Please visit the new BioModels platform to access the latest content. This website is no longer updated and will be retired on 31 May 2019.
BioModels Database logo

BioModels Database


Kaiser et al., (2014). Cecum lymph node dentritic cells harbor slow-growing bacteria phenotypically tolerant to antibiotic treatment.

June 2015, model of the month by Nick Juty
Original models: BIOMD0000000527.

The use of antibiotics in the treatment of bacterial infections has a long and distinguished history, frequently being a newsworthy topic, largely due to the increasing incidence of antibiotic resistant bacterial strains. Bacterial resistance to particular antibiotics can be determined by either genetically encoded determinants (part of the pathogen's genome) or by 'noninherited' means, the latter allowing genetically susceptible bacteria to survive in what would otherwise be non-viable conditions. Such 'phenotypic tolerance' is most often displayed during the stationary phase of bacterial growth, but has shown to be reversible, once more making the organism susceptible to the anti microbial agent, during exponential growth. Both empirical and anecdotal evidence suggest that slow pathogenic growth rates in vivo are responsible for the reduction in observed antibiotic efficiency, when compared to their action in vitro.

The authors [1, BIOMD0000000527] used a mouse model for Salmonella enterica (Figure 1) to determine the underlying mechanism responsible for phenotypic tolerance of the bacterium with the antibiotic ciprofloxavin. Salmonella enterica is a Gram-negative bacterium from the same family as Escherichia coli (Gamma proteobacteria). It is a rod-shaped flagellated cell and is known to cause infectious disease in animals, and is thought to be responsible for around 30% of all food-borne infections.

Figure 2

Figure 2Tissue penetration of antibiotic: Cells infected with Salmonella were treated with ciprofloxacin or dextran, and imaged using confocal microscopy. Profiles show that ciprofloxacin efficiently penetrated intracellular compartment, while dextran did not, meaning that there was high bioavailability of the antibiotic within the tissues under analysis. Figure taken from [1].

The mouse model used by the authors allowed the spread of the pathogen systematically, to allow the consideration of 'complicated' Salmonella infection. Following a series of experiments, focus was directed to the cecum draining lymph node (cLN). While Salmonella was found at other locations such as the spleen, cLN is a more convenient location from which to isolate tolerant bacteria, with less chance of sample contamination, and contains a larger population of the pathogen.

Throughout the experiments, a greater than 50 fold concentration of antibiotic above the established ex vivo Minimum Inhibitory Concentration (MIC) was used, which ensured clearance of the bacteria from the gut lumen. However, while the antibiotic was shown to be active in cLN within a few hours, a small number of bacteria (50-1,000) were shown to remain viable there for up to 10 days. This was identified as a subpopulation of tolerant bacteria. The in vivo biocidal activity of this target antibiotic concentration was verified using pharmacokinetic analysis, while tissue penetration was confirmed using confocal microscopy (Figure 2). Furthermore, colonies formed from cLN re-isolates were shown to still be ciprofloxacin-sensitive, highlighting that the bacterium had not undergone any genetic modification, and were shown to be capable of re-initiating infection. Further fluorescence activated cell sorting (FACS) and microscopy techniques suggested that classical dentritic cells (CD103+CX3CR1−CD11c+; cDCs) may represent an important reservoir of tolerant cells. The significance of the dentritic cell population was further established through experimental systems to both reduce and increase this cellular population, which was found to be reflected in the pathogen load.

It was hypothesized that slow growth rate of the pathogen in the cLN may be responsible for antibiotic tolerance. Further evidence was needed to establish the actual growth rate of the pathogen in this situation, in vivo, which is quite a challenge. To this end, the authors used a novel strategy employing a defined mixture of tagged, isogenic Salmonella wild-type and 'marked' cells. The marked cells contain a variety of short but defined neutral nucleotide insertions, quantifiable by RT-PCR. This approach was used to estimate the replication rate of the pathogen in the cLN by analysing the infection data with a stochastic birth-death model, extended by immigration [2].

The model predicted the proportion of strains migrating to the cLN, and their population size as a function of the rate of immigration, μ, replication, r, and clearance, c. From this, it was possible to infer bacterial cLN immigration and replication rates. This approach was verified using wild-type and knockout mice.

These parameters were then used to estimate Salmonella population composition at the beginning of antibiotic treatment (Figure 3, 4), and extended to determine growth rate of the tolerant fraction (Figure 3). Analysis of the population dynamics suggest that the doubling time of bacteria increased to 44 hours at 5-10 days post infection (4-9 days post antibiotic administration) (Figure 5), with clearance rate dropping to below 0.01 per hour. This demonstrates that these bacteria enter a tolerant state characterised by very slow replication times and very low clearance rate.

Figure 5

Figure 5Model-derived parameters: Parameters were derived from fitting the model to experimental data. It can be seen that the replication rate for the bacteria falls from 4.6 during the first days of treatment, to 0.4 by day 10. Figure taken from [1].

Figure 1

Figure 1 Salmonella enterica is a Gram-negative, flagellated bacterium that is for infecting around 1.3 million people per year, resulting in around 500 deaths. It has several virulence factors, including an enterotoxin, which is responsible for its most notorious symptoms, diarrhea and vomiting. Image from [MicrobeWiki].

While the majority of human food poisoning cases it causes are noncomplicated, Salmonella can spread beyond gut-draining lymph nodes and cause life-threatening complications. In those instances, antibiotics such as ciprofloxacin are typically used for treatment, since fluoroquinolone antibiotics such as ciprofloxavin, display broad range antimicrobial activity, have rapid response rates in vitro, and display excellent tissue penetration. However, while in vitro biocidal activity can be measured in the span of a few hours, in vivo treatment often spans over a week, with relapses a frequent occurrence.

Figure 3

Figure 3The original model accounted for immigration from the cecum, during the first 24 hrs post infection, and replication within the cLN. In this work, following ciprofloxacin treatment, immigration is removed, and clearance incorporated. It was estimated that during day 1, 298 cells enter the CLN, and grows at a rate 2.82 per day. Since tratment kills gut bacterial, influx of bacteria (immigration) will be null during the treatment days (right side). Figure taken from [1].

Figure 4

Figure 4Experimental data to inform model: Mice were infected with a mixed population of Salmonella containing marked cells which could be identified through RT-PCR. After 1 day, these were treated with ciprofloxacin for up to 10 days. Colony Forming Units (CFU) were determined, and individual tags abundances established, and the data used to fit the model. X indicates experimental data, grey lines indicate simulations. Figure taken from [1].

The authors were also interested in finding a strategy to reduce antibiotic tolerant pathogen populations in wild-type hosts; there has been speculation that the instigation of a host immune response during antibiotic treatment may be beneficial. It was found that a single dose of lipopolysaccharide (LPS) administered during antibiotic treatment. This triggered an immune response within two hours, and was sufficient to reduce the number of tolerant bacteria, often to levels below the experimental level of detection. Further investigations suggest this may be through the actions of indirect (paracrine) signaling, and demonstrate that persistent, tolerant pathogen load in tissues are indeed susceptible to innate immune response, suggesting that for this organism, this treatment regime may be worth exploring further.

It seems likely that, given the diversity of pathogen types, the plethora of potential niches within the host, the multitude of virulence factors, and the wide-ranging growth rates, there will be a number of different pathogen-specifc adaptions, and potentially some common mechanisms that can be explored using studies such as presented here, which elegantly combine experimental work with mathematical modeling to investigate an otherwise almost intractable problem.

Bibliographic references

  1. Kaiser et al. Cecum lymph node dentritic cells harbor slow-growing bacteria phenotypically tolerant to antibiotic treatment. PloS Biol. 12(2):31001793, 2014.
  2. Kaiser et al. Lymph node colonization dynamics after oral Salmonella Typhimurium infection in mice. PloS Pathog. 9(9):e1003532, 2013.