Supplementary material, Reblova et al, Loop E, Part I.
Principal Component Analysis (PCA)
PCA carried out on the simulated trajectories helps to identify the main collective motions and deformations during the simulation. Methodological principles of PCA method have been described e.g. in Sherer et al., 1999, in detail in Wlodek et al., 1997. Structural deformation during the simulation can be described as a vectorial representation - each deformation mode is represented by the eigenvector. The relative contribution of the mode in the global motion within the trajectory is indicated by its eigenvalue. The leading mode (denoted as 1st mode) is characterized by the highest eigenvalue. Projection of the trajectory on the selected PCA mode identifies the time-dependent behavior of the selected motion and determines characteristic PCA coefficient values. Motion associated with a given PCA mode may be visualized using artificial trajectory mapping changes between the minimum and maximum PCA coefficient values.
Table S1.
PCA analysis of the selected simulations. The relative contribution of the leading 3 modes in the global motion is represented as a proportion of the selected eigenvalue and the sum of all eigenvalues.
*Only trajectory portion with closed U77/A99 base pair (see the main text) was analyzed
**Initial 2 ns of the simulation were not included into analysis
Name of simulation / Proportion of the 1st mode (%) / Proportion of the 2nd mode (%) / Proportion of the 3rd mode (%)EbacMg / 18 / 11 / 10
EbacNa1* / 25 / 17 / 9
EbacNa2 / 23 / 16 / 10
EchlNa** / 23 / 12 / 11
Table S1 shows the proportion of the three leading PCA modes into the global motion. Evidently, the eigenvalues of the leading components are relatively small indicating absence of any dominant collective mode that could characterize the deformations occurring during the individual simulations. Nevertheless, we have inspected in detail the main PCA modes. PCA analysis has detected substantial oscillations of the inter-phosphate distances across the deep groove in all simulations. These deep groove width changes are further correlated, in the filtered PCA trajectories, with bending motions between the central non-Watson-Crick and flanking canonical segments. The bending is mainly localized in the A-tract region via A104/G72 and G98/A78 sheared base pairs. Nevertheless, participation of the central region in the global bending is not negligible.
Following Figures S1-S3 provide the graphical representation of the motions as indicated by the PCA analysis.
<Figure-S1, S2, S3>
Figure Legends:
Figure S1:
Motion associated with the 1st PCA mode in the EbacMg simulation as visualized by VMD code (maximum and minimum of the projection coefficient (Sherer et al., 1999; Wlodek et al., 1997). Arrows highlight the substantial deformations as narrowing of the deep groove, and bending of the upper end of the molecule, which is mainly localized near the sheared A104/G72 base pair.
Figure S2:
Motion associated with the 1st PCA mode in the EbacNa2 simulation as visualized by VMD code (maximum and minimum of the projection coefficient (Sherer et al., 1999; Wlodek et al., 1997). Arrows highlight the substantial deformations as narrowing of the deep groove in the central non-Watson-Crick region, and bending of both ends of the molecule.
Figure S3:
Motion associated with the 1st PCA mode in the EbacNa2 simulation as visualized by VMD code (maximum and minimum of the projection coefficient (Sherer et al., 1999; Wlodek et al., 1997). The same as Figure S2, however, side-view is used to highlight the bending. This Figure more clearly displays the bending motion with participation of not only the sheared G/A base pairs, but also of the central non-Watson-Crick region.
Selected structural parameters in the crystal and simulated structures
Presentation of global helical parameters calculated using 3DNA code. The same average structures were used as for calculating the PDB files.
Note, that the structures appear to be very well described by the global helical twist defined as the angle between consecutive C1' - C1' vectors. On the other hand variations of the remaining parameters are large (compared with the actual differences between the molecular structures) and do not provide an unambiguous description of the structural differences. Local structural parameters (tilt, roll, ..., not shown) would lead to a smaller differences, however, there again is no consensus yet on how to treat the parameters of steps with non-Watson-Crick pairs consistently.
<Figure-S4>
Unfolding of the Loop E in absence of salt
<Figure-S5>
Figure Legend:
Transformation of the trans W.C./Hoogsteen U/A base pair into trans Sugar-Edge/Hoogsteen (sheared) U/A base pair.