Biochemical Characterization of Cardiolipin Synthase Mutations Associated with Daptomycin Resistance in Enterococci

Abstract

Daptomycin (DAP) resistance in enterococci has been linked to mutations in genes that alter the cell envelope stress response (CESR) (liaFSR) and changes in enzymes that directly affect phospholipid homeostasis, and these changes may alter membrane composition, such as that of cardiolipin synthase (Cls). While Cls substitutions are observed in response to DAP therapy, the effect of these mutations on Cls activity remains obscure. We have expressed, purified, and characterized Cls enzymes from both Enterococcus faecium S447 (residues 52 to 482; Cls447a) and Enterococcus faecalis S613 (residues 53 to 483; Cls613a) as well as Cls variants harboring a single-amino-acid change derived from DAP-resistant isolates of E. faecium. E. faecium Cls447a and E. faecalis Cls613a are tightly associated with the membrane and copurify with their substrate, phosphatidylglycerol (PG), and product, cardiolipin (CL). The amount of PG that copurifies with Cls is in molar excess to protein, suggesting that the enzyme localizes to PG-rich membrane regions. Both Cls447a^H215R and Cls447a^R218Q showed an increase in V_max (μM CL/min/μM protein) from 0.16 ± 0.01 to 0.26 ± 0.02 and 0.26 ± 0.04, respectively, indicating that mutations associated with adaptation to DAP increase Cls activity. Modeling of Cls447a to Streptomyces sp. phospholipase D indicates that the adaptive mutations Cls447a^H215R and Cls447a^R218Q are proximal to the phospholipase domain 1 (PLD1) active site and near the putative nucleophile H217. As mutations to Cls are part of a larger genomic adaptation process, increased Cls activity is likely to be highly epistatic with other changes to facilitate DAP resistance.

INTRODUCTION

Enterococci are important pathogens responsible for serious infections, such as urinary tract infections, bacteremia, skin and soft tissue infections, endocarditis, and meningitis, and they are a major contributor to nosocomial infections in the United States (1, 2). Enterococcus faecalis and E. faecium are responsible for the majority of clinical infections, and their treatment has become increasingly problematic as more than 80% of E. faecium isolates are now resistant to ampicillin and vancomycin (1). In 2008, Rice (3) defined the “no ESKAPE” pathogens (for E. faecium, Staphylococus aureus, Klebsiella pneumoniae, Acinetobacter baumanii, Pseudomonas aeruginosa, and Enterobacter spp.) as organisms associated with U.S. hospital infections for which new antibiotics are urgently needed (4). Fortunately, ampicillin resistance in E. faecalis remains rare, but increased resistance to other antibiotics, such as aminoglycosides and vancomycin, is likely to continue with their increased use (1, 5). Vancomycin-resistant enterococci (VRE) present a substantial clinical threat, and FDA-approved therapies such as quinupristin-dalfopristin and linezolid have important limitations due to toxicity and bacteriostatic effect which may lead to therapeutic failure.

Daptomycin (DAP) is a cyclic lipopeptide antibiotic that received approval by the FDA in 2003 for treatment of Gram-positive infections in skin and soft tissue (5) and subsequently was approved for S. aureus bacteremia and right-sided endocarditis in 2006 (6). DAP has demonstrated rapid in vitro bactericidal activity against clinically significant strains of Gram-positive bacteria, including VRE, methicillin-resistant Staphylococcus aureus (MRSA), glycopeptide intermediate-susceptible S. aureus (GISA), penicillin-resistant Streptococcus pneumoniae (PRSP), and coagulase-negative staphylococci (CoNS) (7–11). While DAP susceptibility among enterococci remains >99%, continued use of DAP will likely increase the frequency of resistance worldwide (12). Several studies have already reported the development of resistance in both S. aureus (6, 13–16) and enterococci (2, 17–19). Recent evidence strongly suggests that DAP inserts into the plasma membrane in a calcium-dependent manner and subsequently disrupts the functional integrity of the cell membrane, altering cell division. Fluorescence studies have shown that DAP forms membrane-associated oligomers on liposomes and bacterial cells, suggesting that it is phosphatidylglycerol (PG)-dependent oligomerization of DAP that contributes to its effectiveness in altering membrane homeostasis (20). In S. aureus, DAP resistance was correlated with mutations to lysylphosphatidylglycerol synthetase (MprF), again suggesting a specific relationship between PG and DAP susceptibility (21, 22). DAP appears to have a multifaceted effect on cell wall homeostasis that includes leakage of small ions, such as intracellular potassium, leading to loss of membrane potential, and by changes to the cell envelope ultrastructure leading to rapid cell death (22, 23).

Recently, genetic changes responsible for DAP resistance in enterococcal species have been identified using comparative genomic sequencing of clinical strain pairs (20, 24–26). DAP resistance was associated with mutations in genes encoding (i) components of the LiaFSR three-component regulatory cell envelope stress response pathway and (ii) phospholipid biosynthesis enzymes cardiolipin synthase (Cls) and GdpD (glycerophosphodiesterase) (20, 24, 25). Tran et al. (26) have shown recently that exchange of the susceptible cls allele for the resistant one belonging to E. faecium R446 (encoding Cls^R218Q) did not affect DAP susceptibility, supporting the idea that changes to Cls function in a broader context with other adaptive changes, such as alterations in the LiaFRS signaling pathway (20, 24, 25). Interestingly, Cls substitutions have also been observed in S. aureus strains that have developed nonsusceptibility to DAP during therapy, suggesting that changes in Cls are important mediators of DAP resistance (15).

Cls is a membrane-associated protein comprised of two putative transmembrane helices and a pair of phosphatidylcholine-hydrolyzing phospholipase domains, or PLDs (27). PLD superfamily members have one or two conserved HKD motifs at the active site with a canonical spacing of HxK(x)₄D (28) (see Fig. S1 in the supplemental material). Cls catalyzes the formation of cardiolipin (CL) from two PG molecules. Based upon its homology to phospholipase D, it is likely that the Cls active site is made up of functional groups contributed from both PLD1 and PLD2 with His217 as the putative active-site nucleophile (29, 30). As shown in Fig. 1, Cls mutations associated with DAP resistance in enterococci are located predominately in three regions of the Cls primary sequence: (i) an N-terminal transmembrane helical region, (ii) a short linker region joining the two N-terminal transmembrane helices to the PLD1/2-containing C-terminal region, and (iii) a region proximal to the PLD1 catalytic site. Recently, we have shown (31) that development of resistance to DAP in both E. faecium S447 and E. faecalis S613 is accompanied by distinct changes in phospholipid composition that include a decrease in PG and an increase in a glycerophosphoglycolipid, identified as glycerolphospho-diglycodiacylglycerol (GP-DGDAG). However, no substantial change in cardiolipin content was identified (31). Moreover, using an allelic replacement approach, we also showed that introduction of a gene encoding a Cls with an R218Q substitution did not affect DAP susceptibility in a clinical strain of E. faecium (26), suggesting that Cls mutation alone is not sufficient for DAP resistance.

Fig 1.

Schematic of Cls domain organization indicating relative positions and sequences of mutations associated with DAP resistance in enterococci. The wild-type Cls sequence is at the top of each list, followed by the mutations identified from comparative sequencing (underlined) using the numbering of Arias et al. (24). Adaptive mutations to Cls are found in PLD1, the linker region joining the two putative transmembrane helices to PLD1/2, and an N-terminal transmembrane helical region. The H215R mutation is located one amino acid away from the conserved signature motif of PLD1, while R218Q occurs between the conserved H217 and K219 residues required for catalytic activity in other organisms.

To date, biochemical characterization of Cls mutations associated with DAP resistance has not been performed. Moreover, physicochemical characterization of bacterial cardiolipin synthases has been hindered by the inability to produce sufficient amounts of active and highly purified protein for study. Previous reports are based on observations made with crude membrane or partially purified preparations of Cls (32–35). In this study, we report the expression, purification scheme, and characterization of functional cardiolipin synthases from E. faecalis (Cls613) and from E. faecium (Cls447) as well as the mutants Cls447^H215R and Cls447^R218Q, which are associated with DAP resistance. A comparison of the catalytic activities of E. faecium Cls447a to those of Cls447a^H215R and Cls447a^R218Q shows that mutations associated with DAP resistance increase Cls activity.

MATERIALS AND METHODS

3-[(3-Cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS), n-decyl-β-d-maltopyranoside (DM), n-dodecyl-β-d-maltoside (DDM), n-octyl-β-d-glucopyranoside (OG), cyclohexylpentyl-β-d-maltoside (Cy5), n-octylphosphocholine (FOS-8), n-nonylphosphocholine (FOS-9), n-decylphosphocholine (FOS-10), n-undecylphosphocholine (FOS-11), n-dodecylphosphocholine (FOS-12), octylphospho-N-methylethanolamine (FOS-mea-8), and decylphospho-N-methylethanolamine (FOS-mea-10) were purchased form Affymetrix, Inc. (Santa Clara, CA). Triton X-100, nonylphenoxypolyethoxylethanol (NP-40), and n-decyl-N,N-dimethyl-3-ammonio-1-propanesulfonate (Zw 3-10) were purchased from Sigma-Aldrich. l-α-Phosphatidylglycerol sodium salt from chicken egg (PG) and cardiolipin sodium salt from bovine heart (CL) were obtained from Avanti Polar Lipids Inc. (Alabaster, AL).

Expression of Cls and Cls variants.

Constructs designed for the expression of Cls and Cls variants were made synthetically (Gene Script, Piscataway, NJ) using codons optimized for expression in E. coli and did not include the two putative transmembrane helices. The resulting constructs were synthesized and designated Cls613a (residues 53 to 481 of Cls from E. faecalis S613) and Cls447a (residues 52 to 482 of E. faecium S447). Sequences for Cls variants were obtained from E. faecalis R712 (MIC of 12 μg/ml), E. faecium R446 (MIC of 32 μg/ml), and E. faecium R499 (MIC of 48 μg/ml) (24). Synthetic genes were subcloned between the NcoI/HindIII sites of the pUC57 (Gene Script, Piscataway, NJ) multicloning region. Cls613a and Cls447a also included a His tag and tobacco etch virus (TEV) protease cleavage site to facilitate purification. For expression, Cls613a and Cls447a were subcloned into the NcoI/HindIII sites of pET28a (Novagen, NJ) and expressed in Escherichia coli Rosetta (Novagen Inc., Madison, WI) or BL21(DE3) Star cells (Life Technologies, Grand Island, NY). Site-directed mutagenesis of wild-type Enterococcus faecium Cls447 (with substitutions R218Q and H215R and insertion of MPL at position 110 derived from putative Cls enzymes present in DAP-resistant isolates of E. faecium [24]) was performed using the Stratagene QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) by following the manufacturer's protocol. All constructs were confirmed by DNA sequencing.

E. coli Rosetta cells containing pET28a(Cls613a) and pET28a(Cls447a) were grown at 37°C in M9 minimal medium (supplemented by 1 mM MgSO₄, 0.1 mM CaCl₂, 0.5% [wt/vol] glucose, 2 × 10⁻⁴% [wt/vol] thiamine) containing 50 mg/ml kanamycin and 25 mg/ml chloramphenicol to mid-log phase (optical density at 600 nm [OD₆₀₀] of 0.5 to 0.6) and induced by addition of isopropyl β-d-1-thiogalactopyranoside (IPTG) to 0.4 mM for 18 h at 16°C.

Purification of E. faecium Cls447a, Cls447a^H215R, and Cls447a^R218Q.

Frozen cell pellets were resuspended in buffer A consisting of 50 mM Tris, pH 7.4, 0.5 M NaCl, 20 mM imidazole, 10 mM β-mercaptoethanol (BME), 0.2 mM phenylmethanesulfonyl fluoride (PMSF), 20% (vol/vol) glycerol, and EDTA-free Complete protease inhibitor cocktail tablet (1 tablet/50 ml extraction solution; Roche Diagnostics Corp., Indianapolis, IN). Cells were disrupted using an EmulsiFlex-C3 high homogenizer (Avestin) at 17,000 lb/in². The lysate was centrifuged for 60 min at 34,000 rpm (4°C). The supernatant was then applied to a 5-ml HiTrap affinity (Ni²⁺) column (GE Healthcare Life Sciences). The column was extensively washed with buffer A and eluted with a step elution gradient from 20 to 500 mM imidazole. The fractions containing Cls were pooled and dialyzed against 50 mM Tris, pH 8.5, 100 mM NaCl, 10 mM BME, 0.2 mM PMSF, 1 mM EDTA, and 20% (vol/vol) glycerol overnight at 4°C. The dialysate was applied to a 5-ml HiTrap Q-XL Sepharose column equilibrated in 20 mM Tris, pH 8.5, 100 mM NaCl, 0.2 mM PMSF, 5 mM BME, and 1 mM EDTA (pH 8.0). The protein was eluted with a linear gradient of 0.1 to 1 M NaCl. The peak fractions were pooled, concentrated, and loaded onto a Superdex 200 column (HiLoad 16/60; GE Healthcare) for the final purification step, and the peak fractions were pooled and retained for further analysis. Sample purity was >95% as judged by SDS-PAGE.

Purification of E. faecalis S613 Cls613a.

Membrane vesicles were prepared from cell pellets resuspended in buffer A, followed by disruption using an EmulsiFlex-C3 high homogenizer (Avestin) at 17,000 lb/in² in order to reduce membrane damage. After the removal of any unbroken cells by low-speed centrifugation, membranes were pelleted by ultracentrifugation for 2 h at 34,000 rpm (4°C). The efficacy of detergent extraction was tested using detergents of various properties (CHAPS, DM, DDM, OG, FOS-12, Triton X-100, NP-40, Zw 3-10, and Cy5) (Affymetrix, Inc.) by taking the protein pellet after ultracentrifugation and resuspending it in buffer A containing the appropriate detergent at a concentration of 1 to 2% (wt/vol). Solubilization efficiency in various detergents was assessed by SDS-PAGE analysis of the amount of protein extracted from the high-speed spin pellet (see above). Ni-nitrilotriacetic acid (NTA) agarose purification was not recommended at concentrations of detergent higher than 2%, and therefore the concentration of detergents was limited to a final detergent concentration of ≤2% (wt/vol). Detergents with various physical properties, including CHAPS, DM, DDM, OG, FOS-12, Triton X-100, NP-40, Zw 3-10, and Cy5, were tried (36, 37). We determined a critical solubilization concentration (CSC) required to disrupt a membrane system into a predominantly micellar dispersion (37). SDS-PAGE was used to monitor the efficiency of solubilization at 23 and 4°C as a function of time and FOS-12 concentration. Resuspensions were rocked slowly overnight at 23 and 4°C before being centrifuged at 34,000 rpm to remove insoluble protein. Detergent-solubilized protein was then quantitated by SDS-PAGE analysis. Purification of E. faecalis Cls613a required extraction from an insoluble fraction using FOS-12 at 0.5% (wt/vol). Solubilized Cls613a was applied to 5 ml Ni-NTA resin (Qiagen) per 10 ml of supernatant, preequilibrated with 5 column volumes of binding buffer A containing 0.5% (wt/vol) FOS-12. The supernatant containing the solubilized Cls613a was mixed with the preequilibrated NTA resin and allowed to bind for at least 4 h by gentle mixing at 4°C. Nonspecifically bound proteins were removed from the resin by two washes of buffer A that included 30 and 40 mM imidazole and 0.1% (wt/vol) FOS-12. The column was then washed and the protein eluted with buffer A containing 500 mM imidazole and 0.1% (wt/vol) FOS-12. Following concentration (Vivaspin 20 at a 10,000 molecular weight cutoff), Cls613a was applied to a Superedex-200 column (HiLoad 16/60; GE Healthcare) for further purification. Sample purity was assessed by SDS-PAGE to be >95%.

Determination of Cls activity.

The assay mixture consisted of the reaction buffer (50 mM Tris, pH 7.4, 20% [vol/vol] glycerol) and various concentrations of PG (0, 0.65, 1.29, 2.59, 3.89, 5.19, 6.48, and 9.08 mM). Reactions were started by addition of Cls to a final concentration of 60 μM in a final volume of 7 μl. Samples were incubated at 37°C for 60 min and applied to silica gel plates for thin-layer chromatography (TLC). The TLC plate was developed with chloroform-hexane-methanol-acetic acid (50:30:10:5, vol/vol/vol/vol). The lipids were stained with iodine vapor. Each plate contained PG and CL standards to permit normalization of staining and quantitation across TLC plates. Spot intensities and local background corrections were made using the Image J software package (38). Select TLC plates contained standards of different cardiolipin concentrations for more precise estimation of CL from the intensity data. The relationship between the intensity of the spots and each standard concentration was used to estimate CL formation (μM). The kinetic parameters (K_m and V_max) were determined by fitting the data to the Michaelis-Menten equation using a nonlinear least-square fitting routine in GraphPad Prism version 5.01 (GraphPad Software, La Jolla, CA). Data shown are the averages from 5 experiments, performed for each mutant and the wild type, plus standard errors.

Secondary-structure estimation of Cls by CD.

Circular dichroism (CD) studies of E. faecium Cls447a and E. faecalis S613 were performed with a Jasco J-810 spectropolarimeter equipped with a Peltier temperature controller. CD measurements from 200 to 260 nm were performed at 5 μM protein in two buffer systems, buffer A (10 mM NaP, pH 7.4, 0.1% FOS-12 [wt/vol], 20% glycerol [vol/vol] for Cls613a) and buffer B (10 mM NaP, pH 7.4, 20% glycerol [vol/vol] for Cls447a) using a 0.1-cm-path-length quartz cuvette. The appropriate buffer absorbance values were recorded and the CD spectra were baseline corrected by subtraction of the appropriate buffer (buffer A or B). All measurements were performed in triplicate. The secondary-structure estimation from CD spectra was performed with the program K2D2 (39).

RESULTS

Expression and purification of E. faecium Cls447a and E. faecalis Cls613a.

Expression of Cls613a and Cls447a was robust at 37°C, but the majority of the protein was insoluble. At 16°C, about 17% of the expressed Cls447a protein was readily soluble and active, while the remaining material stayed strongly associated with the membrane fraction (see below). Cls447a expression produced substantially more soluble protein than Cls613a. Purification of Cls447a routinely produced >500 μg of Cls/liter of cell culture. As shown in Fig. 2a, after gel filtration chromatography, Cls447a was more than 95% pure and did not require exogenous lipids or detergent to retain activity and solubility up to 10 mg/ml. Overproduction of soluble Cls447a^R218Q and Cls447a^H215R was less efficient due to a smaller fraction of soluble protein, yielding about 350 and 30 μg protein/liter of culture, respectively. The purities of the expressed proteins Cls447a, Cls447a^R218Q, and Cls447a^H215R were assessed to be greater than 95% by SDS-PAGE (Fig. 2a). Interestingly, purified Cls from the soluble fraction copurified with lipids whose mobility by TLC was identical to that of PG and CL (Fig. 3). Using commercially available CL and PG as mobility and intensity standards, we estimated the ratio of Cls to copurifying lipids to be ∼1:1:8 (Cls:CL:PG).

Fig 2.

Expression and purification of E. faecium 447 Cls447a (residues 52 to 482) and E. faecalis S613 Cls613a (residues 53 to 483). (A) SDS-PAGE analysis of purified Cls447a, Cls447a^H215R, and Cls447a^R218Q. Cls447a and the mutants have the expected molecular mass of 51.5 kDa. (B) SDS-PAGE analysis of purified Cls613a. (C) Subcellular fractionation by differential centrifugation of total cell extract indicated that ∼17% of Cls447a from E. faecium was present in the high-speed supernatant (HSS), while approximately 96% of E. faecalis Cls613a was found in the high-speed pellet (HSP) and could only be purified in large amounts by detergent extraction with 0.5% (wt/vol) FOS-12.

Fig 3.

Total activity of wild-type E. faecium Cls447a and variants was measured using thin-layer chromatography (TLC). The major lipid formed was identified as CL by cochromatography with commercial standards using a chloroform/hexane/methanol/acetic acid system. Lipids were stained with iodine vapor. The Image J software package was used to quantify spot intensity. Lane 1 is an equimolar mixture of PG and CL; lanes 2 to 9, CL standards at various concentrations (mM); lanes 10 to 17, Cls447a activity assay as a function of increasing PG concentrations at 37°C. Addition of exogenous PG produces new CL, demonstrating that Cls has retained activity. As shown in lane 10, purified Cls447a copurifies with E. coli CL and PG.

In contrast to Cls447a from E. faecium, E. faecalis Cls613a could only be isolated at scales of ≤100 μg/liter of induced cell culture (Fig. 2b). We estimated that 4% of expressed Cls613a was soluble and required detergent extraction of the insoluble membrane pellet by 0.5% (wt/vol) FOS-12 to produce amounts sufficient for CD and activity assays (Fig. 2c). A range of detergents at various concentrations was evaluated. Surprisingly, of nine different detergents surveyed, only FOS-12, a zwitterionic detergent dodecylphosphocholine, was able to modestly solubilize Cls613a. Similar detergents containing a comparable phosphate group (FOS-8, FOS-9, FOS-10, FOS-11, FOS-mea-8, and FOS-mea-10) were also able to solubilize Cls613a but at much lower efficiency, presumably due to the shorter length of their alkyl chains. Based on our detergent results and the structural studies of phospholipase D from Streptomyces spp. with different phospholipid substrates (29), FOS-12 may mimic binding of the substrate, making it well suited to solubilization or refolding of a small amount of the membrane-associated Cls613a in the insoluble pellet. This is consistent with our finding that FOS-12 is an inhibitor of Cls activity (see the next section). Although the overall yield was more than 4-fold less than that of E. faecium Cls447a, Cls613a from E. faecalis could be extracted for purification at 4°C with 12 to 18 h of extraction in 0.5% (wt/vol) FOS-12. Purification of an E. faecium 447 Cls mutant with an insertion of MPL at residue 110 (Cls447^MPLin) was unsuccessful due to very low expression and solubility. Unlike the case for E. coli Cls, 1% Triton X-100 was unable to solubilize either Cls447a or Cls613a.

In vitro enzymatic activity of purified E. faecium Cls447a and variants associated with daptomycin resistance.

Physicochemical characterization of cardiolipin synthases has been hampered by difficulties with their expression and purification. Previous studies have made use of crude membrane or partially purified preparations of Cls using a mixed micelle assay (32–35). Purification of highly purified protein provided the opportunity to use in vitro kinetics to assess the role of Cls and adaptive variants Cls447a^R218Q and Cls447a^H215R in altering cardiolipin synthesis.

To establish the optimal assay conditions for this enzyme and understand its basic properties, Cls activity was measured as a function of pH, divalent metals, and temperature. The optimum pH was fairly broad, with a maximum around pH 7.5, and was comparable to many other phospholipid-synthesizing enzymes (32, 33, 35). The divalent metals Mg²⁺, Co²⁺, and Cd²⁺ had no measurable effect on enzyme activity at 4 mM, indicating that metal ions are not required for activity. The activity of E. faecium Cls447a was inhibited by FOS-12 at 0.1% (wt/vol), although it proved to be quite effective at resolubilizing Cls447a and Cls613a. Presumably, the similarity of FOS-12 to PG makes it a good competitive inhibitor of activity. The optimal temperature for Cls447a activity was 37°C and was stable for up to 1 h at 45°C but was lost within 10 min at 55°C.

After optimization of the assay conditions, the total activity of Cls447a and the purified variants was measured using TLC (Fig. 3) in quintuplicate. Analysis of variance was used to perform a pairwise comparison of the product quantification made from the TLC plate spot intensities for wild-type Cls447a, Cls447a^R218Q (P = 0.12), and Cls447a^H215R (P = 0.014) (see Fig. S2 in the supplemental material). The major lipid formed was identified as CL by cochromatography with a CL standard. The relationship between the intensity of the spots and each standard concentration was then used to determine CL formation (μM) and estimate the kinetic parameters (K_m and V_max). As shown in Fig. 4 and Table 1, the increase in CL production was more pronounced in Cls447a^H215R, while Cls447a^R218Q still demonstrated a trend, albeit weak, toward increased activity in the final estimation of V_max. Unlike the soluble Cls447a, Cls613a could only be isolated in large amounts by extraction of the high-speed ultracentrifugation pellet with FOS-12. However, Cls613a could not be readily assayed as FOS-12 inhibits its enzymatic activity.

Fig 4.

Enzyme activities of adaptive mutants Cls447a^R218Q and Cls447a^H215R are increased relative to wild-type Cls447a. The initial velocity (labeled Activity) is defined as the concentration of CL (μM) produced in 1 min/μM of protein at various PG (mM) concentrations. Symbols represent the averages from the assay at each concentration of PG, and the error bars show the standard errors. Plots of initial velocity versus [PG] were fit to the Michaelis-Menten equation, and the fitted curves are presented as solid lines. All assays were performed in quintuplicate.

Table 1.

Activity of Cls447a and variants associated with DAP resistance at 37°C

Protein	Km (mM)	Vmax (μM CL/min/μM protein)
Cls447a	3.55 ± 0.58	0.16 ± 0.01
Cls447aR218Q	6.25 ± 1.76	0.26 ± 0.04
Cls447aH215R	3.66 ± 0.84	0.26 ± 0.02

Since we could not purify sufficient amounts of Cls613a, catalytic activity was tested from the small amount of purified protein that was soluble in the absence of FOS-12 (∼30 μg). We were able to show that Cls613a was able to synthesize CL from PG. Like Cls447a, Cls447a^R218Q, and soluble Cls447a^H215R, Cls613a copurified with PG and CL.

Cls447a and Cls613a are strongly associated with the cell membrane.

Despite removal of the two putative N-terminal transmembrane helices, we noted that active Cls447a enzyme was still strongly associated with membrane fractions during purification. To verify that Cls447a retained its association with the membrane, subcellular fractionation by differential centrifugation of total cell extracts was performed. As shown in Fig. 2C, E. faecalis Cls613a was found mostly exclusively (96%) in the high-speed pellet, and in the case of E. faecium Cls447a about 83% of protein was associated with membranes after centrifugation at 33,000 rpm. To directly determine the nature of this membrane association, an integral membrane protein assay was performed (40). Membrane fractions were treated with 0.1 M Na₂CO₃ at pH 11, 2 M urea, or 1% SDS. SDS (1%) typically releases integrally bound proteins from the membrane (40, 41), while milder extraction conditions are more consistent with peripheral proteins. Only treatment with SDS could quantitatively release Cls447a and Cls613a from the membrane (see Fig. S3 in the supplemental material). Taken together, our results indicate that both of the proteins Cls613a and Cls447a are strongly associated with the lipid bilayer even after the removal of hydrophobic membrane-spanning regions.

Estimation of secondary structure and folding of E. faecium Cls447a and E. faecalis 613a by CD.

CD was used to estimate protein folding and secondary-structure content of purified Cls447a and Cls613a (42). As shown in Fig. 5, Cls447a and Cls613a have UV-CD spectra characteristic of a predominantly alpha-helical structure. The alpha-helical content of Cls447a was estimated to be 21.5% using the program K2D2 (39). Cls613a extracted with FOS-12 also exhibited a strong secondary structure and was estimated to have about 20.5% alpha-helical content, in good agreement with that observed for Cls447a. The strong secondary structure of the FOS-12-extracted Cls613a suggests that the protein is folded, as it did not display behavior of very low ellipticity above 210 nm that is often characteristic of disordered proteins (43).

Fig 5.

Circular dichroism suggests E. faecium Cls447a and E. faecalis Cls613a are well folded and have largely alpha-helical secondary structures. Purified Cls447a (dotted line) and Cls613a (continuous line) (5 μM) were estimated to be about 21.5 and 20.5% α-helix, respectively. Purification of E. faecalis Cls613a required extraction with 0.5% FOS-12, therefore the spectrum of Cls613a was performed in the presence of FOS-12.

Structural modeling of the E. faecium Cls447a active site.

In the absence of a high-resolution structure for Cls, homology modeling using Phyre 0.2 (44) to phospholipase D from Streptomyces sp. (Protein Data Bank accession code 1v0w) was used to provide a structural context for the daptomycin-associated mutants Cls447a^H215R and Cls447a^R218Q (Fig. 6). Successful structural modeling relies in great part on the extent of homology between the template structure and the sequence being examined. The structural consensus of the region spanning Cls447a PLD1 and PLD2 (residues 106 to 463) to Streptomyces species (1v0w) and Salmonella enterica (1bys) phospholipase D is very strong. Structural alignment of the Cls447a sequence to the Streptomyces species (1v0w) phospholipase D structure produced a model with an overall RMSD of 1.55 Å (root-mean-squared deviation of the model of 1v0w) and an E value of 7.588806e⁻¹⁸, indicative of an excellent fit of the structural template to the sequence query (Fig. 6A) (44, 45). Based upon the model, the Cls447a active site is made up of contributions from both PLD1/2 with His217 (the conserved His in the HKD motif of PLD1) in position as the active-site nucleophile and His399 from PLD2 in position to act as a general acid protonating the leaving group (28, 29) (Fig. 6B). However, in the absence of a crystallographic or nuclear magnetic resonance structure of Cls447a and more kinetic studies, the role of each PLD in the reaction mechanism cannot be resolved readily.

Fig 6.

Structural model for the E. faecium 447 Cls active site showing the positions of the active-site mutations H215R and R218Q. (A) Based upon the strong sequence homology within the PLDs of Streptomyces sp. phospholipase D (1v0w) and E. faecium 447 Cls, a structural model for Cls447a was built using the program Phyre and then manually aligned in CCP4MG (9, 23, 37). The E value between E. faecium Cls447 and Streptomyces sp. phospholipase D was 7.588806e⁻¹⁸. (B) The H215R and R218Q mutations (red sticks) are proximal to the active site, including the putative catalytic residue His-217 that makes up part of the conserved HKD motif of the PLD family. Wild-type sequence is shown as green sticks. The position of the substrate phosphate from Leiros and coworkers is shown in red (40).

DISCUSSION

Cardiolipin synthase has been shown to have an important role in modulating the physical properties of membranes in response to environmental changes (46). CL promotes formation of membrane subdomains with unique properties that can serve to localize and regulate assembly of protein complexes with important biological consequences for cell division and membrane transport. The synthesis of CL from PG has the potential to alter membrane properties directly thorough the synthesis of CL or, more indirectly, by altering the amount of PG in the membrane. In response to DAP, mutations in Cls have been identified in many species, including E. faecalis S613, E. faecium 447, and S. aureus (15, 24, 25). Mishra et al. (31) have shown that development of DAP resistance in both E. faecalis and E. faecium are associated with a significant reduction in PG cell-membrane content. Indeed, in E. faecium 447 a 57% decrease in PG levels was observed, and while strain 447 has other genomic changes (26), there is a consistent pattern across both enterococcal and staphylococcal strains for mutation in cardiolipin synthase that affect DAP susceptibility. In E. faecium an allelic replacement of the wild-type Cls gene for with of Cls^R218Q did not increase DAP resistance, consistent with a more indirect role of Cls in DAP adaptation (26). Isolation and purification of the E. faecium adaptive Cls variants Cls447a^H215R and Cls447a^R218Q indicate that these adaptive mutations increase Cls activity, although the trend to higher activity is stronger for Cls447a^H215R than Cls447a^R218Q. Since Cls447a copurifies with an ∼8-fold molar excess of E. coli PG and about one equivalent of CL, the concentration of copurifying lipids is ∼0.5 mM and may lead to a modest underestimation of Cls activity for all of the proteins over the range of our kinetic assays (0.65 to 9.08 mM). In vivo, the pool of PG is substantially reduced in E. faecium and E. faecalis in response to DAP challenge (31), and the adaptive changes observed in Cls that increase its V_max by a modest 1.6-fold could compensate for the decreased substrate (PG) and lead to a restoration of the CL pool.

Using Streptomyces sp. phospholipase D as a structural model for PLD1/2 of Cls447a clearly suggests that mutations H215R and R218Q are proximal to the PLD1 active site (Fig. 6B). The side chain at position 215 projects directly into the ligand binding pocket and is within 10 Å of the putative active-site nucleophile, H217, while the side chain of position 218 projects away from the active site. However, any mutation at position 218 could have a strong effect on the positions of the residues at the binding pocket as well as H217, since the substitution of R→Q must be accommodated within a well-packed and only partially solvent-exposed region of the protein. Changes at position 218 could easily lead to shifts in the positions of nearby active-site main-chain residues and consequently their side chains, altering activity. Other adaptive mutations to Cls, such as deletion of Lys at position 61 in a stretch of three Lys residues in E. faecalis (24) and Asn13 to Ile or Ser in E. faecium (24, 25), are within a region that links the transmembrane helices to the PLD1/2 domains and therefore may alter Cls activity through quite different means, although their net effect may still be to increase Cls activity.

Active and well-folded Cls447a retains a very strong membrane association comparable to that of an integral membrane protein and copurifies with CL and PG. Interestingly, Cls447a, Cls447a^H215R, and Cls447a^R218Q all copurify with a molar excess of a lipid species that comigrates with PG standards by TLC (Fig. 3). Using authentic standards, we estimated the molar ratio of PG to Cls447a and CL to be ∼8:1:1, suggesting that Cls brings a small patch of lipid with it during purification. The strong molar excess of PG suggests that Cls localize to PG-rich regions of the membrane. Additional incubation of purified protein with the endogenous lipids does not alter the PG/CL ratio, suggesting that the reaction is at equilibrium. Addition of exogenous PG produces new CL, demonstrating that the purified protein has retained activity (Fig. 3). Using purified Cls, it should be possible to directly test Cls affinity for PG-rich vesicles in vitro. Secondary-structure estimates from CD are consistent with a folded, largely α-helical structure consistent with its homology to phospholipase D. An efficient purification protocol for Cls will now allow further biochemical studies on the Cls enzymatic mechanism as well as structural studies. Taken together, our studies suggest that Cls is tightly associated with the membrane, and that mutations in the enzyme in response to DAP therapy increase its catalytic activity. In turn, increases in Cls activity may act directly on the pool of PG or, since Cls mutations are typically found with other genomic changes, such as changes in liaFRS signaling, it may compensate for earlier changes in membrane biochemistry to increase cellular fitness to DAP. Further work will be required to elucidate the precise molecular mechanism that links Cls mutations to DAP resistance.