|Home | About | Journals | Submit | Contact Us | Français|
Staphylococcus aureus uses the two-component regulatory system GraRS to sense and respond to host defense peptides (HDPs). However, the mechanistic impact of GraS or its extracellular sensing loop (EL) on HDP resistance is essentially unexplored. Strains with null mutations in the GraS holoprotein (ΔgraS) or its EL (ΔEL) were compared for mechanisms of resistance to HDPs of relevant immune sources: neutrophil α-defensin (human neutrophil peptide 1 [hNP-1]), cutaneous β-defensin (human β-defensin 2 [hBD-2]), or the platelet kinocidin congener RP-1. Actions studied by flow cytometry included energetics (ENR); membrane permeabilization (PRM); annexin V binding (ANX), and cell death protease activation (CDP). Assay conditions simulated bloodstream (pH 7.5) or phagolysosomal (pH 5.5) pH contexts. S. aureus strains were more susceptible to HDPs at pH 7.5 than at pH 5.5, and each HDP exerted a distinct effect signature. The impacts of ΔgraS and ΔΕL on HDP resistance were peptide and pH dependent. Both mutants exhibited defects in ANX response to hNP-1 or hBD-2 at pH 7.5, but only hNP-1 did so at pH 5.5. Both mutants exhibited hyper-PRM, -ANX, and -CDP responses to RP-1 at both pHs and hypo-ENR at pH 5.5. The actions correlated with ΔgraS or ΔΕL hypersusceptibility to hNP-1 or RP-1 (but not hBD-2) at pH 7.5 and to all study HDPs at pH 5.5. An exogenous EL mimic protected mutant strains from hNP-1 and hBD-2 but not RP-1, indicating that GraS and its EL play nonredundant roles in S. aureus survival responses to specific HDPs. These findings suggest that GraS mediates specific resistance countermeasures to HDPs in immune contexts that are highly relevant to S. aureus pathogenesis in humans.
Host defense peptides (HDPs) represent a critical first line of immune protection against Staphylococcus aureus infections (1,–3). Distinct HDPs are deployed via constitutive or inducible processes by neutrophils, keratinocytes, platelets, or other human tissues that this organism commonly encounters (3, 4). Moreover, distinct HDPs appear to have evolved for optimal host defense in specific immunologic and anatomic compartments (4, 5). In turn, pathogenic S. aureus strains have coevolved specific and rapidly adaptive systems to sense and respond to host cues to achieve immune avoidance or subversion.
Recently, we showed that the two-component regulatory system (TCRS) GraSR plays a key role in S. aureus survival in the face of HDPs (6,–8). Also known to be a component of the antibiotic peptide sensor (APS) (9, 10), GraSR upregulates adaptive resistance genes such as mprF and dltA, which encode proteins that modulate net surface charge and influence the composition of the S. aureus envelope (11, 12). In a previous report (7), we demonstrated that S. aureus strains with deletions in graS (ΔgraS) or its extracellular sensor loop (ΔEL) become hypersusceptible to several cationic peptides, including calcium-complexed daptomycin (DAP) (a cationic lipopeptide complex), polymyxin B (PMB) (a prokaryotic cationic cyclopeptide), and human neutrophil peptide 1 (hNP-1). Importantly, we observed a direct correlation between induction of mprF and dltABCD and survival in the face of PMB, but not hNP-1. Thus, certain HDPs may trigger more efficient sense/countermeasure response functions mediated by GraS.
These prior findings provided a logical basis for our hypothesis that GraS or its extracellular loop (EL) mediates critical adaptive countermeasures to specific mechanisms of HDP action in distinct anatomic contexts. Thus, the present study was designed to explore the role of the GraS holoprotein versus its EL in mediating resistance responses to relevant HDPs under pH conditions reflecting bloodstream (pH 7.5) or phagolysosomal (pH 5.5) settings.
Table 1 summarizes the methicillin-resistant Staphylococcus aureus (MRSA) strain panel used in this study, generated from the community-acquired MRSA strain MW2. MW2 (USA400) is a prototypic and well-characterized clinical isolate with a known genome that has been demonstrated to be virulent in multiple animal models. graS mutant strains were generated by in-frame deletion of target genes by allelic replacement using the temperature-sensitive plasmid pMAD as previously detailed (7). Organisms were cultured in brain heart infusion broth (BHI) (Becton Dickinson) at 37°C with shaking overnight (~16 h), subcultured under identical conditions for 3 h to logarithmic phase, harvested by centrifugation, washed, and resuspended to the appropriate CFU (culture confirmed) by spectrophotometry for each specific assay.
Peptides representing relevant immunologic contexts were studied. Human neutrophil peptide 1 (hNP-1) (Peptides International, Louisville, KY) is a prototypic α-defensin found in human neutrophil-specific granules and is important in phagolysosomal killing of S. aureus. Human β-defensin 2 (hβD-2; Peptides International) is a predominant host defense peptide (HDP) elaborated by epithelial tissues throughout the body. The mimetic peptide RP-1 is a synthetic 18-amino-acid congener engineered in part from the microbicidal α-helix domains of the platelet factor 4 family of kinocidins (8). RP-1 was synthesized, purified, and authenticated as previously detailed (9). Each study peptide has demonstrated in vitro antimicrobial activity against S. aureus (3, 7, 8, 10).
A soluble graS extracellular loop (sEL) sensor domain mimetic has been previously shown to alter the survival of S. aureus in response to certain HDPs (7). This molecule was designed to contain three sensor motifs interposed by diglycine hinges (DYDFPIDSL-GG-DYDFPIDSL-GG-DYDFPIDSL). A “nonsense” peptide of the same composition but having a randomized (scrambled) sequence was also generated and was included in assays as a control. The sEL and nonsense peptides were synthesized, purified, and authenticated as previously described (8, 9).
The susceptibility of graS mutants to distinct HDPs was assessed using an established radial diffusion method (10). In brief, logarithmic-phase cells adjusted to 106 CFU/ml were seeded into 10 ml of buffered 1% agarose. Piperazine-N,N′-bis(2-ethanesulfonic acid) (PIPES) or 2-(N-morpholino)ethanesulfonic acid (MES) buffer was used to adjust assay conditions to pH 7.5 or pH 5.5. These pH conditions were chosen to represent relevant anatomical contexts, namely, bloodstream or acidic phagolysosome, respectively. Based on pilot studies, 10 μg of each HDP, alone or in combination with an equivalent mass (10 μg) of graS EL mimetic or nonsense peptide, was added to wells in the seeded underlay matrix and incubated at 37°C for 3 h. The sEL peptide, nonsense peptide, and vehicle (phosphate-buffered saline [PBS]) alone were included in each assay as internal controls. After 3 h of incubation, plates were overlaid with nutrient medium (Trypticase soy) and incubated for 24 h at 37°C. Zones of inhibition (ZOI) were measured to the nearest millimeter in diameter. A minimum of two independent experiments were conducted on separate days for statistical analysis.
Six-parameter multicolor flow cytometry was used to analyze four specific mechanisms of HDP action and two global effects associated with these mechanisms in the parental, ΔgraS, or ΔEL S. aureus strains: (i) perturbation of cell membrane (CM) energetics (ENR) (e.g., transmembrane potential), (ii) CM permeabilization (PRM), (iii) annexin V binding (ANX) (negatively charged phospholipids and CM turnover (13, 14), (iv) caspase-like/metacaspase-like cell death protease induction (CDP) (11, 14, 15), (v) osmodisruption (forward scatter [FSC]), and (vi) membrane invagination/chromosomal condensation/cytoplasmic refractivity (side scatter [SSC] granularity). The following fluorophores were used with a FACSCalibur instrument (Becton Dickinson): 3,3-dipentyloxacarbocyanine (DiOC5) (excitation, 484 nm; emission, 660 nm) (Invitrogen, Carlsbad, CA) for ENR, propidium iodide (PI) (excitation, 535 nm; emission, 620 nm) (Sigma, St. Louis, MO) for PRM, annexin V-allophycocyanin conjugate (ANX-V) (excitation, 650 nm; emission, 660 nm) (Invitrogen, Carlsbad, CA) for ANX, and CellEvent caspase-3/7 green (C-3/7) (excitation, 502 nm; emission, 530 nm) (Invitrogen, Carlsbad, CA) for CDP. FSC and SSC were measured in parallel. Logarithmic-phase organisms were adjusted to 106 CFU/ml in PIPES (pH 7.5) or MES (pH 5.5) and exposed to 20 μg of each HDP of interest for 1 h at 37°C. Based on extensive pilot data, this peptide concentration was used to achieve approximately 50% survival given the high inoculum of bacteria exposed. A triple-stain cocktail containing DiOC5 (0.5 μM), PI (5.0 μg/ml), and ANX-V (2.5 μl/ml) in 50 mM potassium-containing minimal essential medium (K+ MEM) (without phenol red; Sigma) was added to each sample following incubation. Samples were stained at room temperature for 15 min before flow cytometry. Parallel samples were incubated with 30 μl of C-3/7 reagent for 30 min at 37°C following peptide exposure. After incubation, 400 ml of PBS was added to remove any background signal. Sodium dodecyl sulfate (SDS) (10%, wt/vol; Ambion) (a nonspecific perturbant of ENR and PRM) or buffers alone (K+ MEM or PBS) were included as controls in each experiment. Fluorescence of a minimum of 10,000 cells was acquired from each sample, and results from a minimum of two independent studies conducted on different days were used for statistical analysis.
The Mann-Whitney U test was used as appropriate to determine significant differences in susceptibility phenotypes and mechanisms of HDP action in S. aureus strains.
The susceptibility of S. aureus strains to prototypic HDPs under different pH conditions is summarized in Fig. 1 and and2.2. The wild-type strain was susceptible to hNP-1 and RP-1 at both pH 7.5 and 5.5. However, this strain was susceptible to hBD-2 only at pH 7.5. The susceptibility was greatest for RP-1, substantially less for hNP-1, and least for hBD-2. Notably, the wild-type strain was more susceptible to the HDPs at pH 7.5 than at pH 5.5 (Fig. 1A and andC).C). These susceptibility rankings were consistent for HDPs under both pH conditions, even though absolute susceptibilities were reduced at pH 5.5.
At pH 7.5, the ΔgraS and ΔEL mutants were significantly more susceptible to hNP-1 or RP-1 than the parental strain (Fig. 1 and and2)2) (P < 0.05). The ΔgraS mutant trended toward greater hBD-2 susceptibility than the parental strain, but this did not reach significance. At pH 5.5, the mutants were significantly more susceptible to all HDPs than the respective controls at pH 7.5. Notably, only the ΔgraS mutation conferred significantly greater susceptibility to hBD-2 at pH 5.5, while the ΔEL mutation did not achieve significance. Complementation of mutants largely restored susceptibility to wild-type-equivalent levels (Fig. 1 and and22).
The exogenous soluble EL (sEL) did not exert intrinsic anti-S. aureus efficacy alone (Fig. 2). At pH 7.5, it significantly protected the ΔgraS and ΔEL mutants against hNP-1 (P < 0.05) but not hBD-2 or RP-1. At pH 5.5, the sEL protected against hNP-1 and hBD-2 (Fig. 2). Interestingly, the sEL did not protect any strain against RP-1 at either pH.
The comparative impact of HDP mechanisms on the panel of S. aureus strains is shown in Fig. 3 and and44 and in Fig. S1 and S2 in the supplemental material. The specific mechanisms of action of the study HDPs versus SDS relative to the ΔgraS or ΔEL mutants are detailed below.
In the wild-type strain, hNP-1 caused significant increases in side scatter (SSC), energetics (ENR), and annexin V binding (ANX) compared to those for untreated controls (P < 0.05) (Fig. 3 and and4).4). At pH 7.5, both the ΔgraS and ΔEL mutants exhibited decreased SSC and ANX compared to the wild type (Fig. 3). These effects reached significance in the ΔgraS and ΔEL strains for SSC or ANX, respectively. These results corresponded to biological relevance as evidenced by significantly increased susceptibility of the mutants to hNP-1 (Fig. 1 and and2).2). Interestingly, at pH 5.5, only the ΔgraS mutant exhibited reduced SSC and ANX versus the parent strain (Fig. 3). Complementation largely reverted hNP-1 mechanisms to wild-type equivalence. No detectable impact of hNP-1 in ΔgraS or ΔEL mutants was observed regarding forward scatter (FSC), ENR, cell membrane permeabilization (PRM), or cell death protease activation (CDP) at either pH (Fig. 3 and and44).
Exposure to hBD-2 at pH 7.5 caused significant increases in SSC, ENR, and ANX in wild-type S. aureus (Fig. 3 and and4).4). Relative to that for the wild type, ΔgraS and ΔEL mutations led to further increases in ANX (Fig. 3) (P < 0.05). Only the ΔgraS mutant displayed significantly reduced ENR versus the parental strain. In contrast to the case for hNP-1, increased ANX was not associated with significantly greater susceptibility to hBD-2 in the ΔgraS mutant at pH 7.5 (Fig. 1 and and2).2). The ΔEL mutant also exhibited greater ANX than the wild type, but this effect did not translate to greater hBD-2 susceptibility. At pH 5.5, hBD-2 caused significantly increased ENR in the ΔEL mutant (Fig. 3 and and4).4). Complementation reverted hBD-2-induced mechanisms to wild-type equivalence in both mutant strains and at pH 5.5 appeared to hypercompensate with respect to ANX in both revertants and with respect to SSC and ENR in the ΔEL complemented strain (Fig. 3). No detectable impact of hBD-2 was observed regarding FSC, PRM, or CDP at either pH for either the ΔgraS or ΔEL mutants (Fig. 3 and and44).
At pH 7.5, RP-1 exerted multiple actions in the wild-type strain, including significant increases in SSC, PRM, ANX, and CDP (Fig. 3 and and4).4). Oppositely, RP-1 caused a significant decrease in ENR in the wild-type strain at pH 7.5. Importantly, both the ΔgraS and ΔEL (albeit to a lesser extent) mutants exhibited significant increases in PRM and CDP, with concurrent significant decreases in SSC and ENR compared to those for the wild-type control (Fig. 3 and and4)4) (P < 0.05). These effects translated to increased RP-1 susceptibility compared to the control (Fig. 1 and and2).2). At pH 5.5, overall the effects of RP-1 were less extensive that those at pH 7.5 in the wild-type strain. However, both the ΔgraS and ΔEL mutants exhibited significant reductions in SSC and ENR and increases in PRM and CDP compared to the parental strain at pH 5.5 (Fig. 3) (P < 0.05). Only the ΔgraS mutant exhibited a significant decrease in ANX at pH 5.5. Complementation largely failed to restore other RP-1-induced mechanisms to wild-type levels, suggesting that RP-1 functions through mechanisms that are less amenable to GraS-mediated countermeasures that those of hNP-1 or hBD-2.
As anticipated, exposure of strains to the nonspecific detergent sodium dodecyl sulfate (SDS) caused a significant reduction in ENR and a significant increase in PRM at pH 7.5 or 5.5. These effects are clearly seen in the mechanistic signature map (Fig. 4). By comparison, each HDP exhibited effects distinct from that of SDS, which were pH dependent (Fig. 4; see Fig. S1 and S2 in the supplemental material).
The concept of adaptive bacterial systems for sensing and responding to host cues to achieve immune avoidance or subversion is well established (16). Among these mechanisms, the GraSR multicomponent system appears to be integral to S. aureus defense against innate immune effector HDPs (6, 7). The current investigation builds upon our previous work by exploring the mechanisms of resistance mediated through GraS and/or its EL in response to HDPs of relevant immune context and under cognate physiologic conditions (3, 4, 17). Several key findings which shed new light on the role of GraSR in the pathogenic relationship of S. aureus with the human host emerged from these studies.
Overall, the present data established that wild-type and mutant S. aureus strains were more susceptible to HDPs at pH 7.5 than at pH 5.5. This finding is consistent with our prior reports, demonstrating that S. aureus susceptibility to α-defensins and certain other HDPs is greater at pH 7.5 than at pH 5.5 (5, 13). Importantly, the current data revealed that distinct HDPs, even structurally related defensins of different immune compartments, exert unambiguously distinct actions against S. aureus. Furthermore, these effects appear to be pH dependent. For example, in the wild-type strain, hNP-1-mediated killing at either pH 7.5 or 5.5 correlated significantly with increases in ANX and SSC. These findings are consistent with our prior report showing that the HDP thrombin-induced platelet microbial protein 1 (tPMP-1) induces cytoplasmic membrane invagination and nucleic acid condensation correlating with chromosomal condensation, cytoplasmic refractivity, and inhibition of macromolecular synthesis (18, 19). hBD-2 exerted anti-S. aureus efficacy predominantly at pH 7.5. This efficacy was associated with significant hyperpolarization (ENR) and ANX in the wild-type strain. The staphylocidal effects of these mechanisms appear to be pH dependent, given that similar events observed at pH 5.5 were not associated with hBD-2 efficacy. In contrast, the platelet kinocidin mimetic RP-1 exerted multiple actions at both pH 7.5 and 5.5. The key mechanistic signature of RP-1 efficacy in wild-type S. aureus included simultaneous increases in SSC, PRM, ANX, and CDP, with a significant decrease in ENR. These mechanisms were amplified at pH 7.5 versus 5.5., consistent with greater RP-1 efficacy at pH 7.5. Importantly, the mechanistic signature of each HDP mentioned above was clearly distinct from that of the nonspecific detergent SDS. This observation affords cogent evidence that HDPs exert specific, targeted, and context-dependent effects that confront S. aureus and its GraSR system.
The ΔgraS and ΔEL mutants exhibited significant and equivalent increases in susceptibility to hNP-1 under pH conditions designed to reflect bloodstream (pH 7.5) or phagolysosomal (pH 5.5) immune contexts. Mechanistically, increases in hNP-1-induced killing of these mutants at pH 7.5 corresponded to decreased SSC and ANX compared to those for the wild-type strain. These results suggest that the GraS holoprotein confers adaptive responses to S. aureus that yield intracellular countermeasures manifesting as increased CM turnover, including changes in phospholipid location and/or composition. At pH 7.5, the finding that these events were equivalent in the EL and graS mutants strongly suggests that, at least for hNP-1, the EL is sufficient for the sensing function of the GraSR system. At pH 5.5, hNP-1 exhibited less efficacy than at pH 7.5 for all strains tested. However, the ΔgraS mutant exhibited greater susceptibility to hNP-1 at pH 5.5 than the wild type, and in inverse relation to ANX. Further, the ANX response was dependent on the GraS system, as both the ΔgraS and ΔEL mutants were deficient in mounting a protective ANX response observed in the wild-type parent. This theme was amplified at pH 5.5, a condition likely encountered within the neutrophil phagolysosome.
Several new insights also emerged regarding the interaction of hBD-2 with the GraS system in S. aureus. At pH 7.5, neither the ΔgraS nor the ΔEL mutation rendered S. aureus more susceptible to hBD-2. However, at pH 5.5, the ΔgraS mutation led to an increase in susceptibility to this HDP; a lesser effect was also observed with the ΔEL mutant. This pattern of results suggests novel intersections of GraS function and hBD-2 activity under distinct pH conditions. First, efficacy at pH 7.5 corresponded to impairment of ANX in the ΔgraS mutant, indicating that GraS mediates important CM turnover required for survival in response to hBD-2. Second, given that neither the charge nor the conformation of hBD-2 is significantly different at pH 7.5 versus 5.5 (13, 17), the GraSR system rather than the peptide hBD-2 may function more differently under distinct pH conditions. Third, the observation that the full ΔgraS deletion, and not the ΔEL mutation alone, conferred distinct vulnerability to hBD-2 indicates that the GraS holoprotein is more influenced by pH than is its EL. Several mechanisms underlying this relationship may exist, including conformational dynamics, the impact of a ΔpH gradient (electron motive force), or the influence of charge or osmotic conditions targeting the holoprotein rather than the EL sensor motif. Thus, it appears that at neutral pH (e.g., in the bloodstream), the holoprotein and the EL domain of the sensor are functionally equivalent. However, at acidic pH (e.g., in phagolysosomes), EL sensing of an HDP may not be as effectively transduced by the GraS holoprotein and thus may fail to convey an adaptive survival response(s) to the organism. This finding is supported by the observation that the sEL protected the ΔgraS mutant against hBD-2 at pH 5.5. Therefore, the GraS holoprotein and its EL motif appear to play complementary, nonredundant roles in sensing specific hBD-2 and conveying adaptive responses to S. aureus.
Compared to defensin hNP-1 or hBD-2, the platelet kinocidin mimetic peptide RP-1 exerted a much more complex mechanistic signature that correlated with overall greater susceptibility of S. aureus under pH conditions reflecting the bloodstream (pH 7.5) or phagolysosome (pH 5.5). The ΔgraS mutant, and to a lesser extent the ΔEL mutant, exhibited significant increases in key mechanisms of RP-1 action compared to controls. These mechanisms included increases in CM functions (increased PRM and decreased ENR) and the rapid induction of cell death-like pathway(s) (e.g., increased ANX and CDP) compared to the case for the wild type. Consistent with these findings, the sEL failed to protect any S. aureus strain from killing by RP-1 at either pH. Collectively, these findings indicate that RP-1 exerts multiple effects against which the GraS holoprotein or its EL afford protective responses. As RP-1 is not a defensin and differs markedly from hNP-1 or hBD-2 in physicochemical parameters, it is likely that HDPs of distinct structure exert differential actions against which the GraSR system may or may not convey effective resistance. Thus, the considerably greater efficacy compared to other study HDPs suggests that multiple S. aureus countermeasures are required for survival in response to RP-1 in the bloodstream or upon interactions of platelets with neutrophils in the acidic phagolysosome (3).
The current findings are consistent with prior data which suggested that GraSR is involved in stress responses and potential mechanisms of action of other peptide anti-infective agents, including daptomycin, LL-37, and the experimental agent brilacidin (20). Other studies have suggested that GraX as well as distinct two-component regulatory systems, such as VraSR and NsaSR, are also involved in S. aureus responses to antibiotics (21, 22). Likewise, Stk1/Stp1 signaling appears to cross talk with GraSR in modulating S. aureus cell wall charge (23). GraSR may have special relevance to sensing peptide-based agents, including vancomycin (glycopeptide), daptomycin (lipopeptide), and structurally or functionally similar host defense peptides.
In a broader view, the present results also support hypotheses regarding novel mechanisms by which peptide anti-infectives may target S. aureus via the GraSR or other systems (30,–33). The current data suggest that certain HDPs induce cell death pathways in S. aureus such as exist in many other pathogenic bacteria (14, 15, 24,–28). For example, when exposed to certain HDPs, S. aureus exhibits a sharp increase in annexin V binding (ANX) consistent with access to hydroxylated or otherwise negatively charged lipids (29). S. aureus is not known to make phosphatidylserine, the classic binding ligand of annexin V; whether a novel phosphatidylserine intermediate or like lipid emerges as a result of HDP exposure remains to be determined. Likewise, the present data suggest that certain peptides lead to activation of proteases associated with cell death (CDP) which cleave amino acid motifs of caspase- or metacaspase-like substrates. These findings align with responses to HPD actions reflecting pathways (14, 15) that are initiated by altered membrane integrity (e.g., permeability [PRM] and potential [ENR]), leading to chromosomal condensation (e.g., SSC [18; this study) and perturbed macromolecular synthesis (19) and yielding rapid and irreversible cell death (see Fig. S2 in the supplemental material). GraS-mediated countermeasures to these mechanisms appear to involve cell membrane potential (ENR) equilibration and membrane lipid turnover (ANX) to limit further HDP-mediated injury and minimize HDP access to intracellular targets. Thus, deficient GraS holoprotein or EL-mediated functions render S. aureus less able to adaptively respond to HDPs and more vulnerable to killing by certain HDPs. While HDP-mediated cell death in S. aureus has been understudied to date, the current findings suggest coordinated programs that have the potential to be targeted by novel anti-infective peptides or other agents.
Limitations of the current studies should also be understood. Most importantly, studying the effects of HDPs in vitro cannot fully recapitulate the complex conditions under which they function in vivo. In addition, only selected mechanistic parameters were studied in the current investigations. Although our novel approach to define mechanistic signatures led to new insights into HDP interactions with S. aureus, there are likely other mechanisms of HDP action that are counteracted by GraS/EL which contribute to differences in susceptibility under distinct conditions. Studies using increasingly specific GraS or EL mutations, broader GraSR TCS disruptions, and more specific mechanistic probes should afford a more complete understanding. Investigation of potentially unique structure-mechanism relationships of HDPs with the GraSR system is in progress in our laboratory. Nonetheless, the present results offer new insights into the host-pathogen relationships between HDPs and S. aureus adaptive resistance responses, which are likely more complex than has been understood previously.
This study was supported in part by grants from the U.S. National Institutes of Health (1 R21/R33-AI-111661-01 [M.R.Y.], 2 R01-AI091801-06 [A.L.C.], and 5 RO1-AI-039108-17 [A.S.B.]).
Supplemental material for this article may be found at http://dx.doi.org/10.1128/IAI.01030-15.