Toward a detailed description of the thermally induced dynamics of the core promoter

Abstract

Establishing the general and promoter-specific mechanistic features of gene transcription initiation requires improved understanding of the sequence-dependent structural/dynamic features of promoter DNA. Experimental data suggest that a spontaneous dsDNA strand separation at the transcriptional start site is likely to be a requirement for transcription initiation in several promoters. Here, we use Langevin molecular dynamic simulations based on the Peyrard-Bishop-Dauxois nonlinear model of DNA (PBD LMD) to analyze the strand separation (bubble) dynamics of 80-bp-long promoter DNA sequences. We derive three dynamic criteria, bubble probability, bubble lifetime, and average strand separation, to characterize bubble formation at the transcriptional start sites of eight mammalian gene promoters. We observe that the most stable dsDNA openings do not necessarily coincide with the most probable openings and the highest average strand displacement, underscoring the advantages of proper molecular dynamic simulations. The dynamic profiles of the tested mammalian promoters differ significantly in overall profile and bubble probability, but the transcriptional start site is often distinguished by large (longer than 10 bp) and long-lived transient openings in the double helix. In support of these results are our experimental transcription data demonstrating that an artificial bubble-containing DNA template is transcribed bidirectionally by human RNA polymerase alone in the absence of any other transcription factors.

Figure 1. Core promoter sequences analyzed by PBD Langevin dynamics simulations.

Experimentally verified transcriptional start sites (TSS) are shown in large letters. Common promoter sequence elements are indicated by colored boxes. For illustrative purposes, sequences that fit the element definitions but are not properly positioned relative to the TSS are also shown as colored letters. Deviations from the consensus sequence are indicated in gray. The sequences were obtained from the Eukaryotic Promoter Database (EPD, http://www.epd.isb-sib.ch/). The identity of each promoter is described in column 1, the sequence is shown in column 2, and the mode of regulation in column 3.

Figure 2. Schematic representation of a DNA bubble with length l [bp] and amplitude tr [Å] at postition n.

Figure 3. Probability for DNA collective openings of mammalian core promoters, calculated from PBD Langevin dynamic simulations.

The probability was determined from the lifetimes of all open states above a given length and amplitude, normalized over the time of the simulation (Eq. 2). (A) Probability for opening (vertical axis) starting at specific nucleotide positions (horizontal axis), as a function of bubble length [bp]. Probability values are colored to the same scale between promoters for comparison. Nucleotide positions are labeled relative to the TSS (+1). Promoter identity and bubble amplitude thresholds are shown at the top. The thresholds are chosen individually for each promoter, as the smallest values for which the TSS region begins to exhibit maximum probability. (B) Probability for opening (vertical axis) starting at specific nucleotide positions (horizontal axis), as a function of bubble amplitude [Å]. Probability values are colored to the same scale between promoters for comparison. Nucleotide positions are labeled relative to the TSS (+1). Promoter identity and bubble length thresholds are shown at the top of the panels. The thresholds are chosen individually for each promoter, as the smallest values for which the TSS region begins to exhibit maximum probability.

Figure 4. Average lifetimes of DNA collective openings of core promoter sequences, as a function of length[bp].

For clarity, the same promoter profiles are shown from a different angle in the panels at the right. Nucleotide positions are shown relative to the TSS (+1). The TSS is marked with a vertical line. The color scale represents the average lifetimes [ps]. The identity of the promoters is shown above the panels.

Figure 5. Supercoiling and artificial mismatch bubbles enable transcription from the P5 promoter according to the Usheva, Shenk (1996) experiment.

(A) Artificial mismatch bubbles enable bidirectional transcription from the P5 promoter by human PNAP2 in the absence of transcription factors. All reactions received 2 units of purified RNAP2 and different amount of synthetic linear ds DNA template with the AAV P5 promoter as indicated at the top of the lines. The DNA template in reactions 1, 2, and 3 contains 5 bp long mismatches creating a “bubble” in the region of the transcription start site. The reaction in line 4 received ds DNA with no mismatch. The ³²P- labeled reaction RNA products have been separated by gel electrophoresis based on difference in the size of the transcripts. The position of the specific RNA transcripts is shown on the left: tr1- transcripts that initiate at the bubble and terminates at the 5′-prime end of the DNA template; tr2 – transcripts initiated at the bubble and terminated at the 3′-end of the template. The migration of DNA size markers was used to determine the position of the specific transcripts (not shown). Schematic diagram of the experiment is presented at the left. The bidirectional transcription from the mismatched DNA template (gray) is labeled with black arrows. The promoter region is labeled with red and the polymerase with blue (P). (B) Bubble lifetime as a function of length and amplitude at 310°K, shown for individual base pairs of both, the wild type (wt) P5 and the mutant (mt) P5 variant. Each square presents the average lifetimes (color scale) of all bubbles at a given amplitude (vertical axis) and length (horizontal axis), containing a given base pair (top right). Transcription starts at base pair +1.

Figure 6. Collective opening profiles of the collagen nonpromoter sequence calculated from the PBD Langevin dynamic simulations.

(A) Probability for collective opening (vertical axis) of ten base pairs starting at specific nucleotide position within the collagen intron (horizontal axis), as a function of bubble amplitude [Å]. For comparison the profile of the collagen promoter is also presented (bottom panel). Probability values are colored to the same scale between the promoter and the intron sequences, as shown below the plots. Nucleotide positions in the collagen promoter are labeled relative to the TSS (+1). The sequence identity is shown at the top. (B) Probability for opening (vertical axis) of amplitude threshold (tr)≥1 Å, starting at specific nucleotide positions (horizontal axis), as a function of bubble length [bp]. Probability values are colored to the same scale, as shown below the plots. The sequence identity is shown at the top. (C) Average lifetimes of DNA collective openings of amplitude tr≥1 Å (vertical axis), starting at specific nucleotide positions (horizontal axis), as a function of length [bp]. The average lifetimes of collective openings for the collagen promoter are shown below. The TSS is marked with a vertical line. The color scale shown below the plots represents the average lifetimes [ps].

Figure 7. Average base pair separation coordinates for the AAV P5 promoter.

Average base pair separation coordinates [Å] calculated from the Langevin dynamic trajectories of the AAV P5 promoter (black line) and a transcriptionally silent mutant (red line). For comparison, the average coordinates calculated with Monte Carlo simulations are also shown (gray line). The P5 sequence is shown under the plot. The transcriptional start site (TSS) is marked with a blue line. Mutated residues that silence transcription are shown in red letters.