β-Lactam-Induced Cell Envelope Adaptations, Not Solely Enhanced Daptomycin Binding, Underlie Daptomycin-β-Lactam Synergy in Methicillin-Resistant Staphylococcus aureus

ABSTRACT

Methicillin-resistant Staphylococcus aureus (MRSA) is a serious clinical threat due to innate virulence properties, high infection rates, and the ability to develop resistance to multiple antibiotics, including the lipopeptide daptomycin (DAP). The acquisition of DAP resistance (DAP-R) in MRSA has been linked with several characteristic alterations in the cell envelope. Clinical treatment of DAP-R MRSA infections has generally involved DAP-plus-β-lactam combinations, although definable synergy of such combinations varies in a strain-dependent as well as a β-lactam-dependent manner. We investigated distinct β-lactam-induced cell envelope adaptations of nine clinically derived DAP-susceptible (DAP-S)/DAP-R strain pairs following in vitro exposure to a panel of six standard β-lactams (nafcillin, meropenem, cloxacillin, ceftriaxone, cefaclor, or cefoxitin), which differ in their penicillin-binding protein (PBP)-targeting profiles. In general, in both DAP-S and DAP-R strains, exposure to these β-lactams led to (i) a decreased positive surface charge; (ii) decreased cell membrane (CM) fluidity; (iii) increased content and delocalization of anionic phospholipids (i.e., cardiolipin), with delocalization being more pronounced in DAP-R strains; and (iv) increased DAP binding in DAP-S (but not DAP-R) strains. Collectively, these results suggest that β-lactam-induced alterations in at least three major cell envelope phenotypes (surface charge, membrane fluidity, and cardiolipin content) could underlie improved DAP activity, not mediated solely by an increase in DAP binding. (Note that for ease of presentation, we utilize the terminology “DAP-R” instead of “DAP nonsusceptibility.”)

INTRODUCTION

Methicillin-resistant Staphylococcus aureus (MRSA) is a problematic pathogen with high associated mortality rates, principally related to its innate virulence properties as well as its ability to evolve resistance to multiple antibiotics (1). The development of clinical MRSA resistance to “last-line antibiotics,” including daptomycin (DAP), has been well chronicled, resulting in treatment failures (2–4). DAP’s mechanism of action is multifactorial, involving both cell wall synthesis inhibition (5, 6) as well as cell membrane (CM) targeting (7). DAP’s CM interaction involves complexing with calcium, oligomeric aggregation, and insertion into the CM as a “functional” cationic peptide (8–10). Accordingly, alterations in genetic pathways responsible for cell envelope homeostasis have been linked to the acquisition of the DAP resistance (DAP-R) phenotype (7, 11, 12).

Several recent clinical studies have utilized combinations of DAP plus selected β-lactams to treat infections caused by DAP-R MRSA (13, 14). The use of such combinations is supported by extensive studies showing in vitro synergistic activity, in particular against MRSA strains resistant to either antibiotic alone (15, 16). Moreover, such combinations may also prevent the development of DAP-R in MRSA strains by forestalling the emergence of single nucleotide polymorphisms (SNPs) in the multiple-peptide resistance factor gene, mprF, which is involved in the maintenance of positive surface charge (17). Furthermore, β-lactams can reduce the relative positive surface charge and enhance DAP CM binding in some MRSA strains (18, 19); however, the latter events do not appear to be essential for DAP–β-lactam synergy in vitro, suggesting that other mechanisms are in play (20). In addition, as MRSA strains become progressively more resistant to DAP in vitro and in vivo, they tend to become more β-lactam susceptible (the “seesaw effect”) (21). Despite the above-mentioned observations, the precise mechanism(s) responsible for DAP–β-lactam synergy remains incompletely understood (13, 20, 22, 23).

The goal of the current study was to better characterize potential cell envelope mechanisms by which β-lactam antibiotics may promote DAP–β-lactam synergy. We thus delineated several key cell envelope phenotypes in a well-defined set of isogenic DAP-susceptible (DAP-S)/DAP-R MRSA strain pairs (n = 9) following exposure to subinhibitory concentrations of selected β-lactams with a broad spectrum of penicillin-binding protein (PBP)-targeting profiles (see below). We focused on those envelope metrics previously associated with DAP-R in MRSA (24, 25), including CM order and surface charge, quantified with and without β-lactam exposure. Moreover, we assessed the role of the CM anionic phospholipid content (predominantly cardiolipin [CL]) and its distribution under the same conditions. Finally, the degree of overall DAP binding was quantified following distinct β-lactam exposures.

(This work was presented in part at the 30th European Congress of Clinical Microbiology and Infectious Diseases [ECCMID] meeting, 18 to 21 April 2020 [26].)

RESULTS

Surface charge.

As expected, overall, DAP-R strains exhibited significantly less cytochrome c binding at baseline (i.e., elevated cell surface positive charge) (27) than DAP-S strains (P < 0.05). Following exposure to all distinct β-lactams (except for cefaclor [CEC]), DAP-S and DAP-R MRSA strains exhibited increased binding to cytochrome c (i.e., a more negative relative surface charge than the respective strains with no antibiotic preexposure) (Fig. 1A). Exposure to cloxacillin (LOX) led to the greatest increase in cytochrome c binding (more negative surface charge) versus all other β-lactams (P < 0.05).

FIG 1.

Membrane characteristics previously associated with DAP-R. (A) Relative surface charge of DAP-S and DAP-R strains with or without β-lactam conditioning via a cytochrome c binding assay. A higher percentage of cytochrome c binding indicated a more negative cell surface charge. (B) Membrane fluidity of DAP-S and DAP-R strains with or without β-lactam conditioning via a membrane-polarizing spectrofluorometric assay. The polarization index is inversely correlated with membrane fluidity; i.e., a higher polarization index equates with decreased membrane fluidity. *, P < 0.05 versus no antibiotics (NO ABX).

CM order.

Exposure to all β-lactams (except for CEC) resulted in more rigid CMs than in untreated strains, in both DAP-S and DAP-R strains (Fig. 1B). Comparing β-lactam treatments to one another, LOX exposure caused the greatest impacts on CM order (more rigidity) (P < 0.05) (Fig. 1B).

NAO content.

The relative overall N-acrylamide orange (NAO) content (as a principal measure of CL, the primary phospholipid [PL]) following β-lactam exposure is shown in Fig. 2A. These data show a general, albeit modest, trend of increased CL contents in both DAP-S and DAP-R strains exposed to this panel of β-lactams, with meropenem (MEM) and LOX being the most impactful.

FIG 2.

Average anionic PL content quantified by NAO staining (A) and quantification of confocal images (B) of DAP-S and DAP-R strains, with or without β-lactam treatment. *, P < 0.05 versus no antibiotics.

Anionic phospholipid distribution.

We investigated potential differences in anionic phospholipid localizations (predominantly CL) by confocal microscopy in all 9 strain pairs. Confocal images were acquired following growth to the exponential phase, with or without exposure to the selected β-lactams. Fluorescence quantification of the confocal images further validated the apparent increase in the anionic phospholipid content following β-lactam exposure (Fig. 2B). At baseline, DAP-R strains tended to contain more CL per cell than DAP-S strains. In DAP-S strains, all β-lactam exposures led to significantly increased CL contents per cell. In DAP-R strains, exposures to MEM, LOX, or cefoxitin (FOX) led to similarly increased CL contents per cell versus untreated cells (P < 0.05).

Figure 3 shows images employing one representative DAP-S/DAP-R strain pair. Before antibiotic exposure, cells showed concentrated CL clusters, which tended to be at either the poles or the cell septal division plane. In contrast, following β-lactam exposures, cells showed a more global, nonseptal distribution of CL around the CM circumference. This perturbed CL distribution was more apparent and distinct in DAP-R than in DAP-S strains.

FIG 3.

Confocal images of one representative DAP-S/DAP-R strain pair (CB 1663/CB 1664) with or without β-lactam conditioning. NAO was imaged in the green channel, and NucSpot was imaged in the red channel.

DAP binding.

The binding of BODIPY-DAP was quantified via fluorescence intensity from confocal images, with or without β-lactam exposure. At baseline, DAP-R strains on average actually bound more DAP than their DAP-S counterparts. Among DAP-S strains, exposure to all β-lactams studied yielded significantly increased DAP binding (versus untreated strains) (Fig. 4A) (P < 0.05). In contrast, among DAP-R strains, the impact of β-lactam exposures tended to be more variable; thus, nafcillin (NAF) and ceftriaxone (CRO) resulted in significantly decreased DAP binding, while CEC exposure caused significantly increased DAP binding (P < 0.05).

FIG 4.

Quantification of the fluorescence intensity of BODIPY-DAP confocal images of DAP-S (A) and DAP-R (B) strains, with or without β-lactam treatment. Average binding (± the standard deviation) for DAP-S/DAP-R strains is presented. *, P < 0.05 versus no antibiotics.

Overview of cell envelope parameter impacts of β-lactam exposures.

Since clinically, DAP–β-lactam combinations are geared to treat DAP-R MRSA strains, Table 1 summarizes the overall cell envelope alterations identified among our 9 DAP-R strains following β-lactam exposures. Based on the actual data generated, this table lists those cell envelope changes predicted to improve DAP activity.

TABLE 1.

Overview of changes in cell envelope characteristics following β-lactam treatment of DAP-R strains compared to untreated strains and adaptations favorable for DAP activity

Parameter	Value for druga
	NAF	MEM	LOX	CRO	CEC	FOX
Phenotype
Surface charge	−	−	−	−	+	−
Membrane fluidity	NC	−	−	−	+	−
CL content, cultures	NC	+	+	NC	NC	NC
CL content, cells	NC	+	+	NC	NC	+
DAP binding	−	NC	NC	−	+	NC
Potential mechanism of DAP potentiation	Decreased surface charge	Decreased surface charge	Decreased surface charge	Decreased surface charge	Increased DAP binding	Decreased surface charge
		Decreased membrane fluidity	Decreased membrane fluidity	Decreased membrane fluidity		Decreased membrane fluidity
		Increased CL content	Increased CL content

^a+, increased; −, decreased; NC, no change.

DISCUSSION

The exact mechanisms underlying the development of DAP-R in MRSA appear to be heterogeneous and multifactorial. Relevant to the current investigation, a number of studies have linked specific alterations in the cell envelope with the acquisition of the DAP-R phenotype (7, 11, 12, 27–29). Clinically, the addition of different β-lactams (e.g., nafcillin and ceftaroline) to DAP treatment has proven effective in many well-characterized cases of recalcitrant MRSA infections caused by DAP-R strains (14, 16, 30). However, the mechanism(s) behind the salutary outcomes using these combinations is not well understood. As the cell envelope plays an integral and mechanistic role in DAP’s activity, as well as in DAP-R, we hypothesized that β-lactam exposure may lead to CM changes that could potentially enhance DAP’s efficacy. Several interesting observations emerged from our analyses.

First, exposures to most (5/6) of the tested β-lactams led to a significant decrease in the relative positive surface charge (versus untreated strains), with LOX being the most impactful in this regard. Of note, previous studies have linked DAP-R with an increased cell surface charge, commonly mediated by either gain-of-function mutations in mprF (31, 32) or dysregulation of dltABCD expression (11). The resultant increase in the relative positive surface charge was proposed to lead to the repulsion of the bioactive, positively charged DAP-calcium complex (33). The decreased positive surface charge seen in the current study following β-lactam exposures would likely enhance the interaction of the positively charged DAP complex with the cell envelope.

Second, the DAP-R phenotype in MRSA typically correlates with an alteration of membrane order (fluidity/rigidity). One prevailing theme is that an optimal CM order exists for DAP to exert its maximal activity. Thus, DAP-R strains derived either in vitro or in vivo tend to exhibit extremes of CM order (more rigid or more fluid than the respective DAP-S parental strains) (18). The hypothesis that DAP activity can be influenced substantially by altered CM order was further underscored by Müller et al., who found that DAP binding occurred selectively at more fluid lipid domain sites (34).

In the current investigation, when exposed to β-lactams, our study strains (both DAP-S and DAP-R) generally showed increased rigidity versus their respective non-β-lactam-treated control strains. It seems reasonable to assume that this β-lactam-mediated effect reverts the CM order to a state more favorable for DAP’s mechanism of activity. Whether this impact relates to modifications of DAP binding, oligomerization, and/or membrane insertion remains to be determined.

Third, the relationship between CL content and distribution with DAP activity has been most well defined among DAP-R enterococci (35). In such strains, CL appears to delocalize from the divisome and redistribute more globally throughout the CM. In the present study, an overall increase in CM CL was observed following β-lactam exposures. It is possible that such CL increases could contribute to a decreased surface charge, with enhanced DAP activity related to increased surface attraction. However, such increases in CL could also result in a compensatory decrease in the CM concentration of phosphatidylglycerol (PG), both an initial anionic anchoring point for DAP binding as well as a pivotal phospholipid for DAP CM oligomerization (36–38).

Fourth, quantification of the overall CL content demonstrated that exposure to MEM in DAP-S strains or MEM and LOX in DAP-R strains led to significant increases in overall CM CL. Moreover, the per-cell CL content showed that all β-lactams in DAP-S strains as well as MEM, LOX, and FOX in DAP-R strains led to significantly increased CL contents. Of note, such increases in the CL content could well explain, at least in part, the relative decrease in the positive surface charge that we observed.

A recent study by Jiang et al. found that point mutations in cls2, linked to the development of DAP-R, led to increased CL contents in the CM, accompanied by compensatory reductions in the PG content; they further hypothesized that certain DAP-R-associated mutations in the transmembrane region of cls2 may also lead to changes in the CL distribution (37). However, the latter study focused on laboratory-derived strains with only cls2 point mutations with a gain-of-function phenotype. Their investigations did not address clinically derived strains, which likely also contain mutations in other CM homeostasis genes (e.g., in mprF, which increases the content of positively charged lysyl-PG, which can potentially repel the active DAP-calcium complex) (25, 39, 40). Thus, the exact mechanism(s) by which the apparent β-lactam-enhanced CL content and redistribution in MRSA contribute to increased DAP activity remains elusive.

Fifth, confocal microscopy suggested that β-lactam exposures resulted in a delocalization of CL away from the divisome and to a more generalized circumferential distribution. There appear to be two potential “competing” outcomes of such CL delocalization. On the one hand, such CL-rich CMs are more susceptible to bending and stretching in terms of dividing due to limited lateral interaction with other phospholipids (37). In such strains, it is possible that this β-lactam-induced CL delocalization might improve DAP’s insertion and CM pore formation, enhancing its microbicidal activity. On the other hand, delocalization of CL away from the divisome (a principal site of action) could impede DAP’s activity, as noted above for DAP-R enterococci (35).

Sixth, several previous studies looking at DAP–β-lactam combinations suggested that increased DAP binding likely explained the observed synergy (23, 41). However, Berti et al. and Dhand et al. clearly demonstrated that the extent of DAP binding did not uniformly correlate with DAP–β-lactam synergy (20, 23). In support of the latter notion, our current results do not show a consistent relationship between DAP binding and β-lactam exposures; thus, in general, DAP binding was enhanced by β-lactams in DAP-S strains but overall decreased in DAP-R strains. In particular, two of the tested DAP-R strains showed 2-fold decreases in DAP binding following β-lactam exposures, driving down the average reported binding. Previous studies in select strains have shown NAF exposure to increase DAP binding in MRSA (20). However, our more comprehensive strain set analyses demonstrated important strain-to-strain variations and, on average, lower DAP binding after β-lactam exposures in DAP-R strains. We hypothesize that DAP-R strains likely have fundamentally exaggerated responses to β-lactam stress versus DAP-S strains, perhaps in a way that globally decreases DAP binding to CMs (19, 21).

Finally, comparing distinct β-lactams, our data suggest that discrete mechanisms of cellular responses to specific agents underlie DAP–β-lactam synergy in DAP-R strains. For example, NAF exposures led to a decrease in the positive surface charge; this suggested that the β-lactam-mediated reversal of the usual increased surface charge in DAP-R strains is one key synergy mechanism of treatment with NAF plus DAP. In comparison, CRO and FOX yielded decreased surface charge and increased CM rigidity, implying that such a CM profile favors DAP activity. MEM and LOX exposures resulted in increased CL contents, in addition to the reversal of other DAP-R-associated phenotypes. As changes in surface charge, CM order, and CL content were noted following MEM or LOX exposure, it is likely that all of these phenotypes play a combinatorial role in DAP synergy. In contrast, CEC conditioning led to a significant increase in DAP binding, suggesting that CEC-DAP synergy is primarily based on the enhancement of DAP binding.

Collectively, our findings imply that the mechanisms underlying interactions between distinct β-lactams and DAP vary from strain to strain, providing an explanation for the broad range of phenotypic responses observed with different DAP–β-lactam combinations.

It should be noted that this study was specifically designed to utilize uniformly bacteriostatic β-lactam concentrations based on in vitro MICs. This strategy yielded supraphysiological β-lactam antibiotic exposures in some instances, which are not achievable in most clinical settings in vivo. Further studies correlating the above-described phenotypic metrics to strain-specific synergy and genetic profiles, employing more clinically relevant β-lactam antibiotic concentrations, are under way.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

This study focused on the analysis of nine previously characterized clinical bloodstream DAP-S/DAP-R isogenic MRSA strain pairs (2, 19, 31). These strain pairs were prioritized to include DAP-R strains with or without mprF mutations and have been previously characterized for certain phenotypic and genotypic characteristics (19, 31).

The following β-lactams with wide-ranging PBP selectivity were employed: cloxacillin (LOX) (PBP-1), meropenem (MEM) (PBP-1), nafcillin (NAF) (PBP nonselective), ceftriaxone (CRO) (PBP-2), cefaclor (CEC) (PBP-3), and cefoxitin (FOX) (PBP-4) (42). Exposure concentrations of these β-lactams were chosen based on extensive pilot experiments to determine drug levels for each single antibiotic that exerted a sublethal (bacteriostatic) impact in vitro, defined as a <2-log₁₀-CFU/ml reduction in growth over a 24-h period (see Table S1 in the supplemental material). Bacteriostatic concentrations were selected to compile an overarching analysis of all β-lactam-induced adaptations (rather than using fixed β-lactam concentrations for all strains based on human-achievable serum levels).

Surface charge.

The relative net positive cell surface charge was quantified using the spectrophotometrically based cytochrome c binding assay (a highly cationic molecule) as previously described (31, 32). Study strain pairs were grown overnight in the presence or absence of sublethal concentrations of the individual β-lactams listed above. Bacterial suspensions were then centrifuged at 5,000 rpm for 5 min, the supernatant was then removed (containing the β-lactam of interest), and cells were resuspended in fresh medium. Cells were then washed twice with MOPS (morpholinepropanesulfonic acid) buffer (20 mM; pH 7.0), adjusted to an optical density at 600 nm (OD₆₀₀) of 1.0, and collected from 1-ml aliquots via centrifugation. Cell pellets were resuspended in 200 μl MOPS buffer and combined with 50 μl of cytochrome c (2.5-mg/ml solution). Samples were incubated for 10 min at room temperature and separated by centrifugation. The OD₅₃₀ was determined in the supernatant, and the magnitude of cytochrome c binding was then determined using a standard curve. The more cytochrome c remaining in the supernatant is a measure of a relative increase in the relative bacterial positive surface charge.

Quantification of anionic phospholipids.

DAP-S/DAP-R MRSA pairs were grown overnight to stationary phase in the presence or absence of bacteriostatic concentrations of the above-mentioned β-lactams. CL is the principal anionic phospholipid species in the MRSA CM; we utilized an anionic phospholipid-specific dye (N-acrylamide orange [NAO])-based spectrofluorometric assay as a surrogate for CM CL content. As negative controls for this NAO binding assay, we employed the cardiolipin synthase knockout mutants N315 Δcls1 and N315 Δcls2 (43) (Fig. S1). For this assay, 1.0 × 10⁷ CFU/ml of each MRSA strain were exposed to 20 μM NAO and then incubated at 4°C for 20 min. The NAO fluorescence intensity was measured using spectrofluorometry (excitation at 525 nm and emission at 640 nm).

Anionic phospholipid localization.

Anionic phospholipid localization was visualized using stimulated emission depletion (STED) fluorescence microscopy. Cells were grown to exponential phase with shaking at 37°C in LB medium overnight in the presence or absence of bacteriostatic concentrations of each distinct β-lactam. NAO was then added at 20 μM for 1 h at room temperature. Cells were washed and resuspended in phosphate-buffered saline (PBS). For nucleus staining, 1 μl of NucSpot Live 650 nuclear stain was added to a 1-ml suspension, and the mixture was incubated for 10 min at room temperature. The cells were concentrated 20-fold at the last step, and 3 μl was placed on a glass slide. Slides were set with Prolong diamond antifade mountant and a number 1.5 glass coverslip. Images were collected using a Leica SP8 3× STED superresolution confocal microscope using standard filter sets for either green fluorescent protein (GFP) (495-nm excitation and 510- to 579-nm emission, with 592-nm depletion) to visualize NAO or Cy5 (633-nm excitation and 667- to 742-nm emission, with 775-nm depletion) to visualize NucSpot according to the manufacturer’s instructions. Images were processed with Huygens software deconvolution wizard. In addition to CL visualization to identify effects on its CM localization, overall average CL quantification within individual cells was performed by measurement of the integrated fluorescence density in ImageJ for 30 cells in total in the NAO channel, and the corrected cell total NAO fluorescence was then calculated.

CM order (fluidity/rigidity).

Strains were grown at 37°C for 72 h in tryptic soy broth (TSB), replacing the medium and β-lactam antibiotic every 24 h, with or without exposure to the average unbound concentration of each selected β-lactam. CM fluidity/rigidity was then measured using the fluorescent probe 1,6-diphenyl-1,3,5-hexatriene (DPH). Methods for DPH incorporation into the CM, measurement of fluorescence polarization, and calculation of the polarization index have been previously described in detail (44). A BioTek Synergy H1 hybrid multimode reader with excitation at 360 nm and emission at 426 nm was used.

BODIPY-DAP fluorescence microscopy.

To quantify DAP binding (in the presence or absence of each β-lactam), cells were harvested at exponential phase from LB cultures supplemented with 50 μg/ml Ca²⁺. Cells were incubated with 16 μg/ml BODIPY-labeled DAP as previously described (23). The cells were concentrated 20-fold, and 3 μl was placed on a glass slide. Slides were set with Prolong diamond antifade mountant and a number 1.5 glass coverslip. Images were collected using a Leica SP8 3× STED superresolution confocal microscope using a 489-nm laser line and 510- to 579-nm emission, with 660-nm depletion. ImageJ was utilized to measure the integrated fluorescence density of 30 cells, and the corrected cell total fluorescence was calculated.

Statistical analysis.

The two-tailed Student t test was used for statistical analysis of β-lactams compared to untreated strains. One-way analysis of variance (ANOVA) was used for β-lactam comparisons.