Collective Resistance in Microbial Communities by Intracellular Antibiotic Deactivation

Abstract

The structure and composition of bacterial communities can compromise antibiotic efficacy. For example, the secretion of β-lactamase by individual bacteria provides passive resistance for all residents within a polymicrobial environment. Here, we uncover that collective resistance can also develop via intracellular antibiotic deactivation. Real-time luminescence measurements and single-cell analysis demonstrate that the opportunistic human pathogen Streptococcus pneumoniae grows in medium supplemented with chloramphenicol (Cm) when resistant bacteria expressing Cm acetyltransferase (CAT) are present. We show that CAT processes Cm intracellularly but not extracellularly. In a mouse pneumonia model, more susceptible pneumococci survive Cm treatment when coinfected with a CAT-expressing strain. Mathematical modeling predicts that stable coexistence is only possible when antibiotic resistance comes at a fitness cost. Strikingly, CAT-expressing pneumococci in mouse lungs were outcompeted by susceptible cells even during Cm treatment. Our results highlight the importance of the microbial context during infectious disease as a potential complicating factor to antibiotic therapy.

Fig 1. Experimental setup to determine passive resistance.

Antibiotic-susceptible cells (Ab^S) constitutively expressing luc are grown together with antibiotic-resistant cells (Ab^R, which do not express luc). Only when the concentration of the antibiotic in the medium is reduced by enzymatic deactivation of resistant cells will the genetically antibiotic-susceptible cells be able to grow and produce light.

Fig 2. Cm deactivation during mixed population assays.

(a) Plate reader assay sets in quadruplicate (average and standard error of the mean [s.e.m.]) measuring luminescence (symbols with color outline) and cell density (corresponding grey symbols) of S. pneumoniae Cm^S growing in the presence of 3 μg ml⁻¹ Cm, in presence (+) or absence (−) of Cm^R cells. (b) Development of the count of viable Cm^S cells (colony-forming units per ml [CFUs ml⁻¹]) during the cultivation assay presented in a, determined via plating in the presence of kanamycin; average values of duplicates are shown. (c) Culture supernatant (S) samples after 0, 1, 2, and 4 h of Cm^R cultivation (inoculation at optical density OD 0.001) in the presence of 5 μg ml⁻¹ Cm, analyzed for Cm content by high-performance liquid chromatography (HPLC) separation and ultraviolet (UV) detection at 278 nm. (d) Luminescence and cell density profiles of Cm^S cells treated with 3 μg ml⁻¹ Cm (inoculation at OD 0.001) in dependency of the inoculum size of Cm^R cells. (e, f), Cm^S luminescence and growth analysis (e) in Cm-supplemented medium (3 μg ml⁻¹) that was pretreated with Cm^R cell pellet (P), S, and culture lysate (L), and controls without (C⁻) and with Cm (C⁺); (f) schematic overview of the assay (see also Methods and S1 Data).

Fig 3. Interspecies collective resistance.

Still images (overlay of phase contrast and fluorescence microscopy) of a time-lapse experiment of S. pneumoniae Cm^S, cocultivated with a strain of the pneumococcal niche competitor S. aureus (strain LAC pCM29) that expresses CAT and GFP, growing on a semi-solid surface supplemented with 3 μg ml⁻¹ Cm. Scale bar 10 μm.

Fig 4. Population dynamics of bacterial communities.

(a) Simulated growth trajectories for Cm^R and Cm^S populations subject to antibiotic stress and resource competition. (b) Dynamic of intracellular Cm (y_r and y_s) and growth-limiting resource (z). Simulation time is scaled relative to the mean residence time of cells in a chemostat, which is equal to the generation time at steady state. At low population densities, the Cm^R strain can grow, whereas Cm^S cannot, due to a high concentration of Cm. However, the invasion of Cm^R lowers antibiotic stress, generating permissive conditions for the growth of Cm^S cells. The chemostat is then rapidly colonized by both strains (shortly after t = 180) until the resource becomes limiting. From that moment onwards, total cell density changes little, while the relative frequencies of the two strains continue to shift. Eventually, a stable equilibrium is reached, at which the cost and benefit of CAT expression (i.e., reduced growth rate efficiency for Cm^R cells versus their lower intracellular Cm concentration) balance out. Inset (c), The dark-red dot pinpoints the parameter set used in the simulation shown in a and b: r = 20.0, η = 0.9, k_z = 4.0, c = 1.0, p = 50.0, h_Y = 0.25/Y₀, k_Y = 2.5/Y₀, d = 30.0/Y₀ and Y₀ = 0.8. These parameters were selected to lie in a restricted area of parameter space (highlighted in red) where stable coexistence between Cm^S and Cm^R cells is observed Alternative model outcomes, which were identified by a numerical bifurcation analysis (see S1 Text and S4 Fig), include establishment of Cm^S only (area S), establishment of Cm^R only (area R), no bacterial growth (area N), and competition-induced extinction (area E, where Cm^S bacteria first outcompete Cm^R bacteria and subsequently are cleared by the antibiotic; see S5 Fig).

Fig 5. Cross-protection in a mouse pneumonia model.

(a) Eight-wk-old female CD1 mice were infected intratracheally with Cm^S pneumococci or an equivalent amount of Cm^S + Cm^R pneumococci in a one-to-one ratio. One h post infection, mice were treated with one intraperitoneal injection of Cm 75 mg kg⁻¹ followed by two additional doses spaced 5 h apart. Control mice received an injection of the vehicle alone. n = 14 for Cm^S control; 13 for Cm^S Cm-treated; 13 for Cm^S + Cm^R control; and 14 for Cm^S + Cm^R Cm-treated. Data plotted as average and s.e.m. of two independent experiments combined. Dashed line ‘inoc’ denotes the initial inoculum. *p < 0.05; one-way ANOVA with Tukey's multiple comparison post-test. (b) Bacterial colonies recovered from the Cm^S + Cm^R control and Cm^S + Cm^R Cm-treated mice were individually picked and used to inoculate THY media in 96-well plates. These 96-well plates were then used to inoculate 96-well plates with THY media containing either 15 μg ml⁻¹ Cm or 100 μg ml⁻¹ kanamycin to determine whether or not the original bacterial colony was Cm^S or Cm^R. n = 9 for Cm^S + Cm^R control and 14 for Cm^S + Cm^R Cm-treated. Data plotted as average and s.e.m. of two independent experiments combined. *p = 0.04; Mann–Whitney U test (see S1 Data).