Single-molecule analysis reveals the mechanism of transcription activation in M. tuberculosis

See allHide authors and affiliations

Science Advances  23 May 2018:
Vol. 4, no. 5, eaao5498
DOI: 10.1126/sciadv.aao5498


The σ subunit of bacterial RNA polymerase (RNAP) controls recognition of the −10 and −35 promoter elements during transcription initiation. Free σ adopts a “closed,” or inactive, conformation incompatible with promoter binding. The conventional two-state model of σ activation proposes that binding to core RNAP induces formation of an “open,” active, σ conformation, which is optimal for promoter recognition. Using single-molecule Förster resonance energy transfer, we demonstrate that vegetative-type σ subunits exist in open and closed states even after binding to the RNAP core. As an extreme case, RNAP from Mycobacterium tuberculosis preferentially retains σ in the closed conformation, which is converted to the open conformation only upon binding by the activator protein RbpA and interaction with promoter DNA. These findings reveal that the conformational dynamics of the σ subunit in the RNAP holoenzyme is a target for regulation by transcription factors and plays a critical role in promoter recognition.


The multidrug-resistant forms of Mycobacterium tuberculosis (Mtb), the pathogen causing human tuberculosis, are a major public health problem worldwide. Development of new antituberculosis therapeutics requires knowledge of the specific regulatory mechanisms of gene expression, which allow Mtb to survive antibiotic treatment and thus to cause recurrent infection. The first step of gene expression, transcription initiation, is performed in bacteria by the multisubunit DNA-dependent RNA polymerase (RNAP), which is composed of the catalytic core (subunits 2αββ′ω) and the promoter specificity subunit σ. The σ subunit controls promoter recognition, DNA melting, and initiation of RNA synthesis. All bacteria contain at least one σ subunit, which belongs to the σ70 family, and “optional” alternative σ’s (1). Transcription initiation by the RNAP holoenzyme containing the σ70-family subunit is a spontaneous process driven by the interplay between σ, the RNAP core enzyme, and DNA. Free σ in solution adopts a compact “closed” conformation that is incompatible with promoter DNA binding due to steric occlusion of its promoter-binding domains, σ2 and σ4 (24). Ensemble luminescence resonance energy transfer (LRET) studies demonstrated that, during assembly of the RNAP holoenzyme, σ undergoes a core-induced conformational change involving repositioning of the σ2 and σ4 domains (57). As a result, σ adopts an “open” conformation, observed in the crystal structures of the RNAP holoenzyme, where domains σ2 and σ4 contact a coiled-coil region of the β′ subunit (β′CC) and the β subunit flap domain (β-Flap), respectively, and the distance between σ2 and σ4 domains matches the distance between the −10 and −35 promoter elements (8, 9). To initiate transcription, RNAP first forms an unstable “closed complex” (RPc) with the promoter, which isomerizes into a transcriptionally competent “open complex” (RPo) (10). Recent studies demonstrated that RNAP of Mtb (MtbRNAP) differs from RNAP of the paradigm bacteria, Escherichia coli (EcoRNAP), because it requires an accessory RNAP-binding protein, RbpA, to form stable RPo (1113). The nature of this factor dependency remains unknown. RbpA in Mtb is an essential protein that interacts with σ2 domain and stimulates transcription initiation by MtbRNAPs containing either the principal σA or the stress-response σB subunit (12, 14). It is implicated in control of the pathogen physiological states and likely in the development of latent tuberculosis (15, 16).

Here, we used single-molecule Förster resonance energy transfer (smFRET) to explore the molecular basis for RbpA requirement by MtbRNAP. We compared conformational states of the principal σ subunit of E. coli, σ70, and the σB subunit of Mtb in solution, upon RNAP holoenzyme assembly and promoter binding. σB differs from σ70 by the lack of the N-terminal autoinhibition domain σ1.1 and the lack of the nonconserved domain (NCD) (Fig. 1A). The promoter-binding domains of both σ subunits display high structural similarity and can recognize identical promoter consensus elements (12). We show that σ70 and σB subunits exist in open and closed states even after binding to their lineage-specific RNAP core. However, Mtb σB remains preferentially in the closed conformation unless it is stabilized in the open conformation by binding of RbpA and promoter DNA.

Fig. 1 The σ70 subunit in the RNAP holoenzyme exhibits conformational heterogeneity.

(A) Scheme of the σ subunits. Positions of the Cys substitutions are underlined. (B) Structure of the EcoRNAP-σ70 holoenzyme [Protein Data Bank (PDB) code: 4IGC (35)] and (C) its complex with us-fork. The subunits of the RNAP core are shown as molecular surfaces, and the σ subunit is shown as ribbons in cyan. The Cα atoms of the σ subunit residues labeled by fluorophores are shown as spheres in green (σ domain 4) and magenta (σ domain 2). The distance between the Cα atoms is indicated. (D) Sequence of the us-fork DNA. The −10 and −35 promoter elements are underlined and shaded. (E) EPR histograms for free σ70, EcoRNAP-σ70 holoenzyme, and its complex with us-fork (+us-fork). Two conformations of σ, corresponding to the open and closed states, are shown schematically on the top. The black thick lines show Gaussian fits of smFRET efficiencies for individual subpopulations, and the dashed lines represent the sum of Gaussians for the overall population. Mean peak EPR and R are shown on the right.


The σ70 subunit in the EcoRNAP holoenzyme undergoes conformational fluctuations

To probe σ conformation, we monitored distances between donor and acceptor fluorescent probes introduced in domains σ2 and σ4. The σ70 derivative with cysteines (Cys) at positions 442 and 579 and the σB derivative with Cys at positions 151 and 292 (corresponding to positions 440 and 581 in σ70, respectively) were randomly labeled with DY-547 (donor) and DY-647 (acceptor) fluorescent probes (Fig. 1, A and B, and fig. S1). Previous studies (57) and control experiments demonstrated that the modifications did not significantly affect σ activities such as holoenzyme assembly, promoter binding, and promoter melting but displayed quantitative defects in run-off transcription (fig. S2). Note that we performed all biochemical assays [transcription, KMnO4 probing, and electrophoretic mobility shift assay (EMSA)] in the hundred nanomolar protein concentrations range, with a threefold excess of the σ subunit and RbpA over the RNAP core.

To perform the smFRET measurements, we used confocal optical microscopy with pulsed interleaved laser excitation and multiparameter fluorescence detection (PIE-MFD) (17, 18). This setup allowed monitoring distances between fluorescent probes in single σ molecules diffusing in solution, either free or in complex with ligands (RbpA, RNAP core, and/or DNA). The smFRET measurements, which necessitate low picomolar concentrations of the labeled σ subunit, were performed with the ~104-fold excess of the core RNAP and RbpA over σ to ensure maximum efficiency of the holoenzyme formation. Only molecules labeled with both donor and acceptor were selected for calculation of their apparent and corrected FRET efficiencies (EPR and E) and donor-acceptor distances (R) (figs. S1E and S4).

First, we compared donor-acceptor distances in the free σ70 subunit with that in the EcoRNAP-σ70 holoenzyme (Fig. 1E). Free σ70 displayed asymmetric distribution of EPR values that can be fitted with two overlapping Gaussians. The main peak (EPR = 0.8) corresponds to the donor-acceptor distance, R ~ 42 Å, which is in good accordance with those (36 to 42 Å) obtained in ensemble LRET experiments (table S1) and supports the view that free σ70 preferentially adopts a compact, closed, conformation (4, 5). However, the presence of the second peak (EPR = 0.61) indicates that free σ70 undergoes conformational fluctuations, moving apart (opening) and moving together (closing) of the σ2 and σ4 domains, on a time scale longer than milliseconds (our observation time).

Binding of σ70 to the EcoRNAP core is expected to induce an ~20 Å increase in the donor-acceptor distance (table S1) (57). After the addition of the EcoRNAP core, σ70 exhibited a broad distribution of FRET efficiencies that can be fitted to three Gaussians corresponding to three subpopulations of molecules. Two subpopulations, with EPR = 0.62 and 0.36, correspond to open conformations of σ70 with a 10 and 25 Å increase in interdomain distance, respectively. However, 40% of σ70 molecules still adopted a closed conformation (EPR = 0.81) despite being bound to the EcoRNAP core, as demonstrated by species-specific fluorescence correlation spectroscopy (ssFCS) (fig. S5C) (19). We conclude that the open conformation of σ70 in the EcoRNAP holoenzyme is not stable and remains in equilibrium with the closed conformation.

Promoter binding should stabilize the open conformation of the σ subunit (Fig. 1C). We thus tested EcoRNAP-σ70 interaction with the synthetic fork junction DNA template (us-fork), derived from the sigAP promoter (Fig. 1D) for which RNAP can efficiently form RPo-like complexes even at suboptimal temperatures (22°C in our case) (20). Upon us-fork binding, the fraction of molecules at EPR = 0.6 (R = 54 Å) is increased to 71%, which suggests that interaction with −10 and −35 elements of the promoter stabilizes σ70 in the open conformation. Thus, in the promoter-bound RNAP-σ70 holoenzyme, the donor-acceptor distance was found to be 11 Å greater than that in σ70 adopting a closed conformation (table S2). The minor high FRET (EPR = 0.83; ~11% of molecules) and low FRET subpopulations (EPR = 0.36; ~18% of molecules), still observed in the presence of DNA, likely correspond to the unbound EcoRNAP holoenzyme.

The σB subunit in the MtbRNAP holoenzyme retains closed conformation

In the next set of experiments, to explore whether this unexpected conformational heterogeneity observed in the E. coli RNAP holoenzyme is conserved in Mtb, we monitored conformational changes in the σB subunit (Fig. 2). Free σB exhibited a bimodal distribution of EPR values similar to the one observed for free σ70 with EPR = 0.83 (72% of molecules) and 0.67 (28% of molecules) (Fig. 2A). Addition of RbpA, which can bind free σB in solution (12, 14, 21), resulted in a strong reduction of the peak at EPR = 0.67, indicating that RbpA binding can restrain σB conformational dynamics. Surprisingly, binding of σB to the MtbRNAP core had little effect on EPR distribution (Fig. 2A). Only a minor subpopulation of molecules (20%) displayed the FRET efficiency expected for the open σB conformation (EPR = 0.47). The ssFCS analysis confirmed that the major subpopulation, at EPR = 0.81, corresponds to σB bound to the MtbRNAP core (fig. S5, A and B). Thus, we conclude that, in the MtbRNAP holoenzyme, most of the σB molecules remained in a closed conformation characteristic of free σB. Addition of RbpA to MtbRNAP-σB holoenzyme resulted in a strong increase of the peak at EPR = 0.47 (Fig. 2A), showing that binding of RbpA to MtbRNAP-σB partially stabilizes an open conformation of σB with a donor-acceptor distance that is ~24 Å greater than that observed in free σB. The fact that a subpopulation of MtbRNAP-σB containing σB in a closed state (EPR = 0.8) was still observed in the presence of RbpA likely reflects an equilibrium between RbpA-bound and free MtbRNAPs. The calculated donor-acceptor distance in MtbRNAP-σB (R = ~78 Å) was in the range of the distances observed in structures of the RNAP holoenzyme (table S3). Because MtbRNAP was poorly active in transcription initiation without RbpA (11, 12), we supposed that MtbRNAP activity depends on the ability of σB to adopt an open state, with the distance between σ2 and σ4 domains optimal for promoter recognition. To test this hypothesis, we used a chimeric holoenzyme, reconstituted from σB and EcoRNAP core, which has been shown to form a stable promoter complex without RbpA (12). Accordingly, smFRET analysis performed on EcoRNAP-σB (Fig. 2C) demonstrated that most σB molecules adopt an open conformation with EPR = 0.52. A minor subpopulation of σB molecules (~15%) displayed high FRET efficiency (EPR = 0.86) that likely corresponds to free σB or a residual fraction of σB retaining the closed conformation in the holoenzyme. Thus, we conclude that the chimeric enzyme was active in promoter binding because the EcoRNAP core is able to stabilize the open conformation of σB, whereas MtbRNAP lacks this potency in the absence of RbpA.

Fig. 2 RbpA is required for “opening” of σB in the MtbRNAP holoenzyme.

(A) EPR histograms for free σB, RbpA-σB complex, and MtbRNAP-σB and RbpA-MtbRNAP-σB complex. (B) Structure of Mycobacterium smegmatis RNAP in complex with RbpA [PDB code: 5TW1 (22)]. Labeling as in Fig. 1B. Residue numbering corresponds to the σB subunit. (C) EPR histogram of chimeric EcoRNAP-σB holoenzyme.

RbpA and promoter DNA stabilize the open conformation of σB

Next, we examined the impact of us-fork binding on σB conformation in the MtbRNAP holoenzyme (Fig. 3). The MtbRNAP holoenzyme containing labeled σB (Dy-σB) was tested for its ability to bind unlabeled us-fork in the EMSA. MtbRNAP containing unlabeled wild-type σB was used as a control in EMSA with the Cy3-labeled us-fork (Fig. 3B). Holoenzymes containing either unlabeled or labeled σB formed detectable stable competitor-resistant complexes with us-fork. Complex formation was stimulated by RbpA ≈ 2.5- to 5-fold (for unlabeled and labeled σB, respectively; Fig. 3C), suggesting that RbpA stabilizes RPo but is not essential for promoter binding in conditions when promoter melting is bypassed. On the contrary, RbpA was essential for stable complex formation on the full-length homoduplex DNA sigAP promoter (12). We conclude that, in the absence of RbpA, MtbRNAP forms unstable promoter complexes because σB tends to adopt a closed conformation inappropriate for binding to −10 and −35 elements. Accordingly, chimeric EcoRNAP-σB, which contains σB in open conformation (Fig. 2C), formed competitor-resistant complexes with us-fork as efficiently as the RbpA-MtbRNAP-σB complex (Fig. 3B).

Fig. 3 Interaction with promoter DNA stabilizes the open conformation of the σB subunit.

(A) Structure of M. smegmatis RNAP in complex with RbpA and us-fork (blue, template strand; red, nontemplate strand) (PDB code: 5TW1). (B) Native gel electrophoresis analysis of the promoter complex formation between labeled (+Cy3) and unlabeled (−Cy3) us-fork DNA and RNAP holoenzymes containing σB [wild-type (WT)] and donor-acceptor–labeled σB (Dy). (C) Quantification of the gel shown in (B). Values were normalized to that obtained for MtbRNAP-σB in the presence of RbpA. (D) EPR histograms for MtbRNAP-σB in complex with us-fork without (+us-fork) or with RbpA (+us-fork + RbpA). (E) EPR histograms for MtbRNAP-σB in complex with sigAP promoter without (+sigAP) or with RbpA (+RbpA).

Binding of us-fork to MtbRNAP-σB resulted in the appearance of the subpopulation of σB in an open conformation (EPR = 0.41) that became prevalent (Fig. 3D). In agreement with the stabilization effect of RbpA in EMSA, its addition increased the fraction of σB in the open conformation at the expense of the closed state (EPR = 0.81) (Fig. 3D). Control experiments performed on the us-fork labeled with the black hole quencher (BHQ2) (fig. S6) demonstrated that only the subpopulation at EPR = 0.41 corresponds to DNA-bound MtbRNAP-σB, whereas the subpopulation at EPR = 0.81 corresponds to unbound MtbRNAP-σB species. The difference in EPR values between promoter-bound MtbRNAP-σB (EPR = 0.41) and MtbRNAP-σB in complex with RbpA (EPR = 0.47) (Fig. 2A) likely reflects a shift in the mean dye positions due to steric clash with DNA (fig. S4, D and E). We conclude that RbpA and us-fork act cooperatively to stabilize σB in open conformation, as found in RNAP structures. RbpA, which inserts between σ2 and σ4 domains (Fig. 3A) (22), stabilizes promoter complexes by preventing the collapse of σB to the closed state and additionally bridging us-fork and RNAP through protein-DNA contacts (13). Interaction of σB with promoter elements also hinders transition to the closed state.

Finally, we explored whether binding of MtbRNAP to the 106–base pair (bp) sigAP promoter homoduplex DNA also favors the formation of the open σB conformation in the MtbRNAP holoenzyme (Fig. 3E). Unlike the us-fork, the sigAP promoter had no effect on the distribution of EPR values, which remained similar to that observed for free holoenzyme. Thus, we conclude that MtbRNAP was unable to form a stable complex with the sigAP promoter, in agreement with the published EMSA and deoxyribonuclease I footprinting results (12). We only observed the low FRET peak with EPR = 0.41 in the presence of RbpA (Fig. 3E, panel +RbpA). This peak is a hallmark of the open σB conformation found in the RPo-like complex of MtbRNAP with us-fork (Fig. 3D). The ratio between high FRET (unbound MtbRNAP) and low FRET (bound MtbRNAP) subpopulations was shifted toward high FRET, as compared to the distribution of EPR values observed in the presence of us-fork. The discrepancy between the results obtained with us-fork and the promoter fragment is expected, because smFRET data acquisition was performed at 22°C, a temperature that is suboptimal for promoter melting and RPo formation (10). Thus, the equilibrium was shifted toward unstable closed promoter complexes (RPc) and dissociation of MtbRNAP from the promoter. The us-fork complex formation does not need the DNA melting step and is therefore largely temperature-independent (20).

It has been shown that binding of RNAP to the “extended −10” class promoter does not require interaction between the σ4 and the −35 element (23) and does not require an open conformation of σ (24). Our model predicts that MtbRNAP should form a stable complex with the extended −10 promoter in the absence of RbpA. To test this prediction, we used a derivative of the sigAP us-fork that harbors the extended −10 motif and lacks the −35 element (fig. S5A). The EMSA assay demonstrated that MtbRNAP efficiently binds the extended −10 us-fork in the absence of RbpA (Fig. 4, A and B). The experiment was performed using different orders of RbpA addition to test whether the activator can stabilize preformed MtbRNAP-DNA complexes. Addition of RbpA had no effect on binding of the MtbRNAP to the extended −10 us-fork. The complex of MtbRNAP with the wild-type sigAP us-fork was stabilized irrespective of the order of RbpA addition. To determine whether RbpA is also dispensable for transcription initiation at the extended −10 promoter, we performed a run-off transcription assay using the sigAP promoter derivative harboring the extended −10 motif (Fig. 4, C and D). In agreement with the result of EMSA, MtbRNAP was able to initiate transcription from the extended −10 promoter without RbpA. However, RbpA stimulates transcription even at the extended −10 promoter, likely through direct interaction with DNA as reported before (13, 22). RbpA activates MtbRNAP less efficiently when added to the reaction together with nucleotide triphosphates (NTPs) after the incubation of MtbRNAP with promoter DNA. This order-of-addition effect likely reflects slow kinetics of RPo formation on the sigAP promoter and not the interference between DNA and RbpA upon MtbRNAP binding. In support of this idea, no order-of-addition effect of RbpA was observed upon binding of MtbRNAP to us-fork. On the basis of the above results, we propose two pathways of RbpA loading to the promoter complex and MtbRNAP activation: direct binding to free MtbRNAP and binding to preformed promoter complex (see model in Fig. 5). However, we cannot exclude that activation involves dissociation of MtbRNAP from the promoter, its association with RbpA, and following rebinding in an active form.

Fig. 4 Effect of promoter architecture on MtbRNAP activity.

(A) Native gel electrophoresis analysis of the MtbRNAP complex formation with Cy3-labeled us-fork and us-fork harboring extended −10 element (Ext −10). RbpA was added either before (R + F) or after (F + R) the us-fork DNA. (B) Quantification of the gel shown in (A). Values were normalized to that obtained in the presence of RbpA for each template separately. (C) Run-off [32P]RNA products synthesized in multiple-round transcription from the wild-type sigAP promoter and its derivative harboring extended −10 element. (D) Quantification of the RNA products shown in (C). Values were normalized to that obtained for the wild-type sigAP promoter in the presence of RbpA. RbpA was added either before (R + P) or after (P + R) the promoter DNA.

Fig. 5 The “dynamic” model for the σ subunit activation and promoter recognition.

(A) Activator-independent mechanism used by E. coli RNAP. (B) Activator-dependent and activator-independent mechanisms used by MtbRNAP.

The results of our study establish that σ subunits in RNAP holoenzymes can adopt different conformational states and that the capacity of the RNAP core to stabilize the σ open state varies between bacterial species. The conformational flexibility of σ in RNAP likely originates from the conformational flexibility of the RNAP core, for example, movement of the β′ clamp and β-flap domains (2426) and/or partial disruption of the σ-core contacts. This conformational flexibility of the vegetative σ subunits may allow RNAP to adapt to a large spectrum of promoter architectures. RbpA and other RNAP-binding transcriptional factors can control RNAP activity by switching σ between open/closed states. In support of this model, the crystal structure of the Thermus thermophilus RNAP with the phage protein gp39 revealed the σA subunit in a closed conformation with a ~33 Å distance between domains σ2 and σ4 (24). We speculate that the intrinsic tendency of σ to adopt closed conformations plays a critical role in the process of RPo formation at the −10/−35-type consensus promoters but not at the extended −10-type promoters (Fig. 5). Because most Mtb promoters lack a pronounced −35 motif (27), stabilization of the σ open conformation by RbpA is required to compensate the weak interactions at the −35 element and favor RPo formation.


Proteins and DNA templates

Recombinant Mtb and E. coli RNAP core enzymes, σ subunits, and RbpA protein were expressed and purified as described (12, 28). Fork junction DNA templates were prepared by annealing of two oligonucleotides corresponding to template and nontemplate DNA strands of the sigAP promoter from Mtb (Fig. 1D) (29). The 106-bp sigAP promoter DNA fragment (spanning positions −56 to +50) was prepared as described (12). The sigAP derivative harboring the sequence 5′-TGTG-3′ at positions −17 to −14 and substitutions −30A→G; −31T→C; −33T→C was prepared by annealing of two synthetic oligonucleotides and amplification using sigAP-specific primers (12). The substitutions Thr151→Cys and Gly292→Cys in σB and the substitution Gln579→Cys in σ70 were constructed using a site-directed mutagenesis kit (Agilent Technologies). To construct double-Cys mutant of σ70, we used a pET28 plasmid encoding a single Cys derivative of σ70, Cys442, provided by R. Gourse (30).

Labeling of the σB and σ70 subunits

Protein labeling was performed as described by Kim et al. (31), with modifications. Purified σ subunit (500 μg) dissolved in 1 ml of reduction buffer [20 mM tris-HCl (pH 7.9), 200 mM NH4Cl, 2.5 mM EDTA, 0.2% Triton X-100, and 10 mM dithiothreitol (DTT)] was incubated for 2 hours at 4°C. Solid (NH4)2SO4 powder was added to the sample to 70% of saturation and gently agitated until dissolved. The precipitate was pelleted by centrifugation at 17,000g for 5 min at 4°C. The pellet was briefly washed with ice-cold labeling buffer A [0.1 M sodium phosphate (pH 6.8), 200 mM NaCl, 1 mM EDTA, 5% glycerol, and 0.2% Triton X-100] containing (NH4)2SO4 at 70% of saturation. For complete removal of DTT, the washing step was repeated twice. The pellet was dissolved in 200 μl of labeling buffer to a final protein concentration of ~2 mg/ml. The fluorescent dye derivatives DY547P1-maleimide (donor) and DY647P1-maleimide (acceptor), purchased from Dyomics GmbH, were dissolved in dimethylformamide, added to the protein sample at an 8:1 (dye/protein) molar ratio, and incubated at room temperature for 1 hour, followed by overnight incubation at 4°C. The reaction was quenched by adding 0.5% (v/v) of β-mercaptoethanol. To remove the excess of unincorporated dyes, the protein was passed through 10-ml Sephadex G-25 glass column equilibrated with storage buffer [20 mM tris-HCl (pH 7.9), 300 mM NaCl, 0.5 mM EDTA, 0.2 mM β-mercaptoethanol, and 0.05% Triton X-100]. Labeling efficiency was calculated using the following equation: (moles dye per mole of protein) = Adye/(εdye × Cprotein), where Adye is the absorbance value of the dye at the absorption maximum, εdye is the molar extinction coefficient of the dye at the absorption maximum (εDy547 = 150,000 M−1 cm−1; εDy647 = 250,000 M−1 cm−1), and Cprotein is the σ subunit concentration in mole per liter. Labeling efficiencies of the σ70 subunit for donor and acceptor were ~94 and 45%, respectively. Labeling efficiencies of the σB subunit for donor and acceptor were ~80 and ~ 40%, respectively.

Run-off transcription, KMnO4 probing, and EMSA

RNAP holoenzymes were reconstituted by mixing 100 nM core RNAP and 300 nM of the σ subunit in transcription buffer (TB) [20 mM tris-HCl (pH 7.9), 50 mM NaCl, 5 mM MgCl2, 0.5 mM DTT, 0.1 mM EDTA, and 5% glycerol] and then incubating for 10 min at 37°C. When indicated, RbpA was used at 300 nM. The reaction mixtures were incubated with sinP3 and sigAP promoter DNA templates (40 nM) at 37°C for 10 min. Transcription was initiated by the addition of adenosine 5′-triphosphate, guanosine 5′-triphosphate, and cytidine 5′-triphosphate (to a final concentration of 25 μM each), 3 μCi of [α-32P]uridine 5′-triphosphate (UTP) (PerkinElmer Life Sciences), and 1 μM UTP and was carried out for 5 min at 37°C. The transcription assays on sigAP were performed in the presence of 100 μM GpC primer (Eurogentec). For the order-of-addition experiments, RbpA was added to the reaction mixture either before DNA template or together with NTPs. Reactions were stopped by adding 1 volume of the stop solution (8 M urea and 20 mM EDTA). [32P]RNA products were resolved on a 24% polyacrylamide gel electrophoresis (PAGE)/7 M urea denaturing gel. The KMnO4 probing reactions were performed in 20-μl TB. RNAP holoenzymes were reconstituted by incubating 200 nM core RNAP and 600 nM of the σ subunit in TB for 10 min at 37°C. The sigAP promoter DNA fragment (40 nM) labeled by fluorescein at the 5′-end of template strand was added to the reactions and incubated for another 10 min at 37°C. The samples were treated with 5 mM KMnO4 for 60 s and processed as described before (32). DNA fragments were analyzed on 10% PAGE/7 M urea sequencing gel. For the EMSA experiments, the synthetic fork junction DNA template (50 nM) was incubated with 100 nM RNAP, 300 nM σ subunit, and 300 nM RbpA in TB for 30 min at 22°C in the presence of poly(deoxyinosinic-deoxycytidylic acid) (10 μg/ml). For the order-of-addition experiments, RbpA was added to the reaction mixture either before us-fork DNA or after 5 min of incubation with us-fork DNA. Samples were resolved on 6% native 0.5 × tris-borate EDTA–PAGE. All gels were scanned by Typhoon 9200 Imager (GE Healthcare) and quantified using ImageQuant software.

smFRET sample preparation

The donor-acceptor–labeled σB subunit at 25 pM and the donor-acceptor–labeled σ70 subunit at 75 pM were prepared in the filtered (0.1 μm) FRET buffer [20 mM tris-HCl (pH 7.9), 150 mM NaCl, 5 mM MgCl2, 5% glycerol, and bovine serum albumin (0.1 mg/ml)]. When indicated, ligands, MtbRNAP core, EcoRNAP core, and RbpA were added to a final concentration of 500 nM. To reconstitute EcoRNAP-σ70 and MtbRNAP-σB holoenzymes with and without RbpA, the samples were incubated at 37°C for 10 min. When indicated, us-fork DNA, either unlabeled or labeled with black hole quencher (BHQ2) at position −18 of the nontemplate DNA strand (Q-fork; fig. S6), was added to a final concentration of 500 nM and incubated for 10 min at room temperature. Complexes of MtbRNAP with the sigAP promoter were formed in the FRET buffer containing 50 mM NaCl as described above and incubated at 37°C for 10 min before data acquisition. The sigAP promoter was added to a final concentration of 500 nM.

smFRET measurements and data analysis

smFRET measurements were performed on homebuilt confocal PIE-MFD microscope, as described (17). Samples were placed in transparent nonbinding 384-well plates (Corning), and acquisitions were performed for at least 3 hours at room temperature (22°C). Collected data were analyzed with the “Software Package for Multiparameter Fluorescence Spectroscopy, Full Correlation and Multiparameter Fluorescence Imaging” developed in the C. A. M. Seidel laboratory ( (33). A single-molecule event was defined as a burst containing at least 30 photons. Further selection of the relevant species containing donor-acceptor was performed by selecting only the molecules satisfying the following criteria: (i) number of photons detected in the acceptor channel upon acceptor excitation >20, (ii) excited state lifetime of the acceptor upon acceptor excitation 0.55 ns < τA < 2.65 ns, and (iii) stoichiometry ratio S > 0.4 (fig. S1E) (34). Photobleached molecules were also eliminated using the methods described by Kudryavtsev et al. (18). For each photon burst, the apparent donor-acceptor FRET efficiencies (EPR) were calculated as EPR = (IA)/(ID + IA) (where IA and ID are intensities detected on the donor-acceptor channels) and plotted. For each identified subpopulation, donor excited lifetime analysis was used to calculate donor-acceptor smFRET efficiency (E) using the following equation: (E = 1 – τDAD), where τDA is the donor excited lifetime in the selected donor-acceptor subpopulation, and τD is the donor excited lifetime of the donor-only subpopulation. The EPR histograms were analyzed using the OriginPro software. Most of the data were fitted with a sum of two or three Gaussian functions without fixing their center position or widths. The EcoRNAP-σ70 data were fitted with a sum of three Gaussian functions by fixing the width of the high FRET peak to 0.2.

Molecular brightness and FCS analysis

To verify that the RNAP holoenzyme assembled with the labeled σ subunit remained monomeric under our experimental conditions, we calculated the molecular brightness (MB) of the acceptor dye (which does not depend on the FRET efficiency) for each specific smFRET subpopulation. The analysis demonstrated that MB of the RNAP holoenzyme was identical to the one measured for the free σ subunit, suggesting that no oligomerization occurs upon holoenzyme assembly (fig. S3). For the ssFCS analysis, donor-acceptor cross-correlation curves on selected subspecies were fitted using the following modelEmbedded Image(1)where N is the average number of molecules in the observation volume; tD is the diffusion time; ω0 and z0 are the 1/e2 radii of the laser focus volume perpendicular to and along the optical axis, respectively; and tc is the correlation time. FCS data were analyzed with the “Software Package for Multiparameter Fluorescence Spectroscopy, Full Correlation and Multiparameter Fluorescence Imaging” developed in the C. A. M. Seidel laboratory (33). No parameter was fixed during the fitting procedure.

smFRET distance calculations and molecular modeling

To determine accurate donor-acceptor distances (RDA), we took advantage of the available crystal structures of RNAPs from E. coli, Thermus sp., and M. smegmatis in complex with promoter DNA (8, 9, 22, 3537). In these structures, the mean distance between the donor and acceptor attachment points in σB (Cα atoms of Cys151 and Cys292) is 69 ± 0.6 Å and between the donor and acceptor attachment points in σ70 (Cα atoms of Cys442 and Cys579) is 60.7 ± 0.3 Å (tables S2 and S4). Real donor-acceptor distances should be different from the distances between dye attachment points due to the contribution of the linker, the fluorophore geometry, and constraints arising from steric clash of the dyes with protein and DNA. To take these factors into account, we modeled real donor-acceptor distance distribution in the crystal structures of the RNAP using FRET positioning and screening (FPS) software (fig. S4) (38). To model dye accessible volume (AV) clouds and dye position distributions for the σ70-Cys442-Cys579 derivative, we used atomic coordinates from the crystal structures of the E. coli RNAP holoenzyme (PDB code: 4IGC) and Thermus aquaticus RNAP in complex with us-fork DNA (PDB code: 4XLQ). To assess the effect of the us-fork DNA binding on RDA, a model of E. coli RNAP with us-fork was built by fitting 4XLQ into 4IGC in UCSF Chimera software (39). We observed that the calculated RDA distances were 4 to 7 Å shorter than the distances between Cα atoms of Cys442 and Cys579. Addition of the us-fork DNA to E. coli RNAP restricted dye mobility and resulted in ~2 Å increase in RDA. To model dye AV clouds and dye position distributions for the σB-Cys151-Cys292 derivative, we used coordinates from a crystal structure of M. smegmatis RNAP in complex with us-fork DNA (PDB code: 5TW1). To assess the effect of the us-fork DNA binding on RDA, us-fork DNA was removed from the structure. We observed that the calculated RDA distances were 8 to 14 Å longer than the distances between Cα atoms of Cys151 and Cys292. Addition of DNA restricted dye mobility and resulted in ~6 Å increase in RDA. The FPS-calculated RDA distances were used as reference in the following smFRET distance calculations. Using an experimental E value and RDA calculated for these complexes, we calculated an “experimental” R0 (R0) as followsEmbedded Image(2)

The R0 values (indicated by asterisks in fig. S4) were used for the calculation of R for the other species of unknown structure (tables S2 and S3).


Supplementary material for this article is available at

fig. S1. Incorporation of the fluorescent dyes into σ70 and σB subunits.

fig. S2. Activity of the σ70 and σB double-Cys mutants and their labeled derivatives.

fig. S3. Test of protein oligomerization using MB analysis on the labeled species.

fig. S4. Calculation of the donor-acceptor distances using FPS software.

fig. S5. Analysis of the RNAP holoenzyme assembly by the ssFCS.

fig. S6. Interaction of RNAP with an us-fork template.

table S1. Apparent distances (Rap) between domains σ2 and σ4 determined in ensemble LRET experiments.

table S2. Summary of the donor-acceptor distances for σ70 C442-C579.

table S3. Summary of the donor-acceptor distances for σB C151-C292.

Reference (40)

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: We thank A. Kapanidis for critical reading of the manuscript and helpful discussion. Funding: Access to the France-BioImaging infrastructure was supported by a grant from the French National Research Agency (ANR-10-INBS-04, “Investments for the future”) and by the “GIS-IBiSA: Groupement d’Intérêt Scientifique Infrastructures en Biologie Sante et Agronomie.” This work was supported by ANR-16-CE11-0025-01 “Mycomaster” to K.B., LABEX EpiGenMed (ANR-10-LABX-12-01, “Investments for the future”), fellowship from Infectiopole Sud to R.K.V., and Erasmus-Svaagata fellowship to A.S.P. Author contributions: K.B. conceived and supervised the project. K.B. and E.M. designed experiments. R.K.V., A.-M.C., A.S.P., and Z.M. performed experiments. K.B., E.M., A.-M.C., and R.K.V. performed data analysis. K.B. wrote the manuscript with contributions from E.M., R.K.V., and A.-M.C. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article