Role of the Staphylococcus aureus Extracellular Loop of GraS in Resistance to Distinct Human Defense Peptides in PMN and Invasive Cardiovascular infections

ABSTRACT

GraS is a membrane sensor in Staphylococcus aureus that induces mprF and dltABCD expression to alter the surface positive charge upon exposure to cationic human defense peptides (HDPs). The sensing domain of GraS likely resides in the 9-residue extracellular loop (EL). In this study, we assessed a hospital-acquired methicillin-resistant S. aureus (HA-MRSA) strain (COL) for the specific role of two distinct EL mutations: F38G (bulk) and D/35/37/41K (charged inversion). Activation of mprF by polymyxin B (PMB) was reduced in the D35/37/41K mutant versus the D35/37/41G mutant, correlating with reduced surface positive charge; in contrast, these effects were less prominent in the F38G mutant but still lower than those in the parent. These data indicated that both electrostatic charge and steric bulk of the EL of GraS influence induction of genes impacting HDP resistance. Using mprF expression as a readout, we confirmed GraS signaling was pH dependent, increasing as pH was lowered (from pH 7.5 down to pH 5.5). In contrast to PMB activation, reduction of mprF was comparable at pH 5.5 between the P38G and D35/37/41K point mutants, indicating a mechanistic divergence between GraS activation by acidic pH versus cationic peptides. Survival assays in human blood and purified polymorphonuclear leukocytes (PMNs) revealed lower survival of the D35/37/41K mutant versus the F38G mutant, with both being lower than that of the parent. Virulence studies in the rabbit endocarditis model mirrored whole blood and PMN killing assay data described above. Collectively, these data confirmed the importance of specific residues within the EL of GraS in conferring essential bacterial responses for MRSA survival in infections.

INTRODUCTION

Our laboratory and others have shown previously that the two-component regulatory system (TCRS), GraRS, in Staphylococcus aureus plays a key role in mediating resistance to distinct cationic host defense peptides (HDPs) by inducing transcription of downstream resistance genes, such as mprF and dltABCD, to alter the net surface positive charge and create a charge repulsion milieu (1–7). Recently, it was demonstrated that graRS-mediated transcription of mprF and dltABCD in S. aureus appears to be induced by specific HDPs derived from platelets (e.g., RP-1 and tPMPs) and a cationic peptide antibiotic of bacterial origin (polymyxin B [PMB]) (1, 8, 9) but not to the human skin peptide, hBD-2. These data suggested GraS selectively senses and responds to specific cationic antimicrobial peptides in the context of distinct infection sites, depending on their genetic backgrounds and recognition of key structural and mechanistic HDP determinants.

GraS is a membrane sensor that contains two transmembrane segments framing a single 9-amino-acid extracellular loop (35-DYDFPIDSL-43), designated EL here. Deletion of graS (ΔgraS) or its EL (ΔEL) in community-acquired methicillin-resistant S. aureus (CA-MRSA) strain MW2 (USA400) abolished induction of mprF and dltABCD transcription upon exposure to selective cationic peptides, leading to increased susceptibilities to cationic PMB and calcium-daptomycin (1, 9, 10). The important role of the GraS EL and specific amino acids on the induction of mprF and dltABCD was recently revealed by using residue-specific S. aureus mutants of MW2 (1). These results demonstrated that in CA-MRSA strain MW2, the 9-amino-acid EL is likely an important determinant of the GraS membrane sensor for detecting and initiating staphylococcal adaptive responses to certain cationic peptides. Although GraRS is highly conserved in staphylococcal genomes, CA-MRSA and hospital-acquired MRSA (HA-MRSA) strains do harbor different pathogenic phenotypes and genotypes (11). In the case of MW2 and HA-MRSA isolate COL, the 346-residue GraS is almost identical, with only one conserved substitution in residue 59 (I to L) within the 2nd transmembrane segment. Likewise, the GraR response regulators, at 224 residues long, are almost identical between these two strains, with a D-to-Q substitution in strain COL at residue 148. The D148 residue in GraR in MW2 is not the predicted phosphorylation residue usually located at the N terminus.

In the current investigation using strain COL, we examined two new and different sets of mutations, within the EL, namely F38G (bulky residue) and D-35-37-41K (charge inversion), with respect to survival in the HDP, hNP-1, and induction of mprF with PMB. The goal was to assess additional sense-response relationships between GraS and HDPs differing in structure and mechanism of action, albeit in a different strain background. Besides susceptibility to a range of relevant cationic peptides (e.g., PMB, hNP-1, and LL-37) and induction of mprF with PMB, survival of these mutants in whole blood and polymorphonuclear leukocytes (PMNs) was also evaluated. The impact of EL point mutations on GraS protein expression and stability was also examined in Western blot analysis using GraS deletion and point mutants. As HA-MRSA is frequently found in bloodstream infections, the in vivo impact of these mutants was also validated in a rabbit infective endocarditis (IE) model, featuring multiple target organ virulence profiles.

RESULTS

Impact of GraS EL mutations on in vitro susceptibility to cationic peptides and HDPs.

Given the divergence in phenotypic profiles and clinical impacts between CA-MRSA and HA-MRSA, we wanted to determine if the graS mutant of HA-MRSA strain COL was also susceptible to positively charged, peptide-based antibiotics, such as PMB and calcium-daptomycin. As shown in Table 2, the graS mutant of COL was highly susceptible to PMB (16-fold MIC decrease) (120 versus 7.5 μg/ml) compared to the parent, concordant with our previous findings for the graS mutant of MW2 (1) and in agreement with the sequence conservation of the graRS locus between these two strains. The graS mutant of COL also exhibited a 3-fold increase in susceptibility to calcium-daptomycin compared to the parent (similar to the graS mutant of MW2 versus parent [1]). Likewise, introduction of the vector pG164_Tara into the graS mutant did not alter susceptibilities of the mutant to PMB and daptomycin versus the parental strain COL(pG164_Tara). By comparison, no significant difference in susceptibility to vancomycin (a noncharged peptide antibiotic) was detected between the graS mutant and COL, like the MW2 strain (1). Collectively, these results reflect phenotypic similarities in HDP-specific susceptibility in graS mutants between HA-MRSA and CA-MRSA genetic backgrounds.

TABLE 2.

MICs of the Staphylococcus aureus study strains

Strain	MIC (µg/ml) for:
	DAP	VAN	PMB
COL	1.5	3	120
ΔgraS	0.5	3	7.5
ΔgraScomp	1.5	3	120
F38G	1.5	3	60
P39H	0.75	3	60
F38A/P39A	0.75	3	60
F38G/P39G	0.75	3	60
D35K/D37K/D41K	0.5	3	15
D35G/D37G/D41G	0.75	3	15
COL(pG164_Tara)	1.5	3	60
ΔgraS(pG164_Tara)	0.75	3	7.5
ΔgraS(pG164_Tara::graRS)	1	3	30
ΔgraS(pG164_Tara::graRSF38G)	1	3	30
ΔgraS(pG164_Tara::graRSD35/37/41K)	0.75	3	15

To determine specific structural and physiochemical features of the EL (35-DYDFPIDSL-43) in the GraS sensor that confer resistance to structurally distinct cationic peptides and HDPs in strain COL, we focused on specific mutations that were not evaluated before in CA-MRSA, MW2. Specifically, we used the graS mutant of COL with the plasmid pEPSA::graRS as a backbone to introduce mutations for analysis. We concentrated on F38G and D35/37/41K to target the effect of bulk and charge inversion mutations on GraS-related phenotypes, respectively. In addition, we also evaluated the double F38A/P39A and F38G/P39G mutants, which are new in this study, recognizing that the P39A mutation that had a decreased conformational strain had only a moderate effect on PMB sensitivity in prior studies (1). As detailed in Table 1, the structural impacts of all our mutations included altered bulk (F38G), conformational strain (P39A), charged neutralization (D35-37-41G), or charge inversion (D35-37-41K) targeting EL residues F38, P39, D35, D37, and D41, both singly and in double and triple combinations (1).

TABLE 1.

Strains used in this study

Strain or plasmid	Description	Point mutation structural impact	Reference or source
S. aureus
COL	MRSA, wild-type strain with empty pEPSA5 plasmid		20
ΔgraS	ΔgraS in-frame deletion mutant of COL with empty pEPSA5 plasmid		Present study
ΔgraScomp	ΔgraS complemented with pEPSA5 expressing graRS from COL wild-type strain		Present study
F38G	ΔgraS complemented with pEPSA5 expressing graRS but carrying F→G at position 38 generated by site-directed mutagenesis	Decrease in hydrophobicity and steric bulk	Present study
P39H	ΔgraS complemented with pEPSA5 expressing graRS carrying P→H at position 39 generated by site-directed mutagenesis	Decrease in conformational strain	Present study
F38A/P39A	ΔgraS complemented with pEPSA5 expressing graRS carrying F→A and P→A at positions 38 and 39, respectively. Mutations generated by site-directed mutagenesis	Decrease in steric bulk and conformational strain	Present study
F38G/P39G	ΔgraS complemented with pEPSA5 expressing graRS carrying F→G and P→G at positions 38 and 39, respectively. Mutations generated by site-directed mutagenesis	Decrease in steric bulk and conformational strain	Present study
D35/37/41K	ΔgraS complemented with pEPSA5 expressing graRS carrying D→K at positions 35, 37, 41. Mutations generated by site-directed mutagenesis	Charge inversion (negative-to-positive)	Present study
D35/37/41G	ΔgraS complemented with pEPSA5 expressing graRS carrying D→G at positions 35, 37, 41. Mutations generated by site-directed mutagenesis	Charge neutralization (negative-to-neutral)	Present study
COL(pG164-Tara)	Wild-type COL strain with empty pG164-Tara plasmid	pG164_tara containing arabinose promoter driving T7 polymerase expression and the T7 promoter is inducible by IPTG for gene expression	Present study
ΔgraS(pG164_Tara)	ΔgraS with empty pG164_Tara plasmid		Present study
ΔgraS(pG164_Tara::graRS)	ΔgraS with pG164_Tara expressing wild-type GraRS	ΔgraS complemented by pG164_Tara::graRS	Present study
ΔgraS(pG164_Tara::graRSF38G)	ΔgraS with pG164_Tara expressing GraRSF38G	ΔgraS carrying pG164_Tara::graRS with F38G mutation	Present study
ΔgraS(pG164_Tara::graRSD35/37/41K)	ΔgraS with pG164_Tara expressing GraRSD35/37/41K	ΔgraS carrying pG164_Tara::graRS with D35/37/41K mutation	Present study
E. coli DH5α	Host strain for construction of recombinant plasmids
Plasmids
pEPSA5	Shuttle vector for ectopic gene expression in S. aureus		30
pMAD	Allelic replacement vector to generate S. aureus mutant strain		28
pG164_Tara	Shuttle vector with IPTG-inducible T7 promoter and T7 polymerase gene driven by an arabinose-inducible promoter		This study

In these COL mutant strains, the double (F38A/P39A and F38G/P39G) point mutants exhibited a small reduction in MICs to daptomycin (7.5 versus 1.5 µg/ml) and PMB (60 versus 30 µg/ml) versus the parent COL (Table 2). This change was similar to the single mutants (F38G and P39H) in PMB sensitivity, indicating the double mutation did not confer additional sensitivity to PMB versus respective single mutants. With regard to the triple point mutants (D35/37/41G or D35/37/41K), both exhibited increased susceptibility to PMB comparable to the graS mutant, while only the triple D-to-K mutant was more sensitive to daptomycin at the level of the ΔgraS mutant. These findings indicate that charge inversion is more crucial than charge neutralization to daptomycin susceptibility; susceptibility to PMB remains relatively indifferent to charge alterations.

Susceptibility of the GraS EL mutants to skin-derived HDPs.

HA-MRSA commonly causes skin and soft tissue, as well as vascular, infections in the hospital setting (e.g., surgical site and catheter infections). Thus, we sought to determine the susceptibility of the COL strain set to HDPs relevant to skin, PMNs, or platelets, including hBD-2 (skin), LL-37 (skin and PMNs), hNP-1 (PMNs), and RP-1 (a congener of platelet tPMP-1) (Fig. 1A). Although all constructs, except P39H, revealed comparable susceptibility to LL-37 versus the parent, the ΔgraS deletion mutant became more susceptible to hNP-1 than the parent (restored in the complemented mutant). Among the point mutants, only the D35/37/41K and D35/37/41G triple mutants exhibited susceptibility to hNP-1, although this effect was more prominent with mutation in charge neutralization (D to G) than in charge inversion (D to K). Similarly, susceptibility to hBD-2 was increased in both D-to-K and D-to-G triple mutants at a level comparable to the ΔgraS mutant. We have also presented these data as a cluster analysis to demonstrate correlations between study strains and HDPs (Fig. 1B). Together, these findings suggest charge and steric bulk impact EL sensing of hBD-2 differently than hNP-1. Nevertheless, it should be emphasized that MIC determinations are not equivalent to sensing (e.g., mprF activation), as the latter event is relatively transient, lasting 30 min to 2 h (unpublished observation), while the MIC is an overnight readout, attributable to a combination of efflux activity and residual sensing effects.

FIG 1.

In vitro susceptibilities of the HA-MRSA COL strain set to hNP-1, hBD-2, and LL-37. (A) In vitro zone of inhibition (ZOI) assays were carried out with hNP-1, hBD-2, and LL-37 (10 μg/well) as described previously (10, 40). RP-1 was used as a positive control for antimicrobial peptide efficacy. The 9-amino-acid extracellular loop (-DYDFPIDSL-) is likely an important determinant of the GraS membrane sensor function. Specific mutation and controls were represented by the color scheme. These data represent the means (±SD) from three independent experiments. *, P < 0.01; ^†, P < 0.05 versus COL parental strain. ddH2O, double-distilled water. (B) Cluster analysis of correlations among study strains and HDPs. Relative correlational distances (clades) and susceptibility (color scale) illustrate impacts of comparative mutations and their relationships to specific HDPs. Unsupervised cluster analysis was performed by applying unit variance scaling using ClustVis 1.0 (41).

Effect of GraS EL mutations on induction of mprF expression.

Our cadre of mutants was exposed to PMB under sublethal conditions (30 μg/ml for 30 min) in vitro, followed by analyses of mprF transcript levels by quantitative reverse transcription-PCR (qRT-PCR) (Fig. 2). As predicted (1, 9), the expression of mprF (∼150-fold increase) was significantly upregulated in the parental strain COL upon exposure to PMB. In contrast, the ΔgraS strain did not exhibit any induction of mprF by PMB; complementation with a multicopy plasmid, pEPSA5::graRS, only partially restored this defect (Fig. 2), primarily due to the overexpression of graRS in the complemented mutant.

FIG 2.

Induction of mprF (A) and dltA (B) transcription in HA-MRSA COL strain set by polymyxin B. Gene expression analyses with qRT-PCR were performed on RNA samples from cultures of the study strains exposed to polymyxin B (30 µg/ml) for 30 min during exponential growth. +, PMB treated (30 µg/ml); −, PMB nontreated. Fold induction of mprF (A) by PMB: COL, 150.4; ΔgraS, 1.0; ΔgraS_comp, 14.8; F38G, 14.8; P39H, 7.7; F38A/P39A, 4.4; F38G/P39G, 13.0; D35/37/41G, 3; D35/37/41K, 0.3. Fold induction of dltA (B) by PMB: COL, 70.3; ΔgraS, 0.9; ΔgraS_comp, 15.1; F38G, 12.5; P39H, 13.5; F38A/P39A, 4.2; F38G/P39G, 12.0; D35/37/41G, 0.9; D35/37/41K, 1.2. *, P < 0.01 versus condition without PMB treatment. †, P < 0.01 versus graS mutant complemented with pEPSA5::graRS (ΔgraS_comp). P < 0.01 when comparing the D-to-K triple mutant versus D-to-A triple mutant.

The F38G mutation exhibited reduced mprF activation versus the parent, but the degree of reduction was comparable to the complemented ΔgraS strain, which carried the plasmid pEPSA5::graRS, accompanied by a mild growth defect due to overexpression. Notably, the double F38A/P39A mutant significantly decreased expression of mprF in the presence of PMB compared with the complemented mutant (Fig. 2). This result contrasts with F38G/P39G, with higher levels of mprF expression with PMB versus the F38A/P39A mutant but significantly less than the parental strain. This finding indicates that steric bulk (F38A) and relaxation of conformational restraint (F38G) of phenylalanine at position 38 in the context of the native 3-dimensional structure of the EL of GraS principally influences inductive responses of GraS to PMB. Remarkably, the D35/37/41K mutation in COL led to significantly lower mprF induction with PMB than the complemented mutant at a level approaching the graS mutant but lower than the D35/37/41G mutation (Fig. 2). Overall, we found the mprF response to HDPs to be robust (1, 7, 9). Collectively, these data indicate induction of mprF by PMB is dependent on the sensor kinase, GraS, with its anionic aspartic acid (D) residues, as well as the influences of F38 and P39 on the net 3-dimensional structure of the EL in GraS-mediated responses to specific HDPs. As the importance of P39 in EL of GraS has been previously evaluated (1), we focus on D35/37/41K and F38G mutations in the COL strain in this study.

GraS expression in COL constructs.

To confirm stable expression of mutated GraS proteins in our COL constructs, Western blot analyses were performed on the strains expressing F38G and D35/37/41K mutations. As mentioned above, complementation of the ΔgraS mutant with pEPSA5::graRS led to a mild growth defect. To avoid this issue in ensuing experiments, an inducible shuttle vector, pG164_Tara, a derivative of pSK236 containing an isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible T7 promoter and a T7 RNA polymerase gene driven by an arabinose-inducible pBAD promoter, was deployed for tight expression. After induction with IPTG and arabinose, bacterial cells were lysed, followed by analysis with SDS-PAGE, and probed by Western blotting with murine anti-GraS antibody raised in our lab. As shown in the gel in Fig. 3A, equivalent amounts of protein samples were loaded into each lane, but without an internal constant marker, these data are meant to be qualitative rather than quantitative. The blot revealed the purified His₆-tagged GraS band (<41 kDa), which was used as a positive control (Fig. 3B, lane 1). This band is absent from the graS mutant but present in the complemented mutant, the F38G mutant, and the D35/37/41K mutant. These data indicated intact expression of either the native or mutated GraS protein in our COL constructs.

FIG 3.

SDS-PAGE (A) and Western blot analysis (B) of S. aureus strains with native and mutated GraS proteins. Lysates of ΔgraS strains with empty pG164 plasmid or pG164 plasmids expressing wild-type GraS, GraS_F38G, or GraS_D35/37/41K. The parental strain COL was used as the positive control. The blot was probed with a mouse anti-GraS antibody raised in our lab at 1:1,000 dilution and then a secondary horseradish peroxidase-conjugated goat anti-mouse antibody at 1:5,000 dilution. The experiments were carried out three times, and representative images of SDS-PAGE and blotting are presented. Notice that GraS is lower in molecular weight than the purified His₆-tagged GraS (41 kDa). The extraneous bands are likely protein A or degradation products of protein A.

Correlation of net cell surface charge with GraS EL mutations.

A consequence of mprF expression is the lysination of membrane lipid (phosphatidylglycerol) (1, 12), leading to a gain in net positive surface charge. We can deduce the relative surface positive charge based on the amount of positively charged cytochrome c bound if we set the parental strain COL at 100%. Using this scheme, we found that deletion of graS in COL resulted in a decrease in net surface positive charge versus the isogenic parent COL or complemented mutant strain (P < 0.001) (Fig. 4). Reductions in cationic surface charge were also observed in the triple mutant D35/37/41K versus the wild type and the complemented mutant (P < 0.01), approaching that of the graS mutant with pG164_Tara and correlating with reduced mprF expression (Fig. 2). In contrast, the F38G mutant retained a surface positive charge that was comparable to that of the parental strain as well as the ΔgraS mutant complemented with pG164_Tara::graRS (Fig. 4).

FIG 4.

Binding of positively charged cytochrome c to whole cells from the HA-MRSA COL strain set. Bacterial cultures were grown overnight in MHB containing 0.02% arabinose and 1 mM IPTG. The percentage of cytochrome c bound after 15 min of incubation with a particular S. aureus strain at room temperature. Data represent the means (±SD) from three independent experiments. The COL parental strain was normalized to 100% for comparison. *, P < 0.01 versus COL parental strain. P < 0.01 when the tripled D-to-K mutant was compared to the complemented mutant.

The effect of acidic pH on mprF expression in selective graS mutants.

Recent studies suggest that GraS can sense pH to activate mprF expression, especially in the context of phagolysosomes within phagocytic cells (13). Accordingly, we wanted to ascertain if our mutants of interest would activate mprF in acidic versus neutral pH. Using the strain set delineated in Fig. 4, we found by qRT-PCR that mprF expression in parental strain COL was indeed higher at pH 5.5 (phagolysosomal pH) than at 7.5 (Fig. 5), while activation was intermediate at pH 6.5, corroborating with the findings of Flannagan et al. (13). In contrast, response to all pH levels with mprF activation was absent in the graS mutant but restored upon complementation (Fig. 5). Remarkably, both graS mutants carrying pG164_Tara::graRS-F38G (F38G mutant) or pG164_Tara::graRS-D/35/37/41K (D-to-K mutant) exhibited similarly reduced levels of mprF activation versus the complemented mutant at pH 5.5; both mutants are also less responsive at pH 6.5 and 7.5, although the magnitude of decrease was higher at pH 6.5 (49% reduction versus complemented mutant) than at pH 7.5 (29% decrease versus the complement) (Fig. 5). This finding highlighted the differential effect of pH and PMB on mprF activation, with the effect more prominent for PMB than acidic pH. In addition, the F38G and D35/37/41K mutants responded similarly to acidic pH versus the complemented mutant, while the response to PMB revealed lower mprF activation in the D35/37/41K mutant than the F38G mutant (Fig. 2).

FIG 5.

Effect of acidic pH on mprF transcription in MRSA COL strain set. Bacterial cells were exposed to growth in pH 5.5, 6.5, or 7.5 for 1 h prior to RNA extraction. qRT-PCR was conducted on cDNA (0.5 μg) derived from extracted RNA, using mprF primers. Each assay in each group was conducted in triplicate, with the results expressed as the means with one standard deviation. First group (from left), COL; 2nd group, graS mutant with pG164_Tara; 3rd group, graS mutant with pG164_Tara::graRS; 4th group, graS mutant with pG164_Tara::graRS F38G; 5th group, graS mutant with pG164_Tara::graRS D35/37/41K. The asterisks indicate P values of <0.01 (**) and <0.05 (*), respectively.

Survival of GraS EL mutants in human whole blood.

To determine if our graS mutant and selective EL point mutants were more effectively killed by human whole blood (where a variety of HDPs would be expected to exist), we exposed log-phase bacteria to whole blood for 1 h, followed by quantitative subculture. As shown in Fig. 6, the graS mutant survived at ∼25% of the initial inoculum, much lower than the parent, COL (60% survival) (P < 0.01). This survival defect was restored to the parental level upon graRS complementation. Of interest, the F38G mutation, which decreases EL hydrophobicity and steric bulk, led to a lower survival percentage than the parent but higher than the graS mutant. Notably, the D-to-K triple point mutant (D35/37/41K) also exhibited a lower survival rate (∼35%, P < 0.05) than the parent and F38G mutant but higher than the graS mutant (Fig. 6).

FIG 6.

Survival of HA-MRSA COL strain set in human blood. Bacterial cells were grown to mid-exponential phase (OD₆₀₀, 0.3 to 0.4) in MHB, and 0.02% arabinose and 1 mM IPTG were added to induce transcriptions of T7 polymerase and graRS, respectively. Data represent means (±SD). *, P < 0.01; **, P < 0.05.

Role of PMNs in killing EL mutants of GraS.

As PMNs are the major phagocytic cells in human blood, we compared our findings with whole human blood with the intracellular survival of our constructs within PMNs. As shown in Fig. 7, ∼22% of the graS mutant survived within human PMNs ex vivo compared with 36% survival of the parent COL strain. As with the whole-blood assay, the survival defect of the graS mutant in PMNs was restored upon graS complementation. Of note, the survival rate of the triple D-to-K mutant in PMNs was comparable to that of the ΔgraS mutant, while the F38G mutant survived at a higher rate than the triple D-to-K mutant but significantly less than the parent. Together, these observations suggest that the charge (D to K) and bulk (F38G) of the GraS EL are critical factors in conferring GraS the ability to sense and respond to the intra-PMN environment, promoting survival of the infecting MRSA strain.

FIG 7.

Survival of HA-MRSA COL strain set with purified human neutrophils. Purified human PMNs were infected with selective MRSA strains at an MOI of 4 and incubated for 1 h with rotation. Reactions were plated on TSB plus chloramphenicol and surviving CFU enumerated. Data are expressed as means with standard deviations. *, P < 0.01.

Impact of EL mutations on in vivo virulence in experimental IE.

In previous studies with the CA-MRSA strain MW2 in the rabbit IE model (1), its isogenic graS mutant exhibited ∼5- to 6-log₁₀ CFU/g reductions in target tissue counts versus the parental strain. To relate these findings to HA-MRSA strain COL, the numbers of CFU in cardiac vegetations, kidneys, and spleens of its isogenic graS strain were determined in the same experimental IE model. Similar to MW2, the ΔgraS mutant of strain COL displayed much lower CFU densities in all three target tissues versus the parent at 24 h postinfection (reductions ranging from ∼3 to 5 log₁₀ CFU/g versus parent COL, P < 0.01) (Table 3). The ΔgraS strain complemented with the native graRS displayed near-parental CFU densities in all three target tissues. We also ascertained the in vivo virulence of the triple D-to-K (D35/37/41K) and the F38G mutants (not evaluated previously in vivo in the MW2 mutant [1]). The triple D-to-K mutant exhibited significantly reduced target tissue CFU densities versus the parental strain. The F38G mutant had moderately diminished CFU densities in all three target tissues versus the parent, but it was higher than that of the graS mutant.

TABLE 3.

Comparative in vivo virulence of the ΔgraS strain

Straina	Mean ± SD log10 CFU/g of tissue
	Vegetation	Kidney	Spleen
COL(pG164_Tara), n‡ = 6	8.81 ± 0.43	5.59 ± 0.54	5.26 ± 0.26
ΔgraS, n = 6	3.19 ± 0.26b,c	2.09 ± 0.30b,c	2.45 ± 0.60b,c
ΔgraS(pG164_Tara::graRS), n = 4	7.25 ± 0.43	5.33 ± 0.56	5.83 ± 0.29
ΔgraS(pG164_Tara::graRSF38G), n = 6	6.60 ± 0.50	4.28 ± 0.55b	4.69 ± 0.44
ΔgraS(pG164_Tara::graRSD35/37/41K), n = 6	4.57 ± 0.67b,c	2.98 ± 0.57b,c	3.14 ± 0.86b,c

^an, number of animals.

^bP < 0.001 versus COL strain.

^cP < 0.0003 versus ΔgraS(pG164_Tara::graRS) strain.

DISCUSSION

We and others have previously shown that the GraRS TCRS promotes resistance to distinct HDPs via upregulation of downstream target genes, including mprF and dltABCD, to modify the relative surface positive charge (6, 9, 14–17) to reduce HDP binding via electrostatic repulsion. In a prior study, we conducted site-directed mutagenesis of the 9-amino-acid EL sensor of GraS in CA-MRSA strain MW2, in part because the ΔEL mutant had increased susceptibility to multiple HDPs comparable to the ΔgraS mutant (1). In that study, we introduced a limited set of EL mutations, including P39A (reduced conformation strain) and D35/37/41G (charge neutralization); these investigations showed that (i) the triple mutant was defective in activating mprF expression in vitro upon PMB induction and (ii) these mutants exhibited a notable reduction in virulence versus the parent MW2 in the rabbit IE model (1, 9). These results implied that the physicochemical signature conferred by specific EL amino acid residues is crucial to sense-response functions that result in activation of GraS and its downstream effectors (1, 9). Given the divergence in clinical virulence between CA-MRSA and HA-MRSA strains (11), we initially sought to ascertain the effects of double mutations (F38A/P39A and F38G/P39G) as well as additional new EL mutations (F38G and D35/37/41K) in a prototypic HA-MRSA strain, COL, not evaluated previously (1).

While all single and double EL point mutants of COL exhibited small but incremental increases in susceptibility to PMB and daptomycin (Table 2), it is notable that the triple point mutants behaved divergently toward these two cationic peptide antibiotics. More specifically, both triple point mutants revealed similar hypersusceptibility to PMB (15 versus 120 μg/ml for the parent), while the triple charge inversion mutant (D to K) was more susceptible to daptomycin than the charge neutralization mutant (D to G). This finding is interesting for two reasons. First, alteration of charge (neutralization and inversion) has identical effects on PMB susceptibility, whereas charge inversion leads to increased susceptibility to daptomycin. Second, increased daptomycin susceptibility of the triple D-to-K mutant implies that anionic charge contributes, at least in part, to sensing of cationic antimicrobial agents, including HDPs.

We also examined the in vitro susceptibilities of the COL strain set to three prototypical HDPs that are relevant to skin infections (hBD-2, LL-37, and hNP-1), a very common clinical syndrome in MRSA isolates. Like the ΔgraS mutant, only the D-to-K and D-to-G triple mutants were more susceptible to hBD-2 than the parent, COL. Thus, both charge and conformational signatures of the EL of GraS appear to contribute to the ability of the GraS EL to survive such HDPs. Interestingly, the susceptibility to LL-37 is generally similar among various mutants and, thus, appears to be independent of GraS in the zones of inhibition (ZOI) assay. However, it should be stressed that the ZOI assay, similar to the MIC assay, evaluates a combination of efflux (attributable to VraFG efflux pump function, downstream of graRS) (18) and the residual effects of HDP sensing (e.g., mprF), while initial HDP sensing by itself is usually transient in nature. However, the selectivity of efflux by VraFG on specific HDPs is not known.

We subsequently focused on F38G and charge inversion of D35/37/41K mutations in the EL of COL to explore the specific physicochemical signatures necessary for activating mprF. As anticipated, the D-to-K triple mutant, similar to the D-to-G triple point mutant, showed little induction of mprF expression with PMB, similar to the ΔgraS mutant (Fig. 2). Notably, the reduction in mprF expression was larger with the D-to-K mutant than the D-to-G mutant versus the parent and complemented mutant. We also explored the effect of steric bulk within the EL of GraS on mprF expression. While the F38G/P39G mutant exhibited levels of mprF transcript comparable to the complemented ΔgraS mutant with pEPSA5::graRS, the F38A/P39A double mutant (which reduces the overall bulk and conformational strain in the EL) led to significantly reduced expression of mprF in the presence of PMB. We surmise that the steric bulk of phenylalanine (F38) contributes to the ability of the EL to convey its sense-response functions on induction by selected cationic agents such as PMB or HDPs.

PMNs are recognized as key components of innate immunity in S. aureus infections, being recruited early to the site of infections (e.g., skin). Phagolysosomes of PMNs, in turn, contain many HDPs that are released into this compartment (i.e., predominantly hNP-1 and, to a lesser extent, LL-37). In a recent study, Flannagan et al. showed that acidic pH evokes GraS signaling within phagocytic cells, which, in turn, elicit an adaptive response, including activation of mprF (13). Using mprF expression levels as a readout for GraS signaling, we confirmed the results of Flannagan et al., showing that mprF expression was highest for the parent at pH 5.5, progressively decreasing at pH 6.5 and then pH 7.5; of note, the graS mutant lacked this acidic pH response (Fig. 5). Notably, the F38G mutant and the D-to-K triple mutant exhibited an intermediate level of mprF expression at pH 5.5 between the parent and the graS mutant; interestingly, this effect was still notable at pH 6.5 and 7.5 but was more muted. This response to pH, mediated by GraS, differs from induction of mprF by PMB, where the effect of the D-to-K triple mutant was much lower than that of the F38G mutant and approaching that of the graS mutant (Fig. 2). We speculate that the PMB effect is dictated mostly by charge recognition between the EL of GraS and this cationic peptide. In contrast, the pH effect may be a more generalized membrane effect. In this regard, we have recently shown that membrane-membrane interaction between GraS and VraG (the membrane permease) may modulate interaction between GraS and HDPs (18).

We also examined the impact of F38G and D35/37/41K on bacterial survival in human whole blood. The survival rate of the graS mutant was much lower than the parental COL and its graS mutant complemented with graRS (Fig. 6). The F38G mutant displayed significantly decreased survival versus the parental COL and graRS-complemented variants but higher than that of the triple mutant strain. These data suggest that the three aspartic acid and phenylalanine residues in the EL of GraS are important for S. aureus survival in whole blood.

We also evaluated the survival of the parent, D38G, and D-to-K point mutants within human PMNs ex vivo. As seen in Fig. 7, survival of the D-to-K triple mutant was equivalent to that of the ΔgraS mutant (<8%) but significantly lower than those of the wild-type and complemented strains. These findings indicate that the anionic aspartic acid residues in the EL of GraS in strain COL are critical to intra-PMN survival, likely in part due to the EL sensor detecting HDP signals within phagolysosomes (e.g., hNP-1) via electrostatic interactions. Although higher than the D-to-K triple mutant in survival, the F38G mutant also exhibited significantly reduced survival versus the parental COL strain (Fig. 7). This finding implies that the physicochemical signature of the EL of GraS requires steric bulk motifs such as phenylalanine in detecting HDPs such as hNP-1 inside PMNs to generate adaptive response (e.g., via mprF activation) to ensure survival of S. aureus within neutrophils.

To translate the above-described findings in vivo, we used the standard aortic valve IE model infected with either the parental HA-MRSA COL or its isogenic graS variant strain set. Importantly, the present study evaluated a new set of mutants that were not evaluated in previous studies with CA-MRSA strain MW2 (1). As predicted from the in vitro and ex vivo data described above, deletion of graS led to a striking microbiologic impact, with reductions of ∼3- to 5-log₁₀ CFU/g in bacterial burdens in all three target tissues. The F38G and D-to-K triple mutants also displayed marked reduction in virulence in the IE model. Collectively, these in vivo results strongly support a key role for GraS in virulence networks across both CA-MRSA and HA-MRSA strains. Moreover, we gained further insights into the role of charge and bulky residues within this GraS sensor, affording its structure-function signature and conferring MRSA survival in relevant target tissues in vivo.

A major dilemma in GraS-mediated sensing of HDPs and other cationic molecules (e.g., PMB; daptomycin) is how a small 9-residue EL can sense this divergent array of peptides. Structural prediction indicates that the GraS EL is flexible, without any well-developed secondary structure. Incidentally, GraS belongs to a unique subset of histidine kinases called intramembrane histidine kinases (IM-HK), which are common in many species of Gram-positive bacteria (e.g., Firmicutes and Actinobacteria), with each histidine kinase comprising two transmembrane helices framing a very short extracellular loop (<10 residues) for sensing (18). A large number of IM-HKs have been found to lie adjacent to ABC transporters (VraFG in the case of GraRS) genetically linked to the TCRS (16). To address the question of HDP sensing by the 9-residue EL of GraS, we recently reported that the membrane permease VraG interferes with GraS-mediated sensing by deploying a 200-residue extracellular loop to inhibit HDP sensing (18). However, we do not know if VraG binds to HDP first or disrupts the GraS-HDP complex in order for sensing to occur. These works are currently in progress.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

Strains used in this study are listed in Table 1. HA-MRSA strain COL has been studied extensively in vitro and is virulent in several animal models, including IE (19, 20). All S. aureus strains were grown in tryptic soy broth (TSB; Difco Laboratories, Detroit, MI) or Mueller-Hinton broth (MHB; Difco Laboratories) as indicated, depending on the individual experiments. Escherichia coli DH5α was grown in Luria-Bertani medium (Fisher Scientific). Liquid cultures were grown in Erlenmeyer flasks at 37°C with shaking (250 rpm) in a volume that was no greater than 10% of the flask volume. All antibiotics, purchased from Sigma Chemical Co., were used at the following concentrations: chloramphenicol, 10 µg/ml; erythromycin, 5 µg/ml; and ampicillin, 100 µg/ml.

Cationic peptides and HDPs.

Purified hNP-1, hBD-2, and the cathelicidin LL-37 (representative of HDPs encountered in human cutaneous infections) were purchased from Peptide International (Louisville, KY). RP-1 (an 18-amino-acid congener modeled in part upon the α-helical microbicidal domains of the platelet factor-4 family of platelet kinocidins) was synthesized and authenticated as detailed before (21, 22). The antistaphylococcal mechanisms of RP-1 recapitulate those of native thrombin-induced platelet microbicidal protein-1 (tPMP-1) (21). PMB was purchased from Sigma Chemicals Co. (St. Louis, MO). Peptides hNP-1, LL-37, RP-1, and hBD-2 were used for in vitro killing assays as described before (23, 24). PMB was employed in selected gene induction experiments as described previously (1, 8, 9).

Antibiotic MIC.

The MIC to PMB was determined by the standardized microdilution assay in MHB medium, per CLSI protocols, using ∼1 × 10⁵ CFU/ml in 96-well microtiter plates, with MIC assessments done at 48 h (1, 9, 25). The MICs of daptomycin and vancomycin were determined by standard Etest according to the manufacturer’s recommended protocols. For the strains carrying pG164_Tara plasmid (see descriptions of strain construction below for detail), bacterial cultures were grown overnight in the presence of 1 mM IPTG and 0.02% arabinose for gene induction. A minimum of three independent experimental assays were performed to determine MICs of each antibiotic.

HDP susceptibility.

The susceptibility of distinct graS mutants to individual HDPs was assessed using an established radial diffusion method (10). In brief, logarithmic-phase cells adjusted to 1 × 10⁶ CFU/ml were seeded into 10 ml of 1% agarose buffered to pH 7.5 using piperazine-N,N′-bis(2-ethanesulfonic acid) (PIPES, 10 mM). After 3 h of incubation, plates were overlaid with nutrient medium (Trypticase soy) containing 1% agarose 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.

DNA manipulations.

Genomic DNA was isolated from S. aureus COL as described previously (26). Plasmid DNA purification was performed using an E.Z.N.A miniprep kit (Omega Biotek, Norcross, GA). Electroporation of recombinant plasmids into S. aureus was carried out using the procedures of Schenk and Laddaga (27). Successful transformants were verified with restriction digest of recombinant plasmids, followed by PCR and sequencing of inserts.

Construction of S. aureus mutants.

The ΔgraS mutant strain was generated in strain COL with an in-frame deletion using an allelic replacement approach with temperature-susceptible plasmid pMAD as described previously (8, 28). Briefly, PCR was used to amplify an ∼2-kb fragment comprising a 1-kb fragment upstream and another 1-kb fragment downstream of graS using genomic DNA as the template. The PCR fragment was cloned into pMAD, resulting in pMAD-ΔgraS. The recombinant shuttle vector was transformed first into E. coli and then RN4220 as an intermediary and finally into S. aureus strain COL followed by temperature shifts between 43°C and 30°C to promote homologous recombination, as described previously (29). The mutation in graS was then confirmed by restriction digest and PCR followed by sequencing.

For construction of residue-specific point mutations within the EL loop of graS, we used pEPSA5 (30), a shuttle plasmid for ectopic expression in S. aureus, containing graRS as a template, whereby specific mutations were introduced into graS using the Q5 site-directed mutagenesis kit (New England Biolabs, Ipswich, MA). The recombinant plasmid with specific mutations within the EL loop of GraS was then transformed into the ΔgraS mutant of strain COL as described above (used in Fig. 1 and 2). All genetic constructs were confirmed by restriction digest and PCR, followed by sequencing. Table 1 also describes the rationale behind each of the strategic EL point mutations used in this investigation in terms of impact on structure-function metrics.

One unique feature in using the pEPSA5 system is its relative leakiness, even in the absence of the xylose inducer. This was particularly the case in the graS mutant complemented with a multicopy plasmid expressing graRS (pEPSA5::graRS) induced with xylose. This strain has a mild growth defect, presumably due to high-level expression of GraRS. For the experiments in Fig. 1 and 2, we did not need to deploy xylose for moderate expression with respect to induction. To avoid any growth issue for ensuing critical experiments related to F38G and D35/37/41K mutations within the EL loop of GraS (used in Fig. 3 and 7), we used a very tight expression system, called pG164-Tara, a pSK236-derived shuttle plasmid constructed by our lab carrying an IPTG-inducible T7 promoter for graRS expression (native or with point mutations) and a pBAD promoter for expression of the T7 RNA polymerase gene upon arabinose induction. Briefly, ΔgraS mutant strains expressing wild-type graRS, graRS_F38G, or graRS_D35/37/41K were constructed in shuttle plasmid pG164_Tara and then introduced into RN4220 and into the ΔgraS mutant (Table 1).

Preparation of RNA.

To assess induction of mprF and dltABCD by selected cationic peptides, RNA samples were isolated from cultures of distinct EL mutants and the isogenic parent exposed to either PMB or hNP-1. Briefly, overnight cultures of the strain sets diluted to an optical density at 600 nm (OD₆₀₀) of 0.1 were grown for 2.5 h (∼1 × 10⁸ to 5 × 10⁸ CFU/ml) and then exposed to PMB (30 μg/ml) for 30 min before RNA harvest. The sublethality of these peptide concentrations over the 30-min exposure period (≥99% survival) was confirmed by quantitative culture as described before (9). Total cellular RNA was isolated from the S. aureus cell pellets using the RNeasy kit (Qiagen, Valencia, CA) and the FASTPREP FP120 instrument (BIO 101, Vista, CA) according to the manufacturer’s recommended protocols.

Quantification of transcript levels by qRT-PCR.

Quantitative real-time PCR (qRT-PCR) analyses were performed as described previously (31). Briefly, two micrograms of DNase-treated RNA was reverse transcribed using the SuperScript III first-strand synthesis kit (Invitrogen) according to the manufacturer’s protocols. The primers for amplifying the mprF mRNA were qRT-mprF-F and qRT-mprF-R (9). The dltA and gyrB genes were similarly detected using respective specific primers as described previously (31).

Net cell surface charge determination.

To quantify relative cell surface positive charges that may contribute to HDP resistance (6, 7, 14, 23, 32), cytochrome c binding assays were performed as described before (8, 12). To induce expression of the wild-type and mutated graS genes, S. aureus cultures were exposed to 1 mM IPTG and 0.02% arabinose for 2 h before the assays. The binding of cytochrome c (pI 10; Sigma) is estimated from the amount of cytochrome c remaining in solution following exposure to study strains for 15 min. Larger amounts of residual cytochrome c in the supernatant correlate with a more positive bacterial surface charge (23). A minimum of four independent experiments was performed for each study strain.

Effect of acidic pH on mprF expression.

To evaluate mprF expression of selective graS point mutants under relevant acidic conditions, we exposed wild-type and mutant cells to growth in pH 7.5 (neutral), 6.5 (mildly acidic), and 5.5 (phagolysosomal pH) for 1 h, followed by harvesting and RNA extraction. In brief, wild-type cells were grown to an OD₆₀₀ of ∼0.5 at 37°C, washed with phosphate-buffered saline (PBS), and then suspended in 10 ml of TSB medium with 10 mM morpholineethanesulfonic acid, pH 5.5 or 6.5, or 10 mM Tris-HCl, pH 7.5, for 1 h. Mutant cells carrying the empty vector pG164_Tara or the recombinant vector were grown to an OD₆₀₀ of ∼0.4 at 37°C and then induced by 0.02% arabinose and 1 mM IPTG for 30 min. The induced cells were washed and resuspended in 10 ml of TSB with various pH media (as described above), including the inducers, arabinose, and IPTG, and incubated for 1 h. qRT-PCR was then conducted as described above with respective cDNA and primers for amplifying mprF (9). As a control, gyrB in cDNA was similarly detected using specific primers.

Whole-blood killing assay.

Survival of S. aureus strains in human whole blood was determined as described previously, with minor modifications (33). Ethical approval for using human blood was obtained from the IRB committee of Dartmouth College. Briefly, staphylococcal strains were grown in MHB (1:100 dilution from overnight cultures) until an OD₆₀₀ of 0.3 to 0.4, and then 0.02% arabinose and 1 mM IPTG were added to induce expression of cloned genes for 30 min. After the induction, suspensions of staphylococcal strains (10⁶ CFU in 10 µl MHB) were mixed with blood (90 µl) in triplicate and incubated at 37°C for 1 h. The cultures were then serially diluted with PBS and plated onto TSA plates to enumerate CFU.

Susceptibility to killing by human PMNs.

Human PMNs were obtained and processed according to a previously described protocol approved by the IRB committee of Dartmouth College (34). Briefly, fresh heparinized venous blood was diluted with one volume of RPMI and overlaid on 10 ml Hypaque-Ficoll. Blood components were separated by a 40-min spin at 400 relative centrifugal force. Supernatant, mononuclear band, and gradient above the PMN-erythrocyte pellet were discarded. The pellet was resuspended in Hanks’ balanced salt solution (HBSS) without Mg²⁺/Ca²⁺, and 25 ml of 3% dextran was added. Tubes were allowed to sit for 20 min at room temperature, and the PMN layer was collected and centrifuged. Contaminating erythrocytes were lysed with sterile H₂O and tonicity restored with saline solution. PMNs were washed with HBSS and resuspended to 1 × 10⁵ cells/ml in RPMI supplemented with 10% FBS. Staphylococcal cells grown to mid-log phase (OD₆₀₀ of 0.3 to 0.4) were exposed to 1 mM IPTG and 0.02% arabinose for 30 min to induce gene expression. Next, bacterial cells were opsonized in 10% human serum for 30 min at 37°C. PMNs were infected with opsonized bacterial cells at a multiplicity of infection (MOI) of 4 and rotated at 37°C for 1 h. The initial bacterial inoculum was verified by serial dilutions. Aliquots of PMNs and bacteria were serially diluted in RPMI containing 0.1% Triton X-100 to lyse the PMNs, plated on TSA plates, and incubated at 37°C overnight. Colonies were enumerated, and percent survival of the initial inoculum was calculated.

Western blot analysis for detection of GraS.

Preparation of cell lysates and Western blot assays were performed as described before (35). Bacterial cells grown to an OD₆₀₀ of ∼0.6 were exposed to 1 mM IPTG and 0.02% arabinose for 3 h to induce expression of GraRS. After induction, bacterial cells were pelleted at 4°C, washed with PBS, and then lysed as described before to yield cell lysates (35).

For Western blot analyses, SDS-PAGE was first carried out with 100 µg of lysate proteins, and the protein samples were transferred to polyvinylidene difluoride (PVDF) membranes using an iBlot 2 gel transfer system (Invitrogen). The PVDF membranes were prepared for transfer by fixing in acetone at 4°C for 30 min, heating at 50°C for 30 min, wetting in methanol at room temperature for 1 min, and then rinsing with Tris-buffered saline with Tween 20 (TBST; 150 mM NaCl and 10 mM Tris, pH 8.0, with 0.1% [vol/vol] Tween 20) (36). Following blocking with 5% bovine serum albumin in TBST for 1 h at room temperature, the PVDF membranes were incubated with mouse anti-GraS antibody (raised in our lab) at a 1:1,000 dilution in TBST for 1 h. The membranes were then washed with TBST three times, incubated in goat anti-mouse secondary antibody (1:5,000; Jackson ImmunoResearch), and washed again with TBST three times. Finally, chemiluminescence was detected using a ChemiDoc XRS+ system (Bio-Rad).

Experimental model of rabbit IE.

Animals were maintained in accordance with the American Association for Accreditation of Laboratory Animal Care criteria. The Institutional Animal Care and Use Committee (IACUC) of the Los Angeles Biomedical Research Institute at Harbor-UCLA Medical Center approved these animal studies (IACUC 13951). A well-characterized catheter-induced rabbit model of aortic IE was used as described previously (6, 37, 38) to assess the role of the GraS EL sensor in relative S. aureus virulence in vivo. Briefly, female New Zealand White rabbits (2.0 to 2.5 kg of body weight; Harlan Laboratories, Indianapolis, IN) underwent indwelling transcarotid-transaortic valve catheterization with a polyethylene catheter to induce sterile aortic valve vegetations. Twenty-four hours after catheter placement, animals were infected intravenously with ∼1 × 10⁶ CFU/animal, the 95% infective dose (ID₉₅) for the parental COL strain established in pilot studies.

For intrinsic virulence comparisons, five strains were selected for in vivo virulence study: COL strain with empty pG164_Tara, the graS knockout (ΔgraS) strain, and the ΔgraS strain complemented with pG164_Tara containing native graRS or with pG164_Tara carrying graRS with two EL point mutations (F38G and D35/37/41K). To induce expression of the wild-type and mutated GraS proteins, S. aureus cultures were exposed to 1 mM IPTG and 0.02% arabinose for 2 h to induce expression prior to harvest followed by intravenous injections. At sacrifice, cardiac vegetations, kidneys, and spleen were aseptically excised and quantitatively cultured as detailed before (39). The mean log₁₀ CFU/g of tissue (± standard deviations [SD]) was calculated for each target tissue in each group for statistical comparisons. The lower limit of microbiologic detection in the target tissues is ≤1 log₁₀ CFU/g of tissue.

Maintenance of plasmid (pG164 or recombinant pG164::graRS) within the ΔgraS complemented strain during in vivo studies was verified (>99% of the constructs grew on TSA plates containing chloramphenicol versus TSA plates without chloramphenicol) (1).

Statistical analysis.

For phenotypic and genotypic studies, including PMN killing assays, two-tailed Student's t test was used between comparable groups, with a P value of <0.05 considered significant. Statistics were performed with GraphPad Prism 6.0. For the experimental endocarditis study, bacterial tissue CFU densities among the various groups were compared using the Kruskal-Wallis analysis of variance (ANOVA) test with the Tukey post hoc correction for multiple comparisons. Significance was determined with a P value of <0.05.