Phenol-soluble modulins – critical determinants of staphylococcal virulence

Abstract

Phenol-soluble modulins (PSMs) are a recently discovered family of amphipathic, alpha-helical peptides that have multiple roles in staphylococcal pathogenesis and contribute to a large extent to the pathogenic success of virulent staphylococci, such as Staphylococcus aureus. PSMs may cause lysis of many human cell types including leukocytes and erythrocytes, stimulate inflammatory responses, and contribute to biofilm development. PSMs appear to have an original role in the commensal lifestyle of staphylococci, where they facilitate growth and spreading on epithelial surfaces. Aggressive, cytolytic PSMs seem to have evolved from that original role and are mainly expressed in highly virulent S. aureus. Here we will review the biochemistry, genetics and role of PSMs in the commensal and pathogenic lifestyles of staphylococci, discuss how diversification of PSMs defines the aggressiveness of staphylococcal species, and evaluate potential avenues to target PSMs for drug development against staphylococcal infections.

Introduction

Staphylococcus aureus is a dangerous pathogen that is a frequent cause of hospital- and community-associated infections on a global scale (Lowy, 1998, Deleo et al., 2010). Many S. aureus hospital-associated infections present as infections of the respiratory tract, while skin and soft tissue infections dominate among community-associated infections (Lowy, 1998, Deleo et al., 2010). However, S. aureus can also cause a series of other diseases, including endocarditis or osteomyelitis. Often, S. aureus infection can proceed to septicemia and become life-threatening. Furthermore, S. aureus is a frequent cause of biofilm-associated infections, in particular those developing on indwelling medical devices (Otto, 2008). Finally, S. aureus can cause food poisoning.

S. aureus strains may produce a diverse and large repertoire of virulence factors, many of which are encoded on mobile genetic elements (MGE)s (Novick et al., 2001). MGE-encoded virulence factors can be transferred from strain to strain by horizontal gene transfer and are usually present only in a minority of strains. Some virulence factors appear to have been acquired from coagulase-negative staphylococci (CoNS) such as S. epidermidis (Otto, 2013). However, some S. aureus virulence factors are encoded on the core genome, such as alpha-toxin and phenol-soluble modulins (PSMs), which are thus produced by virtually all strains.

Toxins, secreted molecules that directly harm the human or animal host, are arguably the most important and dangerous virulence determinants of S. aureus. In particular, S. aureus produces a series of cytolytic toxins, which are proteins that cause lysis of host cells such as red and white blood cells. Those that lyse white blood cells (leukocytes) are called leukocidins. Leukocidins of S. aureus belong to the beta-barrel forming family, comprising alpha-toxin and the bicomponent leukocidins (Szmigielski et al., 1999). Of note, these toxins have recently been shown to exert their cytolytic activities dependent on recognition by specific receptors on the surface of the host cell (Wilke & Bubeck Wardenburg, 2010, Alonzo et al., 2011, Spaan et al., 2013). PSMs, the subject of the present review, are a family of peptides with multiple functions in pathogenesis. They gained recognition mostly due to their pronounced cytolytic activity toward leukocytes (Wang et al., 2007), which is why they can be regarded as a new class of staphylococcal leukocidins.

Antibiotic resistance is frequent among S. aureus and severely complicates treatment of S. aureus infections (Lowy, 2003). Penicillin resistance is common among S. aureus; and resistance to methicillin as the antibiotic of first choice against penicillin-resistant S. aureus is of great concern. Many countries report methicillin resistance rates among hospital isolates that exceed 50%. Community-associated methicillin-resistant S. aureus (CA-MRSA) strains, first reported in the late 1990s, represent an additional severe threat to public health systems (Deleo et al., 2010). Antibiotic resistance is also common among many other staphylococcal species, which are in general less aggressive than S. aureus, but are often involved with chronic infections. S. epidermidis in particular is the most frequent cause of infections on indwelling medical devices (Otto, 2009).

Staphylococci are colonizers of the human or animal skin and mucous surfaces. S. epidermidis is the most frequent skin colonizer in humans (Otto, 2009), while S. aureus colonizes the nares and ano-rectal areas of about one third of the population (Rimland & Roberson, 1986, Wertheim et al., 2005). Importantly, staphylococcal infections are believed to originate mainly from colonizing bacteria (von Eiff et al., 2001), making colonization a risk factor for infection. However, we know much less about the physiology of staphylococci during their commensal state as compared to our knowledge of the molecular underpinnings of staphylococcal infectivity. Of note, it has been argued that the physiological status of staphylococcal commensal life may resemble that seen in biofilms (Sugimoto et al., 2013), which are bacterial agglomerations embedded in an extracellular matrix commonly associated with device-related infections (Otto, 2008).

In this review article, we will present the current knowledge on PSMs, a recently discovered family of short, amphipathic peptides produced by S. aureus and other staphylococci, which have been recognized as premier determinants of staphylococcal virulence and physiology.

PSMs – history, definition, and structure

The term “phenol-soluble modulins” (PSMs) was coined by Seymour Klebanoff in 1999, when he and his coworkers discovered immune-modulatory peptides of S. epidermidis that partitioned into the phenol phase during hot phenol extraction (Mehlin et al., 1999). He described a pro-inflammatory “complex” of three peptides, PSMα, PSMβ, and PSMγ, with PSMγ being equivalent to the previously described δ-toxin of S. epidermidis. The Klebanoff group reported a series of pro-inflammatory activities of the PSM “complex” (Mehlin et al., 1999, Liles et al., 2001), many of which we believe today were due to impurities, as pure (synthetic) peptides and isogenic deletion mutants of psm genes were not used in those early studies. Furthermore, PSMs have now been found to signal through a specific receptor (FPR2, see below) (Kretschmer et al., 2010), which is different from the TLR2/TLR6 receptor reported to be activated by the PSM “complex” back then.

PSMs can be classified according to their length. PSMs of the α-type are ~ 20–25 amino acids in length, and those of the β-type are about double that size (43–45 amino acids). The PSMs of the β-type commonly have a negative net charge, while most, but not all, α-type PSMs have a neutral or positive net charge. All PSMs form an amphipathic α-helix, with the α-helical part stretching over the entire length of α-type PSMs and forming the C-terminal part of the β-type PSMs, as predicted by primary and secondary structure bioinformatics analysis (Fig. 1). The structure of S. aureus δ-toxin, which belongs to the α-type, was determined by nuclear magnetic resonance (NMR) analysis (Tappin et al., 1988), confirming amphipathy and α-helicity. These features explain why PSMs are phenol-soluble. PSMs are nowadays more easily identified by their behavior during reversed-phase high-pressure liquid chromatography (RP-HPLC) rather than by their partitioning into the phenol phase. Namely, they elute at exceptionally high concentration of organic solvent, much later than most other proteins or peptides (Wang et al., 2007). This feature is also likely due to the pronounced amphipathy of PSMs and their tendency to aggregate. Finally, it is important to stress that PSMs usually contain an N-terminal N-formyl methionine, indicative of the fact that they are secreted without a signal peptide (Wang et al., 2007). Accordingly, psm genes only encode the amino acids found in the secreted PSM peptides. Some portions of secreted PSMs may lack the N-terminal N-formyl group, which is due to the activity of the cytoplasmic enzyme N-deformylase (Somerville et al., 2003). Interestingly, this appears to happen at a much more pronounced extent in S. aureus than S. epidermidis (Cheung et al., 2010).

Fig. 1. Structure of PSMs.

Shown is a structural representation of PSMα3 modeled after NMR data available for δ-toxin (a) and an α-helical wheel representation of the same peptide (b) as examples. Hydrophobic amino acids are in yellow/brown, hydrophilic amino acids in blue (positive charge) or red (negative charge). The two C-terminal asparagine residues are in pink. Note that in the α-helix formed by PSMα3 hydrophilic and hydrophobic residues occupy opposite locations, making the α-helix distinctly amphipathic.

Purification and measurement of PSMs

As performed in the early studies by the Klebanoff group, a crude PSM preparation can be obtained as precipitate by hot phenol extraction followed by dialysis against distilled water (Mehlin et al., 1999). Precipitation with trichloroacetic acid (TCA) or extraction with 1-butanol can be used as alternatives to enrich PSMs from the culture supernatant (Otto et al., 2004, Joo et al., 2011). Crude PSM extract is further purified by RP-HPLC using a gradient of organic solvent such as acetonitrile or 1-propanol. Some PSMs (e.g., PSMα and PSMβ of S. epidermidis), which elute together at the first RPLC step, need one or more additional RP-HPLC steps for purification (Mehlin et al., 1999, Yao et al., 2005, Joo & Otto, 2014). As a special case of PSM purification, fibrous PSM complexes can be obtained from S. aureus biofilms cultured in drip bioreactors with a defined media, namely peptone-NaCl-glucose (Schwartz et al., 2012).

The amount of PSMs can be measured in complex bacterial culture supernatant by RP-HPLC with a proper column and eluent system (Fig. 2) (Joo & Otto, 2014). Due to the exceptionally late retention time of PSMs during RP-HPLC, PSMs are separated quite easily from other proteins in the culture supernatant. However, the fact that many PSMs do not separate from each other under those conditions necessitates coupling with mass spectrometry (MS) to quantify single PSMs accurately. The concentration of each PSM is determined by integration of the extracted ion chromatogram of the two major m/z ratios among multiply-charged ions of each PSM (Otto et al., 2004, Joo & Otto, 2014). For absolute quantification, purified or synthetic PSMs should be used to calibrate (Yao et al., 2005). As a useful tool for qualitative detection of PSMs, Nizet and colleagues have used an imaging mass spectrometry technique with MALDI-TOF mass spectrometry to discover PSMα derivatives (Gonzalez et al., 2012).

Fig. 2. Analysis of PSMs by RPLC/MS.

Shown is RPLC chromatography routinely used for RPLC/MS analysis of PSMs. Culture supernatant of an S. aureus strain (USA300) was directly applied to a 2.3 × 30 mm C8 reversed phase column and a brief gradient from 10 to 50% followed by an extended gradient from 50 to 90% water/0.1% triflouroacetic acid (TFA) to acetonitrile/0.1% TFA as described in Wang et al. (Wang et al., 2007) was run. Note that PSMs are separated from other proteins, but not all PSMs completely separate from other PSMs, necessitating coupling to MS for measurement of single PSM quantities. WT, wild-type USA300 (LAC); Δpsmα, psmβ, hld, isogenic mutant not expressing any PSM.

Occurrence of PSMs in staphylococci

PSMs occur in many staphylococcal strains, with the PSM profile apparently being specific for and conserved within a certain species (Vuong et al., 2004, Wang et al., 2007, Queck et al., 2009, Rautenberg et al., 2011). Although this would represent a possibly valid tool to distinguish staphylococcal species, PSM profiling has not yet been used for that purpose. It has been noted that PSM production is more frequent among pathogenic than non-pathogenic species (Rautenberg et al., 2011), but whether there is a clear correlation between pathogenicity and PSM production on the species level needs further verification.

For some PSMs, such as the δ-toxins and PSMβ peptides of S. aureus, S. epidermidis and many other staphylococci, amino acid sequences are fairly conserved between species. Some previously described peptides, such as the gonococcal growth inhibitor peptides from S. haemolyticus (Beaudet et al., 1982, Watson et al., 1988) or the slush peptides from S. lugdunensis (Donvito et al., 1997), are now recognized to belong to the β-type PSMs. Genes encoding δ-toxin and β-type PSMs can be found by similarity searches in many recently sequenced staphylococcal genomes (Tab. 1). In contrast, most α-type PSMs, such as the PSMα peptides of S. aureus, are too short to be routinely annotated in genome sequences and in general too short to yield meaningful results in similarity searches. Performing tblastn similarity searches with all S. aureus and S. epidermidis α-type PSMs, we detected some genes in staphylococci other than S. aureus and S. epidermidis that are similar to PSMε of S. epidermidis and were not previously described (Tab. 1). However, these searches did not yield any significant results for other α-type PSMs; and for the reasons outlined above, they are not likely to detect all α-type psm genes in a genome. For most staphylococcal species, we therefore do not have a complete list of PSM peptides, although we know from RP-HPLC/MS analyses that they produce PSMs (Rautenberg et al., 2011).

Table 1.

PSM homologues identified by similarity searches

Species/strain	PSM name/annotated as (suggested name in brackets)	Amino acid sequence	Gene1	Protein1	Length (amino acids)	Net Charge
PSMβ homologues
S. aureus	PSMβ1	MEGLFNAIKDTVTAAINNDGAKLGTSIVSIVENGVGLLGKLFGF			44	−2
	PSMβ2	MTGLAEAIANTVQAAQQHDSVKLGTSIVDIVANGVGLLGKLFGF			44	−2
S. epidermidis	PSMβ1	MSKLAEAIANTVKAAQDQDWTKLGTSIVDIVESGVSVLGKIFGF			44	−2
	PSMβ2	MEQLFDAIRSVVDAGINQDWSQLASGIAGIVENGISVISKLLGQ			44	−4
S. warneri SG1	PSMβ1 (PSMβ1)	MEGLIKAIKDTVEAGVNNDGAKLGTSIVGIVENGVGVLSKLFGF	A284_07765	AGC90870.1	44	−2
	PSMβ1 (PSMβ2a)	MTKLAEAIANTVQAAQGHDGAKLGTSIVSIVENGVSVLGKLFGF	A284_07760	AGC90869.1	44	−1
	PSMβ1 (PSMβ3a)	MTKLAEAIANTVKAAQGHDGAKLGTSIVSIVENGVSVLGKLFGF	A284_07755	AGC90868.1	44	0
	PSMβ1 (PSMβ3b)	MTKLAEAIANTVKAAQGHDGAKLGTSIVSIVENGVSVLGKLFGF	A284_07750	AGC90867.1	44	0
	antibacterial protein (PSMβ2b)	MTKLAEAIANTVQAAQGHDGAKLGTSIVSIVENGVGVLSKLFGF	A284_07745	AGC90866.1	44	−1
S. haemolyticus JCSC1435	unnamed protein (PSMβ1a)	MSKLVQAISDAVQAGQNQDWAKLGTSIVGIVENGVGILGKLFGF	SH1745	BAE05054.1	44	−1
	unnamed protein (PSMβ2a)	MEKIANAVKSAIEAGQNQDWTKLGTSILDIVSNGVTELSKIFGF	SH1744	BAE05053.1	44	−2
	unnamed protein (PSMβ1b)	MSKLVQAISDAVQAGQNQDWAKLGTSIVGIVENGVGILGKLFGF	SH1743	BAE05052.1	44	−1
	unnamed protein (PSMβ2b)	MEKIANAVKSAIEAGQNQDWTKLGTSILDIVSNGVTELSKIFGF	SH1742	BAE05051.1	44	−2
	unnamed protein (PSMβ3)	MQKLAEAIAAAVQAGQDKDWGKMGTSIVGIVENGISVLGKIFGF	SH1741	BAE05050.1	44	−1
S. pseudinterme dius ED99	putative hemolysin-H1C (PSMβ1)2	MTELFDAIKNTIEAGINHDWVTLGTSIADIVANGINVISGFFG	SPSE_1635	ADX76893.1	43	−5
	putative hemolysin-H1C (PSMβ2)2	MGDLFESITNIVTSAIQHDWVNMGAAIVATVQHGVELIAKLFGLG	SPSE_1546	ADX76805.1	45	−4
S. pseudinterme dius HKU10-03	hypothetical protein (PSMβ2)2	MGDLFESITNIVTSAIQHDWVHMGAAIVATVQHGVELIAKLFGLG	SPSINT_0962	ADV05490.1	45	−4
	hypothetical protein (PSMβ1)2	MTELFDAIKNTIEAGINHDWVTLGTSIADIVANGINVISGFFG	SPSINT_0872	ADV05400.1	43	−5
S. saprophyticus subsp. saprophyticus ATCC 15305	putative antibacterial protein (PSMβ1)	MEGLFEAIKNTVQAGIAADGAKIGSGVLDIISNGASLIGKALGL	SSP1598	BAE18743.1	44	−2
	putative antibacterial protein (PSMβ2)	MADLFNAIKETVQAGIAGDGAKLGTSIVSIVENGVGVLSKLFGF	SSP1597	BAE18742.1	44	−2
	putative antibacterial peptide (PSMβ3)2	MTGIVEAIGNAVNVGLAHDWATMGVSIADVLAKGVDSVLGFFK	SSP0130	BAE17275.1	43	−3
S. lugdunensis N920143	SlushA	MSGIVDAISKAVQAGLDKDWATMATSIADAIAKGVDFIAGFFN	SLUSHA	CCB52870.1	43	−3
	SlushB	MSGIIEAITKAVQAGLDKDWATMGTSIAEALAKGIDAISGLFG	SLUSHB	CCB52869.1	43	−3
	SlushC	MDGIFEAISKAVQAGLDKDWATMGTSIAEALAKGVDFIIGLFH	SLUSHC	CCB52868.1	43	−3
S. carnosus TM300	phenol soluble modulin (PSMβ1)	MEGLFNAIKGTVEAAINQDGGQLAVNIVDIVQNGIQIVSKFIGA	SCA_0784	CAL27694.1	44	−3
	phenol soluble modulin (PSMβ2)	MLTKLAQAIVDTVNAAQAQDPAKLGTSIVDIVQNGVGIVGKLFGF	SCA_0785	CAL27695.1	45	−1
S. simulans bv. staphylolyticus NRRL B-2628 plasmid pACK1	unnamed protein product (PSMβ-pACK1)	MSGIVDAISQAVQSGLEQDWVTMGTNIAKALAQGIDVISGFFG			43	−4
δ-toxin homologues
S. epidermidis	δ-toxin	MAADIISTIGDLVKWIIDTVNKFKK			25	0
S. aureus	δ-toxin	MAQDIISTIGDLVKWIIDTVNKFTKK			26	0
S. carnosus TM300	(δ-toxin)	MAGDIVGTIGDFVKLIIETVNKFTNK			26	−1
S. simulans	(δ-toxin)	MAGDIVGTIGEFVKLIIETVQKFTQK	AJ223775.1	CAA11542.1	26	−1
S. intermedius	(δ-toxin)	MAGDIISTIVDFIKLIAETVKKFTK	AY860843.1	AAW55662.1	25	0
S. pseudinterme dius HKU10-03	(δ-toxin)	MAADIISTIVEFVKLIAETVEKFIK			25	−2
S. pseudinterme dius ED99	(δ-toxin)	MAADIISTIVEFVKLIAETVAKFIK			25	−1
S. warneri SG1	(δ-toxin 1)	MAADIISTIGDLVKLIINTVKKFQK	A284_04270	AGC90171.1	25	1
	(δ-toxin 2)	MTADIISTIGDFVKWILDTVKKFTK	A284_04275	AGC90172.1	25	0
PSMε(S. epidermidis) homologues
S. epidermidis	PSM ε	MFIINLVKKVISFIKGLFGNNENE			24	0
S. pseudinterme dius ED99	(PSM ε)	MFIIDLIKKVIEFLKGLFGNK			21	1
S. pseudinterme dius HKU10-03	(PSM ε)	MFIIDLIKKVIEFLKGLFGNK			21	1
S. lugdunensis N920143	(PSM ε)	MFIVNLIKKVIEFIKKLFGNK			21	3
S. lugdunensis HKU09-01	(PSM ε)	MFIVNLIKKVIEFIKKLFGNK			21	3
S. warneri SG1	(PSM ε)	MQFITDLIKKAVDFFKGLFGNK			22	1

¹if annotated. The psm genes in S. aureus and S. epidermidis are found and annotated in many strains. See Fig. 3 for gene numbers in strains USA300 and RP62A.

²encoded separately. Other psmβ genes are commonly grouped in one genetic locus.

S. aureus and S. epidermidis are the only species for which peptide purification approaches led to a complete list of produced PSMs and PSM-encoding genes (Mehlin et al., 1999, Vuong et al., 2004, Yao et al., 2005, Wang et al., 2007) (Fig. 3). S. aureus produces four PSMα peptides, encoded in the psmα locus, two PSMβ peptides, encoded in the psmβ locus, and the δ-toxin, which is encoded within RNAIII, the effector molecule of the Agr system (Wang et al., 2007). S. epidermidis produces the α-type PSMs PSMα, PSMδ, PSMε, and the δ-toxin (Mehlin et al., 1999, Vuong et al., 2004, Yao et al., 2005). PSMα and PSMδ are somewhat similar to the S. aureus PSMα peptides and encoded together in a locus that resembles the S. aureus psmα locus, indicating that they are evolutionarily related. Furthermore, S. epidermidis produces two PSMβ peptides. The S. epidermidis psmβ locus contains another gene, psmβ3, but we have never found a secreted peptide corresponding to that gene (Vuong et al., 2004, Queck et al., 2009). In some strains including the genome-sequence strains RP62A and ATCC12228, the psmβ1 gene is duplicated, leading to enhanced production of PSMβ1 (Wang et al., 2011).

Fig. 3. PSMs in S. aureus and S. epidermidis.

Amino acid sequences are shown at the top. Numbers at the right show the net charge of the peptides at pH 7.0, rounded to whole numbers, and considering N-formylation. Genes are shown at the bottom. Gene numbering is according to strains S. aureus USA300 FPR3757 (Diep et al., 2006) and S. epidermidis RP62A (Gill et al., 2005), respectively.

Notably, all known psm genes are encoded on the core genome or on MGEs that are present in virtually all strains of the species. As a noticeable exception, the PSM-mec peptide is encoded in staphylococcal cassette chromosome mec (SCCmec) types II, III and VIII, representing a rare example of combined presence of antibiotic resistance and virulence factors on a staphylococcal MGE (Queck et al., 2009, Chatterjee et al., 2011). In addition to the MRSA strains harboring these SCCmec elements, PSM-mec is frequently produced by methicillin-resistant S. epidermidis (Queck et al., 2009); and psm-mec genes are also found in other methicillin-resistant staphylococci (Monecke et al., 2012). The presence of a psmβ homologue on a plasmid of S. simulans ssp. staphylolyticus is a further exception to the common encoding of psm genes on the genome. Of note, to clearly distinguish PSM peptides carrying the same name, a PSM peptide should always be described together with the species, such as in “S. epidermidis PSMβ1”.

Interestingly, the PSM production pattern reflects the clearly distinct “lifestyles” of S. aureus and S. epidermidis. Despite the capacity of S. epidermidis to produce strongly aggressive and cytolytic PSMs such as PSMδ (Cheung et al., 2010), the organism shifts the production of PSMs to less aggressive forms, such as the PSMβ peptides (Otto, 2009, Wang et al., 2011). In contrast, many virulent S. aureus strains produce high amounts of the strongly cytolytic PSMα peptides and only low amounts of PSMβ peptides (Wang et al., 2007).

Regulation of PSMs

Similar to several other major staphylococcal exotoxins, PSMs are positively regulated by the global regulatory quorum sensing system, accessory gene regulator (Agr) (Wang et al., 2007, Queck et al., 2008) (Fig. 4). Notably, Agr thus connects the response to cell density (quorum-sensing) with virulence (Novick, 2003). The agr locus is comprised of a divergent promoter system that transcribes two important mRNAs. RNAIII is a regulatory RNA that controls the transcription of several major exotoxins and surface proteins, and also contains the gene coding for the PSM δ-toxin, named hld for hemolysin delta. RNAII codes for two genes belonging to a classical two-component signal transduction system, a histidine kinase (agrC), and a response regulator (agrA), and two genes that code for an auto-inducing peptide (AIP) precursor (agrD) and an AIP maturation and export protein (agrB). AgrA becomes activated when AgrC detects a critical concentration of secreted AIP. In turn, activated AgrA induces the transcription of rnaiii and its own operon (rnaii), constituting a positive feedback loop. A functional Agr regulatory system is essential for S. aureus virulence as demonstrated by severe attenuation of pathogenicity in isogenic agr mutants (Abdelnour et al., 1993, Cheung et al., 1994, Gillaspy et al., 1995, Cheung et al., 2011).

Fig. 4. Regulation of PSMs by Agr.

The Agr quorum-sensing circuit, as described in the text, is shown at the top. Direct regulation of the psmα and psmβ operons by AgrA is shown left (blue shadowing) and RNAIII-dependent Agr control of other target genes at the right (yellow shadowing). The likely evolution of the connection of quorum-sensing, RNAIII-dependent and –independent control is apparent in the encoding of the PSM δ-toxin (hld) by RNAIII. Agr also controls the psm-mec gene in an RNAIII-independent manner, but binding of AgrA to the psm-mec promoter has not yet been demonstrated.

In contrast to the regulation of virtually all Agr target genes via RNAIII, psm genes are regulated by direct binding of AgrA to their promoters (Queck et al., 2008). This represents an exceptionally direct control by the Agr system, indicating that cell density-dependent regulation of PSMs preceded a later connection to further virulence determinants. The fact that the PSM δ-toxin is encoded within rnaiii, which thus probably evolved around the δ-toxin gene hld, further supports this notion. Although the regulation of PSM transcription is dominated by AgrA (Queck et al., 2008, Chatterjee et al., 2011), additional levels of PSM regulation exist. For example, the biofilm mode of growth appears to lead to enhanced relative production of β-type PSMs over α-type PSMs, as shown in S. epidermidis (Wang et al., 2011), but how differential production of α- versus β-type PSMs is achieved mechanistically is not known. Recent studies show that the PSMs, in addition to other toxins, are degraded by aureolysin, a protease which is regulated by several global regulators, which include Agr, staphylococcal accessory regulator (SarA), and SaeRS (Zielinska et al., 2011, Cassat et al., 2013, Rasigade et al., 2013). Accordingly, SarA impacts PSM production by a general impact on protease activity (Zielinska et al., 2011).

The psm-mec gene is embedded in a gene encoding a regulatory RNA (Kaito et al., 2011), in a way very similar to the connection of hld with RNAIII (Fig. 5). The psm-mec regulatory RNA decreases agrA activity (Kaito et al., 2013), and thus has an indirect effect on PSM production. However, the effect of the psm-mec regulatory RNA on PSM production is quite low in extent, reaching a maximal factor of ~ 3, and is only seen in some strains (Kaito et al., 2013). Lastly, Agr also regulates transcription of psm-mec itself in an RNAIII-independent manner (Chatterjee et al., 2011).

Fig. 5. PSM-mec.

The PSM-mec peptide and the psm-mec RNA surrounding the PSM-mec-encoding gene are encoded on SCCmec elements of types II, II, and VIII, of which SCCmec type III is shown here as example.

S. aureus can survive within a variety of host cells to hide from the immune system and disseminate to other sites in the host (Gresham et al., 2000, Kubica et al., 2008, Giese et al., 2011, Thwaites & Gant, 2011, Cassat et al., 2013). Importantly, intracellular environments were found to activate the expression of PSMs (Geiger et al., 2012, Surewaard et al., 2012). When phagocytosed by polymorphonuclear leukocytes (PMNs, or neutrophils), S. aureus quickly produces high levels of the intracellular signaling molecule, (p)ppGpp, which activates the production of PSMs by an unknown mechanism (Geiger et al., 2012). Unlike the Agr quorum sensing system, which requires time for the accumulation of enough AIP for activation of AgrA (Carnes et al., 2010, Pang et al., 2010), activation of the stringent response is independent of AgrA. It is likely that both the stringent response (Geiger et al., 2012) and Agr (Surewaard et al., 2012) are induced by S. aureus to ensure that PSMs are produced immediately and continually in intracellular environments, maximizing the capacity of S. aureus to trigger phagosomal escape and subsequent cell lysis.

Finally, it was reported that PSM transcription is up-regulated, in addition to other important virulence factors, in the presence of calf serum (Oogai et al., 2011). It is tempting to speculate that S. aureus may up-regulate PSM production as a means to compensate for the observed sequestration of PSMs by lipoproteins in human serum (Surewaard et al., 2012). Taken together, while the Agr global regulatory system is by far the main effector of PSM production, other factors, which include the regulatory psm-mec mRNA, proteolytic degradation, and various responses to the intracellular and extracellular environments, all play important roles affecting PSM concentrations in vitro and in vivo.

Export of PSMs

The fact that PSMs carry an N-terminal N-formyl methionine indicates that they are secreted without a signal peptide using a non-canonical protein export system. Recently, the PSM export system was identified to be an ABC transporter, which was named phenol soluble modulin transporter (Pmt) (Chatterjee et al., 2013) (Fig. 6). The pmt locus encodes four genes, pmtA-D, of which pmtA and pmtC code for ATPases, whereas pmtB and pmtD encode membrane-associated proteins. In S. aureus, Pmt was shown to be able to export all types of PSMs into the surrounding milieu. Pmt is present in all Staphylococcus species and absent from non-staphylococcal bacterial genomes, suggesting that Pmt is specific for staphylococci, as is PSM production. Notably, Pmt is essential for bacterial survival, because in its absence PSMs accumulate in the cytoplasm, causing acute growth deficiency and cellular defects. As expected, Pmt is required for mediating PSM-associated virulence phenotypes. Due to its key impact on virulence and additional essentiality for bacterial survival, and because it is highly conserved among the staphylococci, Pmt appears to be a exceptionally promising target for interference with staphylococcal pathogenesis (Chatterjee & Otto, 2013).

Fig. 6. PSM export.

The genetic locus encoding the four Pmt ABC transporter components is shown at the top (a). Numbers refer to gene numbers in the USA300 FPR3757 genome (Diep et al., 2006). (b) Pmt functions. (c) Consequences of Pmt absence/blocking.

PSMs in the natural habitat of staphylococci

PSMs have been attributed at least two main functions that are related to their detergent-like features and may facilitate the life of staphylococci in their natural habitat, that is, on the epithelial surfaces of humans and other mammals. First, the observed PSM-dependent spreading of S. aureus on agar plates most likely translates to PSMs having a key function in the spreading of staphylococci on epithelial surfaces (Tsompanidou et al., 2013). Second, PSMs may assist in the emulsification of nutrients in those environments, although this remains to be demonstrated. Of note, PSMs are produced at the high levels that these functions presumably require. In fact, ~ 60% of the secreted protein mass in S. aureus in-vitro culture was shown to be PSMs (Chatterjee et al., 2013). Finally, PSMs are responsible for biofilm structuring (Wang et al., 2011, Periasamy et al., 2012), which is discussed further below. As biofilm-like bacterial agglomerations are believed to exist on epithelial surfaces, this PSM function likely also contributes to the physiology of staphylococci in their commensal state. However, in general we still lack detailed information on the role of PSMs in the non-infectious lifestyle of staphylococci.

PSMs as cytolysins

Several PSMs have an extraordinary ability to lyse many eukaryotic cell types, such as bone cells (osteoblasts) (Cassat et al., 2013, Rasigade et al., 2013), endothelial and epithelial cells (Giese et al., 2011), monocytes (Wang et al., 2007), erythrocytes (Cheung & Otto, 2011), and most importantly, PMNs (Wang et al., 2007, Cheung et al., 2010, Forsman et al., 2012). The wide variety of cell types that are subject to PSM-mediated lysis is in accordance with the notion that PSM cytolytic activity is due to membrane perturbation in a receptor-independent fashion (Kretschmer et al., 2010) (Fig. 7). However, not all PSMs show this pronounced cytolysis of eukaryotic cells. For instance, the PSM β-type peptides are generally much less cytolytic than the α-type peptides. Whether this is due to their commonly negative net charge is not known. Notably, S. aureus PSMα3 and S. epidermidis PSMδ are by far the most potent cytolytic PSMs described to date, with pronounced cytolytic potencies observed in the low micromolar range (Wang et al., 2007, Cheung et al., 2010). Interestingly, the cytolytic activities of the PSMs are enhanced in the presence of S. aureus β-toxin (Cheung & Otto, 2011) and PVL (Hongo et al., 2009) in addition to antimicrobial peptides (Cogen et al., 2010), signifying that the PSMs may not always exert their cytolytic effects independently.

Fig. 7. PSM-mediated cytolysis.

Likely mechanism of receptor-independent membrane attachment and disintegration, and subsequent pore formation by PSMs. This model is based on experimental evidence obtained with S. aureus δ-toxin, which forms receptor-independent short-lived pores in artificial membranes (Verdon et al., 2009). (b) Mechanism of intracellular neutrophil killing by PSMα peptides of S. aureus (Surewaard et al., 2013).

Much of our understanding about the lytic nature of the PSMs stems from extensive research performed with S. aureus δ-toxin, which forms small transient pores in artificial lipid vesicles (Pokorny et al., 2002). Phosphatidylcholine (POPC), which is a major phospholipid constituent of eukaryotic membranes (Devaux & Morris, 2004), is generally used to study membrane perturbation. A more detailed study of the interaction of the PSMs of S. aureus and S. epidermidis with POPC vesicles revealed that POPC-lytic activities of PSMs did not correlate with those previously described for leukocytes and erythrocytes (Duong et al., 2012). In particular the PSMβ peptides showed strongly lytic behavior toward POPC phospholipid vesicles, implying that the pore-forming activities of PSMs are more complex than originally thought and possibly dependent on the membrane lipid composition. Some PSMs can lyse bacterial protoplasts (Kreger & Bernheimer, 1971) and even whole bacteria, such as Streptococcus pyogenes and Legionella pneumophilla (Verdon et al., 2008, Cogen et al., 2010, Joo et al., 2011). This is discussed further below.

A hallmark of a typical subcutaneous S. aureus infection is the rapid infiltration of PMNs. PMNs are important host innate immune cells that have a key role in controlling local infections by taking up bacteria by phagocytosis (Amulic et al., 2012). Phagocytosed bacteria, compartmentalized into phagosomes, are exposed to a battery of antimicrobial agents. However, virulent S. aureus strains such as CA-MRSA strains have developed strategies to evade killing by cells of the host innate immune system (Voyich et al., 2005, Kobayashi et al., 2010, Rigby & DeLeo, 2012). These strategies include the ability to survive within the PMNs, which provides protection from detection by the immune system and also represents a means of transportation to disseminate to other sites of the body (Rogers, 1956, Gresham et al., 2000, Thwaites & Gant, 2011). Several recent studies have shown that PSMα peptides are responsible for the lysis of neutrophils after ingestion (Geiger et al., 2012, Chatterjee et al., 2013, Surewaard et al., 2013) (Fig. 7). Furthermore, expression levels of PSMs correlate with cytotoxicity within PMNs (Rasigade et al., 2013). Among the classical bicomponent S. aureus leukocidins, only LukAB (also known as LukGH) (Ventura et al., 2010, Dumont et al., 2011) contributes to PMN lysis after phagocytosis (Ventura et al., 2010, DuMont et al., 2013). PSMα expression is also important for the phagosomal escape or survival of S. aureus within osteoblasts (Rasigade et al., 2013), professional phagocytes other than neutrophils, and non-professional phagocytes such as endothelial cells (Grosz et al., 2013), indicating a general crucial role in intracellular survival of S. aureus.

Recognition of PSMs by the host

The PSMs are potent pro-inflammatory agents that mediate their activities primarily through the interaction with human formyl peptide receptor 2 (FPR2) (Kretschmer et al., 2010) (Fig. 8). FPR2 belongs to a family of G protein-coupled receptors that aid cells of the innate immune system to recognize pathogen-associated molecular patterns (PAMPs) produced by bacteria. In humans, three FPR paralogs exist; FPR1, FPR2/ALX (previously known as FPR-like 1), and FPR3. FPR1 and FPR2 are expressed on monocytes, macrophages, PMNs, immature dendritic cells and some other cell types, while FPR3 is only present on monocytes and macrophages (Migeotte et al., 2006).

Fig. 8. PSM receptor interaction.

In the nanomolar range, PSMs bind to and activate FPR receptors, with by far the strongest activation occurring with FPR2. Pro-inflammatory, FPR2-mediated activities of PSMs include neutrophil activation, chemotaxis, and release of specific cytokines such as IL-8. N-formylated and N-deformylated PSMs activate FPR2 receptors, with activation by formylated PSMs being somewhat stronger. The S. aureus virulence factors CHIPS and FLIPr block recognition by FPR1 and FPR2, respectively.

The FPRs detect N-formyl peptides, such as the N-formylated PSMs. However, they were shown to also recognize non-formylated forms of PSMs (Kretschmer et al., 2011), which are secreted under certain conditions (Somerville et al., 2003). At nanomolar concentrations, the PSMs activate many FPR2 receptor-dependent effector functions, which include the induction of chemotaxis, exocytosis and superoxide generation, pro-inflammatory cytokine production, up-regulation of CD11b expression, and the induction of intracellular calcium flux (Fu et al., 2006, Wang et al., 2007, Queck et al., 2009, Kretschmer et al., 2010, Chaffin et al., 2012, Forsman et al., 2012). These pro-inflammatory activities are essential for an effective immune response to invading pathogens. The activation of FPR2-dependent pro-inflammatory activities by non-formylated PSMs from S. aureus and S. epidermidis is generally less pronounced than that by PSMs with an N-formyl methionine residue (Kretschmer et al., 2011).

The specificity of the PSMs for FPR2 was discovered with the help of two small protein inhibitors (the chemotaxis-inhibitory protein of S. aureus (CHIPS) (de Haas et al., 2004) and FPR2/ALX-inhibitory protein (FLIPr) (Prat et al., 2006)). These inhibitors, secreted by S. aureus, block the ability of FPR1 and FPR2 to recognize formylated peptides, respectively. Based on the specificity of FPR2 for the detection of PSMs, it is thought that the innate immune system may discriminate between virulent S. aureus strains, such as CA-MRSA, and commensal staphylococci, which generally have a reduced PSM output (Kretschmer et al., 2010). Notably, FPR2-specific inhibition by FLIPr dramatically reduces chemotaxis and IL-8 release in neutrophils as stimulated by culture filtrate of the CA-MRSA strain USA300 (Kretschmer et al., 2010), emphasizing the importance of PSM recognition in the interaction of the innate immune system with S. aureus. Evidently, PSMs and FLIPr have an opposite impact on FPR2 activation. Most likely, production of FLIPr and PSMs is controlled by the bacteria in a way that allows establishment of an infection without immune recognition. This is achieved by the strict control of PSM production by the Agr quorum-sensing system (Queck et al., 2008), delaying PSM production until the bacterial density is high. Then, strong PSM production likely outcompetes FLIPr to promote inflammation via FPR2. However, the in-vivo kinetics of FLIPr and PSM production still need to be investigated. Similarly, it is yet unknown whether PSM-dependent induction of inflammatory responses is part of the bacterial pathogenesis program or PSMs represent pathogen-associated molecular patterns that the immune system recognizes to launch a defensive response.

While the PMNs are critical for innate immune defense, macrophages and dendritic cells (DCs) play an essential role in linking innate and adaptive immunity by serving as antigen presenting cells (APCs). In particular, APCs process and present antigens to T-lymphocytes, which then initiate the stimulation of an adaptive immune response. Interestingly, a recent study demonstrated that PSMs modulate murine DC function; PSM-stimulated DCs increased in vitro priming of regulatory T cells, which led to a suppression of Th1 immune responses (Schreiner et al., 2013). Murine macrophages stimulated with isogenic psmα and hld CA-MRSA mutants showed decreased production of IL-6, IL-12 and TNFα cytokines as compared to stimulation with the corresponding wild-type strain (Spentzas et al., 2011) confirming that both PSMα peptides and δ-toxin are important mediators of inflammatory cytokine production. Taken together, the PSMs appear to be capable of manipulating the host immune system in various ways to promote survival of S. aureus during infection.

PSMs in different infection types

PSMs in skin and soft-tissue infections and sepsis

In humans, skin and soft tissue infections represent ~ 90% of all clinical manifestations of community-associated S. aureus disease (Deleo et al., 2010). However, S. aureus can cause severe invasive disease resulting in bacteremia (sepsis). Even under antibiotic therapy, invasive S. aureus disease is associated with high mortality (Klevens et al., 2007). Notably, on average, greater levels of PSMs are detected in the supernatants of CA-MRSA strains as compared to HA-MRSA strains, suggesting a correlation with the virulence potential of those strains (Wang et al., 2007, Li et al., 2010). Accordingly, isogenic psmα CA-MRSA mutants were severely attenuated in their ability to cause subcutaneous abscesses in the skin of mice compared to the wild-type strain (Wang et al., 2007) (Fig. 9a). Furthermore, mice infected with isogenic psmα CA-MRSA mutants were significantly delayed in reaching moribund status compared to mice infected with the wild-type strain in a bacteremia model of infection (Wang et al., 2007), confirming a link between high expression of PSMα peptides and the virulence potential of CA-MRSA. In contrast, psmβ deletion mutants had none and hld mutants only a slight impact on disease progression in these mouse models. Similarly, in a rabbit model of skin infection, a psmα deletion mutant was strongly impaired in forming abscesses, to an extent similar to that caused by an agr deletion mutant and clearly exceeding that of a lukSF (PVL) deletion mutant, which did not show a significantly different ability to form abscesses compared to the wild-type strain in that model (Kobayashi et al., 2011) (Fig. 9b).

Fig. 9. PSMα peptides and skin infection.

Representative pictures of abscesses formed by the USA300 CA-MRSA strain and its isogenic psmα deletion mutant in mice (Wang et al., 2007). (b) Comparative analysis of virulence determinants in a rabbit skin infection model (Kobayashi et al., 2011). The abscess volume is shown on the y-axis. Significantly smaller abscesses compared to the wild-type strain were found with the hla, psmα, and agr, but not lukSF mutants.

As for the impact of the psm-mec locus on virulence, remarkable differences were reported in murine models of skin and bacteremia infections (Queck et al., 2009, Kaito et al., 2013). In S. aureus strains that produce relatively high levels of PSM-mec, a significant attenuation in virulence was observed in isogenic S. aureus psm-mec mutants, suggesting that the PSM-mec peptide contributes to virulence similarly to the α-class PSMs (Queck et al., 2009). In contrast, deletion of the psm-mec locus caused an increase in virulence in some HA-MRSA isolates (Kaito et al., 2013) as a result of the negative effect of the psm-mec RNA on Agr, but not in others (Chatterjee et al., 2011, Kaito et al., 2013). The impact that the psm-mec locus has on virulence thus is highly strain-dependent and appears to result from a combination of the PSM-mec peptide, which has a positive effect, and the psm-mec RNA, which has a negative effect on virulence.

PSMs in biofilm-associated infection

Biofilms are surface-attached agglomerations with a distinct physiology and pronounced capacity to withstand attacks by host defenses and antibiotics (Costerton et al., 1999). Biofilm-associated infections often develop on indwelling medical devices; staphylococci are the most frequent pathogens involved with such infections (Otto, 2008). The factors that determine aggregation of cells in a biofilm, such as the production of an extracellular matrix, are fairly well understood. This includes staphylococci, in which the matrix consists of proteins, extracellular DNA, teichoic acids, and a specific exopolysaccharide called polysaccharide intercellular adhesin (PIA) or poly-N-acetylglucosamine (PNAG), according to its chemical composition. In contrast, the factors that contribute to the structuring of biofilms and the detachment of cells from biofilms, a process vital for the systemic dissemination of a biofilm infection, have remained poorly understood until recently (Otto, 2013).

S. epidermidis in particular, often regarded as an opportunistic pathogen, is the leading cause of chronic catheter-related infections in hospital settings (Otto, 2009, Rogers et al., 2009). S. epidermidis has a strong aptitude to form biofilms on inert surfaces (Kogan et al., 2006, Rohde et al., 2007). PSMβ peptides are strongly produced in S. epidermidis, and in a relatively higher proportion compared to other PSMs during biofilm as compared to planktonic growth (Wang et al., 2011). PSMβ peptides were shown to structure S. epidermidis biofilms, leading to the characteristic biofilm structure with channels deemed important for nutrient delivery to all biofilm layers (Wang et al., 2011) (Fig. 10). Furthermore, these peptides were found to be crucial for the detachment of biofilm clusters in vitro and dissemination from biofilms on indwelling devices in vivo, as demonstrated in a mouse model of indwelling device-associated infection (Wang et al., 2011). In S. aureus, all PSM peptides were shown later to contribute to these phenotypes in vitro and in vivo (Periasamy et al., 2012). The extent to which other PSMs of S. epidermidis contribute to biofilm structuring and detachment has not yet been analyzed using gene deletion mutants. The finding that the PSM surfactant peptides are responsible for biofilm structuring and detachment are in line with similar findings in other bacteria (Branda et al., 2001, Boles et al., 2005, Angelini et al., 2009), indicating a common mechanism involved in these phenotypes that is dependent on surfactant molecules. However, the chemical nature of the surfactants underlying biofilms structuring is different in different bacteria. Notably, the experiments on biofilm dissemination in S. epidermidis and S. aureus biofilm-associated infection were the first to show the impact of biofilm-structuring surfactants on dissemination from in-vivo biofilms.

Fig. 10. PSMs and biofilm structuring.

Confocal laser scanning microscopy pictures of biofilms formed by an S. epidermidis wild-type strain (strain 1457) and an isogenic deletion mutant of the psmβ operon (Wang et al., 2011). Note the thicker biofilm formed by the mutant and the absence of channels in the biofilm.

A recent report demonstrated that PSMs can polymerize to form fibrillar, amyloid structures that help stabilize S. aureus biofilms in vitro (Schwartz et al., 2012). Interestingly, the amyloid structures isolated from S. aureus cultures consist of only deformylated PSMs. However, further research is needed to investigate whether these structures exist in vivo and how they fit into the pathogenic lifestyle of S. aureus and S. epidermidis during biofilm infection.

PSMs in bone infection (osteomyelitis)

S. aureus is the leading cause of bone infections (osteomyelitis) in adults and infants (Arnold et al., 2006, Vander Have et al., 2009). CA-MRSA-related osteomyelitis in infants is often associated with increased interval and severity compared to HA-MRSA-related osteomyelitis (Martinez-Aguilar et al., 2004, Arnold et al., 2006, Bocchini et al., 2006). Intracellular residence within osteoblasts appears to be essential for the chronic presence and dissemination of S. aureus (Tuchscherr et al., 2010, Tuchscherr et al., 2011). Recently, the global S. aureus regulators Agr, SarA, and SaeRS were shown to regulate the expression of aureolysin, a protease responsible for degrading important S. aureus exoproteins (Zielinska et al., 2011, Gonzalez et al., 2012, Cassat et al., 2013, Rasigade et al., 2013). Although there is conflicting evidence supporting the roles SaeRS and SarA in aureolysin-mediated degradation of exoproteins (Cassat et al., 2013, Rasigade et al., 2013), it is clear that PSMα peptides are crucial for cytotoxicity toward murine and human osteoblasts in vitro (Cassat et al., 2013, Rasigade et al., 2013) and that they are subject to aureolysin-mediated degradation during bone infection (Cassat et al., 2013). Furthermore, PSMα peptides were essential for bone remodeling and destruction in a mouse model of osteomyelitis infection as visualized by high-resolution microcomputed tomography (Cassat et al., 2013). Taken together, these results implicate the PSMα peptides as important virulence determinants in the pathogenesis of osteomyelitis.

Delta-toxin and atopic dermatitis

Atopic dermatitis (AD) is a chronic inflammatory skin disease that commonly affects infants (Williams & Flohr, 2006). Mast cells are believed to be key contributors to the pathology of AD and other allergic diseases through the release of inflammatory mediators from intracellular granules activated by antigen cross-linked to Fc3RI receptor-bound IgE (Galli et al., 2011, Galli & Tsai, 2012). Although more than 90% of AD patients are tested positive for S. aureus (Rudikoff & Lebwohl, 1998), the link between S. aureus and AD pathogenesis has remained unclear. However, a recent study implicates a link between S. aureus colonization and allergic skin disease and correlates these observations with the potent mast cell-degranulating activity of δ-toxin (Nakamura et al., 2013). Notably, other S. aureus PSMs lacked such activity. Importantly, non-cytolytic concentrations of both formylated and non-formylated forms of δ-toxin are equally potent in stimulating mast cell degranulation, these activities are not dependent on FPR2, and the intracellular signaling pathways stimulated by δ-toxin leading to mast cell degranulation are different from those induced by antigen cross-linked to IgE. The ability of δ-toxin to induce mast cell degranulation was also confirmed in vivo using in murine models of inflammatory skin disease. Furthermore, skin colonization with an isogenic S. aureus hld mutant promoted less IgE and IL-4 production and inflammatory skin disease when compared to mice colonized with the wild-type δ-toxin expressing S. aureus strain. Moreover, enhancement of IgE production and skin inflammation by δ-toxin was abolished in mast cell-deficient mice and restored by adoptive transfer of mast cells (Nakamura et al., 2013).

Commensal staphylococci such as S. epidermidis, which are not linked to AD, also produce δ-toxin peptides that are highly similar to that produced by S. aureus (Scheifele et al., 1987). S. epidermidis δ-toxin showed some, but reduced activity compared to S. aureus δ-toxin in causing mast cell degranulation (Nakamura et al., 2013). Whether δ-toxin producing S. epidermidis isolates contribute to allergic skin disease requires further investigation.

Relationship of PSMs with other S. aureus toxins

Clearly, the PSMs are not the only toxins that contribute to skin and soft tissue and many of the other infections mentioned above. For instance, α-toxin (Hla), an important pore-forming and pro-inflammatory (Thelestam et al., 1973, Bhakdi & Tranum-Jensen, 1991, Walev et al., 1993, Ezepchuk et al., 1996, Valeva et al., 1997, Hocke et al., 2006, Liang & Ji, 2006, Bartlett et al., 2008) toxin, contributes to abscess formation in mice (Inoshima et al., 2011) and rabbits (Kobayashi et al., 2011). Furthermore, bi-component pore-forming leukotoxins are increasingly recognized as important virulence determinants involved in S. aureus pathogenesis. These leukotoxins include LukED, LukAB (LukGH) and PVL (Diep et al., 2010, Alonzo et al., 2011, Dumont et al., 2011). They are cytolytic towards cells of myeloid origin of different animal species. Notably, cytolysis triggered by those toxins has recently been found to be receptor-dependent (Alonzo et al., 2013, DuMont et al., 2013, Spaan et al., 2013), as has α-toxin-dependent cell lysis and pro-inflammatory activities (Wilke & Bubeck Wardenburg, 2010, Inoshima et al., 2011). Studies using isogenic S. aureus deletion mutants showed that LukED and LukAB contribute to the virulence of S. aureus in murine sepsis (Alonzo et al., 2011) and renal abscess models (Dumont et al., 2011), respectively. The contribution of PVL to the pathogenesis especially of CA-MRSA-induced disease is controversial (Otto, 2010, Otto, 2011).

While PSMs play a key role as virulence determinants during many types of S. aureus infection, the pathogenesis of S. aureus disease is multi-factorial. The PSMs may work in concert with other toxins, possibly in a synergistic fashion as reported for PVL and PSMs (Hongo et al., 2009), to exert maximal pathogenesis as in the case of skin and soft tissue infections. Notably, unlike α-toxin, LukAB, LukED and PVL, the cytolytic activity of the PSMs very likely does not require a receptor (Kretschmer et al., 2010). Thus, the non-specific lytic nature of the PSMs may provide a distinct advantage over the aforementioned toxins, inasmuch as they may attack a wide variety of cell types, and offer a possible explanation as to why the PSMs contribute to pathogenicity in a variety of S. aureus diseases.

PSMs as antimicrobials

Although periodically, antimicrobial activities of δ-toxin analogs and other PSM-like peptides of staphylococci have been reported (Beaudet et al., 1982, Dhople & Nagaraj, 1995, Dhople & Nagaraj, 2005), more intensive studies on such activities of PSMs have only been performed more recently (Tab. 2). In these studies, PSMδ and δ-toxin of S. epidermidis were found to be active predominantly against Streptococcus pyogenes (Cogen et al., 2010). Furthermore, S. epidermidis δ-toxin recruited host antimicrobial peptides (AMPs), such as LL-37, CRAMP, and hBDs, to reduce survival of S. pyogenes in a synergistic fashion (Cogen et al., 2010). In another study, most intact PSMs exhibited very low or no antimicrobial activity on S. pyogenes, but proteolytically processed PSMα1 and PSMα2 showed significantly increased activities (Joo et al., 2011). The N-terminal two amino acids, formyl-methionine and glycine, of these PSMs are missing in the antimicrobially active derivatives. Whether these processed PSMs contribute to bacterial interference in vivo is unclear. Also, which proteases are involved in the processing is not known, but recent findings on additional proteolytically processed derivatives of PSMα1 and PSMα4 implicates aureolysin. Overall, the biological significance of the antimicrobial activities found in some PSMs is unclear. The facts that these activities are rare and the antimicrobially active processed PSMα peptide derivatives are only produced in low amounts support the notion that they are a side effect of the general membrane-damaging effect of PSMs. The features that most PSMs exhibit suggest that they evolved predominantly to harm eukaryotic rather than prokaryotic membranes. Nevertheless, that staphylococci are protected from the antimicrobial effects of PSMs by the Pmt export system (Chatterjee et al., 2013) indicates that they may at least theoretically contribute to bacterial interference with other genera, which remains to be clearly demonstrated in real-life bacterial communities.

Table 2.

PSMs with antimicrobial activities.

Strain	PSM type	Derivatization	Target	MBC (μM)	Note	Reference
CoNS isolate	unknown	-	Neisseria gonorrhoeae G-10		16 kD protein dissociated to 1.4 kD	(Beaudet, et al., 1982)
S. epidermidis	δ-toxin	intact	S. aureus 113	4–8	KAM2 used except for S. pyogenes (25% Todd-Hewitt broth, 75% PBS)	(Cogen, et al., 2010)
			S. pyogenes, NZ131	16
			Escherichia coli	8–16
	PSMδ	intact	MRSA, Sanger 252	16–32
			MRSA, USA300	32
			S. aureus,113	8
			S. pyogenes NZ131	16
			Escherichia coli	8
S. aureus	δ-toxin	synthesized and modified 16 to 26 residues-derivatives	Escherichia coli	2.1–3.91	intact S. aureus δ-toxin is not antimicrobial	(Dhople & Nagaraj, 1995, Dhople & Nagaraj, 2005)
			S. aureus	3.7–221
	PSMα1	N-terminal two residues are absent	S. pyogenes	8	THY2 used	(Joo, et al., 2011)
		N-terminal five residues are absent	S. pyogenes	11		(Gonzalez, et al., 2012)
			S. epidermidis, ATCC35984	42
	PSMα2	N-terminal two residues are absent	S. pyogenes	8	THY used	(Joo, et al., 2011)
			Staphylococcus epidermidis 12228	16	TSB2 used
	PSMα3	Synthesized alanine exchange derivatives: F3A, V4A, F8A, F10A, F11A, L14A, F18A, L19A	S. pyogenes, Micrococcus luteus	3		(Cheung, et al., 2013)
	PSMα4	N-terminal three residues are absent	S. pyogenes	10	aureolysin-dependent	(Gonzalez, et al., 2012)

¹MIC (μM)

²KAM, killing assay media (Dorschner, et al., 2006); THY, Todd-Hewitt broth containing 0.2% yeast extract; TSB, tryptic soy broth

³determined by growth inhibition zones

Structure-function analysis of PSMs and role of cytolytic versus pro-inflammatory capacity during infection

Recently, an alanine substitution peptide library of PSMα3 was used to investigate which structural determinants are responsible for the diverse biological activities of PSMs (Cheung et al., 2013) (Fig. 11). PSMα3 was chosen because it is the most cytolytic PSM and significantly contributes to all phenotypes associated with PSMs, including hemolysis, biofilm structuring and pro-inflammatory activities (Wang et al., 2007, Kretschmer et al., 2010, Cheung & Otto, 2011, Periasamy et al., 2012). The results obtained with the alanine substitution peptide library of PSMα3 identified several important variants that have unique phenotypes. Namely, PSMα3 L7A and K17A showed reductions in cytolytic but retained pro-inflammatory activities, whereas PSMα3 N21A and N22A showed the reverse phenotype. PSMα3 K6A and PSMα3 K12A showed marked reductions in both cytolytic and pro-inflammatory activities. To investigate whether the in-vitro pro-inflammatory phenotypes of those derivatives were also reflected in vivo, a murine peritonitis model was used to measure the numbers of infiltrating leukocytes after exposure to select PSMα3 alanine variants. Interestingly, only PSMα3 L7A and K17A induced levels of leukocyte influx similar to those induced by wild-type PSMα3. S. aureus constructs expressing these PSMα3 derivatives from the genome will be valuable to investigate the biological role of PSM-receptor interaction in S. aureus disease in vivo. Furthermore, the identified non-toxic PSMα3 derivatives may serve to develop a toxoid vaccine.

Fig. 11. PSM structure-function analysis.

An alanine exchange library of the PSMα3 peptide was created and used to determine changes in PSM function caused by structural changes (Cheung, et al., 2013). Overall, large hydrophobic side chains impacted biofilm-structuring capacities (such as some phenylalanine residues shown in regular font) and predominantly cationic amino acids impacted cytolytic and pro-inflammatory functions. In addition, the C-terminus exclusively affected receptor-mediated but not cytolytic activities. Finally, removal of large hydrophobic side chains such as all phenylalanine residues led to derivatives that showed antimicrobial activities, indicating that these large hydrophobic side chains are involved in preventing cytolysis of bacterial versus eukaryotic membranes in PSMs.

Changes at the hydrophobic side of PSMα3, especially the exchange of large side chains such as phenylalanine, led to a reduction in biofilm-structuring activities and sometimes to antimicrobial activity toward S. pyogenes not observed in native PSMα3 (Cheung et al., 2013). The latter suggests that large hydrophobic side chains in PSMs prevent the lysis of bacterial versus that of eukaryotic cells. Accordingly, PSMα3, which has only strong capacity to lyse eukaryotic cells, has five phenylalanine residues, whereas PSMδ of S. epidermidis, which lyses both eukaryotic and prokaryotic cells, has only one.

Diversification of PSMs and PSM production

Based on recent findings, it has become evident that PSMs have undergone divergent evolution to develop different functions. For example, many α-type, but not β-type, PSMs are cytolytic and only the δ-toxin leads to mast cell degranulation (Wang et al., 2007, Cheung et al., 2010, Nakamura et al., 2013). The PSMβ peptides do not appear to have a distinct role, unless this role has not yet been identified. However, their lack of cytolytic function may be beneficial for commensal bacteria such as S. epidermidis in their effort to use PSM peptides in a “non-aggressive” fashion primarily to structure biofilms and facilitate life on epithelial surfaces, without simultaneously harming host cells.

While the capacity to produce strongly cytolytic PSMs is not limited to S. aureus, as we know from the cytolytic PSMδ of S. epidermidis (Cheung et al., 2010), only S. aureus appears to produce high amounts of cytolytic, “aggressive” PSMs to lyse host cells and thereby evade elimination by host defenses. Additionally, S. aureus uses increased production of PSMs to augment virulence, as seen with highly virulent S. aureus strains such as CA-MRSA as compared to HA-MRSA (Wang et al., 2007, Li et al., 2009). Whether S. epidermidis or other staphylococcal species up-regulate production of “aggressive” PSMs under conditions of systemic invasion or specific disease conditions is not known.

Targeting PSMs for anti-staphylococcal drug development

Historically, inactivated bacterial toxins proved to be successful candidates for active immunization to protect against bacterial disease, of which, the diphtheria, tetanus, acellular pertussis (DTaP) vaccine is a classic example (Adkins et al., 2012). However, unlike in S. aureus, the toxins produced by Corynebacterium diphtheria, Clostridium tetani, and Bordetella pertussis are, for the most part, the only and primary toxins that are responsible for disease development. In contrast, S. aureus produces a plethora of toxins that have different contributions to different S. aureus diseases. As an additional complication, the toxins produced by S. aureus are not uniformly expressed among all isolates, making the identification of a toxin target for immunotherapy against S. aureus disease challenging (Proctor, 2012). With the exception of α-toxin (Kennedy et al., 2010) and PVL (Brown et al., 2009), few other S. aureus toxins have been thoroughly investigated as candidates for active vaccination against a variety of pathogenic staphylococci (Spellberg & Daum, 2011).

Targeting the PSMs for anti-staphylococcal therapy and drug development would be advantageous for several reasons: the PSMs form a major constituent of the S. aureus exoprotein repertoire; they have an integral role in different facets of S. aureus disease; they show little variation in amino acid sequences between S. aureus strains; and the genome-encoded psm genes are present in all sequenced S. aureus strains (Wang et al., 2007). Therefore, neutralizing the cytolytic and pro-inflammatory activity of all PSMs would reduce their potency against host cells and possibly their overall contribution towards S. aureus disease progression.

Although protective humoral immune responses to S. aureus infection are generally not observed (Spellberg & Daum, 2011), a recent serological study showed a correlation between increased antibody levels to S. aureus exotoxins (α-toxin, PVL, S. aureus enterotoxin C1, δ-toxin, and PSMα3) and a lower incidence of sepsis in hospitalized adults with invasive S. aureus infections (Adhikari et al., 2012). These data indicate that the PSMs are immunogenic, since naturally circulating anti-PSM antibodies are detected in human adults (Adhikari et al., 2012). Furthermore, this observation correlates with studies that show robust anti-PSM antibody responses in mice and rabbits after active immunization with PSMs (Nolte & Kapral, 1981, Wang et al., 2011, Spaulding et al., 2012). In addition, rabbits immunized with a combination of PSMs show weak protection in a bacteremia infection model (Spaulding et al., 2012). The protective effect may be improved with the discovery of non-toxic PSMα3 derivatives (Cheung 2013), as high concentrations of PSMs are known to be cytotoxic in vitro (Wang et al., 2007).

The benefits of passive administration of anti-PSM antibodies were demonstrated in a study, in which antibodies against the S. epidermidis PSMβ peptides effectively blocked the dissemination of S. epidermidis from bacteria-coated catheters in mice (Wang et al., 2011). However, passive immunization with anti-PSM antibodies may not be generally beneficial in biofilm-related S. aureus and S. epidermidis infections, because – while they may inhibit dissemination - they may also lead to more compact and thicker biofilms (Wang et al., 2011). Accordingly, Agr-dysfunctional, and therefore PSM-negative, S. epidermidis and S. aureus strains are frequently isolated from biofilm-associated infections (Vuong et al., 2004, Traber et al., 2008). Thus, the benefits of passive vaccination with anti-PSM antibodies in the case of biofilm-associated staphylococcal disease need to be thoroughly investigated further. In contrast, passive immunization with antibodies directed against cytolytic PSMs may be more adequate in acute forms of staphylococcal disease. Problems with this approach are the following. First, PSMs are produced at high levels; likely requiring a high concentration of antibodies, at least if the primary mode of function is sequestration rather than promotion of opsonophagocytosis, whose efficacy in the case of S. aureus vaccines is debatable (DeLeo & Otto, 2008). Second, PSMs may exert much of their contributions to pathogenesis in the intracellular environment (Surewaard et al., 2012), which may be difficult to reach for antibodies in sufficient amounts. Third, the variety of PSMs produced by S. aureus would require an antibody cocktail, further complicating such an approach.

Keeping these difficulties in mind, developing drugs or mAbs against Pmt, the dedicated PSM ABC transporter (Chatterjee et al., 2013), has several key advantages over targeting single PSMs or other S. aureus toxins (Chatterjee & Otto, 2013). Inhibiting PSM transport would reduce the overall pathogenicity of S. aureus in vivo because of the interference of the secretion of all PSMs. Furthermore, because the fact that the pmt genes are found genome-encoded among all genome-sequenced staphylococci, it is probable that other pathogenic staphylococci may also be targeted by a potential drug inhibiting Pmt. Additionally, deletion of the pmt genes is harmful to S. aureus viability (Chatterjee et al., 2013) suggesting that blocking Pmt may have a direct bactericidal in addition to a virulence-diminishing effect.

Lastly, while lysis of immune cells very likely is of key importance for the strong impact of α-type PSMs on staphylococcal virulence, the biological role of PSM pro-inflammatory capacity and receptor-mediated recognition of PSMs by the human innate immune system is not understood. If indeed FPR2-dependent activation turns out to be crucial for the pathogenesis of staphylococcal disease, FPR antagonistic molecules, such as CHIPS and FLIPr, may be evaluated as potential drugs against staphylococcal infections.

In conclusion, the PSMs, FPR2-receptor mediated signaling, and especially Pmt are all possible targets for anti-staphylococcal drug development. However, as attractive as these single targets may be, successful protection against S. aureus may depend on disarming multiple targets that are involved in S. aureus virulence (Cheung & Otto, 2012).

Conclusion/open questions

PSMs are a fairly recently discovered group of peptides. Naturally, therefore, many aspects of PSM biology and biochemistry remain to be analyzed. First, our knowledge on psm genes and PSM peptides is restricted to S. aureus, S. epidermidis, and the psmβ-like genes that can be found by similarity searches in other staphylococci. A comprehensive analysis of PSMs of most staphylococcal species is still missing. Second, how the physico-chemical features of PSMs determine their function in surface spreading, emulsification, biofilm formation, and interaction with membranes during cytolysis remains to be more clearly analyzed. Most of this unknown and the only information we have on PSM structure is derived from experiments on S. aureus δ-toxin. One important question to be answered is the role of PSM aggregation, in particular in the reported form of ordered amyloid fibers, during biofilm formation and in-vivo biofilm infection. Third, the structural determinants of FPR2 activation by PSMs are not understood, as is the role of PSM-receptor interaction in pathogenesis, especially in comparison with the non-receptor-mediated cytolytic function of PSMs. Regarding cytolysis, further details of the interaction of PSMs with different cell types, including professional and non-professional phagocytes before and after ingestion, and the role of these interactions during infection and in possible synergy with other toxins need to be elucidated in more detail. Finally, PSMs may be used as vaccine targets, while small molecule inhibitors that target PSM export could induce killing bacteria during infection. However, while this newly recognized family of virulence determinants is clearly a promising therapeutic target, drug development efforts are still in their infancy.