The interaction of ω₂ with the RNA polymerase β’ subunit functions as an activation to repression switch

Abstract

The ω gene is encoded in broad-host range and low-copy plasmids. It is genetically linked to antibiotic resistance genes of the major human pathogens of phylum Firmicutes. The homodimeric forms of ω (ω₂) coordinate the plasmid copy number control, faithful partition (ω₂ and δ₂) and better-than-random segregation (ζϵ₂ζ) systems. The promoter (P) of the ωϵζ operon (P_ω) transiently interacts with ω₂. Adding δ₂ facilitates the formation of stable ω₂·P_ω complexes. Here we show that limiting ω₂ interacts with the N-terminal domain of the β’ subunit of the Bacillus subtilis RNA polymerase (RNAP-σ^A) vegetative holoenzyme. In this way ω₂ recruits RNAP-σ^A onto P_ω DNA. Partial P_ω occupancy by ω₂ increases the rate at which RNAP-σ^A complex shifts from its closed (RP_C) to open (RP_O) form. This shift increases transcription activation. Adding δ₂ further increases the rate of P_ω transcription initiation, perhaps by stabilizing the ω₂·P_ω complex. In contrast, full operator occupancy by ω₂ facilitates RP_C formation, but it blocks RP_O isomerization and represses P_ω utilization. The stimulation and inhibition of RP_O formation is the mechanism whereby ω₂ mediates copy number fluctuation and stable plasmid segregation. By this mechanism, ω₂ also indirectly influences the acquisition of antibiotic resistance genes.

INTRODUCTION

Resistance to glycopeptides, macrolides, pleuromutilins, phenicols, linezolid and other antibiotics among Gram-positive cocci is generally linked to plasmids whose copy number control, partitioning and post-segregational killing is regulated by the ω cassette (1,2). This cassette contains sequences that encode the ω or ω2 gene products (1). It is thus important to understand how homodimeric ω (ω₂) (or ω2 [ω2₂]) functions, not only because of its intrinsic biological interest, but also because of its relevance to antibiotic resistance transmission (1,2). Plasmids of the inc18 family are commonly found in Enterococcus and Streptococcus. These plasmids have a broad host range in Firmicutes. Here, ω forms an operon with ϵ and ζ. Meanwhile, ω2 forms an operon with the ermB gene (1).

A transcriptional analysis of few inc18 plasmids (e.g. pSM19035, pIP501 and pAMβ1) revealed that ω₂ controlled the expression of the copy control gene copS. Also mediated by ω₂ was the expression of plasmid partition genes, such as δ and ω, and toxin-antitoxin systems, such as the ωϵζ operon (Figure 1A) (1). The inc18 plasmids persist in the population through a variety of mechanisms controlled by ω₂. The ω₂ protein is a ParB homologue and binds to parS centromers. In concert with δ₂ (a ParA ATPase), ω₂ is involved in accurate plasmid partitioning and coupling plasmid replication to faithful segregation (1,3–5). All three parS sites (Figure 1A) can cause partition-mediated incompatibility (5). Furthermore, toxin ζ stabilizes bacterial plasmids by programming the death of any host cell that fails to inherit a plasmid copy during cell division (1,6). Toxins ζ and RelE are the most ubiquitous toxins in nature. In contrast, much less is known about the purpose of the ω2 gene product, which is truncated in some members of the family (e.g. pSM19035) (Figure 1A) (1).

Figure 1.

Interaction of ω₂, δ₂ or RNAP-σ^A with the P_ω operator sites. (A) Genome organization of the relevant region of plasmid pSM19035. The promoters (P), the mRNAs and the genes are symbolized by boxes, wavy lanes and rectangles, respectively. The plasmid replication origin (ori) is labelled in orange. The direction of replication is denoted by a black arrows below. Protein ω₂-mediated transcriptional repression is indicated (ω₂, red ovals). The upstream regions of P_copS, P_δ and P_ω (red box) which constitute the cis-acting parS centromeric sites magnified. The ω₂ cognate sites consist of a variable number of contiguous 7-bp heptad repeats (iterons) symbolized by ▸ (in the direct orientation) or ◂ (in the inverted orientation). The number of repeats and their relative orientations are indicated. The genes involved in replication (copS, repS, RNAIII and γ), dimer resolution (β), faithful partition (δ and ω), stable segregation (ϵ and ζ) are indicated. The antibiotic resistance gene ermB and the truncated version of ω2 (‘ω2’) are also indicated. (B) A structural model of ω₂-bound to P_ω DNA which derived from the crystal structure of the complex of the minimal operator site and ω₂Δ19 (PDB ID 1IRQ, 2BNW and 2BNZ). Pink/purple ω₂ molecules form a left-handed protein-matrix winding around the nearly linear operator DNA. The DNA is represented in grey with the −35 and −10 elements in yellow. (C) Model of RNAP-σ^A which is derived from crystal structures of the homologous protein from T. aquaticus and T. thermophilus (PDB ID 1IW7, 2A6H) together with the P_ω DNA from PDB ID 2CAX. Colour coding: brown and light brown refer to α₂; blue, β; red, β’ and green, σ subunit. (D) Model of δ₂ binding to DNA. The atomic coordinates of (δ·ATPγS·Mg²⁺)₂ were taken from the 2OZE PDB entry. The modelled structures were prepared and visualized with PyMOL version 1.5.0.4. (E) Interactions of ω₂ and δ₂ with P_ω DNA and each other: 1. ω₂ (in purple) transiently interacts with P_ω DNA forming complex C1; 2. The interaction of ω₂ with δ₂-apo (in grey), leads to functional transition of ω₂ (in green) and formation of the durable C2 complex; 3. In the presence of ATP, δ₂ (in blue) binds to C1 to generate C3. (F) Complexes formed by ω₂ and RNAP-σ^A upon binding to P_ω DNA: 1. RNAP-σ^A bound to P_ω DNA forms complex RC1; 2. the interaction of RNAP-σ^A with limiting concentrations of ω₂ leads to a functional transition of ω₂ and formation of C2; and 3. RNAP-σ^A bound to C1 makes RC2.

Streptococcus pyogenes monomeric ω (71-residue long, 7.9 kDa) has an unstructured N-terminal domain (NTD, residues 1–24) followed by a ribbon-helix-helix (RHH) fold (residues 25–71). The latter facilitates the formation, in solution, of a dimer that has a pseudo-2-fold symmetry (7–9). The RHH domain recognizes the parS centromers embedded in the promoter regions of the cop, δ, ω and ω2 genes (Figure 1A and Supplementary Figure S1) (4). The operator binding sites are comprised of a series of 6–10 unspaced heptad repeats (5′-^T/_AATCAC^T/_A-3′) in a forward orientation. Alternatively, they consist of two or three repeats of the following: two heptads in a forward orientation followed by one in an inverse orientation (→→←) (Supplementary Figure S1). Both the ω₂ and the NTD lacking ω₂ΔN19 mutant transiently bind promoters and repress promoter utilization both in vivo and in vitro (4,9–13).

The minimal ω₂ binding site is comprised of two contiguous heptads in a forward (→→) or inverted (→←) orientation. It has higher affinity for the latter (see 10). The structure of the complex of ω₂ bound to →→ DNA is very similar to the one of ω₂ bound to →← DNA. In neither case does ω₂ distort the DNA when binding to it (9,14). These structures show that a pair of positively charged antiparallel β strands from ω₂ insert into the major groove of DNA. The β strands make specific and sequence-dependent contacts with symmetric or asymmetric repetitive sequences that deviate 0.3 Å with respect to the central C-G pair of each repetition (8,9,14). In a full cognate site, ω₂ is displaced ∼7-bp and rotated 252º with respect to its neighbouring dimer. The negatively charged sugar-phosphate DNA backbone faces the positively charged surface of the protein (Figure 1B) (9).

Protein ω₂ transiently binds with high affinity (apparent dissociation constant [K_Dapp] = 5 ± 1 nM) and cooperativity to P_ω DNA (Figure 1E, condition 1 [C1]) (13). The physical interaction of the apo form of δ₂ with ω₂ bound to P_ω DNA facilitates a structural transition in ω₂ that might involve folding of its unstructured NTD (Figure 1E, condition 2) (see 13,15). ω₂ stably binds P_ω DNA with high affinity (K_Dapp = 0.7 ± 0.1 nM), forming the C2 complex (ω₂·P_ω DNA) (Figure 1E, condition 2). The C2 complex is stable, with a half-life of >30 min, whereas the C1 complex is transient, with half-life of <1 min (10,13,15). However, despite the difference in stability, C1 and C2 have a similar mobility in a PAGE at low protein concentrations.

The δ₂ protein is a U-shaped ATPase that in its ATP-bound form, binds non-specifically to DNA (Figure 1D) (16). In the presence of ATP, δ₂ interacts with C1 to form the C3 complex (Figure 1E, condition 3) (13). Since it lacks the unfolded NTD, ω₂ΔN19 cannot facilitate C2 and C3 formation (13). Given that δ₂ (a ParA ATPase) works together with ω₂ (a ParB centromeric binding protein) bound to parS (e.g. P_ω DNA) to promote faithful plasmid segregation (12,13), it is likely that δ₂ also contributes to ω₂-mediated transcription regulation.

Transcription initiation by the multisubunit RNA polymerase (RNAP) is an intricate multistep process (17–20). Bacterial RNAP exists in two forms: i) the ubiquitous core enzyme, which consists of the dimeric form of α (α₂), the monomeric form of β, β’, and one or more of small non-essential subunits; this carries out processive transcription elongation followed by termination; and ii) the RNAP-σ holoenzyme, in which a dissociable σ subunit, essential for promoter recognition, has joined the core enzyme (21–23). The Bacillus subtilis vegetative RNAP-σ^A holoenzyme binds to specific −10 and −35 promoter (P) elements to form an unstable closed binary complex (RP_C) (21,24–27). A RNAP-σ^A-assisted isomerization step then occurs. This is mediated by kinetically unstable intermediates (RP_I). This, in turn, leads to P melting of ∼14-bp (−12 to +2) in the DNA surrounding the transcription start site. This process yields the catalytically active, open RNAP-σ^A·P DNA complex (RP_O) (17,28). The structures responsible for the functions associated with RPo formation are predominantly located in the σ, β and β’ subunits of the RNAP-σ (18,22,23,25,28). In the presence of nucleotide triphosphates, an initiation complex (RP_INIT) is formed. This complex is a prerequisite for displacement of RNAP-σ^A from the promoter through an elongation complex (RP_E) (24,25,28). RNAP subunits δ, ϵ and ω are not essential for this process and their roles are therefore poorly understood.

As a result of the association of regulatory elements to promoter-embedded operator sequences, gene regulation is often achieved at the level of transcription initiation (29). The ω₂ protein interacts with its cognate sites as a left-handed protein helix wrapped around a nearly linear P_ω DNA (9). In this structure the −35 and −10 elements are free to interact with RNAP-σ^A (Figure 1B, yellow regions). A model of RNAP-σ bound to P_ω DNA suggests that ω₂ might repress transcription by steric hindrance (Figure 1C). However, preliminary results indicate that ω₂ forms a ternary complex with RNAP-σ^A and P_copS DNA (4). These data suggest that ω₂ regulates transcription through a mechanism that does not exclude the RNAP-σ^A from the RP_C. It is assumed that this mechanism also applies to P_ω and P_δ. In this study, we aimed to unravel the mechanism of ω₂-mediated transcriptional regulation of P_ω DNA, in vitro, and P_δ utilization, in vivo. We first characterized the effect that ω₂ binding to P_ω DNA had on RNAP-σ^A promoter recognition. We also tested whether or not modifying the stoichiometry of ω₂, δ₂ and RNAP-σ^A resulted in variations in their affinity for P_ω DNA. Also investigated was whether or not transcription activation or repression by ω₂ required direct contacts with RNAP-σ^A and δ₂. Another important question was whether or not this binding was cooperative. Based on the results of this study, we present a model that explains how ω₂-mediated transcriptional regulation functions.

MATERIALS AND METHODS

Bacterial strains and plasmids

The E. coli strains DH5α (Invitrogen) and ER2566 (New England Biolabs) and the B. subtilis strains BG214, BG508 (4) and NIG2001 (30) were used. The BG508 strain carries P_δ fused to a promoter-less lacZ gene. This construct was integrated as a unique copy into the amyE locus of the B. subtilis chromosome (4). In the NIG2001 strain, the wild-type (wt) rpoC gene was substituted in the B. subtilis genome with a version that had a His-tag coding sequence fused to the 3′-end (30). The P_ω bearing pCB30 plasmid was used for promoter analysis, and pHP14 was used for cloning purposes (4). The plasmids used for gene over-expression were pT712ω bearing ω, pCB746 bearing δ (4,11,16), and pT712ωD56A bearing the ωD56A gene (this work). The single mutations in the ω gene were obtained by gene synthesis (Genscript). BG508 cells bearing pHP14 carrying either the ω, δ, ωδ, ωΔN19, ω2, ωK52A, ωE53A, ωD56A, ωR64A or ωK70A genes were used for the β-galactosidase assays. The native promoters of these genes were also incorporated into the constructs.

DNA, RNA, proteins and reagents

Plasmid DNA was purified as described (4). The multiple mutations in heptads 1, 1 plus 2, 7 and 7 plus 6 were obtained by in vitro synthesis (Genscript). DNA restriction and modification enzymes and RNaseA were purchased from Boehringer Mannheim and the nucleotides were purchased from Sigma. Gel-purified DNA fragments were end-labelled as described (4). The amount of DNA was quantified using molar extinction coefficient of 6500 M⁻¹ cm⁻¹ at 260 nm and was expressed in moles of DNA molecules.

The RNAP-σ^A was purified using Ni-NTA and Q-sepharose columns as described (30). The ω₂ and δ₂ proteins were purified as described (4,11,16). Protein ωD56A was purified as wt ω₂. Protein concentrations were calculated using molar extinction coefficients at 280 nm of 2980, 2980, 38 850 and 236 000 M⁻¹ cm⁻¹ for ω₂, ωD56A, δ₂, and RNAP-σ^A, respectively. Concentrations were expressed in molarity of protein monomers for RNAP-σ^A and of dimers for δ₂, ω₂ and ω₂ derivatives. Note that unless otherwise stated, δ₂ is in the ATP·Mg²⁺-bound form.

B. subtilis BG508 harbouring different plasmids was grown to OD₆₀₀ = ∼0.5 and aliquots were used for β-galactosidase assays (4). The cultures were pelleted and resuspended in buffer B (10 mM Na₂HPO₄/NaH₂PO₄ pH 7.2, 50 mM β-mercaptoethanol, 1 mM MgCl₂, 10 mM KCl) containing 0.4 μg/ml lysozyme (4). After a 5 min incubation at 37ºC, the lysates were clarified by centrifugation for 5 min at 12,000 g and assayed for β-galactosidase activity, as described (31).

Protein cross-linking was used to study potential protein–protein interactions. For this, bisdisuccinimidyl suberate (DSS) was employed as the crosslinking agent and SDS-PAGE was used to visualize the result (13). Two-dimensional gel electrophoresis (2D) was performed essentially as described (32). The resolved proteins were transferred onto a 0.45 μm polyvinylidene fluoride membrane (PVDF, Millipore). Rabbit polyclonal anti-ω₂ and anti-RNAP-σ^A antibodies were obtained using standard techniques (4).

Far-western blotting was used to probe the direct interaction between ω₂ and RNAP-σ^A. The prey used were ω₂ (1 μg), RNAP-σ^A (1 μg) and bovine serum albumin (BSA, 5 μg used as a control); these were resolved by SDS-PAGE and transferred to a PVDF membrane. The protein was renatured by incubation of the membrane in TBS containing 0.05% Tween, 10% glycerol and 5 mM β-mercaptoethanol followed by a blocking step with 5% skim milk, as described (4). The efficiency of the protein transfer was checked by Ponceau staining. The membrane was then incubated with 2 μg/ml of bait protein. To detect interactions between the bait and prey, rabbit polyclonal antibodies against the bait were employed as described (33).

Tryptic digestion of gel-purified protein bands, spotting onto the MALDI-targets, and MALDI-TOF-TOF of the spotted peptides were carried out as previously described (34).

In vitro transcription experiments

A 423-bp P_ω DNA sequence (5 nM) was used as a template for in vitro transcription run-off assays. 20 μl reaction mixtures containing 20 nM B. subtilis RNAP-σ^A, variable concentrations of ω₂, δ₂ or both, 0.5 mM each of ATP, CTP, GTP and UTP plus 3000 Ci/mmol [α-³²P]-UTP in buffer C (25 mM Tris-HCl, pH 8, 6 mM MgOAc, 5 mM DTT), and 20 U RNasin (Promega) were prepared. After 6 min of incubation at 37°C, the reactions were stopped by adding 10 μl of formamide. RNAs were analysed by 8% denaturing (d) polyacrylamide gel electrophoresis (PAGE), and autoradiographed. Chemical sequencing reactions of the purines were run in parallel to determine the sizes of the cDNAs.

Protein-DNA complexes

For electrophoretic mobility shift assays (EMSA), the 423-bp [α³²P]-P_ω DNA (0.1 nM) was incubated either with a variety of concentrations of ω₂, δ₂ or RNAP-σ^A or with a constant concentration of one component and a range of concentrations of the others. Incubations were performed in buffer D (50 mM Tris-HCl pH 7.5, 50 mM NaCl, 10 mM MgCl₂) for 15 min at 37°C in a 20 μl reaction. Mixtures were subjected to 6% PAGE in 1xTAE at 4ºC. Gels were dried prior to autoradiographical analysis.

In order to obtain K_Dapp values from the EMSA experiments, the relative concentrations of free DNA and protein·DNA complexes were densitometrically determined under non-saturating conditions using differently exposed autoradiographs of the EMSA gels. The protein concentration needed to trap 50% of the free, labelled DNA containing the same molar concentration of heptads, in complexes is approximately equal to the K_Dapp under conditions where the DNA concentration is much lower than the K_Dapp.

Reaction conditions similar to those used for EMSA were employed in footprinting experiments. The 423-bp [α³²P]-P_ω DNA (1 nM) was incubated with variable protein concentrations and treated with DNaseI, as previously described (4). The samples were resolved by 6% dPAGE and the gel was dried prior to autoradiographical analysis. For KMnO₄ footprinting, the samples were treated with 1 mM KMnO₄ for 0.5 min at 37º C, after which the DNA was cleaved with piperidine (35).

RESULTS

RNAP-σ^A facilitates ω₂·P_ω DNA complex formation

In order to unravel the mechanism of ω₂-mediated regulation of P_ω utilization, the interaction of ω₂ and/or RNAP-σ^A with P_ω DNA was assayed by EMSA (Figure 2A and 2B). P_ω DNA has 7 discrete ω₂ cognate sites. A 423-bp DNA segment containing P_ω bound ω₂ with high affinity to form ω₂·P_ω DNA also known as the C1 complex (K_Dapp of 6 ± 1.7 nM) (Figure 2C filled circles). When P_ω DNA was replaced with a non-specific DNA, the affinity of ω₂ was low, with a K_Dapp > 500 nM (data not shown) (4). This confirmed previously reported data indicating a high affinity of ω₂ for P_copS, P_δ and P_ω (10). In addition, binding was found to be cooperative (Figure 2C). P_ω also bound RNAP-σ^A with high affinity to form RNAP-σ^A·P_ω DNA, also known as the RC1 complex (K_Dapp of 29.4 ± 9 nM, Figure 2C filled rombs).

Figure 2.

Cooperative binding of ω₂ and RNAP-σ^A to P_ω DNA. (A) EMSA of (0.1 nM) 423-bp [α³²P]-P_ω DNA incubated with 0.75 nM and 6 nM ω₂ (lanes 2 and 3), increasing concentrations of RNAP-σ^A (1.9, 3.7, 7.5, 15, 30 and 60 nM, lanes 4–9), or fix ω₂ (0.75 nM) and increasing concentrations of RNAP-σ^A (1.9–60 nM) (lanes 10–15) in buffer D. (B) EMSA of [α³²P]-P_ω DNA with of 0.75 nM and 6 nM ω₂ (lanes 9 and 10, respectively) or with 7.5 nM (lanes 1–7) or 15 nM (lanes 11–17) RNAP-σ^A and increasing concentrations of ω₂ (0.19, 0.37, 0.75, 1.5, 3 and 6 nM) in buffer D. (C) Graph showing the percentage of P_ω DNA bound to the proteins based on densitometric data of bands from the above gels. The signals present in the protein-DNA complex and in the free-DNA (FD) were determined by densitometry. The data presented here are averages and standard deviations of the results of at least four independent experiments.

In the presence of limiting ω₂ (0.75 nM) concentrations, ω₂·P_ω DNA complex formation was not observed (Figure 2A, lane 2 and 2C filled circles). Increased concentrations of RNAP-σ^A enhanced recruitment of ω₂ to the promoter region by at least 8-fold. This was observed as an increase in the formation of high affinity C2 complexes. This effect was detected using limiting RNAP-σ^A concentrations (3.7 nM) (Figure 2A, lanes 11–15). The ‘cooperative’ binding leading to C2 complex formation could not be attributed to molecular crowding, because the BSA control did not function as a substitute for RNAP-σ^A (data not shown). The stable C2 and the transient C1 had a similar mobility. However, at higher concentrations of ω₂, the mobility of C2 was even further diminished (Figure 2A, lanes 3 versus 11). The C1 complex had a half-life of <1 min (10). The formation of the ω₂·P_ω DNA complex was also enhanced at least 8-fold (K_Dapp of 0.6 ± 0.2 nM) when a fixed limiting concentration of RNAP-σ^A (∼4-fold below K_Dapp) and increasing concentrations of ω₂ were incubated with P_ω DNA (Figure 2B, lanes 1–6 and 2C open circles).

To distinguish cooperative binding from a mechanism whereby a protein–protein interaction preceded binding to P_ω DNA, the experiment was modified by doubling the concentration of RNAP (∼2-fold below K_Dapp). In the presence of sub-stoichiometric concentrations of RNAP-σ^A (15 nM), the ω₂·P_ω DNA complex formation became enhanced by at least 10-fold. This resulted in a stoichiometry of ∼2 ω₂/P_ω DNA (Figure 2B, lanes 12–17 and 2C, open circles). This is consistent with the known characteristics of ω₂ binding to P_ω DNA which had a stoichiometry of ∼1 ± 0.2 ω₂ /heptad. The minimal ω₂ binding site consisted of two contiguous heptads (9–12). It is therefore likely that ω₂ interacts with RNAP-σ^A, and that such an interaction induces a conformational change in the former that increases its apparent affinity for P_ω DNA. This favours the formation of the C2 complex. Meanwhile, in the absence of RNAP-σ^A, ω₂-bound to P_ω forms the C1 complex with an 8-fold lower apparent affinity (K_Dapp of 6 ± 1.7 nM).

A low concentration of ω₂ facilitates the formation of the RNAP-σ^A·P_ω DNA complex

Limiting concentrations of ω₂ enhanced the recruitment of RNAP-σ^A to P_ω DNA (Figure 2B, lanes 1–3 and 12–14). 7.5 nM RNAP-σ^A was the limiting concentration of RNAP-σ^A necessary to detect the RC2 complex (RNAP-σ^A·P_ω DNA·ω₂). This was ∼4-fold less than the K_Dapp (Figure 2C open versus filled rombs). It is therefore likely that ω₂ interacts with RNAP-σ^A and facilitates a functional transition of RNAP-σ^A. To test whether or not ω₂ and RNAP-σ^A co-localize in a RC2 complex, ω₂ and RNAP-σ^A were incubated with P_ω DNA and subjected to DNase I footprinting analysis. The ω₂ protein protected nucleotides −22 to −75 (with a numbering relative to the +1 transcription start site) (Supplementary Figure S2A, lanes 4–5). Meanwhile, RNAP-σ^A made a weak but significant contact with a segment located between positions −53 to +18. In parallel, a clearly hypersensitive site appeared at position −37. This is denoted by a dotted line square in Supplementary Figure S2A, lanes 6–9. Addition of a limiting concentration of ω₂ of ∼4-fold below the K_Dapp resulted in the disappearance of the hypersensitive site. This effect was reversed upon increasing the concentration of RNAP-σ^A (Supplementary Figure S2A, lanes 10–13). It was also reversed by increasing the concentration of ω₂ to ∼2-fold below the K_Dapp (Supplementary Figure S2B, lanes 13–16). Judging by the fading out of the hypersensitive site at position −37, it is likely that limiting concentrations of ω₂ or RNAP-σ^A reposition RNAP-σ^A on the P_ω DNA.

Addition of sub- to stoichiometric concentrations of ω₂ led to an increase in the formation of RC2 (RNAP-σ^A·P_ω DNA·ω₂) complex formation by at least 3-fold (K_Dapp of 9.5 ± 3.4 nM) (Figure 2B, lane 17 and 2C, open rombs). Equilibrium was thus reached at about ∼4 ω₂/RNAP-σ^A/P_ω DNA. These results suggested that there was not a sufficient number of ω₂ molecules to occupy the seven P_ω heptads. Since ω₂ binds with a slightly higher affinity and cooperativity to heptad pairs in the →← than in the →→ orientation (10), we favour the hypothesis that the →← heptads at positions −41 to −27, which overlap the −35 element and its neighbours, might be the ones recruited by ω₂ to interact with RNAP-σ^A. Indeed, sub- to stoichiometric concentrations of ω₂ bound to P_ω DNA protected this region from DNase I attack (Supplementary Figure S2B, lanes 3–6). Meanwhile at stoichiometric concentrations RNAP-σ^A made weak but extensive contacts with the upstream region, which had the same exposed hypersensitive site at position −37 (Supplementary Figure S2B, lanes 8–11). At sub-stoichiometric concentrations of ω₂, RNAP-σ^A made extensive contacts with the upstream −35 region. The hypersensitive site at position −37 remained exposed (Supplementary Figure S2B, lanes 13–16). This hypersensitive site was lost in the presence of stoichiometric concentrations of ω₂. Meanwhile, protection from DNase I was only observed in a stretch of DNA between positions −72 to −21 (Supplementary Figure S2B, lanes 18–21). In light of these results, it is likely that: (i) ω₂ physically interacts with RNAP-σ^A; and (ii) depending on the experimental conditions, ω₂ either displaces RNAP-σ^A from or re-localises with it on P_ω DNA.

The interaction between ω₂ and δ₂ does not affect RNAP-σ^A binding to P_ω DNA

Protein δ₂ bound non-specific DNA (K_Dapp 130 ± 20 nM) in the presence of ATP. This lead to the formation of the DC complex (Supplementary Figure S3A, lanes 6–8) (13). Binding of δ₂ to non-specific DNA increased 3 to 4-fold in the presence of ω₂·P_ω DNA (Supplementary Figure S3A, lanes 10–12, and S3C, empty squares). In the absence of ATP, δ₂ only augmented the affinity of ω₂ for P_ω DNA by 6- to 10-fold (C2 formation) (Supplementary Figure S3B, lanes 10–15 and Figure 1E, condition 2). This was consistent with observations that: (i) the presence of apo-δ₂ decreased the off rate of ω₂ from ω₂·P_ω DNA complexes; (ii) the presence of ω₂ significantly increased the half-life of the δ₂·P_ω DNA complexes in the presence of ATP (13); and (iii) upon interacting with δ₂, ω₂ that is bound to P_ω DNA (parS) undergoes a structural transition that might involve the formation of an α-helix in the normally unstructured NTD (see 15).

To test whether the interaction of ω₂ with δ₂ or RNAP-σ^A are mutually exclusive or if the interaction between δ₂ and ω₂ affects ω₂-mediated recruitment onto the P_ω DNA of RNAP-σ^A, limiting concentrations of ω₂ (∼8-fold below K_Dapp) and/or RNAP-σ^A (∼4-fold below K_Dapp) were incubated with increasing δ₂ concentrations and subjected to an EMSA (Supplementary Figure S3A and S3B, lanes 10–15). In the presence of ATP, assembly of the ternary C3 complex (δ₂·ω₂·P_ω DNA) occurred (Supplementary Figure S3A, lanes 10–15). However, these results were not observed when ATP was omitted (Supplementary Figure S3B, lanes 10–15). Meanwhile, assembly of the ternary RC2 complex (ω₂·P_ω·RNAP-σ^A) was observed without ATP (Supplementary Figure S3A and S3B, lanes 13–15). Finally, in the presence of ATP, δ₂ did not significantly affect the affinity of RNAP-σ^A for P_ω in the presence of ATP (Supplementary Figure S3D, lanes 8–16).

Whether ω₂ functions as an activator or repressor is dependent on its concentration

Protein ω₂ binds P_copS, P_δ and P_ω DNA with a stoichiometry of ∼1 ± 0.2 ω₂ /heptad (9,12). We previously mapped the pSM19035 transcription start sites of P_copS, P_δ and P_ω (see Supplementary Figure S1). In that study, we showed that 7.5 – 15 ω₂/P_copS, P_δ or P_ω DNA represses promoter utilization (4). To gain insight into the mechanism by which ω₂ regulates promoter utilization, we performed transcription run-off experiments using RNAP-σ^A (at K_Dapp) in the presence of increasing concentrations of ω₂. Linear P_ω DNA containing seven heptads was used as the template (Figure 3A). As expected, 282-nt mRNA transcripts were produced (Figure 3B, lane 1). This result is consistent with the location of the initial nucleotides of P_ω, which had been previously mapped in vivo (4).

Figure 3.

RNAP-σ^A mediated transcription as a function of the presence or absence of ω₂ and δ_2. (A) A 423-bp DNA segment containing P_ω is depicted. The heptads are labelled and their relative orientations are represented by arrows. The positions of the −35 and −10 elements are indicated with filled rectangles. The transcription start site is represented with a solid arrow bent 90º. The ω and part of the ϵ gene are indicated as well. (B) Run-off experiments: the 423-bp P_ω DNA (5 nM) was the template in an in vitro transcription experiment using [α³²P]-UTP in buffer C. RNAP-σ^A (20 nM) was present in all cases. Results shown for transcription in the absence (lane 1) or presence of increasing concentrations of either ω₂ (3.7, 7.5, 15, 30 and 60 nM, lanes 2–6) or δ₂ (37, 75, 150, 300 nM, lanes 7–10). Also shown are results of assays in the presence of either 150 or 300 nM δ₂ (lanes 11–16 or 17–22 respectively) with increasing concentrations of ω₂. (C) Quantification of mRNA synthesis in the presence of increasing concentrations ω₂ alone or in the presence of a fix concentration of δ₂ (150 or 300 nM). Shown here are the analysed results from five independent experiments.

In the presence of limiting concentrations of ω₂ (0.9 and 1.8 ω₂/P_ω DNA), transcriptional activation was modest (1.4- to 1.7-fold), but reproducible (Figure 3B, lanes 2–4, and 3C). Protein ω₂ might preferentially bind the heptads of the P_ω DNA overlapping the −35 element. To test this hypothesis, the heptad 7, heptads 6 plus 7 and, as controls, heptads 1 or 1 plus 2 were inactivated (Figure 3A). As documented in Supplemental material Annex 1, the selective occupancy of heptads 6 and 7 versus 1 and 2 plays a minor role, if at all, in ω₂-mediated activation of P_ω utilization.

Stoichiometric concentrations of 7.5 ω₂/P_ω DNA inhibited P_ω expression by 4- to 8-fold. At slightly saturating conditions (15 ω₂/P_ω DNA or ∼2 ω₂/heptad) mRNA synthesis halted completely (>50-fold) (Figure 3B, lanes 5 and 6). Concentrations of ω₂ equal to or higher than those required to repress P_ω did not affect the expression of the unrelated promoter (P_cro of phage A2) (data not shown). We could hence rule out RNase contamination or any other nonspecific effect as the reason for the lack of RNA synthesis. It is therefore likely that ω₂ has a dual activity: at limiting concentrations it facilitates the P_ω-RNAP-σ^A interaction, but at stoichiometric concentrations and higher, transcriptional repression results.

Limiting concentrations of ω₂ facilitate the transition from RP_C to RP_O and stoichiometric concentrations of ω₂ block this shift

To discern the mechanism of ω₂-mediated repression of P_ω, we investigated RP_O complex formation and abortive initiation (RP_INIT) effects in the presence of variable concentrations of ω₂. For this, we carried out KMnO₄ footprinting assays in the presence or absence of GTP and ATP. Up to 9-nt long transcripts were synthesized in these assays (see Supplementary Figure S1).

No oxidized thymines were detected on the template strand after 50 s of KMnO₄ exposure (Figure 4, lanes 1, 7, 13 and 19). In the absence of nucleotide precursors, the position in the P_ω DNA of non-base-paired thymines preferentially attacked by KMnO₄ revealed that RNAP-σ^A promoted spontaneous formation of a RP_O complex centred at position −11T and −10T of the template strand, rather than an extended P melting of ∼14-bp (from −12 to +2) (Figure 4, lanes 2 and 8). KMnO₄-promoted cleavage of RNAP-σ^A bound template increased in the presence of sub-stoichiometric concentrations of ω₂ (Figure 4, Supplementary Figure S4A, lanes 3–4). Cleavage was inhibited at saturating concentrations of ω₂ (Figure 4, Supplementary Figure S4A, lanes 5–6). However, increased cleavage was not observed when sub-stoichiometric ω₂ concentrations were added to reactions containing RNAP-σ^A, i.e. with a pre-formed RP_O (Figure 4, lanes 9–10).

Figure 4.

Effect of ω₂ on the formation of RP_O at P_ω. The 423-bp [α³²P]-P_ω DNA (1 nM) was pre-incubated with increasing concentrations of ω₂ (7.5, 15, 30 and 60 nM; lanes 2–6 and 14–18) or with 7.5 nM RNAP-σ^A (lanes 8–12 and 20–24) in buffer C. A second protein was added along with the initiating nucleotides, GTP and ATP (as indicated). DNA melting was probed by KMnO₄ footprinting as a way of observing the open complex. The positions hypersensitive to KMnO₄ are labelled (RP_O and RP_INIT) and depicted at the bottom of the figure. The coordinates are relative to the transcription start point. Chemical sequencing reactions for purines (G +A) are shown and the relevant regions of P_ω depicted to the right of the figure.

In the presence of ATP and GTP, RNAP-σ^A promoted abortive initiation and synthesis of up to 9-nt oligonucleotides (RP_INIT) (data not shown). KMnO₄ attack revealed the formation of an extended single-stranded bubble that cleaved at positions −10T, −6T, −5T and +7T (Figure 4, lanes 14 and 20). When ω₂ and RNAP-σ^A were left out, no cleavage was observed (Figure 4, lanes 13 and 19). Pre-incubation of P_ω DNA with sub-saturating concentrations of ω₂ followed by addition of RNAP-σ^A resulted in a significant increase (∼2.5-fold) in KMnO₄ cleavage (Figure 4, lane 15 and 21 and Supplementary Figure S4B). At higher ω₂ concentrations, the reaction became inhibited (Figure 4, lanes 17–18, Supplementary Figure S4A, lanes 15–16 and S4B). However, RNAP-σ^A-mediated RP_O formation was hardly, if at all, affected by addition of ω₂ to the preformed RNAP-σ^A·P_ω complexes (Figure 4, lanes 21–24).

Altogether, these data suggest that: (i) ω₂ does not repress transcription by sterically hindering the interaction between RNAP-σ^A and P_ω DNA; (ii) in the absence of the nucleotides cofactors, RNAP-σ^A forms a short RP_O complex on P_ω centred at position −11T and −10T; (iii) limiting concentrations of ω₂ push RP_O to begin RNA synthesis (RP_INIT); (iv) stoichiometric concentrations of ω₂ inhibit RP_O formation, with ω₂ blocking the isomerization of RP_C to RP_O; and (v) ω₂ has no apparent effect on pre-formed RP_O. We cannot rule out that saturating concentrations of ω₂ may inhibit transcription by steric occlusion or relocation of RNAP-σ^A on preformed ω₂·P_ω complexes (see Supplementary Figure S2B, lanes 18–21).

Protein δ₂ represses P_ω expression

In the presence of ATP, δ₂ binds non-specific DNA with a K_Dapp = 130 ± 20 nM) (Supplementary Figure S3A, lanes 6–8) (13). As shown in Figure 3B (lane 1 versus 7–10), limiting concentrations of δ₂ did not affect transcription of P_ω. Meanwhile, stoichiometric and saturating concentrations of δ₂ inhibited P_ω utilization (2.8- to 3.5-fold). Ultimately utilization was blocked entirely.

To unravel the mechanism of δ₂-mediated P_ω repression, KMnO₄ cleavage experiments were performed. Pre-incubation of P_ω DNA with sub-stoichiometric to stoichiometric concentrations of δ₂ followed by addition of RNAP-σ^A did not alter the pattern of KMnO₄ cleavage obtained in the absence of δ₂ (Supplementary Figure S5, lanes 1–4 versus 5). This result suggested that δ₂ represses P_ω (Figure 3, lanes 9–10) through gene silencing, i.e. halting RNAP elongation (Figure 3, lanes 9–10). The same model has been as proposed for other ParAB systems to which δ₂ and ω₂ belong (36,37). This model is also consistent with the observation that non-specific binding of δ₂ to DNA might occlude RNAP-σ^A clearance or affect RNAP mediated elongation.

Protein δ₂ acts as a co-activator of P_ω expression

Upon coming into contact with ω₂·P_ω complexes, δ₂ protein bound to non-specific DNA relocates onto ω₂·P_ω to form ternary C3 complexes (Figure 1E, condition 3) (12). The C3 complex is characterized by a longer half-life than is C1 (ω₂·P_ω DNA). To test whether C3 might affect transcription, P_ω run-off experiments were performed (Figure 3B, lanes 11–22). At 150 nM, a concentration of δ₂ equivalent to its K_Dapp, addition of limiting concentrations of ω₂ (1.8 to 3.7 ω₂/P_ω DNA) significantly stimulated mRNA synthesis by >3-fold (Figure 3B, lanes 12–14 and 3C). Nevertheless, synthesis was attenuated and ultimately blocked at higher concentrations of ω₂ (Figure 3B, lanes 15–16).

At 300 nM, δ₂ significantly reduced P_ω utilization by ∼14-fold (Figure 3B, lanes 1 versus 17). Addition of limiting concentrations of ω₂ (1.8–3.7 ω₂/P_ω DNA or 0.2–0.5 ω₂/heptad) significantly stimulated P_ω dependent mRNA synthesis by >6-fold (Figure 3B, lanes 18–20, and 3C). As expected, ω₂ blocked P_ω utilization when used at the slightly saturating concentrations of 15 ω₂/P_ω DNA (Figure 3B, lane 22). It is likely, therefore, that whether ω₂ acts as a transcriptional activator or a repressor hinges on its concentration (Figure 3C). Moreover, while by itself repressing P_ω utilization, δ₂ apparently behaves as a transcriptional co-activator. This is consistent with the observations that: i) the half-life of ω₂·P_ω complex increased ∼30-fold in the presence of δ₂ (10,13); and ii) upon interacting with δ₂, ω₂ bound to P_ω DNA, a rearrangement of its unstructured NTD occurs (see 15).

Protein ω₂ interacts with the NH₂-terminal half of the RNAP-σ^A β’ subunit

RNAP-σ^A and ω₂ cooperatively bind P_ω DNA cooperatively and create higher-order nucleoprotein complexes that reflect the combinatorial control of gene expression (Figures 2–4). This effect is likely attributed to direct protein–protein interactions between adjacent DNA-binding factors that promote the assembly of higher-order complexes. To determine whether or not RNAP-σ^A and ω₂ physically interacted, RNAP-σ^A was bound to a Ni²⁺ agarose column through coordination with the C-terminal histidine tag of its β’ subunit (30). The RNAP-σ^A-bound matrix retained ω₂, even in the absence of P_ω DNA. These proteins also co-eluted from the matrix during an elution step (Supplementary Figure S6).

Initial identification of the RNAP-σ^A subunit(s) responsible for association to ω₂ was achieved by carrying out far-western blots of RNAP-σ^A, ω₂ and BSA as control. These proteins were resolved by SDS-PAGE under conditions under which the small ω protein (7.9 kDa) migrated with the front. After renaturation of prey proteins and a membrane blocking step, the bait, ω₂, was added. Protein–protein interactions were detected using anti-ω₂ polyclonal antibodies. As expected, ω₂ interacted with itself and also with the co-migrating β and/or β’ subunits of RNAP-σ^A. No other subunits gave a signal (Figure 5A). Due to a similarity in mass, β (133.6 kDa) and β’ (134.2 kDa) subunits could not be distinguished. We therefore took advantage of the fact that all B. subtilis RNAP-σ^A subunits, except for β’ (pI 8.8), have acidic pIs and repeated the far-western experiments, this time using two-dimensional (2D)-PAGE. Protein ω₂ interacted with β’ and several proteins in unexpected spots (termed 1–2, 3–4 and 5). These had masses of ∼34 kDa and were located in the basic region of the gel (Figure 5B, Ab-anti ω₂ condition). Corresponding polypeptides were extracted from the gel, subjected to mass spectrometry analysis, and identified as RNAP-σ^A β’ subunit NTDs with slight variations in the C-termini (Figure 5C). Taking into account the sizes observed, it was assumed that the Sw2 structural module (residues 316–342) was missing from these NTD variants of the β’ subunit.

Figure 5.

Far-western blotting of ω₂ and RNAP-σ^A. (A) 1 μg ω₂, 1 μg RNAP-σ^A or 5 μg BSA (as control) were resolved by SDS-PAGE and either stained with Coomasie blue or transferred onto nitrocellulose membranes. Membranes underwent a renaturing step and incubated with the specific ω₂ and RNAP-σ^A antibodies (Ab). (B) 1 μg RNAP-σ^A was resolved by iso-electric focusing in a pH 3–11 gradient followed by an SDS-PAGE. Protein was then transferred to a membrane, renatured, incubated with the ω₂ bait and highlighted with either Ab-anti ω₂ or Ab-anti RNAP-σ^A. (C) Basic, ∼34 kDa polypeptides (1–5) that had reacted with Ab-anti ω₂ were gel purified and identified by mass spectrometry. The regions identified are shown in this figure. The sequence coverage was > 40%.

The central and C-terminal regions of ω₂ appears not to interact with RNAP-σ^A

The ω₂ protein has three functional regions: (i) the unstructured NTD (residues 1–24), which is essential for the ω₂·δ₂ interaction (11,15); (ii) the β-sheet domain (residues 28–32), which is required for ω₂ recognition of its cognate DNA site (9,14); and (iii) the α-helix α1 (residues 34–46) which, in concert with the α2 helix (residues 51–64) contributes to monomer-monomer and dimer-dimer interfaces (8,9). The function, if any, of the C-terminal region (residues 65–71) is unknown (7,8).

Significantly, the 79-residue long ω2 (9.0 kDa) shares a 98% identity with the 71-residue ω for the first 55 residues (Supplemental material Annex 2). However, they only share an 18% identity in the last 24 residues (Supplementary Figure S7A). The ω2 protein repressed P_δ utilization in vivo nearly as efficiently as wt ω₂ (Supplemental material Annex 2, Supplementary Figure S7B). Similar results were observed when the P_δ-lacZ fusion was replaced by the P_ω-lacZ fusion (data not shown). Furthermore, plasmid-borne ω2 and δ genes which were transcribed from P_ω (parS2) and P_δ (parS1) (Figure 1A), were necessary and sufficient for stabilizing of an otherwise unstable plasmid in B. subtilis cells (38, our unpublished results). It is likely that: (i) the dimer is the functional unit of ω2; (ii) ω₂2 interacts with δ₂ just like ω₂; and (iii) the different C-terminal domains of ω and ω2 are not involved in gene repression.

Protein docking experiments predicted that the unstructured ω NTD folded into an α-helical structure that interacted with the β’ NTD of RNAP-σ^A (Supplementary Figure S8). In contrast, several charged amino acids located in the coiled region between the α-helices α1 and α2 (residues 47–52) and in the α2 helix itself (residues 51–64) could make contacts with the oppositely charged residues in the β’ subunit of RNAP-σ^A. To test this hypothesis (Supplemental material Annex 3), the charged residues that were accessible in these domains, K52, E53, D56, R64 and K70, were replaced with alanine. Subsequently, the in vivo behaviour of these mutants was investigated (Supplemental material Annex 3). With the exception of ωD56A, the ω mutant variants repressed P_δ transcription in vivo as efficiently as wt ω₂ (Supplementary Figure S7B) (Supplemental material Annex 3). The D56A mutant only reduced P_δ transcription ∼6-fold. As described in Supplementary Annex 3, purified ωD56A also formed dimers, albeit in a far smaller proportion than wt ω₂ (Supplementary Figure S7C). Since in the dimeric form of ω, the β-sheet domain adopts an antiparallel configuration before binding P_ω, the primary defect of ωD56A might be a poor ability to dimerize. Therefore, it was not further analysed. It would be very interesting to determine whether or not the ω₂ NTD can, by itself recruit RNAP-σ^A to the ω₂·P_ω DNA complex.

DISCUSSION

Direct contacts between ω₂ and RNAP-σ^A stimulate RP_C complex formation and its subsequent isomerization to RP_O. This appears to be the mechanism by which limiting concentrations of ω₂ activate P_ω transcription. This process is further enhanced by δ₂ (Figure 3B). However, stoichiometric concentrations of ω₂ have the opposite effect: here the recruited RNAP-σ^A inefficiently isomerizes into RP_O and represses P_ω or P_δ transcription both in vivo and in vitro. It would be highly interesting to determine whether or not ω₂ functions as an activator to repressor switch of P_copS. However, preliminary results indicate that ω₂·P_copS DNA forms a ternary complex with RNAP-σ^A and that ω₂ regulates transcription through a mechanism that does not exclude the RNAP-σ^A from the RP_C (4).

The dual activity of the ω₂ regulator

Plasmid-encoded ω₂, from Gram-positive cocci, is the only one out of the more than 2000 RHH₂ proteins that can either activate or repress the utilization of a single promoter (P_ω) in a concentration-dependent manner. This is true at least in a simplified in vitro system. The majority of RHH₂ proteins are predicted to be transcriptional repressors (39). However, four of them act both as activators and as repressors. These are: a P22-Arc variant, Mer, AmrZ [AlgZ] and NikR. They bind to a variety of promoters, functioning as repressors for some while behaving as activators for others (40–43). In the case of other putative regulators, various metal cofactors and different stoichiometries might also influence the effect they have on promoter functioning (40–43).

The activator to repressor switch function of ω₂ appears to be managed by the diverse modes of ω₂ binding to the operator region of P_ω. These modes of binding have distinct effects on the initial activity of RNAP-σ^A. Under limiting concentrations, ω₂ promotes RNAP-σ^A binding to operator sequences that overlap the −35 sequence. This results in the stimulation of RP_C formation and in an increase in the rate of isomerization from RP_C to RP_O. It is thus likely that: (i) ω₂ increases the local concentration of both proteins, leading to a ternary ω₂·P_ω·RNAP-σ^A complex; (ii) this ternary complex facilitates the rate of isomerization from RP_C to RP_O and increases P_ω dependent mRNA synthesis; and (iii) δ₂ may act as a co-activator by increasing the half-life of ω₂·P_ω DNA complexes (see 13).

With stoichiometric concentrations of ω₂ full operator occupancy was achieved, and a different outcome was observed. Under these conditions, ω₂ assembles into a left-handed matrix that wraps around right-handed, straight P_ω DNA (see Figure 1B). This assembly makes P_ω DNA accessible to RNAP-σ^A (RP_C formation), while simultaneously inhibiting isomerization to RP_O (transcriptional repression) (Figure 4, lanes 17–18). In vitro, we observed that δ₂ contributed to P_δ repression (Figure 3B), and the presence of both ω₂ and δ₂ transcriptional repression of P_δ and P_ω was elevated in vivo with respect to wt ω₂ alone (Supplementary Figure S7B, data not shown). When stoichiometric concentrations of ω₂ were added to preformed RP_O complexes, a moderate effect on P_ω utilization was observed (see Figures 3 and 4). This suggested that RNAP-σ^A transcription was influenced by ω₂ that was bound to its cognate promoters. In sum, we describe here a previously uncharacterized mechanism of transcription regulation in bacteria belonging to the phylum Firmicutes.

Several different models can be considered in order to explain the specific transcriptional repression resulting from the full occupancy of the operator sequences by ω₂: (i) relocation of RNAP-σ^A to a position unfavourable for efficient RP_O formation; (ii) ‘locking’ RNAP-σ^A into a conformation unfavourable for RP_O formation; (iii) blocking the interaction between the β’ NTD and the DNA, which may be an essential step for RP_O formation; and (iv) inhibition of RNAP-σ^A mediated transcription by hindering any putative upstream element. We favour the first model because RP_C formation is stimulated by the interaction between ω₂ and RNAP-σ^A, while the subsequent isomerization step producing stable RP_O (through the unstable intermediate, RP_I) becomes inhibited. Also consistent with this model is the fact that ω₂ establishes interactions with operator sites when RNAP-σ^A is already bound to P_ω DNA. Meanwhile, ω₂ fails to inhibit pre-formed RP_O, Also in line with this model is the observation that, in the presence of limiting amounts of ω₂ bound to P_ω DNA, RNAP-σ^A moves onto P_ω DNA (Supplementary Figure S2A), and that while under saturating concentrations of ω₂, RNAP-σ^A moves off of P_ω DNA (Supplementary Figure S2B).

The interplay between the ω₂ regulator and the β’ subunit of RNAP holoenzyme

The work presented here establishes that ω₂ is a global regulator of plasmid biology through its effect on replication, faithful partitioning and better-than random segregation (see Introduction). Importantly, ω₂ represents an exception to the accepted prokaryotic transcription regulation paradigm, which asserts that, there are proteins that can act as either activators or repressors, but that the same protein cannot act as both. Since ω₂ regulates the expression of plasmid encoded genes that are harboured in different Firmicutes bacteria, it presumably recognizes the RNAP β’ NTD of all of them. This recognition might be limited to the first 316 amino acids, the length of the shortest polypeptide of the β’ subunit that binds ω₂ (Figure 5B). This would not be surprising because the β’ subunits of all Firmicutes share a high degree of sequence identity. Consistent with this, ω₂ does not regulate transcription of a genetically distant bacterium (e.g. E. coli) (16). These facts imply that the regions of the Firmicutes β’ subunits that show a significant degree of sequence divergence are not involved in ω₂ binding, such as residues 124–165 and 178–208 (22% and 13%, respectively). Presumably, residues 260–271, which match almost perfectly between the β’ subunits of E. coli and B. subtilis, is a region also not involved in ω₂ binding.

Few proteins have been observed to interact with the RNAP β’ subunit. Most of these are encoded by proteobacterial phages. The phage Xp10-p7 factor interacts with the first 10 residues of the NTD of β’ (44). The Mu-C protein binds to part of region F (b7) (45). T7-Gp2 recognizes part of the jaw (b9-b10) and σ 1.1 domains (46,47). And lastly, N4-SSB interacts with part of region H (b11) at the CTD (48). These regulators do not share a specific target domain and have different modes of action: N4-SSB and Mu-C specifically act as transcription activators (45,48), while T7-Gp2 and Xp10-p7 are repressors (44,46). In contrast, ω₂ has a dual function (Figure 3B). Of these regulators, only T7-Gp2, Xp10-p7, and ω₂ act during the early stages of RNAP-σ isomerization. Meanwhile, T7-Gp2 and Xp10-p7 directly interact with RNAP-σ rather than binding to P DNA as does ω₂ (44,46,47).

Biological implication of ω₂-mediated transcription regulation

A growing number of plasmid-encoded genetic determinants for resistance to diverse antimicrobials among streptococci, enterococci and staphylococci has been shown to be regulated by ω-like cofactors. They act either as part of the ωϵζ operon or on their own as part of the ω cassette (ω or ω2 genes) (1). In conjunction with δ₂ and RNAP-σ, the biological role of ω₂ as a dual regulator is to control vital plasmid functions in Firmicutes. It corrects the downward fluctuations in plasmid copy number through regulation of the synthesis of CopS (also termed CopF, CopR). CopS is a repressor of the initiator RepS (also termed RepE, RepR) protein. ω₂ also controls the synthesis of the toxin-antitoxin module, which in turn restricts the survival of plasmid-free segregants. The ω₂ protein mediates the synthesis of the partition system by regulating the expression of δ₂ and ω₂. Protein ω₂2 manages the expression of the ermB gene (4,16,49). With the help of δ₂ (the ParA ATPase), the ω₂ (the ParB centromere binding) protein also safeguards plasmid faithful segregation via the ParAB system. The regulation of Firmicutes RNAP-σ^A by the ω cassette is a newly characterized mechanism through which bacterial transcription of a large number of antibiotic resistance genes is regulated.

SUPPLEMENTARY DATA

Supplementary Data are available at NAR Online.

FUNDING

Funding for open access charge: Dirección General de Investigación-Ministerio de Economía y Competitividad [BFU2012-39879-C02-01 to J.C.A.].

Conflict of interest statement. None declared.