Reading of the non-template DNA by transcription elongation factors

Summary

Unlike transcription initiation and termination, which have easily discernable signals such as promoters and terminators, elongation is regulated through a dynamic network involving RNA/DNA pause signals and states- rather than sequence-specific protein interactions. A report by Nedialkov et al. (in press) provides experimental evidence for sequence-specific recruitment of elongation factor RfaH to transcribing RNA polymerase (RNAP) and outlines the mechanism of gene expression regulation by restraint (“locking”) of the DNA non-template strand. According to this model, the elongation complex pauses at the so called “operon polarity sequence” (found in some long bacterial operons coding for virulence genes), when the usually flexible non-template DNA strand adopts a distinct hairpin-loop conformation on the surface of transcribing RNAP. Sequence-specific binding of RfaH to this DNA segment facilitates conversion of RfaH from its inactive closed to its active open conformation. The interaction network formed between RfaH, non-template DNA, and RNAP locks DNA in a conformation that renders the elongation complex resistant to pausing and termination. The effects of such locking on transcript elongation can be mimicked by restraint of the non-template strand due to its shortening. This work advances our understanding of regulation of transcript elongation and has important implications for the action of general transcription factors, such as NusG, which lack apparent sequence-specificity, as well as for the mechanisms of other processes linked to transcription such as transcription-coupled DNA repair.

In the cell the process of transcribing genomic DNA into RNA molecules forms the basis of gene expression and its regulation. It also serves as a starting point for many other cellular processes, not the least among them are the mechanisms responsible for the detection and repair of DNA damage. Transcript elongation, which begins after RNAP escapes the promoter and ends at the terminator, is by far the longest step of RNA synthesis. Our understanding of elongation has evolved significantly over the past two decades (Nudler, 2012), yet the elucidation of regulatory signals during transcript elongation and how the elongation factors recognize and act upon them remain subjects of active investigation.

The report by Nedialkov et al. (in press) highlights a novel mechanism of transcription elongation regulation by the bacterial virulence regulator RfaH (Bailey, Hughes, & Koronakis, 2000). They propose that elongation of transcription can be directed towards processive NTP addition or forced into one of the off-states by “locking” the non-template (NT) DNA strand in alternate distinct conformations, and that RfaH, by binding to the transcription elongation complex (TEC), can lock the NT strand in a persistent on-pathway state. RfaH makes a perfect subject for the study of regulators interacting with the NT strand of the TEC: its binding to the exposed NT strand is well-documented (Artsimovitch & Landick, 2002; Sevostyanova, Svetlov, Vassylyev, & Artsimovitch, 2008; Zuber et al., 2018). Despite study by primarily only a single research group, RfaH structure, function, and mechanism of action are elucidated in greater detail than perhaps any other elongation factor (Belogurov, Sevostyanova, Svetlov, & Artsimovitch, 2010; Burmann et al., 2012; Hu & Artsimovitch, 2017; Sevostyanova, Belogurov, Mooney, Landick, & Artsimovitch, 2011). Studies of the role of RfaH in regulation of transcription have unusually broad implications, as RfaH belongs to the only family of transcription factors (NusG/Spt5) conserved in all three domains of life (NandyMazumdar & Artsimovitch, 2015; Werner, 2012).

NusG, an RfaH paralog and a general elongation factor, has also been proposed to bind the TEC NT strand (Belogurov & Artsimovitch, 2015), but there is little evidence for such binding in the fragmented and often contradictory compendium of NusG studies. The recent structural models of NusG bound to TEC (Said et al., 2017) or RNAP core (Liu & Steitz, 2017), do not explicitly address the question of NusG binding to the NT strand, because the relevant part of the transcription bubble is either disordered or absent altogether (Fig. 1). To date, the mechanistic explanations of NusG anti-pausing activity have focused on its interactions with TEC, resulting in disparate models of NusG acting by stabilizing the closed conformation of the RNAP clamp, antagonizing off-pathway bending of the bridge helix, or facilitating refolding of the trigger loop (Grohmann & Werner, 2010; Jovanovic et al., 2011; Martinez-Rucobo, Sainsbury, Cheung, & Cramer, 2011; Nayak, Voss, Windgassen, Mooney, & Landick, 2013; Weixlbaumer, Leon, Landick, & Darst, 2013). It bears noting that these models were inspired by divergent structures (assumed to represent different functional states of TEC), observed in a variety of RNAP complexes, which were often incomplete, of low resolution and/or quality, and/or non-functional.

Figure 1. NusG in structural models of the TEC.

NusG (purple cartoon)–TEC (teal* cartoon) coordinates, extracted from 5ms0, were superposed onto NusG (green cartoon)-RNAP (hidden), extracted from 5tbz. *Template and non-template strands of DNA are, respectively, yellow and red cartoons with rings. *RNA is the green cartoon with rings.

The role of the exposed NT strand in regulation of transcription has been largely overlooked for several possible reasons: the exposed NT DNA usually appears disordered in crystallographic models, or has been partially or entirely omitted by design to simplify the TEC formation and/or crystallization; in vitro, TEC is often assembled on a pre-formed transcription bubble, where the NT strand is not complementary to the template, thereby preventing re-annealing of DNA; another popular method to start transcription in vitro, primer extension, dispenses with the downstream NT strand altogether; and ontologically, the study of transcription elongation has been dominated by the search for protein-protein interactions and protein-based enzymatic functions. Recently, the importance of clamp closing and opening in toggling TEC between processive elongation and pausing, as well as the bridge helix invading the NTP condensation space and the fraying of the RNA 3′ end from DNA-RNA hybrid (Martinez-Rucobo et al., 2011; Zhang & Landick, 2016), has been challenged by a number of structural and biochemical studies (Guo et al., 2018; Kang et al., 2018). A report by Artsimovitch and colleagues (in press) provides experimental evidence supporting an alternate mechanism of switching the TEC between these two states wherein the NT strand can be restrained (“locked”) in one of the binary on-/off-pathway conformations through interactions with RfaH.

Recruitment of RfaH to the TEC is mediated by sequence-specific interactions with the solvent-accessible NT DNA strand of the ops (operon polarity suppressor) regulatory site found in a subset of E. coli operons (Bailey et al., 1997; Artsimovitch & Landick, 2002). A recent report by Zuber et al. outlines the structural basis for this sequence specificity: the RfaH N-terminal domain (NTD) binds to the ops NT DNA, which adopts a hairpin-like structure and makes specific contacts with 4 nucleotides comprising the hairpin loop (including a sequence-specific H-bond network involving 2 central bases) (Zuber et al., 2018). The rigid hairpin structure of the solvent-accessible NT strand and the network of NT DNA-RfaH NTD interactions provide the framework for restraining this otherwise flexible single-stranded DNA. Sequence-specific interactions appear to be essential only for the initial recruitment of the full-length RfaH, which then continues to suppress transcription pausing and termination throughout the entire operon. It is likely that the extra energy provided by the more extensive sequence-specific interaction network is needed to offset the cost of dissociation of RfaH domains (closed-to-open state transition). In vitro, the RfaH NTD domain by itself behaves as an ops-independent transcription factor, similar to E. coli NusG (which predominantly exists in the open state), indicating that once the NT strand is locked in the on-pathway conformation, the requirements for sequence specificity and hairpin formation may be relaxed. The C-terminal domain (CTD), upon dissociation from the NTD, undergoes transition from an all-α to an all-β fold and is free to interact with other proteins (e.g. ribosomal S10) (Burmann et al., 2012). As a result of this structural transformation and alternate set of protein-protein interactions, the CTD no longer competes with DNA and RNAP for binding to the NTD (Figure 2). A detailed evaluation of the rearrangement of the initial recruitment complex into the long-lived anti-pausing state awaits a comparison between the structures of RfaH complexed with TEC at the ops site and after escape from it. The lack of sequence specificity of NusG action on TEC appears not to be universal, as its orthologs from B. subtilis and S. cerevisiae were reported to be recruited to a specific site (Crickard, Fu, & Reese, 2016, Mondal, Yakhnin, & Babitzke, 2017). In view of these findings it would be very instructional to obtain and evaluate structural models of E. coli and B. subtilis NusG proteins bound to their cognate TECs.

Figure 2. Model of RfaH bound to the TEC.

RfaH (purple cartoon)–TEC (teal* cartoon) coordinates, extracted from the model in Nedialkov et al., (in press). *Template and non-template strands of DNA are, respectively, yellow and red cartoons with rings. *RNA is the green cartoon with rings.

Nedialkov et al. reported biochemical evidence and computational analysis of the RfaH-bound TEC that support the role of the RfaH NTD in restraining the solvent-accessible NT strand and the mechanistic link between this restraint and anti-pausing effects of RfaH in vitro. They demonstrated that the RfaH-dependent locked state of the NT DNA was different from that in the “scrunched” TEC, which, similar to the RfaH-bound complex, produced an extended Exo III protection pattern and was implicated in transcriptional regulation (Duchi et al., 2016; Strobel & Roberts, 2014). The mechanistic importance of restraining/locking of the NT DNA was further supported by evidence that the RfaH effects on transcription elongation could be successfully mimicked by reducing the length of the NT strand exposed on the surface of the TEC by 4 or 5 nucleotides. These data are a strong indicator that NT toggling between flexible and restrained states is the physical switch between the pausing-prone and pausing-resistant states of TEC. Dependence of this transition on RfaH-RNAP (mainly the RNAP gate loop (NandyMazumdar et al., 2016)) interactions was also demonstrated, providing further support of a model in which the NTD interactions with both DNA and RNAP are required for RfaH recruitment to the TEC and the subsequent rendering of the TEC resistant to pausing and termination.

Nedialkov et al. hypothesized that NT DNA locking/unlocking might be employed by other TEC-binding regulators, such as UvrD. According to the new model of bacterial transcription-coupled DNA repair (TCR), UvrD binds to the RNAP and traverses the transcribed regions until the TEC encounters and stalls at a site of DNA damage. UvrD then pulls the TEC backwards from the site of the damage using its helicase motor activity, simultaneously recruiting other proteins involved in nucleotide excision DNA repair (UvrA and UvrB) (Epshtein et al., 2014; Kamarthapu et al., 2016; Pani & Nudler, 2017). The exact mechanism of signal transduction from the stalled TEC to UvrD that would result in activation of the UvrD helicase function remains poorly understood. The locking/unlocking of the NT strand during toggling between the active and paused/stalled states of TEC offers a plausible mechanism for such signaling: UvrD remains in the inactive form bound to the active TEC during transcription (the NT strand is locked by NusG and thus inaccessible) through protein-protein interactions with RNAP; upon encountering a DNA damage site, the TEC pauses/stalls, thereby “unlocking” the NT strand, which (together with the upstream DNA fork) then becomes a substrate for UvrD binding and helicase action (Fig. 3). Notably, UvrD was shown to destabilize NusG binding to the TEC in vitro (V. S. and E. N., unpublished observation), which, given the proposed role of NusG in locking the NT strand in actively transcribing TEC, is consistent with this mechanism of TEC-UvrD signaling. Although this model of UvrD activation during TCR is hypothetical, it does offer a number of testable predictions: e.g., by analogy of its resistance to the action of RfaH, the TEC in which the NT strand has been artificially shortened from 9 to 4–5 nucleotides should be less efficient in the activation of UvrD-dependent backtracking.

Figure 3. Role of the NT DNA “locking” in the regulation of transcription and the TCR.

The NT DNA strand exposed on the surface of the TEC can adopt two distinct conformations: “locked”, which is favored by TEC-bound elongation factors RfaH and NusG, and biased towards processive, pause-resistant transcript elongation, and “unlocked”, flexible conformation associated with the paused TEC. The latter conformation can be recognized by DNA repair factor UvrD, which then switches into the “activated” state and initiates TCR.

Altogether, the findings reported by Nedialkov et al. (in press) provide important mechanistic insights into the action of regulators of transcription elongation (general and operon-specific), as well as pathways branching from the TEC, such as TCR.