Many members of the genus Staphylococcus are important human pathogens. Staphylococcus aureus in particular causes a multitude of frequently severe and life-threatening diseases, with acute disease promoted by a series of secreted toxins and other virulence determinants [1]. Coagulase-negative staphylococci, most notably Staphylococcus epidermidis, are a premier cause of hospital-associated infections on indwelling medical devices [2,3].

While virulence of S. aureus is clearly multi-factorial, the phenol-soluble modulin (PSM) peptide family has recently been identified as a key contributor to infection with highly virulent S. aureus strains, such as community-associated methicillin-resistant S. aureus (CA-MRSA) [4,5]. The PSMα peptides in particular strongly impact the capacity of CA-MRSA strains to lyse human neutrophils and other cell types, and promote skin infection and bacteremia. Less virulent hospital-associated strains characteristically produce smaller amounts of those peptide toxins [6].

In addition to their cytolytic potential, PSM peptides promote inflammatory responses by activating the FPR2 receptor [7]. Furthermore, they contribute to biofilm structuring, detachment, and the systemic dissemination of biofilm-associated infection [8,9]. Moreover, some PSMs may exhibit antimicrobial functions, for example toward Streptococcus pyogenes [10,11].

In contrast to the receptor-mediated pro-inflammatory roles of PSMs, the mechanistic basis of other PSM functions, such as cytolysis and biofilm maturation, is less well understood. Because cytolysis by PSMs is of utmost importance for staphylococcal virulence, the mechanism of PSM cytolytic activity is of key interest. Most likely, cytolysis is driven by the surfactant properties of PSMs, which are assumed to be critical for the formation of pores in cytoplasmic membranes and based on the characteristic amphipathic α-helical structure of PSMs. In fact, the PSM δ-toxin has been shown to lyse synthetic vesicles by the formation of short-lived pores [12]. However, other PSMs have never been analyzed for their potentials to lyse phospholipid vesicles.

Phosphatidylcholine (POPC) is a major constituent of human cytoplasmic membranes. Membranes of red and white blood cells for example contain about equal amounts of cholesterol and phospholipids, among which POPC is the most abundant at 30% or higher of total membrane phospholipid [13,14]. Notably, POPC is present mostly in the outer leaflet of the human cytoplasmic membrane [15], and thus a predominant component of the membrane part interacting with exogenous toxins such as PSMs. In contrast, bacterial membranes do not contain cholesterol and have a phospholipid composition strongly different from that of eukaryotic membranes. S. aureus membranes for example contain mostly phosphatidylglycerol, diphosphatidylglycerol, and lysylphosphatidylglycerol, but no POPC [16].

Here, we analyzed the interaction of all PSM peptides known in S. aureus and S. epidermidis, and of psm isogenic deletion mutants of a CA-MRSA strain, with synthetic POPC vesicles. Our analyses demonstrate pore-forming activity of all PSM peptides. Importantly, PSM peptides without considerable reported cytolytic activities toward human cells, such as PSMβ peptides, lysed POPC vesicles very efficiently. This indicates that the interaction of PSMs with eukaryotic membranes is governed by more complex interactions than previously believed. The elucidation of such interactions will be an important task of future investigations aimed to answer the key question of why eukaryotic cells are much more susceptible to PSM cytolytic activity than bacterial cells.

To analyze whether and to which degree PSM peptides lyse POPC vesicles, we set up a fluorescein release assay. We tested final PSM peptide concentrations in the range of 0.25 to 2 μM and chose 0.5 and 1 μM for the final assays, as with those concentrations quantitative differences were most evident. The relative capacities of the peptides to lyse the vesicles were measured over time and curves were fit to a one-phase association. Association constants were determined to express differences in lytic capacities.

1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) vesicles were prepared as described previously [12]. Briefly, a POPC (Avanti Polar Lipids) lipid film, which was prepared after rotary evaporation and lyophilization, was rehydrated in a solution of 20 mM 3-(N-morpholino)propanesulfonic acid (Sigma), 100 mM potassium chloride (Sigma), 0.01 mM ethylene glycol tetraacetic acid (Sigma), 0.02% (w/v) sodium azide, 50 mM carboxyfluorescein (Fluka), and 100 mM potassium chloride (buffer 1). The hydrated lipids were passed through a polycarbonate and nitrocellulose membrane multiple times in a Mini-Extruder (Avanti Polar Lipids, Inc.). Lipid fractions were collected after the extruded material was added to a Sephadex G-25 column equilibrated with buffer 1 without carboxyfluorescein.

Lipid concentrations were measured using the Bartlett assay as described previously [12]. Purified lipid vesicles were diluted to 200 μM in equilibration buffer and used within a day of preparation.

Synthetic N-formylated S. aureus and S. epidermidis PSMs (GL Biochem Shanghai Ltd.) were constituted and diluted in dimethyl sulfoxide.

Culture filtrates, from clinical CA-MRSA isolate USA300 (clone LAC) and isogenic psm and agr deletion mutants, were collected from bacterial cultures in tryptone soy broth (TSB) inoculated 1:100 from pre-cultures grown overnight, and grown for 8 h at 37 °C, with shaking at 200 rpm. PSM production in the culture filtrates was analyzed by high-performance liquid chromatography and confirmed to be as described [11].

Carboxyfluorescein release from POPC lipid vesicles incubated with PSMs or culture filtrates was measured by fluorescence (excitation at 470_nm, emission at 520_nm) in a 96 well plate fluorimeter (VICTOR3, Perkin Elmer).

Curve fitting was performed using Graph Pad Prism Version 5.04. All assays were performed in triplicate and error bars show ±SEM.

Our results demonstrate that all PSM peptides of S. aureus and S. epidermidis have the capacity to lyse synthetic phospholipid (POPC) vesicles. Of note, the POPC-lytic activity of S. aureus culture filtrate is exclusively dependent on PSMs. Furthermore, we noted considerable differences in the lytic capacities of different PSM peptides, which did not correlate with previously described capacities to lyse eukaryotic cells. These noticeable discrepancies, most pronounced for PSMβ peptides, indicate that the lytic activities that PSMs have toward human cells is not simply determined by surfactant-type interaction of PSMs with phospholipid bilayers. Elucidation of the molecular basis of these discrepancies, which may include differential composition of bacterial versus eukaryotic membranes or the requirement for specific docking structures, is an important task of future research.