Table 1.
TATA-Box location is highlighted in bold.
Fig 1.
Covariance matrix of the twist fluctuations obtained for the central 43 base-pair steps during MD-simulations (at-sequence).
In the unrestrained case (left panel), coupling exists only between nearest neighbor sites. Upon global unwinding (right panel, σ = 0.067), also distant sites become coupled to a significant extent.
Fig 2.
(A) Twist-distribution sampled for a CpG nucleotide step during unrestrained MD simulation. The probability distribution is bimodal due to the sampling of the two possible backbone substates termed BI and BII. BI states are associated with an on average lower twist and have higher variance than BII states for the associated CG base pair step. (B) Changes in the overall BI population at different levels of global unwinding (for AT and CG steps). At a σ = 0.067 an increased BI level is observed that relaxes for higher σ due to local DNA melting and relaxation of the rest of the dsDNA.
Fig 3.
Setup of the simulation systems (as stick models with the backbone indicated as cartoon).
Yellow circles indicate positions of the C1’ atoms (4th and 47th bp), which constitute the torsional reaction coordinate, illustrated as dashed line (unwinding is indicated as curved arrow in the middle of the left panel). Step-wise unwinding eventually leads to melting of the TATA-box segment (located in the lower half of the DNA molecule, right panel). C/G bases are shown in green, A/T bases in blue.
Fig 4.
Snapshots of two selected 8 bp segments of the DNA at different levels of global unwinding (gc-sequence).
Structures of the TATA-box containing segment are shown in the upper series. In the lower panels structures of a distant GC-rich segment are illustrated (same color-coding as in Fig 3).
Fig 5.
Average changes in twist at each base pair step (relative to unrestrained DNA, at-sequence) upon global unwinding for a superhelical stress σ slightly below melting (blue) and above melting in the TATA box segment (red line).
The average deviations (with respect to fully relaxed unrestrained DNA) and standard deviations were calculated as averages of three adjacent base-pair steps.
Fig 6.
Change in total twist in the TATA box segment upon global unwinding for the at-sequence (A) and gc-sequence (B).
In the regime I (harmonic regime) a continuous unwinding is observed that changes abruptly during the melting phase (regime II). During this phase-transition global unwinding is largely stored in the TATA-box segment. Further unwinding causes continuous unwinding of the TATA-Box (regime III) with a steeper slope than in regime I (due to the much smaller twist persistence length of melted vs. intact DNA). Global twisting and twisting of the TATA-Box segment have been computed with a protocol described in chapter 5 in S1 File (same as in Table 2).
Table 2.
Deformation of the TATA-Box under different levels of unwinding, averaged over both sequences.
Parameters have been obtained by computation of the rigid-body transformation from the C-G base-pair prior to the TATA-Box to the first C-G base-pair after it. Our procedure is explained in chapter 5 in S1 File.
Fig 7.
Average elastic energy per degree of freedom and changes in BI/BII conformational substate population with respect to unrestrained simulations for each of the central 43 bp-steps of the heterogeneous at-sequence (left column) and the gc-sequence (right column).
Each vertical dashed line in the panels corresponds to a base pair (sequence indicated on the x-axis). The black lines indicate the average deformation energy at each base pair step. The cyan bars indicate an increase in BII, orange bars an average increase in BI states relative to the unrestrained simulation. The profiles have been generated for different levels of global DNA unwinding. Upper panels show simulation results of Umbrella windows with low unwinding. Middle panels illustrate moderate unwinding simulations, where the DNA duplex has still remained structurally intact. Bottom panels reflect simulations at strong unwinding, which results in TATA-Box melting for both sequences. In order to calculate , we determined deformation energies (Eq 4) for every frame and subsequently calculated averages (over three adjacent base-pair steps). Note that a harmonic energy function is insufficient in the denatured phase, corresponding values therefore only have a qualitative meaning. Error bars have been calculated as standard errors by splitting the simulation into bins of 25 ns.
Fig 8.
Display of the sequence-dependent impact of global unwinding.
In the absence of global stress, all sequences undergo equal levels of deformation (, A, highlighted in blue). Global unwinding then causes G-C sequences to absorb most of the stress (red color in B), while promoter like sequences remain rather relaxed (light blue color in B). At strong levels of unwinding, global stress is absorbed through melting of the TATA-Box (red in C), whereby distant sequences are relaxed again (blue in C).
Fig 9.
Free energy profiles for global unwinding of the DNA molecules (σ) obtained from US simulations (PMF) and calculated by Ising models.
Error bars of the PMFs have been calculated by thermodynamic integration and block-averaging [48].
Table 3.
Stacking energies of the ten different base pair steps used in the Ising model.
For base pairing we used Ebp = 0.64 and Ebp = 0.12 for AT and GC pairs, respectively. All values have been adapted from [49] and are in units of kcal/mol.
Fig 10.
Calculated melting probabilities along the sequence using the Ising model at a global supercoiling stress of σ = 0.067.