Probing a label-free local bend in DNA by single molecule tethered particle motion

Being capable of characterizing DNA local bending is essential to understand thoroughly many biological processes because they involve a local bending of the double helix axis, either intrinsic to the sequence or induced by the binding of proteins. Developing a method to measure DNA bend angles that does not perturb the conformation of the DNA itself or the DNA-protein complex is a challenging task. Here, we propose a joint theory-experiment high-throughput approach to rigorously measure such bend angles using the Tethered Particle Motion (TPM) technique. By carefully modeling the TPM geometry, we propose a simple formula based on a kinked Worm-Like Chain model to extract the bend angle from TPM measurements. Using constructs made of 575 base-pair DNAs with in-phase assemblies of one to seven 6A-tracts, we find that the sequence CA6CGG induces a bend angle of 19° ± 4°. Our method is successfully compared to more theoretically complex or experimentally invasive ones such as cyclization, NMR, FRET or AFM. We further apply our procedure to TPM measurements from the literature and demonstrate that the angles of bends induced by proteins, such as Integration Host Factor (IHF) can be reliably evaluated as well.


SI text
List of abbreviations ∥ : 2D-vector positions of the particle measured experimentally ∥ Amplitude of motion of the particle defined as 〈 ∥ 〉 ∥ Amplitude of motion of the particle corrected from the blurring effect ∥ 2D-vector positions of the simulated particle ∥ Amplitude of motion of the simulated particle defined as 〈 ∥ 〉 End-to-end distance of the DNA molecule Radius of the labelling particle

DNA constructs
The sequences of the 88bp long insert ( Fig. 1) are the following ones:

Assembly of the HT-TPM biochip
Patterning of anchoring sites on functionalized coverslips Regular arrays of rhodamine-labelled neutravidin (Molecular Probes) were obtained using a standard micro-contact printing protocol. Briefly, a stamp made of PDMS with squared pillars of 0.8 µm size and 3 µm pitch was inked with a 20 µg/mL neutravidin solution in PBS (Euromedex) for 1 min, then washed with deionized water and dried under nitrogen flow. The stamp was then brought into close contact with epoxidized glass support for 1 min during which the protein is transfered onto the surface.

Formation of DNA-bead complexes
Polystyrene carboxylated beads (Merck) of (150 ± 3) nm radius were covalently coated with antidigoxigenin antibodies (Roche) using EDAC (Sigma-Aldrich), activation and storage buffers (Ademtech). A 100 pM solution of suspended functionalized beads was mixed with an equal volume of a solution of 50 pM DNA bearing a digoxigenin on one end for 20 min at room temperature in PBS buffer supplemented with 1mg/mL of pluronic 127 (Sigma-Aldrich) and 0.1 mg/mL BSA (Sigma-Aldrich), noted TPM buffer. This lead to pre-assembled DNA-bead complexes.

Assembly of the fluidic observation chamber for HT-TPM experiments
A 250 µm thick silicone tape was cut and used as a spacer between the patterned coverslip and an epoxidized glass slide with 2 holes for inlet and outlet to obtain a working flow cell. The soformed analysis chamber was rinsed and incubated with TPM buffer for 30 min at room temperature. The DNA-bead complexes solution was introduced in the flow cell and incubated over night at 4°C.
Prior to visualization, the flow cell was extensively rinsed by injecting 30 chamber volumes of TPM buffer. Then, for each condition, movies of 5 min were recorded on different zones in the same flow cell and analyzed. To ensure reproducibility, experiments were repeated on different days.

Instrumentation for microscopy imaging
The tethered beads were visualized using a dark-field microscope (Axiovert 200, Zeiss) equipped with a x32 objective and an additional x1.6 magnification lens and acquired for 5 min at room temperature, at a recording rate of 25 Hz and with a duration of acquisition of 40ms, on a CMOS camera Dalsa Falcon 1.4M100. The field of observation covers an area of ~ 215 µm x 160 µm.

Single particle tracking
The software Nanomultiplex co-developed with Magellium Toulouse (request should be addressed Rexp||raw. In a general manner, we will use indifferently 〈 〉 and in the following. The averages, performed in the calculation of the asymmetry factors and amplitudes of motion, are taken over a sliding window of 5 s along the time trace. We invite the reader to refer to (Plenat T, Tardin C, Rousseau P, Salome L (2012) High-throughput single-molecule analysis of DNA-protein interactions by tethered particle motion. Nucleic Acids Research 40(12):e89-e89) for the detailed calculations of Rexp||raw of the bead.

Procedure of analysis for HT-TPM experimental data
In order to quantify the small differences expected on Rexp||raw, we set up a two-step procedure that is described in detail below. Briefly, it consists in selecting traces fulfilling several criteria of validity and applying corrections for detector temporal averaging to their Rexp||raw. All this procedure was performed with homebuilt Mathematica scripts (available upon request).

Criteria of validity of the DNA-bead complexes
First, we discard the trajectories that have mean asymmetry factors above 1.35, calculated as the average of the asymmetric factors measured along the time trace, or that have a mean amplitude of motion smaller than 1 nm or higher than 1000 nm. Then the probability distribution of the average of for each trajectory is built with the remaining trajectories and fitted by a Gaussian distribution centered on a mean value, called mean with a standard deviation (sd). As we noticed that a few trajectories had Rexp||raw averages standing out of the Gaussian distribution, we added a second step of validation to eliminate the misformed tethers with an average Rexp||raw outside the interval (mean ± 2.5 sd). Using this criterion, no more than 1.3% of valid trajectories were eliminated during this additional step.
In total, about 12% trajectories were eliminated and the final number of valid trajectories eventually ranged between 348 and 3496 (See Table 1), depending on the DNA construct.

Correction of time averaging effect
Finite exposure time of detectors, , , equal here to 40 ms, can lead to a blurring effect in single molecule (or particle) tracking experiments, as investigated for example in (Manghi M, et al. (2010) Probing DNA conformational changes with high temporal resolution by tethered particle motion. Physical Biology As previously, the trajectories with an average ∥ outside the interval (mean ± 2.5 sd) were eliminated.

Calculation of the amplitude of the motion of an ensemble of particles
Lastly, the experimental value of the amplitude of the motion of an ensemble of particles was obtained by fitting the probability distribution of ∥ with a Gaussian, which gave us its mean value as its center. The error on the amplitude of the motion of an ensemble of particles was calculated by using the bootstrap method of R software (R Foundation for Statistical Computing, Vienna, Austria). Doing so, we find a typical error of 0.4 nm (See Table 1).

DNA coarse-grained model
We Since the polymer motion is limited to the upper half-plane, we impose a "hard wall" boundary condition for monomer spheres and for the particle. All spheres interact via stretching and bending forces. We invite the reader to refer to (Manghi M, et al. (2010) Probing DNA conformational changes with high temporal resolution by tethered particle motion. Physical Biology 7(4):046003) for the details of these Kinetic Monte Carlo simulations.

Simulated bent DNA
The bent sequences used in the experiments are simulated by setting a fixed angle between three successive monomers located at the center of the DNA molecule. A full range of angles were studied in successive simulations: 0,18,30,45,50,60,72,90,120 and 180 degrees.
Extraction/Calculation of the particle to anchor 2D-distance of the simulated particle The 2D-vector of the particle position ∥ is measured throughout simulations and utilized to estimate the amplitude of motion defined as 〈 ∥ 〉 , the average being taken along the trajectory.

Theory for a local bend
We consider a homogeneous polymer of length L using the Worm-Like Chain (WLC) model by  (12):1106-1122). We suppose that a kink is located at distance l from one end and locally induces a spontaneous curvature with an angle θ, see below.
We denote the tangent vector at curvilinear position s by t(s) , and the tangent-tangent correlation function is given by 〈t(s) . t(s′) 〉 = 9[\]−|P − P Q |/B _`.
The mean-square end-to-end distance is defined as : 〈 〉 = a bP

Theory for a local stiffer insert
We use the same model and assume now that a stiffer insert of length m and persistence length B _n is inserted at position o, see below: The mean-square end-to-end distance is again 〈 〉 = a bP Experimental data (symbols) obtained after deconvolution of the particle effects using the minimal model and considering the size of labelling particle equal to 150 nm (•) and 155 nm (○), and two series of fits with R WLC and θ 1 (−) (already shown in figure 3), or R WLC , θ 1 and a rigidity term α (…) as free parameters. The results are shown in S3 Table. S5