Staphylococcus aureus leucocidin ED contributes to systemic infection by targeting neutrophils and promoting bacterial growth in vivo


E-mail; Tel. (+1) 212 263 9232; Fax (+1) 212 263 9180.


Bloodstream infection with Staphylococcus aureus is common and can be fatal. However, virulence factors that contribute to lethality in S. aureus bloodstream infection are poorly defined. We discovered that LukED, a commonly overlooked leucotoxin, is critical for S. aureus bloodstream infection in mice. We also determined that LukED promotes S. aureus replication in vivo by directly killing phagocytes recruited to sites of haematogenously seeded tissue. Furthermore, we established that murine neutrophils are the primary target of LukED, as the greater virulence of wild-type S. aureus compared with a lukED mutant was abrogated by depleting neutrophils. The in vivo toxicity of LukED towards murine phagocytes is unique among S. aureus leucotoxins, implying its crucial role in pathogenesis. Moreover, the tropism of LukED for murine phagocytes highlights the utility of murine models to study LukED pathobiology, including development and testing of strategies to inhibit toxin activity and control bacterial infection.


Staphylococcus aureus bacteraemia, coupled with subsequent dissemination to and damage of distant tissue sites, is responsible for significant morbidity and mortality in the USA and throughout the world (Klevens et al., 2006; 2007). A hallmark of highly virulent S. aureus strains is their ability to effectively kill neutrophils (Voyich et al., 2005), key innate immune effector cells absolutely required for control of infection (Bodey et al., 1966; Pincus et al., 1976; Dale et al., 1979; Lekstrom-Himes and Gallin, 2000). S. aureus uses multifaceted and often redundant mechanisms to ensure broad protection against attack by host neutrophils (Foster, 2005; Nizet, 2007).

One major mechanism by which S. aureus targets and kills neutrophils in vitro is through the production of bi-component pore-forming leucotoxins (Menestrina et al., 2003). Strains associated with human infections can produce up to four different bi-component leucotoxins: γ-haemolysin (HlgACB), LukSF-PVL, LukAB/HG and LukED. Among these toxins, HlgACB is believed to play a role in septic arthritis and weight loss upon systemic infection (Nilsson et al., 1999), contribute in part to the inflammatory response observed in the rabbit eye in vivo (Supersac et al., 1998), as well as contribute modestly to community-acquired methicillin-resistant Saureus (CA-MRSA) survival in human blood and virulence upon systemic infection of mice (Malachowa et al., 2011). Studies of the contribution of PVL to S. aureus pathogenesis on the other hand have led to conflicting conclusions due in part to the toxin's species specificity, but PVL is believed to contribute to pneumonia (Voyich et al., 2006; Labandeira-Rey et al., 2007; Diep et al., 2010; Loffler et al., 2010). Recently, LukAB/HG, a new member of the S. aureus leucotoxin family, was shown to contribute to neutrophil killing; promote survival of S. aureus in human whole blood; restrict neutrophil-mediated killing; and promote CA-MRSA pathogenesis (Ventura et al., 2010; Dumont et al., 2011). Among the leucotoxins, LukED is the least characterized. LukED exhibits toxicity toward PMNs in vitro and induces dermonecrosis when purified toxin is injected into rabbits (Gravet et al., 1998; Morinaga et al., 2003). Despite all the effort devoted to the study of S. aureus leucotoxins, the direct mechanism of action of these toxins during the course of infection has not been defined.

In this work, we identify LukED as a major virulence factor involved in bloodstream infection with S. aureus. Our studies demonstrate for the first time that LukED plays a critical role in S. aureus lethality for mice. In stark contrast to other staphylococcal leucotoxins, we observed that LukED effectively targets and kills murine phagocytes, including neutrophils ex vivo. Investigation into the in vivo mechanism of action of LukED demonstrated that the toxin promotes disease progression via its potent cytotoxic effects on phagocytes recruited to haematogenously seeded infection sites. These results underscore the potential role of LukED as a critical virulence factor required for bloodstream infection with S. aureus, including highly pathogenic MRSA.


Using global regulators to dissect the contribution of secreted factors to S. aureus systemic infection

In an effort to identify individual virulence factors involved in bacteraemia, we first investigated the contribution of the accessory gene regulatory (Agr) system to the lethality observed upon S. aureus systemic infection. The Agr system regulates the differential expression of S. aureus secreted and surface proteins in a quorum dependent manner (Novick and Geisinger, 2008). Mice were infected systemically with S. aureus Newman, a highly virulent clinical methicillin sensitive Saureus (MSSA) strain (Duthie and Lorenz, 1952), as well as an isogenic Δagr mutant lacking the entire agr locus. We observed that the Δagr mutant was significantly attenuated for virulence compared with animals infected with wild type, as the animals did not succumb to infection (Fig. 1A).

Figure 1.

The regulatory input of Agr upon Rot directly influences bloodstream infection with S. aureus. A and B. ‘Survival’ of mice infected systemically with Newman wild type (WT), n = 12 (A) and n = 17 (B); Δagr, n = 6; ΔagrΔrot, n = 6; or Δrot, n = 12. Mice were injected with ∼1 × 107 cfu of the indicated strains via the retro-orbital venous plexus and monitored for ‘survival’ over time as described in Experimental procedures. C and D. Immunoblots of leucotoxins produced by the indicated strains. (−) Control indicates proteins isolated from gene deletion mutants of the particular leucotoxin probed in each panel. For LukE immunoblots an ΔhlgΔlukED or an ΔrotΔhlgΔlukED mutant was included due to cross-reactivity of the anti-LukE antibody with subunits of γ-haemolysin. Statistically significant differences between curves were determined by Log-Rank (Mantel Cox) test and P-values are shown (***P ≤ 0.0005).

The regulation of a number of important virulence factors (including S. aureus cytotoxins) by the Agr system is mediated in an RNAIII-dependent manner. RNAIII is a regulatory RNA molecule, expressed upon Agr activation, which influences translation of target mRNAs (Novick et al., 1993; Novick and Geisinger, 2008). A major mechanism by which RNAIII modulates virulence factor expression is via its regulatory control over the transcription factor Rot (Repressor of toxins) (McNamara et al., 2000; Said-Salim et al., 2003; Geisinger et al., 2006; Boisset et al., 2007). RNAIII blocks translation of Rot by binding to rot mRNA (Geisinger et al., 2006; Boisset et al., 2007). The inhibitory binding of RNAIII to rot mRNA facilitates optimal expression of otherwise Rot-repressed cytotoxins (Said-Salim et al., 2003). To test whether the virulence defect of the Δagr mutant is dependent on Rot-regulated factors, we infected mice with an ΔagrΔrot double mutant strain. We observed that the deletion of rot in the Δagr strain fully restored virulence in mice (Fig. 1A), consistent with a previous report using a rabbit endocarditis model (McNamara and Bayer, 2005). To directly probe the contribution of Rot to S. aureus pathogenesis, a Δrot mutant with a normal functioning agr locus was tested in the systemic infection model. The Newman Δrot mutant strain was found to be hypervirulent compared with wild type (Fig. 1B). Taken together these results suggest that the regulatory input of Agr upon Rot directly influences bloodstream infection with S. aureus.

LukED is critical for the virulence of a Δrot mutant

Virulence factors responsible for death due to S. aureus bloodstream infection are poorly defined. The enhanced virulence of the Δrot mutant suggests that a Rot-repressed factor contributes to lethality in S. aureus bloodstream infection. A major group of Rot-repressed factors are cytotoxin-encoding genes (McNamara et al., 2000; Said-Salim et al., 2003). To gain insight into the cytotoxin(s) potentially responsible for the enhanced virulence of Δrot and the ΔagrΔrot double mutant strains, we monitored cytotoxin abundance in culture supernatants via immunoblotting. We observed that toxin levels were markedly reduced in the Δagr mutant compared with the wild-type strain (Fig. 1C), a phenotype rescued by deleting rot in the Δagr strain (Fig. 1C). In contrast to other toxins, the LukE subunit of the bi-component leucotoxin LukED was strikingly overproduced by the ΔagrΔrot mutant (Fig. 1C). Similarly, we observed that the Δrot strain produced increased amounts of LukE, while no major difference was observed for the other cytotoxins (Fig. 1D).

The increased production of LukE in the absence of Rot and the associated hypervirulence of a Δrot mutant led us to hypothesize that LukED was involved in the increased virulence of the Δrot strain. To directly test this hypothesis we constructed double mutants lacking both rot and each of the four major leucotoxin genes/operons present in strain Newman (hla, hlgACB, lukAB/HG and lukED) (Dumont et al., 2011), and challenged mice systemically with each strain. We observed that deletion of lukAB/HG and hlgACB caused modest but statistically significant reductions in the Δrot hypervirulent phenotype (Fig. 2A–C) consistent with previously published roles for these toxins in the pathogenesis of S. aureus (Nilsson et al., 1999; Dumont et al., 2011; Malachowa et al., 2011). In stark contrast, a ΔrotΔlukED double mutant was markedly reduced for virulence (Fig. 2D), suggesting that lukED plays a critical role in the hypervirulence exhibited by the Newman Δrot mutant.

Figure 2.

Identification of virulence factors critical to S. aureus bacteraemia-mediated death. ‘Survival’ of mice infected with Newman wild type (WT), Δrot, and rot/leucotoxin double mutants, all n = 6. Mice were injected with ∼1 × 107 cfu of the indicated strains via the retro-orbital venous plexus and monitored for ‘survival’ over time as described in Experimental procedures. Statistically significant differences between curves were determined by Log-Rank (Mantel Cox) test and P-values are shown (*P < 0.05; **P ≤ 0.005).

LukED promotes disease progression during systemic infection

To evaluate whether LukED directly contributes to lethality in S. aureus bloodstream infection, we constructed a ΔlukED mutant in strain Newman, as well as a complementation strain wherein lukED and its native promoter sequence were ectopically integrated into the chromosome (Fig. 3A). To verify that the ΔlukED mutant was altered only in LukED production, the toxin profile was analysed by immunoblotting. We observed that only LukE production was altered by the ΔlukED mutant, a phenotype fully complemented in the ΔlukED::lukED strain (Fig. 3B). The strains were then used to challenge mice systemically as described above. These experiments revealed that the ΔlukED mutant was markedly attenuated for virulence compared with wild type, a phenotype completely restored in the complement strain (Fig. 3C).

Figure 3.

LukED contributes to lethality in S. aureus bloodstream infection. A and B. PCR and immunoblot confirmation of the Newman ΔlukED mutant and its complement ΔlukED::lukED. For LukE immunoblots, an ΔhlgΔlukED mutant was included due to cross-reactivity of the anti-LukE antibody with subunits of γ-haemolysin. C and D. ‘Survival’ curves of MSSA (strain Newman) and MRSA (USA500 strains) infected mice. Mice were injected with ∼1 × 107 cfu of Newman wild type (WT), n = 6; ΔlukED, n = 6; ΔlukED::lukED, n = 6; USA500 (BK2371 or BK2395) wild type (WT), n = 10; or an isogenic ΔlukED mutant from each USA500 strain, n = 10, via the retro-orbital venous plexus and monitored for ‘survival’ over time as described in Experimental procedures. Statistically significant differences between ‘survival’ curves were determined by Log-Rank (Mantel Cox) test and P-values are shown (**P ≤ 0.005; ***P ≤ 0.0005).

LukED is critical for the pathogenesis of USA500 MRSA strains

To evaluate the contribution of LukED to the virulence of modern strains, we first determined whether the major clones of MRSA currently causing infections in the USA, pulse field electrophoresis types USA100, USA200, USA300, USA400 and USA500 (Klevens et al., 2007), contained the lukE/D genes. All strain types, excluding USA200, contained the lukE/D genes (Table 1) (Diep et al., 2006). Additionally, all lukED-containing strains were capable of expressing lukE mRNA as determined by qRT-PCR (Fig. S1A). Among the lukED-positive strains, USA300, USA400 and USA500 are considered to be the most virulent in animal models (Li et al., 2009; 2010). We observed that S. aureus USA500 and Newman were far more virulent in mice systemically infected with 1 × 107 colony-forming units (cfu) compared with USA300 and USA400 strains (Fig. S1B). USA500 strains are associated with both hospital and community acquired infections (Diep et al., 2006; Klevens et al., 2007; Li et al., 2009). However, virulence factors involved in the pathogenesis of USA500 are poorly defined. To evaluate the contribution of lukED to USA500 infection, we constructed ΔlukED mutants in two independent USA500 clinical isolates (i.e. BK2371 and BK2395) and subsequently tested their virulence potential. Deletion of lukED markedly attenuated the virulence potential of both strains (Fig. 3D), suggesting that LukED is a major determinant of USA500 virulence.

Table 1. S. aureus clinical isolates used in this study.
StrainYearSpecimen sourcePFGE typespa-typespa motifCCaagr grouplukEDCitation
  • a. 

    spa-type deduced clonal complex (CC). Isolates broadly grouped according to their relatedness by PFGE.

  • b. 

    Other designations: NRS384, NCBI reference sequence NC_007793.

  • c. 

    Other designation: NRS123.

  • −, absent; +, present; LAC, Los Angeles County clone; PFGE, pulsed-field gel electrophoresis.

BK23821996WoundUSA1002T1-J1-M1-B1-M1-D1-M1-G1-M1-K152+Roberts et al. (1998)
BK24051996Tracheal aspirateUSA1002T1-J1-M1-B1-M1-D1-M1-G1-M1-K152+Roberts et al. (1998)
BK25161996WoundUSA20016W1-G1-K1-A1-K1-A1-O1-M1-Q1-Q1-Q1303Roberts et al. (1998)
BK25321996SputumUSA20016W1-G1-K1-A1-K1-A1-O1-M1-Q1-Q1-Q1303Klevens et al. (2007)
LACb2002WoundUSA3001Y1-H1-G1-F1-M1-B1-Q1-B1-L1-O181+Kennedy et al. (2008)
NRS6472005BloodUSA3001Y1-H1-G1-F1-M1-B1-Q1-B1-L1-O181+Klevens et al. (2007)
BK188102005PneumoniaUSA3001Y1-H1-G1-F1-M1-B1-Q1-B1-L1-O181+Kennedy et al. (2008)
NRS1931999Pleural fluidUSA400194U1-J1-F1-K1-K1-P1-F1-K1-P1-E113+CDC (2003)
MW2c1998BloodUSA400131U1-J1-J1-J1-J1-F1-E113+CDC (2003)
BK23711996WoundUSA5007Y1-H1-G1-C1-M1-B1-Q1-B1-L1-O181+Roberts et al. (1998)
BK23951996WoundUSA5007Y1-H1-G1-C1-M1-B1-Q1-B1-L1-O181+Roberts et al. (1998)
Newman1952OsteomyelitisN/A1Y1-H1-G1-F1-M1-B1-Q1-B1-L1-O181+Duthie and Lorenz (1952)

LukED facilitates bacterial replication in vivo

To determine the specific contribution of LukED to haematogenous infection, we monitored colonization, bacterial replication and abscess formation in the kidneys of animals infected systemically with S. aureus. Compared with Newman wild type or the ΔlukED::lukED complemented strain, a ΔlukED mutant exhibited significantly reduced abscess formation after 96 h (Fig. 4A). We reasoned that a reduction in abscess formation could indicate either: (i) an inability of the ΔlukED mutant to seed the kidney of infected animals, or (ii) an inability of the ΔlukED mutant to replicate in seeded kidneys due to better control of infection by immune cells. To test both possibilities, we monitored bacterial burden to the kidney at 16 and 96 h (Fig. 4B). Total cfu in the kidneys early after infection were identical for all strains. In contrast, at 96 h a ΔlukED mutant exhibited a 15-fold decrease in bacterial burden compared with both wild type and the complemented strain. Consistent with the observed differences in bacterial burden, animals infected with the ΔlukED mutant also exhibited reduced markers of inflammation (IL-6 and GCSF) in the serum at 96 h (Fig. 4C). Collectively, these findings suggest that LukED contributes to the virulence of S. aureus by promoting bacterial proliferation within haematogenously seeded tissue.

Figure 4.

LukED facilitates bacterial replication in vivo. A. Gross pathology of abscess formation in kidneys of mice infected for 96 h with Newman wild type (WT), ΔlukED, and ΔlukED::lukED. White arrows point to abscesses in wild type and ΔlukED::lukED infected organs. B. Enumeration of cfu from infected kidneys. Ten mice per group were injected with ∼1 × 107 cfu of Newman wild type (WT), ΔlukED, and ΔlukED::lukED. After 96 h, kidneys were removed and homogenized, followed by plating of serial dilutions onto solid media for enumeration of bacterial burden. C. IL-6 and GCSF levels in the serum of animals from Panel (B). Statistical significance between cfu and serum cytokine levels was determined by 1-way anova with Tukey's multiple comparison test (**P ≤ 0.005; ***P ≤ 0.0005).

LukED targets and kills neutrophils by damaging their plasma membrane

One mechanism by which LukED could promote S. aureus virulence is through the killing of neutrophils. To determine whether LukED is cytotoxic towards primary murine neutrophils, we isolated peritoneal elicited cells (PECs) from animals infected with S. aureus. Infection of the peritoneum induced a robust infiltration of neutrophils (CD11b+/Ly6G+, ∼66%) (Fig. 5). Isolated PECs were intoxicated with purified recombinant LukE, LukD, or an equimolar mixture of LukE and LukD (LukED). Intoxication with high doses of any single toxin subunit (10 µg ml−1) exhibited negligible cytotoxic effects towards PECs (Fig. 5). In contrast, intoxication with both subunits significantly reduced the number of viable PECs from ∼50% to ∼15% (Fig. 5A). Within the PEC population neutrophils were specifically targeted, as over 85% of Ly6G+/CD11b+ cells were killed (Fig. 5B).

Figure 5.

LukED kills phagocytes elicited to the site of infection ex vivo. A. Profile of innate immune cells recruited to the peritoneum upon infection with S. aureus and subsequently intoxicated with the indicated amounts of LukE, LukD or LukED. Representative FACS plots of cells in the live gate are shown. The graphical depiction represents the average percent viable PECs within the live gate. B. Killing of primary murine neutrophils (CD11b+/Ly6G+) by LukED. Representative FACs plots are shown. The graphical depiction displays percent viable neutrophils relative to the total viable neutrophils recovered from unintoxicated controls (set to 100%). Statistical significance was determined by 1-way anova with Tukey's multiple comparison test (***P ≤ 0.0005).

LukED intoxicated PECs, but not PECs intoxicated with single subunits, exhibited characteristic morphological alterations associated with membrane permeabilization and cell death (nuclei swelling, cell rounding and membrane halos) (Fig. 6A). Other indicators of rapid cell death included a dose-dependent decrease in metabolic activity as measured via CellTiter (Fig. 6B), and overt membrane destabilization as determined via lactate dehydrogenase release into culture medium within 1 h of intoxication (Fig. 6C). Additionally, ethidium bromide uptake, an assay typically used as an indicator of pore formation (Finck-Barbancon et al., 1993), was observed as early as 15 min post intoxication and continued to increase throughout the first hour of intoxication (Fig. 6D). Similar results were also observed for phagocytes isolated from the peritoneum of mice after thioglycollate treatment, from the bone marrow of naïve mice, and from whole blood-derived primary human neutrophils (Data not shown and Fig. S2). Together, these results demonstrate that LukED is toxic to murine neutrophils due to membrane damage that leads to rapid cell death.

Figure 6.

LukED targets and kills neutrophils by damaging their plasma membrane. Intoxicated PECs were evaluated for viability, membrane damage and pore formation via (A) light microscopy at 40 × magnification (B) metabolic activity (CellTiter) (C) LDH release (CytotoxOne), and (D) ethidium bromide incorporation into cellular DNA. For microscopic imaging of intoxicated cells (A and D) a toxin dose of 5 µg ml−1 was used. For all other assays of viability, membrane integrity and pore formation a dose–response of equimolar ratios of LukE and LukD was added to cells and measurements made as described in the Experimental procedures.

LukED targets and kills phagocytes in vivo

Although staphylococcal bi-component leucotoxins are known for their ability to kill immune cells in vitro, the mechanism by which these toxins contribute to S. aureus pathogenesis in vivo is poorly defined. We embarked on experiments to examine whether LukED promotes pathogenesis in vivo by killing phagocytes recruited to haematogenously seeded infection sites (in this case, murine kidneys). Mice were infected with Newman wild type, ΔlukED or ΔlukED::lukED strains and after 96 h, kidneys were removed, and single cell suspensions prepared for flow cytometric analysis. Cells were stained with a fixable viability dye (PacBlue), and α-CD11b antibody (to detect phagocytes). Approximately 40% of total cells (including kidney parenchymal cells as well as infiltrating immune cells) stained PacBlue+ regardless of whether they were infected with WT, ΔlukED or ΔlukED::lukED indicating similar sample processing for all organs (Fig. S3). Further analyses of CD11b+/PacBlue+ cells (total non-viable phagocytes) revealed a significant reduction in overall cell viability (∼90% PacBlue+) for both wild type and ΔlukED::lukED infected animals (Fig. 7A). In stark contrast, mice infected with the ΔlukED mutant exhibited greater proportions of viable phagocytes in infected kidneys (only ∼50% PacBlue+) (Fig. 7A). These results suggest that LukED directly impacts the viability of phagocytic cells at the site of tissue infection.

Figure 7.

LukED promotes S. aureus virulence in vivo via its potent toxicity towards neutrophils. A. Determination of percent viable CD11b+ phagocytic leucocytes in S. aureus infected kidneys. B. Immune cell profile (neutrophils; GR1+/CD11b+, B lymphocytes CD3-/B220+, and T lymphocytes CD3+/B220-) of mice treated with a neutrophil depleting antibody (1A8) or an isotype control antibody (2A3) followed by systemic infection with wild-type S. aureus. C. ‘Survival’ curves of mice treated with 2A3, 1A8, or untreated followed by systemic infection of ∼1 × 108 cfu wild type (WT), n = 10 for 2A3 or 1A8 and n = 6 for untreated; or a ΔlukED mutant, n = 10 for 2A3 or 1A8 and n = 6 for untreated. Statistically significant differences between ‘survival’ curves were determined by Log-Rank (Mantel Cox) test and P-values are shown (*P < 0.05; **P ≤ 0.005).

LukED promotes S. aureus virulence in vivo by killing phagocytes

If the primary contribution of LukED to S. aureus pathogenesis is neutrophil killing, depletion of neutrophils before infection should result in comparable virulence characteristics between wild type and the ΔlukED mutant strain. To test this hypothesis, we specifically depleted neutrophils using the 1A8 anti-Ly6G antibody (Daley et al., 2008). The 1A8 antibody was efficient at depleting neutrophils (Gr-1+/CD11b+), while the 2A3 isotype control antibody was not (Fig. 7B) (Daley et al., 2008; Blomgran and Ernst, 2011). In contrast, the antibodies have no effect on lymphocytes (B cells CD3-/B220+; T cells B220-/CD3+) (Fig. 7B). Following antibody administration, mice were infected systemically with ∼1 × 108 cfu wild type or the ΔlukED mutant. Animals treated with 2A3 (isotype control antibody) exhibited survival patterns similar to those already described, confirming that lukED is critical for the full virulence of S. aureus. In contrast, when mice were depleted of neutrophils, the virulence of the wild type and the ΔlukED mutant were indistinguishable (Fig. 7C). In this experiment, we observed that administration of 2A3 control antibody resulted in slower kinetics of animal death after infection with wild type compared with neutrophil-depleted animals, an effect presumably due to subtle influences of the control antibody on the murine immune response. It thus remained possible that an infectious dose of S. aureus resulting in 100% lethality within 30 h might also lack a distinguishable phenotype between wild type and a ΔlukED mutant irrespective of the presence/absence of neutrophils. To rule out this possibility we also infected untreated animals with 1 × 108 cfu of S. aureus Newman, and ΔlukED, and measured survival over time (Fig. 7C). The majority of animals (5 out of 6) infected with wild type rapidly succumbed to infection within 36 h, while those infected with a ΔlukED mutant remained markedly attenuated. Collectively, these results demonstrate that LukED targets neutrophils in vivo to promote S. aureus virulence.


To cause severe disease S. aureus must efficiently avoid rapid killing by host neutrophils, which mediate the initial response to infection. The mechanism(s) by which the bacterium averts neutrophil killing is multifaceted and incompletely understood, but is believed to rely heavily upon secreted proteins that can inhibit the function of and/or kill these critical immune cells (Wang et al., 2007; Dumont et al., 2011). In this study, we conclusively demonstrate that LukED contributes to the pathophysiology of S. aureus by killing neutrophils in vivo facilitating bacterial growth at the site of infection. Thus, our findings extend the complex and integrated role of toxins in S. aureus immune cell killing and highlight LukED as a critical virulence factor involved in the lethality observed in S. aureus bacteraemia.

Why LukED had not been previously implicated as a major virulence factor in S. aureus is not certain, although we speculate it may stem from the redundant cytotoxic activities of toxins present in S. aureus culture supernatant toward human phagocytes (Wang et al., 2007; Ventura et al., 2010; Dumont et al., 2011; Malachowa et al., 2011). In addition, the expression and production of leucotoxins in S. aureus is heavily influenced by growth medium and growth conditions, which in turn, modulate the cytotoxicity of S. aureus culture supernatants (Malachowa et al., 2011). Such findings suggest caution in the interpretation of in vitro studies using culture supernatants. An advantage of the study design implemented in this work is its minimal reliance on ex vivo and in vitro phenotypic analyses to infer in vivo functionality. LukED is thus far the only Staphylococcal leucotoxin found to exhibit potency towards murine phagocytes ex vivo and in vivo. Thus, the utility of LukED-based studies using mouse models will certainly prove an advantageous means by which to further elucidate the true functional role of bi-component leucotoxins during host infection.

Both LukE and LukD are 100% conserved at the amino acid level in sequenced S. aureus strains (Fig. S4), suggesting that the major biological function of the toxin is similar to that described in this study. Previous reports indicate the existence of a variant LukED toxin (LukEDv) (Morinaga et al., 2003). Upon closer examination we have confirmed that the ‘variant’ sequence is conserved in nearly all sequenced strains, including S. aureus Newman (Fig. S4). Contrary to the originally described lukE/D sequences in strain Newman (Gravet et al., 1998), the sequences of lukE/D in the Newman genome sequence (Accession #: NC_009641) are 100% identical to lukEDv (Fig. S4) (Morinaga et al., 2003; Baba et al., 2008). We thus propose that LukED and LukEDv are in essence one and the same.

The lukE/D genes are present in ∼87% of tested strains, including MSSA and MRSA (Gravet et al., 1998; 1999; 2001; Morinaga et al., 2003; Diep et al., 2006), underscoring its potential pivotal role in pathogenesis. Consistent with this observation, antibodies directed against LukED have been found in patients suffering from diverse S. aureus infections (Verkaik et al., 2010), suggesting that LukED is produced during the course of human infection. Additionally, epidemiological evidence links lukED to S. aureus associated impetigo and diarrhoea (Gravet et al., 1999; 2001). It remains to be determined whether MRSA strains other than USA500 (e.g. USA100, USA300 and USA400) rely as heavily upon LukED for systemic infection. However, the observation that deletion of lukED significantly attenuates highly virulent strains of S. aureus supports the premise that selectively inhibiting LukED may prove valuable in the development of novel treatment strategies to combat S. aureus systemic infection.

Experimental procedures

Bacterial strains and culture conditions

Staphylococcus aureus strains used in this work are described in Tables 1 and 2. Cultures were grown in either tryptic soy broth, or RPMI supplemented with 1% casamino acids as describe previously (Torres et al., 2010; Dumont et al., 2011). Overnight cultures were routinely incubated at 37°C with shaking (180 r.p.m.) and subcultured 1:100 for 3–5 h under these same conditions. Due to the lack of antibiotic selection strategies for the USA500 strains we were unable to utilize available tools to complement ΔlukED mutants. Thus, construction of two independent mutants was used to validate the phenotype of USA500 (Fig. 3D).

Table 2. S. aureus strains used in this study.
VJT21.44E. coli LysY LaqQEscherichia coli LysY LaqQ + pET14b-6X-His-LukD This study
VJT20.78E. coli LysYLaqQEscherichia coli LysY LaqQ + pET14b-6X-His-LukE This study
VJT3.81S. aureus NewmanParental StrainWT (Newman)Duthie and Lorenz (1952)
VJT7.17S. aureus NewmanS. aureus Newman Δagr:tetΔagrBenson et al. (2011)
VJT9.98S. aureus NewmanS. aureus Newman Δrot::Tn917ΔrotBenson et al. (2011)
VJT10.03S. aureus NewmanS. aureus Newman Δagr::tetΔrot::Tn917ΔagrΔrotBenson et al. (2011)
VJT8.16S. aureus NewmanS. aureus Newman ΔlukEDΔlukEDDumont et al. (2011)
VJT8.91S. aureus NewmanS. aureus Newman ΔlukABΔlukABDumont et al. (2011)
VJT7.12S. aureus 8325-4S. aureus 8325-4 hla::erm hlb::φ42E hlg::tetΔhlaΔhlbΔhlgNilsson et al. (1999)
VJT7.09S. aureus NewmanS. aureus Newman ΔhlgACB::tetΔhlgACBSupersac et al. (1998)
VJT23.61S. aureus NewmanS. aureus Newman ΔlukED::pJC1.112lukEDΔlukED::plukEDThis study
VJT15.93S. aureus NewmanS. aureus Newman Δrot::Tn917ΔlukABΔrotΔlukABThis study
VJT18.96S. aureus NewmanS. aureus Newman Δrot::specΔhlg::tetΔrotΔhlgThis study
VJT21.37S. aureus NewmanS. aureus Newman Δrot::specΔhla:ermΔrotΔhlaThis study
VJT15.96S. aureus NewmanS. aureus Newman Δrot::Tn917ΔlukEDΔrotΔlukEDThis study
VJT21.92S. aureus NewmanS. aureus Newman ΔlukEDΔhlg::tetΔlukEDΔhlgThis study
VJT21.93S. aureus NewmanS. aureus Newman Δrot::ermΔlukEDΔhlg::tetΔrotΔlukEDΔhlgThis study
VJT27.33S. aureus USA500USA500 BK2371WT (USA500-1)Roberts et al. (1998)
VJT27.40S. aureus USA500USA500 BK2371 ΔlukED::kanBK2371 ΔlukEDThis study
VJT27.36S. aureus USA500USA500 BK2395WT (USA500-2)Roberts et al. (1998)
VJT27.43S. aureus USA500USA500 BK2395 ΔlukED::kanBK2395 ΔlukEDThis study

Generation of mutant and complemented strains

All mutants not previously described were constructed via transduction of marked mutations using phage 80α (Table 2). Mutant strains (lukED::kan) were generated using the allelic replacement strategy previously described (Bae and Schneewind, 2006). Plasmids for allelic replacement of lukED were constructed using pCR2.1 and pKOR-1. A kanamycin resistance cassette (aphA3) was amplified from plasmid pBT-K (kindly provided to us by Dr Anthony Richardson) using oligonucleotide pair VJT524 (5′-TCCCCCCGGG-CTTTTTAGACATCTAAATCTAGGTAC) and VJT525 (5′-TCCCCCCGGG-CTCGACGATAAACCCAGCGAAC) and subsequently digested with XmaI and subcloned into the pCR2.1 vector containing sequences flanking the lukED locus (an internal XmaI site was previously generated between both flanking sequences to facilitate the insertion of antibiotic resistance markers). A PCR amplicon of the resultant lukED flanking sequences containing the internal kanamycin resistance gene was then recombined into pKOR1 resulting in the pKOR-1ΔlukED::kan plasmid. Further allelic replacement was carried out in strain Newman according to previously described methods and subsequently introduced into all other strains via transduction.

A lukED complementation strain was generated by cloning into plasmid pJC1112, which stably integrates into the SaPI-1 site of S. aureus resulting in single copy chromosomal complementation. To construct pJC1112, plasmid pJC1001, which carries the SaPI 1 attachment site on a temperature sensitive pT181 (cop634) replicon was digested with HpaI and subsequently re-ligated thereby removing the pT181 replicon and making it a suicide plasmid in S. aureus. To construct the pJC1112-lukED complementation vector, a PCR amplicon containing the lukED operon and upstream 791 bp was generated using primer pairs VJT605 [5′-CCCC-CTGCAG(PstI)-GATAGGTGAGATGCATACACAAC] and VJT299 [5′-CCCC-GGATCC(BamHI)-TTA-TACTCCAGGA TTAGTTTCTTTAG] and was subsequently digested and subcloned into pJC1112. The resultant plasmid was designated pJC1112-lukED and was subsequently integrated into the S. aureus SaPI-1 site.

Murine systemic infection with S. aureus

All animal infections were performed according to protocols approved by the NYU School of Medicine Institutional Animal Care and use Committee. Female ND4 Swiss Webster mice (∼6 weeks old) (Harlan laboratories) were used in all experiments (Dumont et al., 2011). Mice were first anesthetized via intraperitoneal injection with 250 µl Avertin (2,2,2-tribromoethanol dissolved in 2-methyl-2-butanol and diluted to a final concentration of 2.5% v/v in sterile saline) followed by infection via the retro-orbital venous plexus with 100 µl PBS containing ∼1 × 107 cfu of S. aureus (Dumont et al., 2011) except for neutrophil depletion studies in which ∼1 × 108 cfu were injected. For ‘survival’ curves, mice were observed at 3–5 h intervals and examined for signs of morbidity (hunched posture, ruffled fur, lack of movement, paralysis, and an inability to acquire food/water). At these prescribed end-points mice were immediately sacrificed and survival curves were plotted over time. To measure bacterial burden to infected kidneys, mice were sacrificed at 16 or 96 h post infection and kidneys were isolated, homogenized and serial dilutions were plated onto tryptic soy agar plates to enumerate cfu. For all other mouse experiments animals were sacrificed at either 16 or 96 h post infection and tissue/blood samples were collected for processing. All animal experiments were performed at least twice with groups of six or greater animals (see figure legends for specific cohort sizes).

Serum cytokines/chemokines

Serum was collected from mice infected as described above and IL-6 and GCSF were quantified using cytometric bead arrays (Becton Dickson; BD).

Isolation of PECS and intoxications

Mice were injected intraperitoneally with 1 × 107 cfu of S. aureus strain Newman. Sixteen hours post injection, the peritoneal cavity was flushed with PBS containing gentamicin (50 µg ml−1), penicillin (100 µg ml−1), and streptomycin (100 µg ml−1). Isolated PECs were subsequently washed, filtered, counted with trypan blue exclusion, and intoxicated with purified recombinant LukED followed by antibody staining for FACS analysis. Cell viability, membrane permeability and pore formation were measured via CellTiter (Promega), CytotoxOne (Promega), and EtBr incorporation, respectively, on an EnVision 2103 plate reader (Perkin-Elmer). Light and fluorescent microscopy images were acquired using an Axiovert 40CFL microscope (Zeiss).

Characterization of primary immune cells

Kidneys from infected mice were dissected and single cell suspensions generated (Torres et al., 2007). Cells were incubated with CD16/CD32 Fc blocker and subsequently stained with the following antibodies and dyes at the described dilutions: pacific blue viability dye (1:1000; Invitrogen), α-CD11b-PE-Cy7 (1:200; BD), and α-CD3-APC (1:250; BD). For characterization of PECs, cells were processed and blocked as described above and subsequently stained with the following antibodies: anti-CD11b-PE-Cy7 (1:200), and anti-Ly6G-FITC (1:500; BD). All samples were analysed on an LSR-II flow cytometer (Bectin-Dickson, BD). For quantification of immune cells isolated from infected kidneys a total of five independent mice were infected with each strain and percentages of cells acquired were averaged. PECs were isolated on two independent occasions from six mice and intoxications were subsequently conducted in triplicate at each toxin dose.

In vivo neutrophil depletion studies

Groups of six mice were injected with 300 µg of either anti-Ly6G (1A8) antibody or an isotype control (2A3) antibody intraperitoneally 48 h before infection as described previously (Blomgran and Ernst, 2011). A total of 1 × 108 cfu of either wild type (Newman) or ΔlukED was injected retro-orbitally and ‘survival’ was monitored over time. Spleens from control animals infected for 16 h with S. aureus were isolated, stained with anti-GR1-PE (1:1500; BD), anti-CD11b-PE-Cy7 (1:200), anti-CD3-APC (1:250), and anti-B220-FITC (1:500; BD) antibodies, and analysed by FACS to confirm neutrophil depletion in 1A8, but not 2A3 treated animals.

Generation of α- LukE and HlgC polyclonal sera

Rabbit polyclonal α-LukE and α-HlgC sera were generated using recombinant proteins as previously described (Dumont et al., 2011).

Protein purification

6X-His-LukE and 6X-His-LukD single subunit expression vectors were kindly provided by Dr Naoko Morinaga (Chiba University, Japan) and subsequently transformed into the Escherichia coli LysYLaqQ expression strain (New England Biolabs). Eight hundred millilitres of cultures were incubated at 37°C, 220 r.p.m. for 3.5 h followed by cooling to 16°C and induction with 0.1 mM IPTG for 16 h at 16°C, 220 r.p.m. Bacterial pellets were sonicated on ice and cell lysates clarified by centrifugation at 10000 r.p.m. for 30 min followed by incubation with 1 ml of Ni-NTA resin for 1 h. Bound protein was washed and eluted with Tris-buffered saline (TBS) supplemented with 500 mM Imidazole and subsequently dialysed into TBS + 10% glycerol. One hundred microlitres aliquots of filter sterilized protein were stored at −80°C until use.

Western blot

Strains were grown as described above. Cell-free culture supernatants containing soluble secreted proteins were subsequently collected, filter-sterilized, and the proteins precipitated with TCA as described previously (Dumont et al., 2011). All protein samples were run on 10% SDS-PAGE gels at 80 V for approximately 3 h. Proteins were transferred to nitrocellulose at 1 Amp for 1 h followed by blocking in phosphate buffered saline containing 0.1% Tween (PBST). Primary antibody dilutions were as follows: LukA (1:5000), LukE (1:10000), HlgC (1:5000), and Hla (1:5000; Sigma). Mouse anti-rabbit secondary antibody conjugated to AlexaFluor-680 was used at a 1:25000 dilution. Western blots were scanned on an Odyssey Imager (Licor).

Quantitative reverse transcriptase PCR (qRT-PCR)

Total RNA was prepared from the indicated S. aureus strains (Table 1) grown for 5 h in RPMI to an OD of ∼1.3. Twenty millilitres of bacterial culture was subsequently mixed 1:1 with a solution of 50% ethanol/50% acetone and frozen at −80°C until use. RNA was extracted from bacterial cells using an RNeasy purification kit according to the manufacturers protocol (Qiagen). Quality of the RNA was evaluated on an agarose-formaldehyde gel, and RNA abundance quantified using a nanodrop spectrophotometer. A SYBER green-based (Qiagen) ddCt relative quantification of gene expression assay was set up using 16S rRNA amplification as the endogenous control for all samples as described previously (Benson et al., 2011). The following primers were used: 16 s rRNA (TGAGATGTTGGGTTAAGTCCCGCA, CGGTTTCGCTGCCCTTTGTATTGT) and lukE (GAAATGGGGCGTTACTCAAA, GAATGGCCAAATCATTCGTT). Dissociation curves were determined for each primer set and relative quantification of gene expression was determined for duplicate reactions of each strain tested using Applied Biosystems Real Time Quantitative PCR Software (Applied Biosystems). Relative gene expression was determined by comparing all strains to the isogenic Newman lukED knockout strain that produces no transcript.


Statistics for survival curves was calculated using a Log-Rank (Mantel-Cox) test for statistical significance between curves. For all other experiments either 1-way anova with Tukey's multiple comparison test, or a two-tailed Student's t-test was used to determine statistical significance between groups. In all cases a P-value of less than 0.05 was considered statistically meaningful.


We thank members of the Torres Laboratory, Dr Nancy Freitag, and Dr Joel Ernst for critically reading this manuscript. We are grateful to Dr Naoko Morinaga for the kind gift of the LukE and LukD expression vectors; Dr Timothy Foster for the gift of the Δhla and Δhlg mutant S. aureus strains, Drs Joel Ernst and Ludovic Desvignes for help with the neutrophil depletion studies, and Lina Kozhaya and Stephen Rawlings for assistance in setting up FACS experiments. Several of the S. aureus strains used in this work were obtained from the NIH-NIAID supported Network on Antimicrobial Resistance in S. aureus (NARSA). This research was supported in part by New York University School of Medicine Development Funds (VJT), Grant 1R56AI091856-01A1 from the National Institute of Allergy and Infectious Diseases (VJT), Grant 5R01AI022159-26 from the National Institute of Allergy and Infectious Diseases (RPN), and an American Heart Association Scientist Development Grant (09SDG2060036) (VJT). MAB was supported in part by an American Heart Association predoctoral fellowship (10PRE3420022). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.