Fig 1.
Transcriptional activation via the Mediator complex.
The schematic diagram shows the gene-specific transcription factor GCN4 bound sequence-specifically to a target site ("Upstream Activation Site"; "UAS"). The basal transcriptional machinery (including RNA polymerase II [RNAPII]; basal transcription factors TFIIB, TFIID, TFIIE, TFIIF and TFIIH; the multi-subunit "Mediator Complex") assembles around the transcription start site indicated by an arrow. The GAL11 subunit of the Mediator is specifically identified. The regulatory protein-protein interaction of the GCN4-activation domain (cAD) with GAL11 is represented by a two-headed arrow.
Fig 2.
Transcriptional activation of GCN4 via Mediator GAL11.
A. Domain organization of GCN4 (top) and GAL11 (bottom). The positions of the two tandem activation domains (N-terminal AD, central AD [cAD]) and the DNA-binding domain (DBD) of GCN4 are illustrated. A sequence logo based on GCN4 orthologs from 29 different species (see S1 Text for sequence alignments) shows the absolute conservation of three large hydrophobic residues (corresponding to W120, L123 and F124 in Saccharomyces cerevisiae GCN4). GAL11 contains three domains,ABD1, ABD2 and ABD3, that each can bind GCN4 independently. B. Structure of GAL11-ABD1. Cartoon presentation of the uncomplexed structure showing positions of the four separate α-helices (α1 to α4 from N- to C-terminus). C. Surface view showing the location of three deep pockets that display a capacity for binding hydrophobic side chains of partner proteins. Liquorice representations of probe clusters of small organic compounds are shown fitted into three distinct pockets (red = "Pocket#1"; blue = "Pocket#2"; magenta = "Pocket#3") mapped by a computational solvent mapping program [32]. The model shows GAL11-ABD1 residues 164–233. D. Model of the GCN4-cAD bound to GAL11-ABD1. The GAL11-ABD1 and GCN4 structures are shown in silver and purple, respectively. Five different positions, reflecting orientations at 200 ns intervals of aMD_no1, are drawn for the GCN4 cAD in cartoon representation.
Table 1.
Summary of MD simulations.
Fig 3.
Differential GAL11-ABD1 pocket occupancy by GCN4-W120 and F124.
Panels A—C show the GAL11-ABD1 in grey surface representation and the GCN4-cAD as a purple ribbon. The key hydrophobic residues GCN4-W120 and F124 are shown as liquorice models in light blue and green, respectively. The time-frame shown in the right corner of each panel corresponds to the 4 x 1000 ns aggregate trajectory shown in Panel D. A. GCN4-F124 located in the ABD1 pocket near the beginning of aMD_no2. B. Co-occupancy of GCN4-W120 and F124 inside the ABD1 pocket. Note that the cAD α-helix is additionally rotated by 90° in comparison to (A) and (C). C. GCN4-W120 located within the ABD1 pocket towards the end of aMD_no4. Note that this simulation, just like the other three started out with GCN4-F124 in the pocket. D. Distance measurement [Å] between residue GAL11-A216, which forms the floor of the ABD1 pocket, and GCN4-W120 (blue trace) or F124 (green trace). For this analysis the four aMD simulations trajectories were concatenated to a single trajectory before processing to allow cross-comparison of the data under identical conditions.
Fig 4.
Changes in helical orientation of the GCN4-cAD relative to GAL11-ABD1.
A. The three panels show the GAL11-ABD1 in silver surface representation and the GCN4-cAD as a purple ribbon. The key hydrophobic residues GCN4-W120 and F124 are shown as liquorice models in light blue and green, respectively. The time-frame shown in the right corner of each panel corresponds to the 4 x 1000 ns aggregate trajectory shown in Panel D. All representations are aligned to the long α-helix 4 of ABD1 (see Fig 2B for annotation of helices; residues GAL11-Q211 and L232 were chosen as vector endpoints for α-helix 4). The helical axis of GCN4 is marked by a white arrow. B. Measurements of the angle of α-helix 4 of ABD1 in relation to the GCN4-cAD α-helix during simulations GAL11-ABD1/GCN4-cAD _aMD_no1, no 2, no3 and no4. For the cAD α-helix, residues GCN4-S117 and F124 were selected. C. Time-dependent representation of the 90° rotation performed by cAD (cartoon representation, red to blue) in GAL11-ABD1/GCN4-cAD _aMD_no3. The different helical orientations (shown in cartoon representation) at different time points throughout the one microsecond simulation are superimposed on each other. The color of the helix indicates the time point according to the color gradient scale shown on the right. The GCN4 helix undergoes a number of different orientations (~100°, ~150°, ~50° and ~0° according to the criteria defined in B).
Fig 5.
Distribution of the 13 experimentally-based models within aMD simulation phase space.
Each of the blue dots defines the position in phase space of a snapshot from the four combined aMDs (GCN4-cAD/GAL11-ABD1_aMD_no 1 to 4) sampled at 1 ns intervals. The red dots define the coordinates of the 13 models calculated using identical parameters (the numbers in grey identify the model numbers as used in PDB#2LPB). The distances and angles were calculated according to the criteria described in the legends of Figs 3 and 4. A. Plot of the distance of F124 relative to Pocket#1 (horizontal axis; distance in Å) versus the angle of the α-helical part of GCN4-cAD relative to GAL11-Abd1 α-helix 4 (vertical axis; angle in degrees). B. Plot of the distance of W120 (horizontal axis) versus the distance of F124 relative to Pocket#1 (vertical axis; both distances in Å). C. Plot of the distance of W120 relative to Pocket#1 (horizontal axis; distance in Å) versus the angle of the α-helical part of GCN4-cAD relative to GAL11-Abd1 α-helix 4 (vertical axis; angle in degrees). There are five PDB#2LPB model structures (#6, 7, 8, 12 and 13) that are located at the periphery of the phase space simulated by the aMD simulations using this particular parameter set. These outliers are characterized by a relatively large distance of W120 from Pocket#1 (~10–15 Å) and by narrow angle (~40–80°) of the GCN4-cAD α-helical portion relative to GAL11-Abd1 α-helix 4. This set of conformations may either require more extensive sampling to be reached from 2LPB-model#1 as a starting structure, or represent conformations that are not energetically favorable under the forcefield conditions used in the simulations.
Fig 6.
Molecular mechanics analysis of the GAL11-ABD1/GCN4-cAD interaction.
A. The decomposition of the van der Waals contribution is shown as a contour plot. The horizontal axis represents the amino acid sequence of the GCN4 activation that was represented in the simulations and in the models in PDB#2LPB. The vertical axis represents snapshots at 1 ns intervals from the four aMD simulations. The ΔG value of the van der Waals contribution of each residue at each time point is color-coded according to the scale shown (substantial contributions to ΔG are green and dark blue). The data derived from independent simulations (indicated on the left; aMD_no1 is represented by frames 1–1000, aMD_no2 by frames 1001–2000 etc.) are shown on the same plot to facilitate the detection of constant and variable features. Residues making significant energetic contributions are highlighted by a red-dotted line aligned to the amino acid sequence. B. Snapshot of GAL11_GCN4_aMD_no2 (frame 935) demonstrating the hydrophobic interaction between GAL4-M107 (magenta liquorice representation) with the leucine pair GAL11-L169 and L227 (light blue liquorice representation). A movie showing the dynamics of this interaction is available (S2 Movie).
Fig 7.
Structural basis of pocket #1 occupancy by GCN4-cAD residues W120 and F124.
GCN4-cAD residues W120 and F124 are displayed in liquorice representation in light blue and green, respectively, on a portion of the GCN4 α-helix shown as a purple ribbon. Key residues participating in the formation of Pocket #1 on the GAL11-ABD1 surface are drawn as van der Waals structures and labelled according to amino acid identiy (V170 white; M173 pink; Y220 red; K217 and K 221 in light blue). V170 and M173 are part of GAL11-ABD1 α-helix 1 (see Fig 2B for helix nomenclature) marked in dark grey. K217, K221 and Y220 are on GAL11-ABD1 α-helix 4 shown in light grey. The changing orientation of the GCN4 α-helical axis is marked by a purple arrow. A. Structure near beginning of simulation (30 ns of GAL11-ABD1/GCN4-cAD _aMD_no1). GCN4-F124 occupies Pocket #1 formed by GAL11-Y220 and -K217 (among other residues not shown here), while GCN4-W120 engages in hydrophobic interactions with the non-polar section of the GAL11-K221 side-chain. B. Structure after 120 ns simulation (GAL11-ABD1/GCN4-cAD _aMD_no1). GCN4-F124 has moved towards GAL11-M173, freeing up Pocket #1 for GCN4-W120 to move in while maintaining hydrophobic engagements with GAL11-K217 and -K221. C. Structure after 220 ns simulation (GAL11-ABD1/GCN4-cAD _aMD_no1). GCN4-F124 binds now to GAL11-M173 and -V170, and Pocket #1 is now fully occupied by GCN4-W120.
Fig 8.
α-helicity of coactivator-bound and free GCN4-cAD.
The trajectories of four aMD simulations (GCN4_aMD_no1 to no4) were combined to allow comparisons across the entire range. The secondary structure is color-coded (pink: α-helix; dark blue: 310 helix; turquoise: turn; white: coil; yellow: β-sheet). The amino acid sequence is on the vertical scale (N-terminus at top; the position of key residues is marked by red dotted lines) and time in nanoseconds of aMD on the horizontal scale. A. Secondary structure analysis of the GCN4-cAD bound to GAL11-ABD1. Except for aMD_no1 (GCN4_aMD_no1), which shows the presence of 310 helices at the beginning of the simulation, all other aMDs are almost exclusively α-helical in the region including the bulky hydrophobic residues W120, L123 and F124. There is evidence for relatively stable N- and C-terminal borders (indicated by dotted dark blue lines at residues S117 and D125, respectively), especially in aMD_no2 and 3 for the central helix ("Helix #1") that encompass W120, L123 and F124. In addition, there is evidence for the presence of an additional transient α-helix ("Helix #2") surrounding residues M107, F108, Y110 and L113. B. Secondary structure analysis of de novo folded GCN4-cAD. There is a widespread formation of short-lived α-helices at different positions and lengths that include residues W120, L123 and F124.
Table 2.
Sequences of cAD-like motifs used in MD simulations.
Fig 9.
Secondary structures of cAD-like07 and cAD-like96.
The complete sequence of the simulated polypeptide chain is displayed vertically on the left of each graph (Panels A, B and D). The horizontal axis represents time of aMD simulation in nanoseconds. Horizontal red dotted bars mark the positions of the conserved hydrophobic residues. A color code for secondary structure is shown beneath Panel D. A. Formation of α-helical structures during aMD simulation of cADlike07, displaying poor transactivation potential. B. Formation of α-helical structures during aMD simulation of cADlike96, with the highest observed transactivation potential. Note the more extensive formation of stable α-helices and their long-term stability in cADlike96 in comparison to cADlike07. Also, while the N-terminal boundary of the helical structures in cADlike96 is relatively sharply defined (encompassing two aspartic acid residues), the boundary in cADlike07 is much less stable and extends further N-terminal (proline residues). A comparison of the two graphs shows that there are major differences between the two cADlike domains in the extent of α-helical content and stability, as well as the position and stability of the boundaries of these helices. C. Post hoc correlation of α-helicity with transactivation potential. The α-helical propensity (vertical axis) of the cADlike sequences described in Warfield et al. (Table S2 in [42]) was predicted with the Agadir algorithm (http://agadir.crg.es [44]) and plotted against their corresponding transactivation potential (-fold induction of ARG3 mRNA; black dots). The red-dotted line marks the first-order linear regression (r² = 0.89) and the black lines demarcate the 95% prediction interval (including all data points). D. Secondary structure analysis of cAD-like96 bound to GAL11-ABD-1. The trajectories from four independent one microsecond aMD simulations (GAL11-ABD1/cAD-like96_aMD_no1 to no4; Table 1) were combined. The cAD-like96 displays stable α-helicity and helical boundaries. The conserved bulky hydrophobic residues (W94, L97, and F98; highlighted with red horizontal dotted lines) are more than 99% embedded with an α-helical context.
Fig 10.
Molecular mechanics analysis of the GAL11-ABD1/cAD-like96 interaction.
The decomposition of the van der Waals contribution is shown as a contour plot. The horizontal axis represents the amino acid sequence of the cAD-like96 activation. The vertical axis represents snapshots at 1 ns intervals from the four aMD simulations. The ΔG value of the van der Waals contribution of each residue at each time point is color-coded according to the scale shown (substantial contributions to ΔG are green and dark blue). The data derived from independent simulations (indicated on the left; aMD_no1 is represented by frames 1–1000, aMD_no2 by frames 1001–2000 etc.) are shown on the same plot to facilitate the detection of constant and variable features. Residues making significant energetic contributions are highlighted by a red-dotted line aligned to the amino acid sequence.
Fig 11.
Conformational analysis of the GAL11-ABD1/cADlike96 complex.
The same quantitation criteria as described in the legend for Fig 4 were applied. A. Examples of different helical orientations of cADlike96 relative to the GAL11-ABD1 interaction surface. B. Quantitation of the helical orientation.