) but it certainly drives me crazy when trying to determine the best approach for measuring conjugation – there are limitless possibilities. For these reasons, we have mostly stuck with E. coli K-12 (which I feel is at least a better choice than DH10B) as a donor/recipient for these experiments in an effort to first establish a baseline from which to pull what’s left of my hair out in future experiments.
I guess the point of this blog is to spew my frustrations with these types of experiments and remind myself the cautions of interpreting their results. Comments are welcome!
You have probably heard in the news about CRE, or carbapenem resistant Enterobacteriaceae. The MN Department of Health has a very nice page on CRE here. CRE has certainly gotten a lot of hype about being the next major “superbug” – and this is with good reason. But I wonder if the media going to the well one too many times for the “superbug” catchphrase is going to hurt us all in the long run? Is CRE really a superbug? It is important to note that CRE actually represent any Enterobacteriaceae that are resistant to carbapenems. This can encompass a wide variety of bacteria causing different diseases. The major player is Klebsiella pneumoniae, which is most commonly associated with urinary tract infection and resulting sepsis. This type of infection, if untreatable with antibiotics, can certainly be deadly. This is the worst case scenario, and there are a range of infections in between that can involve CRE and may or may not require antibiotic treatment. I think there are some important points to consider when discussing CRE.
CREs are mostly created through the conjugative transfer of plasmids that encode a carbapenemase. These can include KPC, NDM, and VIM. The latter two have emerged in the US only recently, but KPC in K. pneumoniae has been around for a while. The dangers of these genes being carried on plasmids is that they can move between bacteria via bacterial conjugation, and these plasmids can carry resistance to additional antibiotics. What is not well understood at this point is the extent to which conjugation occurs and precisely where these plasmids tend to spread. Plasmids encoding KPC, NDM, and VIM are diverse and confer different arrays of resistance phenotypes. It is clear that selective pressures are driving the integration of these genes into a variety of plasmid types, but it is less clear why certain plasmid types (such as IncA/C) are able to carry these genes and so many other resistance genes at the same time, with little cost to the bacterial host. One thing is clear – our labeling of these plasmids as “promiscuous” (like I have done many times) needs to be done with caution. If plasmids such as IncA/C were as promiscuous as I originally thought they were, they would have easily swept E. coli and Salmonella populations by now and we would be experiencing a pandemic of MDR E. coli and Salmonella carrying IncA/C plasmids (and CRE bacteria). This is not the case and probably will not ever be the case. There is something that limits the dissemination of these plasmids among E. coli and Salmonella. Could it be a quorum sensing mechanism that limits the conjugation of this plasmid? Maybe. But my main point is that CRE is certainly concerning. It is a fascinating research topic. However, I am hesitant to call CRE a superbug because my view of a superbug is a potential global killer. Let’s hope that CRE does not reach that status.