Archaeal and eukaryotic RNAPs (DNA-dependent RNA polymerases) are complex multi-subunit enzymes. Two of the subunits, F and E, which together form the F/E complex, have been hypothesized to associate with RNAP in a reversible manner during the transcription cycle. We have characterized the molecular interactions between the F/E complex and the RNAP core. F/E binds to RNAP with submicromolar affinity and is not in a dynamic exchange with unbound F/E.
- fluorescence resonance energy transfer (FRET)
- RNA polymerase
RNAPs (DNA-dependent RNA polymerases) transcribe the genomes of bacteria, archaea and eukarya in a highly regulated manner. All multi-subunit RNAPs are evolutionary related and this is reflected in their structure and subunit composition . The archaeal RNAP has a higher subunit complexity than the bacterial enzyme and is closely related to eukaryotic RNAP II in terms of subunit composition, structure and requirements for basal transcription factors that govern its activities during the transcription cycle . A detailed characterization of the eukaryotic and archaeal RNAPs is a demanding intellectual and technical challenge that has long been beyond our reach. With the advent of structural information for RNAP and methods to reconstitute RNAPs from individual recombinant subunits we can now characterize the molecular mechanisms of RNAP in great detail .
RNAPs are functionally compartmentalized into subunits concerned with enzyme assembly (subunits D, N, L and P, homologous with eukaryotic RPB3, 10, 11 and 12) and catalysis (subunits A′, A″, B′ and B″, homologous with RPB1 and 2). In addition, four small subunits have specialized functions and are not strictly required for basic RNAP activity in vitro, including promoter-directed transcription (subunits H, K, F and E, homologous to RPB5, 6, 4 and 7) . All types of eukaryotic RNAPs (I, II, III and IV/V) include homologues of the H, K, F and E subunits, which emphasizes their importance for RNAP functionality in vivo [3,5]. Subunit H (RPB5) forms the ‘lower jaw’ of RNAP and contacts the downstream DNA template; RNAPs that lack subunit H display a reduced transcription activity, for unknown reasons [4,6]. Overall the RNAP forms a compact structure resembling a crab's claw. Subunits F and E (RPB4 and 7) form a stalk-like protrusion from the main body of the RNAP (Figure 1). In the archaeal system the F/E complex is involved in DNA melting and mediates the action of the basal factor TFE (transcription factor E) during transcription initiation [7–9]. Subunit E (RPB7) includes an S1/OB (S1/oligonucleotide-binding) fold that binds RNA in vitro and probably interacts with the emerging transcript or the DNA template in vivo during transcription [10–14]. In the yeast Saccharomyces cerevisiae the association of RPB4/7 with RNAP II correlates with the growth phase. During exponential growth only 20% of RNAP II contains RPB4/7, whereas all RNAPs are saturated with RPB4/7 in the stationary growth phase . In vitro transcription experiments using yeast RNAP II have shown that RPB4/7 is required during transcription initiation but dispensable for transcription elongation . Edwards et al.  speculated that RPB4/7 was predominantly associated with initiating RNAPs and that it dissociated from elongating enzymes. A prerequisite for this hypothesis is that RPB4/7, and RPB4/7-like complexes such as archaeal F/E, are reversibly associated with RNAPs and it suggests that the association and dissociation of RPB4/7 is important for RNAP function . Two independent whole genome distribution analyses of yeast RNAP II subunits identified RPB4/7 both at promoters and within intragenic regions, implying that RPB4/7 is associated both with initiating and elongating RNAPs [17,18]. Finally, there is no direct evidence for a reversible association of RPB4/7-like complexes with RNAP. In summary, the F/E (RPB4/7) RNAP subunits play multiple roles during the transcription cycle but there is conflicting evidence regarding their association with RNAP.
Eukaryotic RNAPs are not amenable to in vitro reconstitution and therefore do not allow a detailed analysis of RNAP assembly under strictly controlled conditions. We are using the euryarchaeal Methanocaldococcus jannaschii RNAP as a hyperthermophilic model system for eukaryotic RNAP II and have previously developed methods to reconstitute this RNAP from individual recombinant subunits . This model system can be perturbed, by mutagenesis, to allow the incorporation of fluorescent probes into the RNAP subunits. In the present study we have focused on subunits F/E and H because they are the most prominent candidates to be reversibly associated with RNAP. F/E and H are soluble under native conditions and are not strictly necessary for transcription initiation and therefore are, in theory, dispensable for RNAP function under certain conditions in vivo. In order to characterize RNAP assembly and to investigate the association of F/E (RPB4/7) and subunit H (RPB5) with RNAP we have generated fluorescently labelled F/E and H and monitored their interactions with RNAP using both native gel electrophoresis and fluorescence anisotropy in solution. We take advantage of the fact that active archaeal RNAPs can be assembled without subunits H, K, F and E. Soluble fluorescently derived subunits F/E and H can subsequently be incorporated into RNAPs lacking these, or other, subunits under native conditions, and the interactions between RNAP subunits can be quantified. Our results show that: (i) F/E and H interact with RNAP in the submicromolar range; (ii) the association of the F/E complex with RNAP requires subunit K and the association of subunit H with RNAP requires subunit A″; and (iii) neither the F/E complex nor subunit H are in a dynamic equilibrium with unbound F/E or H respectively. The latter result unequivocally proves that the archaeal F/E complex cannot reversibly associate with RNAP spontaneously.
Recombinant protein production
M. jannaschii RNAP was purified and assembled as described previously in . Single cysteine residues were engineered into RNAP subunit H at position Lys11, subunit F at position Ala25 and subunit E at position Val49 using an SOE (splice by overlap extension) PCR strategy. The subunit HK11C was expressed in Escherichia coli strain BL21(DE3) Rosetta 2, extracted with P300 lysis buffer as described in , and purified by heat inactivation at 65°C for 20 min and by gel filtration (using a S-100 HR 16/60 column; GE Healthcare). Subunit FA25C was expressed as a GST (glutathione transferase)-fusion protein and purified using a GST-Trap column (GE Healthcare) according to the manufacturer's instructions. Subunit EV49C was purified by inclusion body purification as described in  and subsequent solubilization in a buffer containing denaturant (P300 buffer containing 6M urea).
Proteins were labelled using Alexa Fluor® 488 (A488), Alexa Fluor® 594 (A594) or Alexa Fluor® 647 (A647) (Invitrogen). Typically 1–2 mg of protein was labelled at a single cysteine residue using an Alexa Fluor® maleimide derivative. The single cysteine variant subunits HK11C and FA25C were labelled under native conditions with a 5-fold molar excess of dye over protein for 2 h at room temperature (21 °C) in a buffer containing 50 mM Tris/HCl, pH 7.5, and 1 M NaCl. To remove the excess of unincorporated dye, labelled subunit HK11C was purified by gel filtration using a Nap 5 column (GE Healthcare) and buffer containing 50 mM Tris/HCl, pH 7.5, and 1 M NaCl. Unlabelled subunit HK11C was separated from fluorescently labelled subunit HK11C by anion-exchange chromatography (MonoQ; GE Healthcare). Subunit EV49C was labelled under denaturing conditions (buffer supplemented with 6 M urea) and subsequently heterodimerized with subunit F. In brief, labelled subunit EV49C was combined with subunit F (molar ratio of 3:2 of E/F) and was assembled using a denaturation–renaturation approach. Subunits FA25C and EV49C were combined in buffer containing 6 M urea and the denaturant was removed by stepwise dialysis against urea-free buffer using a dialysis frame (Perbio slide-a-lyser, 0.5–3.0 ml). Excess of subunit E and misfolded F/E complexes were removed by a heat-inactivation step. Excess dye and unlabelled F/E were removed using an anion-exchange column (MonoQ; GE Healthcare). The purity and labelling efficiency of the fluorescently labelled proteins was assessed by SDS/PAGE and absorption spectroscopy.
Native gel electrophoresis and analysis
A488-labelled F/E complex (200 nM) was incubated with RNAPΔF/E (RNAP without the F/E complex), RNAPΔK/F/E (RNAP without the K subunit and F/E complex) or wild-type RNAP at a concentration of 125, 250, 375, 500, 750 or 1000 nM in binding buffer (10 mM Tris/HCl, pH 7.5, and 1 M NaCl) for 20 min at 65 °C. A488-labelled subunit H (200 nM) was incubated with RNAPΔH (RNAP without the H subunit), RNAPΔA″/H (RNAP without the A″ and H subunits) or wild-type RNAP at a concentration of 0, 125, 250, 375, 500, 750 or 1000 nM. Following the addition of native Tris/glycine loading buffer the reaction mixture was separated on a Tris/glycine gel (12% gel) for 45 min at 200 V, and fluorescently labelled complexes were visualized on a Fuji FLA-2000 scanner using an excitation wavelength of 473 nm and a bandpass filter of 520 nm.
Fluorescence anisotropy measurements
Fluorescence anisotropy measurements were carried out on a FluoroMax-4 device (Horiba Jobin Yvon) with a thermostated cuvette holder at 65 °C in a 200 μl quartz cuvette. A594-labelled F/E complex (50 nM) or A647-labelled subunit H (50 nM) was incubated with increasing concentrations of RNAPΔF/E, RNAPΔK/F/E, wild-type RNAP or RNAPΔH, RNAPΔA″/H or wild-type RNAP in a total volume of 125 μl buffer (50 mM Tris/HCl, pH 7.5, and 1 M NaCl). Titration with buffer only was performed as a control. Fluorescence anisotropy was measured over time (10 min for each titration step) using an excitation wavelength of 588 nm (10 nm slit) and an emission wavelength of 612 nm (10 nm slit).
Anisotropy was measured with an integration time of 1 s. The Kd values were calculated from direct fitting of the curves obtained from direct binding assays in the program Prism (GraphPad) using a one-site-binding model. At least two independent experiments were conducted to determine Kd values.
The F/E complex is stably associated with archaeal RNAP
In order to characterize the association of the RNAP F/E complex with the RNAP core we incubated a constant amount of A488-labelled F/E complex with increasing amounts of RNAPΔF/E and separated unincorporated F/E complex from RNAP-bound F/E complex by native gel electrophoresis. Figure 2(A) shows the dose-dependent incorporation of fluorescently labelled F/E complex into a distinct band that corresponds to RNAP. Incubation of the non-labelled F/E complex in a 500-fold excess over RNAPs with A488-labelled F/E complex did not result in a competition of the labelled RNAP signal (Figure 2B). This demonstrates that the RNAP-bound F/E complex cannot be exchanged or displaced by free F/E. Previous work from our laboratory has shown that RNAP lacking subunit K was not able to respond to subunits F/E in functional assays [8,9]. When we tested the binding of fluorescently labelled F/E complex to RNAPΔK/F/E no incorporation was observed (Figure 2A) demonstrating that the association of F/E with RNAP is strictly dependent on subunit K, and explaining the lack of F/E response of RNAPΔK/F/E in our previous transcription experiments [8,9]. This functional defect of RNAPΔK/F/E is congruent with structural information of the archaeal RNAP since subunit K makes contributions to the interactions between RNAP and the F/E complex (Figure 1A).
In order to characterize the interaction between subunits F/E and the RNAP core in solution in a detailed quantitative manner we carried out fluorescence spectroscopy. A constant amount of A594-labelled F/E complex was incubated with increasing concentrations of RNAPΔF/E and the fluorescence anisotropy was recorded. Figure 2(C) shows that the relative fluorescence anisotropy of the labelled F/E complex increases in a dose-dependent manner by the addition of RNAPΔF/E. This is the result of a decrease in the rotational diffusion (‘tumbling’) of the F/E complex upon incorporation into the RNAP core, which increases the size of the particles by more than one order of magnitude (from 34 kDa to 370 kDa). Fitting of the data to a curve resulted in a Kd of 352±79 nM. Titration of A594-labelled F/E complex with RNAPΔK/F/E or wild-type RNAP does not affect the anisotropy confirming that: (i) subunit K is strictly required for F/E binding; and (ii) that RNAP-bound F/E complex is not in a dynamic equilibrium with free F/E complex in solution. Whereas the native gel electrophoresis was carried out at ambient temperatures, the fluorescence anisotropy was recorded at the more physiologically relevant temperature of 65 °C. The two methods gave the same result and thus the temperature is not likely to have caused artefactual results. Control experiments with buffer only resulted in no overall change in anisotropy (results not shown).
RNAP subunit H is stably and not reversibly associated with archaeal RNAP
Having successfully applied fluorescence techniques to characterize the interactions between the F/E complex and RNAP core we extended our analysis to the auxiliary subunit H. A488-labelled subunit H is incorporated into a distinct low-mobility RNAP complex in a dose-dependent manner in a native gel electrophoresis experiment (Figure 3A). Incubation of fluorescently labelled subunit H with wild-type RNAP did not result in the incorporation of A488-labelled subunit H (Figure 3A). This result suggests that, similar to the F/E complex, RNAP-bound subunit H is not in a dynamic equilibrium with unbound subunit H. This finding is further supported by a displacement experiment. A 500-fold molar excess of unlabelled subunit H is not able to exchange or displace A488-labelled subunit H that is incorporated into RNAP (Figure 3B).
Close inspection of the two crenarchaeal RNAP structures (PDB codes 2PMZ  and 2WAQ/2WB1 ) reveals that subunit H makes extensive contacts with subunit A″ (Figure 1B). When we incubated fluorescently labelled subunit H with increasing concentrations of RNAPΔA″/H no incorporation of subunit H occurred, demonstrating a strict requirement for the A″ subunit (Figure 3A). In parallel, we characterized the association of subunit H with RNAP in solution using fluorescence anisotropy. Figure 3(C) shows that the fluorescence anisotropy of A488-labelled subunit H increases by the addition of RNAPΔH in a dose-dependent manner with nanomolar affinity (Kd=21±4 nM). The anisotropy does not increase by the addition of wild-type RNAP (containing subunit H) or RNAPΔA″/H (Figure 3C). The latter two results confirm that RNAP-bound subunit H is not in equilibrium with unbound subunit H, and that subunit A″ is strictly required for the incorporation of subunit H into the RNAP. Control experiments with buffer only resulted in no overall change in anisotropy (results not shown).
Complex multi-subunit RNAPs such as the archaeal and eukaryotic enzymes consist of 10–17 subunits. The individual subunits contribute to the activities of RNAP in different ways, including RNAP assembly, DNA and RNA binding, catalysis and regulatory interactions with basal factors. RNAPs are dynamic molecules that consist of a rigid core with flexible elements, and during the transcription cycle RNAPs reversibly associate with basal transcription factors that modulate its properties and structure. The mobile RNAP clamp and RPB4/7-like complexes are emerging as potential regulatory hotspots that could be involved in both transcription initiation and elongation [3,7,16,21]. Since the observation that RNAP II, in exponentially growing S. cerevisiae cells, is subsaturated with respect to RPB4/7, various hypotheses on the molecular mechanisms of RNAP have been proposed that conceptually require the reversible association of RPB4/7 with RNAP. On the basis of crystallographic evidence the association–dissociation of RPB4/7 with RNAP II has the potential to modulate the position of the flexible RNAP-clamp domain. During transcription initiation a closure of the clamp could be involved in DNA melting . During transcription elongation a closure of the clamp domain could stabilize the ternary RNAP/DNA/RNA complex and thereby affect RNAP processivity. X-ray crystallography of RNAPs has given us important insights into the ternary organization of RNAP and the protein interaction network between the subunits. However, crystal structures cannot alone provide the basis for our understanding of the dynamics of RNAP and transcription complexes. Recombinant in vitro transcription systems can make important contributions in this context. We are using a hyperthermophilic archaeal model system for eukaryotic RNAP II because of its superior biochemical tractability. This system allowed us to construct archaeal RNAP subunits that are homologous with RPB4/7, F/E, and to use fluorescent probes to analyse their interactions with RNAP. Our results demonstrate that the archaeal F/E subunits bind to RNAP with moderate affinity (Kd=352±79 nM), are stably associated with RNAP and are not in a dynamic equilibrium with free F/E. This conclusion does not support hypotheses about an opening and closing of the RNAP clamp due to association and dissociation of RPB4/7-like complexes. We cannot rule out the possibility that the association of F/E (RPB4/7) with RNAP is affected differently by nucleic acids and basal factors in the context of the transcription initiation complex and elongation complex. However, a recent whole-genome chromatin immunoprecipitation microarray (‘ChIP on chip’) analysis of the S. cerevisiae RPB4/7 has shown that the occupancy profiles of RPB4/7 and another RNAP subunit, RPB3, are nearly identical and not influenced by the position of RNAP along the gene . This result suggests that RPB4/7 is continually associated with RNAP in vivo, independent of the phase of the transcription cycle. We furthermore show that F/E binding to RNAP is crucially dependent on subunit K (RPB6) and this explains previous observations  that the action of basal factor TFE is not only dependent on RNAP subunits F/E, but also subunit K. This is in good agreement with a study showing mutants of the homologous RPB6 subunit result in decreased binding of RPB4/7 to RNAP II .
In parallel, with the F/E–RNAP interaction experiments we analysed binding of RNAP subunit H (RPB5) using A647-labelled subunit H. Subunit H interacts with the archaeal RNAP with high affinity (Kd of 21±4 nM), its incorporation is strictly dependent on subunit A″ and, once incorporated into the RNAP, subunit H is not in equilibrium with unbound subunit H. The observation that subunit H interacts with RNAP with an affinity that is one order of magnitude higher than the F/E complex reflects their interactions with subunits A″ and K respectively. This is congruent with the structural information obtained on two crenarchaeal RNAPs [20,21]. Whereas the subunit E forms a narrow wedge under the RNAP clamp via subunit K, the H subunit is embedded in subunit A″.
In conclusion, the interaction analysis using fluorescently labelled archaeal RNAP subunits strongly suggests that once the RNAPs are assembled they form a remarkably stable complex that is not subject to subunit exchange and therefore not compromised by instabilities that may arise from this exchange.
Dina Grohmann and Angela Hirtreiter planned and conducted experiments. Finn Werner conceived and designed the study and obtained funding.
This work was supported by the Wellcome Trust [grant number 079351/Z/06/Z]; and by the Biotechnology and Biological Sciences Research Council [grant number BB/E008232/1].
We would like to thank Gabriel Waksman and Achillefs Kapanidis for helpful discussions on fluorescence methods.
Abbreviations: A488, Alexa Fluor® 488; A594, Alexa Fluor® 594; A647, Alexa Fluor® 647; GST, glutathione transferase; RNAP, DNA-dependent RNA polymerase; RNAPΔF/E, RNAP without the F/E complex; RNAPΔK/F/E, RNAP without the K subunit and F/E complex; RNAPΔH, RNAP without the H subunit; RNAPΔA″/H, RNAP without the H and A″ subunits; TFE, transcription factor E
- © The Authors Journal compilation © 2009 Biochemical Society