This study was carried out in accordance with the recommendations of approved protocols at the NIH Blood Bank or with a protocol (633/2012BO2) approved by the Institutional Review Board for Human Subjects, NIAID, NIH with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. S. haemolyticus isolates were collected in an anonymous fashion, stored and analyzed according to German law and standards for research use of human biological material.

Bacteria were grown in tryptic soy broth (TSB) at 37°C with shaking. The S. haemolyticus strain ATCC29970 was used in all experiments. Clinical S. haemolyticus isolates were grown from blood cultures obtained from patients hospitalized in the University Medical Center Hamburg-Eppendorf. Species identification was performed by MALDI-TOF mass spectrometry fingerprinting using a MALDI-TOF Biotyper instrument (Bruker Daltonics, Bremen, Germany). All isolates were stored at −80°C for subsequent characterization.

PSM peptides were synthesized by commercial vendors at >95% purity (S. aureus PSMα3, American Peptide Company; S. haemolyticus PSMα, Peptide 2.0; S. haemolyticus PSMβ1, β2, and β3, Atlantic Peptide) with the N-terminal N-formyl methionine modification that is introduced to all bacterial proteins and remains in PSMs due to the lack of a signal peptide.

PSMs were analyzed by a reversed-phase high performance chromatography/electrospray ionization mass spectrometry (RP-HPLC/ESI-MS) method using an Agilent 1260 Infinity chromatography system coupled to a 6120 Quadrupole LC/MS in principle as described (Joo and Otto, 2014), but with a shorter column and a method that was adjusted accordingly. A 2.1 × 5 mm Brownlee SPP (2.7 μm) guard column was used at a flow rate of 0.5 ml/min. After sample injection, the column was washed for 0.5 min with 90% buffer A [0.1% trifluoroacetic acid (TFA) in water]/10% buffer B (0.1% TFA in acetonitrile), then for 3 min with 25% buffer B. Then, an elution gradient was applied from 25 to 100% buffer B in 2.5 min, after which the column was subjected to 2.5 min of 100% buffer B to finalize elution.

To purify S. haemolyticus PSMs, two-step reversed phase chromatography on an AKTA 100 purifier system (GE Healthcare) was used. Forty milliliters of a stationary-phase culture of S. haemolyticus was injected on an HR 16/20 column packed with SOURCE PHE (GE Healthcare) material (~16 ml bed volume). The flow rate was 4 ml/min. After injection, the column was washed with 3 column volumes of buffer A (0.1% TFA in 10% acetonitrile/90% water). Afterwards, bound material was eluted with a linear gradient over 15 column volumes from 100% buffer A to 100% buffer B (0.1% TFA in 90% acetonitrile/10% water). Fractions in the elution range of PSMs were pooled, lyophilized, and re-dissolved in 2 ml water.

Pooled and re-dissolved PSM fractions were injected onto a ZORBAX SB-C18 9.4 × 25 cm (Agilent) column, which was run at 4 ml/min. A linear gradient was applied over 80 min from 100% buffer A to 100% buffer B. Peak fractionation was used and PSM-containing peaks were further analyzed for containing peptides and their purity by RP-HPLC/MS.

PSMα of S. haemolyticus was subjected to N-terminal Edman sequencing by the NIAID Peptide Synthesis and Analysis Laboratory after removal of the N-terminal N-formyl modification by boiling in 25% TFA for 2 h at 55°C (Shively et al., 1982).

The structure of synthetic PSM peptides was analyzed by CD spectroscopy on a Jasco spectropolarimeter model J-720. Solutions of PSM peptides, each at 1.0 mg/ml, were prepared in 50% trifluoroethanol. Measurements were performed in triplicate. Resulting scans were averaged, smoothed, and the buffer signal was subtracted.

Hemolysis (lysis of erythrocytes) was measured by incubating samples with human erythrocytes (2% in Dulbecco's phosphate-buffered saline, DPBS) for 1 h at 37°C as previously described (Wang et al., 2007). Hemolysis was determined by measuring OD_540nm using an ELISA reader.

Lysis of human neutrophils by PSMs was determined essentially as described (Voyich et al., 2005). PSMs were incubated in 96-well tissue culture plates with 10⁶ neutrophils and plates were incubated at 37°C for 1 h. Then, neutrophil lysis was determined using a Cytotoxicity Detection Kit (Roche Applied Sciences) by release of lactate dehydrogenase (LDH).

Neutrophils were subjected to a brief hypotonic shock with pyrogen-free water, washed, and suspended at 5 × 10⁶ cells/ml in Hank's Balanced Salt Solution containing 0.05% human serum albumin. Chemotaxis of neutrophils was determined by using fluorescently-labeled neutrophils that migrated through a membrane fitted into an insert of a 24-well microtiter plate transwell system (Costar) containing a prewetted 3-μm-pore-size polycarbonate filter as described (de Haas et al., 2004).

Statistical analysis was performed using Graph Pad Prism version 6.02. All error bars depict the standard deviation.

As a preliminary test for the presence of PSMs in S. haemolyticus, we prepared stationary-phase (8 h) culture filtrate and subjected it to the rapid RP-HPLC/MS analysis of PSMs we routinely use in our laboratory. Stationary-phase cultures were used because PSM production has been shown to be strictly regulated by the accessory gene regulator (Agr) quorum-sensing system in S. aureus and S. epidermidis (Vuong et al., 2004a; Queck et al., 2008); and based on genome sequence analysis showing presence of Agr it is fair to assume that this is also the case in S. haemolyticus (Takeuchi et al., 2005). Analysis of culture filtrate showed that S. haemolyticus produces peaks in the elution range of PSMs with masses suggesting the presence of several PSMs (Figure 1). Notably, this analysis confirmed that S. haemolyticus in contrast to S. aureus and S. epidermidis does not produce a δ-toxin, as genome analysis had suggested (Takeuchi et al., 2005).

To further characterize those putative S. haemolyticus PSMs, a larger-scale purification was performed. Culture filtrate was subjected to initial RP chromatography on SOURCE PHE material and the fractions in the range where PSMs elute were lyophilized, resuspended, and subjected to high-resolution RP-HPLC on C18 material (see Section Methods for details; Figure 2). With this procedure, the four PSMs could be separated and isolated in sufficient amounts for clear mass spectrometric analysis using RP-HPLC/MS of the single fractions, as well as N-terminal sequencing.

All PSMs carry an N-terminal N-formyl modification at the N-terminal methionine, as present in all primary bacterial translation products, due to the lack of a signal peptide (Cheung et al., 2014a). In S. aureus, peptide N-deformylase removes this modification to a substantial degree, resulting in considerable amounts of N-deformylated PSMs in the culture filtrate (Cheung et al., 2010). In S. haemolyticus, similar to S. epidermidis (Czekaj et al., 2015), N-deformylation only occurred to a very minor degree, as evident from only small peaks with masses corresponding to N-deformylated PSMs.

Masses obtained by HPLC/MS and in case of the α-type peptide, the entire sequence obtained by Edman sequencing, together with genome analysis, allowed clear identification of the amino acid sequences of the four PSMs of S. haemolyticus and the genetic loci in which they are encoded (Figures 3A,B). The three PSMβ peptides of S. haemolyticus (PSMβ1, PSMβ2, PSMβ3), which correspond in amino acid sequence to the previously described gonococcal growth inhibitor peptides, are encoded in an apparent operon in front of gene SH_RS08485. The genes encoding PSMβ2 and PSMβ3 are duplicated, a situation also known for some other PSM-encoding loci, notably the S. epidermidis psmβ locus whose encoded PSMβ peptides share some similarity with the S. haemolyticus PSMβ peptides (Cheung et al., 2014a).

While the genes encoding the longer PSMβ peptides can often be found by computer similarity searches such as BLAST, on the peptide or DNA level, this is virtually impossible for the short α-type PSMs. Identification of the genetic locus encoding PSMα of S. haemolyticus thus depended on the amino acid sequence we obtained from Edman sequencing. This allowed attributing production of S. haemolyticus PSMα to a gene in the vicinity of the ndhF gene, a location, which also encodes the PSMα peptides of S. aureus and two α-type PSMs of S. epidermidis (Cheung et al., 2014a), suggesting a common origin (Figure 3B). It is noteworthy that an alignment with similarity analysis and phylogenetic analysis of the S. aureus, S. epidermidis, and S. haemolyticus PSMs showed that PSMs are sometimes very similar to other PSMs in the same locus, but sometimes more to PSMs in the corresponding locus in another species (Figures 3C,D). Thus, it appears that gene duplication events happened in evolution after, as well as before, the transfer of the psmα and psmβ loci between species.

Next, we computed α-helical wheels and measured α-helicity by circular dichroism (CD) analysis, to confirm that the S. haemolyticus PSMs also fulfill the structural requirements, i.e., formation of amphipathic α-helices, to be called PSMs. Indeed, all four S. haemolyticus PSMs are α-helical according to CD, as they showed the maximum at 193 nm and the two minima at 208 and 222 nm that are characteristic for α-helices (Figure 4A). Furthermore, when the sequences were arranged in α-helical wheels by bioinformatics analysis, they all showed opposite arrangement of hydrophilic and hydrophobic amino acids, characteristic of amphipathic α-helical peptides (Figure 4B).

With the exception of the difference that is due to the presence of the mobile genetic element-encoded PSM-mec, based on evidence obtained with S. aureus and S. epidermidis, the PSM pattern of a given staphylococcal species is conserved among different isolates and strains (Wang et al., 2007; Queck et al., 2009). Therefore, we analyzed whether the PSM pattern is also conserved in different S. haemolyticus strains. To that end, we analyzed 10 S. haemolyticus bacteremia blood culture isolates obtained from the Eppendorf University Hospital in Hamburg, Germany, in addition to the ATCC29970 strain used for analysis and purification. Eight of the ten strains showed the same PSM pattern in terms of presence of the four PSM peptides, with only slight differences in production (Figure 5). In one of the eight strains, no PSMβ1 production was observed, with the remaining pattern being conserved, suggesting a possible point mutation or deletion abolishing production of that peptide. Two strains were devoid of PSM production, a situation with similar frequency also seen in S. aureus and S. epidermidis (Vuong et al., 2004b; Traber et al., 2008), originating from spontaneous mutations in the Agr system, which strictly controls PSM production (Queck et al., 2008). Notably, when tested on human or sheep blood, the PSM-deficient strains lacked hemolytic capacities, while PSM producers were all hemolytic, indicating that PSM production underlies hemolytic activity in S. haemolyticus. Interestingly, hemolytic activity was in general more pronounced toward human than sheep blood in this assay.

Hemolytic activity is what gave S. haemolyticus its name; however, hemolytic agents were never clearly identified in S. haemolyticus, although it was suspected that they may be peptides (Loyer et al., 1990). To investigate whether the identified PSMs of S. haemolyticus have hemolytic activity that could explain the hemolytic properties of that species, we measured lysis of human erythrocytes. All S. haemolyticus PSMs showed hemolytic capacity, with that of the PSMα peptide in particular reaching levels known for the most hemolytic peptides in other staphylococci, PSMα3 of S. aureus and PSMδ of S. epidermidis (Wang et al., 2007; Cheung et al., 2010), as demonstrated by direct comparison with S. aureus PSMα3 (Figure 6). This peptide is thus a good candidate to account for the hemolytic capacity of S. haemolyticus; however, also the PSMβ peptides may have a significant contribution to the hemolytic capacity of S. haemolyticus given that they are produced at high levels (see Figure 2). Hemolytic capacity was similar in this assay toward human vs. sheep blood.

Phagocytosis and elimination by neutrophils is the preeminent way host defense deals with invading staphylococci (Rigby and DeLeo, 2012). The psmα locus of S. aureus has been shown to be crucial for the elimination of neutrophils and thus to represent a key part of the immune evasion capacity of S. aureus (Wang et al., 2007). While S. epidermidis PSMs other than PSM-mec have not yet been investigated for their contribution to that phenotype using isogenic deletion mutants, cytolytic activity toward human neutrophils is found in particular for S. epidermidis PSMδ (Cheung et al., 2010). In contrast to α-type PSMs, β-type PSMs are commonly found to have only very low cytolytic activity toward leukocytes (Wang et al., 2007; Cheung et al., 2010). Thus, we measured lysis of human neutrophils by release of lactate dehydrogenase for all S. haemolyticus PSMs. As expected, we detected only very low cytolytic activity toward human neutrophils by S. haemolyticus PSMβ peptides. However, the cytolytic activity of S. haemolyticus PSMα was very high, reaching activity levels exerted by the strongly lytic S. aureus PSMα3 (Figure 7).

All PSMs described to date bind to FPR2 and activate neutrophils (Kretschmer et al., 2010). This is likely to be interpreted as a way of the human immune system to recognize staphylococcal invaders. Arguably the most crucial phenotype that is triggered by that interaction is neutrophil chemotaxis, i.e., the attraction of neutrophils to the site of staphylococcal infection. Although, PSM-FPR2 interaction is poorly understood, different structural features than those that promote receptor-independent cytolysis likely determine the receptor-dependent pro-inflammatory activities (Cheung et al., 2014b). Thus, it is not uncommon to find similar pro-inflammatory capacities in PSMs that differ very much in their cytolytic capacities. Indeed, when we measured chemotaxis as an important inflammatory phenotype that is triggered by PSMs in human neutrophils, we found that all S. haemolyticus PSMs were similarly pro-inflammatory (Figure 8).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.