All bacteria strains and plasmids used in this work are listed in Table 1. Most S. aureus genetic constructs were created in HG003 strain background (Herbert et al., 2010; Sassi et al., 2014). S. aureus bacterial cultures were grown on Mueller Hinton Broth (MHB) media until OD₆₀₀ of 0.5–0.7 at 37°C with shaking. Bacteria containing plasmids with chloramphenicol or tetracycline resistance were grown in the presence of 15 μg/ml of chloramphenicol and 3 μg/ml of tetracycline, respectively.

Plasmid pOP172 was obtained from GENEWIZ. Synthetic codon optimized S. aureus YjbH gene with HA₃ tag (HA₃-YjbH) was synthesized and cloned into pUC57 vector resulting in pOP172. To create pOP173 plasmid, HA₃-YjbH fragment was KpnI/PstI digested from pOP172 and cloned into pMK4-pHU vector (Andrey et al., 2010). S. aureus strains carrying sfGFP alone or sfGFP-YjbH under control of IPTG inducible promoter were kindly provided by Claes von Wachenfeld (Lund University). The pRAB11-MazF plasmid (pAR1884), expressing mazF gene under control of anhydrotetracycline (ATc) inducible promoter was constructed by amplification of HG003 mazF gene using primers carrying Bgl2 and EcoR1 restriction sites. Amplified product was cut with Bgl2 and EcoR1 and ligated the into pRAB11 vector (Helle et al., 2011). To induce the expression of MazF, 0.2 μM of ATc was added to the cell cultures and incubated for 10–180 min at 37°C with shaking.

Cultures of 25 ml with OD₆₀₀ of 0.7 were collected and washed three times with 1 ml of Phosphate-buffered saline (PBS) buffer. Cells were lysed in the presence of 400 μl of lysis buffer 1 (LB 1) [PBS, 200 μg/ml lysostaphin, 200 μg/ml DNAse I, protease inhibitors (Roche)] for 20 min at 37°C, chilled on ice, and sonicated 10 times with 30-s cycles using Cell Disrupter B-30 (Branson). Extracts were clarified by centrifugation for 10 min at 14000 × g 4°C. Total protein concentration was measured in supernatants (SN) by the Bradford protein assay. This method permitted to obtain of about 1 mg of total protein with a concentration of about 2.5–3.0 mg/ml. Samples were mixed with Laemmli Sample Buffer (SB) and analyzed by SDS polyacrylamide gel electrophoresis (SDS-PAGE) with following Coomassie Blue staining or western blot.

The protocol for isolation of aggregated proteins from S. aureus was developed on the base of method described for yeast (Panasenko and Collart, 2012). Cells were grown on MHB until OD₆₀₀ of 0.5–0.8. Control bacterial culture was left at 37°C without treatment. Other cultures were incubated at 37°C for 30 min with 5 mM of diamide, 10% ethanol or with antibiotics [oxacillin (10–40 μg/ml), vancomycin (10–40 μg/ml), kanamycin (50–400 μg/ml), tetracycline (60 μg/ml), erythromycin (20 μg/ml)]. For heat shock (HS), cells were incubated at 53°C for 30 min. After cultures treatment, 15 OD₆₀₀ units were harvested at 4000 g for 5 min, and washed with 1 ml of PBS. Pellets were resuspended in 0.3 ml (20 μl per 1 OD unit) of lysis buffer 2 (LB2) [20 mM Na-phosphate, pH 7.6, 10 mM DTT, 1 mM EDTA, 0.1% Tween 20, 1 mM PMSF, protease inhibitor cocktail (Roche), 200 μg/ml DNAse, 200 μg/ml lysostaphin] and incubated at 37°C for 20 min. Chilled samples were sonicated 10 times with 40-s cycles, using Cell Disrupter B-30 (Branson), and centrifuged for 20 min at 200 × g at 4°C. Total protein concentration was measured in supernatants by the Bradford protein assay. Whole cell extracts were adjusted to identical protein concentration of 1.0 mg/ml. 30 μl of supernatants were boiled with 10 μl of 4× SB (total proteins). Anti-Pbp2 antibody was used as a control for the equal protein amount in whole cell extracts. Equal amount of whole cell extracts (200 μl) were centrifuged at 16000 × g for 20 min at 4°C to pellet the aggregated proteins. After removing supernatants (soluble fractions), insoluble proteins were washed twice with washing buffer (20 mM Na-phosphate, pH 7.6, 2% of NP-40, 1 mM PMSF, protease inhibitor cocktail (Roche), sonicated (10 s at duty cycle 40%), and centrifuged at 16000 × g for 20 min at 4°C. Insoluble (aggregated) proteins were boiled in 50 μl of SB. 60 μl of soluble fractions were boiled with 20 μl of 4× SB. 15 μl of samples were separated by gradient (4–12%) SDS-PAGE, and analyzed by Coomassie Blue staining or western blot.

Cells expressing either sfGFP-YjbH or only sfGFP under IPTG inducible promoter were grown on MHB with 3 μg/ml of tetracycline overnight. Cultures were diluted 1/100 in MHB with 3 μg/ml of tetracycline and with 1 mM of IPTG and grown 37°C for 5 h. Cells were incubated at 37°C for 30 min with 5 mM of diamide, or with antibiotics [oxacillin (20–40 μg/ml), tetracycline (20–60 μg/ml), kanamycin (200–400 μg/ml)]. For heat shock (HS) cells were incubated at 53°C for 30 min. Control culture was left at 37°C without treatment. 1.5 ml of cultures were chilled and centrifuged at 4°C 10000 × g for 1 min. Pellets were resuspended in 30 μl of PBS and 7 μl of cell suspension was loaded on 1% agar, and analyzed with fluorescent microscope.

RNAs were purified using RNeasy Plus Mini Kit (Qiagen) and QIAshredder Kit (Qiagen). DNA was removed using QIAGEN DNase Kit (Qiagen). TaqMan real-time quantitative polymerase chain reaction (RTqPCR) was performed with Platinum Quantitative RT-PCR ThermoScript One-Step System (Invitrogen). All RNAs were tested for the absence of DNA contaminations. Primers and MGB Double-Dye probes for trfA, rsbW, 16S RNA, and gyrB (Table 2) were designed using Primer Express software (version 1.5; Applied Biosystems), obtained from Eurogentec and used in a concentration 0.05–0.1 μM. For each pair of primers, primer efficiency was calculated and primers and probes were used in concentrations that give the same primer efficiency with housekeeping gene (16S RNA or gyrB). To quantify RTqPCR data the 2^(−ΔΔCT) method has been used (Rao et al., 2013), where fold change of target gene expression in a target (treated) samples relative to a reference (non-treated) samples was normalized to a reference gene (16S RNA or gyrB). Thus, the relative gene expression in non-treated samples was set to 1. The errors for the ΔΔCT were obtained by least square error propagation of the standard deviation for the individual RTqPCR measurements performed in triplicates.

The induction of the MazF expression by ATc was performed for 10 min. Total RNAs were purified as described above in biological duplicates. Ribosomal RNAs were depleted with RiboZero kit (Illumina). Libraries were created using the Illumina TruSeq stranded mRNA kit. 1st strand cDNAs were synthesized with random primer. Libraries were sequenced at Fasteris SA. Results of RNA-seq were normalized for sequencing depth [normalization estimated by edgeR (Robinson and Oshlack, 2010; Robinson et al., 2010)], and by the length of the gene, in kilobases and presented in RPKM (Reads Per Kilobase Million).

Anti-HA (anti-influenza hemagglutinin; Sigma) antibodies were used at the dilution 1:3000. Anti-Spx was a gift from Dorte Frees (University of Copenhagen) (Stahlhut et al., 2017) and used in dilution 1:3000. Anti-MazE and anti-MazF antibodies were a gift from Patrick Viollier (University of Geneva) and used in dilution 1:500. Anti-Pbp2 antibodies were a gift from Mariana Gomes de Pinho (ITQB NOVA, University of Lisbon) and used in dilution 1:8000.

MazE antitoxin is regulated by an upstream pathway activated by yet unknown stimuli and unknown mechanisms (Figure 1). We hypothesized that YjbH adaptor protein may be involved in stimulus sensing. In B. subtilis it was demonstrated that YjbH is aggregated upon heat shock and diamide-induced oxidative stress, and this aggregation is accompanied by an increase in level of transcriptional regulator Spx (Larsson et al., 2007; Engman and von Wachenfeldt, 2015). We first asked whether YjbH in S. aureus is also prone to aggregation and if so, in which conditions. We used cells expressing IPTG inducible sfGFP alone or sfGFP fused to YjbH (sfGFP-YjbH). After IPTG-induction of sfGFP constructs, bacterial cells were subject to treatment with heat shock, diamide and ribosome-targeting antibiotic, kanamycin, and observed with the fluorescent microscopy (Figure 2). A bacterial strain carrying sfGFP alone show a fluorescent signal distributed homogeneously in all cells before and after tested conditions (Figure 2, upper panel, GFP). In non-treated cells, expressing sfGFP-YjbH fusion, fluorescent signal was also distributed homogeneously. In contrast, distinct fluorescent foci were visible after 30 min of treatment with diamide, heat shock, and kanamycin (Figure 2, lower panel, GFP-YjbH). The same foci were detected after treatment with the cell wall antibiotic, oxacillin, and ribosome-targeting antibiotic, tetracycline (data are not shown).

We speculate that visible foci appeared after stress treatment are aggregated sfGFP-YjbH protein, as was described for B. subtilis (Engman and von Wachenfeldt, 2015). To verify this hypothesis, we developed a method to isolate insoluble proteins in S. aureus. We treated wild-type cells with different stress conditions, such as heat shock (53°C), ethanol (10%), diamide (5 mM), and different antibiotics. Whole cell extracts were adjusted to the same protein concentration. Equal amounts of total protein were separated into soluble and insoluble fractions by centrifugation. Precipitated proteins were washed twice by sonication in buffer containing 2% of non-ionic detergent NP-40 to disrupt the membranes. Proteins remained insoluble after that treatment, are likely to be aggregated proteins (Engman and von Wachenfeldt, 2015), therefore we use the term aggregates for them. Insoluble proteins were resuspended in the sample buffer containing 2% of SDS and analyzed by SDS-PAGE and Coomassie Blue staining.

No difference was observed in soluble protein fractions from non-stress or stress conditions (Figure 3A). However, significant difference was in insoluble fractions. Heat shock, ethanol and kanamycin, induced massive protein aggregation compared to non-stressed conditions, suggesting that under these conditions many proteins are altered and prone to aggregation. Oxidative stress induced by diamide, and treatment with cell wall antibiotics oxacillin and vancomycin, did not lead to notable protein aggregation.

To prove that visible foci at Figure 2 are indeed insoluble sfGFP-YjbH, we constructed a plasmid expressing HA-tagged YjbH protein expressed from a constitutive HU promoter. The plasmid was transformed into either wild-type cells or cells deleted for yjbH (ΔyjbH). HA-YjbH expression was detected in both strains (Figure 3B). Spx accumulation was observed upon yjbH deletion, as previously described in B. subtilis (Larsson et al., 2007) and in S. aureus (Renzoni et al., 2011). Spx level was decreased upon YjbH plasmid overexpression (Figure 3B). Thus, HA-tagged YjbH protein was indeed functional and compensated the yjbH deletion.

The effect of different environmental stresses on YjbH protein levels was then analyzed. Exponentially growing ΔyjbH bacteria expressing HA-YjbH were subjected to the different stresses, such as heat, ethanol, diamide and antibiotics. Total protein extracts were adjusted to the same protein concentration and fractionated. Fractions of soluble and insoluble proteins were analyzed by SDS-PAGE and western blot for the presence of YjbH and Spx.

In total protein extracts the levels of YjbH were not changed under stress to non-stress conditions (Figure 3C and Supplementary Figure S1, Total proteins). However, the distribution of YjbH in soluble and insoluble fraction was affected (Figure 3C and Supplementary Figure S1, Soluble and Insoluble proteins). In non-stressed conditions YjbH mainly remained soluble (Figure 3C, lane 1). Only a minor fraction of YjbH was insoluble. In contrast, a significant amount of YjbH was insoluble after exposure to heat shock, ethanol, diamide, and ribosome-targeting antibiotics, such as kanamycin, tetracycline, and erythromycin (Figure 3C, lanes 2–4, 7–9). At the same time, less YjbH remained in soluble fraction under these conditions. It was shown earlier, that insoluble YjbH is not membrane associated but rather an aggregate (Engman and von Wachenfeldt, 2015). Cell wall antibiotics, oxacillin and vancomycin, did not cause significant YjbH aggregation and it remained mainly soluble (Figure 3C, lanes 5–6). Taken together, these observations suggest that in S. aureus YjbH is prone to aggregation upon environmental stimuli like it does in B. subtilis. However, not all tested stress conditions induced similar YjbH aggregation. Importantly, YjbH aggregation was observe under oxidative stress caused by diamide that did not result in dramatic increase of general protein aggregation and looked similar to untreated conditions (Figure 3A, right, lanes 1 and 4). Thus, YjbH aggregation may be selective and exploited to regulate certain environmental responses.

In B. subtilis heat and diamide induced YjbH aggregation and consequent Spx stabilization was observed due to decreased proteolysis of Spx by ClpXP (Garg et al., 2009; Engman and von Wachenfeldt, 2015). To analyze if stress conditions resulting in YjbH aggregation in S. aureus leads to accumulation of Spx protein, we followed steady state Spx protein levels by western blot. We observed increased levels of Spx in case of ethanol and diamide treatment (Figure 3C, Total proteins, lanes 3 and 4), when YjbH was aggregated (Figure 3C, Insoluble proteins, lanes 3 and 4). However, upon oxacillin and vancomycin treatment we also observed increased levels of Spx (Figure 3C, Total proteins, lanes 5 and 6) while YjbH remains mainly soluble in these conditions (Figure 3C, Soluble proteins, lanes 5 and 6). At the same time, heat shock and ribosome-targeting antibiotics, that caused aggregation of YjbH, did not lead to increased levels of Spx in total extracts (Figure 3C, Total proteins, lanes 2, 7–9). Thus, not all stress conditions leading to YjbH aggregation, resulted in Spx increased levels. This observation suggests that Spx steady state levels are not exclusively modulated by YjbH.

To understand better the link between YjbH solubility and Spx levels we next analyzed the distribution of Spx between soluble and insoluble fractions during various stresses. We found that even if the total amount of Spx was not induced upon heat shock or ribosome-targeting antibiotics, the majority of protein was found in aggregates and a very low amount was detected in soluble fractions [Figure 3C, compare Soluble and Insoluble fractions (lanes 2, 7–9)]. Under diamide, oxacillin or vancomycin stresses, where increased Spx levels were observed in the total extracts, also higher amounts were found in the soluble fraction compared to the aggregated fraction [Figure 3C, compare Soluble and Insoluble fractions (lanes 4–6)]. These results demonstrate that diamide and cell wall antibiotics, oxacillin and vancomycin, increase the total levels of Spx. Spx mostly remains soluble and, thus, probably functional under these stress conditions. Heat shock and ribosome-targeting antibiotics, kanamycin, tetracycline, and erythromycin, treatments did not increase total levels of Spx, and Spx was found aggregated in probably an inactive form.

Spx is a transcriptional factor, which interacts with the alpha−subunit of the RNA polymerase and induces transcription of many genes, including trfA (Jousselin et al., 2013). In B. subtilis it was shown that oxidation of Spx molecule leads to formation of intramolecular S-S bond between Cys10 and Cys13, that modulates the activity of Spx (Antelmann and Helmann, 2011; Rojas-Tapias and Helmann, 2018). These amino acid residues are conserved in S. aureus. However, it is unknown how Spx oxidation or aggregation may affect its activity in S. aureus. To determine how functional was Spx upon stress conditions we followed its activity by analyzing trfA transcription using RTqPCR. We analyzed expression of trfA transcript in conditions that cause increased Spx protein levels, such as yjbH deletion, and probably modulate Spx activity, such as diamide. RNAs were purified from diamide treated and non-treated cells, and trfA mRNA levels were measured by RTqPCR. In non-treated conditions we observed an increase of trfA transcript levels in the yjbH deletion mutant compared to the wild-type cells (about three times) (Figure 4A). Diamide treatment increases trfA transcription in both wild-type and yjbH deleted strains. However, the difference in trfA transcription in wild type and yjbH deletion mutant on diamide was less, that in untreated conditions (4.2 and 6.2 compare to 3 and 1). Diamide treatment led to higher induction of trfA in wild-type cells compare to ΔyjbH cells (Figure 4A, blue columns 1 and 4.2 compare to orange columns 3 and 6.2). These data are expected, because deletion of yjbH results only in increase of Spx protein levels, while, diamide treatment results in both, increase of Spx protein levels and oxidation of Spx protein, leading to higher levels of active Spx and consequently, higher trfA transcription.

Next, we tested if solubility affects functional state of Spx. We analyzed expression of trfA under heat and oxidative stress, where we observed different solubility of Spx. We found, that trfA mRNA levels were decreased upon heat shock stress while increased upon oxidative stress (Figure 4B). These observations corroborate our hypothesis that Spx is aggregated and less functional after heat shock, but not after diamide treatment where Spx was found soluble.

The correlations between YjbH aggregation, protein levels of active/inactive Spx and trfA transcription after heat or oxidative stresses, predict a different end-point effect on MazE antitoxin levels and MazF toxin activity. We, therefore, analyzed YjbH, Spx and MazE protein levels after diamide or heat shock treatment at different time points. Soluble and insoluble proteins were isolated from cell extracts adjusted to the same total protein concentrations.

In total protein extracts the levels of YjbH were similar before and after both stress conditions (Figure 5, Total proteins). YjbH was mostly found in the soluble fractions before stress (time 0). After 10 min of diamide treatment YjbH aggregated and, concomitantly, Spx was stabilized (Figure 5, left panel, diamide, Spx in Total proteins). As diamide treatment induces trfA transcription through increased amounts of functional Spx (Figure 4B), we expected an increase of MazE proteolysis. Indeed, after 10 min of diamide treatment we observed a dramatic reduction of MazE protein levels suggesting an increase of trfA-dependent MazE proteolysis by ClpCP (Figure 5, left panel, diamide, MazE in Total proteins). At the same time upon diamide treatment the level of MazF was increased, compared to untreated cells (Figure 5, left panel, diamide, MazF in Total proteins).

After heat shock YjbH remained soluble for longer than upon diamide stress, but finally aggregated after 60 min of treatment (Figure 5, right panel, YjbH in Soluble and Insoluble proteins). Despite YjbH aggregation, Spx levels remained low and unchanged during all times of heat shock treatment compared to non-stressed conditions (Figure 5, right panel, heat shock, Spx in Total proteins). In agreement with inactive Spx that was found mostly aggregated after heat shock (Figure 3C), we observed a consequent reduced transcription of trfA gene (Figure 4B). A decreased trfA transcription suggests stabilization of MazE antitoxin, as reduction of TrfA will reduce MazE proteolysis. However, we did not observe stabilization of MazE proteolysis after heat shock. Levels of MazE protein were reduced compared to non-stress conditions, but not as much as after diamide treatment (Figure 5, right panel, MazE in Total proteins). These observations suggest that other potential factors must therefore be affected by heat shock and that could explain reduced MazE protein levels without an increase of TrfA.

The decreased MazE levels upon diamide or heat shock stress predict not only stabilized MazF, but also a higher MazF endoribonuclease activity compared to non-stressed condition. To investigate MazF activity we created a plasmid expressing MazF under control of inducible promoter. We transformed this plasmid in the strain deleted for mazEF TAS to model the situation when MazF is not inhibited by MazE. MazF protein was visible after 10 min of induction (Figure 6A). To investigate MazF endoribonuclease activity we analyzed mRNA levels of rsbW gene, one of the previously described MazF endoribonuclease targets (Schuster et al., 2015). We expect a higher cleavage or decreased mRNA levels of rsbW upon MazF toxin expression or conditions increasing MazF activity. The level of rsbW transcript was slightly increased in ΔmazEF cells compared to wild-type cells, and it was dramatically decreased when mazF transcription was induced in ΔmazEF cells (Figure 6B). In-depth analysis of reads covering rsbW transcript, clearly showed a reduction of reads mapping the identified MazF cleavage motif (UACAU) present in rsbW gene (Figure 6C). These results corroborate previous observations that rsbW mRNA is a target of MazF toxin (Schuster et al., 2015) and show that rsbW transcript levels are decreased upon MazF overexpression.

To analyze if diamide or heat shock stress affect MazF endoribonuclease activity, we measure levels of rsbW mRNA target by RTqPCR using a probe hybridizing exactly on the MazF cleavage motif. For control and normalization, we used gyrB gene that was previously shown to be a non-target of MazF (Fu et al., 2009). We observed two-fold and a five-fold reduction of rsbW mRNA levels upon heat shock or diamide treatment, respectively (Figure 6D).

Taken together, these results show that decreased levels of MazE antitoxin upon diamide and heat shock resulted in modulation of MazF toxin activity. The slight decrease of rsbW mRNA cleavage, despite the reduced levels of MazE upon heat shock, can be explained by residual MazE sufficient to maintain MazF inactive. In contrast, MazE levels upon diamide are absent or highly reduced thus resulting in liberated highly active MazF toxin.

To show that decreased levels of MazE protein upon diamide treatment (Figure 5), affecting MazF activity, is dependent on trfA adaptor, we analyzed rsbW levels in trfA deleted strain with or without diamide treatment (Figure 6E). As expected, the deletion of trfA resulted in increased levels of rsbW in both conditions, confirming the role of TrfA in MazE degradation and, consequently, MazF activation. We further analyzed whether the deletion of mazEF affects rsbW levels in cells treated or not with diamide. Deletion of mazEF in non-treated conditions did not dramatically change rsbW levels, being in agreement with data obtained by RNA-seq (Figure 6B). These results were expected as MazF activity is inhibited by MazE in wild type or not present in ΔmazEF mutant. However, upon diamide treatment we observed strong stabilization of rsbW in ΔmazEF mutant, that may be explained by some additional effects of diamide on rsbW expression by unknown mechanisms.