Host-induced cell wall remodeling impairs opsonophagocytosis of Staphylococcus aureus by neutrophils

ABSTRACT The bacterial pathogen Staphylococcus aureus responds to the host environment by increasing the thickness of its cell wall. However, the impact of cell wall thickening on susceptibility to host defenses is unclear. Using bacteria incubated in human serum, we show that host-induced increases in cell wall thickness led to a reduction in the exposure of bound antibody and complement and a corresponding reduction in phagocytosis and killing by neutrophils. The exposure of opsonins bound to protein antigens or lipoteichoic acid (LTA) was most significantly reduced, while opsonization by IgG against wall teichoic acid or peptidoglycan was largely unaffected. Partial digestion of accumulated cell wall using the enzyme lysostaphin restored opsonin exposure and promoted phagocytosis and killing. Concordantly, the antibiotic fosfomycin inhibited cell wall remodeling and maintained the full susceptibility of S. aureus to opsonophagocytic killing by neutrophils. These findings reveal that host-induced changes to the S. aureus cell wall reduce the ability of the immune system to detect and kill this pathogen through reduced exposure of protein- and LTA-bound opsonins. IMPORTANCE Understanding how bacteria adapt to the host environment is critical in determining fundamental mechanisms of immune evasion, pathogenesis, and the identification of targets for new therapeutic approaches. Previous work demonstrated that Staphylococcus aureus remodels its cell envelope in response to host factors and we hypothesized that this may affect recognition by antibodies and thus killing by immune cells. As expected, incubation of S. aureus in human serum resulted in rapid binding of antibodies. However, as bacteria adapted to the serum, the increase in cell wall thickness resulted in a significant reduction in exposure of bound antibodies. This reduced antibody exposure, in turn, led to reduced killing by human neutrophils. Importantly, while antibodies bound to some cell surface structures became obscured, this was not the case for those bound to wall teichoic acid, which may have important implications for vaccine design.

S taphylococcus aureus frequently infects wounds caused by surgery or insertion of intravenous access devices (1,2).These infections can result in S. aureus seeding into the bloodstream, leading to bacteremia and subsequent metastatic dissemination to sites including the heart, bones, and joints (1,(3)(4)(5).Despite antibiotic therapy and a potent immune response, S. aureus infections have a high rate of relapse and frequently become chronic or recurrent (3,5).
The binding of neutrophils to opsonins on the surface of S. aureus occurs via dedicated receptors and triggers phagocytosis of the pathogen followed by the subsequent exposure of ingested bacteria to a raft of bactericidal products including reactive oxygen species, antimicrobial peptides, and proteases (11,12).
The staphylococcal cell envelope is a dynamic structure that responds to hostinduced stresses (32)(33)(34)(35).Consequently, S. aureus has a thicker cell wall in vivo than when growing in vitro (36), a phenotype that is replicated when staphylococci are exposed to human serum or present within endothelial or osteoblast cells (37)(38)(39)(40).In the case of serum, cell wall thickening is triggered when S. aureus detects the presence of the host defense antimicrobial peptide LL-37 via the GraRS two-component system (37).This results in significantly greater quantities of both peptidoglycan and WTA in the cell wall, relative to bacteria grown in laboratory culture medium (37).Importantly, the changes to the cell envelope triggered by human serum are distinct from those that occur during bacterial entry into stationary phase and are also not triggered by incubation of S. aureus in PBS or cell culture medium, that is, serum-induced changes are not simply due to a lack of nutrients or lack of staphylococcal replication, but represent a specific response to the host environment (37).
Host-induced changes to the cell wall are important for the ability of the pathogen to cause and sustain infection.Cell wall thickening has been shown to reduce susceptibil ity to antibiotics, while mutant strains lacking various cell wall synthetic enzymes are less virulent in infection models (32,34,35,37,40).However, it is unknown whether host-induced changes to the bacterial cell wall affect the detection and killing of S. aureus by the host immune system.To address this, we examined the impact of hostinduced changes to the staphylococcal cell envelope on subsequent interactions of S. aureus with neutrophils.This revealed that cell wall thickening constitutes a previously unrecognized mechanism of immune evasion that functions by significantly reducing the exposure of opsonins bound to proteins and LTA, thereby reducing opsonophago cytic killing.

Host-induced changes to S. aureus reduce killing by neutrophils
To understand the impact of the host environment on staphylococcal susceptibility to host defenses, we either grew bacteria to exponential phase in tryptic soy broth (TSB grown) to represent standard laboratory conditions or incubated S. aureus in 100% human serum (serum-incubated) to mimic host conditions as previously described, which triggers cell wall thickening (37) (Fig. 1A).This model uses pooled human serum, which avoids variability in anti-staphylococcal antibody levels between donors (14).Since the serum is not heat inactivated, it contains functional immunoglobulins and complement components.In addition to triggering cell wall thickening via the GraRS system, the serum also suppresses both the growth of S. aureus and activation of the Agr quorum-sensing system that regulates the expression of many virulence factors (37,(41)(42)(43)(44)(45)(46)(47)(48).
Following TSB growth or serum incubation, we then measured the survival of bacteria prepared under each condition during incubation with purified ex vivo human neutro phils from male and female healthy donors in the presence of fresh serum (10%) to provide antibody-and complement-mediated opsonization (9,49).We examined four distinct wild-type S. aureus strains to represent both methicillin-resistant (USA300, Col) and methicillin-susceptible (SH1000, Newman) organisms (50)(51)(52)(53).
For all four of the S. aureus strains tested, exponential phase bacteria were efficiently killed over time, with <5% of bacteria remaining viable after 2 h incubation with neutrophils (Fig. 1B through E).However, serum-incubated bacteria survived at levels up to five times greater than that seen for exponential bacteria for all strains (Fig. 1B  through E).In addition to demonstrating that serum incubation reduced staphylococcal susceptibility to host immune defenses, the high level of consistency observed across all four strains indicated that this is a conserved phenotype.
To understand whether the bacterial growth phase was important for the reduced susceptibility of serum-incubated bacteria to neutrophil-mediated killing, we repeated the assay using USA300 grown to stationary phase.We found that serum-incubated stationary phase cells were significantly less susceptible to neutrophil-mediated killing compared to TSB-grown stationary phase S. aureus, with similar levels of survival to exponential phase bacteria (Fig. 1B; Fig. S1).Therefore, the protective effect of serum incubation on S. aureus survival during exposure to neutrophils was not dependent on the bacterial growth phase.

Host-induced changes to S. aureus reduce opsonin exposure and opsonopha gocytosis
Having found that serum incubation reduced staphylococcal susceptibility to host defenses relative to TSB-grown bacteria, we next determined the mechanism(s) responsible.Given the consistency in survival data across all four strains examined, we focused on the USA300 lineage since it is both well characterized and clinically important (50).
We started by assessing whether the increase in survival of serum-incubated bacteria was due to impaired phagocytosis, using two distinct assays.Bacteria were grown in broth or incubated in serum, before being washed in PBS and then incubated with neutrophils in the presence of fresh serum to enable opsonization (Fig. 2A).The first phagocytosis assay was a flow cytometry-based approach that determined how many fluorescently labeled bacteria were associated (or not) with neutrophils (49) (Fig. S2).This revealed that the majority of both broth-grown and serum-incubated S. aureus were associated with neutrophils after 30-min incubation with the immune cells.However, while <3% of broth-grown bacteria remained unbound to neutrophils, >20% of serum-incubated bacteria were free (Fig. 2B).This finding was replicated in a second phagocytosis assay that measured the viability of free and neutrophil-associated bacteria (54), with >10% of serum-incubated S. aureus unphagocytosed compared with <1% of broth-grown S. aureus cells (Fig. 2C; Fig. S3) (55,56).Using this second assay, we also found that serum-incubated stationary phase bacteria were phagocytosed less efficiently than TSB-grown stationary phase cells (Fig. S4).
Combined, these two assays demonstrated that serum-incubated S. aureus was significantly better at evading phagocytosis than broth-grown bacteria.An additional finding was that there were equal number of viable serum-incubated and TSB-grown S. aureus cells associated with neutrophils, regardless of growth phase (Fig. 2C; Fig. S4).This indicated that serum-incubated S. aureus cells were as susceptible to the microbicides produced by the neutrophils as broth-grown cells.Therefore, we concluded that the enhanced survival of serum-incubated bacteria compared with broth-grown bacteria (Fig. 1) was due to enhanced evasion of phagocytosis, rather than resistance to the antibacterial products of neutrophils.
To understand why more serum-incubated S. aureus cells were able to evade phagocytosis compared with TSB-grown cells, we first considered whether serum caused clumping of bacteria that precluded phagocytosis.However, using microscopy, we found that bacteria incubated in serum for 16 h did not form large clumps, relative to brothgrown S. aureus, which ruled out bacterial aggregation as an explanation for reduced phagocytosis (Fig. S5).
We then examined the degree of opsonization of bacteria by antibody and comple ment using western blotting.In keeping with previous work (57,58), for this experiment, we used a mutant strain of USA300 lacking Spa and Sbi to avoid interference caused by these immunoglobulin-binding proteins (Fig. S6).TSB-grown or serum-incubated bacteria were washed in PBS and then incubated, or not, in fresh serum to enable opsonin binding as used in the opsonophagocytosis assays described above (Fig. 1A) before detection of bound antibody and complement component C3 (Fig. 2D).
Despite their reduced phagocytosis by neutrophils, there was more antibody and complement bound to serum-incubated cells than to TSB grown, suggesting that a lack of bound opsonins did not explain the immune evasion phenotype of serum-incubated bacteria (Fig. S7) (59).
To understand why serum-incubated cells had high levels of bound antibody and complement but low levels of phagocytosis, bacteria were prepared as described above for opsonophagocytosis assays and then the levels of surface-exposed antibody and the complement component C3 quantified using flow cytometry (Fig. 2D; Fig. S8).TSB-grown bacteria that had been exposed to PBS instead of serum acted as a negative control and confirmed that antibodies used in the assay did not bind non-specifically to S. aureus cells (Fig. 2E and F).We then showed that, as expected, TSB-grown bacteria that were incubated in human serum for 30 min were very strongly bound by both IgG and the complement component C3 (Fig. 2E and F).
Next, we examined serum-incubated bacteria and found that they had a significantly reduced level of exposed opsonins, compared with TSB-grown cells, regardless of whether they had been incubated in fresh serum for 30 min or not (Fig. 2E and F).Therefore, despite prolonged incubation in serum and high levels of bound antibody and complement (Fig. S7), serum-incubated cells had significantly reduced exposure of opsonins on their cell surface relative to TSB-grown bacteria that had been opsonized.
Taken together, these experiments revealed that serum-incubated bacteria are better able to survive exposure to neutrophils than broth-grown S. aureus because they are less likely to be phagocytosed, in keeping with the lower surface exposure of bound IgG and complement.

Cell wall accumulation impairs opsonophagocytosis by concealing IgG bound to LTA and protein
Since the cell envelope of S. aureus accumulates peptidoglycan and WTA during incubation in serum (37), we tested whether this concealed some of the bound antibody and complement.To do this, serum-incubated bacteria were subsequently incubated for 20 min with a range of sub-lethal concentrations of the enzyme lysostaphin, which cleaves peptidoglycan, to partially remove the cell wall.The lysostaphin was then removed by washing and bacterial viability was confirmed by CFU counts.This limited cell wall digestion resulted in a significant, dose-dependent increase in exposure of bound IgG and complement, demonstrating that some of the bound opsonins were concealed by the accumulation of cell wall polymers during incubation in serum (Fig. 3A  and B).
We then tested whether the concealment of bound IgG by accumulated cell wall explained the reduced phagocytosis of serum-incubated bacteria relative to TSB-grown S. aureus.In keeping with increased IgG and complement exposure, limited lysostaphin treatment increased the phagocytosis of serum-incubated S. aureus by neutrophils (Fig. 3C and D).
Since human serum contains IgG that recognizes multiple S. aureus surface structures, we next sought to understand whether the reduced opsonization observed for serumincubated bacteria was specific to a particular antibody target.Bacteria were grown in TSB and then incubated briefly in serum (30 min) or serum incubated (16 h).Surfaceexposed IgG was then eluted from bacteria and assessed for its binding to each of the major surface structures by ELISA.
Serum-incubated bacteria had similar levels of anti-WTA IgG on their surface com pared to exponential phase bacteria and only slightly lower levels of anti-peptidoglycan IgG (2.5-fold difference) (Fig. 3E).However, surface exposure of IgG targeting other surface structures was greatly reduced in serum-incubated compared to exponential phase cells, with anti-LTA IgG 9-fold lower, anti-membrane-associated proteins 18-fold lower, and anti-cell wall-associated proteins 48-fold lower (Fig. 3E).We did not examine anti-capsular polysaccharide antibodies in these assays as USA300 is deficient in this polymer (50).As such, the lower surface IgG exposure in serum-incubated cells com pared to those in the exponential phase is primarily due to a loss of exposure of antibody bound to LTA and surface proteins.
Partial digestion of peptidoglycan using lysostaphin restored surface exposure of IgG bound to LTA and proteins to similar levels observed for TSB-grown bacteria (Fig. 3E).Therefore, accumulation of cell wall in serum-incubated bacteria preferentially conceals IgG bound to LTA and surface proteins, while anti-WTA and anti-peptidoglycan antibod ies remain strongly exposed.
Finally, we showed that increasing opsonin exposure via partial lysostaphin digestion of the cell wall, with the enzyme washed away before incubation with immune cells, rendered serum-incubated staphylococci as susceptible to neutrophil-mediated killing as TSB-grown bacteria (Fig. 3F).
Taken together, the experiments described here demonstrate that serum-incubated S. aureus cells are bound by high levels of antibody and complement but the accumulation of cell wall conceals some of these bound opsonins, reducing phagocytosis and killing by neutrophils.

Antibiotic-mediated inhibition of peptidoglycan accumulation maintains opsonin exposure and efficient opsonophagocytosis
To further test whether serum incubation reduced phagocytosis via cell wall-mediated concealment of bound opsonins, and to explore potential therapeutic approaches to enhance neutrophil-mediated killing, we first used the antibiotic fosfomycin to block the serum-induced accumulation of peptidoglycan, as we have done previously (37).This antibiotic targets MurA, which catalyzes the production of the peptidoglycan precursor UDP N-acetylmuramic acid in the cytoplasm (60).This inhibits peptidoglycan synthesis and prevents serum-induced cell wall thickening from occurring and has been used clinically in anti-staphylococcal combination therapies (37,61).
As observed previously, serum-induced changes to S. aureus resulted in a significant reduction in opsonization, as determined by exposure of IgG and complement, relative to TSB-grown bacteria (Fig. 4A and B).However, the presence of fosfomycin in serum significantly reduced opsonin concealment, maintaining IgG and complement exposure at similar levels to that seen for TSB-grown bacteria (Fig. 4A and B).
Further analysis of IgG exposure confirmed that fosfomycin treatment preserved the exposure of IgG bound to all major surface structures relative to bacteria that had not been treated with the antibiotic (Fig. S9).Similar findings occurred with another inhibitor of peptidoglycan synthesis, oxacillin, which acts on penicillin-binding proteins (62), whereas antibiotics that targeted fatty acid biosynthesis (AFN-1252) or DNA gyrase (ciprofloxacin) did not increase IgG exposure relative to serum-incubated cells that had not been exposed to antibiotics (63, 64) (Fig. S9).Therefore, in support of our previous findings, we concluded that the accumulation of peptidoglycan during serum incubation significantly reduces the exposure of IgG bound to LTA and surface proteins.
The increased exposure of IgG and complement on the surface of bacteria incubated in serum containing fosfomycin, restored phagocytosis to levels seen with TSB-grown bacteria, as determined by both phagocytosis assays (Fig. 4C and D).Furthermore, bacteria that were incubated in serum with fosfomycin were killed by neutrophils as efficiently as TSB-grown bacteria, whereas serum-incubated bacteria not exposed to fosfomycin survived at significantly higher levels (Fig. 4E).In keeping with our analysis that AFN-1252 did not prevent IgG concealment during host adaptation, this antibiotic did not promote neutrophil-mediated killing of serum-incubated S. aureus (Fig. S10).
S. aureus anchors proteins to peptidoglycan via sortase enzymes (SrtA and SrtB)(65) and so we assessed whether the impact of fosfomycin on serum-incubated cells was due to interference with this process.However, serum-incubated mutants defective for SrtA or SrtB, which cannot anchor proteins to peptidoglycan, survived incubation with neutrophils as well as serum-incubated wild-type bacteria (Fig. S11).
Taken together, these findings provided additional evidence that serum-induced cell wall accumulation conceals opsonins bound to LTA and surface proteins, which, in turn, compromises phagocytosis and killing by neutrophils.They also indicate that the antibiotic fosfomycin, in addition to its antibacterial activity, may aid the clearance of infection by preventing the concealment of opsonins.

DISCUSSION
The binding of antibodies and complement to the bacterial cell surface enables the detection and destruction of pathogens by phagocytic immune cells (10,11,13).The data presented here demonstrate that S. aureus can conceal a subset of bound opsonins via cell wall accumulation, significantly reducing opsonophagocytosis and killing by neutrophils, a previously unrecognized mechanism of immune evasion (Fig. S12).
Cell wall remodeling occurs in response to host stresses and protects against antibiotics and host defense peptides.It involves the accumulation of peptidoglycan and WTA (36,37,66), and it is therefore unsurprising that exposure to antibodies targeting these two polymers was least affected.By contrast, the exposure of antibodies bound to surface proteins and LTA was significantly reduced by serum-induced changes to the cell envelope, in keeping with their localization within the cell wall itself (67).Previous work indicated that WTA can block antibodies from binding to antigens within the cell wall (68).Although WTA accumulates in the wall during serum incubation, it is currently unknown whether this contributes to the concealment of IgG bound to LTA or proteins.Unfortunately, since cell wall accumulation is dependent on D-alaninelabeled WTA (37), we could not use a WTA-deficient mutant to explore the role of this polymer in reducing opsonin exposure.However, our work did show that inhibition of peptidoglycan accumulation preserved antibody exposure and opsonophagocytic killing by neutrophils, demonstrating a key role in cell wall accumulation.
Host-induced peptidoglycan accumulation is due to a combination of peptidoglycan synthesis and inhibition of autolytic activity (37).Recent work has revealed that mutants lacking the Atl autolysin have defective surface exposure of staphylococcal surface proteins, inhibiting their recognition by reactive antibodies (69).Exposure of surface proteins was restored using enzymatic digestion of peptidoglycan, providing additional evidence that peptidoglycan accumulation can obscure surface antigens and prevent their detection by antibodies.However, the impact of concealment of surface proteins on opsonophagocytosis has not been investigated previously.
Several experimental vaccines have been developed based on surface proteins in an attempt to generate high serum titers of opsonizing antibodies.Unfortunately, despite very promising data from animal infection experiments, none of these vaccines have shown efficacy in humans (70,71).Several plausible reasons for this discrepancy have been proposed, including the host specificity of staphylococcal immune evasion factors and previous staphylococcal infection directing the host toward non-protective immunity (14,(71)(72)(73)(74).
Another difference between model infection of animals and natural infection in humans is the physiological state of the bacteria.For many animal infections, bacteria are grown in TSB immediately prior to administration into the animal and will therefore have high levels of multiple protein antigens exposed on their surface, which facilitates rapid opsonophagocytosis (47,72,75).By contrast, natural invasive infection typically begins with colonization of superficial sites such as an inserted IV catheter and so bacteria may be in a very different physiological state from those grown in laboratory media when they enter the bloodstream and are thus less well recognized by antibodies targeting surface proteins (73,74,(76)(77)(78).As such, the addition of WTA as a vaccine antigen may provide a reasonable level of protection against bacteria that have accumulated cell walls and thus have reduced exposure to surface proteins.
Previous work has indicated that the thickened cell wall associated with vancomycin resistance reduces staphylococcal susceptibility to intracellular killing by neutrophils (79).However, our data did not show a difference in staphylococcal survival within neutrophils, with similar numbers of intracellular viable broth-grown and serum-incuba ted bacteria.Instead, the survival advantage of host adaptation appeared to be due to enhanced evasion of phagocytosis.It has also been reported that the staphylococcal cell envelope changes as bacteria enter the stationary phase, including increased cell wall thickness and reduced cell wall-associated protein content (36,75).Therefore, we investigated whether the growth phase affected serum-induced changes to the propensity of S. aureus to evade phagocytosis.These experiments showed that serum incubation promoted evasion of phagocytosis of S. aureus grown to both exponential and stationary phases.As such, the growth phase at which bacteria encounter serum is irrelevant to subsequent cell wall remodeling and immune evasion.
We do not yet know whether these findings apply to other Gram-positive pathogens.However, since previous work has shown that serum triggers cell wall thickening in Enterococcus faecalis and viridans group streptococci, it is possible that our findings with S. aureus represent a broadly conserved mechanism of immune evasion (80).
Cell wall thickening in S. aureus is triggered by bacterial sensing of the host defense antimicrobial peptide LL-37 via the GraRS system (37).Since LL-37 is present in most tissues and among the earliest host responses to infection or trauma (81)(82)(83), we hypothesize that S. aureus has evolved to sense this AMP as an early indicator that it is subject to immune attack and provides an opportunity to employ defensive measures against the impending arrival of neutrophils.In support of this hypothesis, GraRS, the two-component system that detects LL-37, is activated in the early stages of staphylo coccal skin colonization, while S. aureus mutants lacking GraRS are significantly less virulent than wild-type strains in invasive infection models (84)(85)(86).
In addition to protecting against opsonophagocytosis, LL-37 exposure triggers reduced susceptibility to the antibiotics daptomycin and vancomycin (37,87), suggesting S. aureus employs strategies that are broadly protective against the twin threats of host immunity and antibiotic therapy.This is similar to our previous work showing that induction of the S. aureus general stress response regulated by the alternative sigma factor SigB can promote the survival of bacteria exposed to host defenses and various classes of antibiotics (88).Further support for the link between S. aureus-immune interactions and antibiotic tolerance comes from studies showing that oxidative stress conferred by phagocytic cells reduces staphylococcal susceptibility to antibiotics (89,90).
While the immune response may compromise the efficacy of antibiotic therapy under certain circumstances, our study also highlights how antibiotics and the immune response can work synergistically by showing that fosfomycin blocked LL-37 induced cell wall thickening and thereby maintained exposure of bound opsonins, leading to efficient opsonophagocytic killing.In addition, previous work has suggested that fosfomycin also promotes the killing of S. aureus via enhanced production of the neutrophil oxidative burst (91).However, while we exposed S. aureus to fosfomycin in serum, this was removed by washing prior to incubation with neutrophils and thus does not explain the enhanced killing effect observed in our assays.This strongly suggests that there are at least two mechanisms by which fosfomycin and neutrophils synergize against S. aureus and a greater understanding of this may contribute to more effective therapeu tic approaches that reduce the high incidence of relapsing or chronic staphylococcal infections (92).
In summary, we show that S. aureus cells are heavily opsonized upon initial exposure to serum.However, S. aureus responds to serum by accumulating peptidoglycan, which conceals bound opsonins, reducing phagocytosis and killing by neutrophils.

Bacterial strains and growth conditions
Bacterial strains used in this study are shown in Table 1.Strains were grown at 37°C on tryptic soy agar (TSA) or in TSB with shaking (180 r.p.m.) supplemented with erythromy cin (10 µg mL −1 ) or kanamycin (90 µg mL −1 ) when required.

Construction of strains
The JE2 sbi::Tn/spa::kan double mutant was constructed via transduction of the kanamycin resistance marker from Newman spa::kan (93) into the sbi::Tn mutant present in the NARSA transposon mutant library (50) using φ11.

IgG Fc binding assay
The Fc portion of human Immunoglobulin G (1 mg, Abcam) was labeled with biotin (Thermo Scientific EZ-Link Sulfo-NHS-Biotin) before the removal of unbound biotin by dialysis.The labeled Fc portion was then incubated with PBS-washed bacterial cells for 30 min (10 µg protein and 10 9 CFU S. aureus in 1 mL PBS).Unbound immunoglobulin fragment was removed by three rounds of washing with PBS before cells were incuba ted with streptavidin-alkaline phosphatase for 30 min.Cells were then washed with three rounds of PBS before incubation in 200 µL p-nitrophenol phosphate substrate solution for ELISA (Merck) for 10 min.Cells were then pelleted by centrifugation and the supernatant recovered and A 405 determined.

Generation of TSB-grown and serum-incubated bacterial cultures
To generate TSB-grown bacteria, cultures were grown for 16 h in TSB to stationary phase.These were then diluted to 10 7 CFU mL −1 in fresh TSB and incubated for 2 h at 37°C until 10 8 CFU mL −1 was reached.For some experiments, bacteria were used directly from stationary phase cultures.
Where appropriate, the cell walls of serum-incubated cultures were degraded by lysostaphin.To do this, 1 mL aliquots of serum-incubated bacteria were washed in PBS and resuspended in 1 mL PBS supplemented with indicated concentrations of lysosta phin (between 0.04 and 4 µg mL −1 ).Bacteria were incubated statically for 20 min at 37°C before being washed by 3 rounds of centrifugation in PBS.

Purification of neutrophils
Neutrophils were extracted from 45 mL human blood and collected in heparin tubes to prevent coagulation.Blood (15 mL) was carefully layered over 30°C PolymorphPrep (20 mL) and centrifuged for 1 h at 500 × g to separate the different cell types.Neutrophils were collected, washed with Hanks balanced salt solution (HBSS), and adjusted to 5 × 10 6 viable cells mL −1 in HBSS.Based on microscopy and trypan blue staining, we estimate purity at >95% and viability at >98%.

Determination of bacterial killing by neutrophils and phagocytosis by CFU counts
Neutrophils were adjusted to 5 × 10 6 cells mL −1 in HBSS supplemented with 10% human serum, 0.1 mM CaCl 2 , and 0.1 mM MgCl 2 .In the case of lysostaphin-treated bacteria, 10% serum was omitted from the HBSS.TSB-grown/serum-incubated bacteria were generated as described above, washed three times in PBS, and then added to neutrophils at 5 × 10 6 CFU mL −1 .Tubes were incubated with end-over-end mixing at 37°C for 3 h and at each time point (0, 0.5, 1, and 2 h) aliquots were removed, serially diluted 10-fold in PBS with multiple rounds of pipetting to break up bacterial aggregates, and plated to enumerate CFU ml −1 .Previ ous work has shown that this approach gives ~100% recovery of the inoculum when neutrophil-mediated killing is blocked, providing confidence that all viable bacteria are recovered, regardless of, for example, aggregate formation (49).
In addition, the number of phagocytosed/unphagocytosed bacteria was also enumerated at the 0.5 h time point.A 500 µL aliquot of the neutrophil/bacteria mixture was taken and centrifuged at 500 × g for 1 min to pellet the neutrophils, along with any neutrophil-associated bacteria.The supernatant (containing unphagocytosed bacteria) was serially diluted 10-fold in PBS and plated for CFU counts and the pellet was resuspended in 500 µL PBS, serially diluted 10-fold in PBS, and plated for CFU counts.The CFU mL −1 values of the pellet and the supernatant were divided by the CFU mL −1 of the starting inoculum to generate the percentage of CFU mL −1 neutrophil-associated and unphagocytosed, respectively.
To validate the experimental conditions used in this assay, two control experiments were run.First, bacteria that were prepared as described above were subjected to centrifugation and the CFU counts pre-and post-centrifugation were quantified to determine whether bacteria were pulled out of suspension under these conditions.Second, to understand whether bacteria associated with neutrophils were intracellu lar, bacteria were incubated with neutrophils as described above, before subsequent incubation with or without lysostaphin (40 µg mL −1 ) to kill extracellular bacteria.Neutrophils were then washed and CFU counts were determined before washing to remove the lytic enzyme.In a pilot experiment, neutrophils were lysed with Triton X-100 (0.1%) to determine whether this affected the recovery of CFU.However, this detergent was not used in other experiments.

Measurement of phagocytosis by flow cytometry
To measure phagocytosis by flow cytometry neutrophils and bacteria were prepared as described above except that immediately before bacteria were added to the neutrophils, the bacteria were incubated with 10 µg mL −1 fluorescein isothiocyanate for 30 min at room temperature, and then washed three times in PBS.
As above, 5 × 10 6 CFU mL −1 bacteria were added to 5 × 10 6 cells mL −1 in HBSS supplemented with 10% human serum, 0.1 mM CaCl 2 and 0.1 mM MgCl 2 .In the case of lysostaphin-treated bacteria, 10% serum was omitted from the HBSS.After a 30-min incubation at 37°C with end-over-end mixing in the dark, cultures were fixed by the addition of an equal volume of 4% paraformaldehyde (PFA).Samples were then analyzed by flow cytometry using an Amnis CellStream.Bacteria were detected using the 488 nm laser and at least 10,000 bacterial events were recorded.Events with FITC ≥2 × 10 3 were counted as bacteria.Events with an FCS of ≥3,000 were counted as neutrophil-associated and <3000 were counted as free.

Measurement of IgG and complement surface exposure by flow cytometry
TSB-grown and serum-incubated cultures were prepared as described above.A spa/sbi double mutant was used to prevent non-specific antibody binding.Aliquots (500 µL) were incubated for 30 min at room temperature in either PBS or 10% human serum.Samples were washed by three rounds of centrifugation in PBS (13,000 × g for 1 min) and blocked for 1 h in 4% BSA in PBS.Samples were washed once in PBS before IgG was detected with a 1:1,000 dilution of goat anti-human IgG antibody labeled with the BV421 fluorophore (Jackson ImmunoResearch) or C3 was detected with a 1:1,000 dilution of goat anti-human C3 F(ab′) 2 labeled with FITC (Protos Immunoresearch).Antibody incubations were carried out statically for 1 h at room temperature in the dark.Samples were washed with PBS by three rounds of centrifugation (13,000 × g for 1 min) and fixed in 4% PFA.Samples were analyzed by flow cytometry using an Amnis CellStream.IgG was detected using the 405 nm laser and C3 using the 488 nm laser.At least 10,000 bacterial events were recorded and the median value was recorded.

Measurement of IgG and complement by western blotting
Cultures of TSB-grown and serum-incubated bacteria (1 mL at 10 8 CFU mL −1 ) were prepared as described above, washed by three rounds of centrifugation in PBS (13,000 × g for 1 min), and resuspended in 100 µL PBS.A spa/sbi double mutant was used to prevent non-specific antibody binding.Lysostaphin (10 µg mL −1 ) was added and bacteria were incubated statically for 1 h at 37°C.Sample buffer (187.5 mM Tris-HCl [pH 6.8], 6% SDS, 30% glycerol, 0.03% bromophenol blue, and 15% beta-mercaptoethanol; 50 µL) was added and samples were incubated at 95°C for 10 min before 15 µL was loaded onto 10% polyacrylamide gels.Gels were run in Tris-Glycine running buffer (25 mM Tris, 192 mM glycine, 0.1% SDS, pH 8.4) at 100 V for 10 min followed by 200 V for 50 min before being transferred onto PVDF membranes (10 V for 60 min).Membranes were blocked for 1 h at room temperature in 5% milk and 1% BSA in TBST.IgG was detected using 1:10,000 dilution of donkey anti-human IgG conjugated to HRP (Abcam) and C3 was detected by 1:5,000 dilution of rabbit anti-C3 (Abcam) followed by 1:10,000 dilution of goat anti-rabbit IgG conjugated to HRP (Abcam).Blots were developed using SuperSignal West Pico PLUS chemiluminescent substrate (Thermo Scientific) and imaged using the Bio-rad ChemiDoc MP imaging system.

Characterization of IgG bound to cells
Bacteria were grown to exponential phase and incubated in serum for 30 min or 16 h, followed or not by partial cell wall digestion using lysostaphin as described above.Cells (10 8 ) were washed three times in PBS before the bound antibody was eluted using 200 µL antibody elution buffer (Pierce) for 5 min.Cells were then removed by centrifugation and the eluted antibody solution was neutralized with 100 µL protein A binding buffer (Pierce).
To determine the binding ligands of bound antibodies, 10 µg purified cell surface components LTA (Sigma), WTA (37), peptidoglycan (37), membrane proteins (14), or cell wall proteins (14) were immobilized onto the wells of a Nunc Maxisorp ELISA plate by incubation at 4°C for 16 h.The remaining binding sites were blocked with PBS containing 3% bovine serum albumin before the addition of the eluted antibody samples (200 µL).Wells containing eluted antibodies were incubated at ambient temperature for 1 h, washed three times with PBS, and then 200 µL PBS containing anti-human antibodies conjugated to alkaline phosphatase (Abcam, 1:2,000 dilution) was added for 1 h.Wells were again washed three times with PBS and bound alkaline phosphatase quantified using a p-Nitrophenol phosphate substrate solution for ELISA (Merck) and A 405 readings.

Statistical analyses
CFU counts were log 10 transformed and displayed as the geometric mean ± geometric standard deviations (95).Other data are displayed as the mean ± standard deviation or median ± 95% CI.For all experiments, three or more independent replicates were performed as indicated by individual data points.Data were analyzed by one-way ANOVA, two-way ANOVA, or Kruskal Wallis, with appropriate post hoc multiple compari son test as detailed in figure legends using GraphPad Prism (V8.0).

FIG 1
FIG 1 Incubation in human serum promotes tolerance to killing by neutrophils.Bacteria were grown in TSB or incubated in 100% human serum before incubation with human neutrophils in the presence of 10% human serum, and staphylococcal survival was measured over time (A).Survival of TSB-grown and serum-incubated cultures of S. aureus USA300 (B), Col (C), Newman (D), and SH1000 (E) during a 2 h incubation with purified human neutrophils.Data represent the geometric mean ± geometric standard deviation of at least three independent biological replicates.Data were analyzed by a two-way ANOVA with Sidak's post hoc test (*P < 0.05; serum incubated vs TSB grown at indicated time points).

FIG 2
FIG 2 Incubation in human serum reduces opsonophagocytosis and surface-exposed IgG and complement.(A) S. aureus USA300 was grown in TSB or incubated in 100% serum, before incubation with purified human neutrophils for 30 min before the percentage of neutrophil-associated (phagocytosed) and non-neutrophil associated (free) bacteria were determined.To do this, two different approaches were used: flow cytometry (B) and CFU counts (C).(D) To assess opsonization, TSB-grown or serum-incubated S. aureus cells were incubated for 30 min with PBS only or PBS containing 10% serum before the levels of surface-exposed IgG (E) or C3 (F) was determined by flow cytometry.Data in panels B and C show values for each independent experiment, with bars representing the mean ± standard deviation.In E and F, each biological repeat is represented by the median fluorescence value of 10,000 bacterial cells.Data in panels B, C, E, and F were analyzed by two-way ANOVA with Sidak's post hoc test (*P < 0.05; n.s, P ≥ 0.05 for the indicated comparisons).

FIG 3
FIG 3 Cell wall accumulation impairs opsonophagocytosis by concealing bound opsonins.Cultures of S. aureus USA300 were incubated in 100% human serum for 16 h before treatment with indicated concentrations of lysostaphin for 20 min to partially digest the cell wall before levels of surface-exposed (A) IgG and (B) complement component C3 were determined by flow cytometry.In addition, these cells were incubated with purified human neutrophils for 30 min before the percentage of neutrophil-associated and non-neutrophil-associated (free) bacteria were determined by (C) CFU counts and (D) flow cytometry.(E) Bound IgG was eluted from the surface of S. aureus grown to exponential phase in broth and then incubated in serum for 30 min (TSB), or serum-incubated for 16 h (SI) or serum-incubated cells that were subsequently subjected to lysostaphin treatment (SI + Lys) as determined by ELISA.Eluted IgG was then assessed for binding to WTA, peptidoglycan, LTA, membrane-associated proteins, or cell wall-associated proteins.(F) Survival of serum-incubated lysostaphin-treated cells during a 2-h incubation with purified human neutrophils.Lysostaphin was washed away before incubation with neutrophils.Data in panels A-D represent the mean ± standard deviation of the indicated number of independent biological replicates.In panels A and B, each biological repeat is represented by the median fluorescence value of 10,000 bacterial events.Data in panels A and B were analyzed by one-way ANOVA with Dunnett's post hoc test.Data in panels C, D, and F were analyzed by two-way ANOVA with Dunnett's post hoc test (*P ≤ 0.05; lysostaphin-treated vs non-lysostaphin treated).Data in panel E represent the median ± 95% CI of four independent biological replicates and were analyzed by the Kruskal-Wallis test and Dunn's post hoc test to establish statistically significant differences between groups (**P < 0.01; *P < 0.05; ns, P ≥ 0.05 for the indicated comparisons).Data in panel F represent the geometric mean ± geometric standard deviation of four independent experiments.

FIG 4
FIG 4 Antibiotic-mediated inhibition of cell wall accumulation maintains opsonin exposure and efficient opsonophagocytosis.S. aureus was TSB grown, incubated in 100% human serum (16 h) or incubated in 100% serum supplemented with fosfomycin (16 h) and subsequently incubated for 30 min with PBS (Continued on next page)

FIG 4 (
FIG 4 (Continued) or 10% serum before the levels of surface-exposed (A) IgG and (B) complement component C3 were determined by flow cytometry.In addition, these cells were incubated with purified human neutrophils for 30 min before the percentage of neutrophil-associated and non-neutrophil-associated (free) bacteria were determined by (C) flow cytometry and (D) CFU counts.(E) Survival of TSB-grown, serum-incubated, and serum + fosfomycin-incubated cultures during a 2-h incubation with purified human neutrophils.Data in panels A-D represent the geometric mean ± geometric standard deviation of the indicated number of independent biological replicates.Data in panel E are presented as the geometric mean ± geometric standard deviation of three independent experiments.In panels A and B, each biological repeat is represented by the median fluorescence value of 10,000 bacterial events.Data in panels A and B were analyzed by two-way ANOVA with Tukey's post hoc test.Data in panels C-E were analyzed by two-way ANOVA with Dunnett's post hoc test (*P < 0.05; n.s, P ≥ 0.05; comparisons are indicated in panels A-D and serum/serum +fosfomycin-incubated vs TSB grown at each time point in panel E).

TABLE 1
Strains used in this study