Research ArticleMICROBIOLOGY

Applying ecological resistance and resilience to dissect bacterial antibiotic responses

See allHide authors and affiliations

Science Advances  05 Dec 2018:
Vol. 4, no. 12, eaau1873
DOI: 10.1126/sciadv.aau1873
  • Fig. 1 Response of an ESBL-producing population to cefotaxime, a β-lactam.

    (A) Schematic of antibiotic response of an ESBL-producing population. In the absence of antibiotic, bacteria reproduce and consume nutrients. Upon the introduction of an antibiotic, some of the bacteria undergo lysis and release Bla and a small amount of recyclable nutrients into the environment. The released Bla degrades the antibiotic (blue inhibition arm). (B) Time course of antibiotic response. In the absence of an antibiotic (black curve), bacteria grow to a carrying capacity without any delay. In the presence of sufficient antibiotic (gray curve; A = 100 μg/ml cefotaxime), the population displays the characteristic crash, as the cells lyse, and recovery after the Bla released from lysed cells degrades the antibiotic. (C) ESBL substantially degraded cefotaxime in a short time window. The supernatant from a culture of sensitive cells still contained substantial concentrations of cefotaxime, as depicted by the zones of inhibition in the lawn of sensitive cells (strain MC4100, left column). The supernatant from the culture containing ESBL-producing bacteria did not contain substantial concentrations of cefotaxime, as depicted by the full lawns (right column). Arrows indicate where supernatant was placed on the agar plate. (D) Populations previously exposed to cefotaxime exhibited the same temporal dynamics. Culture was treated with a range of antibiotic concentrations. After 24 hours, bacteria from the recovered population were used to reinoculate fresh media with or without cefotaxime (25 μg/ml). During the second round of treatment, time courses from the populations previously exposed to cefotaxime (0, 2, or 200 μg/ml) were identical, suggesting that the population recovery was unlikely due to mutants or phenotypic variants with increased tolerance. (E) Bla production is not induced by cefotaxime. We used fluorocillin to determine that the isolate’s Bla production is not significantly increased by the addition of antibiotic. Here, the Bla activity present in a population after 3 hours of exposure to a range of antibiotic concentrations was quantified by the rate at which fluorocillin was hydrolyzed and produced green fluorescence. One-way analyses of variance (ANOVAs) indicate that the increase in fluorescence recorded was insignificant when compared to the control.

  • Fig. 2 Quantifying resilience and resistance.

    (A) Time courses are used to quantify a population’s resilience. When no antibiotic was added (black curve), the population grew up unperturbed and reached a target threshold density (here, 50% of the carrying capacity) in time = T50%. If the antibiotic concentration added was very low (blue curve; A = 0.5 μg/ml cefotaxime), then the population reached the threshold density in a similar time to the untreated population. As the antibiotic concentration increased, the degree of lysis increased and affected the time necessary for the treated population to reach the threshold (Embedded Image). We characterized the population’s resilience for a range of antibiotic concentrations as the inverse ratio of the times to the half-maximal carrying capacity (Embedded Image). (B) Net growth rate quantifies population’s resistance. When no antibiotic was added, the population’s net growth rate decreased over time as it consumed the available nutrients and approached stationary phase. When a low dose of antibiotic was added, the net growth was not notably altered (blue curve). In this instance, the treated population’s minimum net growth rate is recorded as ρA and compared to the untreated population’s net growth rate at the same time (ρ0). As the antibiotic concentration increased, the net growth rate curve of the treated population deviated more from the untreated curve. For each antibiotic concentration, ρA was recorded as the net growth rate at the point of maximum negative deviation from the untreated population and normalized by ρ0. We characterized a population’s resistance as the ratio of recovery times (ρA0). (C) Resistance and resilience as functions of the cefotaxime concentration. At low doses of cefotaxime, the population was resistant and resilient, showing little disturbance after exposure [see (A) and (B)]. As the antibiotic concentration was increased, resistance and resilience decreased due to the increase in cell lysis causing the net growth to decrease and the time to the half-maximum density increased. Once the population underwent a crash, the resistance was minimized and resilience became the dominating factor for survival. (D) The resistance-resilience map defines a phenotypic signature. Using the same data as in (A) to (C), the resistance-resilience framework can visualize the shift in a population’s antibiotic response. When the antibiotic concentration was 0 or very low, the population’s response displayed high resistance and resilience. Once the antibiotic concentration increased to 5 μg/ml, the population’s response shifted to a position where resistance was minimized and resilience dominated the antibiotic response. With further increase in antibiotic concentration, the resistance level continued to decrease. An effective treatment should minimize both resistance and resilience. Dot colors reference the antibiotic concentration used at that point, and arrows indicate the direction of increasing antibiotic concentrations.

  • Fig. 3 Modeling reveals key determinants of resistance and resilience.

    (A) Simulated time courses of an ESBL-producing isolate with and without an antibiotic. The characteristic “crash and recovery” is generated once the antibiotic concentration is high enough. (B) Sensitivity analysis reveals determinants of resistance and resilience. Total effect indices (ST) for resistance and resilience are reported for each parameter (a = 100 μg/ml). Resistance is most affected by the lysis rate (γ). The remaining parameters did not substantially affect the system’s resistance but did affect resilience. The most influential parameters included the maximum lysis rate (γ), Bla activity (κb), the turnover rate of Bla (db), and the amount of nutrients released during lysis (ξ). (C) Modulating resistance and resilience by tuning Bla activity. We altered Bla activity in the model (left column) or experimentally added clavulanic acid (right column) in combination with a range of antibiotic concentrations. Here, a low Bla activity corresponds to κb = 0 in the model or clavulanic acid (0.005 μg/ml) in the experiment. A high Bla activity corresponds to κb = 0. 35 in the model or no clavulanic acid in the experiment. Reducing Bla activity increased the population’s sensitivity, causing both resistance and resilience to decrease at lower concentrations of antibiotic.

  • Fig. 4 Diverse phenotypic responses by different ESBL-producing isolates.

    (A) Responses to combinations of Bla inhibitor and antibiotic concentrations. In general, with increasing antibiotic concentrations, the population responses start out as both resistant and resilient (top right of each subplot). The resistance is minimized with low doses of antibiotic, giving way to a response driven by resilience. As antibiotic concentration increased, the corresponding resilience decreased. With the addition of clavulanic acid, the concentration at which the response lost its resistance and then resilience is lowered. For isolates I, II, and IV, the highest clavulanic acid concentration (0.5 μg/ml, green curve) causes the antibiotic response to shift directly from resistant and resilient to sensitive with the lowest dose of cefotaxime (0.5 μg/ml). Isolate III was not as affected by clavulanic acid, requiring the highest concentration of the Bla inhibitor in combination with a high concentration of antibiotic to prevent the population from recovering (indicated by arrow). The color of the curves indicates the concentration of clavulanic acid. The color of the dots indicates the concentration of cefotaxime, and arrowheads indicate the direction of increasing antibiotic concentrations. (B) Dependence of MIC on Bla inhibition. For each ESBL-producing isolate, the MIC corresponds to the lowest concentration of cefotaxime needed to prevent recovery in 48 hours. One isolate, isolate III, was not very susceptible to clavulanic acid, suggesting that this intervention method would not be optimal in this case. Here, different line patterns represent the isolate, and the symbol color represents the concentration of clavulanic acid, as defined in (A). The gray lines indicate that the population recovered under all tested concentrations of cefotaxime tested at that concentration of clavulanic acid; therefore, the MIC could not be calculated (isolate I, clavulanic acid < 0.005 μg/ml; isolate IV, clavulanic acid < 0.5 μg/ml).

  • Table 1 Defining antibiotic responses.

    Each row represents a different response to an antibiotic delivered at time zero. In the first row, all individual cells can withstand the antibiotic, which results in the population growing to carrying capacity, unperturbed. In the second row, only a subpopulation of cells is killed. This result manifests as an initial decline in the population density followed by full recovery, after the antibiotic is removed in the allotted time. A third scenario entails a greater sensitivity compared with the second scenario, due to greater sensitivity of individual cells or lower capability of the population in removing the antibiotic. In this case, the population only partially recovers during the allotted time after the initial decline. In the final row, all cells are killed, leading to the population extinction. Currently, antibiotic sensitivity analyses only consider whether bacteria can recover from a set dose of antibiotic in a standard period (red dot). Bacteria that display full recovery are considered resistant, partial recovery are intermediate, and no recovery are sensitive. However, temporal dynamics (blue curve) reveal differences in how a population recovers. Living cells, blue; dead cells, gray.


    Embedded Image

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/4/12/eaau1873/DC1

    Section S1. Model development

    Section S2. Sensitivity analysis

    Fig. S1. Collective antibiotic tolerance.

    Fig. S2. Varying exogenous Bla.

    Fig. S3. Time course and rate of change curves of isolate I.

    Fig. S4. Time courses and rate of change curves generated by the model.

    Fig. S5. Quantifying Bla activity in different components of culture.

    Fig. S6. Sensitivity analysis reveals parameters affecting resistance and resilience.

    Fig. S7. Time courses showing the effects of Bla inhibition.

    Fig. S8. Framework can be applied to heterogeneous populations.

    Fig. S9. Schematic for methods.

    Table S1. ESBL-producing isolates screened.

    References (4252)

  • Supplementary Materials

    This PDF file includes:

    • Section S1. Model development
    • Section S2. Sensitivity analysis
    • Fig. S1. Collective antibiotic tolerance.
    • Fig. S2. Varying exogenous Bla.
    • Fig. S3. Time course and rate of change curves of isolate I.
    • Fig. S4. Time courses and rate of change curves generated by the model.
    • Fig. S5. Quantifying Bla activity in different components of culture.
    • Fig. S6. Sensitivity analysis reveals parameters affecting resistance and resilience.
    • Fig. S7. Time courses showing the effects of Bla inhibition.
    • Fig. S8. Framework can be applied to heterogeneous populations.
    • Fig. S9. Schematic for methods.
    • Table S1. ESBL-producing isolates screened.
    • References (4252)

    Download PDF

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article