Report

Opening and Closing of the Bacterial RNA Polymerase Clamp

See allHide authors and affiliations

Science  03 Aug 2012:
Vol. 337, Issue 6094, pp. 591-595
DOI: 10.1126/science.1218716

Clamping Down

Crystal structures of RNA polymerase show that a “clamp” region which surrounds the DNA binding site can adopt conformations ranging from a closed to an open state. Chakraborty et al. (p. 591) used single-molecule fluorescence energy transfer experiments to detect the clamp's conformational changes in solution during the transcription cycle. The results support a model in which a clamp opening allows DNA to be loaded into the active-center cleft and unwound. Direct interactions with DNA likely trigger clamp closure upon formation of a catalytically competent transcription initiation complex.

Abstract

Using single-molecule fluorescence resonance energy transfer, we have defined bacterial RNA polymerase (RNAP) clamp conformation at each step in transcription initiation and elongation. We find that the clamp predominantly is open in free RNAP and early intermediates in transcription initiation but closes upon formation of a catalytically competent transcription initiation complex and remains closed during initial transcription and transcription elongation. We show that four RNAP inhibitors interfere with clamp opening. We propose that clamp opening allows DNA to be loaded into and unwound in the RNAP active-center cleft, that DNA loading and unwinding trigger clamp closure, and that clamp closure accounts for the high stability of initiation complexes and the high stability and processivity of elongation complexes.

RNA polymerase (RNAP) has a structure that resembles a crab claw, with two “pincers” surrounding a cleft that contains the RNAP active center and serves as the binding site for DNA (Fig. 1A) (14). RNAP is a multisubunit protein. The largest subunit (β′ in bacterial RNAP) forms one pincer, termed the “clamp.” The second-largest subunit (β in bacterial RNAP) forms the other pincer. Crystal structures of RNAP in different crystal contexts indicate that the RNAP clamp can adopt different conformational states, ranging from an open state to a closed state (Fig. 1A) (19). The open and closed states differ by a ≥20° swinging motion of the clamp about a hinge region, referred to as the “switch region,” located at the base of the clamp, and by a ≥20 Å displacement of residues at the tip of the clamp. It has been hypothesized that the RNAP clamp adopts different conformational states not only in crystals, but also in solution, and that clamp conformational dynamics is important for function.

Fig. 1

Determination of RNAP clamp conformation in solution. (A) Measurement of smFRET between fluorescent probes incorporated at the tips of the RNAP β′ pincer (clamp) and the RNAP β pincer. Open (red), partly closed (yellow), and closed (green) RNAP clamp conformational states are as observed in crystal structures (Protein Data Bank accessions codes 1I3Q, 1HQM, and 1I6H). σ and the β′ nonconserved domain are omitted for clarity in this and subsequent figures. (B) Incorporation of fluorescent probes at the tips of the RNAP β′ pincer (clamp) and the RNAP β pincer, by unnatural amino acid mutagenesis to incorporate 4-azidophenylalanine at sites of interest in β′ and β, followed by Staudinger ligation to incorporate fluorescent probes at 4-azidophenylalanines in β′ and β, followed by in vitro reconstitution of RNAP from labeled β′ and β and unlabeled α* (covalently linked α-N-terminal–domain dimer) and ω (10). Plasmids, genes, and proteins are shown as ovals, open bars, and closed bars, respectively. (C) Relation between smFRET efficiencies, E, and RNAP clamp conformational states (10). The red boxes denote the open (model 4; red), closed (model 11; green), and collapsed (model 15; blue) clamp states observed in this work for RNAP-σ70 holoenzyme (Fig. 2).

In this work, we used single-molecule fluorescence resonance energy transfer (smFRET) (1015) to measure bacterial RNAP clamp conformation in solution at each step in transcription initiation and elongation. We monitored distances between a fluorescent probe, serving as donor, incorporated at the tip of the clamp and a fluorescent probe, serving as acceptor, incorporated at the tip of the β pincer (Fig. 1A).

To incorporate probes site-specifically, we used a procedure comprising unnatural amino acid mutagenesis (16), Staudinger ligation (17, 18), and RNAP reconstitution (Fig. 1B and figs. S1 to S3) (19). The procedure involved (i) preparation of β′ and β subunits containing 4-azidophenylalanine at the sites of interest (accomplished by expressing engineered genes containing nonsense codons at sites of interest using cells that contained an engineered suppressor–transfer RNA (tRNA)/aminoacyl-tRNA-synthase pair in media that contained 4-azidophenylalanine) (fig. S2A); (ii) incorporation of the fluorescent probes Cy3B (donor) and Alexa647 (acceptor) by azide-specific chemical modification (accomplished by Staudinger ligation using phosphine derivatives of fluorescent probes) (fig. S2, B and C); and (iii) in vitro reconstitution of RNAP core (β′βαIαIIω) or RNAP holoenzyme (RNAP core plus specificity factor σ).

To monitor distances between fluorescent probes, we used confocal optical microscopy with alternating laser excitation (ALEX) (1115) to quantify smFRET in single molecules of RNAP freely diffusing in solution (figs. S4 to S5). ALEX allows filtering of data to consider only molecules that contain both the donor and the acceptor, which eliminates complications due to incompletely labeled and incompletely assembled complexes (1115). The results provide equilibrium population distributions of donor-acceptor smFRET efficiencies, E, and enable the calculation of donor-acceptor distances, R (1115). To relate smFRET results to RNAP clamp conformations, we compared observed smFRET efficiencies, E, to calculated smFRET efficiencies, Ecalc, for each of a set of 18 structural models with RNAP clamp conformations ranging from fully open to fully closed in 2° increments (Fig. 1C and figs. S6 and S7).

Transcription initiation and transcription elongation are multistep reactions (20): (i) RNAP binds to promoter DNA, yielding an RNAP-promoter closed complex, in which DNA is double-stranded and is not yet loaded into the RNAP active-center cleft [RPc; where “closed” refers to the fact that the DNA is double-stranded (dsDNA)]; (ii) RNAP loads DNA into, and unwinds DNA in, the RNAP active-center cleft, yielding an RNAP-promoter open complex [RPo; where “open” refers to the fact the DNA is partly single-stranded (ssDNA)]; (iii) RNAP synthesizes the first ~10 nucleotides (nt) of RNA using a “scrunching” mechanism, in which RNAP remains stationary on promoter DNA and pulls downstream DNA into itself (14, 21), as an RNAP-promoter initial transcribing complex (RPitc); and (iv) RNAP escapes the promoter and synthesizes the rest of the RNA using a “stepping” mechanism, in which RNAP translocates relative to DNA (22), as a transcription elongation complex (RDe).

In a first set of experiments, we examined RNAP clamp conformation in free RNAP holoenzyme containing the principle σ factor, σ70, and in free RNAP core. We observed that RNAP-σ70 holoenzyme exhibited a broad, multimodal distribution of smFRET efficiencies (Fig. 2A, first panel). The distribution could be fitted with three Gaussians, corresponding to three subpopulations: (i) a major subpopulation with mean E = 0.15 and mean R = 81 Å, corresponding to an open clamp state; (ii) a minor subpopulation with mean E = 0.28 and mean R = 69 Å corresponding to a closed clamp state in which the clamp is rotated inward by ~14°; and (iii) a minor subpopulation with mean E = 0.40 and mean R = 64 Å, corresponding to one or more “collapsed” clamp state, more closed than any RNAP crystal structure to date, in which the clamp is rotated inward by ~22° (Fig. 1C, red boxes; Fig. 2A, first panel; and table S1). The observed open, closed, and collapsed clamp states have RNAP active-center–cleft solvent-accessible widths of, respectively, ~20 Å (sufficient to accommodate dsDNA), ~12 Å (sufficient to accommodate ssDNA, but insufficient to accommodate dsDNA), and ~8 Å (insufficient to accommodate either dsDNA or ssDNA). We conclude that the RNAP clamp can adopt open, closed, and collapsed states in solution. We further conclude that the open state, the dimensions of which allow loading of dsDNA into the active-center cleft, is the predominant state in free RNAP-σ70 holoenzyme in solution. Identical results were obtained with RNAP core (Fig. 2D). We conclude that clamp conformational dynamics are an intrinsic property of RNAP core and are not dependent on association of RNAP core with σ70.

Fig. 2

RNAP clamp conformation in σ70-dependent transcription initiation and elongation. Histograms and Gaussian fits of observed donor-acceptor smFRET efficiencies, E (at left); mean E, mean R, and percentage, P, for each subpopulation (in inset at left); inferred structural states of the RNAP clamp (at right; colored as in Fig. 1C); and inferred extents of closure of the RNAP clamp (at far right; in degrees ±SEM relative to the open state). The vertical red lines denote mean E of the open, closed, and collapsed states in RNAP-σ70 holoenzyme. (A) RNAP clamp conformation in RNAP holoenzyme, RPo, RPitc (4 nt of RNA), and RDe (14 nt of RNA). (B) Control three-color FRET experiments with third probe on σ70 (data filtered to include only molecules containing σ70). (C) Control three-color FRET experiments with third probe on DNA (data for RPo filtered to include only molecules containing DNA). (D) RNAP clamp conformation in RNAP core.

In the next set of experiments, we examined RNAP clamp conformation at each step in σ70-dependent transcription initiation and in transcription elongation. We started with a sample of RNAP-σ70 holoenzyme (Fig. 2A, first panel). Upon the addition of DNA and formation of RPo (fig. S1A), the distribution of smFRET efficiencies became narrow and unimodal, and only a subpopulation corresponding to a closed clamp state, with the clamp rotated inward by ~16°, was observed (Fig. 2A and figs. S7 to S10, second panels). We conclude that, after the loading of DNA into, and unwinding of DNA in, the RNAP active-center cleft to form RPo, the RNAP clamp is closed. We speculate that direct interactions between negatively charged DNA in the active-center cleft and the positively charged inner face of the clamp trigger clamp closure.

Upon the further addition of a nucleoside triphosphate (NTP) subset yielding RPitc (fig. S1A), no further change in the distribution of smFRET efficiencies was observed; only the subpopulation corresponding to a closed clamp state, with the clamp rotated inward by ~16°, was observed (Fig. 2A, third panel). Equivalent results were obtained for RPitc containing 2, 4, or 7 nt of RNA (figs. S8 and S9). We conclude that, upon the transition to RPitc—despite the synthesis of up to 7 nt of RNA through a scrunching mechanism that entails the unwinding and repositioning of up to 5 base pairs (bp) of DNA (14, 21)—no change in clamp conformation occurs.

Upon the further addition of an NTP subset yielding RDe (fig. S1A), no further change in distribution was observed; only a subpopulation corresponding to the closed clamp state, with the clamp rotated by ~16°, was observed (Fig. 2A, fourth panel). Equivalent results were obtained with RDe containing a 14-nt RNA product (which does not extend beyond the RNAP RNA exit channel) and RDe containing a 15-nt RNA product (which extends beyond the RNAP RNA exit channel) (figs. S8 and S9), and with RDe containing σ70 and not containing σ70 (figs. S8 and S9) (13). We conclude that, upon the transition to RDe—despite transition to a stepping mechanism of RNA synthesis (22), filling of the RNA exit channel, and optional release of σ70 (13)—no change in clamp conformation occurs.

Control experiments validate these observations. To rule out complications from complexes lacking σ70, we performed three-color ALEX experiments (15) analyzing complexes containing a third fluorescent probe, Alexa488, on σ70 and considering only molecules that contained all three fluorescent probes (Fig. 2B). To rule out complications from complexes lacking DNA, we performed analogous three-color ALEX experiments, analyzing complexes containing a third fluorescent probe, Alexa488, on DNA (Fig. 2C). To rule out complications due to RNAP-probe interactions, we performed experiments using reversed donor and acceptor probe sites, different probe sites, and/or different donors and acceptors (fig. S10).

In the next set of experiments, to assess RNAP clamp conformation in intermediates in formation of RPo, we analyzed transcription initiation by RNAP holoenzyme containing the alternate σ factor σ54 (Fig. 3). In contrast to RNAP-σ70 holoenzyme, which forms unstable RPc (20), RNAP-σ54 holoenzyme forms a stable RPc and undergoes isomerization from RPc to RPo only upon the binding of an AAA+ adenosine triphosphatase, such as NtrC1, and adenosine triphosphate (ATP) hydrolysis (Fig. 3A) (23). This enables trapping and analysis of RPc and of intermediates in the isomerization of RPc to RPo (Fig. 3A and fig. S11) (23). We analyzed clamp conformation in RNAP-σ54 holoenzyme, RPc (formed by adding DNA to RNAP-σ54 holoenzyme), RPc+NtrC1 (formed by adding NtrC1 to RPc), RPi1 (formed by adding the ground-state ATP analog ADP-BeFx to RPc+NtrC1), RPi2 (formed by adding the transition-state ATP analog ADP-AlFx to RPc+NtrC1) and RPo (formed by adding ATP to RPc+NtrC1) (Fig. 3A). RNAP-σ54 holoenzyme exhibited the same distribution of clamp states as RNAP-σ70 holoenzyme: a major subpopulation with an open clamp, a minor subpopulation with a closed clamp (clamp rotated inward by ~14°), and a minor subpopulation with a collapsed clamp (clamp rotated inward by ~24°)—consistent with the conclusion that clamp conformation in free RNAP is an intrinsic property of RNAP core (Fig. 3B, first panel). Upon the formation of RPc, RPc+NtrC1, RPi1, and RPi2, the open clamp state continued to predominate, the closed clamp state continued to be observed as a minor subpopulation, and the collapsed state disappeared (Fig. 3B, second through fifth panels). Upon the formation of RPo, the distribution narrowed, and only a closed clamp state, with the clamp rotated inward by ~12°, was observed (Fig. 3B, sixth panel). We conclude that the RNAP clamp predominantly is open in RPc and in all intermediates on the pathway of isomerization of RPc to RPo, and closes only upon loading of DNA into, and unwinding of DNA in, the RNAP active-center cleft to form RPo. These results support the conclusion that direct interactions between loaded, unwound DNA and the clamp trigger clamp closure.

Fig. 3

RNAP clamp conformation in σ54-dependent transcription initiation. (A) Intermediates in σ54-dependent transcription initiation (23). (B) RNAP clamp conformation in RNAP-σ54 holoenzyme, RPc, RPc+NtrC1, RPi1, RPi2, and RPo.

The antibiotics myxopyronin (Myx), corallopyronin (Cor), and ripostatin (Rip) inhibit the isomerization of RPc to RPo through interactions with the RNAP switch region, the hinge that mediates opening and closing of the RNAP clamp (2426). Bacteriophage T7 Gp2 inhibits the isomerization of RPc to RPo through interactions with a site within the RNAP active-center cleft, on or near the inner face of the clamp (27, 28). It has been hypothesized that these inhibitors function by preventing clamp opening and thereby preventing loading of DNA into, and unwinding of DNA in, the active-center cleft (“hinge-jamming” hypothesis) (24, 2628). We assessed effects of these inhibitors on RNAP clamp conformation in RNAP-σ70 holoenzyme in solution. We find that all four inhibitors depopulate the open clamp state, leaving only closed and collapsed clamp states (states in which the clamp is rotated inward by ~8° to ~26°) (Fig. 4, A and C). Analogous experiments assessing the effects of one of the four inhibitors, Myx, on σ54-dependent transcription initiation show that Myx also depopulates the open state in RNAP-σ54 holoenzyme and in RPc (Fig. 4C and fig. S12). In contrast, the inhibitors rifampin (Rif), streptolydigin (Stl), and CBR703 (CBR), which inhibit steps in transcription subsequent to isomerization of RPc to RPo (29), do not affect clamp conformation (Fig. 4, B and C). A Gp2 mutant defective in binding to RNAP (27) does not affect clamp conformation (fig. S13). We propose, consistent with the hinge-jamming hypothesis, that Myx, Cor, Rip, and Gp2 prevent clamp opening and inhibit the isomerization of RPc to RPo by preventing loading of DNA into, and unwinding of DNA in, the RNAP active-center cleft (Fig. 4C).

Fig. 4

Effects of inhibitors on RNAP clamp conformation. (A) Effects of inhibitors that block the isomerization of RPc to RPo: Myx, Cor, Rip, and Gp2. (B) Effects of inhibitors that block steps subsequent to the isomerization of RPc to RPo: Rif, Stl, and CBR. (C) Summary. The major clamp conformation at each step in transcription initiation and transcription elongation stage is shown in black. Minor clamp conformations are shown in gray. Myx, Cor, Rip, and Gp2, which inhibit the isomerization of RPc to RPo (vertical red line), inhibit clamp opening (slanted red lines); Rif, Stl, and CBR, which inhibit steps subsequent to the isomerization of RPc to RPo (vertical blue lines), do not inhibit clamp opening.

We infer that the RNAP clamp is predominantly, and likely obligatorily, open in free RNAP and RPc, but is closed—rotated inward by ~12 to ~16°—in RPo, RPitc, and RDe. We propose that clamp opening enables the loading of DNA into, and the unwinding of DNA in, the RNAP active-center cleft during the isomerization of RPc to RPo. We propose that clamp closure is induced or stabilized by direct interactions between DNA inside the active-center cleft and the clamp, and that clamp closure is responsible for the high stability of RPo and the high stability and processivity of RDe.

We speculate that the clamp may reopen in subsequent steps in the transcription cycle that are marked by low stability and low processivity, including transcription pausing and transcription termination. Consistent with this hypothesis, a crystal structure of RDe containing an RNA hairpin associated with pausing and termination shows an open clamp (7). Further consistent with this hypothesis, the transcription factors NusG and Spt4/5, which inhibit pausing and termination, bridge the clamp and the opposite RNAP pincer, in a manner likely to inhibit clamp opening (30).

Supplementary Materials

www.sciencemag.org/cgi/content/full/337/6094/591/DC1

Materials and Methods

Figs. S1 to S13

Tables S1 and S2

References (3157)

References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
  2. Acknowledgments: We thank P. Schultz and R. Mehl for plasmids and A. Reitmar for assistance with instrument set-up. This work was funded by NIH grants GM037554 to C.L. Brooks III, GM069937 to B.T.N., GM069709 to S.W., and GM41376 and AI072766 to R.H.E., and a Howard Hughes Medical Institute Investigatorship to R.H.E.
View Abstract

Navigate This Article