|Home | About | Journals | Submit | Contact Us | Français|
DNA is the carrier of genetic information, and as such, is at the center of most essential cellular processes. To regulate its physiological function, specific proteins and motor enzymes constantly change its conformational state with well controlled dynamics. Twenty-five years ago, Schafer, Gelles, Sheetz and Landick employed the Tethered Particle Motion (TPM) technique for the first time to study transcription by RNA polymerase at the single-molecule level. TPM has since then remained one of the simplest, most affordable and yet incisive single-molecule techniques available. It is an in vitro technique which allows investigation of DNA-protein interactions that change the effective length of a DNA tether. In this chapter, we will describe a recent strategy to multiplex TPM which substantially increases the throughput of TPM experiments, as well as a simulation to estimate the time-resolution of experiments, such as transcriptional elongation assays, in which lengthy time averaging of the signal is impossible due to continual change of the DNA tether length. These improvements allow efficient study of several DNA-protein systems, including transcriptionally active DNA-RNA polymerase I complexes and DNA-gyrase complexes.
Single-molecule approaches reveal biologically important molecular heterogeneity in conformational dynamics, molecular interactions, or even intermediate steps of reactions which are obscured in ensemble (bulk) measurements. Single-molecule experiments are designed to reveal the conformation and/or location of individual molecules as well as the dynamics of transitions among different molecular states. Tethered Particle Microscopy (TPM), is a conceptually simple yet elegant technique which monitors changes in the amplitude of the Brownian motion of a DNA-tethered bead in solution (Nelson et al., 2006). It is an optical method that requires tracking the position of a polystyrene bead tethered to the glass surface of a microscope flow chamber by a single DNA molecule. One distinguishing characteristic and advantage of TPM over other single-molecule techniques is the absence of significant externally applied force which may perturb interactions between the DNA and other molecules of interest. Indeed, external loads may slow and eventually inhibit the activity of DNA-processing enzymes (Neuman, Abbondanzieri, Landick, Gelles, & Block, 2003).
Therefore, TPM is well suited to observe changes in the conformations of nucleic acids, such as protein-induced looping and condensation, or in the elasticity of DNA due to protein binding as well as ionic strength. It can also be used to monitor the activity of DNA processing enzymes, such as RNA polymerases.
Since TPM has many potential applications, a few laboratories have set about optimizing protocols for experiments (Fan, 2012; Finzi & Gelles, 1995; Laurens et al., 2012; Lindner et al., 2011; Manzo & Finzi, 2010; Norregaard et al., 2013; Pouget et al., 2004; Sandip Kumar, 2014; Schafer, Gelles, Sheetz, & Landick, 1991; Sitters et al., 2015; Vanzi, Broggio, Sacconi, & Pavone, 2006; Yin, Landick, & Gelles, 1994). Recently, Garini et al. proposed to use confocal microscopy to observe the excursion of the labeled bead in three-dimensions (Nir, Lindner, & Garini, 2012). Hidden Markov models using variational Bayesian methods have also been implemented to improve resolution and objectively assign conformational states (Johnson, van de Meent, Phillips, Wiggins, & Linden, 2014). Other investigators have concentrated on the enhancements to increase the throughput of the technique. For example, Plénat et al. reported a method which allowed the observation of precisely arranged arrays of hundreds of tethered beads simultaneously (Plénat, Tardin, Rousseau, & Salomé, 2012). With this improvement, they reported the activity of the T7 bacteriophage exonuclease enzyme which due to its low processivity had previously been examined only using bulk assays.
However, the strategy they developed to achieve high throughput is based on equipment, including software and nanofabricated chambers, that may not be readily available to most laboratories. This chapter reports a more modest but still enabling level of multiplexed, simultaneous bead tracking, which can be achieved in a simple and cost-effective manner. To demonstrate the utility and the limits of this high throughput method, experiments to monitor transcriptional elongation by RNA polymerase I or the wrapping of DNA by E. coli gyrase are described.
In transcription experiments, stalled complexes of RNA polymerase and DNA were immobilized on the microchamber surface. When all four ribonucleotides were added, transcriptional activity resumed and lengthened the DNA tether, which increased the amplitude of the Brownian motion of the tethered bead. In supercoiling experiments, the DNA wrapping activity of E. coli gyrase around its C-terminal domain (CTD), which occurs prior to cleavage and strand passage of the T-segment, was monitored through changes in the excursion of a DNA-tethered bead. Wrapping and unwrapping DNA generated a telegraphic-like time traces of DNA length changes. The following updates previous reports on the TPM technique (Dunlap, Zurla, Manzo, & Finzi, 2011) to include more recent developments like multiplexing.
In this section, we describe the details of the sample preparation including the design of the DNA construct, the construction of a functionalized microchamber, and the immobilization of the DNA molecules for observation in a differential interference contrast microscope.
The DNA fragments used were produced using the polymerase chain reaction (PCR). Primers labeled with biotin or digoxigenin allowed attachment of the DNA to the microchamber surface and/or a microbead (Finzi & Dunlap, 2013).
Microchambers were freshly prepared before each experiment. The materials required are: 50 × 24 and 22 × 22 mm glass coverslips (Fisherbrand, catalog numbers 12545F and 12548B), double sided tape, vacuum grease, ethanol, and deionized water. The assembled microchambers will have a rectangular channel with a volume of approximately 25 μL.
RNA Polymerase with a 3-HA tag at the C-terminal domain (D. A. Schneider, 2012) was immobilized on a microchamber surface coated with human influenza hemagglutinin (HA) antibody. To study the wrapping of DNA by gyrase, DNA was immobilized via a digoxigenin-antidigoxigenin linkage. This section describes coating surfaces with antibodies.
Reconstitution of the transcription complex from purified components requires: RNA Polymerase I, Rrn3p, TATA binding protein (TBP) and core factor (CF). These components were purified using published methods (David A. Schneider et al., 2007). Preparation of the single-molecule assays includes the following steps:
Pol I-DNA-bead complexes formed inside the microchamber are as shown in Figure 4. After introducing all four NTPs, elongation produces an increase in DNA tether length.
High-throughput TPM can be set up using either differential interference or dark-field to produce sufficient contrast to track the motion of the polystyrene microspheres. For DIC, a Leica DM LB2 upright microscope (Leica Microsystems, Wetzlar, Germany) with a Leica 506287 oil immersion objective (63X, NA 1.4 and a Leica Optivar lens (541 517 HC, 0.33X–1.6X) set at 0.63X demagnification level gives a reasonably large field of observation. The components of this system are diagramed in Figure 5.
A total magnification of 40X is sufficient to produce images of beads that are 6–8 pixels across (Figure 6). For dark-field microscopy an objective with 20X magnification was used, and up to 100 randomly distributed tethered particles (many particles in the dark field image are immobile) could be observed per field of view.
A CM-140GE video camera (JAI, Copenhagen, Denmark) with gigabit Ethernet (GigE) connection was used to stream 1390 × 1040 pixel video to the computer at 30 frames per second. The frame rate with 1-ms exposure value was sufficient to observe the beads without blurring due to the fast bead motion (Han et al., 2009; Wong & Halvorsen, 2006). In our tests, video was streamed at 125 MB/s through a Cat 6e network cable between the camera and an Intel Pro 1000 series network adaptor interfaced to a NI-IMAQ driver to manage the data bus. This minimized the load on the CPU for data transfer from the camera. The computer had a 3 GHz AMD quad core CPU, 4 GB of DDR3 RAM memory, a 500 GB hard disk for operating system and software, and a 2 TB, 7200 rpm Western Digital Black series hard disk for video recording.
Labview with NI Vision Acquisition Software (National Instruments, Austin, TX) with an NI-IMAQdx High Performance driver was programmed to capture, display and save the stream as uncompressed AVI video file. The raw data format was an uncompressed movie file. A typical file size is nearly 40 GB for around 15 minutes of recording.
A sequence of Cartesian coordinates was established for each bead in a video. To begin, the locations of the tethered beads were determined by eye during playback of the video file with the open source video player (VLC, VideoLAN, Paris France). Then the VideoReader class in MatLab (Version R2013, MathWorks, Natick, MA) was used to access an initial frame in the video file, and using the imcrop function of the image acquisition toolbox, a region of interest in which to locate the position of each bead through the remaining video sequence was manually registered in a custom Matlab routine. There are several methods for locating particles in images, each with some advantages and some drawbacks (Chenouard et al., 2014). A recent algorithm based on radial symmetry is robust and quick and was implemented to track the tethered beads (Parthasarathy, 2012).
The time series of xy coordinates were analyzed using methods similar to those described previously (Kumar, et al., 2014) to select beads and determine the magnitude of the confined Brownian motion exhibited. First, the mechanical drift was eliminated by subtracting the apparent motion of stuck beads. Then, a radial distribution of the positions was plotted to verify symmetrical excursion in the XY plane as shown in Figure 7A. A bead tethered by more than one DNA molecule, most often diffuses through a range of xy positions scattered in an ellipsoidal rather than a circular pattern. Therefore, the shape of the excursion pattern of a tethered bead, such as that shown in Figure 7B, served to identify beads tethered by single DNA molecules. Then, the covariance matrices of the coordinates of the selected beads were calculated, and only beads with a diagonal ratio smaller than or equal to 1.07 were considered for further analysis.
The two-dimensional projection of the instantaneous positions of the bead with respect to the anchor point, excursions, were averaged over short time intervals as shown in Equation 1,
In which x and y are the coordinates of the bead in one video frame of the sequence. xt and yt are moving averages of x and y across the time interval, t, including that frame. This time interval can be determined experimentally and must be of sufficient length to follow the bead as it diffuses throughout the available hemisphere (Han, et al., 2009). It also depends on the diffusion medium, as shown previously (Kumar, et al., 2014). Four seconds is adequate in low viscosity buffers, and such moving averages reveal the coordinates of the attachment point of the tether to the glass.
The average excursion values for DNA tethers attached to stalled complexes (556 bp) were statistically similar in DIC and dark field microscopy, as shown by their average values and standard deviations (155 ± 12 nm) and (160 ± 9 nm), respectively (Figure 8).
Any changes in tether length versus time due to DNA translocation by RNA polymerase should reveal elongation rates. To smooth the variation in this signal, the excursion during elongation was estimated using a moving average of 20 s as shown in Figure 9. Furthermore, in order to translate excursion values (nm squared) to tether lengths (bp), an experimental calibration curve was used (Kumar, et al., 2014). A calibration curve can be constructed using excursion or the square of excursion values measured for known contour lengths of DNA tethers. A curve based on the square of the excursion values has the advantage of being linear.
Although the example shown in Figure 9 concurs with expectation, there are noteworthy limits to the analyses of TPM data recorded for dynamically varying tether lengths. For example, pausing by RNA polymerases is a well subscribed feature of transcription (Artsimovitch & Landick, 2000; Forde, Izhaky, Woodcock, Wuite, & Bustamante, 2002; Neuman, et al., 2003). However, diffusion of the sub-micron sized, tethered bead requires 1–4 seconds to sample adequately the space of the available hemisphere and constitutes a large noise component that obscures small or transient underlying tether length changes.
One approach to determining analytical limits is to establish the accuracy with which transcriptional starts and stops can be detected. Simulations of tethered particle motion are useful for this endeavor and fairly simple to implement. At each instant of diffusive motion, the tethered particle responds to randomly oriented translational force transmitted from the thermal bath and a restoring force toward the anchor point due to elasticity of the worm-like chain. Using the Stokes-Einstein relation, D = kbT/(6πηR), and an expression for the tension in a worm-like chain (Bustamante, Marko, Siggia, & Smith, 1994), these forces and associated displacements are easily calculated and sequences of positions visited by a bead undergoing tethered particle motion can be created for tethers of arbitrary lengths including those that increase over time. Such simulations were coded in MatLab and comparisons of the variance of x and y positions from these simulations with actual experimental data were used for validation. After that, simulations were used to evaluate the accuracy with which the beginning and the end of an interval of linearly increasing tether length might be detected.
Figure 10 shows a representative “transcriptional” event (ramp) and the corresponding TPM signal that might result. There is a large variance in the data that grows as the tether lengthens. Within a MatLab fitting routine, minimization of the residuals between the data and a doubly jointed function built from an initial plateau, a linear increase, and a final plateau was used to estimate the coordinates of the joints. Just these two points are sufficient to specify the best estimate for the underlying “transcriptional” event. Figure 11 shows the correspondence between start and end times that were normally distributed around 50 and 100 seconds respectively with standard deviations of 4 s. Clearly there is a correlation, and two conclusions can be drawn. First is that the beginning and end points of a transcriptional interval cannot be pinpointed with great accuracy. The 95% confidence interval for estimates of the start of “transcription” is 10 s while that of the end spread across 18 s. This indicates that tether length changes corresponding to 100–200 bp, equal to one half of these intervals multiplied by a typical in vitro speed of 20 bp/s, might go undetected depending on the tether length. Secondly, the accuracy of these estimates degrades as the tether length increases.
It is important to consider these limitations when interpreting TPM data. Given the high intrinsic variance, information about the tether length requires averaging the excursion, ρ, over a relatively long time interval, which undermines precise pinpointing when those fluctuations change in magnitude. This essential time-averaging is why it is challenging to identify accurately the beginning and end of the period of linear growth of the tether length shown in Fig. 10. In a nutshell, for enzymes that produce progressive changes in the tether length, relatively larger changes over relatively longer time periods compared to the necessary averaging time can be reliably measured, while relatively small changes occurring relatively quickly may be difficult to detect with certainty.
The time resolution may be much better for dynamic experiments with enzymes that produce larger, abrupt tether length changes, especially in relatively short tethers. The DNA construct designed to study the wrapping of DNA around gyrase was 336 bp, and wrapping was expected to reduce the tether length by 40 to 100 bp of DNA. Although gyrase-catalyzed strand passage, which produces coiling or uncoiling, does require ATP, wrapping by gyrase does not, and therefore shortening of the tether due to wrapping was observed after addition of the enzyme alone to the microchamber. Since wrapping is reversible, the DNA tether length exhibited telegraphic-like signals (Figure 12).
The methods presented above can be used to assemble large, randomly distributed arrays of microspheres tethered by single molecules for simultaneous observation, multiplexing. While more sophisticated means to assemble and observe catalytic enzyme activity on single DNA molecules have been reported in the literature and may achieve higher densities, the methods presented above may suffice and do not require microfabrication technology. Certainly, care must be taken to avoid over-interpreting the results, since diffusion is an inherently noisy process and beads with 200–500 nm radius may diffuse rather slowly compared to the reaction of interest. Nonetheless, multiplexing is critical for achieving enough throughput to investigate the activity of proteins that are difficult to repair or reconstitute or that exhibit low activity in vitro.
We thank Qing Shao for the illustration of the microchamber and Dan Kovari for useful discussion. This work was supported by grants the National Institutes of Health GM084070 to L.F. and D.D.D. and GM084946 to D.A.S.