Main

Staphylococcus aureus is one of the most important bacterial pathogens that has affected human health to date1. The organism colonizes approximately 30% of the human population1, but once it invades deeper tissues the clinical manifestations of S. aureus range from mild skin and soft tissue infections (SSTI) to more debilitating infections, such as sepsis, endocarditis and pneumonia1. Severe infections with S. aureus have a poor prognosis2, which is complicated by resistance to commonly used antibiotics3. As no vaccines are currently approved for S. aureus, there is substantial interest in further understanding the pathophysiology of this bacterium. Virulence factors are crucial for the success of S. aureus in the human host4, as these factors control many aspects of its commensal and pathogenic lifestyles5,6,7,8,9.

An important group of staphylococcal virulence factors are bi-component leukocidins, which are pore-forming toxins (PFTs) that kill immune cells (also known as leukocytes)7. Among leukocytes, phagocytes are required for the containment of S. aureus infection by the host9 and are considered to be the major target of leukocidins7. Leukocidins can also target natural killer cells, dendritic cells and T lymphocytes10 (Table 1), which suggests that these toxins can disrupt both innate and adaptive immune responses. In addition to their leukocidal activity, some leukocidins are able to lyse erythrocytes11 (Table 1). For historical reasons, these bi-component toxins are referred to collectively as leukocidins or leukotoxins12. Nevertheless, S. aureus secretes other toxins that are also able to target phagocytes, lymphocytes and erythrocytes, including α-toxin, β-toxin and small cytotoxic peptides known as phenol-soluble modulins (PSMs)8,13.

Table 1 Leukocidins produced by human S. aureus isolates and their respective myeloid and erythroid receptors
Full size table

Bi-component leukocidins are a collection of PFTs produced by many staphylococci. PFTs are generally secreted as inactive monomeric subunits that multimerize on binding to the membrane of a target host cell, which results in the formation of a pore that spans the phospholipid bilayer and induces cell death (Fig. 1). On the basis of the secondary structure of the membrane-spanning domains, PFTs are classified into α-helical PFTs or β-barrel PFTs14. β-Barrel PFTs are further classified into cholesterol-dependent cytolysins or haemolysins; the staphylococcal bi-component leukocidins belong to the haemolysin class of toxins. S. aureus isolates that are associated with human infections can produce up to five different leukocidins: Panton–Valentine leukocidin (PVL or LukSF–PV), γ-haemolysin AB and γ-haemolysin CB (HlgAB and HlgCB), leukocidin ED (LukED) and leukocidin AB (LukAB; also known as LukGH)7 (Table 1).

Figure 1: Pore formation by leukocidins.
figure 1

a | Crystal structures of single leukocidin protein components and multimeric β-barrel leukocidin pores show high structural similarity. In the soluble form, hydrophobic residues in the β-barrel stem of the S component and the F component are covered by the cap. The rim domain of the S component, which is responsible for the initial binding to the host target cell, is involved in receptor recognition. Hetero-oligomerization of the S component with the F component induces a conformational change that results in insertion of the hydrophobic stem into the membrane of the target cell. The resulting octameric β-barrel pore consists of alternating four S components and four F components. γ-Haemolysin A (HlgA) is shown in red and HlgB is shown in blue. Structural information was acquired from the RCSB Protein Data Bank, with accession numbers 2QK7 (unbound HlgA), 1LKF (unbound HlgB) and 3B07 (single HlgA and HlgB from HlgAB octamer). The major structural domains were coloured using PyMOL software. b | Binding and pore formation of different leukocidins to their respective receptor targets. Differences in the events between leukocidins that target chemokine receptors (Panton–Valentine leukocidin (PVL), leukocidin ED (LukED), HlgAB, HlgCB; left side) versus the leukocidin that targets CD11b (LukAB; right side) are highlighted. For PVL, LukED, HlgAB and HlgCB, the initial binding event of the S component to its specific receptor enables the secondary binding of the polymerizing F component, hetero-oligomerization and pore formation. In the rim domain of the S component (green), divergent region 1 (DR1) of LukE determines the recognition of CC-chemokine receptor 5 (CCR5), whereas DR4 of LukE determines the recognition of CXC chemokine receptor 1 (CXCR1) and CXCR2. The bottom loops in the rim domain of LukS–PV are essential for targeting C5a anaphylatoxin chemotactic receptor 1 (C5aR1). The interaction of C5aR1 and C5aR2 with LukS–PV and HlgC is multifactorial and involves the amino termini and extracellular loops of the receptors. Sulfated tyrosine residues (orange) in the N termini of the receptors C5aR1 and Duffy antigen receptor for chemokines (DARC) are essential for their interaction with PVL (for C5aR1) and HlgAB and LukED (for DARC). Uniquely, LukAB is secreted as a pre-assembled dimer. The dimerization of LukAB results in the leukocidin having a high affinity for the I-domain of its receptor CD11b. Receptor recognition by LukAB is mediated by a divergent carboxy-terminal extension of LukA (grey spike). The actual number of receptors per pore is unknown. Images in part a courtesy of B. W. Bardoel, University Medical Center Utrecht, The Netherlands.

PowerPoint slide

Full size image

The other leukocidins that are known to be produced by S. aureus are leukocidin MF′ (LukMF′)15 and leukocidin PQ (LukPQ)16; however, these toxins are associated with zoonotic infections and are rarely found in human isolates of S. aureus (Box 1). Although the first description of leukocidin activity in S. aureus culture supernatants was published around 1895, (the discoveries that lead to the identification of the leukocidins were recently reviewed elsewhere; see Ref. 7), the molecular mechanisms that cause pore formation have remained incompletely understood.

Bi-component leukocidins are thought to protect S. aureus from being killed by host phagocytes; however, the necessity of the apparently redundant range of phagocyte-targeting toxins is incompletely understood. The ability of the bi-component leukocidins to target and kill human leukocytes in vitro has been established and supported by more 100 years of research12; however, whether they target and kill human leukocytes in vivo to promote S. aureus infection remains controversial3. This controversy is a result of the poor understanding of the molecular mechanisms that underlie the differential cellular tropism and species specificity exhibited by these toxins (Table 1). Moreover, an incomplete awareness that animal leukocytes are not as susceptible to leukocidins as human leukocytes has hindered progress in understanding the contribution of the bi-component leukocidins to staphylococcal pathophysiology12,17.

During the past five years, the receptors that are targeted by staphylococcal leukocidins have been identified, which has helped to explain the cellular tropism and species specificity of these toxins. The identification of these receptors has also enabled the role of each individual leukocidin to be assessed and has improved our understanding of the complex interplay between different leukocidins during infection. In addition to their cytotoxic potential, the discovery of the leukocidin receptors has revealed new insights into how leukocidins modulate immune cell functions. Similarly, the identification of an erythroid receptor for leukocidins has also provided new insights into the link between virulence and nutritional immunity. This Review provides an overview of the similarities and differences in the interactions between leukocidins and their respective receptors, and discusses the implications of receptor identification for the mechanisms of action of leukocidins, their diverse roles during pathogenesis in vivo and their potential as targets for therapeutic interventions.

Leukocidins and receptors

Leukocidin structures. The bi-component leukocidins are PFTs that share a highly conserved structure18 (Fig. 1a). All of the subunits of the different leukocidins have an approximate molecular weight of 33 kDa. Leukocidins have two subunits that are classified as the host cell targeting S component (for slow migration in chromatography columns: LukS–PV, LukE, HlgA, HlgC and LukA), and the polymerization F component (for fast migration: LukF–PV, LukD, HlgB and LukB)7 (Table 1). In contrast to the S components and F components of other leukocidins, which are secreted as two soluble and independent monomers, LukAB is pre-assembled as a soluble heterodimer19.

Both the S component and the F component have three important domains — a cap, a rim and a stem (Fig. 1a). In the soluble form, the hydrophobic stem region is packed within the cap domain. The oligomerization of leukocidins is thought to induce a conformational change in each component that causes the extension and unfolding of the stem domain, which penetrates the plasma membrane of target cells18. During oligomerization, a ring-shaped pre-pore is formed that enables the multimeric β-strand domains of the stem regions to be inserted into the cell membrane, which results in the formation of a pore that is 1–2 nm in diameter18,20 (Fig. 1a). Structural studies of HlgAB and HlgCB have revealed that these leukocidins form an octameric pore that has alternating HlgA or HlgC and HlgB subunits18,20 (Fig. 1a). Similarly, LukAB was recently found to form a hetero-octameric pore21. On the basis of the crystal structure of the LukAB heterodimer, the sites of interaction between LukA and LukB help to explain why this toxin is pre-assembled in solution21,22. Three salt bridges between the interfaces of the cap and rim domains of LukA and LukB, which are not found in the other leukocidins, are required for the formation of LukAB dimers in solution. Hetero-octamers have been proposed as the preferred stable conformation for other leukocidins, such as PVL and LukED. Hexameric and heptameric pores have also been observed, but they have been hypothesized to represent intermediate structures during the formation of the stable pore23,24; however, this hypothesis remains to be tested experimentally.

For all of the bi-component leukocidins except LukAB, which binds to its target as a pre-assembled dimer, the initial binding of the S component to the host cell is followed by the recruitment of the F component25,26,27 (Fig. 1b). During the multimerization of the S component and the F component, a pre-pore is formed that eventually spans the entire membrane. F components have also been shown to interact with the surface of neutrophils independently of the S component28,29,30; nevertheless, the cellular target, or targets, and the importance of this binding to pore formation by the bi-component leukocidins remain to be fully elucidated. Most leukocidins can form functional pores using both cognate (for example, LukED and HlgAB) and non-cognate combinations of subunits (for example, LukE–HlgB and HlgA–LukD)31,32,33 (Table 1). This is true for PVL, LukED, HlgAB and HlgCB, but not for the pre-assembled LukAB dimer, which does not interact with any of the other leukocidin subunits owing to its structural divergence21 (Table 1). The functional consequences of cognate and non-cognate interactions between the bi-component leukocidins are discussed below.

Leukocidin receptors. The distinct cellular and species specificities of the bi-component leukocidins (Table 1) have provided historical clues for the involvement of specific proteinaceous host receptors. In 2011, three receptors were described for LukED and PVL in different laboratories. For LukED, CC-chemokine receptor 5 (CCR5) was identified as a receptor10. Concurrently, C5a anaphylatoxin chemotactic receptor 1 (C5aR1) and C5aR2 were identified as the receptors for PVL34 (Table 1). Subsequently, receptors were identified for all of the bi-component leukocidins: in addition to CCR5, LukED targets CXC chemokine receptor 1 (CXCR1) and CXCR2 (Ref. 35); HlgCB, similarly to PVL, targets C5aR1 and C5aR2 (Ref. 36); HlgAB targets CXCR1, CXCR2 and CCR2 (Ref. 36); LukAB targets CD11b (also known as integrin αM)37; and LukED and HlgAB both target the Duffy antigen receptor for chemokines (DARC; also known as ACKR1)38 (Table 1). The presence of the cognate leukocidin receptors on the surface of host target cells is required for leukocidin toxicity10,34,35,36,37,38. Moreover, the interspecies variations in the sequence and structure of leukocidin receptors are responsible for the divergent susceptibility of immune cells from different mammalian species to leukocidins34,36,37,39 (Box 2; Table 1).

The majority of the receptors that are targeted by bi-component leukocidins belong to the families of complement and chemokine receptors (Box 1), of which most are class A rhodopsin-like G protein-coupled receptors (GPCRs) — C5aR1, C5aR2, CXCR1, CXCR2, CCR2, CCR5 and DARC40. These structurally and functionally related seven-transmembrane-spanning receptors are involved in transducing extracellular signals to the interior of the cell through cytosolic G proteins. By contrast, DARC is an atypical chemokine receptor that is not coupled to a G protein41,42. With the exception of DARC, these receptors are expressed in specific leukocytes at very high levels. Key cellular functions that are regulated by G proteins include cellular activation and migration40. CD11b is more distantly related to the family of GPCRs and a component of the Mac-1 integrin. The Mac-1 integrin is highly expressed on phagocytes and is involved in many crucial cellular functions, such as phagocytosis, cell-mediated killing and chemotaxis43.

Leukocidins and receptor interactions. The identification of the leukocidin receptors has led to detailed molecular studies of leukocidin–receptor interactions. Targeting of the receptors by leukocidins is mediated by S components, which bind to their receptors with high affinity (all within a nanomolar range)10,34,35,36,37,38. The S components share 70–90% amino acid sequence identity44 (Box 2), and the alignment of leukocidin amino acid sequences has revealed divergent regions (DRs) in the rim of the S components35 (Fig. 1a). For LukED, DR1 is involved in the targeting of CCR5 by LukE45 (Fig. 1b), whereas DR4 determines the targeting of CXCR1 and CXCR2, but not of CCR5, by LukE35 (Fig. 1b). The identification of different regions in one leukocidin that are involved in the recognition of different receptors enabled the generation of mutant LukED toxins that only target CCR5 or CXCR1 and CXCR2, which subsequently enabled the relative importance of the different leukocidin–receptor interactions to be assessed in vivo35. For PVL, a cluster of amino acids in the bottom loops of the rim domain is essential for the recognition of C5aR1 by LukS–PV46 (Fig. 1b). It is likely that the amino acids that are conserved in LukS–PV and HlgC determine the specificity of these toxins for C5aR1 and C5aR2, which are their shared receptors. The molecular mechanism by which HlgAB targets different receptors is not understood. For LukAB, the LukA subunit has unique amino-terminal and carboxy-terminal extensions. Remarkably, one conserved amino acid within the C-terminal extension, a glutamic acid at position 323 in LukA, was found to be involved in the interaction between LukAB and CD11b19 (Fig. 1b). The salt bridge that mediates the dimerization of LukA–LukB in solution also seems to be essential for this leukocidin to efficiently bind to its receptor21.

The molecular determinants in the receptor for the leukocidin–receptor interaction have also been investigated. For PVL and its receptor C5aR1, and for HlgAB and LukED and their shared erythroid receptor DARC, the sulfation of tyrosine residues in the N-terminal region of the receptor seems to be essential for leukocidin–receptor interactions34,38 (Fig. 1b). Sulfated N-terminal tyrosine residues possibly define a conserved interaction site for leukocidins. By taking advantage of the differential species-specific engagement of PVL and HlgCB with their shared receptors C5aR1 and C5aR2, it was shown that different regions of the receptors are involved in binding and pore formation by these leukocidins47. The human specificity that is exhibited by PVL is determined by the second extracellular loop of C5aR1. In contrast to PVL, the specificity of HlgCB for C5aR1 is determined by the first and third extracellular loops of the receptor47. Differences in the interaction between PVL and HlgCB with C5aR1 have also been hypothesized, based on the use of specific C5aR1 antagonists in vitro. Although the toxicity of PVL can be inhibited by C5aR inhibitors, many of the inhibitors cannot neutralize the toxicity of HlgCB47. The specificity of LukAB for human CD11b is determined through its interaction with the receptor I-domain37, which is the binding site for many of the CD11b ligands (Fig. 1b). With the exception of the conserved interaction of HlgAB and LukED with the sulfated N-terminal tyrosine residues of DARC, both toxins differentially interact with this receptor38. An N-terminal receptor residue that is involved in LukED-mediated, but not HlgAB-mediated, toxicity is also involved in the binding of the natural ligand interleukin-8 (IL-8; also known as CXCL8), which supports the notion that IL-8 competes for receptor binding with LukE but not HlgA38.Competition between S components and natural ligands for receptor binding has been shown for PVL, HlgAB, HlgCB and LukED10,34,35,36,38.

Although the interactions between leukocidin subunits and their respective receptors have been investigated, the number of receptors that contribute to the formation of the leukocidin pore is unknown. Similarly, how the leukocidins transition from a receptor-bound state to form octameric pores remains to be elucidated. Nevertheless, recent studies have revealed similarities and differences in the interactions between bi-component leukocidins and their respective receptors. These differences challenge the presumed functional redundancy of this family of toxins and provide an explanation for their cellular tropism and species specificity (Box 3).

Leukocidin genomic organization, regulation and expression. S. aureus has considerable variation in its gene content between strains, both in the core genome and the accessory genome48. The hlgACB and lukAB loci are located in the core genome and are present in more than 99.5% of human isolates of S. aureus. By contrast, genes that encode other leukocidins, such as PVL and LukED, are not as widely distributed49,50. PVL is located in the temperate phage ΦSa2 (in the accessory genome)51 and is found in less than 2% of all clinical isolates; however, the majority of community-acquired methicillin-resistant S. aureus (CA-MRSA) isolates in the United States contain the genes that encode PVL52. LukED is encoded in the stable S. aureus pathogenicity island vSaβ53, which is present in about 70% of all clinical isolates. For PVL, HlgCB, LukED and LukAB, the S component and the F component are found in an operon and are co-transcribed from a single promoter. By contrast, hlgA is transcribed as a single gene that is adjacent to the hlgCB locus54.

The expression of bi-component leukocidins is complex and only partially understood. However, it is clear that S. aureus combines 'self-sensing' and 'host sensing' to regulate the expression of these toxins. Globally, the quorum sensing accessory gene regulator (Agr) system55, which is a self-sensing system, regulates the expression of leukocidin genes by controlling the production of the repressor of toxins (Rot), which is a helix–turn–helix (HTH)-type transcriptional regulator56,57. In addition, several transcriptional regulators that control the Agr–Rot system, such as SarA, have been found to indirectly regulate the expression of certain leukocidins56. Interestingly, when S. aureus comes into contact with neutrophils or with human blood the expression of leukocidins is induced58,59,60,61,62. In this context, the host-sensing SaeRS two-component regulatory system, which consists of a histidine protein kinase (SaeS) and a response regulator (SaeR), is responsible for the observed enhanced expression of leukocidins58,59,60,61,62. Interestingly, the upregulation of lukAB expression following contact with neutrophils (a response that is mediated by the SaeRS system)58,60,61 contributes to the ability of S. aureus to target and kill neutrophils63. Recent studies also suggest that S. aureus differentially regulates the expression of leukocidins by sensing metabolic shifts through a metabolite-sensing HTH-type transcriptional regulator known as RpiRc, which provides a link between metabolism and virulence64. Although our current understanding of leukocidin gene expression is incomplete, it is clear that the expression of leukocidins is regulated in response to environmental and metabolic cues.

A question that has concerned toxin researchers is whether the concentrations of toxin that are used in in vitro studies are biologically relevant during infections in vivo. In mammals, S. aureus is exposed to a range of environments and stresses that can influence gene expression in ways that are challenging to replicate in vitro. It is also unknown whether all leukocidins are expressed at the same time in vivo. Studies that have investigated the response of S. aureus following exposure to human blood or blood components have revealed the selective expression of leukocidin genes, in which the expression of HlgAB, HlgCB and LukAB is induced62. Another study found that LukAB seems to be the most upregulated leukocidin when S. aureus is exposed to human neutrophils63. These observations correlate with the contribution of these leukocidins to the survival of S. aureus (HlgAB and HlgCB)62 or growth (LukAB)63,65,66 in ex vivo models. Most studies that have tried to quantify leukocidin levels in vivo have focused on PVL. Pus samples from human infections contain up to 399 μg ml−1 PVL, a concentration that is about 1,000 fold higher than is required for PVL to kill a human neutrophil67, which supports the notion that S. aureus produces sufficient amounts of these toxins to target immune cells. Recent studies that used in vivo imaging systems65, mass spectometry68 and RNA sequencing (RNA-seq)69 have confirmed that leukocidins are produced during infection. These data are consistent with the observation that humans that are infected with S. aureus develop leukocidin-specific antibodies70,71. Thus, these toxins are likely to contribute to pathogenesis in a leukocyte- targeting-dependent manner. The various effects of leukocidins on cells at different concentrations are discussed in more detail in the following section.

Molecular mechanisms of action

Host cell death. Leukocidins can kill target cells at low concentrations (∼1 nM) in vitro, and cell susceptibility is associated with receptor levels10,34,35,37,38,47,72. The formation of pores ultimately leads to the death of target cells through cell lysis, as it results in the leakage of divalent cations that are crucial for cell homeostasis (Fig. 2a). During the past decade it has become clear that numerous PFTs use cellular pathways to enhance their ability to lyse cells. One such pathway is the inflammasome pathway73. Leukocidins, such as HlgAB, HlgCB, PVL and LukAB, are able to activate the NOD-, LRR- and pyrin domain-containing 3 (NLRP3) inflammasome in macrophages and monocytes74,75,76,77, which promotes their lytic and pro-inflammatory activities. Leukocidin-mediated inflammasome activation seems to be dependent on the presence of cognate cellular receptors, which are required for leukocidins to form pores, as was recently shown for LukAB in human monocytes76. The mechanism by which each leukocidin activates NLRP3 is not fully understood, but it is thought that toxin pores in the plasma membrane result in the leakage of potassium ions from the cytoplasm of target cells, resulting in the activation of NLRP3 (Ref. 78) (Fig. 2a). The ability of leukocidins to form pores in the plasma membrane33 provides an explanation as to how these toxins that target different receptors can all activate NLRP3. It is also possible that differences in leukocidin potency can be explained by differences in the signalling pathways that are used by the different cellular receptors (for example, GPCR versus integrin).

Figure 2: Leukocidins in pathophysiology.
figure 2

a | Cell death by leukocidins. Perturbation of cell homeostasis caused by the leakage of cations through pores and by the mobilization of ions through ion channels results in osmotic imbalance and inflammatory cell death, as nuclear factor- κB (NF-κB) stimulation and inflammasome activation lead to the release of pro-inflammatory cytokines and the assembly of endogenous amino-terminal gasdermin D (GasD) pores. b | The modulation of host cell signalling by leukocidins. Depending on the targeted receptor, single leukocidin S components in the absence of an F component can functionally antagonize G protein-coupled receptors by preventing signalling that is induced by the endogenous receptor ligand. The significance of functional antagonism by single S components during infection is unknown. At sublytic concentrations, leukocidins prime neutrophils in a receptor-specific manner, which results in the increased production of reactive oxygen species, enhanced degranulation and phagocytosis, and enhanced bactericidal activity of phagocytes. c | Leukocidins promote the growth of Staphylococcus aureus. Haemoglobin is the most abundant source of iron in mammals, and haem iron is the preferred source of iron for metabolism in S. aureus. Duffy antigen receptor for chemokines (DARC) is the erythroid receptor for γ-haemolysin AB (HlgAB) and leukocidin ED (LukED). Targeting of DARC by HlgAB and LukED results in haemolysis, which promotes the growth of S. aureus in a haemoglobin acquisition-dependent manner. d | Antagonism of leukocidins through the formation of inactive hybrid complexes as a result of sequestration of the S component from its cognate F component. Antagonism of cytotoxicity by non-cognate paring has been described for LukED and PVL. Cas1, caspase 1; IL-1β, interleukin-1β; GasD, gasdermin D; Pro-IL-1β, pro-interleukin-1β.

PowerPoint slide

Full size image

Following activation, NLRP3 forms a cytoplasmic protein complex with apoptosis-associated speck-like protein containing a CARD (ASC) and caspase 1 that is known as the NLRP3 inflammasome, which can trigger pyroptosis (a necrosis-like cell death pathway) and the increased production of pro-inflammatory cytokines by macrophages74,75,76,79. How leukocidin-mediated pores and the NLRP3 inflammasome work together to promote cell lysis is not fully understood. Recently, it has been shown that intracellular lipopolysaccharide (LPS)-induced pyroptotic cell death through caspase 4 (in mice) and caspase 4 and caspase 5 (in humans) mediates the cleavage of gasdermin D80,81. The cleavage of gasdermin D results in the release of the N-terminal membrane-targeting domain, which promotes the assembly of the protein into pores that disrupt the membranes of mammalian cells82,83. Moreover, the observation that caspase 1 can also cleave gasdermin D80,84 suggests that gasdermin D-mediated pore formation could also be involved in leukocidin-mediated pyroptotic cell death. If proven correct, this would suggest that host cells respond to damage from PFTs by producing their own intracellular pore-forming protein to cause cell death (Fig. 2a). Detailed mechanistic studies of the pathways that are involved in the cell death of neutrophils — traditionally considered to be the major target of the leukocidins — are lacking and will be required to fully understand the role of leukocidins in promoting cell death.

Modulation of host cell signalling. In addition to killing cells, leukocidins can alter cell signalling pathways at sublytic concentrations (below 1 nM) in neutrophils and macrophages75,76,79, alter the activation status of neutrophils by priming the cells34,85, and trigger the formation of neutrophil extracellular traps86 (Fig. 2b). Among the toxins, PVL and LukAB have been studied the most in the context of host cell signalling. Interestingly, PVL and LukAB can alter the cell in different ways. For example, PVL primes human neutrophils, which results in an enhanced production of superoxide in response to a secondary stimulus and enhanced phagocytic capacities85, whereas LukAB does not prime human neutrophils86. As single subunits, LukE, LukS–PV, HlgC and HlgA functionally antagonize their respective receptors in vitro10,34,36 (Fig. 2b). Owing to the clustering of S component and F component genes in a single operon, the significance of functional antagonism by a single S component during infection has not been studied. Nevertheless, studies on the non-lytic roles of leukocidins have highlighted the complexity of how each leukocidin can alter the function of their target cell. These effects are probably due to differential signalling of the targeted receptors. Future studies are required to further understand the effects that leukocidins have on leukocytes, and the contribution of these non-lytic effects to the pathophysiology of S. aureus.

More than just leukocytes. In addition to inducing cell death in phagocytes, HlgAB and LukED can lyse erthyrocytes32,38,87 (Table 1). HlgAB and LukED promote the growth of S. aureus following erythrocyte lysis38,87 (Fig. 2c), which suggests that these toxins are involved in nutrient acquisition. Indeed, the lysis of erythrocytes results in the release of haemoglobin, which is the preferred iron source for the growth of S. aureus6. Iron is an essential metal for metabolism in many bacteria and is required for the pathogenesis of S. aureus. The ferric-uptake regulator (Fur), a major regulator of iron acquisition, regulates the expression of LukED, HlgAB and HlgCB88, which further indicates that leukocidins may have a role in nutrient acquisition during infection with S. aureus38,87.

DARC is the erythroid receptor for leukocidins38, and the S. aureus-mediated lysis of human erythrocytes is DARC-dependent (Fig. 2c; Table 1). The effect of the haemolytic activity of LukED and HlgAB in vivo has been challenging to study, as mutants that diminish the haemolytic activity of leukocidins without altering the leukocidal activity of the toxins remain to be identified. However, mice that were infected with isogenic S. aureus mutants that lacked LukED and HlgAB had the same phenotype as mutants that lacked components of the iron-regulated surface determinant (Isd) haem–iron acquisition system89, which provided a link between haemolysis and the pathophysiology of S. aureus infection38.

Additive versus antagonistic activities of S. aureus leukocidins. Many clinical isolates of S. aureus encode all five bi-component leukocidins. The assembly of cognate and non-cognate leukocidins, LukE with LukD or LukE with HlgB, accordingly, could result in the assembly of 13 different toxin complexes that have a range of cytotoxic activities32,33 (Table 1). The differences in activities between cognate and non-cognate leukocidin subunit pairs could be due to differences in cellular receptor binding and in the formation of pores in cell membranes. Studies that investigated the killing of human monocytes by S. aureus in vitro revealed that isogenic strains that lack all of the leukocidins are less efficient at killing human monocytes than any single mutant strain76. Moreover, LukAB and PVL have an additive effect in the cytotoxic potential of S. aureus. As such, the leukocidin milieu at the site of infection could influence the contribution of any particular leukocidin in vivo63,66.

Leukocidins have also been shown to synergize with other virulence factors to contribute to immune subversion and pathogenesis. For example, the killing of neutrophils by PVL is enhanced by the action of PSMs8,90. Similarly, LukAB has been found to synergize with PVL and PSMs to promote the escape of S. aureus from macrophages after phagocytosis91. Very few studies have investigated the synergy between leukocidins and other virulence factors in vivo. Early studies demonstrated that α-toxin, an important PFT for the pathogenesis of S. aureus13, and HlgAB and HlgCB together caused septic arthritis more readily than strains that just produced either α-toxin or HlgAB and HlgCB alone92. Similar results were observed in a rabbit model of endophthalmitis93. More recently, a study that aimed to identify virulence factors that are involved in subverting macrophage functions demonstrated that S. aureus growing as a biofilm can evade phagocytosis owing to the combined activities of LukAB and α-toxin94. Importantly, these in vitro observations were confirmed in vivo, as the production of LukAB and α-toxin resulted in enhanced pathogenesis in a mouse model of orthopaedic implant infection94. Thus, it is clear that leukocidins are able to synergize with other S. aureus virulence factors to enhance their cytotoxic potential and their contribution to virulence.

Bi-component leukocidins have also been found to antagonize each other87. This antagonism is most evident for PVL and LukED. PVL has been shown to block the LukED-mediated lysis of erythrocytes by forming complexes with LukED at the plasma membrane of erythrocytes that impair pore formation87 (Fig. 2d). In a series of in vivo bloodstream infections in mice, it was found that S. aureus strains that produced PVL and LukED were less virulent than strains that produced LukED alone87. Interestingly, these findings could explain previous studies in which deletion of the pvl locus resulted in increased virulence in several murine87,95,96,97 and rabbit models of S. aureus infection98,99.

The virulence of S. aureus infection is remarkable, given that the organism colonizes many individuals as a commensal1. In a murine model of intranasal infection, it was shown that uncoupling the LukED antagonism by PVL promoted colonization of the lungs by S. aureus87 (Fig. 2d). Thus, in addition to contributing to acute infection, interactions between leukocidins could contribute to colonization. However, the role of bi-component toxins in colonization remains to be fully understood.

Together, these findings highlight the possibility that bi-component leukocidins exert different activities that result in different phenotypes when they are expressed at the same time, and that to fully understand the contribution of these toxins to the pathophysiology of S. aureus we need to study them as a group.

Therapeutics

Owing to the increase in antibiotic resistance and the lack of new classes of antibiotic that are available to treat bacterial infections, there is a clinical need to develop alternative antimicrobial strategies and an effective S. aureus vaccine. An important consideration in the development of new antibiotics and vaccines that target S. aureus is that bi-component leukocidins kill phagocytes that are required for the clearance of the pathogen9.

The cellular receptors that are used by leukocidins are candidate drug targets for severe S. aureus infections (Fig. 3). Blocking interactions between receptors and leukocidins by means of receptor antagonists has protected neutrophils against the action of some leukocidins in vitro10,36,38; however, the development of a combination therapy in which several receptor–leukocidin interactions are targeted could lead to adverse reactions resulting from suppressed host immune responses.

Figure 3: Leukocidins and anti-staphylococcal therapies and vaccines.
figure 3

Antagonists of the receptors that are targeted by leukocidins can protect target cells against cytotoxicity. Bispecific monoclonal antibodies (mAbs) can neutralize both S components and F components. Broadly neutralizing mAbs target several pore-forming toxins (PFTs). A leukocidin AB (LukAB)-neutralizing mAb targets the LukAB dimer in solution. Fusion mAbs combine opsonophagocytic activity with the neutralization of leukocidins and can potentially act intracellularly. Toxoids that are derived from mutant proteins induce a broadly neutralizing antibody response. GPCR, G protein-coupled receptor; HlgA, γ-haemolysin A; PV, Panton–Valentine; S. aureus, Staphylococcus aureus.

PowerPoint slide

Full size image

Several monoclonal antibodies (mAbs) that neutralize one or more toxins are currently being evaluated in clinical and preclinical trials22,100,101 and could be used to treat severe infections during the active phase or for the prevention of infections in high-risk groups (Fig. 3). An engineered bi-specific tetravalent mAb was shown to neutralize PVL and HlgCB in vitro and in an inflammation model in rabbits102. More recently, a mAb that neutralizes α-toxin, PVL, HlgAB, HlgCB and LukED was reported to inhibit leukocidin-induced cytotoxicity in vitro and to confer protection against S. aureus infection in mice103 and rabbits104. This broad-neutralizing mAb recognizes a shared conformational epitope in α-toxin and the leukocidin F components. Owing to leukocidin structural dissimilarities, this mAb does not neutralize LukAB103. However, the identification of a series of mAbs that specifically neutralize LukAB was recently described22; this included naturally occurring mAbs that were isolated from children who were infected with S. aureus105. The role of LukAB in the intracellular compartment, in which it promotes bacterial escape following phagocytosis37,63, is likely to pose challenges for the neutralization of this toxin. Nevertheless, this obstacle could be overcome by combining opsonization and leukocidin neutralization using a fusion antibody platform101 (Fig. 3).

Many individuals have large quantities of antibodies that target bi-component leukocidins70,106. Studies that have investigated protection by pre-existing leukocidin-specific antibodies have reported conflicting results97,106,107,108,109, which has made it difficult to evaluate the potential of a vaccine that solely targets leukocidins. The complex pathophysiology of S. aureus, together with the absence of a single dominant virulence factor, necessitates a multivalent approach, and targeting bi-component leukocidins could improve the capacity of phagocytes to clear infection (Fig. 3). It remains to be determined whether a vaccine that induces broad immunity against all staphylococcal β-barrel PFTs is technically feasible. Recently, a vaccine candidate that was based on a mutant LukS–PV toxoid resulted in a cross-neutralizing immune response against all human leukocidins, except LukAB, in mice71,110. A successful multivalent vaccine will probably need to be complemented with additional targets, such as other immune modulators or cell surface proteins4,100,111. We hypothesize that previous active and passive immunization strategies against S. aureus have been unsuccessful in clinical trials owing to the exclusive focus on opsonophagocytosis and the reliance on suboptimal animal models of infection101,111,112. Currently, there are no reliable animal models to study the pathophysiology of S. aureus48,111 (Box 3).

Outlook

The identification of host receptors that determine the cellular tropism and species specificity of bi-component leukocidins has advanced our understanding of how these toxins target and kill host cells. Additional advances include uncovering the consequences of receptor binding on host cell function. The notion that leukocidins have multiple cellular effects in a concentration-dependent manner (lytic versus sublytic effects), suggests that the level of toxins produced during infection with S. aureus could have broad effects on the host immune response. Moreover, the discovery that leukocidins can target signalling receptors suggests that these toxins can act as ligands, or even as antagonists, during infection10,34,35,36. It is possible that an uncharacterized role of the leukocidins is to block the function of their targeted receptors, thus preventing the migration of leukocytes.

Another important consideration is that the activities and effects of leukocidins could vary in different target cells. A careful examination of how bi-component leukocidins target and kill different human phagocytes is required, particularly the pathways that are involved in the death of different phagocytes. The expression profiles and the tissue distribution of the known leukocidin receptors support the view that the targets of these toxins are leukocytes. However, the receptors are also expressed in non-myeloid cells, which suggests that leukocidins have the potential to contribute to the pathogenesis of S. aureus in a manner that does not involve targeting leukocytes.

S. aureus is a major pathogen of clinical significance, and leukocidins are a prime example that S. aureus has evolved as a human-specific pathogen. New in vivo models are required that better mimic the human host (Box 3). The use of humanized mice has been informative113,114; however, more work is needed to fully recapitulate the function of human neutrophils in mice. The identification of human specific receptors and recent advances in genome engineering with CRISPR–Cas115 suggest that developing a mouse model that is compatible with all leukocidins could be possible. Such a model would overcome the current lengthy and variable process of murine humanization, and could enable the study of the role of leukocidins as a group to be assessed in S. aureus infection. These advances are likely to have major implications for the development of effective vaccines and novel therapeutics to prevent and treat infection with S. aureus.

A question that has been challenging to address is why does S. aureus produce so many different toxins that target leukocytes? These toxins are differentially regulated in vitro and in vivo63,116, which supports the hypothesis that S. aureus senses host cues to coordinate the production of leukocidins in a tissue-dependent manner. Moreover, the understanding that leukocidins are able to synergize and antagonize the activities of other leukocidins, stresses the need for further studies that consider that the contribution of leukocidins to infection could change depending on the infection site, the clinical isolate of S. aureus, and the host.

From the perspective of the host, the genes that encode two leukocidin receptors, CCR5 and DARC, are under selective pressure. Cells of individuals that lack these receptors are resistant to leukocidins that target CCR5 and DARC10,38. Moreover, numerous polymorphisms have been found in genes that encode other leukocidin receptors117. Future work that investigates the effect of these polymorphisms on host susceptibility to leukocidins could provide insight into observed differences in the susceptibility of humans to infection with S. aureus.

With recent advances in our understanding of how S. aureus bi-component leukocidins target host cells, we are now able to address how these toxins contribute to the pathophysiology of S. aureus, which could lead to the development of novel anti-staphylococcal agents.