The nucleotide addition cycle of RNA polymerase is controlled by two molecular hinges in the Bridge Helix domain
© Weinzierl. 2010
Received: 20 September 2010
Accepted: 29 October 2010
Published: 29 October 2010
Skip to main content
© Weinzierl. 2010
Received: 20 September 2010
Accepted: 29 October 2010
Published: 29 October 2010
Cellular RNA polymerases (RNAPs) are complex molecular machines that combine catalysis with concerted conformational changes in the active center. Previous work showed that kinking of a hinge region near the C-terminus of the Bridge Helix (BH-HC) plays a critical role in controlling the catalytic rate.
Here, new evidence for the existence of an additional hinge region in the amino-terminal portion of the Bridge Helix domain (BH-HN) is presented. The nanomechanical properties of BH-HN emerge as a direct consequence of the highly conserved primary amino acid sequence. Mutations that are predicted to influence its flexibility cause corresponding changes in the rate of the nucleotide addition cycle (NAC). BH-HN displays functional properties that are distinct from BH-HC, suggesting that conformational changes in the Bridge Helix control the NAC via two independent mechanisms.
The properties of two distinct molecular hinges in the Bridge Helix of RNAP determine the functional contribution of this domain to key stages of the NAC by coordinating conformational changes in surrounding domains.
RNA polymerases (RNAPs) play a central role in the regulation of gene expression. Like the majority of the enzymes involved in fundamental biological information-processing functions (for example, replication, transcription, recombination, repair), RNAPs are probably best viewed as intricate molecular machines. The movement of nucleic acid substrates, coupled with various types of active site chemistries, requires a precisely orchestrated sequence of conformational changes of protein domains during the transcription cycle (for recent reviews see [1–4]).
The nanomechanical mechanisms guiding the structural rearrangements of domains within the active site are still very poorly understood. Thus far, models of the fundamental reaction catalyzed by RNAPs, the nucleotide addition cycle (NAC), have predominantly been derived from a series of crystal structures that contain RNAPs as apoenzymes (for example [5–9]), or complexed with various substrates and inhibitors (for example [10–15]). Such structures, revealing (among other features) pre- and post-translocation states of RNAPs, have provided the basis for various hypotheses concerning the molecular mechanism of the NAC [1–4, 16, 17]. There are, however, two potential shortcomings associated with such approaches. First, in order to 'freeze' the RNAPs in a crystallizable conformation, substrate analogs or inhibitors need to be chosen that stop the reaction cycle at a specific point. This may result in the adoption of 'off-pathway' conformations that do not represent normal enzyme states. A second, more fundamental, problem is that short-lived intermediate structures cannot be captured in crystals because they are thermodynamically or kinetically unstable. Yet, it is likely that an awareness of the existence and functional significance of such intermediates will be required to develop a deeper understanding of the mechanisms operating within molecular machines.
While the combination of structural observations and mutagenesis data clearly highlights the functional contribution of the C-terminal portion of the Bridge Helix towards controlling the rate of the NAC, the role of the N-terminal portion of the Bridge Helix has thus far remained enigmatic. The primary sequence of this region is exceptionally highly conserved during evolution; for example, the sequences of the N-terminal 15 amino acids are identical between the archaeon Methanocaldococcus jannaschii and humans, and differ by only a single residue from yeast (Figure 1B). Such a high degree of structural conservation over more than two billion years of evolution can be partially accounted for by the fact that the Bridge Helix N-terminus is tightly surrounded by other domains and may therefore be spatially and evolutionarily constrained due to the need to maintain an extensive network of protein-protein interactions (Figure 1A; Additional files 1, 2 and 3; [12, 15, 25, 26]). In apparent agreement with this view, all available X-ray structures of RNAPs show the N-terminal portion of the Bridge Helix in a rigidly α-helical conformation, suggesting the absence of significant conformational changes. For this reason, none of the current models of RNAP function consider the Bridge Helix N-terminus to play any dynamic role during the NAC [1–4, 16, 17].
New evidence presented here, based on a combination of high-throughput mutagenesis studies and molecular dynamics simulations, demonstrates that such a static view of the Bridge Helix N-terminus is untenable. The results show that this region contains a highly localized molecular hinge, and that the conformation of this site has a substantial influence on the rate of the NAC. In combination with the previously identified C-terminal hinge region, the data reinforces the overarching concept that the Bridge Helix plays a predominantly nanomechanical role during the translocation stage of the NAC by coordinating conformational changes in surrounding domains.
The N-terminus of the Bridge Helix of the RNAP from the euryarchaeon Methanocaldococcus jannaschii was dissected by high-throughput mutagenesis [19, 20]. In this automated approach, each amino acid within the target region (Figure 1B) was replaced with all other 19 residues to reveal local structural requirements. The mutants were then assayed in robotic promoter-independent transcription assays, which provide a consistent measure of the synthetic rate of the NAC and correlate directly with results obtained from a variety of promoter-dependent, abortive- and elongation transcription assays ([18–20]; S. Wiesler and ROJW, unpublished data).
Extensive database searches, covering completed genome sequences of a large variety of pro- and eukaryotic species, show that naturally occurring proline substitutions in the Bridge Helix primary sequences are exceptionally rare. The only known instances of proline residues occurring naturally anywhere in the N-terminal part of the Bridge Helix are found in the bacteria Orientia tsutsugamushi [28, 29], and certain isolates of Arcobacter butzleri  and Bacillus subtilis . In each case, the substituted position is precisely orthologous to mj A' M808 (Figure 2C). In the C-terminal half of the Bridge Helix, the highly divergent plant RNAPIV and RNAPV enzymes display a strong tendency for a proline residue at the position orthologous to archaeal S824 (Figure 2C; [32, 33]).
The presence of two distinct hinges in the Bridge Helix raises the question whether BH-HN and BH-HC are involved in the same mechanism during the NAC. Kinking of either of the two hinges will result in considerable local distortions, predicted to include a spatial redeployment of amino acid side chains and changes in the overall length, flexibility and general topology of the Bridge Helix domain. Kinking of BH-HN could result in altered interactions with adjacent domains, such as the βD-II, Link and F-Loop domains (Figure 1A; Additional files 1, 2 and 3), whereas hinge movements in BH-HC are expected to affect the position and/or mobility of the DNA-RNA hybrid and Trigger Loop conformation (Figure 1A; ).
In order to investigate this unexpected tolerance to the presence of two adjacent prolines in positions 823 and 824 in more detail, a complete substitution series of the residues around position mj A' S824-P was prepared, generating an assortment of systematic double-mutants (Figure 5A; mj A' Q823-X/S824-P and S824-P/G825-X; with X denoting 19 different variants). All substitutions N-terminal to S824-P (i.e. mj A' Q823-X/S824-P) displayed an almost completely invariant degree of extensive superactivity that was indistinguishable from the original S824-P mutant (Figure 5B). This result is remarkable because previous mutagenesis of Q823 revealed a broad spectrum of activity, ranging from substantial loss of function (Q823-C or I) to superactivity (Q823-D or E; ). It is evident that in the double-mutants the chemical nature of the side-chain residue in position 823 exerts no further functional influence, presumably because of the major distortion already caused by the proline substitution in position 824. Once such a gross structural alteration has occurred in S824-P, any additional changes in the adjacent N-terminal residue become structurally irrelevant.
Because Q823-P/S824-P displayed no loss of superactivity, the residue immediately N-terminal to the double-proline substitution was also permutated, resulting in variants containing three adjacent substitutions in the BH-HC hinge region (Figure 5A; mj A' A822-X/Q823-P/S824-P). The majority of these substitutions, including the triple proline mutant A822-P/Q823-P/S824-P (Figure 5C; Additional file 4B), still displayed clearly detectable superactivity, albeit at a slightly reduced level. The tolerance of the BH-HC hinge to radical mutagenesis, as previously observed in the two-amino acid deletion screen, is therefore also reflected in the unexpectedly high tolerance to multiple proline substitutions in that region. This geometric freedom is, however, also spatially limited: substitutions in positions C-terminal to S824-P (Figure 5A) were mostly inactive, indicating that despite the structural flexibility of the positions N-terminal to S824-P, the C-terminal positions are functionally constrained (Figure 5D; a similar trend is also apparent in the two amino acid deletion scan data [see Figure 4B]).
The results presented here reveal several new surprising insights, including compelling evidence for the existence of a molecular hinge region in the N-terminal portion of the Bridge Helix and evidence for an unexpectedly large degree of tolerance to radical structural changes in the C-terminal part of this domain. It is apparent that the Bridge Helix domain displays a much greater conformational freedom than anticipated from currently available X-ray structures of RNAPs. Few, if any, of the residues of the Bridge Helix appear to make any specific contribution to catalysis other than through defining the nanomechanical properties intrinsic to the α-helical structure. The implications for mechanistic models aimed at describing the NAC are manifold, ranging from a re-evaluation of the structural basis of the RNAP translocation mechanism, to highlighting the hitherto neglected role of highly conserved domains in the catalytic site, and to obtaining a better understanding of the evolutionary diversity of Bridge Helices in different organisms.
Currently, we have only a limited understanding of the forces acting on the Bridge Helix that could drive localized conformational changes. Attempts to model the full NAC using molecular dynamics studies are severely limited by the large size of multi-subunit RNAPs and the immense computational effort that would be required to simulate the molecular events expected to last from 10s to 100s of milliseconds for the extension of a nascent transcript by a single nucleotide (approximately 30 ms/rNTP incorporation under optimal in vivo conditions; for example, ). The study of the intrinsic structural properties of individual domains by fully atomistic computer simulations reveal, however, interesting nanomechanical properties that have functional implications for the RNAP translocation mechanism [50–52]. The Bridge Helix domain contains intrinsically unstable α-helical regions that undergo spontaneous kinking motions, even in the absence of externally applied forces (Figure 3A). At least two of these unstable regions correspond precisely to the biochemically-mapped BH-HN and BH-HC regions. Strategically-placed glycine residues, such as mj A' G809, G810 (for BH-HN) and G825 (for BH-HC) provide the structural basis for forming these molecular hinges, with surrounding residues determining additional kinking parameters, such as the likelihood of kinking and/or the half-life of the kink after its isomerization (Figure 2A; Heindl et al., unpublished observations). Interestingly, the simulations also highlight a potential third structurally labile region near the center of the Bridge Helix domain, spanning residues mj A' Q817 to R820. The relative sensitivity of this sequence to proline substitutions (Figure 2B) suggests, however, that this area of instability behaves functionally differently to BH-HN and BH-HC. It is possible that structural fluctuations in the central part of the Bridge Helix play a more dynamic role in supporting short-lived conformational changes that can either compensate for major structural rearrangements due to BH-HN and BH-HC kinking, or act as a store of 'fast' motions to lubricate the kinking of the hinge regions kinetically . Although such hypotheses are currently beyond experimental proof, it is interesting to note that several, structurally unexplained superactivity mutants map to this central sequence (for example, mj A' D816-N; Q817-S/T/C/K and V819-K;Figure 2A ), and that this region is also exceptionally tolerant to radical twisting of the α-helical axis induced by deletions of two-amino acid segments (Figure 4B). Furthermore, data from anisotropic network mode analysis suggests that rigid body movements of the clamp domain may exert forces onto the center of the Bridge Helix via the Switch domains, potentially linking transcription to a ratchet-like translocation mechanism (Additional file 5; ). The Bridge Helix thus appears to have evolved specific nanomechanical properties that result in the controlled and highly localized isomerization of its conformation in response to allosteric alterations in the surrounding protein domains and nucleic acid substrates.
The Bridge Helix N-terminus is tightly surrounded by other domains, such as the β-D loop II , the experimentally uncharacterized Link domain [25, 55], and the F-Loop . As evident from the exceptionally high degree of sequence conservation (Additional files 1A, 2A and 3A), each of these domains is likely to play key roles in the NAC. The β-D II domain is a loop-like structure that interacts extensively in a side-way interaction with the central part of the Bridge Helix, while simultaneously maintaining direct physical contact with the most recently incorporated nucleotide (i-1 position). The interaction between the β-D II domain and nascent transcript also creates an extended binding pocket for the rNTP (additional file 1B, C). Similarly, the highly conserved Link domain is strategically placed to interact with the Bridge Helix N-terminus, β-D II domain, nascent transcript (i-1 and i-2 positions) and the incoming rNTP (Additional file 2B). Finally, an N-terminal extension of the Bridge Helix, the F-Loop, forms an extensive cap-like structure that contacts the Link domain and the tip of the Trigger Loop (Additional file 3B; ). The differential response of superactive substitutions in BH-HN and BH-HC to the presence of Mn2+ in the catalytic site supports the view that conformational changes in these regions cause a distinct effect in the catalytic site of RNAP. While the C-terminal Bridge Helix operates predominantly by influencing Trigger Loop conformation, kinking of the N-terminus via BH-HN most likely alters the positions and/or conformations of the β-D II and Link domains, which are in direct physical contact with the nucleotide and nucleic acid substrates.
The application of a high-throughput in vitro mutagenesis approach to the N-terminal portion of the M. jannaschii bridge helix domain has revealed a range of new insights that could not be anticipated from previously available structural and genetic data. The data sets (Figures 2A, 4B and 5B) clearly illustrate that many of the most interesting insights were derived from substitutions that would almost certainly not have been designed using a rational, structure-led approach (for example, mj A' M808-P; ΔD816/Q817; A822-P/Q823-P/S824-P). Furthermore, widely used methods, based predominantly on alanine-scanning mutagenesis , are also limited in their capacity to uncover some of the most interesting phenotypes (Figure 2A). It is therefore clear that automated high-throughput methods for generating site-directed mutants and assaying their phenotypic consequences will play an increasingly important role in exploratory investigations of protein structure/function relationships as part of a diverse strategy aimed at obtaining new insights into complex biological systems .
Combinatorial permutation libraries, containing all 19 variants with codon replacements optimized for expression in E. coli were purchased for mj A' H806, A807, M808, G809, G810, R811, E812, G813, Q823-X/S824-P, A822-X/Q823-P/S824-P, S824-P/G825-X, and S824-P/G825-P/Y826-X from GeneArt (Regensburg, Germany). The mutations, located within a BstBI-SbfI fragment of the codon-optimized C-terminus of the Bridge Helix  were transferred to a pET21a bacterial expression vector for the production of full-length, intein-free mj A' subunits. The presence of the desired mutations in the expression constructs was verified by DNA sequencing. DNA constructs containing the two amino-acid deletions across the Bridge Helix were purchased from GeneArt as synthetic gene fragments and transferred to bacterial expression plasmids as described above.
The robotic procedures for high-throughput growth of bacterial expression strains, recombinant subunit purification and automated assembly into recombinant RNAPs (the 'RNAP Factory') have been described previously [19, 20, 58]. Briefly, bacterial constructs expressing the Bridge Helix mutants were transformed into chemically competent Acella cells (ΔendAΔrecA derivatives of E. coli BL21 [DE3]; EdgeBio, Gaithersburg, Maryland, USA). After growth for 16 to 18 hours at 37°C in 24-well plates in autoinduction medium (OverNight Express, Novagen, Nottingham, UK) the bacteria were harvested and used for a centrifugation-based robotic inclusion body purification protocol. The mj A' inclusion bodies were automatically solubilised in the presence of 8 M urea and quantitated at 562 nm with the bicinchoninic acid assay. Recombinant RNAPs containing the mj A' variants were assembled in a 96-well format dialysis cell using an urea-gradient from 6 M to urea-free spanning 16 hours at room temperature . Each mutant subunit was expressed, purified and assembled in vitro at least in quadruplicate to assure consistency and reproducibility. The assembled RNAPs were harvested and used immediately for robotic transcription assays (see below). The assembly efficiencies of key mutants (including, among others, mj A' M808-P; A807-P/M808-P; M808-P/G809-P; S824-P; Q823-P/S824-P; S824-P/G825-P; A822-P/Q823-P/S824-P) were compared to assembly rates achieved with the wildtype mj A' subunit by assaying the reconstituted polymerases at limiting and saturating template DNA concentrations (see Additional file 19 in  for details and examples); no differences between wild-type and mutant enzymes were detected. For replacement of Mg2+ ions in the catalytic site with Mn2+ (for the transcription assays shown in Figure 6), the in vitro assembly process was carried out as described above, but with dialysis- and transcription buffers containing 10 mM Mn2-O-acetate instead of Mg2-O-acetate.
The robotically implemented high-throughput trichloroacetic acid (TCA) precipitation assays, measuring the incorporation of (α-32P) rUTP into TCA-insoluble products, were carried out exactly as previously described [19, 20]. Briefly, assay mixtures were incubated for 45 minutes at 70°C in thin-wall PCR plates before precipitating the radiolabeled transcripts by the addition of ice-cold TCA solution. After incubation for 30 minutes at 1°C, the nucleic acid precipitates were collected by vacuum filtration on a 96-GF/F glass fiber filter plate (Whatman, Maidstone, UK) and extensively washed with further aliquots of ice-cold TCA solution. After drying the filters, scintillant (MicroScint-O; Perkin-Elmer, Cambridge, UK) was added and the amount of incorporated (α-32P) rUTP quantified with a microplate counter (TopCount NXT, Packard, Cambridge, UK).
Molecular dynamics (MD) simulations were performed using GROMACS . In preparation for MD simulations, the archaeal Bridge Helix was modelled on the S. cerevisiae 'active elongation' RNAPII structure (PDB #2E2H) using the SwissModel server in automated mode . The simulation production runs were executed in a fully solvated atomistic production mode without restraints. The energies of the modelled structures were initially minimized in vacuum using GROMACS with an AMBER force field (http://ambermd.org/) on a CPU cluster of the National Grid Service (NGS). During pre-processing the system was warmed to 200K under the control of a Berendsen thermostat with a coupling constant of 1.0 ps. All structures were energy-minimized in pre-equilibrated simulation boxes filled with TP3 water, and sodium and chloride ions were added to a final concentration of approximately 150 mM. For production runs the temperature was increased to 300 K (27°C). The equations of motion were integrated using a step-size of two femtoseconds. The trajectories generated by 27 independent 200 picoseconds simulation runs, were analyzed using STRIDE , as implemented in VMD . The frequencies of particular residues adopting a 'coil' conformation during 5 ps analysis windows were plotted relative to the Bridge Helix sequence.
molecular hinge located within the carboxy-terminal portion of the Bridge Helix
molecular hinge located within the amino-terminal portion of the Bridge Helix
nucleotide addition cycle
This work was supported by a Wellcome Trust project grant [078043/Z/05/Z] to ROJW. I would like to thank Dominic Conquest for help with the subcloning of the M808-X and G813-X expression clone libraries into expression vectors. I would also like to thank Hans Heindl for advice on the molecular dynamics simulations, Noam Weingarten for programming and Tamas Kiss and Gabor Terstyanszki for advice and provision of computing resources on the UK National Grid Service. I also appreciate the helpful comments on the manuscript received from Martin Buck, Patricia Burrows and Simone Wiesler.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.