Single-molecule approach for studying RNAP II transcription initiation using magnetic tweezers

Abstract

The initiation of transcription underlies the ability of cells to modulate genome expression as a function of both internal and external signals and the core process of initiation has features that are shared across all domains of life. Specifically, initiation can be sub-divided into promoter recognition, promoter opening, and promoter escape. However, the molecular players and mechanisms used are significantly different in Eukaryotes and Bacteria. In particular, bacterial initiation requires only the formation of RNA polymerase (RNAP) holoenzyme and proceeds as a series of spontaneous conformational changes while eukaryotic initiation requires the formation of the 31-subunit pre-initiation complex (PIC) and often requires ATP hydrolysis by the Ssl2/XPB subunit of the general transcription factor TFIIH. Our mechanistic view of this process in Eukaryotes has recently been improved through a combination of structural and single-molecule approaches which are providing a detailed picture of the structural dynamics that lead to the production of an elongation competent RNAP II and thus, an RNA transcript. Here we provide the methodological details of our single-molecule magnetic tweezers studies of transcription initiation using purified factors from Saccharomyces cerevisiae.

1. Introduction

Single-molecule approaches allow for the observation of individual protein-DNA complexes to be followed as a function of time. These experimental techniques can be subdivided into fluorescent-based and manipulation-based strategies, where the later combines the application of forces and torques with measurements of changes in distances germane to the system under study. The manipulative modalities include AFM, optical tweezers, and magnetic tweezers and each have been successfully applied to a wide range of biochemical systems including DNA and protein polymers [1–5], replication [6], transcription [7–9], translation [10,11], cytoskeletal motors [12,13], topo-isomerization [14–16], and nucleosome remodeling [17]. The main advantages of pursuing single-molecular analyses include the observation of unsynchronized dynamic events on the biochemical pathway of interest and the measurements of observable distributions instead of means. The application of forces further allows one to control the state of the molecules under study. For example, in the case of a DNA substrate in the magnetic tweezers, both the tension and the super-helicity may be controlled allowing for the influence of these variables on the activity of the system to be determined [18–20].

In our recent work, we have applied single-molecule magnetic tweezers to the study of transcription initiation in Eukaryotes [21]. The control of messenger RNA (mRNA) concentration is primarily controlled via the regulation of transcription initiation. Each initiation event leads to the production of a single RNAP II enzyme tracking along a gene and generating a single mRNA molecule with a nucleotide sequence complementary to that of the template DNA strand. This process exhibits rich and complex regulation including the action of promoters [22], enhancers and activators [23], co-activators [24], insulators [25], and protein modifications [26]. However, at the center of each initiation event is the requirement to unwind the promoter DNA to allow for the loading of the single-stranded template in the active-site cleft of RNAP II. In Bacteria, this crucial step proceeds spontaneously via a coupled conformational change involving RNAP, the sigma-specificity factor, and the DNA itself [27]. In Eukaryotes, the formation of a ~1.5 mega-Dalton complex called the pre-initiation complex (PIC) is required to stimulate promoter unwinding [28]. In the yeast S. cerevisiae, this complex consists of TATA-box binding protein (TBP), TFIIB, TFIIE, TFIIF, TFIIH, and RNAP II [28–30]. In a functional sense, these minimum components of the PIC are analogous to the RNAP holoenzyme (core RNAP + sigma factor) in Bacteria in that it is the minimum protein complex needed to recognize promoters, unwind DNA, and begin transcription elongation in vitro. Here, in contrast to the spontaneous DNA opening catalyzed by the bacterial RNAP holoenzyme, PIC-dependent initiation is often dependent on ATP-hydrolysis by a DNA translocase subunit of TFIIH called Ssl2 (XPB in Metazoa) [31–34]}.

In the yeast system, an additional phase of initiation exists compared to the human system wherein the PIC searches for an appropriate start-site prior to beginning processive transcription elongation. This search is thought to take place in a directional manner starting from 40 bp downstream of the TATA-box up to as far downstream as 120 bp [35–37]. This start-site scanning mechanism is ATP-dependent through the activity of Ssl2, the same motor responsible for initial DNA unwinding, and doesn’t depend on transcription by RNAP II [38]. Thus, in the context of S. cerevisiae PICs on TATA-dependent promoters, a minimal initiation mechanism consists of PIC formation, ATP-dependent DNA unwinding, ATP-dependent start-site scanning, and finally, NTP-dependent RNA polymerization leading to promoter escape. Key questions in the field center around the molecular mechanisms behind each of these steps. How does ATP-hydrolysis by Ssl2 lead to the unwinding of a region of DNA two turns upstream from where it physically contacts the promoter? How are the nucleic acid binding properties of Ssl2 and RNAP II coordinated to facilitate scanning? What is the conformation of the protein-nucleic acid complex during scanning? What is the nature of a start-site that allows for processive elongation to begin?

We recently adapted a method that was originally developed and used to study the spontaneous DNA unwinding that occurs during bacterial transcription initiation [39] to study the ATP-dependent process of promoter opening and start-site scanning using eukaryotic pre-initiation complexes from S. cerevisiae [21]. Here, we review the basics of magnetic trapping including how the instrument may be used to study DNA unwinding in general. We provide detailed protocols for the collection of data on the eukaryotic pre-initiation complexes, the analysis of the acquired data sets, and briefly describe past results and future directions.

2. Magnetic Trapping

2.1. Overview

A magnetic trapping microscope consists of an inverted microscope with the added ability to apply a magnetic field of controlled strength and orientation to a sample in the field of view [19,20,40]. The magnetic field may be produced by permanent magnetics of varying geometries or by electromagnets [41]. In a common geometry (and in our microscope), two magnets are positioned above the flow cell on a pedestal that may be moved vertically and rotated (Fig. 1A). A force is applied to super-paramagnetic beads proportional to the gradient of the magnetic field and the magnetic susceptibility of the beads. This force may be increased or reduced by moving the magnets closer to or further away from the sample respectively. In addition, rotating the magnets leads to the application of a torque on the beads due to the dipole nature of the magnetic field. In the standard magnet geometry used here, this torque is much larger than any restoring torque in the system and therefore effectively clamps the angular position of the bead such that rotation of the magnets leads directly to rotation of the bead. In the study of protein-DNA complexes, the magnetic bead is typically attached to the free end of a nucleic acid polymer whose other end is attached to the cover glass of the flow cell. These attachments are usually mediated via non-covalent streptavidin-biotin and antibody-antigen interactions although covalent strategies also exist [42].

Figure 1.

Magnetic tweezers setup. (A) Diagram of the magnetic tweezers instrument and flow system. The flow cell is placed in a custom holder attached to an automated xy-stage. A pair of permanent magnets (NS) is suspended on a rotatable shaft above the flow cell via a cantilever on an automate stage to control magnet height (h) and thus the applied force. The magnets can also be turned (t) to rotate the magnetic field. The flow cell is imaged with a 20X air objective mounted on a piezo-controlled stage. The flow cell is illuminated from above by a LED light source. The image is collected on a CCD camera. Dashed line represents the light path. Buffer can be pumped to wash the flow cell and reagents can be injected via the inline sample loop. (B) Custom built flow cell holder that utilizes magnetic force to make a leak proof connection between the inlet and outlet tubing and the flow cell ports. (C) A field of 1 μm beads tethered to the flow cell surface by DNA is shown in a typical experimental field of view. The boxed region has been magnified to show a 0.8 μm reference bead (*) and DNA tethered bead as they appear during tracking. Scale bar is 55 μm.

2.2. The microscope

Our magnetic tweezers instrument is similar to typical instruments described in the literature (Fig. 1A). The flow cell containing tethered DNA molecules is placed in a custom holder (Fig. 1B) attached to XY stages that are controlled via motorized actuators (Newport, #TRB25CC). Above the flow cell suspended on a hollow shaft in a collar is a set of 125 mm³ cubic NdFeB permanent magnets separated by a 1 mm gap. The shaft is connected to a drive motor (Physik Instrument (PI), #M126.PD1) enabling the magnets to be rotated and the whole magnet assembly is mounted on a Z-stage controlled via a motorized actuator allowing the magnets to be raised and lowered (PI, #M126.PD1). A LED light source (627 nm, Luxeon, #SP-12-R5) is aligned above the magnet shaft. The light passes through the shaft and magnet gap, illuminating the flow cell. An air immersion objective (Zeiss, 20X EC plan-neofluar, #420350–9900-000) mounted on a piezo stage (PI, #E-665.CR) collects the light and a tube lens (300 mm, Newport, # KPX112AR.16) projects the image onto a CCD camera (1690 ×1710 pixels, Mikrotron, EoSens 3CL). With this optical setup we achieve an observation window size of 372 × 376 μm with a magnification of 36x (Fig. 1C). The instrument is controlled via lab-written software developed in LabView (National Instruments) and has an effective frame-rate of 40 Hz.

2.3. Positional feedback

The instrumental drift over short time windows (~10 min) is generally manageable without the need for feedback correction. With our instrument, longer acquisition windows over tens of minutes to hours require xyz-feedback correction to be able to continually track the DNA tethers. Xyz-feedback correction is achieved by using x- and y-stages with automated screw drives along with a piezo controlled objective to adjust the z coordinate. Every 10,000^th frame or approximately once every couple of minutes, the xyz-position of a reference bead is tracked. If it exceeds predetermined limits (0.2 μm for x and y, 1 μm for z), the program moves the stages and objective back to the original set point. We have found with these stage motors that a slow movement velocity (0.005 mm/sec) significantly reduces artifacts in the data due to stage movement.

2.4. Flow cell holder and plumbing

The custom holder for the flow cell is designed to facilitate connection of tubing to allow pump driven flow for buffer exchange, removing untethered magnetic beads and flowing in protein factors (Fig. 1B). It is comprised of two parts made out of Delrin plastic. The bottom part attaches to the xy-stages and has four small permanent magnets. The top part has two ports with O-rings and four iron screws positioned to be aligned with the magnets on the bottom. The flow cell is placed in the recessed area of the bottom and the top snaps in place held together by the force of the four magnets. The O-rings make a seal between the flow cell glass and tubing port. The strength and location of the magnets that hold the flow cell together were selected such as not to interfere with the magnetic tweezers assay but to provide sufficient force to make a leak-free connection up to flow rates of 50 μl/min. Detailed CAD files for the custom holder are available upon request.

2.5. Flow cell construction

The single molecule magnetic tweezers experiments are conducted in a functionalized, passivated flow cell constructed from two glass coverslips and a parafilm spacer containing a cut-out channel region (Fig. 2). The following sections describe how we prepare flow cells for our studies of RNAP II transcription initiation. The flow cell dimensions are specific for our instrumental design and can be changed to accommodate other geometries.

Figure 2.

Flow cell assembly and surface passivation. The flow cell is constructed from two cover glasses and a parafilm spacer with a cut-out channel (grey). Two holes are drilled in one cover glass for inlet and outlet ports. The parafilm spacer is sandwiched between the cover glass, using the drilled holes as a guide and heated to assemble the flow cell. The flow cell surface is passivated with a bifunctional PEG layer covalently attached to the glass via the reaction of silane with surface hydroxyls and to the antibodies via NHS-ester generated amine linkages. Reference amine-coated polystyrene beads (0.8 μm) are then covalently attached to the surface via EDC-activated carboxyls on the antibody.

2.5.1. Cover slip preparation and flow cell assembly

Micro cover glass (Ted Pella, INC, #1.5 (0.16–0.19mm), 22×50 mm) is used for the top and bottom surface of the flow cell. The top cover glass has two holes drilled providing an inlet and outlet port for the flow cell channel. We use a Dremel rotary tool in a drill press with a diamond encrusted spherical bit (0.9 mm, course grit, DiamondBurs.Net LLC) to drill the holes. Where the hole is to be drilled, a drop of water (~30 μl) is pipetted on the cover glass. The water cools the glass and bit during drilling, helping to prevent cracking. Each hole is drilled at high speed in two steps as we have found this generates holes smaller than the bit diameter which prevents leaks at our port interfaces and reduces cover glass cracking. To do this, first lower the bit into the water drop and apply light pressure on one side of the glass. When the water drop becomes cloudy, stop drilling, dry the glass, and turn the glass over to drill from the other side after applying a fresh drop of water. After each drilling step clean the drill bit with a wire brush to remove compacted glass. This cleaning step increases the life of the bit and reduces cover glass cracking.

The drilled and undrilled cover glass are placed in a Teflon rack in a 100 ml beak. The glass surfaces are then cleaned by adding freshly prepared piranha solution (75 mL sulfuric acid: 25 mL hydrogen peroxide) to the beaker and incubating for 15 min. in a fume hood at room temperature. One should use extreme care when handling the piranha solution as it is extremely corrosive and a strong oxidizer. Where appropriate personal protection equipment to protect eyes, skin, and clothes from coming in contact with the piranha solution. The cover glasses then are rinsed with milli-Q purified water (700 mL) and dried with nitrogen gas. The flow cell is assembled by placing a parafilm spacer, cut using a stencil of the channel, on the drilled cover glass. The undrilled cover glass is then placed on the spacer and lined up with the drilled cover glass. The flow cell is sealed by heating on a hot plate (~100 C) for 10 – 20 s while gently pressing on the glass to bond the parafilm to the glass (Fig. 2). Multiple flow cells can be prepared this way and stored dry under desiccation at room temperature until needed. Typically, our flow cells have a channel volume of 10 – 15 μl.

2.5.2. Surface functionalization and passivation

The assembled flow cell surfaces are passivated with polyethylene glycol (PEG) and functionalized with an antibody specific for digoxigenin (Fig. 2). The antibody binds to the digoxigenin labeled DNA handle establishing the DNA tether anchor to the flow cell surface. We and others use a heterobifunctional PEG derivative (nanocs, INC., cat. #: PG2-NSSL-5K) containing silane at one end of the PEG chain (MW 5000) and an NHS ester moiety at the other end [42]. The silane moiety forms a covalent bond to the glass surface while the NHS ester allows for the covalent attachment of antibody through an amide bond. The hetero-functional PEG (1.4 mg) is dissolved in DMSO (100 μl) and 20 μl are pipetted into a single flow cell channel. The flow cell is placed in a humidified chamber and incubated at room temperature (~23–25 C) in the dark for one hour. The PEG solution is then removed by gently blowing nitrogen gas (high purity) through the flow cell. The flow cell is then washed by pipetting milli-Q water (100 μl) and wicking it through the channel with a rolled-up kim-wipe. The water is followed by PBS buffer wash (100 μl, 1X, pH 7.4). An antibody solution (30 μl, 100 μg/ml, Anti-Digoxigenin, Roche, #11333089001) in PBS is pipetted into the flow cell channel and allowed to incubate in a humidified chamber for 30 min at room temperature to allow covalent coupling of the antibody to the PEG surface via the NHS moiety. After antibody treatment the flow cell is washed with 1M TRIS-HCl (pH 8.3, 100 μl) and incubated in a humidified chamber for 30 min at room temperature to block unreacted NHS groups.

2.5.3. Reference bead attachment and flow cell storage

To account for instrumental drift in the xyz displacement of the DNA tethered bead, we monitor the xyz displacement of a reference polystyrene bead that is covalently attached to the PEG: antibody. We attach amino-coated polystyrene beads (0.8–1 μm, Spherotech, #AP-08–10) covalently to the surface via EDC (1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide) activated surface exposed carboxyl groups on the deposited antibody (Fig. 2). The flow cell is washed with 0.1 M MES buffer (pH 4.3, 150 μl). EDC (1mg, Thermo Scientific, #22980) is dissolved in 0.1 M MES buffer (pH 4.3, 100 μl) containing amino coated polystyrene beads (final bead dilution is 1/4000). Bead:EDC solution is pipetted into the flow cell and incubated in a humidified chamber for 1 hour at room temperature. Post incubation, the flow cell is washed with milli-Q water (150 μl) and incubated for an additional 10 min. This step is critical for restoring carboxyls on the antibody that were modified with EDC but did not react with the amine coupled beads. We have observed significantly reduced DNA tether densities that cannot be compensated with increased DNA concentration if the EDC modification is not reversed. The flow cell is then washed with storage buffer (PBS, 0.1 mg/ml BSA (denatured, NEB, #B9000S), 0.1 % (v/v) Pluronic F127 (Invitrogen, #P6866), 150 μl) and stored in humidified chamber at 4 C. We generally use the flow cells within a week, but they can be stored in this way for at least a month without loss of reference beads and without reduction in surface activity for forming DNA tethers.

3. Bead tracking and data collection

Methods for the 3D-tracking an out of focus bead have been described in detail previously [19,43]. Briefly, the center of the bead can be readily tracked in x and y by using image processing methods for finding the centroid of the bead image whereas z-coordinate tracking requires an empirical calibration of the bead image as a function of distance from the focus of the objective. This calibration, often referred to as a “z-stack” or “zLUT (z Look Up Table),” is a series of images taken at different microscope objective distances. During the calibration, the tethers are held at maximum force (i.e. ~4 pN) to minimize fluctuations of the bead. Images are collected as the objective is moved by known distance increments (i.e.100 nm) by the piezo electric controlled stage. A smooth calibration series to use in quantitating experimental z-distances is generated by interpolating between the reference series of images. Once the calibration movie is collected for a given field of view, all non-overlapping beads can be tracked in z, allowing the DNA tethers’ physical properties such as length and torsional constraint to be determined.

In addition, magnetic tweezers require force calibration to determine the applied force at a given magnet height. The methods for force calibration have also been previously described in detail [18,44]. Briefly, these methods empirically determine the force on each bead by assuming that tethered beads behave as thermally buffeted inverse pendula. By measuring the length of the tether (L) holding the bead to the surface and quantitating the lateral excursions of the bead <Δx>², the force (F) may be determined as L⊕k_BT/2<Δx>² where ke is Boltzmann’s constant and T is temperature in Kelvin. This relationship is valid as long as the corner frequency of bead fluctuations does not exceed the Nyquist frequency of the camera as is the case here [44]. Although at higher forces, a more complicated dependence of force to magnet height can be observed [44], at the forces (~0.1 – 4 pN) and distances (1 – 15 mm) used here, the force is exponentially dependent on magnet-bead distance with a decay constant of between 0.8 – 0.9 mm⁻¹ for the1 micron paramagnetic beads (Dynabeads MyOne Streptavidin T1, Invitrogen #65601) used here.

Based on our magnification and a bead size of 1 μm, this setup would be predicted to reach a z-tracking precision of ~2 nm for the position of a bead stuck to the coverslip [45], but the thermal fluctuations of a bead by 2 kb of DNA held at 0.3 pN dramatically limit the precision to ~10–15 nm as judged from data collected on naked DNA tethers and filtered with a 1 s window [21].

4. Detecting DNA unwinding

4.1. Different strategies

Magnetic tweezers and optical tweezers techniques have been used to study a number of nucleic acid motor proteins including helicases, DNA polymerases, and RNA polymerases [6,9,14,16]. Often DNA/RNA unwinding is followed via the change in end-to-end distance generated by the difference between double-stranded and single-stranded persistence length. Processive motors take many steps along the nucleic acid resulting in large changes in extension that can be followed directly by monitoring the end-to-end distance of nucleic acid tethers. Often, to achieve a high linear resolution, these assays are performed at relatively high forces where the end-to-end distance of the nucleic acid polymer is approximately equal to its contour length (i.e. 4 pN) and bead fluctuations are reduced relative to low force.

In contrast, promoter opening during transcription initiation by RNA polymerase is a more challenging system, since only a turn or so of DNA is unwound to form the open DNA complex and the amount of unwound DNA changes by amounts as small as a single base-pair. Experiments pioneered by Terence Strick and colleagues on E. coli RNA polymerase demonstrated how magnetic tweezers could be used to measure promoter DNA opening at base-pair resolution and low force [9,39].

4.2. Magnetic tweezers assay for promoter unwinding

The topology of a torsionally constrained DNA molecule can be described by the number of times one strand crosses the other or the number of links (Lk). These links can take on one of two forms. Twist (Tw) refers to the winding of each strand around the axis of the double-stranded molecule and writhe (Wr) refers to the crosses of the double-stranded axis itself. The sum of twist and writhe must equal the linking number and, while twist and writhe may change, the linking number must remain constant if no covalent bonds are broken (ΔLk = ΔTw + ΔWr = 0, Fig. 3). The magnetic tweezers promoter opening assay takes advantage of this conservation. Since ΔTw = −ΔWr, any changes in the twist (i.e. DNA unwinding) results in a compensatory change in the writhe (i.e. the DNA duplex-duplex crossings). Due to the bending stiffness of dsDNA, a single writhe in the DNA molecule results in a significant change in the end-to-end extension of a DNA molecule (~40–50 nm). Thus, in a torsionally constrained DNA molecule the DNA unwinding signal can be monitored by changes in the end-to-end distance of the DNA generated via the addition or subtraction of DNA writhe. To convert changes in end-to-end distance to number of base-pairs unwound, a calibration rotation-extension curve is used. This curve is generated by rotating the magnets to produce known amounts of writhe in the absence of any protein factors. The decrease in extension per magnet turn reveals the expected length change for the unwinding of a single turn of DNA (10.5 bp for B-form DNA).

Figure 3.

Detecting changes in DNA topology with the magnetic tweezers. Writhes (Wr) can be imparted into a torsionally constrained DNA when the magnets are rotated, resulting in a decreased DNA extension. Due to the conservation of linking number for a torsionally constrained DNA any change in the number of twists (Tw) must be compensated by a change in the number of writhes. For DNA with negative super-helicity, removing a twist results in removing a writhe causing an increase in the DNA extension. In contrast, for DNA with positive super-helicity, removing a twist results in adding a writhe causing a decrease in the DNA extension. Regardless of super-helicity if DNA is compacted the DNA extension will decrease. The arrows indicate the change in DNA extension from the initial supercoiled state.

Furthermore, the direction of the change in DNA extension upon DNA unwinding depends on the initial super-helicity of the DNA. Whereas unwinding negatively supercoiled DNA results in an extension increase, unwinding positively supercoiled DNA leads to an extension decrease. In contrast, regardless of the super-helical density, any process that compacts the DNA in the absence of unwinding results in a decrease in DNA extension (Fig. 3). Therefore, by conducting experiments on both negatively and positively supercoiled DNA, one can unambiguously determine the contribution of DNA unwinding and DNA compaction to the observed signal. Specifically, if the length changes are equal and opposite under the two conditions, then the signal can be interpreted as stemming entirely from DNA unwinding.

This assay has been used to great effect in studies of bacterial transcription initiation. In these studies, the lifetimes of closed and open RNAP-promoter complexes were revealed along with their dependencies on super-helical density, the presence of the initiating nucleotide, the presence of ppGpp, and promoter sequence [39]. In subsequent work, the dependence of the size of the transcription bubble on the length of the RNA transcript during initial transcription was determined [9]. These experiments provided clear evidence supporting a model of DNA scrunching for the last stages of promoter escape where the bubble grows concomitantly with nucleotide addition between the second and ninth incorporation. We asked a similar question for the eukaryotic system: How does the size of the open DNA bubble evolve over time as the system progresses from closed complex to promoter escape.

5. Magnetic Tweezers Studies of Eukaryotic Transcription Initiation

5.1. Challenges presented by the eukaryotic system

The eukaryotic transcription system has a number of challenges that are not present in the bacterial system. First, RNAP II requires a cast of general transcription factors to facilitate its binding to the promoter via the formation of a preinitiation complex (PIC) and unwind the DNA to form an open complex. The minimal factors required to form a transcriptionally active PIC are TATA binding protein (TBP), TFIIB, TFIIF, TFIIE, TFIIH and RNAP II. In the magnetic tweezers assay, PIC assembly is slow, requiring long data acquisitions. Second, the RNAP II core promoter has weak sequence specificity for directing PIC formation. Simple TATA-like sequences are sufficient to bind TBP from which the rest of the factors can be recruited to form the PIC. In the magnetic tweezers assay DNA tethers are at least 2 kb in length increasing the likelihood of having multiple TATA-like sequences present. Third, depending on the promoter, the efficiency of the minimal yeast RNAP II PIC as judged by template usage is under 20 %. Thus, many torsionally constrained DNA molecules need to be surveyed to generate sufficiently large data sets of PIC-dependent activity. These challenges were overcome with minor modifications to a typical magnetic tweezers setup and careful engineering of the DNA template.

5.2. DNA template design and construction

Bacterial RNAP holoenzymes bind to −35 and −10 regions of gene promoters and elimination of these well-defined regions in magnetic tweezers experiments resulted in no observed DNA opening [39]. In contrast, the eukaryotic RNAP II transcription system has a variety of core-promoter sequence elements that facilitate PIC formation; however, not all promoters have all of the elements [46]. Furthermore, the sequence elements have low sequence complexity increasing the likelihood of element-like sequences that could direct preinitiation complex formation at non-promoter sequences. This is particularly challenging for the magnetic tweezers assay, since the DNA tether is 2 kb in length. In our studies of S. cerevisiae RNAP II promoter opening of the TATA-dependent His4 promoter, DNA sequences outside the His4 promoter were designed to lack TATA-like sequences so as to direct PIC formation to the His4 promoter specifically. In vitro transcription assays were used to verify that transcription was only initiated at the His4 promoter transcription start site rather than at non-specific sites. The full sequence used in our experiments can be found in the original article [21].

The DNA tethers consist of 2 kb of DNA containing the TATA-box and promoter region flanked by two 1 kb DNA handles which facilitate the anchoring of the DNA to the flow cell surface and to the paramagnetic bead (Fig. 4). The following is a general overview of how we and others generate the DNA construct. Each DNA region is generated by PCR using DNA primers engineered with unique endonuclease restriction sites. The selection of specific endonucleases can vary depending upon the specific DNA promoter region being studied. One DNA handle is generated by incorporating biotin-labeled uridine deoxynucleotides (biotin-dUTP. Roche, #11093070910) and the other is generated by incorporation of digoxigenin-labeled uridine deoxynucleotides (dig-dUTP, Roche, #11093088910) during PCR with Taq DNA polymerase. The DNA handles are PCR amplified from an 1kb region of plasmid DNA with ~50% GC content. Any DNA sequence can be used for generating the DNA handles, since the accessibility of the DNA is limited through the multiple attachment points between the DNA handles and the glass and bead surfaces. The degree of label incorporation during the PCR is controlled by adjusting the ratio of labeled to unlabeled dUTP. We typically use a ten-fold excess of dNTP over labeled dUTP, allowing a label to be incorporated approximately every tenth thymidine. The DNA region containing the promoter is also amplified via PCR. The three DNA regions are then digested with endonucleases in separate reactions. Finally, the three pieces are pooled together and ligated with DNA ligase (T4 DNA ligase, NEB, #M0202L) via standard protocols. In the ligation reaction, excess handle DNA is used to drive the formation of the full-length product, which is subsequently agarose gel purified removing the excess handle DNA. We purify the full-length ligation product on 0.8% ultrapure agarose (Invitrogen, #16500) in 0.5x TBE (Tris/borate/EDTA) buffer electrophoresed at 120 volts. The DNA band on the gel is excised and the DNA electro-eluted from the gel slice in 0.5x TBE using the Elutrap system (Whatman, #10447705) at 175 volts. The DNA is them concentrated using the Qaigen PCR clean-up kit, using a single spin column and eluting the DNA in a 30 μl volume. We have found storing the concentrated DNA (> 1 nM) in buffer containing 1 mM EDTA and 50 % (v/v) glycerol helps to preserve the DNA from acquiring nicks in the phosphodiester backbone upon prolonged storage at −20 C. DNA preparations stored under these conditions yielded similar torsionally constrained DNA tether populations well over a year.

Figure 4.

DNA construct with TATA-dependent promoter. Diagram of the TATA-dependent promoter containing DNA used to make DNA tethers in the magnetic tweezers assay. The DNA is attached to the flow cell surface through the anti-dig:digoxigenin interaction and attached to the paramagnetic bead through the biotin:streptavidin interaction of the 1kb DNA handle regions. The intervening DNA contains the RNAP II TATA-dependent promoter (red). TATA-like sequences (green) had to be mutated to insure promoter specificity.

5.3. Populating the flow cell with DNA tethers

We generate DNA tethers on the flow cell surface similarly as previously described methods [19,47]. Briefly, the biotin and digoxigenin labeled DNA construct is mixed with excess MyOne Streptavidin T1 beads (1 μm diameter, Invitrogen #65601) to maximize the number of beads with a single tether. The magnetic beads (5 μl, 10mg/ml) are washed with 200 μl BW buffer (PBS, 1 mg/ml BSA (denatured, NEB, #B9000S)). A magnet is applied to the tube to pellet the beads and the wash buffer is removed. The washed magnetic beads are resuspended in 10 μl BW buffer. A 50 pM dilution of DNA is prepared in storage buffer. A 0.5 μl drop of DNA solution is pipetted into an Eppendorf tube and the 10 μl of washed beads are added to the drop of DNA. Immediately, the DNA:bead solution is diluted up to ~100 μl by adding 90 μl storage buffer. The DNA:bead solution (~30 μl) is pipetted into the flow cell and incubated for 20 min at room temperature, allowing the bead:DNA to settle and the digoxigenin handle of the DNA to bind to the antibody on the surface. The flow cell is then placed into flow cell holder and positioned on the xy-stage of the instrument where the inlet and outlet ports are then attached. The flow cell bead:DNA solution is exchanged by flowing 200 μl of yTRxn buffer (10 mM HEPES, pH 7.6, 100 mM potassium glutamate, 10 mM magnesium acetate, 3.5% (v/v) glycerol, 0.1% (v/v) Pluronic acid F127, and 0.1 mg/ml BSA (denatured, NEB, #B9000S)) at a flow rate of 10 μl/min, thus removing untethered beads. Periodically, a magnet stick is held over the flow cell during flow to remove any magnetic beads nonspecifically associating with the surface. Usually this protocol yields a good DNA tether density where for most of the DNA tethers there is no bead overlap. However, if there is significant bead overlap or DNA tethers are too sparsely populated, the DNA concentration during the bead:DNA incubation step can be adjusted accordingly. In our field of view, this procedure typically results in 100–150 trackable beads (Fig. 2C).

5.4. Data collection and analysis

With our instrumental setup, DNA, and flow cell preparation we typically observe 50–100 torsionally constrained tethers for a given field of view. To maximize our temporal resolution we do not track all of the beads in real time during an experiment as increased computation reduces the effective frame rate. Rather, we track a reference bead to follow the instrumental drift during the long movie acquisition and inform automated stage-based feedback. The xy-stage and z-piezo allows us to implement a xyz-feedback correction to keep the field of view trackable over the long duration of acquisition. Post-acquisition, the movie is converted into a series of JPEGs and tracked off-line using NanoBLOC software [43].

The xyz-data for each DNA tethered bead is drift corrected by subtracting the average of the reference bead traces. The DNA end-to-end extension (z-data) is then converted to base-pairs unwound using the slope of the DNA extension dependence on magnet rotation. Data collected on both negative and positive supercoiled DNA are analyzed to assess if the observed changes in DNA extension are solely due to DNA unwinding or if DNA compaction contributes. DNA tethers showing potential DNA-opening activity are screened further. Specifically, the drift corrected xyz-data for each tether is screened to make sure the DNA only portion of the trace is consistent (i.e. expected extension, torsional behavior, etc.). Any traces showing irregularities (i.e. surface interaction, high noise, etc…) are excluded from further analysis. DNA tethers showing potential PIC dependent activity in DNA extension are furthered screened to ensure the observed activity is not due to artifacts (i.e. surface interaction, free beads/particulates or stage movement). DNA tether data sets are then filtered with a 1 s moving average window to smooth the data. With throughput still a challenge, we then use between 20–30 tethers showing activity in each NTP condition to generate probability distributions of bubble sizes that can be fit to quantitate the number and size of open states.

5.5. Characterization of DNA tether torsional constraint

To assess which DNA tethers possess the requisite torsional constraint, the magnets are positioned to apply ~0.3 pN of force on the beads and are then turned counter-clockwise 20 times. A movie is then recorded to track the end-to-end distance of the DNAs as the magnets are turned clockwise 40 times, removing the 20 positive writhes and imparting 20 negative writes to the DNA molecule. At low forces (0.1 – 0.5 pN), the end-to-end extension of a single DNA tether will decrease as a function of magnet turns and will be symmetric (Fig. 5A). In contrast, a nicked DNA tether will not exhibit any change in extension during magnet rotation, as generated writhes are able to relax via rotation of the un-nicked strand. Furthermore, the slope of the obtained rotation-extension curve (nm/turn) can be used to convert end-to-end distance changes (nm) to turns of DNA unwound and subsequently to base-pairs unwound assuming B-form DNA with 10.5 bp/turn (i.e. bp unwound = (ΔL*10.5)/slope, Fig. 5B).

Figure 5.

RNAP II promoter opening experiment. (A) DNA rotation time traces at 0.3 and 1 pN. At 0.3 pN the DNA is rotated from +20 turns to −20 turns then the force is increased to 1 pN and rotated from −20 turns to +20 turns (magnet velocity: 1 turn/sec) The symmetric profile at 0.3 pN and asymmetric profile at I pN is indicative of a single torsionally constrained DNA tether. (B) Using the magnet rotation velocity, the time traces in (A) can be converted to a rotation-extension plot. Here, DNA extension as a function of turns is shown under 0.3 pN of tension. The slope of the linear region is used to convert the change in DNA extension to base-pairs unwound. The raw data (grey circles) are filtered with a 1 s moving window (black line).

After turning the magnets under low force, the magnets are lowered to increase the force to ~1 pN. Under this increased tension, a DNA molecule with negative writhes will unwind, resulting is an increase in the extension of the DNA tether to that of the relaxed DNA molecule. Here, as the magnets are turned counter-clockwise 40 times, the unwound DNA simply rewinds with no observed change in extension. Subsequently, the DNA begins to compact as net positive writhes begin to form in the DNA molecule (Fig. 5A, B). This asymmetry of the rotation-extension curve confirms the presence of a single torsionally constrained tether. In contrast, multiple DNA molecules tethered to a bead result in a distinct rotation-extension profile as DNA molecules wrap around each other [15].

This procedure is performed with real-time tracking of around 10 tethers to estimate the quality of the DNA tether preparation prior to flowing protein. However, the rotation extension curves for each tether is analyzed afterwards prior to interpretation of any protein-dependent signals as a control.

5.6. Observing NTP and ATP-dependentpreinitiation complex activity

A schematic of a typical magnetic tweezers experiment can be seen in Fig. 6A. Before adding the general transcription factors and RNAP II to the flow cell, a movie (~20 min) of the DNA tethers with both negative and positive writhes are collected at 0.3 pN force to serve as measure of the closed DNA state under both conditions (Fig. 6A(1)). While collecting another movie the general transcription factors: TATA-box binding protein (TBP, 223 nM), TFIIB (31 nM), TFIIE (4 nM), TFIIF (13 nM) and TFIIH (5 nM) along with RNAP II (10 nM) and NTPs (0.5 mM) in yTRxn buffer are flowed into the flow cell at 10 μl/min and 24 – 25 C. During flow the DNA has 0 – 2 positive writhes to ensure the flow force on the bead does not facilitate DNA unwinding. Once the protein has been added the pump is turned off and the flow is stopped by closing an inline valve. The magnets are then turned to the desired super-helicity throughout the remainder of the movie (i.e. −6 turns, Fig. 6A(2)). We typically collect a 1 – 1.5 hr movie. A representative experiment is shown in Fig. 6B. PIC promoter opening is detected on both negative and positive supercoiled DNA by observing the change in tether extension as discussed above (Fig. 6B, black arrow) and in this case persists for the remainder of the experiment. A closer look at a DNA unwinding event (Fig. 6C) shows both the raw data (grey circles) and the smoothed data (black line). Here, DNA extension has been converted to base-pairs unwound using the slope of the rotation extension curve (Fig. 5B).

Figure 6.

(A) Schematic of an experiment which is broken into 3 phases: (1) DNA only control, (2) PIC activity, and (3) Post buffer wash. During the DNA only control, DNA extension under both negative and positive super helicity is observed for a base line. Then PIC and NTPs are flowed in and the DNA is turned to the desired super helicity and observed (2). Finally, to assess if the PIC successfully initiated RNAP II to form an elongation complex the flow cell is washed with buffer to remove factors and NTP and the DNA is turned to the desired super-helicity to see if the extent of DNA opening is still present (3). (B) Representative RNAP II promoter opening experiment in the presence of NTPs. The data is filtered with 1-sec moving average window to facilitate observing changes in the data. Red bars indicate regions of negative super-helicity and green bars indicate regions of positive super-helicity and do not correspond exactly to the schematic in (A). Shaded regions indicate flow. Arrowhead indicates a change in DNA extension consistent with DNA unwinding. The unwound state persists for the remainder of the experiment even when rotating the magnets, imparting positive super-helicity; rotating the magnets, returning back to the original negative super-helicity; and even after washing out PIC factors and NTPs, suggesting RNAP II transitioned to elongation. (C) A closer look at a DNA unwinding event where the raw data (grey circles) and filtered data (black line) have been converted from DNA extension to base-pairs unwound using the slope of the rotation extension curve (Fig. 5B).

5.7. Testing if open state is consistent with an elongation complex

The magnetic tweezers assay can readily detect changes in DNA topology; however, it does not determine where those changes occur along the DNA template. In Bacteria, a small difference in the number of base-pairs unwound between open and elongation complexes allowed for promoter escape to be directly assayed [39]. As this was not clearly observed in our experimental traces, we tested for the presence of productive elongation complexes in two ways using the fact that once a bona fide elongation complex is formed, it is stable even after the removal of ATP, factors, and NTPs while open complexes are not [33,48,49]. The first approach tested the stability of open DNA to a buffer wash removing excess factors and NTPs (Fig. 6A(3)). If open DNA is the result of an elongation complex, the extent of open DNA before and after a buffer wash should be the same as was the case in a subset of traces. A second approach made use of a chain terminating ribonucleotide such as 3’dCTP to stall RNAP II during elongation and prevent the elongation of the entire template and subsequent dissociation of the polymerase. This control increased the percentage of tethers that showed stable DNA opening signals after buffer washes again consistent with the presence of a RNAP II elongation complex [21].

5.8. Modulating PIC dsDNA translocase activity and RNAP II transcription

In our studies of yeast RNAP II transcription initiation, we observed two open DNA states. The smaller open state (6 bp) is ATP-dependent, while the larger open state (13 bp) is NTP-dependent. To test if the smaller open state was on path to the larger open state, we used saturating dATP (0.5 mM) and sub-saturating NTPs (0.05 mM) to modulate the dsDNA translocase activity of Ssl2 and transcription activity separately. RNAP II cannot readily incorporate dATP into a growing RNA; however, the dsDNA translocase of TFIIH can use dATP to power translocation [32,34]. By using dATP and reducing the NTP concentration, we increased the prevalence of the smaller open state in the presence of NTPs allowing us to observe direct transitions between the small and large open DNA states [21].

6. Future Outlook

The experiments described above have revealed new aspects of the transcription initiation mechanism in the eukaryotic system. Specifically, we were able to detect an intermediate unwound DNA state, dependent on ATP hydrolysis, that is subsequently expanded in a transcription-dependent manner. However, there are many unanswered questions that remain. For example, due to the limited throughput of the current experiment, the kinetics of the reaction are difficult to estimate. In particular, the full distribution of the lifetime of the intermediate state, its dependence on ATP concentration, promoter architecture, TSS position, and the presence of transcriptional regulators will be of crucial importance in understanding the kinetics of both basal and regulated transcription initiation. To be able to collect data in a more efficient manner, an increase in either the efficiency of the reaction itself and/or the number of observations that are possible within a single experiment must be achieved. In the first case different promoter constructs or additional activating factors or proteins from different eukaryotic model systems can be explored. In the second case, it has already been shown that one may significantly increase the number of observable tethers by patterning attachment sites on the coverslips via PDMS stamping [50].

In addition to improvements in throughput, several magnetic tweezers labs have incorporated fluorescence detection into their setups [51,52]. This improvement would allow for the simultaneous monitoring of the size of the transcription bubble via the approach described here and the presence or absence of particular factors. This kind of experiment would be useful for ascertaining the makeup of the PIC as a function of progress through the initiation mechanism by using labeled general transcription factors or labeled RNAP II. Pushing this idea even further, fluorescence resonance energy transfer experiments may reveal the conformational changes that actually lead to DNA unwinding. Incorporating these approaches will open up the ability to ask many mechanistic questions using this technique. In particular, the regulation of transcription must be dictated via both promoter sequence and transcriptional regulators. Exploring the large parameter space represented by these two variables will require the application of these advances in experimental technique.

In all eukaryotic systems initial DNA unwinding occurs at the site of PIC assembly. While most experiments have shown that initial unwinding on linear templates is ATP-dependent, recent structural work has thrown some doubt on this issue [53,54]. These studies first observed open complexes in cryo-EM structures in the absence TFIIH. Follow up experiments looked at DNA distortions in vitro via 2-aminopurine fluorescence and performed Ssl2 depletion studies in vivo to look for how TFIIH-dependence varies genome-wide. One conclusion from this work was that the stability of the unwound region as dictated by base-pairing energy determines a promoters TFIIH-dependence. However, according to this model and the stability we have calculated for our HIS4-based promoter, we would observe ATP-independent unwinding which we do not observe. Sorting out promoter-sequence-dependent and experimental- condition-specific results represents a clear direction for future work. In particular, distinguishing between productive unwinding that leads to subsequent transitions on the path to initiation from rapid, but unproductive isomerizations between open and closed DNA remains difficult. Resolving this issue may require pushing the time resolution of the magnetic tweezers by using smaller beads with quicker relaxation times. In addition, a brighter light source coupled with a fast CMOS camera and GPU-based frame acquisition has been demonstrated to allow for much faster frame rates [55].

One outstanding mechanistic question that remains is to physically dissect the transcription start site scanning reaction from that of initial DNA unwinding. Intriguingly, in S. cerevisiae the initial site of DNA unwinding is not the site of transcription initiation. Instead, up to 120 bp of downstream DNA is moved relative to the polymerase active site until a start-site is encountered [35–37]. Since our data suggest that scanning takes place in the context of the smaller transcription bubble [21], we would predict that DNA templates with different distances to start-sites should have correspondingly different lifetimes for the small bubble unless the process is rate limited by some step other than translocation, but this remains to be tested. Furthermore, the human PIC is not known to perform the scanning reaction. Experiments comparing and contrasting both the sizes and kinetics of bubble intermediates across species should reveal more in the way of mechanistic differences that may explain what on the surface appears to be different behaviors between two well-conserved molecular machines.

Highlights:

• Single-molecule magnetic tweezer assay to follow DNA unwinding by eukaryotic transcription initiation complexes in real time is described in detail.

• An intermediate-sized 6 bp DNA bubble generated by ATP hydrolysis by the Ssl2 subunit of TFIIH was detected.

• Subsequent NTP-dependent transcription expands this bubble to 13 bp.

• Future directions may include separating initial unwinding from start-site scanning, increasing the throughput and time resolution of the approach, and the incorporation of fluorescence.