6S RNA-Dependent Susceptibility to RNA Polymerase Inhibitors

ABSTRACT

Bacterial small RNAs (sRNAs) contribute to a variety of regulatory mechanisms that modulate a wide range of pathways, including metabolism, virulence, and antibiotic resistance. We investigated the involvement of sRNAs in rifampicin resistance in the opportunistic pathogen Staphylococcus aureus. Using a competition assay with an sRNA mutant library, we identified 6S RNA as being required for protection against low concentrations of rifampicin, an RNA polymerase (RNAP) inhibitor. This effect applied to rifabutin and fidaxomicin, two other RNAP-targeting antibiotics. 6S RNA is highly conserved in bacteria, and its absence in two other major pathogens, Salmonella enterica and Clostridioides difficile, also impaired susceptibility to RNAP inhibitors. In S. aureus, 6S RNA is produced from an autonomous gene and accumulates in stationary phase. In contrast to what was reported for Escherichia coli, S. aureus 6S RNA does not appear to play a critical role in the transition from exponential to stationary phase but affects σ^B-regulated expression in prolonged stationary phase. Nevertheless, its protective effect against rifampicin is independent of alternative sigma factor σ^B activity. Our results suggest that 6S RNA helps maintain RNAP-σ^A integrity in S. aureus, which could in turn help bacteria withstand low concentrations of RNAP inhibitors.

INTRODUCTION

Staphylococcus aureus is a commensal Gram-positive bacterium but also an opportunistic pathogen responsible for diseases ranging from benign (mostly cutaneous forms) to life-threatening (visceral or osteoarticular forms) infections (reviewed in references 1 and 2). Due to the emergence of resistant strains, mainly methicillin-resistant S. aureus (MRSA) and vancomycin-intermediate S. aureus (VISA), S. aureus has become a high-priority target for the discovery of new antibiotics (3).

In standard antibiotic treatment regimens, if antistaphylococcal penicillins (i.e., penicillinase-resistant penicillins) and glycopeptides give unsatisfactory results, combination therapy with rifampicin may be considered, particularly in complicated prosthetic device-associated infections (4). Rifampicin, a rifamycin derivative, is an inhibitor of bacterial RNA polymerase (RNAP) (5–7). The molecule binds to the RNAP β-subunit in the DNA/RNA channel to prevent transcription by steric hindrance. This effect occurs during a narrow window, just after the synthesis of the first ribonucleotides; rifampicin is ineffective on transcripts once they are elongated (7).

Highly conserved among bacteria, the core RNA polymerase contains four essential subunits (two α, β, and β′) and one accessory subunit (ω) (8–10). Among Gram-positive bacteria, RNA polymerase includes two other accessory subunits, δ and ε (ε is specific to Firmicutes). These accessory subunits may enhance transcriptional specificity and recycling of RNAP. A sigma factor subunit completes the core enzyme: when present, the complex is called RNAP holoenzyme. σ factors recognize bacterial promoters and participate in adaptation to changing growth conditions (11). The σ factors are associated with specific transcriptional programs whose function and features may differ among species (11). The number of σ factors varies between species. For example, seven σ factors have been identified in Escherichia coli and Salmonella enterica (12) and 22 in the spore-forming bacterium Clostridioides difficile (13). In contrast, S. aureus possesses only four σ factors, σ^A, σ^B, σ^H, and σ^S. σ^A is the vegetative factor responsible for transcription of housekeeping genes (14), σ^B is the main alternative sigma factor contributing to stress adaptation (15–17). The last two factors are expressed only in response to specific conditions: σ^H is involved in the regulation of competence (18, 19) and σ^S in the response to miscellaneous environmental stresses (20). A number of transcriptional factors participate together with σ factors in modulating bacterial transcription (21).

Small RNAs (sRNAs) are recognized as ubiquitous elements that fine-tune gene expression at transcriptional and posttranscriptional levels (22, 23). sRNAs are well studied in Gram-negative bacteria. However, in Gram-positive bacteria, including S. aureus, their roles in virulence, metabolism, and antibiotic resistance are less well understood, although there is no doubt about their involvement in these processes (24–26). The majority of characterized sRNAs interact with mRNAs. However, some sRNAs interact directly with protein complexes. This is the case for 6S RNA, one of the first-described sRNAs, identified in Escherichia coli in 1967 (27) and sequenced in 1970 (28). In E. coli, 6S RNA binds preferentially to RNAP associated with the σ⁷⁰ factor. 6S RNA accumulates during exponential growth and reaches its maximum levels in stationary phase (29). The 6S RNA/RNAP interaction leads to inhibition of numerous E. coli σ⁷⁰-dependent promoters and consequently reorients transcription dependent on alternative sigma factors, allowing adaptation to many environmental conditions (reviewed in references 30 and 31). Although 6S RNA is conserved among bacteria (32), its role(s) and function(s) in many of them remain unknown.

We recently developed a platform to assess S. aureus sRNAs required for fitness based on an sRNA mutant library (33). Using this platform, we identified a rifampicin susceptibility phenotype associated with the lack of 6S RNA, pointing to a possible new mechanism of resistance against low rifampicin concentrations. We showed that this phenotype is restricted neither to rifampicin nor to S. aureus but extends to other RNAP inhibitors and bacterial species. Characterization of 6S RNA in S. aureus indicates its partial involvement in σ^B-dependent transcription regulation at late stationary phase, rather than during transition from exponential to stationary phase. Additional experimental evidence suggests that S. aureus 6S RNA has a role in RNAP holoenzyme cohesion.

RESULTS

Absence of 6S RNA confers increased susceptibility to rifampicin in S. aureus and S. enterica.

At the beginning of this study, we examined the possible involvement of sRNAs in processes underlying S. aureus susceptibility to antibiotics. To uncover sRNA-associated phenotypes, our laboratory previously developed a fitness assay based on competition between sRNA-tagged deletion mutants within a library that includes mutants of most S. aureus bona fide sRNAs, defined as those expressed by an autonomous gene without antisense transcription (33, 34). Deletions were designed to remove most sRNA gene sequences, leading to inactive sRNAs, in most cases keeping intact promoters and terminators. Briefly, the fitness of individual sRNA deletion mutants growing within a collection of mutants is tested by comparing their proportion in the presence or absence of different compounds. The accumulation or reduction of individual strains is identified by monitoring the tagged sequences. This method distinguishes strains showing even subtle growth differences. Three identical libraries containing 74 putative sRNA mutants and 3 control mutants were challenged with rifampicin at a sublethal concentration (6 μg L⁻¹). After 3 days of growth, one mutant was underrepresented ∼100-fold compared to the other mutants when normalized to the same libraries grown in the same medium, without rifampicin (Fig. 1). The mutant with reduced fitness due to the presence of rifampicin carried a deletion of the ssrS gene (referred to here as the ssrS^Sa strain), which encodes 6S RNA, an sRNA known to interact with RNAP, the rifampicin target (5–7).

FIG 1.

Fitness loss of the ssrS^Sa mutant in the presence of a sublethal concentration of rifampicin. (A) Scheme of fitness experiment sampling. Three libraries were cultured for 3 days in tryptic soy broth (TSB) with or without rifampicin. Cultures were diluted 1:1,000 at 24 and 48 h. After each dilution step (0, 24, and 48 h), samples were withdrawn for tag counting under both growth conditions, when the OD₆₀₀ reached 1 (samplings 1, 3, and 4) and after the first overnight growth (sampling 2), as indicated. (B) Results of the competition assay between S. aureus sRNA mutants in the presence of 6 μg L⁻¹ rifampicin. Mutant strain names are on the y axis; the x axis shows the proportion of each mutant within the population grown in the presence of rifampicin normalized to the inoculum and to the corresponding sample grown in the absence of rifampicin. For each mutant, four histograms are shown; the color code corresponds to samplings indicated in panel A. Locus 2 and 3 mutants have tag insertions in loci likely not transcribed and not expected to alter the strain fitness. Error bars represent the experimental standard deviations between the three libraries. (Inset) Enlargement of bars for four relevant sRNA mutants: ssrS^Sa, sau60, and the control strain loci 2 and 3.

We asked whether ssrS^Sa mutant susceptibility to rifampicin observed in the fitness experiment is detectable in monocultures. Serial dilutions of overnight cultures of the wild-type (WT) and ssrS^Sa strains were spotted on solid medium containing low levels of rifampicin (5 μg L⁻¹); under this condition, the ssrS^Sa mutant was 100-fold more sensitive to rifampicin than the parental strain (Fig. 2A). This susceptibility was reversed by insertion of an ssrS^Sa copy at an ectopic chromosomal locus (ΔssrS^Sa ecto-ssrS^Sa) (see Fig. S1 in the supplemental material; Fig. 2A, left panel). A longer growth lag in rifampicin-containing liquid medium was also observed for the ssrS^Sa mutant than for the wild-type or complemented strains (Fig. 2C). These two tests indicated that the rifampicin susceptibility phenotype was solely due the absence of 6S RNA. This phenotype is observed within a narrow window of rifampicin concentrations, below the MIC (12 μg L⁻¹). We conclude that 6S RNA protects S. aureus cells against sublethal concentrations of rifampicin.

FIG 2.

ssrS deletions confer a conserved rifampicin susceptibility phenotype from S. aureus to S. enterica. Three independent clones were grown overnight for each indicated strain. (A) Serial dilutions of overnight S. aureus (HG003 strain and its derivatives) cultures were spotted on BHI agar with or without 5 μg L⁻¹ rifampicin (RIF). (B) Serial dilutions of overnight S. enterica (LT2 strain and its derivatives) cultures were spotted on LB agar with or without 5 μg mL⁻¹ rifampicin. (C) Growth kinetics of S. aureus strains (HG003 and its derivatives) in BHI with or without 5 μg L⁻¹ rifampicin. OD₆₀₀ is an arbitrary value due to plate reader conditions, not representative of absorbance measurements of S. aureus in flasks. Error bars represent standard deviations from three experiments.

As 6S RNA is widely conserved in the bacterial kingdom, we examined its protective role against rifampicin in the enteric pathogen Salmonella enterica, a Gram-negative species. The ssrS gene of S. enterica (ssrS^LT2) was deleted (Fig. S2). As in S. aureus, ssrS^LT2 deletion led to a rifampicin susceptibility phenotype compared to its parental strain (Fig. 2B, left panel). However, this phenotype was only partly complemented by insertion of the ssrS^LT2 wild-type gene at a chromosomal ectopic position (Fig. 2B, left panel).

6S RNA protection against rifampicin is partially interchangeable between S. aureus and S. enterica.

Since S. aureus and S. enterica ssrS mutants show a similar rifampicin susceptibility phenotype, we investigated whether the 6S RNA genes would be functional in heterologous backgrounds. For this, gene swaps were performed, replacing (i) the native S. enterica LT2 ssrS^LT2 gene with the S. aureus ssrS^Sa homolog (S. enterica ΔssrS^LT2::ssrS^Sa) (Fig. S2) and (ii) the native S. aureus ssrS gene (ssrS^Sa) with the S. enterica ssrS^LT2 homolog (S. aureus ΔssrS^Sa::ssrS^LT2) (Fig. S1).

The ssrS^Sa gene failed to compensate the S. enterica ΔssrS^LT2 strain rifampicin susceptibility (Fig. 2B, left panel). This is possibly due to reduced synthesis of staphylococcal 6S RNA in the S. enterica background, as suggested by the results of Northern blot analysis (Fig. S3A). Interestingly, however, the reverse swap in S. aureus ΔssrS^Sa partially restored growth in rifampicin (Fig. 2A, left panel). Complementation of ssrS^Sa by ssrS from an evolutionarily distant species suggests that different 6S RNAs shield against rifampicin using similar mechanisms.

6S RNA protects RNAP against different RNAP inhibitors.

The family of RNAP inhibitors comprises molecules with different mechanisms of action. We chose two RNAP inhibitors, rifabutin and fidaxomicin, and a putative RNAP inhibitor, aureothricin, to test the impact of ssrS^Sa on drug susceptibility.

Rifabutin, a spiropiperidyl rifamycin, is a rifampicin analog (35, 36). Fidaxomicin (also known as lipiarmycin [37, 38] and tiacumicin B [39]) is a narrow-spectrum antibiotic (40) that inhibits transcription initiation by locking RNAP through an open-clamp state that prevents an efficient interaction with the promoter (41–44). Aureothricin is a member of the dithiolopyrrolone group and has broad-spectrum activity (45). However, the mechanism of action of this molecule remains unclear. For each drug, the appropriate sublethal concentrations to use were first established using S. aureus strain HG003. The ssrS^Sa mutant showed an ∼4-log-fold-greater susceptibility to rifabutin than HG003, almost entirely complemented by ssrS^Sa ectopic expression (Fig. 3A). The ssrS^Sa mutant was moderately negatively affected by fidaxomicin compared to the parental strain, with visibly smaller colonies (Fig. 3A). The increased fidaxomicin susceptibility phenotype was not fully complemented by ectopic expression of ssrS^Sa. Unlike with rifabutin and fidaxomicin, no aureothricin hypersusceptibility was associated with the absence of 6S RNA (Fig. 3A).

FIG 3.

Susceptibility to RNAP inhibitors. (A) Serial dilution of S. aureus overnight cultures plated on solid medium containing rifabutin (RB), fidaxomicin (FDX), aureothricin (AUR), or no antibiotic. The numbers 1, 2, and 3 indicate independent clones. The antibiotic concentrations used were below the MIC. (B) Serial dilutions of C. difficile overnight cultures plated on solid medium containing FDX or no antibiotic. Pictures are representative of four replicates. Thiamphenicol was added in all plates (15 μg mL⁻¹) to maintain the plasmid. p, empty vector pMTL84121; pssrS^Cd, pMTL84121-ssrS^Cd.

Fidaxomicin is mainly active against C. difficile, a major human intestinal pathogen (40). We decided to test whether ssrS deletion also impacts RNAP inhibitor susceptibility in this species. For this, we constructed a ΔssrS derivative (ΔssrS^Cd) of C. difficile 630Δerm. C. difficile ΔssrS^Cd was 1,000-fold more susceptible to fidaxomicin than its parental strain (Fig. 3B, left panel). A plasmid carrying the ssrS^Cd gene introduced in the ΔssrS^Cd strain complemented the phenotype by restoring wild-type level of susceptibility to fidaxomicin (Fig. 3B, left panel).

We conclude that ssrS-related susceptibility to antibiotics is a common feature of different RNAP inhibitors in evolutionarily distant bacterial species. The mechanism associated with this susceptibility phenotype is likely the same for different RNAP-targeted antibiotics and different species.

Growth phase-dependent expression of ssrS in S. aureus.

The expression profile of 6S RNA differs according to species (30). We performed Northern blotting experiments to evaluate 6S RNA expression in S. aureus (Fig. 4A). 6S RNA was strongly expressed and accumulated to 20-fold-higher levels in stationary phase, as determined in S. aureus HG003. The expression profile of S. aureus 6S RNA was similar to that reported in Salmonella (46), E. coli (29), and Bacillus subtilis 6S-1 RNA, which carries a second 6S RNA (47–50).

FIG 4.

ssrS gene expression and 6S RNA sequence in S. aureus. (A) ssrS^Sa expression. Cultures of HG003 grown in BHI were sampled at OD₆₀₀ of 1, 4, and 7 and ON (20 to 24 h incubation). A Northern blot probing for 6S RNA and transfer-messenger RNA (tmRNA) (for normalization) was performed. A quantification of 6S RNA normalized to tmRNA is presented. The standard deviation is based on biological triplicates. (B) Identification of 6S RNA ends by 5′-3′ RACE mapping. Sequences were analyzed separately at different time points (OD of 7, ON [20 to 24 h incubation], and day 4 [D4]) and compiled (mix). Colored letters represent extremities found in analyzed sequences. A color scale indicates the frequency at each 5′ or 3′ end. The highest frequencies are indicated below the corresponding nucleotides.

In overnight (ON) samples where 6S RNA is the most abundant, a second, faster-migrating band was also observed (Fig. 4A and Fig. S3A). A second band was reported in C. difficile even during exponential phase (51). To determine a potential alternative 6S RNA form in S. aureus, also previously suggested (52), we performed 5′-3′ RACE (rapid amplification of cDNA ends) on samples collected at different time points during growth in rich medium: at an optical density at 600 nm (OD₆₀₀) of 7 (corresponding to entry into stationary phase), ON (i.e., after the first overnight growth), and on day 4. At all sampling points, the major transcription start site (TSS) is the same (Fig. 4B) and in agreement with the site determined by global TSS mapping (53). Concerning the 3′ end, the longest form ending with a T is the most abundant in OD 7 samples, representing 28.6% of analyzed sequences. These data confirm that the size of the longest, most abundant form is 231 nucleotides (nt) (predicted at 230 nt [52]). Samples from overnight or day 4 cultures exhibited shorter forms, which may result from processing or degradation by 3′ exonucleases.

Moderate impact of 6S RNA on the global S. aureus transcription profile.

The role of 6S RNA in transcriptional regulation was suggested early (29) and then validated by transcriptomic analysis in E. coli (54–56); in transcriptome sequencing (RNA-seq) data, 35 genes were at least 2-fold differentially expressed in a 6S RNA-deficient strain compared to the parental strain at the onset of stationary phase. To determine whether S. aureus 6S RNA could play a similar role, the transcriptional profile of the ΔssrS^Sa mutant was compared with that of its parental strain by RNA-seq on samples collected at an OD₆₀₀ of 7, which corresponds to the entry into stationary phase of S. aureus (Table 1). Transcriptome analyses were performed on biological triplicates, and features with a P value of <0.05 were retained for interpretation. Surprisingly, the transcriptional profiles of parental and ΔssrS^Sa strains were highly similar. Only three genes were >2-fold downregulated in ΔssrS^Sa (fold change [FC] < 0.5) (Table 1). They encode a hypothetical epoxyqueuosine reductase (QueH/SAOUHSC_02911), a hemin transporter (HrtA/SAOUHSC_02640; its cofunctional partner HrtB/SAOUHSC_02641 is also downregulated), and a 30-amino-acid peptide (SAOUHSC_01817) of unknown function. Other genes related to transporters, cell wall metabolism, and redox state were also significantly reduced but with a lower fold change. All of these genes are regulated by σ^A except bstA, a σ^B DNA-damage-induced gene encoding a putative DinB superfamily protein. Taken together, these results suggest that 6S RNA does not redirect transcription during the stationary-phase transition in a sigma-dependent manner.

TABLE 1.

Transcriptomic analysis of ΔssrS^Sa versus HG003 in late exponential phase^a

Locus tag/gene name	FC	P adj	Function or relevant datab	Classification	Regulation	TUc
SAOUHSC_02911	0.48	2.2E−28	Epoxyqueuosine reductase	tRNA modification	σA
SAOUHSC_02640/hrtA	0.48	4.2E−10	Hemin efflux ATP-binding protein HrtA	Transport and binding	σA, HssR	a
SAOUHSC_01817	0.49	5.9E−10	Integral component of membrane		σA	b
SAOUHSC_01736	0.52	1.0E−08	Unknown; downstream ssrS (34)		σA	c
SAOUHSC_02641/hrtB	0.53	3.7E−08	Hemin efflux system permease protein HrtB	Transport and binding	σA, HssR	a
SAOUHSC_01818/ald2	0.55	8.0E−07	Alanine dehydrogenase	Energy metabolism	σA, CcpA	b
SAOUHSC_00874	0.56	3.5E−10	Thioredoxin-like protein	Hypothetical protein	σA
SAOUHSC_02297	0.57	3.7E−14	S1 RNA-binding domain-containing protein	Protein synthesis	σA	d
SAOUHSC_02590	0.60	3.7E−09	Amino acid permease	Transport and binding	σA, CcpA, CodY
SAOUHSC_02296	0.62	1.7E−08	SprT-like protein		σA	d
SAOUHSC_00561/vraX	0.64	4.2E−10	VraX		σA
SAOUHSC_00704	0.64	2.0E−08	ABC-2 transporter		σA
SAOUHSC_03028/bstA	0.64	1.2E−03	DinB-like protein		σB
SAOUHSC_01735/tcdA	0.65	1.9E−06	ThiF domain-containing protein	Cofactor biosynthesis	σA	c
SAOUHSC_02656	0.65	4.0E−10	Cytochrome c oxidase-like protein		σA
SAOUHSC_00157/murQ	1.51	1.6E−03	N-Acetylmuramic acid-6-phosphate etherase	Cell envelope	σA, MurR, CcpA	e
SAOUHSC_00156/mupG	1.52	1.2E−03	6-Phospho-N-acetylmuramidase	Cell envelope	σA, MurR, CcpA	e
SAOUHSC_01121/hla	1.56	1.6E−05	Alpha-hemolysin	Virulence/toxin	σA, SaeR, CcpA, RNA III
SAOUHSC_02169/chp	1.60	1.8E−04	CHIPS	Virulence	σA
SAOUHSC_00961/comK1	1.85	4.3E−08	Competence protein		σA, CodY

^aFold change (FC) represents the gene expression ratio between ΔssrS^Sa and its parental strain at an OD₆₀₀ of 7. The top portion of the table contains genes with FC of <0.66, and the bottom shows genes with FC of >1.5. FC and adjusted P value (P_adj) were determined using the DESeq2 method.

^bHypothetical proteins are in italics. CHIPS, chemotaxis inhibitory protein of S. aureus.

^cTU, transcription unit.

Transcriptome results, obtained early in the stationary phase, did not provide evidence linking σ^B transcriptional activity to the presence of 6S RNA. We used a reporter fusion strategy to pursue this question: the gene encoding the fluorescent protein mAmetrine was placed under the transcriptional control of the σ^B-regulated SAOUHSC_00624 promoter (57) (pPsigB-mAmetrine) (Fig. 5A). The HG002 strain is an HG003-isogenic strain containing an 11-bp deletion in the rsbU gene, which encodes a σ^B activator (58), used here as a negative control. No significant difference in fluorescence was observed between ssrS^Sa and parental strains when fluorescence was measured for the first 18 h of growth in rich liquid medium. Thus, in keeping with transcriptomic findings, we conclude that 6S RNA does not appear to redirect transcription during transition from exponential to stationary phase in S. aureus, which differs from what was reported for E. coli (56, 59). Interestingly, however, after 18 h of culture, mAmetrine expression continued to increase in the parental strain HG003, compared to markedly lower expression in the ΔssrS^Sa strain. This result, suggesting that 6S RNA could be important for efficient σ^B–dependent gene expression during starvation, remains to be investigated.

FIG 5.

6S RNA and σ^B interplay in late stationary phase and in rifampicin response in S. aureus. (A) Fluorescence and OD₆₀₀ were monitored simultaneously in three strains (HG003 [parental], HG003 ΔssrS^Sa, and HG002) expressing a fluorescent protein (mAmetrine) under the control of the σ^B promoter of SAOUHSC_00624 from a plasmid (pPsigB-mAmetrine). HG002 (rsbU strain equivalent to the σ^B strain) is a negative control. Error bars represent standard deviations for biological triplicates. (B) Spot test comparing HG003 and HG002 (parental and ΔssrS^Sa strains, respectively) with a sublethal concentration of rifampicin (3.13 μg L⁻¹). Arbitrary values are shown as OD₆₀₀. Experiment was done with independent duplicates.

Since the absence of 6S RNA in S. aureus leads to an increased susceptibility to rifampicin, we questioned if this phenotype was related to σ^B regulation. HG002 is deficient in σ^B activity, illustrated by the absence of yellow pigmentation (60). We first noticed a greater susceptibility to rifampicin in HG002 than HG003 (Fig. 5). This observation indicates that lower σ^B activity per se confers increased rifampicin susceptibility, as described for a C. difficile sigB mutant (61). This effect is probably due to a stress adaptation deficiency related to the absence of σ^B regulation. To determine the effect of 6S RNA in this genetic context, the ssrS^Sa deletion was introduced into HG002. The resulting strain (HG002 ΔssrS^Sa) was considerably more susceptible to rifampicin than the parental strain HG002 (Fig. 5B). This observation indicates that the absence of 6S RNA leads to increased rifampicin susceptibility through a pathway that is independent of σ^B activity.

6S RNA plays a role in RNAP stability in S. aureus.

6S RNA binds to RNAP-σ⁷⁰ in E. coli (29, 47). As RNAP holoenzyme is a protein complex with accessory subunits (especially σ factors), an element binding to this complex could directly influence its stability or composition. We first performed an electrophoretic mobility shift assay (EMSA) in S. aureus with radiolabeled 6S RNA (6S [³²P]RNA), purified sigma factors (σ^A-His and σ^B-His), and RNAP (with His-tagged RpoC) to assess the interaction with 6S RNA (Fig. 6A). A ³²P-labeled unrelated sRNA, SprB (62), was used as a control. No interactions with SprB were detected. In contrast, our results showed interaction between 6S RNA and RNAP coupled to the vegetative sigma factor, σ^A, and to a lesser extent between 6S RNA and RNAP-σ^B.

FIG 6.

6S RNA and RNAP holoenzyme interactions in S. aureus. (A) EMSA with 6S [³²P]RNA (6S RNA), RNAP, σ^A, and σ^B. All the proteins were His tagged and purified. [³²P]SprB (SprB) is a control RNA. (B) Immunodetection of σ^A performed by Western blotting of samplings at OD₆₀₀ of 1 and 7, ON, and at day 3 (D3). Quantification of σ^A is relative to the amount of RNAP β/β′ subunits. Experiments were carried out in biological triplicates and analyzed by one-way ANOVA and Tukey’s HSD test [F_OD1(2,6) = 1.54, P = 0.288; F_OD7(2,6) = 5.21, P = 0.049, T_OD7(adjusted P value [P_adj] = 0.045, 95% confidence interval (CI) = 0.026 to 2.00); F_ON(2,6) = 4.50, P = 0.064; F_D3(2,6) = 8.91, P = 0.016, T_D3(P_adj = 0.013, 95% CI = −0.871 to −0.138)]. Significant differences of σ^A/(β/β′) means between strains (P < 0.05) are indicated by a star. W, wild type (HG003); Δ, ssrS^Sa mutant; e, ΔssrS^Sa ecto-ssrS^Sa. (C) Growth curves of HG003, ssrS^Sa mutant (ΔssrS^Sa) and complemented (ΔssrS^Sa ecto-ssrS^Sa) strains in BHI. Strains were cultured in independent triplicates from ON, 2-day, or 3-day precultures. Error bars represent standard deviations.

We questioned if the absence of 6S RNA could alter RNAP holoenzyme composition. The amounts of σ^A and β/β′ subunits were evaluated in ssrS^Sa and parental strain cultures at different time points. Western blots were performed with antibodies raised against σ^A and RNAP (Fig. 6B). Interestingly, σ^A pools were lower in the ssrS^Sa mutant than the parental strain and complemented strains under all tested conditions. At day 3, σ^A pools were significantly decreased by nearly 2-fold in the ssrS^Sa mutant compared to the parental strain. These results suggest that 6S RNA plays a role in RNAP holoenzyme stability and could act as a protective belt for RNAP-σ^A. We hypothesized that a reduced amount of σ^A could modify strain outgrowth. Levels of growth of HG003 (parental strain), HG003 ΔssrS^Sa and its complemented mutant HG003 ΔssrS^Sa ecto-ssrS^Sa from precultures that had grown ON, for 2 days, and for 3 days were compared (Fig. 6C). Surprisingly, no growth difference was observed between the three strains in BHI, regardless of the preculture age. Despite the significant effect on σ^A levels, 6S RNA is not an essential factor for S. aureus growth in rich medium.

DISCUSSION

Here, we demonstrated that the absence of 6S RNA in S. aureus leads to a fitness loss in the presence of low rifampicin concentrations. This marked phenotype was associated with only one sRNA gene (ssrS^Sa) of 77 tested mutants in a competition experiment. This phenotype is conserved from Gram-positive to Gram-negative bacteria, suggesting a common protective effect.

In S. aureus, the rifampicin susceptibility phenotype was fully restored by ectopic gene complementation, indicating that it was solely due to the absence of 6S RNA. In S. enterica, however, similarly done complementation of the ssrS^LT2 deletion was only partial, while the native and ectopic copies had similar expression levels (Fig. S3A). In E. coli, the ssrS and ygfA genes are in an operon (48, 63) and mature 6S RNA results from processing of 5′ and 3′ transcript ends (64, 65); similar organization and regulation are expected in S. enterica. Two hypotheses may explain the incomplete complementation of the ΔssrS^LT2 rifampicin susceptibility phenotype: (i) the ectopic ssrS^LT2 copy could be subjected to a slightly different processing pathway (not detected in the gel in Fig. S3A) and (ii) ΔssrS^LT2 could affect ygfA expression (however, no growth defect has been observed for the mutant so far).

Despite weak similarity between S. aureus ssrS^Sa and the cognate S. enterica ssrS^LT2 gene, we observed partial complementation of the rifampicin susceptibility phenotype in S. aureus ΔssrS^Sa by the ΔssrS^LT2 allele. The lack of the reverse complementation (ssrS^Sa into S. enterica ΔssrS^LT2) might be ascribable to the lower expression of ssrS^Sa in S. enterica (Fig. S3A) and/or to any of the hypotheses raised above for the S. enterica ΔssrS^LT2 ecto-ssrS^LT2 phenotype.

Susceptibility of the ssrS mutants was not observed for all the compounds tested. Differences in the mechanisms of action, binding sites, and drug entry efficiencies could explain this observation (42, 66). The ΔssrS^Sa mutant showed increased susceptibility of S. aureus to rifampicin, rifabutin, and fidaxomicin. Similarly, cognate ΔssrS S. enterica and C. difficile mutants showed marked sensitivities to rifampicin and fidaxomicin, respectively. These drugs bind RNAP close to sites interacting with DNA, suggesting that 6S RNA interaction with the enzyme may, at least partially, prevent antibiotic access to their sites. Based on this reasoning, our results suggest that aureothricin, for which toxicity was unaffected by ssrS^Sa deletion, does not bind RNAP at the interface with DNA.

Our findings suggest differences in the regulatory roles of 6S RNA in S. aureus compared to those reported for E. coli (54). In the latter species, 6S RNA interaction with the RNAP holoenzyme is proposed to coordinate transcriptional regulation with growth (29). Accordingly, numerous transcriptome analyses performed under different conditions indicated that in E. coli, many 6S RNA-regulated genes were related to translational/transcriptional machinery or amino acid metabolism (29, 54–56, 67–70). In contrast, our S. aureus transcriptomic analysis and promoter assay revealed no obvious 6S RNA-related differences in expression during the transition to stationary phase. Two major features of S. aureus could explain this phenomenon. The first is lower diversity of sigma factors in S. aureus, which has only four σ factors, among which σ^A and σ^B control the majority of transcribed genes. Given that the main alternative sigma factor σ^B is involved in stress response and not only in stationary-phase adaptation, σ^B promoters could be less sensitive to 6S RNA during the transition phase. The second feature is the compensatory effect of a coregulator. The levels of the alarmone ppGpp increase in E. coli ssrS mutants and might compensate for the lack of 6S RNA (55, 56, 59). This possibility provides an attractive explanation for the phenomenon in S. aureus, as ppGpp is synthesized in response to nutrient starvation and drives growth adaptation (71). Further experiments are needed to explore this pathway in S. aureus.

Our promoter assay (Fig. 5A) suggests a 6S RNA-dependent expression of σ^B-promoters in late stationary phase, after 18 h of culture. Among its known roles in E. coli, 6S RNA also influences transcription during long-term starvation (54, 67). Whether S. aureus 6S RNA interacts with σ^B for alternative promoter expression during late stationary phase remains unclear. In particular, in the absence of a functional σ^B, the absence of 6S RNA still generates rifampicin inhibition, indicating that this phenotype was not due to a lack of reprograming transcription from σ^A to σ^B. The relationship between 6S RNA and σ^B remains to be characterized.

We showed that S. aureus 6S RNA interacts directly with RNAP-σ^A, raising the question of whether this could directly affect the holoenzyme stability. Of note, the Δsau60 mutant exhibits a moderate reduction in fitness in the presence of rifampicin (Fig. 1B). Δsau60 corresponds to a deletion within the intergenic sequence upstream of rpoB encoding the β subunit of RNA polymerase; this deletion may alter the ratio of RNAP subunits and possibly the RNAP stability, leading to a rifampicin susceptibility phenotype. However, this attenuated phenotype was not detected by a spot test.

In E. coli, the majority of 6S RNA is coupled to RNAP-σ⁷⁰ (29). In Streptococcus pneumoniae, 6S RNA bound to RNAP was recently proposed to be a stockpile for inactive RNAP (72). In S. aureus, the absence of 6S RNA leads to a reduced amount of σ^A in prolonged stationary-phase cultures. A similar effect was observed with σ⁷⁰ in E. coli (29) and in the soluble sigma fraction of Synechocystis sp. (73). In S. aureus, σ^A is unstable (74); our results indicate that it is probably stabilized by RNAP core enzyme and 6S RNA. Knowing that σ^A is the vegetative sigma factor in S. aureus, decreased levels in the ΔssrS^Sa strain could have a negative impact on growth, and particularly on outgrowth recovery. In comparison, in B. subtilis, which expresses two different 6S RNAs, outgrowth is delayed in cells lacking 6S-1 RNA (75), whereas no extended lag phase was noticed in E. coli 6S RNA-deficient cells (29). Similar to E. coli (29), no lag linked to ssrS was observed during S. aureus outgrowth from stationary phase, suggesting that reduced σ^A pools in the ssrS^Sa mutant are enough to manage growth restart and that 6S RNA is not essential for growth in rich medium.

RNAP inhibitors remain in use in combination therapies against difficult-to-treat infections (4). Antibiotic concentrations below the MIC are encountered by bacteria under many environmental conditions, including hosts undergoing antimicrobial treatments (76). We demonstrated that 6S RNA provides protection against low concentrations of RNAP inhibitors. 6S RNA is highly conserved, and the effects of ssrS deletion on RNAP inhibitor susceptibility were observed in unrelated pathogens. 6S RNA may significantly enhanced fitness to RNAP inhibitors under these conditions. Our studies indicate the importance of 6S RNA in stabilizing RNAP interactions with σ^A and suggest that it plays its main roles in prolonged stationary phase. Our findings give insight into the mode of action of 6S RNA in an important pathogen and suggest the need to develop strategies that prevent low-level rifampicin from persisting in the antibiotic-treated host.

This protective effect is possibly due to steric hindrance, as the presence of 6S RNA would reduce the accessibility of the RNAP to its inhibitors. A second nonexclusive proposal is that the destabilization of σ^A associated with the absence of 6S RNA affects the transcriptional program to adapt to low concentrations of RNAP inhibitors. Note that in S. aureus, this shielding effect is not associated with the sigma stress factor σ^B.

MATERIALS AND METHODS

Bacterial strains and culture.

All strains used in this study and their genotypes are listed in Table S1 in the supplemental material. Strains were cultured at 37°C, with 180-rpm agitation for liquid cultures except for C. difficile. The latter was cultured under anaerobic conditions (5% H₂, 5% CO₂, 90% N₂) with 7.5 μg mL⁻¹ (precultures) or 15 μg mL⁻¹ (plates) thiamphenicol for plasmid selection. E. coli and S. enterica (serovar Typhimurium) strains were cultured in lysogeny broth (LB), S. aureus strains in brain heart infusion (BHI) or tryptic soy broth (TSB), and C. difficile in BHI. When necessary, media were supplemented with antibiotics.

S. aureus mutants were constructed in the HG003 or HG002 background (58) by allelic exchange using pIMAY (77) derivatives (except strains for RNAP purification), as described elsewhere (33). Plasmids used in this study are described in Table S2. Most plasmids constructed for this study were obtained by Gibson assembly (78), using primers listed in Table S3, and cloned in E. coli IM08B (79). An ectopically complemented S. aureus mutant was obtained from the ssrS^Sa mutant by a 2-step crossover recombination at locus 2 using a pIMAY derivative as described above. Genetic features of ΔssrS^Sa, ΔssrS^Sa ecto-ssrS^Sa and ΔssrS^Sa::ssrS^LT2 S. aureus mutants used in this study are described in Fig. S1.

To purify S. aureus σ factors, sigA and sigB from HG003 were PCR amplified using primers F-SigA/R-SigA-His and F-SigB/R-SigB-His, respectively, and cloned into the NdeI/XhoI restriction sites of the pET-21C vector. The resulting plasmids pET-21C-sigA and pET-21C-sigB were transformed into E. coli strain BL21(DE3) pLysS, leading to strains producing σ^A and σ^B that were His₆ tagged in their C-terminal portions upon IPTG (isopropyl-β-d-thiogalactopyranoside) induction.

A HG003 strain expressing a chromosomally encoded His-tagged RpoC for the purification of the RNAP core enzyme was constructed as followed. The recombinational transfer of the histidine sequence into rpoC gene was achieved by two-step PCR. A sequence encoding 10 histidines was added upstream from the termination codon, in frame with RpoC (β′ subunit of RNA polymerase). The rpoC-his fragment was generated by long-flanking homology PCR using the primers listed in Table S3 and cloned between the BamHI and PstI restriction sites of temperature-sensitive pBT2 vector (80) to obtain pBT2-rpoC-His plasmid. The resulting plasmid was electroporated into S. aureus RN4220 and then transferred to HG003 strain. The gene encoding His-tagged RpoC protein was integrated into the S. aureus HG003 chromosome by double-crossover recombination as described elsewhere (81) to obtain the HG003 rpoC-his strain.

S. enterica serovar Typhimurium mutants were constructed in the background of strain LT2-derived MA7455 (82) using λRed recombineering (83). The ΔssrS^LT2 ecto-ssrS^LT2 strain (ectopic complementation of the ΔssrS^LT2 mutation) was constructed by inserting an ssrS^LT2 copy fused to a chloramphenicol resistance cassette at the neutral chiPQ locus in ssrS^LT2 mutant as described above. See Fig. S2 for construction.

C. difficile mutants were constructed in the 630Δerm background (84). The knockout mutant was obtained using an allelic chromosomic exchange following the published method (85) with the primers CM57/CM58 and CM59/CM60 and pMSR vector pDIA7052. To complement the ssrS^cd deletion mutant, the ssrS^Cd sequence and its promoter region were PCR amplified using the primer pair CM77/CM78 and cloned into pMTL84121 to produce pDIA7065. Two E. coli strains were used as intermediates: NEB 10-beta for plasmid construction (Table S2) and HB101 RP4 for conjugation.

MIC determination.

Antimicrobial susceptibility testing by broth microdilution for rifampicin, rifabutin, fidaxomicin, and aureothricin MIC determination in S. aureus was performed as described elsewhere (86). The S. enterica serovar Typhimurium rifampicin MIC was determined as described elsewhere (87).

Fitness experiment.

The fitness experiment was performed as described elsewhere (33), with three independent sRNA-tagged mutant libraries grown simultaneously in TSB with or without 6 μg L⁻¹ rifampicin and sampled at four points: at an OD₆₀₀ of 1, ON, at an OD₆₀₀ of 1 after the first dilution, and at an OD₆₀₀ of 1 after the second dilution. All the mutants were tag sequenced with an adapted Illumina protocol. The amount of each mutant was normalized to the total amount of bacteria with and without rifampicin and to the inoculum (Fig. 1A). In total, three mutants [locus1, rsaD(tag26), and teg146 mutants] were discarded from the analysis because they were under-represented in the assembled library. Concerning sprD and sau5949, only two values were taken into account in the third dilution sampling.

Spot test.

Overnight cultures were 10-fold serially diluted in NaCl 0.9% (S. aureus and S. enterica) or BHI (C. difficile) until a dilution of 10⁻⁸ was reached and spotted on agar plates containing different sublethal antibiotic concentrations, namely, less than the MIC (rifampicin, <12 μg L⁻¹ for S. aureus and <12 μg mL⁻¹ for S. enterica; rifabutin, <15.6 μg L⁻¹; aureothricin, <6.25 μg mL⁻¹; fidaxomicin, <4 mg L⁻¹ for S. aureus and <30 μg L⁻¹ for C. difficile). Pictures were taken after ON growth or 24 h for C. difficile.

Growth curves.

S. aureus strains were cultured in microplates from ON triplicate cultures diluted 1/1,000 in BHI with or without 5 μg L⁻¹ rifampicin. Two- and three-day cultures were also used as precultures for Fig. 6C. Absorbance at 600 nm (OD₆₀₀) was measured over time with a plate reader (Clariostar).

Fluorescence measurement.

mAmetrine expression (excitation wavelength [λ_ex] = 425 ± 15 nm; emission wavelength [λ_em] = 525 ± 15 nm) was monitored over time in microplates by a plate reader (Clariostar), simultaneously with absorbance measurement of overnight triplicates cultures diluted 1:1,000 in TSB to limit autofluorescence.

RNA extraction.

Strains were cultured until the desired OD₆₀₀ was reached. After centrifugation, pellets were frozen in dry ice-ethanol. RNAs were then extracted by phenol-chloroform treatment as described elsewhere (88). When necessary, RNAs were incubated with Turbo DNase treatment (Thermo Fisher Scientific) prior to a second phenol-chloroform extraction.

Northern blotting.

Ten micrograms of total RNAs per well was separated on 1.3% agarose or 10% Tris-borate-EDTA (TBE)–urea polyacrylamide gels (Criterion Precast gels) as described elsewhere (89). For polyacrylamide gels, electrophoresis in 1× TBE was followed by transfer to Hybond-N⁺ membranes in 0.5× TBE using a TE70 ECL semidry transfer unit (Amersham Pharmacia Biotech). Probes (Table S3) were [α-³²P]dCTP labeled.

RNA-seq and transcriptomic analysis.

RNAs (DNA free) extracted from triplicate cultures sampled at an OD₆₀₀ of 7 were sequenced using a NextSeq 500/550 high-output kit v2 (75 cycles). Sequences were aligned to the reference genome (CP000253, NCTC8325) with the Bowtie2 tool and quantified with the Feature Counts program. Differential gene expression analysis was performed using the DESeq2 algorithm (90).

5′-3′ RACE.

5′ and 3′ ends were determined using the circularization method described in reference 51. Ten micrograms of total RNA of the wild-type strain (HG003) was extracted from samplings at an OD₆₀₀ of 7, ON, and on day 4. Primers used to amplify the 5′-3′ junction with Phusion high-fidelity DNA polymerase (Thermo Fisher Scientific) are listed in Table S3. A CloneJET PCR cloning kit (Thermo Fisher Scientific) was used to clone the final PCR products.

Purification of σ^A, σ^B, and RNAP core enzyme.

For σ purification, E. coli strains BL21(DE3) pET-21C-sigA and BL21(DE3) pET-21C-sigB were grown in 1 L of LB broth at 37°C to an OD₆₀₀ of 0.5. After induction with 1 mM isopropyl-1-thio-β-d-galactopyranoside for 3 h, bacteria were collected by centrifugation, resuspended in 10 mL of buffer A (10 mM HEPES [pH 7.5], 200 mm NaCl, 1 mM MgCl₂, 20 mM imidazole, 5% glycerol) with 0.25 mg mL⁻¹ lysozyme per gram of pellet and frozen and thawed twice. Lysates were treated with DNase I (100 U mL⁻¹) for 20 min at 30°C, and supernatants containing σ^A-His or σ^B-His proteins were obtained by centrifugation at 8,000 × g for 15 min. For RNA polymerase purification, a fresh overnight culture of S. aureus HG003 rpoC-his was used to inoculate 500 mL of BHI at an OD₆₀₀ of 0.1 and grown for 5 h at 37°C. The culture was harvested by centrifugation at 4,000 × g for 15 min. The cell pellet was resuspended in 10 mL of buffer A with 1 mg of lysostaphin and DNase I (100 U mL⁻¹). After incubation for 20 min at 37°C, the lysate was clarified by centrifuging for 30 min at 40,000 × g. For affinity purification of σ^A-His, σ^B-His, and His₁₀-tagged RNAP, a HiTrap Talon column (5 mL; GE Healthcare) was connected to an AKTA Prime chromatography system (GE Healthcare) equilibrated with buffer B (10 mM HEPES [pH 7.5], 1 M NaCl, 20 mM imidazole, 5% glycerol). After loading the lysate containing either σ^A-His, σ^B-His, or His₁₀-tagged RNAP, the column was washed with 100 mL of buffer B. His-tagged proteins were then eluted using an imidazole gradient, dialyzed in buffer C (10 mM HEPES [pH 7.5], 100 mM NaCl, 1 mM MgCl₂, 5% glycerol), and then subjected to a second step of purification on heparin column. Proteins were loaded on a HiTrap heparin HP column (1 mL; GE Healthcare) equilibrated with buffer C. After a wash with 20 mL of buffer C, proteins were eluted using a NaCl gradient, dialyzed in buffer C, and then concentrated in a centrifugal concentrator with a 10-kDa-molecular-weight-cutoff membrane (Merck Millipore).

EMSA.

ssrS^Sa and sprB were in vitro transcribed from PCR product templates containing a T7 promoter (primers listed in Table S3) with a MEGAscript T7 transcription kit (Thermo Fisher Scientific). RNAs were separated by 8% polyacrylamide–7 M urea gel electrophoresis and eluted overnight in G50 elution buffer (20 mM Tris-HCl [pH 7.5], 2 mM EDTA and 0.25% SDS). RNAs were precipitated in cold ethanol and 0.3 M sodium acetate and dephosphorylated using calf intestinal alkaline phosphatase (New England Biolabs), according to manufacturer protocol. The RNAs obtained were 5′ radiolabeled with T4 polynucleotide kinase (New England Biolabs) and [γ-³²P]ATP (ATP) and purified with MicroSpin G-50 columns (Amersham Pharmacia Biotech). RNAP (140 nM) alone or preincubated 10 min at 37°C with 420 nM sigma factors was mixed with 4 nM radiolabeled 6S RNA or SprB in buffer D (15 mM HEPES [pH 7.5], 100 mM NaCl, 1 mM MgCl₂, 5% glycerol, 100 μg mL⁻¹ BSA, 200 μg mL⁻¹ E. coli tRNA). Complex formation was performed at 37°C for 10 min, and samples were loaded on 5% polyacrylamide–5% glycerol gels under nondenaturing conditions. Gels were dried and visualized using a Typhoon phosphorimager (Molecular Dynamics).

σ^A and β/β′ subunit quantification.

S. aureus strains were cultured in triplicate until the desired OD₆₀₀ was reached. Frozen pellets were lysed in 50 mM Tris-HCl buffer (pH 7.5) with glass beads. The total protein amount in the supernatant was determined by Bradford protein assay. Western blot electrophoresis was performed with 3 μg proteins per well, using 8% bis-Tris Plus polyacrylamide gels (Bolt; Invitrogen). Transfer and hybridization followed iBlot and iBind manufacturer instructions (Invitrogen), respectively. Membranes were prehybridized at 4°C ON with human serum (1:10,000 dilution) to saturate unspecific binding. Rabbit primary antibodies were used for immunodetection of σ^A (anti-σ^A; 1:5,000 dilution) and β/β′ subunits (anti-RNAP; 1:10,000 dilution). A horseradish peroxidase (HRP)-conjugated goat anti-rabbit immunoglobulin (Advansta; 1:4,000 dilution) was chosen as the secondary antibody. Pictures were taken with a charge-coupled device (CCD) camera. The statistical analysis for the comparison of σ^A normalized to β/β′ quantity between strains was done by a one-way analysis of variance (ANOVA), followed by Tukey’s honestly significant difference (HSD) test to identify pairs with significantly different amounts of σ^A.

Data availability.

The data for this study have been deposited in the European Nucleotide Archive (ENA) at EMBL-EBI under accession number PRJEB50160 (https://www.ebi.ac.uk/ena/browser/view/PRJEB50160).

This work is dedicated to the memory of our colleague Brice Felden, who passed away unexpectedly on 5 March 2021.