S. aureus deficient for the BrnQ1 branched-chain amino acid transporter do not replicate in human macrophages

S. aureus are internalized by professional phagocytes such as macrophages, where they survive and replicate within acidic phagolysosomes [18]. To test the possible involvement of BCAAs in the S. aureus – macrophage interaction we established a macrophage infection model, involving a short phagocytosis time, followed by a lysostaphin pulse-treatment to eradicate extracellular bacteria (Fig 1a). Here we purposefully avoided prolonged use of lysostaphin or gentamicin as antimicrobials as these can be internalised by macrophages and thus could impact the survival of intracellular bacteria [12,31–33]. Using macrophage colony-stimulating factor (M-CSF)-differentiated human monocyte-derived macrophages, we analysed the ability of individual S. aureus USA300 mutants retrieved from the Nebraska Transposon Library [34] to replicate and egress from infected macrophages. We analysed mutants deficient for the BCAA transporters BrnQ1, BrnQ2, BrnQ3 and BcaP, as well as mutants lacking the ilvE, ilvD and leuA genes, which are involved in BCAA biosynthesis [21]. A mutant in codY, the repressor of the aforementioned genes [26,27], known to exhibit a hypervirulent phenotype [35], was included as control (Fig 1b). This analysis revealed that by 24h post-infection (p.i.) all strains except the brnQ1 mutant, had emerged from infected macrophages and massively replicated extracellularly, resulting in host cell death and the formation of micro-colonies within the infected wells of cell culture plates (Fig 1c). Since the high number of bacteria recovered at 24h p.i. is not solely due to intracellular replication, but rather to extracellular replication of those bacteria that escaped macrophage restriction, we termed this phenomenon unrestricted growth (URG).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. The BCAA transporter BrnQ1 is required for S. aureus replication in human macrophages.

(a) Experimental procedure for macrophage infection: primary human macrophages were infected with S. aureus, at a multiplicity of infection (MOI) of 5 for 30 min. Extracellular S. aureus were removed by a 15 min treatment with 30 μg/ml lysostaphin and two subsequent rinses with DPBS. Infected macrophages were further incubated in antibiotic-free media. “T0” corresponds to 45 min p.i. (b) S. aureus mutants in genes involved in either BCAA-uptake (brnQ1-3 and bcaP) or biosynthesis (ilvE, ilvD, leuA) were chosen for the analysis. Gene identifiers correspond to USA300_FPR3757(old locus ID) [30]. (c) Phase-contrast micrographs of macrophages at 24h p.i., which stayed uninfected, were infected with wild-type S. aureus strains JE2, Cowan I, and RN4220, or mutants in the indicated genes. All mutants were retrieved from the Nebraska Transposon Mutant Library (NTML) generated in the USA300 JE2 background (USA300 LAC-derivative). Unrestricted growth (URG) of the bacteria is observed by formation of bacterial micro-colonies (white arrows). Shown are representative images from two independent experiments. Scale bar = 100 µm. “NE” designates mutant strain identifier in the NTML. (d) Intracellular replication in human macrophages. Data are shown as mean ±SD from independent experiments (JE2 WT and brnQ1: n = 7, brnQ1 + pbrnQ1 and Cowan I: n = 3). (e) Macrophage infection with brnQ1 mutants in the genomic backgrounds of S. aureus strains 6850, SH100 and RN4220. White arrows indicate bacterial microcolonies. Shown are representative images from three independent experiments (Scale bar: 100 µm). (f) Live cell imaging of macrophages infected with mRFP-expressing JE2 WT or JE2 brnQ1 (magenta, white arrows). Host cell death (green fluorescence, green arrows) is detected by CellEvent Caspase-3/7 Green Detection Reagent. Shown are micrographs extracted from time-lapse series representative of 3 independent experiments (n = 3). Scale bar = 50 µm. (see also S1 Movie). (g) Transmission electron micrographs of intracellular JE2 WT or JE2 brnQ1 confined to the phagosomal compartment (white arrows) within the macrophage, at 10h p.i. Shown are representative micrographs, from one experiment. Statistical analysis: (d) two-way ANOVA with Dunnett’s multiple comparisons test (each sample vs JE2 WT, at each time-point), using log10-transformed data; **p < 0.01; ***p < 0.001; ****p < 0.0001. (URG = unrestricted growth). (a) contains icons which were modified (colour and design) from https://bioicons.com/, as follows: petri-dish-lid-yellow icon by Servier https://smart.servier.com/ is licensed under CC-BY 3.0 Unported https://creativecommons.org/licenses/by/3.0/.

https://doi.org/10.1371/journal.ppat.1013291.g001

Remarkably, even S. aureus strains with reduced virulence, Cowan I and RN4220 [36], emerged from infected macrophages and exhibited URG (Fig 1c).

To further investigate the contribution of brnQ1 to S. aureus growth within macrophages, we transduced the brnQ1::Tn transposon insertion (NE945) into a parental S. aureus JE2 strain, hereafter referred to as JE2 brnQ1 and again assessed the ability of this strain to infect macrophages and display URG. These data revealed that, in contrast to the parental JE2 strain, JE2 brnQ1 failed to replicate. Genetic complementation of brnQ1 in trans (JE2 brnQ1 + pbrnQ1) (S1 Fig) rescued the ability of JE2 brnQ1 to egress from macrophages (Fig 1d). While at 6h p.i. bacterial loads are comparable among all strains under scrutiny (S2 Fig), at 12h p.i. we recovered approximately 2-fold more bacteria compared with “T0” (after lysostaphin pulse, i.e. at 45 min p.i.). Both parental JE2 and JE2 brnQ1 + pbrnQ1 as well as S. aureus Cowan I reached URG by 24h p.i. In contrast, the number of recovered colony-forming units (CFU) for JE2 brnQ1 dropped to ~4% of the initial infection inoculum by 24h p.i. (Fig 1d). Furthermore, absence of brnQ1 abolished URG in several S. aureus strains: 6850, SH1000 and RN4220 (Fig 1e). Taken together, these data reveal that a functional BrnQ1 transporter is a general requirement for S. aureus growth inside macrophages, and not simply a JE2-specific phenotype.

To investigate whether the JE2 brnQ1 growth defect within macrophages is specific to the intracellular environment, we next analysed bacterial growth in the medium used for macrophage infection assays. In an undefined medium such as TSB, which contains an abundance of amino acids and peptides, growth of JE2 brnQ1 is indistinguishable from JE2 WT and the complemented strain JE2 brnQ1 + pbrnQ1 (S3a Fig). In contrast, in a chemically defined medium (CDM) containing 1 mM BCAA, JE2 brnQ1 displayed delayed growth when compared to controls (S1 Fig). In RPMI1640, where the concentrations of Val, Leu and Ile are ~ 117 µM (Val) and 131 µM (Leu and Ile, each), JE2 brnQ1 also displays a growth delay in comparison to JE2 WT and JE2 brnQ1 + pbrnQ1 (S3b Fig). This growth delay is still evident in Infection Medium (i.e., RPMI1640 + 10% v/v FBS), indicating that even in the presence of FBS the bacterium’s nutritional requirement for BCAAs is not fully met (S3c Fig). Importantly, in all culture media where JE2 brnQ1 initially displays delayed growth, the mutant eventually achieves the same bacterial density as parental JE2 (S3b and S3c Fig). Taken together, these data show that brnQ1-deficient S. aureus displays delayed growth in media with low BCAA availability (≤ 1 mM), however, in macrophages, growth of the mutant is completely halted.

Replication of S. aureus within macrophages precedes cell death of infected phagocytes [12]. We therefore performed live cell imaging to monitor the intracellular replication of JE2 WT and JE2 brnQ1 in the presence of a fluorogenic caspase 3/7 substrate (Fig 1f and S1 Movie). JE2 WT commenced replicating within macrophages at approximately 8h p.i., which was followed by bacterial egress from the infected macrophage and replication in the extracellular milieu. Activation of effector caspases was evident as green fluorescence at ~ 11h p.i. Shortly thereafter, caspase activation was detectable in uninfected bystander cells. In contrast, JE2 brnQ1 did not replicate, remained confined inside macrophages and failed to trigger caspase activation during the assayed time (i.e., up to 20h p.i.) (S1 Movie).

In non-professional phagocytes, such as epithelial cells, JE2 brnQ1 exhibited similarly impaired intracellular replication, accompanied by reduced cytotoxicity against the host, despite major differences in infection dynamics in this cell type [i.e., bacterial phagosomal escape precedes cytosolic replication [7,37]] (S4 and S5 Figs and S2–S5 Movies and S1 Text).

Notably, culture supernatants from brnQ1 bacteria are as cytotoxic as JE2 WT supernatants, indicating that in principle, BrnQ1 is dispensable for secreted toxin production if bacteria grow in a rich medium (S6a Fig). Additionally, brnQ1 mutants show wild-type haemolysis, on sheep blood agar, which supports the observation that brnQ1 mutants can be virulent under the right growth conditions (S6b and S6c Fig).

Next, we analysed infected macrophages by transmission electron microscopy (TEM). We observed that, at 10h p.i., both JE2 WT and JE2 brnQ1 were localised to the lumen of intact phagosomal compartments (Fig 1g). Overall, these data demonstrate that wild-type S. aureus overcomes intra-macrophage confinement whereas the brnQ1 mutant cannot. Moreover, under our experimental conditions, 10h p.i. is a timepoint at which both wild-type and the brnQ1 mutant, which fails to replicate, are still confined within membrane-bound phagosomes.

Transcriptome analysis of intracellular S. aureus reveals dysregulation of iron metabolism in brnQ1 mutants

To identify the cause of the restricted intracellular growth of brnQ1 mutants, we next performed dual RNA sequencing (RNAseq) of infected macrophages at 10h p.i. We hypothesized that this time point would be crucial for the bacterial adaptation to the phagolysosomal environment, since S. aureus JE2 WT and brnQ1 mutant still reside within a membrane-bound vacuole but only the WT strain can initiate replication shortly thereafter. Consistent with our previous observations [38], we detected a similar host gene expression signature, irrespective of whether macrophages were infected with JE2 brnQ1 or JE2 WT (Figs 2a and S7a), In contrast, we detected profound differences in gene expression between JE2 WT and JE2 brnQ1. Among bacterial transcripts, we found that 80 genes were significantly downregulated in the JE2 brnQ1 mutant inside macrophages as compared to WT. Only 6 transcripts were upregulated in JE2 brnQ1 (Figs 2b and S7b and Table 1). Most of the downregulated genes in intracellular JE2 brnQ1, belonged to operons that encode iron acquisition systems (Figs 2c and S8). These included several isd heme utilization genes (isdA, isdB, isdCDEF and isdG), srtB, staphyloferrin B siderophore biosynthesis genes (sirAB and sbnABCDEFGH) and the staphyloferrin A biosynthesis gene sfaC (Table 1). Several other genes, including chemotaxis-inhibiting protein (chp), thermonuclease (nuc), von Willebrand factor-binding protein (vwb), superoxide dismutase (sodM), Panton-Valentine leukocidin (PVL) subunit lukS, were downregulated. The cysteine protease staphopain A (scpA), as well as genes encoding superantigen-like proteins Ssl2, 4, 7, 11, 12, 13, 14, which are known virulence factors, were similarly downregulated as compared to wild-type.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. List of regulated genes for S. aureus JE2 in primary human macrophages at 10h p.i. (JE2 brnQ1 compared to JE2 WT).

https://doi.org/10.1371/journal.ppat.1013291.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Intracellular S. aureus JE2 brnQ1 mutants do not respond to iron-starvation inside macrophages.

Dual RNAseq of primary human macrophages infected with either JE2 WT or JE2 brnQ1, at 10h p.i.: (a) Human transcripts and (b) Bacterial transcripts. Horizontal lines represent a p-value cutoff < 0.05. Vertical lines represent cutoffs of log₂fold changes of either <-1 and > 1 for human transcripts or <-1.5 and > 1.5 for bacterial transcripts. Transcripts of iron uptake systems, which are prominently downregulated in intracellular JE2 brnQ1 vs JE2 WT, are depicted in green. Differential transcript abundance is expressed as JE2 brnQ1 vs JE2 WT (DESeq2). Data are shown from 3 independent experiments (n = 3). (c) Schematic of the ferric uptake regulator Fur regulon of S. aureus (information retrieved from RegPrecise). Coloration of the boxes underneath the open reading frames indicates log₂fold changes in expression of JE2 brnQ1 vs JE2 WT according to the heatmap. Gene identifiers correspond to USA300_FPR3757 (old locus ID). (d) RT-qPCR analysis of selected genes found to be downregulated or upregulated (e) in intracellular JE2 brnQ1. Data are shown as mean -ΔCt values ±SD from independent experiments (n = 4, different donors). (f) Venn diagram showing the overlap between the set of significantly downregulated genes in JE2 brnQ1 shown in b and the Fur regulon retrieved from RegPrecise. Statistical analysis: (d, e) one-way ANOVA with Dunnett’s multiple comparisons test (vs JE2 WT for each gene); ns = not significant; **p < 0.01; ****p < 0.0001.

https://doi.org/10.1371/journal.ppat.1013291.g002

We further validated by RT-qPCR the marked downregulation of heme- and siderophore-mediated iron acquisition genes in intracellular JE2 brnQ1 as compared to intracellular WT JE2 (Fig 2d). It is interesting to note that a few genes showed modest upregulation in the brnQ1 mutant (Table 1). Again, using RT-qPCR we confirmed that, in the brnQ1 mutant, two loci in the opp3 oligopeptide transporter operon [26] and a gene which encodes a putative transporter for p-aminobenzoyl-glutamate (PABA) (SAUSA300_RS13380/ SAUSA300_2417), a precursor of folic acid [28], were indeed upregulated in intracellular JE2 brnQ1 (Fig 2e). Importantly, all changes in gene expression observed in JE2 brnQ1 relative to JE2 WT were restored to wild-type levels when JE2 brnQ1 was complemented with brnQ1 on a plasmid (Fig 2d and 2e).

Taken together, these data indicate that wild-type S. aureus can adapt to the intracellular milieu of the macrophage as evidenced, for instance, by upregulation of many genes including Fe-regulated loci, while the brnQ1 mutant displays a global defect in gene expression.

The S. aureus-containing phagosome has been shown to be iron depleted [39]. By comparing transcriptomes of intracellular bacteria at 10h p.i. to bacteria grown for 10h in Infection Medium, we reaffirm the observation that S. aureus is indeed iron-starved in this niche, as indicated by upregulation of iron-regulated loci by intracellular JE2 WT (S9 Fig).

In S. aureus, transcription of genes encoding iron uptake proteins is controlled by the ferric uptake regulator Fur [40]. Fur is a transcription factor that uses Fe²⁺ as corepressor. Under conditions of iron starvation, when Fur is not bound to Fe²⁺, Fur is inactive resulting in de-repression of target genes [41]. The observation that many of the most significantly downregulated genes in the brnQ1 mutant function in iron acquisition, prompted us to consult a list of Fur-regulated targets retrieved from the RegPrecise database [42]. This analysis revealed that 18 of the 80 (22.5%) genes that were significantly downregulated in JE2 brnQ1 indeed belonged to the Fur regulon (Fig 2f). To exclude that inadvertent mutations in fur were responsible for our observations, we sequenced the fur locus from JE2 WT and JE2 brnQ1 and found that the gene was intact in both strains.

Intracellular S. aureus brnQ1 mutants are metabolically quiescent at 10h p.i. and fail to respond to environmental cues

Since the brnQ1 mutant carries an intact fur gene, yet fails to regulate iron acquisition loci, we speculated that the brnQ1 mutant may lack the ability to respond appropriately to environmental cues within the phagosome. Ostensibly this defect is exclusively due to BCAA starvation. Therefore, we posited that, even upon inactivation of the fur gene, Fur-regulated genes would still not be de-repressed in the intracellular environment.

To test this, we generated fur and brnQ1/fur mutants in the JE2 background. To validate our strains, we first analysed isdB expression, as a representative iron-regulated gene within the Fur regulon, in Infection Medium (iron replete; contains BCAAs). Under this condition, the inactivation of fur resulted in increased expression of isdB, that was comparable between both fur mutant backgrounds (Fig 3a). Moreover, when we cultivated both JE2 fur and JE2 brnQ1/fur in iron-rich medium (RPMI1640 supplemented with 20 µM FeSO₄), total intracellular iron (Fe²⁺ and Fe³⁺) was ~ 3-fold higher for both fur backgrounds as compared to either JE2 WT or JE2 brnQ1 (S10 Fig). This confirms that inactivation of fur, in both wild-type and mutant backgrounds, results in constitutive de-repression of iron-regulated operons in planktonic bacteria.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Loss of BrnQ1 function renders intracellular S. aureus unresponsive to environmental cues and metabolically quiescent, but a supply of exogenous BCAAs restores responsiveness and promotes intracellular replication.

(a) RT-qPCR analysis of isdB expression for bacteria grown in Infection Medium for 1 hour or (b) in intracellular bacteria within infected primary human macrophages, at 10h p.i. isdB was selected as representative iron-regulated gene to report on Fur regulon status. Data are shown as mean -ΔCt values ±SD from independent experiments: (a) n = 3; (b) n = 4. (c) Schematic representation of the set-up for experiments using aTc induction of intracellular JE2 with a plasmid allowing anhydrous tetracycline (aTc)-inducible expression of the cyan-fluorescent protein Cerulean. After 1h p.i. or 10h p.i., infected cells were treated with either vehicle control (-) or 300 ng/mL aTc (+) for 2h to allow Cerulean expression. (d) Representative confocal images of infected primary human macrophages at 1h p.i. and (f) 10h p.i. upon induction (Magenta: Cerulean; Green: Vancomycin BODIPY-FL/S. aureus; Gray: Phalloidin/Actin; Blue: Hoechst/DNA. Scale bar: overviews: 5 µm; zoomed inset: 1µm). (e) Quantification of intracellular Cerulean-positive bacteria, as percent of total bacteria at 1h p.i. and (g) 10h p.i. in samples exposed to aTc (+) or vehicle control (-) for 2h. Data are shown as mean ±SD from independent experiments (n = 3). (h) Intracellular replication of JE2 brnQ1 or (i) JE2 brnQ1/fur in primary human macrophages, without or with BCAA supplementation. Excess BCAAs (1mM each) were supplemented 24h prior to and during the infection. Data are shown as mean ±SD from independent experiments (n = 3). Graphed URG value represents an imputed value (median of all quantified URG experiments which are shown in Fig 1d). (j) RT-qPCR analysis of isdB expression in intracellular bacteria in infected primary human macrophages, at 10h p.i. Excess BCAAs (1mM each) were supplemented 24h prior to and during the infection. Data are shown as mean -ΔCt values ±SD from independent experiments: n = 4; except JE2 brnQ1/fur + BCAA: n = 3. Statistical analysis: (a, b) one-way ANOVA with Tukey’s multiple comparisons test; (e, g, h, i, j) one-way ANOVA with Sidak’s multiple comparisons test (h and i: for the 10h p.i. time point only); ns = not significant; *p < 0.05; **p < 0.01; ***p < 0.001; (URG = unrestricted growth). (c) contains icons which were modified (colour and design) from https://bioicons.com/, as follows: petri-dish-lid-yellow icon by Servier https://smart.servier.com/ is licensed under CC-BY 3.0 Unported https://creativecommons.org/licenses/by/3.0/. The confocal-scanning-laser-microscope-CSLM icon by DBCLS https://togotv.dbcls.jp/en/pics.html is licensed under CC-BY 4.0 Unported https://creativecommons.org/licenses/by/4.0/.

https://doi.org/10.1371/journal.ppat.1013291.g003

Using these strains, we analysed transcription of isdB inside macrophages and observed that JE2 WT and JE2 fur show comparable isdB expression. This result was expected as under the iron-starved conditions encountered within the phagosome, isdB should be fully de-repressed in JE2 WT similarly to JE2 fur, where iron-regulated genes are constitutively expressed. Intracellular JE2 brnQ1/fur showed isdB expression levels similar to JE2 brnQ1 alone despite lacking the Fur repressor (Fig 3b). Furthermore, isdB expression in both brnQ1 and brnQ1/fur mutants was significantly decreased relative to JE2 WT or JE2 fur, again indicating that brnQ1 deficiency is enough to prevent appropriate induction of gene expression (Fig 3b).

The inability of JE2 brnQ1 to remodel its gene expression in the intracellular milieu hinted that the bacteria may enter a state of intracellular quiescence. To test this, we assessed whether JE2 brnQ1 bacteria were capable of de novo protein synthesis within macrophages. We measured de novo synthesis of the cyan-fluorescent protein Cerulean upon intracellular induction with anhydrotetracycline (aTc) as previously described [43]. In a control experiment, both JE2 WT and JE2 brnQ1 responded to aTc induction in TSB medium, showing that, in principle, both strains can produce Cerulean (S11 Fig). To test whether Cerulean production can be induced by intracellular bacteria, aTc was added to infected macrophages at 1h or 10h p.i. for 2h (Fig 3c). Induction with aTc at 1h p.i. resulted in Cerulean fluorescence for both JE2 WT and JE2 brnQ1 (Fig 3d). At this time, approximately 15% of all internalized bacteria, as detected by BODIPY FL Vancomycin staining, also showed Cerulean fluorescence, while in vehicle-treated cells the bacteria remained Cerulean negative (Figs 3e and S12). In contrast, when Cerulean production was induced at 10h p.i., ~ 10% of the JE2 WT bacteria became Cerulean-positive, whereas only very rarely did JE2 brnQ1 become Cerulean-positive and was comparable to vehicle treated controls (Figs 3f, 3g and S12).

The inactivation of brnQ1 alone, the dominant BCAA transporter, was sufficient to halt intracellular bacterial growth. This indicated that the S. aureus-containing phagosome must be scarce in BCAAs. To test this directly, we supplemented the Infection Medium with 1 mM of each BCAA and observed that the defect in intracellular replication was rescued, both in the case of JE2 brnQ1 (Fig 3h) and JE2 brnQ1/fur (Fig 3i), which now overcame macrophage restriction and grew unrestrictedly. Moreover, addition of excess BCAAs rendered intracellular JE2 brnQ1 and JE2 brnQ1/fur responsive to the iron-deplete state of the phagosome, as shown by significantly higher isdB expression levels 10h p.i. (Fig 3j).

During growth in CDM lacking all three BCAAs, the growth of S. aureus is preceded by an extended lag phase of ~20h. However, omission of Val and Leu only, surprisingly completely halted bacterial growth (S13a Fig). We have therefore hypothesized that provision of these 2 amino acids alone would be sufficient to rescue the growth deficit of S. aureus brnQ1 in macrophages. Indeed, supplementation with 1 mM Val and Leu, but not the single amino acids, restored growth of the brnQ1 mutant in RAW 264.7 macrophages (S13b and S13c Fig).

Taken together, we posit that by 10h p.i. S. aureus brnQ1 mutants have assumed a state of quiescence where the bacteria fail to adapt to the intracellular niche as evidenced by altered gene expression and impaired de novo protein synthesis. Importantly, this quiescence is caused by BCAA starvation and can be overcome if Val and Leu become available, indicating that BCAA acquisition underpins S. aureus growth and survival inside macrophages.

Inactivation of the CodY repressor rescues growth of brnQ1-deficient S. aureus in an IlvD and BcaP dependent manner

The brnQ1 mutant fails to grow in BCAA-depleted phagosomes, although S. aureus carries genes that encode BCAA biosynthesis enzymes (e.g., the ilv operon) [21], as well as a second BCAA transporter (bcaP) [29], which are expressed upon inactivation of the repressor protein CodY [21, 26, 29, 44] and could potentially ensure BCAA supply.

To understand the role that CodY plays in the acquisition of BCAAs during infection, we created a brnQ1/codY double mutant and measured bacterial replication in macrophages. In contrast to the brnQ1 single mutant, the brnQ1/codY double mutant behaved similarly to the WT strain and exhibited URG both in primary human (Fig 4a) and murine RAW 264.7 macrophages (Fig 4b). Restored intracellular growth of brnQ1-deficient S. aureus upon codY inactivation, indicated that the bacteria must now be able to meet their BCAA requirement.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Intracellular growth deficiency of S. aureus brnQ1 mutants can be rescued by inactivation of the CodY repressor or BcaP overexpression.

Intracellular growth of a brnQ1 mutant in the S. aureus USA300 background (∆brnQ1) and a brnQ1/codY double mutant (∆brnQ1/codY::Tn) in (a) primary human macrophages and (b) murine RAW 264.7 macrophages. (c) Intracellular growth in RAW264.7 macrophages of a triple S. aureus USA300 brnQ1/codY/ilvD mutant (∆brnQ1/codY::Tn/ilvD::Tn-Kan^R). (d) Growth in infected RAW 264.7 macrophages of S. aureus brnQ1 carrying either a vector control (pRMC2), plasmid-encoded BCAA transporter BcaP (ΔbrnQ1 + pbcaP) or plasmid encoded brnQ1 (ΔbrnQ1 + pbrnQ1). (e) Growth in infected RAW 264.7 macrophages of S. aureus brnQ1, a double brnQ1/bcaP mutant (ΔbrnQ1/ ΔbcaP), a triple brnQ1, bcaP, codY mutant (ΔbrnQ1/ΔbcaP/codY::Tn) and a quadruple mutant carrying the additional ilvD mutation (ΔbrnQ1/ΔbcaP/codY::Tn/ilvD::Tn-Kan^R). USA300 WT serves as control. All data are shown as mean log10 value ± S.D. for the calculated fold change in CFU/mL at 20h relative to 1.5h p.i., for each bacterial strain. Each data point plotted represents a biological replicate derived from at least three independent experiments (n ≥ 3). Statistical analysis: (a, b, c, e) one-way ANOVA with Tukey’s multiple comparison test; (d) Brown-Forsythe and Welch ANOVA with Dunnet’s T3 multiple comparisons test. ns = not significant; *p < 0.05, ***p < 0.001, ****p < 0.0001.

https://doi.org/10.1371/journal.ppat.1013291.g004

Conceivably, this could be accomplished through endogenous BCAA synthesis. To test this, we created a brnQ1/codY/ilvD triple mutant and assessed its ability to grow in CDM without Val. Growth experiments revealed that the brnQ1/codY mutant, as expected, could grow in the absence of exogenous Val yet the brnQ1/codY/ilvD mutant could not (S14 Fig). This confirms that, as opposed to the triple mutant, the brnQ1/codY mutant must be able to synthesize Val to allow growth in Val-deplete media.

Next, we infected RAW 264.7 macrophages with the mutants. As expected, brnQ1-deficient S. aureus failed to grow out of macrophages whereas WT and brnQ1/codY S. aureus overcame macrophage restriction and displayed URG. Remarkably, the brnQ1/codY/ ilvD mutant also grew out of macrophages indicating that, despite the inability to synthesize BCAAs, the bacteria still access these amino acids inside the macrophage (Fig 4c).

The observation that the intracellular growth of S. aureus brnQ1 can be restored by exogenous addition of Val and Leu during infection, indicated that the bacteria must retain the ability for BCAA uptake. BcaP is an obvious candidate as it has previously been shown that it transports all three BCAAs and that it can partially compensate for loss of brnQ1 [29]. Therefore, we next infected macrophages with a brnQ1 mutant strain that overexpresses bcaP or a respective vector control and we observed that S. aureus brnQ1 overexpressing bcaP could grow unrestrictedly without BCAA supplementation (Fig 4d). These data suggested that increased bcaP expression, due to codY inactivation [29], could explain why the brnQ1/codY/ilvD triple mutant maintained the ability to overcome macrophage restriction. Hence, we utilized a brnQ1/bcaP double mutant, again inactivated either codY alone, or codY and ilvD in this background and infected RAW 264.7 macrophages. Similar to a brnQ1 single mutant, the brnQ1/bcaP double mutant was unable to replicate (Fig 4e). Inactivation of codY in the brnQ1/bcaP background restored growth in macrophages however, an additional ilvD mutation significantly impaired growth (Fig 4e). Taken together, these data indicate that limited phagosomal BCAA levels can in principle be overcome in S. aureus by increased expression of CodY-regulated BCAA synthesis genes and/or BCAA transporters.

S. aureus brnQ1 mutants survive for inordinate duration within human macrophages

Since brnQ1-deficient S. aureus failed to replicate in macrophages during the first 24h of infection, we next assessed long-term intracellular survival in primary human macrophages. Often, studies on long-term intracellular S. aureus infections rely on sustained antibiotic treatment of the infected host cells to limit the effects caused by extracellular bacteria that may egress from host cells during the experiment [13,45–49]. Here we did not maintain antibiotic treatment on infected macrophages, and this allowed us to assess whether URG would occur for S. aureus brnQ1. Bacteria were quantified by CFU counting at 1, 7, 14, 21 and 28 days p.i. unless we observed URG (Fig 5a). Within the first 24h of infection, the number of JE2 brnQ1 bacteria decreased to ~3.5% of the initially phagocytosed bacterial population (i.e., lysostaphin-protected bacteria at “T0”). After this initial decline, the abundance of JE2 brnQ1 remained steady whereas JE2 WT grew unrestrictedly (URG) by day 1 p.i. (Fig 5b). Remarkably, viable JE2 brnQ1 were recovered from infected macrophages at the endpoint of the assay after 28 days p.i. Across 14 independent experiments, we observed that typically JE2 brnQ1 survived for 28 days within the macrophages. However, on occasion, URG was observed at day 2 or 15 p.i. albeit with less frequency (Fig 5b). Consistent with the CFU recovery data, only few bacteria were detected microscopically inside macrophages infected with JE2 brnQ1 from 24h p.i. onwards and ultrastructural analysis revealed brnQ1 mutants to be localised inside intact membrane-enclosed vacuoles (Fig 5c). We hypothesized that the variability in occurrence of URG might arise in a stochastic manner, resulting, for example, from loss of macrophage bactericidal capacity, death of individual host cells with subsequent release of bacteria, or heterogeneity within the bacterial subpopulations. To address this, we set up experiments where macrophages derived from the same donor were seeded in multiple technical replicates (e.g., up to 80 wells/experiment), infected with JE2 brnQ1 and observed daily for 28 days to monitor URG. We observed variability within technical replicates, with URG occurring at various time points during the infection (S15a Fig). Furthermore, brnQ1 mutants in different S. aureus backgrounds similarly survived for up to 28 days without eliciting visible cytotoxic effects against host cells (S15b, S15c and S15d Fig). These data collectively indicate that the duration of infection prior to URG is stochastic and donor-independent.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. S. aureus brnQ1 mutants survive for at least 28 days inside macrophages but are distinct of small colony variants (SCV).

(a) JE2 brnQ1 intracellular replication in primary human macrophages, assayed over a period of 28 days. Data are shown as mean values ±SD from independent experiments (T₀: n = 7; T_0.5: n = 5; T_d1: n = 6; T_d7 and T_d21: n = 2; T_d14 and T_d28: n = 3). (b) Maximum time until unrestricted growth (URG) in macrophages infected with JE2 brnQ1 across 14 biological replicates (n = 14), using 13 different blood donors (arrows: same donor, but used in an independent experiment). JE2 WT serves as control (URG at 24h p.i.). The experiment was terminated 28 days p.i. regardless of URG occurrence. (c) Transmission electron microscopy of JE2 brnQ1-infected macrophages at 24h p.i., 48h p.i., 7 and 28 days p.i. (white arrows). Shown are representative micrographs from one experiment. (d) URG recovery experiment, whereby during a 28-day infection in macrophages, JE2 brnQ1 clones (i.e., micro-colonies formed upon overgrowth from infected macrophages) were recovered upon URG (days 21, 23 and 27 p.i. respectively). Recovered JE2 brnQ1 clones were re-inoculated in Infection Medium and growth was analysed. OD₆₀₀ was measured every 18 min, for 20h. (e) Recovered JE2 brnQ1 were used for infection experiments in primary human macrophages, for 48h. Graphed URG value for JE2 WT control, represents an imputed value (median of all quantified URG experiments which are shown in Fig 1d). (f) Circular genome alignment of two JE2 brnQ1 clones recovered from macrophages upon URG (URG 4 and URG5 shown in d and e using the USA300_FPR3757 complete genome and JE2 WT for comparison (parental strain, laboratory stock). (g) JE2 brnQ1 mutants that did not show URG were recovered from viable macrophages at 28 days p.i. and colony size, haemolysis on sheep-blood agar, and colony pigmentation were assessed. Colony diameter measurements of JE2 brnQ1 which were recovered from macrophages, were compared with bacteria grown to late exponential phase in TSB. Data points correspond to single colonies, recovered from independent experiments (n = 2). Statistical analysis: (g) two-tailed Mann-Whitney test; ns = not significant.

https://doi.org/10.1371/journal.ppat.1013291.g005

S. aureus can grow in the absence of Leu or Val in vitro, albeit after an extended lag phase (S13a Fig). Moreover, growth in the absence of Val was shown to select for heritable loss-of-function mutations in codY or the 5’ untranslated region (5’UTR) of ilvD, to allow de novo BCAA synthesis and subsequent bacterial replication [21]. Spontaneous selection of such suppressor mutants in intracellular JE2 brnQ1 is a possible scenario that could explain the differences in URG observed across experiments and/or technical replicates. Furthermore, our data have revealed that inactivation of CodY is sufficient to rescue the growth of brnQ1 S. aureus in macrophages and in CDM lacking Val (Figs 4 and S14a).

To determine whether brnQ1 mutants that initiated URG had acquired genetic changes, we recovered five JE2 brnQ1 clones upon egress from infected macrophages, sub-cultured the clones in Infection Medium and infected again macrophages. Each recovered brnQ1 isolate again displayed delayed growth in Infection Medium (Fig 5d) and impaired growth upon macrophage infection (Fig 5e) suggesting that the egress events were not the result of heritable mutations. Consistently, whole genome sequencing of two such brnQ1 isolates revealed no mutations in either codY or the 5’UTR of ilvD (Fig 5f).

Intracellular S. aureus brnQ1 mutants are not classical persisters

Within a bacterial subpopulation, transient growth arrest and metabolic quiescence, accompanied by antibiotic tolerance, are hallmarks of persistence [50–53]. Intracellular persisters of S. aureus can arise upon antibiotic pressure [43,54,55]. Several mechanisms triggering intracellular S. aureus persisters were proposed, such as a transient activation of the stringent response (SR) [43]. In S. aureus, SR can be induced by amino-acid starvation (especially Leu/Val deprivation [27]) and results in rapid synthesis of the (p)ppGpp alarmone leading to profound transcriptional reprogramming, especially the repression of proliferation-related genes and activation of the stress-response and starvation-related regulons and a halt of bacterial division [27]. However, upon consulting our transcriptomic data (S1 Dataset), we did not observe changes in the expression of translation factors, ribosomal proteins or tRNAs in JE2 brnQ1 compared to wild-type, at 10h p.i. Similarly, the expression of relA, relP and relQ, encoding the bifunctional Rsh (p)ppGpp synthase/hydrolase (relA) and the RelP and RelQ synthases [56] remained unaltered (S8 Fig). A study found that persistence in intracellular S. aureus in the presence of antibiotics involved multiple protective mechanisms: transcriptional reprogramming of the cell wall stress stimulon, SOS response to preserve bacterial genome integrity or the heat-shock stimulon [43]. In case of intracellular JE2 brnQ1, however, we found none of these pathways to be differentially regulated (S1 Dataset and S8 Fig).

A particular type of persistence described for S. aureus is the so-called SCV phenotype [45,57–59]. SCVs exhibit slow growth and metabolic alterations [60] and are often the result of adaption to an intracellular lifestyle [59]. SCVs form slow-growing colonies on agar plates and lack haemolysis and pigmentation [60] however, JE2 brnQ1 recovered from macrophages after 28 days of infection exhibited normal colony size as compared to the control (i.e. brnQ1 bacteria grown in TSB), displayed haemolysis, and were pigmented (Fig 5g) indicating JE2 brnQ1 recovered from macrophages are not SCVs.

In summary, growth of S. aureus within macrophages requires high affinity transport of BCAAs. BCAA deprivation, for instance mediated through brnQ1 inactivation, results in S. aureus entering a state of quiescence that is distinct from SCV phenotypes. In the BCAA-starved state, S. aureus brnQ1 stay viable within intact macrophage vacuoles for inordinate duration, even in the absence of antimicrobials.

Ethics Statement

Leukocyte samples (i.e., leukoreduction system chambers from anonymous healthy donors) were obtained from the Institute for Clinical Transfusion Medicine and Haemotherapy, of the University of Würzburg, Germany for research purposes and require no specific approval by an ethics committee. Whole human blood was also drawn from healthy adult volunteers upon verbal consent in accordance with protocols approved by the University of Western Ontario’s Research Ethics Board, London, Ontario, Canada.

Bacterial culture conditions

Bacterial strains are listed in S1 Table. All liquid cultures were grown aerobically at 37°C with 200 rpm shaking. E. coli was grown in Luria-Bertani broth (LB) or on LB agar containing 100 µg/mL ampicillin for plasmid maintenance when necessary. S. aureus strains were routinely grown in tryptic soy broth (TSB), or on TSB agar plates (TSA) if not stated otherwise. Media were supplemented with appropriate antibiotics, when necessary: 10 µg/ml chloramphenicol (plasmid encoded antibiotic resistance); and/or 5 µg/ml erythromycin, 5 µg/mL tetracycline or 50 µg/mL kanamycin (for chromosomally encoded antibiotic resistance).

Construction of bacterial strains and plasmids

All used strains (S1 Table), plasmids (S2 Table) and oligonucleotides (S3 Table) can be found in the Supporting material. Strains used for macrophage infection experiments shown in Fig 1c were retrieved from the Nebraska Transposon mutant library [34].

The S. aureus insertional transposon mutant of brnQ1 (NE945) was retrieved from the Nebraska Transposon mutant library, transposon insertion at the correct position was confirmed by PCR, using primers brnQ1_fwd and brnQ1_rev and the mutation was further transduced via phage φ11 or φ80 into the genetic background of wild-type S. aureus strains to yield JE2 brnQ1, 6850 brnQ1, SH1000 brnQ1 and RN4220 brnQ1 respectively.

Plasmid pbrnQ1 was generated to complement the brnQ1 gene in trans under the transcriptional control of its native promoter, using Fast Cloning [80]. Briefly, the open reading frame of brnQ1 gene (locus ID: SAUSA300_0188/ SAUSA300_RS00985) was amplified from S. aureus JE2 genomic DNA together with its native promoter region included in the 400 bp upstream of the start codon, using primer pairs brnQ1_Comp_Inf_fwd and brnQ1_Comp_Inf_rev. The pALC2085 vector [81] was linearized by PCR using primer pairs p2085_Inf_for brnQ1_fwd and p2085_Inf_for brnQ1_rev. Both insert and vector were amplified using an 18-cycle amplification protocol and Phusion DNA-polymerase (Thermo Fisher Scientific, Cat. No. F530L). 1 µL DpnI enzyme (Thermo Fisher Scientific, Cat. No. ER1701) was added to each 50 µL un-purified PCR reaction, vector and insert were mixed in a 1:1 (v/v) ratio and DpnI digest was carried out for 1h at 37°C. 2 µL of the reaction was directly transformed into electrocompetent E. coli DH5α and plated on LB agar plates supplemented with 100 µg/mL ampicillin. The plasmid pbrnQ1 was isolated from E. coli and the sequence was confirmed by Sanger sequencing (SeqLab, Göttingen). Subsequently, plasmids were electroporated into S. aureus RN4220 and further transduced into S. aureus JE2 using phage φ11. A similar procedure was used to generate a control, by amplifying a promoter-less brnQ1 using primer pairs brnQ1_Comp_Inf_fwd and brnQ1_P-less_Comp_Inf_rev and cloning it into the pALC2085 backbone to yield pP_less_brnQ1. Plasmids pbcaP and pbrnQ1 shown in Fig 4d were previously described [24,29] and carry bcaP or brnQ1, respectively, that were cloned into the expression plasmid pRMC2 under the control of their respective native promoters.

Red-fluorescent or cyan-fluorescent expressing S. aureus JE2 were generated by transducing the pSarAP1-mRFP or pCerulean, respectively, from recombinant S. aureus RN4220 strains using phage φ11.

S. aureus JE2 fur and JE2 brnQ1/fur were generated by transducing the tet cassette from S. aureus MJH010 fur::tet [40] using phage φ11.

Strains in USA300 carrying either codY (locus ID: SAUSA300_1148/ SAUSA300_RS06210) or ilvD (locus ID: SAUSA300_2006/ SAUSA300_RS11035) mutations were created by transducing either codY::Tn (NE1555) or ilvD::Tn-Kan^R using phage φ80 into the recipient strain.

The replacement of the erythromycin resistant cassette in strain JE2 NE718 (ilvD) was done using the plasmid pKAN as previously described [82]. In brief, pKAN plasmid was transduced into JE2 NE718 (ilvD) using phage φ80. Replacement of the Ery^R cassette for Kan^R was confirmed by resistance phenotype (i.e., the strain became Ery^R and Kan^R) and by PCR analysis.

Isolation of primary human monocytes from human peripheral blood

Primary human macrophages were derived from peripheral blood mononuclear cells (PBMCs) isolated from whole blood leukoreduction system (LRS) chambers [83] using the SepMate-50 system (StemCell Technologies, Cat. No. 85450) and Ficoll-Paque PLUS density gradient medium (Sigma Aldrich, Cat. No. GE17-1440-03), according to manufacturer’s instructions. Monocytes were then purified from the PBMC fraction using the EasySep Human CD14 Positive Selection Kit II (StemCell Technologies, Cat. No. 17858) according to the manufacturer’s instructions. In some instances PBMCs were isolated from whole human blood using lympholyte-poly (Cedarlane labs) according to the manufacturer’s instructions and as described [39].

Macrophage differentiation and polarization

All macrophage cultivation steps, and infection experiments were done in RPMI1640 (Thermo Fisher Scientific, Cat. No. 72400054) supplemented with FBS (10% v/v; heat inactivated: 56°, 30 min) henceforth termed Infection Medium, unless otherwise indicated. Purified monocytes were seeded in Nunc UpCell cell culture plates (Thermo Fisher Scientific, Cat. No. 174902) at a density of 5-7x 10⁶ cells/plate in Infection Medium supplemented with 50 ng/mL macrophage-colony-stimulating factor (M-CSF; StemCell Technologies, Cat. No. 78057) and incubated at 37°C in a humidified incubator with 5% CO₂ for 7 days. Following the 7-day differentiation protocol, macrophages were detached from the Nunc UpCell by incubation at 20°C, following the manufacturer’s instructions. Cells were collected, centrifuged, re-suspended in Infection Medium, counted using a Neubauer counting chamber and seeded in cell culture well plates. In some instances isolated PBMCs were adhered directly to 12 well tissue culture treated plates and differentiated in the presence of recombinant human M-CSF at 10 ng/mL (PeproTech, Cat. No. 300–25) as previously described [39]. Differentiated macrophages were used for infections from day 7 to day 9.

Macrophage infection assay

For standard infection experiments, 10⁵ cells were seeded into 24 well microtiter plates, 24h prior to infection. Bacterial overnight cultures (TSB) were diluted to an OD₆₀₀ = 0.5 and grown in TSB for 45 min at 37 °C, under continuous shaking at 200 rpm. Bacteria were washed with Dulbecco’s phosphate-buffered saline (DPBS, Thermo Fisher Scientific, Cat. No. 14190169), diluted in Infection Medium and counted using a Thoma counting chamber (Hartenstein, Germany, Cat. No. ZK04). For counting, 10 µL of the above-described suspension were loaded onto the Thoma chamber (small counting square area: 0.0025 mm²; depth 0.1 mm). Bacteria were allowed to sediment for 10 min, then counted in at least 16 small squares and the “chamber factor” of 4 x 10⁶ was used to estimate no. of bacteria/mL. Non-opsonized bacteria were used to infect macrophages at a multiplicity of infection (MOI) of 5. For the synchronization of infection, plates were centrifuged at 200 x g, for 5 min, without brakes and then incubated for 30 min at 37°C in an incubator with humidified atmosphere with 5% CO₂. After 30 min, extracellular bacteria were removed by 15 min incubation in Infection Medium containing 30 µg/ml lysostaphin (AMBICIN L, AMBI), followed by washing twice with sterile DPBS and further incubation in Infection Medium without antibiotics until the end of the experiment. For long-term experiments, viable infected macrophages were cultivated for up to 28 days in the absence of lysostaphin (except for the initial 15 min pulse).

In some instances, primary human and RAW 264.7 macrophage infections were performed using 12 well tissue culture treated dishes as previously described [12]. Here macrophages were infected at an MOI of 5 in serum-free RPMI1640. After synchronization by centrifugation at 277 x g for 2 min, phagocytosis was allowed to occur for 30 min. At this time the medium was replaced with serum-free RPMI1640 containing 100 μg/mL gentamicin for 1 hr, to kill the extracellular bacteria, followed by washing twice with sterile DPBS and further incubated with Infection Medium. For RAW 264.7 macrophage experiments, Infection Medium with 5% (v/v) FBS was used instead.

Intracellular bacterial replication assay (CFU)

To enumerate intracellular bacteria, at the required time points, cells were washed once with DPBS and lysed using alkaline water (pH11) for 5 min at room temperature. Dilution series of the complete cell lysate were plated on TSA and recovered bacterial colony forming units (CFU) were counted. “T0” represented in all graphs corresponds to 45 min p.i. (i.e., 30 min phagocytosis and 15 min lysostaphin treatment). In some instances, after gentamicin treatment and DPBS washing, infected macrophages were lysed using 0.5 mL of 0.1% (v/v) Triton X-100 diluted in DPBS. Lysis at this time equals 1.5h p.i. and the CFU/mL represents the number of phagocytosed or gentamicin protected bacteria. At 20h p.i. supernatants from all wells were collected individually and centrifuged to pellet any bacteria that may have emerged from infected macrophages. While centrifuging, wells containing macrophages were lysed using Triton X-100 as described immediately above, and this lysate was used to resuspend any pelleted bacteria. At each time point bacteria were serially diluted onto TSA and the CFU determined. To calculate the fold change at 20h p.i., the CFU/mL at 20 h was divided by CFU/mL at 1.5h. The data were expressed as the Log10 fold change.

Exogenous BCAA complementation

Macrophages were seeded 24h prior to infection in either Infection Medium or Infection medium supplemented with 1 mM of each branched-chain amino acid (Ile, Leu, Val) and all infection steps were carried out in the respective medium. Samples were further processed for CFU counting or RNA extraction at 45 min (“T0”), 10h and 24h p.i. (CFU) or 10h p.i. (RNA). In some instances, 1mM of either Val, Leu or both were added to RPMI1640 + 5% FBS that was added back to RAW 264.7 macrophages immediately after gentamicin treatment. CFU determination was performed as described above.

Live Cell Imaging

4.5 x 10⁴ macrophages/well were seeded in 8 well chamber µ-slides (ibidi, Cat. No. 80826) 24h prior to infection. Infection with mRFP-expressing bacteria was performed at an MOI of 5 as described above. Activation of effector caspases 3/7 in infected macrophages was visualised 5h p.i. by the addition of 5 µM of the fluorogenic substrate Cell Event Caspase-3/7 Green Detection Reagent (Thermo Fisher Scientific, Cat. No. C10723).Time-lapse imaging was performed in Imaging Medium (Infection Medium without phenol red), in the absence of antibiotics, on a Leica TCS SP5 confocal laser scanning microscope (Leica GmbH, Wetzlar, Germany; software LAS AF version 2.7.3.9723) using a 40x oil immersion objective (Leica HC PL APO, NA = 1.3), equipped with a pre-warmed (37°C) live-cell incubation chamber (Life Imaging Systems) and perfused with a humidified atmosphere containing 5% CO₂. The LAS AF software was used for setting adjustment and image acquisition. All images were acquired at a resolution of 1024x1024 and recorded in 8-bit mode, at 30 min time intervals. Z-stacks were imaged with a step size of 0.4 µm. Image-processing was performed using Fiji [84].

Phase contrast microscopy

Phase contrast microscopic images of live, infected cells were acquired with a LEICA DM IRB microscope connected to a SPOT Pursuit USB camera using a Leica C Plan L 20 x/0.30 PH1 objective and VisiView software (Visitron).

Transmission Electron Microscopy (TEM)

TEM sample preparation and imaging were carried at the Imaging Core Facility of the Biocenter, University of Würzburg, Germany. Macrophages were seeded on cover slips and infection was performed as described. Infected samples were washed once in DPBS, fixed with 2.5% glutaraldehyde (in 50 mM sodium cacodylate [pH 7.2], 50 mM KCl and 2.5 mM MgCl₂) for 2h at 4°C. Samples were subsequently washed twice for 5 min with 50 mM cacodylate buffer (pH 7.2) and were further fixed with 2% buffered OsO₄ (Science Services) and contrasted with 0.5% uranyl acetate in distilled water and finally embedded in Epon (Serva) after ethanol-based dehydration. Ultrathin sections were generated from embedded samples and contrasted with 2% uranyl acetate in ethanol and lead citrate. TEM imaging was performed on a JEOL JEM-2100 instrument.

Preparation of cells for RNA extraction

For total RNA extraction (host and pathogen), 3 x 10⁶ macrophages grown in 6-well cell culture plates (i.e., 1.5 x 10⁶ macrophages were seeded per well and 2 wells/condition were pooled) were infected at MOI of 5, as described. At 10h p.i., cells were washed once with DPBS, and RNAprotect Cell Reagent (Qiagen, Cat. No. 76526) was added immediately (1 part Infection Medium and 5 parts RNAprotect). Samples were stored at 4°C for a maximum of one week before further processing.

Experiments with bacteria grown in Infection Medium were carried out in parallel, using the same bacterial inoculum. Bacteria were similarly grown in 6-well plates at 37°C in a humidified incubator with 5% CO₂, without shaking. After 10h, bacteria were collected by centrifugation and pellets were flash-frozen in liquid N₂ and stored at -80°C.

For experiments depicted in Fig 3a, bacteria were grown overnight in TSB, washed 2x in DPBS and resuspended in Infection Medium at OD₆₀₀ = 1. Bacteria were grown for 1h at 37°C with 200 rpm shaking, collected and flash-frozen in liquid N₂ and stored at -80°C.

RNA extraction

RNA extraction from either S. aureus-infected macrophages or cultures of S. aureus was done using the RNeasy Mini Kit (Qiagen, Cat. No. 74104). Infected cells stored in RNAprotect Cell Reagent were collected in RNase-free tubes, by centrifugation (6,200 x g, 5 min, 4°C). Frozen bacterial pellets were allowed to thaw on ice for 5 min. Next, 600 µL RLT Buffer (kit) supplemented with β-mercaptoethanol (10 µL) was added to the pellets, samples were disrupted in a FastPrep FP120 instrument (6 m/s for 45 sec, at 4°C) using Lysing Matrix B tubes (MP Biomedicals, Cat. No. 116911100), immediately cooled down and centrifuged (10,000 x g, 2 min, 4°C). Following lysis, manufacturer’s instructions for the RNeasy Mini kit were followed. TURBO DNA-free DNase (Thermo Fisher Scientific, Cat. No. AM1907) was used to digest contaminating DNA (1h, 37°), according to manufacturer’s instructions. For routine applications, concentration and purity were determined using a Nanodrop instrument and non-denaturing agarose gel electrophoresis, using ethidium bromide staining [85]. The absence of DNA contamination was verified by qPCR (40 cycles using gyrB-specific primers and RNA samples as template).

RNA sequencing

RNA-seq of S. aureus grown in Infection Medium as well as dual RNAseq of S. aureus-infected macrophages were conducted by the Core Unit SysMed at the University of Würzburg (Germany). RNA quality was checked using a 2100 Bioanalyzer with the RNA 6000 Nano kit (Agilent Technologies, Cat. No. 5067–1511). Ribosomal RNA (rRNA) was depleted from 200 ng of total RNA as input using Lexogen’s RiboCop rRNA Depletion Kit protocol according to manufacturer’s recommendation with the following modifications: for mixed human and bacterial samples, 5 µl of RiboCOP V1.2 probes were mixed with 5 µL of RiboCOP META probes, 4 µl hybridization buffer, and 21 µL of RNA sample. Following rRNA depletion, DNA libraries suitable for sequencing were prepared using CORALL Total RNA-Seq Library Prep protocol according to manufacturer’s recommendation with 14 PCR cycles. After quality control, libraries were sequenced on an Illumina NextSeq platform (1*75 bp single end run). Demultiplexed FASTQ files were generated with bcl2fastq2 v2.20.0.422.

Sequencing reads were trimmed for Illumina adapter sequences using Cutadapt version 2.5 with default parameters. In addition, Cutadapt was given the –nextseq-trim = 20 switch to handle two colour sequencing chemistry and reads that were trimmed to length 0 were discarded. Reads were filtered with FastQScreen [www.bioinformatics.babraham.ac.uk/projects/fastq_screen/]. The splice-aware aligner STAR [86] was used for the mapping of reads originating from human cells to the human genome (GRCh38.p13), whereas the Reademption pipeline [87] was used to map reads originating from S. aureus to the reference (GCF_000013465.1_ASM1346v1). Read counts for each gene were generated and the count output was utilized to identify differentially expressed genes using DESeq2 v1.24.0 [88].

RT-qPCR

1 µg total RNA was converted to cDNA using the RevertAid First Strand cDNA Synthesis kit (Thermo Fisher Scientific, Cat. No. K1622) according to manufacturer’s instructions, using random hexamer primers. RT-qPCR was performed one a StepOne Plus instrument (Applied Biosystems). Each reaction mix consisted of GreenMasterMix (Genaxxon, Cat. No. M3052.0500), 900 nM of each primer and 50 ng (infected samples) or 10 ng (bacteria alone) cDNA. Relative gene expression was normalized to expression of the gyrB reference gene and represented as -ΔCt values [89]. Oligonucleotides used in RT-qPCR experiments are listed in S3 Table.

Genome sequencing

DNA was isolated using the Promega One4All kit on a KingFisher (ThermoFisher Scientific) device. DNA concentration was measured using Qubit (dsDNA HS assay kit, Thermo Fisher Scientific, Germany) and sequencing libraries were generated using the Nextera XT library Prep Kit (Illumina, USA). Data acquisition was performed on a NextSeq550Dx (Illumina, USA) during a 2 × 150 bp paired-end sequencing run using a mid-output cassette. The derived FastQ files were quality-controlled, and the genomes were assembled using SKESA (V2.4.0 on Seqsphere+ (client and server version 9.0.0, Ridom GmbH, Münster). Contigs were aligned and reordered using progressiveMauve [90] and the genomes were annotated using prokka [91].

Cerulean induction in infected macrophages

Macrophages were seeded on coverslips and infected as described with JE2 WT or JE2 brnQ1 carrying pCerulean (anhydrous tetracycline (aTc)-inducible expression of the codon-adapted fluorescent protein Cerulean [81]). At either 1h or 10h p.i. Cerulean was induced by the addition of 300 ng/mL aTc, and samples were incubated for further 2h before fixing with 4% paraformaldehyde (PFA, Morphisto, Cat. No. 11762.01000). Fixed cells were washed 3x in DPBS, permeabilized with 0.2% Triton-X100 (in DPBS) for 5 min, followed by incubation with 50 mM ammonium chloride for 5 min. Staining was done for 20 min, with a solution containing 1 µg/mL BODIPY FL Vancomycin (Thermo Fisher Scientific, Cat. No. V34850), Phalloidin-iFluor647 (Abcam, Cat. No. 176759; diluted according to manufacturer’s protocol) and 4 µg/mL Hoechst 34580 (Thermo Fisher Scientific, Cat. No. H21486), in DPBS. Cover-slips were mounted on microscopy slides with Mowiol and air-dried overnight.

Image acquisition was performed on a Leica TCS SP5 CLSM using a 40x oil immersion objective (NA 1.4). Per sample at least 10 fields of view were recorded with each covering an area of 388 × 388 µm. Images were recorded at a resolution of 1024 x 1024 pixels at 8-bit grey scale depth.

Imaging was done using the following excitation/emission (Ex/Em) settings: Cerulean: Ex/Em = 458/460–494 nm (Argon laser); BODIPY FL Vancomycin: Ex/Em = 488/499–586 nm; (Argon laser); Hoechst 34580: Ex/Em = 405/423–477 nm (diode laser); Phalloidin-iFluor 647 Ex/Em = 633/634–710 nm (HeNe laser).

Image analysis was conducted using Fiji [84], using an in-house macro toolset, which uses BODIPY-FL- channel to identify the macrophage-associated bacteria, whose positions are saved as individual regions-of-interest (ROI). Subsequently, the mean fluorescence intensities in each ROI were determined for both the Cerulean channel and the BODIPY-FL channel. The results were exported from Fiji as text files with comma-separated values (csv). The csv-files fere transformed into text-format using RStudio (Integrated Development for R. RStudio, PBC, Boston, MA) and subsequently imported the data as dot plots into the flow cytometry software Flowing2 (Turku Bioscience Centre, Turku Finland) to determine percent of total bacteria expressing Cerulean.

Fluorescence-based proliferation assay

Fluorescence proliferation assays were performed as previously described [18,92] using GFP expressing S. aureus USA300 and ∆brnQ1. In brief, bacteria were labelled with Cell Proliferation Dye eFluor 670 |(eBioscience, Cat. No. 65-0840-85) and used to infect macrophages that were grown on 18 mm glass coverslips (No. 1) at an MOI of 5 as described above. At 1.5, 10 and 20h p.i. macrophages were stained with 1 µg/mL tetramethylrhodamine-conjugated wheat germ agglutinin (TMR-WGA, Thermo Fischer Scientific, Cat. No. W849). Stained cells were fixed with 4% (v/v) PFA at room temperature for 20 min and mounted on glass microscope slides using ProLong Gold Antifade Mountant (Thermo Fischer Scientific, Cat. No. P36934). Confocal fluorescence microscopy was performed using a Zeiss LSM 880 with Fast AiryScan equipped with a Plan-Apochromat ×63/1.4 oil DIC M27 objective and Definite Focus 2. GFP, TMR-WGA and eFluor 670 were excited with available laser lines at 488 nm, 561 nm, and 633 nm, respectively. Post-image processing (i.e., contrast enhancement and image cropping) was conducted using Fiji [84].

Bacterial colony morphology analysis

S. aureus JE2 brnQ1 were recovered from infected macrophages by lysis with pH11 water. Colony pigmentation was assessed by plating the lysate onto TSA plates. Haemolysis was assessed by plating lysates onto Columbia agar with 5% defibrinated sheep blood. After 30h incubation at 37°C, plates were photographed and colony diameter was measured with Fiji [84]. Fresh bacteria grown in late-exponential phase in TSB served as control.

Bacterial growth curves

Bacterial growth curves were recorded in 48-well plates using a TECAN MPlex plate reader. S. aureus strains were grown overnight in the appropriate culture medium: TSB, RPMI1640 or Infection Medium. Next day, cultures were diluted to an OD₆₀₀ = 0.1 in fresh corresponding medium and used to inoculate 400 µL/well. Sterile medium was used as reference. For experiments depicted in S1 and S13a Figs, bacteria were grown overnight in TSB and washed 4x in DPBS before inoculation at OD₆₀₀ = 0.1, in a chemically defined medium (CDM) [[93] with 25 mM glucose]. BCAAs were excluded from the CDM formulation, as needed. Absorbance at 600 nm (OD₆₀₀) was measured every 18 min for 20h or 50h, under continuous shaking at 200 rpm. For experiments depicted in S1 Fig, bacterial doubling times were estimated from growth curves.

Bacterial growth in valine-free CDM

For experiments depicted in S14 Fig, CDM was prepared as previously described [24] and Val was omitted when needed. Bacteria were grown overnight in TSB, pelleted and washed in 1 mL of sterile saline (0.9% w/v NaCl). Bacteria were diluted into saline at an OD₆₀₀nm of 0.1 and 10 μL of this suspension was used to inoculate 2mL culture volumes to give a starting OD₆₀₀nm of ~ 0.0005. The bacteria were grown in 13 mL polypropylene snap cap tubes at 37°C with constant shaking at 250 rpm. The end point OD₆₀₀ was measured at 24h. As controls, either 1 mM Val or 0.1% (w/v) tryptone was added to the culture medium.

S. aureus infection of non-professional phagocytic cells

HeLa cells (HeLa 229, ATCC CCL-2.1) and 16HBE14o- human bronchial epithelial cells were routinely cultured in Infection Medium, at 37 °C and 5% CO₂. For infection, 10⁵ cells/well were seeded in 12-well microtiter plates 24h prior to infection. Infections were carried out with a multiplicity of infection (MOI) of 5 as described for macrophage infections. In this case, however, bacteria and host cells were co-cultivated for 1h, after which extracellular bacteria were removed by a 30 min treatment with 20 µg/ml lysostaphin (AMBICIN L, AMBI) followed by further incubation in medium containing 2 µg/ml lysostaphin until the end of the experiment.

Phagosomal escape assay

Phagosomal escape was determined as previously described [94]. Briefly, HeLa YFP-CWT or 16HBE14o- YFP-CWT cells were infected with S. aureus at MOI of 5 in 24 well µ-plates (ibidi, Cat. No. 82406). For experiments depicted in S4b, S4c and S5b Figs, mRFP-expressing bacterial strains were used. For experiments shown in S5c Fig, bacteria were stained with SNARF-1 (Thermo Fisher Scientific, Cat. No. S22801) prior to infection. For this, logarithmic growth-phase bacteria grown in TSB were washed once with DPBS and stained with 8 µM SNARF-1 in a final volume of 500 µL DPBS, for 20 min, at RT, protected from light. Afterwards, samples were washed twice in DPBS and de-stained in fresh DPBS for 15 min, at 37°C, protected from light. After de-staining, samples were again washed twice in DPBS, re-suspended in Infection Medium, counted using a Thoma chamber as described, and used for infection. To ensure that the SNARF-1 staining does not interfere with bacterial growth, growth curves using either SNARF-1 stained or unstained bacteria were performed in both TSB medium and RPMI1640. Infection was performed as described. At 3, 6 and 8h post infection, cells were washed, fixed with 4% PFA overnight at 4°C, treated with 50mM NH₄Cl in DPBS for 5 min, permeabilized with 0.1% (v/v) Triton X-100 and nuclei were stained with 4 µg/mL Hoechst 34580. Images were acquired with an Operetta automated fluorescence microscopy system (Perkin Elmer Operetta High-Content Imaging System) and analysed with the built-in Harmony Software. Phagosomal escape was indicated by the co-localization of the YFP-CWT (phagosomal escape) and mRFP/SNARF signals (total intracellular bacteria).

Live cell imaging experiments shown in S4 Fig and S2–S5 Movies were performed as described for macrophages, using mRFP-expressing bacteria.

Cytotoxicity assay (LDH release)

The cytotoxicity of intracellular bacterial strains against epithelial cells (S5e Fig) and the cytotoxicity of bacterial supernatants against HeLa cells (S6a Fig), were analysed by measuring the amount of released lactate dehydrogenase (LDH). At the required time points, spent medium of infected/treated cells (~2x10⁵ cells/sample) was collected, centrifuged for 5 min at 100 x g and 100µl of supernatant (in technical duplicates) was used to determine LDH release using the Roche Cytotoxicity Detection Kit^PLUS (Sigma Aldrich, Cat. No. 4744934001), according to manufacturer’s instructions. Spent medium of uninfected cells served as negative control and lysed cells served as positive control.

S. aureus culture supernatant treatment of epithelial cells

Bacteria were grown for 24h in TSB supplemented with the appropriate antibiotics, if necessary. Cultures were adjusted to an OD₆₀₀ = 4, in a final volume of 1 mL, centrifuged at 20,000 x g and supernatants were filtered through 0.22 µm-membrane filters. 2 x 10⁵ seeded HeLa cells were treated with 10% (v/v) supernatants (in Infection Medium) and incubated for 4h at 37°C and 5% CO₂. Cytotoxicity was measured by quantifying LDH released into the supernatant. 10% sterile TSB medium was used as negative control.

Iron quantification in bacterial pellets

A colorimetric Iron Assay kit (Sigma Aldrich Cat. No. MAK025) was used to measure bacteria-associated iron. Overnight S. aureus JE2 cultures in RPMI1640 were transferred to RPMI1640 supplemented with 20 µM FeSO₄ (FeSO₄ heptahydrate dissolved in technical grade H₂SO₄), at OD₆₀₀ = 0.1 and grown for 24h at 37°C, under shaking at 200 rpm. OD₆₀₀ = 0.5 equivalents of bacterial cultures were pelleted by centrifugation and lysed for 30 min at 37°C in 110 µL Lysis Buffer (200 µg/mL lysostaphin, 1.2% Triton-X100, in water). Lysed samples were mixed with 110 µL Assay Buffer (kit) and incubated for 10 min at 80°C. Samples were subsequently centrifuged (13,000 x g, 10 min, 4°C) and 100 µL supernatant was used to measure total iron (Fe²⁺ and Fe³⁺), according to manufacturer’s instructions.

Cerulean induction in bacterial cultures

Induction of the Cerulean fluorescent protein in S. aureus JE2 during growth in TSB medium was measured by monitoring fluorescence and OD₆₀₀, in a TECAN MPlex plate reader, in black 96 well plates (Thermo Fisher Scientific, Cat. No. 165305). Overnight bacterial cultures were diluted to OD₆₀₀ = 0.1 in fresh TSB, supplemented with 300, 150 ng/mL aTc or vehicle control (ethanol). Absorbance at 600 nm and fluorescence (Ex/Em_{433/504 nm}) were measured every 18 min for 16h. Results are expressed as arbitrary fluorescence units (Cerulean/OD₆₀₀ ratio).

Random unrestricted growth (URG)

Bacterial unrestricted growth (URG) observed in infected macrophages, was analysed as follows: Primary human macrophages were seeded in 96-well plates at a density of 1.5 x 10⁴ cells/well and infected with S. aureus WT or its isogenic brnQ1 mutant as described. Macrophages originating from a single donor were employed for each experiment to exclude donor effects on experimental outcome. Plates were observed daily and the number of wells, as well as the time point when URG occurred were recorded and plotted in a Kaplan-Meier survival plot, where each well represented a test subject. Wild-type S. aureus-infected macrophages and uninfected macrophages were used as positive (URG at 24h) and negative controls (no URG), respectively.

Assessment of secreted protein profiles of different S. aureus strains

Single bacterial colonies were grown in TSB overnight. Cultures were diluted to 0.1, drop plated on SCMA plates and incubated at 37°C for 18-20h. Precipitation zones were measured with a standard metric ruler.

Statistical analysis

Data were analysed using GraphPad Prism Software (GraphPad Software, Version 10.2.3). For statistical analysis, at least three independent experiments were performed (sample size is indicated for each graph). All data are presented as means with standard deviation (±SD). P-values ≤0.05 were considered significant. When two groups were analysed, a diagnostic t-test was run first. If residuals were normally distributed (Shapiro-Wilk test), samples were analysed with an unpaired, two-tailed t-test. If residuals were not normally distributed, the non-parametric Mann-Whitney test (unpaired, two-tailed) was used instead.

When more than two groups were compared, a diagnostic one-way ANOVA was carried out first to check for normality of residuals (Shapiro-Wilk test), followed by Brown-Forsythe test to check for equal variance. If the ANOVA assumptions were met, the test was followed by an appropriate post-hoc test, as indicated for each graph. The Brown-Forsythe and Welch ANOVA test were used whenever the condition for equal variance was not met. If the ANOVA assumptions were not met, the non-parametric Kruskal-Wallis test, followed by Dunn’s multiple comparisons test was applied instead.

A two-way ANOVA was applied for time-course CFU experiments on log10-transformed CFU data, followed by an appropriate post-hoc test, as indicated for each graph.

No matching or pairing was assumed in any ANOVA test.