Lipidomic Analysis of the Outer Membrane Vesicles from Paired Polymyxin-Susceptible and -Resistant Klebsiella pneumoniae Clinical Isolates

Gram-negative bacteria produce outer membrane vesicles (OMVs) as delivery vehicles for nefarious bacterial cargo such as virulence factors, which are antibiotic resistance determinants. This study aimed to investigate the impact of polymyxin B treatment on the OMV lipidome from paired polymyxin-susceptible and -resistant Klebsiella pneumoniae isolates. K. pneumoniae ATCC 700721 was employed as a reference strain in addition to two clinical strains, K. pneumoniae FADDI-KP069 and K. pneumoniae BM3. Polymyxin B treatment of the polymyxin-susceptible strains resulted in a marked reduction in the glycerophospholipid, fatty acid, lysoglycerophosphate and sphingolipid content of their OMVs. Conversely, the polymyxin-resistant strains expressed OMVs richer in all of these lipid species, both intrinsically and increasingly under polymyxin treatment. The average diameter of the OMVs derived from the K. pneumoniae ATCC 700721 polymyxin-susceptible isolate, measured by dynamic light scattering measurements, was ~90.6 nm, whereas the average diameter of the OMVs isolated from the paired polymyxin-resistant isolate was ~141 nm. Polymyxin B treatment (2 mg/L) of the K. pneumoniae ATCC 700721 cells resulted in the production of OMVs with a larger average particle size in both the susceptible (average diameter ~124 nm) and resistant (average diameter ~154 nm) strains. In light of the above, we hypothesize that outer membrane remodelling associated with polymyxin resistance in K. pneumoniae may involve fortifying the membrane structure with increased glycerophospholipids, fatty acids, lysoglycerophosphates and sphingolipids. Putatively, these changes serve to make the outer membrane and OMVs more impervious to polymyxin attack.

Principle component analysis (PCA) score plots and the heat map revealed significant global lipidomic differences between the OMVs of the polymyxin-susceptible and -resistant K. pneumoniae strains (Figures 1 and 2). Notably, following treatment with a clinically relevant concentration of polymyxin B (2 mg/L) we observed marked global lipidome perturbations in the OMVs of the polymyxin-susceptible K. pneumoniae strains; whereas the OMVs of the resistant strains showed moderate global lipidome perturbations in response to polymyxin B treatment of the cells. For univariate analyses, all of the putatively identified lipids were further analysed to reveal those showing at least 2-fold differences (p < 0.05, FDR < 0.05, one-way ANOVA test) in relative abundance ( Figure 3). The cluster algorithm and fold-change analysis highlighted that, compared to the untreated controls, polymyxin B treatment (2 mg/L) of the polymyxin-susceptible K. pneumoniae ATCC 700721 significantly reduced the phosphatidylcholine, phosphatidylethanolamine and 1-acyl-glycerophosphocholine content of its OMVs. Additionally, the sphingolipids namely, sphingosine, N-acyl-sphingosine (ceramide), N-acyl-sphinganine(dihydro-ceramide), sphingomyelin, glucosyl-ceramide and lactosyl-ceramide were significantly reduced following polymyxin B treatment ( Figure 3Ai). Moreover, certain saturated fatty acids (e.g., hexadecanoic acid and octadecanoic acid), and polyunsaturated fatty acids (α-linolenic acid and arachidonic acid) were also reduced in the OMVs of the polymyxin B treated susceptible isolate. Polymyxin B treatment of the its paired polymyxin-resistant K. pneumoniae ATCC 700721 laboratory isolate significantly increased the content of lysoglycerophosphates, phosphatidylcholines and phosphatidylethanolamines in its OMVs (Figure 3Aii). Similarly, to the polymyxin-susceptible ATCC 700721 isolate, most of the glycerophospholipid and fatty acid content of the OMVs isolated from the polymyxin B treated polymyxin-susceptible clinical isolates (K. pneumoniae BM3 and FADDI-KP069) were significantly reduced compared to untreated controls (Figure 3Bi,Ci). In particular, glycerophospholipids (e.g., phosphatidylethanolamines, phosphatidylcholines, lysophosphatidylcholines, and lysoglycerophosphates) were remarkably reduced in response to polymyxin B treatment. In addition, fatty acids (e.g., docosanoic acid, octadecenoic acid and hexadecanoic acid); and sphingolipids (mainly dihydro-ceramides) were also significantly reduced in response to polymyxin B be treatment. In contrast, the majority of glycerophospholipids, fatty acids and sphingolipids content of OMVs isolated from their paired polymyxin-resistant K. pneumoniae BM3 a FADDI-KP069 isolates were significantly increased in response to polymyxin B treatment (Figure 3Bii,Cii). Notably, all of the polymyxin B-resistant strains secreted OMVs significantly are richer in glycerophospholipids, fatty acids, lysoglycerophosphates and sphingolipids compared to polymyxin B-susceptible isolates even when grown in the absence of polymyxin B (Figure 4). Glycerophospholipids, fatty acids, glycerolipids and sphingolipids play a crucial role in maintain outer membrane integrity, bacterial survival and pathogenesis [40]. Phospholipids (including glycerophospholipids) are essential components of bacterial membranes and they are responsible for maintaining membrane integrity and the selective permeability of the outer membrane [41]; they contribute to cationic peptide resistance, protect bacteria from osmotic stress and regulate flagellum-mediated motility [42]. In addition, sphingolipids are involved in maintaining normal bacterial growth and membrane integrity; and trigger bacterial pathogenesis via induction of the host immune system [43].

Transmission Electron Microscopy Imaging and Dynamic Light Scattering Size Estimation of K. pneumoniae OMVs
Dynamic light-scattering (DLS) analysis revealed that the average hydrodynamic radius of the OMVs derived from the K. pneumoniae ATCC 700721 polymyxin-susceptible isolate is ~90.6 nm; the profile was symmetrical and the OMV scatter ranged from ~30-500 nm ( Figure 5A). The average hydrodynamic radius of the OMVs isolated from the paired K. pneumoniae ATCC 700721 polymyxinresistant isolate was ~141 nm and the OMV scatter ranged from ~30 to 1000 nm ( Figure 5C), which indicates that the resistant isolae sheds larger OMVs than the susceptible one. Polymyxin B treatment (2 mg/L) of the K. pneumoniae cells resulted in the production of OMVs with slightly larger average particle size in both the susceptible (average diameter ~124 nm, OMV scatter ~30-900 nm; Figure 5B) and resistant (average hydrodynamic radius ~154 nm, OMV scatter ~30-1500 nm; Figure 5D) strains. Notably, the OMV scatter profile in the resistant strain is asymmetrical, with and without polymyxin B treatment. In line with the DLS data [44,45], transmission electron microscopy imaging of K. pneumoniae OMVs revealed a similar size distribution wherein the polymyxin-resistant K. pneumoniae ATCC 700721 strain produced larger OMVs than the susceptible strain ( Figure 6). Moreover, the OMVs isolated from the polymyxin-resistant isolate stained darker with the TEM contrast reagent uranyl acetate, which enhances the contrast by interaction with lipids; in line with the lipidomics findings, this would suggest that the OMVs of the resistant strains contain more lipids. Similarly, in Salmonella enterica, LPS remodelling in the outer membrane in response to polymyxins or other environmental PhoP/Q-PmrA/B activating conditions, has been shown to stimulate the biogenesis of larger-diameter OMVs [36][37][38][39].

Transmission Electron Microscopy Imaging and Dynamic Light Scattering Size Estimation of K. pneumoniae OMVs
Dynamic light-scattering (DLS) analysis revealed that the average hydrodynamic radius of the OMVs derived from the K. pneumoniae ATCC 700721 polymyxin-susceptible isolate is~90.6 nm; the profile was symmetrical and the OMV scatter ranged from~30-500 nm ( Figure 5A). The average hydrodynamic radius of the OMVs isolated from the paired K. pneumoniae ATCC 700721 polymyxin-resistant isolate was~141 nm and the OMV scatter ranged from~30 to 1000 nm ( Figure 5C), which indicates that the resistant isolae sheds larger OMVs than the susceptible one. Polymyxin B treatment (2 mg/L) of the K. pneumoniae cells resulted in the production of OMVs with slightly larger average particle size in both the susceptible (average diameter~124 nm, OMV scatter~30-900 nm; Figure 5B) and resistant (average hydrodynamic radius~154 nm, OMV scatter~30-1500 nm; Figure 5D) strains. Notably, the OMV scatter profile in the resistant strain is asymmetrical, with and without polymyxin B treatment. In line with the DLS data [44,45], transmission electron microscopy imaging of K. pneumoniae OMVs revealed a similar size distribution wherein the polymyxin-resistant K. pneumoniae ATCC 700721 strain produced larger OMVs than the susceptible strain ( Figure 6). Moreover, the OMVs isolated from the polymyxin-resistant isolate stained darker with the TEM contrast reagent uranyl acetate, which enhances the contrast by interaction with lipids; in line with the lipidomics findings, this would suggest that the OMVs of the resistant strains contain more lipids. Similarly, in Salmonella enterica, LPS remodelling in the outer membrane in response to polymyxins or other environmental PhoP/Q-PmrA/B activating conditions, has been shown to stimulate the biogenesis of larger-diameter OMVs [36][37][38][39].

Materials
Polymyxin B was supplied by Betapharma (Shanghai, China). All chemicals were purchased from Sigma-Aldrich (Melbourne, VIC, Australia) at the highest research grade; ultrapure water was from Fluka (Castle Hill, New South Wales, Australia). Stock solutions of polymyxin B (10 mg/L) were freshly prepared in ultrapure water and filtered through 0.22 µm syringe filters (Sartorius, Melbourne, Victoria, Australia).

Bacterial Isolates and Growth Conditions
All bacterial strains used in this study are described in Table S1. Resistance to polymyxin B was defined as MICs of ≥8 mg/L [46]. A total of six different K. pneumoniae isolates were studied: The clinical isolates K.  Table S1. All bacteria were stored at −80 • C in tryptone soya broth (TSB, Oxoid, Melbourne, Australia). Prior to experiments, parent strains were subcultured onto nutrient agar plates (Medium Preparation Unit, University of Melbourne, Victoria, Australia). Overnight broth cultures were subsequently grown in 5 mL of cation-adjusted Mueller-Hinton broth (CaMHB, Oxoid, West Heidelberg, Victoria, Australia), from which a 1 in 100 dilution was performed in fresh broth to prepare mid-logarithmic cultures according to the optical density at 500 nm (OD 500nm = 0.4 to 0.6). All broth cultures were incubated at 37 • C in a shaking water bath (180 rpm).

Minimum Inhibitory Concentration (MIC) Microbiological Assay
MICs were performed according to the Clinical and Laboratory Standards Institute (CLSI) guidelines [47]. MICs were determined for all isolates in three replicates on separate days using broth microdilution method in cation-adjusted Mueller-Hinton broth (CAMHB) in 96-well polypropylene microtitre plates. Wells were inoculated with 100 µL of bacterial suspension prepared in CaMHB (containing 10 6 colony-forming units (cfu) per mL) and 100 µL of CaMHB containing increasing concentrations of polymyxin B (0.25-256 mg/L). The MICs were defined as the lowest concentration at which visible growth was inhibited following 18 h incubation at 37 • C. Cell viability was determined by sampling wells at polymyxin B concentrations greater than the MIC. These samples were diluted in normal saline and spread plated onto nutrient agar. After incubation at 37 • C for 20 h, viable colonies were counted on these plates. The limit of detection was 10 cfu/mL.

Isolation of Outer Membrane Vesicles (OMVs)
Mid-logarithmic cultures (6 L) of each isolate were grown at 37 • C with shaking (1800 rpm) and cell-free supernatants were collected through centrifugation (15 min at 10,000× g, 4 • C). Where indicated, polymyxin B was added to the culture volume at a final concentration of 2 mg/L. The OMV containing supernatants were filtered through 0.22-µm membrane (Sigma-Aldrich) to remove any remaining cell debris, then concentrated through a tangential filtration concentrator unit (Pall Life Science, Ann Arbor, MI, USA) and collected using 100 kDa Pellicon filtration cassettes (Millipore, Melbourne, Australia). Also, a portion of the supernatant was plated for growth on agar plates overnight at 37 • C to make sure that the supernatant is free of bacterial cells. OMVs in the cell-free supernatants were then pelleted down by ultracentrifugation at 150,000× g for 2 h at 4 • C in a Beckman Ultracentrifuge (SW28 rotor). Purified OMVs were concentrated re-suspended in 1 mL sterile PBS and the concentration was determined by Bio-Rad (Gladesville, NSW, Australia) protein assay.

Lipidomics Analysis
OMV lipids were extracted with the single-phase Bligh-Dyer method (CHCl 3 /MeOH/H 2 O, 1:3:1, v/v) [48]. For further analysis, samples were reconstituted in 100 µL of CHCl 3 and 200 µL of MeOH, centrifuged at 14,000× g for 10 min at 4 • C to obtain particle-free supernatants. LC-MS for lipidomic analysis was conducted on a Dionex U3000 high-performance liquid chromatography system (HPLC) in tandem with a Q-Exactive Orbitrap mass spectrometer (Thermo Fisher, Melbourne, Australia) in both positive and negative mode with a resolution at 35,000. The mass scanning range was from 167 to 2000 m/z. The electrospray voltage was set as 3.50 kV and nitrogen was used as collision gas. The Ascentis Express C 8 column (5 cm × 2.1 mm, 2.7 µm, Sigma-Aldrich, 53831-U) was maintained at 40 • C, and the samples were controlled at 4 • C. The flow rate was 0.2 mL/min at first 24 min, but increased to 0.5 mL/min from 25 min to 30 min. The multi-step gradient started from 100% to 80% mobile phase A over the first 1.5 min, then to 72% mobile phase A at 7 min, over the next 1 min, the gradient changed to 65% mobile phase A, from 8 min to 24 min, the gradient reached a final composition of 35% mobile phase A and 65% mobile phase B. This was followed by a washing step from 65% to 100% mobile phase B over the next 1 min, and maintained for 2 min. A 2-min re-equilibration of the column with 100% A was performed between injections. Untargeted lipidomic analyses were performed through mzMatch [49]; and IDEOM [50] (http://mzmatch.sourceforge.net/ideom.php). Raw LC-MS data files were converted to mzXML format through a proteowizard tool, Msconvert. Automated chromatography peaks were picked by XCMS [51], and then converted to peakML files, which were combined and filtered by mzMatch based on the intensity (1000), reproducibility (RSD for all replicates < 0.8), and peak shape (codadw > 0.8). The mzMatch program was used for retrieving intensities for missing peaks and the annotation of related peaks. Unmatched peaks and noises were rejected through IDEOM. The database used in IDEOM included KEGG, MetaCyc and Lipidmaps [52]. Univariate statistics analysis was performed using a Welch's T-test (p < 0.05), while multivariate analysis was conducted using the metabolomics R package.

Transmission Electron Microscopy (TEM)
Carbon-coated Formvar copper grids were placed on a drop of OMV suspension (1 mg/mL protein) for 5 min then washed three times with PBS and fixed in 1% glutaraldehyde for 4 min. Grids were then washed three times with PBS, two times with Milli-Q water and stained for 20 s with 4% uranyl acetate. Grids were finally washed with Milli-Q water and incubated on ice for 10 min in methyl-cellulose with 4% uranyl acetate (9:1). Grids were then air-dried and viewed with a Tecnai Spirit (T12) transmission electron microscope, and the images were acquired using TIA software (FEI, Melbourne, Australia).

Dynamic Light Scattering
The particle size of the OMVs was measured using dynamic light scattering (DLS). OMVs were diluted with PBS to a protein concentration of 0.05 mg/L and the scatter was recorded using a Zetasizer NanoS (Malvern, PA, USA) at 173 • with a laser of wavelength 632 nm. Data were analysed with Zetasizer Software (V7.11, Malvern, UK) to obtain the average hydrodynamic radius.

Conclusions
In this study, we show that polymyxin B treatment of the susceptible K. pneumoniae strains significantly reduced the glycerophospholipid, fatty acid, lysoglycerophosphate and sphingolipid content of their OMVs, compare to the untreated control. On the other hand, in the OMVs of their paired polymyxin-resistant strains these lipids were increased both intrinsically and in response to polymyxin B treatment. In view of these findings, it is reasonable to hypothesize that the outer membrane remodelling associated with polymyxin-resistance in K. pneumoniae entails fortifying the membrane with increased glycerophospholipids, fatty acids, lysoglycerophosphates and sphingolipids, which are lipids to which polymyxins cannot avidly bind. It is important to mention that polymyxins primarily target the lipid A in the Gram-negative outer membrane-hence their narrow spectrum of activity against Gram-negative bacteria that do not express LPS. These outer membrane changes may be accompanied by the modification of the lipid A with cationic moieties and/or a reduction in the lipid A content, which, together with the increased content of the aforementioned lipids, serve to make the K. pneumoniae outer membrane and OMVs more impervious to polymyxin attack.