Report

A NusE:NusG Complex Links Transcription and Translation

See allHide authors and affiliations

Science  23 Apr 2010:
Vol. 328, Issue 5977, pp. 501-504
DOI: 10.1126/science.1184953

Transcription and Translation in Train

In bacteria, translation of messenger RNA into proteins by the ribosome usually begins soon after the ribosome binding site emerges from RNA polymerase. Now there is evidence for direct coupling between transcription and translation in bacteria. Proshkin et al. (p. 504; see the Perspective by Roberts) show that the trailing ribosome controls the rate of transcription by preventing RNA polymerase from spontaneous backtracking, which allows precise adjustment of transcriptional yield to translational needs under various growth conditions. Burmann et al. (p. 501; see the Perspective by Roberts) provide a potential mechanism for coupling by showing that the transcription factor NusG, which binds RNA polymerase through its amino-terminal domain, competitively binds either a ribosomal protein or the Rho transcription termination factor through its carboxy-terminal domain. Rho binding might occur after release of the ribosome from messenger RNA, thus linking termination of transcription and translation.

Abstract

Bacterial NusG is a highly conserved transcription factor that is required for most Rho activity in vivo. We show by nuclear magnetic resonance spectroscopy that Escherichia coli NusG carboxyl-terminal domain forms a complex alternatively with Rho or with transcription factor NusE, a protein identical to 30S ribosomal protein S10. Because NusG amino-terminal domain contacts RNA polymerase and the NusG carboxy-terminal domain interaction site of NusE is accessible in the ribosomal 30S subunit, NusG may act as a link between transcription and translation. Uncoupling of transcription and translation at the ends of bacterial operons enables transcription termination by Rho factor, and competition between ribosomal NusE and Rho for NusG helps to explain why Rho cannot terminate translated transcripts.

Escherichia coli NusG is a two-domain protein (fig. S1) (1) that is essential for cell viability (2). NusG homologs are found in all known bacteria, and the 27–amino acid NusG carboxy-terminal domain (CTD) Kyprides-Onzonis-Woese (KOW) motif is found in proteins from archaea and eukaryotes (35). A sequence highly homologous to NusG amino-terminal domain (NTD) followed by KOW motifs appears in human transcription factor hSpt5 (6). NusG suppresses RNA polymerase (RNAP) pausing and increases elongation rates in vitro. In vivo, it acts in concert with NusA, NusB, and NusE to promote read-through of terminators within ribosomal rrn operons and on the phage λ chromosome, a process that additionally requires the λ N protein (7). NusG activates Rho transcription termination factor in vitro and is necessary for most Rho-mediated termination events in vivo (8, 9). NusG-NTD binds to RNAP and increases the rate of transcription elongation but cannot stimulate termination (1, 10).

The rates of transcription and translation are correlated over a range of different growth rates (11), and NusG was suggested to be involved in this correlation (12). Thus, depletion of NusG slowed the rate of lacZ translation without affecting the rate of lacZ transcription elongation (12). The dual capacity of NusG to act in transcription as well as in translation is shared by the 30S ribosomal subunit protein NusE, which doubles as a component of some transcription elongation complexes (TECs) (13). As a transcription factor, NusE is loaded by NusB onto the boxA sequence within nut RNA (1416) and becomes part of an antitermination complex that includes NusA, NusG, and other cellular factors (7, 17). The NusB:NusE:RNA ternary complex is proposed to associate with RNAP through NusE (7, 18).

Genetic evidence supports an interaction between NusG and NusE. Thus, the nusG4 (S163F) mutation restores λ N antitermination in a nusE71 (nusEA86D) strain (19). We asked whether this genetic interaction reflects a direct physical contact between the proteins. For all experiments, we used the NusE∆loop variant (15), referred to here as NusE. NusE∆loop is fully active for transcription, although it cannot support translation (fig. S2) (20), and its crystal structure is known in the NusB:NusE complex (15). We analyzed mixtures of NusG and the NusB:NusE complex by size exclusion chromatography. A mixture of NusB:NusE and NusG eluted earlier from the column than either NusB:NusE or NusG alone (fig. S3) (20), consistent with formation of a NusB:NusE:NusG complex. To confirm the interaction and to map contact surfaces, we investigated complex formation by NMR. Titration of isolated 15N-labeled NusG-NTD or NusG-CTD with NusB:NusE complex caused chemical shift changes in the 1H, 15N-HSQC (heteronuclear single quantum coherence) nuclear magnetic resonance (NMR) spectrum of NusG-CTD but not of NusG-NTD (figs. S4 and S5). Reverse labeling (15N-NusE or 15N-NusB and unlabeled NusG-CTD) revealed that NusE is the recognition protein in the NusB:NusE complex (figs S4 and S5), suggesting direct NusG-CTD:NusE interaction.

From the chemical shift changes upon titration, we could estimate the dissociation constant for the NusB:NusE:NusG-CTD (molecular mass of 32.3 kD) interaction as Kd = 50 μM (fig. S6). Comparison of secondary chemical shifts and characteristic nuclear Overhauser enhancement spectroscopy (NOESY) cross-peak patterns of NusB:NusE and NusG-CTD with the corresponding data of the NusB:NusE:NusG-CTD complex revealed no substantial conformational changes in any of the participating proteins, indicating that only minor side chain rearrangements are necessary to form the interaction surfaces (Fig. 1). Isotope-filtered NOESY spectra (fig. S7) (21) revealed unambiguous intermolecular contacts: NusG-I164:NusE-M88, NusG-I164:NusE-I100, NusG-T169:NusE-Q99, NusG-P140:NusE-V84, NusG-R167:NusE-V98, and NusG-E172:NusE-S101 (22) (Fig. 2A). Docking of the rigid domains with flexible side chains using the NOESY-derived intermolecular NOEs as distance restraints yielded a distinct conformation of the complex without any violation of the restraints (fig. S8). The NusE-binding region on NusG-CTD is composed of loops between β strands β1 and β2, β3 and β4, and residues from β strand β4. Hydrophobic amino acids P140, F141, F144, I164, and F165 show close contacts to hydrophobic residues of NusE. NusG-R167 is very close to NusE-D87, and NusG-F165 at the tip of the loop between β strands β3 and β4 is buried deeply in a hydrophobic pocket on NusE (Fig. 2, B and C). The NusE hydrophobic pocket is composed of residues from helix α2 and the carboxy-terminal β-strand β4. There were no significant effects on resonances of NusB in the NMR spectra of the NusB:NusE complex upon binding of NusG-CTD. These data together with mapping of the chemical shift changes on the sequences and the known three-dimensional structures (Fig. 1) show that the NusE-binding interface is opposite to the NusB:NusE interaction region, in proximity to the ribosome-anchoring flexible loop, R46 to T67 (fig. S2).

Fig. 1

(A) Mapping of chemical shift changes {∆δnorm [in parts per million (ppm)] > 0.2, red; > 0.1, orange; and > 0.04, yellow} on the structures of NusG-CTD [gray, Protein Data Bank identification code (PDB ID) 2JVV (1)] and the NusB:NusE complex [dark gray and light gray, respectively; PDB ID 3D3B (15)]. Strongly affected residues are labeled (22) and shown as sticks. Coloring scheme is as follows: carbon, red; hydrogen, white; oxygen, red; nitrogen, blue; sulfur, yellow. The light gray sphere in NusE denotes the Cα position of S46, which in this construct replaces residues 46 to 67 of wild-type NusE (20). Numbering is based on the residue numbers of wild-type NusE (1 to 46 and 68 to 103). (B) Surface representation of the structures shown in (A). Orientations relative to (A) are indicated. (C) HSQC-derived chemical shift changes versus sequence position. (Left) NusG-CTD chemical shift changes on titration with NusB:NusE; (right) NusE (in the NusB:NusE complex) chemical shift changes on titration with NusG-CTD; missing residues of the NusE ribosome-binding loop are indicated by small double bars on the sequence axis (right). Dotted line, significance level of ∆δnorm[ppm] = 0.04; red bars, NusG-CTD signals disappearing on complex formation.

Fig. 2

(A) Experimental basis for the structure determination of the NusE:NusG-CTD complex. For clarity, NusB was omitted. Carbon, marine (NusE) and green (NusG); hydrogen, white; oxygen, red; nitrogen, blue; and sulfur, yellow. Black lines represent unambiguously assigned NOEs between NusG-R167-Hδ:NusE-V98-Hγ, NusG-I164-Hδ:NusE-M88-Hε, NusG-E172-Hβ:NusE-S101-Hβ, and NusG-P140-Hα:NusE-V84-Hβ. These intermolecular NOEs allowed unambiguous determination of the relative orientation of the two proteins by rigid body minimization. (B) Possible interactions in the NusE:NusG-CTD interface between negatively charged NusE-D97 and positively charged NusG-R167. NusG-I164 is sandwiched by the hydrophobic side chains of NusE-M88 and NusE-I100. (C) NusG-F165 and NusG-F141 bind to a NusE hydrophobic pocket. NusG-F144 is in close proximity to the interaction site and possibly participates indirectly in this interaction. Several NusE hydrophobic side chains are close to NusG phenylalanine residues. The orientation relative to Fig. 1A is indicated.

NusE residues M88 and D97 and NusG residues P140, F165, and R167 in the NusE:NusG-CTD interface are highly conserved among different bacteria (fig. S9), underscoring the importance of this interaction for bacterial viability. Although F165 was proposed to be required for an intramolecular NTD-CTD domain interaction in Aquifex aeolicus NusG (23), the NusE:NusG-CTD complex structure reveals that F165D or F165T mutations disturb the NusE:NusG-CTD interface.

Coregulation of transcription and translation was initially identified within the attenuation system that controls expression of amino acid biosynthetic operons (24), and polarity was shown to be the result of premature Rho-dependent transcription termination induced by a translation-terminating mutation (25). The reported correlation between the rates of transcription and translation of the infB and lacZ genes, however, implies a direct linkage between the two processes (11). The interaction surface of NusE with NusG (~1100 Å2) is still accessible when NusE forms part of the 30S ribosomal subunit (Fig. 3A). Thus, NusE could mediate simultaneous formation of a NusG-NTD:RNAP complex and a NusG-CTD:ribosome complex (Fig. 3B). This analysis suggests direct physical coupling of transcription and translation via NusG (Fig. 3C), a notion that is supported by the observation that NusG depletion decreases translation elongation rates (12).

Fig. 3

(A) Possible interaction between NusG and NusE/S10 in the 30S ribosomal subunit. The NusE (marine) NusG-CTD (green) heterodimer was aligned to the structure of the E. coli 30S ribosomal subunit (ribosomal proteins, brown; 16S rRNA, rose). Landmark features of the 30S subunit (head, body, foot) are labeled. S10 is part of the head region of the 30S subunit in close proximity to the entrance site of the mRNA. The path of the mRNA is shown in gold. The orientation relative to Fig. 1A is shown. (B) Schematic representation of the assembly of the λ N–mediated antitermination complex (20). NusA domains AR1 (acidic repeat 1), AR2 (acidic repeat 2), and SKK (S1-KH1-KH2) are in light gray; RNAP αCTD, gray; boxA, boxB, and spacer region of nut RNA, gold; λ N, dark gray; NusB, brown; NusE, marine; NusG-CTD, green; NusG-NTD, dark green; and β’-clamp helices of the RNAP, black. (C) Model of the physical coupling between transcription and translation. Coloring as in (A) and (B); the 50S ribosomal subunit is shown in light gray and the A-site tRNA in dark gray. NusG links RNAP and the ribosome so that efficient and fast translation of the nascent mRNA can occur. Coordinates for the complete 70S ribosome, the mRNA, and the A-site tRNA were kindly provided by T. M. Schmeing and V. Ramakrishnan, Medical Research Council, Cambridge, United Kingdom (30).

We next investigated whether the physical coupling of transcription and translation via NusG interferes with the known Rho-related functions of NusG (8, 9). HSQC titrations of 15N-NusG-NTD or 15N-NusG-CTD with Rho showed a nearly complete loss of signals for NusG-CTD but no effect for NusG-NTD (fig. S10). The resonances of NusG-CTD are broadened beyond detection by the dramatic increase of the rotational correlation time on NusG-CTD:Rho complex formation (molecular mass of NusG-CTD is 6.9 kD; molecular mass upon addition of Rho hexamer is 288.9 kD). Thus, Rho binds to NusG-CTD but not to NusG-NTD. Signals of residues from the highly flexible N terminus of NusG-CTD [R123, K125, and T126 (Fig. 4A)] could be observed in the presence of Rho, suggesting that they are located at the surface of the complex and do not contribute to binding. These residues are part of the linker region between NusG-CTD and NusG-NTD in full-length NusG, and their high flexibility combined with their absence from the NusG-NTD:Rho binding interface indicates that NusG-NTD remains flexibly linked to NusG-CTD in the complex with Rho. Indeed, all detectable signals of full-length 15N-NusG after addition of Rho could be assigned exclusively to residues of NusG-NTD and the linker region. HSQC displacement experiments show the NusG-CTD:Rho and the NusG-CTD:NusE interactions to be mutually exclusive (Fig. 4, B and C).

Fig. 4

(A) 1H, 15N HSQC spectra of 15N-NusG-CTD titrated with Rho. Spectrum of free NusG-CTD, gray, and with an additional equivalent of Rho, black. All gray signals disappeared after Rho addition, and the remaining signals (black) are part of the spacer region between the NusG domains. (B) 1H, 15N HSQC spectra of a displacement titration of a 1:1 of 15N-NusG-CTD:Rho complex (gray) with NusB:NusE. Addition of 20 equivalents of NusB:NusE, black. All newly appearing signals either belong to the spacer and the C terminus of NusG or are on the side of the spacer region pointing away from the interaction surface (E154 and N145). (C) 1H, 15N HSQC spectra of a displacement titration of a 1:1 15N-NusG-CTD:NusB:NusE complex (gray) with Rho. Addition of one equivalent of Rho, black. The resulting spectrum is identical to the spectrum observed with the NusG-CTD:Rho complex [part (A) of this figure].

Chromatin immunoprecipitation (ChIP) chip analysis indicates that Rho collocates with elongating RNAP soon after transcription initiation (26, 27). How Rho is recruited to the TEC is not clear. Nevertheless, termination does not occur until transcription reaches the operon terminus. Our findings suggest a straightforward explanation for this delay in termination. We suggest that during coupled transcription-translation NusG-CTD is bound to ribosomal NusE and is therefore unavailable for binding to Rho. Release of ribosomes at the ends of operons frees the NusG-CTD to interact with and stimulate Rho. This mechanism may complement the occlusion of RNA to Rho by translating ribosomes.

The different relative affinities of NusG for NusB:NusE-CTD (Kd = 50 μM) and for Rho [Kd = 12 nM (28)] as reflected in the displacement experiments (Fig. 4) appear to argue against the above model. However, NusE:NusG-CTD interaction takes place within a complex with TEC that includes many other factors. Other interactions with the TEC may additionally lower the NusE:NusG-CTD Kd. For example, during processive antitermination NusB:NusE binds boxA RNA (1416), linking these factors with RNAP and enhancing the overall stability of the antitermination complex. Consistent with the idea that the NusE:NusG interaction is stabilized on RNAP, we found that overproduction of NusG-CTD in wild-type E. coli is not toxic, suggesting that it does not efficiently compete with wild-type NusG for binding to TEC (1). In contrast, NusG-CTD did compete with a NusG mutant with reduced affinity for NusE. Thus, overexpression of NusG-CTD was lethal in a nusGF165A mutant strain, in which a key NusE interface residue on NusG is altered. We suggest that NusG-CTD titrates isolated NusE and/or ribosome-bound NusE in the mutant cells.

Lastly, our data also explain the puzzling observation that ribosome-bound NusE still supports antitermination (29). According to our model, this activity entails RNAP-ribosome coupling through NusG.

Supporting Online Material

www.sciencemag.org/cgi/content/full/328/5977/501/DC1

Materials and Methods

Figs. S1 to S10

Table S1

References

References and Notes

  1. See the supporting material available on Science Online.
  2. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.
  3. We thank R. Heissmann for excellent technical assistance. Financial support by the Deutsche Forschungsgemeinschaft to P.R. (Ro617/16-1), M.C.W. (Wa1126/3-1) and B.L.S. (NSF MCB-0744422) as well as by the NIH to M.E.G. (GM037219) is gratefully acknowledged.
View Abstract

Navigate This Article