Heteroresistance at the single-cell level: adapting to antibiotic stress through a population-based strategy and growth-controlled interphenotypic coordination

Abstract

Heteroresistance refers to phenotypic heterogeneity of microbial clonal populations under antibiotic stress, and it has been thought to be an allocation of a subset of "resistant" cells for surviving in higher concentrations of antibiotic. The assumption fits the so-called bet-hedging strategy, where a bacterial population "hedges" its "bet" on different phenotypes to be selected by unpredicted environment stresses. To test this hypothesis, we constructed a heteroresistance model by introducing a blaCTX-M-14 gene (coding for a cephalosporin hydrolase) into a sensitive Escherichia coli strain. We confirmed heteroresistance in this clone and that a subset of the cells expressed more hydrolase and formed more colonies in the presence of ceftriaxone (exhibited stronger "resistance"). However, subsequent single-cell-level investigation by using a microfluidic device showed that a subset of cells with a distinguishable phenotype of slowed growth and intensified hydrolase expression emerged, and they were not positively selected but increased their proportion in the population with ascending antibiotic concentrations. Therefore, heteroresistance--the gradually decreased colony-forming capability in the presence of antibiotic--was a result of a decreased growth rate rather than of selection for resistant cells. Using a mock strain without the resistance gene, we further demonstrated the existence of two nested growth-centric feedback loops that control the expression of the hydrolase and maximize population growth in various antibiotic concentrations. In conclusion, phenotypic heterogeneity is a population-based strategy beneficial for bacterial survival and propagation through task allocation and interphenotypic collaboration, and the growth rate provides a critical control for the expression of stress-related genes and an essential mechanism in responding to environmental stresses.

Importance: Heteroresistance is essentially phenotypic heterogeneity, where a population-based strategy is thought to be at work, being assumed to be variable cell-to-cell resistance to be selected under antibiotic stress. Exact mechanisms of heteroresistance and its roles in adaptation to antibiotic stress have yet to be fully understood at the molecular and single-cell levels. In our study, we have not been able to detect any apparent subset of "resistant" cells selected by antibiotics; on the contrary, cell populations differentiate into phenotypic subsets with variable growth statuses and hydrolase expression. The growth rate appears to be sensitive to stress intensity and plays a key role in controlling hydrolase expression at both the bulk population and single-cell levels. We have shown here, for the first time, that phenotypic heterogeneity can be beneficial to a growing bacterial population through task allocation and interphenotypic collaboration other than partitioning cells into different categories of selective advantage.

FIG 1

A plasmid-bearing ESBL gene leads to heteroresistance. (a) PAPs of four clinical isolates. Four isolates were analyzed, including two E. coli (Eco) strains (carrying the CTX-M-14 and TEM-10 ESBL genes) and two K. pneumoniae (Kpn) strains (producing CTX-M-65 and SHV-2). Ceftriaxone was the antibiotic used for all strains. (b) Construction of the plasmid expressing CTX-M-14. (c) Fluorescent image of cells from overnight culture of the resistant strain R4. Growing cells show significant differences in CTX-M-14-GFP expression. The inset shows a histogram of GFP intensity for the same cell population.

FIG 2

High-intensity group (HIG) and low-intensity group (LIG) cells show differences in resistance to antibiotics. (a) Sorting gates of HIG and LIG cells. (b) Etest of LIG (left) and HIG (right) cells. (c) Copy numbers of the plasmid and CTX-M-14 transcripts in HIG and LIG cells. (d) Dynamic GFP intensity histograms of HIG and LIG cells after sorting. Cells of both subgroups were cultured in antibiotic-free medium and sampled for FCM assay every 30 min for 4 h immediately after sorting. The red dotted line indicates the peak position of the whole population under antibiotic-free conditions, and the blue dotted lines indicate peak positions of the sorted LIG and HIG subpopulations.

FIG 3

Effects of antibiotic on the GFP intensity of the R4 strain. (a) Growth curves of the R4 strain cultured in medium containing serially diluted ceftriaxone. (b) Average fluorescence intensity (left y axis; a.u., arbitrary units) and MIC measured with Etest (right y axis) at the time point of 3 h of ceftriaxone treatment as shown in panel a. (c) Histograms of GFP intensity after 3 h of ceftriaxone treatment as shown in panel a. The y axis in these histograms indicates the percentage of the maximum cell count (% Max), and the color lines indicate the ceftriaxone concentrations (μg/ml) used. Dotted lines labeled with “L” (low), “M” (moderate), or “H” (high) indicate the positions of the major peaks, which shift rightward as the ceftriaxone concentration increases. (d) Fluorescence images of cells after 2 h of ceftriaxone treatment as shown in panel a.

FIG 4

Survival rates of resistant (R4) and control sensitive (S4) strains at bactericidal ceftriaxone concentrations. (a) Correlation of death rates of R4 cells with their initial GFP intensities after 2 h of ceftriaxone treatment at the minimal bactericidal concentration (8,192 µg/ml). Cells were binned according to their initial GFP intensities. (b) Correlation of death rates of R4 cells with their initial growth rates after the same treatment as for panel a; cells were binned according to their initial growth rates. (c and d) Graphs corresponding to panels a and b for S4 treated at its minimal bactericidal concentration of ceftriaxone (128 ng/ml). Death count (black squares) and survival rate (blue squares) were plotted together.

FIG 5

Single-cell observation of phenotype transition under antibiotic stress. (a) Box charts showing average GFP intensities and growth rates of the R4 strain at the end of 2-h treatment in four escalating ceftriaxone concentrations. For each antibiotic dosage, 50 cells were randomly selected and observed. (b) The distribution of cellular GFP intensity versus the growth rate (frequency in color scale) of R4 cells after the same antibiotic treatment as shown in panel a. (c and d) Corresponding graphs for the S4 strain. (e) Dynamics of growth rate (GR) and GFP intensity (I) of two daughter cells that have departed from a single mother cell. The daughter cell C1 shows an exponentially increasing GFP intensity and reduced growth rate, whereas the daughter cell C2 exhibits a relatively lower GFP intensity but a higher growth rate. (f) Scheme of growth-centric feedback loops that control the expression of hydrolase in a stringent way.