Balancing mcr-1 expression and bacterial survival is a delicate equilibrium between essential cellular defence mechanisms

Abstract

MCR-1 is a lipid A modifying enzyme that confers resistance to the antibiotic colistin. Here, we analyse the impact of MCR-1 expression on E. coli morphology, fitness, competitiveness, immune stimulation and virulence. Increased expression of mcr-1 results in decreased growth rate, cell viability, competitive ability and significant degradation in cell membrane and cytoplasmic structures, compared to expression of catalytically inactive MCR-1 (E246A) or MCR-1 soluble component. Lipopolysaccharide (LPS) extracted from mcr-1 strains induces lower production of IL-6 and TNF, when compared to control LPS. Compared to their parent strains, high-level colistin resistance mutants (HLCRMs) show reduced fitness (relative fitness is 0.41-0.78) and highly attenuated virulence in a Galleria mellonella infection model. Furthermore, HLCRMs are more susceptible to most antibiotics than their respective parent strains. Our results show that the bacterium is challenged to find a delicate equilibrium between expression of MCR-1-mediated colistin resistance and minimalizing toxicity and thus ensuring cell survival.

Fig. 1

Molecular model of colistin binding to lipid A. a Schematic of phosphoethanolamine transfer to the 1-PO₄ group of hexa-acylated lipid A as catalysed by MCR-1. b Models of colistin (blue sticks) binding to lipid A (left) or phosphoethanolamine-lipid A (right) (spheres coloured green, red, blue and orange for C, O, N and P atoms, respectively). The model is based on the NMR and docking studies of polymixin B binding to lipid A with lipid A coordinates from PDB 3fxi and colistin coordinates adapted from the NMR structure of polymixin B bound to lipid A. The positively charged Dab colistin residues closely interact with the negatively-charged 1′ and 4′ phosphate groups of lipid A, reducing the net-negative charge of lipid A. The hydrophobic leucine residues and tail of colistin A interact with the fatty acid tails of lipid A, allowing colistin A to insert into, and disrupt, the bacterial outer membrane. b (right), model of colistin binding to phosphoethanolamine-lipid A indicates addition of positively charged phosphoethanolamine onto the 1′-PO₄ of lipid A likely interferes with the interaction of positively charged Dab8 and Dab9 side chains with the phosphate group, preventing colistin binding to the outer membrane of Gram-negative bacteria. Figure created using Pymol (https://www.pymol.org/)

Fig. 2

Effects of mcr-1 overexpression on bacterial growth and fitness in vitro. a Overproducing mcr-1 causes variable effects on growth rate depending on the concentrations of l-arabinose (n = 3). b The expression levels of mcr-1 gene induced by increasing concentrations of arabinose were measured by qRT-PCR (n = 2). c Relative fitness of mcr-1 overexpressing strain mcr-1/pBAD competing control strain pHT315 under increasing concentrations of l-arabinose (0.0002% vs 0%, 0.002% vs 0%, 0.02% vs 0%, 0.2% vs 0%). Error bars represent the SD (n = 6). The differences in fitness were tested using non-parametric Mann–Whitney test, ** indicates the p values is <0.05. The average relative fitness and p values are listed in Supplementary Table 6

Fig. 3

TEM micrographs of untreated and treated E.coli. In a and b, TEM micrographs of untreated control cells (E. coli TOP10 with pBAD minus mcr-1, and E. coli TOP10 (mcr-1/pBAD) without l-arabinose induction, respectively); both cells are intact with a well-defined inner and outer membrane, and showed a highly homogeneous electron density in cytoplasm region (d and e). c TEM micrographs of mcr-1 overproducing cells; the damaging outer membrane and some completely lysed cells were observed (f)

Fig. 4

The toxic effects of mcr-1 overexpression on cell viability. a–c Confocal laser scanning microscopy images of cells treated with/without l-arabinose and stained with LIVE/DEAD^® (n = 3). Live and dead cells presented green and red colour, respectively. Scale bar is 15 µm. d Ratio of dead to live bacteria (biomass) obtained from CLSM z-stack images through COMSTAT analysis of E. coli biofilms grown for 16 h in LB broth, followed by ±l-arabinose (0.2% w/v; 8 h) treatment, where the biofilms were stained with LIVE/DEAD^® (n = 4). The COMSTAT data was assessed using one-way analysis of variance (ANOVA) followed by Tukey–Kramer multiple comparisons post hoc test. Statistical significance was set at p < 0.05

Fig. 5

The toxic effects of mcr-1 mutants on cell viability. a–c Confocal laser scanning microscopy images of cells treated with l-arabinose (0.2% w/v; 8 h) and stained with LIVE/DEAD^® (n = 4). Live and dead cells presented in green and red colour, respectively. d and e Confocal laser scanning microscopy images of control cells treated without l-arabinose for 8 h and stained with LIVE/DEAD^® (n = 4). Live and dead cells presented in green and red colour, respectively. f Ratio of dead to live bacteria (biomass) obtained from CLSM z-stack images through COMSTAT analysis (n = 4). The COMSTAT data was assessed using one-way analysis of variance (ANOVA) followed by Tukey–Kramer multiple comparisons post hoc test, '+' and '−' indicate bacteria treated with or without l-arabinose, respectively

Fig. 6

Competition assays and G. mellonella killing models in HLCRMs. a Fitness of full-length mcr-1, calalytically inactivated mcr-1(E246A), and mcr-1 soluble domain. Fitness of blaTEM-1b as a negative control. b Relative fitness of wild-type mcr-1-positive strains and their derivatives. c Relative fitness of mcr-1/pBAD-positive E.coli TOP10 strains and their derivatives. In all fitness figures, error bars represent the SD (n = 6). The differences in fitness were tested using non-parametric Mann–Whitney test, * indicates 0.01 < p value < 0.05 and ** indicates p value < 0.01. The average relative fitness and p values are listed in Supplementary Table 6. d HLCRMs displays impaired virulence in the G. mellonella infection models. All results represent means of three independent experiments with 10 larvae per treatment. Mortality bar charts were plotted using the Kaplan–Meier method (GraphPad Software). Error bars represent the SD (n = 3) and p value for strains PN16 (p = 0.0056, t = 5.427, d.f = 4), PN21 (p = 0.0029, t = 6.509, d.f = 4) and PN23 (p = 0.0003, t = 2.183, d.f = 4) were calculated by student t test. ** indicates 0.001 < p value < 0.01, *** indicates p value < 0.001

Fig. 7

Expression of cytokines in LPS-mediated THP-1 macrophage. a, b and c indicated IL-6 (n = 2) and TNF-alpha (n = 3) productions in the modified LPS (grey) or normal LPS (red)-treated cell culture, respectively. Error bars represented SD

Fig. 8

G. mellonella killing data and TEM micrographs of strain PN21. a G.mellonella mortality rate in PN21 parental strain (D0) and its mutant (D14). Error bars represent the SD (n = 3) and p value was calculated by Student’s t test (t = 6.509, d.f = 4). b and c indicated TEM micrographs of parental strain PN21(D0) and PN21(D14) mutant, respectively. Membranes for both strains are intact with a highly homogeneous electron density in cytoplasm region

Fig. 9

Kaplan–Meier plots showing the percent survival of G. mellonella over 72 h post infection with MCRPEC and non-MCRPEC human clinical strains. Survival curves were plotted using the Kaplan–Meier method (GraphPad Software). Error bars represent the SD (n = 3) and p value for a (t = 8.654, d.f = 4), b (t = 8.050, d.f = 4), c (t = 12.07, d.f = 4 for strain ff112 with pMCR-1(PN16) and t = 3.479, d.f = 4 for strain ff112 with pMCR-1(PN23)) and d (t = 3.801, d.f = 4) were calculated by Student’s t test