Gain-of-Function Mutations in the Phospholipid Flippase MprF Confer Specific Daptomycin Resistance

Ever since daptomycin was introduced to the clinic, daptomycin-resistant isolates have been reported. In most cases, the resistant isolates harbor point mutations in MprF, which produces and flips the positively charged phospholipid LysPG. This has led to the assumption that the resistance mechanism relies on the overproduction of LysPG, given that increased LysPG production may lead to increased electrostatic repulsion of positively charged antimicrobial compounds, including daptomycin. Here we show that the resistance mechanism is highly specific and relies on a different process that involves a functional MprF flippase, suggesting that the resistance-conferring mutations may enable the flippase to accommodate daptomycin or an unknown component that is crucial for its activity. Our report provides a new perspective on the mechanism of resistance to a major antibiotic.

IMPORTANCE Ever since daptomycin was introduced to the clinic, daptomycinresistant isolates have been reported. In most cases, the resistant isolates harbor point mutations in MprF, which produces and flips the positively charged phospholipid LysPG. This has led to the assumption that the resistance mechanism relies on the overproduction of LysPG, given that increased LysPG production may lead to increased electrostatic repulsion of positively charged antimicrobial compounds, including daptomycin. Here we show that the resistance mechanism is highly specific and relies on a different process that involves a functional MprF flippase, suggesting that the resistance-conferring mutations may enable the flippase to accommodate daptomycin or an unknown component that is crucial for its activity. Our report provides a new perspective on the mechanism of resistance to a major antibiotic.
protein structure modulates the substrate range of the flippase to accommodate daptomycin or another membrane-embedded substrate that is crucial for daptomycin activity.

RESULTS
Distinct point mutations at the junction of MprF synthase and flippase lead to DAP-R. The most frequently identified MprF mutations associated with DAP-R are located at the junction of the flippase domain and synthase domain or in the synthase domain of the protein ( Fig. 1A; see also Table S1 in the supplemental material). These strains often contain additional point mutations in other chromosomal loci, such as yycFG (walKR), rpoB, rpoC, vraS, and dltA (6,7), raising the issue of whether the documented MprF mutations are in fact sufficient for mediating the DAP-R phenotype. In order to elucidate the contribution of individual mutations to DAP-R in a defined genetic background, the most frequently identified mutations were introduced into mprF harbored on a plasmid, which was then transferred to the S. aureus 113 (SA113) mprF mutant. Two mutations at the junction between the flippase domain and the synthase domain (T345A and V351E) led to significantly increased, clinically relevant DAP-R (MIC of 3 g/ml) compared to the parental MprF sequence (MIC of 1 g/ml) (Fig. 1B). In contrast, other mutations in this region of the protein (S295L, P314L, and S337L) and two mutations in the synthase domain (I420N and L826F) did not alter daptomycin susceptibility, suggesting that these mutations contribute to DAP-R only in combination with additional mutations. Thus, specific mutations at the junction between the synthase domain and the flippase domain of MprF can reproducibly confer DAP-R in S. aureus.
DAP-R-conferring point mutations in MprF do not alter the cellular LysPG level or membrane leaflet distribution. Basal levels of CAMP resistance mediated by MprF depend on the protein's capacity to synthesize LysPG and translocate a substantial amount of this lipid to the outer membrane leaflet, where it repulses harmful cationic proteins by electrostatic interaction (17,18,23,24). In order to elucidate if the DAP-R-associated point mutations at the junction between the synthase domain and  (Table S1). The MprF synthase and flippase domains are shown in gray and black, respectively. (B) Impact of daptomycin resistance-associated point mutations expressed in the S. aureus ⌬mprF mutant on daptomycin susceptibility. The recently characterized clinical daptomycin-resistant isolate, strain 703 (4), served as a control and reference for clinically relevant daptomycin MICs. Values that are significantly different from the values determined for the S. aureus ⌬mprF mutant expressing wild-type MprF (pRBmprF) are indicated (***, P Ͻ 0.0001). The means plus standard errors of the means (SEM) of results from at least five independent experiments are shown. WT, wild type. the flippase domain ( Fig. 1B) might lead to increased activity of one of the two protein domains, the levels of LysPG production and distribution between the inner and outer membrane leaflets of S. aureus with native versus mutated mprF were compared. LysPG contents were determined by thin-layer chromatography (TLC) and staining of LysPG with the phosphate group-specific dye molybdenum blue (20,21). None of the S. aureus strains expressing a mutated mprF gene exhibited altered LysPG production ( Fig. 2A), indicating that these point mutations confer DAP-R via a mechanism other than increasing LysPG synthesis. Of note, the expression of the cloned mprF variants was controlled by the constitutive Bacillus subtilis promoter vegII, which may explain why they displayed slightly reduced LysPG production compared to the wild type. The localization of LysPG in the cytoplasmic membrane was determined by incubating intact S. aureus cells expressing MprF with wild-type sequence or with the T345A mutation with the fluorescent dye fluorescamine, which reacts with the free amino group of LysPG at the outer membrane leaflet but cannot access the inner leaflet. Thin-layer chromatography and quantification of fluorescamine-labeled versus nonlabeled LysPG allowed inner-leaflet and outer-leaflet LysPG to be distinguished (25,26). Experiments performed with wild-type MprF and T345A-MprF led to the same percentage of LysPG in the outer membrane (ca. 40%) (Fig. 2B), indicating that DAP-R is not associated with an increased capacity of MprF to translocate LysPG. Thus, the signature mutations in MprF leading to DAP-R do not seem to alter either of the two documented activities of MprF.
The DAP-R-conferring MprF point mutation T345A does not alter the S. aureus surface charge. Daptomycin is thought to integrate into the bacterial cytoplasmic membrane upon binding of calcium ions in a manner similar to that seen with many typical CAMPs (19). Most bacteria achieve protection against a broad range of CAMPs by introduction of positive charges that modify the bacterial surface charge and diminish the affinity for CAMPs, thereby allowing bacteria to tolerate substantial CAMP concentrations (27). Aside from the modification of membrane phospholipids with lysine or other amino acids, the neutralization of negatively charged teichoic acids (TAs) with D-alanine is a particularly widespread CAMP repulsion mechanism found in several bacterial divisions (27,28). D-Alanylation of teichoic acids is mediated by the DltABCD system, which is composed of four proteins responsible for activation, transfer, and linkage of cytosolic D-alanine residues onto the backbone of teichoic acids (29). In order to determine if mutated, DAP-R-conferring MprF affects the Dlt system, we quantified the teichoic acid D-alanylation of S. aureus expressing wild-type MprF compared to T345A-MprF (Fig. 3A). While a dltA knockout mutant serving as a negative control showed the complete absence of D-alanylation, we did not observe a difference between strains expressing wild-type MprF and strains expressing T345A-MprF. In order to analyze if the daptomycin resistance-causing point mutations in MprF could somehow affect the overall S. aureus surface charge in a LysPG or wall teichoic acid (WTA) alanylation-independent manner, the capacities of S. aureus expressing MprF with wild-type sequence or a T345A mutation to bind the cationic protein cytochrome c or calcium-bound annexin V was compared. These model proteins were proven to allow a sensitive assessment of changes in the surface charge of S. aureus in several previous studies (17,18,21,30). The lack of mprF had a profound impact on the capacity of S. aureus to bind cytochrome c or annexin V (Fig. 3B), demonstrating the suitability of the assays. However, the T345A mutation did not alter the binding behavior of annexin V significantly (Fig. 3C), and both DAP-R-conferring mutations (T345A and V351E) did not alter the binding of cytochrome c (Fig. 3B). Thus, the DAP-R-conferring point mutations in MprF do not lead to a general alteration of the cell surface charge.
MprF point mutation T345A causes cross-resistance only to daptomycin and the related lipopeptide antibiotic friulimicin B. The DAP-R-conferring point mutations in MprF do not seem to be based on a canonical CAMP resistance strategy, which raises the issue of how specific the resistance mechanism may be. A variety of cationic, membrane-active antibiotics from different classes and with different modes of action, including the calcium-dependent lipopeptides daptomycin and friulimicin B, the calcium-independent lipopeptide polymyxin B, the nonlipidated peptide bacitracin, the glycopeptide vancomycin, and the lantibiotics nisin and gallidermin (31,32), were analyzed for their capacity to inhibit growth of S. aureus expressing MprF with wild-type sequence or a T345A mutation. The ability of these compounds to inhibit S. aureus growth was reduced in the presence of a functional MprF protein (Fig. 4A), which indicates that surface charge alterations have a strong impact on the capacity of these agents to inhibit S. aureus. In contrast, the neutral antibiotic oxacillin was not affected by MprF. However, the T345A and V351E mutations in MprF exclusively led to crossresistance to friulimicin B, which is the closest relative of daptomycin among the tested antibiotics ( Fig. 4) but has a different target (10). Of note, the addition of calcium to calcium-independent antibiotics did not lead to differences in the levels of inhibition of S. aureus expressing wild-type versus T345A-MprF by any of these compounds (Fig. S1). Thus, mprF-mediated DAP-R does not lead to broad-spectrum cross-resistance to cationic antibiotics and antimicrobial peptides but is restricted to compounds with a specific, daptomycin-related structure. Moreover, the different targets of daptomycin and friulimicin B suggest that the resistance mechanism does not involve the target of daptomycin.
T345A-mediated DAP-R depends on the presence of a functional MprF flippase domain. The T345A point mutation does not alter the LysPG flippase activity of MprF, but it may enable the flippase to translocate other substrate molecules in addition to Bound annexin (%) LysPG. We have previously identified conserved amino acids in the flippase domains of MprF proteins and have shown that they are essential for flippase activity (20).    (Fig. 5A), which confirms that LysPG could not be translocated from the inner layer to the outer layer of the cell membrane in these strains. T345A was not able to increase the daptomycin MIC in combination with mutations D71A and R206A and led to an only slightly increased MIC with mutation R112A (Fig. 5B), indicating that the functionality of the MprF flippase is crucial for the capacity of T345A to confer DAP-R. The T345A point mutation reduces intramolecular interactions of MprF domains. The point mutations in MprF leading to DAP-R do not occur at conserved amino acid positions, but they involve a variety of sites at the junction of the LysPG synthase and flippase domains (Fig. 1B). The various domains of MprF have been found to undergo several complex intramolecular interactions (20), which may be altered by the T345A point mutation. In order to test this hypothesis, the impact of T345A on the capacities of full-length MprF or of the synthase and flippase domains to interact were compared in the bacterial two-hybrid system, which has been proven to be suitable for elucidating intramolecular MprF interactions (20). Full-length MprF proteins with native sequence and those with the T345A point mutation showed similar capacities to interact (Fig. 6A). However, the flippase domain and an extended version of the flippase domain, which was previously shown to be required for full flippase activity (20), interacted with the synthase domain much less efficiently when the T345A mutation was present. Thus, the T345A mutations in MprF (leading to specific resistance to structurally related lipopeptide antibiotics) are associated with reduced intramolecular interactions. Such mutations do not seem to affect the efficiency of the flippase functionality but might instead extend the range of molecules that the flippase is able to translocate (Fig. 6B).

DISCUSSION
Point mutations leading to resistance to antibiotics are a common phenomenon occurring during therapy with almost any antimicrobial compound (33). Resistance levels conferred by such mutations often lead to only moderately increased MICs; however, these can be sufficient to compromise the efficacy of antibiotic therapies. The mechanisms of antibiotic resistance are diverse, ranging from modified target molecules to decreased uptake or gain of function of enzymes that inactivate the antimicrobial compound (33). Elucidation of how the signature DAP-R-associated point mutations compromise the antibiotic's activity has remained elusive (6). Since such mutations often occur within MprF, a protein which is known to electrostatically repulse CAMPs by synthesizing and translocating cationic LysPG (24), it is tempting to speculate that an increase of LysPG levels in the outer leaflet of the cytoplasmic membrane may be the major consequence of these DAP-R-conferring point mutations. By introducing the most frequently identified mprF mutations among clinically derived DAP-R strains into an S. aureus strain with a defined genetic background, we found, surprisingly, that MprF-Mediated Daptomycin Resistance ® many of them were not able to cause DAP-R. Yet it is possible that such mutations contribute to DAP-R in a more complex manner involving, for instance, additional changes in multiple genetic loci. Notably, our study results indicate that DAP-Rconferring point mutations at the junction of the flippase and synthase domains of MprF such as T345A and V351E do not alter the level or translocation of LysPG. Other research groups have reported the association of altered LysPG production and/or translocation with these reported mprF point mutations (3,4,14,34). However, since most of those studies analyzed strains that had been under in vitro or in vivo selection pressure, it is possible that they harbored additional point mutations that either preexisted in the parental isolates or were acquired during daptomycin exposure that could have influenced MprF activity and DAP-R, for example, by additional modifications of the cell envelope or by other, less obvious modifications (see Table S2 in the supplemental material) (35). The T345A point mutation was selected for more-detailed analyses and was found to not alter LysPG production or translocation, D-alanylation of teichoic acids, or the overall cell surface charge. T345A conferred resistance to only two structurally related lipopeptide antibiotics, namely, daptomycin and friulimicin B, whereas the activity of other lipopeptide or peptide antibiotics was not affected. Since daptomycin and friulimicin B do not share the same target (10), the resistance mechanism appears to be based on specific interactions of MprF with the structurally related lipopeptide antibiotics rather than with a target molecule. Moreover, T345A could confer resistance only when the flippase domain was functional, suggesting that flippase functionality may have been extended (rather than compromised), leading to DAP-R. We have previously shown that the MprF flippase is capable of flipping two different phospholipids species, Ala-PG and LysPG (21), which indicated that the flippase has relaxed substrate specificity for similar substrates. As proposed for other phospholipid flippases (36) and as suggested by our recent structural investigation of MprF (20), the membraneintegrated domains of MprF likely associate to form a channel in order to accommodate phospholipid substrate molecules and to facilitate their translocation. Thus, the affinity of MprF domains for each other may determine the substrate specificity of the channel and, in the case of reduced domain interactions, may extend the substrate specificity of the flippase to accommodate and translocate either daptomycin and friulimicin B or another membrane-embedded molecule whose orientation in the membrane is crucial for the activity of these antibiotics (Fig. 6B). A change in MprF flippase specificity would be in agreement with all of our findings; however, this is particularly difficult to demonstrate directly because it would likely affect the orientation of substrate molecules in the membrane rather than their presence or absence. Moreover, as shown previously for other phospholipid transporters (36), the translocation process is probably very fast and may be reversible. Indeed, all our attempts to demonstrate that T345A-mutated MprF affects the membrane integration or orientation of daptomycin led to inconclusive results. Thus, future highly sophisticated and time-consuming biophysical technology performed with in vitro-reconstituted MprF-containing membrane vesicles will likely be necessary to study MprF-mediated altered daptomycin translocation dynamics in the membrane. Our finding that T345A does not alter the LysPG synthase and flippase activity of MprF was unexpected and points to a novel resistance mechanism against daptomycin, which warrants further in-depth investigation. MprF is the first bacterial phospholipid flippase to have been described, but its mode of action remains only superficially understood. Our study reveals critical details of its role in a novel resistance mechanism with important implications for basic bacterial membrane-associated processes and for the development of inhibitors which may block DAP-R to maintain the efficacy of this important therapeutic compound.

MATERIALS AND METHODS
Bacterial strains and mutagenesis of mprF. The common laboratory strain, methicillin-susceptible S. aureus SA113 (ATCC 35556) and its mprF knockout derivative SA113⌬mprF have been described recently (17). Point mutations in mprF were introduced by site-directed mutagenesis in Escherichia coli using E. coli/S. aureus shuttle vector pRB474 bearing mprF via the use of a QuikChange kit (Stratagene, La Jolla, CA, USA) (see Table S3 in the supplemental material), as described recently (20). Mutated derivatives of mprF were cloned in pRB474mprF and transferred into strain SA113⌬mprF. Expression of the pRB474mprF variants was mediated by the use of constitutive Bacillus subtilis promoter vegII. Plasmids were maintained with 10 g/ml chloramphenicol in all studies, with the exception of the MIC assays. Plasmids used in this study are given in Table S3. Primers used in this study are given in Table S4.
Determination of susceptibility to antimicrobial agents. The MICs of daptomycin, bacitracin, polymyxin B, vancomycin, and oxacillin were determined with MIC test strips from Liofilchem according to the manufacturer's advice. The MICs of friulimicin B, nisin, and gallidermin were determined by broth microdilution in Mueller-Hinton broth (MHB) in a 24-well plate under shaking conditions. Friulimicin B and daptomycin MICs were determined in the presence of 50 mg/liter CaCl 2 . Other MICs were determined in the presence of 50 mg/liter CaCl 2 when indicated.
Translocation of LysPG. The distribution of LysPG in the inner leaflet and outer leaflet of the membrane was determined as described recently (26). Briefly, S. aureus overnight cultures were diluted 1:100 and grown for 12 h in brain heart infusion (BHI) medium. Cells were harvested and washed several times, and the cell pellet was incubated with the membrane-impermeative, amino-reactive dye fluorescamine (0.52 M) to specifically label outer-leaflet LysPG. The reaction was stopped after 30 s, and after several washing steps, the phospholipids were extracted and separated in two dimensions via thin-layer chromatography. Fluorescamine-labeled outer-leaflet LysPG was identified with a UV lamp, while unlabeled inner-leaflet LysPG was identified with amino-group reactive ninhydrin. Both lipid species were extracted from the TLC plates and digested with perchloric acid for 3 h in order to liberate and quantify the phosphate content with a colorimetric agent and to quantify the phospholipid content spectrophotometrically at a wavelength of 660 nm.
Quantification of D-alanine from teichoic acids. Bacteria were grown to early stationary phase in basal medium (BM) complemented with 0.36% glucose for 6 h and washed twice with ammonium acetate buffer (20 mM, pH 4.8, 4°C) as described recently (38). A 1-ml volume of a suspension with an OD 600 of 30 was incubated with NaOH (0.1 M; final volume of 100 l) for 1 h of shaking at 37°C to hydrolyze the D-alanine esters. HCl (100 l) served as a stopping reagent, and cell debris was removed by centrifugation and sterile filtration. The D-alanine content of the teichoic acid polymers was assayed by high-performance liquid chromatography (HPLC) upon precolumn derivatization of the amino acid by the use of ortho-phthalaldehyde (OPA). The sample and reagent (OPA diluted 1:10 in 1 M sodium borate buffer, pH 10.7) were drawn into the autosampler injection needle (Agilent 1200 HPLC system; Waldbronn, Germany) and shaken for 90 s before injection. The amino acid derivatives were separated on a reversed-phase column (Grom-Sil OPA-1; Alltech-Grom GmbH, Rottenburg-Hailfingen, Germany) (150 mm by 4.6 mm, 3-m pore size) at a flow rate of 1.1 ml/min using a linear-gradient elution from 0% to 60% buffer B for 15 min and were detected at 340 nm. Buffer A was 25 mM phosphate buffer (pH 7.2) containing 0.75% tetrahydrofuran (THF), while buffer B was composed of 35% MeOH-15% acetonitrile (ACN)-25 mM phosphate buffer. A minimum of three independent runs were performed. Peak areas were quantified based on a D-alanine standard curve.
Repulsion of cationic cytochrome c. Differences in the bacterial capacity to repulse cationic proteins were determined by comparing the levels of binding of the red-colored cationic protein cytochrome c as described previously (18,39). Exponential-phase bacteria were harvested and washed twice with sodium acetate buffer (20 mM, pH 4.6), and the bacterial cell suspension was adjusted to an OD 600 of 3. Aliquots of 1.5 ml were pelleted, resuspended in 750 l cytochrome c solution (Sigma; 0.25 mg/ml in sodium acetate buffer), and incubated at 37°C with shaking for 15 min. Suspensions were pelleted, the resulting supernatant was diluted 1:5 with sodium acetate buffer, and absorbance was measured at 410 nm.
Binding of annexin V to negatively charged phospholipids. To validate the experimentation of the phospholipids, particularly for the assay examining translocation of LysPG, we performed the annexin V-Ca ϩϩ assay, which measures the levels of binding to phosphatidyl serine present on the outer layer of cell membrane ("flipped") (40). This assay has been very commonly used in eukaryotic systems to unravel apoptotic reactions, because of the ability of annexin V-Ca ϩϩ to bind to and demonstrate the translocation of phosphatidylserine in the outer layer of the CM. We utilized this method as an indirect measure of the relative levels of outer CM-flipped, positively charged LysPG (the higher the level of positively charged LysPG that is flipped to the outer CM, the lower the level of negatively charged PL species that are available for annexin V-Ca ϩϩ binding) (40)(41)(42). Briefly, S. aureus cells were grown overnight in BHI broth. Postcentrifugation, the cell pellet was washed twice and resuspended in binding buffer to adjust the OD 600 to 0.5 (ϳ10 8 CFU/ml). A 5-l volume of allophycocyanin (APC) annexin V was added to the cells, and the cells were subjected to gentle vortex mixing and incubated at room temperature for 15 min in the dark (30). The cells were then quantified by flow cytometry for analysis of surface-bound fluorophore (30) (excitation and emission wavelengths of 650 nm and 660 nm, respectively; 10,000 events acquired). Data are represented in relative fluorescent units.
MprF domain interactions. MprF-domain interactions were analyzed with a bacterial two-hybrid kit (BACTH system kit; Euromedex), as described recently (20). Briefly, E. coli BTH101 was transformed with mprF variants (Table S1) and fused to adenylate cyclase fragments T25 and T18 of Bordetella pertussis, and protein interactions resulting in cyclic AMP (cAMP) production and subsequent expression of the lac and mal operons in E. coli were quantified by determining ␤-galactosidase activity in triplicate (20).
FIG S1, EPS file, 0.1 MB.  We declare that we have no conflict of interest.