Monday, February 27, 2012

The magic of antibiotic tolerance

Bactericidal antibiotics are effective at killing proliferating bacteria as long as they don't carry mutated or acquired genes that encode resistance to the antibiotics. Unfortunately even antibiotic-sensitive bacteria can tolerate antibiotics under some circumstances. Bacteria that are in a nondividing "dormant" state often survive antibiotic exposure.  When the antibiotic is removed and growth resumes, the bacteria regain susceptibility to antibiotics.

At first glance antibiotic tolerance appears to be a passive process in which nondividing cells survive simply because the target of the antibiotic is inactive.  However, this is not correct.  Antibiotic tolerance requires an active response by the bacteria.  The nondividing bacteria that survive antibiotic treatment are called persisters.  Persisters may account for infections that are difficult to eradicate with antibiotics.

ResearchBlogging.orgPersisters were first identified in 1944 by Joesph Bigger, who was testing the effectiveness of the new miracle drug penicillin on cultures of Staphylococcus.   However, interest in antibiotic tolerance waned as antibiotic resistance came to be a problem.  In the 1980s Harris Moyed revisited the issue of antibiotic tolerance, and his group harnessed the power of bacterial genetics to isolate mutants that exhibited abnormally high frequencies of persister formation and to map the mutations that caused the novel trait.

As Joseph Bigger discovered, very rare persisters can even be identified in growing cultures, around 1 in 100,000 bacteria.  How can these rare bacteria behave so differently from those surrounding them if antibiotic tolerance doesn't involve alterations to the bacterium's DNA?  The current thinking is that the persister state is triggered randomly in a small fraction of proliferating bacteria due to random fluctuations in the expression of a small number of persister genes.  In the rare bacterium (about 1 in 100,000), the persister genes will be expressed at a high enough level to induce the persister state and slow growth of the bacterium.

Certain environmental cues can also enhance the development of persisters.  For example, the fraction of persisters in a culture increases substantially as nutrients are depleted.  At stationary phase, when the bacteria stop increasing in number, at least 1% of the bacterial cells become antibiotic tolerant.

One of the active responses that stimulates persister cell formation is called the stringent response, which rapidly generates the unusual nucleotides ppGpp and pppGpp when bacteria are starving for nutrients.  Historically these derivatives of GDP and GTP were known as "magic spots" because they appeared as novel radioactive spots on thin-layer chromatograms when E. coli starved for amino acids were labeled with 32P-phosphate. (p)ppGpp has wide-ranging effects on bacterial physiology.  The best-known activity of (p)ppGpp is its attachment, along with the protein DksA, to RNA polymerase during amino acid starvation, causing transcription of rRNA and tRNA genes to cease and transcription of amino acid biosynthetic operons to increase.  This make sense since there's no point in making more ribosomes until more amino acids are available to support protein synthesis and growth of the bacteria.

Magic spot!  Source

(p)ppGpp production is also necessary to generate the rare persisters in cultures of proliferating bacteria.  When the two genes coding for the enzymes that make (p)ppGpp, relA and spoT, are deleted, persister cells become even more rare in growing E. coli cultures.

A recent study published in the journal Science looked at the role of the stringent response in inducing antibiotic tolerance in biofilms, in which nondividing bacteria are embedded in a matrix secreted by the bacteria.  These studies were conducted with E. coli and Pseudomonas aeruginosa strains with deletion mutations in relA and spoT.  When the biofilms formed by the mutant and wild-type strains were treated with different antibiotics, the mutants turned out to be much more sensitive to the antibiotics even though the bacteria were not dividing (data for P. aeruginosa shown in bar graphs below).  Similar results were obtained when standard cultures of the P. aeruginosa mutant and wild-type strains were grown to stationary phase and then treated with antibiotics.

Figure 1E from Nguyen et alP. aeruginosa was allowed to form biofilms on polycarbonate membrane filters resting on agar culture medium.  The biofilms were incubated with different antibiotics or with no antibiotic (control).  To determine the number of bacteria surviving antibiotic treatment, the bacteria in the biofilms were dispersed by vigorous mixing and plated onto agar culture medium to determine cfu counts.  +SR, complementation of the ΔrelA ΔspoT mutant with intact copies of relA and spoT; *P ≤ 0.05, ***P ≤ 0.0005 versus wildtype.

Bactericidal antibiotics, regardless of their target, are now known to enhance production of reactive oxygen species (ROS) within bacteria.  If not detoxified by enzymes such as catalase and superoxide dismutase, ROS can fatally damage the bacteria by reacting with their DNA, protein, and lipids.  ROS is generated within bacteria during the course of their normal metabolic activities, even when antibiotics are not present.  When the authors measured the amounts of hydroxyl radical generated within untreated E. coli and P. aeruginosa biofilms, the ΔrelA ΔspoT mutants were burdened with higher levels of hydroxyl radicals than the wild-type strains. Why do the mutant strains have higher levels of hydroxyl radicals?  Further analysis of the bacteria in the biofilms showed that the ΔrelA ΔspoT mutants also had lower levels of catalase and superoxide dismutase activity.  It's possible that bactericidal antibiotics triggered production of lethal amounts of ROS that the mutants could not handle due to their insufficient production of catalase and superoxide dismutase, but this needs to be confirmed by further experimentation using bacterial strains with the superoxide dismutase and catalase genes knocked out.

The Science study also looked at infected laboratory mice treated with antibiotics.  Wild-type and ΔrelA ΔspoT P. aeruginosa strains were grown to stationary phase, and a lethal dose of the bacteria were injected into the abdominal cavity of mice.  Four hours later the animals were treated with the antibiotic ofloxacin.  The antibiotic was more effective at preventing lethal infection by the ΔrelA ΔspoT mutant than those caused by the wild-type strain.  Similarly, ofloxacin was more effective at reducing the number of bacteria in a biofilm chamber implanted underneath the mouse skin when the biofilm was composed of the ΔrelA ΔspoT mutant.

One must keep in mind that multiple pathways to persister cell formation have been identified (see figure below).  Note that for the animal experiments persisters were allowed to develop in the test tube prior to animal inoculation.  It is possible that the stringent response would not be involved at all if persisters were allowed to develop during infection instead.

Figure 4 from Lewis, 2010.  The redundant pathways to persister formation are shown.  FMN, flavin mononucleotide pool; pmf, proton motive force; TAs, toxin/antitoxin modules

Featured paper

Nguyen, D., Joshi-Datar, A., Lepine, F., Bauerle, E., Olakanmi, O., Beer, K., McKay, G., Siehnel, R., Schafhauser, J., Wang, Y., Britigan, B.E., & Singh, P.K. (2011). Active starvation responses mediate antibiotic tolerance in biofilms and nutrient-limited bacteria Science, 334 (6058), 982-986 DOI: 10.1126/science.1211037

Key references

Bigger, J.W. (1944). Treatment of staphylococcal infections with penicillin by intermittent sterilisation The Lancet, 244 (6320), 497-500 DOI: 10.1016/S0140-6736(00)74210-3

Moyed H.S., & Bertrand K.P. (1983). hipA, a newly recognized gene of Escherichia coli K-12 that affects frequency of persistence after inhibition of murein synthesis. Journal of bacteriology, 155 (2), 768-775 PMID: 6348026

Cashel, M., & Gallant, J. (1969). Two compounds implicated in the function of the RC gene of Escherichia coli Nature, 221 (5183), 838-841 DOI: 10.1038/221838a0

Paul, B., Barker, M.M., Ross, W., Schneider, D.A., Webb, C., Foster, J.W., & Gourse, R.L. (2004). DksA: a critical component of the transcription initiation machinery that potentiates the regulation of rRNA promoters by ppGpp and the initiating NTP. Cell, 118 (3), 311-322 DOI: 10.1016/j.cell.2004.07.009

Korch, S.B., Henderson, T.A., & Hill, T.M. (2003). Characterization of the hipA7 allele of Escherichia coli and evidence that high persistence is governed by (p)ppGpp synthesis Molecular Microbiology, 50 (4), 1199-1213 DOI: 10.1046/j.1365-2958.2003.03779.x

Kohanski, M.A., Dwyer, D.J., Hayete, B., Lawrence, C.A., & Collins, J.J. (2007). A common mechanism of cellular death induced by bactericidal antibiotics Cell, 130 (5), 797-810 DOI: 10.1016/j.cell.2007.06.049

Lewis, K. (2010). Persister cells Annual Review of Microbiology, 64 (1), 357-372 DOI: 10.1146/annurev.micro.112408.134306

Related posts


  1. Thank you for posting about the issue of persisters. I have a lot to think about after reading this.

    I was wondering if you have read this particular paper yet?

    Persistence of Borrelia burgdorferi in Rhesus Macaques following Antibiotic Treatment of Disseminated Infection (2012) Monica E. Embers, Stephen W. Barthold, Juan T. Borda, Lisa Bowers, Lara Doyle, Emir Hodzic, Mary B. Jacobs, Nicole R. Hasenkampf, Dale S. Martin, Sukanya Narasimhan, Kathrine M. Phillippi-Falkenstein, Jeanette E. Purcell, Marion S. Ratterree, Mario T. Philipp

    full text:

    A few people have made note that the authors used a large dose of spirochetes for needle inoculation and that the antibiotic used, ceftiofur, is not one that has been tested for efficacy in animal models of borreliosis.

    I managed to find a few papers on equine borreliosis and the use of ceftiofur - some behind pay walls - but nothing else. (If you know of other studies, please let me know.) The criticism leveled at the researchers is that ceftiofur does not have the same pharmacological (PK, PD) effect on Rhesus macaques that ceftriaxone would - when this non-human primate study was intended to mirror a comparable study on humans which used ceftriaxone for treatment.

    What are your thoughts on this? Are these criticisms valid? How would you find out that ceftiofur and ceftriaxone were pharmacological equivalents for this purpose?

    Is there a way to measure what the right volume of spirochetes is for needle inoculation? I've read a few papers on this and the number of infectious organisms needed to establish an infection vary by host and strain of Borrelia.

    I think that even if this study is not an exact analog to a human treatment trial, the outcome in and of itself for it being an animal model study is interesting: Even with all the antibiotics which were given the macaques with Borrelia burgdorferi infections, live spirochetes could still be transmitted (xenodiagnosis) post-treatment and there was RNA transcription detected (lp 28-1, OspA - refer to the study) as well.

    I think more research in this area is needed.

  2. Apparently you don't allow active links. The link to copy for the full text is here:

    1. I didn't realize until now that ceftiofur and ceftriaxone were different antibiotics (though closely related). I had assumed incorrectly that "Ceftiofur" was a brand name for ceftriaxone. It would certainly be a deficiency in the paper since the PK/PD of ceftiofur in rhesus monkeys and even the MIC are unknown. The other possibility is that they really did use ceftriaxone and simply made an error when they were writing up the Methods section. It just wouldn't make sense otherwise - they write "ceftriaxone" throughout the rest of the paper, and they supposedly tailored the drug regimen to the one used by Klempner et al., who administered ceftriaxone to their subjects. The only possible reason I can come up with for their use of ceftiofur is that they were asked to use the alternative drug by their institute's IACUC (animal committee).

      There may be yet another issue with the two month doxycycline treatment that followed the ceftiofur/ceftriaxone. Doxycycline appears to decay faster in the macaques (half-life of 6 hours, my guess from their numbers on p. 2) than it does in humans (15-25 hrs from table in Wormser's 2009? review). I don't know if this difference would account for some of their results, but it's something else to consider.

      The only proper way to figure out how many spirochetes to inject is to get a lot of macaques and inject different numbers of Bb into different groups (an ID50-type experiment). Their IACUC would not be likely to permit this. They really need to use ticks to inoculate the animals.

    2. Thanks, Microbe Fan - you responded just as I was posting another comment. Good timing!

      So far two people which I have interacted with online about this issue (Dr. Phillip Baker, who was Program Officer for Lyme disease research while this Rhesus macaque study was being conducted and a bacteriologist who identifies as "Henry" on a forum) have pointed out the difference between ceftiofur and ceftriaxone. Otherwise, I wouldn't have known there was a difference either or asked - I assumed as you did. But I also read the paper and saw the authors' consistent use of ceftriaxone throughout the paper so it was odd when ceftriofur popped up. If all it is was an error - well, it's amazing how much argument it is generating within the Lyme disease community.

      Your point about doxycycline is well taken. If the half life is that much less, how does one know it reached MIC/MBC during those two months? It may not have.

      Even then, though, if ceftiofur is pharmaocologically similar enough in action to ceftriaxone (and let's say they did use ceftiofur, hypothetically) - if spirochetes remained behind which were metabolically active, xenodiagnosis studies succeeded, and RNA transcription from spirochetes was present... does this indicate that this infection can persist, that it has persisters?

      I think that it's an interesting study, but in order to truly replicate what happens in a real life situation, the researchers would have to use ticks with a low passage infectious strain of Bb and not needle inoculation. I think that in the paper, the researchers even state this as their intention for a followup study.

  3. Just pointing out a few items in your article:

    "When the biofilms formed by the mutant and wild-type strains were treated with different antibiotics, the mutants turned out to be much more sensitive to the antibiotics even though the bacteria were not dividing (data for P. aeruginosa shown in bar graphs below). Similar results were obtained when standard cultures of the P. aeruginosa mutant and wild-type strains were grown to stationary phase and then treated with antibiotics."

    This suggests - at least hypothetically - that if a host is infected with mutant bacteria which persists that varying the type of antibiotic to treat the infection works well. The same also appears to be true for non-mutant bacteria, too. Does this mean the model of treating bacterial infections with the same antibiotic for a given period of time and stopping all antibiotics may not be the optimal treatment plan? That switching off to different antibiotics may be a better approach? I would think this would be more an issue with bacteria which are slow to divide to begin with - rather than ones which divide quickly.

    "Wild-type and ΔrelA ΔspoT P. aeruginosa strains were grown to stationary phase, and a lethal dose of the bacteria were injected into the abdominal cavity of mice. Four hours later the animals were treated with the antibiotic ofloxacin. The antibiotic was more effective at preventing lethal infection by the ΔrelA ΔspoT mutant than those caused by the wild-type strain. Similarly, ofloxacin was more effective at reducing the number of bacteria in a biofilm chamber implanted underneath the mouse skin when the biofilm was composed of the ΔrelA ΔspoT mutant."

    It's interesting to note that both strains are in a stationary or dormant phase when the antibiotics are used - but the antibiotics are more effective on the mutants. What are the particular qualities of ofloxacin which make it more likely to stop these mutant strains than wild-type strains?

    By the way, here are two articles recently posted to Science Daily others may want to read which are related (first directly and somewhat indirectly) to the topic of tolerance: - Quote from article: "While there is still much to be learnt about how and under which circumstances tolerance to infection is employed by the host, most of what is currently known about the molecular mechanisms underlying this host defense strategy comes from work carried out at the Instituto Gulbenkian de Ciência by the group led by Miguel Soares." - Frontal Attack or Stealth? How Subverting the Immune System Shapes the Arms Race Between Bacteria and Hosts. Quote from article: "Our findings suggest that the low ID50 (more infectious) bacteria be classed as working by stealth; the high ID50 bacteria are those that resort to frontal attacks."

    Interesting stuff. Suggest reading the papers on which the the articles are based.

    1. I'll address your question about ofloxacin first. The authors would say that there's really nothing special about ofloxacin. The other bactericidal antibiotics that they tested in vitro would also have killed the nondividing mutant bacteria more effectively than the nondividing wild-type bacteria in mice. The property of oxfloxacin that may account for this is its ability to generate lethal amounts of reactive oxygen in the nondividing mutant bacteria - but the other bactericidal antibiotics have this property as well even though their initial target on the bacterium differs.

      As for your first point, I don't think that this work suggests that switching antibiotics would help since all bactericidal antibiotics lead to the same mode of killing (generation of lethal amounts of reactive oxygen).

    2. ...and C.O., thanks for sending me the two links. The second one sounds especially interesting.

  4. Microbe Fan,

    Thanks for responding to my questions. I realized right after I posted it that maybe the antibiotic didn't matter in outcome - thanks for confirming that. At some point I have something to ask you about ROS and iNOS, but it can wait.

    You're welcome for the links. And yes, I agree - I thought the second one might be particularly of interest to you and other readers here.

  5. Update:

    Thought you might want to know that Embers et al have issued a correction on PLoSONE. They used ceftriaxone for the entire study, and not ceftiofur.