De novo purine biosynthesis is required for intracellular growth of Staphylococcus aureus and for the hypervirulence phenotype of a purR mutant.

Staphylococcus aureus is a noted human and animal pathogen. Despite decades of research on this important bacterium, there are still many unanswered questions regarding the pathogenic mechanisms it uses to infect the mammalian host. This can be attributed to it possessing a plethora of virulence factors and complex virulence factor and metabolic regulation. PurR, the purine biosynthesis regulator, was recently also shown to regulate virulence factors in S. aureus, and mutations in purR result in derepression of fibronectin binding proteins (FnBPs) and extracellular toxins, required for a so-called hypervirulent phenotype. Here, we show that hypervirulent strains containing purR mutations can be attenuated with the addition of purine biosynthesis mutations, implicating the necessity for de novo purine biosynthesis in this phenotype and indicating that S. aureus in the mammalian host experiences purine limitation. Using cell culture, we show that while purR mutants are not altered in epithelial cell binding, compared to WT S. aureus, purR mutants have enhanced invasion of these non-professional phagocytes, consistent with the requirement of FnBPs for invasion of these cells. This correlates with purR mutants having both increased transcription of fnb genes, resulting in higher levels of surface-exposed FnBPs to promote invasion. These data provide important contributions to our understanding of how the pathogenesis of S. aureus is affected by sensing of purine levels during infection of the mammalian host.

defect in a rabbit endocarditis model, but the purF mutation did render the bacterium highly susceptible to vancomycin treatment (9).
Recently, it was demonstrated that inactivation of the transcriptional repressor of purine biosynthesis, PurR, results in hypervirulent S. aureus in a mouse bacteremia model (10,11). In purR-deficient S. aureus, transcription of purine biosynthesis genes and known virulence factor genes, including those encoding fibronectin binding proteins (FnBPs), is increased (10,11). This purR mutant-dependent hypervirulent state was found to be mediated by aberrant upregulation of FnBPs, whose expression is normally repressed by PurR. Since several known virulence factors, including exotoxins (11), are controlled by PurR, it is unclear whether FnBP expression alone is sufficient for hypervirulence of purR S. aureus or whether the concurrent substantial increase in pur gene transcription is also required. Moreover, the specific events that occur in vivo that lead to increased virulence are unknown.
As FnBPs are required for the invasion of nonphagocytic cells by S. aureus (12)(13)(14), we sought to determine if purR mutants demonstrate increased invasion, which could in part account for their increased pathogenesis. Furthermore, we hypothesized that the increase in de novo purine biosynthesis may confer a growth advantage during intracellular replication in macrophages, allowing faster escape of purR mutant S. aureus from Kupffer cells and quicker dissemination to other organs. Here, we demonstrate that purR-deficient S. aureus has an increased capacity to invade epithelial cells and concurrently requires de novo purine biosynthesis for intracellular replication in the absence of exogenous purines. Moreover, a systemic murine infection model mirrors these findings and demonstrates that the ability to synthesize purines de novo is essential for the pathogenesis of purR-deficient S. aureus, regardless of increased FnBP expression.

RESULTS
De novo purine biosynthesis is required for S. aureus replication and pathogenesis in vivo. Previously, we showed that FnBPs are essential for the hypervirulence of a purR mutant (10). However, it was not known whether FnBP expression is sufficient for this phenotype or whether the concurrent increase in pur gene expression contributes to rapid lethality in mice. In an attempt to address this at the outset of this study, we assessed the virulence of an S. aureus USA300 purK::⌽N⌺ mutant (15) (the mutation results in a block in the purine biosynthesis pathway [ Fig. S1]), along with a purK::⌽N⌺ ΔpurR double mutant (see below), in relation to those of the wild type (WT) and a ΔpurR mutant. To do this, we infected mice intravenously (i.v.) with each of the four strains using a well-established model of murine bacteremia. While WT-infected animals steadily lost weight over the course of the 4 days of infection, animals infected with the purR mutant required sacrifice at 24 h postinfection (hpi), as previously demonstrated (10) (Fig. 1a), and this correlated with significant increases in bacterial burden, versus those of the WT, in the heart and kidneys at 24 hpi (Fig. 1b). In contrast, animals infected with the purK mutant did not lose weight (Fig. 1a) or show outward signs of disease, even by 96 hpi, and had significantly lower bacterial burdens in the heart and kidneys (Fig. 1c). Most importantly, we found that including the purK::⌽N⌺ mutation in the ΔpurR strain converted what was a hypervirulent purR mutant strain into an attenuated strain (Fig. 1). In fact, not only was removal of de novo purine biosynthesis sufficient to eliminate the hypervirulence of a purR mutant, but also it reduced bacterial virulence, as evidenced by weight loss and bacterial burdens, to levels lower than those of the WT. Together, these data demonstrate that de novo purine biosynthesis is required for the pathogenesis of S. aureus, as well as for the hypervirulence associated with purR inactivation during systemic disease.
Lack of de novo purine biosynthesis is without effect on serum-and FnBPdependent hyperclumping of S. aureus. Given that the purK mutation completely abrogated the hypervirulence of the purR mutation, we next investigated whether this may be due simply to effects on growth in the absence of purines or whether the inability to synthesize purines affected FnBP-dependent bacterial clumping in serum, which we previously correlated with hypervirulence (10). The purK mutant, irrespective of whether it also contained a purR mutation, demonstrated attenuated growth in tryptic soy broth (TSB) (Fig. 2a). Provision of purK in trans partially restored the growth defect of the single purK mutant, and we attribute this to the facts that purK is the second gene in the 11-gene operon and the transposon insertion exerts a polar effect on downstream gene transcription. Consistent with this notion, when the double mutant was complemented with the same purK expression plasmid, we observed full restoration of growth, ostensibly because the purR mutation results in significantly increased transcription of the complete pur operon (10).
To further investigate the requirement of de novo purine biosynthesis for S. aureus growth, we analyzed growth of the WT and four different purine biosynthesis gene mutants, obtained from the Nebraska transposon library (15), in a chemically defined medium lacking purines. While no differences in endpoint growth were evident between the WT and a purR mutant, none of the mutants that are deficient for purine biosynthesis were able to grow under identical conditions (Fig. 2b). In agreement with the idea that disruption of the purine biosynthesis pathway was the only reason for significantly diminished growth, provision of IMP restored growth of each mutant to WT levels, in a dose-dependent manner (Fig. 2c). The only exception to this was the purA mutant, which has a defect in the biosynthetic pathway at a step after IMP ( Fig.   FIG 1 De novo purine biosynthesis is required for S. aureus pathogenesis in vivo. Groups of 6 to 8 8-week-old female BALB/c mice were infected with 1 ϫ 10 7 CFU of S. aureus via tail vein injection. (a) Animal weight was recorded daily and is shown as percentage of weight loss from initial weight. Data are means Ϯ SEMs from 2 experiments with a total of 14 animals. (b) At 24 hpi animals were sacrificed, organs harvested, and CFU per organ determined. Data are means Ϯ SEMs for 8 animals per group. The dotted line represents the limit of accurate detection. (c) At 96 hpi animals were sacrificed, organs harvested, and CFU per organ determined. Data are means Ϯ SEMs from 2 experiments with a total of 14 animals. The dotted line represents the limit of accurate detection. Animals infected with the ΔpurR mutant met early euthanasia criteria at 24 hpi. *, P value Ͻ 0.05; **, P value Ͻ 0.01; ****, P value Ͻ 0.0001, based on one-way analysis of variance (ANOVA) with Bonferroni posttest. S1) and therefore should not be supported by the presence of the metabolite. Overall, these findings demonstrate the importance of the pur operon for S. aureus growth. Moreover, these results demonstrate that the purK mutant, which is complemented by both purK in trans and by the metabolite IMP, serves as a useful mutant to study the role of purine biosynthesis in further detail, especially in the context of a purR mutation. Therefore, this mutant was used throughout the remainder of this study. We next assessed whether the inability to synthesize purines inhibited the serumand FnBP-dependent clumping phenotype of a purR mutant (10). As demonstrated in Fig. 2d, purR S. aureus displayed the characteristic hyperclumping phenotype when cultured in the presence of serum; this is characterized by a drop in culture optical density at 600 nm (OD 600 ) as bacteria settle to the bottom of the culture tube within minutes. Cultures of the WT and the purK mutant in serum resulted in modest clumping that is characteristic of S. aureus with WT FnBP expression. In contrast, the purR purK S. aureus strain displayed archetypal purR-dependent hyperclumping, despite displaying an overall reduction in growth (Fig. 2d). This phenotype was strictly due to inactivation of purR, as provision of purR in trans eliminated hyperclumping, whereas provision of purK in trans did not (Fig. 2d). Taken together, these data show that in the context of a purR mutation, disruption of purine biosynthesis does not abrogate FnBP-dependent hyperclumping, indicating that elevated FnBP expression due to loss of purR is without effect.
An S. aureus purR mutant demonstrates enhanced invasion of nonprofessional phagocytes. As a way to explain the hypervirulence, we hypothesized that purR mutants have an increased capacity to invade nonprofessional phagocytes, since it is well established that S. aureus uses FnBPs as a means to invade these cells (12)(13)(14). We have demonstrated purR mutants overexpress FnBPs, at least transiently, in early stages of growth (10). To define this, we performed an invasion assay using the human lung epithelial cell line A549.
Using this system, we assessed the ability of WT or purR-deficient S. aureus to adhere to and invade A549 cells. Bacteria were grown to two different growth phases: an OD 600 of 0.6, at which FnBP expression is elevated in the purR mutant compared to that in the WT, and an OD 600 of 2.0, at which no transcriptional differences in fnb genes were previously reported (10). We observed no obvious trends in bacterial adhesion outside of a small decrease in the adhesive capacity of strains lacking FnBPs (Fig. 3a). When bacteria were grown to an OD 600 of 0.6, mutants lacking FnBPs showed no invasion, confirming that entry into epithelial cells absolutely depends on FnBP expression. At this growth phase, we observed no significant increase in the ability of the purR mutant to invade epithelial cell (Fig. 3b). Interestingly, we also observed similar levels of invasion between the WT and each of the purK and purR purK mutants, further supporting our contention that the inability to synthesize purines does not affect levels of FnBP expression and, consequently, the invasive capacity of these bacteria. In contrast, when the bacteria were grown to an OD 600 of 2.0, purR mutants demonstrated a significantly higher capacity to invade host cells than that of the WT (Fig. 3b, right). Furthermore, even though purK mutants invaded host cells similarly to the WT, the mutant lacking both purR and purK showed increased invasion over that of the WT. To confirm these data, we visualized infected A549 cells by fluorescence microscopy, using bacteria expressing green fluorescent protein (GFP) from a plasmid. We were able to determine the frequency with which each of the aforementioned strains was found inside epithelial cells ( Fig. 3c and Fig. S2), using antibody-based staining to differentiate intracellular from extracellular bacteria. This analysis corroborated the results from bacterial counts, with WT and purK mutants demonstrating similar levels of invasion ( Fig. 3d). In contrast, bacteria lacking purR, irrespective of the purK mutation, display enhanced invasion of A549 cells. Moreover, this analysis revealed that purR mutants are more invasive at both growth phases tested (Fig. 3d), likely because this technique is more sensitive than counting CFU. There is an apparent discordance between the growth phase-dependent transcriptional upregulation of FnBPs in purR-deficient S. aureus and the growth phase during which we saw increased invasion. This could be due to the effect of purR deficiency on other proteins that may affect FnBP expression. In S. aureus FnBPs have been shown to be posttranslationally targeted and removed from the bacterial cell surface by the action of secreted proteases, including aureolysin (16) and V8 (17). Interestingly, secreted proteases have decreased transcription in a purR mutant of S. aureus (10,11), indicating that differential posttranslational regulation of surface-exposed FnBPs may occur in purR mutants. Together, these data demonstrate that the hyperinvasive phenotype of a purR mutant is due to an early transcriptional increase in FnBP expression, followed by an effect on proteins that decrease the levels of FnBPs on the bacterial cell surface.
Purine biosynthesis mutants demonstrate decreased replication in epithelial cells. The invasive capacity of S. aureus is strictly dependent on FnBPs, but the ability to replicate intracellularly is a multifactorial process. We speculated that the increased pool of purines in a purR mutant may provide an advantage in the restricted intracellular environment. Therefore, we sought to determine the ability of S. aureus to replicate in epithelial cells. Examination of bacterial burden at different times postinfection demonstrated that replication began by approximately 8 hpi; thus, we routinely measured intracellular replication at 10 hpi for the most reliable data (Fig. S3). Replication of the purR mutant showed trends similar to those observed for invasion, with purR-deficient bacteria showing increased levels of replication when the inoculum was grown to an OD 600 of 2.0 (Fig. 4a). To test the dependence of intracellular replication on de novo purine biosynthesis, we also examined the replication of the purK and purR purK mutants. We observed a marked defect in intracellular replication for both the purK and purR purK mutants (Fig. 4a), demonstrating a strict reliance on purine biosynthesis for this intracellular replication.
We sought to examine this in further detail using fluorescence microscopy. Our lab has established a fluorescence-based proliferation assay (18) in which bacteria are surface labeled with eFluor 670 at the outset of infection and lose the dye as they undergo replication. We used this system to examine intracellular replication at a single-cell level, where bacteria that have proliferated will be GFP positive and eFluor 670 negative ( Fig. 4b and Fig. S4). Analysis of images acquired at 10 hpi demonstrated that more replication occurred in purR-infected cells than in WT-infected cells (Fig. 4c). Furthermore, despite the low-level increase in CFU, the purK and purR purK mutants were still capable of some intracellular replication (Fig. 4c). In fact, the numbers of cells containing replicating bacteria were not very different (Fig. 4c), suggesting that repli-cation occurs for the purK mutant bacteria but to levels significantly lower than those seen with the WT. Overall, these data demonstrate that mutations in the pur pathway hinder but do not prevent the ability of S. aureus to replicate intracellularly and that a purR mutant shows improved intracellular replication compared to that of the WT.
The effect of a purR mutation on intracellular replication is due to increased invasion. In addition to increased invasion of purR mutants, we also saw increased intracellular replication in epithelial cells. To determine whether purR bacteria truly proliferate more efficiently within host cells or whether this was simply due to increased host cell invasion, we devised an experiment to circumvent FnBP-dependent bacterial uptake. To do this, we employed COS7 fibroblasts stably expressing human Fc␥IIa receptor (COSIIA cells), which confers on these cells the ability to phagocytose IgGbearing targets (19) (Fig. 5a). As this cellular model requires opsonization (Fig. 5a), the bacteria employed in this study carried deletions of the spa and sbi genes, to eliminate nonspecific IgG binding. Furthermore, to eliminate any confounding effect of FnBP expression, the bacterial strains also carried deletions of the fnbAB genes.
At the outset, we showed that the COSIIA cells were indeed capable of phagocytosing IgG-coated beads, verifying their ability to phagocytose in an IgG-dependent manner (Fig. 5b). In order to opsonize bacteria with IgG, we performed a biotinylation step to coat the bacteria and followed that with opsonization using an antibiotin antibody. We were able to confirm biotinylation of the bacterial cell surface through fluorescent avidin staining, and IgG opsonization was confirmed via staining with a specific fluorescent secondary antibody (data not shown). Having verified opsonization of the bacteria, we then sought to examine the ability of these strains to grow intracellularly. Examination of the replicative capacity of S. aureus in these cells demonstrated that there was no appreciable difference between the replication of WT and purR mutant cells (Fig. 5c). In contrast, little to no replication was observed for the purK or purR purK double mutant, in agreement with the findings for epithelial cells (Fig. 5c). Taken together, these data demonstrate that purR mutants do not grow at an accelerated rate within host cells but rather suggest that the replicative advantage displayed by purR S. aureus within epithelia is due to their enhanced ability to invade more host cells.
De novo purine biosynthesis is required for S. aureus replication in macrophages. The above-described findings demonstrate a defect of pur biosynthesis mutants in the ability to replicate in nonprofessional phagocytes. However, it is well established that S. aureus must replicate in Kupffer cells, the resident liver macrophages, in order to establish infection in murine systemic models (20,21). Therefore, as we have seen a growth defect of pur pathway mutants in vitro and in epithelial cells, we were interested to study their replication in macrophages, which are the bottleneck to systemic infection in mice. Our lab has a well-established gentamicin protection assay in RAW 264.7 macrophages (22), which we utilized to interrogate the ability of purK and purR purK bacteria to grow in macrophages.
At 18 hpi intracellular replication of both WT and purR mutant bacteria could be seen, with no obvious differences between the two strains (Fig. 6a). In contrast, the purK and purR purK mutants showed no replication even at 24 hpi, a time by which WT and purR S. aureus had killed the host cells and were freely replicating in the culture medium (Fig. 6a). These data suggest that mutants unable to synthesize purines are severely restricted within professional phagocytes.
We chose the intermediate point of 18 hpi, at which intracellular replication of the WT and purR mutant could be reliably detected, and investigated the behavior of the purine biosynthesis mutants in more detail. To determine whether this growth impairment is specific to purK or if it is a general response of mutants of the pur pathway, we performed similar infections using purA and purF mutants. As demonstrated in Fig. 6b, intracellular growth of purF and purA mutants was also inhibited, indicating that general defects in purine biosynthesis compromise bacterial growth within macro- phages. This conclusion was further supported by fluorescence imaging, which revealed that the purK and purR purK mutants failed to grow despite being phagocytosed ( Fig. 6c and d and Fig. S5).
Our findings indicate that the growth defect of purine biosynthesis mutants can be restored through the addition of exogenous purines (Fig. 2c). As we detected no replication of pur mutants in macrophages, we chose to assess if intracellular growth could also be rescued by supplying IMP exogenously. Indeed, the addition of 100 M IMP to the culture medium fully restored the growth of the purK, purR purK, and purF mutants but not of the purA mutant (Fig. 7a); the purA mutation affects the pathway after the IMP step (Fig. S1). Analysis of fluorescence images supported these data, with replication of the purK and purR purK mutants being readily detectable in RAW 264.7 macrophages in the presence of IMP (Fig. 7 and Fig.  S6). Importantly, the addition of IMP did not compromise RAW 264.7 cell viability (Fig. S7), and therefore, the observed replication was due to availability of purines, not sudden macrophage death. Indeed, exogenous IMP allowed the purK and purR purK mutants to replicate to levels higher than those of the WT (Fig. 7b and c). Together, these findings demonstrate that de novo purine biosynthesis is required for replication of S. aureus in macrophages.

DISCUSSION
The ability of pathogenic bacteria to synthesize and/or acquire nutrients is integral to their survival and capacity to cause disease. Purines are essential components of life, and the importance of these macromolecules is highlighted by the fact that most free-living organisms are capable of de novo purine biosynthesis. Here, we demonstrate that in S. aureus, de novo purine biosynthesis is essential for in vitro and in vivo virulence, including hypervirulence associated with inactivation of the purine biosynthesis repressor PurR.
We report the inability of a number of pur mutants to grow in a chemically defined medium, which can be alleviated by the addition of exogenous purines (Fig. 2). This is consistent with previous reports for S. aureus (23), but we were interested to note that exogenous purines, while required for pur mutants, did not enhance the growth of WT S. aureus. Dependence on purines has also been demonstrated for Francisella tularensis (4), Brucella abortus (5), and E. coli (6), among other bacteria. S. aureus has been shown to replicate in macrophages (22,24) and osteoclasts (25), but little is known about the nutritional requirements of S. aureus in the intracellular environment. Studies on S. aureus metabolism have identified that mutations affecting in glycolysis (pfkA and pyk) (26) and lactate dehydrogenase (27) result in reduced survival in RAW 264.7 macrophages, but no data are available on nucleotide biosynthesis. We observed decreased intracellular replication of S. aureus pur biosynthesis mutants in epithelial cells (Fig. 4) and virtually no replication in macrophages (Fig. 6). Interestingly, WT S. aureus has been shown to inhibit nucleotide biosynthesis of A549 cells (28), but it is presently unclear if that can influence the levels of purines in the phagosome. This is the first report demonstrating that intracellular growth of S. aureus requires purine biosynthesis, although similar findings have been reported for a number of intracellular pathogens. Indeed, purL, purH, and purE mutants of B. abortus have been shown to be attenuated in macrophages (5), and purD and purF mutants of the same bacterium have reported defects during replication in RAW 264.7 macrophages and HeLa cells (29). All these findings suggest the requirement for purine synthesis to grow within the intracellular environment and indicate that purines are unavailable for bacterial utilization.
Mutations in purR lead to hypervirulence through overexpression of the FnBPs. Nevertheless, these mutants also display upregulation of a number of other genes, including the purine biosynthesis operon (10,11). In order to gain a better appreciation of S. aureus pathogenesis, it is critical to know whether upregulation of FnBPs alone is sufficient for this hypervirulence or whether concurrent upregulation of purine biosynthesis is also needed. We observed severe attenuation of individual purine biosynthesis mutants during murine infections, congruent with previous reports (7,30). However, we also demonstrated similar findings for a purR purK mutant, which behaved like an attenuated pur mutant rather than the hypervirulent purR mutant (Fig. 1). This was further reflected in disease progression, where a purR purK mutant did not have improved bacterial replication later in disease.
To date, only one study has examined S. aureus with mutations in the purine biosynthetic operon and purR. The authors demonstrated that purR purA and purR purH mutants were more virulent than purA and purH single mutants, respectively (11). Nevertheless, in both cases, the purR pur mutants were still attenuated compared to WT bacteria, indicating an overall reduction in bacterial pathogenesis. Moreover, that study only examined the bacterial burden in the kidneys of infected mice at 20 hpi. We saw a similar trend in kidney samples at 24 hpi (Fig. 1b) and 96 hpi (Fig. 1c), although our data did not reach statistical significance. Therefore, our data consistently show that de novo purine biosynthesis is required for the establishment and progression of S. aureus bacteremia. Indeed, this fits well with our findings from macrophage infections, as replication in the resident liver macrophages (Kupffer cells) is required for the systemic spread of S. aureus from the liver (20,21).
S. aureus is not alone in the requirement for de novo purine biosynthesis for full pathogenesis. There are reports of purine biosynthetic mutants of Salmonella enterica serovar Typhimurium, E. coli, and Bacillus anthracis having decreased growth in human serum (31), and S. aureus purA and purB mutants have been shown to have the same defect (23). Furthermore, animal models have demonstrated that purF and purA mutants of F. tularensis are also severely attenuated in mice (4), purD and purF mutants of B. abortus have decreased persistence (29), and a purF mutant of uropathogenic E. coli was attenuated in a mouse bladder colonization model (6). Overall, these findings suggest that purine availability is significantly limited during infection, and purine biosynthesis inhibitors could potentially be used in combination therapy with antibiotics to increase bacterial clearance. This is of particular importance in bacteria such as S. aureus, in which antibiotic resistance is rampant.
The ability of S. aureus to acquire nutrients is paramount to its replication and subsequent success as a pathogen. Here, we further demonstrate the essential role that de novo purine biosynthesis plays in the pathogenesis of S. aureus. Furthermore, we decouple the elevated expression of FnBPs and pur biosynthesis genes in a purR mutant and demonstrate their individual roles in virulence. These findings represent an important step toward the understanding of S. aureus biology during infection and the interplay between nutrient acquisition, virulence factor expression, and disease severity.

MATERIALS AND METHODS
Tissue culture. Human lung epithelial A549 cells were purchased from the ATCC and maintained in Dulbecco's modified Eagle's medium (DMEM) with 10% (vol/vol) fetal bovine serum (FBS) at 37°C and 5% CO 2 and passaged twice a week. RAW 264.7 macrophages were purchased from the ATCC and maintained in RPMI 1640 medium with 5% (vol/vol) FBS at 37°C and 5% CO 2 and passaged every 2 days. COSIIA cells (19), stably transfected with Fc␥IIa receptor, were a gift from Sergio Grinstein and were maintained in DMEM with 10% (vol/vol) FBS and 500 g/ml of G418 at 37°C and 5% CO 2 and passaged twice a week.
Bacterial growth. Bacterial strains used in this study are listed in Table 1. E. coli was grown in Luria-Bertani (LB) broth, and S. aureus was grown in tryptic soy broth (TSB) at 37°C, shaken at 200 rpm, unless otherwise stated. Where appropriate, media were supplemented with erythromycin (3 g/ml), chloramphenicol (12 g/ml), lincomycin (10 g/ml), ampicillin (100 g/ml), or tetracycline (3 g/ml). Where necessary, bacteria were incubated with eFluor 670 dye (0.5 g/ml) in PBS for 5 min, followed by the addition of TSB (18). Cells were then pelleted, washed twice with PBS, and resuspended in PBS to a density of 2 ϫ 10 7 CFU/ml. Fifty microliters of that suspension was added to a well of confluent A549 cells containing 700 l of SF DMEM. Plates were pelleted at 1,000 rpm for 1 min and incubated at 37°C and 5% CO 2 for 15 min. Cells were then washed once with PBS and fresh SF medium was added for a further 15 min at 37°C and 5% CO 2. Cells were then treated with 150 g/ml of gentamicin for 30 min at 37°C and 5% CO 2 , extensively washed to remove the gentamicin, and kept in SF DMEM for For infection of COSIIA cells, cells in 12-well tissue culture plates were maintained in DMEM plus 10% (vol/vol) FBS in the absence of antibiotics. On the day of the infection, cells were washed with PBS and maintained in SF DMEM for at least 1 h prior to infection. S. aureus lacking spa, sbi, and fnbAB were grown in TSB O/N, and 500 l was pelleted and washed 4 times with PBS (pH 8.0). Bacteria were then resuspended in PBS (pH 8.0) with succinimidyl ester biotin and incubated at room temperature (RT) for 45 min. Cells were washed twice with PBS and incubated with mouse anti-biotin antibody (5 g/ml; Jackson ImmunoResearch) at RT for 30 min. Cells were pelleted, washed twice with PBS, and normalized to a density of 2 ϫ 10 7 CFU/ml. A total of 50 l of that suspension was added to a well of confluent COSIIA cells containing 700 l of serum-free DMEM. Plates were pelleted at 1,000 rpm for 1 min and incubated at 37°C and 5% CO 2 for 30 min. Cells were then treated with 150 g/ml of gentamicin for 30 min at 37°C and 5% CO 2 , extensively washed to remove the gentamicin, and kept in SF DMEM for the desired duration of infection, as indicated throughout. At the desired times, medium was removed and cells lysed in PBS plus 0.1% (vol/vol) Triton X-100, scraped from the well, and plated for CFU determination.
IgG bead opsonization. Silica beads (3.14 m; Bangs Laboratories) were opsonized with human IgG (0.8 mg/ml) for 1 h. IgG᎑opsonized beads were added to individual wells containing COSIIA cells and then centrifuged at 277 ϫ g for 1 min to synchronize binding of targets to the cells. After phagocytosis for 30 min, cells were washed vigorously to remove unbound silica beads. Beads remaining extracellular were detected by staining for 3 min with anti᎑human fluorophore᎑conjugated secondary antibodies (0.75 g/ml) prior to fixation with 4% (vol/vol) PFA.
Macrophage infections. For infection of RAW 264.7 macrophages, the protocol as established by Flannagan et al. (22) was used. Where necessary, cells were supplemented with IMP after the removal of gentamicin. For propidium iodide (PI) stains, cells were treated with 1 g/ml of PI in RPMI 1640 for 5 min prior to live-cell imaging (22). Where necessary, bacterial cells were incubated with eFluor 670 dye (0.5 g/ml) in PBS for 5 min, followed by the addition of TSB (18). For fluorescence analysis, cells were stained with tetramethylrhodamine (TMR)-conjugated wheat germ agglutinin (WGA) (1 g/ml) for 5 min, washed with PBS, and fixed with 4% (vol/vol) PFA for 20 min.
Clumping assays. Clumping assays were performed as previously described (10). Briefly, O/N cultures grown in TSB were diluted to an OD 600 equivalent of 0.01 and grown in TSB or TSB with 10% (vol/vol) heat-inactivated horse serum for 3.5 h at 37°C. Tubes were then allowed to sit for 5 min on the bench, and the optical density of the center of the culture was measured.
PCR and construct generation. S. aureus strain USA300 LAC, cured of the 27-kb plasmid that confers antibiotic resistance, was used as the WT strain for mutant generation, unless otherwise stated. Primers used in this study are listed in Table 2. For mobilizing transposon insertion mutations into various genetic backgrounds, phage transduction was performed according to standard techniques. Phage lysate was prepared from the donor strain using phage 80␣, recipient strains were infected, and transductants were selected using appropriate antibiotics (10). Insertions were confirmed by PCR. Markerless deletions were constructed using the pIMAY system, as previously described (32). For complementation, the full-length genes were amplified using primers listed in Table 2, ligated into pALC2073, and transformed into E. coli. All plasmids were passaged through RN4220 prior to transfer to the strain of interest.
Mouse infections. All animal experiments were performed in compliance with guidelines set out by the Canadian Council on Animal Care. All animal protocols (protocol 2017-028) were reviewed and approved by the University of Western Ontario Animal Use Subcommittee, a subcommittee of the University Council on Animal Care. Six-to 8-week-old female BALB/c mice (Charles River Laboratories) were injected via tail vein with 100 l of bacterial culture containing 1 ϫ 10 7 CFU of bacteria. To prepare the bacteria, strains were grown to an OD 600 2 to 2.5 in TSB, washed twice with PBS, and resuspended to the desired numbers in PBS. Infections were allowed to proceed for up to 96 h before animals were