https://orcid.org/0000-0001-6751-4098
Abstract
Glycopeptides (GPs), lipopeptides (LPs) and lipoglycopeptides (LGPs) are related antimicrobials important for the management of invasive MRSA infections. Cross-resistance among these antibiotics in MRSA is well documented, as is the observation that susceptibility of MRSA to β-lactams increases as susceptibility to GPs and LPs decreases (i.e. the seesaw effect). Efforts to understand the relationship between GP/LP/LGP cross-resistance and the seesaw effect have focused on the PBPs, but the role of lipid metabolism has not been investigated.
Since the cell membrane is structurally and metabolically integrated with the cell wall and anchors associated proteins, including PBPs, we examined the relationship between membrane lipid composition and the phenomena of cross-resistance among GPs/LPs/LGPs and the β-lactam seesaw effect.
We selected for daptomycin, vancomycin and dalbavancin resistance using the USA300 strain JE2 and evaluated the resulting mutants by WGS, MS-based lipidomics and antimicrobial susceptibility testing to assess the relationship between membrane composition, cross-resistance, and the seesaw effect.
We observed cross-resistance to GPs/LPs/LGPs among the selected strains and the seesaw effect against various β-lactams, depending on the PBP targets of the particular β-lactam. We found that modification of membrane composition occurs not only in daptomycin-selected strains, but also vancomycin- and dalbavancin-selected strains. Significantly, we observed that the abundance of most phosphatidylglycerols positively correlates with MICs of GPs/LPs/LGPs and negatively correlates with the MICs of β-lactams.
These studies demonstrate a major association between membrane remodelling, cross-resistance and the seesaw effect.
Introduction
Vancomycin, a glycopeptide (GP), remains the first-line antibiotic against invasive MRSA infections.1 Several alternatives to vancomycin are available, including the lipoglycopeptides (LGPs), such as dalbavancin, and the lipopeptide (LP) daptomycin. However, cross-resistance among GPs/LGPs/LPs has been observed.2,3
GP- and LGP-resistant MRSA frequently acquire mutations in genes that regulate cell-wall metabolism including yycGF (or walKR)4 and vraTSR.5 Daptomycin exposure often leads to gain-of-function mutations in mprF (fmtC),6 which converts the major membrane lipids, phosphatidylglycerols (PGs), with a negative charge, to the positively charged lysyl-phosphatidylglycerols (LysylPGs), effectively decreasing daptomycin binding.7,8 Dalbavancin resistance selected by dalbavancin exposure remains uncommon, but such isolates have exhibited reduced susceptibility to vancomycin and daptomycin associated with vraT (yvqF) mutations.9
Many studies have observed that the susceptibility of MRSA to β-lactams increases as the susceptibility to vancomycin or daptomycin decreases, a phenomenon known as the seesaw effect.3,10–12 While the exact mechanism of the seesaw effect remains uncertain, some evidence suggests that it arises from changes in the expression or activity of PBPs.13–15 However, there is no clear genotype associated with the seesaw effect and there has been no attempt to select for GP/LP/LGP resistance in the same genetic background to explore the emergence or absence of this phenomenon.
The cell wall and cell membrane are metabolically and structurally integrated. PGs donate their glycerol phosphate headgroup to the essential cell-wall component, lipoteichoic acid (LTA), to form its polymeric backbone.16 On the other hand, diglucosyl diacylglycerols (DGDGs) serve as the lipid anchor of LTA to the cell membrane. Thus, we hypothesize that differences in the emergence of cross-resistance among GPs/LPs/LGPs and the β-lactam seesaw effect will produce characteristic alterations in lipid metabolism as these antimicrobials target various components of the cell envelope.
In this study, we assessed the occurrence of cross-resistance among GPs/LPs/LGPs and the β-lactam seesaw effect in MRSA strains that had been selected for resistance to daptomycin, vancomycin and dalbavancin (in this study ‘resistance’ is used to indicate non-susceptibility, reduced susceptibility or resistance). We then explored the emergence of cross-resistance and the β-lactam seesaw effect and evaluated the global lipid profiles of all study strains to correlate cross-resistance and the seesaw effect with the composition of the cell membrane.
Materials and methods
In vitro selection of resistance
Daptomycin-, dalbavancin- and vancomycin-non-susceptible mutants were selected using the serial passage method from the well-characterized USA300 MRSA strain JE2 (BEI Resources) as described in the Supplementary data (available as Supplementary data at JAC Online).17
Antibiotic susceptibility testing
Susceptibility testing was carried out in accordance with CLSI guidelines, or by Etest® in the case of telavancin, as described in the Supplementary data.18
WGS
DNA extraction, sequencing libraries and sequence analysis were performed/prepared as described previously19,20 and in the Supplementary data.
Sample preparation, lipidomic analysis and data analysis
Sample preparation, hydrophilic interaction LC (HILIC)-ion mobility-MS (HILIC-IM-MS)-based lipidomic analysis and data analysis were carried out as described previously and in the Supplementary data.17
Data availability
WGS data from this study are available from the NCBI Sequence Read Archive (SRA; http://www.ncbi.nlm.nih.gov/sra) under BioProject number PRJNA547605.
Results
In vitro selection for GP/LP/LGP resistance
From JE2, mutants were selected with 4-, 256- and 4-fold reduced susceptibility to daptomycin (JE2-Dap2), dalbavancin (JE2-Dal2) and vancomycin (JE2-Van4), respectively (Figure 1a).
(a) Strains used in this study and their antibiotic susceptibilities and genetic profiles. (b) Susceptibility of JE2-derived strains to antibiotics targeting different PBPs and peptide-based antibiotics. Log2 fold change calculated relative to the MIC for the parent JE2 strain. VAN, vancomycin; DAP, daptomycin; DAL, dalbavancin; NAF, nafcillin; LEX, cefalexin; PBPns, non-selective PBP inhibitor.
WGS was performed on all selected strains (Figure 1a). Mutations in vraT of the vraTSR operon and/or yycG of the yycGF operon were observed in JE2-Dal2 and JE2-Van4. A mutation in mprF was identified in JE2-Dap2. Additionally, JE2-Van4 contained mutations in ssaA, which may play a role in peptidoglycan autolysis, and vraG, which has been implicated in vancomycin resistance previously.5
GP/LP/LGP cross-resistance
Cross-resistance was universally observed by MIC testing among all strains (Figure 1a). JE2-Dal2 displayed cross-resistance to vancomycin and daptomycin, with 8- and 4-fold reduced susceptibility, respectively. JE2-Van4 and JE2-Dap2 displayed modest cross-resistance to daptomycin and vancomycin, respectively, but dalbavancin MICs increased 64- and 32-fold, respectively.
Susceptibility to β-lactams
The susceptibilities to six β-lactams with varied PBP specificities were determined [Figure 1b and Table S1 (available as Supplementary data at JAC Online)]. The JE2-Dal2 strain displayed the seesaw effect with most β-lactams, with increased susceptibility to nafcillin (non-selective PBP inhibitor; 8-fold), meropenem (PBP1; 4-fold), ceftriaxone (PBP2; 2-fold) and cefalexin (PBP3; 4-fold). JE2-Dap2 and JE2-Van4 showed increased susceptibility to some, but not all, β-lactams. Notably, all study strains displayed the seesaw effect with the PBP3-specific cefalexin, but none displayed a clear seesaw effect with PBP4-targeting cefoxitin. These results are consistent with previous reports that the seesaw effect is PBP isoform-dependent.13–15
Membrane lipids in GP/LP/LGP resistance
Principal component analysis (PCA) of the lipidomic data showed clustering of biological replicates and clear intergroup separation (Figure S1). A summary of the major lipids observed in the JE2-derived mutant strains is shown in Figure 2(a) and in Table S2. For JE2-Dap2, most PGs were unchanged relative to the parent strain, but the minor long-chain PGs 36:0 and 37:0 were elevated. In JE2-Dal2, short-chain PGs 30:0 and 31:0 were decreased, whereas PGs 32:0 to 37:0 were elevated to varying degrees. In JE2-Van4, a trend of increased PG 36:0 and 37:0 relative to the parent strain was also observed. Thus, the long-chain PGs 36:0 and 37:0 were consistently increased for all passaged strains.
(a) Lipidomic results for S. aureus JE2 and JE2-derived daptomycin-resistant (Dap2), dalbavancin-resistant (Dal2) and vancomycin-resistant (Van4) strains. Major species of PGs, DGDGs, LysylPGs, CLs and FFAs from HILIC-IM-MS analysis are shown. Results are row-centred and scaled by unit variance scaling. Statistical significance was determined by Student’s t-test, two-tailed with equal variance, relative to the parent strain (Par). *P ≤ 0.05 and **P ≤ 0.005. (b) Pearson correlation coefficients, r, for the relationship between levels of individual lipid species and MICs of various β-lactam and peptide-based antibiotics for JE2 and JE2-derived strains (n = 4). *P ≤ 0.05, two-tailed, relative to Par. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.
The amount of LysylPGs was 1.6- to 5.2-fold higher in JE2-Dap2 than in the other JE2-derived strains, which is consistent with the presence of mutation in mprF.8 LysylPGs 34:0 and 35:0 were also significantly higher in JE2-Dal2 and JE2-Van4 relative to the parent strain, but the extent of the increase was small. LysylPGs 36:0 and 37:0 were not detected in any JE2 strains. Interestingly, we also observed significant increases in all cardiolipins (CLs) in the JE2-Dap2 strain relative to the parent strain (Figure 2a).
All major species of DGDGs were significantly decreased in JE2-Dap2. In JE2-Dal2, DGDGs changed in both directions. In JE2-Van4, only DGDGs 30:0, 31:0 and 33:0 were decreased. For free fatty acids (FFAs), there was a consistent trend of increase in FFAs 15:0, 17:0, 18:0, 19:0 and 21:0 in the resistant strains relative to the parent strain.
Correlation of lipid levels and MICs
Pearson correlation coefficients were determined between the levels of individual lipid species and antibiotic susceptibility (Figure 2b and for P values see Figures S2 to S6).
Overall, the PGs were more significantly and consistently correlated with MICs than the other lipid species. In general, the MICs for the peptide-based antimicrobials were positively correlated with the levels of PGs 32:0 to 37:0 and negatively correlated with PGs 30:0 and 31:0. Conversely, the MICs of the β-lactams (except ceftaroline) were negatively correlated with PGs 32:0 to 37:0 and positively correlated with PGs 30:0 and 31:0 as might be expected from strains demonstrating the seesaw effect.
The broad-scale correlation between ceftaroline MIC with DGDGs, LysylPGs and CLs appears to be driven by the JE2-Dap2 strain, which displayed overall decreases in DGDGs and increases in LysylPGs and CLs. Notably, the long-chain compounds in both DGDGs (37:0) and FFAs (21:0) displayed stronger correlation with MICs than other lipid species.
Discussion
The JE2-Dap2 strain characterized here contained a mutated mprF gene and showed substantially increased levels of LysylPGs, consistent with previous studies on daptomycin-resistant strains.8 The levels of most PGs did not show significant changes except significant increases in the minor long-chain PGs (36:0 and 37:0), a phenomenon that was also observed in a highly daptomycin-resistant strain of MRSA.17 Significant increases in CLs and decreases in DGDGs were also observed in JE2-Dap2. A common observation is the increase in the levels of PGs 36:0 and 37:0 and odd-chain fatty acids in all three strains. The increased content of branched-chain and long-chain fatty acids also leads to increased fluidity of the membrane. Increases in membrane fluidity have been frequently observed in daptomycin-resistant MRSA.7 It is possible that increased membrane fluidity may be underlying the cross-resistance to daptomycin in JE2-Dal2 and JE2-Van4. However, the exact mechanism between membrane composition alterations and β-lactam susceptibility is unclear at this time. Recently, Renzoni et al.12 suggested a potential mechanism being the defective lipidation, mediated by PGs, in the chaperone protein, PrsA, which is involved in the post-translational folding of PBP2a.
To our knowledge, this work represents the first evaluation of membrane lipid composition in vancomycin- and dalbavancin-non-susceptible MRSA, as well as the first study to explore a relationship between membrane lipids and the β-lactam seesaw effect. Our results provide compelling evidence that: (i) membrane lipid composition is implicated in resistance to vancomycin and dalbavancin, as well as daptomycin; (ii) the alteration of membrane lipid composition in resistant strains is dependent upon their fatty acyl composition; and (iii) the abundance of most long-chain fatty acyl PGs positively correlates with the occurrence of cross-resistance and negatively correlates with β-lactam susceptibilities. As the factors underlying the seesaw effect continue to be explored, the contribution and significance of membrane lipids to β-lactam, GP, LP and LGP resistance should not be overlooked.
Funding
This work was supported by grants from the University of Washington School of Pharmacy Faculty Innovation Fund, the University of Washington Royalty Research Fund (A128444) to L.X. and B.J.W., the start-up fund to L.X. from the Department of Medicinal Chemistry at the University of Washington, the National Institute of Allergy and Infection Diseases (R21AI132994 and R01AI136979) to B.J.W. and L.X., and the Cystic Fibrosis Foundation (SINGH19R0) to S.J.S.
Transparency declarations
B.J.W. has received research grants from commercial sources, including Shionogi. All other authors: none to declare.
Author contributions
B.J.W. and L.X. conceived the study. K.M.H. and L.X. designed and performed the lipidomic experiments and analysed the MS data. B.J.W., K.M.H., T.S. and N.K.A. performed the resistance passage studies and other microbiological aspects of the experiments. S.J.S., A.W., E.A.H., K.P. and K.M. performed WGS and analysis. All authors reviewed the data, prepared the manuscript and approved the final version.
References
CLSI. Performance Standards for Antimicrobial Susceptibility Testing—Twenty-Ninth Edition: M100.
Author notes
Brian J. Werth and Libin Xu Equal contribution to this study from both authors.
