Biochemical Journal

Research article

Domain interactions of the transcription–translation coupling factor Escherichia coli NusG are intermolecular and transient

Björn M. Burmann , Ulrich Scheckenhofer , Kristian Schweimer , Paul Rösch


The bacterial transcription factor NusG (N-utilization substance G) is suggested to act as a key coupling factor between transcription and translation [Burmann, Schweimer, Luo, Wahl, Stitt, Gottesman and Rösch (2010) Science 328, 501–504] and contributes to phage λ-mediated antitermination in Escherichia coli that enables read-through of early transcription termination sites. E. coli NusG consists of two structurally and functionally distinct domains that are connected through a flexible linker. The homologous Aquifex aeolicus NusG, with a secondary structure that is highly similar to E. coli NusG shows direct interaction between its N- and C-terminal domains in a domain-swapped dimer. In the present study, we performed NMR paramagnetic relaxation enhancement measurements and identified interdomain interactions that were concentration dependent and thus probably not only weak and transient, but also predominantly intermolecular. This notion of two virtually independent domains in a monomeric protein was supported by 15N-relaxation measurements. Thus we suggest that a regulatory role of NusG interdomain interactions is highly unlikely.

  • antitermination
  • molecular dynamics
  • NMR
  • N-utilization substance G (NusG)
  • paramagnetic relaxation enhancement (PRE)
  • phage λ


Escherichia coli NusG (N-utilization substance G) is essential for cell viability [1] and is found in all known bacteria. The CTD (C-terminal domain) contains a KOW motif that is also found in archaeal [2] and eukaryotic [3] proteins. The exact role of NusG in transcription regulation remained obscure until recently, when NusG was identified as the long-sought-after coupling factor between transcription and translation in E. coli [4], in addition to its better known role in the TEC (transcription-elongation complex) where it increases the elongation rate of RNAP (RNA polymerase) in vivo and in vitro [57] by suppression of transcriptional pausing [7], which was shown recently to be due to promoting forward translocation of the RNAP [8]. NusG, along with the other Nus factors A, B and E (ribosomal protein S10), RNA and RNAP, is also a component of phage λ protein N-mediated antitermination [911] and the phage HK022 protein Nun-mediated termination complex [12,13]. The regulatory functionality of NusG can be attributed to direct interaction with the RNAP [9,14,15].

Additionally, NusG is involved in transcription antitermination in rRNA [16,17] and in termination at certain ρ-dependent sites where it recruits ρ [12,18,19]. NusG also prevents backtracking of the RNAP [7], but detailed information on the final steps of ρdependent termination is still lacking [20]. Direct NusG–ρ interaction was reported [18,19], and the NusG-CTD was identified as the interaction domain [4]. In addition to its role in transcription, NusG is also involved in translational regulation [21], and, again, the NusG-CTD was identified as the ribosome-interaction domain [4].

The structure of E. coli NusG [20] is highly similar to the structures of the homologous Aquifex aeolicus [22,23] and Thermus thermophilus [24] proteins. A. aeolicus NusG is reported to form a domain-swapped dimer under certain crystallization conditions [23]. However, such a dimer is not observed in a different crystallization setup [22], and ultracentrifugation analysis showed no indication of a non-monomeric form of NusG as an isolated protein [19]. Interestingly, however, the NusG paralogue RfaH is reported to exist in a closed conformation with both domains tightly interacting, and this tight interaction is abolished only upon binding of RfaH to the ops (operon polarity suppressor) site of the non-template DNA strand, when the RfaH-CTD leaves the binding pocket on the RfaH-NTD (N-terminal domain), thus providing an RfaH-activation step [25,26]. Although E. coli NusG and RfaH exhibit the same folding topology of their NTDs [20,26] (NusG-NTD, PDB code 2K06; RfaH, PDB code 2OUG), their CTDs show reversed topologies: whereas RfaH-CTD is completely α-helical, NusG-CTD exhibits an anti-parallel β-barrel-like structure [20] (NusG-CTD, PDB code 2JVV).

The area masked on the RfaH-NTD by its CTD consists of several surface-exposed hydrophobic residues [15,27], and surface hydrophobic residues are more numerous in RfaH than in NusG [20], where they may be involved in additional protein interactions [22]. These lines of results raise the question of which properties were possessed by the common ancestor of these proteins, in particular, whether it was in a closed or open form, and whether domain opening provided a regulatory mechanism. In the present study, we analysed whether E. coli NusG domains interact and whether a closed conformation, exists, if only at a low density.

To address these questions and to gain insight into the role of the individual E. coli NusG domains within the TEC, their relative motional behaviour and their mutual interactions, we employed chemical shift and PRE (paramagnetic relaxation enhancement) perturbation titration as well as relaxation experiments.


Cloning, expression and purification of full-length NusG and its individual domains

Cloning, expression and purification was performed on the basis of published methods [19,20]. NusG was cloned via Bpu1102I and NdeI restriction sites into the E. coli expression vector pET11A (Novagen). E. coli strain BL21(DE3) (Novagen) harbouring the recombinant plasmid was grown at 37°C in LB (Luria–Bertani) medium containing ampicillin (100 μg/ml) until a D600 of 0.8 was reached and the cells were induced with a final concentration of 1 mM IPTG (isopropyl β-D-thiogalactopyranoside). Cells were harvested 4 h after induction, resuspended in four times the pellet weight of lysis buffer [50 mM Tris/HCl, pH 7.5, 150 mM NaCl, 1 mM DTT (dithiothreitol) and half of a protease-inhibitor tablet (Complete™, EDTA-free; Roche)], and lysed using a micro-fluidizer (Microfluidics). After centrifugation (12000 g, 45 min), polyethylenimine (Fluka) was added dropwise with continuous stirring to the supernatant to a final concentration of 0.6%. The lysate was incubated for 20 min and centrifuged at 12000 g for 30 min. Ammonium sulfate was added dropwise with stirring to the supernatant to a final concentration of 60%. The lysate was centrifuged at 12000 g for 30 min and the pellet was dissolved in 30 ml of buffer A (50 mM Tris/HCl, pH 7.5). The lysate was then dialysed against 2×4 litres of buffer A for 4 h each. The lysate was applied to a HeparinFF column (GE Healthcare) using a step gradient with increasing NaCl concentrations (0–1 M). For further purification, the eluted fractions containing NusG were pooled and concentrated with Vivaspin concentrators [Vivascience; MWCO (molecular-mass cut-off) 5 kDa]. The concentrated sample was applied to an S75 gel-filtration column (50 mM Tris/HCl, pH 7.5, and 150 mM NaCl; GE Healthcare). The fractions containing NusG were pooled and dialysed against buffer as used for NMR measurements (NMR buffer: 10 mM potassium phosphate, pH 6.4, and 50 mM NaCl). The identity and structural integrity of the purified protein was analysed by SDS/PAGE (19% gel) as well as by NMR spectroscopy. NusG-NTD (amino acids 1–124) was cloned through Bpu1102I and NdeI into the E. coli expression vector pET11A (Novagen). The same procedure as described above for full-length NusG was used for expression and purification of NusG-NTD. NusG-CTD (amino acids 123–181) was cloned and purified as described previously [4].

Full-length NusG mutation S16C

For NusGS16C, the following primers were used: 5′-primer, GTCGTTCAGGCGTTTTGCGGTTTTGAAGGCCGC; 3′-primer, GCGGCCTTCAAAACCGCAAAACGCCTGAACGAC. The mutation was introduced by using the QuikChange® protocol (Stratagene). Expression and purification were as for full-length NusG, except that 1 mM DTT was added to all buffers.

Random spin labelling of the ϵ-amino groups of lysine residues

Spin labelling of the lysine residues with OXYL-1-NHS (1-oxyl-2,2,5,5-tetramethylpyrroline-3-carboxylate N-hydroxysuccimide ester; Toronto Research Chemicals) was performed using a slightly modified published protocol [28]. Briefly, 500 μM protein solution in NMR buffer was washed with 10 mM sodium carbonate buffer, pH 9.2, for buffer exchange in Vivaspin concentrators (Vivascience; MWCO 5 kDa). A 10-fold excess of OXYL-1-NHS in DMSO was added, followed by incubation for 1 h at room temperature (25°C) and an additional 1 h at 4°C in the dark. To remove unreacted OXYL-1-NHS, the sample was washed with 20 ml of NMR buffer in a Vivaspin concentrator. For reduction of the spin label, ascorbate (500 mM stock) was added directly to the NMR tube to a final concentration of 5 mM.

Site-specific spin labelling of the NusG cysteine mutants

Spin labelling of the cysteine residues introduced into NusGS16C with MTSL [(1-oxyl-2,2,5,5-tetramethyl-Δ3-pyrroline-3-methyl) methanethiosulfonate; Toronto Research Chemicals] was performed using a slightly modified published protocol [29]. Briefly, 500 μM protein solution in NMR buffer was washed with 10 mM sodium acetate buffer, pH 4.5, for buffer exchange in Vivaspin concentrators (Vivascience; MWCO 5 kDa). After addition of DTT to a final concentration of 5 mM, the sample was kept at 4°C for 1 h. For removal of DTT, the solution was eluted isocratically with 10 mM sodium acetate from a HiTrap desalting column (GE Healthcare). A 10-fold excess of MTSL dissolved in acetonitrile was added, followed by incubation overnight at room temperature in the dark. To remove unreacted MTSL, the sample was washed again with 20 ml of NMR buffer in a Vivaspin concentrator. For reduction of the spin label, ascorbate (500 mM stock) was added directly to the NMR tube to a final concentration of 5 mM.


NMR experiments were recorded at sample temperatures of 298 K on Bruker Avance 600 MHz and 700 MHz (equipped with a cryogenically cooled probe) NMR spectrometers. Assignments for NusG-NTD and full-length NusG were from previous work [20]. The isolated NusG-CTD (1H,15N)-HSQC (heteronuclear single-quantum coherence) resonances matched the corresponding signals of full-length NusG perfectly and were trivially assigned, and the assignments were verified by triple-resonance NMR experiments.

PRE (R2,para=R2,spinlabelR2,no spinlabel) and their errors were determined by a two-point scheme using an HSQC experiment with an additional spin-echo period during the first INEPT (insensitive nuclei enhanced by polarization transfer) procedure [30]. Spin-echo intervals were set to 0.1 ms and 10.2 ms. 15N longitudinal (R1) and transverse (R2) relaxation rates were determined by standard methods at 1H frequencies of 600.2 MHz and 700.2 MHz respectively. Relaxation curves were fitted to mono-exponential decays using the program package CURVEFIT (Arthur G. Palmer, Department of Biochemistry and Molecular Biophysics, Columbia University, New York, U.S.A.). Rotational correlation times were determined from the R2/R1 ratios using the program TENSOR2 [31].

Molecular dynamics simulation

In order to test the possibility of an intramolecular interdomain interaction of E. coli NusG analogous to the swapped dimer interaction in A. aeolicus NusG, we calculated the time-dependent structure with molecular dynamics simulations. An initial structural model of E. coli NusG with domain interaction was constructed by fitting the solution structures of the E. coli domains to A. aeolicus NusG. The simulation using XPLOR (XPLOR-NIH 2.1.2 [32]) was based on our standard protocol for protein structure calculation [33]. During the calculation, backbone coordinates were held fixed for all residues except the linker region between the NTD and CTD (residues 117–126) and the sequence region 99–104 to reduce steric clashes between the NTD and CTD in the initial structure. Additionally, residues in the flexible region Glu48–Arg62 were removed in silico to avoid steric conflicts. No further restraints were applied. Existence of an intramolecular interdomain interaction was accepted on the basis of the absence of violation of restraints (deviation of bond lengths and angles), and 50 structures out of a total of 80 calculated were accepted on the basis of this criterion.


NMR titrations suggest independence of the NusG domains

(1H,15N)-HSQC chemical shifts of resonances from the isolated NusG-CTD and isolated NusG-NTD were virtually identical to the chemical shifts of the respective domains in full-length NusG (Supplementary Figure S1 at, rendering persistent domain interactions within full-length NusG highly unlikely. In order to further clarify whether or not transient domain interactions exist for the more general case of the isolated domains, we resorted to observation of HSQC chemical-shift perturbations of an isotope-labelled domain, NusG-CTD or NusG-NTD, upon addition of the unlabelled potentially interacting domain, a method that is well established as a tool to study the interplay between molecules [34]. NMR spectroscopy, due to its inherent insensitivity and its resulting requirement for sample concentrations in the high micromolar range, is useful for detecting even weak (Kd~μM–mM) interactions, and perturbations of amide group resonance shifts as detected by (1H,15N)-HSQC are very sensitive even to subtle structural changes. As Phe165 of the NusG-CTD was claimed to interact strongly with a NusG-NTD hydrophobic cavity [23], significant chemical-shift changes are expected on domain contact, at least for Phe165 and residues in its vicinity, that is residues in the loop between strands β3 and β4. Titration of unlabelled NusG-NTD to 15N-labelled NusG-CTD to a 2-fold excess and vice versa, however, did not result in observable chemical-shift changes in the NMR experiments (Supplementary Figure S2 at Thus, with the level of sensitivity provided by HSQC perturbation experiments, domain interaction could be detected neither within the full-length protein nor for the isolated domains under our experimental conditions. This observation strongly correlates with our previous conclusion that the area around Phe165 is not involved in mutual interactions of NusG domains, but rather is a key residue in NusE–NusG complex formation [4].

15N relaxation reveals decoupled domain reorientation

In order to study the degree of motional independence of the NusG domains we determined their 15N-relaxation rates in the context of the full-length protein [35]. In a two-domain protein, concerted tumbling of domains can be described by a single rotational diffusion tensor, whereas independent tumbling of domains requires description with different rotational diffusion tensors [36]. In an isotropic rotation model, differences in these tensors are directly reflected in differences in the average relaxation rates.

15N transverse and longitudinal relaxation rates were determined at 14.1 T, with a sample concentration of 200 μM to reduce aggregation (Figures 1A and 1B). In the HSQC spectrum of the full-length protein, virtually all residues of NusG-CTD were observed, the average longitudinal relaxation rate (R1) was 1.35±0.09 s−1, and the transversal relaxation rate (R2) was 15.4±1.9 s−1, corresponding to a rotational correlation time τc of 10.1 ns in an isotropic model. Although the NusG-NTD signals were considerably weaker, 67 non-overlapping signals were observed and the average R1 was 1.1±0.2 s−1 and R2 was 19.0±4.6 s−1, corresponding to τc=13.1 ns in the isotropic model. Analysis of R1/R2 distributions offers an elegant method to detect interdomain motion on a timescale faster than the overall tumbling [37]. The R1/R2 ratios form the basis for determination of the rotational diffusion tensor by NMR relaxation, and for compact globular proteins a uniform distribution of R1/R2 ratios is characteristic [35]. The bimodal distribution of the R1/R2 ratios (Figure 1C) that reflect the two domains demonstrates their different rotational reorientation behaviour, arguing against a stable domain interaction on the nanosecond timescale. Although the difference in relaxation rates demonstrates independent movement of the two domains to a certain degree, τc values of the NTD and CTD in full-length NusG are higher than those of the isolated domains (13.1 ns compared with 8.9 ns, and 10.1 ns compared with 4 ns; see Supplementary Figure S3 at for relaxation rates of the individual domains). This increase of rotational correlation times of individual isolated domains up to values found in multidomain proteins without any domain interaction has been observed for several proteins [3841]. The spatial exclusion due to the presence of a second domain connected by a linker is probably hampering the unrestricted rotation of a domain, resulting in a longer correlation time [39,41]. The large increase of the rotational correlation times of both NusG domains indicates that motional decoupling of NusG-CTD and NusG-NTD through the five-residue linker is imperfect, and a significant contribution of this effect in full-length NusG to the relaxation mechanisms is expected. The interdomain motion alters the overall shape of full-length NusG and therefore affects the overall tumbling. As a consequence, complete uncoupling of domain dynamics and overall rotation is unlikely, and these coupled motions render a more detailed analysis at least very difficult [39]. In intact NusG, Val162 and Ile164 of NusG-CTD exhibit significantly enhanced R2 rates of 21.5 s−1 as compared with the domain average of 15.4 s−1, a difference not found in isolated CTD. Most probably, the enhanced transverse relaxation can be attributed to a chemical exchange contribution, and as these residues are located in the domain interface of the swapped-dimer crystal structure, this might be regarded as initial evidence of transient domain interaction in solution.

Figure 1 15N-Relaxation rates for full-length NusG

(A) Longitudinal relaxation rates (R1) for full-length NusG. (B) Transversal relaxation rates (R2) for full-length NusG. Residues 45–65 and 115–125 are highly flexible and located in strongly overlapping regions of the protein, so their relaxation rates could not be determined. (C) The distribution of R1/R2 is bimodal for full-length NusG. NusG-CTD is shown as black bars, and NusG-NTD is shown as grey bars.

Only highly sensitive PRE experiments show signs of interaction between the isolated domains

Fast exchange on the NMR timescale leads to observation of population-averaged parameters, and states that are extremely weakly populated often cannot be detected even by HSQC chemical-shift perturbations or changes in relaxation rates. Paramagnetic interactions, however, provide a means to detect even weak and transient interactions between molecules. Paramagnetic centres dramatically increase the relaxation rates of nearby nuclei and, as a consequence, the presence of even very minor concentrations of paramagnetic labels in proximity to observed nuclei can enhance the relaxation rate of the latter to an observable degree [42]. Thus PRE observed in a non-spin-labelled protein in the presence of a spin-labelled protein points to an at least transient proximity of both molecules. The experiment is straightforward, as a non-spin-labelled protein, for example NusG-CTD, can be made easily detectable in HSQC spectra by 15N enrichment, and the introduction of paramagnetic centres into proteins, for example NusG-NTD, is possible by random labelling of mostly surface-exposed lysine residues with OXYL-1-NHS [43]. In such an experiment, addition of OXYL-1-NHS–NusG-NTD to 15N-NusG-CTD (Supplementary Figure S4 at causes observable and specific PRE in the latter, and vice versa (Figure 2). Mapping of the resonances with increased relaxation rates on to the three-dimensional structure of NusG-CTD shows very clearly the region around Phe165 and the region Pro140 to Asn145, close to Phe165 in the turn between β1 and β2, to be the ones most seriously affected (Figure 2).

Figure 2 PRE-detected interactions between the isolated domains

(A) Surface representation of the individual E. coli NusG domains (NusG-NTD, PDB code 2K06; and NusG-CTD, PDB code 2JVV [20]) with secondary-structure elements. Phe65 and Phe165 are shown as blue sticks, and amino acids with an R2,para (HN) effect >20 Hz in the titrations of the individual domains are shown in red. (B) R2,para (HN) rates and their errors determined by a two-point scheme [30] for each amino acid upon titration with the individual spin-labelled domains are in grey; signals disappearing due to extensive line broadening are in red. The dotted line indicates a 20 Hz significance level.

The outcome of the inverse experiment, that is addition of OXYL-1-NHS–NusG-CTD to 15N-NusG-NTD (Supplementary Figure S4), was less clear-cut. The hydrophobic patch around Phe65 was clearly affected, together with several residues in helix α3 and residues close to Phe65 that form an apolar surface surrounded by polar residues. However, several isolated residues in helices α1 and α2 at the opposite side of the molecule as well as several residues of the upper loop regions facing away from the putative interaction site were affected (Figure 2). Combined, the PRE results yield a picture in which a plug around Phe165 fits nicely into a socket around Phe65. Although this is a reasonable contact surface, and although these sites correspond well to those found in the swapped dimer crystal structure [23], our negative results with HSQC chemical-shift mapping indicate the interaction to be rather weak and transient. Under the assumption of chemical-shift changes in the range typically observed in this experimental setup {e.g. NusG-CTD–NusE [4], λN (protein N of phage λ)–AR1 (acidic repeat 1) [44] and LckSH3 (Lck Src homology 3)–Tip (tyrosine kinase-interacting protein) [33]}, no significant chemical-shift changes point to less than 10% bound species in the NMR samples, which corresponds to a Kd higher than 5 mM.

PREs of site-specific full-length NusG are concentration dependent

Chemically modifying cysteine residues with MTSL provides a means to observe the PRE induced by site-specific labels. Thus, in order to clarify whether or not the weak domain interaction observed by random labelling in solution is intra- or inter-molecular, we performed additional PRE experiments on a full-length NusG construct, containing MTSL–Cys S16C. PRE originating from MTSL–Cys was detected in the unchanged CTD, clearly indicating an interaction between NusG-NTD and NusG-CTD, thus neither supporting nor ruling out the formation of a domain-swapped dimer or an intramolecular interdomain interaction within one molecule. To further probe these domain interactions, we studied the concentration dependence of the PRE induced by MTSL–Cys16 on the CTD, as intramolecular domain interactions would clearly be expected to result in a concentration-independent PRE due to unchanged internuclear distance distributions in a stable structure with interacting domains, whereas a concentration-dependent PRE would demonstrate intermolecular domain interactions by reduction of the fraction of interacting molecules upon dilution and a decrease of the observed ensemble averaged PRE. The PRE thus determined at different protein concentrations (Figure 3) shows a direct dependence of the relaxation enhancement on sample concentration, thus clearly ruling out an intramolecular interaction and supporting the model of a transient domain-swapped dimer formation, which is extremely sparsely populated.

Figure 3 PRE in a site-specific spin-labelled S16C full-length NusG is concentration dependent

R2,para (HN) rates and their errors of spin-labelled S16C full-length NusG determined at a concentration of 150 μM (dark grey) and at a concentration of 50 μM (light grey) are shown for selected residues of the CTD.

E. coli NusG NTD–CTD interaction analogous to that observed within the swapped dimer in A. aeolicus NusG, but within a single molecule, would only be possible if no sterical restraints existed in the linker sequence. To test whether or not such steric and dynamic restraints exist, we performed molecular dynamics simulations using a structural model with interacting domains to explore the conformational space of the linker region. Only a slight rearrangement of the loop preceding helix α3 of NusG-NTD would be necessary to remove all sterical clashes in the initial model based on the solution structure of E. coli NusG. In addition, all 50 analysed lowest-energy structures are without violation of bonds or angles and provide possible linker conformations that could allow domain interactions within one NusG molecule (Figure 4). Therefore an intramolecular domain interaction cannot be ruled out solely on the argument of the linker being too short. Nevertheless, the linker region must adopt a very extended conformation for the domains to be able to directly interact. This is expected to lead to a high degree of rigidity in the linker on the timescale of molecular reorientation due to the low number of possible conformational states that the linker can adopt in this extended state. On the other hand, the 15N R2 values of Gly119 (7.6 Hz) and Asp120 (6.4 Hz) are significantly lower than those for the other residues in the respective domains (Figure 1B). The reduction in R2 is characteristic of enhancement of mobility on the sub-τc (sub-nanosecond) timescale, demonstrating experimentally the flexibility of the linker region in contrast with the rigidity of the linker needed for direct interaction as shown by the molecular dynamics simulation. This renders a domain interaction within one molecule highly unlikely.

Figure 4 Superposition of the 20 lowest-energy structures of molecular dynamics simulation with intramolecular interdomain interactions

NTD, green; CTD, red; linker region (residues Gln117–Thr126), blue. The flexible region Glu48–Arg62 (broken line in the left-hand structure) was removed in silico to avoid steric conflicts.


Our NMR studies on domain interactions of E. coli NusG show that these interactions are not detectable by standard chemical-shift mapping, and 15N-relaxation measurements reveal movements of the two domains to be independent to a very high degree. Domain interactions could only be observed with the extremely sensitive technique of PRE measurements, and it needed a combination of this technique and site-directed paramagnetic labelling to show that these weak interactions were inter- rather than intra-molecular. Thus a low population of molecules seems to adopt a bimolecular conformation compatible with the proposed swapped dimer observed in X-ray crystallography. Intramolecular domain interaction, however, would be required for an auto-inhibitory function, and therefore a biologically relevant form, to reduce the population of the active open state. The mode of regulation found for the NusG paralogue RfaH [26] can thus be ruled out to apply to NusG, which may explain why NusG maintains its overall function without additional activation signals, whereas RfaH needs the ops site to render the protein functional. Current opinion is that NusG and RfaH were derived from the same ancestor [15], possibly a NusG protein that exists in closed conformation and that may be regulated by domain interactions. It may be speculated further that regulation of NusG through domain interactions is no longer necessary when RfaH is present. This hypothesis awaits further structural investigation, because, in addition to the proposed swapped dimer for the A. aeolicus [23] and the non-interacting T. thermophilus [24] NusGs, detailed structural information on other thermophilic NusGs is not yet available.


The experiments and data analysis were conducted by Björn Burmann and Kristian Schweimer. Ulrich Scheckenhofer established several purification protocols and performed part of the NMR assignment. Björn Burmann, Kristian Schweimer and Paul Rösch wrote the manuscript. Paul Rösch designed and co-ordinated the study.


This work was supported by the Deutsche Forschungsgemeinschaft (DFG) [grant number Ro617/16-1 (to P.R.)].


We thank Ramona Heissmann for excellent technical assistance.

Abbreviations: CTD, C-terminal domain; DTT, dithiothreitol; HSQC, heteronuclear single-quantum coherence; MTSL, (1-oxyl-2,2,5,5-tetramethyl-Δ3-pyrroline-3-methyl) methanethiosulfonate; MWCO, molecular-mass cut-off; NTD, N-terminal domain; NusG, N-utilization substance G; ops, operon polarity suppressor; OXYL-1-NHS, 1-oxyl-2,2,5,5-tetramethylpyrroline-3-carboxylate N-hydroxysuccimide ester; PRE, paramagnetic relaxation enhancement; RNAP, RNA polymerase; TEC, transcription-elongation complex


View Abstract