prkA is not an essential gene in L. monocytogenes

In many pathogens, PASTA kinases contribute to the response to cell wall stress, intrinsic β-lactam resistance, and virulence. To identify targets and pathways modulated by PrkA in L. monocytogenes, we performed parallel phosphoproteomic and suppressor analyses in the context of cell wall stress. First, we set out to define the phosphoproteome of wild-type L. monocytogenes and a conditional ΔprkA mutant (ΔprkA_cond, described previously [28]) in which prkA is deleted from its native locus and a trans-encoded copy of prkA is controlled by a theophylline-inducible riboswitch. Wild type and ΔprkA_cond were exposed to a sub-inhibitory concentration of the β-lactam ceftriaxone (CRO), to which the ΔprkA_cond mutant is ~100-fold more sensitive than wild type in the absence of inducer [22]. This preliminary phosphoproteomic analysis revealed 241 unique arginine, aspartic acid, cystine, glutamic acid, histidine, lysine, serine, threonine, or tyrosine phosphopeptides in wild-type L. monocytogenes in the presence of CRO (S1 Table). Combining phosphopeptides that differed only by missed cleavages, oxidations, or deamidations, 23 unique phosphosites were absent or significantly less abundant in the ΔprkA_cond mutant, suggesting they may be direct PrkA targets (S1 Table).

Of note, one of the top ten most abundant phosphopeptides found in the ΔprkA_cond mutant samples mapped to PrkA, suggesting that the inducible riboswitch controlling the expression of prkA is leaky. Multiple independent groups have constructed conditional ΔprkA mutations due to an inability to make an unmarked deletion, suggesting, as is the case in M. tuberculosis, that the PASTA kinase PrkA is essential. To determine if PrkA is essential and to minimize phosphorylation events by PrkA in our phosphoproteomics experiments, we attempted to construct a new strain with a marked deletion mutant of prkA replaced with an erythromycin resistance cassette under a constitutive promoter, ΔprkA::erm, without introducing a trans-encoded copy of prkA. Unlike in the case of an unmarked deletion, we were able to recover ΔprkA::erm mutants, suggesting that PrkA is not essential. In contrast to what we have previously observed for the ΔprkA_cond strain, the ΔprkA::erm mutant had a slight growth defect in BHI compared to wild type (S1A Fig), perhaps suggesting why construction of unmarked clean deletions of prkA have been previously unsuccessful. This result also supports the notion that some amount of prkA is expressed in the ΔprkA_cond strain in the absence of inducer. Consistent with a critical role for PrkA in cell envelope stress responses, the ΔprkA::erm mutant was similarly sensitive as ΔprkA_cond to a variety of stressors, including CRO, ampicillin, lysozyme, and LL-37 (S1B Fig and Table 1). Expression of prkA from a constitutive promoter in trans restored both the growth defect in BHI (S1C Fig) and intrinsic resistance to CRO (S1D Fig) of the ΔprkA::erm mutant.

Download:

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

Table 1. MICs of cell envelope-targeting agents against ΔreoM strains in BHI.

https://doi.org/10.1371/journal.ppat.1009881.t001

Our previous studies on prkA had required the use of a conditional mutant [22,28] and Wamp et al. [52] recently reported that prkA is essential in L. monocytogenes. Therefore, to ensure that our marked ΔprkA::erm mutant was not recovered due to second site suppressor mutations, we transduced the ΔprkA::erm allele into a fresh wild-type 10403S background and performed whole-genome sequencing on 10 randomly selected transductants (ΔprkA::erm T1-T10) in addition to the ΔprkA::erm parent. The ΔprkA::erm parent contained one mutation in Lmrg_02080 (Table 2), annotated as lieB, a putative ABC transporter permease. One transductant contained a six base pair deletion at the Lmrg_02823 locus, encoding a putatively PG-anchored protein (LPXTG-containing) of unknown function. In two transductants, the prophage encoded at the comK locus had excised from the genome (Table 2). Importantly, in the remaining seven transductants, no mutations were detected relative to the parent wild-type strain, demonstrating that our ability to make the marked ΔprkA::erm deletion strain was not due to compensatory secondary mutations elsewhere in the genome. All 10 transductants grew similarly to the ΔprkA::erm parent both in BHI (S1A Fig) and in a CRO MIC assay (S1B Fig). We conclude therefore that prkA is not an essential gene in L. monocytogenes.

Download:

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

Table 2. Mutations in ΔprkA::erm strain and 10 ΔprkA::erm transductants compared to parent wild-type 10403S.

https://doi.org/10.1371/journal.ppat.1009881.t002

Phosphoproteomic analysis reveals PrkA-dependent phosphorylations during β-lactam stress

We selected ΔprkA::erm T1 to perform phosphoproteomic analysis because this transductant contained no mutations relative to wild type other than the deletion of prkA. Wild-type and ΔprkA::erm T1 cultures were treated with half an MIC of CRO and harvested at mid-log phase. Lysates were subjected to phosphopeptide enrichment, followed by LC-MS/MS analysis for phosphoproteome determination. To fully characterize the phosphoproteome of L. monocytogenes, we searched for arginine, aspartic acid, cystine, glutamic acid, histidine, lysine, serine, threonine, or tyrosine phosphorylations, and identified 127 unique phosphopeptides in wild-type L. monocytogenes present during β-lactam stress (S2 Table). We next separately searched for only serine and threonine phosphorylation events in wild-type and ΔprkA::erm samples to identify likely PrkA targets (Fig 1A and S3 Table). We defined PrkA-dependent phosphopeptides as those that were significantly (P value < 0.05) less abundant in the ΔprkA mutant compared to in wild type. Using this cutoff, we identified 39 PrkA-dependent phosphopeptides (Fig 1A and S3 Table). Combining phosphopeptides that differed only by missed cleavages, oxidations, or deamidations, there were 26 unique phosphorylation sites across 23 proteins (Fig 1B). Among this list of putative PrkA substrates were Listeria homologs of several confirmed or putative PASTA kinase substrates in other organisms, namely HPr [57], ReoM [42,52,53], GpsB [58], and HtrA [42] (Fig 1B). The remaining 18 proteins represented novel PASTA kinase substrates. The proteins identified cover a broad range of biological functions, among them cell division (GpsB, FtsH), metabolism (ROK family glucokinase, CysK), and stress response (HtrA, MnSOD, catalase) (Fig 1C, left panel). They include cytoplasmic, membrane-associated, and ribosomal proteins (Fig 1C, middle panel). At least twelve of the putative substrates have catalytic activity, and several have receptor and signal-transducing activity (Fig 1C, right panel). These data are thus consistent with a role for PrkA having pleiotropic effects on bacterial physiology during cell wall stress in L. monocytogenes.

Download:

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

Fig 1. PrkA phosphorylates proteins involved in diverse processes in the context of cell wall stress.

(A) Volcano plot showing log2 fold-change of abundance ratios versus -log10 P value of phosphorylated peptides with XCorr scores > 2.0 in the ΔprkA::erm mutant compared to wild type. The green box highlights peptides that were two-fold or less abundant and significantly different (P < 0.05) and the red box highlights peptides that were two-fold or more abundant and significantly different in the ΔprkA::erm mutant; some dots represent more than one phosphopeptide (indicated). (B) List of proteins containing phosphosites that were significantly less abundant in the ΔprkA::erm mutant compared to wild type. Differentially phosphorylated peptides from (A) that differed only by oxidations, deamidations, and/or missed cleavages were combined into the same phosphosite. Amino acids in brackets indicate ambiguous phosphosites. (C) GO term analysis of the proteins from (B). Some proteins are listed in more than one annotation per graph.

https://doi.org/10.1371/journal.ppat.1009881.g001

Mutations suppressing cell wall stress sensitivity of a ΔprkA mutant also restore virulence-associated defects

In parallel with the initial phosphoproteomics experiment, we performed a screen for mutations that could suppress the sensitivity of the ΔprkA_cond mutant to CRO to identify pathways modulated by PrkA during cell wall stress. We constructed an EMS-mutagenized library of the ΔprkA_cond mutant, targeting ~1 mutation/genome as calculated by a rifampicin resistance assay [59]. 10⁶ CFU of the library was plated on agar plates containing 2 μg/ml CRO, which completely inhibited the growth of the parent strain ΔprkA_cond but not wild type. We isolated 21 CRO-resistant suppressors for further analysis, naming them ΔprkA_cond CRO Suppressor (PCS) mutants. Because the ΔprkA_cond parent strain contains a trans-encoded prkA allele under the control of a theophylline-inducible riboswitch [28], we initially sequenced the riboswitch in the 21 PCS mutants to first rule out riboswitch mutations, which we presumed would result in increased prkA expression and be the cause of suppression. Of the 21 mutants, 9 (43%) contained riboswitch mutations. We moved forward with analysis of three randomly selected PCS mutants with wild-type riboswitch sequences: PCS2, PCS3, and PCS16. To confirm that these suppressors are indeed more resistant to CRO than the ΔprkA_cond parent, we measured the MIC of CRO for PCS2, PCS3, and PCS16. The MIC of CRO was restored to that of, or within two-fold of, wild type for all three suppressors (Fig 2A). The three suppressor mutants also had MICs for ampicillin (AMP) comparable to wild type (Table 3). Two of the three PCS mutants were additionally more resistant to lysozyme than ΔprkA_cond with the exception of PCS16, which was 8-fold more sensitive (Table 3).

Download:

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

Fig 2. Suppressors of CRO sensitivity also suppress ex vivo virulence defects of a ΔprkA mutant.

(A) Bars indicate median MICs of CRO for the indicated strains; n = 3. *, P < 0.05 compared to wild type, and #, P < 0.05 for the indicated comparisons, by one-way ANOVA with Tukey’s multiple comparisons test. (B) Intracellular bacteriolysis in immortalized Ifnar^-/- macrophages. Macrophages were infected with the indicated strains carrying the pBHE573 reporter vector at an MOI of 10, and luciferase activity was measured 6 hours post-infection. Error bars indicate standard error of the mean (SEM); n = 5. (C) Host cell death in primary C57BL/6 bone marrow-derived macrophages (BMDMs). BMDMs were infected with the indicated strains at an MOI of 10, and lactate dehydrogenase activity in cell supernatants was measured 6 hours post-infection. Error bars indicate SEM; n = 3. (B-C) *, P < 0.05 compared to wild type, and #, P < 0.05 compared to ΔprkA_cond by one-way ANOVA with Tukey’s multiple comparisons test. (D) Intracellular growth in C57BL/6 BMDMs. BMDMs seeded on glass coverslips were infected with the indicated strains at an MOI of 10, and CFU per coverslip was enumerated at the indicated time points. Data are representative of three biological replicates. (E) Plaque formation in immortalized murine fibroblasts (L2 cells). L2s were infected with the indicated strains at an MOI of ~0.5, and plaque sizes were normalized to those of wild type on day 6 of infection. Error bars indicate SEM; data are averaged from a minimum of 15 plaques from three biological replicates. N.D., not detected. *, P < 0.05 compared to wild type by one-way ANOVA with Tukey’s multiple comparisons test.

https://doi.org/10.1371/journal.ppat.1009881.g002

Download:

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

Table 3. MICs of cell envelope-targeting agents against PCS mutants in BHI.

https://doi.org/10.1371/journal.ppat.1009881.t003

We next assessed whether the three PCS mutants selected for their ability to suppress cell wall stress sensitivity could also relieve the severe phenotypes of a ΔprkA mutant during infection. Specifically, the ΔprkA_cond parent strain lyses more frequently than wild type in the cytosol of host cells, which induces more inflammasome-mediated death of host cells [22]. To measure cytosolic killing of the PCS mutants compared to ΔprkA_cond, we employed a luciferase-based reporter system previously described [6]. In this assay, L. monocytogenes carries a plasmid-borne luciferase gene under the control of a host-driven promoter, so that luciferase is only produced by the infected cell upon release of the plasmid from lysed bacteria and luciferase activity can be quantified to measure relative cytosolic bacteriolysis. A strain lacking LLO, the pore-forming cytolysin that enables escape from the vacuole, does not reach the cytosol and thus does not report in this assay (Fig 2B). A strain engineered to express two Listeria phage lysis proteins upon entry into the cytosol (holin-lysin) lyses at high levels, resulting in high luciferase activity (Fig 2B). Consistent with prior observations, the ΔprkA_cond mutant lysed at ~10-fold higher levels than wild-type Lmo (Fig 2B). All three PCS mutants displayed significantly less bacteriolysis than ΔprkA_cond, restoring lysis levels to that near wild type (Fig 2B). Correspondingly, while the ΔprkA_cond mutant caused ~3-fold higher cell death than wild-type L. monocytogenes as measured by LDH release, the PCS mutants caused less cell death than the ΔprkA_cond mutant (Fig 2C).

Consistent with prior observations [22], the ΔprkA_cond mutant was also unable to replicate in the cytosol of macrophages; indeed, bacterial burden was lower at the end of the growth curve than the start, supporting the idea that this mutant is killed in the cytosol (Fig 2D). In contrast, PCS3 and PCS16 were able to replicate at rates similar to wild type and reach similar burdens to wild type after 8 hours (Fig 2D). Growth of PCS2 was restricted, with bacterial numbers remaining static over the course of the growth curve (Fig 2D). Notably, however, PCS2 burdens did not decrease like those of the ΔprkA_cond parent (Fig 2D). Consistent with previous results [22], the ΔprkA_cond mutant was unable to form plaques in a monolayer of fibroblasts while wild type does so robustly. Plaque formation requires phagosomal escape, bacterial replication in the cytosol, and cell-to-cell spread. All three PCS mutants were able to plaque, albeit to differing degrees (Fig 2E). PCS16 formed almost wild-type-sized plaques, while PCS3 and especially PCS2 had diminished plaquing ability, forming plaques ~40% and ~5% of the size of wild type, respectively (Fig 2E). The minimal size of plaque formation by the PCS2 mutant is likely attributable to its slow intracellular growth (Fig 2D). Cumulatively, these data demonstrate that mutations that suppress sensitivity to cell wall-targeting antibiotics can restore some virulence-related phenotypes of a ΔprkA mutant, suggesting that cell wall stress responses regulated by PrkA are critical for L. monocytogenes virulence.

PCS mutations map to the ReoM pathway controlling MurA stability

To identify the mutations responsible for suppression, PCS2, PCS3, and PCS16 were submitted for whole genome sequencing. Six mutations relative to the ΔprkA_cond parent were found across the three genomes, two of which were common to all three mutants (Table 4). PCS2 had an additional unique missense mutation in the proteolytic subunit clpP, and PCS3 had a unique missense mutation in the ATPase subunit clpC (Table 4). Of note, clpP and clpC mutants alone have defects in intracellular replication [60,61], consistent with our observations that PCS2 and PCS3 only partially restore ΔprkA_cond defects in intracellular growth and plaque formation (Fig 2D and 2E). PCS16 additionally contained mutations in Lmrg_02076, a GNAT-family N-acetyltransferase, and 12 base pairs upstream of the reoM start codon (Table 4). Presumably, the mutation upstream of reoM reduced its expression, similar to a ΔgpsB suppressing mutation in the RBS of reoM recently isolated by Wamp et al. that is phenocopied by deletion of the reoM CDS [52]. Thus, each of the three PCS isolates contained a mutation in the ReoM pathway that controls MurA degradation, suggesting that PrkA-mediated phosphorylation of ReoM is important for the response to cell wall stress both in vitro and ex vivo.

Download:

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

Table 4. Mutations in PCS mutants compared to the ΔprkA_cond parent strain.

https://doi.org/10.1371/journal.ppat.1009881.t004

We next sequenced the genomes of the remaining 9 PCS mutants with wild-type riboswitch sequences to identify other potential pathways modulated by PrkA in the presence of CRO. These 9 mutants contained 1 to 6 mutations relative to the ΔprkA_cond parent. Three PCS mutants contained mutations in clpC, two mutants contained mutations in clpP, four mutants contained mutations in the CDS or immediately upstream of reoM, and three mutants contained mutations in the CDS or immediately upstream of murZ. MurZ is a MurA paralog, and through an unknown mechanism, loss of function murZ mutations also result in increased MurA stability [62]. While an intergenic SNP could affect a small RNA or other regulatory element, we postulate that the SNP in PCS1 upstream of murZ decreases murZ expression, similar to a SNP upstream of reoM that decreases reoM expression [52]. Therefore, each of the twelve PCS mutants sequenced contained a mutation in reoM, clpC, clpP, or murZ (Table 4). Of note, while 8 of the PCS mutants contained mutations in both Lmrg_00741 and Lmrg_01945, the remaining 4 mutants did not, suggesting that these mutations were not required to compensate for disruption of the ReoM pathway. Furthermore, two PCS mutants, PCS7 and PCS20, contained only one mutation relative to the ΔprkA_cond parent strain, suggesting that disruption of the ReoM pathway and concomitant stabilization of MurA is sufficient for rescuing growth of the ΔprkA_cond mutant in the presence of CRO. Cumulatively, consistent with Wamp et al., these findings suggest that the ReoM pathway is a critical PrkA-controlled adaptation to cell wall stress. Furthermore, our ex vivo virulence assays point to an important role for PrkA-mediated regulation of ReoM in promoting adaptation of L. monocytogenes to the eukaryotic cytosol.

PrkA-mediated regulation of ReoM is required for PG synthesis during cell wall stress

ReoM was one of the most abundant PrkA-dependent phosphoproteins we identified in L. monocytogenes during exposure to CRO, and all PCS mutants sequenced contained mutations in the ReoM pathway that modulates MurA levels. Therefore, we chose to move forward with investigating the role of ReoM as a PrkA substrate during cell wall stress and infection. We constructed an in-frame deletion of reoM (ΔreoM) and subsequently transduced the ΔprkA::erm allele into the ΔreoM mutant (ΔreoM ΔprkA::erm). The ΔreoM mutant was as resistant as wild type to all stressors tested and was slightly more resistant than wild type to CRO (Fig 3A), AMP, and LL-37 (Table 1). Strikingly, the double mutant ΔreoM ΔprkA::erm restored the MIC to within 2-fold of wild type for CRO (Fig 3A), AMP, lysozyme, and LL-37 (Table 1). Complementation of the ΔreoM ΔprkA::erm strain with a trans-encoded copy of reoM controlled by its native promoter (pPL2-P_reoM-reoM) restored sensitivity of the strain to all compounds tested (S2A Fig and Table 1). These results are consistent with the model put forth by Wamp et al. [52] and suggest that deletion of ReoM bypasses the requirement for phosphorylation by PrkA, preventing ClpCP-mediated degradation of MurA and resulting in increased PG synthesis to protect against cell wall insults.

Download:

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

Fig 3. Loss of reoM restores PG synthesis defects of a ΔprkA mutant of L. monocytogenes during cell wall stress.

(A) Bars indicate median MICs of CRO for the indicated L. monocytogenes strains; n = 3. *, P < 0.05 compared to wild type, and #, P < 0.05 for the indicated comparisons by one-way ANOVA with Tukey’s multiple comparisons test. (B-C) The indicated strains of L. monocytogenes were grown in the presence of 1 mM alkyne-d-ala-d-ala for 1 hour in the absence and presence of 0.5 MIC of CRO. Cells were subjected to a copper-catalyzed alkyne-azide cycloaddition reaction with a modified Alexa Fluor 488, and fluorescence was subsequently analyzed by flow cytometry, with a minimum of 1.5 x 10⁵ events read per sample. (B) Representative histograms for Alexa Fluor 488 signal. (C) Median fluorescence intensity (MFI) was normalized to that of WT minus CRO. *, P < 0.05 compared to wild type minus CRO by two-way ANOVA with Tukey’s multiple comparison’s test. #, P < 0.05 for the indicated comparisons by two-way ANOVA with Sidak’s multiple comparison’s test.

https://doi.org/10.1371/journal.ppat.1009881.g003

Our findings together with the model put forth by Wamp et al. [52] suggests that ΔprkA mutants are defective in cell wall synthesis and that deletion of reoM in a ΔprkA background should restore this defect by increasing flux into the PG synthesis pathway. We directly tested this hypothesis by measuring PG labeling in wild-type, ΔprkA::erm, ΔreoM, and ΔreoM ΔprkA::erm strains during cell wall stress as previously described [63,64]. Alkyne-conjugated d-alanine-d-alanine dipeptide (alk-d-ala-d-ala) was added to exponentially growing bacteria for 1 hour in the absence or presence of 0.5 MIC of CRO. The alk-d-ala-d-ala probe is incorporated into nascent PG and subsequently can be detected by copper-catalyzed azide-alkyne cycloaddition to a modified fluorophore (Fig 3B). In the absence of CRO, wild-type L. monocytogenes and ΔprkA::erm were similarly labeled, suggesting that these strains synthesized similar amounts of PG in the absence of stress, consistent with a minimal growth defect of ΔprkA::erm mutants in these conditions (Fig 3C). The ΔreoM mutant was significantly more labeled than wild type even in the absence of CRO, consistent with increased flux through MurA in the absence of reoM (Fig 3C) [52]. Upon exposure to CRO, labeling of wild-type L. monocytogenes increased to a similar magnitude as that of the ΔreoM mutant in the presence of CRO (Fig 3C), which was similar to its unstressed levels. In contrast, labeling of the ΔprkA::erm mutant did not increase upon exposure to CRO (Fig 3C). Both in the absence and presence of CRO, the ΔreoM ΔprkA::erm mutant was labeled at levels similar to that of the ΔreoM mutant (Fig 3C). These results strongly support the notion that PrkA is required for PG synthesis during cell wall stress and that deletion of its phosphosubstrate ReoM restores PG synthesis defects of a ΔprkA mutant. In addition, these observations support a model in which PG synthesis is disrupted and constitutively high in an ΔreoM mutant.

Deletion of reoM restores virulence-associated defects of a ΔprkA mutant

To assess whether PrkA-mediated regulation of ReoM is important for cytosolic survival, we measured intracellular bacteriolysis using the luciferase-based reporter system. Similar to the ΔprkA_cond mutant [22], the ΔprkA::erm mutant lysed at significantly higher rates than wild-type L. monocytogenes (Fig 4A). The ΔreoM mutant alone lysed at similar rates to wild type, indicating inhibition of excessive peptidoglycan synthesis by ReoM is not necessary for cytosolic survival (Fig 4A). The ΔreoM ΔprkA::erm mutant also lysed at wild-type levels (Fig 4A), suggesting PrkA-mediated control of PG synthesis through ReoM is important for responding to cell wall stress in the cytosol. Re-introduction of reoM via pPL2-P_reoM-reoM into the ΔreoM ΔprkA::erm mutant resulted in increased cytosolic killing (S2B Fig). As an ex vivo measure of virulence, we also assessed the ability of ΔprkA::erm, ΔreoM, and ΔreoM ΔprkA::erm to form plaques in a monolayer of fibroblasts. Like the ΔprkA_cond mutant [22], the ΔprkA::erm mutant was unable to plaque (Fig 4B). The ΔreoM mutant alone exhibited a plaquing defect, forming plaques ~50% of the size of wild type (Fig 4B), and this defect was complemented by pPL2-P_reoM-reoM (S2C Fig). This finding implies that ReoM is required during the intracellular L. monocytogenes lifecycle and points to a potential role for ReoM in virulence. The ΔreoM ΔprkA::erm mutant partially suppressed the plaquing defect of ΔprkA::erm, forming plaques equivalent in size to the ΔreoM mutant (Fig 4B). Plaquing by the double mutant was substantially diminished by re-introduction of reoM via pPL2-P_reoM-reoM (S2C Fig). Of note, the complement strain ΔreoM ΔprkA::erm pPL2-P_reoM-reoM formed ~80% fewer plaques than the other strains, suggesting that reintroduction of reoM restored the ΔprkA mutant phenotype. The small number of plaques that did form in the ΔreoM ΔprkA::erm pPL2-P_reoM-reoM complement strain formed plaques roughly half the size of the ΔreoM ΔprkA::erm parent strain (S2C Fig). To account for this observation in our statistical analysis, we imputed values of 0 for plaque sizes up to the next fewest number of plaques formed by any strain in that experiment, which was ΔreoM pPL2-P_reoM-reoM. Altogether, these observations indicate that deletion of reoM rescues virulence phenotypes associated with loss of prkA in the cytosol, suggesting that PrkA-mediated control of PG synthesis is crucial in this environment. Additionally, these results suggest that ReoM is required for virulence independent of its role as a PrkA substrate.

Download:

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

Fig 4. Loss of reoM restores ex vivo virulence defects of a ΔprkA mutant.

(A) Intracellular bacteriolysis in immortalized Ifnar^-/- macrophages. Macrophages were infected with the indicated strains carrying the pBHE573 reporter vector at an MOI of 10, and luciferase activity was measured 6 hours post-infection. Error bars indicate SEM; n = 4. *, P < 0.05 compared to wild type and #, P < 0.05 for the indicated comparison by one-way ANOVA with Tukey’s multiple comparisons test. (B) Plaque formation in immortalized murine fibroblasts (L2 cells). L2s were infected with the indicated strains at an MOI of ~0.5, plaques were stained on day 4 of infection, and sizes were normalized to those of wild type. Error bars indicate SEM; data are averaged from a minimum of 81 plaques from three biological replicates. N.D., not detected. *, P < 0.05 compared to wild type by one-way ANOVA with Tukey’s multiple comparisons test.

https://doi.org/10.1371/journal.ppat.1009881.g004

PrkA-mediated regulation of ReoM is critical for L. monocytogenes infection

We next investigated the impact of PrkA-mediated ReoM regulation on virulence of L. monocytogenes. Wild-type, ΔprkA::erm, ΔreoM, ΔreoM ΔprkA::erm, and the corresponding reoM complementation strains were used to infect C57BL/6 mice intravenously, and bacterial burdens were measured 48 hours post-infection. Consistent with our prior findings with the ΔprkA_cond mutant [22], CFU isolated from mice infected with the ΔprkA::erm mutant numbered at or below the limit of detection in both the spleen and the liver (Fig 5). The ΔreoM mutant was nearly as virulent as wild type in the spleen (Fig 5A) but exhibited a ~3-log defect in the liver, which could be complemented by pPL2-P_reoM-reoM (Fig 5B). Mice infected with the ΔreoM ΔprkA::erm mutant harbored nearly as many CFU as those infected with wild-type L. monocytogenes in both the spleen and the liver (Fig 5). The ΔreoM ΔprkA::erm mutant was attenuated by reintroduction of reoM via pPL2-P_reoM-reoM (Fig 5). These results suggest that increasing PG synthesis through deletion of reoM is sufficient to largely restore the severe virulence phenotype of a ΔprkA mutant, but that excessive PG synthesis caused by lack of ReoM in a wild-type strain is detrimental to virulence.

Download:

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

Fig 5. PrkA-mediated regulation of ReoM is essential during infection.

(A-B) Nine-week-old C57BL/6 mice were infected intravenously with 1 x 10⁵ CFU of the indicated strains. Forty-eight hours post-infection, spleens (A) and livers (B) were harvested and CFU enumerated. Solid bars indicate medians and dotted lines indicate limits of detection. Data are representative of two independent experiments of 5 mice each. *, P < 0.05 compared to wild type by Kruskal-Wallis test with Dunn’s multiple comparisons test.

https://doi.org/10.1371/journal.ppat.1009881.g005

PASTA kinases are important for the response to myriad stressors and conditions that could be encountered in the host during infection [39]. Given that deletion of reoM did not fully restore virulence of the ΔprkA mutant in vivo, we hypothesized that other PrkA phosphorylation-dependent pathways may also contribute to pathogenesis. Therefore, we initiated an in vivo suppressor screen to gain insight into the adaptations that PrkA regulates during infection. Twelve wild-type mice were each infected with 1x10⁷ CFU of an EMS-mutagenized library of the ΔprkA::erm strain, representing ~100-fold coverage of each possible G:C to A:T transition in the L. monocytogenes genome in total. Seventy-two hours post-infection, spleens and livers were plated to enumerate CFU and isolate suppressors. Each organ contained CFU above the limit of detection, with some burdens approaching wild-type levels of ~10⁷ (S3 Fig). Following purification, one isolate from the spleen of each mouse, named ΔprkA::erm virulence suppressor (PEVS) mutants 1–12, was submitted for whole-genome sequencing. Two PEVS isolates had frameshift mutations in reoM, three isolates had mutations in reoY, and six isolates had frameshift mutations in murZ, all eleven of which resulted in premature stop codons (Table 5). The twelfth isolate contained a missense mutation in murA that did not map to the active site or other known regions of protein-protein interactions (Table 5). Thus, all twelve isolates contained a mutation in the ReoM pathway of MurA regulation, most of which were the only mutation relative to the ΔprkA::erm parent (8/12 isolates). Remarkably, although G:C to A:T transitions are present in the genomes of many of the in vivo suppressor isolates, none of the 12 mutations mapping to reoM, reoY, murZ, or murA are EMS-induced mutations, indicating that there is strong selective pressure in the context of the host to disrupt this pathway in the absence of prkA. Cumulatively, these results suggest that regulation of ReoM by PrkA, which results in stabilization of MurA and increased PG synthesis, is a primary role of PrkA during L. monocytogenes infection.

Download:

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

Table 5. Mutations in PEVS mutants compared to the ΔprkA::erm parent strain.

https://doi.org/10.1371/journal.ppat.1009881.t005

The PASTA kinase/ReoM pathway is conserved in Staphylococcus aureus

Originally identified in E. faecalis, ReoM homologs are conserved among Gram-positive PASTA-kinase containing organisms [65], and it was previously shown that the ReoM pathway is also intact in the model organism Bacillus subtilis [52]. To determine whether the function of ReoM and the downstream pathway that controls degradation of MurA is conserved and PASTA-kinase regulated in the important human pathogen S. aureus, we created an in-frame deletion of its reoM homolog (SAUSA300_1574). Consistent with previous findings [31,32], a mutant of S. aureus lacking the PASTA kinase Stk1 is more sensitive than wild type to the clinically relevant β-lactam antibiotic oxacillin (Fig 6A). The ΔreoM mutant alone was as resistant as wild type to oxacillin (Fig 6A). Similar to our findings in L. monocytogenes, deletion of reoM in the Δstk1 background restored resistance of the Δstk1 mutant to the cell wall-targeting antibiotic to wild-type levels (Fig 6A). We also determined that loss of reoY (SAUSA300_1353) (Fig 6B) or clpP (Fig 6C) restored intrinsic oxacillin resistance to the Δstk1 mutant, suggesting that the ReoM pathway functions similarly in S. aureus as it does in L. monocytogenes. Altogether, these findings suggest that ReoM is an important, conserved target of PASTA kinases upon activation by cell wall stress.

Download:

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

Fig 6. The PASTA kinase/ReoM pathway is conserved in S. aureus.

The indicated S. aureus strains were grown in increasing concentrations of oxacillin for 12 hours. Endpoint growth measured by OD₆₀₀ is reported. Error bars indicate SEM; (A) n = 3, (B-C) n = 4.

https://doi.org/10.1371/journal.ppat.1009881.g006

Ethics statement

All experiments involving animals were approved by the Institutional Animal Care and Use Committee of the University of Wisconsin-Madison.

Bacterial strains, cloning, and growth conditions

L. monocytogenes was routinely grown at 37°C in brain heart infusion (BHI) broth or on BHI + 1.5% agar plates and frozen in BHI + 40% glycerol. S. aureus was routinely grown at 37°C in tryptic soy broth (TSB) or on TSB + 1.5% agar plates and frozen in BHI + 40% glycerol. E. coli was routinely grown at 37°C in LB or on LB + 1.5% agar plates and frozen in LB + 40% glycerol. E. coli XL1-Blue was used for sub-cloning and SM10 or S17 were used for conjugating plasmids into L. monocytogenes. Antibiotics were used at the following concentrations: 200 μg/ml streptomycin, 50 μg/ml gentamicin, 10 μg/ml chloramphenicol, 30 μg/ml kanamycin, 2 μg/ml erythromycin, 1 μg/ml anhydrotetracycline.

Primers used for this study are listed in Table 6, plasmids in Table 7, and strains in Table 8. In-frame deletions were generated in L. monocytogenes as previously described [73] using pKSV7-oriT [74]. The pKSV7-ΔprkA::erm plasmid was constructed by Gibson Assembly with linearized pKSV7 and 3 PCR-generated fragments: a ~1 kb fragment upstream of the prkA ORF, amplified with JLK57 and JLK55 from L. monocytogenes 10403S gDNA; an erm cassette with a constitutive promoter, amplified with JLK56 and JLK4 from pPL2e-riboE-prkA [28]; and a ~1 kb fragment downstream of the prkA ORF, amplified with JLK5 and JLK58 from 10403S gDNA. The pKSV7-ΔreoM plasmid was constructed by Gibson Assembly with linearized pKSV7 and a gBlock (IDT) comprised of ~500 bp up- and downstream of the reoM ORF. The pIMK2-prkA plasmid was constructed by Gibson Assembly with linearized pIMK2 [75] and a PCR fragment of the prkA CDS amplified from 10403S gDNA using primers JLK162 and JLK163. The pPL2-P_reoM-reoM plasmid was constructed by Gibson Assembly with linearized pPL2 [76] and a PCR fragment of reoM and its native promoter amplified from 10403S gDNA using primers JLK105 and JLK106. The pJB38-ΔreoM plasmid was constructed using Gibson Assembly with pJB38 [77] linearized with restriction enzymes EcoRI and KpnI and PCR fragments of ~1kb up and downstream of the reoM ORF amplified with primer pairs CMG55/CMG56 and CMG57/CMG58, respectively, from genomic DNA purified from S. aureus USA300 JE2. All plasmid inserts were verified by Sanger sequencing. Phage transduction in L. monocytogenes was performed as previously described [78] using U153 phage. S. aureus transposon mutants were generated by transducing transposons from the Nebraska Transposon Mutant Library [79] into our laboratory stock of wild-type USA300 JE2. Phage transduction in S. aureus was performed as previously described [79,80] using ϕ85 phage.

Download:

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

Table 6. Primers used in this study.

https://doi.org/10.1371/journal.ppat.1009881.t006

Download:

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

Table 7. Plasmids used in this study.

https://doi.org/10.1371/journal.ppat.1009881.t007

Download:

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

Table 8. Strains used in this study.

https://doi.org/10.1371/journal.ppat.1009881.t008

Sample preparation for phosphoproteome analysis

L. monocytogenes strains for phosphoproteome determination were grown overnight (16–18 hours) in BHI at 37°C with shaking. Cultures were back-diluted 1:50 into 50 ml BHI containing half an MIC of ceftriaxone (see Table 4) in a 250 ml beveled flask and grown at 37°C with shaking to an OD₆₀₀ of ~0.5. Cultures were pelleted at 3220 x g for 10 minutes at 4°C, washed once in an equal volume of PBS, and pelleted again. Pellets were resuspended in 1 ml buffer (50 mM Tris, pH 8.0, 10 mM DTT, 0.1% SDS, 100 μM PMSF, phos-stop tablet [Roche]), and cells were lysed by 2 minutes of vortexing with zirconia beads (BioSpec Products). Lysates were pelleted by spinning at 21,130 x g for 10 minutes at 4°C, and the supernatant was filtered through a 0.22 μm PES filter and stored at -80°C.

Enzymatic “in liquid” digestion and phosphoenrichment

Clarified and filtered lysates from L. monocytogenes were TCA/acetone precipitated (10% TCA, 40% acetone final) for protein extraction then pellets ([330 μg) re-solubilized and denatured in 100 μl of 8 M urea/50 mM NH₄HCO₃ (pH 8.5)/1 mM Tris-HCl. 2.5 μl of 1 mg/ml BSA was spiked as an internal standard and samples were diluted to 300 μl final volume with 160 μl of 25 mM NH₄HCO₃ (pH 8.5), 25 μl of Methanol, and 12.5 μl of 25 mM DTT. Samples were sonicated in a sonication bath for 3 minutes then incubated at 56°C for 15 minutes during the reduction step, subsequently cooled on ice to room temperature, then 15 μl of 55 mM IAA was added for alkylation and incubated in darkness at room temperature for 15 minutes. The reaction was quenched by adding 40 μl of 25 mM DTT and subsequently 5 μl of 100 mM CaCl₂ and 25 μl of Trypsin stock was added (400 ng/μl Trypsin Platinum from Promega Corp. in 100 mM Tris-HCl [pH 8]) plus 115 μl of 25 mM NH₄HCO₃ (pH 8.5) to 500 μl final volume. Digestion was conducted overnight at 37°C. Reaction was terminated by acidification with 2.5% TFA (Trifluoroacetic Acid) to 0.3% final concentration and samples were diluted to 800 μl total volume with 24 μl of 10% HFBA and 206 μl of milliQ water (final concentrations of 0.3% HFBA, 0.22% TFA and 3% methanol). These samples were desalted and concentrated using Strata-X 33 μm Polymeric Reverse Phase SPE cartridges (Phenomenex) per manufacturer protocol, where in short, methanol conditioned and water equilibrated cartridges had samples applied in 2x 400 μl aliquots, washed quickly with milliQ water and eluted in 2x 300 μl of 70/29/1% ACN/H₂O/FA then dried to ~20 μl in the speed-vac and brought back to 150 μl total volume with 1 M glycolic acid in 80% ACN and 5% TFA, loading buffer. Phosphoenrichments were carried out in two stages, first using 1 mg of titanium dioxide magnetic microparticles (MagReSyn TiO₂ from ReSyn Biosciences) then unbound material was enriched with 1 mg of Zirconium-ion (Zr⁴⁺) functional magnetic microparticles (MagReSyn Zr-IMAC from ReSyn Biosciences) and combined with the initial enriched eluate. Sample handling for both enrichments was according to the manufacturer protocols and evaluation of enrichments was internally standardized with a 5 pmol spike per each digest of AngioII-phosphate peptide standard.

NanoLC-MS/MS and data analysis

Samples were analyzed on Orbitrap Fusion Lumos Tribrid platform, where 3 μl of each enriched sample was injected using Dionex UltiMate3000 RSLCnano delivery system (ThermoFisher Scientific) equipped with an EASY-Spray electrospray source (held at constant 50°C). Chromatography of peptides prior to mass spectral analysis was accomplished using capillary emitter column (PepMap C18, 2 μM, 100 Å, 500 x 0.075 mm, Thermo Fisher Scientific). A NanoHPLC system delivered solvents A: 0.1% (v/v) formic acid, and B: 80% (v/v) acetonitrile, 0.1% (v/v) formic acid at 0.30 μL/min to load the peptides at 2% (v/v) B, followed by quick 1 minute gradient to 5% (v/v) B and gradual analytical gradient from 5% (v/v) B to 37.5% (v/v) B over 76 minutes followed by sharper gradient from 37.5% (v/v) B to 95% (v/v) B over 25 minutes when it concluded with a 5-minute flash-out at 95% (v/v) B. As peptides eluted from the HPLC-column/electrospray source survey MS scans were acquired in the Orbitrap with a resolution of 120,000 followed by HCD-type MS2 fragmentation into Ion Trap (32% collision energy) under ddMSnScan 1 second cycle time mode with peptides detected in the MS1 scan from 350 to 1600 m/z; redundancy was limited by dynamic exclusion and MIPS filter mode ON.

Lumos acquired MS/MS data files were searched using Proteome Discoverer (2.4.1.15) Sequest HT search engine against Listeria monocytogenes 10403S database (2,797 total sequences) plus cRAP common lab contaminant database (116 sequences). Static cysteine carbamidomethylation, variable methionine oxidation plus asparagine, glutamine deamidation, and serine, threonine phosphorylation with 2 tryptic miss-cleavages and peptide mass tolerances set at 10 ppm with fragment mass at 0.6 Da were selected. Peptide and protein identifications were accepted under strict 1% FDR cut offs with high confidence XCorr thresholds of 1.9 for z = 2 and 2.3 for z = 3. Strict principles of parsimony were applied for protein grouping. Chromatograms were aligned for feature mapping and area-based quantification using unique and razor peptides. Samples were normalized to spiked-in AngioII-phosphate. Phosphopeptides were filtered for “contaminant”, “Quan Values”, and “Phospho” modifications. Gene ontology (GO) terms were acquired for proteins in Proteome Discoverer.

CRO suppressor isolation

The ΔprkA_cond strain was mutagenized by a 10-minute exposure to 17.8 mM ethyl methanesulfonate (EMS) following a protocol previously described [59,81]. 10⁶ CFU of this library was plated on BHI agar plates containing 2 μg/ml CRO and grown at 37°C. Twenty-one colonies were picked either 24 or 48 hours post-plating and screened secondarily on 2 μg/ml CRO plates (all 21 isolates grew on the second selection plate). Isolates were subsequently picked into 96-well plates containing 100 μl BHI broth, grown overnight at 37°C, pelleted by centrifugation, resuspended in 100 μl BHI + 40% glycerol, and stored at -80°C.

Whole genome sequencing and SNP identification

L. monocytogenes isolates for whole-genome sequencing were grown overnight (16–18 hours) in BHI. Genomic DNA was purified using the MasterPure Gram-positive DNA purification kit (Epicentre) per the manufacturer’s instructions, except that 5 U/μl mutanolysin was used instead of lysozyme. DNA was submitted to the Microbial Genome Sequencing Center (MiGS, Pittsburgh, PA) for whole-genome sequencing using the Illumina NextSeq 2000 platform (150 bp paired-end reads). Reads were mapped onto the L. monocytogenes 10403S reference sequence (RefSeq accession number NC_017544) using Bowtie2 (version 2.3.5.1) [82]. Single nucleotide polymorphisms (SNPs) were called using SAMtools and BCFtools (version 1.9) [83,84].

Growth assays

L. monocytogenes strains were grown overnight (16–18 hours) in BHI at 30°C on a slant. For BHI growth assays, cultures were inoculated 1:50 into 100 μl of BHI in 96-well plates. Optical density (OD₆₀₀) was measured every 15 minutes in an Eon Microplate Spectrophotometer (BioTek Instruments, Inc., Winooski, VT), and plates were incubated at 37°C with 180 RPM shaking between time points. For L. monocytogenes MIC assays, cultures were inoculated 1:50 into 100 μl BHI containing serial 2-fold dilutions of the indicated compound in 96-well plates. Optical density (OD₆₀₀) was measured at 0 and 12 hours post-inoculation in a Synergy HT Microplate Spectrophotometer (BioTek Instruments, Inc., Winooski, VT), and plates were incubated on a 37°C shaker platform (250 RPM) between time points. The MIC was defined as the lowest concentration of CRO at which the endpoint OD₆₀₀ minus the starting OD₆₀₀ was <0.1, the lowest OD₆₀₀ with visible turbidity. For S. aureus MIC assays, Clinical and Laboratory Standards Institute protocols were followed with minor modifications [85]. Frozen stocks of S. aureus were streaked onto fresh TSA and incubated at 37°C for 15 hours. Single colonies were selected in biological duplicate, inoculated into 5 mL TSB, and grown for 15 h at 37°C with 180 RPM shaking. Strains were back diluted 1:100 in 100 μl cation-adjusted Muller Hinton broth supplemented with 2% sodium chloride (CA-MH) containing 2-fold dilutions of oxacillin 96-well plates and incubated at 37°C with 180 RPM shaking. After 12 hours, final OD₆₀₀ was measured in the Eon spectrophotometer.

Bacteriolysis assays

Intracellular lysis of L. monocytogenes was measured by luciferase reporter activity as previously described [6]. Briefly, 5 x 10⁵ immortalized Ifnar^-/- BMDMs were seeded into 24-well plates overnight. L. monocytogenes strains carrying the pBHE573 reporter construct [6] were grown overnight (16–18 hours) in BHI at 30°C on a slant. Cells were infected at an MOI of 10. One hour post-infection, media was removed from cells and replaced with fresh media containing gentamicin. Six hours post-infection, cells were lysed with TNT buffer, and luciferase activity was measured using luciferin reagent as previously described [6]. Luminescence was measured in a Synergy HT Microplate Spectrophotometer (BioTek Instruments, Inc., Winooski, VT).

LDH assays

Induction of host cell death by L. monocytogenes was measured by lactate dehydrogenase (LDH) release as previously described [86]. Briefly, 5 x 10⁵ BMDMs were seeded into 24-well plates overnight with 100 ng/ml Pam3CSK4 pre-treatment. L. monocytogenes strains were grown overnight (16–18 hours) in BHI at 30°C on a slant. Cells were infected at an MOI of 10. Half an hour post-infection, media was removed from cells and replaced with fresh media containing gentamicin and Pam3CSK4. Six hours post-infection, LDH release from infected cells was measured as previously described [87,88] and normalized to 100% lysis (determined by addition of Triton X-100 to a control well at a final concentration of 1%).

Intracellular growth assays

Bone marrow-derived macrophages (BMDMs) were isolated from C57BL/6 mice as previously described [89]. BMDMs were seeded at 5 x 10⁶ cells/60 mm dish with coverslips and allowed to adhere overnight. L. monocytogenes strains were grown overnight (16–18 hours) in BHI at 30°C on a slant. BMDMs were infected at an MOI of 0.2. Half an hour post-infection, media was removed and replaced with fresh media containing gentamicin. The number CFU per coverslip was determined at the indicated time points post-infection as previously described [90]. Shown is a representative growth curve from three biological replicates.

Plaquing assays

Plaque assays were conducted using the L2 cell line (murine fibroblasts [91]) essentially as previously described [92]. Briefly, 1.2 x 10⁶ cells were seeded in 6-well plates overnight. L. monocytogenes strains were grown overnight (16–18 hours) in BHI at 30°C on a slant. Cells were washed twice with pre-warmed PBS and then infected at an MOI of ~0.5 One hour post-infection, the media was removed, cells were washed thrice with pre-warmed PBS, and then fresh media containing gentamicin and 0.7% agarose was added to the wells. A second agarose plug was added ~48 hours post-infection. Three to five days post-infection, agarose plugs were removed, cells were stained with 0.3% crystal violet for 10 minutes, and wells were washed twice with double distilled water and allowed to dry completely. Stained wells were imaged, and plaque sizes were measured using ImageJ software (version 1.53).

Peptidoglycan labeling in vitro and quantitation by flow cytometry

Peptidoglycan synthesis by L. monocytogenes grown in broth was measured by copper-catalyzed azide-alkyne cycloaddition (CuAAC) reactions essentially as previously described [63,64]. Cells growing exponentially in BHI were back-diluted to OD₆₀₀ 0.3 in 1 ml BHI medium, with or without the addition of 0.5 MIC of CRO (as determined in Table 4). 1 mM of alkyne-d-alanine-d-alanine (EDA-DA, [64]) was added and cultures were grown for 1 hour to mid-log phase (final OD₆₀₀ 0.5–0.9) at 37°C with shaking. ~100 μl aliquots (normalized by OD₆₀₀) were washed twice with an equal volume PBS, fixed in ice-cold 70% ethanol for 10 minutes at -20°C, and again washed twice with PBS. Cells were resuspended in 50 μl of CuAAC reaction mixture (128 μM TBTA, 1 mM CuSO₄, 1.2 mM freshly prepared sodium ascorbate, and 20 μM AFDye 488 Azide Plus fluorophore [Click Chemistry Tools] in PBS + 0.1% Triton X-100 + 0.01% BSA) and incubated with shaking at room temperature for 30 minutes in the dark. Cells were washed twice with PBS and finally resuspended in FACS buffer. Samples were acquired using an LSRII flow cytometer (BD Biosciences, San Jose, CA) with FACSDiva software (BD Biosciences) and analyzed using FlowJo software (Tree Star, Ashland, OR). Median fluorescence intensities (MFIs) are reported.

Mouse infections

Intravenous mouse infections were performed as previously described [86]. L. monocytogenes strains were grown overnight (16–18 hours) in BHI at 30°C on a slant. Overnight cultures were back-diluted 1:5 into BHI and grown at 37°C with shaking to mid-log phase (OD₆₀₀ 0.4–0.6). Cultures were then washed with PBS and adjusted to 5 x 10⁵ CFU/ml. Nine-week-old C57BL/6 mice (Charles River Laboratories) were infected with 1 x 10⁵ CFU in 200 μl PBS. Forty-eight hours post-infection, animals were euthanized by CO₂ asphyxiation according to AVMA standards. Spleens and livers were harvested, homogenized in PBS + 0.1% NP-40, and plated for CFU. Infections were performed in biological duplicate, one with female mice and one with male mice.

In vivo suppressor screen

The ΔprkA::erm mutant was mutagenized by a 3-minute exposure to ethyl methanesulfonate (EMS) following a protocol previously described [59,81]. One ml of the library was thawed, washed in PBS and resuspended in BHI, back-diluted 1:5 into BHI, and grown at 37°C with shaking for one doubling (measured by OD₆₀₀). Cultures were then washed with PBS and adjusted to 5 x 10⁷ CFU/ml. Twelve 24-week-old female C57BL/6 mice (Charles River Laboratories) were each infected intravenously with 1 x 10⁷ CFU of this library in 200 μl PBS. Seventy-two hours post-infection, spleens were harvested, homogenized, and plated on BHI plates with streptomycin. Colonies from each mouse were picked into 96-well plates containing 100 μl BHI broth, grown overnight at 37°C with shaking, pelleted by centrifugation, resuspended in 100 μl BHI + 40% glycerol, and stored at -80°C. One isolate from the spleen of each mouse was subsequently streaked onto BHI plates with streptomycin, grown overnight in BHI broth, and subjected to whole genome sequencing and SNP analysis as described above.

Statistics

Statistical analyses were performed with GraphPad Prism (version 9). Means from more than two groups were compared using one-way analysis of variance (ANOVA) or Kruskal-Wallis tests as appropriate with the post-tests indicated in the figure legends. For the phosphoproteomics experiments, P values were calculated using the Proteome Discoverer software (version 2.4.1.15). For bacteriolysis assays, the holin-lysin strain was not included in statistical analyses as it is an engineered control and is thus not biologically meaningful.