Alpha-toxin-elicited CX₃CL1 release in Staphylococcus aureus pneumonia impairs bactericidal function of human monocytes

ABSTRACT

Staphylococcus aureus is an important human pathogen causing severe invasive infections. Pathogenesis is attributed to a wide array of virulence factors, including several potent exotoxins such as the pore-forming α-toxin. In this study, we found that patients with S. aureus respiratory tract infections had elevated CX₃CL1 levels in airway fluid and plasma. Using human-organotypic lung models, we observed that stimulation of lung epithelium with α-toxin induces an intensified CX₃CL1 expression apically in the epithelium as well as the release of CX₃CL1. Blocking α-toxin or ADAM10 activity in organotypic lung using an α-toxin-blocking antibody or a specific ADAM10 inhibitor confirmed their role in modulating CX₃CL1 cleavage and release. Analyses of CD14⁺ human monocytes in combination with a CX₃CR1 inhibitor revealed that α-toxin-mediated CX₃CL1 release induces CX₃CL1-dependent chemotaxis. In line with these data, lung tissue from patients with S. aureus respiratory tract infection showed elevated CX₃CL1 and CD14 staining as compared with tissue from patients with non-infectious lung diseases. Functional studies of monocytes showed that CX₃CL1 released by lung models resulted in upregulated CD83 and downregulated CD86, as well as impaired killing of phagocytosed S. aureus. Furthermore, stimulation of monocytes with soluble CX₃CL1 hampered their reactive-oxygen and nitric-oxide production. Taken together our data show that S. aureus triggers the release of lung epithelial CX₃CL1, and we identify an immunomodulatory effect of α-toxin involving its cytotoxic and ADAM10-interacting properties, inducing CX₃CL1 release leading to impaired monocyte effector function.

IMPORTANCE

Exotoxins are essential virulence factors for the pathobiont S. aureus and contribute toward severe invasive infections such as pneumonia. S. aureus α-toxin is a pore-forming exotoxin that causes host cell lysis and severe lung pathology. We found that α-toxin drives the release of membrane-bound chemokine CX₃CL1 by involving ADAM10-mediated proteolytic activity. Furthermore, the release of CX₃CL1 modulated immune responses locally, as demonstrated by enhanced monocyte migration and reduced capacity of monocytes to kill ingested bacteria. CX₃CL1-induced reduction in bacterial killing coincided with impaired production of reactive oxygen and nitric oxide species. This reveals a novel mechanism in the pathogenesis of S. aureus lung infections involving α-toxin-induced release of CX₃CL1, leading to impaired bacterial killing by monocytes.

INTRODUCTION

The pathophysiology of Staphylococcus aureus pneumonia involves a variety of virulence mechanisms mediated by secreted degradative enzymes, immune evasion, and cytotoxic factors such as exotoxins (1–3). The pore-forming exotoxin α-toxin is one of the most well-studied virulence factors of S. aureus. It has been shown that α-toxin mediates cell damage directly via its pore-forming capability and contributes to lethality in pulmonary infection (4, 5). S. aureus strains lacking α-toxin show reduced cytotoxicity and virulence in invasive lung disease models, and the more α-toxin S. aureus produces, the greater the airway epithelial damage (5, 6). Consequently, using the mouse and rabbit S. aureus pneumonia model, α-toxin has been shown to be a key virulence determinant causing severe lung tissue pathology (7, 8). Following the identification of A disintegrin and metalloproteinase domain-containing protein 10 (ADAM10) as a receptor for α-toxin (4), major advances have been made in understanding the pathogenesis and host cell specificity of S. aureus. Binding of α-toxin to ADAM10 expressed by epithelial cells (9), activation of intracellular signaling pathways following pore formation has also been shown to increase the enzymatic activity of ADAM10 (10). The role of ADAM10 in α-toxin-mediated lethality has been partly attributed to the cleavage of E-cadherin in the lung tissue of mice infected with S. aureus (11). Furthermore, the targeting of ADAM10 by α-toxin can hijack its regular homeostatic function by increasing its enzymatic activity and causing cleavage of tight junction proteins (12–14). Similarly, α-toxin-mediated epithelial damage is not limited to its receptor engagement. Oligomerization of α-toxin to form a heptameric β-barrel pore in the membrane (15), followed by pore formation and influx of ions such as Ca²⁺, can lead to cell lysis (9, 12, 14, 16). Subsequent pore formation in the airway epithelial cells can lead to several downstream inflammatory responses such as increased levels of pro-inflammatory mediators, increased epithelial permeability, and increased vascular leakage (17–19). Since there is an extensive list of potential ADAM10 substrates, perturbations to the tightly regulated activity of ADAM10 may have consequences in the pathogenesis of S. aureus pneumonia beyond ADAM10’s role as an α-toxin docking protein and its cleavage of E-cadherin (20). ADAM proteins have recently been reviewed and highlighted in the regulation of immune cells (21). One ADAM10 substrate that regulates immune cell function is the chemokine CX₃CL1 (also known as fractalkine). CX₃CL1 is produced by various cell types, including lung epithelial cells. In its membrane-bound form, CX₃CL1 promotes cell-cell adhesion, and it induces chemotaxis as a soluble molecule. The proteolytic cleavage of membrane-bound CX₃CL1 by a metalloproteinase is a unique mechanism of chemokine release and has been implicated in the recruitment and transmigration of immune cells to CX₃CL1-expressing mucosal tissue, such as the respiratory tract (22, 23). Increased levels of CX₃CL1 have been reported in infectious and inflammatory diseases (24, 25), but whether S. aureus respiratory tract infections and α-toxin are associated with CX₃CL1 release is not known. Additionally, whether α-toxin interaction with ADAM10 at sub-lytic concentrations in mucosal tissue modulates immune cell functions and contributes to the pathogenesis locally is poorly understood.

Here, we examined patients with S. aureus lung infections and used a human in vitro organotypic model of S. aureus pneumonia (6). We found that in S. aureus lung infection, release of lung epithelial membrane-bound CX₃CL1 mediated by α-toxin involves its cytotoxic and ADAM10-interacting properties and that soluble CX₃CL1 modulates monocyte phenotype and function. This study reveals an α-toxin-ADAM10-CX₃CL1 axis as a novel component of α-toxin-mediated modulation of effector functions of immune cells and of pathogenesis in S. aureus respiratory infections.

RESULTS

Local and systemic levels of CX₃CL1 are increased in patients with S. aureus respiratory tract infection

To investigate the relationship between S. aureus respiratory tract infections and CX₃CL1, plasma, and lung airway fluids (bronchioalveolar lavage and tracheal aspirates) from prospectively enrolled patients with confirmed S. aureus respiratory tract infection were collected and analyzed for soluble CX₃CL1. Compared with other severely ill non-infectious ICU prospectively enrolled patients or healthy volunteers, we observed on average 10 times higher levels of CX₃CL1 in the S. aureus respiratory tract infections group in both lung fluids and plasma, respectively (Fig. 1A and B; Tables S1 and S2 ). Next, we compared the levels of CX₃CL1 in frozen plasma samples from healthy volunteers and a retrospective cohort of patients with confirmed S. aureus bloodstream infection, without respiratory tract infections. Healthy volunteers had undetectable levels of CX₃CL1 (P = 0.66, 14 out of 19 below limit of detection, LOD), whereas patients with S. aureus bloodstream infection had slightly elevated levels (P < 0.01) (Fig. 1C). Notably, compared with the patients with S. aureus respiratory tract infection, many of the patients with S. aureus bloodstream infection had undetectable levels of CX₃CL1 (10 out of 23 below LOD). We observed no association between CX₃CL1 levels and clinical parameters such as presence of co-morbidity, age, or severity/outcome in the S. aureus patient cohort (Fig. S1A through C). Taken together, these data suggest that S. aureus respiratory tract infection increases CX₃CL1 levels both locally and systemically.

Fig 1.

CX₃CL1 levels in samples from Staphylococcus aureus-infected patient samples. (A, B) CX₃CL1 protein levels in lung airway fluids (bronchioalveolar lavage and tracheal aspirates) from ICU control patients and patients with S. aureus respiratory tract infection (A), and in plasma from healthy volunteers and patients with S. aureus respiratory tract infection (B). The bars show mean ± SEM. (C) Plasma levels of CX₃CL1 in S. aureus bacteremia (bloodstream infection) patients and healthy volunteers. Values below the limit of detection (LOD) were set to 0.1103 $\left(\mathrm{L}\mathrm{O}\mathrm{D}/\sqrt{2}\right)$ and are shown in gray. Statistical significances were calculated using the Mann-Whitney unpaired t-test (A, B) and the Wilcoxon signed-rank test with median value compared with a hypothetical value of 0.156 (i.e., in comparison with LOD) (C).

S. aureus-associated CX₃CL1 release is driven by α-toxin

As CX₃CL1 levels were particularly high in patients with S. aureus respiratory tract infection, we performed follow-up analyses using in vitro lung tissue models developed to study S. aureus pneumonia (6). The tissue models were stimulated with bacterial culture supernatants derived from the community-associated MRSA strain USA300, which has been associated with severe lung infections as well as two additional clinical S. aureus strains—strain NP796 isolated from a patient with a severe necrotizing pneumonia and strain LE2332 isolated from a patient with lung empyema (26). In the unstimulated lung model, CX₃CL1 was evenly expressed throughout the epithelium, closely reflecting the CX₃CL1 expression pattern in uninfected human lung tissue (Fig. 2A). Although unstimulated and LE2332-stimulated lung models displayed similar patterns of CX₃CL1 immunostaining, the lung models stimulated with the high toxin-producing strains USA300 or NP796 supernatants revealed a unique pattern of intensified CX₃CL1 staining at the apical side of the epithelium compared with the deeper mucosal layers, which displayed relatively diffuse or no staining (Fig. 2B). By analyzing the fluorescent ratio of CX₃CL1 staining on the top (apical) and bottom (basal) half of the mucosal layer, we observed a significantly stronger apical expression of CX₃CL1 in response to NP796 and USA300 supernatants as compared with that of the LE2332-treated and unstimulated models (Fig. 2C). Importantly, these responses correlated with elevated levels of CX₃CL1 in the lung model supernatants (Fig. 2D). This suggests that the membrane-bound CX₃CL1 is shed from the cell surface and released in response to S. aureus.

Fig 2.

Alpha-toxin-mediated tissue redistribution and release of CX₃CL1. (A) Representative mean intensity projections of immunofluorescence microscopy images of tissue sections from the lung tissue model and human lung tissue. The sections were stained with the nuclear stain 4´,6-diamidino-2-phenylindole, dihydrochloride (DAPI) (blue) in combination with antibodies detecting CX₃CL1 (green). The scale bars in images equal 50 µm. (B) Immunofluorescence staining of CX₃CL1 (green) and cell nuclei (blue, DAPI) in sectioned lung model stimulated with S. aureus LE2332, NP796, or USA300 bacterial culture supernatants (diluted 1:100), and cell culture media was used as a negative control in the unstimulated models. The scale bars in images equal 100 µm. (C) Measurement of the spatial re-distribution (ratio of apical to basolateral mean fluorescence intensity, MFI) of CX₃CL1 in lung tissue stimulated with bacterial culture supernatants (diluted 1:100). (D) CX₃CL1 levels in culture supernatants from unstimulated lung tissue model (black bar) or lung tissue models stimulated with LE2332, NP796, or USA300 bacterial culture supernatants (diluted 1:100) (gray). (E) Measurement of the spatial re-distribution (ratio of apical to basolateral MFI) of CX₃CL1 in lung models stimulated with various concentrations of α-toxin as indicated. (F) CX₃CL1 levels in culture supernatants of unstimulated lung models, or of lung models stimulated with α-toxin suspension (100 ng/mL) or NP796 bacterial culture supernatant (1:100) treated with anti-α-toxin antibodies (200 µg/mL, MEDI4893*) (striped bars). As control (black bar), an isotype-matched antibody was used. (G) CX₃CL1 levels in culture supernatants of unstimulated lung model (black bars) or of lung models stimulated with wild-type α-toxin (gray) or the mutated α-toxin H35L (white). ELISA determination of CX₃CL1 levels was performed 24 h post-stimulation. All experiments were performed at least three times. In panels D–G, the bars show the mean ± SD of three individual experiments. Statistically significant differences were determined by Kruskal-Wallis with uncorrected Dunn’s test in panels C, D, and G, linear regression analysis in panel E, and ordinary two-way ANOVA with Sidak’s multiple comparison test in panel F.

An enhanced β-hemolytic phenotype on blood agar was evident for USA300 and NP796 strains (Fig. S1D), consistent with the previously reported increased production of α-toxin in these strains compared with LE2332 (26). An equally prominent β-hemolytic phenotype was found for S. aureus isolates from patients analyzed for CX₃CL1 in lung airway fluids (Fig. S1E). Thus, indicating that the above-noted effect on CX₃CL1 could be α-toxin mediated. Stimulation of lung models with increasing concentrations of α-toxin revealed a dose-dependent response in the apical/basal ratio of CX₃CL1 expression in the lung models (R² = 0.70, P < 0.001) (Fig. 2E). Furthermore, the addition of anti-α-toxin antibody (MEDI4893*) completely mitigated the NP796- and α-toxin-induced CX₃CL1 release in the lung model supernatants (Fig. 2F), suggesting that CX₃CL1 released in S. aureus respiratory tract infection is driven by α-toxin. This is further supported by the reduced CX₃CL1 levels in the BAL fluid of mice with staphylococcal respiratory tract infection that were administered the anti-α-toxin antibody MEDI4893* (Fig. S2). Using H35L, a mutated α-toxin that is unable to form stable pores, we observed no change in the amount of soluble CX₃CL1, suggesting the induction of CX₃CL1 is dependent on α-toxin’s pore-forming properties (Fig. 2G).

Alpha-toxin-induced CX₃CL1 release is blocked by inhibiting ADAM10 activity

It has been suggested that the sheddase activity of ADAM10 can increase after calcium influx as a consequence of α-toxin-mediated pore formation (17), thus leading to a loss of tissue integrity due to an increased rate of E-cadherin cleavage by ADAM10 (13). ADAM10 is also known to be responsible for the shedding of membrane-bound CX₃CL1 in the steady state and during inflammation (Fig. 3A) (22). Therefore, we investigated the role of ADAM10 in S. aureus-induced CX₃CL1 release by using an inhibitor (GI254023X) of ADAM10 activity. Pretreatment of lung tissue models with GI254023X blocked the release of CX₃CL1 induced by either α-toxin and NP796 supernatant (Fig. 3B). In addition, in an in vivo murine model of S. aureus pneumonia, we observed reduced CX₃CL1 levels in the BAL fluid of mice treated with GI254023X prior to infection (Fig. S3A). In the lung tissue models, GI254023X prevented the altered expression pattern of CX₃CL1 (Fig. 3C and D). Furthermore, we demonstrated that blocking with anti-α-toxin antibody revealed a similar CX₃CL1 expression pattern in either α-toxin or NP796 supernatant-treated lung model as that of the unstimulated model (Fig. 3C and D). In addition, we analyzed expression of E-cadherin, another factor known to be modulated by α-toxin-mediated ADAM10 activity. The tissue model stimulated with either α-toxin or NP796 supernatant exhibited a loss in E-cadherin expression compared with the unstimulated model (Fig. S3B and C). The pretreatment with GI254023X or anti-α-toxin blocking antibody rescued the loss of E-cadherin expression in both α-toxin and NP796 supernatant-treated lung model (Fig. S3B and C).

Fig 3.

Alpha toxin-induced CX₃CL1 release acts via ADAM10. (A) Schematic drawing of the proposed pathway through which CX₃CL1 is released from lung tissue exposed to α-toxin stimulation, and the proposed action of an anti-α-toxin antibody, MEDI4893*, or the ADAM10 inhibitor, GI254023X. (B) CX₃CL1 levels in culture supernatants of unstimulated lung models (cell culture media) (black bars), or of lung models pre-treated with 10 µM of ADAM10 inhibitor, GI254023X (gray bars), 2 h post-treatment, the models were stimulated with α-toxin (100 ng/ml) or NP796 bacterial culture supernatant (1:100), or DMSO (GI254023X diluent). ELISA determination of CX₃CL1 levels in supernatants of lung models was performed after 24 h of stimulation. (C) Immunofluorescence staining for CX₃CL1 (green) and cell nuclei (blue, DAPI) in sectioned lung model stimulated with S. aureus α-toxin (100 ng/mL) or NP796 bacterial culture supernatants (1:100), in the presence of either MEDI4893* (200 µg/mL) or GI254023X (10 µM) or unstimulated models (cell culture media). (D) Measurement of the spatial re-distribution (ratio of apical to basolateral mean fluorescence intensity) of CX₃CL1 in lung models stimulated with α-toxin suspensions (100 ng/mL) or NP796 bacterial culture supernatants (1:100), alone (untreated, circle) or treated with anti-α-toxin antibodies (200 µg/mL, MEDI4893*, open square), or the ADAM10 inhibitor, (10 µM, GI254023X, closed square). All experiments were performed at least three times. In panel B, the bars show the mean ± SD of three individual experiments. In panel C, the figure shows data from one representative experiment. In panel D, data are presented as the mean value ± SEM. The scale bars in images equal 100 µm. The statistical significances were calculated by ordinary two-way ANOVA, with Sidak’s multiple comparisons test in panel B and uncorrected Fisher’s LSD in panel D.

To assess whether ADAM10 metalloprotease enzyme activity and CX₃CL1 release are a consequence of α-toxin-mediated epithelial cell lysis (16HBE), we performed dose-response experiments where 16HBE cells were stimulated with increasing concentrations of α-toxin. Using the LDH release assay, the cytotoxicity (15–10%) was weak at 100 ng/mL, whereas it increased strongly at higher α-toxin concentrations (Fig. S4A). Furthermore, ADAM10 metalloprotease enzyme activity was observed already at α-toxin concentrations from 50 ng /mL or higher (Fig. S4B). The release of CX₃CL1 was detected after stimulation with a concentration of α-toxin at 100 ng/mL or higher (Fig. S4C). In all three assays, the α-toxin-mediated effects were inhibited with either the anti-α-toxin antibody MEDI4893* or the ADAM10 inhibitor GI254023X at all concentrations tested (Fig. S4A through C). Furthermore, α-toxin at concentrations above 1,000 ng/mL results in cell lysis despite the presence of the ADAM10 inhibitor GI254023X. These findings show that even lower α-toxin concentrations trigger ADAM10 activity and subsequent CX₃CL1 release; however, there is a clear link between increased α-toxin concentration and cytotoxicity, as well as CX₃CL1 release. In the next step, we aimed to determine whether the increased release of CX₃CL1 was specific to S. aureus α-toxin or if cytotoxins from other bacteria could elicit a similar effect. To this end, we performed dose–response experiments in which 16HBE cells were stimulated with increasing concentrations of four different bacterial pore-forming toxins: α-toxin, δ-toxin, streptolysin O, pneumolysin, and a non-cytolytic mutant of pneumolysin (pneumolysin^C428G). LDH release assays showed that pneumolysin and streptolysin O exhibited cytotoxicity profiles similar to that of α-toxin, with modest cytotoxicity (10–20%) at 100 ng/mL, which increased substantially at higher toxin concentrations (Fig. S4D). In contrast, δ-toxin and pneumolysin^C428G induced only low levels of cytotoxicity, even at the highest concentrations tested (Fig. S4D). CX₃CL1 release was detectable following α-toxin stimulation starting at 100 ng/mL, whereas pneumolysin- and streptolysin O-induced CX₃CL1 release was observed only at concentrations of 500 ng/mL or higher (Fig. S4E). Delta-toxin induced CX₃CL1 release only at 1,000 ng/mL, whereas pneumolysin^C428G did not induce any detectable CX₃CL1 release at any tested concentration (Fig. S4E). Furthermore, the relationship between cytotoxicity and CX₃CL1 release was examined by comparing the effects of α-toxin with those of the other toxins individually (Fig. S4F and G). We observed that detectable levels of CX₃CL1 release were only induced by pneumolysin, δ-toxin, and streptolysin O at concentrations associated with higher cytotoxicity (<20%). In contrast, α-toxin induced CX₃CL1 release at both cytotoxic (lytic) and non-cytotoxic (sub-lytic) concentrations (Fig. S4F and G).

Alpha-toxin-induced CX₃CL1 release results in monocyte migration

Histological and immunohistochemical analysis of S. aureus-infected lung tissue showed evidence of inflammation with substantial infiltration of immune cells, including CD14-positive cells (Fig. 4A; Fig. S5). Large amounts of bacteria were detected in lung tissue samples derived from S. aureus-infected patients with lung involvement as evident by a positive Brown-Brenn staining (Fig. 4A). Immunostaining for CD14 was prominent in areas with bacteria (Brown-Brenn positive) and coincided with cell-associated as well as more diffuse CX₃CL1 staining patterns (Fig. 4A; Fig. S5). CD14 and CX₃CL1 were also relatively more prominent in the tissue of S. aureus-infected patients compared with control tissue from individuals with other lung conditions (Fig. 4A and B; Fig. S5). As soluble CX₃CL1 is a known chemotactic factor for monocytes (22), and to test whether the CX₃CL1 shedding in the lung model coincides with enhanced migration of monocytes, we introduced blood monocytes into the lung models and monitored their motility with confocal live imaging following stimulation of the models with α-toxin or NP796 supernatant. Both α-toxin and NP796 stimulation induced increased motility of monocyte-derived cells within the lung models (Fig. 5A). To further test whether CX₃CL1 released from stimulated lung models was responsible for the observed monocyte migration, we performed transwell migration assays using stimulated lung model supernatants on monocytes and neutrophils pre-treated with either a CX₃CL1 receptor (CX₃CR1) antagonist (AA-1-2008) or vehicle control. Both monocytes and neutrophils migrated in response to supernatants from α-toxin-stimulated lung models; however, addition of the CX₃CR1 antagonist (AA-1-2008) effectively reduced monocyte, but not neutrophil, migration to baseline levels (Fig. 5B; Fig. S6). Furthermore, CX₃CL1 added to cell culture media induced CX₃CR1-dependent monocyte migration (Fig. 5B). However, we did not observe any CX₃CR1-dependent migration of neutrophils given the same treatment (Fig. S6). Together, this suggests that S. aureus stimulation of lung tissue leads to the release of chemokines that attract both neutrophils and monocytes, with CX₃CL1 as the major chemokine contributing to monocyte chemotaxis under these conditions.

Fig 4.

Histological and immunohistochemical qualitative analysis of S. aureus respiratory tract infection. Lung tissue biopsies were stained with Brown-Brenn reagents (gram bacteria), anti-CD14, or anti-CX₃CL1 antibodies. (A) Immunohistochemical analysis for gram⁺ S. aureus (Brown-Brenn, blue), CD14 (brown), and CX₃CL1 (brown) of whole biopsy sections (top row) from an S. aureus-infected patient, with higher magnifications (20×, bottom row) of boxed areas. For the S. aureus-stained tissue, a 40× magnification of one selected area in the 20× image is shown (insert). CD14- or CX₃CL1-positive cells within this area of consecutive sections are indicated by arrows. The scale bars in images equal 5 mm (top row) or 100 µm (bottom row). (B) Brown-Breen staining and immunohistochemical analysis for CD14 (brown) and CX₃CL1 (brown) of whole biopsy sections (top row) from a non-infected fibrosis patient, with higher magnifications (20×, bottom row) of boxed areas. The scale bars in images equal 5 mm (top row) or 100 µm (bottom row).

Fig 5.

Monocyte migration in response to α-toxin-induced release of CX₃CL1. (A) Live imaging and total displacement determination of monocyte-derived macrophage-like cells in lung tissue models stimulated with α-toxin (100 ng/mL) or NP796 bacterial culture supernatant compared to unstimulated models (cell culture media) as indicated. (B) Migration of human blood monocytes for 2 h in a transwell assay in response to supernatants from unstimulated lung model (cell culture media), lung models stimulated for 24 h with α-toxin (100 ng/mL), or medium supplemented with recombinant CX₃CL1 (10 ng/mL), in the absence (DMSO, the diluent of AA-1-2008, black bars) or in the presence of the CX₃CL1 receptor antagonist (AA-1-2008, 10 nM, gray bars). Pre-treatment with the CX₃CR1 antagonist (AA-1-2008, 10 nM) or DMSO (vehicle) was for 2 h prior to the migration assay. (C) Migration of human blood monocytes for 2 h in a transwell assay in response to supernatants from lung models stimulated with α-toxin (100 ng/mL) or NP796 bacterial culture supernatant (1:100), treated with an isotype control-matched antibody (black bars, 200 µg/mL) or anti-α-toxin antibodies (striped bars, 200 µg/mL, MEDI4893*). Supernatants from anti-α-toxin antibody-treated conditions were supplemented with recombinant CX₃CL1 (10 ng/mL, blue bars). (D) Migration of human blood monocytes for 2 h in a transwell assay in response to supernatants from unstimulated lung models (black bars, DMSO GI254023X diluent in cell culture media), or of lung models pre-treated with the ADAM10 inhibitor (gray bars), GI254023X (10 µM) and then stimulated for 24 h with α-toxin (100 ng/mL) or NP796 bacterial culture supernatant (1:100), or medium from GI254023X-treated models supplemented with recombinant CX₃CL1 (10 ng/mL, blue bars). (E, F) Live imaging of monocyte-directed migration toward a CX₃CL1 gradient in a collagen matrix, total displacement (E) and velocity (F) for 2 h, is shown. In the migration chambers, CX₃CL1 (10 ng/mL) was added on both sides (black), excluded (gray), or added on one side (blue). Statistical significances were calculated by ordinary one-way ANOVA, with Tukey’s multiple comparisons test with a single pooled variance in panels A, C, and D, ordinary two-way ANOVA, with Bonferroni’s multiple comparisons test in panel B, and Kruskal-Wallis, with Dunn’s multiple comparisons test in panel F.

Supernatants from α-toxin- or NP796-stimulated lung models pre-treated with either anti-α-toxin-neutralizing antibodies or an ADAM10 inhibitor failed to induce comparable levels of monocyte chemotaxis to the control supernatants (Fig. 5C and D). The addition of exogenous CX₃CL1 to supernatants from lung models pre-treated with ADAM10 inhibitor or anti-α-toxin antibodies restored the migratory capacity of the monocytes in the transwell migration assay (Fig. 5C and D). To test whether CX₃CL1 also induces monocyte migration in a collagen matrix, we performed 3D migration assays in a chamber that accommodates stable chemokine gradients for up to 48 h. There was a clear preferential migration of monocytes toward CX₃CL1 when added to one side of the µ-slide chamber (Fig. 5E and F; Video S1). These data suggest that soluble CX₃CL1 released by α-toxin-stimulated lung models is both necessary and sufficient for the chemotaxis of human monocytes.

CX₃CL1 induces CD83 upregulation and impaired phagocytic killing capability of monocytes

To investigate S. aureus-induced responses beyond migration, we stimulated monocytes with lung model-derived supernatants containing CX₃CL1 and analyzed expression of the co-stimulatory molecules CD83, CD86, and CD80 and the co-inhibitory molecule PD-L1, all of which are typically found on human myeloid cells. Monocytes were identified as described in Figure S7A, and the analyses demonstrated that lung model supernatants stimulated with α-toxin or NP796 induced the upregulation of CD83 without a concomitant upregulation of CD86 or CD80 (Fig. 6A, B and D); all membrane proteins of the immunoglobulin receptor superfamily are involved in the regulation of antigen presentation (27). The CD83 effect was reversed when α-toxin or ADAM10 was blocked prior to stimulation of the lung model (Fig. 6A). Upregulation of CD83 was also seen when CX₃CL1 was added to unstimulated lung model supernatants, suggesting CX₃CL1-induced alteration of co-stimulatory molecules on monocytes (Fig. 6A). Alpha-toxin-stimulated lung model supernatants also induced a downregulation of CD86 as well as an upregulation of PD-L1 (Fig. 6B and C), a protein belonging to the family of immune checkpoint molecules and a co-inhibitory receptor regulating T cell activation. The alterations in CD86 and PD-L1 surface expression were reversed when α-toxin or ADAM10 was blocked during stimulation with supernatants from α-toxin-stimulated lung models (Fig. 6B and C). Unstimulated lung model supernatants spiked with CX₃CL1 or NP796-stimulated lung model supernatant had no detectable effect on PD-L1 surface expression (Fig. 6C). Furthermore, the changes in CD80 expression observed with NP796-stimulated lung model supernatants could not be reproduced with CX₃CL1-spiked lung model supernatants (Fig. 6D).

Fig 6.

CX₃CL1 modulates expression of co-stimulatory/inhibitory molecules. (A–D) Monocytes were stimulated or left unstimulated for 24 h and subsequently processed for flow cytometry analysis. MFI for CD83 (A), CD86 (B), PD-L1 (C), and CD80 (D) on HLA-DR⁺ CD14⁺ monocytes cultured with supernatants from unstimulated lung models (black bars, cell culture media), or unstimulated lung model supernatants spiked with 10 ng/mL CX₃CL1 (blue bars), or lung models stimulated with α-toxin (100 ng/mL) or NP796 bacterial culture supernatant (1:100) (black bars) as indicated. For CX₃CL1 stimulation of monocytes, CX₃CL1 was added at 10 ng/mL to unstimulated lung model supernatants (blue). Lung model supernatants used to stimulate monocytes were also generated by treating the α-toxin suspension or NP796 supernatant with anti-α-toxin antibodies (striped bars, 200 µg/mL, MEDI4893*), or by treating the lung model with the ADAM10 inhibitor (gray bars, GI254023X, 10 µM) prior to stimulation with either α-toxin or NP796 supernatant. The bars show the mean ± SEM of 8 independent donors, and statistical significances were calculated by the Wilcoxon signed-rank test and the Friedman test, uncorrected Dunn’s test.

To study the effects of CX₃CL1 on monocytic anti-microbial capabilities, we infected CX₃CL1-pretreated monocytes with S. aureus and quantified phagocytic uptake as well as killing of intracellular bacteria. We used the two S. aureus strains, Cowan I and LE2332, instead of NP796, since this strain caused rapid lysis of the monocytes. Intracellular bacteria were identified using imaging flow cytometry (ImageStreamX) and a GFP-expressing Cowan I strain (Fig. 7A). We designated bacteria residing inside the monocytes after 1 h as phagocytosed. The intracellular killing of GFP-producing Cowan I was estimated by calculating the percentage of bacteria that were detectable 4 h post-infection. Monocytes pre-treated with CX₃CL1 exhibited a decreased ability to kill intracellular bacteria, whereas their ability to phagocytose remained unchanged (Fig. 7B). To verify that the GFP signal detected by imaging flow cytometry represented live bacteria, we lysed the monocytes and plated the bacteria. The data confirmed similar uptake by monocytes and reduced intracellular bacterial killing by monocytes pre-treated with CX₃CL1 (Fig. 7C). Pre-treatment with lung model supernatants from NP796-stimulated models yielded similar results to CX₃CL1-pretreated monocytes with respect to CFU counts, and the effect was abrogated by blocking α-toxin or ADAM10 prior to stimulation of the models (Fig. 7D). The decreased ability to kill phagocytosed bacteria in the presence of CX₃CL1 was consistent with a decrease in both ROS and NO production (Fig. 7E and F; Fig. S8). Taken together, we observed phenotypic and functional changes that suggest impaired monocyte function following exposure to CX₃CL1 as a consequence of ADAM10-mediated activity induced by α-toxin.

Fig 7.

CX₃CL1 modulates monocyte function. (A–C) Monocytes stimulated with CX₃CL1 (10 ng/mL) or left unstimulated were infected with S. aureus Cowan I-GFP, fixed and stained with Draq5, and analyzed by imaging flow cytometry or CFU assay. (A) Representative images showing monocytes containing bacteria. (B) Image flow quantification of the proportion of Cowan I-GFP⁺ monocytes after 1 h (left) and after 4 h (right). (C) Percentage phagocytosed Cowan I or LE2332 (1 h, left) and percentage killing of phagocytosed Cowan I or LE2332 (4 h, right) after stimulation with media (white), CX₃CL1 (10 ng/mL, blue). The bars show the mean ± SEM of four independent donors. (D) Percentage phagocytosed Cowan I or LE2332 (1 h, left) and percentage killing of phagocytosed Cowan I or LE2332 (4 h, right) after stimulation with lung model supernatants as indicated. The lung model supernatants were obtained by stimulating lung models for 24 h with NP796 bacterial culture supernatant alone (checkered, 1:100), NP796 bacterial culture supernatant treated with MEDI4893* (gray striped, 200 µg/mL), or NP796 bacterial culture supernatant in the presence of GI254023X (gray, 10 µM). (E, F) Monocytes stimulated with CX₃CL1 (10 ng/mL) for 18 h or left unstimulated were infected with S. aureus strain LE2332 (MOI 1) and analyzed for ROS and NO production. Representative histograms (E) of ROS and NO production, and representative bar graphs (F) of ROS and NO production by LE2332-infected monocytes. The bar graphs show the mean ± SEM of five independent donors. In panels B–D, data were presented as whisker plots with a box indicating the interquartile range and error bars indicating the highest and lowest values of four independent donors, and statistical significances were calculated by the Mann-Whitney test in panels B and C and the Kruskal-Wallis with uncorrected Dunn’s test in panel D. In panel F, statistical significances were calculated by Wilcoxon matched-pairs signed-rank test.

DISCUSSION

In this study, we investigated the mechanism and consequences of S. aureus-associated release of CX₃CL1, a unique membrane-bound chemokine that requires metalloprotease processing before secretion. We found that CX₃CL1 levels increase both locally and systemically in patients with S. aureus respiratory tract infection. Using a 3D human in vitro staphylococcal pneumonia model, we show that α-toxin-induced activation of ADAM10 as well as α-toxin-induced cytotoxicity leads to increased CX₃CL1 release, which enhances monocyte motility and impairs intracellular killing capacity by monocytes. The reduced bactericidal phenotype was accompanied by an increased CD83 and a decreased CD86 surface expression. The results imply that α-toxin alters the inflammatory milieu by which CX₃CL1 affects immune cell functions beyond cellular adhesion and migration which contributes to S. aureus intracellular survival and pathogenesis.

S. aureus toxins interact with host tissue constituent cells as well as immune cells. To fully appreciate the role of S. aureus toxins in human disease, studies of S. aureus infections have to take into consideration the host specificities as well as the cellular and molecular interactions occurring within the relevant tissue. In this study, we found elevated levels of CX₃CL1 in patients with S. aureus infections—especially in infections involving the respiratory tract. Exploiting the human lung model enabled us to study the mechanisms by which S. aureus triggers CX₃CL1 release in human-derived lung epithelial cells. The blocking effect of the anti-α-toxin-specific antibody demonstrated that it was specifically α-toxin among the various S. aureus toxins that led to the redistribution and release of CX₃CL1 in the lung model. Furthermore, using a mutated α-toxin variant, our results support the dependency of α-toxin’s pore-forming activity in the activation and release of CX₃CL1 from the membrane. In addition to α-toxin, other S. aureus pore-forming toxins, for example, Panton-Valentine leukocidin (PVL) and γ-hemolysins, may also affect signaling pathways in the respiratory tract (28–30). However, the availability of receptors for the S. aureus pore-forming toxins can explain cellular and tissue tropism; ADAM10 is highly expressed on epithelial cells, whereas the receptors for PVL and γ-hemolysins are predominantly found on human myeloid cells. Whether the different pore-forming toxins of S. aureus exhibit coordinated activities to thwart host defenses and establish a successful infection needs to be investigated further. Additionally, by using cytotoxins from other bacteria, such as pneumolysin and streptolysin O, we demonstrate that CX₃CL1 release is not solely mediated by ADAM10 activation but can also be triggered by increased cellular toxicity. Notably, α-toxin induced a higher level of CX₃CL1 release in epithelial cells compared with the other pore-forming toxins, and α-toxin induced CX3CL1 release also at a non-cytotoxic concentration. This differential effect of α-toxin may result from a synergistic effect of inducing ADAM10 metalloproteinase activity and cytotoxic properties. To encompass the potential effects of various cells on tissue responses during toxin exposure, models with extended cell complexity need to be developed. This is particularly relevant considering that these toxins, initially recognized for their ability to disrupt eukaryotic membrane barriers and cause target host cell lysis, are now understood to also exert subtle changes in cell activity and host physiology, even at sub-lytic concentrations (9, 31, 32). This notion is in line with our results demonstrating that although there is a clear association between the degree of α-toxin-mediated cytotoxicity and CX₃CL1 release, the release was noted even at sub-lytic toxin concentrations.

The release of CX₃CL1 can be mediated by several different metalloproteinases, including ADAM10 and ADAM17 (22, 33–35), and here, we show for the first time that this ADAM10-CX₃CL1 axis is triggered by S. aureus virulence factors, most notably α-toxin. As ADAM10 is known to cleave membrane-bound CX₃CL1 and also serves as the main receptor for α-toxin, we examined the effect of the ADAM10-inhibitor GI254023X on α-toxin-induced CX₃CL1 release. We observed that both GI254023X and the anti-α-toxin antibody blocked the release of CX₃CL1 in our toxin-stimulated in vitro model. Since ADAM10 and ADAM17 share substrates, including CX₃CL1 (36), we cannot rule out the involvement of ADAM17 in the release of CX₃CL1. However, GI254023X has a 1,000-fold higher affinity for ADAM10 and is generally regarded as an ADAM10-specific inhibitor. Therefore, the effect mediated by GI254023X should, to a high degree, be attributed to the loss of ADAM10 activity.

Although it cannot be excluded that freezing as well as the timing of plasma collection may impact the levels of CX₃CL1 detected, the data indicate that S. aureus infection with lung involvement leads to particularly high CX₃CL1 levels both locally and systemically. In line with the reasoning of a lung-infection contribution, we observed increased CX₃CL1 levels in patients with confirmed S. aureus respiratory infection, whereas a large number of plasma samples from our retrospective cohort of critically ill patients with bloodstream S. aureus infection but without respiratory tract involvement had undetectable levels of CX₃CL1. We did not find an association with co-morbidities or with the severity of disease. This observation is in contrast to a study by Pachot et al. in which a population of critically ill patients with septic shock was examined, and where an increase in CX₃CL1 and positive correlation with disease severity was observed (37). In this context, it is interesting that as many as 50% of the patients included in the study by Pachot et al. had pulmonary infection listed as the primary site of infection. Because the study by Pachot and colleagues did not take etiology into account, direct comparisons must be made with caution. However, the high percentage of lung focus in that study is in line with our data, suggesting that the measured increase in CX₃CL1 might be particularly pronounced due to respiratory tract infection caused by S. aureus.

Alpha-toxin-stimulated lung models induced migration in both monocytes and neutrophils. In this context, CX₃CL1 was one of the most significant chemokines for monocyte migration, as evident by the finding that CX₃CR1 receptor inhibition decreased monocyte migration to baseline levels. Inhibiting CX₃CR1, on the other hand, had no effect on neutrophil migration, suggesting that other chemokines released from lung tissue are important for neutrophil migration in S. aureus infection. The independence of neutrophil migration from CX₃CL1 can probably be explained by CXCL8, a prototypical neutrophil chemoattractant, released from S. aureus and α-toxin-stimulated lung models (6). The fact that CX₃CL1 plays a minor role in human neutrophil migration was further strengthened by the observation that the addition of CX₃CL1 did not act as a chemoattractant for neutrophils. This is also in line with the fact that neutrophils express relatively low levels of CX₃CR1 (38), whereas human monocytes are highly positive for CX₃CR1 (39).

In addition, we observed that CX₃CL1 induced phenotypic and functional changes in monocytes, including an upregulation of CD83 expression. Although often used as a marker for activated dendritic cells (DCs), soluble CD83 has been shown to interfere with (DC)-induced T cell stimulation (40). It has also been suggested that monocytes, in contrast to DCs, produce and release CD83, which may have immune suppressive functions by interacting with membrane-bound CD83 on activated DCs (41, 42). In further support of a suppressive phenotype, we also observed a reduction in the activation marker, CD86, on the surface of CX3CL1-treated monocytes. Notably, the CX₃CL1-treated monocytes revealed a reduced capacity to kill phagocytosed S. aureus. This functional impairment was linked to a reduction of both NO and ROS. These effector molecules have a direct antimicrobial effect, and the products generated during the respiratory burst may also act as second messengers in redox signaling and cellular activation processes (43, 44).

It is known that CX₃CL1 is constitutively expressed in epithelial and endothelial tissues and has functions related to cell anchoring and adhesion (39). Furthermore, ADAM10 constitutively sheds CX₃CL1 at low levels (22), which strongly suggests that ADAM10 and CX₃CL1 have functions related to maintaining tissue homeostasis. Our observations of released CX₃CL1 and its effects on monocytes after α-toxin stimulation suggest that S. aureus is exploiting CX₃CL1’s homeostatic functions to promote bacterial survival. There is accumulating evidence that the CX₃CL1-CX₃CR1 axis is involved in various diseases, including atherosclerosis (25) and rheumatoid arthritis (45), as well as pulmonary disorders (46, 47). In the respiratory tract, CX₃CL1 has also been proposed to support the migration of monocytes in interstitial lung disease (48) and to promote pulmonary fibrosis by attracting macrophages with a pro-fibrotic phenotype (49). The CX₃CL1-CX₃CR1 axis is also suggested to be involved in the pathogenesis of critically ill and septic shock patients, as evidenced by a link between reduced CX₃CR1 mRNA expression in circulating leukocytes and increased mortality (24, 37). Our study is the first to show that S. aureus induces shedding of CX₃CL1 by lung epithelial cells via α-toxin-mediated cytotoxicity and activation of ADAM10, with the potential to modulate immune responses locally and impair monocytes’ antimicrobial function, thereby contributing toward S. aureus pathogenesis and survival during respiratory infections.

MATERIALS AND METHODS

Patients and samples

Lung tissue, bronchioalveolar lavage (BAL), tracheal aspirates, and blood of patients were collected from patients at the University Hospital Zurich, Switzerland, after informed consent was obtained. All blood and tracheal aspirate samples were processed and immediately analyzed for CX₃CL1. Collection of patient samples was done in accordance with the Declaration of Helsinki and performed with permission from the Regional Ethics Board in Zurich, Switzerland.

Patients with S. aureus bacteremia were recruited at the Department of Infectious Diseases, Örebro University Hospital, Sweden, after signed informed consents were obtained. The demographic data, together with the comorbidities, are provided in Table S1. Sepsis was defined according to the Sepsis-3 definition. Blood samples were collected 1 or 2 days after hospital admission, that is, when blood culture signaled positive, and further processed for the collection of plasma for freezing and storage at −80°C. The study was done in accordance with the Declaration of Helsinki and performed with permission from the Regional Ethics Board of Uppsala, Sweden.

Blood samples from healthy volunteers (controls) were collected in University Hospital Zurich or in the Department of Clinical Immunology and Transfusion Medicine at the Karolinska University Hospital and were further processed for isolation of monocytes and collection of plasma. Monocytes were analyzed fresh, and plasma was analyzed fresh or after freezing and storage at −80°C. All healthy volunteers were included after providing signed informed consent. Blood collections were performed in accordance with the Helsinki declaration and with permission from the Regional Ethics Board, in Stockholm, Sweden, or the Cantonal Ethics Board in Zurich, Switzerland.

Bacterial strains

Clinical S. aureus isolates used in this study were collected from the pleural fluid of a necrotizing pneumonia case (NP796) and one lung empyema case (LE2332) at Global Hospitals, Hyderabad, India (6). USA300 (11358) (50) and Cowan I (ATCC 12598) (51) were used as reference strains. In addition, four clinical S. aureus strains (1408, 1439, 1441, and 1525) from patients with S. aureus respiratory tract infection were used in the hemolysis assay. S. aureus Cowan I (ATCC 12598) was transformed with the reporter plasmid pCN56 encoding the fluorescence-activated cell sorting-optimized mutant version gfpmut2 of the green fluorescent gene of A. victoria encoding the GFP reporter and a constitutive blaZ promoter (52). The blaZ promoter was integrated into the multiple cloning site of pCN56 by using the restriction enzymes SphI and BamH1. Before the transformation of S. aureus Cowan I, the plasmid was multiplied in E. coli IM30B (53) to bypass restriction systems.

Lung tissue model setup

The lung tissue model was set up as previously described (54). Briefly, lung tissue models were established using 8 × 10⁴ MRC-5 cells (human lung fibroblast cell line) at passage <30 cultured in a collagen matrix for 7 days, followed by the addition of 1.2 × 10⁵ 16HBE14o cells (human bronchial epithelial cell line) per model. Epithelial cells were confluent after 3 days and were then air-exposed for 7 days. Complete DMEM (DMEM supplemented with 10% heat-inactivated fetal calf serum, 2 mM L-glutamine, 10 mM HEPES) was changed every 2–3 days. Lung tissue models were exposed to stimuli in a 50 µL droplet on the apical side. For models containing monocyte-derived cells, monocytes were isolated as described, stained with PKH26 Red Fluorescent Cell Linker Kit for General Cell Membrane (Sigma-Aldrich, Saint Louis, MO, USA) according to the manufacturer’s instructions, and added at a concentration of 5 × 10⁵ monocytes per model the day before adding epithelial cells.

Isolation of monocytes and neutrophils

Human peripheral blood mononuclear cells (PBMCs) were isolated from buffy-coated or whole blood from healthy volunteers, using Lymphoprep (Axis-Shield, Oslo, Norway) gradient centrifugation according to the manufacturer’s instructions. Monocytes were isolated from PBMCs using the EasySep Human monocyte enrichment kit without CD16 depletion (StemCell Technologies, Vancouver, British Columbia, Canada) according to the manufacturer’s instructions. Human neutrophils were isolated from fresh whole blood obtained from healthy volunteers using Polymorphprep (Axis-Shield, Oslo, Norway) by performing density gradient separation according to the manufacturer’s instructions.

Hemolysis assay

For the hemolysis assays, human red blood cells (RBCs) were isolated from heparinized venous blood obtained from healthy adult volunteers. RBCs were washed twice in sterile saline (0.9% NaCl) and centrifuged at 500 × g for 10 min. RBCs were diluted to 5% in PBS. 100 µL of RBCs was incubated with 100 µL of bacterial culture supernatant in a 96-well plate for 30 min at 37°C. A dilution series of bacterial culture supernatant was prepared using PBS, with 1:1 being the highest concentration. Plates were centrifuged for 5 min at 500 × g, and the supernatants were transferred to a sterile 96-well plate; RBC lysis was evaluated by determining the absorbance at 415 nm. PBS alone and H₂O were used as negative and positive controls, respectively.

Bacterial supernatant preparation

Bacterial culture supernatants for the stimulation experiments were prepared as previously described (26). In short, the strains were cultured overnight at 37°C in 25 mL casein hydrolysate and yeast extract (CCY) medium. Cell-free supernatants were prepared through centrifugation at 3,350 × g, followed by filter sterilization.

Lung tissue model stimulation

Once established, lung tissue models were stimulated with 1:100 dilutions of bacterial supernatants in PBS or cell culture media. Antibody blocking of α-toxin with MEDI4893* was performed by pre-incubating the bacterial culture supernatants for 1 h before stimulating the tissue models. Similarly, tissue models were pretreated with ADAM10 inhibitor (GI254023X) for 2 h at 37°C before stimulations. All stimulations and treatment of tissue model prior to stimulation were done by adding a 50 µL drop on the apical side of the model. For the unstimulated tissue models, 50 µL of the cell culture media was added to the apical side as a negative control. Prior to cryosectioning, models were submerged in 2M sucrose solution for 90–120 min, followed by cutting out and embedding in O.C.T. (Sakura Finetek Europe, Zoeterwoude, Netherlands) and subsequently frozen at −20°C for 24 h and then transferred to −80°C until used. The cell lines MRC5 and 16HBE14o- were routinely tested for mycoplasma contamination.

Immunostaining and confocal microscopy

Cryosectioning of lung tissue models was performed as previously described (6). Briefly, 10 µm sections were obtained using a MICROM cryostat HM 560 MV (Carl Zeiss, Oberkochen, Germany) and fixed in 2% freshly prepared paraformaldehyde in PBS for 15 min at room temperature. Immunofluorescence staining of lung tissue model sections for confocal microscopy (54) was performed as previously described. Mouse anti-E-cadherin (2 µg/mL, clone HECD-1; Invitrogen, Carlsbad, CA, USA) and mouse anti-CX₃CL1 (2 µg/mL, clone MM0207-8J23; Abcam, Cambridge, United Kingdom) were used to stain the sections of lung tissue models. Specific staining was detected by Alexa Fluor 488-conjugated donkey anti-mouse IgG (3.3 µg/mL) (Molecular Probes, Eugene, OR, USA). Staining was visualized using a Nikon A1 confocal microscope (Nikon Instruments, Tokyo, Japan). The mean fluorescence intensity (MFI for E-cadherin) and apical/basal MFI ratio (for CX₃CL₁) in 4–5 fields per tissue section were determined using Image J analysis software (Fiji).

Histology and immunohistochemistry

Patient tissue was fixed, sectioned, and stained as described previously (55). Patient tissues were first fixed in 4% buffered formalin and then paraffin-embedded (Leica, Muttenz, Switzerland). Two micrometer sections were stained with Brown-Brenn to visualize bacteria in the tissue. Immunohistochemistry staining was used to detect human CD14 and human CX₃CL1 (Abcam; MM0207-8J23) on paraffin-embedded biopsies. Whole-slide scanning and photomicrography were performed with a NanoZoomer 2.0-HT digital slide scanner (Hamamatsu, Houston, TX).

ELISA

The levels of CX₃CL1 were determined according to the manufacturer’s instructions using the Human CX₃CL1/Fractalkine DuoSet ELISA and DuoSet ELISA Ancillary Reagent Kit (R&D Systems, Minneapolis, MN, USA). Supernatants were diluted 1:1 with reagent diluent for all measurements and performed in duplicates. For CX₃CL1, the limit of detection was 0.156 ng/mL. Values below LOD were set to 0.1103 $\left(\mathrm{L}\mathrm{O}\mathrm{D}/\sqrt{2}\right)$. The levels of CD83 (pg/mL) secreted in the culture medium were determined with the Human CD83 DuoSet ELISA kit (R&D Systems, Minneapolis, MN, USA) according to the manufacturer’s instructions. Supernatants were analyzed undiluted, and all measurements were performed in duplicates.

In vivo blocking of α-toxin and ADAM10

Blocking of α-toxin in an experimental in vivo model of S. aureus respiratory tract infection was performed as described in the legend for Fig. S2. S. aureus frozen stock cultures were thawed and diluted to the appropriate inoculum in sterile PBS, pH 7.2 (Invitrogen) (56). Specific-pathogen-free 7- to 8-week-old female C57BL/6 J mice (The Jackson Laboratory, Bar Harbor, ME) were briefly anesthetized and maintained in 3% isoflurane (Butler Schein Animal Health) with oxygen at 3 L/min and infected intranasally. Anti α-toxin MEDI4893* or c-IgG was administered in 0.5 mL intraperitoneally (IP) 24 h prior to infection. Animals were euthanized with CO₂ 24 h post-infection, and BAL fluid was collected for CX₃CL1 measurement. All animal studies were approved by the AstraZeneca Institutional Animal Care and Use Committee and were conducted in an Association for Accreditation and Assessment Laboratory Animal Care (AAALAC)-accredited facility in compliance with U.S. regulations governing the housing and use of animals.

Seven- to 9-week-old female C57BL/6 J mice (Janvier Labs, France) were pre-treated with GI254023X diluted in 0.1 M carbonate buffer to a 0.014 M stock, and ~200 mg kg-1 or PBS with carbonate buffer as a control in a final volume of 100 µL via intraperitoneal injection. Pre-treatment was applied daily for 3–5 days prior to the infection with S. aureus. On the day of the infection, mice were anesthetized with ketamine/xylazine injected intraperitoneally (ketamine 65–90 mg/kg and xylazine 10–13 mg/kg). The anesthetized mice were infected intranasally with the S. aureus strain USA300 (5 × 10⁷ CFU). At the end point, the mice were euthanized with CO₂, and BAL fluid was collected by washing the lungs with PBS for CX₃CL1 measurement. The protocols (ZH050/18) were approved by the institutional animal care and use committee of the University of Zurich, and all experiments were conducted in accordance with the Cantonal Veterinary Office Zurich.

ADAM10 metalloprotease enzymatic activity assay

ADAM10 enzymatic activity was measured on the basis of cleavage of fluorescent substrate using the Mca-P-L-A-Q-A-V-Dpa-R-S-S-S-R-NH2 Fluorogenic Peptide Substrate III (R&D Systems). 16HBE cells were seeded at 50,000 cells per well in a 96-well plate. The cells were washed and stimulated with 10, 50, 100, 250, 500, and 1,000 ng/mL of α-toxin for 2 h in the absence or presence of anti-α-toxin antibodies (200 µg/mL, MEDI4893*) or the ADAM10 inhibitor, GI254023X (10 µM). 16HBE cells were pre-treated with GI254023X for 4 h or α-toxin was preincubated with MEDI4893* for 4 h before stimulations. The ADAM10 enzymatic activity was determined by incubating the cells in 25 mM Tris buffer at pH 8.0 with 10 µM fluorogenic peptide substrate for 30 min at 37°C in a 100 µL final reaction volume. Fluorescence intensity was read on a SpectraMax iD3, Molecular Devices plate reader.

Lactate dehydrogenase (LDH) activity assay

Cytotoxic responses were assessed by measuring LDH release into the tissue culture medium by cells that had been stimulated with bacterial culture supernatants or pure toxins. The cells were stimulated with 10, 50, 100, 250, 500, and 1,000 ng/mL of recombinant toxins, including α-toxin (Sigma-Aldrich), streptolysin O (Abcam), δ-toxin (IBT bioservices), pneumolysin, and a non-cytolytic mutant of pneumolysin (pneumolysin^C428G) (57) for 2 h. LDH was measured using the Promega CytoTox 96 Nonradioactive Cytotoxicity assay kit according to the manufacturer’s protocol. The absorbance was read at 490 nm using a SpectraMax iD3, Molecular Devices plate reader. Percent cytotoxicity was determined in relation to the lysis control.

Live imaging

For live imaging, the models were prepared as previously described (58). Briefly, the models were kept for 4 days post-air exposure and fixed upside down on a 6-well No. 0 coverslip 10 mm glass diameter uncoated plate (MatTek, Ashland, MA, USA). Models were kept in a humidified CO₂ chamber during the whole imaging process, and images were taken every 20 min for approximately 16 h. For each model, a z-stack of approximately 120 µm around the epithelial layer was acquired with a 547 nm laser. Image processing and analysis of reconstructed three-dimensional Z-stack was done using Imaris (Bitplane, Andor Technology, Belfast, United Kingdom). Cells were tracked using identical settings across all stimulations with the ImarisTrack module.

Transwell migration assay

The chemotactic effect of CX3CL1 (10 ng/mL) and lung conditioned media (supernatants collected from stimulated lung tissue models) was assessed using a Transwell migration assay. The lung conditioned media (diluted 1:20 in RPMI 1640 media) were added to the outer chamber of 24-well transwell plates (3 µm Costar, Corning Inc., Corning, NY, USA). Neutrophils and monocytes were seeded at 3 × 10⁵ to 5 × 10⁵ cells/well in the upper chamber of a 24-well Transwell plate and incubated for 2 h at 37°C in a CO₂ incubator. The CX₃CR1 antagonist, AA-1-2008, was added (10 nM) to the cells 2 h prior to transmigration assays. To quantify the cellular migration, the cells were collected and mixed with a specified amount of Count Bright absolute counting beads (10,000–20,000 beads/sample, Molecular Probes). Migration counts were obtained by using a BD LSRII Fortessa cell analyzer (BD Bioscience, Franklin Lakes, NJ, USA) or Attune NxT (Thermo Fisher, Waltham, MA, USA) with FSC/SSC gating on neutrophils/monocytes and beads, respectively, to obtain the total number of cells that had migrated. FlowJo software version 10.5.3 (Tree Star, Ashland, OR, USA) was used for flow cytometry analyses.

Directed migration in µ-slide chemotaxis assay

For the migration chamber experiments, freshly isolated monocytes stained with PKH26 (Sigma-Aldrich) according to the manufacturer’s instructions were resuspended in a collagen mixture. Collagen mixture was made by mixing 20 µL 10× DMEM (ThermoFisher), 5 µL 1M NaOH, 122 µL H₂O, 3 µL 7,5% NaHCO₃, 50 µL DMEM, and 50 µL 3 mg/mL Collagen I. This was mixed with 4.5 × 10⁵ monocytes in a 50 µL cell suspension, resulting in a final concentration of 1.5 × 10⁶ cells/mL. Then, 6 µL of cell-collagen mixture was allowed to polymerize in the µ-slide chambers (ibidi). Before starting the image-acquiring process, medium with (+, 100 ng/mL) or without (−) CX₃CL1 was added to the side-chambers that were in contact with the polymerized cell-collagen mixture. “+/−” indicates chemokine added to one side-chamber, “+/+” indicates chemokine added in both side-chambers, and “−/−” indicates no chemokines in either side chamber. One image was taken every 5 min, and the µ-slide was kept at 37°C and 5% CO₂ during the whole imaging process (2 h).

Flow cytometry analysis

Cells were washed in FACS buffer (2% FCS, 1 mM EDTA in PBS) and stained with Zombie UV viability dye and fluorochrome-conjugated mAbs specific (all from Biolegend, San Diego, CA, USA) for CD45 (clone HI30, Alexa 700), CD83 (HB15e, Brilliant Violet 421), CD14 (clone M5E2, Brilliant Violet 585), CD56 (HCD56, Brilliant Violet 510), CD19 (HIB19, Brilliant Violet 510), CD3 (OKT3, Brilliant Violet 510), PD-L1 (clone 29E.2A3, Brilliant Violet 605), CD80 (clone 2D10, PE), and HLA-DR (clone Y1/82A, PE/Cy5) for 30 min on ice. Cells were then fixed using Cytofix (BD) for 10 min and resuspended in FACS buffer. For analysis, monocytes enriched from buffy coats were identified using FCS/SSC and then selected for single live cells positive for CD45 and HLA-DR, negative for CD3, CD56, and CD19, and positive for CD14 (Fig. S6A). Samples were analyzed with Fortessa LSRII SORP flow cytometer (BD Biosciences), and data were processed with FlowJo 9.7.6 software (Tree Star, Ashland, OR). The geometric MFI of CD83, CD80, CD86, and PD-L1 on the selected monocytes was then reported.

Phagocytosis and intracellular killing assay

Isolated monocytes were resuspended in complete RPMI media with 2 mM L-glutamine and 5% FCS (Gibco) and seeded in a 24-well cell culture plate at 5 × 10⁵ cells/well. Monocytes were then stimulated with rCX₃CL1-10 ng/mL or lung conditioned medium (1:20 dilutions) for 16–18 h at 37°C in a CO₂ incubator. For the phagocytosis assay, cultured monocytes were infected with S. aureus strains (Cowan I and LE2332) at an MOI of 10 for 60 min. The extracellular bacteria were killed by adding flucloxacillin (10 mg/mL) and lysostaphin (2 µg/mL) and incubated for 20 min. The number of phagocytosed bacteria was determined by lysing monocytes in H₂O, followed by serial dilution of lysates, which were then plated on a Columbia blood agar plate incubated at 37 °C for 18–24  h. The bacterial CFU counts were determined the next day. Percentages of phagocytosed bacteria were analyzed and calculated in relation to the inoculum.

Similarly, for the intracellular killing assay, stimulated monocytes were infected with S. aureus strains at an MOI of 10 for 60 min. After the initial infection of 60 min, extracellular bacteria were killed by adding flucloxacillin (10 mg/mL) and lysostaphin (2 µg/mL). The infected monocytes were further incubated for 4 h with the antibiotic- and lysostaphin-containing RPMI media at 37°C in a CO₂ incubator. Finally, the samples from each well were collected; the cells were lysed and serially diluted in sterile H₂O and then plated on Columbia blood agar plates that were subsequently incubated at 37°C for 18–24  h. The bacterial CFU were determined the next day. The percentage of intracellular surviving bacteria was analyzed and calculated in relation to the invasion (60 min time point).

Assays for ROS and NO detection

Monocytes stimulated with rCX₃CL1-10 ng/mL for 18 h at 37°C in a CO₂ incubator were infected with S. aureus strains (Cowan I and LE2332) at MOI 1 for 90 min ROS and NO production by monocytes was quantified by using CellROX Green (ThermoFisher) and DAF-FAM Diacetate (ThermoFisher), according to the manufacturer’s protocol. For analysis, flow cytometry was used, and monocytes were identified as described in Fig. S6A. Samples were analyzed with an Invitrogen Attune NxT flow cytometer (ThermoFisher Scientific). The geometric MFI of ROS and NO on selected monocytes is reported.

Image flow cytometry

For imaging flow cytometry, monocytes were purified and infected with S. aureus Cowan producing GFP (MOI 10) for 1 h or 4 h as described above. Cells were fixed in 4% paraformaldehyde and resuspended in 25 µL PBS. Draq5 (Invitrogen) was added at a final concentration of 5 µM, and 10⁴ cells were acquired using an Amnis ImageStreamX Mk II system (Luminex, Austin, TX, USA). Using the IDEAS 6.0 software (Luminex), the number of bacteria per cell was analyzed essentially as previously described using the feature Spot count_Peak ([M02, Staph, Bright, 6] [59]). The mean spot count in the monocyte population, reflecting the average number of bacteria per cell, is reported. The intracellular killing of GFP-producing Cowan I was estimated by calculating the percentage of bacteria that were detectable 4 h post-infection.

Statistical analysis

Data were analyzed with GraphPad Prism software v.8.1.1 (GraphPad Software Inc., San Diego, CA, USA), and the statistical methods used are indicated in the respective figure legend. Differences were considered to be statistically significant at P < 0.05. For all statistical analyses, *P < 0.05, **P < 0.01, and ***P < 0.001.

DATA AVAILABILITY

The authors declare that the data supporting the findings of this study are available within the article and its supplemental files or from the corresponding authors on request.

ETHICS APPROVAL

Collection of samples from patients and healthy volunteers was done after informed consent was obtained, and the study was done in accordance with the Declaration of Helsinki and performed with permission from the Regional Ethics Board in Zurich, Switzerland; Regional Ethics Board of Uppsala, Sweden; Regional Ethics Board in Stockholm, Sweden; or the Cantonal Ethics Board in Zurich, Switzerland. All animal studies were approved by the AstraZeneca Institutional Animal Care and Use Committee and were conducted in an Association for Accreditation and Assessment Laboratory Animal Care (AAALAC)-accredited facility in compliance with U.S. regulations governing the housing and use of animals. The mouse protocols (ZH050/18) were approved by the institutional animal care and use committee of the University of Zurich, and all experiments were conducted with approval of the Cantonal Veterinary Office Zurich.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/mbio.02689-25.

Supplemental material. mbio.02689-25-s0001.docx.

Supplementary figures, legends, and table.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.