Aminoglycoside heteroresistance in Enterobacter cloacae is driven by the cell envelope stress response

Abstract

Enterobacter cloacae is a Gram-negative nosocomial pathogen of the ESKAPE (Enterococcus, Staphylococcus, Klebsiella, Acinetobacter, Pseudomonas, and Enterobacter spp.) priority group with increasing multi-drug resistance via the acquisition of resistance plasmids. However, E. cloacae can also display forms of antibiotic refractoriness, such as heteroresistance and tolerance. Here, we report that E. cloacae displays transient heteroresistance to aminoglycosides, which is accompanied with the formation of small colony variants (SCVs) with increased minimum inhibitor concentration (MIC) of gentamicin and other aminoglycosides used in the clinic, but not other antibiotic classes. To explore the underlying mechanisms, we performed RNA sequencing of heteroresistant bacteria, which revealed global gene expression changes and a signature of the CpxRA cell envelope stress response. Deletion of the cpxRA two-component system abrogated aminoglycoside heteroresistance and SCV formation, pointing to its indispensable role in these processes. The introduction of a constitutively active allele of cpxA led to high aminoglycoside MICs, consistent with cell envelope stress response driving these behaviors in E. cloacae. Cell envelope stress can be caused by environmental cues, including heavy metals. Indeed, bacterial exposure to copper increased gentamicin MIC in the wild-type but not in the ΔcpxRA mutant. Moreover, copper exposure also elevated the gentamicin MICs of clinical isolates from bloodstream infections, suggesting that CpxRA- and copper-dependent aminoglycoside resistance is broadly conserved in E. cloacae strains. Altogether, we establish that E. cloacae relies on transcriptional reprogramming via the envelope stress response pathway for transient resistance to a major class of frontline antibiotic.IMPORTANCEEnterobacter cloacae is a bacterium that belongs to the WHO high-priority group and an increasing threat worldwide due its multi-drug resistance. E. cloacae can also display heteroresistance, which has been linked to treatment failure. We report that E. cloacae shows heteroresistance to aminoglycoside antibiotics. These are important frontline microbicidal drugs used against Gram-negative bacterial infections; therefore, understanding how resistance develops among sensitive strains is important. We show that aminoglycoside resistance is driven by the activation of the cell envelope stress response and transcriptional reprogramming via the CpxRA two-component system. Furthermore, heterologous activation of envelope stress via copper, typically a heavy metal with antimicrobial actions, also increased aminoglycoside MICs of the E. cloacae type strain and clinical strains isolated from bloodstream infections. Our study suggests aminoglycoside recalcitrance in E. cloacae could be broadly conserved and cautions against the undesirable effects of copper.

Keywords: AMR; ESKAPE; aminoglycosides; cell envelope stress; heteroresistance; persistence.

Fig 1

E. cloacae cultures contain a subpopulation that is resistant to gentamicin (Gm). (A and B) Representative images from disk-diffusion assays (DDA) testing gentamicin sensitivity of E. coli ATCC 11775 and E. cloacae ATCC 13047 (A). Inset shows E. cloacae “halo” inside the zone of clearance, whose diameters are shown in (B). Mean ± SD of independent experiments indicated; cultures were sampled at 20 h (stationary phase; n = 4) or 1.5 h (exponential phase; n = 3) for spreading on plates for disk-diffusion assays. Two-tailed P-values for comparison of outer and inner diameters from Student’s t-tests. (C and D) Time-to-kill curves for gentamicin (20 mg.L⁻¹) (C) and ciprofloxacin (0.5 mg.L⁻¹) (D) showing the percentage survival of E. cloacae over time. Cultures were treated with the antibiotic, samples were collected, washed, and plated on LB agar to quantify viable colony-forming units (CFU).mL⁻¹. Graphs show first-order decay curve fit to data, and rapid (Phase 1) and slow (Phase 2) killing phases are marked. No viable CFUs detected after 5 h post-ciprofloxacin treatment (< 300 CFU.mL⁻¹). Number of independent experiments as follows: C, n = 11; D, n = 4. (E) MIC measured by broth microdilution for E. cloacae grown in LB without or with Gm (20 mg.L⁻¹) for 24 h as in experiments in C. Data from n = 7 (LB) and n = 6 (Gm) independent experiments. Two-tailed P-value from a Mann-Whitney U-test. (F) Population analysis profiling (PAP) assays of E. cloacae cultures in stationary or exponential phase or following gentamicin treatment (20 mg.L⁻¹) for 24 h as labeled. Bacteria were plated on Mueller-Hinton agar plates with the indicated concentrations of gentamicin. Percentage survival relative to plates without antibiotic is plotted. Data from n = 4 independent experiments shown. In C–F, the box is the interquartile range (IQR), the horizontal line is the median, and the whiskers depict 1.5xIQR.

Fig 2

Aminoglycoside exposure of E. cloacae ATCC 13047 triggers SCVs with increased MICs. (A) Schematic depiction of the appearance of SCVs after exposure of E. cloacae to gentamicin (Gm) (20 mg.L⁻¹), and their reversion back to “typical” colony morphotypes after growth in broth without gentamicin. Arrows used to point to colony morphotypes indicated in the legend. The images (N1, N2, N3) show n = 3 independent SCVs grown in LB for 24 h and plated on LB agar. (B) Percentage of SCVs after exposure to the indicated aminoglycoside for the indicated times. E. cloacae cultures in exponential phase were treated with gentamicin (20 mg.L⁻¹), tobramycin (Tob) (20 mg.L⁻¹), or amikacin (Ami) (80 mg.L⁻¹) for the times as indicated. Data from n = 3 independent experiments. (C) Survival of E. cloacae after exposure to amikacin (80 mg.L⁻¹) or tobramycin (20 mg.L⁻¹) at 5 or 24 h as indicated. Data from n = 3 independent experiments. ns, not significant for comparison of CFU at the two time points by mixed-effects analysis of variance. (P > 0.05). (D) MICs measured by broth microdilution for non-SCVs and SCVs against Gm, Ami, kanamycin (Kan), Tob, ceftriaxone (Cef), ciprofloxacin (Cip), colistin (Col), and tetracycline (Tet). Mean (colored square) and SD shown from n = 8 (non-SCV) or n = 10 (SCV) independent experiments; open circles represent all data points. False discovery rate-adjusted two-tailed P-values for comparisons between SCV and non-SCV for each antibiotic from Mann-Whitney U-tests; ns, not significant (P > 0.05). In B and C, the box is the interquartile range (IQR), the horizontal line is the median, and the whiskers depict 1.5xIQR.

Fig 3

Transcriptional reprogramming drives heteroresistance to gentamicin. (A) Gentamicin time-to-kill curves after exposure to gentamicin (8 mg.L⁻¹) for parental E. cloacae (E.clo) or SCV as indicated. In the graph, the box is the interquartile range (IQR), the horizontal line is the median, and the whiskers depict 1.5xIQR from n = 3 independent experiments. (B and C) Principal component analyses (B) and hierarchical clustering (C) of RNA sequencing data from the indicated growth conditions. Each dot in (B) represents an independent biological sample; n = 3 experiments. Color key in (C) indicates raw Z-scores of normalized log₂ counts per million for differentially expressed genes in the indicated conditions. Gm, gentamicin (8 mg.L⁻¹). (D) Venn diagram showing the distribution of up- and downregulated genes in the indicated conditions following RNA sequencing analyses. (E–H) Gene ontology (GO) terms enriched in genes up- or downregulated in the indicated comparisons. SCV, SCV vs parental; Genta, SCV given gentamicin (8 mg.L⁻¹) for 2 h vs SCV without gentamicin. The union size is the number of target genes in the data set. False discovery rate (FDR)-adjusted P-values for GO terms plotted on -log₁₀ scale. (I) Plot showing the statistically enriched transcription factors among differentially expressed genes as indicated. The union size is the number of target genes in the data set. Hits with FDR-adjusted P < 0.05 (Adj P-value) are plotted on a -log₁₀ scale.

Fig 4

Differentially regulated genes predicted to be CpxRA targets. Differentially expressed genes in SCVs versus parental colonies and gentamicin-treated versus untreated SCV conditions as indicated. Log₂ fold-change [log₂(FC)] and false discovery rate-adjusted P-values are listed.

Fig 5

The CpxRA cell envelope stress response pathway is required for gentamicin (Gm) resistance. (A) Growth curves of the indicated E. cloacae strains plotted as mean ± SD from n = 3 experiments. (B) Representative images from gentamicin disk-diffusion assays showing cpxRA-dependent “halo” within the zone of clearance from n = 4 independent experiments. (C) PAP assay showing proportion of surviving bacteria of the indicated E. cloacae genotypes on agar plates with 8 mg.L⁻¹ gentamicin. Data from n = 3 independent experiments. Two-tailed P-values for the comparisons with wild-type (WT) E. cloacae from mixed-effects analysis of variance (ANOVA). (D) Gentamicin time-to-kill curves for the indicated complemented and ΔcpxRA knockout E. cloacae showing the percentage survival over time after exposure to gentamicin (20 mg.L⁻¹). Graph shows first-order decay curve fit to data from n = 3 independent experiments. (E) Growth curves of the indicated E. cloacae WT and cpxA24 strains in LB plotted as mean ± SD from n = 3 experiments. (F) Quantitative Reverse Transcription Polymerase Chain Reaction (qRT-PCR) of CpxRA-target genes degP and yebE (relative to rpoD) in the indicated strains. Data from n = 4 independent experiments. (G) Representative images showing small colony morphology of the WT and cpxA24 strain, plated on LB agar without antibiotics. (H) MICs measured by broth microdilution for the indicated strains against Gm, amikacin (Ami), kanamycin (Kan), tobramycin (Tob), and tetracycline (Tet). Mean (colored square) and SD shown from n = 12 independent experiments. Open circles represent all data points. False discovery rate-adjusted two-tailed P-values of comparisons between WT and cpxA24 are shown from non-parametric two-way ANOVA following aligned-rank transformation. In C, D, F, and H, the box is the interquartile range (IQR), the horizontal line is the median, and the whiskers depict 1.5xIQR.

Fig 6

Exposure to copper increases gentamicin MIC via cpxRA. (A) Gentamicin MIC measured by broth microdilution for the indicated complemented or ΔcpxRA knockout E. cloacae. MICs were measured in the absence or presence of 4 mM copper sulfate. Data from n = 4 independent experiments. False discovery rate (FDR)-adjusted two-tailed P-value for comparison between the MIC of the two genotypes in the presence of Cu²⁺ from non-parametric analysis of variance (ANOVA) following aligned-rank transformation. (B) Cu²⁺ MIC measured by broth microdilution for the indicated complemented or ΔcpxRA knockout E. cloacae. Data from n = 4 independent experiments. ns, not significant, i.e., two-tailed P-value >0.05 for the indicated comparisons from Mann-Whitney U-test. (C) qRT-PCRs showing the fold-change in expression for the Cpx-target genes degP and yebE in the indicated strains treated with or without Cu⁺² (4 mM) for 3 h, relative to the housekeeping reference rho. Data from n = 4 independent experiments. Two-tailed P-values from mixed-effects ANOVAs. (D) Gentamicin MICs for the indicated clinical E. cloacae isolates in the absence (“Control”) or presence of Cu²⁺ (4 mM copper sulfate). Data from n = 4 independent experiments. Gray line indicates EUCAST breakpoint. FDR-adjusted two-tailed P-values are shown for indicated comparisons in the presence of Cu²⁺ for each strain from non-parametric ANOVA following aligned-rank transformation. (E) qRT-PCR showing the expression of Cpx-target genes degP and yebE (relative to rho) in the indicated bloodstream isolates grown without and with Cu⁺² (4 mM) for 3 h. Data from n = 4 independent experiments. Two-tailed P-values from mixed-effects ANOVAs. In A–E, the box is the interquartile range (IQR), the horizontal line is the median, and the whiskers depict 1.5xIQR.