Commensal bacteria augment Staphylococcus aureus infection by inactivation of phagocyte-derived reactive oxygen species

Staphylococcus aureus is a human commensal organism and opportunist pathogen, causing potentially fatal disease. The presence of non-pathogenic microflora or their components, at the point of infection, dramatically increases S. aureus pathogenicity, a process termed augmentation. Augmentation is associated with macrophage interaction but by a hitherto unknown mechanism. Here, we demonstrate a breadth of cross-kingdom microorganisms can augment S. aureus disease and that pathogenesis of Enterococcus faecalis can also be augmented. Co-administration of augmenting material also forms an efficacious vaccine model for S. aureus. In vitro, augmenting material protects S. aureus directly from reactive oxygen species (ROS), which correlates with in vivo studies where augmentation restores full virulence to the ROS-susceptible, attenuated mutant katA ahpC. At the cellular level, augmentation increases bacterial survival within macrophages via amelioration of ROS, leading to proliferation and escape. We have defined the molecular basis for augmentation that represents an important aspect of the initiation of infection.


Introduction
Staphylococcus aureus exists in a polymicrobial environment, primarily as a human commensal organism [1][2][3], but can also cause disease after a breach in host defences, often via localised tissue injury [4]. S. aureus causes a spectrum of disease, from minor skin infections to lifethreatening bacteraemia: infections that are increasingly difficult to treat due to antibiotic resistance [5]. Human innate immune defences, primarily phagocytes, play a crucial role in preventing serious S. aureus disease. However, during infection S. aureus can reside within, and escape from, an intraphagocyte niche [6][7][8]. Similar to other intracellular pathogens [9,10], this can lead to a population bottleneck, where most bacteria are effectively killed by phagocytes, but a small proportion survive, enabling continued infection [11]. This results in the emergence of clonal bacterial populations, which expand from the small numbers surviving the population bottleneck. In the murine sepsis model, liver-resident macrophages known as Kupffer cells are the basis of this population bottleneck and subsequent bacterial clonality [12,13]. Macrophages are crucial for defence against S. aureus, exposing bacteria to an array of bactericidal mechanisms, including ROS, deleterious enzymes and antimicrobial peptides [14]. After phagocytosis, NADPH oxidase (NOX2) produces superoxide (O 2 -) [15], which is converted to hydrogen peroxide (H 2 O 2 ) and hydroxyl radical (�OH). Hypochlorous acid (HOCl) is generated from H 2 O 2 via the enzyme myeloperoxidase (MPO) [16]. Reactive nitrogen species (RNS) are produced by inducible nitric oxide synthase (iNOS), creating nitric oxide (NO�) which can then react with O 2 to form peroxynitrite (ONOO -) [17]. All reactive species cause bacterial damage, but HOCl and H 2 O 2 may be key in vivo, as both are efficacious against biofilms [18,19]. S. aureus uses several approaches to resist ROS/RNS: two superoxide dismutases detoxify O 2 - [20], catalase removes H 2 O 2 [21] , alkyl hydroperoxidase acts to reduce H 2 O 2, ONOOand organic peroxides [22] , and staphylococcal peroxidase inhibitor (SPIN) inhibits MPO therefore blocking HOCl formation [23]. Many S. aureus-ROS studies focus on neutrophils, since chronic granulomatosis disease (CGD) highlights ROS as vital in neutrophil bacterial clearance [24]. Nevertheless, ROS are also important in tissue macrophages [25] for defence against S. aureus [26]. Augmentation is a recently described phenomenon whereby human skin commensals enhance S. aureus pathogenesis [12]. S. aureus bloodstream infection in mice can be augmented by either Gram-positive commensals, their purified peptidoglycan (PGN) or a natural mix of skin flora [12]. In this example of microbial crowdsourcing, only S. aureus benefits, not the non-pathogenic commensals, which succumb. During murine sepsis, augmenting material is co-phagocytosed with S. aureus in Kupffer cells, resulting in increased bacterial survival and the subsequent formation of clonal liver microabscesses [12], with the potential to seed other organs in the body [13]. In Kupffer cells, augmentation is associated with reduced oxidation and, importantly, augmentation is not observed in transgenic mice lacking functional NOX2, defining a pivotal role for ROS in this phenomenon. However, major signalling receptormediated mechanisms (including NOD1, NOD2, TLRs and the inflammasome) did not account for augmentation [12]. To elucidate the molecular mechanism(s) underpinning augmentation, we sought to define the breadth of materials able to enhance S. aureus infection and investigate whether augmentation occurs for other human pathogens. Using in vitro and in vivo studies, we demonstrate that the molecular basis for augmentation is absorption of ROS by augmenting material, shielding S. aureus from macrophage-mediated killing.

A broad range of pathogen-derived materials augment S. aureus infection
Previously, we have shown that S. aureus pathogenesis can be augmented by live Gram-positive skin commensals, purified PGN, or natural skin flora [12]. To determine the breadth of material able to augment S. aureus pathogenesis, we used the murine sepsis model and coinjection of a low S. aureus infectious dose with potential augmenting materials. Increased bacterial numbers in the liver is a key marker of augmented infection, with accompanying weight loss and/or increased kidney bacterial load in severe cases [12]. We first tested Gram-negative bacteria Escherichia coli and Roseomonas mucosa, as part of the human microflora [27,28], with heat-killed (HK) Micrococcus luteus as a positive control. Addition of HK M. luteus, E. coli or R. mucosa significantly increased S. aureus counts in the liver in comparison to S. aureus-only infected mice (Figs 1A-1C, 1E and S1A-S1C). On average S. aureus liver counts are greatly increased from the inocula (of 1x10 6 CFU) to 1.25x10 8 CFU in augmented infections, in comparison to 1x10 6 in control infections. Interestingly, E. coli benefits from the presence of S. aureus, with an increase in E. coli counts in the liver, although these CFU counts are reduced in comparison to the injected E. coli inoculum (Fig 1B). In order to assess whether cross-kingdom materials could augment staphylococcal infection, we tested HK Cryptococcus neoformans and live fungi in the murine sepsis model. Addition of HK C. neoformans significantly increased S. aureus liver numbers (Figs 1D, 1E and S1D). In contrast, Saccharomyces cerevisiae, an occasional human commensal which rarely becomes pathogenic [29,30], did not increase S. aureus liver or kidney numbers, but did enhance mouse weight loss (Figs 1E and S1E-S1G). Together these data demonstrate that S. aureus pathogenicity can be enhanced by a wider range of microorganisms than has previously been shown.

Can infection with other human pathogens be augmented?
Next, we tested whether augmenting material was able to increase the virulence of a range of human pathogens: Enterococcus faecalis, an opportunist pathogen capable of residing within macrophages [31]; Streptococcus pneumoniae, which is able to survive within phagocytes and experiences a population bottleneck which seeds further infection [32]; Pseudomonas aeruginosa, which survives within macrophages [33]; and Streptococcus pyogenes, which can survive and escape from within host cells [34]. During murine sepsis, live M. luteus augmented E. faecalis infection (of 5x10 7 CFU) with a significant increase in liver and lung bacteria compared to E. faecalis alone (Figs 1F, 1G and S1H-S1L). Furthermore, M. luteus PGN augmented a larger E. faecalis inoculum (1x10 8 CFU), although it did not at 5x10 7 CFU (S1M-S1R Fig). In both cases of E. faecalis augmentation, the liver bacterial number never increased higher than the inoculum. Pathogenesis of S. pneumoniae and P. aeruginosa was not increased by the presence of M. luteus PGN in sepsis models (Figs 1G and S2A-S2F). Also M. luteus PGN did not alter mouse weight or S. pyogenes numbers in an intra-muscular leg infection model (S2G and S2H Fig). These data suggest that augmented infections which result in increased pathogen numbers from the inoculum may be specific to S. aureus and a facet of its particular interaction mechanism with the host.

Augmentation requires spatial and temporal co-localisation of S. aureus and augmenting material
Macrophages and ROS are implicated in augmentation [12], suggesting that augmenting material influences S. aureus infection within the phagocyte. To test whether augmentation in the murine sepsis model requires concomitant inoculation of augmentor/pathogen, PGN was injected at a range of timepoints before and after S. aureus infection.
PGN was injected into mice at 24, 6 or 1 hours before, or 0, 6, 24 or 48 hours after S. aureus infection (Fig 2A-2F). S. aureus liver bacterial numbers were significantly increased when PGN was co-injected, but not when PGN was injected at all time-points before (Fig 2A-2C) or after (Fig 2D-2F) S. aureus, suggesting that co-administration is required. No change in weight or kidney bacteria was observed for PGN injected before (Figs 2C and S3A, S3B), while, for PGN injected after S. aureus, no change in kidney bacteria was observed, but significant reductions in mouse weight loss were observed at 6 and 24 hours (Figs 2F and S3C, S3D). Together these data demonstrate that to increase S. aureus pathogenesis, augmenting material must be present concomitantly with S. aureus.
The requirement for concomitant administration of S. aureus and augmenting material suggested that they are likely co-phagocytosed. To examine this, the amount of augmenting Breadth of the augmentation phenomenon, from S. aureus to other pathogens. Co-injection of low dose S. aureus (SA 1x10 6 CFU) with heat-killed M. luteus (HK ML equivalent of 1x10 8 CFU) into mice: liver CFU (n = 5 per group, median value shown, two-tailed Mann-Whitney test), �� p<0.008 B Coinjection of low dose S. aureus (SA 1x10 6 CFU) with E. coli (EC 5x10 6 CFU) into mice: liver CFU, S. aureus, circles; E. coli triangles (n = 8-10 per group, median value shown, two individual two-tailed Mann-Whitney tests), ��� p<0.0003 C Co-injection of low dose S. aureus (SA 1x10 6 CFU) with R. mucosa (RM 2x10 8 CFU) into mice: liver CFU (n = 5 per group, median value shown, two-tailed Mann-Whitney test), � p<0.05 D Co-injection of low dose S. aureus (SA 1x10 6 CFU) with heat-killed C. neoformans (CN 750 μg) into mice: liver CFU (n = 9-10 per group, median value shown, two-tailed Mann-Whitney test), �� p<0.006 E Summary heat-map for microorganisms tested for augmenting ability in S. aureus infection, showing significant changes in liver CFUs and weight change F Co-injection of low dose E. faecalis (EF 5x10 7 CFU) with M. luteus (ML 2x10 8 CFU) into mice: liver CFU (n = 10 per group, median value shown, two-tailed Mann-Whitney test), ���� p<0.0001 G Summary heat-map of alternative pathogens tested for ability to be augmented, with addition of PGN or M. luteus, showing significant changes in liver CFUs and weight change. In all cases liver CFUs were enumerated at 3 days post-infection.   [35]. Augmentation has previously been shown to occur during systemic infection of zebrafish larvae [12]. Larvae were infected with GFP fluorescent S. aureus and/or fluorescently stained PGN. Macrophages phagocytosed injected material in each individually injected group (Fig 2G and 2H) and S. aureus and PGN were co-localised when present within the same macrophage (Figs 2I and S3E). Macrophages were imaged and the area of phagocytosed fluorescent materials was quantified using Fiji. The area taken up by S. aureus within individual macrophages was not altered when PGN was present ( Fig 2J). However, the area of PGN was significantly increased in the presence of S. aureus, in comparison to PGN injected alone ( Fig 2K). Thus, augmentation does not alter the level of S. aureus phagocytosis in vivo, however, it appears that macrophages which engulf S. aureus also phagocytose more augmenting material.

Augmenting material protects S. aureus from ROS in vitro
Since augmentation does not occur in the absence of NOX2, and Kupffer cells have reduced ROS levels in augmented S. aureus infection [12], we hypothesised that inactivation of ROS by augmenting material could be the mechanism by which S. aureus survival is enhanced with an ensuing increase in pathogenesis. We therefore tested whether augmenting material protects S. aureus from specific ROS and RNS in vitro, using H 2 O 2 , sodium hypochlorite (a source of HOCl), peroxynitrite, and methyl viologen (a source of superoxide) (Fig 3A). S. aureus survival in liquid culture in vitro was measured following ROS exposure, with or without M. luteus PGN. Exposure to each ROS led to a significant reduction in S. aureus numbers, while addition of PGN significantly increased S. aureus survival in the presence of H 2 O 2 , HOCl and peroxynitrite, but not methyl viologen (Figs 3B-3D and S4A).
We have previously shown that live bacteria augment S. aureus infection [12], as such, we hypothesised that M. luteus would promote S. aureus survival in the presence of ROS. M. luteus was used at 100 times the concentration of S. aureus. Addition of live M. luteus led to significantly increased survival of S. aureus after H 2 O 2 , HOCl and peroxynitrite treatments, but not methyl viologen (S4B- S4E Fig). It was possible that live M. luteus was mediating augmentation via production of ROS defence enzymes, such as catalase. Addition of HK M. luteus increased S. aureus survival when exposed to HOCl but not H 2 O 2 and peroxynitrite (Figs 3E and S4B-S4D). To determine if the lack of effectiveness of HK M. luteus was due to the availability of ROS active moieties the ratio of augmenting material was raised (ratio of 1:2500), which significantly increased S. aureus survival following exposure to H 2 O 2 or peroxynitrite (Fig 3F and  3G). Thus, both live and HK M. luteus can protect S. aureus from ROS. It appears that H 2 O 2 or peroxynitrite are effectively deactivated by enzymes present in live M. luteus; although these enzymes promote augmentation they are not required.
Thus, augmenting material may act as a buffer to react with, and therefore detoxify, ROS. If this is so, pre-treatment of augmenting material with ROS would diminish its effect. To test this, live M. luteus were pre-treated with ROS prior to inclusion in the in vitro liquid culture assay. HOCl pre-treated M. luteus showed a clear loss of protective ability, with no surviving bacteria, in comparison to live or HK M. luteus (Fig 3E). For both H 2 O 2 and peroxynitrite, the level of S. aureus survival with addition of ROS-treated M. luteus was~10-20%, whereas with HK M. luteus this was~100% (Fig 3F and 3G). This indicates that augmenting material has a finite capacity to react with ROS and, in so doing, loses its ability to protect S. aureus.

Augmenting material restores virulence to ROS susceptible S. aureus in vitro and in vivo
S. aureus mutants lacking oxidative stress resistance mechanisms are susceptible to ROS and attenuated in pathogenesis [20,21]. We used a S. aureus katA ahpC to map which ROS resistance mechanisms are important for S. aureus survival and to test the ability of augmenting material to rescue this strain. The place of the ROS resistance enzymes investigated here in detoxifying the oxidative burst is shown in Fig 4A. S. aureus katA ahpC would be expected to have a reduced ability to detoxify H 2 O 2 , organic peroxides and peroxynitrite, and is more sensitive to peroxides in vitro [21]. S. aureus katA ahpC was protected from H 2 O 2 by live M. luteus, but not by HK or H 2 O 2 -treated cells (Fig 4B), as were individual katA or ahpC mutants (S5A and S5B Fig). When katA ahpC was exposed to HOCl, katA ahpC survival was significantly increased from~0.2% to~100% with the addition of M. luteus and to~68% with HK M. luteus, but not ROS-treated M. luteus which remained at~4.4% survival (S5C Fig). The role of ROS resistance was then tested in vivo using the murine sepsis model. The katA, ahpC and katA ahpC strains were attenuated, with significantly fewer liver bacteria recovered for ahpC and katA ahpC, and kidney bacteria for katA ahpC and katA, with all strains causing significantly reduced weight loss in comparison to wild-type (S5D- S5I Fig).
To test whether ROS-susceptible S. aureus could be augmented in vivo, a low dose (1x10 6 CFU) of katA ahpC was injected with or without HK M. luteus (1x10 8 CFU). In the presence of augmenting material, the katA ahpC strain had an exceptionally large and significant increase in liver bacterial numbers from 0 CFU to~3x10 7 CFU, levels seen in augmented wildtype S. aureus infections, but no significant change in weight loss or kidney bacterial numbers (Figs 4C and S5J, S5K). Thus S. aureus katA ahpC can not only be augmented but also this leads to loss of attenuation in the liver. This further supports the assertion that augmentation occurs in the liver and is associated with the ability of the bacteria to survive ROS, as well as that augmentation also occurs during the initiation of infection. ROS resistance is additionally required for later infection stages, as S. aureus katA ahpC in the presence of augmenting material does not recover to parental bacterial numbers in the kidney.
To examine the role of HOCl in augmented S. aureus murine infection, MPO knock-out mice (MPO -/-) were infected alongside wild-type controls. Interestingly, we observed a significant increase in liver bacterial numbers in MPO -/mice in comparison to wild-type mice ( Fig  4D), demonstrating that MPO -/mice were more susceptible to S. aureus infection. This suggests that MPO activity is a crucial component of the host defence in this model. Augmented infections for both wild-type and MPO -/mice had significantly more liver bacterial numbers and increased weight loss than respective non-augmented infections (Figs 4D and S5L). Therefore, loss of MPO did not inhibit the ability of augmenting material to enhance S. aureus infection, suggesting that HOCl amelioration, at least alone, is not sufficient for augmentation in vivo.

Augmenting material protects S. aureus by inactivating ROS in macrophages
Augmentation occurs at the initiation of infection by circumventing the deleterious effects of ROS in vivo. To determine how these manifest at the cellular level, we used a murine macrophage cell line. Time-lapse imaging of RAW264.7 cells infected with fluorescent S. aureus were used to examine whether bacteria surviving within macrophages may represent the source of the microabcesses that occur as a product of augmentation [12]. In the presence of augmenting material, intracellular S. aureus survival and growth were observed within individual macrophages, which eventually led to host cell death and formation of large extracellular accumulations of bacteria, referred to here as bacterial masses (Fig 5A). Using a high-throughput assay to examine S. aureus mass formation, RAW264.7 cells were infected with S. aureus with or without HK M. luteus at a ratio of 1:10, a lower ratio than was used in the preceding in vitro and in vivo work, to limit cell toxicity. Despite this, the number of masses was significantly increased in the augmented group in comparison to S. aureus infection alone (Fig 5B). We next examined the ratio of augmenting material to S. aureus, using an augmenting material ratio of 10, 5, 2.5, 0.5 and 0.05 to S. aureus, with increased numbers of masses forming in the presence of higher concentrations of augmenting material (Fig 5C). Higher S. aureus levels also led to increased mass formation (Fig 5D). These data demonstrate dose-dependent

PLOS PATHOGENS
Augmentation of S. aureus infection augmentation by HK bacteria of S. aureus survival and proliferation within macrophages. Finally, we used human monocyte-derived macrophages (MDMs) in the time-lapse S. aureus mass formation assay. S. aureus was able to survive, proliferate and escape from MDMs ( Fig  5E). Similarly, S. aureus mass formation from MDMs was significantly increased in the presence of augmenting material (Figs 5F and S6A). Augmenting material therefore increases the capacity of S. aureus to overwhelm human macrophages.
To examine the importance of ROS production, the mass formation assay was evaluated following treatment with NOX2 inhibitors DPI or apocynin, using concentrations which did not inhibit S. aureus growth (S6B and S6C Fig). Addition of DPI or apocynin significantly reduced the level of augmentation compared to the untreated controls (Fig 5G and 5H), but treatment with a specific scavenger of mitochondrial superoxide (mitoTEMPO) did not (Fig 5I). This confirms that ROS production, specifically in phagosomes, is important for augmentation of S. aureus infection within macrophages. Our in vitro assays showed that augmenting material protects S. aureus from ROS. Therefore, we used a specific fluorescent probe, Hydrop, to examine levels of H 2 O 2 within infected RAW264.7 macrophages. The Hydrop assay showed significantly reduced H 2 O 2 levels in RAW264.7 cells infected with S. aureus and HK M. luteus, in comparison to S. aureus alone (Figs 5J and S6D). To further examine how augmentation affects oxidation of the bacteria, RAW264.7 macrophages were infected with GFP S. aureus stained with CellROX, a dye which becomes fluorescent when oxidised by ROS. There were significantly more oxidised S. aureus events in macrophages infected with S. aureus alone than those infected with S. aureus alongside HK M. luteus (Figs 5K and S6E). Together these data demonstrate ROS levels are reduced in the presence of augmenting material, suggesting this material acts to inactivate ROS.

Can augmentation be exploited for vaccine development?
S. aureus vaccine development has been unsuccessful, in part due to animal models not being representative of human disease and high variability in infection outcome [36]. Natural human infection with S. aureus emerges from a polymicrobial environment. We therefore tested if the augmented infection model (where augmenting material represents polymicrobial species) might provide a suitable framework for vaccine development.
The test vaccine consisted of 1 μg ClfA, 50 μg CpG and 1% w/v Alhydrogel an aluminium based adjuvant, components which have been used previously [37][38][39]. Mice were vaccinated on days 0, 14 and 21 before S. aureus infection on day 28, with blood drawn before and after vaccinations (Fig 6A). The vaccine was tested for efficacy, alongside PBS control injections, in three S. aureus infection scenarios; low dose (1x10 6 S. aureus), high dose (1x10 7 S. aureus) and augmented low dose (1x10 6 S. aureus +/-M. luteus PGN), in 2 independent experiments (Figs 6 and S7). Low dose infection caused low numbers (an average of 38 CFU) of bacteria in the liver, with more observed in the high dose infection (an average of 1.43x10 6 CFU), and as expected, the augmentation groups had very high liver bacterial numbers (an average of 4.45x10 7 CFU) ( Fig 6B). Interestingly, vaccination reduced S. aureus liver and kidney bacterial numbers only for the augmentation groups (Figs 6B and S7A, S7B). The second independent experiment also showed that the vaccine was only effective in reducing S. aureus pathogenesis in the augmented scenario (S7C-S7F Fig). These data suggest that using augmentation to examine vaccine efficacy may be a useful strategy, as it mimics natural infection.

Discussion
S. aureus is an insidious pathogen made more concerning due to the spread of antimicrobial resistance and the lack of an available vaccine. Understanding infection dynamics provides a

PLOS PATHOGENS
Augmentation of S. aureus infection route to the identification of disease breakpoints where interventions might be most effective. An effective vaccine should be able to prevent disease establishment, and so understanding the status of the pathogen at this infection initiation stage is crucial. All pathogens exist within a polymicrobial environment from which they emerge to cause disease. S. aureus lives as a human commensal, primarily in the nares where even in this niche it forms only a small proportion of the microbiome [1][2][3]. Thus, all S. aureus infections are initiated from an inoculum that is mostly not the pathogen. With this backdrop, we have identified the augmentation phenomenon, where human-skin commensals or derivatives enhance S. aureus pathogenesis, acting at the level of initial macrophage interaction [12]. The amount of material required to augment S. aureus infection is comparable to the number of bacteria located on human skin or vascular catheters [40,41]. Here we find that augmentation is not specific to S. aureus as it occurs with other opportunist pathogens. Both E. coli and E. faecalis, which survive intracellularly within macrophages [42,43], benefitted from augmenting material. However, in both cases augmentation was evidenced by reduced clearance rather than an increase in pathogen load, suggesting the increase in pathogenesis resulting in increased bacterial burden may be peculiar to S. aureus. We also show S. aureus disease can be augmented by a range of particulate materials from whole bacteria to fungal cell walls, suggesting that augmentation is not mediated by a response to specific components. This is supported by our previous work that demonstrated augmentation not to require any of the major host response pathways such as NOD1 and NOD2 [12]. The hypothesis that augmentation occurs at the initiation of infection was further supported by the requirement for co-inoculation of S. aureus and augmenting material.

Augmentation of S. aureus infection
Augmentation has a profound effect on S. aureus disease, resulting in the ability to reduce the required inoculum by 1000-fold to cause systemic disease in the murine sepsis model [12]. To determine what the molecular mechanism might be, we homed in on those events which occur within the macrophage after phagocytosis, where ROS production is known to be required for augmentation [12]. In vitro, augmenting material protected S. aureus from H 2 O 2 , HOCl and peroxynitrite, suggesting that augmenting material reacts with ROS acting as a buffer, allowing continued S. aureus survival. Augmenting material showed a variable, protective capacity against different ROS. As an example, low dose HK M. luteus (equivalent CFU 5x10 6 ) protected WT S. aureus from killing by HOCl but not the other ROS tested (Figs 3E  and S4C, S4D). Conversely, PGN (25 mg/ml) was able to protect S. aureus from H 2 O 2 , peroxynitrite and HOCl (Fig 3B-3D). Use of a higher dose of HK M. luteus (equivalent CFU 1.25x10 8 ), which was comparable to the number of live cells from which the PGN was derived gave protection to all 3 ROS demonstrating parity (Fig 3F and 3G). Live M. luteus, at a concentration of 10 6 CFU, was able to protect S. aureus from ROS killing by HOCl, H 2 O 2 and peroxynitrite, which is likely due in part to M. luteus ROS resistance enzymes, such as catalase. A variety of biological entities present on augmenting material hold the potential to react with ROS resulting in, for example, oxidation or chlorination [44,45]. Furthermore, pre-treatment of augmenting material with ROS inhibited its protective ability, defining a finite capacity for ROS detoxification. The in vitro data was obtained in an environment very different from that experienced by the bacteria inside phagocytes, let alone in vivo, therefore it was important to make analyses in these more complex milieu. There are a range of ROS, all ultimately originating from superoxide as a product of NADPH oxidase, but which are directly involved in S. aureus killing in macrophages is unknown [26]. S. aureus katA ahpC is susceptible to H 2 O 2 in vitro and is attenuated in vivo. Interestingly, augmentation had a dramatic effect on S. aureus katA ahpC pathogenesis resulting in extremely boosted virulence, to a level compatible to its parent. This embeds the role of ROS resistance at the very earliest stages of disease in order to pass the initial threshold of infection establishment. H 2 O 2 is produced early during oxidative burst [16,17,46] and may therefore constitute a key ROS in controlling S. aureus. We also found that augmenting material protects S. aureus from HOCl and that pre-treatment of the augmentor with this ROS abrogated its protective effect. HOCl is derived from H 2 O 2 by MPO within macrophages but at a proposed lower level than in neutrophils [47], leading to a higher H 2 O 2 concentration in macrophages [48]. Nevertheless, here we demonstrated that MPO is an important host defence enzyme in vivo, where its loss resulted in increased bacterial load in the liver highlighting HOCl as an important ROS in the control of infection. Lack of MPO did not prevent augmentation, in contrast to the loss of NOX2 activity in mice which did [12]. It is therefore likely that augmenting material acts as a sink for ROS in general thereby protecting S. aureus and allowing it to survive this crucial phase in host innate defences.
The effect of augmenting material is to allow S. aureus to survive the ROS assault in the macrophage. Inhibition of ROS in the absence of augmenting material did not greatly enhance S. aureus mass formation in isolated macrophages, possibly indicating that other killing mechanisms, of which there are a variety [26], may compensate in vitro. When S. aureus infection is augmented, absorption of ROS by augmenting material may prevent further maturation of the phagosome and thus activation of downstream bactericidal mechanisms. However, the importance of host ROS in controlling S. aureus infection real-life infections is clearly demonstrated with increased S. aureus pathogenicity in MPO (or NOX2 [12]) deficient mice, as well as the attenuation of S. aureus katA ahpC infection in vivo. A model for the molecular mechanism of augmentation is shown in S8 Fig, where phagocytosis of a threshold number of S. aureus leads to activation of ROS production and bacterial killing. Augmentation results in a bolus of phagocytosed material in addition to the S. aureus that acts to detoxify ROS and so increase

PLOS PATHOGENS
Augmentation of S. aureus infection the chance of bacterial survival, subsequent proliferation and lysis of the phagocyte, releasing a cluster of bacteria able to further multiply to form a microabscess. It is these microabscesses that can then go onto seed other sites in the host leading to a systemic and potentially fatal infection. Augmentation may act, therefore, to increase chances of infection spread by expanding the number of macrophages that are ineffective at controlling the initial infective dose. As the initiation of human infection will come from a polymicrobial environment, augmentation provides a framework to test prophylactic regimen. Indeed, under an augmentation scenario, an experimental S. aureus vaccine reduced bacterial burden. Understanding infection dynamics and the interplay between pathogen, host and other organisms is beginning to give insight into disease progression, and how novel interventions to sway the outcome in the favour for the host may be derived.

Animal husbandry
Mice were housed in designated animal facilities in standard husbandry conditions. BALB/c female mice aged 6-9 weeks (Charles River Laboratories) for all animal work completed at the University of Sheffield. At the University of Calgary female and male MPO -/-Mice and C57 wild-type controls aged 8-11 weeks (Jackson Laboratory) were used. Adult zebrafish were maintained according to standard protocols in UK home office approved facilities, at the Bateson centre Aquaria at the University of Sheffield. Embryos less than 5.2 days post-fertilisation (dpf) of Tg(mpeg1:mCherryCAAX)sh378 [35] were used.

Murine models
The mouse sepsis model was completed by injecting BALB/c mice with a pathogen (S. aureus, E. faecalis, S. pneumoniae or P. aeruginosa), augmenting material (M. luteus, PGN, HK M. luteus, HK C. neoformans, E. coli, R. mucosa or S. cerevisiae), or mixture of a pathogen and an augmentor. These were injected into the tail vein in a volume of 100 μl. Individual experimental figure legends show the CFU and/or amount of augmenting material used. Bacteria were prepared for injection as previously established [49], and serial dilutions of the inoculum were plated to confirm CFUs injected into mice. Mice were monitored daily for health and weight,

PLOS PATHOGENS
Augmentation of S. aureus infection and were euthanised at the experimental end point of 48 or usually 72 hours post-infection (hpi). Liver and kidney CFU were calculated as previously established [12].
For murine intramuscular injections, mice were challenged intramuscularly with S. pyogenes, with or without 500 μg M. luteus PGN in 50 μl PBS and quantitative endpoints compared at 24 hpi. Mice were euthanised and thigh muscle dissected and then homogenised. Bacterial CFU counts were determined by plating of homogenised tissue and blood samples onto the specified agar, with or without dilution in PBS, as appropriate.
In the subcutaneous mouse injections used in the vaccine experiments, mice were scruffed and injected subcutaneously into the scruff with 100 μl of vaccine or PBS control.
For murine blood sampling, mice were warmed to 37 o C to promote tail vein dilation, a small cut was made into the tail vein allowing a small volume of blood to be collected.

Zebrafish infection model
For zebrafish infection and imaging, the established zebrafish infection model was followed [50]. Immediately prior to injection, PGN was stained with Alexa Fluo 405 NHS Ester (Fisher) following established protocols [51]. 2 dpf larvae were anesthetized with tricaine and injected with 1 nl containing 400 CFU of GFP-S. aureus, 5 ng of stained PGN, or both, into the yolk sac circulation valley. Larvae were then recovered, before being mounted in 0.8% low melt agarose (Affymetrix, 32830) in glass-bottom microwell dishes (MatTek, P35 G-1.5-14 C). An Ultra-VIEW VoX spinning disk confocal microscope (Perkin Elmer, Cambridge, UK) was used for imaging larvae at 2 hpi, where 405 nm, 445 nm, 488 nm, 514 nm, 561 nm and 640 nm lasers were available for excitation and a 40x oil objective (UplanSApo 40x oil [NA 1.3] was used. Imaging of macrophages were obtained in the caudal hematopoietic tissue (CHT). Analysis was carried out using Fiji (ImageJ) to measure the area of macrophages, and S. aureus and PGN within macrophages. 47 macrophages from 21 zebrafish larvae which had been co-injected with S. aureus and PGN were assessed for the presence of phagocytosed S. aureus, PGN or both.

Bacterial strains and culture
Microbial strains used in this study are listed in Table 1.

Peptidoglycan and capsule preparation
M. luteus or S. aureus PGN used throughout this study was prepared and purified using established protocols [55]. C. neoformans capsule was isolated after growing cultures for 7 days at 28˚C at 180 rpm. The culture was autoclaved, cells harvested at 6000 g, and supernatant collected. Ice-cold ethanol was added to precipitate the cryptococcal capsule, which was collected after centrifugation again at 6000 g.

Vaccine preparation
Vaccine formulations were made in 1x endotoxin free PBS. Vaccines were used at a final concentration of 1 μg/dose Clumping factor A (ClfA, recombinant, endotoxin purified, mass-
For the production of ClfA, XL1 blue E. coli were used to produce recombinant 6xHis tagged ClfA (residues 40-559). ClfA was purified from cell lysates via nickel affinity chromatography, size exclusion chromatography and confirmed as endotoxin-free. Finally, mass spectrometry was used to verify ClfA (residues 40-559) identity.

In vitro ROS challenge assays
To generate ROS-killed M. luteus, bacteria were treated with ROS (sodium hypochlorite, H 2 O 2 or peroxynitrite) until colonies would no longer form on agar. An aliquot of ROS-killed suspension was then dried and weighed to determine concentration, before freezing to -20˚C in PBS.
S. aureus and M. luteus from overnight broth cultures were adjusted to an optical density at 600 nm (OD 600 ) of 0.05 and 0.1, respectively, in 50 mL TSB. Bacteria were grown at 37˚C (S. aureus) or 30˚C (M. luteus) for 2 h, shaking. Bacteria were harvested and resuspended in PBS to OD 600 1.0 for S. aureus, or 11.3 for M. luteus. S. aureus and M. luteus were diluted to 2x10 5 CFU/mL and 2x10 7 CFU/mL, respectively, in PBS. HK and ROS-killed M. luteus were diluted to equivalent concentrations to live M. luteus, as determined by dry weight (approximately 3.2

Cell culture
Experiments were conducted with RAW264.7 cells (ATCC TIB-71), a leukemic murine macrophage cell line, or primary MDMs derived from human blood. RAW264.7 cells were cultivated in Dulbecco's modified Eagle's medium (DMEM) supplemented with 2 mM L-Glutamine, 100 Units/mL streptomycin, 0.1 mg/mL penicillin, and 10% v/v Foetal bovine serum (FBS) (all media components sourced from Sigma-Aldrich). RAW264.7 cells were passaged into fresh media upon reaching 70-80% confluence. All experiments carried out between passages 5 and 20. MDMs were isolated as described previously [56]. Briefly, peripheral blood mononuclear cells were isolated by Ficoll Plaque (GE Healthcare) density centrifugation, seeded in 24 well plates at 2x10 6 cells/well in RPMI-1640 media (Lonza) supplemented with 2 mM L-Glutamine, 10% v/v newborn calf serum (Gibco) and incubated at 37˚C, 5% CO 2 . Non-adherent cells were removed after 24 h, and adherent cells were fed with fresh RPMI-1640 supplemented with 2 mM L-Glutamine and 10% v/v low endotoxin heat-inactivated foetal bovine serum (Biosera). MDMs were used for experiments at 14 days post-isolation. Media was replaced every 2-3 days for all cells used.

Cell infection
Cell infection assays were carried out as similar to those described previously [12], with modification. RAW264.7 or MDM cells were seeded into 24 well plates (Corning) or white 96 well micro-clear plates (Greiner) and grown to 80% confluence. DMEM supplemented with 2 mM L-glutamine was used in experimental assays for RAW264.7 cells. Cell monolayers were washed with tissue culture PBS (Fisher) three times before infection to remove residual antibiotic.
S. aureus was thawed from a frozen aliquot, as for mouse experiments. Unless otherwise stated, S. aureus was added to cells at a multiplicity of infection (MOI) 5, and HK M. luteus at MOI 50.

S. aureus macrophage survival and mass formation assay
For experiments concerning S. aureus mass formation, GFP expressing S. aureus was added to cells with or without HK M. luteus in 96 well plates and incubated at 37˚C, 5% CO 2 for 2.5 h. Media was removed and fresh infection media supplemented with 20 μg/mL lysostaphin (Biosynexus), alongside 100 μg/ml gentamycin (Fisher) for MDMs, was added and incubated for 0.5-1 h at 37˚C, 5% CO 2 to kill extracellular bacteria. Monolayers were washed with PBS three times, fresh media was added and incubated overnight at 37˚C, 5% CO 2 . Wells were imaged at 24 h post-infection using ImageXpress Micro (Molecular Devices), using a 2x objective lens and FITC filter. Masses were analysed using MetaXpress high-content image acquisition and

PLOS PATHOGENS
Augmentation of S. aureus infection analysis software (Molecular Devices) for average number of masses larger than 20-40 μm per well. Experiments into the effects of chemicals on S. aureus mass formation were carried in the presence of 2 μM diphenyleneiodonium chloride (DPI), 500 μM apocynin or 1 μM Mito-TEMPO, all sourced from Sigma-Aldrich, with media supplemented with solvent used for controls.
In a modification of this experiment, the ratio of S. aureus to HK M. luteus was altered. RAW264.7 cells were infected with GFP S. aureus (MOI 5)

Hydrop dye
For experiments using Hydrop fluorescent dye (Goryo chemical), mCherry S. aureus was stained with Alexa Fluor 555 NHS Ester (Fisher) [51] S. aureus were stored on ice until addition to RAW264.7 cells in a 24 well plate. S. aureus (MOI 5) and/or HK M. luteus (MOI 50) were added to cells in the presence of 1 μM Hydrop dye. 200 nM PMA with Hydrop dye was added to cells for positive control. Wells were imaged at 30 min post-infection on a Nikon Eclipse Ti microscope was used to image cells in a climate controlled set-up (37˚C, Atmosphere: 5% CO 2 / 95% air) with a x20 Lambda Apo NA 0.75 phase contrast objective for brightfield or with GFP or mCherry filters, images were captured with a Andor Neo-5.5-CL3 camera. Analysis was carried out using NIS elements (Nikon) and Fiji (ImageJ). The threshold for GFP images was adjusted within Fiji to exclude background fluorescence. The same threshold was set for all images from an experiment, and this was used to measure the fluorescence levels.

CellROX
For experiments using CellROX deep red reagent (Fisher), S. aureus SJF4620 was first stained with 20 μM CellROX and incubated at 37˚C for 30 min, shaking. Labelled S. aureus (MOI 50) were added to RAW264.7 cells in the presence or absence of HK M. luteus, with unlabelled GFP S. aureus or blank media used for controls. Cells were incubated at 37˚C, 5% CO 2 for 30 min before fixation with 2% w/v paraformaldehyde (Sigma) for 1 h. Following fixation, samples were washed thoroughly with PBS and stained with 300 nM DAPI (Fisher) before further PBS washes. Samples were imaged on ImageXpress Micro (Molecular Devices) using a 20x objective lens, DAPI, FITC and Cy5 filters. Images were analysed using MetaXpress high-content image acquisition and analysis software. A custom module editor was used to identify bacteria in the GFP and Cy5 filters and count the frequency of signal overlap (referred to as 'oxidised objects').

S. aureus infection of macrophages timelapse
For timelapse experiments, GFP S. aureus in the presence or absence of HK M. luteus was added to RAW264.7 or MDM cells in a 24 well plate, and incubated on ice for 60 min, then at 37˚C, 5% CO 2 for 90 min. Antibiotic-containing media was then added to cells: 20 μg/mL lysostaphin for RAW264.7 cells, 20 μg/mL lysostaphin and 100 μg/mL gentamycin for MDMs. This was incubated at 37˚C, 5% CO 2 for 30 min. Wells were washed with PBS three times and replaced with fresh media. MDMs were imaged every 10 min for 18 h, while RAW264.7 cells were imaged every 20 min for 18 h. Imaging was carried out on a Nikon Eclipse Ti microscope was used to image cells in a climate-controlled set-up (37˚C, Atmosphere: 5% CO 2 / 95% air) with a x20 Lambda Apo NA 0.75 phase contrast objective for brightfield or with a GFP filter,

PLOS PATHOGENS
Augmentation of S. aureus infection images were captured with a Andor Neo-5.5-CL3 camera. Analysis was carried out using NIS elements (Nikon) and Fiji (ImageJ).

Statistical analysis
Statistical analysis was carried out in Prism 8.4.3 (GraphPad), with P<0.05 considered significant. Mouse experiments were analysed using two-tailed Mann-Whitney U tests or Kruskal-Wallis one-way analysis of variance (ANOVA) tests with Dunn's post-test, depending on the number of groups compared. Zebrafish experiments were analysed using unpaired t-tests. In vitro ROS challenge assays were analysed using one-way ANOVAs with Tukey post-test. Cell infection assays were analysed by Mann-Whitney U tests or Kruskal-Wallis ANOVA tests with Dunn's post-test. All measurements were taken from distinct samples. A katA, � p<0.05 B ahpC, � p<0.05; ��� p<0.0007 C Following sodium hypochlorite exposure, percentage survival of S. aureus katA ahpC alone (5x10 4 CFU/mL), or with live M. luteus (5x10 6 CFU/mL), heat-killed M. luteus (equivalent of 5x10 6 CFU/mL) or ROS killed M. luteus (equivalent of 5x10 6 CFU/mL), (n = 3, error bars show mean +/-SD, one-way ANOVA test with Tukey's post hoc test), ���� p<0.0001 D-F Injection of high dose (1x10 7 ) of S. aureus control, katA, katA ahpC into mice (n = 10 per group, median value shown, Kruskal-Wallis tests with Dunn's post-test): D weight loss, �� p<0.009; ��� p<0.0003 E liver CFU, � p<0.05 F kidney CFUs, � p<0.05; �� p<0.01 G-I Injection of high dose (1x10 7 ) of S. aureus control and ahpC into mice (n = 10 per group, median value shown, two-tailed Mann-Whitney tests): G weight loss, � p<0.05 H liver CFU, ��� p<0.0002 I kidney CFU J-K Co-injection of low dose (1x10 6 CFU) S. aureus katA ahpC with heat-killed M. luteus (equivalent of 1x10 8 CFU) into mice: liver CFU (n = 5 per group, median value shown, two-tailed Mann-Whitney tests): J weight loss K kidney CFU L Co-injection of low dose S. aureus (SA 2x10 5 CFU) with M. luteus PGN (ML PGN 500 μg) into wild-type (WT) control mice or MPO -/mice: weight loss (n = 9 per group, median value shown, individual two-tailed Mann-Whitney tests), � p<0.05; �� p<0.004. The right-hand side shows an augmented S. aureus infection, highlighting how the presence of augmenting material in the same phagosome as S. aureus results in reduced S. aureus killing due to inactivating ROS produced in the phagosome. Survival S. aureus is then able to proliferate and escape the macrophage. (TIF)