A Klebsiella pneumoniae antibiotic resistance mechanism that subdues host defences and promotes virulence

Abstract

Klebsiella pneumoniae is an important cause of multidrug-resistant infections worldwide. Recent studies highlight the emergence of multidrug-resistant K. pneumoniae strains which show resistance to colistin, a last-line antibiotic, arising from mutational inactivation of the mgrB regulatory gene. However, the precise molecular resistance mechanisms of mgrB-associated colistin resistance and its impact on virulence remain unclear. Here, we constructed an mgrB gene K. pneumoniae mutant and performed characterisation of its lipid A structure, polymyxin and antimicrobial peptide resistance, virulence and inflammatory responses upon infection. Our data reveal that mgrB mutation induces PhoPQ-governed lipid A remodelling which confers not only resistance to polymyxins, but also enhances K. pneumoniae virulence by decreasing antimicrobial peptide susceptibility and attenuating early host defence response activation. Overall, our findings have important implications for patient management and antimicrobial stewardship, while also stressing antibiotic resistance development is not inexorably linked with subdued bacterial fitness and virulence.

Keywords: Klebsiella pneumoniae; mgrB; antimicrobial peptides; polymyxins; virulence.

Figure EV1. Inactivation of mgrB in K. pneumoniae 52145 is not associated with an in vitro fitness cost

1. Growth kinetics of K. pneumoniae 51245 (blue and grey) and 52145‐ΔmgrB strains (red and black) cultured in LB broth (LB) and 2% glucose M9 minimal media supplemented with thiamine and MgSO₄ (M9) over 24 h at 37°C. Values are presented as the mean ± SD of three independent experiments measured in triplicate.

2. Short‐term biofilm assay results for the K. pneumoniae 52145, 52145‐ΔmgrB and 52145‐ΔmgrBCom strains. Results are displayed as per cent biofilm relative to the wild‐type strain with values presented as the mean ± SD of three independent experiments. Level of significance versus 52145 (**P = 0.007) was determined using two‐way unpaired t‐test.

Figure 1. Deletion of mgrB in K. pneumoniae invokes polymyxin resistance and multiple lipid A modifications in a PhoPQ‐dependent manner

1. A

   Minimal inhibitory concentrations to colistin of the K. pneumoniae 52145, 52145‐ΔmgrB, 52145‐ΔmgrBCom, 52145‐ΔmgrB‐ΔphoQGB, 52145‐ΔmgrB‐ΔpmrAB and 52145‐ΔmgrB‐ΔphoQGB‐phoPQCom strains. The broken line represents the European Committee on Antimicrobial Susceptibility Testing MIC breakpoint.

2. B–D

   Negative ion MALDI‐TOF mass spectrometry spectra of lipid A purified from (B) K. pneumoniae 52145, (C) 52145‐ΔmgrB and (D) 52145‐ΔmgrBCom strains. Data represent the mass‐to‐charge (m/z) ratios of each lipid A species detected and are representative of three extractions.

3. E–I

   Activity of the lpxO (E), pagP (F), pmrC (G) and pmrH (H) and phoP (I) promoters in K. pneumoniae 52145, 52145‐ΔmgrB, 52145‐ΔmgrBCom, 52145‐ΔmgrB‐ΔphoQGB carrying lucFF transcriptional fusions. Values (expressed in relative luminescence units) are presented as the mean ± SD of three independent experiments measured in triplicate. ***P < 0.0005; **P = 0.0071; *P = 0.019; versus 52145 determined using two‐way unpaired t‐test.

Figure EV2. Proposed lipid A chemical structures

Proposed lipid A structures follow previously reported structures for K. pneumoniae (Sforza et al, 2004; Clements et al, 2007; Llobet et al, 2011, 2015) and other Gram‐negative bacteria.

Figure EV3. Lipid A modifications in 52145‐ΔmgrB occur in a PhoPQ‐dependent manner

1. A–D

   Negative ion MALDI‐TOF mass spectrometry spectra of lipid A isolated from the K. pneumoniae 52145‐ΔmgrBΔphoQGB (A), 52145‐ΔmgrB‐ΔphoQGB‐ΔpmrAB (B), 52145‐ΔmgrBΔpmrAB (C) and 52145‐ΔmgrB‐ΔphoQGB‐phoPQCom (D) strains. Data represent the mass‐to‐charge (m/z) ratios of each lipid A species detected and are representative of three extractions.

Figure 2. Deletion of mgrB in K. pneumoniae increases resistance to antimicrobial peptides

1. A–D

   Per cent survival of K. pneumoniae 52145, 52145‐ΔmgrB and 52145‐ΔmgrBCom following 1‐h exposure to the following: (A) human neutrophil peptide‐1 (1.2 μM), (B) human β‐defensin‐1 (3 μM), (C) human β‐defensin‐2 (3 μM) and (D) human β‐defensin‐3 (7 μM). Values are presented as the mean ± SD of three independent experiments measured in duplicate. ***P = 0.0006; **P = 0.01–0.001; *P = 0.013; versus Kp52145 determined using two‐way unpaired t‐test.

2. E

   Resistance of K. pneumoniae 52145, 52145‐ΔmgrB and 52145‐ΔmgrBCom to antimicrobial factors produced by G. mellonella after 24‐h exposure to 10⁶ heat‐killed E. coli cells. The experiments were undertaken using a radial diffusion bioassay with data expressed as radial diffusion units (10 units = 1 mm). Values are presented as the mean ± SD of three independent experiments measured in triplicate. ***P < 0.0001; versus 52145 determined using two‐way unpaired t‐test.

Figure 3. Effect of lpxO, pmrF, pagP and pmrC mutation on the polymyxin resistance of K. pneumoniae mgrB mutant

1. A, B

   Etest® minimal inhibitory concentrations to colistin and polymyxin B of K. pneumoniae 52145, 52145‐ΔmgrB, 52145‐ΔmgrBCom compared to the double, triple and quadruple 52145‐ΔmgrB mutant strains.

2. C, D

   Per cent survival of the K. pneumoniae 52145 wild‐type and mutant strains after exposure to 20 μg/ml colistin (C) and polymyxin B (D) over 1 h. Values are presented as the mean ± SD of three independent experiments measured in duplicate. ***P < 0.0001; *P = 0.0195; versus 52145‐ΔmgrB determined using a two‐way unpaired t‐test.

Figure 4. K. pneumoniae mgrB mutant displays increased virulence in the G. mellonella waxworm infection model

1. A–C

   Kaplan–Meier plots showing the per cent survival of G. mellonella over 72 h post‐infection with 10⁵ organisms of the following: (A) K. pneumoniae 52145 (blue), 52145‐ΔmgrB (red), 52145‐ΔmgrBCom (green) and 52145‐ΔmgrB‐ΔphoQGB (grey), (B) clinical K. pneumoniae strains T1a (blue) and T1b (red), and (C) K. pneumoniae 52145 (blue), 52145‐ΔmgrB (red) and 52145‐ΔpmrC‐ΔlpxO‐ΔmgrB‐ΔpmrF‐ΔpagP quintuple‐mutant (green) strains. Forty larvae were infected in each group. Level of significance was determined using the log‐rank (Mantel–Cox) test with Bonferroni correction for multiple comparisons where applicable [α = (A) 0.0008; (B) 0.05; (C) 0.017]. P‐values presented correspond to the difference between (A) 52145‐ΔmgrB and the other 52145 strains and (B) T1a versus T1b.

2. D

   Survival curve of G. mellonella primed with 10⁶ heat‐killed E. coli cells (HKEc) or phosphate‐buffered saline (PBS) and then exposed to PBS (black, grey) or 10⁶ organisms of K. pneumoniae 52145 (blue, magenta) and 52145‐ΔmgrB (red, green) over 72 h. Ten larvae were infected in each group. Level of significance was determined using the log‐rank (Mantel–Cox) test with Bonferroni correction for multiple comparisons (α = 0.008).

Figure EV4. Capsule polysaccharide (CPS) and MgrB each contribute to K. pneumoniae virulence in G. mellonella

1. Kaplan–Meier plot showing the per cent survival of G. mellonella over 72 h post‐infection with 10⁶ organisms of K. pneumoniae 52145‐ΔmanC (blue) and 52145‐ΔmgrB‐ΔmanC (red). Forty larvae were infected in each group. Level of significance was determined using the log‐rank (Mantel–Cox) test (α = 0.05).

2. Activity of the cps promoter in K. pneumoniae 52145 and 52145‐ΔmgrB carrying the cps::lucFF transcriptional fusion. Values (expressed in relative luminescence units [540 nm]) are presented as the mean ± SD of three independent experiments measured in triplicate. Level of significance was determined using two‐way unpaired t‐test.

Figure 5. Virulence of K. pneumoniae mgrB mutant in a murine intranasal infection model and expression of lysozyme in G. mellonella upon infection

1. A–C

   Bacterial loads of K. pneumoniae 52145 and 52145‐ΔmgrB in nasal‐associated lymphoid tissue (NALT), lung and spleen homogenates of infected mice after 24 h. Six mice per group were infected with log₁₀ colony‐forming unit values presented as the mean ± SD. Data were analysed using two‐way unpaired t‐test.

2. D

   Expression of lysozyme produced by G. mellonella after 8 h of infection with K. pneumoniae 52145, 52145‐ΔmgrB, 52145‐ΔmgrBCom, 52145‐ΔmgrB‐ΔphoQGB, 52145‐ΔpmrC‐ΔlpxO‐ΔmgrB, 52145‐ΔmgrB‐ΔlpxO‐ΔpmrF, 52145‐ΔpmrC‐ΔlpxO‐ΔmgrB‐ΔpmrF, 52145‐ΔlpxO, 52145‐ΔmgrB‐ΔlpxO and 52145‐ΔmgrB‐ΔlpxO‐lpxOCom as determined by reverse transcriptase quantitative real‐time PCR. Three larvae per group were infected, and values are presented as the mean ± SD of two independent cDNA preparations measured in duplicate. ***P < 0.0001; *P = 0.035; versus 52145 determined using two‐way unpaired t‐test.

Figure EV5. Expression of insect metalloproteinase inhibitor, gallerimycin and galiomycin produced by G. mellonella during K. pneumoniae infection

1. A–C

   G. mellonella antimicrobial peptide expression determined after 8 h of infection with K. pneumoniae 52145, 52145‐ΔmgrB, 52145‐ΔmgrBCom, 52145‐ΔlpxO, 52145‐ΔmgrB‐ΔlpxO and 52145‐ΔmgrB‐ΔlpxO‐lpxOCom by reverse transcriptase quantitative real‐time PCR. Three larvae per group were infected, and values are presented as the mean ± SD of two independent cDNA preparations measured in duplicate. ***P < 0.0001; versus 52145 determined using two‐way unpaired t‐test.

Figure 6. mgrB inactivation results in downregulation of early inflammatory responses in macrophages upon infection

1. Immunoblot analysis of IκBα and tubulin levels in lysates of iBMDM cells infected with K. pneumoniae 52145, 52145‐ΔmgrB and 52145‐ΔmgrBCom for the indicated times.

2. Immunoblot analysis of phospho‐ERK (P‐ERK), phospho‐p38 (P‐p38), phospho‐JNK (P‐JNK) and tubulin levels in lysates of iBMDMs cells infected with K. pneumoniae 52145, 52145‐ΔmgrB, 52145‐ΔmgrBCom and for the indicated times.

3. Immunoblot analysis of phospho‐ERK (P‐ERK), phospho‐JNK (P‐JNK) and tubulin levels in lysates of iBMDMs cells infected with K. pneumoniae 52145, 52145‐ΔmgrB, 52145‐ΔpmrC‐ΔlpxO‐ΔmgrB‐ΔpmrF‐ΔpagP, 52145‐ΔmgrB‐ΔpmrC, 52145‐ΔmgrB‐ΔpagP, 52145‐ΔmgrB‐ΔpmrF and 52145‐ΔmgrB‐ΔlpxO for 40 min.

4. TNF‐α secretion by iBMDM macrophages stimulated for 6 h with 1 × 10⁵ UV‐killed K. pneumoniae 52145, 52145‐ΔmgrB, 52145‐ΔmgrBCom and 52145‐ΔpmrC‐ΔlpxO‐ΔmgrB‐ΔpmrF‐ΔpagP. ***P < 0.0001; *P = 0.02; versus 52145 determined using two‐way unpaired t‐test (mean ± SD).

Data information: Data are representative of at least three independent experiments.Source data are available online for this figure.