Staphylococcus aureus LukAB cytotoxin kills human neutrophils by targeting the CD11b subunit of the integrin Mac-1

Abstract

Staphylococcus aureus causes diseases ranging from superficial wound infections to more invasive manifestations like osteomyelitis and endocarditis. The evasion of host phagocytes recruited to the site of infection is essential to the success of S. aureus as a pathogen. A single S. aureus strain can produce up to five different bicomponent pore-forming leukotoxins that lyse immune cells by forming pores in the cellular plasma membrane. Although these leukotoxins have been considered redundant due to their cytotoxic activity toward human neutrophils, each toxin displays varied species and cell-type specificities. This suggests that cellular factors may influence which cells each toxin targets. Here we describe the identification of CD11b, the α subunit of the αM/β2 integrin (CD11b/CD18), macrophage-1 antigen, or complement receptor 3, as a cellular receptor for leukocidin A/B (LukAB), an important toxin that contributes to S. aureus killing of human neutrophils. We demonstrate that CD11b renders human neutrophils susceptible to LukAB-mediated killing by purified LukAB as well as during S. aureus infection ex vivo. LukAB directly interacts with human CD11b by binding to the I domain, a property that determines the species specificity exhibited by this toxin. Identification of a LukAB cellular target has broad implications for the use of animal models to study the role of LukAB in S. aureus pathogenesis, explains the toxin’s tropism toward human neutrophils and other phagocytes, and provides a cellular therapeutic target to block the effect of LukAB toward human neutrophils.

The Gram-positive bacterium Staphylococcus aureus causes invasive disease and can infect nearly all host tissues by combating the host immune system with an array of virulence factors (1). Infection with this pathogen can lead to several life-threatening diseases including endocarditis, pneumonia, osteomyelitis, sepsis, and necrotizing fasciitis (2). Treatment of staphylococcal infections is complicated by the emergence of antibiotic-resistant strains (3). In addition, the rise of community-associated methicillin-resistant S. aureus (CA-MRSA) infections is especially concerning due to the high morbidity and mortality of such infections (3).

To overcome host defenses, S. aureus secretes a large repertoire of cytotoxins that cause cell death through osmotic lysis by forming pores in the plasma membrane of target cells (4–6). The bicomponent leukocidins are a family of secreted staphylococcal toxins that form β-barrel pores through the assembly of two separate polypeptides into heterooligomeric complexes (7, 8). A single S. aureus strain can produce up to five different leukotoxins including two versions of γ-hemolysin (HlgAB and HlgCB), leukocidin E/D (LukED), Panton–Valentine leukocidin (PVL), and leukocidin A/B (LukAB, also known as LukGH) (4–6). The sequence similarity among these leukocidins ranges from 60% to 80% with the exception of LukAB, which is only 30–40% similar to the others (9, 10). LukAB is the most recently identified member of the bicomponent leukocidin family (9, 10). This toxin contributes to the cytotoxicity of clinical isolates toward innate immune cells and has been shown to play an important role in the success of S. aureus in both ex vivo and in vivo models of infection (9, 11). LukAB specifically targets phagocytes such as polymorphonuclear cells (PMNs or neutrophils) (9–12), which are an integral part of the host innate immune response to S. aureus (13). The capacity to kill PMNs is conserved among all leukocidins (4, 5), although the actions of LukED (14) and Hlg (15) are not limited to phagocytes (5). Despite their common cellular target, the leukocidins diverge in their potency toward PMNs of different species. In particular, LukAB is extremely effective at killing human PMNs but not murine PMNs (12), a trait shared with PVL (16) but not LukED (14, 17–19). The basis of LukAB’s tropism for human PMNs and other phagocytes has not been determined, but the recent identification of C-C chemokine receptor 5 (CCR5) as a cellular receptor used by LukED to kill lymphocytes, macrophages, and dendritic cells (14) suggests that other bicomponent leukocidins may use specific host factors to target and eliminate specific leukocytes.

To provide an explanation for the cellular tropism and species specificity exhibited by LukAB, we set out to identify cellular determinants required for LukAB-mediated cytotoxicity. We found that the integrin αM/β2 receptor, also known as CD11b/CD18, macrophage-1 antigen (Mac-1), or complement receptor 3 (CR3) is a host molecule required for LukAB-mediated cell killing. LukAB directly interacts with the CD11b subunit of this integrin, specifically the I domain, which is responsible for the species specificity exhibited by this toxin.

Results

LukAB Directly Interacts with Integrin αM/β2 (Mac-1).

To identify host proteins that interact with LukAB, we first performed a pull-down assay with PMN-HL60 cells, which are LukAB-sensitive neutrophil-like cells differentiated from the human promyelocytic leukemia (HL60) cell line (9). Biotinylated PMN-HL60 surface proteins were detergent solubilized and incubated with His-tagged LukAB, followed by isolation of toxin–host protein complexes by affinity chromatography (Fig. 1A). After SDS/PAGE and immunoblot analysis, we observed that a large number of host proteins associated with LukAB (Fig. 1A). The pull-down was repeated with primary human PMNs without biotinylation, and the identity of the cellular factors enriched in the presence of LukAB was determined by mass spectrometry. The most abundant LukAB-interacting cellular surface proteins were CD18 and CD11b (Table S1), the α and β components of the integrin complex known as integrin αM/β2, CR3, or Mac-1, herein referred to as Mac-1. The association of LukAB, but not LukED or PVL, with Mac-1 in a pull-down with detergent solubilized PMN-HL60 surface proteins was confirmed by immunoblot with a CD11b-specific antibody (Fig. 1B). A specific and direct interaction between LukAB and Mac-1 was confirmed by performing pull-down assays with purified recombinant toxin and purified receptor. A total protein stain revealed that LukAB, but not LukED or PVL, could pull down both the CD11b and CD18 subunits of the purified Mac-1 complex, which are about 150 and 95 kDa, respectively (Fig. 1C). Immunoblot further validated the presence of CD11b in the pull-down with LukAB but not the other toxins (Fig. 1D).

Fig. 1.

LukAB directly interacts with the human integrin Mac-1 (CD11b/CD18). (A) Pull-down of biotinylated PMN-HL60 lysates with his-tagged LukAB using nickle-conjugated beads, where samples were transferred to a nitrocellulose membrane and probed with DyLight streptavidin. (B) Immunoblot of a pull-down of PMN-HL60 lysates with his-tagged leukotoxins as described above using an anti-CD11b antibody. (C) Sypro Ruby protein stain of a pull-down of purified Mac-1 with his-tagged leukotoxins as described above and (D) corresponding immunoblot with an anti-CD11b antibody.

To better characterize the direct interaction of LukAB with Mac-1, we performed surface plasmon resonance (SPR) analysis, which indicated that LukAB binds to Mac-1 in a dose-dependent and saturable manner with a dissociation constant (K_d) of ∼38.4 nM (Table 1).

Table 1.

SPR analysis of LukAB/Mac-1 interactions

Protein + LukAB	Dissociation constant, Kd
Human recombinant Mac-1	3.84 × 10−8⋅M (±2.61 × 10−8)
Human recombinant I domain	1.92 × 10−9⋅M (±1.13 × 10−9)
Murine recombinant I domain	No interaction*

^*Affinity of interactions observed was greater than 10 mM.

CD11b Subunit of Mac-1 Is Necessary and Sufficient to Render Cells Susceptible to LukAB.

To determine whether Mac-1 confers susceptibility of host cells to LukAB, HL60s were transduced with lentiviruses containing nontargeting shRNA (NT shRNA) or CD18 shRNA. To enhance the susceptibility of these cells to LukAB, the stably transduced HL60 cell lines were differentiated into PMN-HL60s (Fig. 2A), and the effect of the shRNAs on the cell surface levels of CD18 and CD11b were confirmed by flow cytometry (Fig. 2B). Compared with NT shRNA-expressing cells, the CD18 shRNA-expressing cells displayed markedly reduced levels of CD18 on the cell surface (Fig. 2B). Because CD18 is required for the stability and surface localization of CD11b (20, 21), CD11b was also depleted in the CD18 shRNA cells (Fig. 2B). Thus, by targeting CD18, we generated a Mac-1 depleted cell line. Intoxication of the CD18 shRNA-expressing cells with purified LukAB revealed that Mac-1 is necessary for the formation of LukAB pores (Fig. 2C). In contrast, we observed that PVL formed pores in a Mac-1–independent manner, indicating that LukAB and PVL exploit different cellular determinants to exert their cytotoxicity (Fig. 2C).

Fig. 2.

CD11b is necessary and sufficient for LukAB-mediated killing of host cells. (A) Intoxication of HL60 or PMN-HL60 cells with 10 μg/mL of LukAB for 1 h. Cell viability was measured with the metabolic dye CellTiter. (B) Flow cytometry plot of CD18 and CD11b surface levels in PMN-HL60 cells transduced with a CD18 shRNA lentivirus compared with a NT shRNA lentivirus. (C) Intoxication of the NT and CD18 shRNA-expressing PMN-HL60 cells with 10 μg/mL of LukAB or PVL for 1 h. Pore formation was measured with the fluorescent dye EtBr. (D) CD18 and CD11b surface levels in HL60 cells and PMN-HL60 cells transduced with a CD11b shRNA lentivirus or NT shRNA lentivirus measured as described in B. (E) Intoxication of the NT and CD11b shRNA-expressing PMN-HL60 cells with 10 μg/mL of LukAB or PVL for 1 h. Pore formation was measured with EtBr. (F) Intoxication of CD11b transfected 293T cells with 40 μg/mL of LukAB or PVL for 2 h. Percentage depletion of CD11b⁺ cells was determined by staining cells with an anti-CD11b antibody postintoxication and performing flow cytometry analysis. Bar graph depicts the average of duplicate samples from two independent experiments. FACS plots are from a representative experiment. All other data are represented as the average of triplicate samples ±SD unless otherwise indicated. ***P < 0.0001 by one-way analysis of variance.

In addition to Mac-1, PMN-HL60s are decorated with lymphocyte function-associated antigen 1 (LFA-1, CD11a/CD18) and CD11c/CD18 (p150/95), which are also dependent on CD18 for cell-surface stability (20, 21). As a result, depletion of CD18 also led to a reduction in the surface levels of these β2 integrins (Fig. S1). To ensure that knockdown of the Mac-1 complex was responsible for the increased resistance to LukAB and not the general knockdown of β2 integrins, HL60 cells were stably transduced with CD11b-targeting shRNA. This strategy resulted in marked depletion of surface CD11b with no notable effect on CD18 levels (Fig. 2D). In fact, the Mac-1 levels observed on the cells transduced with the CD11b-targeting shRNA resembled those of the parental HL60 cells (Fig. 2D). We found that depletion of CD11b made the cells resistant to LukAB, but not to PVL pore formation (Fig. 2E). These findings demonstrate that CD11b is necessary for LukAB-mediated killing of target cells.

To determine whether CD11b is sufficient to render cells susceptible to LukAB, we next performed a gain-of-function experiment. It has been shown that HEK293T cells can support CD11b surface localization in the absence of CD18 (22). Therefore, we transiently transfected these cells with either a plasmid encoding CD11b or an empty plasmid, and CD11b surface levels were determined via flow cytometry (Fig. 2F). Intoxication of these cells with LukAB, but not PVL, resulted in depletion of the majority (80–90%) of the CD11b⁺ HEK293T cells, confirming that CD11b is necessary and sufficient to render cells susceptible to LukAB (Fig. 2F).

I Domain of CD11b Is Required for LukAB-Mediated Toxicity Toward Target Cells.

We next sought to determine if LukAB cytotoxicity could be blocked with CD11b-specific antibodies. Before intoxication with LukAB, primary PMNs were pretreated with three antibodies targeting different CD11b epitopes, as well as antibodies against CD18, CD11a, and CD11c. Although all three CD11b antibodies and the CD18 antibody reduced LukAB toxicity, only the LM2/1 CD11b antibody significantly inhibited LukAB activity compared with untreated cells or an isotype control (Fig. 3 A and B). Importantly pretreatment with the CD11b-specific antibodies, including the LM2/1 clone, did not alter Mac-1 surface levels (Fig. S2).

Fig. 3.

LukAB targets the I domain of CD11b to kill cells. (A) Viability of PMNs treated with 10 μg/mL of integrin-specific antibodies (α-CD11a, α-CD11c, and α-CD18), including three different anti-CD11b clones (α-CD11b), or no antibody (No Ab) followed by a 1-h intoxication with 2.5 μg/mL of LukAB. Membrane damage was measured with the fluorescent dye SYTOX green. Results represent the mean from PMNs isolated from eight donors ±SEM. (B) Viability of PMNs treated with 10 μg/mL of LM2/1 or an isotype control then intoxicated and evaluated as described above. Results represent the mean from PMNs isolated from four donors ±SEM. (C) CD11b surface levels on HL60 cells stably transduced with empty vector (EV), WT CD11b, or I-less CD11b virus compared with PMN-HL60s as determined by flow cytometry analysis with an anti-CD11b antibody. Mean fluorescence intensity in relative fluorescence units: EV = 4.96, WT CD11b = 93.1, I-less CD11b = 36.4, and PMN-HL60 = 49. Viability of stably transduced HL60 cell lines described in B compared with PMN-HL60s after 1-h intoxication with 10 μg/mL of LukAB where membrane damage (D) and cellular metabolism (E) were evaluated as described in A and Fig. 2A, respectively. Data are represented as the average of triplicate samples ±SD. *P < 0.05, **P < 0.01, and ***P < 0.0001 by one-way analysis of variance.

The LM2/1 antibody recognizes the CD11b I domain (or A-domain), which is where most endogenous Mac-1 ligands bind through a metal ion-dependent adhesion site (MIDAS) (23). We hypothesized, based on the LM2/1 blocking data, that the I domain of CD11b is required for LukAB-mediated killing of target cells. To address this possibility, we constructed a mutated CD11b where the I domain was deleted using overlap PCR as previously described (24). It has been established that the deletion of the I domain does not affect the interaction of CD11b with CD18 or the interaction between Mac-1 and endogenous ligands that do not require the I domain (24). We transduced HL60 cells with virus made from constructs containing wild-type (WT) CD11b, CD11b lacking the I domain (“I-less”), or an empty vector control. We chose these cells because they are highly resistant to LukAB and have low levels of CD11b (Fig. 2 A and C). We reasoned that if the I domain is necessary for cytotoxicity, exogenous WT CD11b would render these cells as susceptible as PMN-HL60 cells, where as an I-less version of CD11b would not. Following transduction and stable integration, the levels of CD11b on the surface of the HL60 cell lines was evaluated by flow cytometry with an anti-CD11b antibody that recognizes both the WT and I-less versions of CD11b (Fig. 3C). Ectopic WT CD11b rendered HL60 cells susceptible to LukAB as evidenced by increased membrane damage and cell death compared with the empty vector control HL60 cells (Fig. 3 D and E). In contrast, HL60 cells containing ectopic I-less CD11b were markedly resistant to LukAB-mediated cytotoxicity (Fig. 3 C–E and Fig. S3). We found that ectopic WT and I-less CD11b levels fluctuate on the HL60 surfaces, which explains the slight difference in receptor levels observed in Fig. 3C. Nevertheless, HL60 cells containing higher levels of ectopic I-less CD11b compared with native CD11b on PMN-HL60 were still resistant to LukAB (Fig. S3). Collectively, these data demonstrate that the I-domain of CD11b is required for LukAB-mediated killing of host cells.

LukAB Displays Higher Affinity for the Human CD11b I Domain than the Murine CD11b I Domain.

Purified LukAB has been shown to be highly cytotoxic toward human and primate PMNs, intermediately toxic toward rabbit PMNs, and minimally toxic toward murine PMNs (12). These findings suggest that LukAB targets PMNs in a species-specific manner. We observed that murine peritoneal exudate cells (PECs), which are highly susceptible to LukED, are resistant to LukAB (Fig. 4A). PECs mostly consist of recruited PMNs (Ly6G⁺/CD11b⁺), and monocytes and macrophages (Ly6G⁻/CD11b⁺), all of which have high levels of surface CD11b (Fig. 4B).

Fig. 4.

LukAB preferentially binds to the human CD11b I domain compared with the murine CD11b I domain. (A) Pore formation in PECs following a 1-h intoxication with 20 μg/mL of LukAB or 10 μg/mL of LukED as measured with EtBr. Data are represented as the average of triplicate samples ±SD. (B) Flow cytometry analysis showing Ly6G and CD11b surface levels on PECs using anti-Ly6G and anti-CD11b antibodies. (C) Phylogenetic tree of the amino acid sequence alignment of human, gorilla, rabbit, and mouse I domains constructed with DNASTAR MegAlign software using the CLUSTALW method. (D) Dot blot of purified recombinant human or murine CD11b I domain incubated with 5 μg/mL Alexa488–LukAB. Alexa488–LukAB binding was quantified by densitometry. (E) Competition dot blot assay where purified recombinant human CD11b I domain was incubated with 5 μg/mL fluorescently labeled LukAB (Alexa488–LukAB) and 10-fold excess (50 μg/mL) of unlabeled LukAB or unlabeled PVL. Alexa488–LukAB binding was quantified by densitometry.

The species specificity of LukAB together with the necessity of the CD11b I domain for toxin activity (Fig. 3 C and D) led us to examine the conservation of this domain from different species. Alignment of the amino acid sequences of the human, gorilla, rabbit, and mouse CD11b I domains revealed that as expected, gorilla is the most similar to human (98.6% identity), followed by rabbit (79.1% identity), and then murine (78.1% identity) (Fig. 4C). These data correlate with the tropism of LukAB toward PMNs from each species (12). To investigate if these differences could influence LukAB binding to the CD11b I domain, we developed a dot blot assay to detect the LukAB–CD11b I domain interaction. We observed a dose-dependent interaction between active Alexa488-labeled LukAB and the human CD11b I domain (Fig. 4D). Comparison of Alexa488–LukAB binding to human versus murine CD11b I domains revealed that LukAB preferentially binds to the human CD11b I domain (Fig. 4D). SPR analysis determined that LukAB binds to the human CD11b I domain with an approximate K_d of 1.92 nM, whereas no interaction between LukAB and the murine CD11b I domain was detected (Table 1). Furthermore, the binding of LukAB to the human I-domain was specific, as excess unlabeled LukAB, but not unlabeled PVL, could compete off Alexa488–LukAB binding in the dot blot assay (Fig. 4E).

Extracellular S. aureus Use CD11b to Cause LukAB-Mediated Cell Damage During Infection.

To establish a role for CD11b in S. aureus infections, we infected the NT or CD11b shRNA-expressing PMN-HL60 cells with the CA-MRSA pulsed field gel electrophoresis USA300 strain Los Angeles County (LAC) or an isogenic mutant lacking lukAB (ΔlukAB) (9). Consistent with our previous findings (9), WT LAC killed the NT PMN-HL60 cells in a LukAB-dependent manner (Fig. 5A). In contrast, when CD11b surface levels were reduced in these cells by shRNA (CD11b), the WT LAC no longer caused cell damage and instead closely resembled the lukAB mutant strain (Fig. 5A).

Fig. 5.

CD11b renders cells susceptible to LukAB-mediated killing by extracellular S. aureus in ex vivo infections. (A) Viability of the NT or CD11b shRNA PMN-HL60 cells described in Fig. 2B following a 2-h infection with nonopsonized WT CA-MRSA USA300 strain LAC or an isogenic lukAB mutant (ΔlukAB) at the indicated multiplicity of infection (MOI). Membrane damage was measured with SYTOX green. Data are represented as the average of triplicate samples ±SD. (B) Viability of PMNs treated with 10 μg/mL of CD11b-specific antibodies followed by a 1-h infection with the indicated MOI of nonopsonized WT LAC. Membrane damage was measured with SYTOX green. Results represent the mean from PMNs isolated from eight donors ±SEM. *P < 0.05 and ***P < 0.0001 by one-way analysis of variance.

We next performed ex vivo infections of purified human PMNs with WT LAC, where we attempted to block LukAB-mediated cell damage through pretreatment with anti-CD11b antibodies before infection. These experiments revealed that the LM2/1 antibody successfully neutralized USA300-mediated cell damage (Fig. 5B), thus establishing a role for LukAB-mediated targeting of the CD11b I domain during the S. aureus–PMN interaction.

Phagocytosed S. aureus Exploit LukAB-Mediated Targeting of CD11b to Cause Cell Damage and Promote Escape from Within.

We recently established that LukAB-mediated cell damage postphagocytosis promotes the early escape of LAC from within PMNs and subsequent bacterial outgrowth (11). To determine if CD11b contributes to the intracellular cytotoxic activity of LukAB, the NT and CD11b shRNA PMN-HL60 cells were infected with opsonized USA300 and synchronized to promote phagocytosis (11). Importantly, depletion of CD11b did not influence phagocytosis of LAC (Fig. S4). Under these conditions, we observed that knockdown of CD11b abolished cell damage caused by WT LAC (Fig. 6A). Similar to our findings with human PMNs (11), phagocytosed LAC employs LukAB to prevent PMN-HL60–mediated growth restriction (Fig. 6B), and knockdown of CD11b eliminated the growth advantage of WT LAC compared with the ΔlukAB mutant strain (Fig. 6B).

Fig. 6.

LukAB-mediated cellular damage and growth rebound of phagocytosed S. aureus is dependent on CD11b. (A) Viability of NT or CD11b shRNA PMN-HL60 cells described in Fig. 2B following a 90-min infection with various MOI of opsonized WT or ΔlukAB LAC. Membrane damage was measured with SYTOX green. Data are represented as the average of triplicate samples ±SD. (B) Growth of opsonized WT or ΔlukAB LAC upon infection of NT or CD11b shRNA PMN-HL60 cells at an MOI of 10. Bacterial colony-forming units were determined at 1-, 2-, or 3-h postsynchronization and were normalized to input at time 0, which was set at 100%. Results represent the average of triplicate samples from two independent infections ±SD. (C) Localization of CD11b in PMNs postinfection with opsonized GFP–LAC at an MOI of 10, or in uninfected PMNs determined by staining with a fluorescently conjugated anti-CD11b antibody or an isotype control before infection. Cells were fixed postsynchronization and images were captured using an Applied Precision PersonalDV live-cell imaging system. A representative image for each condition is shown. (D) Pretreatment of PMNs with 10 μg/mL, the LM2/1 anti-CD11b antibody, or an isotype control followed by infection with GFP-LAC at an MOI of 10 under lysostaphin treatment. EtBr staining in red is indicative of membrane damage. Images were captured using a fluorescent microscope at 0 and 30 min postsynchronization, and representative images from 30 min are shown. (E) Quantification of EtBr-positive PMNs per field of view obtained from images as in D. Results represent the average of three independent counts ±SEM from infections of PMNs isolated from three donors at (T0) and 30 (T30) min postinfection. ***P < 0.0001 by one-way analysis of variance.

For CD11b to be used by phagocytosed S. aureus to escape from within PMNs, we proposed that CD11b might be in close proximity to phagocytosed S. aureus. It has been established that CD11b is internalized during phagocytosis of zymosans and can be found in the phagosomal membrane (25). To determine the location of CD11b during phagocytosis of S. aureus, human PMNs were prestained with a fluorescently labeled anti-CD11b antibody or a fluorescently labeled isotype control, followed by infection with GFP–LAC. Infected cells were fixed postsynchronization and imaged using an Applied Precision PersonalDV live-cell imaging system with z-stack capability. In uninfected human PMNs, the CD11b staining is predominantly dispersed across the plasma membrane of the cell (Fig. 6C and Fig S5). However, upon infection with LAC, CD11b was also found intracellularly (Fig. 6C and Fig. S5).

We next attempted to block the LukAB-mediated PMN damage caused by phagocytosed LAC using the neutralizing LM2/1 anti-CD11b antibody. For these experiments, PMNs were pretreated with the LM2/1 antibody or an isotype control before infection with GFP–LAC WT, isogenic ΔlukAB, or isogenic ΔlukAB chromosomally complemented with lukAB (11). These experiments were performed in the presence of lysostaphin and anti-LukA to eliminate the potential influence of extracellular bacteria and LukAB, as well as the fluorescent dye ethidum bromide (EtBr) to measure membrane damage (11). Of note, pretreatment with LM2/1 before infection does not block phagocytosis of S. aureus as the amount of GFP–LAC observed within PMNs was similar regardless of LM2/1 treatment (Fig. S6). Phagocytosed LAC causes LukAB-mediated membrane damage at 30 min postsynchronization when PMNs are pretreated with isotype control antibody (Fig. 6D). In contrast, LM2/1 pretreatment resulted in decreased LukAB-mediated membrane damage (Fig. 6 D and E), mimicking the phenotype observed with the lukAB mutant strain.

Discussion

This study describes the identification of CD11b, a component of the Mac-1 integrin, as a cellular molecule exploited by the staphylococcal leukotoxin LukAB to specifically target and kill cells. This conclusion is supported by our findings that LukAB directly interacts with the Mac-1 complex (specifically the I domain of CD11b), and CD11b is necessary and sufficient to render cells susceptible to LukAB as evidenced by knockdown and gain-of-function analyses.

The identification of a cellular target that is specifically used by LukAB and not other bicomponent toxins such as LukED and PVL highlights that the staphylococcal leukotoxins possess nonredundant mechanisms for targeting specific cell types. CCR5 was recently identified as a cellular receptor used by LukED to target and kill lymphocytes, macrophages, and dendritic cells (14). However, monocytes and PMNs are killed by LukED in a CCR5-independent manner, suggesting that additional cellular receptors may be used by LukED to target these cells (14, 17–19). The fact that a single staphylococcal toxin may target multiple receptors and that each toxin may use distinct nonredundant receptors vastly increases the number of cell types that S. aureus can eliminate with an already extensive repertoire of toxins. Despite the lack of overlap in cellular targets identified thus far, the staphylococcal leukotoxins have been shown to exhibit synergistic effects (26). The identification of CD11b as a cellular target for LukAB provides an explanation for the previously reported synergism between LukAB and PVL toward human PMNs (10, 11), as sublytic concentrations of PVL increase CD11b surface levels on these cells (27). From an evolutionary standpoint, the use of multiple, nonoverlapping cellular targets provides an explanation for why S. aureus has such a large arsenal of cytotoxins and is such a well-adapted pathogen when it comes to evading host PMNs.

The killing of innate immune cells such as PMNs is crucial to the pathogenesis of S. aureus. Mac-1 is expressed in all cells targeted by LukAB (9) including PMNs, macrophages, monocytes, and dendritic cells (28), and is involved in multiple cellular functions such as phagocytosis, cellular activation, cell-mediated killing, and chemotaxis (22, 29). The prevalence of Mac-1 on these crucial immune cell types is exploited by multiple pathogens in addition to S. aureus, and may be a conserved method of targeting host phagocytes. For example, the Bordetella pertussis adenylate cyclase toxin (CyaA) is another bacterial toxin that binds to and uses Mac-1 to cause cell death (30). Other pathogens such Bacillus anthracis (31), pneumococcus (32), and gonococcus (33) bind to Mac-1 to promote internalization of the bacterium or spore by the host cell. Both the pneumococcal pilus adhesion protein RrgA and the gonococcal porin and pili interact with Mac-1, like LukAB, via the I domain of CD11b (32, 33).

Here we show that both extracellular S. aureus and phagocytosed S. aureus use LukAB to cause PMN damage during infection by targeting CD11b. The finding that LukAB is the main contributor to leukotoxin-mediated PMN killing during ex vivo infection by S. aureus is consistent with the observation that this toxin is the most highly up-regulated toxin in the presence of PMNs (11, 34). We have recently proposed a model for the escape of S. aureus from within PMNs early in infection that is dependent on LukAB-mediated killing of PMNs (11). In this study we were able to further these findings by showing that CD11b is internalized upon phagocytosis of S. aureus and that CD11b is required for LukAB-mediated damage caused by phagocytosed S. aureus. However, the exact mechanism by which CD11b mediates the intracellular effects of LukAB is not yet known.

The identification of human CD11b I domain as a cellular target of LukAB provides an explanation for the observed species specificity exhibited by this toxin. We determined that the affinity of LukAB toward the human CD11b I domain was in the nanomolar range, whereas no interaction between LukAB and the murine CD11b I domain was observed, consistent with the previously reported limited susceptibility of murine PMNs (12). The difference in binding affinity is most likely explained by the divergent sequence homology between the I domains from these two species, which exhibit 78.1% identity. Of note, we have observed that LukAB contributes to both the infection process and the bacterial burden of USA300 in a low-dose infection model of murine renal abscesses (9). At this point, it is unclear how LukAB promotes disease in murine models considering the marked resistance of murine PMNs to LukAB and the low affinity of this toxin for murine CD11b. A possible explanation is that LukAB may cause disease in the mouse through receptor-independent immunomodulatory mechanisms, as has been as shown for PVL. Similar to LukAB, PVL also displays cytolytic tropism toward human PMNs (16), but has been shown to elicit proinflammatory responses in human PMNs and macrophages (26, 27) as well as murine PMNs and macrophages (35, 36). These findings suggest that the roles of LukAB and PVL in the mouse may be better explained by immunomodulatory functions rather than cytolytic functions, and indicate that mouse models underestimate the true contribution of these toxins to S. aureus pathobiology in humans. Moreover, the species-specific activities of an expanding number of virulence factors produced by S. aureus (e.g., superantigens, chemotaxis inhibiting protein of S. aureus, leukotoxins) (4, 37) highlight the limitations of the murine models currently used to study S. aureus pathogenesis. Thus, improved animal models are paramount for understanding the full virulence potential of S. aureus, which is a prerequisite for the development of effective drugs that can combat this important human pathogen.

Materials and Methods

Isolation of Primary Human PMNs.

Blood samples were obtained from anonymous healthy donors as buffy coats (New York Blood Center). The New York City Blood Center obtained written informed consent from all participants involved in the study. This research was approved by the New York University School of Medicine institutional human subjects board. PMNs were isolated using a Dextran gradient as previously described (11).

Other Procedures.

Detailed procedures are available in SI Materials and Methods.