Differential Induction of Type I and III Interferons by Staphylococcus aureus

Staphylococcus aureus is a leading cause of bacterial pneumonia, and we have shown previously that type I interferon (IFN) contributes to the pathogenesis of this disease. In this study, we screened 75 S. aureus strains for their ability to induce type I and III IFN. Both cytokine pathways were differentially stimulated by various S. aureus strains independently of their isolation sites or methicillin resistance profiles. These induction patterns persisted over time, and type I and III IFN generation differentially correlated with tumor necrosis factor alpha production.

ABSTRACT

Staphylococcus aureus is a leading cause of bacterial pneumonia, and we have shown previously that type I interferon (IFN) contributes to the pathogenesis of this disease. In this study, we screened 75 S. aureus strains for their ability to induce type I and III IFN. Both cytokine pathways were differentially stimulated by various S. aureus strains independently of their isolation sites or methicillin resistance profiles. These induction patterns persisted over time, and type I and III IFN generation differentially correlated with tumor necrosis factor alpha production. Investigation of one isolate, strain 126, showed a significant defect in type I IFN induction that persisted over several time points. The lack of induction was not due to differential phagocytosis, subcellular location, or changes in endosomal acidification. A correlation between reduced type I IFN induction levels and decreased autolysis and lysostaphin sensitivity was found between strains. Strain 126 had a decreased rate of autolysis and increased resistance to lysostaphin degradation and host cell-mediated killing. This strain displayed decreased virulence in a murine model of acute pneumonia compared to USA300 (current epidemic strain and commonly used in research) and had reduced capacity to induce multiple cytokines. We observed this isolate to be a vancomycin-intermediate S. aureus (VISA) strain, and reduced Ifnb was observed with a defined mutation in walK that induces a VISA phenotype. Overall, this study demonstrates the heterogeneity of IFN induction by S. aureus and uncovered an interesting property of a VISA strain in its inability to induce type I IFN production.

INTRODUCTION

Staphylococcus aureus is a versatile bacterium, found as a commensal in 20% to 60% of the population, and it is also responsible for a variety of clinical manifestations, ranging from skin and soft-tissue infections to invasive diseases such as bacteremia, pneumonia, and sepsis (1, 2). Since its emergence in the 1960s, methicillin-resistant Staphylococcus aureus (MRSA) has globally disseminated and has become a major cause of bacterial infections in health care and community settings (3). MRSA strains are often also resistant to several other antibiotic classes, which has significant implications for current and future treatments. Furthermore, individuals colonized with MRSA have an increased risk of subsequent infection and are an important source of person-to-person transmission (4). Notably, in the United States 613,210 S. aureus-related hospitalizations were recorded in 2014, 58% of which were caused by MRSA strains (5).

Type I and type III interferon (IFN) responses are induced during S. aureus infection and contribute to pathogenesis in murine models of acute pneumonia (6–9). In humans, type I IFNs are composed of 17 subtypes, including 13 subtypes of IFN-α, IFN-β, IFN-ε, IFN-κ, and IFN-ω, that all signal through a common receptor, IFNAR (10). Type III IFNs are composed of 4 members, IFN-λ1, IFN-λ2, IFN-λ3, and IFN-λ4, that also all signal through a shared receptor, IFNLR (11, 12). Despite their different receptors, many of the signals leading to type I and type III IFN transcription overlap (13, 14). However, the kinetics of induction of their downstream genes differ. Induction of type I IFN-stimulated genes (ISG) peaks early and then declines, whereas type III IFN induces a more sustained response (15, 16).

Previous studies by our group have shown that type I IFN activation in response to S. aureus is not limited to a single receptor (6, 7). Comparison of the epidemic isolate of MRSA, USA300, along with a clinical isolate identified differential activation of the pathway (7). This difference in cytokine induction was linked to recognition of specific bacterial components, namely, DNA and peptidoglycan, through different pathogen recognition receptors (PRR) (6, 7). It was also observed that a strain with a greater propensity to undergo autolysis induced stronger type I IFN signaling (6). However, the extent of type III IFN signaling by S. aureus strains was not determined.

In this study, we report the heterogeneity of type I and type III IFN induction capacities among S. aureus clinical isolates. We identify and characterize a strain that displays reduced type I IFN induction while preserving type III IFN signaling. We show that the inability of this strain to induce type I IFN is associated with increased vancomycin resistance that reduces bacterial autolysis and protects the bacterium from host cell killing.

RESULTS

S. aureus strains can induce a broad range of interferon levels.

Our prior work identified two distinct S. aureus strains that differentially induce type I IFN production in various cell types (6, 7). To have a better understanding of the global IFN responses induced by S. aureus strains in eukaryotic cells, we measured the type I (IFN-β) and type III (IFN-λ) IFN responses induced by S. aureus USA300 (commonly used in research) and compared them to those induced by a panel of 75 S. aureus clinical isolates. We also monitored tumor necrosis factor alpha (TNF-α) levels as a general readout for inflammation. We observed that the intensity of type I and type III IFN responses varied significantly from strain to strain (Fig. 1A; see also Fig. S1 in the supplemental material). Interestingly, strains that induced a strong cytokine response for one type of interferon did not necessarily induce a significant response for the other type, suggesting differential induction pathways. These data demonstrated that interferon induction is not a conserved response between strains of S. aureus.

FIG 1.

Type I and III interferons are differentially induced in BMDCs by various S. aureus clinical isolates. BMDC were stimulated with a panel of 75 S. aureus strains for 24 h. Cytokines were quantified in the supernatant by ELISA, and concentrations were normalized to cytokine production induced by USA300 stimulation. (A) Heat map showing the mean IFN-β and IFN-λ concentrations induced by S. aureus strains (n = 5 for each strain). (B to D) IFN-λ, IFN-β, and TNF-α levels (respectively) induced by S. aureus strains grouped by their body isolation site (AD site, n = 13; blood, n = 8; nares, n = 11; perianal, n = 6; skin, n = 6; pneumonia, n = 12; wound abscess, n = 2). (E to G) IFN-λ, IFN-β, and TNF-α levels induced by S. aureus strains in relation to their methicillin resistance profile (MRSA, n = 37; MSSA, n = 23). Each dot represents a strain. Data are shown as the pooled means ± standard errors of the means (SEM) from at least 2 independent experiments. Statistical significance was assessed using Kruskal-Wallis test (B to D) or Mann-Whitney test (E to J). *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001.

As bacteria can upregulate or downregulate specific factors to adapt to environmental conditions, we tested if the body site from which clinical strains were isolated could have an impact on the induction of these cytokines. We sorted the clinical strains into 7 distinct categories based on their body isolation sites and compared the levels of IFN-β, IFN-λ, and TNF-α produced by each group (Fig. 1B to D). No statistically significant differences were observed between groups for either cytokine. We also looked at the relationship between clonal complexes and IFN induction but did not find any significant correlation (data not shown). Methicillin resistance has been linked to the production of factors that may affect the general stress response of certain Gram-positive bacteria (4, 17, 18). We separated the clinical isolates based on their methicillin resistance profiles and compared their levels of cytokine induction (Fig. 1E to G). No statistically significant differences were observed between groups for both types of IFN, but methicillin-sensitive S. aureus (MSSA) strains induced more TNF-α than methicillin-resistant S. aureus (MRSA) strains (median concentration of TNF-α induced by MRSA was 0.798 and by MSSA was 1.008; P < 0.01) (Fig. 1G).

While type I and type III IFN are recognized by distinct receptors, their induction relies on many common pathogen recognition receptors and transcriptional regulators (19–23). Therefore, it is expected that their expression is linked. However, as shown in Fig. 1A, strains that induced a strong response for one type of interferon did not necessarily induce a significant response for the other type. We decided to examine these cytokine elicitation patterns in more detail by comparing the induction results for each pair of cytokines throughout the population. TNF-α and IFN-β inductions were positively correlated (Spearman r_s = 0.333, P = 0.009, r_s² = 0.110; Fig. S2A), while TNF-α and IFN-λ induction were negatively correlated (Spearman r_s = −0.405, P = 0.0014, r_s² = 0.164; Fig. S2B). This suggests that strains that induced higher levels of IFN-β also induced a stronger TNF-α response, whereas strains that induced higher levels of IFN-λ induced a lower TNF-α response. Finally, IFN-λ and IFN-β induction patterns did not show any significant correlation (Spearman r_s = −0.197, P = 0.3611, r_s² = 0.014; Fig. S2C). As type I and type III IFN have distinct kinetics of induction (24), we selected 6 representative strains and quantified their patterns of type I and type III IFN induction in bone marrow-derived dendritic cells (BMDC) at different time points (Fig. S2D). After 24 h, IFN-λ and IFN-β induction patterns of the selected strains did not show any significant correlation (Spearman r_s = −0.1857, P = 0.3256, r_s² = 0.034; Fig. S2D), consistent with the previous observation (Fig. S2C). However, after 48 h, a positive correlation of IFN-λ and IFN-β induction patterns was observed (Spearman r_s = 0.488, P = 0.006, r_s² = 0.238; Fig. S2D), corroborating prior experiments (25).

Strain 126 displays a constitutively dampened type I interferon response despite localizing to the same subcellular compartment as USA300.

Our results suggest that over time, the induction of type I and type III interferon is linked. However, in our screen of clinical isolates, we observed that when BMDC were exposed to strain number 126, lower IFN-β levels were produced compared to those of USA300 despite analogous levels of IFN-λ (Fig. S1). This defect could be the result of impaired cytokine induction or degradation. To answer this question, we analyzed the induction of Ifnb mRNA by quantitative reverse transcription-PCR (qRT-PCR) and found that cells stimulated with 126 induced 15-fold less Ifnb (P ≤ 0.01) (Fig. 2A) than USA300. We also investigated if this decrease in type I IFN was due to differences in induction kinetics between the strains. Levels of IFN-β produced by BMDC after incubation with both strains were measured over time (Fig. 2B). USA300 induced stable production of IFN-β for the first 24 h and then dropped by 59% at 48 h. In contrast, IFN-β induction by 126 was significantly reduced compared to that of USA300 (92 to 98% decrease at each time point) (Fig. 2B). Meanwhile, both strains induced similar levels of IFN-λ for 24 h before levels triggered by 126 dropped at 48 h (Fig. 2C). Both strains induced similar levels of TNF-α throughout the experiment, indicating equivalent capacities to induce inflammation (Fig. 2D).

FIG 2.

Strain 126 localizes in the same cellular compartment as USA300 but induces reduced cytokine production. (A) BMDC were stimulated with USA300 (n = 6) or 126 (n = 6). Induction of Ifnb mRNA was analyzed by qRT-PCR after 2 h. (B to D) IFN-β, IFN-λ, and TNF-α levels were quantified by ELISA from the supernatant of BMDC stimulated with either USA300 (n = 5) or 126 (n = 5) after 4, 16, 24, or 48 h. (E and F) AF647-labeled USA300 (n = 6) or 126 (n = 6) was incubated with BMDC. After 5 or 30 min, fluorescence was quantified by flow cytometry (in APC channel). Levels of phagocytosis were determined by the percentage of AF647-positive cells (E) and by the mean fluorescence intensity of the AF647-positive cells (F). Statistical significance was assessed using 2-way RM-ANOVA followed by Sidak’s multiple-comparison test. (G) AF647-labeled S. aureus (green) was visualized inside BMM after 30 min in the presence of AF555 LAMP-1 (red) and DAPI (blue). Data are representative of 2 independent experiments. (H and I) BMDC were incubated with USA300 or 126 in the presence of LysoLive pH-sensor green. After 2 h, fluorescence was observed by flow cytometry. Endosomal acidification was determined by the percentage of FITC-positive cells (H) and by the mean fluorescence intensity of the FITC-positive cells (I). Data are shown as one representative graph from at least 2 independent experiments (control, n = 2; USA300, n = 3; 126, n = 3). All other data are shown as the pooled means ± standard errors of the means (SEM) from at least 2 independent experiments. Statistical significance was assessed using Mann-Whitney test for the qRT-PCR experiment, 2-way RM-ANOVA test followed by Sidak’s multiple-comparison test for the kinetics and phagocytosis experiments, and unpaired Student's t test for the endosomal acidification experiment. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001.

The cellular location of bacteria influences their capacity to activate type I IFN signaling (6, 7, 26). To ascertain if the difference in signaling between the strains was due to differential uptake, we quantified phagocytosis by using flow cytometry. We observed that 126 was phagocytosed 40% more than USA300 after 30 min (P ≤ 0.0001) (Fig. 2E). Furthermore, the cells containing 126 displayed a significantly increased mean fluorescent intensity (MFI; 2.2-fold higher, P ≤ 0.0001) (Fig. 2F), suggesting more 126 than USA300 bacterial cells occupied each mammalian cell. As a control, no statistically significant difference in MFI was observed between the bacterial strains (data not shown), indicating equivalent dye labeling. These results suggest that although both strains can be phagocytosed by mammalian cells, 126 was more readily phagocytosed than USA300. To determine if the subcellular location of the strains was different, we performed confocal microscopy on cells infected with AF647-labeled USA300 or 126 for 30 min and stained for LAMP-1, a lysosome-endosome protein (Fig. 2G). We found that both strains colocalized with LAMP-1-positive compartments within the cell, suggesting that both strains were found in the endosomes. Although both strains were localized to the same compartment, they may be processed or handled differently by the host cell. To determine if endosomal acidification was altered by 126, we used LysoLive pH-sensor (Marker Gene), a fluorescent dye that concentrates in acidic organelles. No significant differences in acidification were observed between cells stimulated by either strain (Fig. 2H and I). These data identify an S. aureus strain that constitutively triggers low levels of IFN-β production. The cause of this dampened induction was not related to the localization of the bacteria within the cell, since both strains were found to be directed to the late endosomes whose pH was unaltered.

Strain 126 is resistant to autolysis and cell wall digestion.

In a previous study, we observed that increased IFN-β induction was related to enhanced release of pathogen-associated molecular patterns (PAMPs) via bacterial autolysis (7). Given strain 126 has limited ability to induce type I IFN signaling, we hypothesized that the autolysis profile of strain 126 would be reduced. To test this, we subjected USA300 and 126 to an autolysis assay and observed that, over time, the optical density (OD) of 126 had only decreased by 25.5% versus 56.47% for USA300 (P ≤ 0.0001) (Fig. 3A). We sought to determine if the decreased autolysis displayed by 126 could also translate into decreased sensitivity to lysostaphin. Lysostaphin is an endopeptidase capable of cleaving pentaglycine bridges located in the cell wall of staphylococci, leading to rapid lysis of bacteria (27). To answer this question, we determined the lysostaphin MICs for both strains. We found that the lysostaphin MIC of USA300 in a microtiter plate assay was 62.5 ng/ml compared to 125 ng/ml for 126 (P ≤ 0.0001) (Fig. 3B). We also conducted bacterial viability tube assays with a range of lysostaphin concentrations. At 0.12 μg/ml, we observed over 1,500-fold more viable 126 bacteria than USA300 (P ≤ 0.0001) (Fig. 3C). Consistent with the enhanced resistance to lysostaphin, 126 was also more resistant to host cell killing. In our extracellular killing assay using BMDC, we observed that while USA300 had 12% of the initial inoculum remaining after 4 h, 126 had proliferated to reach 183% of the initial inoculum (P ≤ 0.0001) (Fig. 3D). These data indicate 126 has a more resilient cell wall that, with reduced autolysis, prevents release of cellular contents to activate type I IFN.

FIG 3.

Low-IFN-β-inducing strain displays increased resistance to autolysis and lysostaphin. (A) Autolysis assay (USA300, n = 3; 126, n = 3). (B and C) Lysostaphin MICs of USA300 (n = 6) and 126 (n = 6) were determined by bacterial growth inhibition (B) and dilution plating (C). Statistical significance was assessed using 2-way RM-ANOVA followed by Sidak’s multiple-comparison test. (D) BMDC were incubated with USA300 (n = 6) or 126 (n = 6). Bacterial survival was determined after 4 h. Statistical significance was assessed using unpaired Student's t test. (E to H) BMDC were stimulated with USA300 (n = 6) or 126 (n = 6), and cytokines were quantified in the supernatant by multiplex ELISA after 24 h. Results are presented as a heat map of mean normalized concentrations (E) or mean IL-1β, IL-10, and MIP-1β concentrations and SEM (F to H). Statistical significance was assessed using unpaired Student's t test. All results are representative of at least 2 independent experiments. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001.

We explored further the diminished capacity of 126 to activate the host response and whether this was limited to type I IFN. BMDC were stimulated with USA300 and 126 for 24 h and cytokines quantified by multiplex enzyme-linked immunosorbent assay (ELISA). We observed that USA300 induced an overall stronger cytokine response than 126 (Fig. 3E and Fig. S3). Proinflammatory cytokines such as interleukin-1β (IL-1β) (77.79% reduction; P ≤ 0.0001) (Fig. 3F) and IFN-γ (70.84% reduction; P ≤ 0.0001) (Fig. 3G) were reduced. This was also the case for anti-inflammatory cytokines; IL-10 was reduced by 57.26% (P ≤ 0.0001) (Fig. 3G). Several other cytokines were also decreased in response to 126 compared to wild-type (WT) USA300 (Fig. S3). Only one cytokine was induced at a statistically significant higher level by 126 than USA300: macrophage inflammatory protein 1β (MIP-1β) (Fig. 3H) was 58.89% higher (P < 0.05). These results suggest that the increased resistance to cellular damage and killing of 126 prevents its degradation and release of products that activate the innate immune response.

Strain 126 has reduced virulence in vivo.

Strain 126 displays increased resistance to cellular degradation as well as dampened ability to induce specific cytokine responses. These factors suggest an increased ability to evade immune defenses in vivo. However, type I IFN signaling has been shown to be deleterious in the context of S. aureus respiratory infections (6, 7). As 126 induces a diminished IFN-β response, it could also be cleared more readily by the host. To test this, we compared the virulence of 126 to USA300 in our model of acute pneumonia. After 24 h of infection, we observed a 5-fold reduction in bacterial burden in the bronchoalveolar fluid (BALF) (P ≤ 0.01) and 54-fold less 126 bacteria in lung tissue (P < 0.0001) than USA300 (Fig. 4A and B). This decreased bacterial burden of strain 126 was also accompanied by a significant decrease in neutrophil and monocyte numbers in the airway (59.90% and 42.33% decrease, respectively, P ≤ 0.01) (Fig. 4C and D). We also observed decreased numbers of Ly6C-negative monocytes and natural killer cells (Fig. 4E and F), while all other cell populations were unchanged between the two strains (Fig. S4A to E). To determine if these differences in cellular recruitment were specific to 126 or just a by-product of its enhanced clearance, we infected mice for 4 h with either USA300 or 126. Even at this early time point, enhanced clearance of 126 was evident, being 13.6-fold less in BALF (P ≤ 0.01) (Fig. 4G) and 8.2-fold less in the lung (P ≤ 0.01) (Fig. 4H). At this 4-h time point, no differences were observed in any of the airway cell populations (Fig. 4I to L and Fig. S4F to J). These data show that 126 is more readily cleared than USA300 and evokes a reduced host response in vivo due to its rapid clearance.

FIG 4.

Strain 126 displays decreased virulence in vivo. Mice were inoculated intranasally with 10⁷ CFU of either USA300 (n = 14) or 126 (n = 15), and the response to infection was assessed 24 h later. Bacteria were enumerated from the BALF (A) and lung homogenates (B) by dilution plating. (C to F) Cell populations were determined using flow cytometry analysis from BAL samples. Statistical analyses were performed using a Mann-Whitney test. (G to L) Mice were infected intranasally with 10⁷ CFU of either USA300 (n = 5) or 126 (n = 6) for 4 h. Bacterial burdens in the BALF (G) and lung homogenates (H) were quantified by dilution plating. (I to L) Cell populations were determined using flow cytometry analysis from BAL samples. Each dot represents a mouse. All data are shown as the pooled means ± SEM from at least 2 independent experiments. Dashed lines mark cell numbers in naive mice. Statistical analysis was performed using a Mann-Whitney test. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001.

VISA strains are low inducers of type I IFN.

Vancomycin-intermediate S. aureus (VISA) strains are known to have a thickened cell wall that prevents diffusion of vancomycin as well as a reduction in autolytic activity (28, 29). Based on our data, we hypothesized that the phenotype of 126 could have enhanced resistance to vancomycin. The vancomycin MIC of 126 was found to be 4 μg/ml, consistent with it being a VISA strain (30), compared to 2 μg/ml for USA300. We confirmed this result using a MIC plate strip test (Etest) and again observed 126 to possess increased resistance to vancomycin, with a MIC of 6 μg/ml, compared to USA300 at 2 μg/ml (Fig. 5A). To determine if lower IFN-β induction was unique to 126 or common to all VISA strains, we incubated BMDC with a panel of VISA strains for 24 h and quantified cytokines in the supernatant. All VISA strains showed a statistically significant reduction in IFN-β levels compared to USA300 (between 2- and 33-fold less than USA300, P < 0.0001) (Fig. 5B). While we observed some variations in TNF-α production (Fig. 5D), all VISA strains induced levels of IFN-λ similar to those of USA300 (Fig. 5C).

FIG 5.

Vancomycin-intermediate S. aureus strains have dampened IFN-β induction. (A) Vancomycin sensitivity profiles were evaluated using E tests. Arrows indicate the lowest antibiotic concentrations inducing growth inhibition. (B to D) BMDC were stimulated with a panel of 7 vancomycin-intermediate S. aureus strains at an MOI of 10 for 24 h (USA300, n = 5; 126, n = 5; VISA 1 to 7, n = 6). Cytokines were quantified in the supernatant by ELISA, and concentrations were normalized to cytokine production induced by USA300 stimulation. Statistical analyses were performed using one-way ANOVA followed by Dunnett’s multiple-comparison test. (E) IFN-β levels were quantified by ELISA from the supernatant of BMDC stimulated with either NRS384 (n = 6), NRS384 WalK^G223D (n = 6), or mock PBS (n = 6) after 24 h. Statistical analysis was performed using unpaired Student's t test. (F and G) BMDC were stimulated with NRS384 (n = 3) or NRS384 WalK^G223D (n = 3). Induction of Ifnb (F) and Tnf (G) was analyzed by qRT-PCR after 2 h. Statistical analysis was performed using unpaired Student's t test. (H) Autolysis assay expressed as percentage of starting OD₆₀₀ after 3 h (NRS384, n = 4; NRS384 WalK^G223D, n = 4). Statistical significance was assessed using unpaired Student's t test. (I) Lysostaphin inhibition of NRS384 (n = 4) and NRS384 WalK^G223D (n = 4) was determined by bacterial growth inhibition. (J) BMDC were stimulated with a panel of vancomycin-susceptible S. aureus strains (VSSA) (n = 7) or with a panel of vancomycin-intermediate S. aureus strains (VISA) (n = 8) at an MOI of 10 for 24 h. Interferon β production was quantified in the supernatant by ELISA. Each dot represents a strain. Statistical analysis was performed using a Mann-Whitney test. (K) Relationship between IFN-β induction and autolysis susceptibility (percentage of starting OD after 3 h) (n = 20). Each dot represents a strain (VISA and VSSA included). (L) Relationship between IFN-β induction and lysostaphin sensitivity (OD₆₀₀ after 16 h of exposure to lysostaphin [62.5 ng/ml]) (n = 20). Each dot represents a strain (VISA and VSSA included). Data are shown as the pooled means ± SEM from at least 2 independent experiments. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001.

WalK is the transmembrane sensor histidine kinase component of a well-conserved two-component regulatory system, WalKR, essential for bacterial survival and involved in regulating peptidoglycan synthesis (31–33). In S. aureus, mutations in the walKR locus are associated with a vancomycin-intermediate resistance phenotype (34). A single-nucleotide change in walK or walR is sufficient to develop a VISA phenotype, thickened cell wall, reduced autolysis, and diminished virulence (35). As further evidence that VISA strains have a reduced capacity to induce type I IFN signaling, we studied a mutant in WalK, WalK^G223D, which exhibits reduced autophosphorylation/phosphotransfer (35). We incubated BMDC with S. aureus NRS384 (WT to mutant) or NRS384 WalK^G223D for 24 h and quantified the amount of type I IFN produced by the cells. We found that the WalK^G223D strain triggered significantly lower production of IFN-β (61.23% reduction compared to the WT, P ≤ 0.0001) (Fig. 5E), and this was also seen at the transcriptional level, with a 49% reduction in Ifnb expression (P ≤ 0.01) (Fig. 5F). However, no significant changes in mRNA levels of Tnf were observed (Fig. 5G). We also found that NRS384 WalK^G223D was significantly less susceptible to autolysis and lysostaphin degradation than its wild-type counterpart (Fig. 5H and I). To test if this observation could be applied to a more general population, we compared the levels of IFN-β induced by a panel of vancomycin-susceptible S. aureus strains (VSSA) and a panel of VISA strains. We detected a significant reduction in IFN-β production by BMDC when the cells were exposed to VISA strains compared to that for VSSA strains (P ≤ 0.01) (Fig. 5J). We also analyzed the relationship between IFN-β induction and autolysis and lysostaphin susceptibility across both VSSA and VISA strains and found a significant correlation between decreased type I IFN induction and autolysis and lysostaphin resistance (Fig. 5K and L). To confirm the relationship between autolysis and IFN-β induction, we used an atl mutant, USA300 JE2 atl::erm, defective in the major autolysin, and its wild-type counterpart, USA300 JE2. We incubated each strain with BMDC and measured the amount of Ifnb mRNA they induced after 2 h (Fig. S5). The autolysis mutant displayed a significant decrease in Ifnb induction compared to that of the WT strain, indicating that autolysis plays a role in type I IFN induction. These results suggest that impaired ability to induce IFN-β production is a common feature of VISA strains and correlates with increased resistance to autolysis.

DISCUSSION

In this study, we provide evidence linking enhanced resistance to vancomycin, as observed in VISA strains, and reduced type I IFN production. This phenomenon was identified through a screen of S. aureus isolates of diverse backgrounds showing that activation of this pathway is not homogeneous among the species.

We started our study by examining a panel of S. aureus clinical isolates for their ability to induce IFN-β and IFN-λ and observed a heterogeneity of induction of both cytokines among the strains. Given the contributing role for interferon signaling in the pathogenesis of S. aureus pneumonia (6–9), we had expected that certain isolates had inherently higher induction capacities, but this spectrum of induction was evident across all sample groups. IFN-β induction appeared to positively correlate with TNF-α induction, whereas IFN-λ induction followed the opposite trend. IFN-β and IFN-λ induction did not show any correlation for the first 24 h of the experiment but showed a positive correlation after 48 h. This supported previous observations that indicated a common path of induction for type I and type III IFN signaling, as well as differences in kinetics between IFN-β and IFN-λ (21–25). However, we did observe a disconnect between type I and type III IFN signaling, such that the strain we focused on, 126, still induced IFN-λ while not inducing IFN-β. A cytosolic DNA sensor, Ku70, has been shown to induce type III IFN signaling preferentially over type I IFN signaling in the context of HIV infection (36). This induction has also been shown to be STING dependent (37). It could be that the two strains, due to their different induction properties, activate signaling through different innate receptors. S. aureus is typically phagocytosed and degraded in the phagolysosomes; however, occurrences of phagosomal and cytosolic persistence have been described (38, 39). Given the enhanced resilience of strain 126, as seen in lysostaphin sensitivity and host cell killing, this could justify the recognition of the bacterium by cytoplasmic sensors and, therefore, the privileged induction of type III IFN signaling.

While IFN-β induction varied across VSSA strains, the VISA isolates tested in this study displayed a reduced propensity to activate the type I IFN response. VISA strains are characterized by a thickened cell wall, reduced autolysis, and diminished virulence (40, 41). The cell wall changes consist of an accumulation of d-Ala-d-Ala targets due to decreased cross-linking of peptidoglycan, increased proportion of nonamidated muropeptides, and decreased alanylation of teichoic acid. As a result, the cell wall is strengthened and vancomycin binding sites are reduced (42–44). This feature could also act to prevent degradation of the bacterial cell. Although we showed that strain 126 localized within the mammalian cell in a location similar to that of USA300, the kinetics at which it is degraded are likely to be slower, as indicated in our killing assay. The decreased autolysis by 126 would also lead to a reduction in release of cell wall fragments capable of activating the immune response. Autolytic activity has been shown to release extracellular DNA (45, 46) and peptides from peptidoglycan (47), both ligands that we have shown to lead to activation of type I IFN signaling by S. aureus through TLR9 and NOD2 (6, 7). It has also been observed by other groups that resistance to lysozyme can inhibit activation of type I IFN signaling (48). USA300 is resistant to lysozyme degradation, with 126 expressing less of the peptidoglycan O-acetyltransferase A that contributes to resistance (data not shown), indicating we were observing a separate mechanism behind decreased interferon activation.

VISA strains have been associated with differential cell adherence, division, biofilm formation, and virulence, leading to other phenotypes, such as growth as multicellular aggregates instead of regularly shaped and well-separated cocci (40). If the bacterial cells are not found as distinct entities but rather as microbial clusters, this could explain why increased phagocytosis was observed for 126 compared to USA300. We observed strain 126 to also have decreased virulence in vivo, which is characteristic of VISA isolates (41). However, in this context, in vivo attenuation cannot directly be assigned to the VISA phenotype. Vancomycin susceptibility along with the differing cell wall properties may be only one of the many factors that differ between the strains. One element of the immune response that would benefit the host would be the decreased type I IFN. As we have shown before, type I IFN signaling contributes to pathogenesis (6, 7). The reduced activation of this pathway by 126 could contribute to its improved clearance due to decreased host immunopathology.

In this work, we show association between VISA strains, reduced autolytic activity, and decreased IFN-β induction. While dampened type I IFN signaling was consistent across all the VISA isolates tested in this study, some VSSA strains, such as strains 121, 122, and 125, also displayed a reduced ability to activate this pathway. An expanded panel of strains quantified for vancomycin susceptibility would need to be tested before a significant correlation can be drawn between induction of type I IFN and vancomycin susceptibility. It is likely that mutations of the bacterial cell wall will have effects on IFN induction. As such, other perturbations of cell wall synthesis, such as inhibition of transpeptidase activity, are likely to affect autolysis (49) and IFN responses. These could explain the reduced type I IFN induction observed for some of the VSSA strains in this study. On the same note, VISA strains can result from mutations of several genes, including walKR, graSR, clpP, or sarA (29). We were able to demonstrate that inactivation of the kinase WalK leads to reduced activation of type I IFN along with reduced autolysis and lysostaphin sensitivity. Further investigation is required to determine if these genetic modifications impact IFN induction in similar ways.

While it is now appreciated that bacterial pathogens can activate type I IFN signaling, less is known about the general capacity to activate the pathway within a given species. We show here that S. aureus exhibits a wide spectrum of type I and III interferon induction magnitudes, showing that activation of interferon signaling is not a universally conserved phenomenon. We focused on a unique strain that had limited type I IFN-inducing properties and identified it as being a VISA strain, further showing this to be a general property of VISA isolates. Lack of type I IFN induction was associated with increased resistance to lysostaphin and host cell killing and decreased autolysis and in vivo virulence. A better understanding of the triggers and mechanisms by which S. aureus mediates these host responses will educate any potential future immunomodulatory treatments.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

The S. aureus strains used in this study are summarized in Table 1. BEI Resources strains were provided by the Network on Antimicrobial Resistance in Staphylococcus aureus (NARSA) for distribution by BEI Resources, NIAID, NIH. S. aureus strains were grown in Luria-Bertani (LB) broth at 37°C with shaking. Experiments were performed with bacteria in exponential growth phase (at an OD₆₀₀ of 1). Vancomycin MICs were evaluated by the MicroScan Gram-positive panel (Beckman Coulter). Extended glycopeptide susceptibilities were determined with vancomycin Etest strips (bioMérieux) at 37°C.

TABLE 1.

Bacterial strains used in this study^a

No. in this study	Strain	Source/reference	Isolation site	Sensitivity profile		Ridom spa type	MLST type
				Methicillin	Vancomycin
USA300	USA300 FPR3757	51	Laboratory strain	MRSA	VSSA	NA	NA
502A	502A	7	Laboratory strain	MSSA	VSSA	NA	NA
2	USA300 LAC	52	Laboratory strain	MRSA	NA	NA	NA
4	Newman	53	Laboratory strain	NA	NA	NA	NA
5	USA500	54	Laboratory strain	NA	NA	NA	NA
6	MW2	55	Laboratory strain	NA	NA	NA	NA
7	N315	56	Laboratory strain	NA	NA	NA	NA
USA300	Je2	57	Laboratory strain	MRSA	NA	NA	NA
USA300	Je2::atl	57	Laboratory strain	MRSA	NA	NA	NA
41	PP1759	This study	CF isolate, sputum	NA	NA	NA	NA
42	PP1771	This study	CF isolate, sputum	NA	NA	NA	NA
43	PP1772	This study	CF isolate	NA	NA	NA	NA
44	PP1775	This study	CF isolate	NA	NA	NA	NA
45	PP1779	This study	CF Isolate, BAL	NA	NA	NA	NA
46	PP1840	This study	CF isolate, BAL	NA	NA	NA	NA
47	PP1842	This study	CF isolate	NA	NA	NA	NA
48	PP1843	This study	CF isolate	NA	NA	NA	NA
49	PP1844	This study	CF isolate, BAL	NA	NA	NA	NA
50	PP1761	This study	CF Isolate, sputum	NA	NA	NA	NA
51	112-538	This study	Skin	MRSA	NA	t008	ST8
52	254-28	This study	Skin	MRSA	NA	NA	ST8
53	66-752	This study	Skin	MRSA	NA	t008	ST8
54	226-8	This study	Skin	MRSA	NA	NA	ST8
55	19-327	This study	Skin	MRSA	NA	t008	ST8
56	218-12	This study	Skin	MRSA	NA	NA	ST8
57	88-316	This study	Skin	MRSA	NA	NA	ST8
58	241-64	This study	Skin	MRSA	NA	NA	NA
59	58-895	This study	Skin	MRSA	NA	t008	ST8
60	225-39	This study	Skin	MRSA	NA	NA	ST8
74	3-179	This study	AD site	MSSA	NA	t021	ST30
75	5-176	This study	Nares	MSSA	NA	t334	NA
76	6-196	This study	Nares	MSSA	NA	t1892	NA
77	15-662	This study	AD site	MSSA	NA	t002	ST5
78	15-663	This study	Perianal	MSSA	NA	t1309	ST672
79	16-658	This study	Nares	MSSA	NA	t922	NA
80	28-205	This study	Nares	MRSA	NA	t064	NA
81	31-921	This study	Nares	MSSA	NA	t002	ST5
82	34-234	This study	Nares	MSSA	NA	t026	NA
83	34-236	This study	AD site	MSSA	NA	t026	ST45
84	43-378	This study	Nares	MSSA	NA	t177	NA
85	43-384	This study	AD site	MSSA	NA	t177	ST81
86	53-253	This study	Perianal	MSSA	NA	t189	NA
87	53-257	This study	Nares	MSSA	NA	t216	ST59
88	63-121	This study	AD site	MRSA	NA	t008	ST8
89	68-549	This study	AD site	MRSA	NA	t008	NA
90	81-903	This study	AD site	MSSA	NA	t228	NA
91	81-904	This study	Perianal	MSSA	NA	t148	ST72
92	81-906	This study	Nares	MSSA	NA	t216	ST87
93	97-727	This study	AD site	MSSA	NA	t002	ST5
94	97-729	This study	Perianal	MSSA	NA	t1577	ST30
95	103-348	This study	Perianal	MRSA	NA	t008	NA
96	107-603	This study	Nares	MSSA	NA	t688	NA
97	107-604	This study	Perianal	MSSA	NA	t688	ST5
98	130-858	This study	Nares	MRSA	NA	t002	ST8
114	626, NRS382	BEI Resources	Blood	MRSA	VSSA	t002	ST5
115	CA-263, NRS647	BEI Resources	Blood	MRSA	VSSA	NA	ST8
116	MN-095, NRS703	BEI Resources	Blood	MRSA	VSSA	NA	ST8
117	96758, NRS383	BEI Resources	Blood	MRSA	VSSA	t018	ST36
118	95938, NRS385	BEI Resources	Blood	MRSA	VSSA	t064	ST8
119	MRSA 18 M342506	BEI Resources	Wound abscess	MRSA	VSSA	NA	NA
120	CA-347, NRS648	BEI Resources	Blood	MRSA	VSSA	NA	NA
121	1045, NRS387	BEI Resources	Wound	MRSA	VSSA	t088	ST5
122	CM05	BEI Resources	Tracheal aspirate	MRSA	VSSA	NA	NA
123	MRSA177	BEI Resources	Skin	MRSA	VSSA	NA	NA
124	P1V44, NRS272	BEI Resources	Sputum	MRSA	VSSA	t770	ST247
125	NRS127	BEI Resources	Sputum	MRSA	VSSA	t002	ST5
126	HIP07930, NRS22	BEI Resources	Blood	MRSA	VISA	t266	ST45
127	MRSA M1277	BEI Resources	Blood	MRSA	NA	NA	NA
128	No. 315, NRS209	BEI Resources	Trachea	MRSA	VSSA	t051	ST247
129	HFH-30522	BEI Resources	Sputum	MRSA	NA	NA	NA
130	MRSA M0055	BEI Resources	Sputum	MRSA	NA	NA	NA
131	A980592, NRS157	BEI Resources	Necrotizing pneumonia	MSSA	VSSA	t3379	ST22
132	CA-78, NRS659	BEI Resources	Lung fluid	MRSA	VSSA	t008	ST8
133	MN-113, NRS704	BEI Resources	Pleural fluid	MRSA	VSSA	NA	ST5
134	TN-74, NRS739	BEI Resources	Pleural fluid	MRSA	VSSA	NA	ST8
135	MRSA M0001	BEI Resources	Sputum	MRSA	NA	NA	NA
136	MRSA M0108	BEI Resources	Bronchoalveolar lavage	MRSA	NA	NA	NA
VISA 1	160013, NRS283	Ricardo Russo	Blood	MRSA	VISA	t018	ST36
VISA 2	HIP10540, NR-45901	Ricardo Russo	Unknown	MRSA	VISA	t451	ST8
VISA 3	LY-1999 0620-0	Ricardo Russo	Blood	MRSA	VISA	t037	ST372
VISA 4	HIP09740, NRS51	Ricardo Russo	Bile	MRSA	VISA	t242	ST5
VISA 5	HIP07920, NRS21	Ricardo Russo	Blood	MRSA	VISA	t064	ST8
VISA 6	HIP5827, NRS3	Ricardo Russo	Peritonitis	MRSA	VISA	t062	ST5
VISA 7	HIP09433, NRS27	Ricardo Russo	Cerebral spinal fluid	MRSA	VISA	t004	ST45
NRS384	NRS384	35	Laboratory strain	NA	VSSA	NA	NA
NRS384 WalKG233D	NRS384 WalKG233D	35	Laboratory strain	NA	VISA	NA	NA

^aNA, not available; CF, cystic fibrosis; AD, atopic dermatitis.

Autolysis assays were performed in phosphate-buffered saline with Triton X-100 (0.2%) and incubated at 37°C without shaking. Lysis was determined as the decrease in OD₆₀₀ over time and indicated as a percentage of the initial OD. For lysostaphin assays, bacterial pellets were resuspended in LB containing specific lysostaphin concentrations and incubated in microplates at 37°C with shaking for 16 h. Lysostaphin MICs were determined as the lowest concentration at which bacterial growth was inhibited.

Mammalian cell culture and assays.

Bone marrow-derived dendritic cells (BMDC) and bone marrow-derived macrophages (BMM) were generated from the bone marrow of WT C57BL/6J mice and differentiated in RPMI medium containing 10% heat-inactivated fetal bovine serum, penicillin-streptomycin, and granulocyte macrophage colony-stimulating factor (20 ng/ml) or macrophage colony-stimulating factor (20 ng/ml), respectively, at 37°C and 5% CO₂ in sterile petri dishes for 7 days. BMDC were stimulated with S. aureus (multiplicity of infection [MOI], 10) for 4, 16, 24, or 48 h for cytokine quantification by ELISA and 2 h (MOI, 100) for RNA experiments. Phagocytosis assays were performed with AF647-labeled (AF647 carboxylic acid succinimidyl ester; Life Technologies) S. aureus (MOI, 10) for 5 and 30 min. Killing assays used S. aureus (MOI, 10)-stimulated BMDC 4 h after infection, supernatants were collected, and bacterial concentrations were assessed by serial dilution plating on LB agar plates. Endosomal acidification assays were performed with S. aureus (MOI, 10)-stimulated BMDC in the presence of LysoLive pH-sensor green (Marker Gene Technologies) for 2 h.

Animal studies.

Male and female 6- to 8-week-old C57BL/6J mice were used for this study. Mice were intranasally infected with 2 × 10⁷ to 5 × 10⁷ CFU of S. aureus while under anesthesia. Mice were euthanized 4 or 24 h after infection, and bronchoalveolar lavage fluid (BALF) was collected by instilling 3× 1 ml PBS into the trachea. Lungs were harvested and passed through a 40-μm cell strainer with 400 μl PBS to obtain a single-cell homogenate. BALF and lung homogenates were used to enumerate bacterial counts through dilution plating on BBL CHROMagar Staph aureus plates (Becton, Dickinson).

Flow cytometry.

Cells from BALF and lung homogenates underwent red blood cell lysis and were incubated with fluorescently labeled antibodies against specific cell surface markers: phycoerythrin (PE)-Cy7-labeled anti-CD11b (M1/70), BV605-labeled CD11c (N418), AF700-labeled CD45 (30-F11), AF647-labeled Siglec-F (E50-2440; BD Biosciences), BV650-labeled NK1.1, BV510-labeled CD103 (2E7), allophycocyanin-Cy7-labeled MHCII (M5/114.15.2), peridinin chlorophyll protein-Cy5.5-labeled anti-Ly6G (1A8), PE-Texas Red Ly6C (AL-2; BD Biosciences), PE-labeled CD200R (OX-110), fluorescein isothiocyanate (FITC)-labeled MARCO (ED31; Bio-Rad), and BV421-labeled CD86 (GL-1). Antibodies were from BioLegend unless otherwise stated. The gating strategy for the flow cytometry panel is outlined in Table 2. Uniform dyed microspheres, Dragon Green (Bangs Laboratories, Inc.), were used for cell counting. Data were collected by flow cytometry using a Fortessa cytometer (Becton, Dickinson), and results were analyzed using FlowJo v.10.

TABLE 2.

Pneumonia model flow cytometry panel

Cell type	Surface markers
Neutrophils	CD45+ Ly6C+ CD11b+ MHCII− Ly6G+
Alveolar macrophages	CD45+ Ly6C− SiglecF+ CD11c+ CD11b−
Interstitial macrophages	CD45+ Ly6C− SiglecF− CD11b+ CD11c+ MHCII+
Ly6C+ monocytes	CD45+ Ly6C+ CD11b+ MHCII− Ly6G−
Ly6C− monocytes	CD45+ Ly6C− SiglecF− CD11b+ CD11c+ MHCII−
NK cells	CD45+ NK1.1+
Eosinophils	CD45+ Ly6C− SiglecF+ CD11b+ CD11c−
Plasmacytoid dendritic cells	CD45+ Ly6C+ CD11b− CD11c+ MHCII+
CD103 dendritic cells	CD45+ Ly6C− SiglecF− CD11b− MHCII+ CD103+
CD11b dendritic cells	CD45+ Ly6C+ CD11b+ MHCII+ CD11c+

RNA analysis.

RNA was isolated using the E.Z.N.A. total RNA kit (Omega Biotek), followed by DNase treatment using DNAfree (Life Technologies). cDNA was synthesized using the high-capacity cDNA reverse transcription kit (Applied Biosystems). qRT-PCR was performed using Power SYBR green PCR master mix (Applied Biosystems) in a Quantstudio 6 thermal cycler (Applied Biosystems). Samples were normalized to β-actin.

Cytokine quantification.

Cytokine levels were quantified using ELISA to IFN-β (PBL Interferon), IFN-λ (R&D Systems), and TNF-α (BioLegend). The cytokine 32-plex Discovery assay was performed by Eve Technologies.

Microscopy.

Differentiated BMM were deposited on glass coverslips at the bottom of 24-well plates and allowed to attach for 16 h. Bacteria were labeled with AF647 for 30 min. Excess dye was removed by washing in PBS 5 times. Labeled bacteria were incubated with BMM (MOI, 10) at 37°C for 30 min. Cells were then washed with PBS, fixed, and permeabilized with Cytofix/Cytoperm buffers (Becton, Dickinson). LAMP-1 was detected with primary LAMP-1 rat antibody (DSHB) and secondary AF555-labeled anti-rat antibody (Life technologies). Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). Images were captured with a Nikon A1R confocal microscope.

Statistical analyses.

Statistical tests were chosen according to the study design. For experiments with one dependent variable, comparing 2 samples with no repeated measure, regular unpaired Student’s t tests were used unless assumption of normality was violated (Shapiro Wilk test, P < 0.05), in which case Mann-Whitney tests were used. For experiments with one dependent variable, comparing more than 2 samples, with no repeated measures, one-way analysis of variance (ANOVA) followed by Dunnett’s multiple-comparison tests were used unless the assumption of normality was violated (Shapiro Wilk test, P < 0.05), in which case Kruskal-Wallis test followed by Dunn’s multiple-comparison tests were used. For experiments with 2 dependent variables comparing more than 2 samples, with no repeated measures, 2-way ANOVA followed by Tukey’s multiple-comparison tests were used. For experiments with 2 dependent variables comparing more than 2 samples, with repeated measures, 2-way repeated measures ANOVA (RM-ANOVA) followed by Sidak’s multiple-comparison tests were used. For the cytokine induction pattern relationship experiment, correlation coefficients were calculated using Spearman’s r. Statistical analyses were performed with Prism software (GraphPad, La Jolla, CA, USA). Determination of normality by Shapiro Wilk tests and correlation coefficients (Spearman’s r) were conducted with R, version 3.5.1 (Core Team 2013), using the R Commander interface (Fox 2005).

Ethics statement.

Animal work in this study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health (50), the Animal Welfare Act, and U.S. federal law. Protocols were approved by the Institutional Animal Care and Use Committee of Rutgers New Jersey Medical School of Newark New Jersey.