Regulation of mprF by Antisense RNA Restores Daptomycin Susceptibility to Daptomycin-Resistant Isolates of Staphylococcus aureus

Abstract

Mutations in mprF have been shown to result in reduced susceptibility to daptomycin and other cationic antibacterials. An mprF antisense-inducible plasmid was constructed and used to demonstrate that depletion of mprF can reestablish susceptibility to daptomycin. Inducing antisense to mprF also resulted in increased susceptibility to vancomycin and gentamicin but, paradoxically, decreased susceptibility to oxacillin. These results suggest that mprF mutations that reduce susceptibility to cationic antibacterials result in a gain-of-function phenotype.

Daptomycin (DAP) is a cyclic lipopeptide that exhibits rapid bactericidal activity against Gram-positive bacteria, including methicillin-resistant and glycopeptide-intermediate Staphylococcus aureus (33). The mechanism of action of DAP involves a calcium-dependent insertion in the bacterial cell membrane, resulting in depolarization and cell death (4, 5, 7, 8). According to the Food and Drug Administration, S. aureus isolates are DAP susceptible (DAP^s) at MICs of ≤1 μg/ml and are DAP resistant (DAP^r) at MICs of ≥2 μg/ml. In vitro, spontaneous emergence of DAP nonsusceptibility is rare (frequency, <10⁻¹⁰) but can occur through serial passage with subinhibitory DAP concentrations (10, 28). However, since its introduction into clinical use, DAP^r S. aureus isolates have been reported in which the MIC increased from 0.5 μg/ml to 2 to 4 μg/ml (1, 9, 11, 12, 18-20, 29, 32, 34).

The precise mechanism of DAP resistance has not been determined, but several genes contributing to this phenotype have been identified, including mprF and yycG (21, 37). mprF (fmtC) encodes a nonessential integral membrane protein that catalyzes the synthesis of lysylphosphatidylglycerol (LPG), a positively charged phospholipid that constitutes up to 38% of the S. aureus membrane (24, 25, 31). It is unlikely that these mprF mutations result in a loss of function, because DAP susceptibility is increased in ΔmprF strains (with a MIC of 0.125 μg/ml versus 0.5 μg/ml for the parent). Instead, it is postulated that the mprF-associated DAP^r phenotype represents a gain of function; this phenotype has been observed for mprF point mutation alleles with other antibacterial agents (10, 13, 21). Although this nonsusceptibility phenotype has not been fully described, it is possible that some of these mutations increase the overall positive charge on the bacterial membrane, resulting in DAP repulsion (10, 13).

Targeted antisense RNA technology is useful for identifying and regulating genes involved in antimicrobial resistance phenotypes (10, 15, 35). The purpose of this study was to demonstrate that modulation of mprF expression using inducible antisense RNA can restore DAP susceptibility in DAP^r S. aureus strains. To do so, mprF and its flanking regions were amplified by PCR, fragmented by sonication, and cloned into the xylose-inducible plasmid pSAX-1E. The plasmid library was transformed into S. aureus RN4220, and clones were screened for xylose-inducible growth sensitivity in the presence of sub-MIC DAP levels. Putative growth-sensitive clones were subsequently screened with various levels of DAP (0 to 1 μg/ml) and xylose (0 to 2%). Sixty-two clones displayed inducible DAP susceptibility. Confirmed by sequencing, all constructs contained mprF fragments in the antisense orientation with good coverage across the gene. Eleven plasmids expressing antisense fragments of <400 bp and located near the 3′ gene terminus were analyzed further. Because targeted antisense RNA attenuates gene expression through posttranscriptional degradation (39), measurement of mprF-specific mRNA levels using Northern analysis provided evidence that the observed phenotype resulted from antisense induction, as shown in Fig. 1. One of the 11 candidate plasmids (pSAX-1EmprF2F2) was used in the study because of its demonstrated specific and marked reduction in mprF mRNA after the addition of xylose, minimal impact on growth in the absence or presence of DAP and xylose, small size (236 bp), and proximity to the 3′ terminus of mprF.

FIG. 1.

Northern blot for mprF in constructs containing candidate plasmids in the absence and presence of 2% xylose. Eleven candidate plasmids (antisense [AS] clones 2D2, 1E3, 1E5, 1A2, 1E4, 2F2, 1H7, 2C9, 1C3, 1B5, and 1B8) containing various mprF antisense inserts in pSAX-1E were assessed as described in the text. mRNA was isolated from strains grown in the absence (−) or presence (+) of 2% xylose. The oligonucleotide probe was specific to mprF, with the arrow depicting the full-length transcript. pSAX-1EmprF2F2 (2F2) was chosen for further study based on the minimal impact the addition of xylose had on growth (optical density [OD] ratio with or without xylose).

The antisense mprF plasmid (or empty vector control [pSAX-1E]) was transformed into a subset of the S. aureus timeline isolates from Friedman et al. (10) and a clinical isolate obtained after failure of DAP therapy (Cubist collection); the isolates include different mprF alleles affecting either the putative flippase (MprF T345A) or synthase domain (MprF L826F) (8). DAP susceptibility was determined in the absence and presence of an inducer (Table 1). Adding xylose to the pSAX-1E vector resulted in no change greater than 2-fold in DAP MICs for all strains tested, consistent with the standard error margin. mprF antisense fragment induction resulted in mild but consistent sensitization to DAP in wild-type strains, with DAP MICs similar to those of ΔmprF constructs (MIC = 0.125 μg/ml). In DAP^r strains, antisense fragment induction decreased the MIC, restoring the DAP^s phenotype. Induction of antisense to mprF also restored DAP susceptibility to a laboratory-derived strain containing mutations in mprF and yycG genes. Consistent with these results, time-kill analyses displayed faster killing in the presence of xylose in the antisense-containing plasmid than in the vector-only control using the same amount of DAP (Fig. 2A). Because mutations in mprF have also been shown to affect the MIC of other cationic antibacterial agents (3, 10, 13, 17, 23, 24, 26, 36), the susceptibility of these constructs to vancomycin (VAN) and gentamicin (GENT) was determined (Table 1). VAN susceptibility was also determined because of reports in published literature suggesting that S. aureus strains with increased DAP MICs have increased VAN MICs (2, 6, 11, 13, 14, 19, 21, 23, 26, 29, 34, 37); GENT susceptibility was tested because previous literature reports indicate that DAP and GENT are synergistic in combination, and although GENT is a protein synthesis inhibitor, it also has cell membrane effects (5, 30). VAN and GENT MICs were decreased after mprF antisense induction in DAP^r strains; GENT MICs were also decreased in DAP^s strains. These data underscore the role of mprF in the nonsusceptibility phenotype for these antibacterial agents and suggest that increased production or distribution of LPG in the cell membrane may explain this gain-of-function phenotype (6, 14, 22). Supporting this, the strains containing an mprF mutation and the antisense plasmid were also tested for the ability to bind the cationic protein cytochrome c. As shown in Fig. 2B, strains induced for mprF antisense had less cytochrome c bound than the empty vector when grown in the presence of xylose (Fig. 2B). In addition, others have reported that distinct changes in membrane structure, function, and charge have been linked with DAP nonsusceptibility development in S. aureus (13). Similar to what we have observed using our antisense induction, there are numerous reports of S. aureus strains with disrupted mprF expression having increased susceptibility to cationic antimicrobial agents; this may be because repulsion between the membrane and the agent is reduced (17, 23, 24, 26, 36).

TABLE 1.

MICs of constructs expressing antisense mprF^a

Strain and relevant mutation(s) present	MIC
	Daptomycin				Vancomycin				Gentamicin				Tetracycline				Oxacillin
	pSAX-1E		pSAX-1E + antisense mprF		pSAX-1E		pSAX-1E + antisense mprF		pSAX-1E		pSAX-1E + antisense mprF		pSAX-1E		pSAX-1E + antisense mprF		pSAX-1E		pSAX-1E + antisense mprF
	−xyl	+xyl	−xyl	+xyl	−xyl	+xyl	−xyl	+xyl	−xyl	+xyl	−xyl	+xyl	−xyl	+xyl	−xyl	+xyl	−xyl	+xyl	−xyl	+xyl
RN4220 (WT)	0.5	0.25	0.5	<0.125	1	0.5	1	0.5	0.5	0.25	1	0.125	0.063	<0.03	0.063	0.0625	<0.03	<0.03	<0.625	<0.03
MW2 (WT)	0.5	0.5	0.25	0.125	0.5	0.5	1	0.5	1	0.5	1	0.125	0.125	0.125	0.125	0.125	16	16	16	16
MW2b	0.5	0.25	0.5	0.125	1	0.5	1	0.5	1	0.5	0.5	0.125	0.125	0.125	0.125	0.125	16	16	16	16
MW2b + MprF T345A	2	2	2	0.5	4	2	4	1	1	0.5	1	0.25	0.063	<0.03	0.063	0.125	0.5	0.5	1	32
MW2b + MprF T345A + YycG R263C	4	2	4	0.25	4	2	4	1	1	0.5	1	0.25	0.063	0.063	0.063	0.125	1	2	0.5	8
Clinical isolate (MprF L826F)	2	4	2	0.5	2	2	2	0.5	1	2	2	0.5	0.063	<0.03	<0.03	0.125	0.063	0.25	0.063	0.5

^aXylose concentrations used were 20 mM for daptomycin and 5 mM for other antibiotics. Lower concentrations of xylose were used for all nondaptomycin tests because growth defects were observed when higher concentrations were used. WT, wild type; −xyl, without xylose; +xyl, with xylose.

^bStrain had a mutation upstream of the acetate coenzyme A ligase.

FIG. 2.

Impact of MprF antisense expression on daptomycin time-kill kinetics and the ability of cells to repel cationic cytochrome c. (A) Daptomycin kill kinetics and cells expressing mprF antisense were compared to cells harboring the vector-only control. Strain designations indicate the MprF allele (WT, T345I, or L826F [the same strains as used for MIC analyses]) and the presence of either the pSAX-1EmprF2F2 plasmid expressing mprF antisense (+as) or the pSAX-1E vector-only control (+vc). Drug-free growth controls for strain MW2 expressing antisense (WT+as, GC) or containing the vector only (WT+vc, GC) are representative of growth controls for the other strains (data not shown). The DAP concentration was held constant for each pair, and xylose was used at 20 mM. (B) Binding of cationic cytochrome c relative to cell-free controls was compared for cells expressing mprF antisense or containing the vector-only control; the MW2 and MW2ΔmprF strains were included as controls. Values shown are averages of at least three independent values; error bars indicate standard deviation.

As a control, the susceptibility of these constructs was tested against an antibacterial that does not target the membrane (tetracycline [TET]) and a noncationic membrane antibacterial (oxacillin [OX]). The TET MIC was unaffected when mprF antisense was induced; however, induction produced an unexpected increase in the OX MIC in DAP^r—but not DAP^s—strains. This observation contradicts reports of mprF-deleted strains in which the MIC is decreased (16) but may be consistent with the seesaw effect observed between OX and other antibacterials, including DAP and VAN (27, 38).

In conclusion, we have demonstrated that antisense RNA regulation of mprF increased the susceptibility to DAP, VAN, and GENT in DAP^s and DAP^r S. aureus strains. It remains unclear from these studies how DAP interacts with MprF. We hypothesize that altered MprF activity affects the overall charge of the membrane, which could make cationic antibacterials more sensitive to increased membrane charge potential. This suggests that the MIC increase does not involve MprF directly but is rather associated with increased membrane charge density. A somewhat counterintuitive finding, however, was the increase in the OX MIC upon antisense mprF induction in DAP^r strains. This research supports the use of antisense RNA technology as a valuable tool for specifically probing the bacterial genome for genes involved in an organism's susceptibility, or lack thereof, to antibacterial agents.