Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2012 Oct 9;109(41):16552-7.
doi: 10.1073/pnas.1120452109. Epub 2012 Sep 25.

Sequence-dependent sliding kinetics of p53

Jason S Leith  1 , Anahita Tafvizi, Fang Huang, William E Uspal, Patrick S Doyle, Alan R Fersht, Leonid A Mirny, Antoine M van Oijen
Affiliations

Sequence-dependent sliding kinetics of p53

Jason S Leith et al. Proc Natl Acad Sci U S A. .

Abstract

Proper timing of gene expression requires that transcription factors (TFs) efficiently locate and bind their target sites within a genome. Theoretical studies have long proposed that one-dimensional sliding along DNA while simultaneously reading its sequence can accelerate TF's location of target sites. Sliding by prokaryotic and eukaryotic TFs were subsequently observed. More recent theoretical investigations have argued that simultaneous reading and sliding is not possible for TFs without their possessing at least two DNA-binding modes. The tumor suppressor p53 has been shown to slide on DNA, and recent experiments have offered structural and single molecule support for a two-mode model for the protein. If the model is applicable to p53, then the requirement that TFs be able to read while sliding implies that noncognate sites will affect p53's mobility on DNA, which will thus be generally sequence-dependent. Here, we confirm this prediction with single-molecule microscopy measurements of p53's local diffusivity on noncognate DNA. We show how a two-mode model accurately predicts the variation in local diffusivity, while a single-mode model does not. We further determine that the best model of sequence-specific binding energy includes terms for "hemi-specific" binding, with one dimer of tetrameric p53 binding specifically to a half-site and the other binding nonspecifically to noncognate DNA. Our work provides evidence that the recognition by p53 of its targets and the timing thereof can depend on its noncognate binding properties and its ability to change between multiple modes of binding, in addition to the much better-studied effects of cognate-site binding.

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Fig. 1.
Fig. 1.
Energy landscapes and cartoons of proteins on DNA in search (S) and recognition (R) modes. (A and B) In S mode, a generic protein (yellow) interacts chiefly with the DNA backbone and experiences a smooth landscape. In R mode, it interacts with the nucleobases, yielding a highly sequence-dependent landscape. (C) Cartoon model for p53, based on EM data (25), indicates the domains responsible for the modalities: green C-terminal domain for the S mode; red core domain for the R mode. Tetramerization domain in orange.
Fig. 2.
Fig. 2.
Experimental setup and initial data analysis. (A) Biotinylated λ DNA is flowed into the cell and adheres to the streptavidin-coated surface. The DNA is stretched by hydrodynamic drag. Labeled p53 proteins are imaged diffusing along the DNA. After a series of protein movies are taken, the DNA is stained and imaged. (B) Kymogram of a single p53 protein diffusing on DNA. Flow direction is up; every fourth frame is shown, giving an apparent frame rate of 120 ms. (C) Trajectories of three particles (gray). The dotted black trace represents the bottommost trajectory (kymogram in B), corrected for drift.
Fig. 3.
Fig. 3.
Data analysis: diffusion coefficients of p53 on λ-phage DNA. (A) Trajectories of selected particles in three representative segments. The trajectories have been spread out horizontally for clarity. Portions of trajectories are colored according to the segment in which they lie: red, green, and blue for segments 1, 2, and 3, respectively. The positions and assigned segments for each particle’s displacements are shown to the right. (B) Squares: squared range of the central 95% of each trajectory, plotted over the trajectory’s midpoint. Colored squares represent particles shown in A. Gray circles: mean squared displacement (MSD) at long (> 100 ms) time windows, Δt, of quantum dots fixed at one-third and two-thirds the contour length from the tether point. Shaded region is the MSD of the DNA at long Δt. (C) Horizontal lines consist of dots plotted on the horizontal axis at the midpoint of each displacement within a trajectory, and on the vertical axis at the D corrected for drift and DNA fluctuations of their respective particle. Colored dots correspond to colored dots in A. (D) Estimated Dexpt for each segment, from averaging all values in C. Colored bars correspond to coloring scheme for A–C. Uncertainties were determined by bootstrapping: The particles contributing to each segment’s D were resampled 1,000 times, and the resulting diffusion coefficients calculated. Black error bars represent a standard deviation in the resampled diffusion coefficients above and below the estimated D; cyan error bars sample only half the number of particles (SI Text, Significance and Consistency of Experimental Results). (E) Thin solid traces are MSD/Δt for particles whose median position lies within segments 1 (red), 2 (green), or 3 (blue). For clarity, only every third particle is shown. Each particle is analyzed as in Methods, and for each particle shown, a corresponding dashed line plotted with slope 2D. Thick solid colored traces are the weighted MSD/Δt for the particles shown. Solid black traces are the MSD/Δt for all particles in all nine segments presented. Inset (same axis units) shows MSD/Δt for the DNA (gray); black and thick blue lines are the same as in the main panel.
Fig. 4.
Fig. 4.
Theory: scoring the λ genome and predicted landscapes. (A) Half-site sequence logo for p53. (B) 1–5: Sequences and positions in bp from the tether of full-sites found in segments 4–6 of λ DNA shown in C and E. Lowercase letters indicate nucleotides that do not match the consensus sequence of RRRCWWGYYY. Sequence 5 is the strongest-scored fullsite among all segments. Asterisk (*): Sequence from among the known p53 RE whose Kd has been measured by Weinberg et al. (40) that has the shortest Hamming distance to sequence 5. (C) Predicted one-mode landscape in segments 4–6. Red, full-sites; blue, half-sites. (D) Key elements of two-mode model. The statistical weights of fully specific, hemi-specific, and nonspecific binding at a position x in making up U(x) [5] are indicated. For the great majority of positions on DNA that lack a half-site, the greatest of these is the term representing fully nonspecific binding. For positions that include a half-site, it is the term representing hemi-specific binding. For full-sites, it is the term representing fully specific binding. Kd’s are of p53 to representative of full-sites, half-sites, and nonsites (24). (E) Predicted two-mode effective landscape. Most positions are dominated by nonspecific binding. The possibility of hemi-specific binding makes half-site binding relatively more important for the two-mode model than for the one-mode model.
Fig. 5.
Fig. 5.
Comparison of theory, simulations, and experiment. (A) (Right) Random third of trajectories with center in segment 4, ordered by increasing estimated D. (Left) Predicted potential wells denoted by red (full-sites) and blue (half-sites) bars, with height of bars proportional to predicted effective energy, U(x). Gray bars are a histogram of observed occupancy of all particles within the segment, with bin widths equal to one-twentieth the segment width. (B) Estimated D for experimental (black bars, same as in Fig. 3D) and predicted D/D0 (red bars) for the predicted landscape, over segments along λ DNA. D/D0 is scaled to match Dexpt’s mean and coefficient of variation. Green trace is the percent error in predicted D/D0 normalized by the mean D/D0, relative to Dexpt normalized by the mean Dexpt. (C) Scatter plot of Dexpt versus D/D0 for all segments. Red circles correspond to values for the predicted landscape based on the two-mode model; cyan x’s correspond to values for the control landscape whose correlation with Dexpt was the median from among the 500 control landscapes. (D) Black bars are identical to those in B. Cyan bars correspond to D/D0 of the control landscape that produced the cyan x’s in C. (E) Summary statistics of experimental data. For each segment, we show the number of particles contributing to the segment’s Dexpt, the number of displacements contributing likewise, the estimates of Dexpt as determined in Data analysis in Methods, and the standard deviation of particle’s D weighted by the number of displacements contributed by that particle. (F) Correlation coefficients and p-values.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Vogelstein B, Lane D, Levine AJ. Surfing the p53 network. Nature. 2000;408:307–310. - PubMed
    1. Tafvizi A, et al. Tumor suppressor p53 slides on DNA with low friction and high stability. Biophys J. 2008;95:L01–3. - PMC - PubMed
    1. Berg OG, Winter RB, von Hippel PH. Diffusion-driven mechanisms of protein translocation on nucleic acids. 1. Models and theory. Biochemistry. 1981;20:6929–6948. - PubMed
    1. von Hippel PH, Berg OG. Facilitated target location in biological systems. J Biol Chem. 1989;264:675–678. - PubMed
    1. Kabata H, et al. Visualization of single molecules of RNA polymerase sliding along DNA. Science. 1993;262:1561–1563. - PubMed