Figure 1.
Target proteins and structural features.
The 3D structures of ARID3AFREE (PDB entry 2KK0), DriFREE (PDB entry 1C20) and Dri-DNA (PDB entry 1KQQ) are shown as yellow, light grey and magenta cartoons, respectively. In the figure at the bottom on the left the structure is colored with different shade of colors from the N- (blue) to the C-terminal extremity (red) and the secondary structural elements are labelled according to ref. [6]. In particular, α0–α7 (residues 231–234, 239–254, 272–282, 285–291, 294–300, 310–321, 324–329, 335–346 in ARID3A or according to the PDB entry 2KK0 numbering 25–28, 33–48, 66–76, 79–85, 88–94, 104–115, 118–123, 129–140) and a very short β-harpin consisting of antiparallel strands β1 and β2 (residues 264–265 and 268–269 in ARID3A or according to the PDB entry 2KK0 numbering 58–59 and 62–63) are shown.
Table 1.
Summary of the MD simulations.
Table 2.
Prediction of backbone chemical shifts by PPM in ARID3AFREE and DriFREE simulations.
Figure 2.
Comparison of LMI matrices describing correlated motions.
The Frobenius norms between average LMI matrices calculated from different replicates of the same target protein are reported. An average LMI matrix achieved from an unfolding simulation at 500K of the same protein is used as a control. The example of DriFREE simulations is reported for sake of clarity. Similar results with even lower Frobenius norms were obtained for ARID3AFREE replicates (average Frobenius norm of 5). Average LMI were calculated with five-ns time-windows, as explained in Materials and Methods. The box on the bottom shows the values of Frobenius when the two halves of each replicate are compared to use as a baseline in the comparison.
Figure 3.
Evolution of the size of the largest cluster (cluster 1) as a function of the Imin values in Dri and ARID3A simulations.
The Icrit value for the PSN analysis was calculated for each of the simulations collected in this study, included wild-type, mutant variants and DNA-bound structures. The analyses point out the consistency of this parameter within proteins sharing the same fold but different primary sequences, as well as in their DNA-bound and DNA-unbound state, as also previously pointed out for similar analysis on single PDB structures [17].
Figure 4.
Hub residues and their location on the 3D structure of ARID domains.
The connectivity degree for each PSN hub of DriFREE (A) and DriDNA (B) simulations are shown as a function of the protein residue in the left panels. Since in a PSN a hub is defined as a residue connected by at least three edges, all the residues with node degree lower than three are set at zero. Black triangles indicate the position of the residues experimentally known to impair DNA-binding [9], whereas a gray triangle indicates Q101/Q111 position. The right panels show the location of these hubs on a reference 3D structure of the corresponding target protein. The structure is depicted as ribbon with rainbow shade of colors according to the node degree. The corresponding figures for ARID3AFREE and 1 µs DriFREE simulations are shown in Figure S4. The residue numbering is referred to Dri in the figure where P47, W78, F96, Q101 and Y109 correspond to P57, W88, F106, Q111 and Y119 in ARID3A.
Figure 5.
Long-range paths are promoted by DNA interaction in Dri simulations.
Left panel) Distribution of the shortest paths of communication identified by the PSN-LMI approach in the different simulations employed in this study. Average values from the different replicates of the same system are shown. Right panel) Differences among the paths of length higher than 8 residues (starting and final residues included) in DriDNA and DriFREE simulations. The paths that are present only in DriDNA or DriFREE simulations are shown as red and blue cylinders, respectively. Nodes of the paths are shown as spheres centered on the Cα atom. The nodes belonging to the paths from K113 to both L56 at the base of L1 (K113→Y109→L32→L106→F28→L56) and L87-I91-P88 in L2 (K113→Y109→L32→L106→L59→L98→L87→I91→P88, Dri numbering) are highlighted by different shades of colors from red to blue/black. K113, L56 and P88 are shown as sticks.
Figure 6.
Shortest paths of long-range communication in ARID3A and Dri.
The edges which are present with highest probability in the paths of long-range communication identified by PSN-LMI approach are indicated for ARID3AFREE (blue) and DriFREE (dark green) simulations by lines of thickness proportional to the occurrence probability. The residues connected by the edges are indicated as spheres, whereas the residues known to compromise the DNA-binding capabilities in ARID3A, along with Q111/Q101 and R109/R99 are shown as sticks if identified in the graph.
Figure 7.
FCA modes of L1 and L2 in wt ARID3AFREE or DriDNA and their mutants.
Upper panels) The two panels show the projection along FCA mode 1 and FCA mode 2 for L1 and L2 in ARID3AFREE and its mutants. ARID3AFREE, ARID3AY119A, ARID3AQ111A and ARID3AQ111N are shown with different shade of colors. Bottom panels) The two panels show the projection along FCA mode 1 and FCA mode 2 for L1 and L2 in DriDNA and its mutants. DriDNA, DriY119A, DriQ111A and DriQ111N are shown with different shade of colors. The projections report one replicate of each system, for sake of clarity. The same analysis was carried out for each combination of replicates of wild type and mutant variants and it provides similar results, as also expected by the high overlap in the essential subspace between individual replicates of the same system (root mean square inner product larger than 0.8).
Figure 8.
Paths of long-range communication mediated by Y119 in wt and Y119A ARID3A.
The shortest paths of communications mediated by Y119 in ARID3AFREE (left panel) and ARID3AY119A (right panel) simulations as identified by the PSN-LMI approach are shown as cylinders of thickness proportional to the probability of occurrence. The Y119 Cα and the L64 Cα are also indicated by orange and yellow spheres, respectively. L1 and L2 are highlighted in yellow.
Figure 9.
Paths of long-range communication mediated by Q111 in wt, Q111A and Q111N ARID3A.
The shortest paths of communications mediated by Q111 in ARID3AFREE, (left panel) ARID3AQ111A (middle panel) and ARID3AQ111N (right panel) simulations as identified by the PSN-LMI approach are shown as cylinders of thickness proportional to the probability of occurrence. The Q111 Cα and the L64 Cα are also indicated by orange and yellow spheres, respectively. L1 and L2 are highlighted in yellow.