HIV-1 Tat Binding to PCAF Bromodomain: Structural Determinants from Computational Methods

The binding between the HIV-1 trans-activator of transcription (Tat) and p300/(CREB-binding protein)-associated factor (PCAF) bromodomain is a crucial step in the HIV-1 life cycle. However, the structure of the full length acetylated Tat bound to PCAF has not been yet determined experimentally. Acetylation of Tat residues can play a critical role in enhancing HIV-1 transcriptional activation. Here, we have combined a fully flexible protein-protein docking approach with molecular dynamics simulations to predict the structural determinants of the complex for the common HIV-1BRU variant. This model reproduces all the crucial contacts between the Tat peptide 46SYGR(AcK)KRRQRC56 and the PCAF bromodomain previously reported by NMR spectroscopy. Additionally, inclusion of the entire Tat protein results in additional contact points at the protein-protein interface. The model is consistent with the available experimental data reported and adds novel information to our previous structural predictions of the PCAF bromodomain in complex with the rare HIVZ2 variant, which was obtained with a less accurate computational method. This improved characterization of Tat.PCAF bromodomain binding may help in defining the structural determinants of other protein interactions involving lysine acetylation.


2
. Multiple sequence alignment of the existing 53 PCAF human bromodomain (BRD) sequences as obtained from bromodomain PROSITE family, accession number PS50014 [2][3][4]. The human PCAF BRD corresponding to the PDB structure 1JM4 is the top sequence. Blue residues in the alignment denote residues that have percentage of identity over 60%. The figure was generated using JalView [5].

S1. MD Simulation of AcK50-BRU Tat full-length
The of AcK50-BRU Tatfull-length model was inserted in a simulation box of 0.12 nm length containing ~9,300 TIP3P water molecules [6] and 11 Clcounterions for neutralizing the system. Periodic boundary conditions were used and a cut-off at 12 Å [7] was adopted for short-range non-bonded interactions. The particle mesh Ewald summation method [8] was used for long-range electrostatic interactions. A dielectric constant of 1 was assumed. All chemical bonds were constrained by using the SHAKE algorithm [9]. The equations of motion were integrated using a time step of 1fs. Temperature and pressure were kept constant at 300 K and 1 atm by coupling the system to external baths [10] with Langevin thermostat (coupling constants st = 0.05) and Berendsen barostat (coupling constants sp = 0.5 ps), respectively. The AMBER force field ff99SB [11] was used. The acetylated lysine topology parameter was adapted according to Machado et al. [12].
After 5,000 steps of minimization (steepest descent algorithm for the first 1,500 steps before switching to the conjugate gradient algorithm for the remaining 3,500 steps), 10 ns of MD simulation were performed. NAMD software [13] was used.
We clusterized [14] the trajectory into 7 clusters (See Section S2). For each cluster, 11 representative structures were selected (77 structures in total, grouped in 7 sets) for AcK50-BRU α Z 3 full-length Tat docking with PCAF BRD. They represent 98% of conformations ( Figure S4). Finally, 7 representatives, having at least 3 Å of backbone RMSD from each other, for a total ensemble of 77 structures were chosen. Figure S3. RMSF of full length AcK50-Tat backbone at 5-6 ns (black); 9-10 ns (red). Figure S4. MD simulation of full length AcK50-Tat in water: RMSD of the protein plotted as a function of simulated time.

S2. Clusterization Procedure
AcK50-BRU Tat full-length MD conformers were clusterized following the De Mori et al. method [45]. We performed the clustering by different cut-offs (Table S1). Cut-off 3.0 Å was chosen as described in Section S1.

S3. Validation of Docking Parameters
Different Ambiguous Interaction Restraints were used to find the best parameters for docking (Table S2).
The conformer with the lowest RMSD of residues in the interaction interface (iRMSD) respecting the NMR structure was chosen ( Figure S5). Figure S5. Cartoon representatives in seven clusters of AcK50-BRU Tat full-length , residues 46-56 are shown in red and AcK50 in sticks.
Our model reproduced almost all of the HCs present in the NMR structure [26]. The only hydrophobic contact not reproduced involved AcK50 in BRU Tat and I764 in PCAF BRD. However, in the NMR structure, PCAF BRD I764 is a high mobility residue [26], suggesting that this interaction might be not as strong as other hydrophobic interactions at the protein/protein interface. Figure S6. The structure of our predicted Tat 46 SYGR(AcK)KRRQRC 56 .PCAF complex is superimposed to that obtained by NMR [26]. AcK50 is shown in sticks. The interface between the peptide and the protein is represented as a grid. The van de Waals surface is represented in dots. Figure S7. MD simulation of Tat.PCAF BRD in water: RMSD of the complex plotted as a function of simulated time. Table S3. Hydrophobic contacts observed for the representative structures of clusters 2, 3 and 4 from docked structures in comparison with in vitro experimental works for Tat 46 SYGR(AcK)KRRQRC 56 (synthesis) [15] or in vivo for full-length acetylated Z2 Tat with PCAF BRD [16]. (black = agreement; red = new contacts). The representative structure of cluster 1 is discussed in the main text. Coverage in the last column is the percentage of the occurrence of HCs over the representative structures. Y761D AcK50 AcK50 25% no effect [15] W746A Absent Table S4. Selected intermolecular HBs in the representative BRU Tat.PCAF models by docking. The residues with asterisk form new HBs. Coverage in the last column is the percentage of the occurrence of HCs over the representative structures.

S5. Tat in Complex with Cellular Partners: The P-TEFb Case
Structural information on full length HIV-1 Tat in complex with cellular partners is not available. Complexes exist instead for Tat peptide [15,[17][18][19] with the length ranges from 9 amino acids [17] to 49 amino acids [19] (Figure S9). HIV-1 Tat peptide is always in an unfolded state [15,17,18], with exception of the complex with P-TEFb, where HIV-1 Tat adopts an α-helix [19]. The same docking protocol discussed in the main text was successfully applied also for Tat.P-TEFb complex ( Figure S8 and Table S5). The best structure from the most populated cluster after docking reproduced well most of the HCs and HBs comparing to X-ray structure (PDBID 3MI9 [19]), with backbone RMSD of 1.8 Å. Figure S8. Molecular representation of Tat.P-TEFb model and interactions. HCs are shown in the dot lines. Residues are in sticks with oxygen and nitrogen atoms in red and blue, respectively. (a) Superposition of Tat in complex with P-TEFb X-ray structure, PDBID 3MI9 [19] (red, orange, grey for Tat, CDK9 and CycT1, respectively) with Tat.PTEFb model (magenta, green, cyan for Tat, CDK9 and CycT1, respectively). The backbone RMSD for the two complexes is 1.8 Å. (b) Intermolecular HBs between Tat (magenta) and CDK9 (green) in Tat Table S5. Intermolecular HBs and hydrophobic contacts between Tat.P-TEFb model comparing to Tat and P-TEFb in X-ray structure. The residues with star are not consistent with HBs found in X-ray structure [19].  Figure S9. Molecular representation of HIV-1 Tat in complex with cellular partners [15,[17][18][19]. The PDBIDs are shown in the bar with the starting and ending position of Tat (red) in the different complexes.