Single-molecule insights into torsion and roadblocks in bacterial transcript elongation

ABSTRACT

During transcription, RNA polymerase (RNAP) translocates along the helical template DNA while maintaining high transcriptional fidelity. However, all genomes are dynamically twisted, writhed, and decorated by bound proteins and motor enzymes. In prokaryotes, proteins bound to DNA, specifically or not, frequently compact DNA into conformations that may silence genes by obstructing RNAP. Collision of RNAPs with these architectural proteins, may result in RNAP stalling and/or displacement of the protein roadblock. It is important to understand how rapidly transcribing RNAPs operate under different levels of supercoiling or in the presence of roadblocks. Given the broad range of asynchronous dynamics exhibited by transcriptional complexes, single-molecule assays, such as atomic force microscopy, fluorescence detection, optical and magnetic tweezers, etc. are well suited for detecting and quantifying activity with adequate spatial and temporal resolution. Here, we summarize current understanding of the effects of torsion and roadblocks on prokaryotic transcription, with a focus on single-molecule assays that provide real-time detection and readout.

Introduction

Topological arenas of transcription

Topological domains in nucleoids typically consist of hierarchical DNA loops formed by abundant proteins that bind both specifically and nonspecifically to DNA [1]. For example, H-NS and IHF proteins bind AT-rich DNA throughout the genomes of Escherichia coli, while HU and StpA bind nonspecifically [2,3]. In E. coli, abundant nucleoid-associated proteins (NAPs) are bound to chromosomal DNA with an average coverage of at least one NAP per 100 base pairs of DNAs [4,5]. Some of these proteins may bend and/or bridge distant DNA segments facilitating DNA looping.

Recent studies indicate that the E. coli genome is organized into four large, defined regions, termed macrodomains, mediated by specific DNA-binding proteins, such as MatP, SeqA, SlmA [6–10]. Transcriptional repressors, such as the AraC, Lac, Deo, Gal repressors, mediate smaller loops of a few hundred base pairs in length that regulate transcription at the single operon or gene level [11–13]. Large or small, protein-mediated loops are likely to be affected by and trap supercoiling. Although there exist thermophilic organisms whose genome is overwound to stabilize the double helix at high temperatures [14,15], the prokaryotic nucleoid is generally unwound [16–19] and forms right-handed (-) plectonemes, regions where the main axis of the double helix is intertwined. Within looped topological domains, like in closed circular plasmids, RNAP translocation coupled to RNA synthesis causes winding of the DNA helix downstream and unwinding upstream [20]. These changes in twist on either side of RNAP cause accumulation of torsional stress and, eventually, may lead to the formation of plectonemes which alter physical distances between DNA elements [21–23].

Transcription and supercoiling

Throughout this review, we use the term “supercoiling” to indicate either variation from the canonical twist value of the DNA double helix of approximately 10.4 bp/turn (change in twist), or the introduction of writhe, or both. Quantitatively, the torsional state of DNA is described by its linking number, Lk. The linking number of relaxed, B-form DNA, Lk₀, is given by the number of base pairs divided by the helical pitch, 10.4 bp/turn, and it equals the twist, Tw₀ (Lk₀ = Tw₀). However, Lk₀ may be changed by winding or unwinding (non-nicked) DNA, thereby changing its twist. This operation introduces torque in the DNA and beyond a critical torque, writhe (plectonemes or supercoils) appears; thus, ΔLk = ΔTw + Wr [24]. The superhelical state, or superhelicity, of a molecule is then given by ΔLk/Lk₀, or (n–n₀)/n₀, where n is the number of turns inserted or removed from the molecule and n₀ is the number of turns present in the molecule in its relaxed state.

Torsional stress can be produced by many enzymes. Helicases can unwind the double-helical genome that is topologically constrained by proteins that wrap, loop [25,26]. Furthermore, some topoisomerases, such as gyrase, introduce or remove plectonemes [27,28]. Transcribing RNAPs generate positive supercoils ahead, and negative behind, as they track along the helical groove of DNA, forming a “twin supercoiled domain” [29] (Figure 1). This torsional stress, in turn, is likely to alter transcription by modulating the structure of the DNA duplex and its binding affinity for proteins in the nucleoid that may facilitate or interfere (roadblocks) with RNAP activity. Indeed, the supercoiling level of DNA modulates the affinity of proteins that interact with DNA affecting transcription, as reported in the case of the OmpR, PecT and H-NS proteins [30–33]. The interplay between supercoiling and transcription is further regulated by enzymes such as topoisomerases which catalyze strand passage reactions to control torsional stress [34–36].

Figure 1.

A schematic illustration of the interplay between RNA polymerase elongation, topological DNA structures, and regulation by representative proteins and enzymes. During transcription of the torsionally constrained DNA segment, negative supercoiling accumulates upstream while positive supercoiling accumulates downstream. Meanwhile, topoisomerases regulate the accumulated torsional stress. lac repressor is used here as an example of a transcriptional roadblock and a DNA loop-mediating protein. Negative supercoiling enhances lac repressor binding whereas positive supercoiling produces a shallower minor groove that attenuates binding, as shown in the blue and red dashed boxes, respectively

Roadblocks

For the purposes of this review, a “roadblock” is a physical barrier along the DNA template, which must dissociate or be dislodged for an elongating RNA polymerase to proceed. Roadblocks can be DNA-binding proteins like repressors (see below), nucleoid architectural proteins [37], proteins that direct replication [38], DNA-based molecular motors [39], RNA-directed dCas9 [40], and those factors which interact with transcription factories inside the dynamic, structured nucleoid [18,41–44]. Roadblocks can cause RNAP to pause and backtrack, processes which were recently reviewed [45].

Some examples of roadblock proteins in E. coli include lac [46], trp [47] and pruB [48], with the lac repressor–operator interaction being one of the best characterized both in vivo and in vitro. The lac repressor bound to the operator O1 has been shown to be a very strong roadblock even in engineered systems [49]. Recently, it was also shown that lac repressor is a strong roadblock when bound at a weak operator (O2) if mediating a loop [50]. This prompts the idea that interactions with two or more sites as seen in both cis – and trans-interactions changes the effectiveness of protein roadblocks. When closing a loop, a roadblock also forms a domain that can trap supercoiling, thus altering the activity of the transcription elongation complex within the loop [51]. Thus, the various DNA topologies that can be induced by roadblocking proteins may exert steric effects and alter binding affinity as well.

In the following, the section “Single-molecule techniques for transcription measurements” reviews which different single-molecule techniques have been used to study transcriptional regulation by DNA looping, supercoiling, and torque and what unique insights they provided. The section “Transcription against torsion and through roadblocks” summarizes insights, derived from single-molecule measurements, on the role of torsion and protein, as well as topological, roadblocks in the regulation of transcription.

“Conclusions and Outlook” highlights how single-molecule methodologies have contributed to an accurate mechanistic picture of the structural and physical parameters underlying transcription and the need of such studies to quantitatively determine the dynamic interplay between transcription and topology.

Single-molecule techniques for transcription measurements

Transcription as well as loops produced by several transcriptional regulators, including the lac and the lambda cI repressors, are amenable to be studied using the tethered particle motion (TPM) technique, which relates the Brownian motion of an attached bead to the contour length of the tethering polymer and provides a simple way to monitor loop formation and breakdown as well as transcription in DNA templates under negligible tension [52–58]. Other techniques, such as atomic force microscopy (AFM) [59,60] (Figure 2b), magnetic tweezers (MTs) [25,61–64] (Figure 2 C-E), optical tweezers [65–67] (Figure 2 F-H), nanofluidic confinement [68], and in vivo single-molecule imaging [69], have also been used to monitor transcription and regulation by protein-mediated looping [70]. The tension on the molecule is a critical parameter that affects the choice of a technique. While RNAP elongation through relaxed loops can be monitored using TPM or AFM, supercoiling is needed to promote protein-mediated looping in DNA under even slight tension from tweezers [23]. The capability to arbitrarily supercoil and stretch DNA with magnetic tweezers [71,72] has made them the instrument of choice for investigating the interplay between tension, torsion, and DNA structure [73–81].

Figure 2.

Single molecule techniques used to study transcription, transcription through roadblocks, or against torsion*. The tethered particle motion technique (a), atomic force microscopy (b), magnetic tweezers* (c), magnetic tweezers with nanorod* (d), magnetic torque tweezers* (e), single and dual trap optical tweezers (f), optical tweezers and rotating pipette (g), angular optical trap* (h), total internal fluorescence and FISH (i)

Supercoiling favors looping

Supercoiling can increase the probability of juxtaposition of protein-binding sites by reducing the physical distance between them, thus enhancing the protein-mediated looping [82,83]. Indeed, MT experiments have shown that negative DNA supercoiling facilitates DNA looping [22,61], and supercoiling can even turn loop formation into a deterministic process, in the sense that a definitive probability correlates to specific conditions of loop length, binding affinity, supercoiling, and observation interval [84]. MTs experiments have also shown that loop formation is very sensitive to tension. In two independent works, it was shown that tensions of 200 fN and 500 fN can increase the time required for formation of a 100 bp loop by one and two orders of magnitude, respectively [85,86]. The random coiling associated with a fluctuating cellular environment was shown to greatly enhance the formation of loops, while tension attenuated looping by canceling the effect of fluctuations [57,87]. Loop formation is also sensitive to protein-protein cooperativity [88,89] as well as the size of bridging proteins and kinks [90].

Measuring torque in single DNA molecules

DNA supercoils result from torsional stress and magnetic tweezing is a straightforward assay with which to quantitatively show how torque changes the double helix. Most configurations of MT apply torsion (Figure 2c), but either the sample or the magnet must be modified for the direct measurement of torque. For example, Celedon et al. introduced a modified MT technique, which uses a cylindrical magnet and a nanorod coupled magnetic bead (Figure 2d) to allow direct measurement of torque [76]. Lipfert et al. introduced another modification of MT, magnetic torque tweezers (MTT, Figure 2e) in which they designed a magnet to apply axial tension and slight rotational trapping and used magnetic beads with small rotational fiducial markers to enable direct measurement of torque [91].

Optical tweezers may utilize one or two traps to apply tension to single DNA molecules (figure 2f), and innovative ways of introducing and measuring torsion using optic-based techniques have been developed. One utilizes the application of high tension to produce intermittent dissociation of anchoring biotin-streptavidin linkages that allows the DNA to unwind for some calibrated time interval before tension is lowered and the linkages re-form [92]. Bryant et al. used a rotating micropipette to build up torsional strain in an optical trap and calculated torque from the angular velocities of “rotor” beads attached along the DNA (Figure 2g) as it torsionally relaxed [93]. Wang and colleagues introduced an angular optical trap (AOT) with laser intensity-modulated force and torque control and high-resolution readout (Figure 2h). The method relies on a nanofabricated quartz cylinder, which can be trapped by a linearly polarized light and rotates with the plane of polarization [66,94–96]. A comparison of the specifics of these techniques may be found elsewhere [97].

Torque induces phase transitions in DNA

These techniques can be used to probe the configurational response of DNA to torque and tension and understand how interactions between DNA and its molecular interactors (including RNAPs) may change as a function of these two parameters. Single-molecule approaches also reveal structural changes that transcription may produce by modulating local or global genome supercoiling and/or tension. The torsional modulus of DNA, and the buckling torque (the value of torque beyond which the DNA writhes to form the first plectonemic loop) were measured to be ∼400 pN·nm[2] and ∼+34 pN·nm, respectively, as reviewed by Ma et al. [93,95,96,98]. Lipfert et al. found that both the buckling torque and torsional stiffness depend on tension within the DNA in agreement with results obtained with traditional MT, MTT, AOT, and MT with nanorod-coupled beads [91]. It is important to keep in mind that these results are highly dependent on cationic conditions [99,100]. Single-molecule techniques were extensively used to study over – and under-wound DNA structures and the experimentally observed phase transitions confirmed the theoretical phase diagram (Figure 3). Under high levels of tension, that may be physiologically unlikely, B-form DNA transitions to the S-form, or to P-form if highly overwound [101,102]. Under low tension B-form DNA transitions to a plectonemic (supercoiled) form (sc), and to L-form (left-handed DNA, supercoiled or not depending on the force) when further unwound [101,102]. The transitions exhibit sequence-dependence [96,98] and physiological levels of tension and torque can produce left-handed DNA [103]. L-DNA is likely to form in transcriptionally active domains with high levels of negatively supercoiled DNA, such as those found by Naughton & Gilbert and by Lal et al. [19,104]. These insights, which could have not been attained using bulk approaches, provide a basis for describing the mechanisms underlying changes in transcription as a function of supercoiling.

Figure 3.

Diagram of force-torque phase transitions. Transitions occur along blacklines. The critical torque associated with a phase transition can be measured as a function of force using single-molecule techniques. See main text for details

Transcription against torsion and through roadblocks

Torsion

Prior to the advent of single-molecule techniques, studies to investigate the torsional effects of transcription could not produce direct observations or direct measurements of transcription-generated torque [29,105,106]. In contrast, using small fluorescent beads stuck on the surface of a larger magnetic bead tethered to the downstream end of a DNA template, Harada et al. directly observed tethered beads rotating continuously and clockwise when NTPs were added to initiate transcription, and RNAP produced positive supercoiling with torque greater than 5 pN-nm [107]. Ma et al. subsequently used AOT (Figure 2h) to study transcription-generated torsion and observed that RNAP generates ~11 pN-nm torque before stalling, an amount sufficient to melt a random DNA sequence. About half of the stalled RNAPs resumed transcription when torque was released after 5 s. Although excessive torsion can be relieved by topoisomerases in vivo [16] or reduced by motor proteins producing opposite winding, localized torsion does accumulate transiently. Levens and colleagues found that transient supercoils generated by transcription along a linear DNA template are sufficient to unwind an upstream supercoil-sensitive element [108,109]. This finding emphasizes that supercoiling may affect DNA structure and regulate DNA-based transactions in a dynamic manner [110].

Among all the stages of transcription, initiation is likely the most torsion-sensitive, because it involves the unwinding of 10 ~ 12 bp, approximately one turn, of DNA at a promoter site. Formation of the open promoter complex can shorten DNA templates by wrapping certain promoters, but this is not a general feature of such complexes [111]. Unwinding one turn of DNA leads to an observable change in the end-to-end extension of a plectonemic DNA molecule, and magnetic tweezer experiments have shown that positive DNA supercoiling destabilizes and slows the formation of open promoter complexes [112]. Torque insufficient to produce plectonemes delayed bubble formation and decreased the lifetime, which refuted models based on positions of plectonemes. Chong et al. [113] used a total internal reflection fluorescence (TIRF) microscope (Figure 2i) combined with the fluorescence in situ hybridization (FISH) method to measure the efficiency of transcription initiation on a torsionally constrained circular DNA template. They found that the rate of initiation drops substantially and eventually stops when Topo I is added to remove negative supercoiling and positive supercoiling generated by multiple rounds of transcription accumulates. However, some promoters show enhanced expression when relaxed compared to when they are negatively supercoiled [114]. Changes in the helical structure of the DNA that alter affinity/activity of RNAP and transcription factors may underlie such changes [115]. The change in the level of supercoiling experimentally detected during initiation, associated with unwinding and drawing in DNA without progressive translocation, supports a “scrunching” model and is inconsistent with inchworm or transient excursion models for RNA polymerase behavior [116]. The conversion of an initiation complex to an elongation complex occurs only after the synthesis of 8–12 nucleotides of RNA following multiple rounds of abortive initiation [117]. Single-molecule assays with fluorescently labeled proteins were also used to examine proposed mechanisms of transcription initiation and promoter escape. During the transition from initiation to elongation, a large percentage of fluorescently labeled RNAPs were found to retain the σ[70] subunit during initial and even subsequent elongation [118,119].

Single-molecule assays were also widely used in the study of enzymes that modulate torsional stress in DNA. MTs have been used to observe the activity of topoisomerases IA [120], IB [121], II α [35], IV [122], V [123], helicases [124,125], and reverse gyrase [34,126]. These enzymes affect DNA-based transactions, such as transcription, modulating supercoiling and consequently the expression levels of certain genes [127,128]. In the case of DNA gyrase, a type II topoisomerase, single-molecule imaging suggests that at any time at least 300 gyrase molecules are stably bound to the E. coli chromosome to relieve the supercoiling induced by transcription [129]. Inhibition of topoisomerase activity causes DNA to become hyper-negatively supercoiled upstream of promoters, thus stimulating transcription from supercoiling sensitive promoters [130]. However, the hyper-negative supercoiling can also lead to the formation of R-loops, which inhibit transcription elongation and/or induce backtracking of RNAP and termination [131,132].

Roadblocks

Here, we defined a roadblock as a steric barrier to elongation, which must be dislodged for RNAP to proceed. Studies using AFM and biochemistry have independently shown that formation of a lac repressor-mediated DNA loop can block transcription elongation much more efficiently than the repressor bound to a single site [26]. In those experiments the DNA was not torsionally constrained. On the other hand, a recent study has shown that elongation along a sequence of a few hundred base pairs requires more time when the sequence is looped and that positive supercoiling generated by transcription likely facilitates displacing the lac repressor to exit the loop [51]. The effect of supercoiling on binding affinity has been reported for a variety of NAPs and other transcription factors [133,134]. For example, both Fis (factor for inversion stimulation) and H-NS (histone-like nucleoid-structuring) have been shown to be sensitive to the topological state of DNA for the control of pel genes in Dickeya dadantii [33]. Instead, for the control of the SPI-1 and SPI-2 genes in Salmonella Typhimurium, while Fis was found to be sensitive to DNA topology, H-NS, which generally stiffens DNA as well as bridges different DNA segments, seems to bind regardless of DNA topology [133]. An example of transcription factor with binding affinity sensitive to DNA topology is the phage 434 repressor which showed an enhancement in binding affinity of up to nearly 5-fold as its binding site went from being relatively underwound to overtwisted [134].

Interesting questions about transcription past a roadblock are whether RNAP actively dislodges roadblocks, or passively takes advantage of their spontaneous dissociation [40,135,136], and if the activity of RNAP and/or the binding affinity of roadblocks is/are dependent on the tension in the DNA template. These questions lack definitive answers, but several studies have revealed some rules for traffic on DNA. Diffusion of DNA loops about the nucleosome to allow repositioning of the histone octamer roadblock was observed for nucleosomes in eukaryotes [137], but as discussed by Vörös et al., is unlikely to occur for site-specific repressors [26]. A widely accepted Brownian-ratchet model of transcript elongation holds that single base pair sliding movements interconvert the elongation complex between pre – and post-translocated states, thus conditions favoring NTP binding might rachet up the pressure of RNAP on DNA-binding roadblocks to dislodge them [138]. Raising the binding affinity of a roadblock (decreasing the off-rate) is shown to produce stronger obstruction than increasing the concentration of roadblocks (increasing the on-rate), suggesting that transit through roadblocks likely involves their dissociation [139]. In addition, a burst of transcription from a strong promoter can increase transit through a roadblock, confirming cooperation by transcription elongation complexes and favoring the mechanism of active dislodgement [136,140,141]. Nonetheless, these results cannot completely exclude the possibility that roadblocks spontaneously dissociate, and trailing RNAPs reactivate the roadblocked RNAP complexes ahead.

Of course, RNAP translocation may align with or against the movements of other molecular motors along DNA (Figure 4). Convoys of RNAPs have been shown to translocate faster than a single RNAP, while divergently transcribing RNAPs tend to arrest and terminate prematurely [142]. Thus, torsion influences roadblocks, and promoter repression leads to premature dissociation of open transcription complexes. This effect is abrogated by the presence of topoisomerase. These results suggest that DNA-binding molecules that alter transcription or supercoiling levels can affect transcription at a distance within topologically constrained domains.

Figure 4.

Supercoils of opposite handedness emanating from tandemly transcribing RNA polymerases create local domains that may alter the binding affinity of DNA-binding proteins

Conclusion and outlook

This review highlights the insights produced by single-molecule investigations on the mechanisms and DNA structures that affect transcription elongation focusing particularly on protein-mediated DNA looping that generates topological domains, protein roadblocks, and DNA supercoiling. Protein mediated loops regulate transcriptions at various levels and their formation is influenced by DNA supercoiling. Supercoiling generated by transcription constitutes a feedback mechanism that can modulate the affinity of DNA-binding proteins and bias DNA looping, but DNA-binding protein obstacles and loops can reciprocally regulate transcription (Figure 5). Thus, there is a dynamic interplay between DNA supercoiling and looping. In addition, supercoiling DNA under tension can trigger phase transitions in the DNA, which may affect transcription and its interactions with proteins. Single-molecule techniques provide exquisite control of tension and torsion in DNA that led to the first quantitative measurements of the positive and negative torques that stall an elongating RNAP. These are not just quantitative trivia; they are surprisingly low thresholds of torsional stress that are routinely generated in the genome by RNA polymerases. They underscore the need for topoisomerases that relax torsion, and single-molecule studies have shown that the availability of gyrase in bacteria regulates transcriptional bursting [113].

Figure 5.

Schematic diagram of the interplay between the protein-mediated topological domains in the prokaryotic genome and transcription-generated supercoiling. The “AND” gates indicate that supercoiling produces functional outcomes when elongation occurs in a topological domain, otherwise supercoiling disperses

Protein roadblocks also affect the production and co-transcriptional folding of nascent RNA by inducing pauses. Single-molecule studies have revealed the dynamics of elongation including the frequency and sequence dependence of RNAP pausing with otherwise unattainable spatial resolution [45,143], and studies of how RNAP effectively transits roadblocks are ongoing [51]. DNA sequences may also constitute important parameters in directing the supercoiling level of DNA by not only encoding the sites of promoters and terminators but also those of roadblocks and topoisomerases. Recent reports on the biogenesis and utilization of diamino purine (DAP) in place of adenine as well as the presence of numerous bacteria and viruses with DAP-substituted genomes has called attention to the physical properties of this alternative form of DNA. In this case too, single-molecule studies have provided the first understanding of the processing of this DNA by topoisomerases and its response to unwinding [144–146] and are uniquely positioned for the investigation of transcription of this type of template.

Single-molecule techniques, which allow arbitrary manipulation of nanomechanical parameters, such as tension and torsion, and resolutions as high as one base pair, can produce mechanistic information about transcription, for example, how and if RNAP transits obstacles found along the DNA template, and how different sequences [147], tertiary structures, and defects or DNA modifications affect elongation. The power of these approaches will allow detailed mechanistic analysis of increasingly complex reconstituted systems.

Funding Statement

This work was supported by the National Institute of General Medical Sciences [R01GM084070].

Disclosure statement

No potential conflict of interest was reported by the authors.