LC-MS/MS characterization of phospholipid content in daptomycin-susceptible and -resistant isolates of Staphylococcus aureus with mutations in mprF

Abstract

Daptomycin (DAP) is a cyclic lipopeptide antibiotic used for the treatment of certain Staphylococcus aureus infections. Although rare, strains have been isolated that are DAP resistant. These strains usually have mutations in mprF, a gene encoding a membrane protein with both lysylphosphatidylglycerol (LPG) synthase and flippase activities. Because ΔmprF strains have increased DAP susceptibility, the mechanism of resistance is not likely due to a loss of mprF function. In this study, we developed an LC-MS assay to examine the effect of different mprF mutations on the ratio of phosphatidylglycerol (PG) to LPG in the membrane. Our assay demonstrated that some, but not all, mutations in the flippase and synthase domains result in small but reproducible increases in the proportion of LPG relative to PG. Techniques described herein represent a higher throughput and more sensitive method for measuring relative phospholipids levels. These results offer guidance in the understanding of how mprF confers DAP resistance; namely, mprF-mediated resistance may be through more than one mechanism, including increased overall LPG synthesis and increased LPG present on the outer leaflet of the cytoplasmic membrane.

Keywords::

• mprF
• daptomycin
• S. aureus resistance
• lysylphosphatidylglycerol

Introduction

Daptomycin (DAP) is a cyclic lipopeptide antimicrobial agent produced from Streptomyces roseosporus that has rapid bactericidal activity against gram-positive bacteria, including several resistant pathogens (Tally et al. 1999). DAP is approved by the United States (US) Food and Drug Administration for the treatment of complicated skin and soft tissue infections, as well as the treatment of bacteremia and right-sided endocarditis caused by methicillin-susceptible and methicillin-resistant Staphylococcus aureus (Enoch et al. 2007). Although the mechanism of action of DAP is not fully elucidated, it is hypothesized to involve calcium-dependent insertion into the bacterial membrane, resulting in rapid depolarization and cell death (Enoch et al. 2007). As with other antibacterials that target the cell membrane, the emergence of non-susceptibility to DAP is rare; the incidence of non-susceptibility of S. aureus to DAP is less than 10^-10 colony-forming units (CFUs) in vitro (Friedman et al. 2006, Silverman et al. 2001). Sequence analyses of various laboratory-derived and clinical DAP-resistant isolates have identified that the majority of resistant strains contain a mutation in the multiple peptide resistance factor (mprF) (data on file, Cubist Pharmaceuticals). Furthermore, conditional knockdowns utilizing antisense RNA of strains harboring the resistant mprF allele were shown to become susceptible, demonstrating that mutations in this gene may be sufficient for decreased susceptibility (Rubio et al. 2010).

MprF is a non-essential, 840-amino acid integral membrane protein that catalyzes the synthesis of the cationic membrane phospholipid lysylphosphatidylglycerol (LPG) (Ernst et al. 2009). The S. aureus cytoplasmic membrane is composed of only three phospholipids: LPG, phosphatidylglycerol (PG), and cardiolipin (CL) (White and Frerman 1967). MprF may also regulate the distribution of LPG between the inner and outer leaflets of the cell membrane (Mishra et al. 2009). In mprF deletion mutants, the sensitivity of S. aureus to cationic antimicrobial peptides and to some cationic antibacterials is increased (Kristian et al. 2003, Jones et al. 2008). It has been postulated that in these deletion mutants, the decreased positive charge on the cell membrane may lead to reduced repulsion – and therefore increased sensitization – to the cationic agents (Kristian et al. 2003, Jones et al. 2008). Because mprF deletion mutants are more susceptible to DAP, the point mprF mutations that generate DAP non-susceptibility cannot be due to a similar loss of function. Likely, the point mutations result in a gain-of-function phenotype in which the membrane composition is altered in such a way as to generate non-susceptibility. Based on a recent report, the C-terminal domain of MprF is responsible for the synthesis of LPG; the N-terminal domain is believed to be responsible for flippase activity that populates LPG in the outer leaflet (Jones et al. 2008, Ernst et al. 2009). Strains containing mutations in mprF with a resulting DAP-resistant phenotype have been mapped in both domains, including the membrane-spanning and soluble segments (Friedman et al. 2006).

The objective of this study was to use extracts from the wild-type (WT) and MprF-knockout (ΔMprF) strains to develop an analytical method for the identification and relative quantification of PG and LPG levels. Using this method, an assay was developed to analyze paired DAP-susceptible and DAP-resistant S. aureus isolates with different point mutations in mprF to gain a better understanding of how specific domains of MprF may play a role in DAP susceptibility.

Materials and methods

Bacterial strains and methods

All bacterial strains used in this study are listed and described in Table I. S. aureus strains were propagated on tryptic soy agar with 5% sheep blood (bioMerieux, Lombard, IL, USA) or in Mueller-Hinton broth (Difco Laboratories, Detroit, MI, USA) supplemented with 50 mg/l Ca²⁺ (using calcium chloride) at 37°C. For antimicrobial testing, broth microdilution minimum inhibitory concentrations (MICs) were performed according to Clinical Laboratory Standards Institute guidelines using Mueller-Hinton broth supplemented with 50 mg/l Ca²⁺. Paired clinical isolates included in the study were obtained from patients at the onset of DAP therapy or after failure while receiving DAP therapy and determined to be clonal based on pulsed field gel electrophoresis (Cubist collection). Strains were subjected to sequencing using mprF-specific primers as listed in Friedman et al. (2006).

Table I. Bacterial strains used in this study^a.

Laboratory strains
Strain no.	Relevant genotype or description	Source
CB1118	MW2: community-acquired MRSA	Said-Salim et al. 2003
CB1551	23 day laboratory serial passage of CB1118, MprF S337L, with potential other mutations (not tested)	Laboratory collection
CB1694	Laboratory serial passage of CB1118, day 2	Chan et al. 2005
CB1695	CB1694, MprF P314L	Chan et al. 2005
CB1701	MZ100	Ironwood Pharmaceuticals
CB1703	MZ100, mprF:Tn917	Ironwood Pharmaceuticals
CB2240	CB1118, ΔMprF	Laboratory collection
Clinical strains
Strain no.	Submitting laboratory or hospital	MprF allele*	YycFG allele*	RpoBC allele*
CB182^b	Strong Memorial Hospital, NY	None	None	RpoB D471Y
CB183^b	Paired with CB182	L341S	None	RpoB D471Y; S464P
CB1482^b	John Stroger Hospital, IL	None	None	RpoB D471G, R484H
CB184^b	Paired with CB1482	S295L	None	RpoB D471G, R484H
CB5011	Albany Medical Center, NY	None	None	ND
CB5012	Paired with CB5011	L826F	None	ND
CB5022	University of Maryland, MD	None	ND	ND
CB5052	Paired with CB5022	T345I	ND	ND
CB2573	Jersey Shore Hospital, NJ	None	ND	ND
CB2574	Paired with CB2573	L776S	ND	ND

^aAll clinical isolate pairs (susceptible and resistant to DAP) were obtained from the same patient and were tested for clonality by pulse-field gel electrophoresis.

*Only the following loci were sequenced: mprF, yycFG and rpoBC. None = no mutations were found. ND = sequencing of this locus not done.

^bParent strain was already rifampin resistant and mutations found in pre- and post-DAP therapy are known rifampin resistance determinants.

Construction of CB2240

MW2 was used as the parent strain for the deletion construction. Chromosomal DNA from MW2 was isolated using Promega Wizard genomic DNA purification kit (Cat # TM050; Madison, WI, USA), according to the manufacturer's specifications. The following primers were used to amplify the 5′ region of mprF: 5′ upstream (CACGGTATATGAAGTTTGTTGGTG) and 5′ downstream (aaACGCGTCGACCCAATCGTAAATATGATGGAACG). The following primers were used to amplify the 3′ region of mprF: 3′ upstream (gtgaaaaaATGAATCAGGAAGTTAAACGTCACAAATAAttaaaatccaagtgc) and 3′ downstream (TTCTAAGTAGCTCTGGACCAC). The resulting polymerase chain reaction (PCR) products were purified and used as templates for a second round of PCR to combine the fragments using homologous sequences at the 3′ end of the 5′ PCR product and the 5′ end of the 3′ PCR product. The following primers were used: Upstream primer (aaACGCGTCGACCCAATCGTAAATATGATGGAACG) and downstream primer (aaggtaCCATGGCGCATGACAGACATCATTTTCATC). The final 2698-bp PCR product was purified and digested with SalI and NcoI and ligated to pMAD to produce pMAD + ΔmprF (Arnaud et al. 2004); sequence analysis confirmed the correct insert and orientation. Electrocompetent cells of MW2 were prepared as described in Luchansky et al. (1988), and transformants were recovered after incubation at 30°C on BHI agar supplemented with 3 μg/ml erythromycin (Sigma; St Louis, MO, USA). Allelic replacement was performed in two steps, as described in Arnaud et al. (2004). PCR was used to confirm that the single integration event took place after growth at the non-permissive temperature (i.e., erythromycin-resistant phenotype) and resolution of the partial duplication (i.e., erythromycin-sensitive phenotype) to yield the final deleted variant, which was confirmed by sequence analysis.

Membrane solubilization and lipid extraction

Lipid extractions were performed using the modified Bligh and Dyer (1959) technique described herein. Paired S. aureus cultures were grown on the same day to mid-log phase and harvested by centrifugation at 4000 rpm at 4°C. Cells were washed three times with cold phosphate-buffered saline, resuspended in 1/50 volume of 10% aqueous sodium chloride, and sonicated in a water bath using glass tubes for 20 min. CHCl₃/MeOH was then added in a 1:1 ratio (8 ml); the suspension was vortexed, sonicated again for 20 min, and centrifuged at 4000 rpm at room temperature. The lower organic layer was retrieved without disturbing or retrieving the proteinaceous layer in between. The aqueous layer was extracted again using 4 ml 100% CHCl₃. After this, the CHCl₃ fractions were pooled and the solvent was removed by rotary evaporation or under nitrogen in glass vials. Each extract was tested three times in the analyses below; values reported are the average of the three tests.

Identification of membrane phospholipids by LC-MS

Membrane preparations of the WT and ΔMprF strains were analyzed using liquid chromatography-mass spectrometry (LC-MS) to separate based on phosphate headgroups using normal-phase chromatography. Abundant masses that were present in both WT and ΔMprF or only WT subsequently were used for further analysis by linear ion trap fragmentation. Membrane preparations were directly infused into the electrospray interface at 10 μl/min and detected in the linear ion trap for interpretation of their mass spectrum. Deuterium-labeled internal standards, as well as C32:0 PG and C32:0 LPG, each containing two fatty acid tails composed of 16 carbons with no double bonds, were synthesized to confirm the fragmentation pattern of each. These included both d₅-dipalmitoyl PG (d5-DPPG; deuterated C32:0 PG [727 amu]) and d₄-lysyl-dipalmitoyl PG (d₄-lysyl-DPPG; deuterated C32:0 LPG [854 amu]), as well as non-deuterated DPPG (C32:0 PG [722 amu]) and lysyl DPPG (C32:0 LPG [850 amu]). Subsequently, the deuterium-labeled standards were used as references for both accurate mass determination and LC-MS/MS quantification as described below.

Measurement of PG:LPG ratios by LC-MS using multiple reaction monitoring

For each strain, the membrane preparation was dissolved in a mobile phase solution containing known concentrations of internal standards at a dilution equivalent of 1:10. The LC-MS using multiple reaction monitoring (LC-MRM) setup included an Alltech Saphira 50 × 1 mm internal diameter, 5-μm diol column with a mobile phase consisting of 84.8:14.8:0.5 (vol/vol/vol) (chloroform:methanol:water) + 20 mM ammonium acetate and 0.1% vol acetic acid with a flow rate of 150 μl/min. The detector (API 4000 Qtrap^™ mass spectrometer; Applied Biosystems) was set to monitor MS/MS fragmentation reactions corresponding to loss of PG [M-172]⁺ of C30:0 through C36:0 PG and losses of LPG [M-300]⁺ for C30:0 through C36:0 LPG. This detection strategy avoided complications from headgroups of other phospholipids. Deuterated and non-deuterated standards were used to calibrate the phospholipids being assessed. For all samples, the injection volume was 1 μl. Using standard curves of C32:0 PG and C32:0 LPG with respective deuterated internal standards, the concentrations of C30:0 to C36:0 PG and LPG were determined. The PG and LPG content were reported as the respective sum of the measured molecular species. The phosphorus content of the extract was determined to normalize total PG and LPG concentrations using modified methods from Chen et al. (1956) and Fiske and Subbarow (1925). Analyte concentration calculations were performed using the following equations: (i) Analyte concentration from C32:0 standard curves (μg/ml) × 10; and (ii) analyte concentration (nmol) = (μg/ml/[M + H]⁺ -1 amu) × diluent volume × 1000.

Results

Characterization of PG and LPG

Four major peaks were observed from WT and ΔMprF extracts after separation by headgroup using LC-MS. The most abundant masses under each peak were selected for further analysis by high-resolution MS shown in italics in Table I. Based on mass and elution time, deuterated standards were used to confirm the lipid identified and subsequently used as the internal standard for quantification. It should be noted that PG (722 amu) and LPG (850 amu) were identified through negative-mode MS/MS, which are shown in Figure 1 and Figure 2. As presented in Figures 1 and 2, MS/MS analysis of the WT extract 721.44 m/z ion was identified as PG containing 17:0 and 15:0 fatty acid tails. MS/MS analysis of the WT extract 849.57 m/z ion was identified as LPG with 17:0 and15:0 fatty acid tails. The WT extract 750.1 m/z ion was identified as PG with 15:0 and19:0 fatty acid tails (data not shown). Parent ions with m/z of 766 and 928 could not be identified and consequently were not pursued further. However, PG and LPG were definitely identified from the WT extracts representing the substrate and product of the MprF enzymatic reaction and importantly, LPG was not found in the ΔMprF extract. Major fragment ion peaks for PG were identified as glycerol phosphate (170.96 m/z), C-15 lipid ester (241.20 m/z), and C-17 lipid ester (269.20 m/z), whereas major fragment ions for LPG were identified as lysyl ester (145.12 m/z), C-15 lipid ester (241.20 m/z), C-17 lipid ester (269.20 m/z), and des-C-15 LPG (625.56 m/z). PGs (722 and 750 amu) were found in both WT and ΔMprF strains. As expected, no molecular ion peak was observed for LPG (850 m/z) in the ΔMprF strain, further validating that the ions observed for the 850 m/z lipid corresponded to LPG.

Table II. Molecular ions and retention times of major lipids observed from WT/CB1701 or ΔMprF/CB1703 extracts.

Sample	WT	ΔMprF	WT	ΔMprF	WT	ΔMprF	WT	ΔMprF
RT, min	3.61	3.64	4.90	5.02	8.96	9.51	9.98	None
	738.0	738.0	694.0	694.0	878.3 m		822.2
	752.0	752.0	708.0	708.0	900.1	900.2	836.2
	754.1	754.2	722.0	722.0	914.2	914.2	850.3
	766.1	766.1	736.0	736.1	928.1	928.2
	780.2	780.2	750.1	750.0	942.2	942.2
	794.2	794.2			956.1	956.2
	839.4 m					972.1 m
						975.0 m

The most abundant ions detected and chosen for identification by fragmentation are highlighted in italics. m, minor peak; ΔMprF/CB1703, MprF-knockout; RT, retention time; WT/CB1701, wild-type.

Figure 1. MS/MS characterization of phosphatidylglycerol (722 amu). Fragmentation of parent 721 [M-H]− ion was performed as described in Materials and methods. The major fragment ions 241 and 269 m/z are consistent with C15:0 and C17:0 fatty acid tails. The 170-m/z fragment is consistent with release of the PG headgroup. Taken together, the identity of the 721-m/z ion is consistent with the structure as shown in the Figure above.

Figure 2. MS/MS characterization of lysylphosphatidylglycerol (850 amu). Fragmentation of the parent 849 [M-H]− ion was performed as described in Materials and methods. The major fragment ions 241 and 269 m/z are consistent with C15:0 and C17:0 fatty acid tails. The 145-m/z fragment is consistent with release of the lysine moiety of the headgroup. The 625-m/z fragment is consistent with the release of the LPG moiety. Taken together, the identity of the 849-m/z ion is consistent with the structure as shown in the Figure above.

LC-MRM for PG and LPG quantitation

There was some variability in the PG:LPG ratio for the different WT strains and clinical strains with susceptibility to DAP (Table III). Mutant strain CB2240 (ΔMprF) demonstrated a PG:LPG ratio greater than 1000, thus indicating a substantial depletion of LPG in the cell membrane. In three of the six pairs analyzed (CB1694/1695, CB1118/1551, and CB2573/2574), the PG:LPG ratio decreased, whereas in four other pairs (CB5022/5052, CB5011/5012, CB182/183, and CB1482/184), no ratio change was observed.

Table III. PG:LPG results of paired DAP-sensitive and DAP-resistant isolates.

Strain	Background	MprF allele	DAP MIC (μg/ml)	PG:LPG ratio
CB1113 CB2240	MW2	Clean deletion	0.5 0.125	3.9 >1,000
CB1694 CB1695*	MW2	P314L	0.5 2	3.1 1.5
CB1118 CB1551*	MW2	S337L	0.5 16	3.7 2.2
CB5022 CB5052	Clinical	T345I	0.5 2	4.0 2.3
CB5011 CB5012	Clinical	L826F	0.5 4	3.8 3.1
CB182 CB183	Clinical	L341S	0.5 4	1.7 1.9
CB1482 CB184	Clinical	S295L	0.5 4	4.0 4.2
CB2573 CB2574	Clinical	L776S	0.5 2	2.0 0.5

*Additional mutations are known to exist in these strains. MW2, DAP-susceptible parental strain of Staphylococcus aureus.

Discussion

In this study, an LC-MS/MS-based method was developed for measuring the relative proportions of PG and LPG in DAP-sensitive and DAP-resistant strains. The validity of this assay was confirmed using a clean ΔMprF knockout, which, as anticipated, demonstrated a substantial increase in PG:LPG ratio. PG:LPG shifts indicative of either greater LPG concentrations or lower PG concentrations were observed for some – but not all – DAP-resistant strains. There were two clinical strains with a mutation in the soluble domain (at amino acids 776 and 826); and both these strains demonstrated a decrease in the PG:LPG ratio, likely affecting LPG synthesis. There were five laboratory and clinical strains with a mutation in the membrane portion of the protein. In three of these strains (at amino acids 314, 337, and 345), there was a decrease in the PG:LPG ratio; in the two other strains (at amino acids 295 and 341), there was no shift in the PG:LPG ratio despite decreased susceptibility to DAP. In this study, the median PG:LPG ratios was 3.75 for susceptible strains (range, 1.7–4.0) and 2.2 for resistant strains (range, 0.5–4.2). In a previously published report based on PG and LPG levels quantified by two-dimensional thin-layer chromatography (2D-TLC), the PG:LPG ratio was 6.8 for susceptible strains and ranged from 2.8–7.0 for resistant strains (Mishra et al. 2009). We also performed cytochrome C binding assays (data not shown) as a readout for overall surface charge to assess a whether a correlation could be found in strains with increased LPG synthesis, but no such correlation could be detected in the strains described herein.

Based on this analysis, there is no absolute correlation between relative LPG concentration and DAP resistance. These results are consistent with other reports that there may be more than one mechanism for MprF-mediated DAP resistance. In strains with decreased PG:LPG ratios, resistance may be due to an increase in the positive charge of the membrane from increased overall LPG content. In strains in which the PG:LPG ratio is not significantly altered, resistance may be due to increased flippase activity that disproportionately overpopulates the outer leaflet with LPG relative to PG. Both these mechanisms would account for the increased cationic charge observed in DAP-resistant isolates (Jones et al. 2008). It also is possible that the mutants examined in this study improved the interaction between the two domains. For example, a mutation in the synthase domain may have improved its interaction with the flippase domain, resulting in more robust transport activity. Results from a recent study, in which varying the levels of LPG using different plasmids constructs of mprF did not affect the level of DAP resistance as long as some LPG was produced, are consistent with a flippase-mediated mechanism of resistance (Ernst et al. 2009).

It is important to note that this technique was used to measure the relative levels of phospholipids; it was not used to quantify actual phospholipid amounts. Other techniques have been used for decades for measuring phospholipids in the bacterial membrane; these techniques include 2D-TLC and a variety of molybdenum-based stains. In our opinion, the assay developed in this study is a marked improvement over existing techniques because it does not require the use of radiolabeled phospholipids that are difficult to handle or stains that may have variable intensities, which could introduce error when determining relative concentrations. In addition, the technique allows for the direct and selective measurement of PG and LPG species against synthetic standards of C32:0 PG, C32:0 LPG with corresponding C32:0 PG-d5, and C32:0 LPG-d4 internal standards, resulting in increased accuracy and precision. This technique is able to identify any unusual peaks and to confirm that the composition of the major lipid species has not changed. Sample preparation is straightforward according to established lipid chemistry. Finally, LC-MS/MS analysis time is approximately 3 min per sample, a significantly shorter run time compared with performing 2D-TLC.

It is important to note that certain limitations were inherent to this study. For example, the lack of availability of a range of mutants spanning MprF did not allow for a survey of the entire protein. Also, the presence of other mutations in the background of the clinical strains could affect the MIC of the mutants; this finding is most apparent in strains that were resistant to DAP but did not demonstrate a change in the PG:LPG ratio. Additionally, the membrane preparation method may have depleted one membrane component to a greater degree than another. This type of error would be consistent for all membrane preparations, and the results from this assay may be used to help develop a whole-cell screen that does not rely on extraction or inclusion of a known concentration of internal standard before extraction. Finally, although studies to date have shown that mutations in mprF increase the MIC of the organism to daptomycin, the molecular mechanism for how these changes in the protein specifically confer resistance remains unknown.

Conclusion

We have developed an assay for determining the relative proportions of PG and LPG in S. aureus. This newly developed assay can be utilized to measure the PG:LPG ratio in DAP-sensitive and DAP-resistant strains, as demonstrated in this study. Among the strains tested, no simple correlation can be determined between mutations in mprF and lipid phenotype; however, a relative increase in the proportion of LPG was demonstrated in some strains. As additional resistant strains become available, this assay can be used to more thoroughly probe mutant strains with amino acid changes in mprF to gain a better understanding of PG:LPG ratios and daptomycin resistance. Because the differences in PG:LPG ratios between DAP-susceptible and DAP-resistant isolates are relatively small, development of additional methods for improved reproducibility and accuracy currently are under way. In addition, it is understood that additional studies investigating the impact of LPG membrane distribution on S. aureus susceptibility are warranted, and this technique could be applied to quantify LPG in the inner and outer leaflets of the membrane if methods that could reproducibly separate or tag these two leaflets were available.

Acknowledgements

The authors thank Jeff Alder, Steve Gilman, and Avanti Polar Lipids; and Brian Falcone, in association with ApotheCom, for providing editorial assistance supported by Cubist.

Declaration of interest: Drs Rubio, Varoglu, Chu, and Silverman and Ms Conrad are current employees of Cubist or were employees of Cubist at the time the study was conducted. Drs Moore and Shaw are employees of Avanti Polar Lipids. The authors alone are responsible for the content and writing of the paper.