Abstract

Host defense against Staphylococcus aureus greatly depends on bacterial clearance by phagocytic cells. LukGH (or LukAB) is the most potent staphylococcal leukocidin towards human phagocytes in vitro, but its role in pathogenesis is obscured by the lack of suitable small animal models because LukGH has limited or no cytotoxicity towards rodent and rabbit compared with human polymorphonuclear cells (PMNs) likely due to an impaired interaction with its cellular receptor, CD11b. We aimed at adapting LukGH for the rabbit host by improving binding to the rabbit homolog of CD11b, specifically its I-domain (CD11b-I). Targeted amino acid substitutions were introduced into the LukH polypeptide to map its receptor interaction site(s). We found that the binding affinity of LukGH variants to the human and rabbit CD11b-I correlated well with their PMN cytotoxicity. Importantly, we identified LukGH variants with significantly improved cytotoxicity towards rabbit PMNs, when expressed recombinantly (10–15-fold) or by engineered S. aureus strains. These findings support the development of small animal models of S. aureus infection with the potential for demonstrating the importance of LukGH in pathogenesis.

Introduction

Staphylococcus aureus is a major human pathogen, both in the hospital setting and in the community, and it is responsible for a wide range of diseases, ranging from mild skin infections to life-threatening infections, such as pneumonia and sepsis. The bacterium can survive and multiply in various biological niches owing to an arsenal of virulence factors that mediate tissue adhesion, immune evasion, and host cell injury [1]. A central piece for the latter two mechanisms is the pore-forming cytotoxins that lyse target cells and induce inflammation: the leukocidins, which attack human phagocytic cells, some of them red blood cells and T cells as well, and α-hemolysin (Hla or α-toxin) that mainly damages epithelial and endothelial cells [25]. S. aureus produces up to five different leukocidins: γ-hemolysins HlgAB and HlgCB, LukSF-PV (PVL), LukED and LukGH (also called LukAB) [2,5,6]. The contribution of each leukocidin to the different disease types has only started to be unraveled (e.g. LukSF-PV in necrotizing pneumonia [7]), mainly due to lack of suitable animal species as model organisms [6]. The recent identification of the cellular receptors for each of the leukocidins was crucial in understanding their cell-type specificity and host tropism ([4, 8]; Table 1).

Table 1
Cell-tropism and species specificity of the bi-component cytotoxins of Staphylococcus aureus
Toxin Cellular receptors Targeted cells EC50 (nM) 
Human Rabbit Mouse 
LukSF-PV C5aR [9,10], C5L2 [9], CD45 [11Monocytes, neutrophils, macrophages 0.9 [9], 0.1–0.5 [12], ∼0.1 [13], ∼1 [14], ∼1 [150.1–0.2 [12], ∼1 [14Resistant [9,12,14,16
LukED CCR5 [17], CXCR1 [18], CXCR2 [18Monocytes, neutrophils, macrophages, T cells, dendritic cells, NK cells 2.4–4 [12], ∼1 [150.03–0.04 [1214–48.3 [12
DARC [19Red blood cells ∼20 [19], <15 [20 <15 [20
LukGH CD11b-I [21Neutrophils, macrophages, monocytes, dendritic cells 0.01–0.03* [13], 0.25 [2235 [22550 [22
HlgAB CXCR1 [23], CXCR2 [23], CCR2 [23Monocytes, macrophages, T cells, neutrophils 0.4–0.6 [120.2 [12Resistant [12, 23
DARC [19Red blood cells ∼1 [19  
HlgCB C5aR [23], C5L2 [23Monocytes, macrophages, neutrophils 0.5–0.6 [12], ∼2.5 [160.1 [12Resistant [12,16,23
Toxin Cellular receptors Targeted cells EC50 (nM) 
Human Rabbit Mouse 
LukSF-PV C5aR [9,10], C5L2 [9], CD45 [11Monocytes, neutrophils, macrophages 0.9 [9], 0.1–0.5 [12], ∼0.1 [13], ∼1 [14], ∼1 [150.1–0.2 [12], ∼1 [14Resistant [9,12,14,16
LukED CCR5 [17], CXCR1 [18], CXCR2 [18Monocytes, neutrophils, macrophages, T cells, dendritic cells, NK cells 2.4–4 [12], ∼1 [150.03–0.04 [1214–48.3 [12
DARC [19Red blood cells ∼20 [19], <15 [20 <15 [20
LukGH CD11b-I [21Neutrophils, macrophages, monocytes, dendritic cells 0.01–0.03* [13], 0.25 [2235 [22550 [22
HlgAB CXCR1 [23], CXCR2 [23], CCR2 [23Monocytes, macrophages, T cells, neutrophils 0.4–0.6 [120.2 [12Resistant [12, 23
DARC [19Red blood cells ∼1 [19  
HlgCB C5aR [23], C5L2 [23Monocytes, macrophages, neutrophils 0.5–0.6 [12], ∼2.5 [160.1 [12Resistant [12,16,23
*

LPS-stimulated cells.

The lack of a suitable animal model for infectious diseases caused by human-adapted pathogens is an inherent challenge for pathogenesis studies and drug discovery. This is typically addressed by mouse humanization, either by knocking out/in the host factors involved in human species specificity [24] or by using immunocompromised mice reconstituted with a human hematopoietic system [25]. Another approach is to engineer the pathogen to express virulence factors that are active towards the chosen animal host. The best-known example is the adaptation of Listeria monocytogenes to mouse by improving binding of the listerial invasion protein InlA to the murine variant of its cognate receptor E-cadherin by protein engineering [26].

LukGH is the most potent leukocidin in in vitro assays and ex vivo models [13,2729], and is present in nearly all S. aureus strains [27,30]. It is expressed during human infections [31], but is inactive or displays limited activity in the established S. aureus models, such as mouse and rabbit [21,22] (Table 1). LukGH is also unique among the leukocidins as it is present as a dimer in solution, before it binds to the target cells [32,33], for reasons and with implications not fully understood. It is therefore conceivable that LukGH followed a distinct evolutionary path, as also suggested by the lowest sequence homology (only up to 40%) to the other leukocidins, which are otherwise up to 82% homologous to each other [34]. Moreover, the sequence conservation of LukGH among different S. aureus isolates is much lower (as low as 82%) than for the other leukocidins (>95% identity) with the sequence variants showing association with clonal lineages [27,30]. The cellular receptor for LukGH was identified as the α subunit of the αM2 integrin (CD11b/CD18, or macrophage-1 antigen, or complement receptor 3) [21]. It has been demonstrated that binding to the I-domain of CD11b correlates with cytotoxicity towards neutrophils, i.e. no binding to mouse CD11b-I [21] and very low activity towards murine PMNs in vitro [22].

In this study, we aimed at adapting LukGH to the rabbit host, which is sensitive to all the other β-barrel pore-forming cytotoxins [12] (Table 1), to allow the study of LukGH contribution to S. aureus pathogenesis. Our approach was to engineer LukGH for increased binding affinity to rabbit CD11b-I (rbCD11b-I) and improved activity towards rabbit PMNs (rbPMNs), which is two orders of magnitude weaker than towards human PMNs (huPMNs) [22] (Table 1). In addition, we performed a fine epitope mapping of the CD11b-I-binding site on LukGH to enable rational LukGH adaptation to other species.

Materials and methods

Production of recombinant LukGH variants

LukGH variants were produced recombinantly in Escherichia coli, as described previously [33], based on the wild-type sequence of the community-associated methicillin-resistant Staphylococcus aureus (CA-MRSA) USA300 (ST8) TCH1516 strain. The three natural sequence variants encoded by the MRSA252 (ST36), MSHR1132 (ST1850), and H19 (ST10) strains were produced as described previously [33]. The lukG gene was cloned into pET44a vector and was expressed as a fusion protein with NusA/His6 at the N-terminus to allow metal ion affinity purification of the complex, whereas LukH was expressed in the untagged form. LukH single and multiple mutants were generated with QuikChange II XL Site-Directed Mutagenesis Kit (Agilent) according to the manufacturer's instructions using the lukH_pET200D/TOPO (or variants thereof) as a template. Mutations to D, V, and S or to H at amino acid positions 263 and 312, respectively, were obtained by site-specific saturation mutagenesis using a QuikChange Multi Site-Directed Mutagenesis kit (Agilent) and lukH-PET200D/TOPO as a template. The lukH mutants and wild-type lukG were co-transformed into E. coli TUNER DE3 cells (Novagen), using two vectors with different antibiotic resistance markers, and protein expression was induced with isopropyl β-d-1-thiogalactopyranoside (IPTG) as described previously [33]. All LukGH variant complexes were expressed as soluble proteins, at 20°C, similarly to wild-type LukGH [33].

Recombinant LukGH proteins were purified as described previously [33] or using batch methods designed for high-throughput purification from 0.5 or 0.125 l cultures. Briefly, bacterial pellets were disrupted using either sonication (0.5 l pellet) or 0.1% n-dodecyl-β-d-maltoside and two freeze–thaw cycles in liquid nitrogen (0.125 l pellet). The LukGH dimers were purified from cell extracts by metal ion affinity and cation exchange chromatography using batch methods. First, soluble cell extracts (obtained by centrifugation of cell lysates) were mixed with Ni Sepharose® 6 Fast Flow beads (GE Healthcare) equilibrated in 20 mM Tris–HCl, pH 7.5 plus 50 mM imidazole. Beads were loaded either on Pierce Disposable Columns (Thermo Fisher) or on Microplate Devices UNIFILTER, 96-well plate (Whatman) and elution with 20 mM Tris–HCl, pH 7.5 plus 500 mM imidazole was performed by centrifugation. After dialyzing into buffer without imidazole, the NusA/His6 tag on LukG was removed with enterokinase (NEB). The untagged LukGH complex was further purified using SP Sepharose® Fast Flow beads (GE Healthcare) equilibrated in 20 mM sodium phosphate, pH 7.5 plus 50 mM NaCl. Beads were treated as described for the affinity purification, and the proteins were eluted with 20 mM sodium phosphate, pH 7.5 plus NaCl (150–300 mM). Protein purity was assayed by SDS–PAGE gels, stability by differential scanning fluorimetry (DSF) and secondary structure by circular dichroism (CD).

CD and DSF analysis

Far-UV (195–250 nm) CD spectra were recorded on a Chirascan (Applied Photophysics) spectrometer in a 0.5 mm cuvette (Applied Photophysics) at 20°C with protein concentrations of 0.1–0.8 mg/ml in 20 mM sodium phosphate, pH 7.5 plus 150–300 mM NaCl. All CD spectra were normalized to a concentration of 0.3 mg/ml.

The melting points (Tm) of the proteins were determined by DSF. The proteins (0.25–0.6 mg/ml) were mixed with Sypro Orange dye and with HEPES, pH 7.5 (50 mM final concentration). The assay was conducted in a qPCR instrument (Bio-Rad CFX96) and the Tm values were determined using the Bio-Rad CFX Manager software.

Recombinant human CD11b-I (huCD11b-I) and rabbit CD11b-I (rbCD11b-I) expression and purification

The I-domains (amino acids 127–321) of huCD11b and rbCD11b (huCD11b-I and rbCD11b-I, respectively) were cloned into the pET24a (Novagen) vector at NdeI/XhoI and NdeI/BamHI sites, respectively, and the plasmids were transformed into E. coli TUNER DE3 cells. Protein expression was induced at 20°C for 20 h with 0.4 mM IPTG. HuCD11b-I and rbCD11b-I were purified by cation (HiTrap® SP FF, GE Healthcare) or anion (HiTrap® Q FF, GE Healthcare) exchange chromatography followed by size exclusion chromatography (HiLoad Superdex75 pg, GE Healthcare). HuCD11b-I was treated overnight with iodoacetamide (20 mM, Applichem) to alkylate-free cysteine and prevent dimer formation. Iodoacetamide was removed on PD-10 columns (GE Healthcare) equilibrated with 50 mM sodium phosphate, pH 7.5 plus 300 mM NaCl. Protein purity and monomer content were assessed by non-reducing SDS–PAGE gel.

Biotinylated huCD11b-I and rbCD11b-I were generated with the amino reactive reagent Sulfo-NHS-LC biotin (Thermo Scientific), according to the manufacturer's instructions with final biotin/protein ratios of 0.15–0.2.

Bio-layer interferometry

Binding of LukGH (wild-type and mutants) to huCD11-I or rbCD11b-I was evaluated by Bio-Layer Interferometry (BLI) (fortèBio Octet Red96 instrument, Pall Life Sciences) in assay buffer (PBS plus 1% BSA and 1 mM MgCl2). Biotinylated CD11b-I (2 μg/ml) was immobilized on streptavidin sensors (fortèBio, Pall Life Sciences) to achieve a final loading of 0.6–1.6 nm. Association of LukGH (50 nM) to the immobilized receptor and dissociation in assay buffer were monitored for 5 min each. Response units (RU) (normalized to the same loading) and, where possible (for monophasic binding curve), equilibrium dissociation constants (Kd) were determined using the Data Analysis 7 software (fortéBio, Pall Life Sciences) by simultaneously fitting the association and dissociation curves to a 1:1 binding model. Steady-state Kd values were determined for LukGH wild-type and LukGH_D312A binding to rbCD11b-I by measuring binding responses at multiple LukGH concentrations (20–400 nM) and fitting the data to the steady-state model (Forte-Bio Analysis Software, Version 7).

Cytotoxicity assays

Cell-based assays were performed using differentiated HL-60 cells, human or rabbit polymorphonuclear cells (PMNs). The HL-60 cells (ATCC® CCL-240TM) were differentiated into phagocytes by treatment with 100 mM dimethylformamide in the culture media (RPMI + 20% FCS + 2 mM l-glutamine + Pen/Strep) for 3–5 days as described previously [35]. The differentiation status of HL-60 cells was confirmed by a significant reduction in CD71 and increase in CD11b expression based on staining intensity with phycoerythrin-conjugated anti-CD71 (clone OKT9, eBioscience) and Brilliant Violet 421-conjugated anti-CD11b (clone ICRF44, BioLegend) monoclonal antibodies. Human PMNs were isolated from heparinized human whole blood, obtained from healthy volunteers, using Percoll® (Percoll Plus, GE Healthcare) gradient centrifugation as described previously [33]. Rabbit PMNs were isolated from rabbit whole blood (pooled blood of three New Zealand White rabbits) anti-coagulated with citrate dextrose solution, using Histopaque®-1077 (Sigma–Aldrich) and HetaSep™ (Stemcell Technologies) as described previously [36]. For the assays with stimulated rbPMNs, purified PMNs (1 × 106 cells/ml) were pre-incubated for 1 h at 37°C, 5% CO2 with lipopolysaccharide (LPS) purified from E. coli O111 (500 ng/ml, List Laboratories).

  • (i) Cytolytic activity of LukGH (wild-type and mutants) was assessed as described previously [13,34]. Shortly, differentiated HL-60 cells, huPMNs, non-stimulated or LPS-stimulated rbPMNs (2.5 × 104 cells/well) were treated with serial dilutions of LukGH (0.002–100 nM for human and 0.005–300 nM for rabbit cells) in RPMI + 10% FBS + l-glutamine assay medium at 37°C, 5% CO2 for 4 h. Cell viability was determined using CellTiter-Glo® Luminescent Cell Viability Assay Kit (Promega) according to the manufacturer's instructions. The percentage viability was calculated relative to mock-treated cells (100% viability) and cytolytic activity was expressed as EC50 value (concentration of cytotoxin at which 50% of cells are killed) calculated by nonlinear regression analysis using Prism 6 (GraphPad). Each independent experiment was performed in triplicates on different batches of PMNs or HL-60 cells.

  • (ii) Cell permeability, in presence of LukGH, was assessed using rbPMNs (1 × 105 cells/well) treated with serial dilutions of recombinant toxin (0.005–300 nM) in RPMI + 10% FCS + l-glutamine + 10 mM HEPES assay medium at 37°C, 5% CO2 for 2 h. After incubation, plates were centrifuged at 1500 g for 5 min, supernatant was discarded, and pellets were re-suspended in PBS + 0.5 µM SYTOX® Green Nucleic Acid Stain (Molecular Probes). Following 10 min of incubation, fluorescence (λex = 485 nm, λem = 528 nm) was quantified on a SynergyTMHT Multi-Mode Microplate Reader (BioTek). The percentage of dead cells was calculated relative to a dead cell control (0.1% saponin) and EC50 values were calculated as described above. Each independent experiment was performed in triplicates on different batches of PMNs.

  • (iii) The cytotoxic activity of culture supernatants (CSs) was determined using huPMNs and rbPMNs. Serial dilutions of CSs from 4- to 512-fold, in RPMI plus 10% FCS and 2 mM l-glutamine assay medium, were prepared in a 96-well plate and pre-incubated for 30 min at room temperature with the monoclonal antibodies ASN-1 (1 μM), ASN-2 (1 μM), ASN-1 (1 μM) + ASN-2 (1 μM), an isotype control antibody (2 μM) or buffer. ASN-1, a human mAb that cross-neutralizes Hla and the other four leukocidins (LukED, LukSF-PV, HlgAB, and HlgCB) [34] and ASN-2, a LukGH-specific neutralizing antibody [37] were produced as described recently [29]. The supernatant–antibody mixtures were added to huPMNs or rbPMNs in 96-well, half-area luminescent plates (Greiner) at 2.5 × 104 cell density. Following incubation at 37°C, 5% CO2 for 4 h, toxicity of CSs was assessed by measuring cellular ATP levels with CellTiter-Glo® Luminescent Cell Viability Assay Kit (Promega) and the percentage viability was calculated relative to mock-treated cells (100% viability). Each experiment was performed using 4–6 CSs prepared on different days and from different colonies.

  • (iv) To determine S. aureus-mediated killing of PMNs by extracellular bacteria, overnight cultures of S. aureus grown in RPMI-CAS were diluted 1:100 and grown to mid-log phase (OD600 nm = 0.5) at 37°C. Bacteria were harvested, washed with PBS to remove secreted toxins, re-suspended in RPMI + 10% FCS + l-glutamine + 10 mM HEPES, and added to 1 × 105 PMNs/well in a 96-well plate at different multiplicity of infection (MOI, 50 and 100) together with ASN-1 (2 μM), ASN-2 (2 μM), ASN-1 (2 μM) + ASN-2 (2 μM), an isotype control antibody (4 μM) or buffer. Reactions were incubated for 2 h at 37°C and 5% CO2. Fluorescence was measured using SYTOX® Green Nucleic Acid Stain (Molecular Probes) as described above. The percentage of dead cells was calculated relative to a dead cell control (0.1% saponin). Each independent experiment was performed in 2–6 replicates on different batches of PMNs.

Chromosomal integration of mutant lukGH

The USA300 CA-MRSA strain TCH1516 (ST8-IV-t622, BAA-1717TM, ATCC®) and a clinical isolate, recovered from an endotracheal aspirate of a mechanically ventilated patient, with high LukGH expression level, LA#5 (ST8-IV-t008) [29,38], were cultured under standard microbiological conditions. lukH_D312K and lukH_E263Q-D312N were introduced by site-directed mutagenesis into the lukGH operon, which was subsequently cloned into the pKFT shuttle-vector at SmaI and BamHI restriction sites [39] (Supplementary Figure S1A). Gene replacement was achieved by homologous recombination according to previously published methods [39,40]. Shortly, the vector-construct was transformed into wild-type S. aureus by electroporation and transformants were grown at 30°C/200 rpm selecting for TetR. After a first temperature-based (42°C), homologous recombination event was induced, the plasmid was cured by repeated passaging at 25°C/200 rpm. The second cross-over event was induced by another temperature shift to 42°C that resulted in the excision of the integrated plasmid (TetS) and either the homologous insertion of the mutated lukGH or a reversion to wild type, which was confirmed by sequencing of the lukGH operon (Supplementary Figure S1B).

Strains were characterized by determining the growth curves and expression of the cytotoxins as follows: overnight cultures of the S. aureus strains TCH1516, LA#5 and rabbit-adapted strains, grown in RPMI + 1% Casamino acids (RPMI + 1% CAS) were diluted to an OD600 nm of 0.03 and grown to a stationary phase at 37°C/200 rpm. OD600 was measured at regular intervals for 8 h and plotted against growth time. Cultures were centrifuged at 5000 g for 10 min and the supernatants were sterile-filtered using 0.1 μm filters (Millex Syringe Filter Units, Millipore). These CSs were further used for Western blot and PMN toxicity assays. Western blot of CSs from S. aureus TCH1516 and LA#5 wild-type, rabbit-adapted strains [lukH_D312K and lukH_E263Q-D312N (lukH_QN)], and control strain (TCH1516 Δ all cytotoxins [34]) grown in RPMI + 1% CAS, to comparable OD600, was performed using same loading amount of CS and corresponding antibodies. The expressions of Hla, LukD, and HlgB were assessed using monospecific human antibodies, LukS-PV was detected with a mouse monoclonal antibody (IBT Bioservices) and LukG expression was measured using a rabbit anti-LukB polyclonal Ab (IBT Bioservices) as described recently [29].

Results

LukGH binding to human and rabbit CD11b-I

We first compared the binding strength of recombinant LukGH (derived from the genome sequence of the USA300 CA-MRSA, TCH1516 strain) to the human (huCD11b-I) and rabbit (rbCD11b-I) receptors, Kd, when LukGH was in solution and the recombinant receptor in immobilized form, using BLI. Binding of LukGH to huCD11b-I appeared to follow a 1:1 binding model, so the equilibrium dissociation constant, Kd, could be calculated from the association and dissociation progress curves, yielding a value of 7.96 × 10−9 M with an association rate constant, kon of 1.0 × 105 1/Ms (Figure 1A). For rbCD11b-I, the kinetic profile was biphasic, so we were unable to determine the association and dissociation rate constants, but binding was sufficiently weak to allow determining Kd using steady-state analysis (Kd = 9.51 × 10−8 M, Figure 1B). In these experiments, we observed an ∼10-fold lower affinity of LukGH for rbCD11b-I compared with the human counterpart (Figure 1A,B). This difference was paralleled by the lower activity towards rbPMNs compared with huPMNs, typically by two orders of magnitude, which is in good agreement with published data [22] (Figure 1C).

Binding of LukGH to hu and rbCD11b-I measured by BLI, and activity towards human and rabbit PMNs.

Figure 1.
Binding of LukGH to hu and rbCD11b-I measured by BLI, and activity towards human and rabbit PMNs.

(A) Representative kinetic profile for LukGH binding to huCD11b-I. Association and dissociation steps are separated by a vertical red line. The binding curve (blue line) was fitted to a 1:1 model and the fit is shown in red. The Kd, kon, and koff are shown in the insert (mean of 10 independent experiments ± SD). (B) Steady-state analysis of LukGH binding to rbCD11b-I measured at different LukGH concentrations. The steady-state Kd is shown in the insert (mean of two independent experiments ± SD). (C) Activity of LukGH towards huPMNs and rbPMNs assessed in a luminescent cell viability assay measuring cellular ATP content at increasing cytotoxin concentrations (mean of two independent experiments performed in triplicates ± SEM.).

Figure 1.
Binding of LukGH to hu and rbCD11b-I measured by BLI, and activity towards human and rabbit PMNs.

(A) Representative kinetic profile for LukGH binding to huCD11b-I. Association and dissociation steps are separated by a vertical red line. The binding curve (blue line) was fitted to a 1:1 model and the fit is shown in red. The Kd, kon, and koff are shown in the insert (mean of 10 independent experiments ± SD). (B) Steady-state analysis of LukGH binding to rbCD11b-I measured at different LukGH concentrations. The steady-state Kd is shown in the insert (mean of two independent experiments ± SD). (C) Activity of LukGH towards huPMNs and rbPMNs assessed in a luminescent cell viability assay measuring cellular ATP content at increasing cytotoxin concentrations (mean of two independent experiments performed in triplicates ± SEM.).

We also tested the binding to rbCD11b-I and activity towards rbPMNs of three additional, most divergent, sequence variants of LukGH TCH1516 from S. aureus strains: MRSA252, MSHR1132, and H19 [33]. We found that the activity of these sequence variants was an order of magnitude lower (Supplementary Table S1); therefore, the TCH1516 variant was used in the subsequent rabbit adaptation studies.

Purification and characterization of LukGH variants

It was shown previously that CD11b-I interacts mainly with the LukH subunit of LukGH [21,33], and particularly with the C-terminal tail of LukH [32]. Importantly, the glutamate at position 323 (E323) in LukH was identified as being crucial for binding to huCD11b-I and activity towards huPMNs [32]. To determine the binding epitope of CD11b-I on LukGH, we investigated the involvement of residues surrounding E323 (based on the LukGH dimer [PDB 5K59] and octamer [PDB 4TW1] structures) in receptor-binding and cytotoxin activity. A total of 21 positions, mostly surface exposed, polar, and charged residues located in the cap or the rim domain of LukH, were subjected to alanine mutagenesis (Figure 2A). While most of the mutated side chains do not interact with any other residues based on the LukGH octamer structure (e.g. R121, K309), some do (e.g. D177, R207, D209, and Y314) (Figure 2A), which is presumably the reason for the decreased stability of certain Ala mutants (vide infra).

Amino acid positions on LukH selected for Ala screening.

Figure 2.
Amino acid positions on LukH selected for Ala screening.

(A) Left panel: Structure of the LukGH octamer (PDB 4TW1) with positions selected for Ala screening shown as spheres. LukG and LukH are shown as green and yellow cartoon, respectively, LukG from the neighboring dimer as purple and remaining monomers as gray cartoon. Right panel: Positions selected for Ala screening are shown as green sticks colored according to the atom (red: oxygen, blue: nitrogen). Polar contacts involving the side chains from selected positions are marked with dashed lines. Residues on LukH interacting with the positions selected for Ala screening are shown as yellow sticks and colored by atom. LukG residues involved in polar contacts are shown as purple sticks and colored by atom. (B) CD spectra of selected Ala mutants in 20 mM sodium phosphate, pH 7.5 plus NaCl (150–300 mM), normalized to 0.3 mg/ml. (C) Melting temperature of Ala mutants measured by DSF at concentrations between 0.25 and 0.6 mg/ml in 50 mM HEPES, pH 7.5. The mutants with at least 2°C difference in Tm, compared with wild-type LukGH (45°C) are marked red.

Figure 2.
Amino acid positions on LukH selected for Ala screening.

(A) Left panel: Structure of the LukGH octamer (PDB 4TW1) with positions selected for Ala screening shown as spheres. LukG and LukH are shown as green and yellow cartoon, respectively, LukG from the neighboring dimer as purple and remaining monomers as gray cartoon. Right panel: Positions selected for Ala screening are shown as green sticks colored according to the atom (red: oxygen, blue: nitrogen). Polar contacts involving the side chains from selected positions are marked with dashed lines. Residues on LukH interacting with the positions selected for Ala screening are shown as yellow sticks and colored by atom. LukG residues involved in polar contacts are shown as purple sticks and colored by atom. (B) CD spectra of selected Ala mutants in 20 mM sodium phosphate, pH 7.5 plus NaCl (150–300 mM), normalized to 0.3 mg/ml. (C) Melting temperature of Ala mutants measured by DSF at concentrations between 0.25 and 0.6 mg/ml in 50 mM HEPES, pH 7.5. The mutants with at least 2°C difference in Tm, compared with wild-type LukGH (45°C) are marked red.

The LukGH complexes were purified (>95% purity) with yields varying from 0.8 to 8 mg/l culture. To exclude the possibility that protein instability and improper folding affect the binding or activity, we determined the Tm and secondary structure (far-UV CD spectra) of the variants (Figure 2B,C; Supplementary Figure S2). All variants, except the LukGH_R207A and _D209A, had Tm values ranging from 44 to 45°C, which are similar to the wild-type protein (Figure 2C). The most striking change in Tm, 10°C lower compared with the wild-type, was seen with LukGH_R207A, seemingly due to loss of interactions between A207 and the surrounding LukH residues (S109, E110, and D271) (Figure 2A). The LukGH_D209A displayed a moderately lower Tm (2.5°C). Both positions were therefore excluded from further analyses. Since no significant changes in Tm or CD spectra were observed for the other variants, we concluded that any change detected in binding to CD11b-I and in cytotoxin activity towards rabbit and human cells was not due to misfolding or decreased stability of the proteins.

Correlation between receptor-binding and cytotoxicity of LukGH Ala variants

All alanine variants of LukGH were tested for binding to hu and rbCD11b-I to identify the residues involved in receptor recognition in both species. Most of them showed weaker binding with biphasic kinetic profiles; therefore, the binding strength was expressed as RU, which are proportional to the amount of bound cytotoxin (Figure 3A). As expected, the LukGH_E323A resulted in loss of binding (RU <0.05 nm) for both hu and rbCD11b-I. In addition, loss of binding to rbCD11b-I and a significant decrease in binding to huCD11b-I were seen with LukGH_Q116A, _T267A, _R294A, _K319A, and _Y321A (Figure 3A). The LukGH variants with K118A, R119A, and K290A mutations showed decreased binding to rbCD11b-I and huCD11b-I, although less pronounced for the latter. Notably, we have identified three variants, LukGH_E263A, _D312A, and _D316A, which exhibited increased binding responses to rbCD11b-I. The binding to huCD11b-I was either significantly (LukGH_D316A) or only slightly (LukGH_E263A and _D312A) decreased.

Cytotoxicity and receptor binding of LukGH wild-type and Ala variants.

Figure 3.
Cytotoxicity and receptor binding of LukGH wild-type and Ala variants.

(A) Binding of Ala mutants to rbCD11b-I or huCD11b-I, measured by BLI. Mutants with response units <0.05 nm (indicated by the dotted red line) are considered non-binders. Mutants that could not achieve complete cytotoxicity towards rbPMNs (cell viability >25%) at 300 nM in in vitro assay as shown in (B) are marked with stars. Data in (A) represent mean ± SEM of a minimum of two independent experiments. (B) Activity of selected LukGH mutants towards rbPMNs was assessed in a luminescent cell viability assay measuring cellular ATP content at increasing cytotoxin concentrations. Solid lines represent nonlinear fit of the data (mean of triplicates ± SEM) and the dotted line represents EC50 value calculated by the fit. (C and D) Correlation graphs between EC50 wild-type/EC50 mutant, obtained as shown in (B), and binding responses measured by BLI (A), for rbCD11b-I and huCD11b-I, are shown in (C) and (D), respectively. Data represent the mean ± SEM of a minimum of two independent experiments performed in triplicate; except for EC50 of LukGH_D316A with only one experiment performed in triplicate. Pearson correlation was calculated using Prism 6 (GraphPad) and is shown as an insert. Wild-type LukGH, as a referent value, is marked red.

Figure 3.
Cytotoxicity and receptor binding of LukGH wild-type and Ala variants.

(A) Binding of Ala mutants to rbCD11b-I or huCD11b-I, measured by BLI. Mutants with response units <0.05 nm (indicated by the dotted red line) are considered non-binders. Mutants that could not achieve complete cytotoxicity towards rbPMNs (cell viability >25%) at 300 nM in in vitro assay as shown in (B) are marked with stars. Data in (A) represent mean ± SEM of a minimum of two independent experiments. (B) Activity of selected LukGH mutants towards rbPMNs was assessed in a luminescent cell viability assay measuring cellular ATP content at increasing cytotoxin concentrations. Solid lines represent nonlinear fit of the data (mean of triplicates ± SEM) and the dotted line represents EC50 value calculated by the fit. (C and D) Correlation graphs between EC50 wild-type/EC50 mutant, obtained as shown in (B), and binding responses measured by BLI (A), for rbCD11b-I and huCD11b-I, are shown in (C) and (D), respectively. Data represent the mean ± SEM of a minimum of two independent experiments performed in triplicate; except for EC50 of LukGH_D316A with only one experiment performed in triplicate. Pearson correlation was calculated using Prism 6 (GraphPad) and is shown as an insert. Wild-type LukGH, as a referent value, is marked red.

To assess whether the change in binding translates into different in vitro toxicities, we measured the activity of all LukGH variants towards rabbit and human neutrophils in ATP-based cell viability assays. As the activity of LukGH towards huPMNs markedly depends on the activation state of the cells and is therefore subject to higher variability [13,41], we used differentiated HL-60 cells instead of huPMNs for these experiments. As there are no rabbit HL-60 counterparts, rbPMNs were used, despite showing some batch to batch variability (lower though than with huPMNs), and EC50 ratios between wild-type LukGH and variants were used, instead of absolute EC50 values, to minimize variability.

All variants showed some level of toxicity towards rbPMNs at the highest concentration tested (300 nM). While some showed improved activity towards rbPMNs (LukGH_E263A, _D312A), others, such as LukGH_E323A or _R294A, caused only a partial viability reduction (30–70%) even at the highest concentration tested (Figure 3B). The variants that showed significantly lower binding to rbCD11b-I (Figure 3A) could not achieve complete cytotoxicity (>25% of cells still viable) at 300 nM. Towards the differentiated HL-60 cells, the activity was either maintained or decreased, and no activity was seen with LukGH_E323A, in agreement with the literature [32]. The most deleterious effects in activity were observed with the mutations that were also identified using rbPMNs, namely E323A, Q116A, T267A, R294A, K319A, and Y321A, but also with I296A and Y314A that did not affect activity towards rbPMNs. When the EC50 values were plotted against corresponding binding responses, we observed, for most of the variants, a positive correlation between receptor-binding strength and cytotoxin activity for both differentiated HL-60 cells and rbPMNs (Figure 3C,D), indicating that the affected amino acids are indeed involved in the interaction with the cellular receptor. With some variants (LukGH_R119A, _R121A,_Y314A, and _K290A), we observed reduced binding to the rabbit receptor, but no change in activity towards rbPMNs, suggesting that binding to the recombinant domain may not necessarily reflect the binding to the cell surface or translate to cytotoxic activity.

The most interesting variants that were identified in this screening were LukGH_E263A and LukGH_D312A that showed significantly higher binding responses and ∼6- and 3-fold improved cytotoxicity towards rbPMNs, respectively (Table 2). An opposite and weaker effect for the same mutations was observed when tested with huCD11b-I and differentiated HL-60: decrease in both binding and activity. The distinct effects observed with E263A, D312A, I296A, and Y314 mutations in binding to huCD11b-I or rbCD11b-I and in their activity towards differentiated HL-60 cells or rbPMNs are likely the result of different amino acids in huCD11b-I and rbCD11b-I involved in interactions with these LukH residues.

Table 2
Binding to hu and rbCD11b-I and activity of wild-type LukGH and selected LukGH variants towards differentiated HL-60 cells or rabbit PMNs
Toxin Response unitsmax (nm)* Kd (M)* [EC50 wild-type/EC50 mutant]* 
Rabbit 
 LukGH 0.53 ± 0.15 9.51 × 10−8 ± 1.8 × 10−10 1.0 
 LukGH_E263A 1.35 ± 0.08 4.22 × 10−8 ± 2.3 × 10−8 6.5 ± 0.8 
 LukGH_D312A 0.96 ± 0.03 8.98 × 10−8 ± 2.2 × 10−9 2.7 ± 0.6 
 LukGH_D312K 1.53 ± 0.10 3.81 × 10−8 ± 7.8 × 10−9 10.5 ± 1.1 (5.7 ± 1.8) (12.6 ± 5.1)§ 
 LukGH_E263Q-D312N 2.61 ± 0.37 2.20 × 10−8 ± 2.4 × 10−9 10.7 ± 2.1 (19.2 ± 4.0)§ 
 LukGH_E263Q-D312K 2.49 ± 0.46 2.10 × 10−8 ± 3.1 × 10−9 12.7 ± 1.8 (8.6 ± 0.7) 
 LukGH_E263A-D312A 1.94 ± 0.17 3.06 × 10−8 ± 6.3 × 10−9 11.5 ± 2.5 (6.8 ± 0.8) 
Human 
 LukGH 1.39 ± 0.28 7.96 × 10−9 ± 2.5 × 10−9 1.0 
 LukGH_E263A 1.03 ± 0.13 1.77 × 10−8 ± 2.8 × 10−9 0.52 ± 0.01 
 LukGH_K288A 1.54 ± 0.11 7.41 × 10−9 ± 1.6 × 10−10 0.81 ± 0.16 
 LukGH_K290A 1.40 ± 0.05 8.61 × 10−9 ± 1.1 × 10−10 1.12 ± 0.26 
 LukGH_D312A 1.23 ± 0.08 1.13 × 10−8 ± 1.4 × 10−9 0.85 ± 0.25 
 LukGH_D312K 1.25 ± 0.07 3.97 × 10−9 ± 1.2 × 10−9 0.8 ± 0.02 
 LukGH_E263Q-D312N 1.13 ± 0.06 9.61 × 10−9 ± 1.1 × 10−9 0.58 ± 0.05 
Toxin Response unitsmax (nm)* Kd (M)* [EC50 wild-type/EC50 mutant]* 
Rabbit 
 LukGH 0.53 ± 0.15 9.51 × 10−8 ± 1.8 × 10−10 1.0 
 LukGH_E263A 1.35 ± 0.08 4.22 × 10−8 ± 2.3 × 10−8 6.5 ± 0.8 
 LukGH_D312A 0.96 ± 0.03 8.98 × 10−8 ± 2.2 × 10−9 2.7 ± 0.6 
 LukGH_D312K 1.53 ± 0.10 3.81 × 10−8 ± 7.8 × 10−9 10.5 ± 1.1 (5.7 ± 1.8) (12.6 ± 5.1)§ 
 LukGH_E263Q-D312N 2.61 ± 0.37 2.20 × 10−8 ± 2.4 × 10−9 10.7 ± 2.1 (19.2 ± 4.0)§ 
 LukGH_E263Q-D312K 2.49 ± 0.46 2.10 × 10−8 ± 3.1 × 10−9 12.7 ± 1.8 (8.6 ± 0.7) 
 LukGH_E263A-D312A 1.94 ± 0.17 3.06 × 10−8 ± 6.3 × 10−9 11.5 ± 2.5 (6.8 ± 0.8) 
Human 
 LukGH 1.39 ± 0.28 7.96 × 10−9 ± 2.5 × 10−9 1.0 
 LukGH_E263A 1.03 ± 0.13 1.77 × 10−8 ± 2.8 × 10−9 0.52 ± 0.01 
 LukGH_K288A 1.54 ± 0.11 7.41 × 10−9 ± 1.6 × 10−10 0.81 ± 0.16 
 LukGH_K290A 1.40 ± 0.05 8.61 × 10−9 ± 1.1 × 10−10 1.12 ± 0.26 
 LukGH_D312A 1.23 ± 0.08 1.13 × 10−8 ± 1.4 × 10−9 0.85 ± 0.25 
 LukGH_D312K 1.25 ± 0.07 3.97 × 10−9 ± 1.2 × 10−9 0.8 ± 0.02 
 LukGH_E263Q-D312N 1.13 ± 0.06 9.61 × 10−9 ± 1.1 × 10−9 0.58 ± 0.05 
*

Mean of a minimum of two independent experiments performed in triplicates ± SD; Kd determined by steady-state analysis, LPS-treated cells, §SYTOX green readout.

Delineation of the CD11b-I-binding epitope on LukGH

Based on the binding and activity data, we have identified positions on LukH, which are likely, directly, or indirectly, involved in the binding of LukGH to CD11b-I with surface area of 2400 Å2 (calculated using The PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC) (Figure 4). The positions considered as direct contacts are those where (a) binding was either improved or completely lost in at least one species and (b) binding and activity were decreased in both species. Indirect contacts were defined as positions where mutations led to at least a 2-fold effect in activity and decreased binding in at least one of the species. Nine out of the fourteen amino acids identified as part of the binding epitope are charged or polar indicating that binding occurs in the extracellular milieu, rather than at the membrane surface. Most of the positions identified as part of the binding epitope are conserved between the natural sequence variants of LukH (Supplementary Table S2).

CD11b binding epitope on LukH based on the mutagenesis data.

Figure 4.
CD11b binding epitope on LukH based on the mutagenesis data.

Left panel: The binding epitope of CD11b-I on the LukH monomer shown on LukGH octamer structure (PDB 4TW1). LukH, LukG, and LukG from a neighboring dimer are colored yellow, green, and purple, respectively, and remaining monomers in gray. The residues involved in binding are colored by atom (red: oxygen, blue: nitrogen). Right panel: Residues in the binding epitope are shown as green sticks and colored according to the atom (red: oxygen, blue: nitrogen). The residues involved in the direct interactions are labeled black and the residues involved in indirect interactions are labeled red.

Figure 4.
CD11b binding epitope on LukH based on the mutagenesis data.

Left panel: The binding epitope of CD11b-I on the LukH monomer shown on LukGH octamer structure (PDB 4TW1). LukH, LukG, and LukG from a neighboring dimer are colored yellow, green, and purple, respectively, and remaining monomers in gray. The residues involved in binding are colored by atom (red: oxygen, blue: nitrogen). Right panel: Residues in the binding epitope are shown as green sticks and colored according to the atom (red: oxygen, blue: nitrogen). The residues involved in the direct interactions are labeled black and the residues involved in indirect interactions are labeled red.

Receptor-binding and PMN activity of LukGH variants mutated at amino acid positions 263 and 312 in the rabbit system

The increase in binding and activity of the LukGH Ala mutants at positions 263 and 312 prompted us to further screen these positions for additional improvement in activity towards rbPMNs. We tested smaller (G), positively charged (K), or the same size, polar but not charged (E263Q and D312N) amino acids at positions 263 and 312. In addition, at positions 263 and 312, five other single mutants (D, V or S and V or H, respectively) and several double mutants were generated (Figure 5A,B; Supplementary Figure S3). The variants were characterized by CD and DSF (Figure 5A; Supplementary Figures S3A and S3B). A difference in CD signals observed for two of the variants, LukGH_D312G and _D312V, indicated changes in the secondary structure relative to the wild type and these proteins were therefore excluded from further analyses (Figure 5A). The CD spectra and Tm values of all other variants, except for LukGH_E263K variant for which Tm could not be measured due to low protein yield, were comparable to those of wild-type LukGH (Supplementary Figures S3A and S3B). When these variants were tested for binding to rbCD11b-I and activity towards rbPMNs, we observed a positive correlation between activity and binding (Figure 5B), as previously seen for the Ala mutants (Figure 3C). Most of the single mutants at positions 263 and 312 had a similar or negative effect compared with the corresponding Ala variants (Figure 5B). The exception was the introduction of K at position 312, which resulted in ∼10-fold improved activity towards rbPMNs, compared with the wild-type LukGH, and an improvement in binding kinetics that could be fitted to a Kd of 3.8 × 10−8 M (Table 2). With the double mutants, we observed, in general, an additive effect of the mutations. The LukGH_E263A-D312A variant, for example, exhibited increased binding and activity when compared with the single mutants LukGH_E263A and LukGH_D312A (Figure 5B; Table 2). However, no significant further improvement in Kd or activity could be achieved with the double mutants over the single mutants when amino acids other than Ala were introduced.

Targeted mutagenesis at amino acid positions 263 and 312 in LukH.

Figure 5.
Targeted mutagenesis at amino acid positions 263 and 312 in LukH.

(A) CD spectra of selected LukGH mutants measured in 20 mM sodium phosphate, pH 7.5 plus NaCl (150–300 mM) and normalized to 0.3 mg/ml. (B) Correlation between EC50 wild-type/EC50 mutant towards rbPMNs and binding responses to rbCD11b-I measured by BLI. Wild-type LukGH is colored red and the LukGH variants selected for chromosomal integration are marked green. Data represent mean ± SEM of a minimum of two independent experiments performed in triplicates; except for LukGH_E263K with only one experiment for EC50 performed in triplicate. Pearson correlation was calculated using Prism 6 (GraphPad) and is shown as an insert.

Figure 5.
Targeted mutagenesis at amino acid positions 263 and 312 in LukH.

(A) CD spectra of selected LukGH mutants measured in 20 mM sodium phosphate, pH 7.5 plus NaCl (150–300 mM) and normalized to 0.3 mg/ml. (B) Correlation between EC50 wild-type/EC50 mutant towards rbPMNs and binding responses to rbCD11b-I measured by BLI. Wild-type LukGH is colored red and the LukGH variants selected for chromosomal integration are marked green. Data represent mean ± SEM of a minimum of two independent experiments performed in triplicates; except for LukGH_E263K with only one experiment for EC50 performed in triplicate. Pearson correlation was calculated using Prism 6 (GraphPad) and is shown as an insert.

We reported recently that activation of huPMNs by LPS induced up-regulation of the CD11b receptor and resulted in increased sensitivity towards LukGH [13]. However, the ability of LPS to alter the surface expression of rbCD11b on rbPMNs and susceptibility to LukGH has not been characterized. To exclude the possibility that LPS contamination of the recombinant cytotoxins contributes to the increased activity of LukGH variants, we tested the activity of selected variants (LukGH_D312K, LukGH_E263A-D312A, and LukGH_E263Q-D312K) with LPS-stimulated rbPMNs. While we have observed that LPS increased the sensitivity of rbPMNs to LukGH (by 5–10-fold), the activity improvement obtained with these mutations was maintained in the presence of saturating amounts of LPS (6–8-fold improvement compared with wild-type LukGH) (Table 2).

Targeted mutagenesis of the CD11b-I-binding epitope

With the D312K and E263Q-D312N mutations in LukH, we were able to increase the activity towards rbPMNs by about one order of magnitude. However, this is still one order of magnitude weaker than the activity of LukGH towards huPMNs [22] (Figure 1C). To further improve the activity against rbPMNs, we generated a set of 26 triple mutants using the following backgrounds: LukGH_E263Q-D312N, _E263Q-D312K, and _E263A-D312A. We mutated LukH residues at positions which were previously identified to be important for binding and activity (i.e. 116, 118, and 294) and other proximal residues (114 and 298) to rationally selected amino acids. We introduced amino acids that appear in natural sequence variants of LukH (Supplementary Table S2), have increased or decreased size, and different charge. These triple mutants showed conserved, decreased, or lack of binding to rbCD11b-I (Figure 6A,B). Interestingly, we observed a complete loss of binding to rbCD11b-I for the triple mutants involving position 114, which was not tested as a single Ala mutant. The CD spectra and Tm values of the LukGH_E263Q-D312N-D114 mutants confirmed that folding and stability were unchanged compared with that of the wild-type LukGH, thus we concluded that the amino acid at position 114 is part of the binding epitope (Figure 4). The variants that showed similar binding responses and Kds as LukGH_E263Q-D312N were tested for activity towards rbPMNs. We detected no further improved activity, in agreement with the binding data. Interestingly, even small changes in Kd (<2-fold) had marked effects on activity (Figure 6B).

Binding and activity of LukGH triple mutants.

Figure 6.
Binding and activity of LukGH triple mutants.

(A) Binding of selected triple mutants to rbCD11b-I was measured by BLI. Data represent mean ± SEM of a minimum of two independent experiments. Abbreviations: QN = LukGH_E263Q-D312N; AA = LukGH_E263A-D312A. (B) Kd and EC50wild-type/EC50 mutant for triple mutants tested for activity towards rbPMNs. Data represent mean ± SEM of a minimum of two independent experiments performed in triplicates; except for LukGH E263Q-D312N-R119K with only one experiment for EC50 performed in triplicates. Abbreviations: QN = LukGH_E263Q-D312N; QK = LukGH_E263Q-D312K.

Figure 6.
Binding and activity of LukGH triple mutants.

(A) Binding of selected triple mutants to rbCD11b-I was measured by BLI. Data represent mean ± SEM of a minimum of two independent experiments. Abbreviations: QN = LukGH_E263Q-D312N; AA = LukGH_E263A-D312A. (B) Kd and EC50wild-type/EC50 mutant for triple mutants tested for activity towards rbPMNs. Data represent mean ± SEM of a minimum of two independent experiments performed in triplicates; except for LukGH E263Q-D312N-R119K with only one experiment for EC50 performed in triplicates. Abbreviations: QN = LukGH_E263Q-D312N; QK = LukGH_E263Q-D312K.

The best variants in terms of binding and activity, and with minimal change on protein structure, were LukGH_D312K and LukGH_E263Q-D312N, with ∼10-fold better activity towards rbPMNs and with Kd values of 20–40 nM (Table 2). We confirmed that the increased toxicity of the LukGH variants is the result of increased membrane damage of rbPMNs by employing the SYTOX green nucleic acid staining [28,42] (Table 2). Binding to huCD11b-I was unchanged and activity towards differentiated HL-60 cells was less than 2-fold reduced (Table 2). We therefore selected these two variants for replacement of the wild-type LukGH in the chromosome of S. aureus for further studies with natively produced cytotoxin.

Integration of rabbit-adapted lukGH into the S. aureus chromosome

Two S. aureus strains were selected for chromosomal integration of rabbit-adapted lukGH variants: a prototype USA300 CA-MRSA (TCH1516 strain) and LA#5, a clinical MSSA isolate with high LukGH expression level in vitro [29,38]. The mutated S. aureus strains were compared with the parental wild-type strains in all studies.

To confirm that the chromosomal integration did not alter protein expression, we compared expression levels for all β-barrel pore-forming cytotoxins (leukocidins and Hla) in CSs of S. aureus strains carrying the wild-type, the single, or double mutant LukGH variants, grown to comparable OD600 (Supplementary Figure S4), by Western blot. The gene replacement did not change the predicted endogenous promoter sequence and therefore LukGH expression was not expected to be affected. LukGH expression levels were comparable in the wild-type and mutant strains and in line with published data: LA#5 expressed significantly more LukGH than TCH1516 [29] (Figure 7A). Likewise, same levels of expression between wild-type and mutant strains were observed for the other cytotoxins, LukD, HlgB, Hla, and LukS-PV (present only in the TCH1516 strain), for both strains (Figure 7A).

Characterization of S. aureus strains expressing the rabbit-adapted LukGH: cytotoxin expression and effect of mAbs on cytotoxin activity.

Figure 7.
Characterization of S. aureus strains expressing the rabbit-adapted LukGH: cytotoxin expression and effect of mAbs on cytotoxin activity.

(A) Western blot of CSs from S. aureus TCH1516 and LA#5 wild-type, rabbit-adapted strains (lukH_D312K and lukH_E263Q-D312N (lukH_QN)), and control strain (TCH1516 Δ all cytotoxins). Control: 0.1 µg recombinant protein. (B) Contribution of monoclonal antibodies ASN-1 and ASN-2 to protection of rbPMNs from 16× diluted CSs prepared in RPMI + 1% CAS (mean of four to six supernatants prepared from different colonies and on different days ± SEM). (C) Activity of CSs of LA#5, LA#5_D312K, and LA#5_E263Q-D312N, prepared in RPMI + 1% CAS, towards rbPMNs in the presence and absence of ASN-1, ASN-2, and an equimolar mixture of both antibodies (mean of four to six supernatants prepared from different colonies and on different days ± SEM). (D) Comparison of protection of rb and huPMNs by the monoclonal antibodies ASN-1, ASN-2, or the equimolar mixture of the two mAbs in the presence of CSs of LA#5 and LA#5_D312K (32× dilution) prepared in RPMI + 1% CAS (mean of four to six supernatants prepared from different colonies and on different days ± SEM). The arrow indicates the contribution of LukGH to CS toxicity. (E) Ex vivo toxicity of LA#5 and LA#5_D312K at MOI 50 (left panel) and MOI 100 (right panel) towards rbPMNs in the presence of ASN-1, ASN-2, an equimolar mixture of both antibodies, a negative control antibody and buffer alone (no mAbs) (mean of two independent experiments ± SEM). The data are expressed as % dead cells relative to control antibody and calculated as follows: [%dead cells/%dead cellscontrol mAb] × 100. The arrow indicates the contribution of LukGH to the toxicity.

Figure 7.
Characterization of S. aureus strains expressing the rabbit-adapted LukGH: cytotoxin expression and effect of mAbs on cytotoxin activity.

(A) Western blot of CSs from S. aureus TCH1516 and LA#5 wild-type, rabbit-adapted strains (lukH_D312K and lukH_E263Q-D312N (lukH_QN)), and control strain (TCH1516 Δ all cytotoxins). Control: 0.1 µg recombinant protein. (B) Contribution of monoclonal antibodies ASN-1 and ASN-2 to protection of rbPMNs from 16× diluted CSs prepared in RPMI + 1% CAS (mean of four to six supernatants prepared from different colonies and on different days ± SEM). (C) Activity of CSs of LA#5, LA#5_D312K, and LA#5_E263Q-D312N, prepared in RPMI + 1% CAS, towards rbPMNs in the presence and absence of ASN-1, ASN-2, and an equimolar mixture of both antibodies (mean of four to six supernatants prepared from different colonies and on different days ± SEM). (D) Comparison of protection of rb and huPMNs by the monoclonal antibodies ASN-1, ASN-2, or the equimolar mixture of the two mAbs in the presence of CSs of LA#5 and LA#5_D312K (32× dilution) prepared in RPMI + 1% CAS (mean of four to six supernatants prepared from different colonies and on different days ± SEM). The arrow indicates the contribution of LukGH to CS toxicity. (E) Ex vivo toxicity of LA#5 and LA#5_D312K at MOI 50 (left panel) and MOI 100 (right panel) towards rbPMNs in the presence of ASN-1, ASN-2, an equimolar mixture of both antibodies, a negative control antibody and buffer alone (no mAbs) (mean of two independent experiments ± SEM). The data are expressed as % dead cells relative to control antibody and calculated as follows: [%dead cells/%dead cellscontrol mAb] × 100. The arrow indicates the contribution of LukGH to the toxicity.

Cytotoxicity of rabbit-adapted S. aureus strains towards rabbit and huPMNs

Next, we wanted to confirm that the activity improvement observed with the recombinant LukGH_D312K and _E263Q-D312N towards rbPMNs was retained with the native protein secreted into the culture medium. Testing the CS activity is important because of the different expression levels of different cytotoxins and their relative contribution to PMN killing [6,29,42]. The activity of the CSs towards rbPMNs was measured in the presence of cytotoxin-neutralizing antibodies to dissect the contribution of LukGH. We employed ASN-2, a LukGH-specific neutralizing antibody [37], ASN-1, a human mAb that cross-neutralizes Hla and the other four leukocidins (LukED, LukSF-PV, HlgAB, and HlgCB) [34] or an equimolar mixture of the two antibodies that allows simultaneous targeting of all five leukocidins and therefore complete inhibition of neutrophil toxicity [29]. The strains were grown in RPMI + 1% CAS to mimic the in vivo milieu (low iron and nutrient content); the growth kinetics and optical densities at the time point when CSs were collected were comparable (Supplementary Figure S4).

We have recently shown that full inhibition of TCH1516 and LA#5 toxicity towards huPMNs is dependent on the concerted activity of ASN-1 and ASN-2 [29]. ASN-1 alone resulted in partial inhibition of the TCH1516 CS, while for the LukSF-PV-negative strain, LA#5, inhibition was largely but not exclusively driven by ASN-2 [29]. When testing the same strains on rbPMNs, we found that ASN-2 was less critical and inhibition was almost exclusively conferred by ASN-1, a consequence of the lower sensitivity of rbPMNs to wild-type LukGH (Figure 7B).

Engineering LukGH for improved binding to rbCD11b-I resulted in increased CS toxicity for the LA#5 strain towards rbPMNs. EC50 values differed by a factor of 1.5 and 2 for LA#5_D312K or LA#5_E263Q-D312N, respectively (Figure 7C). In line with the data generated with huPMNs [29], the combination of the two mAbs was able to completely block toxicity towards rbPMNs (Figure 7C). The increased activity of the rabbit-adapted LukGH became apparent in the presence of ASN-1 when only LukGH was active. While the EC50 in the presence of ASN-1 for the LA#5 wild-type CS was 16× (dilution factor), it was ∼4-fold higher (∼64-fold) for the LA#5_D312K- and LA#5_E263Q-D312N-expressing strains. Notably, cytotoxicity in the presence of ASN-2 remained unchanged between wild-type and mutants, confirming that the expression levels of the other cytotoxins were not affected by the rabbit adaptation and that increased toxicity was a direct effect of LukGH mutagenesis. A direct comparison of the neutralization patterns of human versus rbPMNs revealed that the effect was rabbit-specific. The relative mAb and cytotoxin contribution of the parental and adapted strains were comparable with huPMNs, but differed substantially when tested with rbPMNs (Figure 7D).

Interestingly, LukGH adaptation in the TCH1516 strains, with either D312K or E263Q-D312N mutation, did not result in a significant change in CS toxicity in the absence of antibodies when tested towards either hu or rbPMNs. We observed an effect only in ASN-1 protection level towards rbPMNs for both mutants at the lowest CS dilutions tested (4–8-fold), and as expected, not towards huPMNs (Supplementary Figure S5). We speculate that this is due to the lower expression of LukGH in TCH1516 compared with LA#5 [29], and the presence of LukSF-PV, which has a significant contribution to overall in vitro toxicity.

We have recently shown the dominant role of LukGH during ex vivo infections of huPMNs [13,29] and we wanted to assess how this translates to the rabbit-adapted S. aureus strains with rbPMNs. Rabbit PMNs were infected with either LA#5 or LA#5_D312K at MOI 50 and 100 in the presence or absence of ASN-1 and/or ASN-2. PMN membrane damage was assessed after 2 h by SYTOX green nucleic acid staining. With both strains, we observed mostly LukGH-mediated killing as reflected by the same level of protection in the presence of ASN-2 and ASN-1 + ASN-2 (Figure 7E). Furthermore, we observed lower ASN-1 protection levels towards rbPMNs for LA#5_D312K (Figure 7E), confirming increased LukGH contribution to the killing, as seen with the CSs (Figure 7C,D).

Discussion

The most established animal models for studying S. aureus pathogenesis involve mice and rabbits, which are typically resistant to the bacterium, and therefore require much higher bacterial loads for disease onset and show faster disease progression than humans [7,22,43]. Nevertheless, these models were able to pinpoint the importance of four S. aureus toxins: α-hemolysin, γ-hemolysin, LukED, and LukSF-PV, to S. aureus pathogenesis/virulence [3,7,4347]. This is supported, at least for Hla and LukSF-PV, by clinical data, as inferred from cytotoxin expression profiles [38,4648], presence of neutralizing antibodies and clinical outcome [4951]. LukGH is the only S. aureus leukocidin whose activity and relevance in vivo, in these animal models, cannot be evaluated, despite its established contribution in in vitro and in ex vivo models using human cells [13,2729,42]. It has been proposed that LukGH, alone, or in combination with Hla leads to increased organ bacterial load and biofilm formation in mouse bacteremia models [31,52]. The relevance and specificity of these effects is unclear, since mouse cells are resistant to LukGH [22], and binding of LukGH to mouse CD11b is extremely week [21]. Owing to the high cytotoxicity of LukGH and the important role of receptor up-regulation in LukGH cytotoxicity towards human cells [13,41], an in vivo model that reflects the human sensitivity towards LukGH is important.

The risk associated with humanizing animals for the LukGH receptor, e.g. creating a human CD11b/CD18 knock-in mouse model, is high, as CD11b, particularly its I-domain, is interacting with endogenous host factors, such as complement and blood coagulation factors [53]; the contribution of CD11b during S. aureus infection may be altered independent of its role as LukGH receptor. Therefore, the adaptation of LukGH to the animal hosts by in vitro design, as we describe here, is a closer mimic of a natural infection.

The rabbit is an ideal species for this purpose, since there is a measurable, although weak interaction between LukGH and the rabbit receptor (Figure 1B), accompanied by detectable PMN cytotoxicity (Figure 1C). By targeted mutagenesis of LukH surface-exposed residues predicted to be involved in the interaction with CD11b, we were able to increase rbPMN cytotoxicity of LukGH in a CD11b-I-dependent manner. The effects of these particular mutations are rabbit-specific, i.e. no significant change in cytotoxicity towards huPMNs, and also not related to a change in protein stability or expression level. The increased LukGH contribution to cytotoxicity towards rbPMNs in the supernatants of S. aureus strains expressing the rabbit-adapted LukGH mutants parallels the effects observed with the corresponding recombinant variants. This validates our approach and indicates that such rabbit-adapted strains are viable candidates for studying the role of LukGH in S. aureus pathogenesis in animals.

The rabbit is a suitable model organism, not only because most of the leukotoxins are active on rabbit cells, but also because rabbits are a natural host for S. aureus. S. aureus strains isolated from rabbits and humans differ by as little as one nucleotide [54]. Rabbits as well as humans show a plethora of natural staphylococcal infections: pneumonia, soft tissue infections, and bacteremia. Using a bona fide rabbit S. aureus strain to integrate the LukGH variants described here is another option worth considering in future experiments. The currently sequenced ST121 rabbit isolates carry lukGH but do not express the active LukGH dimer [5456] — either due to stop codons in lukG or lukH or to a point mutation in LukG (E45K, WP_046463168.1), located in the LukGH dimer interface [33,37], that disrupts dimer formation (unpublished data).

It is interesting to note that none of the mutations we found to increase activity towards rbPMNs are present in over 100 published LukGH sequences [30] (Supplementary Table S2). Moreover, the three most distant sequence variants tested, including the one encoded by the livestock H19 strain, did not show any advantage, compared with TCH1516, in lysing rbPMNs. We have also tested most of the naturally occurring mutations in the epitope we have identified (Supplementary Table S2), and none of these showed significantly improved reactivity towards rbPMNs, while some decreased it. Although present in bovine strains, LukGH is not active towards bovine neutrophils [57]. It is therefore not clear if and how LukGH contributes to pathogenesis in animal hosts, which is an interesting aspect to investigate in future studies.

It has been previously demonstrated that another potent leukocidin, LukSF-PV, has an essential role in necrotizing pneumonia in rabbits, causing massive lung damage and inflammation [7]. It remains to be seen whether the same is true for rabbit-adapted LukGH. It is also noteworthy that red blood cell lysis, both by Hla and some of the leukocidins, is much more pronounced with rabbit compared with human cells. Moreover, we observed an ∼100-fold higher cytotoxicity of LukED towards rabbit compared with human neutrophils [12], and LukED is present in ∼60% of S. aureus isolates [30]. Despite the fact that, based on in vitro data, rabbits currently appear as the most suitable small animal laboratory species for studying the role of the bi-component leukocidins in S. aureus pathogenesis (Table 1), the prospect for additional models should not be understated.

By mapping the CD11b-I-binding site on LukH, we have identified 14 residues that are in the contact with the receptor and might be useful in similar approaches to increase LukGH susceptibility to other species, including e.g. mouse and guinea pigs. Previously, a single residue (E323) was shown to significantly affect LukGH binding to CD11b-I and PMN activity [32]. Although the authors investigated some of the positions of the binding epitope we have delineated here (positions 316, 319, and 321 were mutated to alanine) for activity towards huPMNs (at 33 nM LukGH), they concluded these are not important for activity. Our data do not contradict their findings, but highlight possible differences in binding epitopes between hu and rbCD11b-I and indicate that lower LukGH concentrations have to be used to distinguish fine differences in activity (instead of all or none read-outs).

We demonstrated here how targeted mutagenesis can be used to expand the host receptor recognition of a bacterial toxin, even in the absence of structural information for the interaction between the toxin and its receptor. It is envisaged that the same method could be applied for other virulence factors, e.g. for the other leukocidins of S. aureus that bind to extracellular loops of transmembrane receptors, for which obtaining high-resolution structures to determine the exact binding interface is challenging. As this in vitro adaptation can be considered, in a sense, a directed evolution of an isolated toxin–receptor pair, it has significant potential over alternative approaches (such as humanization approaches that alter the host), for the in vivo dissection of the mode of action of drugs targeting such virulence factors in efficacy studies. However, as with any engineering effort, care must be taken that the property desired and followed is the only one affected (e.g. no expansion of the receptor-binding activity of the toxin, conferring artefactual pathogenic properties, as seen with murinized InlA from L. monocytogenes [58]) by such efforts. Confirming the in vivo phenotype with different variants, containing various amino acid substitutions (such as LukGH D312K and E263Q-D312N) for gain of function, and with controls that contain loss of function mutations, is essential.

Abbreviations

     
  • BLI

    Bio-layer interferometry

  •  
  • CD

    circular dichroism

  •  
  • CD11b-I

    I-domain of CD11b

  •  
  • CS

    culture supernatant

  •  
  • DSF

    differential scanning fluorimetry

  •  
  • huCD11b-I

    I-domain of CD11b from human

  •  
  • huPMNs

    human polymorphonuclear cells

  •  
  • IPTG

    isopropyl β-D-1-thiogalactopyranoside

  •  
  • Kd

    equilibrium dissociation constant

  •  
  • LPS

    lipopolysaccharide

  •  
  • PMNs

    polymorphonuclear cells

  •  
  • rbCD11b-I

    I-domain of CD11b from rabbit

  •  
  • rbPMNs

    rabbit polymorphonuclear cells

  •  
  • RU

    response units

  •  
  • Tm

    melting point

Author Contribution

A.B. and E.N. designed the study. N.T. performed the experiments, except the generation of the mutant S. aureus strains that was done by L.S. and M.Z. H.R. designed the cell-based experiments. J.Z. performed cloning of CD11b-I variants and preliminary binding experiments. A.B. and N.T. wrote the manuscript with input from E.N., H.R., and L.S.

Funding

This work was supported by the FFG ‘Basisprogramm’ Grants 832915, 837128, 841918, and 845382 from the Austrian Research Promotion Agency (awarded to Arsanis Biosciences).

Acknowledgements

The CD and DSF measurements were performed at the VBCF Protein Technologies Facility (www.vbcf.ac.at). We thank Gabor Nagy for help with designing the S. aureus chromosomal integration and Karin Gross and Barbara Maierhofer for their technical help with cell-based experiments.

Competing Interests

The authors declare a potential conflict of interest as the work was performed at Arsanis Biosciences GmbH (Vienna, Austria), the wholly own subsidiary of Arsanis, Inc., a biotechnology company developing a monoclonal antibody-based product targeting S. aureus infections. L.S., H.R, M.Z., J.Z., E.N., and A.B. declare a potential conflict of interest as they are shareholders in Arsanis, Inc.

References

References
1
Tong
,
S.Y.C.
,
Davis
,
J.S.
,
Eichenberger
,
E.
,
Holland
,
T.L.
and
Fowler
, Jr,
V.G.
(
2015
)
Staphylococcus aureus infections: epidemiology, pathophysiology, clinical manifestations, and management
.
Clin. Microbiol. Rev.
28
,
603
661
2
Vandenesch
,
F.
,
Lina
,
G.
and
Henry
,
T.
(
2012
)
Staphylococcus aureus hemolysins, bi-component leukocidins, and cytolytic peptides: a redundant arsenal of membrane-damaging virulence factors?
Front. Cell. Infect. Microbiol.
2
,
12
3
Berube
,
B.
and
Wardenburg
,
J.
(
2013
)
Staphylococcus aureus α-toxin: nearly a century of intrigue
.
Toxins
5
,
1140
1166
4
Spaan
,
A.N.
,
van Strijp
,
J.A.G.
and
Torres
,
V.J.
(
2017
)
Leukocidins: staphylococcal bi-component pore-forming toxins find their receptors
.
Nat. Rev. Microbiol.
15
,
435
447
5
Reyes-Robles
,
T.
and
Torres
,
V.J
. (
2017
) Staphylococcus aureus pore-forming toxins. In
Staphylococcus aureus: Microbiology, Pathology, Immunology, Therapy and Prophylaxis
(
Bagnoli
,
F.
,
Rappuoli
,
R.
and
Grandi
,
G
., eds), pp.
121
144
,
Springer International Publishing
,
Cham
6
Alonzo
,
F.
and
Torres
,
V.J.
(
2014
)
The bicomponent pore-forming leucocidins of Staphylococcus aureus
.
Microbiol. Mol. Biol. Rev.
78
,
199
230
7
Diep
,
B.A.
,
Chan
,
L.
,
Tattevin
,
P.
,
Kajikawa
,
O.
,
Martin
,
T.R.
,
Basuino
,
L.
et al.  (
2010
)
Polymorphonuclear leukocytes mediate Staphylococcus aureus Panton-Valentine leukocidin-induced lung inflammation and injury
.
Proc. Natl Acad. Sci. U.S.A.
107
,
5587
5592
8
Seilie
,
E.S.
and
Bubeck Wardenburg
,
J.
(
2017
)
Staphylococcus aureus pore-forming toxins: the interface of pathogen and host complexity
.
Semin. Cell Dev. Biol.
72
,
101
116
9
Spaan
,
A.N.
,
Henry
,
T.
,
van Rooijen
,
W.J.M.
,
Perret
,
M.
,
Badiou
,
C.
,
Aerts
,
P.C.
et al.  (
2013
)
The staphylococcal toxin Panton-Valentine leukocidin targets human C5a receptors
.
Cell Host Microbe
13
,
584
594
10
Laventie
,
B.-J.
,
Guérin
,
F.
,
Mourey
,
L.
,
Tawk
,
M.Y.
,
Jover
,
E.
,
Maveyraud
,
L.
et al.  (
2014
)
Residues essential for Panton-Valentine leukocidin S component binding to its cell receptor suggest both plasticity and adaptability in its interaction surface
.
PLoS ONE
9
,
e92094
11
Tromp
,
A.T.
,
Van Gent
,
M.
,
Abrial
,
P.
,
Martin
,
A.
,
Jansen
,
J.P.
,
De Haas
,
C.J.C.
et al.  (
2018
)
Human CD45 is an F-component-specific receptor for the staphylococcal toxin Panton–Valentine leukocidin
.
Nat. Microbiol.
3
,
708
717
12
Diep
,
B.A.
,
Le
,
V.T.M.
,
Visram
,
Z.C.
,
Rouha
,
H.
,
Stulik
,
L.
,
Dip
,
E.C.
et al.  (
2016
)
Improved protection in a rabbit model of community-associated methicillin-resistant Staphylococcus aureus necrotizing pneumonia upon neutralization of leukocidins in addition to alpha-hemolysin
.
Antimicrob. Agents Chemother.
60
,
6333
6340
13
Janesch
,
P.
,
Rouha
,
H.
,
Weber
,
S.
,
Malafa
,
S.
,
Gross
,
K.
,
Maierhofer
,
B.
et al.  (
2017
)
Selective sensitization of human neutrophils to LukGH mediated cytotoxicity by Staphylococcus aureus and IL-8
.
J. Infect.
74
,
473
483
14
Löffler
,
B.
,
Hussain
,
M.
,
Grundmeier
,
M.
,
Brück
,
M.
,
Holzinger
,
D.
,
Varga
,
G.
et al.  (
2010
)
Staphylococcus aureus Panton-Valentine leukocidin is a very potent cytotoxic factor for human neutrophils
.
PLoS Pathog.
6
,
e1000715
15
Koop
,
G.
,
Vrieling
,
M.
,
Storisteanu
,
D.M.L.
,
Lok
,
L.S.
,
Monie
,
T.
,
van Wigcheren
,
G.
et al.  (
2017
)
Identification of LukPQ, a novel, equid-adapted leukocidin of Staphylococcus aureus
.
Sci. Rep.
7
,
40660
16
Spaan
,
A.N.
,
Schiepers
,
A.
,
de Haas
,
C.J.C.
,
van Hooijdonk
,
D.D.J.J.
,
Badiou
,
C.
,
Contamin
,
H.
et al.  (
2015
)
Differential interaction of the staphylococcal toxins Panton–Valentine leukocidin and γ-hemolysin CB with human C5a receptors
.
J. Immunol.
195
,
1034
1043
17
Alonzo
III
F.
,
Kozhaya
,
L.
,
Rawlings
,
S.A.
,
Reyes-Robles
,
T.
,
DuMont
,
A.L.
,
Myszka
,
D.G.
et al.  (
2012
)
CCR5 is a receptor for Staphylococcus aureus leukotoxin ED
.
Nature
493
,
51
55
18
Reyes-Robles
,
T.
,
Alonzo
,
F.
,
Kozhaya
,
L.
,
Lacy
,
D.B.
,
Unutmaz
,
D.
and
Torres
,
V.J.
(
2013
)
Staphylococcus aureus leukotoxin ED targets the chemokine receptors CXCR1 and CXCR2 to kill leukocytes and promote infection
.
Cell Host Microbe
14
,
453
459
19
Spaan
,
A.N.
,
Reyes-Robles
,
T.
,
Badiou
,
C.
,
Cochet
,
S.
,
Boguslawski
,
K.M.
,
Yoong
,
P.
et al.  (
2015
)
Staphylococcus aureus targets the Duffy antigen receptor for chemokines (DARC) to lyse erythrocytes
.
Cell Host Microbe
18
,
363
370
20
Yoong
,
P.
and
Torres
,
V.J.
(
2015
)
Counter inhibition between leukotoxins attenuates Staphylococcus aureus virulence
.
Nat. Commun.
6
,
8125
21
DuMont
,
A.L.
,
Yoong
,
P.
,
Day
,
C.J.
,
Alonzo
,
F.
,
McDonald
,
W.H.
,
Jennings
,
M.P.
et al.  (
2013
)
Staphylococcus aureus LukAB cytotoxin kills human neutrophils by targeting the CD11b subunit of the integrin Mac-1
.
Proc. Natl Acad. Sci. U.S.A.
110
,
10794
10799
22
Malachowa
,
N.
,
Kobayashi
,
S.D.
,
Braughton
,
K.R.
,
Whitney
,
A.R.
,
Parnell
,
M.J.
,
Gardner
,
D.J.
et al.  (
2012
)
Staphylococcus aureus leukotoxin GH promotes inflammation
.
J. Infect. Dis.
206
,
1185
1193
23
Spaan
,
A.N.
,
Vrieling
,
M.
,
Wallet
,
P.
,
Badiou
,
C.
,
Reyes-Robles
,
T.
,
Ohneck
,
E.A.
et al.  (
2014
)
The staphylococcal toxins γ-haemolysin AB and CB differentially target phagocytes by employing specific chemokine receptors
.
Nat. Commun.
5
,
5438
24
Pan
,
X.
,
Yang
,
Y.
and
Zhang
,
J.-R.
(
2014
)
Molecular basis of host specificity in human pathogenic bacteria
.
Emerg. Microbes Infect.
3
,
e23
25
Prince
,
A.
,
Wang
,
H.
,
Kitur
,
K.
and
Parker
,
D.
(
2017
)
Humanized mice exhibit increased susceptibility to Staphylococcus aureus pneumonia
.
J. Infect. Dis.
215
,
1386
1395
26
Wollert
,
T.
,
Pasche
,
B.
,
Rochon
,
M.
,
Deppenmeier
,
S.
,
van den Heuvel
,
J.
,
Gruber
,
A.D.
et al.  (
2007
)
Extending the host range of Listeria monocytogenes by rational protein design
.
Cell
129
,
891
902
27
Ventura
,
C.L.
,
Malachowa
,
N.
,
Hammer
,
C.H.
,
Nardone
,
G.A.
,
Robinson
,
M.A.
,
Kobayashi
,
S.D.
et al.  (
2010
)
Identification of a novel Staphylococcus aureus two-component leukotoxin using cell surface proteomics
.
PLoS ONE
5
,
e11634
28
DuMont
,
A.L.
,
Nygaard
,
T.K.
,
Watkins
,
R.L.
,
Smith
,
A.
,
Kozhaya
,
L.
,
Kreiswirth
,
B.N.
et al.  (
2011
)
Characterization of a new cytotoxin that contributes to Staphylococcus aureus pathogenesis: the cytotoxin LukAB kills human phagocytes
.
Mol. Microbiol.
79
,
814
825
29
Rouha
,
H.
,
Weber
,
S.
,
Janesch
,
P.
,
Maierhofer
,
B.
,
Gross
,
K.
,
Dolezilkova
,
I.
et al.  (
2018
)
Disarming Staphylococcus aureus from destroying human cells by simultaneously neutralizing six cytotoxins with two human monoclonal antibodies
.
Virulence
9
,
231
247
30
Badarau
,
A.
,
Trstenjak
,
N.
and
Nagy
,
E.
(
2017
) Structure and function of the two-component cytotoxins of Staphylococcus aureus — learnings for designing novel therapeutics. In
Protein Reviews
(
Atassi
,
M.Z.
, ed.), vol.
18
, pp.
15
35
,
Springer
,
Singapore
31
Thomsen
,
I.P.
,
Sapparapu
,
G.
,
James
,
D.B.A.
,
Cassat
,
J.E.
,
Nagarsheth
,
M.
,
Kose
,
N.
et al.  (
2017
)
Monoclonal antibodies against the Staphylococcus aureus bicomponent leukotoxin AB isolated following invasive human infection reveal diverse binding and modes of action
.
J. Infect. Dis.
215
,
1124
1131
32
DuMont
,
A.L.
,
Yoong
,
P.
,
Liu
,
X.
,
Day
,
C.J.
,
Chumbler
,
N.M.
,
James
,
D.B.A.
et al.  (
2014
)
Identification of a crucial residue required for Staphylococcus aureus LukAB cytotoxicity and receptor recognition
.
Infect. Immun.
82
,
1268
1276
33
Badarau
,
A.
,
Rouha
,
H.
,
Malafa
,
S.
,
Logan
,
D.T.
,
Håkansson
,
M.
,
Stulik
,
L.
et al.  (
2015
)
Structure-function analysis of heterodimer formation, oligomerization, and receptor binding of the Staphylococcus aureus bi-component toxin LukGH
.
J. Biol. Chem.
290
,
142
156
34
Rouha
,
H.
,
Badarau
,
A.
,
Visram
,
Z.C.
,
Battles
,
M.B.
,
Prinz
,
B.
,
Magyarics
,
Z.
et al.  (
2015
)
Five birds, one stone: Neutralization of α-hemolysin and 4 bi-component leukocidins of Staphylococcus aureus with a single human monoclonal antibody
.
MAbs
7
,
243
254
35
Romero-Steiner
,
S.
,
Libutti
,
D.
,
Pais
,
L.B.
,
Dykes
,
J.
,
Anderson
,
P.
,
Whitin
,
J.C.
et al.  (
1997
)
Standardization of an opsonophagocytic assay for the measurement of functional antibody activity against Streptococcus pneumoniae using differentiated HL-60 cells
.
Clin. Diagn. Lab. Immunol.
4
,
415
422
PMID:
[PubMed]
36
Siemsen
,
D.W.
,
Malachowa
,
N.
,
Schepetkin
,
I.A.
,
Whitney
,
A.R.
,
Kirpotina
,
L.N.
,
Lei
,
B.
et al.  (
2014
) Neutrophil isolation from nonhuman species. In
Neutrophil Methods and Protocols
(
Quinn
,
M.T.
and
DeLeo
,
F.R.
, eds.), pp.
19
37
,
Humana Press
,
Totowa, NJ
37
Badarau
,
A.
,
Rouha
,
H.
,
Malafa
,
S.
,
Battles
,
M.B.
,
Walker
,
L.
,
Nielson
,
N.
et al.  (
2016
)
Context matters: The importance of dimerization-induced conformation of the LukGH leukocidin of Staphylococcus aureus for the generation of neutralizing antibodies
.
MAbs
8
,
1347
1360
38
Stulik
,
L.
,
Malafa
,
S.
,
Hudcova
,
J.
,
Rouha
,
H.
,
Henics
,
B.Z.
,
Craven
,
D.E.
et al.  (
2014
)
α-Hemolysin activity of methicillin-susceptible Staphylococcus aureus predicts ventilator-associated pneumonia
.
Am. J. Respir. Crit. Care Med.
190
,
1139
1148
39
Kato
,
F.
and
Sugai
,
M.
(
2011
)
A simple method of markerless gene deletion in Staphylococcus aureus
.
J. Microbiol. Methods
87
,
76
78
40
McNamara
,
P.J.
(
2008
) Genetic manipulation of Staphylococcus aureus. In
Staphylococcus: Molecular Genetics
(
Lindsay
,
J.A.
, ed.),
1st ed.
, pp.
89
130
,
Caister Academic Press
,
Norfolk
41
Malachowa
,
N.
,
Kobayashi
,
S.D.
,
Freedman
,
B.
,
Dorward
,
D.W.
and
DeLeo
,
F.R.
(
2013
)
Staphylococcus aureus leukotoxin GH promotes formation of neutrophil extracellular traps
.
J. Immunol.
191
,
6022
6029
42
DuMont
,
A.L.
,
Yoong
,
P.
,
Surewaard
,
B.G.J.
,
Benson
,
M.A.
,
Nijland
,
R.
,
van Strijp
,
J.A.G.
et al.  (
2013
)
Staphylococcus aureus elaborates leukocidin AB to mediate escape from within human neutrophils
.
Infect. Immun.
81
,
1830
1841
43
Kobayashi
,
S.D.
,
Malachowa
,
N.
,
Whitney
,
A.R.
,
Braughton
,
K.R.
,
Gardner
,
D.J.
,
Long
,
D.
et al.  (
2011
)
Comparative analysis of USA300 virulence determinants in a rabbit model of skin and soft tissue infection
.
J. Infect. Dis.
204
,
937
941
44
Alonzo, ,
III,
F.
,
Benson
,
M.A.
,
Chen
,
J.
,
Novick
,
R.P.
,
Shopsin
,
B.
and
Torres
,
V.J.
(
2012
)
Staphylococcus aureus leucocidin ED contributes to systemic infection by targeting neutrophils and promoting bacterial growth in vivo: LukED contributes to S. aureus bacteraemia
.
Mol. Microbiol.
83
,
423
435
45
Siqueira
,
J.A.
,
Speeg-Schatz
,
C.
,
Freitas
,
F.I.S.
,
Sahel
,
J.
,
Monteil
,
H.
and
Prevost
,
G.
(
1997
)
Channel-forming leucotoxins from Staphylococcus aureus cause severe inflammatory reactions in a rabbit eye model
.
J. Med. Microbiol.
46
,
486
494
46
Gillet
,
Y.
,
Issartel
,
B.
,
Vanhems
,
P.
,
Fournet
,
J.C.
,
Lina
,
G.
,
Bes
,
M.
et al.  (
2002
)
Association between Staphylococcus aureus strains carrying gene for Panton-Valentine leukocidin and highly lethal necrotising pneumonia in young immunocompetent patients
.
Lancet
359
,
753
759
47
Lina
,
G.
,
Piémont
,
Y.
,
Godail-Gamot
,
F.
,
Bes
,
M.
,
Peter
,
M.O.
,
Gauduchon
,
V.
et al.  (
1999
)
Involvement of Panton-Valentine leukocidin-producing Staphylococcus aureus in primary skin infections and pneumonia
.
Clin. Infect. Dis.
29
,
1128
1132
48
Kuehnert
,
M.J.
,
Kruszon-Moran
,
D.
,
Hill
,
H.A.
,
McQuillan
,
G.
,
McAllister
,
S.K.
,
Fosheim
,
G.
et al.  (
2006
)
Prevalence of Staphylococcus aureus nasal colonization in the United States, 2001–2002
.
J. Infect. Dis.
193
,
172
179
49
Adhikari
,
R.P.
,
Ajao
,
A.O.
,
Aman
,
M.J.
,
Karauzum
,
H.
,
Sarwar
,
J.
,
Lydecker
,
A.D.
et al.  (
2012
)
Lower antibody levels to Staphylococcus aureus exotoxins are associated with sepsis in hospitalized adults with invasive S. aureus infections
.
J. Infect. Dis.
206
,
915
923
50
Fritz
,
S.A.
,
Tiemann
,
K.M.
,
Hogan
,
P.G.
,
Epplin
,
E.K.
,
Rodriguez
,
M.
,
Al-Zubeidi
,
D.N.
et al.  (
2013
)
A serologic correlate of protective immunity against community-onset Staphylococcus aureus infection
.
Clin. Infect. Dis.
56
,
1554
1561
51
Rasigade
,
J.P.
,
Sicot
,
N.
,
Laurent
,
F.
,
Lina
,
G.
,
Vandenesch
,
F.
and
Etienne
,
J.
(
2011
)
A history of Panton–Valentine leukocidin (PVL)-associated infection protects against death in PVL-associated pneumonia
.
Vaccine
29
,
4185
4186
52
Scherr
,
T.D.
,
Hanke
,
M.L.
,
Huang
,
O.
,
James
,
D.B.A.
,
Horswill
,
A.R.
,
Bayles
,
K.W.
et al.  (
2015
)
Staphylococcus aureus biofilms induce macrophage dysfunction through leukocidin AB and alpha-toxin
.
MBio
6
,
e01021-15
53
Takada
,
Y.
,
Ye
,
X.
and
Simon
,
S.
(
2007
)
The integrins
.
Genome Biol.
8
,
215
54
Viana
,
D.
,
Comos
,
M.
,
McAdam
,
P.R.
,
Ward
,
M.J.
,
Selva
,
L.
,
Guinane
,
C.M.
et al.  (
2015
)
A single natural nucleotide mutation alters bacterial pathogen host tropism
.
Nat. Genet.
47
,
361
366
55
Guerrero
,
I.
,
Ferrian
,
S.
,
Penadés
,
M.
,
García-Quirós
,
A.
,
Pascual
,
J.J.
,
Selva
,
L.
et al.  (
2015
)
Host responses associated with chronic staphylococcal mastitis in rabbits
.
Vet. J.
204
,
338
344
56
Viana
,
D.
,
Selva
,
L.
,
Penadés
,
M.
and
Corpa
,
J.
(
2015
)
Screening of virulence genes in Staphylococcus aureus isolates from rabbits
.
World Rabbit Sci.
23
,
185
195
57
Vrieling
,
M.
,
Boerhout
,
E.M.
,
van Wigcheren
,
G.F.
,
Koymans
,
K.J.
,
Mols-Vorstermans
,
T.G.
,
de Haas
,
C.J.C.
et al.  (
2016
)
LukMF′ is the major secreted leukocidin of bovine Staphylococcus aureus and is produced in vivo during bovine mastitis
.
Sci. Rep.
6
,
37759
58
Tsai
,
Y.-H.
,
Disson
,
O.
,
Bierne
,
H.
and
Lecuit
,
M.
(
2013
)
Murinization of internalin extends its receptor repertoire, altering Listeria monocytogenes cell tropism and host responses
.
PLoS Pathog.
9
,
e1003381

Author notes

*

Present address: Department of Biomedicine, University of Basel, Hebelstrasse 20, 4031 Basel, Switzerland.

Present address: EveliQure Biotechnologies, Helmut-Qualtinger-Gasse 2, Vienna, Austria.