The Small RNA Teg41 Is a Pleiotropic Regulator of Virulence in Staphylococcus aureus

ABSTRACT

Previously, our group demonstrated a role for the small RNA (sRNA) Teg41 in regulating production of the alpha phenol-soluble modulin toxins (αPSMs) in Staphylococcus aureus. Overexpressing Teg41 increased αPSM production while deleting the 3′ end of Teg41 (Teg41Δ3′ strain) resulted in a decrease in αPSM production, reduced hemolytic activity of S. aureus culture supernatants, and attenuated virulence in a murine abscess model of infection. In this study, we further explore the attenuation of virulence in the Teg41Δ3′ strain. Using both localized and systemic models of infection, we demonstrate that the Teg41Δ3′ strain is more severely attenuated than an ΔαPSM mutant, strongly suggesting that Teg41 influences more than the αPSMs. Proteomic and transcriptomic analysis of the wild-type and Teg41Δ3′ strains reveals widespread alterations in transcript abundance and protein production in the absence of Teg41, confirming that Teg41 has pleiotropic effects in the cell. We go on to investigate the molecular mechanism underlying Teg41-mediated gene regulation. Surprisingly, results demonstrate that certain Teg41 target genes, including the αPSMs and βPSMs, are transcriptionally altered in the Teg41Δ3′ strain, while other targets, specifically spa (encoding surface protein A), are regulated at the level of transcript stability. Collectively, these data demonstrate that Teg41 is a pleiotropic RNA regulator in S. aureus that influences expression of a variety of genes using multiple different mechanisms.

INTRODUCTION

Staphylococcus aureus is a human pathogen that can cause a variety of different diseases, ranging from minor skin and soft tissue infections to bacteremia and endocarditis, which can be fatal (1). The virulence factors that contribute to these disease outcomes are regulated by an intricate network of transcription factors, two-component systems, and small regulatory RNAs (sRNAs). Of these categories of regulators, sRNAs are less well characterized, though several S. aureus sRNAs have been found to be important regulators of virulence (2–4), metabolism (5), antibiotic resistance (6), and biofilm formation and persistence (7). Previous work by our group identified the function of the sRNA Teg41, a transcript 205 nucleotides in length that is well conserved among S. aureus isolates (8). We found that Teg41 regulates the adjacent αPSM transcript, which encodes four cytolytic alpha phenol-soluble modulin peptides (αPSMs). PSMs are small, amphipathic peptides that are responsible for non-receptor-mediated host cell lysis, proinflammatory cytokine production, and biofilm formation and spread (9). We constructed a strain in which we deleted 24 nucleotides on the extreme 3′ end of Teg41 (called the Teg41Δ3′ strain) and demonstrated that this strain was severely attenuated in PSM production, virulence, and hemolysis. We also showed that expression of the 3′ end of Teg41 alone is sufficient for αPSM-mediated hemolysis, and the 3′ end complements the loss of Teg41 in the Teg41Δ3′ strain.

In the current study, we further demonstrate the importance of Teg41 in S. aureus virulence regulation by identifying new potential targets for Teg41 and characterizing the nature of Teg41-mediated regulation of these targets. We show that the Teg41Δ3′ strain is more severely attenuated than an ΔαPSM mutant strain in both murine abscess and murine systemic models of infection, highlighting that Teg41 contributes more to virulence that simply regulating αPSM production. This result is confirmed by transcriptome sequencing (RNA-seq) analysis, which shows that transcriptomes from the Teg41Δ3′ and the ΔαPSM mutants are distinct, with 113 transcripts being differently expressed between these two strains. Further RNA-seq analyses, comparing the wild-type (WT) and Teg41Δ3′ strains at different time points of growth, depict substantial alterations in the transcriptome in the absence of Teg41. Genes encoding a number of well-characterized virulence factors, such as the psmβ transcript and spa, are shown to be differentially expressed, and we confirm our previous finding that psmα transcript levels are greatly decreased in the Teg41Δ3′ mutant. A chromosomally expressed single copy of Teg41 (complement strain) perfectly restores expression of altered genes in the Teg41Δ3′ strain, emphasizing the significance of our results. Proteomic analysis is also carried out on cytoplasmic and secreted fractions, which validates the RNA-seq data demonstrating decreased αPSM and βPSM production but increased protein A in the Teg41Δ3′ strain.

Finally, to investigate the mechanism of Teg41-mediated regulation, we employ β-galactosidase assays to investigate whether alterations in transcript levels of Teg41-regulated genes could be explained by changes in promoter activity, or if regulation is posttranscriptional. Interestingly, results show that Teg41-mediated regulation can occur at different levels depending on the specific target. Transcription from the psmα promoter is reduced in the Teg41Δ3′ strain; however, no alteration of spa promoter activity is observed in this strain, suggestive of transcriptional and posttranscriptional regulation, respectively.

Collectively, the data presented here demonstrate that the role of Teg41 in S. aureus extends beyond regulation of αPSM production and Teg41 represents a pleiotropic regulatory RNA. In support of this, a recent computational analysis by Subramanian et al. identified Teg41, designated sRNA95 in their study, as an important regulator of S. aureus pathogenicity that is likely a key node in a variety of regulatory cascades including hemolysis, pathogenicity, and metabolism (10). Our results strongly support this hypothesis by showing that Teg41 is not only a regulator of the αPSMs but also a multifunctional regulator that is required for S. aureus pathogenicity.

RESULTS

The Teg41Δ3′ strain is more severely attenuated than an ΔαPSM mutant in murine abscess and systemic dissemination infection models.

In our previous work (8), we showed that the sRNA Teg41 is able to positively regulate αPSM production. We did so using a strain of S. aureus in which 24 nucleotides at the 3′ end of Teg41 were deleted (Teg41Δ3′ strain). We have been unable to construct a complete deletion of the entire Teg41 gene; however, Northern blot analysis failed to detect any truncated Teg41 product in the Teg41Δ3′ strain (8), suggesting that the 5′ end alone is unstable and that the Teg41Δ3′ strain is functionally similar to a Teg41 mutant. To confirm that the phenotypes associated with the Teg41Δ3′ strain could be complemented with a single chromosomal copy of Teg41 (rather than plasmid-based overexpression), we cloned full-length Teg41, along with its native promoter, into pJC1112, a shuttle vector that integrates at the SaPI1 site of the S. aureus genome (11). We have here designated this strain (AH1263 Teg41Δ3′ pJC1112_Teg41::SaPI1) “Teg41 complement.” In all assays tested, phenotypes associated with the Teg41Δ3′ strain are restored to wild-type levels in the Teg41 complement strain (for example, see hemolysis assay and abscess infection assay results in Fig. S1A to C in the supplemental material).

In a murine abscess infection model, the Teg41Δ3′ strain displayed an attenuated virulence phenotype relative to the wild-type (WT) strain (8). Although attenuation was clearly shown for the Teg41Δ3′ strain, we did not investigate whether the reduction in virulence was exclusively due to Teg41 regulation of the αPSMs or whether Teg41 had additional roles. To investigate this, we conducted both abscess and murine systemic models of infection to compare the virulence of the Teg41Δ3′ strain with that of an ΔαPSM mutant. The rationale behind these experiments is that if Teg41 solely regulates the αPSMs, then the attenuation observed in the Teg41Δ3′ strain should be similar to or less than that observed in the ΔαPSM strain. Moreover, we decided to conduct both infection models because it has been suggested that virulence factors involved in these two models are different (12, 13). Mice were injected subcutaneously with S. aureus, and each day, during a 9-day infection period, abscess size was measured. On days 0.3 (8 h), 5, and 9, the number of bacteria present in abscesses was determined. Abscess size was significantly reduced in mice infected with the Teg41Δ3′ strain as previously observed (8) (Fig. 1A). Interestingly, although the abscess area from mice infected with the ΔαPSM strain showed a reduction compared to the WT, abscesses in this group were larger than those caused by the Teg41Δ3′ strain (Fig. 1A). While no difference in bacterial numbers was observed 8 h after infection (as expected), the number of bacteria recovered from Teg41Δ3′-infected abscesses was lower than those from WT-infected mice (as previously reported) and also lower than those from mice infected with the ΔαPSM mutant at 5 and 9 days postinfection (Fig. 1B). These results suggest that the Teg41Δ3′ strain is more attenuated during skin infection than an ΔαPSM mutant.

FIG 1.

The Teg41Δ3′ strain is more attenuated in vivo than an ΔαPSM strain. (A and B) Female C57BL/6J mice were infected with ~3.6 × 10⁷ CFU of the indicated strains, and infection was allowed to proceed for 9 days. (A) Surface lesion size was quantified over 9 days. Data represent the mean (n = 13 to 23) with standard error of the mean (SEM). Statistical significance was determined by Mann-Whitney test; **** indicates P values of <0.0001 for the Teg41Δ3′ strain versus either WT or ΔαPSM mutant. Data are combined from multiple independent experiments. (B) At 0.3 days (8 h), 5 days, and 9 days postinfection (p.i.), mice were euthanized, and tissue was homogenized to determine bacterial titers. Each symbol represents a single animal (n = 10 to 15), and the line represents the median. Data are combined from multiple independent experiments. Statistical significance was determined by Mann-Whitney test (ns, not significant; *, P < 0.05; ***, P < 0.005; ****, P < 0.001). (C to E) Groups of 16 mice were retro-orbitally injected with the indicated strains. After 7 days, mice were euthanized, and organs were harvested. Bacteria were plated, and CFU were normalized to organ mass. Bacterial loads in kidneys (C), lungs (D), and liver (E) are displayed. Results are shown as median with 95% confidence interval (CI). Statistical tests were performed as follows: for panel A, two-way analysis of variance (ANOVA) with Šídák's multiple-comparison test (****, adjusted P [P_adj] < 0.0001; ***, P_adj < 0.001; **, P_adj < 0.01; *, P_adj < 0.05); for panels B to E, Kruskal-Wallis test with Dunn’s multiple-comparison test (****, P_adj < 0.0001; ***, P_adj < 0.001; **, P_adj < 0.01; *, P_adj < 0.05).

We next performed a systemic model of infection. As a control, we included the Teg41 complement strain along with the WT, the Teg41Δ3′ strain, and the ΔαPSM mutant. We injected 6-week-old female BALB/c mice retro-orbitally with 100 μL of 10⁷ CFU and allowed infection to proceed for 7 days. The infection began with n = 8 in each group in duplicate, and during the 7-day infection, four mice infected with the ΔαPSM mutant strain died, two mice infected with WT S. aureus died, three mice infected with the Teg41 complement died, and one mouse in the Teg41Δ3′-infected group died. After this, surviving mice were euthanized, and brains, hearts, lungs, kidneys, livers, and spleens were harvested, homogenized, and plated to determine recovered CFU in distal organs.

In three of the six excised organs, there is a clear trend whereby a decreased number of Teg41Δ3′ bacteria are recovered from the organs compared to the wild type and, moreover, the ΔαPSM strain (Fig. 1C to E). In excised kidneys, there is approximately a 1-log reduction of bacterial load recovered from mice infected with the Teg41Δ3′ strain compared with mice infected with WT, Teg41 complement, and ΔαPSMs (Fig. 1C). The reduction in recovered bacteria from Teg41Δ3′-infected mice compared to ΔαPSM mutant-infected mice strongly suggests that Teg41 serves additional roles beyond αPSM regulation. We observed a similar trend in CFU recovered from the liver (Fig. 1D) and lungs (Fig. 1E), although the Teg41 complement strain failed to restore CFU levels to wild-type level in the liver. While unexpected, this is the only phenotype examined in which chromosomal complementation of Teg41 failed to completely restore wild-type levels (Fig. S1).

We found no significant difference between WT and ΔαPSM mutant strains in bacterial load recovered from any of the six harvested organs. Although somewhat unexpected, the αPSMs have been reported as key determinants of skin and soft tissue infection; therefore, it is possible they are less important for dissemination in a bacteremia model (see Discussion below) (14). There was no significant difference found in any of the groups in the excised brains (Fig. S1D), hearts (Fig. S1E), and spleens (Fig. S1F). These results clearly demonstrate not only that Teg41 is required for full virulence during murine abscess and systemic infections but also that a strain lacking Teg41 is more attenuated than a strain lacking αPSMs.

The Teg41Δ3′ strain and ΔαPSM mutant have distinct transcriptomic profiles.

The results above strongly suggest that Teg41 has more targets than just the αPSMs. To gain further insight into the differences between the strains, whole-transcriptome analysis was performed (by RNA-seq) comparing the ΔαPSM mutant to the Teg41Δ3′ strain. The rationale behind this experiment was that if the αPSMs are the only target of Teg41, then the transcriptome of an ΔαPSM mutant should look very similar (or identical) to that of the Teg41Δ3′ strain. Each strain was grown in triplicate in tryptic soy broth (TSB) for 6 h (postexponential phase), a time at which the αPSM peptides are being produced (14, 15). Cells were pelleted, total RNA was extracted, and RNA-seq was performed. A principal-component analysis (PCA) was performed to assess global dispersion and relative difference between the strains (Fig. 2A). The three Teg41Δ3′ replicates clustered separately from the three ΔαPSM mutant replicates, indicating that the two strains showed distinct transcriptomes. A differential expression analysis (DEA) was conducted to obtain the names of differently impacted genes. Our analysis showed that a total of 113 transcripts were dysregulated (fold change [FC] of >2 or <2 and P value of <0.05) with 36 transcripts more abundant (more highly expressed, up) in the ΔαPSM mutant and 77 transcripts less abundant (more lowly expressed, down) in the ΔαPSM mutant relative to the Teg41Δ3′ strain (Fig. 2B and Supplemental File S1). Of interest, several virulence factors are dysregulated, such as lukG being expressed at lower levels in the Teg41Δ3′ mutant relative to the ΔαPSM mutant, while spa (surface protein A), sasA, gehA, and hla were expressed at higher levels. This result demonstrates altered regulation of additional virulence factors in the absence of Teg41, strongly suggesting that Teg41 has additional targets in S. aureus.

FIG 2.

Comparison of the transcriptomic profiles of an ΔαPSM mutant with the Teg41Δ3′ strain. Both strains were inoculated in triplicate in TSB and grown for 6 h at 37°C. Cells were harvested, and RNAs were extracted and sequenced. (A) Principal-component analysis (PCA) comparing transcriptomes of the Teg41Δ3′ and ΔαPSM strains. The x axis displays principal component 1, explaining 43.7% of variance, and the y axis shows principal component 2, explaining 19.7% of variance. (B) Volcano plot of differential expression analysis (DEA) between ΔαPSM and Teg41Δ3′ strains. Red dots depict significant changes with a log₂ fold change of >1 and a −log₁₀ P value of >1.3 (P < 0.05) as cutoffs. Black dots represent transcripts not differentially impacted. The 5 most highly up- and downregulated genes are indicated.

Teg41 influences expression of multiple genes in S. aureus.

The previous experiments suggest that Teg41 regulates more than just the αPSMs. To further investigate potential targets of Teg41, RNA-seq was conducted from total RNA extracted from the WT and Teg41Δ3′ strains cultured for 3 h, 6 h, or 9 h, corresponding to exponential, postexponential, and stationary phases in TSB at 37°C (for growth curves of WT and the Teg41Δ3′ strain, see reference 8). For the analysis, we considered genes that displayed changes in expression with an FC of >2 or <2 and a P value of <0.05 between the WT and Teg41Δ3′ strains. Each dysregulated gene was associated with a Cluster of Orthologous Groups (COG), and differences in each categorical group were investigated over time. Several COGs were impacted upon deletion of Teg41, with the greatest impact observed at 3 h (Fig. 3A). At this time point, genes belonging to unknown functions, transport and binding proteins, and sRNAs were the most dysregulated in the Teg41Δ3′ strain relative to WT. At 9 h, genes in unknown function, energy metabolism, and purine, pyrimidine, nucleoside, and nucleotide synthesis groups were the most impacted. At 6 h, the Teg41Δ3′ mutant showed the fewest differences from the WT strain (Fig. 3A).

FIG 3.

Removal of Teg41 causes widespread gene expression changes in S. aureus. Strains were inoculated in triplicate in TSB and grown for 3 h, 6 h, and 9 h at 37°C. Cells were harvested, and RNA was extracted and sequenced. (A) Cluster of Orthologous Group (COG) analysis comparing dysregulated genes between Teg41Δ3′ strain and WT at 3 h (blue), 6 h (red), and 9 h (yellow). The number of genes in each category is represented by concentric circles moving outward from the center (each circle representing 10 genes). (B to D) Volcano plot of DEA between Teg41Δ3′ strain and WT at 3 h (B), 6 h (C), and 9 h (D). Red dots depict significant changes with a log₂ fold change of >1 and a −log₁₀ P value of >1.3 (P < 0.05) as cutoffs. Black dots represent transcripts not differentially impacted. The 5 most highly up- and downregulated genes are indicated. (E) Comparison of genes dysregulated in the Teg41Δ3′ strain at each growth phase. (F) List of the 7 genes that were dysregulated in all three RNA-seq data sets, including linear fold change values obtained in each experiment (i.e., 3 h, 6 h, and 9 h).

To further explore differences between the WT and Teg41Δ3′, differential expression analyses (DEAs) were performed at each time point to highlight individual genes with differential expression (Fig. 3B to D and Supplemental Files S2, S3, and S4). For this analysis genes were considered significantly differentially expressed if there was an FC of >2 or <2 and a P value of <0.05 (for details of additional data analysis, see Materials and Methods). At 3 h, a total of 201 genes were significantly differentially expressed, with 107 genes upregulated and 94 downregulated in the Teg41Δ3′ strain compared to the WT (Fig. 3B and Supplemental File S2). As expected, levels of the psmα transcript were decreased in the Teg41Δ3′ strain with an FC of <18, which is consistent with our previous study (8). Interestingly, 33 sRNAs were dysregulated, with 12 sRNAs showing downregulation in the absence of Teg41 and 21 sRNAs being upregulated. Furthermore, a variety of known virulence factors showed differences in expression, with spa (SAUSA300_0113) being more abundant in the Teg41Δ3′ strain (FC = 4.2) whereas a reduction of the psmβ (SAUSA300_1068) transcript (FC < 6) was observed.

At 6 h, only 78 genes showed differences in expression (Fig. 3C and Supplemental File S3). Of the significantly altered transcripts, 44 genes were upregulated and 34 genes were downregulated in comparison of the WT to the Teg41Δ3′ strain. Only 2 sRNAs showed dysregulation, i.e., sprB and sRNA073 (SAUSA300s029 and SAUSA300s211), in contrast to 33 sRNAs showing altered abundance at 3 h. Downregulation of the psmα transcript was once again observed in the Teg41Δ3′ strain, with a 4-fold reduction compared to the WT. Altered abundance of the psmβ and spa transcripts was also observed with the same expression pattern as at 3 h.

During the stationary growth phase (9 h), 147 genes were differentially expressed (Fig. 3D and Supplemental File S4), including 99 transcripts upregulated and 48 downregulated in the Teg41Δ3′ strain relative to the WT. Remarkably, the psmα and psmβ transcripts once again showed downregulation in the Teg41Δ3′ strain, whereas spa mRNA was again upregulated. Also of interest, alpha-toxin (hla) and leukocidin lukG and lukH transcripts were more highly expressed in the Teg41Δ3′ strain at 9 h.

Across all three of the different RNA-seq data sets, comparing the WT to the Teg41Δ3′ strain during different growth phases, it was evident that the removal of functional Teg41 influences expression of multiple genes. One of the key findings from these data sets is that the psmα and psmβ transcripts were constantly downregulated in the Teg41Δ3′ strain compared to the WT strain. Another key result was that the spa transcript was significantly increased in the Teg41Δ3′ strain. We compared all transcripts dysregulated throughout the different growth phases to obtain a list of genes consistently impacted at all three time points by the removal of Teg41 (Fig. 3E and F). A total of 7 genes showed consistent dysregulation: the psmα and psmβ transcripts were less abundant at each time point in the Teg41Δ3′ strain, whereas spa, SAUSA300_2453 (ABC transporter), and ddh (d-lactate dehydrogenase) were more highly expressed. Two genes, plc and esxA, had expression patterns that varied based on the growth phase (Fig. 3E). The RNA-seq analysis and different comparisons performed here strongly indicate that Teg41 regulates expression of multiple genes in S. aureus, many of which encode known virulence factors.

Teg41 controls protein production in S. aureus.

Recent dual transcriptome/proteome studies indicate that RNA synthesis does not necessarily correlate with protein production (16, 17). Moreover, one mechanism of action of sRNAs is the binding to ribosome binding sites (RBS) to prevent protein translation, and thus, such regulation cannot be observed through transcriptomic analysis alone (10, 18, 19). Consequently, we conducted a whole-proteome analysis of both the cytoplasmic and secreted fractions of the Teg41Δ3′ strain relative to the WT and Teg41 complement. Strains were grown for 6 h (postexponential phase, corresponding to initiation of αPSM production [9]), secreted proteins were trichloroacetic acid (TCA) precipitated, and cytosolic proteins were extracted from corresponding pellets. Proteins demonstrating an FC of <2 or >2 and a P value of <0.05 between the WT and Teg41Δ3′ strain and showing no difference in FC (FC between −2 and 2) between the WT and Teg41 complement were considered significantly altered in the Teg41Δ3′ strain. Analysis of the cytosolic fraction showed 17 proteins were differentially abundant, with 13 proteins less abundant and 4 more abundant in the Teg41Δ3′ strain than in the WT (Supplemental File S5). The PSMα1 and PSMα4 peptides were the most impacted proteins with an FC of −24.02 and an FC of −8.80, respectively, corroborating the RNA-seq analysis. PSMα2 and PSMα3 were not detected, consistent with previous reports showing that these peptides are produced at lower levels than are PSMα1 and PSMα4 (20, 21). The secreted protein profile exhibited more changes, with 84 proteins altered in abundance, including 12 virulence factors (Supplemental File S6). Of interest, PSMα1 and PSMα4 still showed lower levels in the Teg41Δ3′ strain (FC = −25.41 and FC = −10.21, respectively), as did PSMβ2 (FC = −3.23). Interestingly, most of the other virulence factors identified in this analysis displayed higher levels in the Teg41Δ3′ strain than in the WT, such as Sbi (FC = 2.02), FnbB (FC = 2.17), Emp (FC = 2.32), Coa (FC = 2.37), and Spa (FC = 2.74). These virulence factors are associated with binding to host cells/extracellular components and evasion of the host’s immune system (22–26). Proteins uniquely detected or absent in the Teg41Δ3′ strain were also analyzed. For this analysis we employed strict criteria, i.e., proteins must be detected in all three replicates of WT and Teg41 complement strains but not in the Teg41Δ3′ strain, or vice versa. In total, 12 proteins (10 in the secreted fraction and 2 in the cytosol) had this pattern of production. In the secreted fraction 8 proteins were uniquely detected in supernatants from the Teg41Δ3′ strain (and not in WT samples), while 2 proteins were not detected in the Teg41Δ3′ strain but were detected in the WT. Similarly, 2 proteins from the cytosol were not detected in the Teg41Δ3′ strain but were detected in the WT (including PSMβ1) (Supplementary Files S5 and S6).

Samples for the proteomic analysis above were taken from the same cultures used for the 6-h RNA-seq analysis above. When comparing the RNA-seq analysis performed at 6 h with the proteomics results, we observed a low correlation between transcript and protein levels. In total, 104 proteins displayed altered abundance in the proteomic analysis. For 96 of these, no variation in corresponding transcript level was detected, suggesting that Teg41 may posttranscriptionally regulate protein production (Fig. 4). Interestingly, only 8 proteins with altered abundance had corresponding (i.e., both increased or both decreased) changes in transcript level (Fig. 4). SpA, the αPSMs, and the βPSMs consistently showed the same trend across all analyses, with spa/SpA being more abundant and the α/βPSMs being less abundant in the Teg41Δ3′ strain than in the WT. It is possible that Teg41 interacts directly with these transcripts to regulate their expression or indirectly through dysregulation of a transcriptional regulator. These findings clearly point out that Teg41 regulates the production of multiple proteins, including virulence factors, in S. aureus.

FIG 4.

A small subset of protein targets can be explained by transcriptional changes in the Teg41Δ3′ strain. RNA-seq and proteomic data from the WT and Teg41Δ3′ strain grown for 6 h in TSB at 37°C were compared, and a list of common impacted targets upon Teg41 removal was made. Proteins found in both cytosolic and secreted fractions were combined and treated as one protein. FCs for secreted fractions were indicated for proteomic values. RGD, Arg-Gly-Asp.

Depletion of Teg41 impacts promoter activity of the PSMs.

Our previous results suggest that Teg41 regulates the abundance of α/βPSM transcripts in S. aureus. Although we previously predicted that a direct base-pairing interaction occurs between Teg41 and the psmα transcript, this interaction has not been demonstrated. Consequently, it remains unclear if Teg41 interacts directly with the psmα and psmβ transcripts to influence their abundance or if Teg41 regulates their expression indirectly at the transcriptional level through an unknown regulatory protein(s). To answer this question, we investigated the activity of the psmα and psmβ promoters in the Teg41Δ3′ strain. Regions encompassing promoter sequences were cloned into the lacZ reporter fusion vector pJB185 and transformed into the WT, Teg41Δ3′, and Teg41 complement strains. β-Galactosidase activity was assessed from 6-h cultures and normalized to total extracted proteins by the Bradford assay. Interestingly, the αPSM promoter showed decreased activity in the Teg41Δ3′ strain compared to the WT (Fig. 5A). Promoter activity was restored to WT levels in the Teg41 complement strain (Fig. 5A). A similar pattern of activity was observed for the βPSM promoter at 6 h (Fig. 5B). Since sRNAs typically function by base pairing to other RNAs, we consider it unlikely that Teg41 directly binds to the PSM promoters to alter transcription. Thus, these results strongly suggest that dysregulation of the PSMs by Teg41 is mediated indirectly through some as-yet-unknown transcriptional regulator(s).

FIG 5.

Teg41 regulates expression of the αPSMs and βPSMs at the transcriptional level. (A and B) Teg41 influences promoter activity of the αPSMs (A) and βPSMs (B). Bacterial strains were grown in TSB for 6 h at 37°C, and cells were harvested and mechanically lysed. β-Galactosidase assays were performed, values were normalized to total protein concentration, and modified Miller units were calculated. Results are shown as mean ± standard deviation (SD). Significances were computed with one-way ANOVA corrected for multiple comparisons using Tukey’s test. ***, P_adj < 0.001; **, P_adj < 0.01; *, P_adj < 0.05. (C) AgrA and MgrA are required for Teg41-induced hemolytic activity. Bacterial strains were grown overnight in TSB at 37°C. Cell-free supernatants were incubated with human blood, and hemolysis was measured at OD₅₄₃. Results are shown as mean ± SD with one-way ANOVA corrected for multiple comparisons using Tukey’s test. ****, P_adj < 0.0001; ***, P_adj < 0.001. (D) MgrA is a transcriptional activator of the αPSM promoter. Bacterial strains were grown in TSB for 8 h at 37°C, and at each hour, cells were harvested and mechanically lysed. β-Galactosidase activity was recorded as stated for panels A and B. Results are shown as mean ± SD. Significances were computed with one-way ANOVA corrected for multiple comparisons using Tukey’s test at each hour. ****, P_adj < 0.0001; ***, P_adj < 0.001; **, P_adj < 0.01; *, P_adj < 0.05. Significance for Teg41Δ3′ versus WT is shown in red, for agrA versus WT in green, and for mgrA versus WT in purple.

To investigate which regulator(s) could potentially be responsible for altered αPSM expression in the Teg41Δ3′ strain, we monitored αPSM-mediated hemolytic activity in mutants of known regulators of the αPSMs, transformed with a Teg41-overexpressing plasmid. Previously, we demonstrated that overexpressing Teg41 leads to increased hemolysis (and αPSM production) in WT S. aureus (8). We hypothesize that mutating a Teg41-dependent regulator would abrogate increased hemolysis upon Teg41 overexpression. Two well-described regulators are known to transcriptionally control αPSM production: AgrA (27), which acts as an activator, and MgrA, which has been shown to repress psmα expression (28). As described in our previous study (8), overexpressing Teg41 in the WT background increased the lysis of human erythrocytes (Fig. 5C). However, in the ΔagrA and ΔmgrA mutants, overexpressing Teg41 did not increase the hemolytic activity compared to empty-vector control strains (Fig. 5C). Thus, Teg41-mediated regulation of αPSMs appears to require functional AgrA and MgrA, although the precise mechanism of how Teg41 interacts with these two regulators is not yet clear.

Surprisingly, in the ΔmgrA mutant, we observed a decreased in hemolysis (compared to WT), suggesting that MgrA has a positive influence on αPSM production. This contradicts a previous study which described MgrA as a repressor of psmα expression (28). To further investigate this conflicting result, we assessed αPSM promoter activity in the ΔmgrA mutant. We included the ΔagrA mutant as a negative control for this experiment (Fig. 5D). In the WT background, αPSM promoter activity was low during the exponential phase of growth but increased as the bacteria entered postexponential and stationary phases. As expected, the mutation of agrA abrogated increased αPSM promoter activity, compared to the WT strain (Fig. 5D). The absence of MgrA was also shown to reduce psmα promoter activity, in agreement with the hemolysis data (Fig. 5C). Since our result, demonstrating that MgrA positively regulates psmα promoter activity, contradicts a previous study, we wanted to examine the promoter activity of an additional, known MgrA target, in our mgrA mutant strain. To do so, we examined urease promoter activity using a ureA-lacZ promoter fusion (29). Urease expression was previously shown to be repressed by MgrA (30). Results show that ureA promoter activity increases in the mgrA mutant compared to WT (Fig. S2), confirming the expected activity of a known MgrA target in this background. Collectively, our data indicate that both AgrA and MgrA are positive regulators of psmα expression and both are required for Teg41-mediated regulation of psmα expression.

Teg41 regulates protein A production posttranscriptionally.

Staphylococcal protein A was shown to be consistently overproduced in the absence of Teg41, as both RNA-seq and mass spectrometry (MS) analyses displayed more transcript and protein present in the Teg41Δ3′ strain. Teg41-mediated regulation of the PSMs (which were less abundant in the Teg41Δ3′ strain) appears to be (at least in part) transcriptional in nature, and so we next wanted to explore how Teg41 was regulating protein A (SpA) production in S. aureus. A promoter activity assay was used to investigate whether Teg41 regulation of protein A could be explained transcriptionally (as shown above for psmα and psmβ). The spa promoter was cloned into pJB185, and β-galactosidase activity was monitored in WT, Teg41Δ3′, and Teg41 complement strains grown for 3 h. Surprisingly, unlike the α/βPSM promoter activity results, no difference in spa promoter activity was observed between the WT, Teg41Δ3′, and Teg41 complement strains, strongly suggesting that Teg41 regulates protein A posttranscriptionally (Fig. 6A). One mechanism through which Teg41 could impact spa mRNA levels posttranscriptionally (and protein A production) is by reducing spa transcript stability upon binding. To explore this hypothesis, a rifampicin stability assay was conducted. mRNA pools in cells exist in equilibrium between production and degradation. Rifampicin addition prevents transcription of new RNA, allowing the degradation rate of RNA to be specifically monitored. The WT, Teg41Δ3′, and Teg41 complement strains were grown in TSB for 3 h, and aliquots of cells were harvested. Then, rifampicin was added to the cultures and subsequent samples were pelleted at 2.5 min, 5 min, and 10 min. RNA was extracted, and protein A mRNA degradation was monitored by Northern blotting (Fig. 6B and C) and by real-time quantitative PCR (RT-qPCR) (Fig. S3). The overall level of spa transcript was confirmed to be higher in the Teg41Δ3′ strain than in the WT and Teg41 complement strains (Fig. 6B and Fig. S3A, samples taken at 0 min). A densitometric analysis of spa bands in the Northern blot revealed that the level of spa mRNA (Fig. 6C) decreased faster over time in the WT and Teg41 complement strains than it did in the Teg41Δ3′ strain (Fig. 6C). This result was confirmed by qPCR following normalization of the samples to the abundance of spa transcript at t = 0 min (Fig. S3B). Together, these results strongly suggest that Teg41 regulates protein A production by reducing spa mRNA stability.

FIG 6.

Teg41 represses protein A production posttranscriptionally. (A) Teg41 does not influence protein A promoter activity. Bacteria were grown for 3 h, and the β-galactosidase activity was assessed. Results are shown as mean ± SD. (B) spa transcript levels and stability increase upon Teg41 removal. Bacteria were grown for 3 h in TSB prior to rifampicin treatment. Cells were pelleted at T₀ (before rifampicin addition), T_2.5, T₅, and T₁₀ (i.e., 2.5, 5, and 10 min after rifampicin treatment). RNA was extracted at each time point, and spa mRNA levels were monitored by Northern blotting. hup levels were also detected as a loading control. (C) Quantification of spa levels from Northern blots by densitometry. Times indicated represent the half-life of spa mRNA in each strain. Rif, rifampicin; AU, absorbance units. (D) Predicted interaction region between Teg41 (red) and spa (blue). The start codon of spa is underlined, and the predicted RBS sequence is in italic. Prediction was performed using IntaRNA (31). (E) spa transcript levels were detected by Northern blotting in WT S. aureus containing the pCN51 empty vector (1), the Teg41Δ3′ strain containing pCN51 (2), the Teg41Δ3′ strain overexpressing full-length Teg41 (3), and the Teg41Δ3′ strain overexpressing the 3′ end of Teg41 (4). For panels B and E the ladders shown are RNA molecular weight markers. Numbers shown indicate size of marker in nucleotides.

The results above show that the presence of Teg41 in the bacterial cell leads to rapid degradation of the spa transcript. This is suggestive of a mechanism whereby Teg41 binds to the spa transcript and accelerates its degradation. To investigate if a direct base-pairing interaction is possible between Teg41 and spa, which might facilitate this, the sequences of both transcripts were analyzed using IntaRNA (31). The resulting computer analysis predicts an interaction between the 3′ end of Teg41 and the spa mRNA, directly overlapping the spa RBS (−8.7 kcal/mol) (Fig. 6D). This predicted interaction is interesting because (i) our previous work demonstrated that the same region of Teg41 (i.e., the 3′ end) was both necessary and sufficient for regulation of the αPSMs and (ii) binding of a regulatory RNA to the spa RBS is similar to the mechanism of action for RNAIII-mediated regulation of spa (32). To investigate further the predicted interaction between the 3′ end of Teg41 and spa, we examined spa transcript levels in the Teg41Δ3′ strain expressing either full-length Teg41 on a plasmid or just the 3′ end of Teg41. As previously observed, spa transcript levels increased in the Teg41Δ3′ strain compared to the WT strain and complementation with full-length Teg41 reduced spa to WT levels (Fig. 6E). Interestingly, complementation with just the 3′ end of Teg41 also reduced the level of spa transcript to that observed in the WT strain. This result supports the prediction that an interaction occurs between the 3′ end of Teg41 and the spa transcript. It also mirrors our previous results showing that Teg41-mediated regulation of the αPSMs is mediated by the 3′ end of Teg41 (8).

DISCUSSION

Previous work by our group identified Teg41 as a regulator of the αPSMs. Overexpression of Teg41 led to increased αPSM production while disruption of Teg41 led to decreased αPSM production and attenuation of virulence in a murine abscess model of infection (8). The data presented in this follow-up study go on to show that removal of Teg41 also leads to severe attenuation of virulence in a murine systemic model of infection and that the attenuation observed in both abscess and systemic infection models following Teg41 removal is greater than that observed upon removal of the αPSMs. These data, along with RNA-seq and proteomic analysis, demonstrate that Teg41 is a major regulator of virulence factor production in S. aureus and that Teg41-mediated regulation occurs through different molecular mechanisms.

The first evidence indicating that Teg41 was essential to S. aureus pathogenesis was the severe reduction in virulence in both abscess and systemic infections in mice. The ΔαPSM mutant showed little to no attenuation compared to the WT strain in the systemic dissemination infection experiment. Deletion of the psmα gene has previously been demonstrated to impact mouse survival in a sepsis infection model (14); however, the infection model employed in our study has a number of key differences from previous work, most notably a 10-fold reduction in inoculum size (in addition to differences in mouse and bacterial strain and measured outcome of infection). The lack of attenuation observed in our systemic dissemination experiment for the ΔαPSM mutant could be due to inactivation of the PSM peptides by serum lipoproteins when a low inoculum is used (33, 34). Attenuation of virulence was observed for the ΔαPSM mutant in our abscess model of infection, consistent with previous reports that skin and soft tissue infection or lung infections are better suited to investigate the role of PSM peptides in virulence (14, 24, 35). The fact that the Teg41Δ3′ strain showed a significant reduction in virulence in the systemic dissemination infection (while no attenuation was observed for the ΔαPSM mutant) and that the Teg41Δ3′ strain showed greater attenuation than the ΔαPSM mutant in the abscess model strongly indicates that Teg41 regulates additional targets involved in S. aureus pathogenesis (other than the αPSMs). Transcriptomic comparisons between the Teg41Δ3′ strain and ΔαPSM mutant confirm this, highlighting the broader role of Teg41 in regulating S. aureus gene expression. The most striking demonstration of this was when the WT transcriptome was compared to that of the Teg41Δ3′ strain. We observed impacts on gene expression at all three time points tested with a total of 201 genes dysregulated at 3 h, 78 genes altered at 6 h, and 147 genes altered at 9 h. During the exponential phase, notable Teg41-regulated genes included a variety of sRNAs and virulence factors. The psmα, psmβ, sspA, plc, isdH, gehA, and clfB genes, as well as genes involved in staphyloxanthin biosynthesis, were downregulated in the Teg41Δ3′ strain, and all have been previously described as important in S. aureus virulence (36–41). Interestingly, 6 genes (esxA, esaA, essA, essC, esxB, and esaE) belonging to the type VII secretion system (T7SS) were downregulated when Teg41 was absent. T7SS contributes to virulence in different disease models and could explain why the Teg41Δ3′ strain showed attenuation in the mouse model used in the study (42). Thirty-one sRNAs displayed dysregulation upon removal of Teg41. The small RNA sbrC, which is downregulated in the Teg41Δ3′ strain, is regulated by σ^B and thought to play a role in σ^B-mediated stress response and virulence (43, 44). Teg49 (upregulated in the Teg41Δ3′ strain) is derived from the sarA transcript and is involved in regulation of virulence (45). It is still unclear exactly how Teg41 regulates the expression of these transcripts. One possibility is that the rate of RNA synthesis (i.e., transcription) is altered (discussed further below), or alternatively, Teg41 may influence the rate of RNA degradation. Many sRNAs destabilize mRNAs upon binding and are responsible for mRNA degradation by recruitment of RNases such as RNase III (46). Such a mechanism could explain the changes in some transcript levels detected in the Teg41Δ3′ strain; however, it is unlikely that Teg41 directly interacts with >100 separate transcripts.

Another well-characterized mechanism of action of sRNAs is to inhibit translation by binding to the mRNA RBS with no detectable change in mRNA concentration. To investigate if this is a potential regulation mechanism for Teg41, proteomic analysis was performed. The low correlation between proteomic and RNA-seq data sets obtained, after experiments performed on the same cultures, is suggestive of posttranscriptional (and potentially translational) regulation. Ninety-six proteins showed altered production in the Teg41Δ3′ strain, which cannot be explained at the transcriptional level. We speculate that these proteins are likely to be translationally regulated by Teg41 either directly through base-pairing mechanisms or indirectly through other translational regulators. We narrowed down 8 targets that displayed the same regulation pattern between transcriptomics and proteomics data. Among them, the αPSMs, βPSMs, and SpA constantly showed dysregulation at all growth phases investigated by RNA-seq. We hypothesize that one or more of these represent direct targets for Teg41.

In our previous study, we hypothesized that Teg41-mediated regulation of the αPSMs occurred directly via a base-pairing interaction with the psmα transcript, and a direct interaction region was predicted (8). However, the reduction in psmα promoter activity in the Teg41Δ3′ strain reported here strongly suggests that Teg41 primarily regulates αPSM production transcriptionally, indirectly through one (or multiple) transcriptional regulator(s). Although transcriptional regulation of psmα is evident in the Teg41Δ3′ strain, a direct base-paired interaction between Teg41 and the psmα transcript is still possible. When two known regulators of psmα transcription were mutated (i.e., AgrA and MgrA [27, 28]), overexpression of Teg41 did not result in increased lysis of human erythrocytes. Surprisingly, this analysis also implicated MgrA as a positive regulator of the αPSMs, contradicting a previous study which showed MgrA acting as a repressor (28). In their study, Jiang et al. (28) used S. aureus strain NCTC8325, which contains a defective copy of the rsbU gene. RsbU, a phosphatase best known for its role in regulating σ^B activity, is also involved in dephosphorylation of MgrA (47). MgrA can be found in two different forms in S. aureus, either phosphorylated by PknB or dephosphorylated by RsbU. The phosphorylation state of MgrA influences binding to its target promoters (47); therefore, differences observed in MgrA regulation of the αPSMs could be explained by the different strains used. AH1263, the USA300 wild-type strain used in our study, has an active RsbU, and therefore, the phosphorylation state of MgrA in this strain is likely different than that in NCTC8325. Interestingly, tet38, norB, and norC, three other genes regulated by MgrA (47–49), showed alteration in transcript level in the Teg41Δ3′ strain. Since both the βPSMs and αPSMs are known to be regulated by AgrA and MgrA, it seems reasonable to speculate that a similar AgrA/MgrA-mediated mechanism of regulation is responsible for the reduced transcription observed at both promoters. However, since no alteration in mRNA or protein level was detected for AgrA or MgrA, it is clear that further studies are necessary to decipher the connection between AgrA, MgrA, and Teg41.

An unexpected finding from this study is the demonstration that Teg41 is a repressor of protein A production. Transcriptional reporter fusion assays showed that no transcriptional regulation is involved in Teg41-mediated regulation of protein A. RNAIII, the effector molecule of the Agr quorum sensing system, is also known to regulate protein A expression by two distinct mechanisms, one of which is the reduction of mRNA stability upon binding of RNAIII to the RBS sequence of the spa transcript (32, 50). We hypothesize that a similar mechanism may occur with Teg41, as an increase of spa mRNA stability was observed when Teg41 was absent. Computer analysis predicts an interaction between the 3′ end of Teg41 and spa mRNA at the RBS, similar to the RNAIII mechanism of action (Fig. 6D), and our results show that the spa transcript in the Teg41Δ3′ strain could be restored to WT levels by overexpressing either the full length or the 3′ end of Teg41. These results mirror our previous discovery that the 3′ end of Teg41 is both necessary and sufficient for αPSM production (8), strongly implicating the 3′ end of Teg41 in its regulatory activity.

In conclusion, the results presented here clearly demonstrate that Teg41 is a multifunctional pleiotropic regulator in S. aureus influencing gene expression at different levels. As such, we propose that Teg41 can be classed as one of the few S. aureus sRNA regulators (e.g., RNAIII, SprX, and SSR42) that are known to regulate multiple targets and cellular processes in S. aureus (51).

MATERIALS AND METHODS

Strains, plasmids, and growth conditions.

Strains and plasmids used in this study are listed in Table 1. Strains were grown at 37°C in tryptic soy broth (TSB) with shaking at 200 rpm overnight before being synchronized in 10 mL of prewarmed TSB at 37°C for 3 h. Then, strains were inoculated at an optical density at 600 nm (OD₆₀₀) of 0.05 in 25 mL of prewarmed TSB in 250-mL conical flasks for the times indicated in the figure legends at 37°C with shaking. When indicated, induction of cadmium-responsive promoters was achieved by addition of 2.5 μM CdCl₂.

TABLE 1.

List of strains and plasmids used in this study

Strain or plasmid	Description	Reference
Strains
S. aureus
AH1263	USA300 LAC isolate cured of plasmid LAC-p03, wild type	67
JLB162	AH1263 Teg41Δ3′	8
RKC0775	JLB162 with Teg41 inserted into chromosome at the sapI site. Teg41 complement strain.	This study
RKC0753	AH1263 Δpsmα	68
RKC0951	AH1263 agrA::bursa	This study
RKC1132	AH1263 ΔmgrA	69
RKC1219	AH1263 pMK4 empty vector	This study
RKC1220	AH1263 pRKC0486	This study
RKC0782	RKC0753 pMK4 empty vector	This study
RKC1207	RKC0753 pRKC0486	This study
RKC1221	RKC1132 pMK4 empty vector	This study
RKC1222	RKC1132 pRKC0486	This study
RKC1162	RKC0951 pMK4 empty vector	This study
RKC1236	RKC0951 pRKC0486	This study
RKC1037	AH1263 pRKC1032	This study
RKC1083	JLB162 pRKC1032	This study
RKC1154	RKC1132 pRKC1032	This study
RKC1150	RKC951 pRKC1032	This study
RKC1268	AH1263 pRKC1249	This study
RKC1269	JLB162 pRKC1249	This study
RKC1270	RKC0775 pRKC1249	This study
RKC1277	AH1263 pRKC1273	This study
RKC1289	JLB162 pRKC1273	This study
RKC1290	RKC0775 pRKC1273	This study
RKC0630	AH1263 pCN51	8
RKC0631	JLB162 pCN51	8
RKC0615	JLB162 pRKC0473	8
RKC0629	JLB162 pRKC0628	8
E. coli IM08B	Cloning strain	70
Plasmids
pMK4	Gram-positive shuttle vector (Cmr)	71
pJB185	Promoterless codon-optimized lacZ reporter plasmid (Ampr Cmr)	52
pCN51	Cadmium-inducible expression vector	72
pRKC0486	pMK4_Teg41 (vector overexpressing Teg41 under its native promoter)	8
pRKC1032	pJB185 with psmα promoter controlling lacZ expression	This study
pRKC1249	pJB185 with spa promoter controlling lacZ expression	This study
pRKC1273	pJB185 with psmβ promoter controlling lacZ expression	This study
pRKC0473	pCN51_Teg41 (vector overexpressing Teg41 from an inducible promoter)	8
pRKC0628	pCN51_Teg41_3′ (vector overexpressing the 3′ end of Teg41 from an inducible promoter)	8

Construction of lacZ reporter with spa, psmα, and psmβ promoters.

A region of 508 bp encompassing the spa promoter was amplified using Q5 high-fidelity master mix (New England Biolabs [NEB]) and forward primer 5′-AAT TTG AAT TCG AAT CAA TTA TTA GCA GAT AA-3′ and reverse primer 5′-AAT TTG GAT CCA CCT AGT TTA CGA ATT GAA TA-3′. The product was double digested using BamHI/EcoRI at 37°C and ligated into plasmid pJB185 (52). pJB185 containing the psmα promoter was constructed using Q5 high-fidelity master mix (NEB) and forward primer 5′-GCG AAT TCC TAA TCT CTC GCA TAA TTG CTT ATG-3′ and reverse primer 5′-GAA GTC GAC TAA GAT TAC CTC CTT TGC TTA TGA GT-3′, digested with EcoRI and SalI, and ligated into the corresponding sites of pJB185. For the psmβ promoter, the in vivo assembly method was used to clone the region of interest into pJB185 at SalI and EcoRI sites (53). Briefly, a region of 383 bp was amplified using Q5 high-fidelity master mix (NEB) and forward primer 5′-ATG ACA TGA TTA CGA TTC TCC AGC TGA GCT ACC AGG ACA C-3′ and reverse primer 5′-ATA AAG TTA ATC AGT CGA CTG AAA ACA CTC CTT AAA ATT TAA ATT TGA AGA TAA CAA AAA CGT G-3′, and the pJB185 plasmid was linearized by reverse PCR using forward primer 5′-GTC GAC TGA TTA ACT TTA TAA GGA GGA AAA ACA TAT GTC-3′ and reverse primer 5′-GAA TTC GTA ATC ATG TCA TAG CTG-3′. PCR mixtures were treated with DpnI at 37°C before transformation into Escherichia coli IM08B and plated onto lysogeny broth (LB) agar containing ampicillin. Transformants for each construct were PCR screened using primers flanking the inserted region and cultured in LB with ampicillin. Plasmids were extracted and verified by Sanger sequencing.

S. aureus genetic manipulation.

The Teg41 complement strain (RKC0775) was constructed by amplifying a fragment containing Teg41 and its native promoter with forward primer 5′-AAAACTGCAGAGATTACCTCCTTTGCTTATGAG-3′ and reverse primer 5′-CGCGGATCCCCTACAATAGTAGATTCTGTAC-3′. The resulting fragment was digested with PstI and BamHI and ligated into the single-copy shuttle vector pJC1112 (11). pJC1112 containing Teg41 was electroporated in S. aureus containing plasmid pRN7203 (which contains the integrase gene) (10) to facilitate pJC1112 integration into the SaPI1 site.

The agrA::bursa (agrA mutant) strain was constructed via phage transduction (54) from strain NE1532, which was obtained from the Network on Antimicrobial Resistance in Staphylococcus aureus (NARSA) transposon mutant library (55).

To generate S. aureus strains carrying plasmids, AH1263 was made electrocompetent as described in reference 56 and electroporated with 1 μg of plasmids extracted from IM08B. Bacteria were plated onto tryptic soy agar (TSA) with corresponding antibiotics and incubated overnight at 37°C. Transformants were verified by PCR with primers flanking the inserted region and plasmid sequencing. Other derivative strains were obtained via phage transduction from AH1263 carrying the plasmid of interest (54).

RNA extraction.

S. aureus cultures were pelleted and washed with ice-cold phosphate-buffered saline (PBS). Pellets were stored at −80°C until processing. Isolation of RNA was performed as previously described (57). RNA was treated with Turbo DNase (Ambion) for 1 h at 37°C. RNA quantity and integrity were assessed via Bioanalyzer (Agilent 2100 Bioanalyzer), and all samples had an RNA integrity number (RIN) of >9.5. RNA samples were kept at −80°C until use.

RNA sequencing and analysis.

RNA-seq libraries were created with the Illumina stranded RNA library preparation system with RiboZero Plus rRNA depletion and sequenced at the Microbial Genomics Sequencing Center in Pittsburgh, PA, USA. Raw read data were imported and analyzed using CLC Genomics Workbench as described previously (58). Lowly expressed genes were eliminated by excluding genes in which normalized expression values in both samples were <10. Similarly, to eliminate genes with ambiguously aligned reads, we eliminated genes where <80% of the aligned reads were unique.

Samples preparation for mass spectrometry analysis.

S. aureus cultures were centrifuged at 3,000 rpm for 15 min. Supernatants were harvested and filter sterilized through a 0.45-μm filter disk to ensure that all bacterial cells were removed from the sample. The remaining cell pellets were washed with ice-cold PBS and stored at −80°C for intracellular proteome analysis. Trichloroacetic acid (TCA) was added to the culture supernatants (for a final concentration of 10%), and samples were incubated at 4°C overnight. The following day, samples were centrifuged at 11,000 rpm for 10 min and the supernatant was removed. The resulting pellets, containing precipitated proteins, were washed with ice-cold acetone. Samples were prepared by filter-assisted sample preparation (FASP). Samples were resuspended in 4% (wt/vol) SDS, 100 mM Tris (pH 7.4), 100 mM dithiothreitol (DTT), with protease inhibitor cocktail (Thermo Fisher Scientific), and clarified by centrifugation at 17,000 × g for 10 min, and protein concentration was determined by a Pierce 600-nm protein assay (Thermo Fisher Scientific). Samples were then standardized to 100 μg and reduced at 37°C for 1 h. Urea was added to a final concentration of 6 M with 20 mM Tris (pH 8.5), and samples were placed in a 30-kDa M_w protein concentrator column (Millipore Sigma). All centrifugation steps performed from this point on were performed at 12,000 × g for 3 to 5 min until the column was almost empty. Three washes were performed with 8 M urea, 20 mM Tris (pH 8.5) (urea buffer), prior to alkylation with 10 mM iodoacetamide in urea buffer and incubation in the dark at room temperature for 30 min. Washes were performed as described above, followed by three more washes with 100 mM triethylammonium bicarbonate (pH 8) (TEAB). Trypsin was added in TEAB at a 1:100 trypsin-to-protein (1 μg) ratio and incubated at 37°C for 18 h. Digested samples were eluted by centrifugation, desalted using C₁₈ columns (Waters), and resuspended in 2% acetonitrile (ACN)-0.1% formic acid.

MS and data analysis.

Digested peptides (5 μL) were separated on a 50-cm Acclaim PepMap 100 C₁₈ reversed-phase high-pressure liquid chromatography (HPLC) column (Thermo Fisher Scientific) using an Ultimate3000 ultra-HPLC (UHPLC) (Thermo Fisher Scientific) with a 60-min (in-gel digest) or 180-min (whole-proteome) gradient (2% to 32% acetonitrile with 0.1% formic acid). Peptides were analyzed on a hybrid quadrupole-Orbitrap instrument (Q Exactive Plus; Thermo Fisher Scientific) using data-dependent acquisition in which the top 10 most abundant ions were selected for tandem mass spectrometry (MS/MS) analysis. Raw files were searched against the S. aureus USA300 proteome (UniProt identifier [ID]: UP000001939) using MaxQuant (59) (https://maxquant.org/) and the integrated Andromeda search engine. Digestion was set as trypsin/P, variable modifications included oxidation (M) and acetylation (protein N-term), and carbamidomethylation (C) was fixed. Label-free quantification was used, with peptides matched between runs. Other settings were left as defaults. The resulting protein group files were further processed using Perseus (60), and for whole-proteome experiments, this included an imputation step with default settings. An unpaired t test with Welch’s correction was used to establish significant changes in protein abundance (label-free quantitation [LFQ] intensity) between strains. Proteins with a P value less than 0.05 and a fold change greater than 2 up or down were considered significant. Proteins with a P value of <0.05, a fold change strictly greater than 2 up and down, and no difference between the wild-type (RKC0520) and complement (RKC0775) strains were considered significantly altered in the Teg41Δ3′ mutant (JLB162).

Promoter activity assay.

Promoter activity assays were performed as previously described (52). S. aureus was grown at indicated time points and pelleted for 5 min at 21,000 × g. Cells were resuspended into 1.2 mL of Z-buffer (60 mM Na₂HPO₄, 60 mM NaH₂PO₄, 10 mM KCl, 1 mM MgSO₄, 50 mM β-mercaptoethanol, pH 7) and lysed with 0.1-mm-diameter glass beads for 1 min. Cellular debris was pelleted by centrifugation at 21,000 × g for 5 min, and supernatants were diluted into 600 μL of Z-buffer. Next, 150 μL of o-nitrophenyl-β-d-galactopyranoside (ONPG) buffer (0.1 M phosphate buffer, 4 mg/mL of ONPG) was added. Tubes were incubated in a water bath at 37°C, and time until reaction mixture turned slightly yellow was recorded. At this point, 200 μL of 1 M NaCO₃ was added to stop the reaction and the OD₄₂₀ of each reaction mixture was read. Total protein concentration of samples was estimated with the bicinchoninic acid (BCA) assay (Pierce, BCA protein kit; Thermo Scientific) at OD₅₄₃. The modified Miller units were calculated as follows: 1,000 × OD₄₂₀/(time recorded × volume of supernatant diluted × OD₅₄₃), and the data are reported on a per-milligram-of-protein basis.

Cell-free erythrocyte hemolysis assay.

S. aureus supernatant hemolytic activity was measured as previously described (8). Briefly, S. aureus supernatant was incubated with human blood cells in hemolysis buffer (40 mM CaCl₂, 1.7% NaCl) and incubated at 37°C in a tube revolver. Tubes were spun down at 5,000 × g for 1 min, and the hemolytic activity was determined by reading the absorbance of the samples at OD₅₄₃.

mRNA stability assay.

S. aureus cultures were treated with rifampicin (200 μg/mL), and cells were pelleted at 0 min (before addition of rifampicin), 2.5 min, 5 min, and 10 min (after addition of rifampicin). RNAs were extracted as described above, and mRNA stability was monitored by either real-time quantitative PCR (RT-qPCR) or Northern blotting.

Real-time quantitative PCR.

RNA (500 ng) was reverse transcribed using the iScript reverse transcription supermix (Bio-Rad). Obtained cDNAs were diluted 1/10 in nuclease-free water, and qPCR was performed using the iTaq universal SYBR green supermix (Bio-Rad) and primers targeting the spa transcript (forward, 5′-ATCTGGTGGCGTAACACCTG-3′; reverse, 5′-CATTGCGTTGATCAGCATTT-3′) in technical duplicates. The hup mRNA was used (forward, 5′-AGCTGGTTCAGCAGTAGATGC-3′; reverse, 5′-CCTCAAAGTTACCGAAACCAA-3′) as an endogenous control for transcript abundance (due to the fact that hup is highly expressed and highly stable and therefore it does not degrade substantially during the course of the experiment). RNA half-life was calculated according to the following equation: half-life = ln2/k_decay with k_decay the rate constant for decay, obtained from the slope of a semilogarithmic plot of normalized values of each transcript as a function of time, as outlined in our previous publication (61). The degradation of each transcript is normalized to the rate of decay of the hup mRNA. Consequently, negative values are obtained for transcripts more stable than hup and positive values are obtained for transcripts less stable than hup. All analysis was performed with RStudio v1.4.1106 (62).

Northern blotting.

RNA (5 μg/lane) was loaded onto a formaldehyde agarose gel and electrophoresed for 1 h 15 min at 120 V. RNAs were transferred onto a nylon membrane overnight by capillary transfer with 20× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) buffer and UV-cross-linked to the membrane. The presence of rRNA and ladder bands was visualized by staining the membrane with methylene blue (0.04% in 0.5 M acetate solution). To detect spa, a radiolabeled probe was made as follows: PCRs targeting spa (forward, 5′-ACGCGTCGACTTGAAAAAGAAAAACATTTATTCAATTCGTAAAC-3′; reverse, 5′-CGCGGATCCCTTTGGATAAAACCATTGCGTT-3′) and hup mRNA (forward, 5′-ATGAACAAAACAGATTTAATCAATGCAG-3′; reverse, 5′-TTATTTTACAGCATCTTTTAATGCTTTACC-3′) were performed and PCR mixtures were radiolabeled using the Roche random prime labeling kit (Sigma). Approximately 1 μg of PCR product was used with [α-³²P]ATP according to the manufacturer’s protocol. Probes were purified. using Illustra MicroSpin G-25 columns (GE Healthcare). Membranes were prehybridized overnight at 45°C in Ultrahyb-Oligo buffer and then incubated with radiolabeled probe overnight at 45°C. After incubation, membranes were washed with 2×, 1×, and 0.5× SSC buffer and visualized using a phosphorimager screen. For stripping, blots were washed twice with boiling 0.2% SDS for 20 min.

Murine systemic infection.

A systemic model of bacterial dissemination was used as previously described by Spaan et al. (63). Briefly, cultures were grown for 3 h in TSB to an OD₆₀₀ of 1. Bacterial cells were pelleted via centrifugation and resuspended in sterile PBS to prepare an inoculum of 10⁷ CFU per 100 μL. Six-week-old female BALB/c mice were injected retro-orbitally with 100 μL of bacterial culture, and the infection was allowed to proceed for 7 days. Mice were monitored over the course of infection for significant weight loss and/or clinical signs of distress. Following 7 days of infection, the mice were euthanized with CO₂ and kidneys, livers, brains, hearts, lungs, and spleens were harvested. Organs were weighed and homogenized before serial dilution and plating were conducted to quantify recovered bacteria per gram of organ. Systemic infection experiments were performed twice with n = 8 for each strain. Similar results were obtained for each experiment. Data shown are pooled from the two experiments.

Abscess infection.

Abscess infection animal studies were conducted in strict accordance with protocols approved by the University of Kansas Medical Center Institutional Animal Care and Use Committee and were conducted as previously described (64). Briefly, overnight cultures were diluted 1:100 in 10 mL TSB in 50-mL conical tubes with a loose cap and grown at 37°C with shaking at 250 rpm for 3 h. Then, bacteria were harvested, washed, and resuspended in Dulbecco’s phosphate-buffered saline (DPBS). Fifty microliters of cell suspension containing 3.2 × 10⁷ to 4.3 × 10⁷ CFU was injected subcutaneously into 8-week-old female C57BL/6J mice that had been previously depilated. Mice were photographed at time of euthanasia or every other day, and images were used to determine lesion surface area with ImageJ. At the terminal point, the lesion and ~3 mm of surrounding tissue were harvested, bisected, minced, and placed in two 2-mL Lysing Matrix H tubes (MP Biomedicals, Irvine, CA) containing Hanks’ balanced salt solution (HBSS), human serum albumin (HSA), and HEPES solution. The samples were homogenized using a FastPrep-24 5G homogenizer (MP Biomedicals) using the manufacturer’s protocol. After homogenization, the separated samples were combined, serially diluted, and plated on TSA to enumerate CFU.

Ethics statement.

Whole human blood was obtained from donors at Ohio University in agreement with the Ohio University Institutional Review Board. Six-week-old BALB/c mice were ordered from Envigo and held at the Ohio University Office of Laboratory Animal Resources. All animal work was performed by trained lab personnel and approved by the Institutional Animal Care and Use Committee.

Data availability.

The data discussed in this publication have been deposited in NCBI’s Gene Expression Omnibus (65) and are accessible through GEO Series accession number GSE203565. The mass spectrometry proteomics data have been deposited in the ProteomeXchange Consortium via the PRIDE (66) partner repository with the data set identifier PXD034014.