Identification of biologic agents to neutralize the bicomponent leukocidins of Staphylococcus aureus

See allHide authors and affiliations

Science Translational Medicine  16 Jan 2019:
Vol. 11, Issue 475, eaat0882
DOI: 10.1126/scitranslmed.aat0882

Circumventing Staphylococcus aureus subversion

Staphylococcus aureus subverts the host immune system with various mechanisms, including the cytolytic leukocidins. Chan et al. modified a protein scaffold to generate centryins, which bind S. aureus leukocidins in vitro and can protect human cells from leukocidin-mediated lysis. Further modification to extend centryin half-life in circulation led to antibacterial protection in various mouse models. These centryins could one day be used to prevent or possibly treat S. aureus infections in people.


A key aspect underlying the severity of infections caused by Staphylococcus aureus is the abundance of virulence factors that the pathogen uses to thwart critical components of the human immune response. One such mechanism involves the destruction of host immune cells by cytolytic toxins secreted by S. aureus, including five bicomponent leukocidins: PVL, HlgAB, HlgCB, LukED, and LukAB. Purified leukocidins can lyse immune cells ex vivo, and systemic injections of purified LukED or HlgAB can acutely kill mice. Here, we describe the generation and characterization of centyrins that bind S. aureus leukocidins with high affinity and protect primary human immune cells from toxin-mediated cytolysis. Centyrins are small protein scaffolds derived from the fibronectin type III–binding domain of the human protein tenascin-C. Although centyrins are potent in tissue culture assays, their short serum half-lives limit their efficacies in vivo. By extending the serum half-lives of centyrins through their fusion to an albumin-binding consensus domain, we demonstrate the in vivo efficacy of these biologics in a murine intoxication model and in models of both prophylactic and therapeutic treatment of live S. aureus systemic infections. These biologics that target S. aureus virulence factors have potential for treating and preventing serious staphylococcal infections.


Staphylococcus aureus (S. aureus) causes a wide spectrum of diseases in humans, ranging from relatively mild conditions, like impetigo, to life-threatening diseases, including bacteremia and pneumonia (1, 2). Severe infections caused by methicillin-resistant S. aureus (MRSA) carry the highest overall case fatality rate (14%) of all identified bacterial and fungal pathogens (3). Antibiotic treatment of S. aureus infection is often inadequate, and this may be partially explained by deficiencies in arms of the immune system that typically work in concert with antibiotics to eradicate infection. In this regard, S. aureus is successful because of the array of virulence factors it produces that promote immune evasion, including the inhibition of complement activation, blockade and destruction of phagocytic cells, and modification of normal B cell and T cell responses (4). In light of the multitude of virulence factors used by S. aureus to cause disease, it may not be surprising that there have been numerous failed attempts to develop effective vaccines or immunotherapies when therapeutics have been approached from a single-target angle (5). It is of paramount importance that novel preclinical therapies target the components of the multifaceted makeup of the S. aureus pathogenic lifestyle (6, 7).

A crucial component of the virulence of S. aureus is the secretion of a group of pore-forming bicomponent toxins known collectively as leukocidins. These leukocidins include the Panton-Valentine leukocidin (LukSF-PV or PVL), leukocidin ED (LukED), gamma hemolysins (HlgAB and HlgCB), and leukocidin AB (LukAB; also known as LukGH) (8, 9). These toxins subvert the immune system by targeting immune cells in a receptor-dependent manner (9). The bicomponent leukocidins are composed of two subunits, a receptor targeting S-type subunit (for “slow” migration in chromatography columns: LukS-PV, LukE, HlgA, HlgC, and LukA) and a polymerizing F-type subunit (for “fast” migration in chromatography columns: LukF-PV, LukD, HlgB, and LukB) (8, 9). Upon binding, these toxins oligomerize to form an octameric pre-pore that penetrates the plasma membrane of target cells, causing osmotic imbalance and cell lysis (8, 9).

LukED lyses immune cells by targeting C-C and C-X-C chemokine receptor type family members CCR5, CXCR1, and CXCR2 and by targeting the Duffy antigen receptor for chemokines (DARC) on erythrocytes (1013). HlgAB also targets CXCR1, CXCR2, and DARC (12, 14), but in contrast to LukED, it also targets CCR2 (14). LukSF-PV and HlgCB both target the C5aR1 and C5aR2 receptors (14, 15). LukAB is the only bicomponent leukocidin that does not target a chemokine receptor but instead targets leukocytes by engaging the αM/β2 integrin complex (also known as CD11b/CD18, macrophage-1 antigen/MAC-1, or complement receptor 3/CR3) via direct interaction with the I-domain of CD11b (16). The lack of conservation between the human and murine CD11b, C5aR, and CXCR1/2 proteins means LukAB, PVL, HlgAB, and HlgCB have higher affinity for the human receptors than the murine receptors, thus limiting the use of mouse models to study these toxins (9).

To identify biologic agents that neutralize S. aureus leukocidin activity, we screened libraries of centyrins, members of an alternate scaffold class of protein-based biologic agents (17), for their abilities to bind and neutralize bicomponent leukocidins. Centyrins are small (~10 kDa) globular proteins derived from a consensus sequence of the 15 fibronectin type III (FN3)–binding domains of the human tenascin-C protein (Fig. 1A) (18, 19). Libraries based on the consensus FN3 binding sequence of human tenascin-C (herein referred to as TENCON) present diversified amino acid residues at positions within the BC loop, C-strand, CD loop, F-strand, and/or FG loop of the FN3 domain (18, 19) and are used in iterative rounds of in vitro panning against purified target proteins, wherein centyrins with increasing selective binding potency for target ligands are identified (20). This approach selects for centyrins that bind to target ligands with high specificity and affinity. As alternatives to antibodies, centyrins exist as unglycosylated single-chain proteins that lack disulfide bonds and can easily be produced in large quantities in Escherichia coli.

Fig. 1 Structure and sequence alignments of selected potent anti-leukocidin centyrins.

(A) Crystal structure of parental TENCON (Protein Data Bank: 3TES). Purple regions indicate sites of amino acid diversification introduced into panning libraries used to identify centyrins with strong affinities for leukocidins. (B) Amino acid sequence alignment of selected leukocidin centyrins and the parental TENCON sequence. Positions of loop regions are underlined and labeled. Amino acid residues conserved across all sequences are highlighted in yellow. Numbers in parentheses indicate the position of the amino acid in the centyrins.

Here, we describe the identification and characterization of centyrins that bind the bicomponent leukocidins of S. aureus and neutralize their activities in both in vitro and ex vivo systems. Furthermore, we identify a series of albumin-binding consensus domain–centyrin fusion proteins that confer in vivo protection in a panel of murine models of S. aureus infection in the context of both prophylactic and therapeutic interventions.


Selection for centyrins that bind bicomponent leukocidins

To identify centyrins that selectively bind S. aureus bicomponent leukocidins or subunits thereof, we performed a series of in vitro selections involving library variants of the parental TENCON protein (Fig. 1A) (18, 19) with diversified residues at a combination of strand and loop positions (see Materials and Methods). Iterative rounds of selection using CIS display methods (20) facilitated the identification of centyrins that bound each target toxin (Table 1). Centyrins with apparent selectivity in binding the leukocidin targets were purified and characterized by size exclusion chromatography to identify those that remained monomeric in solution. Binding selectivity was then confirmed by both enzyme-linked immunosorbent assay (ELISA) and biolayer interferometry (BLI). The dissociation constants (Kds), as determined by BLI, ranged from 190 pM to 140 nM (table S1). Using the approach described above, we identified a panel of centyrins that bound the leukocidins derived from both the so-called “loop” and “sideways” libraries (Fig. 1, A and B, and fig. S1).

Table 1 Summary of centyrin panning campaign.

View this table:

Centyrins that bind bicomponent leukocidins can neutralize toxin activity

A selection of purified centyrins that exhibited potent and selective leukocidin binding activity were then tested for their abilities to neutralize the cytolytic activity of leukocidins in an assay, where primary human polymorphonuclear leukocytes (hPMNs) were challenged with toxins. In all subsequent experiments, TENCON, the parental scaffold molecule with no detectable binding affinity or activity toward the toxins, serves as the negative control for neutralizing activity. These experiments revealed that centyrins can neutralize purified LukAB (Fig. 2A), LukED (Fig. 2B), HlgAB (Fig. 2C), HlgCB (Fig. 2D), and PVL (Fig. 2E). The specific half-maximal concentration (EC50) was calculated for each of the toxin-centyrin combinations on the basis of the value elicited by the most potent centyrin for each toxin. These EC50 values ranged from 13.9 to 132.2 nM (Fig. 2F). On the basis of their binding patterns, leukocidin-targeting centyrins were further tested for their capacities to cross-neutralize leukocidins. This identified centyrins that cross-neutralized both LukED and HlgAB (Fig. 2G) and both PVL and HlgCB (Fig. 2H). These findings are consistent with the observation that LukED/HlgAB and PVL/HlgCB share common receptors to target and kill hPMNs (11, 12, 15), a finding that coincides with the fact that these toxin pairs also share the greatest amino acid similarity among leukocidins (8, 9).

Fig. 2 Ability of centyrins to block leukocidin-mediated lysis of primary hPMNs.

(A to E) Viability of primary hPMNs after a 1-hour incubation with indicated doses of LukAB (A), LukED (B), HlgAB (C), HlgCB (D), or PVL (E) and in the presence of increasing concentrations of centyrins. Data were normalized to the maximal lactate dehydrogenase release by each individual toxin. (F) EC50 of the selected most potent neutralizing centyrins calculated from nonlinear robust regression fit of the dose-response curve (A to E). (G) EC50 of identified LukED/HIgAb cross-reactive centyrins calculated from nonlinear robust regression fit of a dose-response curve (B and C). (H) EC50 of identified cross-reactive centyrins between PVL/HlgCB calculated from nonlinear robust regression fit of dose-response curve (D and E). (A to H) Results are the means ± SEM, with n = 4 donors.

Beyond exhibiting cytolytic activity against immune cells bearing cognate CC and CXC chemokine receptors, LukED and HlgAB also lyse human erythrocytes by targeting the surface receptor DARC (12). To determine whether the LukED and HlgAB centyrins also neutralize toxin-mediated hemolytic activity, we incubated LukED (Fig. 3A) or HlgAB (Fig. 3B) with primary human erythrocytes in the presence of TENCON or the antitoxin centyrins. These experiments demonstrated that centyrins can neutralize the hemolytic activities of these bicomponent leukocidins.

Fig. 3 Ability of centyrins to block LukED- and HlgAB-mediated hemolysis.

Primary human erythrocytes were incubated with 90% lytic dose (LD90) of LukED (A) or HlgAB (B) in the presence of the indicated centyrin, and hemolysis was evaluated after a 30-min incubation at 37°C by measuring the absorbance of released hemoglobin into the surrounding media at 450 nm. Hemolysis was normalized to Triton X-100 lysed control. Results are the means ± SEM, with n = 6 different donors. EC50 of the most potent neutralizing centyrins are shown to the right.

Centyrins block binding of bicomponent leukocidins to their cognate receptors

To further characterize the inhibitory mechanism(s) of the centyrins, we undertook a series of studies to determine whether the centyrins could block binding of purified bicomponent leukocidins (or subunits thereof) to their host receptors or to target host cells. The interaction of LukAB with purified human CD11b I-domain, the cognate receptor of LukAB (16, 21), was measured using a plate-based ELISA. These studies revealed that the anti-LukAB centyrin (SM1S17) blocks the interaction of LukAB and the CD11b I-domain in a concentration-dependent manner (Fig. 4A). In contrast, TENCON failed to block the LukAB/CD11b I-domain interaction.

Fig. 4 Mechanism of neutralization by anti-leukocidin centyrins.

(A) Binding of LukAB to the human CD11b I-domain in the presence of TENCON or LukAB centyrin (SM1S17). Results are from two independent experiments. (B) Immunoblot of biotinylated toxin (anti-his: 100-ng protein loaded; streptavidin: 50-ng loaded; white line indicates noncontinuous lanes) ladder-like appearance of the toxins is the product of varying degrees of biotinylation. (C) Viability of hPMNs after 1-hour incubation with unlabeled and biotinylated S-subunits (Xb) paired with their cognate F-subunit. Results are the means ± SEM, with n = 4 donors. (D to G) Indicated biotinylated subunits were incubated with hPMNs in the presence of TENCON, the neutralizing centyrin or unlabeled toxin subunit (competitor), and binding of the labeled toxin to the cell surface evaluated via flow cytometry. Results are presented as the median fluorescence intensity ± SEM [(D) n = 5 donors, LukE (unlabeled) n = 2 donors; (E) n = 5 donors, HlgA (unlabeled) n = 2 donors; (F and G) n = 4 donors].

Unlike LukAB, which assembles into a LukA-LukB heterodimer to target its receptor (21), the other bicomponent leukocidins target receptors via their S-subunit (1012, 14, 15). To examine the ability of the most potent centyrins to block binding of their target toxin subunit to hPMNs, we developed a flow cytometry–based assay involving biotinylated leukocidin S-subunits (Fig. 4B). Labeled S-subunits were active when mixed with their respective cognate F-subunit (Fig. 4C) with only minor impairment of toxin activity after biotinylation, indicating that toxins largely preserved their cognate receptor targeting and ability to form octameric pores. Biotinylated S-subunits were then incubated with increasing concentrations of unlabeled S-subunit (competitor), TENCON, or the selected toxin centyrins, and binding of the biotinylated S-subunits to the surface of hPMNs was measured. As expected, the unlabeled S-subunits competed with the binding of the biotinylated S-subunits in a dose-dependent manner, validating the assay (Fig. 4, D to G). Moreover, the selected toxin-specific centyrins blocked binding of their respective subunits to target cells, whereas the TENCON centyrin did not block toxin-hPMN binding (Fig. 4, D to G).

Centyrins protect human phagocytes from toxin-mediated killing

Having identified centyrins that inhibit the in vitro cytolytic activities of their respective bicomponent leukocidin targets, we next investigated whether the centyrins also block killing of hPMNs by S. aureus. For these ex vivo experiments, we infected hPMNs with isogenic derivatives of a representative strain of the community-associated MRSA (CA-MRSA) epidemic clone in the United States, USA300 (22, 23). We used strain AH1263 (hereafter referred to as AH-LAC), a LAC derivative cured of antibiotic resistance plasmids (24). AH-LAC exhibited potent cytotoxicity toward hPMNs, a phenotype fully dependent on the bicomponent leukocidins, as deletion of all five leukocidins (Δluk) rendered the strain unable to kill hPMNs (Fig. 5A). The killing exhibited by wild-type AH-LAC is fully dependent on LukAB, as deletion of this toxin (ΔlukAB) phenocopied Δluk (Fig. 5A). Moreover, a strain that produces only LukAB (Δluk/lukAB+) exhibited similar cytotoxicity compared to wild-type AH-LAC (Fig. 5A). To evaluate the protective potential of the selected centyrins (Table 1), we performed ex vivo hPMN infections as above but in the presence of purified TENCON or the indicated leukocidin-binding centyrins. Consistent with the genetic data shown in Fig. 5A, only the anti-LukAB centyrin (SM1S17) or a mixture of anti-leukocidin centyrins including SM1S17 was able to inhibit the cytolytic activity of AH-LAC against hPMNs (Fig. 5B). The anti-LukAB centyrin (SM1S17), in contrast to TENCON, exhibited concentration-dependent inhibitory activity toward AH-LAC–mediated killing of hPMNs (Fig. 5C). Similar results were observed using an array of methicillin-sensitive S. aureus (MSSA) and MRSA isolates that represented the diversity of S. aureus clonal complex lineages (Fig. 5D). These results further support the notion that in this ex vivo infection model, LukAB mediates cytolytic activity.

Fig. 5 Ability of anti-leukocidin centyrins to protect hPMNs from S. aureus–mediated killing.

(A) Extracellular infection of hPMNs with indicated USA300 strains at a multiplicity of infection (MOI) of 25. Bars indicate means ± SEM, with n = 9 donors. (B) Extracellular infection of hPMNs with wild-type USA300 at MOI of 25 in the presence of indicated centyrins. Bars indicate means ± SEM, with n = 9 donors. ns, not significant; WT, wild-type. (C) Extracellular infection of hPMNs with wild-type USA300 at MOI of 25 in the presence of titrated concentrations of TENCON and SM1S17. Points indicate means ± SEM, with n = 3 donors. (D) Extracellular infection of hPMNs with MRSA and MSSA strains of diverse clonal complexes at MOI of 25 with TENCON or SM1S17. Bars indicate means ± SEM, with n = 6. Statistical analysis probing for neutralizing effect of SM1S17 was done using a one-tailed paired t test. (E) Infection of hPMNs with indicated USA300 isogenic strains pre-opsonized with 20% human serum (MOI of 15) in the presence of indicated centyrins, without or with subsequent centrifugation. Bars indicate means ± SEM, with n = 4 donors. (A to E) Cell viability was determined with CytoTox-ONE homogeneous membrane integrity assay. Statistical analysis where applicable was examined using ordinary one-way analysis of variance (ANOVA) with post hoc Dunnett’s multiple comparison test. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

LukAB is also a key virulence factor in the ability of S. aureus to kill phagocytes following phagocytosis, a process mediated by targeting of CD11b in the phagosome (16, 25, 26). To determine whether the SM1S17 exhibits intracellular blocking activity, wild-type AH-LAC and the ΔlukAB isogenic strain were preopsonized with 20% pooled human serum and the bacteria were incubated with the hPMNs via centrifugation to promote association and uptake (25). As a control, opsonized bacterial strains were also incubated with hPMNs without centrifugation, which results in the majority of the bacteria remaining extracellular (25). In contrast to TENCON, the SM1S17 blocked extracellular killing mediated by opsonized AH-LAC (Fig. 5D), phenocopying the inability of the ΔlukAB strain to kill hPMNs. In contrast to the extracellular infection model, SM1S17 exhibited minimal neutralizing activity after internalization of the bacteria, presumably due to the inability of centyrins to access or accumulate within the intracellular compartment (Fig. 5E; intracellular).

Centyrins prevent bicomponent leukocidin-mediated lethality in mice

We used a murine model of systemic intoxication to further investigate the role of centyrins as anti-infectious agents. In this model, purified LukED and HlgAB cause rapid lethality when administered systemically in mice, providing a robust model of S. aureus toxemia (27). Using this model, TENCON afforded no protection when premixed with a lethal dose of purified LukED (Fig. 6A), whereas premixing LukED with the anti-LukED centyrin SM1S26 at a 1:1 molar ratio of toxin to centyrin resulted in 100% protection (Fig. 6A). In contrast, administration of increased dosage of SM1S26 (33- to 155-fold excess compared to the premixing experiments) 1 hour before LukED intoxication afforded no protection (Fig. 6B).

Fig. 6 Protection from in vivo intoxication with ABDcon-centyrins.

(A) Survival curves of mice after intravenous administration of centyrins at 10× (30 μg) or 1× (3 μg) premixed with a lethal dose of purified LukED (10 μg per subunit); n = 3 mice per group. PBS, phosphate-buffered saline. (B) Survival curves of mice treated with a prophylactic dose of LukE centyrins (~500 μg or ~100 μg) before intravenous administration of a lethal dose of LukED (10 μg per subunit); n = 3 mice per group. (C) Viability of hPMNs following incubation with purified LukED (76 nM, 2.5 μg/ml) and SM1S26 with or without ABDcon. Results indicate means ± SEM; n = 3 donors. (D) Survival curves of mice that had been treated prophylactically with ABDcon-SM1S26 (at molar excesses of 1×, 10×, and 100×) 5 hours before multiple sequential intravenous administrations (indicated by arrows) of purified LukED (6 μg per subunit); n = 3 mice per group. (E and F) Survival curves of mice treated prophylactically with ABDcon-centyrins (at molar excess of 50×) 48 hours before first intravenous administration of lethal dose of purified toxin (6 μg per subunit), LukED (E), or HlgAB (F) with cross-reactive centyrins color matched across panels; surviving mice were administered additional lethal doses of toxin at 24-hour intervals and monitored up to 2 days after intoxication; n = 4 mice per group.

The findings described above are consistent with the observation that centyrins exhibit serum half-lives of ~40 min (28). Centyrins are typically rapidly eliminated from systemic circulation through glomerular filtration, a process associated with therapeutic agents of molecular weight <50 kDa (29). One strategy to overcome this limitation is the genetic fusion of centyrins to a serum albumin-binding domain (ABD) (28). In mice, TENCON exhibits a terminal serum half-life of ~40 min, whereas TENCON-ABD fusion proteins exhibit terminal serum half-lives of 34 to 60 hours in mice and up to 193 hours in cynomolgus monkeys (28). Accordingly, we fused the toxin-binding centyrins to an ABD consensus (ABDcon) sequence. The ABDcon-TENCON and ABDcon-SM1S26 fusion proteins behaved similarly to the parental centyrins in ex vivo neutralization assays (Fig. 6C). Next, the ABDcon-TENCON and ABDcon-SM1S26 fusion proteins were examined using the in vivo systemic intoxication model (Fig. 6D). For these studies, we administered a single prophylactic dose of ABDcon-TENCON or ABDcon-SM1S26 (at 1×, 10×, or 100× molar excess over LukED) 5 hours before LukED intoxication. As expected, all animals in the ABDcon-TENCON group succumbed rapidly to a single-dose LukED intoxication. In contrast, a single dose of the ABDcon-SM1S26 fusion protein protected 100% of the mice when administered at doses of 10× and 100× molar excess after two sequential challenges with LukED, one at 5 hours and the second at 24 hours after initial challenge. Moreover, 33 and 100% protection was also observed with ABDcon-SM1S26 at doses of 10× and 100×, respectively, following a subsequent third challenge with LukED at 48 hours after ABDcon-SM1S26 administration.

A series of different potent LukED and HlgAB neutralizing ABDcon fusion centyrins were generated and examined in vivo. In these studies, the ABDcon-centyrins were administered to mice 48 hours before challenge with LukED and HlgAB to examine the long-lasting protection potential of these molecules. All of the tested ABDcon-LukED centyrins protected 100% of the mice after three lethal challenges with the administered toxin (Fig. 6E). Similar results were observed with a pair of ABDcon-HlgAB centyrins in mice challenged with HlgAB (Fig. 6F). Moreover, one of the tested centyrins, ABDcon-SM1S449, also exhibited cross-reactive neutralizing activity toward LukED and HlgAB in both ex vivo (Fig. 2H) and in vivo models (Fig. 6, E and F).

Centyrins protect against S. aureus infections

To further examine the therapeutic potential of the ABDcon-centyrins, we tested the molecules in a lethal S. aureus bacteremia model. In this model, mice were infected intravenously with either a wild-type MSSA strain Newman (30) or its isogenic mutant ΔlukED (31). The ABDcon-TENCON and ABDcon-SM1S26 were administered 4 hours before infection and 48, 96, and 144 hours after infection. In the lethal infection model where ABDcon-centyrins were given prophylactically, 100% of the mice receiving the ABDcon-TENCON centyrin succumbed to infection by 168 hours after infection, whereas 58% of the mice that had received the ABDcon-SM1S26 centyrin were alive 168 hours after infection (Fig. 7A). These data are consistent with blocking of LukED activity in vivo, as similar results were observed when mice were infected with the ΔlukED strain and treated with either ABDcon-TENCON or ABDcon-SM1S26 (Fig. 7A).

Fig. 7 Protection from in vivo infection with ABDcon-centyrins.

(A) Survival curves of mice infected intravenously with ~4 × 107 colony-forming units (CFU) of S. aureus strain Newman. Centyrins were administered intravenously 4 hours before and 48, 96, and 144 hours after infection. n = 12 to 15 mice per group. (B) Survival curves of mice infected intravenously with ~3 × 107 CFU of USA500. Centyrins were administered intravenously 4 hours before and 24, 72, and 120 hours after infection. n = 9 to15 mice per group. (C) Survival curves of mice infected intravenously with ~2 × 107 CFU of USA500. Treatments were given intravenously 4 hours before and 24, 72, and 120 hours after infection. n = 10 mice per group. (A to C) Statistics were performed using Gehan-Breslow-Wilcoxon test with P values adjusted for multiple comparisons. (D) Mice were infected intravenously with ~1 × 107 CFU of USA500. Treatments were as in (C). CFU in tissues were measured 96 hours after infection. CFU from individual mice is shown ± SEM. n = 12 to 15 mice per group. LOD, limit of detection. (E) Mice were infected with ~8 × 107 CFU of USA300 premixed with ABDcon-centyrins, ABDcon-TENCON, or a mixture of ABDcon-SM1S26 and SM1S163, intraperitoneally (ip). CFU in tissues were measured 48 hours after infection. CFU from individual mice are shown ± SEM. n = 10 to 11 mice per group. Statistical analysis of bacterial burden was examined using two-tailed Mann-Whitney test, *P < 0.05, **P < 0.01, and ****P < 0.0001.

To evaluate whether the ABDcon-centyrins can also protect animals from infections with MRSA, we challenged mice intravenously with the highly virulent USA500 MRSA strain BK2395 (hereafter referred to as USA500) (31, 32). Four hours prior to infection with wild-type USA500, mice were administered the ABDcon-TENCON and ABDcon-SM1S26, and then 24, 72, and 120 hours after infection, mice received the same dosing. We found that ABDcon-SM1S26 notably enhanced the survival of infected mice compared to that of mice infected and treated with ABDcon-TENCON (Fig. 7B).

To expand on the finding described above and to assess whether treatment alone was sufficient to protect mice from infection, mice were infected with USA500 and then administered the ABDcon-centyrins 4, 24, and 48 hours after infection. S. aureus has been shown to disseminate to major organs within 4 hours after intravenous infection (33). We found that mice provided with ABDcon-SM1S26 after infection were notably protected from MRSA lethality (Fig. 7C), further demonstrating the utility of these molecules.

We next sought to evaluate whether, in addition to reducing S. aureus–caused lethality, centyrins could also reduce the bacterial burden in infected organs. After the treatment regimen described above, mice were infected intravenously with a sublethal dose of USA500 and then administered centyrin test article 4, 24, and 48 hours after infection. We observed a significant reduction in bacterial burden in multiple organs of mice treated with ABDcon-SM1S26 compared to that of ABDcon-TENCON–treated mice (P < 0.05, Fig. 7D). The ABDcon-TENCON group infected with the ΔlukED mutant exhibited similar bacterial outputs as observed in the ABDcon-SM1S26 cohort infected with wild-type USA500 (Fig. 7D).

Next, we examined the efficacy of the ABDcon-centyrins in an intraperitoneal murine model of systemic infection used to study leukocidins in USA300 (14). For this model, mice were challenged with the USA300 strain SF8300 (34) in the presence of the ABDcon-centyrins as a single bolus injection. Administration of ABDcon-SM1S26 and ABDcon-SM163, to block both LukED and HlgAB, appreciably reduced bacterial burden in multiple target organs compared to mice administered the ABDcon-TENCON centyrin (Fig. 7E).


Given the crucial contributions of S. aureus pore-forming toxins to pathogenesis, these virulence factors have attracted increasing interest as potential therapeutic targets (6, 9, 35). A number of monoclonal antibodies (mAbs) targeting alpha toxin are currently under clinical development for the prevention of S. aureus pneumonia, including mAbs MEDI4893 and KBSA301 (6, 3638). Similarly, a number of anti-LukAB human mAbs have been described (39, 40). Despite the number of promising molecules currently in the pipeline to treat S. aureus infections, there is a strong precedent for failure in such single-target approaches (6). Thus, it is becoming increasingly apparent that drug developers need to consider the value of integrating multidisciplinary strategies when conceiving antistaphylococcal therapeutics. Here, we describe studies that explore the potential of antibody mimetics known as centyrins, which have been explored in a number of clinical contexts (18, 19, 28, 41), to inhibit S. aureus infectivity by targeting and blocking the leukocytic activity of bicomponent leukocidins.

We used iterative rounds of in vitro selection and affinity-based panning to identify centyrins that selectively neutralize the cytolytic activity of each of the five bicomponent leukocidins of S. aureus. Additional centyrins were identified that exhibit limited cross-neutralization activity. In general, centyrins exhibiting cross-reactive properties targeted leukocidins that share strong amino acid sequence conservation between their subunits (e.g., HlgCB and LukSF-PV) (8). Centyrin-mediated blocking of host receptor engagement appears to be a common mode of action in target neutralization by the leukocidin-neutralizing centyrins we have characterized. Evidence for this mechanism is supported by the observation that the anti-LukAB centyrin (SM1S17) disrupts binding to the I-domain of CD11b, the cognate receptor for LukAB. For the remaining leukocidins, we provide evidence that these centyrins neutralize their targets by interfering with binding of the targeted leukocidin to the plasma membrane of host cells.

Consistent with previously published data (42, 43), S. aureus exhibits LukAB-dependent extracellular killing of hPMNs. This killing can be selectively neutralized in a concentration-dependent manner by the LukAB-blocking centyrin, SM1S17. In contrast, minimal protection of the hPMNs was observed with SM1S17 if S. aureus are internalized, indicating that centyrins have minimal access to the intracellular environment. Although S. aureus has traditionally been regarded as an extracellular pathogen, it has become increasingly clear that S. aureus can survive within phagocytes to promote dissemination and pathogenesis (16, 25, 26, 44). Thus, one limitation of this study is the lack of host cell penetrance exhibited by the centyrins and their subsequent inability to neutralize S. aureus toxicity from within. This limitation also likely affects the functions of non-opsonizing antitoxin antibodies. Clinical use of centyrins would likely require engineering of a mechanism to confer phagosome access to centyrins in the host cell. One such approach would be through the fusion of centyrins to human mAbs that target the surface of S. aureus (6). This strategy was recently proven effective for the delivery of S. aureus targeting antibiotics to the intracellular compartment of host cells (44).

Our in vivo data highlight the ability of the ABDcon-centyrins against LukED and HlgAB to protect mice from challenge with lethal doses of purified LukED or HlgAB, demonstrating that these molecules can protect from very acute toxin-mediated disease. Moreover, the data obtained from the different in vivo infections show centyrin-driven toxin inhibition in both prophylactic and treatment settings, which supports the notion that blocking leukocidins in vivo can improve host-mediated containment of S. aureus infection. This study was partially limited by the natural incompatibility that exists between the remaining S. aureus leukocidins and their cognate receptor in mice. However, we hypothesize that the humanizing of such murine receptors would only bolster the value of our dataset. In the most rigorous examination of our centyrin-fusion constructs, we observed markedly improved survival and reduction of bacterial burdens when the molecules were given as treatments (4 hours after infection) after infection with a highly virulent MRSA strain. However, it remains to be determined how these preclinical data would apply to treatment in a clinical setting, where the etiology of infection can take greater than 20 hours to be identified. Another potential concern regarding the therapeutic use of centyrins is the possibility that these molecules have immunogenic properties and are thus capable of evoking the production of host-derived neutralizing antibodies. However, given that centyrins have parental scaffolds that are of human origin, these molecules are predicted to have low immunogenicity in humans (19).

In summary, we describe here the identification and characterization of a new class of “antitoxin” biologic, centyrins, which selectively neutralize the cytolytic activities of each of the five bicomponent leukocidins produced by S. aureus. These biologic agents with toxin-neutralizing activity have a potential clinical utility in the treatment and prevention of serious staphylococcal infections in which the bicomponent leukocidins are implicated as important pathogenesis factors. As new virulence factors involved in the different arms of S. aureus pathogenesis are identified, a flexible platform is exemplified here, wherein centyrins targeting specific virulence factors can be developed. Furthermore, the engineering of fusion proteins combining centyrins with opsonic antistaphylococcal mAbs provides a potential path forward for the development of therapeutics to combat S. aureus infections with activity in both extracellular and intracellular compartments (6).


Study design

The objective of this study was to identify and characterize centyrins capable of binding to and neutralizing the S. aureus bicomponent leukocidins. We hypothesized that if we could generate such centyrins, then these molecules would be able to reduce the disease burden elicited by the toxins. To examine the neutralizing potential of the centyrins, we used an ex vivo model with primary hPMNs to either intoxicate the hPMNs with purified leukocidin or to infect the hPMNs with live S. aureus.

Expanding upon these results, we sought to investigate in mice whether these neutralizing molecules could prevent the lethal outcomes of either intoxication with purified leukocidin or infection with live S. aureus. Treatments with centyrins were given as indicated, and doses were established on the basis of the titration shown in Fig. 6D. In these experiments, mice were monitored for signs of morbidity using pre-established endpoint criteria (see the “Acute intravenous infection model” section in Supplementary Materials and Methods). To promote randomization, all mice were mixed before the start of the study and the sets of mice (typically three to five mice per cage) were randomly assigned treatment groups. Experimenters were not blinded to the treatment/infection groups once established. Additional study design parameters and details are explained in Supplementary Materials and Methods.

All animal experiments were reviewed and approved by the Institutional Animal Care and Use Committee of New York University Langone Health (protocol nos. IA16-00050, 141201, and 130601), and all experiments were performed according to National Institutes of Health (NIH) guidelines, the Animal Welfare Act, and U.S. federal law.

Buffy coats used to purify primary hPMNs were obtained from anonymous donors, with informed consent from the New York Blood Center. Because all of the samples were collected anonymously before their delivery, the New York University Langone Health Institutional Review Board determined that our study was exempt from further ethics approval requirements.

Statistical analysis

Data were analyzed using GraphPad Prism 7.0. All experiments were analyzed for statistical significance by one-way ANOVA and post hoc Holm-Sidak correction of multiple comparisons (α = 0.05), unless otherwise indicated. All comparisons survival curves were analyzed using Gehan-Breslow-Wilcoxon test, unless otherwise indicated. All P values stated are corrected where appropriate. GraphPad style was used for reported statistical significance as follows: *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.


Methodological Details

Fig. S1. Selected neutralizing LukE/D centyrins from loop diversified library.

Table S1. Reported centyrin Kds calculated from BLI.

Table S2. S. aureus strains used in the study.

Table S3. Primary data (provided as an Excel file).

References (4554)


Acknowledgments: We would like to acknowledge early contributions to this work by R. Brezski, D. Clark, and by members of Janssen Biotherapeutics (JBIO, the biologics discovery and development group of Janssen Research & Development LLC.) for contributions in the identification, purification, and characterization of anti-leukocidin centyrins and centyrin albumin-binding protein fusion proteins. We would like to further acknowledge the contributions from current and former members of the Torres laboratory (NYU) in the design and interpretation of experiments described here. We would also like to thank B. Shopsin, I. Thomsen, and the former NIH-sponsored Network on Antimicrobial Resistance in S. aureus (NARSA) for providing the S. aureus strains used in Fig. 5D. Funding: This work was solely supported by Janssen Research & Development LLC, under the auspices of an exclusive License and Research Collaboration Agreement with NYU. A.L. was partially supported by the NIH-National Institute of General Medical Sciences (NIH-NIGMS) award no. GM007308 and NIH-National Institute of Allergy and Infectious Diseases (NIAID) award nos. AI007180 and AI124606. W.E.S. was partially supported by NIH-NIAID award no. AI007180. K.M.B. was partially supported by NIH-NIGMS award no. GM007308 and an MSTP Vilcek Scholar Award. V.J.T. is also supported by the NIH-NIAID under award nos. R01AI099394 and R01AI105129, and he is a Burroughs Wellcome Fund Investigator in the Pathogenesis of Infectious Diseases. Author contributions: R.C., P.T.B., W.E.S., W.R.S., A.S.L., and V.J.T. designed the experiments. R.C., P.T.B., A.O., W.E.S., F.A., A.L., K.M.B., A.P., J.F., and B.W. performed the experiments. R.C., P.T.B., A.O., W.E.S, F.A., K.B., W.R.S., A.S.L., and V.J.T. analyzed the data. R.C., W.E.S., A.S.L., and V.J.T. wrote the manuscript. All the authors edited the manuscript. Competing interests: P.T.B., A.P., J.F., W.R.S., B.W., and A.S.L. are employees of Janssen Research & Development LLC. V.J.T. is an inventor on patents and patent applications (patent nos. 8431687, 8846609, 9091689, 948172, 9480726, 9644023, 9657103, 9783582, 9783597, and 10087243) filed by New York University, which are currently under the commercial license to Janssen Biotech Inc. All other authors declare that they have no competing interest. Data and materials availability: All data associated with this study are present in the paper or in the Supplementary Materials.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article